commit f57ce6ef9ca735d5cb428b2f12f1f9413a70506a Author: Chris Wilson Date: Wed Aug 21 12:20:29 2013 +0100 2.21.15 release commit ed40a7c3de3bbb178278c05907e59239712b98b6 Author: Chris Wilson Date: Sun Aug 18 14:33:44 2013 +0100 sna/gen6+: Tweak semaphore avoidance for composite operations Signed-off-by: Chris Wilson commit 4486ae2d829781e32652bce84c08e63ee1960bf0 Author: Chris Wilson Date: Sun Aug 18 10:11:49 2013 +0100 sna: Do not force use of the CPU bo if replacing a busy GPU bo To minimise waiting for the SHM pixmap afterwards. Signed-off-by: Chris Wilson commit f1cbf13bb1ad7432d681ef7cbbc17bcf2d23c554 Author: Chris Wilson Date: Sat Aug 17 23:37:57 2013 +0100 sna/video: Initialise the UV offsets to 0 for packed formats Signed-off-by: Chris Wilson commit db41bec0fd4dc8eec1f44a988f1c5a531594604f Author: Chris Wilson Date: Sat Aug 17 23:33:17 2013 +0100 sna/video: Return the planar offset calculations to the planar branch They where accidentally move the packed branch in commit 85e89f2121bad96d34ff8df9456e2fbaa9ff7881 Author: Chris Wilson Date: Fri Aug 16 21:11:33 2013 +0100 sna/video: YUV420 is not supported by sprites, replace it with a RGB passthrough Signed-off-by: Chris Wilson commit fed9ca236a0b5bbd7f1d137aa4ef9c2caf26719c Author: Chris Wilson Date: Fri Aug 16 22:43:26 2013 +0100 sna/video: Turn off passthrough if given a zero name Signed-off-by: Chris Wilson commit 18e274425d728baa62445dc091cae6b1eab94a89 Author: Chris Wilson Date: Fri Aug 16 21:58:29 2013 +0100 sna: Make sure that external scanouts are immediately discarded An issue with passthrough Xv buffers is that they end up in the scanout cache and potentially reused instead of being immediately discarded. Signed-off-by: Chris Wilson commit 6692077aca2dcb7754b23b6404ff2db2f70228dc Author: Chris Wilson Date: Fri Aug 16 21:55:16 2013 +0100 sna/video: Keep a ref to the passthrough overlay bo Otherwise we will destroy it at the end of the frame whilst it is still meant to be shown. Not normally an issue as the next frame is show before it vanishes, but is if the image is shown for an extended period of time. Signed-off-by: Chris Wilson commit d8c9b2c85256c870f9677a590f190856826c821c Author: Chris Wilson Date: Fri Aug 16 21:47:25 2013 +0100 sna/video: Mark the sprite color key as changed when disabling Otherwise we will forgot to send the command to turn off the colorkey. Signed-off-by: Chris Wilson commit 85e89f2121bad96d34ff8df9456e2fbaa9ff7881 Author: Chris Wilson Date: Fri Aug 16 21:11:33 2013 +0100 sna/video: YUV420 is not supported by sprites, replace it with a RGB passthrough As YUV420 is not supported by any of the current sprite implementations drop it. Instead implement some RGB passthroughs. Signed-off-by: Chris Wilson commit c610d37dae7e025027fd09a6889bf94853af8b8c Author: Chris Wilson Date: Fri Aug 16 18:22:45 2013 +0100 sna/video: Expose an attribute to keep the overlay always on top Signed-off-by: Chris Wilson commit 7a6bd55da90382459db072bffbd4aa36d18f9a17 Author: Chris Wilson Date: Fri Aug 16 15:38:18 2013 +0100 sna/video: Don't assert on a client error, return BadAlloc If the client sends a BO that is too small to satifsy the PutImage, then reject with BadAlloc rather than die with an assert. Signed-off-by: Chris Wilson commit cb35a0b19a4a1c9ac4e2bb552a61057c215ce890 Author: Chris Wilson Date: Fri Aug 16 08:47:16 2013 +0100 uxa: Restore the lost copyright block after splitting i965_reg.h from i915_reg.h Reported-by: "Lim, Siew Hoon" Signed-off-by: Chris Wilson commit 445f205a18f7548aa97b04379ff66a4fafb3f79c Author: Chris Wilson Date: Thu Aug 15 00:28:25 2013 +0100 sna: Clip the tiled copy during framebuffer resize If we use the BLT for copies, then we need to supply correctly clipped coordinates. Signed-off-by: Chris Wilson commit 6bd897157b34e3b80dea68fa615d0fba08238486 Author: Sedat Dilek Date: Wed Aug 14 10:01:13 2013 +0200 Correct misspelled caching Signed-off-by: Sedat Dilek commit 5d29daa7df72d9a96e044d0706f3014811389629 Author: Chris Wilson Date: Wed Aug 14 18:16:22 2013 +0100 sna: Make sure the frontbuffer exists before doing pitch checks An unusual path to be sure, to call sna_crtc_set_mode_major before we create a GPU bo for the scanout - but might be possible after a GPU hang, or it appears after trying to set a 0x0 mode. At any rate, make sure the GPU bo exists before dereferencing. Bugzilla: https://bugs.launchpad.net/ubuntu/+source/xorg-server/+bug/1212344 Signed-off-by: Chris Wilson commit a0f90a4c79799780592c004c846a963aad5f28e6 Author: Chris Wilson Date: Wed Aug 14 18:12:15 2013 +0100 sna: Rename sna_pixmap_get_bo() Mark how dangerous this function is by giving it a __ prefix. Signed-off-by: Chris Wilson commit 2be9118e882f59591dc0fb9b98ec25494f020bb4 Author: Chris Wilson Date: Wed Aug 14 17:48:13 2013 +0100 sna: Tile the old image onto the new framebuffer when enlarging This improves the transition from going from a mirrored fbcon framebuffer to an extended mode. Signed-off-by: Chris Wilson commit 394978867edf3746c285491db6be41abfe325d87 Author: Chris Wilson Date: Wed Aug 14 08:38:52 2013 +0100 sna: Check that we successfully install the required hosting midlayer Signed-off-by: Chris Wilson commit 3d818c0a36919add462c05d01a64b297f693fa69 Author: Chris Wilson Date: Wed Aug 14 08:34:46 2013 +0100 sna: unwrap old modesetting API whilst hosted If we are hosted, then we do not have our own modesetting backend and calling it results in fireworks. So leave the old modesetting entry points for the hosting midlayer to control. Signed-off-by: Chris Wilson commit 504b18b079e0a12a9d18b46418411805ef1d8377 Author: Chris Wilson Date: Tue Aug 13 11:01:03 2013 +0100 sna: Fix the assertions for the composite fb path Having updated the offsets for the dst, we also need to update the assertions. Signed-off-by: Chris Wilson commit 5840bfe285e3fd9dc550cbe5fa87437870c92038 Author: Julien Cristau Date: Mon Aug 12 19:31:53 2013 +0200 Link the driver against pixman commit b2df6cc8f0bcac6311f2b1d0c008add9b7d05787 Author: Julien Cristau Date: Mon Aug 12 14:45:04 2013 +0200 configure: use CPPFLAGS for xaa.h check, not CFLAGS CFLAGS aren't passed to the preprocessor, which then can't find the header. Signed-off-by: Julien Cristau commit d661332de946c356c10a75e29a48f2c2e3599fec Author: Julien Cristau Date: Mon Aug 12 14:38:22 2013 +0200 Check for struct sysinfo as well as Non-linux glibc systems have the latter but not the former. Signed-off-by: Julien Cristau commit 5775731b6489c63212914163f51b7a576df70658 Author: Chris Wilson Date: Mon Aug 12 22:24:47 2013 +0100 sna/glyphs: Do not attempt to release the negative cache lookup Following commit 021c59f6243c51871f664e5fdd43aa00a809033a Author: Chris Wilson Date: Sat Aug 10 13:15:48 2013 +0000 sna/glyphs: Micro-optimise glyph drawing we need to be careful during sna_glyph_release() that the atlas points into the glyph cache before releasing it. Fortunately caught by an assertion. Reported-by: Jiri Slaby Signed-off-by: Chris Wilson commit 8da8c88e999d74fb110dddd497f03854adc089cc Author: Chris Wilson Date: Mon Aug 12 13:12:21 2013 +0100 intel: Check for HW_SKIP_CONSOLE before use For the older xserver. Signed-off-by: Chris Wilson commit 4abc9c317b1f79f3f3d687fcf2d255b4d7416c2c Author: Chris Wilson Date: Mon Aug 12 13:10:05 2013 +0100 configure: Correct the spelling of XORG_TESTSET_CFLAGS Reported-by: Julien Cristau Signed-off-by: Chris Wilson commit ce666270c8861e039a06718cc3e9093d63d15187 Author: Chris Wilson Date: Mon Aug 12 13:07:03 2013 +0100 sna: Fix use of obsolete this_atlas in a DBG message Signed-off-by: Chris Wilson commit b3cb8b209d0515c54e1126f8ed6adf3cf7fbf3bd Author: Chris Wilson Date: Mon Aug 12 13:05:00 2013 +0100 configure: Fix reversed check for XORG_TESTSET_CFLAG Signed-off-by: Chris Wilson commit 9645e71cb2580adb78bea1cfa70dfc3f66630834 Author: Chris Wilson Date: Mon Aug 12 11:28:03 2013 +0100 sna: Reorder composite fallback migration to handle src==dst || mask==dst Order is important when moving and marking damaged regions. Signed-off-by: Chris Wilson commit e8dfc5b3f4ffeec93e52a5319b5a3118edf0e94e Author: Chris Wilson Date: Mon Aug 12 10:33:41 2013 +0100 sna: Fix destination offset along memcpy composite fallback fastback The application of dst_x|y was incorrect, and so the drawing could end up in the wrong location for a window. References: https://bugs.freedesktop.org/show_bug.cgi?id=66990 Signed-off-by: Chris Wilson commit 8d48a43ed956ff21d779aff1a11a0f7bf6b65bcb Author: Chris Wilson Date: Sun Aug 11 11:49:04 2013 +0100 sna/gen7: Refine ring selection Don't force us to select BLT too early if we allow ring switching. As the RENDER ring benefits from more cacheing over time (e.g. HSW:GT3e) it becomes much more preferable to use it over the BLT. Since we already have the logic to decide if ring switching is possible/preferred, relax the initial checks on where the current activity is to allow switching between batches. Signed-off-by: Chris Wilson commit 6f3999f38c71689298ac44e5b64830f754199740 Author: Chris Wilson Date: Sat Aug 10 19:44:00 2013 +0100 sna: Avoid confusing conflicting CRTC probes vs a DRM race Only bail we fail to find the CRTC for this output amongst the allowed set for this ZaphodHead - but allow us to find an invalid CRTC. Signed-off-by: Chris Wilson commit 6cd366b79ad15be608f829e884b4880c311cf58b Author: Chris Wilson Date: Sat Aug 10 17:55:07 2013 +0000 sna: Remove the redundant clearing of kgem at startup struct kgem is contained within a cleared superstruct. Signed-off-by: Chris Wilson commit b364305b16d79e32592235004872d38406ce8e9d Author: Chris Wilson Date: Sat Aug 10 13:41:46 2013 +0000 sna/glyphs: Micro-optimise glyphs-to-dst Signed-off-by: Chris Wilson commit 021c59f6243c51871f664e5fdd43aa00a809033a Author: Chris Wilson Date: Sat Aug 10 13:15:48 2013 +0000 sna/glyphs: Micro-optimise glyph drawing When you don't have many cycles to play with, every one counts. Here we make sure we cache negative lookups for large glyphs. Signed-off-by: Chris Wilson commit 4d5483c253a9a29f54f2eed9b40623c8e9de6c22 Author: Chris Wilson Date: Sat Aug 10 08:59:33 2013 +0100 sna: Tune pixmap_inplace() not to use a pinned busy GPU bo for replacement Obviously we can only replace the bo if it is not pinned and so just incur a stall when we could have instead rerouted the rendering through its CPU bo. Signed-off-by: Chris Wilson commit de379943b49b94cc065fd55ad7ca1067e1802a31 Author: Chris Wilson Date: Fri Aug 9 16:37:29 2013 +0100 sna: Allow LLC to reuse any active write buffer The inplace write distinction is not important on LLC, so pick any buffer that is on the GPU and available for reuse. Signed-off-by: Chris Wilson commit 821b34f5db2936394e8969b15c1cdd5144acf6e3 Author: Chris Wilson Date: Fri Aug 9 14:55:09 2013 +0100 sna: Squelch a "valid" mode without a dotclock There is yet another race in drm initialisation where X is starting long before the drm device is completely ready, and is being told that the output has a valid mode, but with bogus settings. Ignore it, and hope it comes to its senses later on. Signed-off-by: Chris Wilson commit b6b5c3f009585151eb772dfc2526843c7cee82b3 Author: Chris Wilson Date: Fri Aug 9 11:34:39 2013 +0100 intel: Disable incompatible features whilst hosted Start adding the infrastructure to disable direct hardware access if X is being run under a system compositor (aka "hosted"). Signed-off-by: Chris Wilson commit ab445c23079a76eaa2a7d499ec4755599012bd2f Author: Chris Wilson Date: Fri Aug 9 10:39:22 2013 +0100 sna: Rewrite the can-mmap-cpu test to handle stolen unclaimed bo Signed-off-by: Chris Wilson commit f1d3d1f8c677b938362eee8b9f4e81f5463bc16a Author: Chris Wilson Date: Fri Aug 9 10:07:59 2013 +0100 sna: Check for EINVAL after mmapping I want to catch this error as it happens to see which (stolen) bo is being erroneously mmapped. Signed-off-by: Chris Wilson commit 039a64e89e0323bd02eec24fba21a3924ee5cc31 Author: Chris Wilson Date: Fri Aug 9 09:13:10 2013 +0100 configure: Only use XORG_TESTSET_CFLAG if defined Repairs the build for f16 which has an older version of xorg-macros. Fortunately, as it doesn't define XORG_TESTSET_CFLAG it also doesn't use it generate noisy output. Signed-off-by: Chris Wilson commit 31b4110b5119ad1643f3e75c5e23a44c4ceb30fe Author: Chris Wilson Date: Thu Aug 8 22:25:57 2013 +0100 sna: Use create2 to allocate a DISPLAY bo even if we have not enough stolen Signed-off-by: Chris Wilson commit 0f82c1a451a2d5763d9cf53b48f55200f7716966 Author: Chris Wilson Date: Thu Aug 8 22:05:34 2013 +0100 sna: Always reuse scanout bo where possible When looking for an inactive scanout and all are busy, select the oldest for reuse. Signed-off-by: Chris Wilson commit bd7e653e1ee009e225b0bbf60dcaebc9b0fa2788 Author: Chris Wilson Date: Thu Aug 8 22:03:16 2013 +0100 sna: Avoid leaking stolen framebuffer bo Framebuffers created from stolen bo were not being released and so the kernel would keep the fb and bo alive, causing the memory to be remain unreusable whilst X lived and us to leak all available stolen memory. Signed-off-by: Chris Wilson commit 097669cd728e612bf3e3c531b83892e37481ee7b Author: Chris Wilson Date: Thu Aug 8 20:44:24 2013 +0100 sna/dri: Do not create a scanout-capable buffer if not flipping Avoid paying the penalty of creating an uncached buffer if we never intend to flip to it. Signed-off-by: Chris Wilson commit cbf1f97d5dadbf569c7d26e78899bda95b28b16d Author: Chris Wilson Date: Thu Aug 8 19:43:24 2013 +0100 configure: Include the experimental features in the configuration summary Signed-off-by: Chris Wilson commit fcac4f7f2b8d06cf2451609739ea342e5e3bfd5a Author: Chris Wilson Date: Thu Aug 8 14:52:26 2013 +0100 sna: Only use the experimental create2 if we support DISPLAY cacheing as well Signed-off-by: Chris Wilson commit c6add09791a44d9b3af79d50f52e6913c46341a8 Author: Raul Fernandes Date: Thu Aug 8 09:26:59 2013 +0100 Micro-optimise box intersections We can shave a few instructions off the routine by incrementally performing the "is-empty" check as soon as we compute the intersection in each dimension. commit c01c66bca2c64ae2d77233b6ccdca26431ee51b8 Author: Chris Wilson Date: Wed Aug 7 10:35:04 2013 +0100 sna: Update experimental create2 struct definition Signed-off-by: Chris Wilson commit 6004870d53221caf05a330fcfc8a72e68b2b8898 Author: Dave Airlie Date: Tue Aug 6 13:03:18 2013 +1000 configure.ac: fix dri enabled with kms only picked Otherwise we'd fail saying DRI1 wasn't possible, when that is exactly what we asked for. [ickle: The breakage was introduced with commit bd6ffd1ad2f903215d2c631252e742a0f6893054 [2.21.14] Author: Chris Wilson Date: Sat Jul 27 15:33:19 2013 +0100 configure: Print a summary of compilation options which modified the search to only take place if UMS was enabled, but missed mollifying the resulting error.] Signed-off-by: Dave Airlie commit a875ba39bea4a951e8cf1450ba9722b7cda95943 Author: Chris Wilson Date: Mon Aug 5 16:15:48 2013 +0100 sna: Rename the attribute macro __packed__ to avoid clang barfing Using __packed__ as shorthand for ___attribute__(__packed__) confuses clang as. (I guess to it expands (__packed__) which gcc skips.) As clang also uses packed in its builtins, we have to find a compromise, and so tightly_packed wins for being a more verbose description without the dangerous leading underscores. Signed-off-by: Chris Wilson commit 2536ad044b259eb3ce3e7b1ccf2c59ab3fe37d06 Author: Chris Wilson Date: Sun Aug 4 12:50:41 2013 +0100 sna: Tweak threading thresholds Signed-off-by: Chris Wilson commit 2cdd57c804706bcf558d16d3370c1880ef482af1 Author: Chris Wilson Date: Sun Aug 4 11:43:54 2013 +0100 sna/trapezoids: Some minor DBG to investigate threaded rendering Signed-off-by: Chris Wilson commit 1f7e20d812ef46fa0cba492e06a8f9f5a326d3c5 Author: Chris Wilson Date: Sun Aug 4 12:26:36 2013 +0100 sna: Define fast function attribute for old gcc or other compilers Also written by Mark Kettenis and reported by Sedat Dilek. Signed-off-by: Chris Wilson commit 0cd154039ab02799dc972d93c415e762226df1aa Author: Chris Wilson Date: Sun Aug 4 10:30:14 2013 +0100 2.21.14 release commit a5bf389908934aa4f9493a4a35b18e4aec9c8103 Author: Chris Wilson Date: Sat Aug 3 14:54:31 2013 +0100 sna: Mark the entire userptr proxy chain as flush/non-reusable Fixes cache bookkeepping when mixing userptr uploads. Signed-off-by: Chris Wilson commit 5287660aafe45859c07874c22dca99c1ff5e555a Author: Chris Wilson Date: Fri Aug 2 13:18:12 2013 +0100 sna: Reserve relocation entries for the deferred VBO Whilst we reserved exec entry slots for the deferred VBO, there were no relocation spaces reserved. So if we submitted a render command followed by a multitude of BLT copies, we could then overrun the relocation array when adding the deferred vbo to the batch. Reported-by: Danny Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=67504 Signed-off-by: Chris Wilson commit 3f04b6f33f05f8a24698f934a23462269b84917d Author: Chris Wilson Date: Thu Aug 1 14:49:36 2013 +0100 sna/dri: Disable TripleBuffering by default for compositors Ideally, the method of swapping is something that the applications have control over, along with how to synchronise to the vertical refresh. Whilst triple buffering is good to reduce jitter for games (at the cost of an extra frame of latency, usually considered a good tradeoff), it prevents the applications from accurately controlling the presentation of animations. One vocal critique is Owen Taylor, who demands accurate swap control for smooth animations in gnome-shell. For example, http://blog.fishsoup.net/2012/11/28/avoiding-jitter-in-composited-frame-display/ In lieu of application control, just apply a quirk for the compositor. Everyone else will just have to wait for DRI3. Signed-off-by: Chris Wilson commit 0ae07158c3550b418d3596cacda33a64ee653790 Author: Chris Wilson Date: Thu Aug 1 11:01:14 2013 +0100 sna: Add a few more details to the log for mode switching Signed-off-by: Chris Wilson commit 246911d742569eed6675698c1e51032ba50917d3 Author: Chris Wilson Date: Thu Aug 1 10:52:44 2013 +0100 sna: Don't force inline string-ops for the general memcpy_blt routine As we need optimal copy code for the general case, where unlike swizzling the run lengths are not known before hand, we need to call the arch specific routines from glibc. Signed-off-by: Chris Wilson commit 8174c56c3ad6f1b0e6cd432c888f3eaca91159b4 Author: Chris Wilson Date: Thu Aug 1 09:19:17 2013 +0100 sna/glyphs: Allow glyphs-to-dst to discard CA masks for a8 glyphs Signed-off-by: Chris Wilson commit c5e3b66c27e8960a39ba0fc9108e04a67e1c2d98 Author: Damien Lespiau Date: Wed Jul 31 18:50:51 2013 +0100 uxa/display: Keep the EDID blob around for the lifetime of an output xf86InterpretEDID() doesn't copy the EDID raw data in xf86MonPtr but just stores the given pointer. The DDX driver needs to make sure that data stays valid. Signed-off-by: Damien Lespiau commit 278ee155d572d3132d27823bb69995a8cc61c2c0 Author: Chris Wilson Date: Wed Jul 31 17:01:47 2013 +0100 sna: Relax readback constraint for CPU mappings of scanouts Signed-off-by: Chris Wilson commit bfe25b8110fc4d4761b57e3ec07b8467890b90a1 Author: Chris Wilson Date: Wed Jul 31 11:00:21 2013 +0100 sna: Fix interaction between WT mapping of scanouts and allocating from stolen If we allocate the scanout from stolen, we cannot then access it via the CPU - so prevent the mapping in those cases. Signed-off-by: Chris Wilson commit 493763301e995d02cb838d14348da46dd26444af Author: Chris Wilson Date: Tue Jul 30 17:38:23 2013 +0100 sna: Allow CPU access to scanouts if WT cached On Iris, we may store the framebuffer in the eLLC/LLC and mark it as being Write-Through cached. This means that we can treat it as being cached for read accesses (either by the GPU or CPU), but must be careful to still not write directly to the scanout with the CPU (only the GPU writes are cached and coherent with the display). Signed-off-by: Chris Wilson commit cdbc097796b24ab42b60c49d9885a35573c96a89 Author: Chris Wilson Date: Tue Jul 30 16:42:54 2013 +0100 sna/gen7: Set appropriate constants for Haswell GT3 GT3 has twice the number of cores and URB as GT2, and so we can use more threads and URB entries. Signed-off-by: Chris Wilson commit 9a390c77dc83909c9a05161d91678a87dd6a0d34 Author: Chris Wilson Date: Tue Jul 30 14:07:45 2013 +0100 sna/gen4: Hide an unused function Whilst we are force flushing vertexes we are not using the threaded emitter, so simply hide it from the compiler to prevent it warning about the unused function. Signed-off-by: Chris Wilson commit a3eea6e61e15a147165400c81eab82a14a9ce2fe Author: Chris Wilson Date: Tue Jul 30 13:22:54 2013 +0100 Suppress misleading warnings for -Wmaybe-uninitialized All the curent reports by gcc are false positives. Instead we have better static analysis tools at our disposal and valgrind. Signed-off-by: Chris Wilson commit 208282d31657f2b718e151397d16b307cdf8f77b Author: Chris Wilson Date: Tue Jul 30 13:19:27 2013 +0100 Suppress a bunch of compiler warnings for using constant strings We want these allocated in ro memory even if the antique API complains. Signed-off-by: Chris Wilson commit 7e3445a330340c1aa2ca5a13d838344438336749 Author: Chris Wilson Date: Tue Jul 30 11:17:52 2013 +0100 sna/dri: Tidy up applying damage to the Pixmap when copying regions Now that we are doing a true move-to-gpu first, we can drop the hacks from damage(). Signed-off-by: Chris Wilson commit c507f621c932393a8b87912f71a86e29a8c1c190 Author: Chris Wilson Date: Mon Jul 29 23:54:31 2013 +0100 sna: Suppress a few harmless warnings Mostly of the only used in debugging path variety. Signed-off-by: Chris Wilson commit cf7d36455fc43e88697f4fb8fd865f3259ba0d1a Author: Chris Wilson Date: Mon Jul 29 23:49:08 2013 +0100 intel: Suppress some extremely noisy warnings Warning about redundant declarations within the xorg headers hides genuine warnings in our own code - disable them until the headers are cleaned up. Signed-off-by: Chris Wilson commit d7465ece29c9ae665a5a9f5bdac764a19062786e Author: Chris Wilson Date: Mon Jul 29 21:55:29 2013 +0100 sna: Cleanup along all error paths in sna_pre_init() Only one out of a dozen error paths actually freed the locally allocated memory, oops. Signed-off-by: Chris Wilson commit 5751b9f1f5beb943db8cb5d260379ae39eb18db4 Author: Chris Wilson Date: Mon Jul 29 14:44:51 2013 +0100 sna/dri: Prevent stale DRI buffers from a NULL dereference If the DRI buffer is stale, the drawable may have been recreated and no longer be associated with DRI. In this case, the pixmap may not be on the GPU, so just subsitute the client's old bo and hope the it catches up and does a GetBuffers in the near future. Signed-off-by: Chris Wilson commit 8619db7224f5c3ec417a1d00a68e981549366adc Author: Egbert Eich Date: Mon Jul 29 13:52:54 2013 +0200 Add support for the dell backlight interface Signed-off-by: Egbert Eich [ickle: Add the failsafe to sna as well.] commit c9d89499806779cd6c62d5d6d34df76279cc5abd Author: Chris Wilson Date: Mon Jul 29 11:51:39 2013 +0100 sna: Composite region is already in dst drawable space So do not offset it again when processing the fallback composite operation. Regression from commit 6921abd81017c9ed7f3b2413784068fbc609a0ea Author: Chris Wilson Date: Thu Jul 18 16:21:27 2013 +0100 sna: Add a fast path for the most common fallback for CPU-CPU blits References: https://bugs.freedesktop.org/show_bug.cgi?id=66990 Signed-off-by: Chris Wilson commit 3cb75950a90c28690f78fd3bf93ff8e23e5d543a Author: Chris Wilson Date: Mon Jul 29 11:03:43 2013 +0100 uxa: Fix incorrect include path Signed-off-by: Chris Wilson commit 1dd79c225cedf5614a77e5db9f9d9700bf957bb8 Author: Chris Wilson Date: Wed Jul 24 14:06:05 2013 +0100 sna: Restore driver info upon FreeScreen Signed-off-by: Chris Wilson commit b0826907dd7844c03080421d79e5f927b1245833 Author: Chris Wilson Date: Sun Jul 28 18:32:47 2013 +0100 sna: Fix limit on stipples to only use immediate operands for <= 32 dwords The check was still based on bytes after the len had been converted to dwords. Signed-off-by: Chris Wilson commit 5e9decc963e01d6e558355af97314c502e314000 Author: Chris Wilson Date: Sun Jul 28 16:42:14 2013 +0100 uxa: Add the missing include to silence the compiler warning Silence compiler warning for no prior declaration of intel_init_scrn(). Signed-off-by: Chris Wilson commit f97b55d283e2fc333e93c88b18d5f3412f87c67b Author: Chris Wilson Date: Sun Jul 28 16:31:57 2013 +0100 intel: Remove the unused SDV PCI IDs As a first step towards working out what to do with the remaining used-once PCI IDs, delete the used-never ones. Signed-off-by: Chris Wilson commit 679416f9a5e949e6f15a5e8712921731a1750cbd Author: Chris Wilson Date: Sun Jul 28 15:45:29 2013 +0100 sna: Allow PutImage to utilise userptr for busy targets Typically, PutImage is not a performance critical path since ShmPutImage uses CopyArea and so PutImage is relegated to small one off transfers. Signed-off-by: Chris Wilson commit bd6ffd1ad2f903215d2c631252e742a0f6893054 Author: Chris Wilson Date: Sat Jul 27 15:33:19 2013 +0100 configure: Print a summary of compilation options Signed-off-by: Chris Wilson commit d92b5ddcff03a593270d060187b56ee2aa8e09b3 Author: Chris Wilson Date: Sun Jul 28 12:10:23 2013 +0100 sna/gen7: Prefer GPU spans for Baytrail as well Signed-off-by: Chris Wilson commit 0b68f1741fe360b0ba10421866aed8fffeea44e7 Author: Chris Wilson Date: Sun Jul 28 12:09:49 2013 +0100 sna/gen7: Use ivb/byt/hsw shorthand for generation checking Signed-off-by: Chris Wilson commit 0e2a1c27b7d7cd5b305fb092a27ef57ba266aebd Author: Chris Wilson Date: Fri Jul 26 15:38:20 2013 +0100 sna/gen7: Rename Valleyview to Baytrail The codename changed midcycle - along more rational lines (all the chips within the platform are now part of the Baytrail family rather than different codenames for each). Signed-off-by: Chris Wilson commit 8d10c32d650df36019d55cfc2e31d7641be4032d Author: Chris Wilson Date: Fri Jul 26 15:37:24 2013 +0100 sna/gen7: Set appropriate thread counts for Valleyview^Baytrail Signed-off-by: Chris Wilson commit a84a9fcb5e1218f685e4469dbe00f8f29cc36bf5 Author: Chris Wilson Date: Fri Jul 26 10:41:00 2013 +0100 intel: Remove the unused bridge PCI-IDs Signed-off-by: Chris Wilson commit 8afb5438f3479df39752cbda680c13d80f81f42f Author: Chris Wilson Date: Fri Jul 26 10:25:36 2013 +0100 intel: Always define the PCI-IDs If the macros conflict with another definition on the system, that is actually a useful warning in this case as it means there exists a typo somewhere. Signed-off-by: Chris Wilson commit a12c352b56e742b8f8aca4d8ca9efa26f268817b Author: Chris Wilson Date: Fri Jul 26 10:14:11 2013 +0100 intel: Remove some unused macros Signed-off-by: Chris Wilson commit e94f66c951ae9e0d4304cf8005537a147bda5d79 Author: Chris Wilson Date: Fri Jul 26 10:07:23 2013 +0100 intel: Move some backend specific macros out of the common header All the IS_GEN/IS_DEVICE are only used by the UXA backend, so move them to its headers. Signed-off-by: Chris Wilson commit 04d5c33b9677fc920ad58522f93d832a42de4744 Author: Chris Wilson Date: Fri Jul 26 10:19:43 2013 +0100 uxa: Colocate architecture source files with backend implementation sources i.e. move the toplevel uxa/*.[ch] into src/uxa/*.[ch] Signed-off-by: Chris Wilson commit 45d4e8dcf9aee37015b1ee026997ed4dabdf112e Author: Chris Wilson Date: Fri Jul 26 09:35:06 2013 +0100 uxa: Clear up the common intel directory Move all the UXA backend specifc files into their own subdirectory. Signed-off-by: Chris Wilson commit ab28526ea43728fb675448515e1519a970fb5f56 Author: Chris Wilson Date: Thu Jul 25 21:55:21 2013 +0100 intel: Only print the unique chipset names When printing out the list of supported chipsets, remove the (many times) repeated names as they are very confusing and obscure the product name you may be searching for. Signed-off-by: Chris Wilson commit 6fd37e644d9c662776aff14cd51a27b7e52472de Author: Chris Wilson Date: Thu Jul 25 22:10:43 2013 +0100 intel: Replace the codename in the user facing string for unknown Haswell parts For these parts we do not yet known the correct marketing name, so replace our use of the codename with the generic term. Signed-off-by: Chris Wilson commit 7567c99ee93f560cde33842cffa26a1ad6677600 Author: Chris Wilson Date: Thu Jul 25 21:40:02 2013 +0100 intel: Set the correct marketing names for Ironlake, Sandybridge and Ivybridge Signed-off-by: Chris Wilson commit 17873be335aea223e93170c9ba6819ab6877b86b Author: Chris Wilson Date: Thu Jul 25 21:37:05 2013 +0100 intel: Remove the SDV from the list of identified chipsets These devices are not for retail and so have no marketing names. Signed-off-by: Chris Wilson commit 4c6871684b0006a89be6760db99fda9b47790226 Author: Chris Wilson Date: Thu Jul 25 23:19:02 2013 +0100 intel: Cross-check an unnamed chipset against the list of known PCI-IDs If we recognise the PCI-ID but do not have a marketing name for it, it means we are deliberating ignoring it - presumably because it is an engineering sample. In this case, we do not want to print out a warning into the logfile so replace the "Unknown chipset" with some info about the future product. Signed-off-by: Chris Wilson commit b6bcb7ae20b92e48b8a2fa096c0e33b6b97b1b30 Author: Chris Wilson Date: Thu Jul 25 21:22:25 2013 +0100 intel: Source our PCI IDs table from the copy in the kernel Rather than duplicating the information we already use in the kernel, we can reuse the pci-id tables so long as we apply a little fuzz. Signed-off-by: Chris Wilson commit bb8484da487008df8b69beb8086167bc272e85bb Author: Chris Wilson Date: Sat Jul 27 16:28:22 2013 +0100 2.21.13 release commit 95e172b365365e9189dd863f1033bc6ea036a713 Author: Chris Wilson Date: Sun Jul 28 12:25:16 2013 +0100 sna/dri: correct DBG typo Signed-off-by: Chris Wilson commit c51cb32b6edbfc9b95cbb5e72a13ac42fdc745ed Author: Chris Wilson Date: Sat Jul 27 16:28:45 2013 +0100 test/dri2: We do not require the GLX includes Drop the GLX includes as they are not needed and obviously break if the GLX are not present on the system (and we fail to check during configure). Signed-off-by: Chris Wilson commit b4dbdd28ae05839645912f8aeb5100f6d05f11a1 Author: Chris Wilson Date: Fri Jul 26 15:49:00 2013 +0100 sna: Try the indirect upload if we elect to use a mmap and it fails This allows us to avoid the occassional kernel bug if mmap() fails and the upload is possible via our indirect upload paths. Signed-off-by: Chris Wilson commit 6f5fd772c7ca656b86394a0f036d4e0cf5b33d8e Author: Chris Wilson Date: Thu Jul 25 08:29:55 2013 +0100 sna/dri: Discard the strict checking for stale bo before performing a blit Instead of checking that the DRI2Buffers match up with the current DRI2Drawable, clip the area that is being copied to the minimum extents of the Drawable, source and destination buffers. This ensures that we never read nor write beyond the extents of the allocated or visible regions. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=67210 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=67305 Signed-off-by: Chris Wilson commit ef2a45731ef55b9fbafe5da67e0251b9b871bda9 Author: Chris Wilson Date: Wed Jul 24 19:44:15 2013 +0100 sna/dri: Restore the comparison of bottom-right extents to the pixmap origin This reverts a portion of commit 6d80bd6a73 so that we do not compare an offset redirected window against its outer frame (e.g. glxgears in a 300x300 under unity sits within a much larger ~330x330 frame). References: https://bugs.freedesktop.org/show_bug.cgi?id=67210 Reported-by: Joseph Yasi Signed-off-by: Chris Wilson commit 42862298bc4d8588433abc8ac3b1ff65ec20e30d Author: Chris Wilson Date: Wed Jul 24 10:15:09 2013 +0100 sna: Remove bogus comment concerning sna_enter_vt This bogus comment confused both Chris Halse Rogers and myself. Suggested-by: Christopher James Halse Rogers Signed-off-by: Chris Wilson commit 6d80bd6a7311926f258cd92f0b22e8d03022b62b Author: Chris Wilson Date: Wed Jul 24 09:03:45 2013 +0100 sna/dri: Cleanup validation of blit extents Prompted by a suggestion by Haihao, clarify the intent behind checking the incoming maximum blit extents against the recorded sizes of the attached bo. Due to the asynchronous nature of DRI2 invalidation, it is possible for the DRI2 buffer to be stale and for its bo to be smaller than required for the client's blit. References: https://bugs.freedesktop.org/show_bug.cgi?id=67210 Signed-off-by: Chris Wilson Cc: haihao commit 74dd10d252df2d826b6db5da7e53bd7c5d76dec0 Author: Chris Wilson Date: Tue Jul 23 09:37:15 2013 +0100 sna/gen5: The cached value of the pipelined pointers key requires 32-bits Storing only the low 16-bits of the key for the pipelined state meant that we forced an update with every new drawing op - with the side effect of flushing the render cache. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=67157 Signed-off-by: Chris Wilson commit a8c7834e1d4d31097d0dd2c931d7662b5314a9b2 Author: Chris Wilson Date: Mon Jul 22 22:03:17 2013 +0100 configure: Supply a default value for dridriverdir Rather than fail to load DRI2 if not found. Signed-off-by: Chris Wilson commit 8d94f030c6af64f82d495a835a3b962480c068e7 Author: Chris Wilson Date: Mon Jul 22 22:01:22 2013 +0100 configure: Fix enabling of DRI2 After setting it to false, we then need to set DRI2 to true after we detect the headers. Signed-off-by: Chris Wilson commit 0498e0c38d0b04d89020e61c13b250150ec19c3a Author: Chris Wilson Date: Mon Jul 22 19:20:16 2013 +0100 sna: Fix DBG compilation Signed-off-by: Chris Wilson commit e8cd5b5ff6f6e78260fb57f4b8171682d857e880 Author: Chris Wilson Date: Mon Jul 22 18:47:36 2013 +0100 configure: Use --enable-dri to control both DRI/DRI2 enabling Signed-off-by: Chris Wilson commit 3e2a1be13914e9ba13aaca06576a4f0e0f6e8fb0 Author: Chris Wilson Date: Mon Jul 22 19:34:32 2013 +0100 sna: Bail if we fail to find the attached CRTC during probing In a Zaphod configuration, the set of CRTCs available for an output is limited, and if that set does not match with the already established linkage, our quick probe function will select no mode. However, since the output is already on, the user definitely would like to continue using it, so fallback to InitialConfiguratio to select an appropriate mode. Reported-by: Nick Bowler Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=67176 Signed-off-by: Chris Wilson commit 7b1a5024df96070bab70744ad7e7d78a41fb0f72 Author: Chris Wilson Date: Mon Jul 22 10:51:05 2013 +0100 sna/gen5: Rework the flush after blend state changes QA complained that the full flush to memory was too much of a performance hit, so let's try a lighter weight non-stalling pipe flush without. References: https://bugs.freedesktop.org/show_bug.cgi?id=67157 Signed-off-by: Chris Wilson commit 2737aaab77cc6233826077b90d001fe45efc376c Author: Chris Wilson Date: Sun Jul 21 20:42:37 2013 +0100 sna: Missing git add to fix typo in assert() from last commit I915_TILING != I915_TILING_Y Signed-off-by: Chris Wilson commit fa961ec99a67266687a1934eea24138cf2f7ad19 Author: Chris Wilson Date: Sun Jul 21 18:00:22 2013 +0100 sna: Allow linear inplace uploads along the tiled X PutImage blt paths Mostly for symmetry with GetImage, this in theory just bypasses a lot of intermedate state checking. Signed-off-by: Chris Wilson commit ff705331da62729e302fb882755a88d756d9c233 Author: Chris Wilson Date: Sun Jul 21 09:47:35 2013 +0100 sna: Unwind BLT setup if we don't emit any ops Often we may find that the BLT operations we wish to do after a XY_BLT_SETUP are either completely clipped out or transparent and so do not generate any subsequent commands. This leaves us with many unused XY_SETUP_BLT. References: https://bugs.freedesktop.org/show_bug.cgi?id=61477 Signed-off-by: Chris Wilson commit 2c78403cb56841f342692b97bb53e995c3ff6b9f Author: Chris Wilson Date: Sat Jul 20 21:55:16 2013 +0100 sna: Remember to apply drawable offsets for composite memcpy Yet another missing chunk from commit 6921abd81017c9ed7f3b2413784068fbc609a0ea Author: Chris Wilson Date: Thu Jul 18 16:21:27 2013 +0100 sna: Add a fast path for the most common fallback for CPU-CPU blits I had the composite offsets, but not the normal window offsets! Reported-by: F.Brown Reported-by: Jiri Slaby Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=67124 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=67064 Signed-off-by: Chris Wilson commit e9823ec2a9f81b2e1c78eb91b695847535b0d0af Author: Chris Wilson Date: Sat Jul 20 20:54:58 2013 +0100 sna: Release cloned pixmap when updating tearfree scanout Signed-off-by: Chris Wilson commit b27b1654145c62493614dd7545dcd9fe6936750e Author: Chris Wilson Date: Sat Jul 20 20:04:00 2013 +0100 sna: Relax assertion that CPU syncs are performed on reffed objects Along some paths, we use a CPU mapping on a temporary object that never gets a user ref - so the assertion is bogus. A lesson I keep forgetting. Signed-off-by: Chris Wilson commit f39d9ed0929f83e65dde8da8f03fae6ba8264e4a Author: Chris Wilson Date: Sat Jul 20 18:08:22 2013 +0100 sna: Correct typo in checking src extents before performing memcpy A regression in commit 6921abd81017c9ed7f3b2413784068fbc609a0ea Author: Chris Wilson Date: Thu Jul 18 16:21:27 2013 +0100 sna: Add a fast path for the most common fallback for CPU-CPU blits And add the full set of assertions to validate the memcpy. Signed-off-by: Chris Wilson commit f915d4a224eb686ad18b62ca0258e056b4a483cc Author: Chris Wilson Date: Sat Jul 20 15:26:04 2013 +0100 sna: Inform the upper layers that we turn the CRTC on after a modeset Signed-off-by: Chris Wilson commit 3f52a89046999c02ed778777eff8e2eb62662d90 Author: Chris Wilson Date: Sat Jul 20 12:16:23 2013 +0100 sna: Fix application of composite offsets along PutImage fallback path Signed-off-by: Chris Wilson commit a998a8a34961ccdcec14369486ae3332009a1276 Author: Chris Wilson Date: Sat Jul 20 02:01:25 2013 +0100 sna: Tidy fast source clip processing for sna_do_copy Signed-off-by: Chris Wilson commit 65a09436e6aef9fe0518e322f92e601d5d3f46c1 Author: Chris Wilson Date: Sat Jul 20 00:20:09 2013 +0100 sna/dri: Fix the strict blitting check not to assume the dst is the front buffer In order to allow the case for copying onto a target other than the front buffer, such as the fake front buffer for example. Signed-off-by: Chris Wilson commit e386ba86ea487a2db62d80a0e60f176e052d6406 Author: Chris Wilson Date: Fri Jul 19 15:22:08 2013 +0100 sna: Perform an explicit check against the region extents RegionNotEmpty() is only valid if we only use the Region API, and as we mix direct operations on the region extents, we need to also do our own final check. Signed-off-by: Chris Wilson commit de472d95affcad0f87685782d29a41fa2a0cada6 Author: Chris Wilson Date: Fri Jul 19 15:13:50 2013 +0100 sna: Check for bpp>=8 before attempting to use memcpy_blt fastpath Fixes regression in commit 6921abd81017c9ed7f3b2413784068fbc609a0ea Author: Chris Wilson Date: Thu Jul 18 16:21:27 2013 +0100 sna: Add a fast path for the most common fallback for CPU-CPU blits Signed-off-by: Chris Wilson commit 4aa0288f16f03723a7aaa30967809d10fdb15a27 Author: Chris Wilson Date: Fri Jul 19 14:10:02 2013 +0100 sna: Return true from get_drawable_deltas() if the pixmap is offset Signed-off-by: Chris Wilson commit 6921abd81017c9ed7f3b2413784068fbc609a0ea Author: Chris Wilson Date: Thu Jul 18 16:21:27 2013 +0100 sna: Add a fast path for the most common fallback for CPU-CPU blits This path will mostly be upload for individual glyph uploads, for which the malloc overhead is significant. Signed-off-by: Chris Wilson commit 4da830864983ffe482388e4e4cfee02d106c895e Author: Chris Wilson Date: Thu Jul 18 15:05:04 2013 +0100 sna: Remove the duplicated pimxap migration for the composite fb path Signed-off-by: Chris Wilson commit 0d56f0eb84cc80fa5497c06a7967650aa20ebe51 Author: Chris Wilson Date: Thu Jul 18 14:49:41 2013 +0100 sna: DBG controls to turn off unwinding partial bo Signed-off-by: Chris Wilson commit 494f4bdcb00dc54f84e503b49fd85c8965f7f9ed Author: Chris Wilson Date: Thu Jul 18 14:19:16 2013 +0100 sna: Allow operation inplace when wedged There are times, such as rendering into the scanout, where continuing to use the GTT is preferrable even when wedged. Signed-off-by: Chris Wilson commit fb058de4e617d7e5058674859993ec635a8d779e Author: Chris Wilson Date: Thu Jul 18 13:37:12 2013 +0100 sna: Treat a source with a CPU bo as being attached. Signed-off-by: Chris Wilson commit 4723a730f4dc2d007577f43fe232c49b305c2695 Author: Chris Wilson Date: Thu Jul 18 13:26:36 2013 +0100 sna: Discard overwritten operations before doing a BLT composite Signed-off-by: Chris Wilson commit b2f32373815e166fc6dceb477139ff2ef27db0ba Author: Chris Wilson Date: Thu Jul 18 12:33:10 2013 +0100 sna: Tidy a few DBG regarding cached uploads Signed-off-by: Chris Wilson commit 1b37a167d9caa7868427cb88b2480f3b64e96cc9 Author: Chris Wilson Date: Thu Jul 18 10:51:42 2013 +0100 sna: Only IGNORE_CPU for blt composite operations if the size is known Some operations we do not know the true extents and so check the whole drawable when considering placement. In this case, the drawing may only partially cover the drawable and so we can not simply ignore existing CPU damage. Signed-off-by: Chris Wilson commit 2d62f7c483a999c24c744c0d257c8524f6fd5d32 Author: Chris Wilson Date: Thu Jul 18 22:00:24 2013 +0100 sna: Also do exposure checking after source clipping in sna_do_copy Hopefully a final regression from: commit 07926bfe507071a3d46a2ec13bb86a36bc225761 Author: Chris Wilson Date: Thu Jul 11 15:28:55 2013 +0100 sna: Remove the temporary region allocation from sna_do_copy References: https://bugs.freedesktop.org/show_bug.cgi?id=67055 Signed-off-by: Chris Wilson commit 2c8c4626908987e1bb5b16c49bc16dabac7d89dc Author: Chris Wilson Date: Fri Jul 19 11:55:09 2013 +0100 sna/dri: Return early is the DRI2CopyRegion is not attached to the GPU This can happen if the DRI client passes in a stale DRI2Drawable - that is the Drawable now references a new Pixmap which the client has not run DRI2GetBuffers against. Signed-off-by: Chris Wilson commit e5036fb15983a1d3547eaab381c18b3edf4310b4 Author: Chris Wilson Date: Fri Jul 19 11:51:40 2013 +0100 sna/dri: Reject invalid DRI2Drawables if STRICT_BLIT is defined Simply reject any attempts to copy using stale references (i.e. the DRI2Drawable has changed structure but the client hasn't yet noticed). Signed-off-by: Chris Wilson commit a45714c9a5ba5cd35fbbee2a49f540fe9164cdfd Author: Chris Wilson Date: Fri Jul 19 11:21:28 2013 +0100 sna/dri: Rearrange some more DBG to come before its assertion Again, useful to know what the assertion fails with. Signed-off-by: Chris Wilson commit e9831b50b58c36af3d2dd196b3abbfaa8e9a22c6 Author: Chris Wilson Date: Fri Jul 19 10:59:03 2013 +0100 sna/dri: Expose the refcnts in DBG before we hit the assertions Useful to try and explain some failures. Signed-off-by: Chris Wilson commit 6c8b15d321044d4a81cb187cc5e1ac094eb82367 Author: Chris Wilson Date: Wed Jul 17 20:58:33 2013 +0100 sna: Wrap cpuid.h More our ifdef out of line from the main code into a header file, where we can also apply a little bit of synatic sugar. Signed-off-by: Chris Wilson commit 08cacc265da8c1213d8bf1107168e37d38a3641d Author: Chris Wilson Date: Wed Jul 17 14:22:16 2013 +0100 sna: Guard cpuid for use only with gcc-4.4 and later Since __cpuid_count() was only introduced into gcc-4.4, we obviously cannot use it with earlier versions or with compilers that do not provide compatible interfaces. Signed-off-by: Chris Wilson commit aa5f14b3400614531aa88eb6b70b64bd1f47974d Author: Chris Wilson Date: Wed Jul 17 14:12:29 2013 +0100 sna: Ensure that buffer_size is a power-of-two for use in ALIGN ALIGN() expects it alignment argument to be a power of two. As buffer size depended upon cache_size, which is not always a power of two, issues could arise with unexpected buffer sizes. Signed-off-by: Chris Wilson commit a764a6e69b23f644957cf3e4e98868464f458758 Author: Chris Wilson Date: Wed Jul 17 10:51:56 2013 +0100 sna: Fix typo in computing box intersection Comparing y2 against y1 for the intersection was a silly typo, especially as the routine for computing the intersection already existed. Fixes regression in commit 34c9b759fbab8d548108e954d55de38c6f5bec31 Author: Chris Wilson Date: Tue Jul 16 19:39:37 2013 +0100 sna: Note that borderClip region may be more than a singular box Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66991 Signed-off-by: Chris Wilson commit 7f76a2bf319f59d463a1f96974b03d7c651847dd Author: Chris Wilson Date: Wed Jul 17 10:22:17 2013 +0100 sna: Fall back to /proc/cpuinfo parsing if cpuid cache size probe fails Older hardware does not support cache size probing via cpuid4, so we need to implement the older algorithm which requires a table based lookup. (And in hindsight, why I thought cache probing via cpuid to be quite hairy.) For the moment, just use the value found in /proc/cpuinfo. Reported-by: Oscar Dario Trujillo Tejada Reported-by: Ferry Toth Signed-off-by: Chris Wilson commit 3f3071c6c91c3907cb0fecf2f69a995feefe7956 Author: Chris Wilson Date: Wed Jul 17 09:02:19 2013 +0100 sna: Set a minimum size for tiled IO And assert that we never ask for a zero-sized upload. Signed-off-by: Chris Wilson commit 0952e7e04167d7b2284fc794ed9fb30afebdaf52 Author: Chris Wilson Date: Wed Jul 17 08:24:19 2013 +0100 sna: Include linear GTT maps when deciding whether a bo is mappable If the linear bo is still in the CPU domain, we can map it through the CPU with no penalty, so treat it as mappable. Signed-off-by: Chris Wilson commit 34c9b759fbab8d548108e954d55de38c6f5bec31 Author: Chris Wilson Date: Tue Jul 16 19:39:37 2013 +0100 sna: Note that borderClip region may be more than a singular box If the child is obscured, then borderClip will contain a list of valid boxes rather a singular extents. I thought this was covered by the clipList, but I was wrong. Reported-by: Jesse Barnes Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66970 Signed-off-by: Chris Wilson commit 1323abcc2eb19df81a000d4a37af6949801a9076 Author: Chris Wilson Date: Tue Jul 16 09:02:01 2013 +0100 sna: Make the assumption bpp>=8 explicit inside the bo size calculation Use an assert to document that we have a few rounding issues if bpp < 8. Signed-off-by: Chris Wilson commit 4a447c514b3193897e0d3ec4b3cf75f346d7438e Author: Chris Wilson Date: Tue Jul 16 08:58:21 2013 +0100 sna: Create a pixmap in ordinary system memory for depth=1 uploads Since we will not sample depth=1 pixmaps from the GPU, we may as well directly allocate these in system memory and avoid tickling the upload cache. This then avoids an issue within the size calculation code which makes the assumption that bpp>=8. Reported-by: Jiri Slaby Signed-off-by: Chris Wilson commit 884dd896e4e959e5a284a6f70da9e5cde1bbb7e9 Author: Chris Wilson Date: Mon Jul 15 10:59:39 2013 +0100 sna: Change message context to PROBED for initial configuration Since we are retreiving the hw values rather than choosing a default for ourselves, it is more consistent to use PROBED rather than INFO for our message. Signed-off-by: Chris Wilson commit f8677a971bdbeb8e472b74385780c41c7a51c522 Author: Chris Wilson Date: Mon Jul 15 10:24:53 2013 +0100 sna/gen4: Allow rendering to Y-tiled surfaces We no longer seem to be hitting the same random hangs, so presumably another w/a is taking effect. Signed-off-by: Chris Wilson commit e83e7e60f41cb29132d5eef60af16b65b1b3cd4e Author: Chris Wilson Date: Sun Jul 14 09:31:39 2013 +0100 2.21.12 release commit 0230ef72cbb2d3bcf7157548ddd52a71c0c771c5 Author: Chris Wilson Date: Fri Jul 12 23:57:42 2013 +0100 sna: Skip copying to the intermediate target if we will completely overwrite it Occasionally when forced to use an intermediate destination surface, we know that we will completely overwrite the contents of the surface and so we can forgo the initial copy from the target. Signed-off-by: Chris Wilson commit 2b16eaefb38b4a6efe50a397e68793ab9a1cb2ef Author: Chris Wilson Date: Fri Jul 12 23:22:59 2013 +0100 sna: Avoid integer overflow when computing the tiled extents When stepping over the tiles, beware that the step may overflow the int16_t region extents, so use an int for the calculation. Reported-by: Andrew Woodward Bugzilla: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1200766 Signed-off-by: Chris Wilson commit 835ceccb27afa18704c34dc3283476340f999317 Author: Chris Wilson Date: Fri Jul 12 23:20:56 2013 +0100 sna/gen5: Add some more DBG Signed-off-by: Chris Wilson commit 39f1954f667bf20cd0c25dfa5fb5267a3f9eb113 Author: Chris Wilson Date: Thu Jul 11 21:57:02 2013 +0100 sna: Replace parsing of /proc/cpuinfo with parsing of cpuid Courtesy of a patch from Chad Versace via Ben Widawsky, actually digging through CPUID for the cache info looks quite easy in comparison to the fragile approach of parsing a linux specific file that may or may not be available. Signed-off-by: Chris Wilson commit c604d1426cc11f9799044710acf1ef7d226d2604 Author: Chris Wilson Date: Thu Jul 11 16:40:51 2013 +0100 sna: Remove the duplicated open-coding of SetScreenPixmap Signed-off-by: Chris Wilson commit 07926bfe507071a3d46a2ec13bb86a36bc225761 Author: Chris Wilson Date: Thu Jul 11 15:28:55 2013 +0100 sna: Remove the temporary region allocation from sna_do_copy Signed-off-by: Chris Wilson commit 519d0007d4add23b14b5a9fd83cf4a1b68c63d03 Author: Chris Wilson Date: Wed Jul 10 18:24:36 2013 +0100 sna: Remove incorrect asserts for checking CPU mappings (after remapping) These assertions were checking that the previous state prior to performing the new mapping was consistent. Given that the checks were occurring after the update in mapping, the asserts were bogus. Signed-off-by: Chris Wilson commit a18f4ba9033cec3cc435c29878b12ceef1defc1f Author: Chris Wilson Date: Wed Jul 10 11:56:50 2013 +0100 sna: Ofast was introduced with gcc-4.6 Thomas Jones reported that the build was failing with gcc-4.5 due to the memcpy routines requesting an unsupported optimisation mode (-Ofast) and supplied this patch to only enable Ofast for gcc-4.6+ Reported-by: Thomas Jones Signed-off-by: Chris Wilson commit ccf0fdd56dc6efd30e425209d753ed4f90562d76 Author: Chris Wilson Date: Tue Jul 9 20:41:51 2013 +0100 sna: Only flush after the BLT operation if we have more than 2 distinct bo In order to preserve the optimisation of discarding incomplete batches, we don't always want to immediately submit the batch after inserting the first command. As we currently only cancel a batch if it only touches the bo being discarded, we can skip the immediate flush if it only accesses one bo and maybe be able to use the undo optimisation later. Signed-off-by: Chris Wilson commit d935912d9c13ec8cf4f641c55846714d4e9ba929 Author: Chris Wilson Date: Tue Jul 9 20:38:32 2013 +0100 sna: Free the source reference after performing a BLT composite Fixes regression from commit 8751c0f5ad202850bdd56dbb4eedb211c023cf23 Author: Chris Wilson Date: Fri Jul 5 17:55:10 2013 +0100 sna: Flush blt copies if no operations pending Reported-by: Andreas Reis Reported-by: Mike Lothian Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66742 Signed-off-by: Chris Wilson commit b202008f8f6470f256d3b9000a9f78b6ca8d932b Author: Chris Wilson Date: Fri Jul 5 13:50:34 2013 +0100 sna: Experiment with a new ioctl to create buffers from stolen memory If there is stolen memory reserved by the BIOS, we want to utilize it in preference to regular system memory. However, given the caveat that it is not suitable for CPU access, rules out most use cases - but it is a good match for framebuffers. Signed-off-by: Chris Wilson commit dbb585b9d2510efa9f8432bc25842fa14a8e495e Author: Chris Wilson Date: Sun Jul 7 09:17:58 2013 +0100 sna/gen4: Remove custom max flush vertices w/a from video path This should now be superseded by using the common w/a. Signed-off-by: Chris Wilson commit 2a59eadf8165bd70780ac16220456c6196ac3ff1 Author: Chris Wilson Date: Sun Jul 7 09:13:49 2013 +0100 sna/gen4: Remove the glyph mask hack and tune the flush w/a Remove the hack from the glyph path to force the use of an auxiliary channel, and reduce the maximum amount of inflight vertices until we can then render glyphs with no corruption (at least in my test case). Signed-off-by: Chris Wilson commit c9c1833c30421e1461f6b03b858bbc187068eb0c Author: Chris Wilson Date: Sun Jul 7 08:58:52 2013 +0100 sna/gen4: Tighten FORCE_FLUSH w/a to spot amalgamation of primitives Inspect whether this rectangle will be added to the previous primitive and so charge it against the current number of inflight rectangles. Signed-off-by: Chris Wilson commit 368c909b29758f996dbbdbec4d471df23f60bc04 Author: Chris Wilson Date: Sat Jul 6 22:27:44 2013 +0100 sna/gen4: Restore the flush-every-vertex w/a This is an abhorrent workaround for some internal GPU brokenness. A slight refinement since earlier times is the recognition that 16 is a magic number limiting the maximum number of inflight rectangles through the GPU. References: https://bugs.freedesktop.org/show_bug.cgi?id=55500 Signed-off-by: Chris Wilson commit 1fbf47ec1bed888c1b7c8dcbeb01d8e067727c82 Author: Chris Wilson Date: Sat Jul 6 18:30:47 2013 +0100 sna: Tune inplace hints for CPU operations with GPU targets Signed-off-by: Chris Wilson commit 5aaab9ea0310d48bb1a1ca20308d1c9721a9de3f Author: Chris Wilson Date: Sat Jul 6 15:10:22 2013 +0100 sna: Relax assertion that the source of the cloned pixmap cannot be mapped I was being overzealous at the time of making the COW and trying to be sure that we would never write through a mapping. Then I started to allow clones to be mapped (for reads) and missed relaxing this assertion. Reported-by: Jiri Slaby Signed-off-by: Chris Wilson commit 8751c0f5ad202850bdd56dbb4eedb211c023cf23 Author: Chris Wilson Date: Fri Jul 5 17:55:10 2013 +0100 sna: Flush blt copies if no operations pending More work to try and keep the GPU busy. Signed-off-by: Chris Wilson commit f5529ea61b5e12e0c2d9fda817948ca0a4cab212 Author: Chris Wilson Date: Thu Jul 4 21:39:33 2013 +0100 sna: Use a stack allocated PixmapRec for the fbcon copy Signed-off-by: Chris Wilson commit ea508c177c961ba2f00129476a22a32ff3ea6f1b Author: Chris Wilson Date: Thu Jul 4 20:19:20 2013 +0100 sna: Set 1024x768 fb in absence of any connected devices No actual initial configration magic is required, all we need to do is set the initial framebuffer size with no connected outputs and leave it to the core to select CompatOutput() the like. Signed-off-by: Chris Wilson commit d36c9542d2bd707838a87c451bf76f091aaf5cba Author: Chris Wilson Date: Thu Jul 4 10:45:09 2013 +0100 sna: Fix gamma query to not request uninitialized values Oops, surprising that ebf4517a0b had any positive effect. Signed-off-by: Chris Wilson commit 59fa53707eed18e5cc151bd2ce05a2dcdd325630 Author: Chris Wilson Date: Thu Jul 4 08:51:58 2013 +0100 sna: Set the initial gamma from the attached monitor configuration Fixes a missed configuration option from commit 8a6a21bff86100144ba7960fc32a299ac54ada83 [2.21.11] Author: Chris Wilson Date: Wed Jun 26 13:29:48 2013 +0100 sna: Use the existing configuration for initial modes Signed-off-by: Chris Wilson commit ebf4517a0bb510e406795c169aa21b7cc615ed31 Author: Chris Wilson Date: Wed Jul 3 22:44:46 2013 +0100 sna: Always read back the gamma, even for a inactive CRTC Even if the CRTC is inactive, we still need to initialise the gamma tables. Reported-and-tested-by: Timo Kamph Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66563 Signed-off-by: Chris Wilson commit 4def5ecf6ebc2c6dc05e60d4595fa3dd4b7e38d2 Author: Chris Wilson Date: Wed Jul 3 21:56:54 2013 +0100 sna: Tidy up the user override enums Since we have a second place dependent upon the hidden ordering of the output options, simplify by copying the complete enum block from hw/xfree86/modes/xf86Crtc.c Signed-off-by: Chris Wilson commit 5adca344b2f95cf2f2b2b3fabf6c99c39b64f25f Author: Chris Wilson Date: Wed Jul 3 21:45:11 2013 +0100 sna: Fix configuration user overrides on xorg-server < 1.14.99.1 Marty Jack found that the enums for the user overrides were off by one (his xorg conf was no longer being applied). This is because I had missed the introduction of ZoomModes in 1.14.99 increasing all the important enum values by one. Reported-by: Marty Jack Signed-off-by: Chris Wilson commit 1445a62da8a08acd8a732176d085fd098f68cec3 Author: Chris Wilson Date: Wed Jul 3 12:59:31 2013 +0100 intel: Add an option for forced rediscovery of output status on startup Specifying Section "Device" Option "ReprobeOutputs" "true" EndSection will restore the old behaviour of scanning each output on startup and picking a spanning mode. The behaviour was changed in commit 8a6a21bff86100144ba7960fc32a299ac54ada83 [2.21.11] Author: Chris Wilson Date: Wed Jun 26 13:29:48 2013 +0100 sna: Use the existing configuration for initial modes Please do notify us of any circumstances that force you to use this flag. References: https://bugs.freedesktop.org/show_bug.cgi?id=66494 Signed-off-by: Chris Wilson commit 2ecff61d369d9f67c6e28d407941f35ed6195156 Author: Chris Wilson Date: Wed Jul 3 12:45:10 2013 +0100 intel: Retire Option "RelaxedFencing" The kernel bugs have long since been fixed and should have been propagated to all stable kernels long ago. Now there should be no need to workaround those bugs. Signed-off-by: Chris Wilson commit c957f4171791447f1869f11fd724dbe92e6a8b91 Author: Chris Wilson Date: Tue Jul 2 23:59:02 2013 +0100 sna: Try harder to do the BLT upload along the fallback path Only in the !fallback path will we try using the render pipeline after the blt, in which case we can try using render rather than forcing a stall. In the fallback path, we are going to incur stalls and readbacks, anyway so ignore them when considering blt. Signed-off-by: Chris Wilson commit 158095ff2b2d38f54ca91d2a728f919cd705ff62 Author: Chris Wilson Date: Tue Jul 2 23:17:32 2013 +0100 sna/gen2: Fix alpha replication in the copy pipeline When copying into an a8 surface we need to replicate the result into the green channel. It helps to tell the GPU from where to source the value to be replicated. Signed-off-by: Chris Wilson commit 2b0d36c63817c08990b8ae4efd60df4a11619e0f Author: Chris Wilson Date: Tue Jul 2 23:16:54 2013 +0100 sna: Assorted DBG This DBG was useful for narrowing down the issue in the next patch... Signed-off-by: Chris Wilson commit 103672869c64a1e42022069f413478c256c0ee00 Author: Chris Wilson Date: Tue Jul 2 20:52:23 2013 +0100 sna: Add the condition that 855gm cannot GTT map Y-tiled surfaces We had the check in a couple of places, but missed a key one that decided whether or not to perform a GTT mapping of a bo. Signed-off-by: Chris Wilson commit 9aa665ae33c3ac54b8896e6f8373dfda2637e392 Author: Chris Wilson Date: Tue Jul 2 15:53:38 2013 +0100 sna: Only preserve the real fbcon upon first starting X We only want to preserve the plymouthd splash screen for flicker free start up, a subsequent regeneration should be cleared instead. Signed-off-by: Chris Wilson commit 21e29e95265331f8a33cd676c2d3124007643135 Author: Chris Wilson Date: Tue Jul 2 15:25:46 2013 +0100 sna: Simplify validation of active CRTCs Atter a modeset or KMS takeover, we do a quick readback of the kernel state in order to verify that it matches our expectations. If we find that a foreign framebuffer is attached, or no mode if set on the output, we then turn off that connection and release any resources associated with that pipe. This patch tries to reduce the number of superfluous requests to turn off a connection. Signed-off-by: Chris Wilson commit 67a6a4bfd9cd1a861911e76096923bfc652189e2 Author: Chris Wilson Date: Tue Jul 2 14:27:03 2013 +0100 sna: Hook into crtc_notify rather than ModeSet ModeSet is called after updating each CRTC, unlike crtc_notify which is called after applying all changes. The last is what we need as if we are called too early we detect that the next CRTC doesn't match our expectations and so we disable it, right before applying the desired mode. References: https://bugs.freedesktop.org/show_bug.cgi?id=66494 Signed-off-by: Chris Wilson commit c361b449cc3ec15819883afc220aad8823c0072d Author: Chris Wilson Date: Tue Jul 2 09:49:48 2013 +0100 sna: Include connector status in the initial probe We cannot simply rely on connector->encoder->crtc status as with the introduction of Haswell or SDVO we may have multiple connectors using the same encoder. So we need to explictly check the connector status first, before determining if the output is connected to an active encoder and CRTC. References: https://bugs.freedesktop.org/show_bug.cgi?id=66488 Signed-off-by: Chris Wilson commit 282507af8e932e49b31d5f20958a61c137df590d Author: Chris Wilson Date: Tue Jul 2 09:34:57 2013 +0100 intel: Move the validation of the KMS device into the open routine Currently we leak the fd should we open the device node and decide that is not a GEM/KMS kernel driver. The simplest way to perform the cleanup upon failure is to move the checking for GEM/KMS into the device open routine. Signed-off-by: Chris Wilson commit 24a7bec7faa7dcc4393ef7f66939ae0fb8acdf36 Author: Chris Wilson Date: Mon Jul 1 22:43:02 2013 +0100 sna: Minor tweaks to make DBG compile again Signed-off-by: Chris Wilson commit ad0afda3fe4f5e408e3610d8b76fdc7d1af33138 Author: Chris Wilson Date: Mon Jul 1 22:26:29 2013 +0100 sna: Fix checking the dirty boxes I forgot how insane the data structure for the list of dirty boxes attached to the damage is. It is neither a simple list, nor does not store the count of boxes within each chunk. Fixes regression from commit 9026bb954646c0425360c2236e26c79d097142cd [2.21.11] Author: Chris Wilson Date: Fri Jun 28 15:59:17 2013 +0100 sna: Inspect the dirty boxes when querying whether damage contains a rectangle A side effect is that we now make sure that there is an upper bound to the amount of searching we do for the no-reduce fast path. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66430 Signed-off-by: Chris Wilson commit 1c8a33a72e29261d6bf5a6c160765cbafa4d0b88 Author: Chris Wilson Date: Mon Jul 1 17:41:41 2013 +0100 sna: Allow scanouts to be untiled if need be Fixes regression from commit 77fa8ab08b441934218ddb8f690be1a919f0ec64 [2.21.11] Author: Chris Wilson Date: Tue Jun 25 22:25:25 2013 +0100 sna: Free just-allocated bo if we fail to set-tiling on CREATE_EXACT Signed-off-by: Chris Wilson commit 9cb6756ce16491e42c6e21bc74a5f7de5432510f Author: Chris Wilson Date: Mon Jul 1 17:27:17 2013 +0100 sna: Always create the clear Picture As we unconditionally use it irrespective of whether we then call accelerated code paths or not. Fixes regression from commit dc18eaa585c36c8e5f5b4ec405a976a835fd2ac3 [2.20.10] Author: Chris Wilson Date: Sun Aug 12 10:34:10 2012 +0100 sna: Make the failure to create render caches non-fatal in the event we need to start with a hung GPU. Signed-off-by: Chris Wilson commit c6c6ae71bedce43ff6617179c015c303cbaee9ae Author: Chris Wilson Date: Mon Jul 1 17:00:17 2013 +0100 sna: Make sure we are consistent in applying USE_INPLACE debug option A few new paths were missing the debug check, and clarify a few indirect uses by performing the explicit check. Signed-off-by: Chris Wilson commit cce2351056eda1282a521b96f90ba12436e86492 Author: Chris Wilson Date: Mon Jul 1 16:13:47 2013 +0100 sna: Debug options for forcing mmap syncs Signed-off-by: Chris Wilson commit 7aff6beb37c91df85c46372d44ae4a5fefae5ae8 Author: Chris Wilson Date: Mon Jul 1 17:57:49 2013 +0100 sna: Improve the message about where to find the hang state Search the few canonical locations for our hang state so that we can be more explicit to the user about what to include. Signed-off-by: Chris Wilson commit e0e84e6686be390ef204117525d4a5c6aebbbe07 Author: Chris Wilson Date: Mon Jul 1 16:49:50 2013 +0100 sna: Mark an inplace getimage as preferring the CPU mapping afterwards Signed-off-by: Chris Wilson commit e648c1f39cc17d39eff077639f23687894724b7b Author: Chris Wilson Date: Mon Jul 1 15:27:38 2013 +0100 sna: Assert that we do not have overlapping damage Double check that the results of the box query match with the slow queries. Signed-off-by: Chris Wilson commit 7c6cc0c9edb8f1ebcc8be5edb78c9d858dce1965 Author: Chris Wilson Date: Mon Jul 1 15:04:58 2013 +0100 sna: Check whether the query box is contained within the damage We can improve our query for whether the rectangle of interest is within the damage by checking whether the damage extents contains the entire box rather than just checking for an overlap. Signed-off-by: Chris Wilson commit 64cb923588df0547880009bbeaeecb7ac3ee8ecf Author: Chris Wilson Date: Mon Jul 1 14:30:21 2013 +0100 sna: Experiment with cpu mappings for migration Signed-off-by: Chris Wilson commit 12c013d25ed0c08f748b0a7250cbcef1cded26b0 Author: Chris Wilson Date: Mon Jul 1 13:28:57 2013 +0100 sna: Do not use userptr for GetImage on unsupported architectures If the system cannot blt to a CPU buffer, we should even attempt to do so for GetImage. Signed-off-by: Chris Wilson commit 7d9163983ea2e960c0a7b55266fcc532b9c6e382 Author: Chris Wilson Date: Sun Jun 30 15:26:57 2013 +0100 2.21.11 release commit 5005bd2d52ab64cbeae099d512d0b65be6c4abaa Author: Chris Wilson Date: Sun Jun 30 15:02:19 2013 +0100 intel: Fix failure code for reporting !drmCheckModesetingSupported The new function returns the fd, not a Bool, so the error code must now be -1. Signed-off-by: Chris Wilson commit f8738d7b4cc1c624d4390ef9ce7426ba457d7dd3 Author: Jonathan Gray Date: Sun Jun 30 19:37:45 2013 +1000 intel: replace direct ioctl use with drm{Set, Drop}Master Use drmSetMaster/drmDropMaster instead of calling the ioctls directly. Fixes compilation on OpenBSD where these ioctls aren't defined. Signed-off-by: Jonathan Gray commit 40301e6d03f6e8d2d2d01e6bb9f1754a7e543a08 Author: Chris Wilson Date: Sun Jun 30 11:12:34 2013 +0100 sna: Store the path used to open the device and pass to DRI Avoid having to search the device tree once again in order to simply recover the path we used to open the device. Signed-off-by: Chris Wilson commit 17da58f904e75d434aaf71e297e15d41153ba954 Author: Chris Wilson Date: Sun Jun 30 11:01:49 2013 +0100 sna: Replace conflicting drmDropMaster Calling drmDropMaster twice along the CloseScreen path is not a good idea. Signed-off-by: Chris Wilson commit 3a787da7e888da7e9943be94bd1cb177fe1495ab Author: Chris Wilson Date: Sat Jun 29 21:57:20 2013 +0100 sna: Allow tiled memcpy on i386 With the split into per-swizzle functions, and with the forced optimisation levels, it appears that i386 doesn't suffer so badly and the tiled memcpy are a viable method. Signed-off-by: Chris Wilson commit 1d9941a7c003587c0bd732fb8b21fee5cefa6f87 Author: Chris Wilson Date: Sat Jun 29 21:56:13 2013 +0100 sna: Add the Ofast option to the critical memcpy routines Always enable gcc to fully optimize the core memcpy routines (provided that optimisations are not entirely disabled, for instance for debugging). Signed-off-by: Chris Wilson commit 84c190db33142f3c1ec347ec0bf87f77ce132d36 Author: Chris Wilson Date: Sat Jun 29 19:06:40 2013 +0100 sna: Fix get_image_inplace to use the pixmap offset The inplace routine assumed that the region to be read was already in pixmap coordinates. Making it so makes the code easier, so do it. Signed-off-by: Chris Wilson commit c7d246ba6f750ee080c38ccc5603d01fcf7fce92 Author: Chris Wilson Date: Sat Jun 29 16:31:34 2013 +0100 sna: Move the clone discard into free-gpu Rather than peppering the discard manually before the call to free the GPU bo, always discard the COW when we actually free the GPU bo. Signed-off-by: Chris Wilson commit 6ab2a3acf71b5204c399c7649e5601c93a99f25f Author: Chris Wilson Date: Sat Jun 29 15:04:09 2013 +0100 sna: Improve checks for coherent access through CPU mappings Refactor the CPU mapping tests to a single function, and remember to test for a pending GPU write (i.e. bo->exec). Signed-off-by: Chris Wilson commit 9026bb954646c0425360c2236e26c79d097142cd Author: Chris Wilson Date: Fri Jun 28 15:59:17 2013 +0100 sna: Inspect the dirty boxes when querying whether damage contains a rectangle This helps in the cases where we have subtracted a small number of rectangles from an all-damage pixmap (such as a number of successive GetImage, PutImage operations). The danger is that we end up searching a long list of dirty boxes - maybe just search the first chunk if that becomes noticeable? Signed-off-by: Chris Wilson commit d635e05c9dd26a397ccf958be091b56d1075e923 Author: Chris Wilson Date: Fri Jun 28 19:24:06 2013 +0100 sna: Promote assert(!priv->mapped) along migration paths With the advent of the more permissive mapping schemes and finer damage tracking, we are more liable to have pixmaps mapped and so can reach the upload path with the pixmap still mapped. Signed-off-by: Chris Wilson commit 2d40500851a7c6d857b17258e4989ddf7401cfbc Author: Chris Wilson Date: Fri Jun 28 15:12:13 2013 +0100 sna: Add asserts around applying clears Signed-off-by: Chris Wilson commit e3ad737ef9d33e924b206741949d59224bfef566 Author: Chris Wilson Date: Fri Jun 28 15:33:54 2013 +0100 sna: Use inplace CPU mapping readback for GetImage on linear buffers Signed-off-by: Chris Wilson commit 626b5e541663f838475eaef2c1bc3ae4d4848165 Author: Chris Wilson Date: Fri Jun 28 14:24:10 2013 +0100 sna: Add debug control for disabling accelerated GetImage Signed-off-by: Chris Wilson commit 2356579cdff36adf58fb69894f646a6e63053a15 Author: Chris Wilson Date: Fri Jun 28 13:18:00 2013 +0100 sna: Assert that the kernel tiling mode matches our bo Signed-off-by: Chris Wilson commit e979d32bb71fef7341ceb9c2b2e80c6dfa50a7b3 Author: Chris Wilson Date: Fri Jun 28 10:59:23 2013 +0100 sna/gen2+: Consider precision in render operation placement Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66297 Signed-off-by: Chris Wilson commit 541f816815e392db9e798d2f940029d82a6b2e0b Author: Chris Wilson Date: Fri Jun 28 10:14:16 2013 +0100 sna: Markup when a gradient is opaque Signed-off-by: Chris Wilson commit 34e6366f2a61c145445056e5fc6c483999c0402c Author: Chris Wilson Date: Fri Jun 28 10:03:47 2013 +0100 sna/blt: Remove a pair of leftover asserts As the variable they were inspecting was removed, the asserts are now breaking the debug build. Signed-off-by: Chris Wilson commit 0c93a0cf41cbfe88b18e9e69bd97bd75cf2404bd Author: Chris Wilson Date: Fri Jun 28 00:35:14 2013 +0100 sna: Compensate redirect drawing subrectangle inside an offset pixmap Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66249 Signed-off-by: Chris Wilson commit e33fbd6d7ff2a9efcc14974a05b5e1eb01bbce34 Author: Chris Wilson Date: Fri Jun 28 08:53:32 2013 +0100 sna/blt: Refine op placement logic for handling current source location Signed-off-by: Chris Wilson commit 48b5ac11a0737f65de2e6290308bd37017cc29a9 Author: Chris Wilson Date: Thu Jun 27 21:39:30 2013 +0100 intel: Use fcntl to try and set CLOEXEC if the open(O_CLOEXEC) fails As suggested by Arkadiusz Miskiewicz. Signed-off-by: Chris Wilson commit 9b3e5c211451ac07bd96cd997ac714bcbe1809b0 Author: Chris Wilson Date: Thu Jun 27 21:08:05 2013 +0100 intel: #ifdef O_CLOEXEC for compilation on squeeze If the system doesn't support O_CLOEXEC, then we simply can't use it. Signed-off-by: Chris Wilson commit caf43fcadb0fcb3d342543f1e7dd78ee2314a627 Author: Chris Wilson Date: Thu Jun 27 16:52:21 2013 +0100 sna: Enable memcpy_from_tiled for the IO paths Should you ever need to read back from a tiled surface and for whatever reason do not have a CPU bo to accelerate the operation, maybe we could use the manual tiling instead (as it is useful elsewhere). Signed-off-by: Chris Wilson commit b5e85e495e55e2537d305b7bebacdf6f97b66199 Author: Roy.Li Date: Thu Jun 27 14:10:14 2013 +0800 uxa: fix the compilation error with xorg-xserver <= 1.10 struct _Screen has no canDoBGNoneRoot when ABI_VIDEODRV_VERSION is less than 10.0 Signed-off-by: Roy.Li commit 41715af4d009bfcb351946ddaa3a3ea3767a1429 Author: Chris Wilson Date: Thu Jun 27 16:36:52 2013 +0100 configure: SNA supports the old Xorgs So allow it to be compiled by default for older Xorgs as well. Signed-off-by: Chris Wilson commit 7ce487617445c81f0178823de8896a2b73bbaaf1 Author: Chris Wilson Date: Thu Jun 27 16:08:43 2013 +0100 sna: Trim the large object threshold Be kinder to smaller machines by lowering the threshold at which treat an object as huge and worthy of avoiding duplication. Signed-off-by: Chris Wilson commit 31467e18d2ccdc42b0601b43b581524859de1373 Author: Chris Wilson Date: Thu Jun 27 16:07:36 2013 +0100 sna: Prefer operating inplace with a very large GPU bo As we strive to only keep one copy when working with very large objects, so try operating inplace on a mapping for CPU operations with a large GPU bo. Signed-off-by: Chris Wilson commit b615ce97ec43ea8fe02e995244c757138abcb2de Author: Chris Wilson Date: Thu Jun 27 10:45:22 2013 +0100 sna: Add a fast path for reading back from tiled X bo This is lower latency than the double copy incurred for first moving the bo to the CPU and then copying it back - but due to the less efficient tiled memcpy, it has lower throughput. So x11perf -shmget500 suffers (by about 30%) but real world applications improve by about 2x. Signed-off-by: Chris Wilson commit 6493c8c65f93ad2554c2512a07ba640e966fd026 Author: Chris Wilson Date: Thu Jun 27 10:45:13 2013 +0100 sna: Implement memcpy_from_tiled functions (for X-tiling only atm) To provide symmetry with the ability to write into an X-tiled mapping of a bo, we add the memcpy_from_tiled to be able to read back from the same bo. Signed-off-by: Chris Wilson commit 60d716b53993b08a2a00c22f523c575e62e0a18d Author: Chris Wilson Date: Thu Jun 27 14:11:00 2013 +0100 sna: Add the probed CRTC mode to the list of output modes Signed-off-by: Chris Wilson commit 263e87d5e1915e6c40fa8bc1b325a36f21f92b30 Author: Chris Wilson Date: Thu Jun 27 13:57:22 2013 +0100 sna: Set the current mode when initialising CRTCs Signed-off-by: Chris Wilson commit 2f2f439c9cfdd394ad2b9a125db51e2dba7d3ff7 Author: Chris Wilson Date: Thu Jun 27 12:58:58 2013 +0100 sna: Fake the output physical width/height from the CRTC size Signed-off-by: Chris Wilson commit 5124f35168c0cc162cf5e8bcaafdf756945f6ca9 Author: Chris Wilson Date: Thu Jun 27 12:32:24 2013 +0100 sna: Support operations inplace on CPU mappings of a region Signed-off-by: Chris Wilson commit 18ee5c20d5f8f1752b89f2377e9b9c3f1c140d53 Author: Chris Wilson Date: Thu Jun 27 11:49:33 2013 +0100 sna: We can read from cloned pixmaps inplace - so long as we don't write Signed-off-by: Chris Wilson commit 36993618b4c5d4fa1fde7800eaaa711b074bd623 Author: Chris Wilson Date: Thu Jun 27 10:48:23 2013 +0100 sna: Tweak ordering of userptr temporary mappings for uploads There are a few more circumstances where the temporary mapping is beneficial, such as pixmaps that are only shadow copies or operations that require reads on the destination. Signed-off-by: Chris Wilson commit c6ad9861f00a3166fd89a7c43d602a4c2a5bde24 Author: Chris Wilson Date: Thu Jun 27 10:59:06 2013 +0100 sna: Fix DBG printing of can_upload_tiled_x() The function changed it parameters, this DBG caller did not. Signed-off-by: Chris Wilson commit 591981a36c333cdf08a91f28fb9670a62b95d31a Author: Chris Wilson Date: Thu Jun 27 10:02:28 2013 +0100 sna: Refactor freeing gpu_bo in manual tiled upload Use the common function so that we correctly check for pinned GPU bo, and adjust the hints afterwards. Signed-off-by: Chris Wilson commit 8f6dcc15c8d6ea726a10ef5df18536f8c769291e Author: Chris Wilson Date: Thu Jun 27 09:27:57 2013 +0100 sna: Detect and handle cloned pixels for manual tiled uploads Signed-off-by: Chris Wilson commit d7be3df2fe632bbc8e4f09709cf3cf7a5ef61015 Author: Chris Wilson Date: Wed Jun 26 17:01:37 2013 +0100 sna: Se the default gamma if left uninitialized by the driver Due to a bug in i915.ko the gamma ramp is left uninitialised... Signed-off-by: Chris Wilson commit 491f4fab21df97af1dae92eaf30e76ff62920fb2 Author: Chris Wilson Date: Wed Jun 26 15:15:17 2013 +0100 sna: Correct typo s/\j/\n/ When is a newline, not a newline? When it is ajar. Signed-off-by: Chris Wilson commit a40dd0ed6db35c891710199cc9d18345ab53b397 Author: Chris Wilson Date: Wed Jun 26 14:50:19 2013 +0100 sna: Explicitly initialise the probed transform for a CRTC Signed-off-by: Chris Wilson commit 01ce3ef9f12d3b6541e166f22e1ae67033237159 Author: Chris Wilson Date: Wed Jun 26 14:26:22 2013 +0100 sna: Set RR_Rotate_0 instead of 0 as desired initial rotation Signed-off-by: Chris Wilson commit c5651254c3c152daf1ed79073779c1bed6ed0a9b Author: Chris Wilson Date: Wed Jun 26 14:18:49 2013 +0100 sna: Fallback to xf86InitialConfiguration if nothing is connected Use the default mechanism for picking a fallback output and mode if nothing is connected at X startup. Signed-off-by: Chris Wilson commit 8a6a21bff86100144ba7960fc32a299ac54ada83 Author: Chris Wilson Date: Wed Jun 26 13:29:48 2013 +0100 sna: Use the existing configuration for initial modes After checking for user overrides (preferred output modes, rotation and placement), copy the current CRTC configuration from the kernel into the desired modes. This should enable X to start without introducing any extra flicker. Signed-off-by: Chris Wilson commit 362b0dc6a3b1692b752d8f250075ccc81debfca3 Author: Chris Wilson Date: Wed Jun 26 12:04:46 2013 +0100 sna/gen4+: Fix determination of intermediate extents Complete logic fail for finding the bounding box of the boxes to be copied. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66168 Signed-off-by: Chris Wilson commit 2212468315f383d09ea17c2edac8666bdb862bc7 Author: Chris Wilson Date: Wed Jun 26 11:26:29 2013 +0100 sna: Avoid allocating a temporary if using rendercpy tiles Signed-off-by: Chris Wilson commit fc5b9a96194583c67705d7d05afc3e04e97e3338 Author: Chris Wilson Date: Tue Jun 25 22:58:31 2013 +0100 sna: Clear mapped state after performing manual tiling Signed-off-by: Chris Wilson commit 3f33abee370bb1ce60bca91f29affc62d06b0bad Author: Chris Wilson Date: Tue Jun 25 22:43:48 2013 +0100 sna: Do not force creation of CPU maps on !llc Signed-off-by: Chris Wilson commit 77fa8ab08b441934218ddb8f690be1a919f0ec64 Author: Chris Wilson Date: Tue Jun 25 22:25:25 2013 +0100 sna: Free just-allocated bo if we fail to set-tiling on CREATE_EXACT If the caller requires an exactly constructed bo, abandon the attempt if we cannot set the tiling as specified. Signed-off-by: Chris Wilson commit 273716029e5318f1670e7111be2085c950995d42 Author: Chris Wilson Date: Tue Jun 25 21:35:09 2013 +0100 sna: Rearrange tiled x upload so that damage accumulation is last Signed-off-by: Chris Wilson commit e033a28bcb37867f3c947475714e1ef45c868825 Author: Chris Wilson Date: Tue Jun 25 21:12:23 2013 +0100 sna: Optimize clears to white Signed-off-by: Chris Wilson commit b77c334a26c55d3028971897dcba21407a7a4743 Author: Chris Wilson Date: Tue Jun 25 19:08:30 2013 +0100 sna: Do not perform tiling inplace if the destination is busy Rather than waiting on the GPU stall, just use the standard mechanisms to queue the uploads. Signed-off-by: Chris Wilson commit 4b0cdf28b884755cb842d3303fc0dd9bec62ab24 Author: Chris Wilson Date: Tue Jun 25 18:09:36 2013 +0100 sna: Be explicit when creating tiled-x pixmaps for manual tiling uplaods Signed-off-by: Chris Wilson commit 98feba417c87dce020df6a6ed157e3546591d8e7 Author: Chris Wilson Date: Tue Jun 25 18:35:45 2013 +0100 sna: Drop master when closing the screen As we reacquire master when initialisation the next gen of the server, to keep the reference counting consistent we need to release our master reference on CloseScreen. Signed-off-by: Chris Wilson commit febfd388235fbe566a505d7d83733305cf344850 Author: Chris Wilson Date: Tue Jun 25 18:05:28 2013 +0100 intel: Use the correct errno value for reporting the drmSetVersion failure Signed-off-by: Chris Wilson commit 2e2c448a77ab9dce4807b159708290cd7ad26b5c Author: Chris Wilson Date: Tue Jun 25 14:22:28 2013 +0100 sna: Don't flag IGNORE_CPU for partial overwritten blocks As we set the flags for the migration extents, but only operate upon the boxes in the region, we need to be careful to be sure to migrate data outside of the boxes overwritten. Fixes regression from commit 94cbe7bf7b7acc9a7f2cb56c5e275af028f3cdc8 [2.21.10] Author: Chris Wilson Date: Thu Jun 20 19:40:44 2013 +0100 sna: Mark overwriting CopyArea as not needing the dst damage If we start passing around regions rather than extents, we coud do finergrained migration decision. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55244 Signed-off-by: Chris Wilson commit 27ebef515ebc034b9b41a0f3d26ad7e6f966305e Author: Chris Wilson Date: Mon Jun 24 13:31:25 2013 +0100 Add the nickle utility scripts to the tarball So that the tarball matches the git checkout. Reported-by: Maarten Lankhorst Signed-off-by: Chris Wilson commit c88d911d4209a8ede5ec705ad925383c69182977 Author: Chris Wilson Date: Mon Jun 24 14:30:00 2013 +0100 sna: Initialise gc->ops->PolyPoint for miWideDash The miWideDash may also call PolyPoint rather than emit spans, so make sure that we correctly prepare sna_poly_point before calling mi*. Reported-by: Tobias Brodel Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66104 Signed-off-by: Chris Wilson commit 72e8ab2b8f7aabee3bd891f08c64f5b2587da8e1 Author: Chris Wilson Date: Mon Jun 24 11:46:38 2013 +0100 sna: Remove unused brw_eu_debug.c Reported-by: Maarten Lankhorst Signed-off-by: Chris Wilson commit 6ca4ae441c61b1f425453cc443bb1d4d10d25aaa Author: Chris Wilson Date: Sun Jun 23 23:41:57 2013 +0100 sna: Check for a request to create an inactive scanout The combination is just about valid and used along some odd paths, so replace the assertion with a regular checks. Signed-off-by: Chris Wilson commit b459ab9eeba4c865f15f0b3e28af318948b4adbd Author: Chris Wilson Date: Sun Jun 23 23:00:55 2013 +0100 sna: Move the reset bo to the right list Signed-off-by: Chris Wilson commit 77ddadf0ec6b867a0638660bf06c82ad31577795 Author: Chris Wilson Date: Sun Jun 23 22:01:12 2013 +0100 sna: Clear 'clear' hint upon uploading into tiled Fixes regression from commit 53c113c3cc2f8527debc185f0819139ca8637637 [2.21.10] Author: Chris Wilson Date: Fri Jun 21 19:27:24 2013 +0100 sna: Allow PutImage to write inplace using manual tiling Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66059 Signed-off-by: Chris Wilson commit d3de043d2de4aa11a85c56faf1658ec4c4699da2 Author: Chris Wilson Date: Sun Jun 23 21:30:56 2013 +0100 sna: Discard proxy upload buffers before overwritting for PutImage Signed-off-by: Chris Wilson commit d68d24dc3727dcc281bc9b45bc605874f3a3c4f2 Author: Chris Wilson Date: Sun Jun 23 19:08:32 2013 +0100 sna: Discard overwritten incomplete operations before migrating Signed-off-by: Chris Wilson commit 60dbd5a5b6cd56b7a33eac3afebe2d94ac8579c7 Author: Chris Wilson Date: Sun Jun 23 18:51:38 2013 +0100 sna: Document swizzling for DBG Signed-off-by: Chris Wilson commit fd375da5caf34f93a4e87670bb0c70fec5b4c55c Author: Chris Wilson Date: Sun Jun 23 17:06:11 2013 +0100 sna: Allow tiled uploads to accumulate damage And for the upload to create the bo as required. Signed-off-by: Chris Wilson commit b0e383363e8707c132a8a1748757d84120a99b31 Author: Chris Wilson Date: Sun Jun 23 10:55:21 2013 +0100 sna/gen5: Elide the forced flush after a drawrect Presume that the non-pipelined drawrect is sufficient to serialise operations to the render cache. Signed-off-by: Chris Wilson commit 16fb786bd0791986fcac8b11a7e182090c5b6249 Author: Chris Wilson Date: Sun Jun 23 09:00:17 2013 +0100 sna/gen5: Force a write flush when changing blend modes Otherwise it appears that the hardware will readback from memory bypassing its render cache after a change of modes. There is probably a lot more subtly to it than this, but this appears to be a good first approximation. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=51422 Signed-off-by: Chris Wilson commit 9eabe28a4046b33772bcfe109346179b81294420 Author: Chris Wilson Date: Sat Jun 22 17:37:19 2013 +0100 Tidy version query for i915.ko Signed-off-by: Chris Wilson commit 7ec0378b64e88923ba1c4eba043488685aa3a795 Author: Chris Wilson Date: Sat Jun 22 17:30:02 2013 +0100 sna: Only open the /dev/dri/cardX device once Merge the device open in the main driver with the probing so that we can open the path explicitly passed in by the PlatformProbe and keep that fd around for the main driver and so avoid a later search. Signed-off-by: Chris Wilson commit 41badce186fe0f6e8f49e30b6c1c251027161e35 Author: Chris Wilson Date: Sat Jun 22 16:34:20 2013 +0100 sna: Add reference counting to drmMaster for Zaphod With Zaphod enabled, we will try to acquire/release the drmMaster several times on each enter/leave VT. This obviously flags an error (and drops the master too early), so track the number of references we still hold for drmMaster. References: https://bugs.freedesktop.org/show_bug.cgi?id=66041 Signed-off-by: Chris Wilson commit 81b35092c39bf6d9851ac37da4b20358405cf314 Author: Chris Wilson Date: Sat Jun 22 09:22:45 2013 +0100 2.21.10 release commit 5189ba727fe48990f00d5787c53d7b15fa1838a4 Author: Chris Wilson Date: Sat Jun 22 13:39:47 2013 +0100 sna: Refine the self-cow copy hack to reduce collateral damage Sergio Callegari found a case in LibreOffice that copied cloned pixmaps onto each other and then modified one of the pair. This was triggering the hack put in to speed up firefox scrolling, unfortunately. Reported-by: Sergio Callegari References: https://bugs.freedesktop.org/show_bug.cgi?id=65665 Signed-off-by: Chris Wilson commit 62e42de300275a668a326357d454062221714fe8 Author: Chris Wilson Date: Fri Jun 21 21:00:23 2013 +0100 sna: Determine swizzling once during initialisation and choose memcpy_to_tiled_x Signed-off-by: Chris Wilson commit 53c113c3cc2f8527debc185f0819139ca8637637 Author: Chris Wilson Date: Fri Jun 21 19:27:24 2013 +0100 sna: Allow PutImage to write inplace using manual tiling Signed-off-by: Chris Wilson commit 48028a7c923fa0d66b01e8e94d4f0742866f78ec Author: Chris Wilson Date: Fri Jun 21 14:29:43 2013 +0100 sna: Inspect availablity of render before prefering to use the GPU Signed-off-by: Chris Wilson commit 71fc83401ec8c560a89a284805e849954ea18ee4 Author: Chris Wilson Date: Fri Jun 21 14:28:38 2013 +0100 sna: Check if we may want to simply upload for a CopyArea Signed-off-by: Chris Wilson commit 7e90e522199c4d6b479554073acb33e9d82fb8cc Author: Chris Wilson Date: Fri Jun 21 14:27:42 2013 +0100 sna: Fix inspection of transfer extents for deciding transport Signed-off-by: Chris Wilson commit 94cbe7bf7b7acc9a7f2cb56c5e275af028f3cdc8 Author: Chris Wilson Date: Thu Jun 20 19:40:44 2013 +0100 sna: Mark overwriting CopyArea as not needing the dst damage Signed-off-by: Chris Wilson commit b3d1118bbee1172f72c946163a37ca4ad5feecce Author: Chris Wilson Date: Thu Jun 20 19:34:18 2013 +0100 sna: Promote the CopyArea to the GPU if it subsumes the CPU damage Signed-off-by: Chris Wilson commit 186ac909ad2099f834e7b2c513a02cd4d041327c Author: Chris Wilson Date: Thu Jun 20 15:32:30 2013 +0100 sna: Fix tiled scanout limits for Valleyview It appears that Valleyview shares the same scanout limits as gen4. Signed-off-by: Chris Wilson commit 112fc5f370f8daf267d7241ca2d669194abaa183 Author: Chris Wilson Date: Wed Jun 19 19:53:14 2013 +0100 sna: Tidy a redundant conditional for CopyArea Signed-off-by: Chris Wilson commit ba642d9ff86fcf8b0b758d65565e3b5b6293b667 Author: Chris Wilson Date: Wed Jun 19 16:27:36 2013 +0100 sna: Fix the pitch limits for gen3 scanout Signed-off-by: Chris Wilson commit 06679c1433cfb21ba925584700e2e97439fdf64a Author: Chris Wilson Date: Wed Jun 19 16:16:14 2013 +0100 Revert "sna: Apply scanout stride limits to tiling selection" This reverts commit cc08f6e0ef54744434fe0fd6d76348ee6099a62d. Prefer to keep the frontbuffer as tiled, and use an per-crtc pixmap in case the frontbuffer is not suitable as the framebuffer. commit f165d2e21358703c5f4ed302a4a57219db482a59 Author: Chris Wilson Date: Wed Jun 19 16:15:32 2013 +0100 sna: Switch to a per-crtc pixmap if the pitch exceeds scanout limitations References: https://bugs.freedesktop.org/show_bug.cgi?id=65099 Signed-off-by: Chris Wilson commit cc08f6e0ef54744434fe0fd6d76348ee6099a62d Author: Chris Wilson Date: Wed Jun 19 15:50:01 2013 +0100 sna: Apply scanout stride limits to tiling selection gen4 has a restricted DSPSTRIDE limit for tiled surfaces lower than the maximum supported size of the CRTC. So we need to double check whether tiling the scanout is supported before attempting to allocate a tiled scanout. Reported-by: Paul Donohue Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=65099 Signed-off-by: Chris Wilson commit f95ed64a11339b801d655ee173f37e01727a4a87 Author: Chris Wilson Date: Wed Jun 19 14:47:46 2013 +0100 sna: Assert that the scanout fits within the allocated buffer Signed-off-by: Chris Wilson commit 6132400365f8f467d838ed26002e74412e1c7298 Author: Jonathan Gray Date: Wed Jun 19 11:30:24 2013 +0100 configure: test for librt (clock_gettime) clock_gettime() is in libc not librt on OpenBSD so check to see if linking librt is required. Signed-off-by: Jonathan Gray commit f567f19b37d00131cda749e31c6d2dc8b52a1d62 Author: Chris Wilson Date: Tue Jun 18 22:17:11 2013 +0100 sna: Allow wedged pixmaps to be accessed via GTT Signed-off-by: Chris Wilson commit c3695c3c6b7bc13b5e642c9d92648e8228411bed Author: Chris Wilson Date: Tue Jun 18 09:41:20 2013 +0100 sna: Specialise mi calls to PolyFillRect Signed-off-by: Chris Wilson commit 40329e34b9ac65d930757d53283b7d8f4066c3a2 Author: Chris Wilson Date: Sat Jun 15 13:25:39 2013 +0100 sna: Add extra assertions to sanity check CPU access is coherent Signed-off-by: Chris Wilson commit abb522e07a2c3bf39e2587ddedc4bb8ac955c536 Author: Chris Wilson Date: Tue Jun 18 09:01:49 2013 +0100 sna: Do not sync after a failed mapping The domain update should be a redundant operation, causing needless extra transitions. Signed-off-by: Chris Wilson commit 6086abca13efdeaf9b24c9fbe425546cbe2cfb12 Author: Chris Wilson Date: Mon Jun 17 22:44:58 2013 +0100 sna: Try falling back through an upload if inplace mapping fails Signed-off-by: Chris Wilson commit a86b217934a9a16843ac17b38cd5b61393f65aa0 Author: Chris Wilson Date: Tue Jun 18 09:45:08 2013 +0100 sna: Mark the base bo as unreusable for temporary mappings Update required after forcing unaligned userptr to use proxies. Signed-off-by: Chris Wilson commit ca913e8f1e4b113af3a30cca50679436df23bef1 Author: Chris Wilson Date: Mon Jun 17 22:44:20 2013 +0100 sna: Unwrap proxies for subpage offset SHM pixmaps during sync Signed-off-by: Chris Wilson commit d54049808daafed12cf99ff6a1a7395136ab1dc6 Author: Chris Wilson Date: Wed Jun 12 19:13:12 2013 +0100 sna: Refactor scanout flushing In preparation for extending the kernel interface to be more explicit about when we need to flush a direct write to the scanout. Signed-off-by: Chris Wilson commit 7ebcf94736926d4c7bb5ba1627be0a62b15c1663 Author: Chris Wilson Date: Sat Jun 15 12:27:19 2013 +0100 sna: Be explicit in not marking an upload buffer as being on the cpu Since the upload buffer is special in that it simultaneously exists on both the CPU and GPU, we need to be extremely careful in marking the cpu hint flag and asserting it. In particular, the new stricter assert was firing in the CompositeTrapezoids fallback path. Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit cafa33a522ac634d465f24d0a1607516e65a0828 Author: Chris Wilson Date: Thu Jun 13 22:51:42 2013 +0100 sna/gen3+: Remove redundant clearing of clear hint in video playback The clear hint is correctly updated when performing the move-to-gpu and so it is being superfluously repeated by the callers. Signed-off-by: Chris Wilson commit 45ca894bce73e199d679b9f147c52ada59e40ccb Author: Chris Wilson Date: Thu Jun 13 16:47:10 2013 +0100 sna: Tighten a few asserts on pixmap migration Tracking to track down where the CPU hint may be set to true but the pixmap remain entirely damaged on the GPU. Signed-off-by: Chris Wilson commit 8f340f90f4b2f269d6308d0bd31fbc2a5f579608 Author: Chris Wilson Date: Thu Jun 13 16:08:49 2013 +0100 sna/gen7: Set sampler swizzle for video sources Otherwise the sampler on Haswell will just read all zeros when trying to playback a video. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=65699 Signed-off-by: Chris Wilson commit 5f20be82563176c45a31a9676b68a3455566bc67 Author: Chris Wilson Date: Thu Jun 13 11:18:42 2013 +0100 sna/gen2+: Promote a conditional dirty into an assertion If the target bo is not bound when we start to emit the composite state for the operation, we are screwed. Signed-off-by: Chris Wilson commit 6655cc1d9735a8433994c762583753b12e2125b4 Author: Chris Wilson Date: Wed Jun 12 18:44:27 2013 +0100 sna: Do not enlarge the read back region if there is already CPU damage We enlarge the read back into to minimise the number of times we have to migrate pixmap data. However, if we then enlarge the read so that it overwrites CPU dirty data, then we can introduce corruption. Signed-off-by: Chris Wilson commit 1f180b89db412c5fa9fe95c8f6634be33a1d8581 Author: Chris Wilson Date: Wed Jun 12 14:23:32 2013 +0100 sna/video: Free the private video (adaptor/port) arrays upon CloseScreen Signed-off-by: Chris Wilson commit 04ea469d4f30f9bb25561f3d04f1463999c8d928 Author: Chris Wilson Date: Wed Jun 12 14:20:10 2013 +0100 sna/video: Catch allocation failure whilst setting up the TexturedAdaptor Missing NULL check for calloc of the port private array. Signed-off-by: Chris Wilson commit dc8f2def5461c75fce6d648036a61186fd00d23f Author: Chris Wilson Date: Wed Jun 12 08:34:48 2013 +0100 sna: Allow the throttling to be woken by SIGINT Rather than completely starve X from handling input events and raising the spectre of miEQ overflow, stop throttling upon SIGINT. We will have waited a small period of time (~100ms) for the GPU to catch up, and will do so again in the near future. This also preserves the hang detection. Signed-off-by: Chris Wilson commit 15b92c98755c709f41e59baeb206e5a3e56e3178 Author: Chris Wilson Date: Tue Jun 11 14:23:57 2013 +0100 sna: Make sure the source is coherent on the CPU before uploading Similar to 7d91051c, but along the render path rather than CopyArea. Makes the presumption that for the upload path the region is stored on the CPU explicit. References: https://bugs.freedesktop.org/show_bug.cgi?id=61628 Signed-off-by: Chris Wilson commit 1392f296a3fc94bb64abe4c3370c2f3cd52bef9f Author: Chris Wilson Date: Tue Jun 11 15:10:06 2013 +0100 sna: Mollify an assert to allow us to call busy_ioctl on a GTT write The code was being careful in making sure that only a BO with an outstanding GPU write was being flushed. We need to relax that assertion to allow us to call busy_ioctl after only GTT writes. Signed-off-by: Chris Wilson commit 7d91051c50210560dbd93a9e36f30d9f74ce9133 Author: Chris Wilson Date: Tue Jun 11 14:23:57 2013 +0100 sna: Make sure the source is coherent on the CPU before uploading Makes the presumption that for the upload path the region is stored on the CPU explicit. References: https://bugs.freedesktop.org/show_bug.cgi?id=61628 Signed-off-by: Chris Wilson commit 2e2db4b88f741ca6cd4312c6c5242261c5de4d06 Author: Chris Wilson Date: Tue Jun 11 00:49:35 2013 +0100 sna: Tidy batch submission around CRTC updates A few places we were redundantly submitting the same bo, and in others doing more work than necessary. Signed-off-by: Chris Wilson commit b9439a56170c4732b340137d306933a742ee6ae9 Author: Chris Wilson Date: Mon Jun 10 22:56:20 2013 +0100 sna: Move the GTT dirty tracking from the pixmap to the bo This allows us to more accurately track when the GTT is dirty. However, the only danger is that we may prematurely flush the scanout and clear the dirty bit and not preserve the flush timer. Signed-off-by: Chris Wilson commit 5f0c5a49538d9995926b8dbda85c8e6d99496f17 Author: Chris Wilson Date: Mon Jun 10 22:48:03 2013 +0100 sna: s/bo->dirty/bo->gpu_dirty/ to avoid later confusion The next step is to introduce bo->gtt_dirty. Signed-off-by: Chris Wilson commit b44307753655fac985404c519a54ca65da323900 Author: Chris Wilson Date: Mon Jun 10 21:34:52 2013 +0100 sna: Make sure all sna_write_boxes* flag the GTT is dirty Better would be to start passing in the sna_pixmap so that we can mark the GTT as dirty in a single location, and only if we do indeed then do a GTT update or pwrite. Signed-off-by: Chris Wilson commit 005075de9d20fe7c2b9a6e184c912559929710ad Author: Chris Wilson Date: Mon Jun 10 14:50:47 2013 +0100 sna: Keep the cached EDID if it hasn't changed Rather than reparsing the EDID each time we probe the modes, if the property hasn't changed keep it. Signed-off-by: Chris Wilson commit 3d760c9b32776bf2f1acde9256388769a9c5a398 Author: Chris Wilson Date: Wed Jun 5 12:52:11 2013 +0100 sna: Streamline connector probing This is a tidy up in preparation for a more controversial patch. In the meantime, take advantage of information that remains static over the lifetime of our connectors. Signed-off-by: Chris Wilson commit b584162e387187a7cdb7f46a3a151e00e7d3de58 Author: Chris Wilson Date: Fri Jun 7 16:32:40 2013 +0100 sna: Set the needs_flush flag on the bo when the scanout is dirty Since the call to the busy_ioctl is predicated on whether the bo needs to be flushed by us, therefore when forcing the call to the busy_ioctl to flush the GTT we need to mark it as requiring needs_flush. Signed-off-by: Chris Wilson commit 7151e437df4437df35a67ae125607e665c66353c Author: Chris Wilson Date: Fri Jun 7 16:09:18 2013 +0100 sna: Track when we write directly to bo through the GTT In the future, we may need to explicitly flush GTT writes to the scanout, so add the infrastructure to do so now. Signed-off-by: Chris Wilson commit 79145b70a7bda2941059ce7e630f84e2b24b465a Author: Chris Wilson Date: Fri Jun 7 15:33:51 2013 +0100 sna/dri: Check rather than assert for a change in back/front sizes The DRI2 midlayer may, or may not, recreate the back buffer after a pixmap change. The result is that the back buffer may or may not be flagged as unsuitable for flipping, so be paranoid and double check. (This behaviour was first introduced in 1.9.0 and looks to be removed again in 1.15.0) Signed-off-by: Chris Wilson commit 2f91643cde954c23633c6f91940f7216796b949c Author: Chris Wilson Date: Fri Jun 7 14:50:55 2013 +0100 sna/dri: Fix transcription error in commit ab47433b20 Reversed a predicate causing it to fail to blit when it had a drawable (and worse dereferencing a NULL pointer) instead of terminating when the Drawable was gone. Signed-off-by: Chris Wilson commit e85271e516f811c60f440de61995cc5baf469a47 Author: Chris Wilson Date: Fri Jun 7 13:35:48 2013 +0100 sna/dri: Do not decouple the flip after chaining up a blit Fixes regression from previous patch. Signed-off-by: Chris Wilson commit ab47433b20a623bf790acfbf2005a8782d177f64 Author: Chris Wilson Date: Thu Jun 6 23:15:14 2013 +0100 sna/video: Select vsync by default based on "SwapbuffersWait" To be strictly identical to the DRI code, we should override the user when the option is set. However, this a baby step for simple bug testing. References: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1176647 Signed-off-by: Chris Wilson commit 7e1e18aca214a8705270972131a1800c81bc3b4d Author: Chris Wilson Date: Thu Jun 6 21:40:30 2013 +0100 sna/video: Fixup formats to select visuals Fixes regression from commit 195a51353c3af7bd253227da5f759f06cea01f73 [2.21.8] Author: Chris Wilson Date: Tue Apr 9 19:13:46 2013 +0100 sna/video: Convert to a pure Xv backend Reported-by: Edward Sheldrake Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=65479 Signed-off-by: Chris Wilson commit 0869acc0a6a5b92f5528049d2bd5ec62121ba114 Author: Chris Wilson Date: Thu Jun 6 21:21:24 2013 +0100 sna/dri: Make the fallback blit async again Previously it was async, and is likely to be anyway as the pageflip failed due to the change in output. To be safe, once again request that the copy does not stall the GPU upon the scanline window. Signed-off-by: Chris Wilson commit 9b52c852aaf42dde3a7f29c065e4d5230eac7a64 Author: Chris Wilson Date: Thu Jun 6 16:40:01 2013 +0100 sna/dri: Handle fast double flips across mode changes The tricky nature of the double flip is that there is no SwapComplete pending and so we must avoid calling it under such circumstances. Signed-off-by: Chris Wilson commit 0da5cc23b18d4e2473a3b6a571fb49d08bb1dd2d Author: Chris Wilson Date: Thu Jun 6 14:11:50 2013 +0100 sna: Fix format specifier for mismatching int/long in DBG Signed-off-by: Chris Wilson commit 2770655371954413c750b030c9a7135baee302d4 Author: Chris Wilson Date: Thu Jun 6 12:04:23 2013 +0100 2.21.9 release commit d3345cd269dc75999b35f3286d21059974dcb24b Author: Chris Wilson Date: Thu Jun 6 12:18:32 2013 +0100 sna: Only emit an error for terminal mmap failures If we repeat the mmap, keep quiet. Only emit the ErrorF in case even the repeated attempts to mmap the bo fail. Signed-off-by: Chris Wilson commit 70fa080d02bff9742523cea52c875d9447a78bfd Author: Chris Wilson Date: Wed Jun 5 23:32:01 2013 +0100 sna: Include the GT details in the backend name for a chipset Signed-off-by: Chris Wilson commit 45c09bfe58c37bbf7965af25bdd4fa5c37c0908f Author: Rodrigo Vivi Date: Wed Jun 5 13:07:13 2013 -0300 Add more correct names for Haswell. As we find out more of the final product names for Haswell chipsets, we need to update the user visible identification strings. Signed-off-by: Rodrigo Vivi commit a9f9fb8e44bb0bdbd004da1fce986585ebcafd28 Author: Chris Wilson Date: Wed Jun 5 09:48:19 2013 +0100 sna/video: Fix redundant initialisation of video->clip It will be created before use anyway, but this silences a compiler warning. Signed-off-by: Chris Wilson commit 5d0ae71f1d667759ed4cfd2b7e793a08b8e3702c Author: Chris Wilson Date: Wed Jun 5 09:12:48 2013 +0100 sna: Do not conflate ignoring an output with an allocation failure If the user explicitly ignores an output through xorg.conf, then xf86OutputCreate returns NULL. This is not to be confused with an allocation error that also returns NULL. The latter is terminal, the former is desired. Fixes regression from commit a9acc8dbb436e1e523e7d7c0f3c41946a2d0f245 Author: Chris Wilson Date: Thu May 30 12:09:52 2013 +0100 sna: Cleanup up error reporting after failure to init KMS interface Reported-by: David Rosca Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=65381 Signed-off-by: Chris Wilson commit 6dacaddb6a28670a52cead4b62c056a8acde8f3a Author: Chris Wilson Date: Mon Jun 3 15:34:37 2013 +0100 sna: Always populate the CPU features string So that we don't print random contents of the stack on ancient 32-bit CPUs with no extended instruction sets. References: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1186800 Signed-off-by: Chris Wilson commit 8a88c01c3737a3570628600b1ab51938f82e226d Author: Chris Wilson Date: Mon Jun 3 14:14:09 2013 +0100 sna: Make copying the glyph size more compact The assembly was ugly with an unwarranted AGU stall. Signed-off-by: Chris Wilson commit 66ad4d6f3c990bd40d816b4a22122bbf64786e4c Author: Chris Wilson Date: Sun Jun 2 12:13:46 2013 +0100 sna/dri: Undo any COW before performing a copy with DRI2CopyRegion Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=65250 Signed-off-by: Chris Wilson commit 9a8ff7861c50b2f8698a968a011d5d833c7d5975 Author: Chris Wilson Date: Sun Jun 2 12:13:28 2013 +0100 sna: Add some more DBG hints to copy-on-write cloning Signed-off-by: Chris Wilson commit c4ad7b14ca71b95af83864b05793ea357f48bb88 Author: Chris Wilson Date: Sat Jun 1 16:58:15 2013 +0100 sna: Prevent adding damage to an already all-damaged GPU bo Keep the bookkeeping straight after propagating the source clear colour: commit 810a0ce2f89ba6b343b5043963626c0f2d50fb22 [2.21.7] Author: Chris Wilson Date: Fri May 10 12:02:10 2013 +0100 sna: Propagate clear color when replacing by a CopyArea Reported-by: Roman Jarosz Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=65247 Signed-off-by: Chris Wilson commit be84b6929705584f30b63d6e613ff3d8e6e3dc1d Author: Chris Wilson Date: Sat Jun 1 16:51:38 2013 +0100 sna/dri: Reorder assert not to fail on a pageflip deferred to after a modeset Signed-off-by: Chris Wilson commit 2e11b520bbdb1af7e977db7fdef2d4c41242d56f Author: Chris Wilson Date: Thu May 30 15:34:08 2013 +0100 sna: Compile fix for non-debug builds sna_display.c: In function 'get_fb': sna_display.c:181:3: warning: implicit declaration of function 'assert_scanout' [-Wimplicit-function-declaration] sna_display.c:181:3: warning: nested extern declaration of 'assert_scanout' [-Wnested-externs] * sigh Signed-off-by: Chris Wilson commit 3a9975729c665cd09a33e33eab608350aaee8160 Author: Chris Wilson Date: Thu May 30 14:15:23 2013 +0100 sna: Restore GPU promotion for large fills Fix another regression from commit ef9dc6fae585d5616446eedc1e6e91173f4064c1 [2.21.7] Author: Chris Wilson Date: Mon May 20 11:08:51 2013 +0100 sna: Undo a few more overwritten operations upon a bo The REPLACES predicate was meant to complement the existing tests, not be compulsory. Found whilst investigating why performance seemed to have dropped off in Phoronix's benchmarks. Reported-by: Michael Larabel Signed-off-by: Chris Wilson commit 08b33ba8f12bb4bc425ca1bc40e2660e446e2c65 Author: Chris Wilson Date: Thu May 30 14:57:56 2013 +0100 sna: Assert that an existing scanout is the desired size Signed-off-by: Chris Wilson commit a9acc8dbb436e1e523e7d7c0f3c41946a2d0f245 Author: Chris Wilson Date: Thu May 30 12:09:52 2013 +0100 sna: Cleanup up error reporting after failure to init KMS interface Signed-off-by: Chris Wilson commit f4fcb0e7d24a104afb642dd8819e67ffcaafbfb1 Author: Chris Wilson Date: Thu May 30 12:02:44 2013 +0100 sna: Report allocations failures during Screen initialisation Just abort instead of trying to limp on if we fail to even allocate memory during initialisation. Signed-off-by: Chris Wilson commit 7f7c8087db6af58109e791763fe33582a1559646 Author: Chris Wilson Date: Wed May 29 12:25:37 2013 +0100 sna: Log which outputs are being configured during a modeset Signed-off-by: Chris Wilson commit fc1e0702b25e647cb423851fb7228989fec28bd6 Author: Daniel Vetter Date: Wed May 29 11:25:28 2013 +0100 sna: fixup up possible_clones kms->X impedance mismatch We need to be a notch more careful with remapping the possible clones mask from the kernel: - SDVO and Haswell DDI have multiple connectors mapping to the same encoder. - The connector list can be reordered compared to the encoder list. This is used to move eDP/LVDS to the front. Also add even more debug output. Signed-off-by: Daniel Vetter commit b7aa05e44b1f88284183c1c73bb074dd5dce74cb Author: Chris Wilson Date: Tue May 28 23:01:30 2013 +0100 sna: Sanity check that CRTC / output combination is valid Signed-off-by: Chris Wilson commit b507796679529b14c99e8937870561cd8eebebb9 Author: Chris Wilson Date: Tue May 28 12:13:02 2013 +0100 Add the known marketing names for the performance Haswell parts Start filling in the names for the parts that have been announced, the Iris branded Haswell GT3 parts. Signed-off-by: Chris Wilson commit b3169768ddf8c8977d2ace3675611b5ec6eba7ee Author: Chris Wilson Date: Tue May 28 12:02:56 2013 +0100 sna: Make the backend identifier more informative This is useful, for example, with the multiple gen7 variants. Signed-off-by: Chris Wilson commit 0839e3316a02380cd64d354ee0ec9d4c2b8eecf4 Author: Chris Wilson Date: Tue May 28 11:18:14 2013 +0100 sna: Call mode update after disabling outputs upon VT switch Perform a second sanity check that all the outputs we expect to be active remain so following the cleanup after VT switching. Signed-off-by: Chris Wilson commit 7d64f596b895071ea253481c97275bf551c013e9 Author: Chris Wilson Date: Mon May 27 21:13:40 2013 +0100 test: Add an easily visible tearing test for video playback Contributed by Matti Hamalainen. commit f38d037b94aacdccb2aaa058bbe48d1c8079d604 Author: Chris Wilson Date: Mon May 27 21:07:20 2013 +0100 sna/video: Correct interpretation of 'sync' The 'sync' parameter past along XvPutImage is not a request for us to perform the operation synchronised to the vblank, rather a hint that the client wishes the operation to be synchronous. In Xorg, this is implemented as an immediate ShmCompletion event so make sure we flush the operation to hw in this case. Fixes regression from commit 195a51353c3af7bd253227da5f759f06cea01f73 [2.21.8] Author: Chris Wilson Date: Tue Apr 9 19:13:46 2013 +0100 sna/video: Convert to a pure Xv backend Reported-by: Matti Hamalainen Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=65048 Signed-off-by: Chris Wilson commit 7e7d0ad15bafc4624f8c2ccf73c08ead5cc6fd6a Author: Chris Wilson Date: Mon May 27 13:41:13 2013 +0100 2.21.8 release commit 51103d8637ba34a0d5f3507e196f54333aa765bb Author: Chris Wilson Date: Sat May 25 15:53:11 2013 +0100 sna: Free the released COW reference when uncloning Reported-by: Jiri Slaby Reported-by: David Gow Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=64978 Signed-off-by: Chris Wilson commit fce4292ab7216a1c610ffc57aa6763e05b458904 Author: Chris Wilson Date: Fri May 24 15:15:13 2013 +0100 sna: Fix cut'n'paste compile error from previous commit Always check for outstanding diff before pushing. Signed-off-by: Chris Wilson commit e2e139923dd000f13602e51d94bc8a265b5e234f Author: Chris Wilson Date: Fri May 24 15:08:13 2013 +0100 sna: Log when the framebuffer changes size Signed-off-by: Chris Wilson commit 820f97ede7a38015a781de72cc265b0b65f86415 Author: Chris Wilson Date: Fri May 24 14:38:06 2013 +0100 sna: Reorder cleanup of flush list and CPU bo to silence an assertion Keep the bookkeeping in order. Signed-off-by: Chris Wilson commit 79e98ea6448f666d1a6dcca19bfe0e84fb62dd46 Author: Chris Wilson Date: Fri May 24 13:08:26 2013 +0100 sna: Reenable COW support Accidentally left it disabled whilst testing. Signed-off-by: Chris Wilson commit f6c35e58c1bb94ccfa04723db76d7164d5772f11 Author: Chris Wilson Date: Fri May 24 12:43:16 2013 +0100 sna: Do not undo a bo if we may fallback If we undo the pending operations to a bo, then fallback we cause corruption. For example, see wine and its 1-bit rendering. Fixes regression from commit 07a4400fffe4f83df3debe62abd2c37b45f8f0c0 [2.21.7] Author: Chris Wilson Date: Fri May 10 11:59:59 2013 +0100 sna: Attempt to discard overwritten operations before CopyArea Also we need to be more careful and consider alu when marking the operation as 'replaces'. Signed-off-by: Chris Wilson commit 818702d3dc48964e4fd11afe8f2fce431ae913c0 Author: Chris Wilson Date: Fri May 24 12:15:42 2013 +0100 sna: Discard COW before migrating CPU damage to the GPU At that point, the GPU bo will no longer be a clone of the source and so we need to decouple it and make ourselves a fresh copy. Signed-off-by: Chris Wilson commit 2b9723767612702f4cfa2f73ffced83c46eccddf Author: Chris Wilson Date: Fri May 24 11:46:36 2013 +0100 sna: Do make a clone of a GPU bo that is itself a cache As we do not attempt to undo the cache status upon the clone. Signed-off-by: Chris Wilson commit dbe2b5896ec2cd00a456236467b221c9fafdd2f7 Author: Chris Wilson Date: Fri May 24 12:03:41 2013 +0100 sna: Correct assertion for reusing COW Reported-by: Jiri Slaby Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 2005b273e10f2b946f79c602b55a7bf585c936de Author: Chris Wilson Date: Fri May 24 11:02:20 2013 +0100 sna: Check for NULL damage before dereferencing Fixes regression from commit 16a64649e9c440ab9457467fe04be25719a73e7c [2.21.7] Author: Chris Wilson Date: Fri May 10 15:48:58 2013 +0100 sna: Basic copy-on-write support for cloning pixmaps Reported-by: Axel Rohde Signed-off-by: Chris Wilson commit ee7ea10148b7843091c3fa103f9f4476979f8b68 Author: Chris Wilson Date: Thu May 23 12:55:21 2013 +0100 sna: Only flags the PolyFillRect as REPLACES if it is unclipped Otherwise we discard a fill operation that is required to initialise the pixmap before drawing commences. Fixes regression from commit ef9dc6fae585d5616446eedc1e6e91173f4064c1 [2.21.7] Author: Chris Wilson Date: Mon May 20 11:08:51 2013 +0100 sna: Undo a few more overwritten operations upon a bo Reported-by: Matti Hamalainen Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=64841 Signed-off-by: Chris Wilson commit 87295b1ef85505689ce326137c2794230fb3f35f Author: Chris Wilson Date: Wed May 22 09:43:16 2013 +0100 sna: Remove the overzealous assert IGNORE_CPU is sometimes only set when there is CPU damage to ignore... Signed-off-by: Chris Wilson commit f2e0454f2df4289521b24de41fb41c68f7abf0ed Author: Chris Wilson Date: Wed May 22 08:58:51 2013 +0100 sna: Fix DBG typo and so fix build with --enable-debug s/hint/flags/ -- used the wrong variable name in an assert and upset the compiler. Signed-off-by: Chris Wilson commit abe85adda51c58d3f3b9c5dea6cec4b13a65cdcd Author: Chris Wilson Date: Tue May 21 13:39:35 2013 +0100 sna: Warn about the consequences of disabling tiling Signed-off-by: Chris Wilson commit 195a51353c3af7bd253227da5f759f06cea01f73 Author: Chris Wilson Date: Tue Apr 9 19:13:46 2013 +0100 sna/video: Convert to a pure Xv backend This is to enable feature work which requires access to Client state. Signed-off-by: Chris Wilson commit 8e42637050275945200797538a34c13c90b295cc Author: Chris Wilson Date: Tue May 21 11:13:03 2013 +0100 sna: Re-enable read-read optimisations Coacher is optimistic that the issue is no longer reproducible on his machine - and whilst I do not understand the root cause, I am confident that the kernel code is correct as is our use. References: https://bugs.freedesktop.org/show_bug.cgi?id=61628 Signed-off-by: Chris Wilson commit e65ee10dfd7155230ac18a6d24a67b68336dc973 Author: Chris Wilson Date: Tue May 21 11:01:50 2013 +0100 2.21.7 release commit ef9dc6fae585d5616446eedc1e6e91173f4064c1 Author: Chris Wilson Date: Mon May 20 11:08:51 2013 +0100 sna: Undo a few more overwritten operations upon a bo Signed-off-by: Chris Wilson commit d6c30d1d4df6bcdfa075bd29da7c8aabee20774c Author: Chris Wilson Date: Fri May 17 12:11:34 2013 +0100 sna: Clear the cow_list when discarding the clone upon pixmap destroy Signed-off-by: Chris Wilson commit 21f17455650d52848467290f609c678e85b8ceab Author: Chris Wilson Date: Fri May 17 11:46:16 2013 +0100 sna: Add the missing ref(bo) when undoing the source clone Signed-off-by: Chris Wilson commit ee166ca856ac02ca8900b9ff44ed3899add2c72a Author: Chris Wilson Date: Fri May 17 11:30:03 2013 +0100 sna: Undo the clone when replacing the DRI pixmap Signed-off-by: Chris Wilson commit 8d31fe771b423a3768cbd05db994d49e941369e1 Author: Chris Wilson Date: Fri May 17 11:14:34 2013 +0100 sna: Transfer ownership of the cloned bo to the pixmaps Fix the leak from the previous commit. Signed-off-by: Chris Wilson commit 5d9315873e02d4acc5ddffc698dbf8984cbd5c42 Author: Chris Wilson Date: Fri May 17 10:51:44 2013 +0100 sna: Avoid replacing pinned bo when undoing a clone Otherwise we end up cloning the scanout only to leave it dangling if the client copies the from the front-buffer and then writes to it. Reported-by: Nick Bowler Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=64675 Signed-off-by: Chris Wilson commit 3efe143e367acfa2ff38c63388d24a8d06331944 Author: Chris Wilson Date: Fri May 17 10:30:51 2013 +0100 sna: Clear mapped hints upon cloning a pair of pixmaps Once cloned, we do not want to use inplace operations and instead force a copy. However, if we do not relinquish the hints when copying across the bo, then those hints become stale and lead to corruption. Signed-off-by: Chris Wilson commit f57a65c35268f215c17f1a02b3af50a6678ce3c1 Author: Chris Wilson Date: Tue May 14 16:15:57 2013 +0100 sna: Correct assertions to allow discarding of cpu hint for inplace ops Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit bb0969e6a5c2bddcd0499d81c0c241003f8b07ef Author: Chris Wilson Date: Tue May 14 15:22:59 2013 +0100 sna: Assert that the mapping is released before closing the GEM handle Signed-off-by: Chris Wilson commit 16a64649e9c440ab9457467fe04be25719a73e7c Author: Chris Wilson Date: Fri May 10 15:48:58 2013 +0100 sna: Basic copy-on-write support for cloning pixmaps Signed-off-by: Chris Wilson commit d89b2647dc0b72d6fa6ff89065acd3776b04cd3f Author: Chris Wilson Date: Fri May 10 18:17:24 2013 +0100 sna: Propagate clears when using the BLT composite routines Signed-off-by: Chris Wilson commit 810a0ce2f89ba6b343b5043963626c0f2d50fb22 Author: Chris Wilson Date: Fri May 10 12:02:10 2013 +0100 sna: Propagate clear color when replacing by a CopyArea Signed-off-by: Chris Wilson commit 07a4400fffe4f83df3debe62abd2c37b45f8f0c0 Author: Chris Wilson Date: Fri May 10 11:59:59 2013 +0100 sna: Attempt to discard overwritten operations before CopyArea Signed-off-by: Chris Wilson commit 3ee42de066e4629f78e254c27d07dc33e16dbc02 Author: Rodrigo Vivi Date: Mon May 13 17:56:30 2013 -0300 Adding more reserved PCI IDs for Haswell. As Chris mentioned there is a tendency for us to find out more PCI IDs only when users report. So let's add all new reserved Haswell IDs. I didn't have better names for this reserved ids and didn't want to use rsvd1 and rsvd2 groups, so I decided to use "B" and "E" that stands for the last id digit. Cc: Chris Wilson Signed-off-by: Rodrigo Vivi commit ee96de8b1e7e4a305ee31c0ece1d9d38df8328f9 Author: Rodrigo Vivi Date: Mon May 13 17:56:29 2013 -0300 Fix Haswell GT3 names. When publishing first HSW ids we weren't allowed to use "GT3" codname. But this is the correct codname and Mesa is using it already. So to avoid people getting confused why in Mesa it is called GT3 and here it is called GT2_PLUS let's fix this name in a standard and correct way. Signed-off-by: Rodrigo Vivi commit 979d2f8d0038aa621e1c75200b10a3819e024a66 Author: Chris Wilson Date: Mon Apr 22 18:00:32 2013 +0100 sna/gen4: Tidy testing for an active vertex buffer id Signed-off-by: Chris Wilson commit 7ba63307058337af5a120ad01c93b423a3e422eb Author: Chris Wilson Date: Mon Apr 22 17:43:21 2013 +0100 sna/gen4: Drop unused gen parameter to SF state setup Signed-off-by: Chris Wilson commit 2217f6356b53263b6ce8f92b5c29c0614d4ef2a5 Author: Chris Wilson Date: Thu May 9 13:46:11 2013 +0100 sna/trapezoids: Fix the determination of the trapezoid origin "src-x and src-y register the pattern to the floor of the top x and y coordinate of the left edge of the first trapezoid," Bugzilla: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1178020 Signed-off-by: Chris Wilson commit 6e98df06fa9d218a6139730140a83a1940b05980 Author: Chris Wilson Date: Thu May 9 13:45:38 2013 +0100 sna: Add more debugging to unaligned trapezoids Signed-off-by: Chris Wilson commit 5d62ec25937ad296114bf890a7a5ac11ff03508c Author: Chris Wilson Date: Thu May 9 13:44:24 2013 +0100 sna/gen7: Add DBG for channel setup for render source Signed-off-by: Chris Wilson commit debdcd6a09464251324238ccbd935a6ade265d94 Author: Chris Wilson Date: Thu May 9 13:43:57 2013 +0100 sna: Add DBG statements for choice of spans vertex emitter Signed-off-by: Chris Wilson commit c6e4088dcb261d89fa0065eb7d9b62eada049dbd Author: Chris Wilson Date: Thu May 9 11:30:05 2013 +0100 sna: Handle cached upload buffers for partial migration to GPU Since the extended use of move_area_to_gpu for partial migration of render sources, we exposed the lack of handling of upload caches along that path. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 262ee1ef1e98cb84fa0af6e679c8cd61dc93f008 Author: Chris Wilson Date: Tue May 7 10:35:53 2013 +0100 sna: Do not attempt to clean an active scanout For simplicity, skip buffers that are still in use by the batch - they will be removed later. Signed-off-by: Chris Wilson commit 5637c173f85a5bb9a77572e4c070e0d612e6f49d Author: Chris Wilson Date: Wed May 1 15:14:55 2013 +0100 sna: Be careful not to preemptively upload portions of a SHM pixmap Only upload the portion of the pixmap being used for this rendering operation, as outside of that may remain undefined and to be written by the client before a future operation. Signed-off-by: Chris Wilson commit ca4a32c20d4c1f91552c02b9008ae16435b92d71 Author: Chris Wilson Date: Wed May 1 10:18:23 2013 +0100 sna: Page align requests to userptr Signed-off-by: Chris Wilson commit ab576a42650d8a743dd91108f774c220d866de95 Author: Chris Wilson Date: Sat Apr 27 12:06:30 2013 +0100 Add all reserved PCI-IDs for Haswell There is a tendency for a product to ship based on a 'reserved' PCI-ID prior to us being notified about it. In other words, the first we find out about such a product is when customers start complaining about their shiny new hardware not being supported... References: https://bugs.freedesktop.org/show_bug.cgi?id=63701 Signed-off-by: Chris Wilson commit 7dfb359677027310f4617b49f3da2321727a076f Author: Chris Wilson Date: Fri Apr 26 13:20:36 2013 +0100 uxa/dri: Fix compile error for unknown 'bool' It appears that it is only accidentally pulled in on some systems, but not all. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=63957 Signed-off-by: Chris Wilson commit 8eaafdb43998b5cdefb20b8098c558fa5decb115 Author: Chris Wilson Date: Wed Apr 24 15:48:14 2013 +0100 sna: Prevent accessing an uninitialised region in move_area_to_gpu() If we skip the migration, we need to avoid using the unitialiased region. There is no bad consequence as both branches of the if are no-ops, but it does silence the static checkers and should make the predicate cheaper. Signed-off-by: Chris Wilson commit b4871f1dac3d3c8d81101ef0f29eb69099ff2dcf Author: Chris Wilson Date: Thu Apr 18 09:51:39 2013 +0100 sna: Rephrase initialisation without a specific backend "generic" sounds more impressive than "no" Signed-off-by: Chris Wilson commit 6668f68054cb33e0cd3425f5403d14092a4a17f8 Author: Chris Wilson Date: Wed Apr 17 15:32:49 2013 +0100 Revert "xgvevent" This reverts commit ddd75d6539dcf692cb76747cd63d1f301180f18a. This is a WIP patch, not ready for upstream. The danger of mixing topic branches. commit 1bf0d869ae07ffc5b9cd79904b421cd37997ae8c Author: Chris Wilson Date: Wed Apr 17 15:24:10 2013 +0100 Prefer i830_dri.so for gen2 chipsets Probe for i830_dri.so and, if it exists, use that driver in preference to i915_dri.so for DRI (i.e. GL clients). This was suggested in the context of distributions supplying i915g as the main DRI driver for gen3, in which case we need to provide a separate DRI driver for gen2. Signed-off-by: Chris Wilson commit ddd75d6539dcf692cb76747cd63d1f301180f18a Author: Chris Wilson Date: Tue Apr 9 19:13:46 2013 +0100 xgvevent commit ddd3cc4ed6ac9b69f178147b49628b7a7f3f3104 Author: Chris Wilson Date: Tue Apr 16 11:55:23 2013 +0100 sna: Add VALGRIND_CFLAGS whilst compiling with --enable-valgrind Signed-off-by: Chris Wilson commit f0b6ae2cfb811a8c234634c878800ca1fb95703f Author: Chris Wilson Date: Mon Apr 15 11:42:20 2013 +0100 sna: Only release the scanout cache whilst DRM_MASTER If we try to release the framebuffers whilst not DRM_MASTER, the operation will fail and we will leak the buffers. So do not bother during the normal expiration to clean the scanout cache if vt switched away. Signed-off-by: Chris Wilson commit d68270ca84dc01a930c2ed0c05cecfac88c8cc4b Author: Chris Wilson Date: Fri Apr 12 14:48:04 2013 +0100 sna/xvmc: silence a compiler warning sna_video_hwmc.c: In function 'create_context': sna_video_hwmc.c:95: warning: assignment from incompatible pointer type Signed-off-by: Chris Wilson commit a64490c794e2e98c77e0c7f9cec6778fd1e06465 Author: Chris Wilson Date: Fri Apr 12 14:28:18 2013 +0100 sna: Add missing ';' Signed-off-by: Chris Wilson commit 9dae6f9f1f169c228929185a8bd94e82afe92574 Author: Chris Wilson Date: Fri Apr 12 11:01:08 2013 +0100 sna: Flush the scanout cache after resizing the display And ensure that any new scanout allocations make the requested size. Signed-off-by: Chris Wilson commit 97fc1d4c5bd268f331aaa9f7144e4fe40837fbb0 Author: Chris Wilson Date: Fri Apr 12 10:23:43 2013 +0100 sna/gen5: Force a MI_FLUSH between using the BLT and RENDER engines There is a workaround that says the first RENDER command following use of the BLT should be a non-pipelined command. To be safe, emit a MI_FLUSH before setting up the invariants. Bugzilla: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1168066 Signed-off-by: Chris Wilson commit aff14a23f98b1230a9053a1b8434f5cc9f766d14 Author: Chris Wilson Date: Fri Apr 12 09:49:47 2013 +0100 sna: Refine assertion about the existence of CPU damage when GPU damaged One particular buffer type is quite cunning and simultaneously sets both CPU/GPU as all damaged (the upload buffer) so be aware when making bold assertions. Reported-by: Jiri Slaby Signed-off-by: Chris Wilson commit afe61281d8f92d6eb841e2522ab18e0d2fe73773 Author: Chris Wilson Date: Wed Apr 10 12:13:37 2013 +0100 sna/gen7: Cache our kernels in L3 Signed-off-by: Chris Wilson commit caf3118f90bad7f88f5f08e49177a1696bf97ccb Author: Chris Wilson Date: Wed Apr 10 12:16:34 2013 +0100 sna: Align uploads to start on page boundaries This reduces the number of loops and restarts required in the kernel. Signed-off-by: Chris Wilson commit 092e30de8a3476ab99446264edaf9b250a41736b Author: Chris Wilson Date: Tue Apr 9 20:43:19 2013 +0100 sna/xvmc: Wrap each output adaptor Each of the overlay, sprite and textured video can support XvMC passthrough, so we need to setup an XvMC adaptor for each of our Xv adaptors. Signed-off-by: Chris Wilson commit 540802595fee7f8b2c73e74c2b44e9a21f6471ae Author: Chris Wilson Date: Wed Apr 10 10:33:48 2013 +0100 sna: Suppress hotplug events whilst VT switched away Whilst we are not DRM master, not only is another server in control of the outputs and responding to the udev event, we ourselves cannot change modes and just cause contention upon the DRM device. Instead inform userspace of the change as soon as we are DRM master again and back in control. Signed-off-by: Chris Wilson commit 7afe2b45556f07b7e3a36e36da5a6d1c60366c25 Author: Chris Wilson Date: Tue Apr 9 20:57:23 2013 +0100 sna: Add a DBG option for testing userptr more thoroughly Signed-off-by: Chris Wilson commit fbe081061a2790e0e6e6a4065af313df0d1d6a84 Author: Chris Wilson Date: Tue Apr 9 09:52:13 2013 +0100 sna: Document fence limits for gen4+ Signed-off-by: Chris Wilson commit 5f1367d611b91008ba28a8c99a6b98a3e84449dc Author: Chris Wilson Date: Tue Apr 9 09:18:41 2013 +0100 sna/video: Textured video passthrough no longer relies upon XvMC Signed-off-by: Chris Wilson commit 21294eaaf16a1e35fefe0f45a237826d38dc62eb Author: Chris Wilson Date: Tue Apr 9 09:10:33 2013 +0100 sna/video: Expand passthrough support for overlay planes The passthrough is actually a generic, albeit private, mechanism that anything can use to pass along a GEM handle for use with Xv. Signed-off-by: Chris Wilson commit 86efddd9e4b921fcfc1a4b7492ba9174b84de64c Author: Chris Wilson Date: Mon Apr 8 10:54:56 2013 +0100 sna: Remove assertions for mapped GPU bo if priv->cpu after GPU bo creation The condition that we have a mapped GPU bo iff priv->cpu is only true if we have a GPU bo at the time of using the CPU -- if we create the GPU bo during move_to_gpu, then that assertion is bogus. Reported-by: Jiri Slaby Signed-off-by: Chris Wilson commit 678279eb373310f1a71a3d74e5a500b343e98830 Author: Chris Wilson Date: Sat Apr 6 15:58:50 2013 +0100 2.21.6 release commit 5332d5a7e055042233e279385bfe1388adfe15fa Author: Chris Wilson Date: Sat Apr 6 15:42:23 2013 +0100 configure: Allow valgrind support to be manually enabled Irrespective of the DDX debug settings, some people wish to run Xorg under valgrind and so prefer to have the cleaner output by making the DDX valgrind aware. (Actually Maarten wants valgrind support enabled by default...) Suggested-by: Maarten Lankhorst Signed-off-by: Chris Wilson commit 091cf6f0477824c5826547c02394752b6dc944ce Author: Chris Wilson Date: Sat Apr 6 09:30:57 2013 +0100 sna: Improve assertions to detect rogue priv->cpu status Signed-off-by: Chris Wilson commit 4a43aa81e27e8a651fde8a4761fd14bd8824d90c Author: Chris Wilson Date: Thu Apr 4 10:53:55 2013 +0100 sna: Restore bo->flush status for large bo Since we started discarding the flush flags on cached bo (in order to prevent DRI flush states leaking), we failed to preserve the flush flag for large bo (which uses it to keep batches trim and other hints). Signed-off-by: Chris Wilson commit ed3dab44a717a1a88470228b5e33f20de1e4ad0d Author: Chris Wilson Date: Tue Apr 2 15:20:52 2013 +0100 sna: Adjust userptr structure for implicit padding Signed-off-by: Chris Wilson commit 4e2fc5aee035c3059ca33dbcafc71dc5988d6b09 Author: Chris Wilson Date: Tue Apr 2 14:29:29 2013 +0100 sna: Relax scanline waits on HSW to be emittable from either ring My overzealous reading of the bspec lead me to the conclusion that the MI_LOAD_SCANLINES command was only available on the blitter ring. This is false, thankfully, and allows us to do vsync'ed Xv. Signed-off-by: Chris Wilson commit 5a36fdcee769195d5c6e642e84a8976114e7c6de Author: Chris Wilson Date: Tue Apr 2 11:01:53 2013 +0100 sna/gen4: Kill stray debugging ErrorF from previous commit Signed-off-by: Chris Wilson commit 3d7e16addb2fb5f35936aafe8e16685a91d30f59 Author: Chris Wilson Date: Tue Apr 2 10:58:52 2013 +0100 sna/gen4: Break the Video rendering loop into 16 rectangle chunks If we feed more than 16 rectangles into the video rendering pipeline, the GPU goes crazy and starts emitting corruption. Lalalala. Bugzilla: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1162046 Signed-off-by: Chris Wilson commit f09aa788d79d36688bcfdd3b49b92367590c5f16 Author: Chris Wilson Date: Tue Apr 2 10:01:21 2013 +0100 DRI2GetMSC: Do not send a bogus ust for when the drawable is not displayed According to the opengl glx_sync_control spec, the Unadjusted System Time (or UST) is a 64-bit monotonically increasing counter that is available throughout the system: http://www.opengl.org/registry/specs/OML/glx_sync_control.txt Therefore, sending 0, even in this corner case, is out of spec. However, we cannot just return FALSE here as that triggers a BadDrawable error to be sent, and as is often the case mishandled, to the client. This results in a certain compositor terminating, for example. As an alternative we can use the monotonic system timestamp which in theory should also be monotonic with the previous and subsequent vblank times. Based on a patch by Daniel Kurtz. Reported-by: Daniel Kurtz Signed-off-by: Chris Wilson commit 4af622edfc18af523e1fa9063379f68374e19b04 Author: Chris Wilson Date: Mon Apr 1 22:44:13 2013 +0100 sna: Try to eliminate pending operations to the bo being replaced When we are replacing a bo with fresh data, we can drop pending operations to it and thereby reduce the complexity of the replacement. Signed-off-by: Chris Wilson commit ef0038d358e613381e03c077e06a87fc49108d87 Author: Chris Wilson Date: Mon Apr 1 22:43:48 2013 +0100 sna: Allow the compiler to inline memcpy for the bitblt routines Signed-off-by: Chris Wilson commit 43181692f752f0a552d2e2c76d8379fe16e521cf Author: Chris Wilson Date: Thu Mar 28 15:41:38 2013 +0000 sna/gen7: Refine is_gt2() for Haswell versus Ivybridge The two similar chipsets do not use the same PCI-ID encoding schema. Fixes regression from commit 235a3981ea9759317b392302a2b2b8f4fafab410 Author: Chris Wilson Date: Tue Mar 26 20:37:14 2013 +0000 sna/gen7: Use GT2 values for GT2 variants Reported-by: zaverel@free.fr Signed-off-by: Chris Wilson commit 96c10bdff95a3f8a68c6623446655c4c3dbf738a Author: Chris Wilson Date: Wed Mar 27 22:10:37 2013 +0000 sna/gen7: Resist the temptation to overprogram the number of PS threads for HSW Signed-off-by: Chris Wilson commit 19dfa72c28c6dc677dbfec3a538d4481985195e5 Author: Chris Wilson Date: Wed Mar 27 16:56:10 2013 +0000 sna/gen4+: Set read-write allocation mode for the target render cache As we often first clear the destination before performing a blend, we get a performance boost if that first write populates the render cache. Signed-off-by: Chris Wilson commit d9b8c2039d1be17af8c56364341fc3e10795f200 Author: Chris Wilson Date: Wed Mar 27 14:49:15 2013 +0000 sna/gen7: Fix MOCS for Haswell The memory attributes changed slightly, and in particular there is now an explicit uncached setting - which of course happened to be the value currently selected. Signed-off-by: Chris Wilson commit 04b28021409bc179f6705802f3ccacb1e3c48cc5 Author: Chris Wilson Date: Tue Mar 26 21:53:04 2013 +0000 sna/gen7: Restore render acceleration for VLV power-on board Signed-off-by: Chris Wilson commit 2054f43e4332e9d0ed61a0d643a1390dffff8d96 Author: Chris Wilson Date: Tue Mar 26 20:38:52 2013 +0000 sna/gen7: Prefer spans for GT2 desktop variants Signed-off-by: Chris Wilson commit 235a3981ea9759317b392302a2b2b8f4fafab410 Author: Chris Wilson Date: Tue Mar 26 20:37:14 2013 +0000 sna/gen7: Use GT2 values for GT2 variants The check was only testing for GT2+ and excluding the normal GT2 devices. See also commit ce9f0448367ea6a90490a28150bfdc0a76500129 Author: Chris Wilson Date: Fri Feb 8 16:01:54 2013 +0000 sna/gen6: Use GT2 settings for both GT2 and GT2+ Signed-off-by: Chris Wilson commit 3dddb01005274376c8cdc5eecad36d4e0723004a Author: Mark Kettenis Date: Tue Mar 26 20:50:53 2013 +0100 sna: Use a more portable way to determine total RAM size The sysinfo function is Linux-specific. sysconf(_SC_PHYS_PAGES), while not truly portable, is available on many more systems, including Linux, Solaris, NetBSD, FreeBSD and OpenBSD. So use that instead. Verified that this results in the same value as the sysinfo call on a handful of Linux systems. Signed-off-by: Mark Kettenis commit e63390df52117e1d3ca9d23a736e9995bc734765 Author: Chris Wilson Date: Tue Mar 26 16:32:13 2013 +0000 sna: Add extra '()' around bitwise &/&& for CPU feature detection By popular demand. Signed-off-by: Chris Wilson commit 1d6f4078c1e405749ff688058ef76c287ab36201 Author: Chris Wilson Date: Tue Mar 26 16:28:26 2013 +0000 sna/gen4+: Reorder code to compile on squeeze Remember to only use sse4_2, avx2 in their restricted sections that check for compiler support. Signed-off-by: Chris Wilson commit 0b479504c82bd81c0b93de0e6529eae928eea3bf Author: Chris Wilson Date: Tue Mar 26 16:12:30 2013 +0000 sna: Check for OS support of AVX/AVX2 before use If the OS has not initialised support for the extended instructions, then their use may result in general protection faults. Based on code by Thiago Macieira. Reported-by: Arjan van de Ven Signed-off-by: Chris Wilson commit 9620f419076202ebc89b4c7f1c06374fb2554f28 Author: Chris Wilson Date: Tue Mar 26 15:44:42 2013 +0000 sna: Fix AVX2 detection It requires use of the cpuid feature eax=7:ecx=0, so it requires the extended version of __cpuid to pass in the subfeature. Signed-off-by: Chris Wilson commit da2bd82b0e10f8a263f7ab4fb49b440349d97ace Author: Chris Wilson Date: Tue Mar 26 15:13:32 2013 +0000 sna/gen4+: Add a couple more advanced SSE variants for composite vertices Signed-off-by: Chris Wilson commit 86bcf4a9f5dff98eab076df5872c3c412c0c13af Author: Chris Wilson Date: Tue Mar 26 15:13:08 2013 +0000 sna: AVX2 is in the ebx result of cpuid Signed-off-by: Chris Wilson commit a5be4210b7e8b4a895a70ba617fda45d2e0188f0 Author: Chris Wilson Date: Tue Mar 26 09:05:48 2013 +0000 sna: Trim removed fields from execbuffer2 for OpenBSD support Reported-by: Mark Kettenis Signed-off-by: Chris Wilson commit 761ab4124e00b3dbab15abe69a9b29d9c3d0da0d Author: Mark Kettenis Date: Mon Mar 25 22:40:35 2013 +0100 sna: Fix OpenBSD backlight control implementation Fix cut'n'paste error such that this actually compiles. Signed-off-by: Mark Kettenis commit 0a9b136ec43c860e6e6e651efea0330ac36a4ea5 Author: Chris Wilson Date: Mon Mar 25 15:34:41 2013 +0000 Remove the GeneralSocket upon CloseScreen Be neat and tidy in case we are shutdown but the server is not regenerated (e.g. a hot-unplug). Reported-by: Maarten Lankhorst Signed-off-by: Chris Wilson commit 6b20924f22a14f69651dcf471c48f8a44d8ad42f Author: Chris Wilson Date: Sat Mar 23 23:08:02 2013 +0000 sna: Convert out-of-bounds sample to CLEAR If asked to perform pixelformat conversion of a sample that resides outside the valid area, we can simply substitute CLEAR. Bugzila: http://bugs.freedesktop.org/show_bug.cgi?id=62681 Signed-off-by: Chris Wilson commit f6582387e7d601272a92c3aaeb656de5ce1ffe75 Author: Chris Wilson Date: Sat Mar 23 10:22:41 2013 +0000 sna: Fix an assertion to check the correct bo After syncing the GPU bo for use, that's what we should test as well. Signed-off-by: Chris Wilson commit 88b6b80c2a5191717f0baeb22cdb70ee8c993e09 Author: Chris Wilson Date: Fri Mar 22 13:36:16 2013 +0000 sna: Fix assertions that the pixmap points into the CPU bo for fallbacks Signed-off-by: Chris Wilson commit ccd6b7e402e731c73163b132a80ae29ae2911613 Author: Chris Wilson Date: Fri Mar 22 11:28:01 2013 +0000 sna: Add backlight control implementation Based on the previous patch by Mark Kettenis commit 364e02c7998813643bc81f25ba077a764872d49e Author: Mark Kettenis Date: Fri Mar 22 11:23:11 2013 +0100 uxa: Add OpenBSD backlight control implementation Signed-off-by: Mark Kettenis commit 8d75b2ece1de1c95047c5dc5a569348abd0e6325 Author: Chris Wilson Date: Fri Mar 22 09:51:46 2013 +0000 sna: Fix computation of clip extents for stippling The clip extents for the stippled BLT missed applying the drawable offset to the lower-right corner, so inevitably every operation ended up being clipped. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=62618 Signed-off-by: Chris Wilson commit 9f8e86800345d19785eae2dd2099dd7d61a7a3b3 Author: Chris Wilson Date: Thu Mar 21 22:48:06 2013 +0000 sna/dri: Disable assertion and associated bookkeeping Due to long standing ignored bugs in DRI2, we have to accept breakage in the driver. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=62614 Signed-off-by: Chris Wilson commit 4adebfed415cf58599b22e873f32e7ce2eaa6542 Author: Chris Wilson Date: Thu Mar 21 08:53:12 2013 +0000 2.21.5 release commit 1aca872ee51e10908dcc4979596ae69732e9a02a Author: Chris Wilson Date: Tue Mar 19 08:26:01 2013 +0000 sna: Haswell reintroduces MI_LOAD_SCAN_LINES Better late than never? Interestingly only available from the BLT ring, which makes accurate waiting for XVideo (which must use the render ring) impossible in the current form - we need to render to a temporary then do a vsynced blit in this case. References: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1156679 Signed-off-by: Chris Wilson commit f132452da14fd09a2a3926cc9c034cb2e8c2f1a9 Author: Chris Wilson Date: Tue Mar 19 07:57:30 2013 +0000 sna: Ignore vsync waits on tiny scanline ranges If the update is only a couple of lines tall, any tear will not be visible - so just ignore programming the wait into the GPU. Signed-off-by: Chris Wilson commit 308f0208de59620190dd3cb65b3243d2e8a7bd87 Author: Chris Wilson Date: Mon Mar 18 15:04:22 2013 +0000 sna: Reset operation state between glyphs We are not resetting sufficient state between operations as we presume that all callers of Composite() currently pass in a blank state. In the long run, we want to remove that burden and do a minimal initialisation. References: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1156387 Signed-off-by: Chris Wilson commit 4a37d57f9633bbd29f308239c1cd956767b277c0 Author: Chris Wilson Date: Mon Mar 18 15:00:01 2013 +0000 sna: Add a pair of sanity checks before creating the redirection target Signed-off-by: Chris Wilson commit 28371a34fa83f70a7af3c8d3bfd6c7cef9e35073 Author: Chris Wilson Date: Mon Mar 18 14:49:58 2013 +0000 sna: Skip processing an all-clipped-out glyph Along the slow path, skip all processing of glyphs that are not visible. This is important as the slow path handles the per-glyph redirection case, which is much more expensive. Signed-off-by: Chris Wilson commit 16dac417c8049d65b3641e0f662865772faad61f Author: Chris Wilson Date: Sun Mar 17 21:56:56 2013 +0000 sna/dri: Fix stale Pixmap detection NB the back buffer is not associated with the Drawable and so has no pixmap field set. Hence comparing the front->pixmap against the back->pixmap was always rejecting the blit. All we can check is that front->pixmap corresponds with the current Drawable and so reject stale configuration. Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 85213d5d450eec5696496128c1acecb5ca13c53b Author: Chris Wilson Date: Sun Mar 17 21:42:48 2013 +0000 sna: Don't remove the flush flag for userptr bo This flag is far too overload with meaning, but for now this prevents us attempting to call free() on a SHM segment. Signed-off-by: Chris Wilson commit c5b901a635a9c8c74017682d17cfcd93031907b4 Author: Chris Wilson Date: Sat Mar 16 10:54:51 2013 +0000 sna/dri: Clear flush flag upon bo destroy Fixes sanity checks that we do not leak the flushing status, nor invoke an operation upon an unflushed bo. Signed-off-by: Chris Wilson commit 45d20e9a65bec8d962a4ec20ee35079935f71b91 Author: Chris Wilson Date: Thu Mar 14 22:09:38 2013 +0000 sna: Add an LLC path for creating snoopable buffers As with LLC we do not actually need to track snoopable as a separate cache state. Signed-off-by: Chris Wilson commit 94cb10c3f2b9bbb6ae3c76faebe9fc88691224a9 Author: Chris Wilson Date: Thu Mar 14 22:02:09 2013 +0000 sna/gen5+: Add missing float casts in computation of scaled src offsets Without the casts, the division ends up as 0 rather than the fractional offset into the texture. The casts were missed in the claimed fix: commit 89038ddb96aabc4bc1f04402b2aca0ce546e8bf3 Author: Chris Wilson Date: Thu Feb 28 14:35:54 2013 +0000 sna/video: Correct scaling of source offsets Reported-by: Roman Elshin Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=62343 Signed-off-by: Chris Wilson commit dad50881d545da665191c6681f2acd0ebc3ddbfc Author: Chris Wilson Date: Thu Mar 14 14:53:13 2013 +0000 sna: Consider placement hints when choosing userptr read path Signed-off-by: Chris Wilson commit 11e2802528aed4ef9ffc8b75045c72ac641e54b1 Author: Chris Wilson Date: Thu Mar 14 14:42:25 2013 +0000 sna: Add handle info to sync DBG Signed-off-by: Chris Wilson commit 20832494be06da9ebc4801647521f95e092188fc Author: Chris Wilson Date: Thu Mar 14 14:03:00 2013 +0000 sna: Use userptr downloads for incomplete GPU damaged pixmaps If the pixmap is not wholly damaged, but still contains the region to be read, then use userptr (if available). Signed-off-by: Chris Wilson commit 92023f39a9c8897e5a978f44b7970773b118f628 Author: Chris Wilson Date: Thu Mar 14 09:42:37 2013 +0000 sna: Add a few more assertions to track userptr through the caches i.e. make sure they don't end up in any caches. Signed-off-by: Chris Wilson commit 80401f4fe54142c16ea7578b587529610b68cb67 Author: Chris Wilson Date: Wed Mar 13 11:01:45 2013 +0000 sna/trapezoids: Minimally replace points for TriFan The realisation dawns that it wasn't the ordering of points within the triangle, but simply that I was replacing the wrong one. commit 716723d655a6ee11b090642b759e2542ad712c17 Author: Chris Wilson Date: Wed Mar 13 10:51:03 2013 +0000 sna/trapezoids: Correct ordering of points within TriFran Signed-off-by: Chris Wilson commit 49374f2155f0c8c80b1c54e968d510d55df49221 Author: Chris Wilson Date: Tue Mar 12 20:12:51 2013 +0000 sna/gen4: Tweak compilation flags to avoid mixed settings across functions Confusing gcc with different flags for supposedly inlined functions is not a good idea. References: https://bugs.freedesktop.org/show_bug.cgi?id=62198 Signed-off-by: Chris Wilson commit ee0ed88a09bc2f8ebe49b1d7f7e209a73e02fee0 Author: Aaron Plattner Date: Tue Mar 12 12:45:58 2013 -0700 uxa: don't crash when freeing an uninitialized screen When intel_scrn_create creates a screen, it sets scrn->driverPrivate to (void *)(match_data | 1). Normally, this is read by I830PreInit and then replaced with a pointer to the intel_screen_private structure. However, it's possible for the server to delete the screen before initializing it, which leads to a crash in I830FreeScreen when it tries to interpret the unaligned match_data pointer as a pointer to a intel_screen_private. Fix this by checking the low bit of the pointer and skipping the teardown code if it's set. Signed-off-by: Aaron Plattner commit b1952e79021759927361d284b157713a651a10b1 Author: Chris Wilson Date: Tue Mar 12 19:58:28 2013 +0000 sna/gen3: Tweak code generation for gen3_emit_composite_primitive_constant__sse2 References: https://bugs.freedesktop.org/show_bug.cgi?id=62198 Signed-off-by: Chris Wilson commit f320e6908f24ff356303c8905c78ac290e24c2e6 Author: Chris Wilson Date: Tue Mar 12 19:53:20 2013 +0000 sna/gen3: Tweak code generation for gen3_emit_composite_primitive_identity_gradient__sse2 References: https://bugs.freedesktop.org/show_bug.cgi?id=62198 Signed-off-by: Chris Wilson commit b1d0ca266cc67df65c81a9dbb9cc83027ff240ea Author: Chris Wilson Date: Tue Mar 12 19:03:44 2013 +0000 sna/gen3: Reduce another use of transforms Prefer the slightly cheaper _sna_get_transformed_scaled(). Signed-off-by: Chris Wilson commit 50374fb4947e6682b6c8aced639b422110b6ec98 Author: Chris Wilson Date: Tue Mar 12 18:39:07 2013 +0000 uxa/glamor: Prevent a crash when trying to load a misconfigured glamor Glamor requires that glamoregl is explicitly loaded via Section "Module" or else it currently crashes. Based on a patch by Michel Dänzer for xf86-video-ati. Signed-off-by: Chris Wilson commit c79a189b6f06a58ddbd92427b6c57a0bdfb1e16c Author: Chris Wilson Date: Tue Mar 12 15:57:17 2013 +0000 sna: Missing git-add for 09862a85eba243b Signed-off-by: Chris Wilson commit a31831bce86745205369faaa297064770f4350f9 Author: Chris Wilson Date: Tue Mar 12 15:26:28 2013 +0000 sna: Improve asserts that the CPU bo is not busy after synchronisation Taking into account that we may not do a full synchronisation. Signed-off-by: Chris Wilson commit 09862a85eba243babde9ca721ca6521566f5c082 Author: Chris Wilson Date: Tue Mar 12 15:12:59 2013 +0000 sna: Mark redirect proxies with a unique id This helps gen3 in particular as it uses the unique_id field of a bo to detect changes in render target. Signed-off-by: Chris Wilson commit 4f8fba31d7317248e0d4710ef5553dc1f9593314 Author: Chris Wilson Date: Tue Mar 12 10:12:12 2013 +0000 sna: Apply the source offset to the transform when fixing up gradients Otherwise pixman will apply the source transform to the offsets. Reported-by: Ognian Tenchev Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=62198 Signed-off-by: Chris Wilson commit 0b143fcb2bdeec9abd1c1d2a9dbe707cb9165b65 Author: Chris Wilson Date: Mon Mar 11 18:11:27 2013 +0000 sna: Improve sna_copy_boxes DBG by printing the bo handles The handles of the bo are easier to track through the DBG as they are more often printed than then their addresses. Signed-off-by: Chris Wilson commit 44d3ffdf534eb1c24ab729b9af31f09a86b82256 Author: Chris Wilson Date: Mon Mar 11 17:14:39 2013 +0000 sna: Only allocate addition space if we need pixel data Signed-off-by: Chris Wilson commit b500e30d5dbcf91542c2831ebd3c48b28a01bf28 Author: Chris Wilson Date: Mon Mar 11 17:14:07 2013 +0000 sna: Mark the userptr as a CPU mapping This helps clarify some recent asserts. Signed-off-by: Chris Wilson commit 17a99f0743836997aa8e4f7eafc0ea4978244f44 Author: Chris Wilson Date: Mon Mar 11 16:41:16 2013 +0000 sna/dri: Free the event on the impossible error path Just in case we end up with bogus data, don't leak. Signed-off-by: Chris Wilson commit 9d097d5140857fad209ee5c2c8b5461aac54e81e Author: Chris Wilson Date: Mon Mar 11 16:39:11 2013 +0000 sna/dri: Free chained swaps upon CloseWindow The assumption that the chained swaps are freed after the next vblank turns out to be wrong, and in effect we would leak the bo if we happened to submit the final swap faster than vrefresh and close the window before the vblank. Signed-off-by: Chris Wilson commit 2f6a9927952e25aa2a1628e66295f0903dcf865c Author: Chris Wilson Date: Mon Mar 11 14:40:52 2013 +0000 sna/dri: Reorder assert to avoid NULL dereference Only try to dereference chain->draw after the check to see if it was already destroyed. Signed-off-by: Chris Wilson commit cffdd1eed0ca344142c165788ce7a31b80f1f55f Author: Chris Wilson Date: Mon Mar 11 14:16:27 2013 +0000 test: Try to exercise races between DRI2SwapBuffers and CloseWindow Signed-off-by: Chris Wilson commit 4fb7be0a0d4bf027ed254399b6b538e979f525f2 Author: Chris Wilson Date: Sun Mar 10 14:34:39 2013 +0000 2.21.4 release commit 25a63b32c7b549ffb3c7f14de4bf2731d044eb39 Author: Chris Wilson Date: Mon Mar 11 10:42:08 2013 +0000 sna: Tighten checking for coherent maps Remember all the special cases where we can use a CPU mmap as a temporary substitute for a GTT mapping. Signed-off-by: Chris Wilson commit 210d474a96d3adf7338b49ee2c234893b54bbb2c Author: Chris Wilson Date: Sun Mar 10 16:40:35 2013 +0000 sna: Tweak CPU mappings to be only used if a read is required Signed-off-by: Chris Wilson commit 35f50a98fde7a4ebe6871bd23475948a2a530633 Author: Chris Wilson Date: Sun Mar 10 15:51:45 2013 +0000 sna/dri: Flatten _sna_dri_destroy_buffer() One level of indentation can be trivially removed with an earlier return. Signed-off-by: Chris Wilson commit b81ee116d36845e55f71be2db391f93c1b681d5d Author: Chris Wilson Date: Sun Mar 10 14:15:46 2013 +0000 sna/dri: Add a couple more basic assertions To catch bad reference counting and loss of flush. Signed-off-by: Chris Wilson commit 93ecd5606eaf2e1db2b844f39bb04d77fa53cf8d Author: Chris Wilson Date: Fri Mar 8 23:38:12 2013 +0000 sna: Only shrink the maximum GPU size to fit into cache That is be very careful that we do not enlarge it past the aperture size on early gen. Signed-off-by: Chris Wilson commit 78c756120db2e5e174925a92c0a0bba3ff70860c Author: Chris Wilson Date: Fri Mar 8 16:39:21 2013 +0000 sna/gen4: Initialise a pair of unimportant variables Should never actually affect any results. Signed-off-by: Chris Wilson commit dcd52d0c186e8b17569335338232298fbbea2b2d Author: Chris Wilson Date: Thu Mar 7 23:03:52 2013 +0000 sna: Fallback if we cannot fit the tiling copy into the aperture If we cannot fit the source/destination and a tile into the aperture, then we cannot perform the copy and must do it using the CPU. Signed-off-by: Chris Wilson commit 50d0ea02b3908156534fc00def55cfca30bbfecc Author: Chris Wilson Date: Thu Mar 7 22:58:21 2013 +0000 sna: Trim tile sizes to fit aperture constraints This is especially important with gen2 where we have no unmappable GTT to utilise. Signed-off-by: Chris Wilson commit cd1c14b172b0794f744b304f59b4fc78f0d6f8d6 Author: Chris Wilson Date: Thu Mar 7 10:16:19 2013 +0000 sna: Remove the unused output and crtc lists Signed-off-by: Chris Wilson commit 1b8fc8715598e7bf756ccfa99a6da59bab48c2a0 Author: Chris Wilson Date: Thu Mar 7 10:12:14 2013 +0000 sna: Disambiguate sna_crtc_resize() sna_crtc_resize() operates upon the whole configuration rather than a single crtc like the other sna_crtc_* so rename it to make it distinct. Signed-off-by: Chris Wilson commit 5f1c2b3b8bb062a4574e5114b09a33b7e4f0c811 Author: Chris Wilson Date: Thu Mar 7 09:40:06 2013 +0000 sna: Supply a fake pipe to run completely headless Signed-off-by: Chris Wilson commit 2aacfbaa5143568906e88e489c72f4ebf065a4ec Author: Chris Wilson Date: Thu Mar 7 09:10:32 2013 +0000 sna: Return early if there are no outputs to setup Instead of reporting failure, just try to run headless. Signed-off-by: Chris Wilson commit b84db39edb6ebce965c1354dc5994bf0a70cea92 Author: Chris Wilson Date: Thu Mar 7 09:07:19 2013 +0000 intel: Relax the has-kms test slightly Do not rely on a fully populated set of CRTCs, but merely note that the GETRESOURCES ioctl returns an error if KMS is not enabled. Signed-off-by: Chris Wilson commit c6a59bee3bf5f42147f75a714f2b9aa26590329e Author: Chris Wilson Date: Wed Mar 6 19:01:50 2013 +0000 sna: Disable read-read optimisations There is still a lurking issue, so punt on the optimisation for now. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=61628 Signed-off-by: Chris Wilson commit 9a7577902ee3a1286ea2a74a79aaeff6c72a6f18 Author: Chris Wilson Date: Wed Mar 6 11:22:40 2013 +0000 sna: Lighten the wakeup handler slightly As the wakeup handler is called more frequently, we want to avoid any of the more heavyweight processing. So trim the wakeup handler down to the check to see if the GPU is idle and so we should immediately flush what we have currently queued. Signed-off-by: Chris Wilson commit 779fc0b20c382e572eb41d3cbf3d639e4eb77874 Author: Chris Wilson Date: Tue Mar 5 22:29:54 2013 +0000 sna: Typo ('unsuported') in DBG messages Signed-off-by: Chris Wilson commit e67aece46301ebb77d5fce7ac687a03713956fef Author: Chris Wilson Date: Tue Mar 5 21:43:20 2013 +0000 sna: Assert that the pixmap is not pinned before releasing the GPU bo Nothing suspicious in this set, just an extra dab of paranoia. Signed-off-by: Chris Wilson commit bc5f81901f1b4451e4187b97f65a8be4c03b2494 Author: Chris Wilson Date: Tue Mar 5 15:53:24 2013 +0000 sna: The GPU bo submit is still required before mapping for inplace operations As we may choose to force the stall as we would be doing a read-back in any event... Signed-off-by: Chris Wilson commit 43e3886c6de3ab89e3629c98287ede77ae2f6876 Author: Chris Wilson Date: Tue Mar 5 15:32:35 2013 +0000 sna: Use drawable_gc_flags() to select placement for PolyArc Signed-off-by: Chris Wilson commit 99b30203ba1c26d89d869691a743488a5dc32935 Author: Chris Wilson Date: Tue Mar 5 13:45:45 2013 +0000 sna: Fix deference of just nullified GPU bo in previous commit One last minute refactor too far, combining the too if(priv->gpu_bo) blocks forgot that the first block tried to free gpu_bo... Signed-off-by: Chris Wilson commit d31dc1343d59f8ef8876e1dc8ddfec9c6be1a602 Author: Chris Wilson Date: Tue Mar 5 13:30:37 2013 +0000 sna: Refactor the is-mappable test for inplace operations By moving the test into the common function for creating a mappable GPU bo, we can also consider recreating that bo when desirable. Signed-off-by: Chris Wilson commit 9fa9234c7061be3cff9a65aac0702f4c3caac40d Author: Chris Wilson Date: Tue Mar 5 11:34:49 2013 +0000 sna: Prefer GPU for self-copies when undamaged on the CPU Signed-off-by: Chris Wilson commit 60ec35b8d25ecfabf1744ea7bc81109d7f2a90e2 Author: Chris Wilson Date: Tue Mar 5 11:14:37 2013 +0000 sna: Be explicit when checking for an idle bo after CPU synchronisation Signed-off-by: Chris Wilson commit ad4901a0354efaed65bb15ef33561df067d300b7 Author: Chris Wilson Date: Tue Mar 5 09:23:07 2013 +0000 sna: Ensure that the tearfree swap is unmapped afterwards After swapping the bo for the screen pixmap, we need to make sure that we don't chase any dangling pointers. References: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1133490 Signed-off-by: Chris Wilson commit 007a95d62fc998b3a558bb80310c82256d00b0bb Author: Chris Wilson Date: Mon Mar 4 13:52:04 2013 +0000 sna: Drop a local variable only used to hold a constant Signed-off-by: Chris Wilson commit 4be725333c11804a45c80a4bc45852e81ea57ed5 Author: Chris Wilson Date: Mon Mar 4 11:55:16 2013 +0000 sna: Ensure we do not attempt to operate inplace on a very large pixmap Otherwise we will fail to map it into the GTT. In theory, this should just result in the same fallback paths, but pushing that knowledge up further should help us make better decisions. Signed-off-by: Chris Wilson commit ad05039cc83b40cf45c8221d69506d6c88c27a86 Author: Chris Wilson Date: Mon Mar 4 10:11:22 2013 +0000 sna/gen3: Always prefer BLT for copies Signed-off-by: Chris Wilson commit 59f030dd27eff5fbf9ddcd22faa3087bc9c5ff54 Author: Chris Wilson Date: Mon Mar 4 10:10:39 2013 +0000 sna/gen3: Perform alpha-fixup upon copy sources This is required for copying from depth-15 surfaces onto other depths. Reported-by: Reinis Danne Signed-off-by: Chris Wilson commit 48a733a35a2de36664a42384d066c06f54cf2ad7 Author: Chris Wilson Date: Sun Mar 3 15:32:23 2013 +0000 sna: Only destroy the cached CLEAR picture if it exists Reported-by: Reinis Danne Signed-off-by: Chris Wilson commit 2583232773f1fa505e0676f07b029b29b90c8ea7 Author: Chris Wilson Date: Sun Mar 3 15:25:01 2013 +0000 sna: Nullify cache pointers upon server regen To make sure we don't chase dangling pointers on the next pass. Signed-off-by: Chris Wilson commit 4dec895b69c2e545b142a3646646e13f7fb06507 Author: Chris Wilson Date: Sun Mar 3 14:44:13 2013 +0000 sna/gen3: Correct typo in assertion Signed-off-by: Chris Wilson commit 862601c05e9788d1416173b730a5d4b6acabf374 Author: Chris Wilson Date: Sun Mar 3 14:00:18 2013 +0000 sna/gen3: Add some debug around rendercopies of the fbcon Signed-off-by: Chris Wilson commit 995abb46754eba7524afcb852159a31ea19add82 Author: Chris Wilson Date: Sun Mar 3 12:12:56 2013 +0000 sna: CPUID return ISA features in ecx/edx Whereas we were wrongly looking in eax. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=61733 Signed-off-by: Chris Wilson commit f9f2b5152f988d4b15f773354f3fadaff02cd3d7 Author: Chris Wilson Date: Sun Mar 3 11:10:51 2013 +0000 sna: Undo preferred depth for gen3 Mesa is buggy. Qt is buggy. Everything is buggy. References: https://bugs.freedesktop.org/show_bug.cgi?id=61441 Signed-off-by: Chris Wilson commit ef5edbf61686388126e62ce360a66dd8652b2493 Author: Chris Wilson Date: Sun Mar 3 09:29:53 2013 +0000 sna: Probe for kernel support of framebuffer formats References: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1135403 Signed-off-by: Chris Wilson commit cd313a8d5d1363929bebac83f81e347b4a9e70f1 Author: Chris Wilson Date: Sat Mar 2 22:52:15 2013 +0000 sna/dri: Guard against failed batch submission Avoid dereferencing a NULL bo if we do not submit a batch for the copy operation. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=61708 Signed-off-by: Chris Wilson commit e4df4e03b147694c65196651f5c13e7199f74579 Author: Chris Wilson Date: Sat Mar 2 22:46:57 2013 +0000 sna: Eliminate a few used-once local variables Signed-off-by: Chris Wilson commit ae3531c3a1d72a73b25c5563b4db029f051262cb Author: Chris Wilson Date: Fri Mar 1 23:46:07 2013 +0000 Fix Haswell CRW PCI-IDs As we missed the PCI-ID for the CRW GT1 variant, we would not have enabled render support for those particular Haswell machines. Reported-by: Kenneth Graunke commit fef6cdae9ebd217843bab2f65c87b59f8a9f782e Author: Chris Wilson Date: Fri Mar 1 15:58:42 2013 +0000 sna: Chain up CloseScreen Remember to call into the chained CloseScreen destructors! Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=56608 Signed-off-by: Chris Wilson commit 8c2dedaa4df2522d6a1725c09e9700c5d753ac93 Author: Chris Wilson Date: Fri Mar 1 15:49:57 2013 +0000 sna: Assert that inplace operations are done on coherent maps Signed-off-by: Chris Wilson commit d8296412b88b1ab4ea48572707408f7eeb52c8fa Author: Chris Wilson Date: Fri Mar 1 15:42:38 2013 +0000 sna: Assert that when caching a pixmap header the pScreen match Signed-off-by: Chris Wilson commit 904af323914e05830f17621a78b6e55b371ae5fc Author: Chris Wilson Date: Fri Mar 1 15:34:45 2013 +0000 sna: Assert that we do not resurrect stale pixmap across a server regen References: https://bugs.freedesktop.org/show_bug.cgi?id=56608 Signed-off-by: Chris Wilson commit d4164de5ccb82068e2858a90b2cd44eef82b6037 Author: Chris Wilson Date: Fri Mar 1 12:13:47 2013 +0000 sna: Assert that the ScrnInfo and ScreenPtr relationship is correct References: https://bugs.freedesktop.org/show_bug.cgi?id=56608 Signed-off-by: Chris Wilson commit 8d97c192a33473df3c9036147a4716783a86d742 Author: Chris Wilson Date: Fri Mar 1 11:59:49 2013 +0000 sna: Move some conditional code out-of-line Just to reduce the number of ifdefs, one at a time. Signed-off-by: Chris Wilson commit 4174c7bb14ea3428d86dc327950e76d0589e0cc4 Author: Chris Wilson Date: Fri Mar 1 09:21:03 2013 +0000 sna: Use depth-15 rather than depth-16 for reduced bw devices Keeping the number of bits identical between the RGB channels helps prevent discoloration. Signed-off-by: Chris Wilson commit e05d2e481434084291503efdb84c944a560d8c29 Author: Chris Wilson Date: Fri Mar 1 09:08:19 2013 +0000 sna: Track number of pixmaps allocated whilst counting bo leaks Signed-off-by: Chris Wilson commit 89038ddb96aabc4bc1f04402b2aca0ce546e8bf3 Author: Chris Wilson Date: Thu Feb 28 14:35:54 2013 +0000 sna/video: Correct scaling of source offsets When applying pan and zoom to a mismatched video, it would inevitably miscompute the origin and scale factors. Reported-by: Matti Hamalainen Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=61610 Signed-off-by: Chris Wilson commit a0a2faefdefbea63669dfeb49f7e701196ab5631 Author: Chris Wilson Date: Wed Feb 27 22:26:19 2013 +0000 sna: Fix syntax for __get_cpuid() It is a function not a macro like cpuid() and takes pointers to its arguments! Signed-off-by: Chris Wilson commit bfc1fda87f442890919100473b819240f9c30c44 Author: Chris Wilson Date: Wed Feb 27 22:19:06 2013 +0000 sna/dri: Fix use of uninitialised pipe along error path > sna_dri.c:2018:6: warning: variable 'pipe' is used uninitialized > whenever 'if' condition is true [-Wsometimes-uninitialized] > if (get_private(front)->pixmap != get_drawable_pixmap(draw)) > ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > sna_dri.c:2150:42: note: uninitialized use occurs here > DRI2SwapComplete(client, draw, 0, 0, 0, pipe, func, data); Reported-by: Sedat Dilek Signed-off-by: Chris Wilson commit f08ad3f8ef1ac3a38a92f793fa1cbc1d2d443095 Author: Chris Wilson Date: Wed Feb 27 17:46:55 2013 +0000 sna: Prettify GCC version detection in headers And fixup a basic error in the process. Signed-off-by: Chris Wilson commit 8aea4ae127f92c07e3ce86b6d2afaa207a6cdce9 Author: Chris Wilson Date: Wed Feb 27 17:38:46 2013 +0000 sna: Improve compatibility of cpuid.h detection Reported-by: Sedat Dilek Signed-off-by: Chris Wilson commit 5c27307219b5a5c1f73dcde9d0b2beed5810bd7d Author: Chris Wilson Date: Wed Feb 27 14:58:35 2013 +0000 sna: Remove the trailing '.' from the previous commit Good Maarten for reporting the bug, bad Maarten for silently fixing up my mistakes when testing! Thanks to Colin Walter and his tinderbox. Signed-off-by: Chris Wilson commit b8a0acd615042bee7bdf49c364a4815ade6b9a61 Author: Chris Wilson Date: Wed Feb 27 13:26:18 2013 +0000 sna: Migrate dirty tracking across ScreenPixmap updates Otherwise chaos ensues: Invalid read of size 8 ==8647== at 0x8477BC1: has_offload_slaves.isra.38 (sna_accel.c:14402) ==8647== by 0x84958E7: sna_accel_block_handler (sna_accel.c:13729) ==8647== by 0x164B13: BlockHandler (dixutils.c:387) ==8647== by 0x2B5273: WaitForSomething (WaitFor.c:210) ==8647== by 0x160790: Dispatch (dispatch.c:357) ==8647== by 0x14F549: main (main.c:298) ==8647== Address 0xf3383e8 is 24 bytes inside a block of size 152 free'd ==8647== at 0x4C2BA6C: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==8647== by 0x2403DF: damageDestroyPixmap (damage.c:1549) ==8647== by 0x1FFD88: XvDestroyPixmap (xvmain.c:372) ==8647== by 0x1FE789: ShmDestroyPixmap (shm.c:273) ==8647== by 0x8507C24: _sna_dri_destroy_buffer (sna_dri.c:448) ==8647== by 0x289986: do_get_buffers (dri2.c:521) ==8647== by 0x289C0F: DRI2GetBuffersWithFormat (dri2.c:690) ==8647== by 0x28B68F: ProcDRI2Dispatch (dri2ext.c:306) ==8647== by 0x160A40: Dispatch (dispatch.c:428) ==8647== by 0x14F549: main (main.c:298) ==8647== ==8647== Invalid read of size 8 ==8647== at 0x8477BCA: has_offload_slaves.isra.38 (regionstr.h:74) ==8647== by 0x84958E7: sna_accel_block_handler (sna_accel.c:13729) ==8647== by 0x164B13: BlockHandler (dixutils.c:387) ==8647== by 0x2B5273: WaitForSomething (WaitFor.c:210) ==8647== by 0x160790: Dispatch (dispatch.c:357) ==8647== by 0x14F549: main (main.c:298) ==8647== Address 0xdfdfdfdfdfdfdfe7 is not stack'd, malloc'd or (recently) free'd Reported-by: Maarten Lankhorst Signed-off-by: Chris Wilson commit 11b72628cb54ab0b78a0969fa8fabb591f6cf93f Author: Chris Wilson Date: Tue Feb 26 19:08:58 2013 +0000 sna/gen2: Add SSE2 fast paths for vertex emission Signed-off-by: Chris Wilson commit f2597c89d023beceddd65e99fa595741f2400218 Author: Chris Wilson Date: Tue Feb 26 18:44:45 2013 +0000 sna: Force GCC to use the SSE unit for SSE2 routines Merely hinting that it was preferred by using sse+387 was not enough for GCC to emit the faster SSE2 code. Signed-off-by: Chris Wilson commit a18ce0f642fa347b61e4ca501bd2f747338a2975 Author: Chris Wilson Date: Tue Feb 26 18:44:35 2013 +0000 sna: Flatten the glyph emitters Signed-off-by: Chris Wilson commit 7bb06b02e67435354778fe87a3e0429fe3750c23 Author: Chris Wilson Date: Tue Feb 26 18:12:06 2013 +0000 sna/gen3: Expand the number of SSE2 routines for basic composite ops Signed-off-by: Chris Wilson commit 417c3f9b8c6b9a50dc1af440c53e94d2c6401251 Author: Chris Wilson Date: Tue Feb 26 17:41:57 2013 +0000 sna/trapezoids: Add a pair of unlikely hints for forced box emission Overflowing the buffer is unlikely, so pass the hint on to the compiler. Signed-off-by: Chris Wilson commit aa7e11a1a451c54d5cbff9a6c242075b08c590ae Author: Chris Wilson Date: Tue Feb 26 15:07:12 2013 +0000 sna: Improve handling of migrated userptr bo Superficially fixes gimp-2.9 Signed-off-by: Chris Wilson commit 32d8908337d45b9598ce07bc46b633ed5e0c7070 Author: Chris Wilson Date: Tue Feb 26 11:42:03 2013 +0000 sna: Flatten unaligned box emission Signed-off-by: Chris Wilson commit aa358fad8c934e6962976ecfdc5e5ebc9d1ad83c Author: Chris Wilson Date: Tue Feb 26 11:37:24 2013 +0000 sna: Bump required GCC for sse2 gcc-4.4.5 (on squeeze) triggers an ICE when using target(sse2). Signed-off-by: Chris Wilson commit 08747292b717fcea84ce3adc455b0161e5517e84 Author: Chris Wilson Date: Tue Feb 26 11:28:05 2013 +0000 sna: Conditionally compile sse2 routines Signed-off-by: Chris Wilson commit 9c63ddd9f14588bcec8275e2a64e56cc5b661a5a Author: Chris Wilson Date: Tue Feb 26 11:28:05 2013 +0000 sna: Conditionally compile sse4_2 routines Signed-off-by: Chris Wilson commit dc885d350ec53d08e0d4d529bb938bf2568d61ae Author: Chris Wilson Date: Tue Feb 26 11:28:05 2013 +0000 sna: Conditionally compile avx routines Signed-off-by: Chris Wilson commit 8cdebf3b72467f63a35888f38cd83658575fcf10 Author: Chris Wilson Date: Tue Feb 26 11:06:45 2013 +0000 sna/gen4: Cluster ISA Otherwise we seem to confuse the poor little compiler. This should also make it easier to use CPP to turn off blocks. Signed-off-by: Chris Wilson commit 8272d849c0895d6fa66ccca972899f56dab71cc6 Author: Chris Wilson Date: Tue Feb 26 10:53:20 2013 +0000 sna/gen4+: All associated CPUs have sse2 at least So mark up the basic functions as SSE2 in case we are compiling for 32-bit only. Signed-off-by: Chris Wilson commit 76d46201d7b7abc7c0b6bced9c6fadaf47ad9134 Author: Chris Wilson Date: Tue Feb 26 10:49:42 2013 +0000 sna: Only use the GPU bo after migrating If we choose not to migrate the damage to the GPU bo, then it will be incoherent. This just flattens the logic out as priv->gpu_bo should be NULL here anyway. Signed-off-by: Chris Wilson commit 1068bf7024d8a6650de020e95efdedcbfd7d3c5e Author: Chris Wilson Date: Tue Feb 26 10:14:03 2013 +0000 sna: Mention if compiled as 64bit as part of the CPU feature detection Signed-off-by: Chris Wilson commit 27c71027b1d76979460ef15bba8e8671e3286a24 Author: Chris Wilson Date: Tue Feb 26 10:00:29 2013 +0000 sna: Ignore SSE4a - not an Intel ISA! Signed-off-by: Chris Wilson commit 2a6f3989ba2075c322cdf6e33f829f4d5a885be0 Author: Chris Wilson Date: Tue Feb 26 09:59:34 2013 +0000 sna/gen3: Allow conditional use of SSE2 Signed-off-by: Chris Wilson commit 13461a18b1605feb17304d52136d100df50ca296 Author: Chris Wilson Date: Tue Feb 26 09:41:14 2013 +0000 sna: Print detected CPU features Signed-off-by: Chris Wilson commit 528dbf9ebb2688f476ef283be59d0f2232159dcb Author: Chris Wilson Date: Tue Feb 26 08:15:47 2013 +0000 sna: Fix build on older GCC for cpuid() We need to double check that the features we look for are supported by the compiler before doing so. Signed-off-by: Chris Wilson commit 49d9c5e9d48d043a8366b6471128c999b5e887b0 Author: Chris Wilson Date: Tue Feb 26 00:27:12 2013 +0000 sna: Reverse inverted assertions Oops, the assertions that we had sufficient free space was inverted. Signed-off-by: Chris Wilson commit 94b95cc2fc63457903a7b7b6eaa09bc27f25750c Author: Chris Wilson Date: Tue Feb 26 00:02:16 2013 +0000 sna/gen4+: Begin specialising vertex programs for ISA Allow use of advanced ISA when available by detecting support at runtime. This initial work just uses GCC to emit varying ISA, future work could use hand written code for these hot spots. Signed-off-by: Chris Wilson commit f095678125b25aeae80d838729a7f89d09007e10 Author: Chris Wilson Date: Mon Feb 25 23:14:12 2013 +0000 test: Correct ignore path for async.avi Signed-off-by: Chris Wilson commit aa045d80ec0d3366745d1671b9f667cec587d804 Author: Chris Wilson Date: Mon Feb 25 23:13:23 2013 +0000 sna: Detect available instruction sets at runtime Signed-off-by: Chris Wilson commit 56fd91fc830d7a210e3a0e70ab41261a1f5baa83 Author: Chris Wilson Date: Mon Feb 25 15:18:46 2013 +0000 sna/gen3+: Restart vertex space checks after lock contention If we end up contending for the vertex lock, we need to double check there is sufficient vertex space left for us. Bugzill: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1124576 Signed-off-by: Chris Wilson commit b12f0da0bedb7faf0c96e44ca2d02baec807ad51 Author: Chris Wilson Date: Mon Feb 25 15:18:46 2013 +0000 sna/gen3+: Assert that nbox is not 0 Various assertions to track down a potential programming error. References: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1124576 Signed-off-by: Chris Wilson commit 6f9646f5beecca2972ef0299b5c7469008cb6fe4 Author: Chris Wilson Date: Mon Feb 25 14:36:32 2013 +0000 sna/trapezoids: Instruct the compiler to flatten the callees whilst rasterising Signed-off-by: Chris Wilson commit 421910ca572a044170af69a90bb934f114a62168 Author: Chris Wilson Date: Sun Feb 24 21:58:30 2013 +0000 sna/gen3: Factor out the per-vertex divide for gradients Signed-off-by: Chris Wilson commit a19da0ea517127052ae49cdd6441e8b6077ca523 Author: Chris Wilson Date: Sun Feb 24 11:22:01 2013 +0000 sna: Reduce DefaultDepth to 16 on older chipsets Signed-off-by: Chris Wilson commit d7bba8512c6db5f58448b513d1f4290938b0abd3 Author: Chris Wilson Date: Sat Feb 23 10:56:08 2013 +0000 sna: Tidy a call to set the gamma before setting up a CRTC Signed-off-by: Chris Wilson commit e2aaf4ed3333231cf8cadcf6362743593026cd19 Author: Carl Worth Date: Fri Feb 22 14:44:09 2013 -0800 Refuse to load driver for depth-8 X server. The driver is not functional at this depth, so tell the user as much and bail out. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=31375 Reviewed-by: Eric Anholt commit cbd8f16ea0e3b1b7eae57edfe343a332ae42373c Author: Chris Wilson Date: Fri Feb 22 21:09:52 2013 +0000 sna: Refuse to create larger than max temporary upload bo The maximum size is determined by available RAM, if we exceed it we greatly increase the risk of swap thrashing. Signed-off-by: Chris Wilson commit 7a264792772512528a4d6138e09b4ee1ace7f322 Author: Chris Wilson Date: Fri Feb 22 20:13:58 2013 +0000 sna: Retire before looking for a large 2D buffer Signed-off-by: Chris Wilson commit 630b114f1ceaf9fb1e35f3aac8c3161119ba5af9 Author: Chris Wilson Date: Fri Feb 22 20:08:36 2013 +0000 sna: Allow search_linear_cache() to reuse large buffers Signed-off-by: Chris Wilson commit d87a0b4c3bc7f1a33edee79e3ed09871461422ea Author: Chris Wilson Date: Fri Feb 22 20:07:56 2013 +0000 sna: Allow picture fixup in channel format if not changing RGB depth Signed-off-by: Chris Wilson commit b129b434f98b9d875e44af206f6901de9bf28334 Author: Chris Wilson Date: Fri Feb 22 12:10:20 2013 +0000 sna: Optimise mmap of GPU bo for read-read operations Signed-off-by: Chris Wilson commit 19bd005056a2083de64753681b96716996e4237d Author: Chris Wilson Date: Fri Feb 22 12:05:04 2013 +0000 sna: Avoid migrating and making the GPU bo busy prior to mmapping it References: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1131134 Signed-off-by: Chris Wilson commit afb77ffadedf6c6f9321dde1d650004714accdc4 Author: Chris Wilson Date: Fri Feb 22 10:48:06 2013 +0000 sna/gen7: Skip CLEAR_PARAMS for the null depthbuffer Signed-off-by: Chris Wilson commit 1a7262e5308c9ab416ed87226ed27a9fc84f0ef9 Author: Chris Wilson Date: Fri Feb 22 10:30:49 2013 +0000 sna/gen7: Only a pipeline stall is required for the CA pass Signed-off-by: Chris Wilson commit b0c83b77d0226cbaf6115eed5d0e2cc8a40ac51d Author: Chris Wilson Date: Thu Feb 21 19:37:51 2013 +0000 sna/dri: Eradicate the DRI2 name exchange mechanism Thinking about the compositor <-> server <-> client inter-exchange demonstrates that we cannot prevent the client rendering into the source texture being show by the compositor. That is a subject for DRI3. Signed-off-by: Chris Wilson commit 6172ff3e350dbfbcd122a911d7fa366ff5d268bc Author: Chris Wilson Date: Thu Feb 21 15:34:47 2013 +0000 sna: Always enable the solid-cache and gradient-cache Figuring out when it might not be used is simply too error prone. Signed-off-by: Chris Wilson commit b7a96df9b0872867c823dc945add5be04d51c30a Author: Chris Wilson Date: Wed Feb 20 18:46:01 2013 +0000 sna: Honour LinearFramebuffer The Option was untested, and unsurprisingly was broken. Signed-off-by: Chris Wilson commit d2442c74b8d41018f260f1da13f3fe5d2795792f Author: Chris Wilson Date: Wed Feb 20 10:53:57 2013 +0000 2.21.3 release commit 2cab7e80eb6955a7f8ea051633f6975a12248b69 Author: Chris Wilson Date: Wed Feb 20 12:36:22 2013 +0000 sna/trapezoids: Clamp cells to valid range Signed-off-by: Chris Wilson commit 14de90b251dd8a6ff106e989580ef01cf5c2944d Author: Chris Wilson Date: Wed Feb 20 12:00:54 2013 +0000 sna/trapezoids: Embed a few cells into the stack Avoid an allocation in the common case where the set of trapezoids is fairly narrow. Signed-off-by: Chris Wilson commit 3eca4ea1a5d8ce04598b8d42e93e0dcb93e42e9a Author: Chris Wilson Date: Wed Feb 20 11:20:54 2013 +0000 sna/trapezoids: Perform the cell allocation upfront As we know the maximum extents of the trapezoids, we know the maximum number of cells we will need and so can preallocate them. Signed-off-by: Chris Wilson commit 9df5e48c582e5c4edffdece75b5395c230a50b09 Author: Chris Wilson Date: Tue Feb 19 10:08:29 2013 +0000 sna/video: Only setup XvMC if we first setup Xv Under certain circumstances, XvScreenInit can indeed fail, so do not bother with creatin XvMC (as it triggers internal assertions if it cannot find our adaptor amongst Xv's). Signed-off-by: Chris Wilson commit b6588c48077600a3e015b6d37b101393a806ae1a Author: Chris Wilson Date: Tue Feb 19 08:59:12 2013 +0000 test: Add generated vsync.avi to gitignore commit 1e2fd66ade6bdbf1e6011f3d59e423fada3f12f6 Author: Chris Wilson Date: Tue Feb 19 08:57:47 2013 +0000 sna: Assert that the GPU damage is NULL before destroy a proxy If the GPU bo is a proxy, then it really is a pointer into a upload buffer for CPU data. In these cases, there should never be any GPU damage lying around. Signed-off-by: Chris Wilson commit d90a123db7ac99cf017167bf89df31c635df7e1e Author: Chris Wilson Date: Mon Feb 18 20:06:45 2013 +0000 sna: Assert that the gpu_bo exists if it is entirely damaged This should help catch the error slightly earlier. Signed-off-by: Chris Wilson commit a03aba6f7f55577e29c6a3267528e2614016222d Author: Chris Wilson Date: Mon Feb 18 16:12:14 2013 +0000 sna: A couple more assertions that we forcibly attach pixmaps correctly Signed-off-by: Chris Wilson commit 1f16d854264ea923303b79379266bd789fd9dd4d Author: Chris Wilson Date: Mon Feb 18 14:30:55 2013 +0000 sna/dri: Prevent swapping a decoupled DRI2Buffer If the DRI2Buffer is no longer valid for the Drawable, for example the window had just been reparent, just complete the swap without triggering any assertions. Signed-off-by: Chris Wilson commit b3ba758a0186c9abc6c0583f52775ea714165134 Author: Damien Lespiau Date: Thu Feb 14 14:20:19 2013 +0000 uxa/gen7: Don't use a message register to store vl Turns out the "new" assembler that uses mesa's opcode emission hits the path that automatically transforms MRF registers into GRF ones in the exa_wm_src_projective shader. The diff with the new assembler is: $ intel-gen4disasm -g7 - - { 0x00600041, 0x208077be, 0x008d03c0, 0x008d0180 }, + { 0x00600041, 0x2e8077bd, 0x008d03c0, 0x008d0180 }, mul(8) m4<1>F g30<8,8,1>F g12<8,8,1>F { align1 }; mul(8) g116<1>F g30<8,8,1>F g12<8,8,1>F { align1 }; Of course, message registers are no more in gen7, so the shader is trying to do something shaddy (ahem!). Instead of using m4, let's make exa_wm_src_projective use g68 for v (aka vl) which makes sense since: 1/ vh is g69 2/ exa_wm_src_affine uses g68 for vl already This commit changes the generated assembly, here's the decoded diff: $ intel-gen4disasm -g7 - - { 0x00600041, 0x208077be, 0x008d03c0, 0x008d0180 }, + { 0x00600041, 0x288077bd, 0x008d03c0, 0x008d0180 }, mul(8) m4<1>F g30<8,8,1>F g12<8,8,1>F { align1 }; mul(8) g68<1>F g30<8,8,1>F g12<8,8,1>F { align1 }; Cc: Kenneth Graunke Reported-by: Xiang, Haihao Signed-off-by: Damien Lespiau Reviewed-by: Kenneth Graunke commit 43ba22ef4a4142f334e9ae2d926250988ecbe8bc Author: Chris Wilson Date: Sun Feb 17 09:25:38 2013 +0000 Confirm that i915.ko has KMS enabled before binding to the device If the kernel has the module, but the KMS module option is not enabled, we cannot function. So after checking to see if the i915.ko is bound, then querying whether it provides any KMS resources. If it has no CRTCs attached, then we need to failover to the VESA/fbdev drivers. Note that this should have been detected by drmCheckModesettingSupported() References: https://bugs.freedesktop.org/show_bug.cgi?id=60987 Signed-off-by: Chris Wilson commit 5a5943e2374a674067e0c48b0fe8a1f2eeff2177 Author: Chris Wilson Date: Fri Feb 15 12:05:14 2013 +0000 NEWS: Xserver 1.6 was not packaged in Xorg 7.7 I misinterpreted Sedat Dilek's advice on how to fix my confusing changelog, so drop the Xorg 7.7 confusion and just refer to the version of Xserver the driver is first compatible with. commit 15e1050ea505c40a288ec197d817d3c6d51693d5 Author: Chris Wilson Date: Thu Feb 14 21:59:36 2013 +0000 NEWS: Fix a couple of typos and factual errors Sedat Dilek corrected my spelling and pointed out that what is known as Xorg 1.6 in the log file is actually better known as releases of Xserver 1.6 and Xorg 7.7. commit 86e025a1842f3c2f319676818b1f4624cd94ebc5 Author: Chris Wilson Date: Wed Feb 13 22:05:34 2013 +0000 sna/gen4: Restore nonrectiinear spans workaround Instead of relying on the macro, the intention was simply to use the prefer_gpu hint. However, I dropped it whilst refactoring ideas from later generations. So restore both the debug control to force spans as well as the intended workaround. Reported-by: Edward Sheldrake Signed-off-by: Chris Wilson commit f2a64dbdfceac985d235c4873f52013633d6cfd8 Author: Chris Wilson Date: Wed Feb 13 17:15:46 2013 +0000 sna: Avoid stalling on a SHM upload when otherwise idle Fixes regresion from commit 09ea1f4402b3bd0e411b90eb5575b3ff066d7356 Author: Chris Wilson Date: Thu Jan 10 16:26:24 2013 +0000 sna: Prefer to use the GPU for copies from SHM onto tiled destinations As the stalls on IVB 64-bit machines at least greatly offset the benefits. As those earlier measurements were made on the same IVB machine but running in 32-bit mode, I need to double-check whether or not this is another 32-bit peculiarity. Signed-off-by: Chris Wilson commit 4bf51ec7bcc7818fbee8643a88aebee0362691b2 Author: Chris Wilson Date: Wed Feb 13 16:32:18 2013 +0000 Revert "sna: Avoid promoting SHM CPU bo to GPU to maintain coherence with SHM clients" This reverts commit f743cd5734ca502aa8bdb0e1327fe84d6ce82755. commit 9861423a76402c260724a752ada293a03ce1a79b Author: Chris Wilson Date: Wed Feb 13 15:11:45 2013 +0000 sna/dri: Fix typo limiting gen4 to TILING_X depth buffers Fixes regression from commit 98b312e579385e6e4adf6bf0abe20f8ca84592af Author: Chris Wilson Date: Wed Jan 23 20:51:35 2013 +0000 sna/dri: Stop feeding I915_TILING_Y to mesa i915c Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=60178 Signed-off-by: Chris Wilson commit 38a3506aa1c9c2e07125b54fc319b6de89febff9 Author: Chris Wilson Date: Wed Feb 13 14:09:38 2013 +0000 sna: Handle scanouts still active at the time of destruction Basically remove the bogus assert, and reorder the list to preferentially reuse the still active scanouts. Signed-off-by: Chris Wilson commit 3b82d93162b9e127c6d055d5007e75fcc31f4af8 Author: Chris Wilson Date: Tue Feb 12 21:53:38 2013 +0000 sna/dri: Validate that the scanout flag is set Signed-off-by: Chris Wilson commit 04e4805f57fdf0581e5e75c29c275c15c79852a3 Author: Chris Wilson Date: Tue Feb 12 21:52:35 2013 +0000 sna: Prefer inplace readback if the GPU doesn't support cacheable BLTs As we must perform the GTT reads anyway, first see if we can copy directly to the destination. Signed-off-by: Chris Wilson commit 2c569d22d2d243badc15d5dbcc09e6759f4ff01a Author: Chris Wilson Date: Tue Feb 12 20:15:33 2013 +0000 sna: Only use the inplace read fallback if we have no CPU bo support Signed-off-by: Chris Wilson commit aeaed0b7810e9c794749a4f59c72790063cddc86 Author: Chris Wilson Date: Tue Feb 12 19:30:44 2013 +0000 sna/dri: Assert that we are not passed deceased DRI2Buffers Signed-off-by: Chris Wilson commit 5fb165b02d47634ccf9fb72ab6fb4d65b89f78d0 Author: Chris Wilson Date: Mon Feb 11 22:15:17 2013 +0000 sna/dri: Assert that the old scanout is clear before overwriting Signed-off-by: Chris Wilson commit 2667b821e193c0badf16420f82f8a99a93c27a46 Author: Chris Wilson Date: Mon Feb 11 22:06:00 2013 +0000 sna/dri: Update flush status for tear-free exported GPU bo Signed-off-by: Chris Wilson commit a4442045df9164651702b166e0de2bfd25e8a76f Author: Chris Wilson Date: Mon Feb 11 21:16:19 2013 +0000 sna/dri: Harden the replacement of the back buffer against alloc failure Signed-off-by: Chris Wilson commit 75406775b83ca5c320e4bc2d2dbd8f1eb0d1677d Author: Chris Wilson Date: Mon Feb 11 13:00:13 2013 +0000 NEWS: Fix marketing name for 945gm The GMA3100 was the desktop G31/G33 which whilst almost identical to the GMA950, notably does not require the extra unfenced alignment. Signed-off-by: Chris Wilson commit 0b116a1d023ad7cc148d91f268dbf71452dbd9dd Author: Chris Wilson Date: Mon Feb 11 12:36:43 2013 +0000 sna: Fix inaccurate use of __attribute__((const)) 'const' is only allowed to use the function parameters and not allowed to access global memory - that includes not allowed to deference its arguments... Thanks to Jiri Slaby for spotting my mistake. Signed-off-by: Chris Wilson commit 47657e5355103b2e61da6f059c560e63fe13b0ed Author: Chris Wilson Date: Mon Feb 11 12:31:18 2013 +0000 sna/dri: Ignore a compiler barf breaking an assertion Manual confirmation with gdb says all is fine and things work again after printing the variables responsible. The curious reader is welcome to read the disassembly to find where the compiler goofed. Signed-off-by: Chris Wilson commit 3540554e5a74ccd2ca746482700a9f5657954227 Author: Chris Wilson Date: Mon Feb 11 11:14:17 2013 +0000 sna/dri: Couple up the flush pixmap after checking for allocation failures Tidy up the error path not to leave the GPU bo pinned and marked for flushing. Signed-off-by: Chris Wilson commit 58b61bcd02467d2b08b2dd194ab6884437b83fa6 Author: Chris Wilson Date: Mon Feb 11 11:05:02 2013 +0000 sna/dri: Tighten the assertions that we have one DRI2Buffer per Pixmap Signed-off-by: Chris Wilson commit a13f806ae28459a5cf671ccbf9cc8725f541a4ba Author: Chris Wilson Date: Mon Feb 11 10:50:20 2013 +0000 sna: Assert that we do not destroy a GPU bo for an active DRI pixmap This is an extra paranoid check that the bo is still pinned. Signed-off-by: Chris Wilson commit c2eb2db83280ae45b8df8a3b6e488b6a1404391a Author: Chris Wilson Date: Mon Feb 11 10:47:13 2013 +0000 sna: Avoid stalling for changing cache levels on an active scanout Signed-off-by: Chris Wilson commit 696347a6260f3741ad3e6238db0fd285384bb753 Author: Chris Wilson Date: Mon Feb 11 09:58:58 2013 +0000 sna/dri: Only clear the GPU bo flush flag when the last export is destroyed Add various assertions around DRI to be sure that we don't change the GPU bo for a pixmap that does not match our flush bookkeeping. Then be more careful during the DRI2Buffer destroy that we don't accidentally remove the flush flag from a shared bo. Signed-off-by: Chris Wilson commit 301aebfa71143fb610eb7b816376a8c1f3e07208 Author: Chris Wilson Date: Sun Feb 10 18:01:33 2013 +0000 configure: Add -lX11 required to link the legacy libI810XvMC.so Signed-off-by: Chris Wilson commit 65c320e677b8ab4d745568b33f381d5865bf8b73 Author: Chris Wilson Date: Sun Feb 10 14:20:59 2013 +0000 2.21.2 release commit 0d75b19979b1ac14353765e2bb84c6a466129109 Author: Chris Wilson Date: Sun Feb 10 15:47:53 2013 +0000 sna: Restore glyphs with xorg-1.12 That simple and innocuous build fix for xorg-1.13 bizarrely causes missing glyphs with earlier Xorgs. Signed-off-by: Chris Wilson commit 9fd0d8873a5a5c4f77904cab0b9909ca941b5dae Author: Chris Wilson Date: Sun Feb 10 14:29:29 2013 +0000 NEWS: fix bug url The dangers of cutting and pasting from git log. Signed-off-by: Chris Wilson commit 617fadf3acf7bf75fb203c1e85fd0ddb98b3dbb9 Author: Chris Wilson Date: Sun Feb 10 14:20:59 2013 +0000 2.21.1 release commit 3169a4e53cf39cc3d5c18ac6add909aa3a58de7e Author: Chris Wilson Date: Sun Feb 10 11:57:14 2013 +0000 sna: Reorder some includes so that compat-api.h comes after the headers it wraps Fixes the build in cases where the compat-api.h was defining macros to subvert the real functions found in the xorg includes Signed-off-by: Chris Wilson commit 71fbad64c5cfe6832a03815bece4c89d15253e1a Author: Chris Wilson Date: Sun Feb 10 10:54:17 2013 +0000 configure: Fix typo in checking for libdrm_intel The package name is libdrm_intel not libdrm_intel-1, an obvious cut'n'paste error from testing for pixman-1. Signed-off-by: Chris Wilson commit 3cbdfb54d1fcfed7745111e861e19b7bbac243cc Author: Chris Wilson Date: Sat Feb 9 19:15:20 2013 +0000 sna: Backport to squeeze - Xorg-1.6, pixman-0.16, libdrm-2.4.21 The principle change is to switch to the old Privates API and undo the Region renames. The downside is that this ignores the critical bugfixes made to the xserver since xorg-1.6 - but I assume that whoever wants to run the latest hardware on the old xservers is also backporting those stability fixes... Signed-off-by: Chris Wilson commit 42a6b25817985e22e7d462be87fbd97973d96a29 Author: Chris Wilson Date: Sat Feb 9 15:30:58 2013 +0000 sna: Fix alignment of the base of partial buffers for pre-G33 chipsets The older chipsets have much more restrictive alignment rules for the base address of tiled but unfenced objects. Bugzilla: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1120108 Signed-off-by: Chris Wilson commit 80044e54634d0836694d5aa6f98ce22fe38d367f Author: Chris Wilson Date: Sat Feb 9 09:57:26 2013 +0000 sna: Promote to GPU is only partially damaged on the CPU but busy on the GPU Signed-off-by: Chris Wilson commit d18cb72a94fad0ee99ab361c21d643c927d29c35 Author: Chris Wilson Date: Fri Feb 8 22:31:19 2013 +0000 sna: Randomly perturb 'wedged' to hunt for faults Signed-off-by: Chris Wilson commit ce9f0448367ea6a90490a28150bfdc0a76500129 Author: Chris Wilson Date: Fri Feb 8 16:01:54 2013 +0000 sna/gen6: Use GT2 settings for both GT2 and GT2+ Signed-off-by: Chris Wilson commit ae5399aaf9ef57d33e8fd957e8a96964897c09b3 Author: Chris Wilson Date: Fri Feb 8 11:31:21 2013 +0000 sna: Force the fallback path for unaccelerated randr damage Signed-off-by: Chris Wilson commit c69b4389abc324533a9a311c17a667bf8a1e1673 Author: Chris Wilson Date: Thu Feb 7 22:54:37 2013 +0000 sna/gen4: Split the have_render flag in separate prefer_gpu hints The idea is to implement more fine-grained checks as we may want different heuristics for desktops with GT1s than for mobile GT2s, etc. Signed-off-by: Chris Wilson commit bec99de812ce6a1bbc2c8e4cfd05f4f74c560ea6 Author: Chris Wilson Date: Fri Feb 8 00:53:10 2013 +0000 sna: Remove the bogus assertions on buffer domains Just a few lines earlier we already have the correct assertion that the buffer was not in the GPU domain, so had these two been correct, they would have still been redundant. Signed-off-by: Chris Wilson commit 8d1d3c6e6102ff20fbff74ec6b3b2e94ee757015 Author: Chris Wilson Date: Thu Feb 7 14:47:07 2013 +0000 sna: Fixup an invalid assertion We may choose to operate inplace on a buffer last used by the CPU if we are discarding all the existing damage. Signed-off-by: Chris Wilson commit c405dba367bdca51221bd2464213199783dc18fe Author: Chris Wilson Date: Thu Feb 7 13:41:42 2013 +0000 sna: Also assert that the GPU is not wedged before continuing a batch Signed-off-by: Chris Wilson commit feeff6fcefccdca5335fea55c2fdbf8a4004c175 Author: Chris Wilson Date: Thu Feb 7 13:33:58 2013 +0000 sna: Force GTT readback if the GPU is wedged Signed-off-by: Chris Wilson commit 8a272971d5971a56f57dde00dceb082d0b142c8c Author: Chris Wilson Date: Wed Feb 6 17:59:10 2013 +0000 sna: Allow inplace uploads to utilise GTT on LLC machines Rather than arbitrarily disable the fallback paths for LLC, allow it to utilise any available GTT buffers for inplace uploads. The best explanation so far is that with the streaming is that we are trashing the LLC. On other machines, the difference is in the noise. Signed-off-by: Chris Wilson commit bc8a2c30c4f6bb9ce751b6717a3a2feaea0d6d4b Author: Chris Wilson Date: Thu Feb 7 10:42:58 2013 +0000 sna: Only try the SRC fixup into the buffer if it is CPU mapped On one particular machine, this operation is behaving as if it is reading back UC memory during the explicit write-only composite. Signed-off-by: Chris Wilson commit 889ed28f52bccdbc54692ea075f95f9635a8d58a Author: Chris Wilson Date: Thu Feb 7 10:42:21 2013 +0000 sna: Correctly align used buffers to the following page boundary Signed-off-by: Chris Wilson commit 974b6a97d78dadf09be8a2c4f61020f15d80d558 Author: Chris Wilson Date: Wed Feb 6 17:02:27 2013 +0000 sna: Fallback to non-LLC paths after an allocation failure for an LLC buffer Signed-off-by: Chris Wilson commit 5c8084ef04cb0a7da064fb1e13c8ef7dae528b1b Author: Chris Wilson Date: Wed Feb 6 16:39:31 2013 +0000 intel: Becareful not to match UMS against future generations Signed-off-by: Chris Wilson commit be241fb25ed0a8d41a642ea811253207f88d0962 Author: Chris Wilson Date: Wed Feb 6 16:38:12 2013 +0000 sna: Free the handle after pwrite buffer allocation failure Having just allocated the handle, we need to free it if we then fail to allocate memory for the buffer. Signed-off-by: Chris Wilson commit 4b3b25f0be33d3af3ccecfb3193fc2d365445fdf Author: Chris Wilson Date: Wed Feb 6 16:37:21 2013 +0000 sna: Flush our caches if we fail to mmap an object The likely cause for a mmap failure is that we hold too many objects open or have exhausted our address space. In both cases, we need to trim our caches before continuing. Signed-off-by: Chris Wilson commit daba1ae3e7f0532cc53d9a5178778dbaec203052 Author: Chris Wilson Date: Wed Feb 6 16:17:36 2013 +0000 sna: Correctly handle failure to CPU map a new allocation If we fail to CPU map, we want to fallback to just using pwrite with normal memory. Signed-off-by: Chris Wilson commit 0adb0b5e1ebcf3ddfeddae99d96912ec4c090832 Author: Chris Wilson Date: Wed Feb 6 16:02:30 2013 +0000 sna: Handle mapped buffer allocation failure for LLC The presumption was that if we had LLC we would have allocated the buffer by that point - however, it was remotely possible to have fallen through and so we need to handle those cases. Signed-off-by: Chris Wilson commit f4cff22afae598f41adf36cd149223d1f7dd6b6e Author: Chris Wilson Date: Wed Feb 6 15:15:36 2013 +0000 sna: Relax the buffer size assertion to only be larger than required Not all paths request alloc pages, a few just request sufficient pages for the original size. So we can only assert that condition is satisfied. Signed-off-by: Chris Wilson commit 8bc593c732a2f1ccd1bdabc071c709a44222db61 Author: Chris Wilson Date: Wed Feb 6 15:11:00 2013 +0000 sna: Make sure we always replace io buffers before inserting into the cache Signed-off-by: Chris Wilson commit 5f72158919098dd5684d1c56d1ba643cc3be2c7d Author: Chris Wilson Date: Wed Feb 6 15:10:23 2013 +0000 configure: XvMC support is optional, so make failure to find xcb non-fatal commit cd6d8f9b9df02934ebfff76cb40410c8ce3887dd Author: Chris Wilson Date: Wed Feb 6 10:37:50 2013 +0000 xvmc: Add the complementary XCB_CFLAGS After splitting the xvmc dependences into xcb and non-xcb, we then also have to add the xcb CFLAGS to build libIntelXVmc.la Reported-by: Julien Cristau Signed-off-by: Chris Wilson commit b96ee47ad97943c3dccd40d9570e29002dc3d85f Author: Paul Menzel Date: Sun Feb 3 13:33:08 2013 +0100 configure.ac: Split out XCB libraries from `XVMCLIB` into `XCB` Building the package under Debian Sid/unstable, `dh_shlibdeps` informs that `libI810XvMC.so.1.0.0` does not need to be linked against `libX11-xcb.so.1`, `libxcb-dri2.so.0`, `libxcb-util.so.0` or `libxcb.so.1` [1]. $ debuild -b -us -uc […] make[1]: Entering directory `/src/xserver-xorg-video-intel' dh_shlibdeps -- --warnings=6 dpkg-shlibdeps: Warnung: debian/xserver-xorg-video-intel/usr/lib/libI810XvMC.so.1.0.0 sollte nicht gegen libX11-xcb.so.1 gelinkt werden (es verwendet keines der Bibliotheks-Symbole) dpkg-shlibdeps: Warnung: debian/xserver-xorg-video-intel/usr/lib/libI810XvMC.so.1.0.0 sollte nicht gegen libxcb-dri2.so.0 gelinkt werden (es verwendet keines der Bibliotheks-Symbole) dpkg-shlibdeps: Warnung: debian/xserver-xorg-video-intel/usr/lib/libI810XvMC.so.1.0.0 sollte nicht gegen libxcb-util.so.0 gelinkt werden (es verwendet keines der Bibliotheks-Symbole) dpkg-shlibdeps: Warnung: debian/xserver-xorg-video-intel/usr/lib/libI810XvMC.so.1.0.0 sollte nicht gegen libxcb.so.1 gelinkt werden (es verwendet keines der Bibliotheks-Symbole) make[1]: Leaving directory `/src/xserver-xorg-video-intel' […] Moving `x11-xcb`, `xcb-dri2` and `xcb-aux` from `XVMCLIBS` into `XCB` and adding `XCB_LIBS` only to the `LIBADD` variables of `libIntelXvMC` makes the warnings go away and the libraries are still built without any issues. make[1]: Entering directory `/src/xserver-xorg-video-intel' dh_shlibdeps -- --warnings=6 make[1]: Leaving directory `/src/xserver-xorg-video-intel' dh_installdeb -O--builddirectory=build/ dh_xsf_substvars -O--builddirectory=build/ dh_gencontrol -O--builddirectory=build/ dpkg-gencontrol: Warnung: Feld Depends von Paket xserver-xorg-video-intel-dbg: unbekannte Substitutionsvariable ${shlibs:Depends} dh_md5sums -O--builddirectory=build/ dh_builddeb -O--builddirectory=build/ dpkg-deb: Paket »xserver-xorg-video-intel« wird in »../xserver-xorg-video-intel_2.19.0-6.1_i386.deb« gebaut. dpkg-deb: Paket »xserver-xorg-video-intel-dbg« wird in »../xserver-xorg-video-intel-dbg_2.19.0-6.1_i386.deb« gebaut. dpkg-genchanges -b >../xserver-xorg-video-intel_2.19.0-6.1_i386.changes dpkg-genchanges: rein binärer Upload - es ist kein Quellcode hinzugefügt dpkg-source --after-build xserver-xorg-video-intel dpkg-buildpackage: Binärpaket(e) hochzuladen (keine Quellen enthalten) Now running lintian... W: xserver-xorg-video-intel: hardening-no-relro usr/lib/libI810XvMC.so.1.0.0 W: xserver-xorg-video-intel: hardening-no-fortify-functions usr/lib/libI810XvMC.so.1.0.0 W: xserver-xorg-video-intel: hardening-no-relro usr/lib/libIntelXvMC.so.1.0.0 W: xserver-xorg-video-intel: hardening-no-fortify-functions usr/lib/libIntelXvMC.so.1.0.0 W: xserver-xorg-video-intel: hardening-no-relro usr/lib/xorg/modules/drivers/intel_drv.so W: xserver-xorg-video-intel: hardening-no-fortify-functions usr/lib/xorg/modules/drivers/intel_drv.so N: 1 tag overridden (1 warning) Finished running lintian. The modules were originally added with the following commit present since tag 2.10.0. commit 3e8f2eae3a586aa29be4858698e666e0ec778cea Author: Eric Anholt Date: Thu Oct 15 13:48:56 2009 -0700 XVMC: Use XCB DRI2 instead of cargo-culting our own copy of Xlib stuff. (v2) [1] https://buildd.debian.org/status/fetch.php?pkg=xserver-xorg-video-intel&arch=i386&ver=2%3A2.19.0-6&stamp=1347825458 Signed-off-by: Paul Menzel commit 93770c709aa7d3719b7c717040b16c8f82d5c207 Author: Paul Menzel Date: Tue Jan 22 10:47:22 2013 +0100 NEWS: Fix a typo: a*n* inadvertent commit a8cfddd280b5220f23565b21c91f3f7dd10bbe91 Author: Chris Wilson Date: Tue Feb 5 22:06:03 2013 +0000 sna: Tidy buffer allocation size assertions Rather than perilously update a local variable with the allocated size, just use the size of the bo in the assertion that is large enough to satisfy the allocation request. Reported-by: Jiri Slaby Signed-off-by: Chris Wilson commit 82dc91e8c24a1fbbf03dcf89a3955319b3399ea0 Author: Chris Wilson Date: Tue Feb 5 21:50:43 2013 +0000 test: Add a very basic blt benchmark Signed-off-by: Chris Wilson commit 9c80a0337ec12b6baab5aab380503e672e925677 Author: Chris Wilson Date: Tue Feb 5 14:56:10 2013 +0000 sna: ValleyView uses the same scanline registers as SandyBridge Signed-off-by: Chris Wilson commit 4c45e3fe456d211afc6ba69878b413a72ef5d0bf Author: Chris Wilson Date: Tue Feb 5 14:45:39 2013 +0000 intel: add more ValleyView PCI IDs Signed-off-by: Chris Wilson commit c6101d9d71a86a579ff9771d456b234a38bd80b7 Author: Chris Wilson Date: Tue Feb 5 11:02:30 2013 +0000 man: Fix a typo s/debuging/debugging/ Signed-off-by: Chris Wilson commit f33c90f7ada238683433d05492434120d06ea1fc Author: Chris Wilson Date: Fri Feb 1 19:34:56 2013 +0000 NEWS: Trivial typo s/utilile/utilise/ commit 6346c844525c2b3a82c16fe10485b901a2b5ddbc Author: Chris Wilson Date: Tue Feb 5 10:17:45 2013 +0000 sna/gen4: Remove old single-thread SF w/a The alternative of disabling GPU spans seems to be far more effective. Signed-off-by: Chris Wilson commit 1565917f10d9fb3c7e2e7e273173c38c364b9861 Author: Chris Wilson Date: Tue Feb 5 10:11:14 2013 +0000 sna/gen4: Disable non-rectilinear GPU span compositing This seems to be the primary victim of the render corruption, so disable until the root cause is fixed. References: https://bugs.freedesktop.org/show_bug.cgi?id=55500 Signed-off-by: Chris Wilson commit 37bc822190f36be7b021167ba4d306bbcd97957b Author: Damien Lespiau Date: Fri Jan 18 14:13:08 2013 +0000 build: Make generation of gen code depend on intel-gen4asm This way, when a new intel-gen4asm is available (because one just hacked on it and has installed a new version for instance) the shaders will be recompiled. This helps catching regressions, testing the latest changes in the assembler haven't broken too many things. Signed-off-by: Damien Lespiau Signed-off-by: Chris Wilson commit 18f8d2291fbb53ac993b926c247ca981e1e5207b Author: Damien Lespiau Date: Fri Jan 18 14:13:07 2013 +0000 build: Use $(AM_V_GEN) to silence the assembly of gen programs Signed-off-by: Damien Lespiau Signed-off-by: Chris Wilson commit eea535b7e2a35ec4dfa50550b674d4212676d2ee Author: Damien Lespiau Date: Fri Jan 18 14:13:06 2013 +0000 build: Make autoreconf honour ACLOCAL_FLAGS When running autoreconf, it's possible to give flags to the underlying aclocal by declaring a ACLOCAL_AMFLAGS variable in the top level Makefile.am. Putting ${ACLOCAL_FLAGS} there allows the user to set an environment variable up before running autogen.sh and pull in the right directories to look for m4 macros, say an up-to-date version of the xorg-util macros. Signed-off-by: Damien Lespiau Signed-off-by: Chris Wilson commit 9640640ab02d5de630e903116c1b104752f8b604 Author: Paul Menzel Date: Sat Feb 2 11:44:54 2013 +0100 configure.ac: Do not include `xext` and `xfixes` in `XVMCLIB` Building the package under Debian Sid/unstable, `dh_shlibdeps` informs that `libIntelXvMC.so.1.0.0` does not need to be linked against `libXext.so.6` or `libXfixes.so.3` [1]. $ debuild -b -us -uc […] make[1]: Entering directory `/build/buildd-xserver-xorg-video-intel_2.19.0-6-i386-9thLfo/xserver-xorg-video-intel-2.19.0' dh_shlibdeps -- --warnings=6 dpkg-shlibdeps: warning: debian/xserver-xorg-video-intel/usr/lib/libIntelXvMC.so.1.0.0 should not be linked against libXext.so.6 (it uses none of the library's symbols) dpkg-shlibdeps: warning: debian/xserver-xorg-video-intel/usr/lib/libIntelXvMC.so.1.0.0 should not be linked against libXfixes.so.3 (it uses none of the library's symbols) dpkg-shlibdeps: warning: debian/xserver-xorg-video-intel/usr/lib/libI810XvMC.so.1.0.0 should not be linked against libXext.so.6 (it uses none of the library's symbols) dpkg-shlibdeps: warning: debian/xserver-xorg-video-intel/usr/lib/libI810XvMC.so.1.0.0 should not be linked against libXfixes.so.3 (it uses none of the library's symbols) dpkg-shlibdeps: warning: debian/xserver-xorg-video-intel/usr/lib/libI810XvMC.so.1.0.0 should not be linked against libX11-xcb.so.1 (it uses none of the library's symbols) dpkg-shlibdeps: warning: debian/xserver-xorg-video-intel/usr/lib/libI810XvMC.so.1.0.0 should not be linked against libxcb-dri2.so.0 (it uses none of the library's symbols) dpkg-shlibdeps: warning: debian/xserver-xorg-video-intel/usr/lib/libI810XvMC.so.1.0.0 should not be linked against libxcb-util.so.0 (it uses none of the library's symbols) dpkg-shlibdeps: warning: debian/xserver-xorg-video-intel/usr/lib/libI810XvMC.so.1.0.0 should not be linked against libxcb.so.1 (it uses none of the library's symbols) dpkg-shlibdeps: warning: package could avoid a useless dependency if debian/xserver-xorg-video-intel/usr/lib/libIntelXvMC.so.1.0.0 debian/xserver-xorg-video-intel/usr/lib/libI810XvMC.so.1.0.0 were not linked against libXext.so.6 (they use none of the library's symbols) dpkg-shlibdeps: warning: package could avoid a useless dependency if debian/xserver-xorg-video-intel/usr/lib/libIntelXvMC.so.1.0.0 debian/xserver-xorg-video-intel/usr/lib/libI810XvMC.so.1.0.0 were not linked against libXfixes.so.3 (they use none of the library's symbols) make[1]: Leaving directory `/build/buildd-xserver-xorg-video-intel_2.19.0-6-i386-9thLfo/xserver-xorg-video-intel-2.19.0' dh_installdeb -a -O--builddirectory=build/ […] Not populating `XVMCLIB` with `xext` and `xfixes` makes the warning go away and the libraries are still built without any issues. [1] https://buildd.debian.org/status/fetch.php?pkg=xserver-xorg-video-intel&arch=i386&ver=2%3A2.19.0-6&stamp=1347825458 Signed-off-by: Paul Menzel commit 9807bba950078d86a25b91064ecfebaa0ee459e3 Author: Chris Wilson Date: Fri Feb 1 18:25:48 2013 +0000 sna: Drop bogus refcnt assertion during kgem_bo_retire() As we may call it kgem_bo_sync(), during preparation of the upload buffer which in turn may operate on an object straight out of the snoop cache and hence not yet referenced (or in some cases, ever). Signed-off-by: Chris Wilson commit a5561f13498066922b54af04cc71549322ce0e3b Author: Chris Wilson Date: Fri Feb 1 18:05:35 2013 +0000 sna: Do not add the INPLACE hint if we have the ASYNC hint set If the caller is preparing to use the GPU to rendering into the CPU bo, it will request an ASYNC migration. In those cases, we do not want to substitute it with an INPLACE operation. Signed-off-by: Chris Wilson commit d3ff1cb9d7f788002337b1e6c4c81c58112b85b1 Author: Chris Wilson Date: Fri Feb 1 13:46:33 2013 +0000 2.21.0 release commit 008f8230a7c47f1249eb51e53b3abf158f2a42bf Author: Chris Wilson Date: Fri Feb 1 01:54:52 2013 +0000 sna: Assert that if we have GPU damage we have a GPU bo Scatter the asserts around the migration points to catch where this invariant may be untrue. Signed-off-by: Chris Wilson commit cf0576f87102b1535268691e7e29661b0f9ee73b Author: Chris Wilson Date: Fri Feb 1 00:19:21 2013 +0000 sna/video: Correct computation of planar frame size The total frame size is less than 3 times the subsampled chroma planes due to the additional alignment bytes. Bugzilla: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1104180 Signed-off-by: Chris Wilson commit 268285d9a64fc47fe81fe5bfbfbd1890dad53e1e Author: Chris Wilson Date: Thu Jan 31 21:57:41 2013 +0000 sna/gen3+: Flush vertex threads before touching global state We need to be careful not just when finishing the current vbo to synchronize with the sharing threads, but also before we emit the batch state that no other thread will try and do the same. Signed-off-by: Chris Wilson commit 1239e012ae6d4f00ce73f32d7244905a601170ea Author: Chris Wilson Date: Thu Jan 31 19:18:17 2013 +0000 sna: Make sure the needs_flush is always accompanied by a tracking request References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 9712f49fddc8be939f77c25fcb907873af44619f Author: Chris Wilson Date: Thu Jan 31 18:08:05 2013 +0000 sna: Remove stale assertion Now the reset is meant to restablish 'rq' if the bo was busy. Signed-off-by: Chris Wilson commit cd7df0004cf6e423d2ae6c0cf83a84e0031161b4 Author: Chris Wilson Date: Thu Jan 31 17:32:57 2013 +0000 sna: Pass width/height to composite for rotated displays This is essential to handle displays that are too large to be rendered normally via the 3D pipeline and so that the bounds of the fixup region are known. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=60124 Signed-off-by: Chris Wilson commit 38376b56cfe0dfc603bce48e37432622ef9a0135 Author: Chris Wilson Date: Thu Jan 31 17:29:10 2013 +0000 sna: Remember to move scanouts to the scanout cache after retiring Reported-by: Jiri Slaby Signed-off-by: Chris Wilson commit 0a08de1f02577aef0da289108270c1b35e5d9703 Author: Chris Wilson Date: Thu Jan 31 16:39:47 2013 +0000 sna: After removing the bo from a batch, check whether it is still busy If we transfer a bo to the current batch, then subsequently discard it, we lose the information about its current active state. Try to recover this information, by querying the kernel and adding it to the flushing list if necessary. Reported-by: Jiri Slaby Signed-off-by: Chris Wilson commit fff0686342f8ec3b3f3510340e073defdf2fb73f Author: Chris Wilson Date: Thu Jan 31 12:40:21 2013 +0000 sna/traps: Thread the fallback rectilinear compositor Signed-off-by: Chris Wilson commit 839542d219bd919c99398d514c1d194d18b78eff Author: Chris Wilson Date: Thu Jan 31 12:08:52 2013 +0000 sna/traps: Allow inplace compositing for non-GPU buffers and rectilinear traps Signed-off-by: Chris Wilson commit e329e04b10c88afb40f2fd8fdad5b24b9f7dfc15 Author: Chris Wilson Date: Thu Jan 31 11:33:37 2013 +0000 sna/traps: Translate the extents for the rasterization threads The single-threaded code used the pre-computed width/height and only required the origin from the bounds. However, the threads need to allocate memory for themselves based on the computed bounds, and so it helps if those bounds are then correct (rather than only the top-left being in local space with the bottom-right in global coordinates). Signed-off-by: Chris Wilson commit 033f75e5bd94e226e719f87ed4e0091845384679 Author: Chris Wilson Date: Thu Jan 31 01:38:01 2013 +0000 sna: Stage retirement through the flushing list If the kernel replies that a bo is still busy, stage its retirement through the flushing list to be certain that we never stall on a subsequent write. Signed-off-by: Chris Wilson commit 5f5711e62cc4c8ca15782376c4047174299e2db0 Author: Chris Wilson Date: Thu Jan 31 01:21:08 2013 +0000 sna: Disable dangerous assertions that depend upon external state Signed-off-by: Chris Wilson commit 42529336fd92d39a5a5eceb07f2838d4be50fa8e Author: Chris Wilson Date: Thu Jan 31 00:58:51 2013 +0000 sna: Prevent falling back to swrast if source is on the GPU Currently if the dst is wholly contained within the CPU, then we try to continue to operate on the GPU. However, if we have FORCE_GPU set, it means that one of the sources for the operation resides on the GPU, and that would require a readback in order to perform the operation on the CPU. Hence, if we try to use a CPU bo and fail, convert back to using the GPU bo if forced. Signed-off-by: Chris Wilson commit c2d06c407e1c2cbbf3f7f6c4989710a799cd43d0 Author: Chris Wilson Date: Wed Jan 30 21:17:42 2013 +0000 sna: Improve DBG output for damaged slave outputs After computing the intersection of the damage with the slave, give the region extents. Signed-off-by: Chris Wilson commit 8867aa6a46c33fd2abf3b3f0b1d6115bad6c8017 Author: Chris Wilson Date: Wed Jan 30 21:15:55 2013 +0000 sna/dri: Handle change of BackBuffer across a pending flip If we encounter a delayed flip with a different back buffer than the current, simply update the info rather than bug out. Signed-off-by: Chris Wilson commit a31fd03bd4c87c48dc3ca15e3082e29348224b8c Author: Chris Wilson Date: Wed Jan 30 17:26:28 2013 +0000 sna: Add a bunch of assertions to make sure we do not misplace scanouts As scanouts are uncached, they need to be treated carefully and decontaminated before being placed in the general cache. So double check that no bo in those caches are still marked as a scanout. Signed-off-by: Chris Wilson commit 6f1b862282ddb4545987fb9f0a45b528b7b7b5ee Author: Chris Wilson Date: Wed Jan 30 15:44:53 2013 +0000 sna: Pass the correct WRITE hint when migrating for rendering into the CPU bo Signed-off-by: Chris Wilson commit 5011ed2e729d46fe3cff5454e15a0fd16441f7e1 Author: Chris Wilson Date: Wed Jan 30 15:44:22 2013 +0000 sna: Only discard the clear hint when writing inplace to the GPU pixmap Signed-off-by: Chris Wilson commit 6312f58014c0bb4afa56855be1e9becc3e3cc3d7 Author: Chris Wilson Date: Wed Jan 30 15:43:05 2013 +0000 sna: Don't force a migration from CPU rendering for a DRI2 flushed pixmap Signed-off-by: Chris Wilson commit 60a3b370aea0cf9ffb4947a73984c877b4695d4e Author: Chris Wilson Date: Wed Jan 30 15:41:51 2013 +0000 sna: Retire the bo after a set-domain(CPU,0) Having relaxed the earlier assertion because the kernel is wrong, we can now retire for READ-READ optimisations. Signed-off-by: Chris Wilson commit 78ad5a742f40c2311bfe90997aebedeb998464e5 Author: Chris Wilson Date: Wed Jan 30 15:40:06 2013 +0000 sna: Relax assertion the the kernel considers the bo idle when we call retire All the callers have explicitly changed the domain upon the bo before calling kgem_bo_retire(), so we still get the occasional sporadic failure as kgem_busy() reports true. Kill the assertion for now. Signed-off-by: Chris Wilson commit 83bcd310d279758542e366348f808d7ca0f6d0bb Author: Chris Wilson Date: Wed Jan 30 13:18:21 2013 +0000 sna: Prefer to use snooped buffers for readbacks Signed-off-by: Chris Wilson commit 496f3ff04453524639a52a3b9dfcb8e198e5e597 Author: Chris Wilson Date: Wed Jan 30 12:21:33 2013 +0000 uxa: Harden against failures to submit batchbuffers If we fail to submit a batchbuffer, the driver is broken and likely to continue to fail to render. Give up, and fallback to swrast so that the session remains usable. References: https://bugs.freedesktop.org/show_bug.cgi?id=59771 Signed-off-by: Chris Wilson commit 04d48fee713e7bbc9cdf4f09855f6663a4bdc59f Author: Chris Wilson Date: Wed Jan 30 11:46:20 2013 +0000 sna: Fix errors found from asserts in a66c5f9ed51e Signed-off-by: Chris Wilson commit bc8b191ef6f5030d17a3b6497d1fd7556756c1ff Author: Chris Wilson Date: Wed Jan 30 09:04:10 2013 +0000 sna: Return early if the Drawable box exactly matches one CRTC If we are trying to find the best coverage, then by definition if the drawable is an exact match for one CRTC, we can stop looking. Signed-off-by: Chris Wilson commit de28027ffc649920268ae6fdd64146f08310e8a4 Author: Chris Wilson Date: Wed Jan 30 08:42:48 2013 +0000 sna/dri: Make sure we discard the existing mappings when swapping GPU bo If the GPU bo is currently mapped to the Pixmap, we need to be sure to invalidate that mapping if we swap the GPU bo (for SwapBuffers). If we forget, we leave a dangling pointer to chase. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=60042 Signed-off-by: Chris Wilson commit cf9b9ac3186299ab2418c55e73e19c81e5f615a4 Author: Chris Wilson Date: Wed Jan 30 08:40:53 2013 +0000 sna: Only discard the mapping prior to the actual read when uploading Signed-off-by: Chris Wilson commit a66c5f9ed51e1dcfc2ab03339795b73617629196 Author: Chris Wilson Date: Wed Jan 30 08:22:00 2013 +0000 sna: Before replacing the devPrivate.ptr assert it is not already mapped Signed-off-by: Chris Wilson commit 3fdd28419adee7145d3925cff2704143a324e9d3 Author: Chris Wilson Date: Tue Jan 29 22:26:15 2013 +0000 sna: Only migrate the sample box if using the BLT engine for a composite Modify the presumption that if we are using a core operation on a shadow pixmap, then we are likely to continue migrating that pixmap back and forth. Signed-off-by: Chris Wilson commit 0c3b0f11d718d915e502582e9fadd5c0577640db Author: Chris Wilson Date: Tue Jan 29 22:24:30 2013 +0000 sna: Verify that we always add the SHM CPU bo to the flush list when using As we need to synchronize that bo before the next reply, we need to keep track of it whenever it is active on the GPU. Signed-off-by: Chris Wilson commit f743cd5734ca502aa8bdb0e1327fe84d6ce82755 Author: Chris Wilson Date: Tue Jan 29 18:04:40 2013 +0000 sna: Avoid promoting SHM CPU bo to GPU to maintain coherence with SHM clients Signed-off-by: Chris Wilson commit 9383c5efe9ace34970abddc5e3c84c32505b537f Author: Chris Wilson Date: Tue Jan 29 17:24:24 2013 +0000 sna/gen3+: Fix a DBG for composite_boxes() Signed-off-by: Chris Wilson commit b02a1ea5573b6f0b58a037dd4788c04c296f7ff3 Author: Chris Wilson Date: Tue Jan 29 09:28:33 2013 +0000 sna: Add GT1/GT2 thread counts for Haswell Signed-off-by: Chris Wilson commit 1dc2d9ede5c7f330ebadf85d987559c8a6cb1c6b Author: Chris Wilson Date: Mon Jan 28 23:14:57 2013 +0000 sna: Add some more paranoia that we correctly map before fallbacks Signed-off-by: Chris Wilson commit 63c71bcd96202e6da44d6776d119a82f0c06d386 Author: Chris Wilson Date: Sun Jan 27 23:17:13 2013 +0000 sna: Fix typo in vertex count for threaded source span emitter Signed-off-by: Chris Wilson commit b0d26ca9312695d05c29503a3f892e7f2c5816dd Author: Chris Wilson Date: Sun Jan 27 21:07:03 2013 +0000 sna: Replace the forced vertex finish with just a wait When completing a batch mid-operation, we need to wait upon the other threads to complete their writes so that memory is coherent before submitting the work to the GPU. This was achieved by forcing the finish, but all that from that is the wait, which makes the handling of threads much explicit and removes the unnecessary vbo refresh. Signed-off-by: Chris Wilson commit b0c3170c1092d01b4937f352a3962854785ee549 Author: Chris Wilson Date: Sun Jan 27 19:09:38 2013 +0000 sna: Add the pixmap to the flushing list when creating for inplace CPU writes Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 73f574945f2cac14f9bafa6395e2c4dbb16fcf5d Author: Chris Wilson Date: Sun Jan 27 16:02:52 2013 +0000 sna: Disable all signals in the render threads X uses them (SIGIO especially) for input handling, and gets rightfully confused if it finds itself in a different thread. Signed-off-by: Chris Wilson commit 9a7bf70365980809d0f02190f2f620a957ff1ba8 Author: Chris Wilson Date: Sat Jan 26 23:03:33 2013 +0000 sna: Enable threaded rasterisation for non-antialiased geometry Signed-off-by: Chris Wilson commit 8178cff5718e69e14d3953a7f754d7585a06838f Author: Chris Wilson Date: Sat Jan 26 14:41:04 2013 +0000 sna: Begin sketching out a threaded rasteriser for spans Signed-off-by: Chris Wilson commit 8ffb3f50b3b4601401da76e2848e059ab63231f4 Author: Chris Wilson Date: Fri Jan 25 10:45:39 2013 +0000 sna: Spawn threads to rasterize trapezoids through pixman Signed-off-by: Chris Wilson commit 0ec2f3a8bac96acc55c8fdb432b97d026abaafb4 Author: Chris Wilson Date: Thu Jan 24 23:10:39 2013 +0000 sna: Spawn threads to composite trapezoids inplace Signed-off-by: Chris Wilson commit 427b7311fe1b66d54518bae45e9fa149bda8a6e8 Author: Chris Wilson Date: Thu Jan 24 22:25:46 2013 +0000 sna: Perform the last threaded composite operation directly The point of the refactor was to execute the last stage of the composite in the master thread, so do so. Signed-off-by: Chris Wilson commit 326dcd75f2202b1af29e986f5efb6b1e133217cb Author: Chris Wilson Date: Thu Jan 24 20:58:53 2013 +0000 sna: Parse cpuinfo to determine the actual number of physical cores/caches Signed-off-by: Chris Wilson commit f597b647180c1e7bf83693060f244926191b7462 Author: Chris Wilson Date: Thu Jan 24 18:45:35 2013 +0000 sna: Tidy construction of data for threaded composite Signed-off-by: Chris Wilson commit 1643c97f8f7b49738b649b5f7d1e574d689d167e Author: Chris Wilson Date: Thu Jan 24 18:24:02 2013 +0000 sna: Use threads for simple mask generation Signed-off-by: Chris Wilson commit d60128c55e8f5f69476d42c20f2fd62ccc0f411e Author: Chris Wilson Date: Thu Jan 24 15:41:29 2013 +0000 sna/dri: Compensate clipExtents for drawable offset The clipExtents is in screen coordinates whereas we just want to confirm that the maximum pixel to be copied lies with the DRI2 buffer, which is relative to the drawable. Reported-by: Matthieu Baerts Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=59806 Signed-off-by: Chris Wilson commit 264b3b72500c5af74d124a214347d45c9cb90a1d Author: Chris Wilson Date: Thu Jan 24 15:06:12 2013 +0000 sna: Refactor to use a common fbComposite fallback Signed-off-by: Chris Wilson commit 8ecfbea9d1f83b2de62bee0f58299e7a90c741d1 Author: Chris Wilson Date: Thu Jan 24 14:46:03 2013 +0000 sna: Experiment with a threaded renderer for fallback compositing Signed-off-by: Chris Wilson commit 778dba90cfc4e801a975bd661c56a565ce60524b Author: Chris Wilson Date: Wed Jan 23 21:32:29 2013 +0000 sna/dri: Don't contribute missed frames to the target_msc Signed-off-by: Chris Wilson commit 50b41cb485ffb38e6bf705a3a62840bb78af669b Author: Chris Wilson Date: Wed Jan 23 21:16:49 2013 +0000 sna/dri: Only reject DRI2 buffers that are too small for the request blit The goal is to reject stale DRI2 buffers that are smaller than the target due to not-yet-handled ConfigureNotify, but not to reject blitting from Windows that are larger than the frontbuffer. Fixes a regression from the overzealous commit b27ecf3059bc066ef59f2a71c1d8d8f0ffec7191 Author: Chris Wilson Date: Mon Nov 12 14:06:06 2012 +0000 sna/dri: Prevent scheduling a swap on stale buffers Signed-off-by: Chris Wilson commit 98b312e579385e6e4adf6bf0abe20f8ca84592af Author: Chris Wilson Date: Wed Jan 23 20:51:35 2013 +0000 sna/dri: Stop feeding I915_TILING_Y to mesa i915c Only i915g handles Y-tiling, and we can't differentiate between the two types of clients. Signed-off-by: Chris Wilson commit 31796400915a06fc789088b7dcfcecd6ea91e195 Author: Chris Wilson Date: Wed Jan 23 19:37:23 2013 +0000 sna: Clean up WAIT_FOR_EVENT on gen2/3 Signed-off-by: Chris Wilson commit ea8148b24d48db4f46205817db8a55dd6ea1a4b3 Author: Chris Wilson Date: Wed Jan 23 17:47:12 2013 +0000 sna/dri: Prefer to use the BLT ring for vsync'ed copies on IVB+ Signed-off-by: Chris Wilson commit 3c3a87a2d4261cbd66602812637328a04787f510 Author: Chris Wilson Date: Wed Jan 23 17:35:50 2013 +0000 sna/gen6: Correct the event definition for secondary pipes for MI_WAIT_FOR_EVENT It helps to wait upon the event we program and enable. References: https://bugzilla.kernel.org/show_bug.cgi Signed-off-by: Chris Wilson commit 88753c5a8c6c9acf086d81828260adf330eebb1a Author: Chris Wilson Date: Wed Jan 23 17:35:50 2013 +0000 sna/gen7: Correct the event definition for secondary pipes for MI_WAIT_FOR_EVENT It helps to wait upon the event we program and enable. References: https://bugzilla.kernel.org/show_bug.cgi Signed-off-by: Chris Wilson commit 2d92d8ec562cb1e6b9dca28074adca670734233c Author: Chris Wilson Date: Tue Jan 22 09:24:04 2013 +0000 sna: Extend rectangular PolyLines to cover corner pixels on ccw paths Reported-by: Joe Peterson Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55484 Signed-off-by: Chris Wilson commit c8817e24a9d97110a961c3803290e38ff5cbfc9a Author: Chris Wilson Date: Tue Jan 22 09:06:50 2013 +0000 sna/gen7: Fix inversion of bool return code from CA pass As we inverted the predicate, we no longer restored the original operation after performing a CA pass - glyph would randomly become white. Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 10f549332e315cfe2cc86aadab94a95ae6757c34 Author: Chris Wilson Date: Mon Jan 21 16:48:34 2013 +0000 sna: Free a non-reusable bo if it expires on the flushing list Still no sure just how the bo ends up there, but as there seems to be the occasional malinger, just free it. Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit d7f0df27edb20b052ad39beb26a0b1924f432618 Author: Chris Wilson Date: Mon Jan 21 16:34:09 2013 +0000 sna: Use the maximum backlight value if we fail to read the current value Signed-off-by: Chris Wilson commit 46a3a68e60a1d0a598ec8ece81088a4e6491de55 Author: Chris Wilson Date: Mon Jan 21 16:29:30 2013 +0000 sna: Assert that if marked as a scanout it is indeed bound. On further review, the invariant must have been violated earlier, so make the assert earlier. Signed-off-by: Chris Wilson commit 0507d55dd1bc8fedae524a410a9e7b53f1dad920 Author: Chris Wilson Date: Mon Jan 21 16:24:49 2013 +0000 sna: Only add bound scanouts to the scanout list If we never used the bo as an actual scanout it will never have had been moved to the uncached domain and so we can return it back to the system cache. Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 5a0bc67ba57cf698e100df617474669ed5d036d6 Author: Chris Wilson Date: Mon Jan 21 11:41:38 2013 +0000 sna: New execbuffer flags for lut-handle and fast-relocs are upstream Now the flags are upstream, we can rely on runtime tests as the interface is now frozen. Signed-off-by: Chris Wilson commit 208ca91a31182e8ddad36e6a735c725362cbd071 Author: Chris Wilson Date: Sun Jan 20 18:02:41 2013 +0000 sna/gen7: Place the vsync commands in the same cacheline Do as told; both the LRI and WAIT_FOR_EVENT need to be in the same cacheline for an unspecified reason. Signed-off-by: Chris Wilson commit 9a3e3abfe9b624af2354c5a69778aee3024fe46c Author: Chris Wilson Date: Sun Jan 20 17:48:31 2013 +0000 sna/gen7: Offset start/end scanlines by one The hardware needs to be programmed with the line before the desired scanline, wrapping around as required. Signed-off-by: Chris Wilson commit e6a64f872bfd026aa1ba1bd44b1298918c819849 Author: Chris Wilson Date: Sun Jan 20 16:59:58 2013 +0000 sna/gen3+: Remove bogus assertion that the vbo in included before finish If we are carrying over a nearly full vbo from one batch to the next, we may indeed finish it prior to writing any new primitives and so the assert is truly bogus. Signed-off-by: Chris Wilson commit 5de919336fc1ba1c4116e18ba0560cdb7b0589f0 Author: Chris Wilson Date: Sun Jan 20 16:36:17 2013 +0000 sna/gen6: Tweak programming scanline values The documentation says that both start/end scanline need to be the line before the desired value, and so to program the first scanline we need to set it to the last scanline. The docs also say that the lower 3 bits are ignored, so tweaked the values programmed accordingly with an extra check that the window is not reduced to 0. Signed-off-by: Chris Wilson commit 2f9ac4e8a17e9d60bbb55c46929c37e92181d804 Author: Chris Wilson Date: Sun Jan 20 15:53:32 2013 +0000 sna/gen3+: And restore non-CA compositing state after the CA pass Signed-off-by: Chris Wilson commit 650c9d5ce80afc1d4c8d9f77f6679f085fa4dc9d Author: Chris Wilson Date: Sun Jan 20 14:58:42 2013 +0000 sna/gen3+: Reset vertex relocation state after discarding the batch Fixes a regression from commit a6ecb6d31d8c543f38fca0be6b0ec82e59dcd8d2 Author: Chris Wilson Date: Wed Jan 16 09:14:40 2013 +0000 sna: Discard the batch if we are discarding the only buffer in it as we may keep a stale relocation for the vertex buffer alive if we attempt to clear the bo using the render engine before discarding it. Signed-off-by: Chris Wilson commit 492952e0d6362a046a666956afdf8f9bc0f2b7e7 Author: Chris Wilson Date: Sun Jan 20 14:55:06 2013 +0000 sna/gen3+: Handle flushing vbo for CA glyphs Signed-off-by: Chris Wilson commit b52c921204df6b2486717fcef05b4a1993aa1071 Author: Chris Wilson Date: Sun Jan 20 14:02:07 2013 +0000 sna: Adapt error detection and handling for invalid batchbuffers Allow the DDX to continue even if the kernel rejects our batchbuffers by disabling hw acceleration - just extends the existing hang detection to also handle the driver producing garbage. Signed-off-by: Chris Wilson commit 8215a278f20d34819536edbda05a108a860fefb9 Author: Chris Wilson Date: Sun Jan 20 12:36:07 2013 +0000 sna/gen3: Always close the vertices for a batch, even if the vbo is empty In the case where we emit a no-op, we may not attempt to finish binding the vbo as it is considered empty. This leaves a stray relocation for the next batch, and also causes it to believe that it has a vbo bound already. Signed-off-by: Chris Wilson commit a88a9b9a59fa2d5fd427fa6e1f74fb9844379264 Author: Chris Wilson Date: Sun Jan 20 12:06:09 2013 +0000 2.20.19 release commit 7822bbacbece6fcb2e12863cd6c7a53ab614c37c Author: Chris Wilson Date: Sun Jan 20 11:43:49 2013 +0000 test: Add script to generate source file for testing vsync Courtesy of an original script by Mark Schreiber, https://bugs.freedesktop.org/show_bug.cgi?id=59606 commit 9329d8755981989ccbe66df6085fbab7c809a2c6 Author: Chris Wilson Date: Sun Jan 20 10:14:21 2013 +0000 sna: Make DEBUG_SYNC a configure option As it is advisable to combined the synchronous rendering debug option with other debugging options, it is more convenient to make it into a configure option: --enable-debug=sync Signed-off-by: Chris Wilson commit c9263f192e2f85dd961bc1c4e9ca8180db874517 Author: Chris Wilson Date: Sun Jan 20 01:39:12 2013 +0000 sna: Apply DEBUG_SYNC prior to emitting error report This is handy for the case where the batch triggers a GPU hang rather than being rejected by the kernel. Signed-off-by: Chris Wilson commit 42ab789cce8423d99864776c6d5ba759c4129b54 Author: Chris Wilson Date: Fri Jan 18 13:56:53 2013 +0000 sna: Clear the non-intersecting damage after skipping the slave update Signed-off-by: Chris Wilson commit 828a3a80aa3f0692e7be2831d58bccf02e2c481d Author: Chris Wilson Date: Fri Jan 18 13:16:23 2013 +0000 uxa: Clip dirty region to slave pixmap before appending damage Fixes regression from commit c789d06cf8a0debc67058d7be1483f5b542e2baa Author: Dave Airlie Date: Mon Jan 7 13:57:21 2013 +1000 intel: fixup damage posting to be done correctly around slave pixmap which causes the entire slave scanout to be readback from uncached memory every time a pixel is modified. Reported-by: Stephen Liang Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=59539 Signed-off-by: Chris Wilson commit e17eaf540b614cdcb8f7349dd01852c3afc5ab05 Author: Chris Wilson Date: Fri Jan 18 13:09:36 2013 +0000 sna: Replace double negative '!RegionNotEmpty' with the equivalent RegionNil Signed-off-by: Chris Wilson commit 2de43a0164ba5364ffd7cb48f0bccc9873e87332 Author: Chris Wilson Date: Fri Jan 18 12:01:54 2013 +0000 sna: Skip an empty slave update Signed-off-by: Chris Wilson commit 38de17f80d780bf219fc3c4018ad9cc8808ba50f Author: Chris Wilson Date: Fri Jan 18 10:16:42 2013 +0000 sna: Remove bogus assertion invalidated by 'read-read' sync If we perform a read-read synchronisation, the kernel may still believe that the bo is busy as it remains on the active lists being read by the GPU. Signed-off-by: Chris Wilson commit 9f68ac60ae37cc72503ec40691d1ae43a476f8e7 Author: Chris Wilson Date: Thu Jan 17 20:00:34 2013 +0000 sna/dri: Explicitly flag sync copies for the backends As gen6/7 need to prevent ring switching and perform a rendercopy if we need to perform a vsync'ed copy. Signed-off-by: Chris Wilson commit 1ee00c408d8142cfaf4202393c2364c9ae73cb6e Author: Chris Wilson Date: Thu Jan 17 13:09:47 2013 +0000 sna/trapezoids: Fix horizontal offset for inplace operation Remember that for an inplace operation we are not dealing with an a8 mask, but rather a x8r8g8b8 surface and so need to step accordingly. Signed-off-by: Chris Wilson commit 0d749f93ea52161e59da1adca1a22e96ba293551 Author: Chris Wilson Date: Thu Jan 17 12:28:18 2013 +0000 sna: Drop the MOVE_WHOLE_HINT for PutImage It is not as clearly beneficial as for GetImage, as for example toolkits may only push the shadows around a window. Signed-off-by: Chris Wilson commit dc643ef753bcfb69685f1eb10828d0c8f830c30e Author: Chris Wilson Date: Thu Jan 17 12:27:55 2013 +0000 sna: Apply read-only synchronization hints for move-to-cpu Signed-off-by: Chris Wilson commit 18035a21e147788bea03ab2175ca03ae951701ce Author: Chris Wilson Date: Thu Jan 17 11:52:10 2013 +0000 sna: Remove the confusion of the pixmap->undamaged This was to track a pixmap that had been used for migration (i.e had in the past been used for mixed rendering). It is no longer used so remove it. Signed-off-by: Chris Wilson commit 46141d277f326ae78f7b0e927a500e0eb1987f1b Author: Chris Wilson Date: Thu Jan 17 10:16:24 2013 +0000 sna: Consider fill style for XPolyRectangle The rectangle outline is not always solid... Signed-off-by: Chris Wilson commit d5c8d38afaba04281157bafe212e93f010ae00f5 Author: Chris Wilson Date: Thu Jan 17 10:10:54 2013 +0000 sna: Refactor to remove a goto from sna_put_zpixmap_blt() The complexity of the function has been moved to move-to-cpu so we can take further advantage of the simplified logic in put_zpixmap to clean up the code by removing an unwanted goto. Signed-off-by: Chris Wilson commit 9552438caa4d295c99a9b8821cf2644739861c6a Author: Colin Walters Date: Wed Jan 4 17:37:06 2012 -0500 autogen.sh: Implement GNOME Build API http://people.gnome.org/~walters/docs/build-api.txt Signed-off-by: Adam Jackson commit 87d773249af18ae8722aacb7306b0eee51a90dbc Author: Adam Jackson Date: Wed Jan 16 13:18:23 2013 -0500 configure: Drop AM_MAINTAINER_MODE Signed-off-by: Adam Jackson commit dbf1cfec9cd4e9efe7650f2940c92b4e51214288 Author: Chris Wilson Date: Wed Jan 16 12:20:48 2013 +0000 2.20.18 release Signed-off-by: Chris Wilson commit 47caffc50b5cdd288ad868fa9a697f0d4e2d28dc Author: Chris Wilson Date: Wed Jan 16 10:49:24 2013 +0000 sna: Restrict upload buffers to reduce sampler TLB misses Signed-off-by: Chris Wilson commit ab36300a22222086b94857f356612106ffbeb480 Author: Chris Wilson Date: Wed Jan 16 09:17:59 2013 +0000 sna: Correct DBG to refer to the actual tiling mode forced Signed-off-by: Chris Wilson commit a6ecb6d31d8c543f38fca0be6b0ec82e59dcd8d2 Author: Chris Wilson Date: Wed Jan 16 09:14:40 2013 +0000 sna: Discard the batch if we are discarding the only buffer in it Signed-off-by: Chris Wilson commit 26db2438e34feb8f28444bf7418869b4ecd870da Author: Chris Wilson Date: Wed Jan 16 09:00:21 2013 +0000 sna: Fix computation of large object sizes to prevent overflow Signed-off-by: Chris Wilson commit 54c1d97d5ab325874e1c7b2639e58111d7a6b93f Author: Chris Wilson Date: Wed Jan 16 09:00:04 2013 +0000 sna: Add DBG for when we add the inplace hint Signed-off-by: Chris Wilson commit 588c5aa6bca441d7c9305fe2fcf268e89b6b617d Author: Chris Wilson Date: Tue Jan 15 22:21:56 2013 +0000 sna: Revert use of a separate CAN_CREATE_SMALL flag Signed-off-by: Chris Wilson commit af85ffdec7047efa452d6bab3a0ee3889dd4f046 Author: Chris Wilson Date: Tue Jan 15 20:37:11 2013 +0000 sna: Avoid serialising on an move-to-cpu for an async operation Signed-off-by: Chris Wilson commit d70be85dc723168a481c1955444afd951c4817bf Author: Chris Wilson Date: Tue Jan 15 20:16:45 2013 +0000 sna: Assert that we never try to mix INPLACE / ASYNC hints for move-to-cpu Signed-off-by: Chris Wilson commit 1287c3a24c277cb42930d8af2943b9f7b016f31d Author: Chris Wilson Date: Tue Jan 15 18:59:15 2013 +0000 sna: Specialise sna_get_image_blt for clears to avoid sync readback Signed-off-by: Chris Wilson commit da4972eec57e662b98a7abced6338ceb8a533a48 Author: Chris Wilson Date: Tue Jan 15 18:34:07 2013 +0000 sna/trapezoids: Avoid the multiply for an opaque source Signed-off-by: Chris Wilson commit 7f968c8c991cff751459939bdb42e14255f529b7 Author: Chris Wilson Date: Tue Jan 15 18:41:00 2013 +0000 sna: Add DBG to use_shm_bo() Signed-off-by: Chris Wilson commit af63fab5047a43716c5df875ddc50f7c877f8a83 Author: Chris Wilson Date: Tue Jan 15 18:21:11 2013 +0000 sna: Hint that a copy from a SHM bo will likely be the last in a batch Signed-off-by: Chris Wilson commit 1be436409222c00ff66c6d747487b77f1037b27a Author: Chris Wilson Date: Tue Jan 15 18:20:29 2013 +0000 sna: Pass the async hint for the upload into the GPU Signed-off-by: Chris Wilson commit 2113f7f440dd2f10e80f0bb3bd5cd155f7e19098 Author: Chris Wilson Date: Tue Jan 15 09:33:03 2013 +0000 sna: Free the SHM pixmaps after b266ae6f6f Since b266ae6f6f protected the static allocations from being reaped in the normal course of events, we need to penetrate those defenses in order to finally free the SHM mappings. Signed-off-by: Chris Wilson commit 441c481630a5cf09a7eb26d5db80b1e60cb2b10f Author: Chris Wilson Date: Tue Jan 15 01:26:19 2013 +0000 sna: Mark uploads with async hints when appropriate Signed-off-by: Chris Wilson commit 6abd442279fd32d1ce9b33a72eabbeb922316151 Author: Chris Wilson Date: Tue Jan 15 00:15:23 2013 +0000 sna: Avoid allocating an active CPU bo unnecessarily If we will not write back the GPU damage to the bo as we intend to overwrite it for the next operation, we can forgo allocating the active CPU bo and skip the synchronisation overhead. Signed-off-by: Chris Wilson commit f235c74cd661970c76e152777e9a2c314a368a56 Author: Chris Wilson Date: Mon Jan 14 15:49:42 2013 +0000 sna: Tweak considering of last-cpu placement for inplace regions Signed-off-by: Chris Wilson commit 70c5e41b519e44e620948d683d3b1111494d2f48 Author: Chris Wilson Date: Mon Jan 14 15:03:59 2013 +0000 sna: Limit temporary userptr uploads to large busy targets or LLC machines Signed-off-by: Chris Wilson commit cf860da1c78244036c59edf934b312cc1367e8aa Author: Chris Wilson Date: Mon Jan 14 12:50:54 2013 +0000 sna: Apply PutImage optimisations to move-to-cpu We can replace the custom heuristics for PutImage by applying them to the common path, where hopefully they are equally valid. Signed-off-by: Chris Wilson commit e4ad4477815abe31b1a2323673da86a6def2f246 Author: Chris Wilson Date: Mon Jan 14 13:12:46 2013 +0000 sna: Use userptr to accelerate GetImage Signed-off-by: Chris Wilson commit 3cc04a8e24f02248b6382c9bc354ea15c42b17b6 Author: Chris Wilson Date: Sun Jan 13 17:34:03 2013 +0000 sna: Initialize src_bo to detect allocation failure sna_accel.c: In function 'sna_put_image': sna_accel.c:3730:18: warning: 'src_bo' may be used uninitialized in this function [-Wmaybe-uninitialized] Signed-off-by: Chris Wilson commit 3f04b0b98d7f861ff58b82c99d33b7eacfcda5f7 Author: Chris Wilson Date: Sun Jan 13 17:31:15 2013 +0000 sna: Check size against aperture before attempting to perform the GTT mapping Signed-off-by: Chris Wilson commit 7a7db06c62228acc6d1c03e800c7afa84e886f5a Author: Chris Wilson Date: Sun Jan 13 13:45:18 2013 +0000 sna: Add a compile flag for measuring impact of userptr uploads Signed-off-by: Chris Wilson commit bcc212dc7a939505a678f97f6700eee99204249f Author: Chris Wilson Date: Sun Jan 13 13:36:09 2013 +0000 sna: Use the pixmap size (not drawable) to determine replacement Signed-off-by: Chris Wilson commit 32f43f618d9b11ea44b3e01a95ac3f239a731ad2 Author: Chris Wilson Date: Sun Jan 13 13:23:24 2013 +0000 sna: Allow large image uploads to utilize temporary mappings Signed-off-by: Chris Wilson commit bf2b2e2f91208412c8b74a95859def501514be43 Author: Chris Wilson Date: Sun Jan 13 12:24:44 2013 +0000 sna: Allow creation of a CPU map for pixmaps if needed Signed-off-by: Chris Wilson commit b266ae6f6f8fb4c494ece532ae4621055e66beb2 Author: Chris Wilson Date: Sun Jan 13 11:30:07 2013 +0000 sna: Relax limitation on not mapping GPU bo with shadow pointers Signed-off-by: Chris Wilson commit a2d82161436e489f23637d793c737bc6950a62b8 Author: Chris Wilson Date: Sun Jan 13 10:17:33 2013 +0000 sna: Correct a few assertions after enabling read-only mappings As these do not flush the active state if we have read-read mappings, we need to be careful with our asserts concerning the busy flag. Signed-off-by: Chris Wilson commit ab01fd696e1137ddfb9a85ae68c15c05900f0e8e Author: Chris Wilson Date: Sat Jan 12 09:17:03 2013 +0000 sna: Experiment with a CPU mapping for certain fallbacks Signed-off-by: Chris Wilson commit 03d392cd1d87e17129c42e4d822d3d1749edb02e Author: Chris Wilson Date: Sat Jan 12 08:51:52 2013 +0000 sna: Tweak max object sizes to take account of aperture restrictions Signed-off-by: Chris Wilson commit d111c464bfbae57bb7141872810c88b88f30c087 Author: Chris Wilson Date: Sat Jan 12 08:15:13 2013 +0000 sna: After a size check, double check the batch before flushing As we may fail the size check with an empty batch and a pair of large bo, we need to check before submitting that batch in order to not run afoul of our internal sanity checks. Signed-off-by: Chris Wilson commit ec77a07b41f1062b941774f3782b51d21e7824dd Author: Chris Wilson Date: Fri Jan 11 11:40:57 2013 +0000 sna/dri: Prefer to preserve the ring of the destination bo Signed-off-by: Chris Wilson commit 42f1026e11527cb62b4522b44e71a4e72582a876 Author: Chris Wilson Date: Fri Jan 11 11:40:16 2013 +0000 sna: Reorder struct kgem_bo to move related data into the same cacheline Signed-off-by: Chris Wilson commit aead71051ed757e7565d395c858bf8ab8f0b0ff6 Author: Chris Wilson Date: Fri Jan 11 01:30:43 2013 +0000 sna: Disable memcpy_to_tiled_x() uploads on 32-bit systems It's far too slow due to the register starved instruction set producing attrocious code and the extra overhead in the kernel for managing memory mappings. Signed-off-by: Chris Wilson commit 220970b1a484e283e2bbb44f79df613ce1ee1146 Author: Chris Wilson Date: Thu Jan 10 19:43:05 2013 +0000 sna: Also prefer to use the GPU for uploads into a tiled bo Signed-off-by: Chris Wilson commit 672e59851c427c63f43cde7dfd1688a72100e3b3 Author: Chris Wilson Date: Thu Jan 10 19:35:29 2013 +0000 sna: Prefer userptr if copying to a tiled bo Signed-off-by: Chris Wilson commit 441ef916ae6569c88b3d6abaf7fea4d69be49d76 Author: Chris Wilson Date: Thu Jan 10 19:14:21 2013 +0000 intel: Throttle harder Filling the rings is a very unpleasant user experience, so cap the number of batches we allow to be inflight at any one time. Interestingly, as also found with SNA, throttling can improve performance by reducing RSS. However, typically throughput is improved (at the expense of latency) by oversubscribing work to the GPU and a 10-20% slowdown is commonplace for cairo-traces. Notably, x11perf is less affected and in particular application level benchmarks show no change. Note that this exposes another bug in libdrm-intel 2.4.40 on gen2/3. Signed-off-by: Chris Wilson commit a37d56f338c5fae832d5eeea1283b6dbde827678 Author: Chris Wilson Date: Thu Jan 10 16:28:24 2013 +0000 sna: Use some surplus bits to back our temporary pixman_image_t Signed-off-by: Chris Wilson commit 09ea1f4402b3bd0e411b90eb5575b3ff066d7356 Author: Chris Wilson Date: Thu Jan 10 16:26:24 2013 +0000 sna: Prefer to use the GPU for copies from SHM onto tiled destinations Signed-off-by: Chris Wilson commit c63147a3c33fd26f5c04a8648881659b4a90df06 Author: Chris Wilson Date: Thu Jan 10 15:15:15 2013 +0000 sna: Allow CPU bo to copy to GPU bo if the device is idle. Signed-off-by: Chris Wilson commit 2933e7595838c28081810d4959ca1e005a0419e1 Author: Chris Wilson Date: Thu Jan 10 13:07:19 2013 +0000 sna: Ignore the last pixmap cpu setting if overwritting all damage Signed-off-by: Chris Wilson commit 934ea64f7ff080b00d00c50ba94f63247d7bb130 Author: Chris Wilson Date: Thu Jan 10 13:06:06 2013 +0000 sna: With a GPU bo and a shm source, do not fall all the way back The normal source upload into GPU bo knows a few more tricks that we may want to apply first before copying into the shadow of the GPU bo. Signed-off-by: Chris Wilson commit 8a8edfe4076ee08558c76eddbb68426e4563888c Author: Chris Wilson Date: Thu Jan 10 03:31:37 2013 +0000 sna: Make sure all outputs are disabled if no CompatOutput is defined If we have to fallback and the configuration is wonky, make sure that all known outputs are disabled as we takeover the console. Signed-off-by: Chris Wilson commit 5449e16c0c2b6ca5af4acf42703164b9d2b2d822 Author: Chris Wilson Date: Thu Jan 10 02:54:41 2013 +0000 sna: Open-code xf86CompatOutput() to avoid invalid pointers config->compat_output needs to be sanitized during device initialization or we may dereference an invalid xf86OutputPtr. Signed-off-by: Chris Wilson commit 8881a14200580db731ca6902b289b08989aaa61e Author: Mickaël THOMAS Date: Mon Jan 7 20:47:51 2013 +0100 Set initial value for backlight_active_level If the "Backlight" option is set, backlight_active_level is not set which results in a default value of 0, causing a black screen upon starting Xorg. commit b8c9598294eaa16e0d1578ad98896f6ec5ba37cf Author: Chris Wilson Date: Mon Jan 7 13:57:21 2013 +1000 sna: fixup damage posting to be done correctly around slave pixmap Copied from commit c789d06cf8a0debc67058d7be1483f5b542e2baa Author: Dave Airlie Date: Mon Jan 7 13:57:21 2013 +1000 This fixes the damage posting to happen in the correct ordering, not sure if this fixes anything, but it should make things more consistent. commit c789d06cf8a0debc67058d7be1483f5b542e2baa Author: Dave Airlie Date: Mon Jan 7 13:57:21 2013 +1000 intel: fixup damage posting to be done correctly around slave pixmap This fixes the damage posting to happen in the correct ordering, not sure if this fixes anything, but it should make things more consistent. Signed-off-by: Dave Airlie commit 5891c89ff2be277d1a833d4bc092b65184c1f3d6 Author: Dave Airlie Date: Mon Jan 7 13:54:47 2013 +1000 intel: drop pointless error printf in the slave pixmap sync code. This is left over and spams logs, get rid. Signed-off-by: Dave Airlie commit 27550e81482229007fa9e0e9769fdd20f3616b23 Author: Chris Wilson Date: Sun Jan 6 17:29:19 2013 +0000 sna/dri: Transfer the DRI2 reference to the new TearFree pixmap Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=58814 Signed-off-by: Chris Wilson commit 1a5e4fb725da2eb25cf7f476290c02e9880a4efc Author: Chris Wilson Date: Sun Jan 6 17:08:56 2013 +0000 sna: Only disable upon a failed pageflip after at least one pipe flips If we have yet to update a pipe for a pageflip, then the state remains consistent and we can fallback to a blit without disabling any pipes. If we fail after flipping a pipe, then unless we disable an output the state becomes inconsistent (the pipes disagree on what the attached fb is). Signed-off-by: Chris Wilson commit dd66ba8e5666a1ce7da0ddc226d074f591e1fa22 Author: Chris Wilson Date: Sun Jan 6 16:13:56 2013 +0000 sna: Try to create userptr with the unsync'ed flag set first Signed-off-by: Chris Wilson commit 9051f43fa3c8d011921ac6ff75b763280f26d98f Author: Chris Wilson Date: Sun Jan 6 15:20:14 2013 +0000 sna/gen4+: Handle solids passed to the general texcoord emitter The general texcoord emitter does handle solids (for the case of a transformed mask) and so we need to be careful to setup the VERTEX_ELEMENTS accordingly. Fixes regression from commit 2559cfcc4cbc1d0d84b048565cad3bfee61df8da Author: Chris Wilson Date: Wed Jan 2 10:22:14 2013 +0000 sna/gen4+: Specialise linear vertex emissio Signed-off-by: Chris Wilson commit 4af910e8be92e0ca241ce1e93e322c712dcbe340 Author: Chris Wilson Date: Sun Jan 6 13:43:55 2013 +0000 sna/gen4+: Trim the redundant float from the fill vertices Signed-off-by: Chris Wilson commit 3244e4b23397f54ca76876dd76ebea9a0abd357e Author: Chris Wilson Date: Sun Jan 6 13:24:23 2013 +0000 Revert "sna/gen4+: Backport tight vertex packing for simple renderblits" This reverts commit 8ff76fad1fadc5e309f9a12c30f883460a432049 and commit 48e4dc4bd4b2980f0f804f572d0e3fc1bb4bc21e. I forgot gen4 and gen5 do not have the 'non-normalized' bit in their sampler states. commit d3be77f87916e38af717bafaf2000becd5180d76 Author: Chris Wilson Date: Sat Jan 5 18:07:50 2013 +0000 sna/trapezoids: filter out cancelling edges upon insertion to edge-list Signed-off-by: Chris Wilson commit 2b4a2f52c47a24c297312d51f9a8299c9a54a697 Author: Chris Wilson Date: Sat Jan 5 17:21:34 2013 +0000 sna/trapezoids: filter out zero-length runs Signed-off-by: Chris Wilson commit 59a7b8b32c694735942fd7e42c1382d91004b0b1 Author: Chris Wilson Date: Fri Jan 4 18:22:14 2013 +0000 sna: Clear up the caches after handling a request allocation failure Signed-off-by: Chris Wilson commit 3c31a9fc210221ba8e7922bec80c15ec39cab7bc Author: Chris Wilson Date: Fri Jan 4 18:11:12 2013 +0000 sna: Embed the pre-allocation of the static request into the device So that in the cache where we are driving multiple independent screens each having their own device, we do not share the global reserved request in the event of an allocation failure. Signed-off-by: Chris Wilson commit b5b3cfb0ad1cc5e66c99035f526946bf41011e13 Author: Chris Wilson Date: Thu Jan 3 23:33:44 2013 +0000 sna: Flush the batch prior to referencing work from another ring In the case where the kernel is inserting semaphores to serialise work between rings, we want to only delay the surface that is coming from the other ring and not interfere with work already queued. Signed-off-by: Chris Wilson commit ea2da97773d858001f98adc880f24b9671c51b2f Author: Chris Wilson Date: Thu Jan 3 16:47:14 2013 +0000 sna: Convert allocation request from bytes to num_pages when shrinking Signed-off-by: Chris Wilson commit 2bd6e4dcd43bb0d836f12232050e73ce1510bb0f Author: Chris Wilson Date: Thu Jan 3 16:38:33 2013 +0000 sna: Add a pair of asserts to validate fls()/cache_bucket() Signed-off-by: Chris Wilson commit f9d2730974a869f15eac599ca865b50a9a9658d9 Author: Chris Wilson Date: Thu Jan 3 15:20:45 2013 +0000 sna: Also recognise __i386__ for fls asm Signed-off-by: Chris Wilson commit 69dde74a003ba0168ceca1558a4cb69097421b92 Author: Chris Wilson Date: Thu Jan 3 15:20:23 2013 +0000 sna: Fix off-by-one in C version of fls Signed-off-by: Chris Wilson commit fc702cdf534a4694a64408428e8933497a7fc06e Author: Matt Turner Date: Wed Jan 2 16:07:54 2013 +0000 sna: Rewrite __fls without dependence upon x86 assembly The asm() prevents SNA from compiling on ia64. Fixes https://bugs.gentoo.org/show_bug.cgi?id=448570 commit bc67bdcec832f4302951f2789456666dee2f496c Author: Chris Wilson Date: Wed Jan 2 13:47:51 2013 +0000 sna/gen6+: Fine tune placement of DRI copies Avoid offsetting the overhead of the render copy only to be penalised by the overhead of the semaphore. So compromise. Signed-off-by: Chris Wilson commit 2559cfcc4cbc1d0d84b048565cad3bfee61df8da Author: Chris Wilson Date: Wed Jan 2 10:22:14 2013 +0000 sna/gen4+: Specialise linear vertex emission Signed-off-by: Chris Wilson commit 0996ed85fd8bd79f41f28908733b85566f9e2b69 Author: Chris Wilson Date: Tue Jan 1 22:53:26 2013 +0000 sna/gen2+: Precompute the affine transformation scale factors Signed-off-by: Chris Wilson commit d36cae801f1dcb06d4f93f2f27cc9b9de73e89c9 Author: Chris Wilson Date: Tue Jan 1 21:03:06 2013 +0000 sna/gen4+: Tidy special handling of 2s2s vertex elements Signed-off-by: Chris Wilson commit 8582c6f0bbe1bf01324b46933ff2f50c65f2a82d Author: Chris Wilson Date: Tue Jan 1 20:53:12 2013 +0000 sna/gen6+: Remove vestigial CC viewport state Signed-off-by: Chris Wilson commit 24264af2912f9abae5aff2a6fb5a50383d9e33be Author: Chris Wilson Date: Tue Jan 1 20:39:23 2013 +0000 sna: Fast path inplace addition of solid trapezoids Signed-off-by: Chris Wilson commit e9a9f9b02978cb2d73c38163827eb7141ebed16c Author: Chris Wilson Date: Tue Jan 1 16:40:28 2013 +0000 sna: Micro-optimise glyph_valid() Note that this requires fixing up the glyph->info if the xserver didn't create a GlyphPicture. Signed-off-by: Chris Wilson commit 372c14aae8f4fd2c5865b9d23cd825dcbc33765f Author: Chris Wilson Date: Tue Jan 1 15:49:12 2013 +0000 sna: Remove some obsolete Options Throttling and delayed-flush are now redundant. Signed-off-by: Chris Wilson commit 65924da91da4bb617df1bb0a7c3e9d4aa475b6b1 Author: Chris Wilson Date: Tue Jan 1 11:40:15 2013 +0000 sna: Tidy compat interfaces Signed-off-by: Chris Wilson commit 0a35d9287397031c95ebd9dc53b68e33e7dcf092 Author: Chris Wilson Date: Tue Jan 1 11:12:02 2013 +0000 sna/gen2: Always try to use the BLT pipeline first Signed-off-by: Chris Wilson commit c1457fbd8a169ee19c8e625ea4e779180eb4b070 Author: Chris Wilson Date: Tue Jan 1 10:49:27 2013 +0000 sna/gen2: Tidy a pair of vertex emitters Switch to the new inline scaled transforms. Signed-off-by: Chris Wilson commit 48a5797c0f227204d0723de0ef34b046964c571e Author: Chris Wilson Date: Mon Dec 31 17:30:40 2012 +0000 sna/gen4: Tweak single-thread SF w/a for solids Allow multiple threads for the rare case of compositing with a solid color. Signed-off-by: Chris Wilson commit e4f6ba6b47c41645a40e314f14047ba0b5f93a01 Author: Chris Wilson Date: Mon Dec 31 14:06:36 2012 +0000 sna/gen6+: Hint that we prefer to use the BLT with uncached scanouts Once again balancing the trade-off of faster smaller copies with the BLT versus the faster larger copies the RENDER ring. Signed-off-by: Chris Wilson commit 6e87e7ddfe0c21e0fb6b3c2cb940a40aa7d4e061 Author: Chris Wilson Date: Mon Dec 31 14:03:16 2012 +0000 sna/dri: Use the default choice of backend for copying the region Notably, if everything is idle, using the BLT is a win as we can emit them so much faster than a rendercopy, and as the target is uncached we do not benefit as much from the rendercache. Signed-off-by: Chris Wilson commit a7988bf77f5a106a48b6e39b6eaf60ef2f8bec11 Author: Chris Wilson Date: Sun Dec 30 14:50:49 2012 +0000 sna/dri: Fix triple buffering to not penalise missed frames Signed-off-by: Chris Wilson commit 736b89504a32239a0c7dfb5961c1b8292dd744bd Author: Chris Wilson Date: Sun Dec 30 10:32:18 2012 +0000 uxa: Align surface allocations to even tile rows Align surface sizes to an even number of tile rows to cater for sampler prefetch. If we read beyond the last page we may catch the PTE in a state of flux and trigger a GPU hang. Also detected by enabling invalid PTE access checking. References: https://bugs.freedesktop.org/show_bug.cgi?id=56916 References: https://bugs.freedesktop.org/show_bug.cgi?id=55984 Signed-off-by: Chris Wilson commit 43336c632beb5d599ec0fc614434b88ef7a26422 Author: Chris Wilson Date: Sat Dec 29 16:47:53 2012 +0000 sna: Seed the solid color cache with an invalid value to prevent false hits After flushing, we *do* need to make sure we cannot hit a false lookup via the last cache. Signed-off-by: Chris Wilson commit f6050382095c3bc4f78bc4ff9e9c6086e58d6b28 Author: Chris Wilson Date: Sat Dec 29 16:41:03 2012 +0000 sna/dri: Gracefully handle failures from pageflip Signed-off-by: Chris Wilson commit 1c2ece369177ea6c3fd2f254b2554ceadf5590de Author: Chris Wilson Date: Sat Dec 29 15:53:23 2012 +0000 sna/gen4+: Try using the BLT before doing a tiled copy Signed-off-by: Chris Wilson commit 09ca8feb3455c979e799ddf26daae8f2de2813e1 Author: Chris Wilson Date: Sat Dec 29 15:42:02 2012 +0000 sna: Move the primary color cache into the alpha cache Signed-off-by: Chris Wilson commit 8c56c9b1da9e078bd5b7ff4ebc5d8b23f593d500 Author: Chris Wilson Date: Sat Dec 29 14:14:41 2012 +0000 sna: Allow a flush to occur before batching a flush-bo Signed-off-by: Chris Wilson commit 2f53fb389c001f68134f514e30e25e91de41fb9d Author: Chris Wilson Date: Fri Dec 28 22:58:02 2012 +0000 sna: DBG compile fixes Signed-off-by: Chris Wilson commit dba83dacd2ccbb2ac23b205ce2a872a889fa30bd Author: Chris Wilson Date: Fri Dec 28 19:23:36 2012 +0000 sna/gen3: Use inline transform+scale function So as to avoid reading back from the vbo (which may be wc mapped). Signed-off-by: Chris Wilson commit f0fca544b0602bc4ed2f68e8d260e0a3745b4bad Author: Chris Wilson Date: Fri Dec 28 18:52:44 2012 +0000 sna/gen4+: Check for a spare exec slot for an outstanding vbo Signed-off-by: Chris Wilson commit c6e850b626f4bb44876c683d596ea38f8f6c30ae Author: Chris Wilson Date: Fri Dec 28 17:14:52 2012 +0000 sna/gen4+: Trim an extraneous coordinate from solid composite emission Signed-off-by: Chris Wilson commit 3fdc9923447538ed65bf9ffa189d7290ce804730 Author: Chris Wilson Date: Fri Dec 28 17:14:52 2012 +0000 sna/gen4+: Trim an extraneous coordinate from solid span emission Signed-off-by: Chris Wilson commit fdd6d222bc92b3e385f5d62f5e03dfd86f290e45 Author: Chris Wilson Date: Fri Dec 28 17:08:00 2012 +0000 sna/gen4+: Tidy emit_spans_affine() gcc produced abysmal code for the inlined emission, so hand unroll it for sanity. Signed-off-by: Chris Wilson commit 5d222d4d21e6e3af5316728e0da49a014e9fea21 Author: Chris Wilson Date: Fri Dec 28 17:08:00 2012 +0000 sna/gen4+: Tidy emit_spans_solid() gcc produced abysmal code for the inlined emission, so hand unroll it for sanity. Signed-off-by: Chris Wilson commit 4528f68eff33a5c2f9c1d884e9b3f7228053e0f4 Author: Chris Wilson Date: Fri Dec 28 16:45:50 2012 +0000 sna: Only allocate a busy CPU bo for a GPU readback Signed-off-by: Chris Wilson commit 99fdd1a1c6aa52688c2c821a90f86700b7ee34b2 Author: Chris Wilson Date: Fri Dec 28 16:33:59 2012 +0000 sna: Mark kgem_bo_retire() as static The exported function is not used, so mark it static and strengthen the assertions. Signed-off-by: Chris Wilson commit 548d284b8cf8cc2b311efe3287e0ae956738189a Author: Chris Wilson Date: Fri Dec 28 14:49:38 2012 +0000 sna: Skip copying fbcon if we are already on the scanout If we are already the scanout, then there is little point copying to ourselves... Should be paranoia. Signed-off-by: Chris Wilson commit 583efd4ba067a0a4319e43ebc18dd81ed9c8db0a Author: Chris Wilson Date: Thu Dec 27 17:59:59 2012 +0000 sna: Sanity check config->compat_output In a headless setup this may be left initialised to -1. Signed-off-by: Chris Wilson commit 7725df8aa1b3eab97618311e3f24769a318bd804 Author: Chris Wilson Date: Thu Dec 27 14:01:59 2012 +0000 sna/gen2,3: Remove gen-specific vertex_offset Remove the duplication of vertex_offset in favour of the common vertex_offset. Signed-off-by: Chris Wilson commit 46af1ff126f3fb1f9470b0cbb19c7c2b09d5b92a Author: Chris Wilson Date: Thu Dec 27 00:40:08 2012 +0000 sna/gen6+: Tidy up ring preferences Remove a few duplicated tests. Signed-off-by: Chris Wilson commit dd5b653aa2c5fe2e062533db35c83a40c1952ea6 Author: Chris Wilson Date: Thu Dec 27 09:54:35 2012 +0000 sna: Do not try and set a 0x0 mode Signed-off-by: Chris Wilson commit 861c2362dd38d7d43fe7ffb181cb197199a1c570 Author: Chris Wilson Date: Wed Dec 26 14:12:42 2012 +0000 sna/gen6+: Tweak to only consider active ring on destination Otherwise we decide to use BLT when hitting the render/sampler cache is preferrable for a source bo. Signed-off-by: Chris Wilson commit f9b6aa3aaf784f9149e091a646673ddf341cd7ca Author: Chris Wilson Date: Wed Dec 26 13:05:52 2012 +0000 sna: Explicitly track self-relocation entries Avoid having to walk the full relocation array for the few entries that need to be updated for the batch buffer offset. Signed-off-by: Chris Wilson commit 90b1b220ee7a3c543301956b01c54a4a04632db4 Author: Chris Wilson Date: Wed Dec 26 12:51:58 2012 +0000 2.20.17 release Signed-off-by: Chris Wilson commit 52fd223fc970118cbdcb31f9574414debc905e9c Author: Chris Wilson Date: Fri Dec 21 21:36:30 2012 +0000 sna/video: Initialise alignment for video ports > 0 We repeatedly set the alignment value on the first port, rather than once for each. Reported-by: Jiri Slaby Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 3793ccf7804cfc870b46c623dfeefbe0c381c1d4 Author: Chris Wilson Date: Fri Dec 21 14:48:07 2012 +0000 sna: Remove assertions that the pixmap is wholly defined when uploading As the user may only write to a portion of a pixmap (thus only creating a small amount of damage) and then attempt to use the whole as a source, we run the risk of triggering an assertion that the whole was defined. Signed-off-by: Chris Wilson commit 07dde33a4f51941b4f612823ea6ea7ca01a6efbc Author: Chris Wilson Date: Fri Dec 21 14:35:32 2012 +0000 sna: Remove a pair of stale assertions For gen2-5, it does not matter what mode the batch is in when we insert the scanline wait. With the more aggressive batch flushing, and relaxed assigned of mode for those generations, we are likely to see that the batch is idle when we go to insert the waits. Signed-off-by: Chris Wilson commit bdd0cca4e1192df0038621925c4e6243ba419a81 Author: Chris Wilson Date: Fri Dec 21 14:20:23 2012 +0000 sna: Refactor test for a rotation matrix Signed-off-by: Chris Wilson commit 347c5a7b33729f1bedd408d2ef24756d51b66f1d Author: Chris Wilson Date: Fri Dec 21 10:40:47 2012 +0000 sna/dri: Refactor get_current_msc between blit/flip paths Signed-off-by: Chris Wilson commit 8a67d3f808fcc7c8c51553b1703e8312f28b87a1 Author: Chris Wilson Date: Fri Dec 21 10:21:06 2012 +0000 sna/dri: Set the correct current_msc for the no readback path If we are asked to render immediately, then in order to pass the tests when comparing it to target, we need to set the current_msc to the ultimate future value, -1. Signed-off-by: Chris Wilson commit 48e4dc4bd4b2980f0f804f572d0e3fc1bb4bc21e Author: Chris Wilson Date: Thu Dec 20 21:54:25 2012 +0000 sna/gen4: Backport tight vertex packing of renderblits Signed-off-by: Chris Wilson commit 08d2b073692836aa22f65f8ba30db5d14550c03e Author: Chris Wilson Date: Thu Dec 20 21:30:32 2012 +0000 sna/gen4: Backport more recent state tracking tweaks Signed-off-by: Chris Wilson commit 8ff76fad1fadc5e309f9a12c30f883460a432049 Author: Chris Wilson Date: Thu Dec 20 20:57:40 2012 +0000 sna/gen5: Backport tight vertex packing for simple renderblits Signed-off-by: Chris Wilson commit 9144c951915a1e0c1899a72161f9f0f1ab9b9ac4 Author: Chris Wilson Date: Fri Dec 21 09:44:52 2012 +0000 sna/dri: Avoid querying the current-msc with swapbbufers wait disabled Signed-off-by: Chris Wilson commit 84c327e17f68c4a56fcb76be1f45ab6d35291b5d Author: Chris Wilson Date: Thu Dec 20 19:44:46 2012 +0000 sna/video: Assert that the frame is initialised References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 4d750219925cb3199ebc6751cdbd2862dfb4cdfe Author: Chris Wilson Date: Thu Dec 20 19:34:41 2012 +0000 uxa/dri: Correct the destination of the blit after chained flip is broken Signed-off-by: Chris Wilson commit ca5c028c2b4d9bf02002acd484054fe427ea8d09 Author: Chris Wilson Date: Thu Dec 20 19:31:44 2012 +0000 glamor: Release the drawable after passing to glamor_push_pixels An unlikely path, but a double prepare instead of a prepare/finish. Signed-off-by: Chris Wilson commit de2de36049e2958a60f63fadffe8f54de8da1e56 Author: Chris Wilson Date: Thu Dec 20 19:29:31 2012 +0000 sna: Check the correct variable for a failed allocation Having already checked 'dst' and just allocated 'src', that is who we should be checking. Signed-off-by: Chris Wilson commit deb908fda74541fba649349db279715b05d0554e Author: Chris Wilson Date: Thu Dec 20 19:22:32 2012 +0000 intel: ODEV_ATTRIB_PATH is no longer printed, so kill the temporary variable Signed-off-by: Chris Wilson commit 0f84ecfc3cd7dfe7f43ff99a6498d2ceccd90225 Author: Chris Wilson Date: Thu Dec 20 12:00:00 2012 +0000 sna/gen4+: Amalgamate all the gen4-7 vertex buffer emission Having reduced all the vb code for these generations to the same set of routines, we can refactor them into a single set of functions. Signed-off-by: Chris Wilson commit 1f4ede0ef8f8a8d07e11781ad05617ecdfcd3faf Author: Chris Wilson Date: Wed Dec 19 20:39:10 2012 +0000 sna: Do not throttle before move-to-cpu The idea being that when creating a surface to perform inplace rasterisation, we won't be using the GPU for a while and so give it time to naturally throttle. Signed-off-by: Chris Wilson commit 5deba2832dc42072d9abaeaa7934bc0e1b28b3ed Author: Chris Wilson Date: Wed Dec 19 20:03:33 2012 +0000 sna: Ignore throttling during vertex close Signed-off-by: Chris Wilson commit f91a24fdba517c8e9df5a074db2c789fbf066bb3 Author: Chris Wilson Date: Thu Dec 20 09:46:32 2012 +0000 sna/video: Remove XvMCScreenInitProc The symbols disappears without warning in xorg-1.14 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=58552 Signed-off-by: Chris Wilson commit ee99511846a0f10abeeba8d25d8fb5bf59621b02 Author: Chris Wilson Date: Wed Dec 19 18:02:50 2012 +0000 sna/gen4+: Tweak preference of GPU placement for spans If the CPU bo is busy, make sure we do not stall for an inplace operation. Signed-off-by: Chris Wilson commit bfd96b092db5e4e0fc2446752deafd1156cf37b3 Author: Chris Wilson Date: Tue Dec 18 20:54:33 2012 +0000 sna/video: Fix presentation of cropped sprites Signed-off-by: Chris Wilson commit 2df1b1abf0728f28d2803a096f945779cbe7c70b Author: Chris Wilson Date: Tue Dec 18 16:07:26 2012 +0000 sna/video: Fix up copying cropped textured video packed data Simply ignore the cropping and copy the whole plane rather than complicate the computation of the packed destination pixels. Signed-off-by: Chris Wilson commit 8d523fa824dcb1987557164d048711c1745de378 Author: Chris Wilson Date: Tue Dec 18 16:07:26 2012 +0000 sna/video: Fix up destination offset for copying cropped textured video planes Oh fun. Textured video expects the source content to be relative to the origin, whereas overlay video expects the source at the origin. Signed-off-by: Chris Wilson commit 7bb4573fcc2cf1b8b6bff5d885a2fa81200d2fd7 Author: Chris Wilson Date: Tue Dec 18 15:48:21 2012 +0000 sna/video: Fix up the image size for copying Yikes, setting image.x2 == image.x1 meant no data was copied whilst the video was clipped. Signed-off-by: Chris Wilson commit 551b400377ddc5eb1e89b8b5827a42e810c8d23d Author: Chris Wilson Date: Tue Dec 18 15:14:00 2012 +0000 sna/video: Amalgamate the computation of source vs dest offsets Signed-off-by: Chris Wilson commit d96a226cc59c641c10153ae3a086a5138c852423 Author: Chris Wilson Date: Tue Dec 18 14:26:18 2012 +0000 sna/video: Fix adjustment of drawable vs source origin wrt to clip Signed-off-by: Chris Wilson commit 79cb6304e983514dd754065e65e2381a903f9bd6 Author: Chris Wilson Date: Tue Dec 18 13:49:59 2012 +0000 sna/xvmc: Clean up to avoid crash'n'burn Signed-off-by: Chris Wilson commit 0d26082303f3f4006ce4974d402c560613081b23 Author: Chris Wilson Date: Tue Dec 18 10:54:28 2012 +0000 sna: Prefer the GPU once again for PolyPoint Signed-off-by: Chris Wilson commit 0e0a2d300633122d6d0f6f82ff110f513b4e64d7 Author: Chris Wilson Date: Tue Dec 18 10:27:04 2012 +0000 sna/gen7: Mark the ring switch before checking bo As we may do a batch submission due to the change of mode. Signed-off-by: Chris Wilson commit f522fbe7c98ffad86126c3666b2d9f7e616480b8 Author: Chris Wilson Date: Mon Dec 17 23:04:25 2012 +0000 sna: Refine check for an unset context switch So it appears that we end up performing a context switch on an empty batch, but already has a mode. This is caught later, too late, by assertions. However, we can change the guards slightly to prevent those assertions without altering the code too greatly. And I can then think how to detect where we are setting a mode on the batch but doing no work - which is likely masking a bigger bug. Reported-by: Jiri Slaby Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 6c50cf4809816dbbd93d54f589a79b0dab996180 Author: Chris Wilson Date: Mon Dec 17 22:27:14 2012 +0000 sna: Untangle the confusion of caching large LLC bo We only use a single cache for very large buffers, so we need to be careful that we set the tiling on them. More so, we need to take extra care when allocating large CPU bo from that cache to be sure that they are untiled and the flags are true. Signed-off-by: Chris Wilson commit e474abea7cf761e78e777db07b41ec99c0b6f59f Author: Chris Wilson Date: Mon Dec 17 15:38:04 2012 +0000 sna: Promote pinned-batches to run-time detection Now that the feature has been committed upstream, we can rely on the runtime detection. Signed-off-by: Chris Wilson commit 4d7e3a9123cf41d2dd97c0a8a0d461c189064822 Author: Chris Wilson Date: Mon Dec 17 12:34:05 2012 +0000 uxa: Fix copy'n'paste of false not FALSE Bugzilla; https://bugs.freedesktop.org/show_bug.cgi?id=58406 Signed-off-by: Chris Wilson commit 7a4d1136bd09bfd4d2657c0b1b64d553eeb6ed4f Author: Chris Wilson Date: Mon Dec 17 09:41:47 2012 +0000 sna/video: Pass along the video source offset Fortunately nobody had yet noticed that all videos were assumed to play with a matching src/dst origin. Signed-off-by: Chris Wilson commit dfe9d18f9f97a77ceeb410307010424c789c8bd1 Author: Chris Wilson Date: Mon Dec 17 01:06:57 2012 +0000 sna: Limit the default upload buffer size to half the cpu cache This seems to help with small slow caches. Signed-off-by: Chris Wilson commit 5b0572503eab235bc7eff20d369241330c41e630 Author: Chris Wilson Date: Sun Dec 16 23:04:55 2012 +0000 sna: Enable support for opting out of the kernel CS workaround Keeping a set of pinned batches in userspace is considerably faster as we can avoid the blit overhead. However, combining the two approaches yields even greater performance, as fast as without either w/a, and yet stable. Signed-off-by: Chris Wilson commit 805f78addf3ffb36c736df680806cf722b18fea9 Author: Chris Wilson Date: Sun Dec 16 22:04:54 2012 +0000 sna: Try to reuse pinned batches by inspecting the kernel busy status Signed-off-by: Chris Wilson commit f1aec676810c4a4c180b342d9a83254e08dd55da Author: Chris Wilson Date: Sun Dec 16 17:37:32 2012 +0000 sna: Precompute the base set of batch-flags This is to make it easier to extend in future. Signed-off-by: Chris Wilson commit c7ac12003bd0c7d85fa47d43ee2734b222d84a61 Author: Chris Wilson Date: Sun Dec 16 15:28:24 2012 +0000 sna: Only flush at the low fence wm if idle Signed-off-by: Chris Wilson commit 4580bbeac0051417cb03f272112b0cfe697e31b3 Author: Chris Wilson Date: Sun Dec 16 15:00:21 2012 +0000 intel: Support debugging through AccelMethod Ease debugging by allowing all acceleration or render acceleration to be disabled through AccelMethod: Option "AccelMethod" "off" -> disable all acceleration Option "AccelMethod" "blt" -> disable render acceleration (only use BLT) Signed-off-by: Chris Wilson commit 58770b7d6401d2d81f7fee1c8c0e788d44149712 Author: Chris Wilson Date: Sun Dec 16 14:59:03 2012 +0000 man: Describe Option "AccelMethod" Signed-off-by: Chris Wilson commit 83609af3681fad58af88387077bf7ce0c001a1da Author: Chris Wilson Date: Sun Dec 16 10:53:26 2012 +0000 sna: Tweak the idle SHM CopyArea path to also replace a busy GPU bo Signed-off-by: Chris Wilson commit 6490585f65bde487da7bc41fa5cb1c5a028d0bf4 Author: Chris Wilson Date: Sat Dec 15 23:26:30 2012 +0000 sna: Do not force use of the GPU for a copy from a SHM pixmap As we will undoubtably flush and sync upon the SHM request very shortly afterwards, we only want to use the GPU for the SHM upload iff it is currently busy. Signed-off-by: Chris Wilson commit 3a08f091875f2f0f49697ba9852077094b3a704b Author: Chris Wilson Date: Sat Dec 15 22:53:44 2012 +0000 sna/gen6+: Tweak prefer-blt-bo Split the decision between where it is imperative to use the BLT to avoid TLB misses and the second case where it is merely preferential to witch. Signed-off-by: Chris Wilson commit ac9ef1fc606e87b48baa47be22bf828dcfe6659f Author: Chris Wilson Date: Sat Dec 15 20:49:56 2012 +0000 sna/gen6+: Keep the bo on its current ring Track the most recent ring each bo is executed on, and prefer to keep it on that ring for the next operation. Signed-off-by: Chris Wilson commit 15ccb7148d15d776a661c1b8c5b9b2360fcae4ad Author: Chris Wilson Date: Sat Dec 15 20:07:56 2012 +0000 sna/gen6+: Apply the is_scanout to the key not value in the binding cache Oops, we never managed to reuse the cached location of the target surface as we entered it into the cache with the wrong key. Signed-off-by: Chris Wilson commit fde25b08922d97ca0d4a69c654bf690edbd53b3d Author: Chris Wilson Date: Sat Dec 15 18:59:53 2012 +0000 sna/trapezoids: Add another inline hint cell_list_alloc() is only called from one place, and the compiler should already be inlining it - but does not appear to be. Hint harder. Signed-off-by: Chris Wilson commit 2a21c8b351052be9c32c5669264fb05a8510c957 Author: Chris Wilson Date: Sat Dec 15 17:56:27 2012 +0000 sna: Include shm hint in render placement The goal is to reduce the preference of rendering to a SHM pixmap - only if it is already active, will we consider continuing to use it on the GPU. Signed-off-by: Chris Wilson commit a467102a9539c7f4fa8d0700ecdcaba49d77b3f7 Author: Chris Wilson Date: Sat Dec 15 10:00:48 2012 +0000 2.20.16 release Signed-off-by: Chris Wilson commit b0f8c823b6cafdfdd064c09d58174f946e290541 Author: Chris Wilson Date: Sat Dec 15 09:28:04 2012 +0000 sna/dri: Fallback to a blit after a failed flip ...rather than force the exchange. Signed-off-by: Chris Wilson commit 2c71a8e08abce74b269687d3a6c1edd7f9d643d3 Author: Chris Wilson Date: Sat Dec 15 09:27:07 2012 +0000 sna/dri: Honour TripleBuffer Option In case anyone ever wants to disable the default. Signed-off-by: Chris Wilson commit 6593ad3fecb3d044ee5ca161176d8ecaa0b4126a Author: Chris Wilson Date: Fri Dec 14 23:48:00 2012 +0000 sna/dri: Store and check size of front/back bo attached to a DRI2 drawable So that we can prevent feeding back a stale bo when the DRI2 client tries to swap an old buffer. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=57212 Signed-off-by: Chris Wilson commit 9df9585cb00958b42461b3139bb7aec32090a869 Author: Chris Wilson Date: Fri Dec 14 15:37:56 2012 +0000 sna: Reduce fence watermarks Further restrict the amount of fenced bo we try to fit into the batch to make it easier for the kernel to accommodate the request. Signed-off-by: Chris Wilson commit 0d3ba44e448c152a570cc469d289ab057fa7be5c Author: Chris Wilson Date: Fri Dec 14 12:47:46 2012 +0000 sna/gen2+: Experiment with not forcing migration to GPU after CPU rasterisation Signed-off-by: Chris Wilson commit d867fd01cb0060342102a79600daf43e3dc44a07 Author: Chris Wilson Date: Fri Dec 14 13:10:54 2012 +0000 sna/gen3: Don't combine primitives if beginning a ca 2-pass Signed-off-by: Chris Wilson commit 3ca503dac2ea6c036e7ebe878b41923541daf225 Author: Chris Wilson Date: Fri Dec 14 12:49:14 2012 +0000 sna/gen3: Remove stray setting of vertex_start It is always done at the beginning of vertex emission. Signed-off-by: Chris Wilson commit 7f76f100e8033497620ee46548df45afff41064a Author: Chris Wilson Date: Thu Dec 13 23:12:42 2012 +0000 sna/gen2: Reorder reuse_source() to avoid NULL dereference for solids Signed-off-by: Chris Wilson commit 044a54c23384756a5dc1895473abf34f7abb3d83 Author: Chris Wilson Date: Thu Dec 13 23:05:30 2012 +0000 sna/gen2: Initialise channel->is_affine for solid In case we hit a path were we avoid reusing the source for the mask and leave is_affine unset for a solid mask. Signed-off-by: Chris Wilson commit 29afd0dc8e893cc4110ee0d70546775dae86ddb3 Author: Chris Wilson Date: Thu Dec 13 22:53:18 2012 +0000 sna/gen2: Assertions Signed-off-by: Chris Wilson commit 4d2abe1e3daac74747535d88dff34b024b87bbe9 Author: Chris Wilson Date: Thu Dec 13 22:09:37 2012 +0000 sna/gen3: Remove incorrect optimisation of an opaque source for CA Signed-off-by: Chris Wilson commit d428dbf7ad7c246acb1c301b73b9df4a1309de03 Author: Chris Wilson Date: Thu Dec 13 14:53:50 2012 +0000 sna/gen2: Program solid mask using the DIFFUSE component Signed-off-by: Chris Wilson commit 9e7311516da81ab45484b291ec668503c5ded0bb Author: Chris Wilson Date: Thu Dec 13 14:40:25 2012 +0000 sna/gen2: Align surface sizes to an even tile Makes this 855gm much happier. Signed-off-by: Chris Wilson commit e646047a563598948206167765eaaf4192cfd77f Author: Chris Wilson Date: Thu Dec 13 14:23:54 2012 +0000 sna: Fix up BLT overwrite detection to use target_handle Signed-off-by: Chris Wilson commit 4f96439e39a4bf4b127af9ccfdc09d061caff9bd Author: Chris Wilson Date: Thu Dec 13 13:15:52 2012 +0000 sna: Fix typo for 830/845 gen Must remember, its octal not decimal. Signed-off-by: Chris Wilson commit f631a56bcb3ff1ce1942b828325a157cef1e0880 Author: Chris Wilson Date: Thu Dec 13 00:55:45 2012 +0000 sna: Only flush the batch after an actual relocation As we may write preparatory instructions into the batch before checking for a flush. Signed-off-by: Chris Wilson commit 74bbf20e6e652cba55d6d0bc17066f4112f8548c Author: Chris Wilson Date: Wed Dec 12 21:56:22 2012 +0000 sna: Improve the initialisation failure path for pinned batches Simplify the later checks by always populating the lists with a single, albeit unpinned, bo in the case we fail to create pinned batches. Signed-off-by: Chris Wilson commit 52c8c9218c8f28fb049b02214d833912a803f911 Author: Chris Wilson Date: Wed Dec 12 21:19:02 2012 +0000 sna: Fix the error path in kgem_init_pinned_batches() to use the right iter Signed-off-by: Chris Wilson commit c7f7dd61fd07dbf938fc6ba711de07986d35ce1f Author: Chris Wilson Date: Wed Dec 12 19:43:19 2012 +0000 sna: Pin some batches to avoid CS incoherence on 830/845 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=26345 Signed-off-by: Chris Wilson commit b154d0dc404a152e1283a013a78be06b8d734867 Author: Chris Wilson Date: Wed Dec 12 18:34:54 2012 +0000 sna/gen2: STIPPLE requires an argument Signed-off-by: Chris Wilson commit 9001263b32efde1361555432914d9ac3ee780511 Author: Chris Wilson Date: Wed Dec 12 12:03:40 2012 +0000 sna/gen3+: Use nearest for unscaled videos If the output is unscaled, then we do not require pixel interpolation (and planar formats are exactly subsampled). References: https://bugs.freedesktop.org/show_bug.cgi?id=58185 Signed-off-by: Chris Wilson commit 2dbe7d91a7f15a3a9ddad696c5088ca98898fca2 Author: Chris Wilson Date: Wed Dec 12 09:50:34 2012 +0000 sna/gen4: Use the single-threaded SF w/a for spans as well Fixes the flickering seen in the fishtank demo, for example. Signed-off-by: Chris Wilson commit 2502218a9340044bb660be68f059971119077e29 Author: Chris Wilson Date: Tue Dec 11 17:47:09 2012 +0000 sna/dri: Fix handling of current_msc > target_msc Signed-off-by: Chris Wilson commit 39d6c65f437d7877c1647b7ecf45e76daabc76a6 Author: Chris Wilson Date: Tue Dec 11 15:30:54 2012 +0000 sna/dri: Query current msc before use Might be worth caching the last-known-value so we can skip the query for an old swap request. Signed-off-by: Chris Wilson commit 162e9a2bfc693db186aa481551cf76b3dc5ee55c Author: Chris Wilson Date: Mon Dec 10 23:11:48 2012 +0000 sna/dri: Disable name exchanges for SwapBuffers The DRI2 protocol is inherently racy. Fortuituously, this can be swept under the carpet by forcing the serialisation between the DRI2 clients by using a blit for the SwapBuffers. References: https://bugs.freedesktop.org/show_bug.cgi?id=58005 Signed-off-by: Chris Wilson commit 0c2287c735f990a98b39d00f28168d7a5df25aba Author: Chris Wilson Date: Mon Dec 10 20:09:22 2012 +0000 sna/dri: Only special case 'divisor && msc-passed' for immediate flipping As Jesse pointed out, it is legal for the client to request that the flip be some frame in the future even with no divisor. Signed-off-by: Chris Wilson commit 2ab29a1688cd313768d928e87e145570f35b4a70 Author: Jesse Barnes Date: Mon Dec 10 14:55:32 2012 -0800 dri2: don't schedule a flip prematurely at ScheduleSwap time If divisor is 0 but the current MSC is behind the target, we shouldn't schedule a flip (which will occur at the next vblank) or we'll end up displaying it early and returning the wrong timestamp. Preserve the optimization though by allowing us to schedule a flip if both the divisor is 0 and the current MSC is equal to or ahead of the target; this avoids a round trip through the kernel. Reported-by: Mario Kleiner Signed-off-by: Jesse Barnes commit 986e13a56a8544d5b32dbcaacbc0ee9cf5d47e27 Author: Chris Wilson Date: Mon Dec 10 17:29:08 2012 +0000 sna: Try installing a fallback config on VT enter in case full desiredMode fails This can happen naturally for 3-pipe config on Ivybridge or if the outputs are rearranged whilst we slept. Instead of failing to change the display on the VT, install at least a fb on the CompatOutput so that hopefully the DE can take over, or give some control to the user. Signed-off-by: Chris Wilson commit 8c3b82f207bc8cf697646d3324cb4103da3b7856 Author: Chris Wilson Date: Mon Dec 10 16:20:52 2012 +0000 sna: Avoid reusing the same 'busy' bit for two different meanings. Oops, I thought the 'busy' bit was now used and apparently forgot it is used to control the periodic flushing... Signed-off-by: Chris Wilson commit d1b479a3404e6b52a23e0443c36d0682cbaf3c2f Author: Chris Wilson Date: Mon Dec 10 11:16:04 2012 +0000 sna: Compromise and only flush a split batch if writing to scanout A compromise between not flushing quick enough and flushing too often, hopefully. Signed-off-by: Chris Wilson commit 3e9120d73c6f0c0e06b617da91cc2edce4434bc3 Author: Chris Wilson Date: Mon Dec 10 11:05:16 2012 +0000 sna: Immediately flush a split batch If we submit a batch early (for example if the GPU is idle), then submit whatever else the client drew immediately upon completion of its blockhandler. This is required to prevent flashing due to visible delay between the clear at the start of the cycle and then the overdraw later. References: https://bugs.freedesktop.org/show_bug.cgi?id=51718 Signed-off-by: Chris Wilson commit aa8c5d8201006397bb32ed6bc28618f9aa77a68a Author: Chris Wilson Date: Thu Dec 6 22:08:08 2012 +0000 sna/sprite: Add a DBG to report whether the kernel supports sprites Signed-off-by: Chris Wilson commit 25c0d440dee45c03f5e45b8e0e45071c0c32f507 Author: Chris Wilson Date: Sun Dec 9 12:11:53 2012 +0000 sna: Move source to CPU prior to referencing for inplace trapezoids Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=56825 Signed-off-by: Chris Wilson commit 3e82fcc8d243b7ffe1a3d3249a5cdb5fd068093f Author: Chris Wilson Date: Sun Dec 9 11:19:13 2012 +0000 sna/gen4+: Refine test for preferring GPU spans Signed-off-by: Chris Wilson commit c8f622726a4463b419d032b379576cfb3bc492df Author: Chris Wilson Date: Sun Dec 9 09:26:03 2012 +0000 sna: Replace remaining kgem_is_idle() with kgem_ring_is_idle() Further experimentation... Signed-off-by: Chris Wilson commit 4e4e10935d2815fb62aeaedbfffe10aad115ec0b Author: Chris Wilson Date: Sat Dec 8 22:39:32 2012 +0000 sna: Flush upon change of target if GPU is idle The aim is to improve GPU concurrency by keeping it busy. The possible complication is that we incur more overhead due to small batches. Signed-off-by: Chris Wilson commit cef11795f627a393d4254845b0a19eefbf6c782c Author: Chris Wilson Date: Fri Dec 7 17:40:37 2012 +0000 sna: Convert the ring from BLT/3D to the internal index for kgem_ring_is_idle() Signed-off-by: Chris Wilson commit 52405b2aed492dc7f76fbf082122842f621e7c06 Author: Chris Wilson Date: Fri Dec 7 17:24:42 2012 +0000 sna: Only inspect the target ring for busyness Signed-off-by: Chris Wilson commit 4b7bbb2a23b03bac63f864c33f47fab88dedbf67 Author: Chris Wilson Date: Fri Dec 7 16:43:32 2012 +0000 sna: Only flush before adding fresh surfaces to the batch Previously, before every operation we would look to see if the GPU was idle and we were running under a DRI compositor. If the GPU was idle, we would flush the batch in the hope that we reduce the cost of the context switch and copy from the compositor (by completing the work earlier). However, we would complete the work far too earlier and as a result would need to flush the batch before every single operation resulting in extra overhead and reduced performance. For example, the gtkperf circles benchmark under gnome-shell/compiz would be 2x slower on Ivybridge. Reported-by: Michael Larabel Signed-off-by: Chris Wilson commit 65a8c23ca1bc8e2ebd087027a30358704d4bf11c Author: Chris Wilson Date: Fri Dec 7 14:56:18 2012 +0000 sna: Only flush at the low apeture watermark if idle Signed-off-by: Chris Wilson commit 4bfc5e90f54be1b0997dec9e81796d67b376a01f Author: Chris Wilson Date: Thu Nov 8 23:42:10 2012 +0000 sna: Mark proxies as dirty on first relocation Signed-off-by: Chris Wilson commit 1d2fa5731b7ecfe34a8af809e45bcd3b0b70c890 Author: Chris Wilson Date: Thu Dec 6 22:15:09 2012 +0000 Remove the default log message Breaks compilation with xorg-1.13 Signed-off-by: Chris Wilson commit 7f4d4afa629bd18be89d7270e6178a865cf9586e Author: Chris Wilson Date: Thu Dec 6 17:08:15 2012 +0000 Fix compilation of UMS probe following 13f47008ec Signed-off-by: Chris Wilson commit d733f7d1f1dc343ac34c4a27ac99d71fc0572bc2 Author: Chris Wilson Date: Thu Dec 6 16:55:00 2012 +0000 sna/gen4+: Add common glyph-to-dst emitters Signed-off-by: Chris Wilson commit 13f47008ec411609968c40b8ec34dd495f14c50b Author: Chris Wilson Date: Thu Dec 6 14:05:33 2012 +0000 Refactor the common probe methods for scrn construction Signed-off-by: Chris Wilson commit 0040eb84c9187476a75202ebb251dd74354e4fc7 Author: Chris Wilson Date: Wed Dec 5 09:25:17 2012 +0000 sna: Don't disable CPU bo if supported on unknown hw Signed-off-by: Chris Wilson commit 14069f48645867a735ebdccb1e27a62364643c38 Author: Chris Wilson Date: Wed Dec 5 09:24:02 2012 +0000 sna: Assume that future hardware only gets more flexible E.g. that BLT can always write to cacheable memory, inflexible fences are a thing of the past, etc. Signed-off-by: Chris Wilson commit b82bfcb54a6af0d1ee17806ef94d9da504cad606 Author: Chris Wilson Date: Tue Dec 4 17:26:47 2012 +0000 sna/gen6+: Cache the scanout targets separately to avoid override PTE caching Signed-off-by: Chris Wilson commit 626dd1324dd2c5b14ca4aff598b5eb1e45550e69 Author: Chris Wilson Date: Tue Dec 4 12:52:50 2012 +0000 sna/gen4: Workaround render corruption with multiple SF threads Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=57410 Signed-off-by: Chris Wilson commit dbdb8fabfbade3f19fd8af3524468b5e6668bb66 Author: Chris Wilson Date: Tue Dec 4 12:37:23 2012 +0000 sna/gen4: Special case solids through the general vertex emitter Signed-off-by: Chris Wilson commit a17354d5ce6aeeab3f6e42aba63fce06ad18c526 Author: Chris Wilson Date: Tue Dec 4 12:01:26 2012 +0000 sna/gen4: Remove unused CC viewport Signed-off-by: Chris Wilson commit b9afb9cb614d2ad44330eed03b3f577a35184a88 Author: Chris Wilson Date: Tue Dec 4 11:14:58 2012 +0000 sna/gen4: Avoid emitting URB_FENCE across a cache-line Old erratum. Signed-off-by: Chris Wilson commit 08c30c48bdd0db500498bd9617c15f37bacd8de9 Author: Chris Wilson Date: Mon Dec 3 13:23:33 2012 +0000 sna: Tidy addition of fake GTF modes for panels Signed-off-by: Chris Wilson commit 6f675eeaeade4728af566891b2afbe5b44fbdc2e Author: Chris Wilson Date: Mon Dec 3 10:47:35 2012 +0000 2.20.15 release Signed-off-by: Chris Wilson commit 37eb7343be1aeeb90a860096756603a577df1a77 Author: Chris Wilson Date: Sat Dec 1 09:40:11 2012 +0000 sna/gen5: Inspired by gen4, reorder the flushing This may not be totally safe, but it is a nicer explanation for random single character corruption. References: https://bugs.freedesktop.org/show_bug.cgi?id=51422 Signed-off-by: Chris Wilson commit a8a99428c14c8aed2082853cc60d0f98a1fa2d86 Author: Chris Wilson Date: Sat Dec 1 09:44:49 2012 +0000 sna/dri: Unknown generations are given the max value and presume i965_dri.so Signed-off-by: Chris Wilson commit 382bb7bf77fca412fdefd7c304f395d1fe9483b5 Author: Chris Wilson Date: Fri Nov 30 16:24:30 2012 +0000 sna/gen6+: Only apply the BLT w/a for self-copies Signed-off-by: Chris Wilson commit 1aee8acacfe5869a072d9f20f3b8290b16683260 Author: Chris Wilson Date: Fri Nov 30 12:17:25 2012 +0000 sna: Unify gen4 acceleration again After disabling render-to-Y, 965g seems just as happy with the new code paths as g4x. Signed-off-by: Chris Wilson commit 5d6dd9c5a7eeb1f879525430ad89ab74d427e469 Author: Chris Wilson Date: Fri Nov 30 12:12:49 2012 +0000 Convert generation counter to octal Signed-off-by: Chris Wilson commit 131600020638ef15166361214cd5e1a0c08c2ea6 Author: Chris Wilson Date: Fri Nov 30 11:43:56 2012 +0000 sna: Prevent gen4 from rendering to I915_TILING_Y It always seems to end up in a hang... Signed-off-by: Chris Wilson commit ee4d1873ff504c2150b51d13864164b02b85dd53 Author: Chris Wilson Date: Fri Nov 30 09:52:49 2012 +0000 sna/g4x: And remove one more flush point Signed-off-by: Chris Wilson commit 1b6c1a30723b1d13e9bd3df0b59a8d75639c89be Author: Chris Wilson Date: Fri Nov 30 09:27:57 2012 +0000 sna: Increase tiling alignment to an even tile Seems to help g4x. Signed-off-by: Chris Wilson commit 6aeb6fdf75fa322d8f5ffe393337d8195d7a6a62 Author: Chris Wilson Date: Fri Nov 30 09:03:53 2012 +0000 sna/g4x: Remove the flush before the pipelined pointer changes This one appears unneeded. Hopefully. Signed-off-by: Chris Wilson commit 8be2c6695509809c0ab0c5c014e11dc733f73006 Author: Chris Wilson Date: Thu Nov 29 19:56:15 2012 +0000 sna/g4x: Emit the flush prior to changing pipelined state Signed-off-by: Chris Wilson commit 2d5d55702bb8eced32d5b8cb3c0cd125fd99d6dc Author: Chris Wilson Date: Wed Nov 28 15:46:29 2012 +0000 sna/gen6+: Override PTE cacheability bits for reads This is primarily useful for enabling the render cache for reads from scanouts. Signed-off-by: Chris Wilson commit 3ed87945c9e83fefcbda053b616856658bf2ac8e Author: Chris Wilson Date: Wed Nov 28 11:05:40 2012 +0000 sna/gen5+: Disable max-vbo address As we do not use this feature, disable it and save the relocation. Signed-off-by: Chris Wilson commit b2c9e9da639a134577485f83e0f66f54e2371b98 Author: Chris Wilson Date: Tue Nov 27 11:32:11 2012 +0000 sna/blt: Avoid clobbering common state before converting to a BLT In case we need to continue on with the render operation, we need to preserve the existing state. Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=57601 Signed-off-by: Chris Wilson commit 1e06d19a00f5a5a05369deeb3c5ae15b282c0f92 Author: Chris Wilson Date: Mon Nov 26 15:30:09 2012 +0000 sna: Disable shadow tracking upon regen References: https://bugs.freedesktop.org/show_bug.cgi?id=56608 References: https://bugs.freedesktop.org/show_bug.cgi?id=52255 Signed-off-by: Chris Wilson commit d21ed3a6aba5ae227cc5ecd164f3c18bc48c69af Author: Chris Wilson Date: Mon Nov 26 10:34:28 2012 +0000 sna: Use a single execobject flag to mark read/write domains Slight modification to the proposed API to only pass the simplified domain tracking now performed by the kernel. Signed-off-by: Chris Wilson commit 7a904ce839933d57176e013cdad147533c33ca2f Author: Chris Wilson Date: Mon Nov 26 08:48:31 2012 +0000 2.20.14 release Signed-off-by: Chris Wilson commit 1367e3f9ef5f606c8927cbde441a2ea41fa6d025 Author: Chris Wilson Date: Sun Nov 25 00:24:45 2012 +0000 sna: Exploit the alpha-fixup of the BLT for texture format conversions Signed-off-by: Chris Wilson commit 6d6399f97cf7cb91dcf89e9a5cd1243f761e4314 Author: Chris Wilson Date: Sun Nov 25 00:05:44 2012 +0000 sna: Transform a simple repeat pattern into range of the BLT Signed-off-by: Chris Wilson commit 39f1e228b74e98d3d87157cf093fc56ca31e6b13 Author: Chris Wilson Date: Sat Nov 24 20:16:29 2012 +0000 sna: Make GPU idle more consistent during wakeup Signed-off-by: Chris Wilson commit 40b6121746c55153de444ccb753df80706ff3a69 Author: Chris Wilson Date: Fri Nov 23 18:21:41 2012 +0000 sna/g4x: Refine the w/a for the broken sf shader Signed-off-by: Chris Wilson commit e0b906b09697b5fe7a5be2fdc52abd9b1c73f96d Author: Chris Wilson Date: Fri Nov 23 18:01:25 2012 +0000 sna/g4x: Use the render pipeline more often for composite operations Signed-off-by: Chris Wilson commit 54d8968da5ae39bfbcad511322926931bce2bda3 Author: Chris Wilson Date: Fri Nov 23 17:47:49 2012 +0000 sna/gen4: Revert changes to 965g[m] The changes tested on g45/gm45 prove to be highly unstable on 965gm, suggesting a radical difference in the nature of the bugs between the two generations. In theory, g4x has additional features that could be exploited over and above gen4 which may prove interesting in the future. Signed-off-by: Chris Wilson commit d560296561f68c3ac841345c0f4ce2c8e7381156 Author: Chris Wilson Date: Fri Nov 23 16:35:06 2012 +0000 sna/gen4: Don't force a flush for the dirty target if the we do not read back Signed-off-by: Chris Wilson commit 4023b2044757a9a67d564be0c8adf4885973a6e3 Author: Chris Wilson Date: Fri Nov 23 15:42:18 2012 +0000 sna/gen4: Force composite(WHITE, glyph) for building the glyphstring mask For reasons that are not apparent, if we don't composite with source/mask for the glyph strings, there appears to be some cache corruption. About as bizarre as the rest of gen4 idiosynacracies. Signed-off-by: Chris Wilson commit f74b62755c6e41097c23cc506984859e556a3415 Author: Chris Wilson Date: Fri Nov 23 14:59:42 2012 +0000 sna/gen4: Set composite op before testing for a BLT compatible op Signed-off-by: Chris Wilson commit 4c922eb52cadb867a0a15929e5a214c84a5992f3 Author: Chris Wilson Date: Fri Nov 23 14:19:59 2012 +0000 sna/gen4: Pass the mask channel explicitly rather than through a dummy mask Signed-off-by: Chris Wilson commit 2e68efa8ec66b4c89e9816bfa15067b398da5e3e Author: Chris Wilson Date: Fri Nov 23 14:04:51 2012 +0000 sna/gen4: Reduce the flush before performing the CA pass Signed-off-by: Chris Wilson commit 43aff6708fe97aa2fae0e30f98fc7cd9d7311b75 Author: Chris Wilson Date: Fri Nov 23 13:37:44 2012 +0000 sna/gen4: Update render fill routines to use the dummy mask Signed-off-by: Chris Wilson commit 352910712266202ef017066891ec383fd037fc4a Author: Chris Wilson Date: Fri Nov 23 13:29:01 2012 +0000 sna/gen4: Move the flush from inside the spans to emit_state() Signed-off-by: Chris Wilson commit 217f3e835b99002669999f818afa0d5c3a1cc852 Author: Chris Wilson Date: Fri Nov 23 12:32:14 2012 +0000 sna/gen4: Backport the more efficient composite box emitters Now that we aren't flushing after every single rectangle, we can strive for a little efficiency. Signed-off-by: Chris Wilson commit d3145e3f8146e7d864d32aec49c44c04d619e56a Author: Chris Wilson Date: Fri Nov 23 12:28:21 2012 +0000 sna/gen4: Use a dummy white mask to avoid the flush w/a when compositing Signed-off-by: Chris Wilson commit 65d530b59b957cc5f303ae819baad8075a555ac0 Author: Chris Wilson Date: Thu Nov 22 08:18:38 2012 +0000 Revert "uxa: Refactor early-exit paths of uxa_try_driver_composite()." This reverts commit fa5c573455cf63090dbb6d167d4e5f1cb23daf72 as it causes a SIGSEGV. Reported-by: lu hua Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=57389 commit d3a49f36395d737698616fe8ba9da7b74cd2d89a Author: Chris Wilson Date: Wed Nov 21 18:41:00 2012 +0000 sna/gen3+: Clear the render.vbo when replacing it for vertex upload As we may trigger a flush and a retire when searching for a vertex buffer for the new vertices, we need to be careful to decouple the destroyed vbo in order to avoid a use-after-free when inspecting the state. Signed-off-by: Chris Wilson commit 1c57a52a1f46e8401429955d8c96fd5095c9012a Author: Chris Wilson Date: Wed Nov 21 18:17:31 2012 +0000 sna: Assert that we do not replace a GPU bo with a proxy Signed-off-by: Chris Wilson commit 8da12a00ee653510e1f1a6fecb28dbb36faa8400 Author: Chris Wilson Date: Wed Nov 21 17:28:46 2012 +0000 sna: Skip uploading unintialiased pixmap data References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 0696ea4bd601ed823dbded03eaef6f316df2a5e8 Author: Chris Wilson Date: Wed Nov 21 17:25:33 2012 +0000 sna: Add the missing assertion for !proxy References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit ee72375ecd4f6d6e756bc361fa512b6675309540 Author: Chris Wilson Date: Wed Nov 21 17:08:53 2012 +0000 sna: Do not migrate uninitialised pixmaps Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 44dad490140d85a4c0dcb916030c36a838670c01 Author: Chris Wilson Date: Wed Nov 21 16:22:35 2012 +0000 sna: Do not dispose of a shadow pixmap Fixes regression from 2249e9edc37811c07e2807d6b4def05585b44c22 Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 9c627a05247690891062a2c0c1c8f7bbc0273104 Author: Chris Wilson Date: Wed Nov 21 16:03:02 2012 +0000 sna: Remove the kgem_bo_is_mappable refcnt assertion from freed paths A few callers of kgem_bo_is_mappable operate on freed bo, and so need to avoid the assert(bo->refcnt). References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit c0c48c7a5aca4d24936efbeaefc7674ada2ef87f Author: Chris Wilson Date: Wed Nov 21 14:46:45 2012 +0000 sna: Add a few refcnt assertions References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 0884777c33d20dbc329b98ad0db5ffb0df93ac8c Author: Chris Wilson Date: Wed Nov 21 14:31:44 2012 +0000 sna: Fix bogus assertion from 03fb9ded43 Signed-off-by: Chris Wilson commit 2249e9edc37811c07e2807d6b4def05585b44c22 Author: Chris Wilson Date: Wed Nov 21 13:05:36 2012 +0000 sna: Dispose of local copy for render sources If we transfer the pixmap to the GPU to use as a render source, presume that we will not need to then touch the local copy (at least for a while) and so return that memory to the system. Signed-off-by: Chris Wilson commit 736bb0f7058bf05ef48cdfe6a30d880de817aff9 Author: Chris Wilson Date: Wed Nov 21 12:16:46 2012 +0000 sna: Tighten a couple of assertions for damage with use_bo Signed-off-by: Chris Wilson commit 03fb9ded43f9bf8b73f99d5b3a8dc592fe22b523 Author: Chris Wilson Date: Tue Nov 20 21:13:07 2012 +0000 sna: Assert that we do not create a proxy with existent GPU damage References: http://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 9ab1d1f94e502e5fde87e7c171f3502f8a55f22b Author: Chris Wilson Date: Tue Nov 20 18:42:58 2012 +0000 sna/dri: Queue a vblank-continuation after flip-completion If a vblank request was delayed due to a pending flip, we need to make sure that we then queue it after that flip or else progress ceases. Reported-by: Jiri Slaby Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=56423 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=57156 Signed-off-by: Chris Wilson commit 7a7a76b359f73a4c4bcda0d88004f4dd5e94a186 Author: Chris Wilson Date: Tue Nov 20 16:05:32 2012 +0000 sna/dri: Avoid a NULL dereference inside a DBG Signed-off-by: Chris Wilson commit fa5c573455cf63090dbb6d167d4e5f1cb23daf72 Author: Eric Anholt Date: Sat Nov 17 13:11:13 2012 -0800 uxa: Refactor early-exit paths of uxa_try_driver_composite(). Saves 200b of code at -O2. commit edefb64d2b1c95b0b678cb222273ab64b2e6db2a Author: Eric Anholt Date: Sat Nov 17 13:11:12 2012 -0800 uxa: Work around uninitialized-value warning. The compiler isn't noticing that localDst only diverges from pDst when the _copy variables have also been set. commit 18b2e2a82724407196001ca853bd83150c66c5bd Author: Eric Anholt Date: Sat Nov 17 13:11:10 2012 -0800 uxa: Add printf attribute to intel_debug_fallback(). Shuts up a bunch of warnings with xorg's shared warning flags, and should give us more informative warnings in our code. commit 2d1e267e662505ca0dd318765464a24739dc5bd8 Author: Eric Anholt Date: Sat Nov 17 13:11:09 2012 -0800 uxa/dri: Factor out the repeated swap fallback code. commit cd2f373da7a14e004c999f9f0efaf88c785d3d3f Author: Eric Anholt Date: Sat Nov 17 13:11:08 2012 -0800 configure.ac: Fix bad syntax for test calls commit b8c01d9bd7ce5656706ebebd16e5a8c5ca0ba487 Author: Chris Wilson Date: Mon Nov 19 15:20:10 2012 +0000 sna/dri: Add an assert that the cached DRI2 buffer is pinned for DRI Signed-off-by: Chris Wilson commit 84b1a02fa9fde02366e0384044526982e70d0e8d Author: Chris Wilson Date: Mon Nov 19 13:41:28 2012 +0000 sna/dri: Avoid setting off-delay after a failed flip (use-after-free) Signed-off-by: Chris Wilson commit b4dd1103a55406bcd65f137c668701074a5c41b6 Author: Chris Wilson Date: Sun Nov 18 12:21:49 2012 +0000 sna/gen6+: Tweak prefer-blt to offset latency when in rc6 Signed-off-by: Chris Wilson commit 0cb8544dc16d4c1e9ae7f1ee74ee26c7501a9ed7 Author: Chris Wilson Date: Sun Nov 18 12:13:46 2012 +0000 Remove useless indirection of intel_chipsets Once upon a time this was used to hide a compiler warning about a pointer mismatch, now the compiler still warns about the cast, making the indirect moot. Signed-off-by: Chris Wilson commit 8509e474f57533fc6afcf213165f4c8633631fb5 Author: Chris Wilson Date: Fri Nov 16 23:02:44 2012 +0000 sna/dri: Clear the last-used-cpu flag when performing CopyRegion Keeps the internal bookkeeping intact after the small bypass. Signed-off-by: Chris Wilson commit 866ed4a26cbbb29ef3845b0aa56383c4d951c65a Author: Chris Wilson Date: Thu Nov 15 15:13:14 2012 +0000 sna/dri: Add a couple of more asserts to track injection of a rogue bo Signed-off-by: Chris Wilson commit d2897cb0136ffec83365c7530ed544b562cac478 Author: Chris Wilson Date: Tue Nov 13 12:43:44 2012 +0000 sna/gen4,5: Fix the opacity shader to use the right vertex attribute Reported-by: Edward Sheldrake Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=57054 Signed-off-by: Chris Wilson commit 440eaa049756e5266f3bd80e2751f1fd0d5f9890 Author: Chris Wilson Date: Tue Nov 13 12:42:58 2012 +0000 sna/gen4: Tidy emission of opacity vertex attribute Just make it more consistent between the various emitters Signed-off-by: Chris Wilson commit b6d2bb961517623d46aa6944307cb998ee125459 Author: Chris Wilson Date: Tue Nov 13 10:05:56 2012 +0000 sna/gen4: Do not prefer inplace non-rectilinear spans As gen4 requires the per-rectangle vertex flush, emitting spans on the GPU is inefficient and so we prefer to composite the mask instead. Signed-off-by: Chris Wilson commit ae293609c7400cd3c753ed3762772264c4741df5 Author: Chris Wilson Date: Tue Nov 13 10:21:29 2012 +0000 sna/gen4: Always initialise redirect Do not assume the caller cleared the composite-op structure for us. Signed-off-by: Chris Wilson commit 2954f15e2bcb590a90c2cb6077c0843ee25a4413 Author: Chris Wilson Date: Tue Nov 13 09:46:19 2012 +0000 sna: Specialise the decision for inplace xor uploads Fixes a regression from commit 0be1d964713ca407f029278a8256d02d925dc9da Author: Chris Wilson Date: Tue Sep 11 21:48:24 2012 +0100 sna: Use inplace X tiling for LLC uploads which introduced the ability to swizzle into CPU maps, but also convinced the xorg path to the same - which for large images blows up. Reported-by: Michael Laß Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=57031 Signed-off-by: Chris Wilson commit 66eb0adffa63ef8ece7621ba90dc96af91549612 Author: Chris Wilson Date: Mon Nov 12 14:45:56 2012 +0000 sna/dri: Apply the can_blit() check for CopyRegion CopyRegion() also needs to check for stale bo in case the pixmap dimensions have changed size and so may cause out-of-bounds read/writes. Signed-off-by: Chris Wilson commit b27ecf3059bc066ef59f2a71c1d8d8f0ffec7191 Author: Chris Wilson Date: Mon Nov 12 14:06:06 2012 +0000 sna/dri: Prevent scheduling a swap on stale buffers If the screen has been reconfigured and the DRI client tries to swap the old buffers (having not processed the invalidate event and retrieved the current names), quietly drop the request. If we try to obey the request, we will end up attaching a back buffer that is the wrong size to the scanout... Signed-off-by: Chris Wilson commit 34aa1e3b2702a55799a5655a3ba10bce4cc2065a Author: Chris Wilson Date: Mon Nov 12 11:22:53 2012 +0000 sna: Compile against ancient libdrm We need to trim the sprite video overlays if the prototypes are not known. Signed-off-by: Chris Wilson commit 8f1afde57dca27e6542b0b8e7c87750f3d6367bf Author: Chris Wilson Date: Sun Nov 11 16:16:20 2012 +0000 2.20.13 release Signed-off-by: Chris Wilson commit b16219a19f48b52dda91f26fcbbbbeda056589ab Author: Chris Wilson Date: Sun Nov 11 11:05:35 2012 +0000 sna: Filter out the full-damage marker when undoing redirection ==25902== Invalid read of size 4 ==25902== at 0x4980E13: _list_del (intel_list.h:218) ==25902== by 0x4980EAB: list_del (intel_list.h:240) ==25902== by 0x4981F4B: free_list (sna_damage.c:403) ==25902== by 0x4985131: __sna_damage_destroy (sna_damage.c:1467) ==25902== by 0x49A5276: sna_render_composite_redirect_done (sna_render.c:1921) ==25902== by 0x49C68FC: gen2_render_composite_done (gen2_render.c:1136) ==25902== by 0x497F90F: sna_composite (sna_composite.c:567) ==25902== by 0x4994725: glyphs_via_mask (sna_glyphs.c:1139) ==25902== by 0x4995FB7: sna_glyphs (sna_glyphs.c:1688) ==25902== by 0x8150EB4: ??? (in /usr/bin/Xorg) ==25902== by 0x813CA38: CompositeGlyphs (in /usr/bin/Xorg) ==25902== by 0x8146DE1: ??? (in /usr/bin/Xorg) ==25902== Address 0x7c079ac2 is not stack'd, malloc'd or (recently) free'd Reported-by: bonbons67@internet.lu Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=56785 Signed-off-by: Chris Wilson commit 69acbb77e8aad3370d5e8d9a9e067c54872d7082 Author: Chris Wilson Date: Sun Nov 11 10:49:59 2012 +0000 sna: Fix printing of uninitialied value in DBG ==25902== Use of uninitialised value of size 4 ==25902== at 0x423098E: _itoa_word (_itoa.c:196) ==25902== by 0x4233F7F: vfprintf (vfprintf.c:1602) ==25902== by 0x42FAFAD: __vsnprintf_chk (vsnprintf_chk.c:65) ==25902== by 0x81DBE8E: Xvscnprintf (in /usr/bin/Xorg) ==25902== by 0x81DC8FB: LogVMessageVerb (in /usr/bin/Xorg) ==25902== by 0x81DCA62: LogVWrite (in /usr/bin/Xorg) ==25902== by 0x81DCA9B: VErrorF (in /usr/bin/Xorg) ==25902== by 0x81DC333: ErrorF (in /usr/bin/Xorg) ==25902== by 0x49B2FA8: trapezoid_span_inplace__x8r8g8b8 (sna_trapezoids.c:5069) ==25902== by 0x49B3407: trapezoid_span_inplace (sna_trapezoids.c:5166) ==25902== by 0x49B4C96: sna_composite_trapezoids (sna_trapezoids.c:5619) Reported-by: bonbons67@internet.lu Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=56785 Signed-off-by: Chris Wilson commit 66e4c8ff40ab8cf722efa4293bb17b0d8f2dfa88 Author: Chris Wilson Date: Sun Nov 11 09:40:09 2012 +0000 sna: Flush pending rendering before enabling an output This is to prevent falling in the trap of the rendering being delayed until the next client renders some new content. Signed-off-by: Chris Wilson commit 94dd0b9ee9f55e7c09b8c0ee18939fa69ce66da2 Author: Chris Wilson Date: Sat Nov 10 16:52:09 2012 +0000 sna/gen2: Fix use of uninitialised redirection ==29553== Invalid read of size 4 ==29553== at 0x4980E1B: _list_del (intel_list.h:218) ==29553== by 0x4980EB3: list_del (intel_list.h:240) ==29553== by 0x4981F53: free_list (sna_damage.c:403) ==29553== by 0x4985139: __sna_damage_destroy (sna_damage.c:1467) ==29553== by 0x49A527E: sna_render_composite_redirect_done (sna_render.c:1921) ==29553== by 0x49C6904: gen2_render_composite_done (gen2_render.c:1136) ==29553== by 0x497F917: sna_composite (sna_composite.c:567) ==29553== by 0x8150C41: ??? (in /usr/bin/Xorg) ==29553== by 0x8142F13: CompositePicture (in /usr/bin/Xorg) ==29553== by 0x8145F58: ??? (in /usr/bin/Xorg) ==29553== by 0x81436F2: ??? (in /usr/bin/Xorg) ==29553== by 0x807965C: ??? (in /usr/bin/Xorg) ==29553== Address 0x9407e188 is not stack'd, malloc'd or (recently) free'd Reported-by: bonbons67@internet.lu Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=56785 Signed-off-by: Chris Wilson commit 0f1c30818c9d782b066147448bbcc9ac95ac834f Author: Chris Wilson Date: Sat Nov 10 16:52:09 2012 +0000 sna: Fix use of uninitialised value in DBG ==29553== Use of uninitialised value of size 4 ==29553== at 0x4230964: _itoa_word (_itoa.c:195) ==29553== by 0x4233F7F: vfprintf (vfprintf.c:1602) ==29553== by 0x42FAFAD: __vsnprintf_chk (vsnprintf_chk.c:65) ==29553== by 0x81DBE8E: Xvscnprintf (in /usr/bin/Xorg) ==29553== by 0x81DC8FB: LogVMessageVerb (in /usr/bin/Xorg) ==29553== by 0x81DCA62: LogVWrite (in /usr/bin/Xorg) ==29553== by 0x81DCA9B: VErrorF (in /usr/bin/Xorg) ==29553== by 0x81DC333: ErrorF (in /usr/bin/Xorg) ==29553== by 0x49434F0: kgem_create_buffer (kgem.c:4887) ==29553== by 0x4943B09: kgem_create_buffer_2d (kgem.c:4969) ==29553== by 0x4943E19: kgem_upload_source_image (kgem.c:5021) ==29553== by 0x49A0567: upload (sna_render.c:505) ==29553== Reported-by: bonbons67@internet.lu References: https://bugs.freedesktop.org/show_bug.cgi?id=56785 Signed-off-by: Chris Wilson commit cc2b13c9c05e57dc5004d93b56f332ea95f0a4ef Author: Chris Wilson Date: Sat Nov 10 11:50:15 2012 +0000 sna: Specify read/write domains for no-relocation fastpath On review (read triggering BUGs), we do need to supply the domain tracking of the buffers that is being replaced from the relocation path. Signed-off-by: Chris Wilson commit 0c4a2bcc3d63ecc02e3a940e38e9a416b51ad0c8 Author: Chris Wilson Date: Sat Nov 10 12:34:52 2012 +0000 sna: Allow snooped buffers to be retained (and reused) between batches Signed-off-by: Chris Wilson commit f5d79b202dd448e61ab6ffce26fe9cbf9051d770 Author: Chris Wilson Date: Sat Nov 10 10:30:04 2012 +0000 sna/gen2: Add a modicum of fallback DBG References: https://bugs.freedesktop.org/show_bug.cgi?id=56785 Signed-off-by: Chris Wilson commit 27327633138dce159ca2e91fe5eac1565bd45e1c Author: Chris Wilson Date: Fri Nov 9 17:08:01 2012 +0000 sna/gen4: Only 965gm suffers the !snoop restriction So fixup the bogus assertion for g4x Signed-off-by: Chris Wilson commit 8d3b5ea135fd8f16da2cbfb98041e32c7001a38f Author: Chris Wilson Date: Fri Nov 9 15:31:03 2012 +0000 xvmc: Use DRMINTEL_LIBS instead of hardcoding -ldrm_intel Reported-by: Maarten Lankhorst Signed-off-by: Chris Wilson commit f040b97b01495aa43f7771ebb8ca5c0d44038bc1 Author: Chris Wilson Date: Thu Nov 8 23:42:10 2012 +0000 sna: Mark no-reloc write buffers If we bypass the relocation processing, we also then bypass the pending-write analysis, so we need to supply those to the kernel ourselves (to maintain gpu-cpu coherency). Signed-off-by: Chris Wilson commit 85ba7e96268dbb8da4bb34078333695a451c6570 Author: Chris Wilson Date: Thu Nov 8 15:56:13 2012 +0000 sna: Experiment with using reloc.handle as an index into the execbuffer Signed-off-by: Chris Wilson commit 93d8dddbb92431d6e2c48a17b71cac9f7047902e Author: Chris Wilson Date: Thu Nov 8 09:41:21 2012 +0000 sna: Set the known offset for the batch as well Signed-off-by: Chris Wilson commit 120fa0ef8d04f5e82e5f7a0636033d3d96efa1e8 Author: Chris Wilson Date: Wed Nov 7 17:41:20 2012 +0000 sna: Support a fast no relocation changed path x11perf -copywinwin10 on gm45 with c2d L9400: before: 553,000 op/s after: 565,000 op/s Signed-off-by: Chris Wilson commit b7d2fcf47a9569d0944097a8be60ca3be72b42f6 Author: Chris Wilson Date: Thu Nov 8 08:55:25 2012 +0000 Remove reliance on hard-coded DRI name This provides for using the existing DDX with future DRI drivers which may break from the traditional names - but only with the help of the user/packager. This scheme needs to be replaced with a robust mechanism for driver loading if AIGLX and co are to be kept. Signed-off-by: Chris Wilson commit cefce9c81585b73db7620e08fcf60c89a8204873 Author: Chris Wilson Date: Tue Nov 6 17:26:34 2012 +0000 sna: Abandon kernels that do not support execbuffer2 Signed-off-by: Chris Wilson commit b2245838c15b54d72557de8facb7cc15d59624ae Author: Chris Wilson Date: Tue Nov 6 16:32:32 2012 +0000 sna/gen4: opacity spans requires the per-rectangle flush w/a Note that this is worsened, but not caused, by: commit e1a63de8991a6586b83c06bcb3369208871cf43d Author: Chris Wilson Date: Fri Nov 2 09:10:32 2012 +0000 sna/gen4+: Prefer GPU spans if the destination is active References: https://bugs.freedesktop.org/show_bug.cgi?id=55500 Signed-off-by: Chris Wilson commit a0540ebff083974688c863e08203e3d71a297340 Author: Chris Wilson Date: Tue Nov 6 16:00:42 2012 +0000 sna/gen4: Remove a couple of old, now redundant, w/a flushes Signed-off-by: Chris Wilson commit aaaa6c356456a4bab595c6e6485893c538064e37 Author: Chris Wilson Date: Tue Nov 6 14:34:51 2012 +0000 sna/gen4: Flush after pipelined pointer updates (inverted logic!) Signed-off-by: Chris Wilson commit 28bda6707d979bca29dbea04e932819de204d920 Author: Chris Wilson Date: Mon Nov 5 22:30:29 2012 +0000 sna: Prevent use-after-free during partial move-to-gpu As we reuse the input parameter 'box' to hold the array of boxes that need to be migrated, we need to be careful that we do not later confuse it with the original input parameter. Otherwise, ==1315== Invalid read of size 2 ==1315== at 0x928B091: box_inplace (sna.h:506) ==1315== by 0x9292278: sna_pixmap_move_area_to_gpu (sna_accel.c:2554) ==1315== by 0x9292C14: sna_drawable_use_bo (sna_accel.c:2774) ==1315== by 0x9356C01: gen7_composite_set_target (gen7_render.c:2448) ==1315== by 0x9357AA2: gen7_render_composite (gen7_render.c:2800) ==1315== by 0x92DB12E: glyphs_to_dst (sna_glyphs.c:552) ==1315== by 0x92DEA8D: sna_glyphs (sna_glyphs.c:1664) ==1315== by 0x4F920E: damageGlyphs (in /tmp/Xorg) ==1315== by 0x4F2FF6: ProcRenderCompositeGlyphs (in /tmp/Xorg) ==1315== by 0x437260: Dispatch (in /tmp/Xorg) ==1315== by 0x426466: main (in /tmp/Xorg) ==1315== Address 0xd637054 is 20 bytes inside a block of size 208,464 free'd ==1315== at 0x4C2A2FC: free (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so) ==1315== by 0x92CCFCD: __sna_damage_destroy (sna_damage.c:1469) ==1315== by 0x928AD74: sna_damage_destroy (sna_damage.h:284) ==1315== by 0x9291CB2: sna_pixmap_move_area_to_gpu (sna_accel.c:2470) ==1315== by 0x9292C14: sna_drawable_use_bo (sna_accel.c:2774) ==1315== by 0x9356C01: gen7_composite_set_target (gen7_render.c:2448) ==1315== by 0x9357AA2: gen7_render_composite (gen7_render.c:2800) ==1315== by 0x92DB12E: glyphs_to_dst (sna_glyphs.c:552) ==1315== by 0x92DEA8D: sna_glyphs (sna_glyphs.c:1664) ==1315== by 0x4F920E: damageGlyphs (in /tmp/Xorg) ==1315== by 0x4F2FF6: ProcRenderCompositeGlyphs (in /tmp/Xorg) ==1315== by 0x437260: Dispatch (in /tmp/Xorg) Reported-by: Matti Ruohonen Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=56591 Signed-off-by: Chris Wilson commit e62b0cbf958d1ad95e4522973253a1ae5c1a4da9 Author: Chris Wilson Date: Fri Nov 2 15:54:20 2012 +0000 sna: Add a small delay during startup if another master is still active There exists a race with plymouthd that can cause the drm device to reject us as the rightful master, and so cause X to fail to load. Try waiting for a couple of seconds for whatever it was to close before giving in. Signed-off-by: Chris Wilson commit e1a63de8991a6586b83c06bcb3369208871cf43d Author: Chris Wilson Date: Fri Nov 2 09:10:32 2012 +0000 sna/gen4+: Prefer GPU spans if the destination is active Trying to avoid using the inplace scanline rasteriser for large shapes. Signed-off-by: Chris Wilson commit 72bcd8f85c98502b13a67d9c606371afe513584c Author: Chris Wilson Date: Thu Nov 1 09:30:18 2012 +0000 sna: Try to reduce ping-pong migration for intermixed render/legacy code paths References: https://bugs.freedesktop.org/show_bug.cgi?id=56591 Signed-off-by: Chris Wilson commit 4e363906a5ef15e1eb0a387cfb6b3445ac185b9d Author: Chris Wilson Date: Wed Oct 31 22:58:59 2012 +0000 sna: Set a valid box when checking for GPU bo for BLT composite ops Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit b924831e445615b82a53b10e1849720e933eddfe Author: Chris Wilson Date: Wed Oct 31 22:55:56 2012 +0000 sna: Preserve mode if flushing before a scanline wait Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 678f9586807071bef813bb69d451f14d2fcbcc04 Author: Chris Wilson Date: Wed Oct 31 11:26:18 2012 +0000 sna: assert that the source is not the GTT mapping when uploading Signed-off-by: Chris Wilson commit 783b8048a6d1a9fd0a73ebf7768ae17dc0b21900 Author: Chris Wilson Date: Wed Oct 31 11:16:09 2012 +0000 sna: Prefer to use the GPU for uploads if continuing on the GPU Signed-off-by: Chris Wilson commit 74c912880c302889f38fe5898c8038a0ba20e5db Author: Chris Wilson Date: Wed Oct 31 10:57:54 2012 +0000 sna: Fix a typo in a DBG Reported-by: Matti Ruohonen Signed-off-by: Chris Wilson commit f48a821aa73cb40a51baafc6cd2b063f1f91d864 Author: Chris Wilson Date: Wed Oct 31 10:53:47 2012 +0000 sna: Add a few DBG tracepoints to screen init/fini Signed-off-by: Chris Wilson commit bf81d552c4be039fbcf3272387828b1a8b3fbdb8 Author: Chris Wilson Date: Wed Oct 31 08:50:44 2012 +0000 sna: Clamp the drawable box to prevent int16 overflow And assert that the box is valid when migrating. References: https://bugs.freedesktop.org/show_bug.cgi?id=56591 Signed-off-by: Chris Wilson commit 31eb704b2ad7c861ec4e61fb9de0e9592fc6d269 Author: Chris Wilson Date: Fri Oct 26 13:57:30 2012 +0100 sna: Ensure that the trap is clipped if it ends within the boundary pixel Reported-and-tested-by: Jiri Slaby Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=56395 Signed-off-by: Chris Wilson commit ef431b2d35c1bf4d77bbcc73688951d22f6aa135 Author: Chris Wilson Date: Thu Oct 25 10:15:39 2012 +0100 uxa: Drain the DRM event queue before server regeneration Adam Jackson notes that what appeared to be my paranoid ramblings in SNA actually served a purpose - it prevents a server crash following server regen if an indirect client happened to be running at the time (e.g. LIBGL_INDIRECT_ALWAYS=1 glxgears). Reported-by: Adam Jackson Signed-off-by: Chris Wilson commit efb8ff16491ecfb4d9c0c6a718684310d949d8d3 Author: Chris Wilson Date: Wed Oct 24 22:56:20 2012 +0100 sna: Add missing ValidatePicture() for flattening alphamaps Reported-by: Armands Liepins Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=56367 Signed-off-by: Chris Wilson commit 1a489142c8e6a4828348cc9afbd0f430d3b1e2d8 Author: Chris Wilson Date: Tue Oct 23 23:43:50 2012 +0100 sna: Disable RandR hotplug events if Xinerama is enabled Since RandR itself is disabled if Xinerama is enabled, for example with ZaphodHeads, calling RRGetInfo() upon a hotplug event generates an assertion. Reported-by: Stephen Liang Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55260 Signed-off-by: Chris Wilson commit d87c2756db1af6e4af15864ab0f44d1454079236 Author: Chris Wilson Date: Tue Oct 23 15:50:56 2012 +0100 sna: Beware 16-bit overflow when computing sample areas Reported-by: Ognian Tenchev References: https://bugs.freedesktop.org/show_bug.cgi?id=56324 Signed-off-by: Chris Wilson commit c7f48684cdebc24128a5fa5678614af3deb14b3b Author: Chris Wilson Date: Tue Oct 23 15:17:56 2012 +0100 sna: Only disallow hw sprite scaling on Haswell Earlier chips (Ironlake, Sandybridge and Ivybridge) have integrated sprite scalers. Signed-off-by: Chris Wilson commit 5c3ea9cf6900855502fcd56214a1b9e180265ff5 Author: Chris Wilson Date: Mon Oct 22 22:35:17 2012 +0100 sna: Update DRI buffer if attached to the framebuffer for TearFree flips Signed-off-by: Chris Wilson commit 4dfc83359d006a4e410e3280003b49683309afc3 Author: Chris Wilson Date: Mon Oct 22 14:56:01 2012 +0100 sna: Tidy udev install/remove and add a couple of lines of DBG References: https://bugs.freedesktop.org/show_bug.cgi?id=55260 Signed-off-by: Chris Wilson commit 4d9687d49c9869b2e88d408e5f451c9a1f8f3389 Author: Chris Wilson Date: Mon Oct 22 13:41:54 2012 +0100 sna: Refactor the common code to enable a timer Signed-off-by: Chris Wilson commit fb729788872ccb429ddde8a9a4281b1933243096 Author: Chris Wilson Date: Sun Oct 21 14:36:48 2012 +0100 sna: Only query the system time if we are processing timers Signed-off-by: Chris Wilson commit c0d6a75f02eb97e5c80a4345ae5c68e9a81d49b6 Author: Chris Wilson Date: Sun Oct 21 14:32:14 2012 +0100 sna: Use the FLUSH_TIMER as the only wakeup timer source Signed-off-by: Chris Wilson commit 7bc829c39a203c17053eb728412f698a429ad9fe Author: Chris Wilson Date: Sun Oct 21 14:24:01 2012 +0100 sna: Remove the unused inactive eviction Signed-off-by: Chris Wilson commit 9fa6e4aa2daee99ff5f6efc11232de22100bac80 Author: Chris Wilson Date: Sun Oct 21 12:48:06 2012 +0100 intel: Sanity check that the platform probes points to a i915.ko GEM device References: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1069031 Signed-off-by: Chris Wilson commit f6eed98fcfea54d50a282ac71ee654645551ae11 Author: Chris Wilson Date: Sun Oct 21 10:46:14 2012 +0100 sna: Mark the to-be-damaged region first, then Process afterwards Damage is processed in two phases, with the actual Damage being appended before the operation is performed so that a copy can be made before modification (e.g. software cursors). Signed-off-by: Chris Wilson commit 60e4e009f90543bfd57f6a4e51ebc5b32b4af33b Author: Chris Wilson Date: Sat Oct 20 17:59:45 2012 +0100 sna: Move the source region to the CPU for a self-copy fallback CopyArea Signed-off-by: Chris Wilson commit 7ff8b1ba543508f3b209f2ade7655c3aa34d546d Author: Chris Wilson Date: Sat Oct 20 16:23:26 2012 +0100 2.20.12 release How embarrassing! My fault for rushing :( Signed-off-by: Chris Wilson commit 695b2ce2d32bde191080129b55f9bf8a9d48ee77 Author: Chris Wilson Date: Sat Oct 20 16:19:21 2012 +0100 uxa: Fixup drm_intel_bo_disable_reuse() typo Signed-off-by: Chris Wilson commit 2083e253b3d1ecc218ab1e523e4026ddd4561112 Author: Chris Wilson Date: Sat Oct 20 16:07:11 2012 +0100 2.20.11 release Signed-off-by: Chris Wilson commit fc0ba65f5efe217f2ab5e920255d2133d7c9e5e8 Author: Chris Wilson Date: Sat Oct 20 09:29:10 2012 +0100 uxa: Disable bo reuse after binding to a scanout On gen6+, bo are expected to be LLC by default. However, as using the bo for the scanout causes it to be moved into the uncached domain, this assumption is then false and we should release the bo back to the system rather than spread the uncached buffers around. The most common allocator of scanouts is for pageflipping which are already non-reusable due to the DRI2 export, so there should actually be little impact. Signed-off-by: Chris Wilson commit f4c32af48b0c92a48131090886a6a6b6c45dbe34 Author: Chris Wilson Date: Fri Oct 19 16:29:19 2012 +0100 sna: Clear the damage along with the BO when forcing the stall for inplace BLT Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=56180 Signed-off-by: Chris Wilson commit 299232bdb69c8c2b6231905e0f45e9cfe74fe09a Author: Chris Wilson Date: Fri Oct 19 15:02:00 2012 +0100 sna: Reorder final checks for using the BO and setting the damage pointer When we return NULL from sna_drawable_use_bo(), the expectation is that the damage pointer is also NULL. However, one SHM path leaked. References: https://bugs.freedesktop.org/show_bug.cgi?id=56180 Signed-off-by: Chris Wilson commit 257abfdabe39629fb458ed65fab11283f7518dc4 Author: Chris Wilson Date: Wed Oct 17 23:34:22 2012 +0100 sna/gen4: Presume we need a flush upon state change similar to gen5+ References: https://bugs.freedesktop.org/show_bug.cgi?id=55627 References: https://bugs.freedesktop.org/show_bug.cgi?id=55500 Signed-off-by: Chris Wilson commit 8238c672984e31ae655353d6412e3395a9cdfbe6 Author: Chris Wilson Date: Wed Oct 17 22:16:29 2012 +0100 sna: secure batches accepted upstream, so simply use runtime detection Signed-off-by: Chris Wilson commit 540666a0a81c7daedbd47830d0932df5e57ec903 Author: Chris Wilson Date: Wed Oct 17 16:59:05 2012 +0100 sna/overlay: Move bo out of GTT domain after binding to overlay plane Signed-off-by: Chris Wilson commit 891bae4aa91e85542dcbe38f6ee92141e3efc801 Author: Chris Wilson Date: Wed Oct 17 11:29:10 2012 +0100 sna: Use the secure batches to program scanline waits on gen6+ Signed-off-by: Chris Wilson commit 41be80a8cae1eb0e294392e5033511bfdf2895c5 Author: Chris Wilson Date: Wed Oct 17 11:25:52 2012 +0100 sna: Enable support for SECURE batch buffers Signed-off-by: Chris Wilson commit ba6c82cd9d8089354b90632ca8edbb35cc09b9c4 Author: Chris Wilson Date: Wed Oct 17 13:54:51 2012 +0100 sna/dri: Defensively check for GTT mmap failure during fallback Signed-off-by: Chris Wilson commit 7927f9a351ead1a5593bc91e465706bdd889bb8d Author: Chris Wilson Date: Tue Oct 16 17:56:30 2012 +0100 sna/gen7: Filter BLEND flags for CA glyphs Fixes regression from commit c51aaa731e5cffc892e59730194ad7c98789b02b Author: Chris Wilson Date: Thu Oct 11 11:36:00 2012 +0100 sna/gen7: Replace bogus state tracking assertion The assumption being that we only used the encoded flags for determining the composite state is false for the magic CA pass. Reported-by: Oleksij Rempel Reported-by: Eyal Lotem Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=56037 Signed-off-by: Chris Wilson commit 2ad4aa195571fe214ccffa55e123507f1be66243 Author: Chris Wilson Date: Tue Oct 16 11:59:28 2012 +0100 sna: Drop fake tiled CPU mapping The only path where this is correct already handles it as the special case that it is, everywhere else it just nonsense. Signed-off-by: Chris Wilson commit b42d81b63f5b6a571faffaadd42c74adce40128a Author: Chris Wilson Date: Sun Oct 14 09:15:38 2012 +0100 2.20.10 release Signed-off-by: Chris Wilson commit 877e9c57c788cb3d4db1e96e519b26ca5d542465 Author: Chris Wilson Date: Fri Oct 12 17:57:28 2012 +0100 sna/overlay: Trim suggested BestSize to fit within the overlay constraints As the maximum reported image sizes are for the source image, we should be careful not to recommend the application use an output Window larger than can be handled by the overlay hardware. So shrink it to fit, whilst preserving the aspect ratio. Signed-off-by: Chris Wilson commit 1ec41590c9d142a77a2fdcfcd9a762aca99d9d86 Author: Paulo Zanoni Date: Thu Oct 11 18:10:17 2012 -0300 Fix possible_clones computation for shared encoders between outputs Libdrm's possible_clones is a mask of encoders. Xorg's possible_clones is a mask of outputs, so we just can't do the following: output->possible_clones = kencoder->possible_clones; This is a problem on Haswell because, at least with the current patches floating on the mailing list, there is more than one connector per encoder. This patch writes the code to properly translate libdrm's encoder mask into Xorg's output mask. Signed-off-by: Paulo Zanoni Signed-off-by: Chris Wilson commit 0b92a5a3062d73666279e17ac2d711314bcd8dfe Author: Chris Wilson Date: Thu Oct 11 12:16:03 2012 +0100 sna: Correct a bogus assertion after inplace PutImage If we are forced to use the GPU bo as the target because the CPU bo is busy, the priv->cpu flag is likely to remain set, so we need to clear it rather than fail the assertion that is false. Signed-off-by: Chris Wilson commit c51aaa731e5cffc892e59730194ad7c98789b02b Author: Chris Wilson Date: Thu Oct 11 11:36:00 2012 +0100 sna/gen7: Replace bogus state tracking assertion Signed-off-by: Chris Wilson commit a56fa2fb4aad172be905c3abeefd987804553809 Author: Chris Wilson Date: Thu Oct 11 10:55:34 2012 +0100 sna: Rearrange switch to suppress a compiler warning in sna_copy_plane_blt() Hide the impossible default case so that static analyzers are not confused. Signed-off-by: Chris Wilson commit 94a98d1c472ce0fc7601421a2bcfbd0524d2bc93 Author: Chris Wilson Date: Thu Oct 11 08:42:21 2012 +0100 sna: Limit available space for gen2 Since pre-g33 chipsets impose massive alignment restrictions on objects within the aperture we need to further restrict the amount of available space to be sure we have sufficient room to accommodate the alignment. Signed-off-by: Chris Wilson commit d3ae85dd01952ecff7b780e30a43571146aa61ea Author: Chris Wilson Date: Wed Oct 10 17:27:38 2012 +0100 sna: Use high/low watermarks for fenced aperture counting This helps with fitting larger operations into the small apertures of gen2, which due to the lax accounting could trigger ENOSPC. Signed-off-by: Chris Wilson commit dadffd0061e2b14db397608e3cedfe9c76e546c8 Author: Dave Airlie Date: Thu Oct 11 14:30:29 2012 +1000 intel: fix fullscreen damage posting on pageflip While playing with MPX and sw cursor I noticed page flips won't end up misrendering some bits, so the sw cursor was replacing the bits on the wrong pixmap. Fix the damage handling to be correct and append damage before swapping the pointers and process damage after. This fixes misrendering with MPX cursors under a fullscreen compositor, that pageflips. Signed-off-by: Dave Airlie [ickle: The secret is that damage is sometimes reported before the rendering in DamageRegionAppend, and sometimes afterwards in DamageRegionProcessPending. For instance, the software cursor operates prior to being rendered over.] Signed-off-by: Chris Wilson commit a1ea19a3edaac6e1e5d240b75b7c5fdcfea0e0a8 Author: Chris Wilson Date: Wed Oct 10 23:20:13 2012 +0100 sna: Prevent sign-extension when manipulating strides Reported-by: Prit Laes Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55823 Signed-off-by: Chris Wilson commit d73f5b5bb1a81421f1fdc3ac3b460a8ad90be0ad Author: Chris Wilson Date: Tue Oct 9 19:41:10 2012 +0100 sna/gen6+: Initialize the damage for fill-boxes In case we need to redirect the rendering for a large render target, we need to initialize the damage pointer. Reported-by: Clemens Eisserer Bugzilla: ttps://bugs.freedesktop.org/show_bug.cgi?id=55812 Signed-off-by: Chris Wilson commit ae6ae91baa0574a7a23ab76afac5e2d827c49c20 Author: Chris Wilson Date: Tue Oct 9 18:36:59 2012 +0100 sna: Check the map exists before trying to release it Reported-by: Clemens Eisserer Bugzilla: ttps://bugs.freedesktop.org/show_bug.cgi?id=55812 Signed-off-by: Chris Wilson commit 8bbfa88a8691f6c1b2903090c3cd3159126ef563 Author: Chris Wilson Date: Tue Oct 9 17:26:39 2012 +0100 sna: Also check the bg rrop for reads when deciding upon fallback placement Reported-and-tested-by: chr.ohm@gmx.net Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55810 Signed-off-by: Chris Wilson commit f8a005d1f984382272ad10cabd88c0422d4e76e8 Author: Chris Wilson Date: Tue Oct 9 00:23:24 2012 +0100 sna: Pass the region, not its pointer, when moving the GC to the CPU Fortunately, the sgc->region was not used along that particular fallback path. Signed-off-by: Chris Wilson commit fb5205a86da09b344dbc20598655e917c263125c Author: Chris Wilson Date: Mon Oct 8 09:12:33 2012 +0100 sna/gen5: Flush pipelined ops when changing state When is a pipelined operation, not pipelined? That is the mystery posed by our hardware! Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=51422 Signed-off-by: Chris Wilson commit bed64c998df538046d77eeea90598f48077dab5f Author: Chris Wilson Date: Mon Oct 8 00:13:03 2012 +0100 sna: Fix check_reloc_and_exec typo Garbage xorg includes hiding genuine compiler warnings ftl once again. Signed-off-by: Chris Wilson commit 2ac3776be85d857a57ce7b742e52cd6091d2befb Author: Chris Wilson Date: Sun Oct 7 22:41:25 2012 +0100 sna: Check that we have sufficient space for a copy when replacing a fill Reported-by: Timo Kamph Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55700 Signed-off-by: Chris Wilson commit 1a5d5b9ae5771d81da16222582acedb1557386c8 Author: Chris Wilson Date: Sun Oct 7 22:41:02 2012 +0100 sna: Check that for batch overflows after advancing a BLT Signed-off-by: Chris Wilson commit b41ecda69f2e6c9f8adf79dd24ded45740bcc015 Author: Chris Wilson Date: Sun Oct 7 22:39:36 2012 +0100 sna: Only reduce the maximum batch size for old kernels Be careful we not increase the batch size to span multiple pages on 865g! Signed-off-by: Chris Wilson commit 3ca7f8be5bcfec3c50c43bd485f0c57b931f8206 Author: Chris Wilson Date: Sun Oct 7 22:39:10 2012 +0100 sna/gen2: Count the number of dwords required for the invariant Signed-off-by: Chris Wilson commit 4d2dba0ba70c2b217a777243e8a89f7a5c5349a9 Author: Chris Wilson Date: Sun Oct 7 10:46:19 2012 +0100 Use path for intel-gen4asm derived from pkg-config As we use pkg-config to determine whether to use intel-gen4asm, we should also use it to locate the right version of intel-gen4asm to use. This allows the user to install the assembler in a non-standard path for cross-builds and similar. Reported-by: Josh Tripplet Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55646 Signed-off-by: Chris Wilson commit 249051d90fb0ab0bea5561a04ee2a5d0a2238ff1 Author: Chris Wilson Date: Sun Oct 7 09:33:06 2012 +0100 sna/gen2: And the other compile failure Typing on Sunday before coffee is a very bad idea. Signed-off-by: Chris Wilson commit d11a15f04dfaf8ba236d7255c52e67d8602ab806 Author: Chris Wilson Date: Sun Oct 7 09:30:38 2012 +0100 sna/gen2: Compile fix Be careful when cutting and pasting assertions! Signed-off-by: Chris Wilson commit d2a26adc8e7b02aea204101f207f740bbde62414 Author: Chris Wilson Date: Sun Oct 7 08:59:32 2012 +0100 sna/gen2: Add a couple of assertions to track down a batch overflow References: https://bugs.freedesktop.org/show_bug.cgi?id=55700 Signed-off-by: Chris Wilson commit 3680aa4976407886eb4be9878d5296d5a1fadccf Author: Chris Wilson Date: Fri Oct 5 12:33:32 2012 +0100 sna: Do not create an unsnooped CPU mapping for readbacks Otherwise we notice that we have a CPU mmap during read synchronized and presume that we need not take any further action. However... Signed-off-by: Chris Wilson commit d717a67ff4a59d0c2cc108cedbecdc20b4c9fb97 Author: Chris Wilson Date: Fri Oct 5 10:27:26 2012 +0100 configure: Make rebuilding of the gen4 assembly files explicit References: https://bugs.freedesktop.org/show_bug.cgi?id=55646 Signed-off-by: Chris Wilson commit 5b74a2dc5270094810ec0cf8039fd3b3ab0815b5 Author: Josh Triplett Date: Thu Oct 4 17:48:24 2012 -0700 Fix out-of-tree builds with intel-gen4asm installed The make rules to compile shaders with intel-gen4asm referenced the .g4a source files without using $(srcdir), which broke out-of-tree builds. Reference .g4a source files via $(srcdir), and add $(srcdir) to m4's include path, fixing out-of-tree builds. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55645 Signed-off-by: Josh Triplett commit 07a46333477898cfc2020ba5da9a6a7a866b21a7 Author: Chris Wilson Date: Fri Oct 5 10:17:45 2012 +0100 sna: Refactor the fallback CopyArea to use the common region computation Signed-off-by: Chris Wilson commit 3ad3a2d839f8e2f5b02f883e2b4131d4c0901fb0 Author: Chris Wilson Date: Thu Oct 4 15:14:08 2012 +0100 sna/gen2: Tidy DBG code for disabling composite operations Signed-off-by: Chris Wilson commit 6054f8e1120d499f70e7bd4809b3fc6994792bcb Author: Chris Wilson Date: Thu Oct 4 14:40:59 2012 +0100 sna: Remove another sna_pixmap->cpu assertion that was not true Signed-off-by: Chris Wilson commit acf354d7fb575c468e86a5935cad095015ccadb1 Author: Chris Wilson Date: Thu Oct 4 14:35:14 2012 +0100 sna: Propagate failures from compositing glyph masks So that we can fallback correctly. This is primarily using for debugging failure paths... Signed-off-by: Chris Wilson commit c7ad655359b746917deeac3efb9eb843465cf4b9 Author: Chris Wilson Date: Thu Oct 4 13:52:55 2012 +0100 sna/gen2: Prevent using the GTT maps with I915_TILING_Y on 855gm Signed-off-by: Chris Wilson commit 0a2c5eb7669c249194b41d624ca05654a8d7035b Author: Chris Wilson Date: Thu Oct 4 13:51:23 2012 +0100 sna: Once again look into assertions around sna_pixmap->cpu Revert back to basics, and clear the CPU flag everytime we use the GPU, rather than try to avoid clearing it along some paths. Signed-off-by: Chris Wilson commit 4d3b849c57d78f0447334c5c0871d8bd78e0af24 Author: Chris Wilson Date: Thu Oct 4 12:14:47 2012 +0100 sna: Avoid using the gpu for uploads whilst wedged Signed-off-by: Chris Wilson commit 4608e482615d4b4599608c7bc17ac8a9f9293840 Author: Chris Wilson Date: Thu Oct 4 11:23:36 2012 +0100 sna/gen2: Allow fine damage tracking for render operations Signed-off-by: Chris Wilson commit 83b8669abc7415202f9e0c764de675ffbcf45dac Author: Chris Wilson Date: Wed Oct 3 23:45:28 2012 +0100 sna/gen2: Setup invariant blend arguments I thought these were completely specified via the LOAD_STATE_IMMEDIATE commands we used whilst seting up the render pipeline. I was wrong. Reported-by: Timo Kamph References: https://bugs.freedesktop.org/show_bug.cgi?id=55455 Signed-off-by: Chris Wilson commit b71af5dc626adfdf7d753e8e61fb13e8b9290d17 Author: Chris Wilson Date: Wed Oct 3 22:53:57 2012 +0100 sna: Clear sna_pixmap->cpu when destroying the CPU damage So that we keep the assertion checks that without CPU damage we can not be on the cpu. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55591 Signed-off-by: Chris Wilson commit 0d42872d1e0bab1808c0a544c7ece049da103825 Author: Chris Wilson Date: Wed Oct 3 14:42:19 2012 +0100 sna: Protect against undefined O_CLOEXEC Reported-by: Giovanni Mariani Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55577 Signed-off-by: Chris Wilson commit a683643ebf9120bd73a7cd740b5fb24f7e8e797b Author: Chris Wilson Date: Wed Oct 3 14:13:13 2012 +0100 sna: Fix an instance of undamaged tracking Signed-off-by: Chris Wilson commit 47c76bb9680eee89d1cef86058782a7073ca4a47 Author: Chris Wilson Date: Wed Oct 3 14:12:50 2012 +0100 sna/gen7: Tidy detection of unbounded render op Signed-off-by: Chris Wilson commit 3e770f09dbe76287293fea0a8a1f231885aa1c1f Author: Chris Wilson Date: Wed Oct 3 13:00:25 2012 +0100 sna/dri: Delay the deactivation of the pageflipping As we need to clflush the scanout buffer as we return it to the bo cache on SNB+, it is costly to terminate the pageflipping as soon as we drop a frame as mesa often fails to keep up to the vrefresh rate. Signed-off-by: Chris Wilson commit b3a502a7057649b284601bdd5110b136ab1f5fbb Author: Chris Wilson Date: Wed Oct 3 11:46:57 2012 +0100 sna: Review sna_pixmap->cpu tracking Notably 52b211cb15b3 was triggering a few assertions. Signed-off-by: Chris Wilson commit bf44d9965b5cc0f0ee31ba2579783d8194fff756 Author: Chris Wilson Date: Tue Oct 2 12:22:51 2012 +0100 sna: Prevent sna_damage_contains_box__no_reduce() from a NULL deref Signed-off-by: Chris Wilson commit 0ae6c2ccb006a99e8d7bbfff77f98cc58f62ce2f Author: Chris Wilson Date: Tue Oct 2 11:44:16 2012 +0100 sna: Actually upload the damage when replacing a busy CPU bo for PutImage Reported-by: Roman Jarosz Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55508 Signed-off-by: Chris Wilson commit 52b211cb15b3bfe33a61cafcd8dcc1c9fbf20243 Author: Chris Wilson Date: Mon Oct 1 23:55:37 2012 +0100 sna: Tweak CPU damage partial upload paths Signed-off-by: Chris Wilson commit 1fc3fa699c0b4cae5259d9e439f84ede4a6012b2 Author: Chris Wilson Date: Mon Oct 1 22:42:26 2012 +0100 sna: Avoid accumulating damage to the CPU if it is already damaged Signed-off-by: Chris Wilson commit 8c2821d77c4b3bdd4d40d4a2c69f935a9eb0fd47 Author: Chris Wilson Date: Tue Oct 2 11:31:40 2012 +0100 sna: Remember to call ValidatePicture() before using newly created Pictures Reported-by: Marco De Michele Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55527 Signed-off-by: Chris Wilson commit 51202798862c79ad417936cfe2d1847b1da4ab87 Author: Chris Wilson Date: Mon Oct 1 13:06:42 2012 +0100 uxa: Remove USE_GLAMOR check from unrealize glyph caches In an interesting twist during teardown, the driver info structure is freed long before the caches. However, the glyph cache teardown was checking to see if glamor was enabled to see if it could skip the teardown. In fact, since we won't have created the glyph caches in that circumstances it was safe any way. Reported-by: Maarten Lankhorst Signed-off-by: Chris Wilson commit ba1dea5e373b8d7b410c07df897acc4d11079f63 Author: Chris Wilson Date: Mon Oct 1 11:19:41 2012 +0100 sna/gen2: Clear STIPPLE setup before rendering with the 3D pipeline One over-zealous removal too many. Reported-by: Timo Kamph Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55455 Signed-off-by: Chris Wilson commit eec12faf82370a8540114287c224953aa66d7bd9 Author: Chris Wilson Date: Sun Sep 30 18:39:06 2012 +0100 sna/dri: Check back/front buffers of a pixmap for format change Highly unlikely. Signed-off-by: Chris Wilson commit 8e23251aad6d56a59e626b623e59bafac3143562 Author: Chris Wilson Date: Sun Sep 30 13:52:34 2012 +0100 sna: Cache repeated use of the temporary CPU mapping for tiled_x uploads Fixes ShmPutImage regression under gnome-shell Signed-off-by: Chris Wilson commit 1756cf4d8e80cea47a28786c51cac4a760d6d0cf Author: Chris Wilson Date: Sun Sep 30 12:26:23 2012 +0100 sna/dri: Only chain to a triple buffer pageflip sequence Signed-off-by: Chris Wilson commit ba525463fb91900e775c91a54368a9e8b270268b Author: Chris Wilson Date: Sun Sep 30 09:03:36 2012 +0100 sna/dri: Improve handling of async/sync pageflip transition Signed-off-by: Chris Wilson commit 2eab5ef751d7e9caabd7df171cdbb5e784059998 Author: Chris Wilson Date: Sun Sep 30 08:49:34 2012 +0100 sna: Make sure the CPU bo is destroyed if we fail to mmap it during alloc Signed-off-by: Chris Wilson commit 45b351412b8ba81f4fef2f64540d87cbf16359fd Author: Chris Wilson Date: Sat Sep 29 21:21:18 2012 +0100 sna: set_cacheing returns a bool not an error-code Signed-off-by: Chris Wilson commit 33f6f753788398efdf20cde4aa0b4ca402eee8fe Author: Chris Wilson Date: Sat Sep 29 16:41:00 2012 +0100 sna/dri: Swallow error from GetMSC Just ignore the error return from WaitVBlank and make up a value for last MSC/SBC similar to an unattached Drawable. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55395 Signed-off-by: Chris Wilson commit a873f6d354ebbc52265e686b614027f537a6363a Author: Chris Wilson Date: Sat Sep 29 16:01:58 2012 +0100 sna: Release the front buffer and drop DRM master early during shutdown If we delay the dropping of DRM master we may complete a VT switch before we are ready and so prevent a second X server from reacquiring master, and dying. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55446 Signed-off-by: Chris Wilson commit a192f8cbc238d84e058b01b7682abe7ef09d3508 Author: Chris Wilson Date: Sat Sep 29 10:51:00 2012 +0100 2.20.9 release Signed-off-by: Chris Wilson commit 16860d448f11ba405050bd1ff02dc63a4b26280c Author: Chris Wilson Date: Sat Sep 29 10:40:51 2012 +0100 uxa: Only destroy the bufmgr if it exists Be wary in case the Screen is destroy before we initialise bufmgr. Signed-off-by: Chris Wilson commit 13f7a9e267fbefd3326569e89b840bef31d77996 Author: Chris Wilson Date: Fri Sep 28 09:25:12 2012 +0100 sna: Restore LLC cache level after using a buffer for scanout (uncached) Signed-off-by: Chris Wilson commit 8bfd31e9bb13bcb7f12e4147bec0da87b8e87dde Author: Chris Wilson Date: Thu Sep 27 18:12:11 2012 +0100 sna/dri: Improve handling of flipping with no outputs Signed-off-by: Chris Wilson commit dddb6e4e63d5cc8b2a1f2ae3ff287922d30460b9 Author: Chris Wilson Date: Thu Sep 27 17:17:30 2012 +0100 sna: Attempt to restore the current mode if pipe is disabled on DPMS on If we attempt to power up the pipe through a DPMS request, but it was previously disabled due to an error, first try re-enabling. Signed-off-by: Chris Wilson commit 88cfd23379950d1fe4e682519c48ef6e3091c2f3 Author: Chris Wilson Date: Thu Sep 27 16:16:39 2012 +0100 uxa: Update mode->fb_id after completing pageflips As intel_crtc_on() depends upon the current value for determining if the pipe is active, we can only change the value afterwards. Signed-off-by: Chris Wilson commit 83990c649cbd3cb563d08f04e1c02232c44997aa Author: Chris Wilson Date: Thu Sep 27 14:58:26 2012 +0100 Remove the last couple of references to mibstore Signed-off-by: Chris Wilson commit cb0b72c4fbb7b1bc941522d8679fef856f2f8936 Author: Chris Wilson Date: Thu Sep 27 14:53:46 2012 +0100 uxa: Fixup the last couple of commits after realising I didn't have uxa enabled Signed-off-by: Chris Wilson commit feef53005b57d6b4489a15f8d0db8245dcb0cfaa Author: Chris Wilson Date: Thu Sep 27 13:36:55 2012 +0100 uxa: Do not attempt to pageflip whilst not master Signed-off-by: Chris Wilson commit 5a45cbacb777e478d8fbda9223b0fb5c705d7249 Author: Chris Wilson Date: Thu Sep 27 13:31:27 2012 +0100 uxa: Confirm the pipe is alive before flipping Signed-off-by: Chris Wilson commit edc5b7f741a4bb8e9a58b5eb2201fa9304f94ef7 Author: Chris Wilson Date: Thu Sep 27 13:27:20 2012 +0100 uxa/dri: Confirm Drawable is still attached to an active pipe before flipping If we schedule a flip to happen on a far distant vblank, it is possible for us to disable the pipes with a modeset, whilst maintaining the fb size, or with a DPMS command. If the pipe is disabled and we still try to flip, then we trigger an EINVAL error from the kernel. Signed-off-by: Chris Wilson commit eb667378543f97d7c7e7767daddbd9b39e07c1f8 Author: Chris Wilson Date: Wed Sep 26 15:12:57 2012 +0100 sna: Typo in DBG Signed-off-by: Chris Wilson commit 557ad78d02dfff3db9e7fc21f05230a8fe482f7c Author: Chris Wilson Date: Wed Sep 26 15:02:17 2012 +0100 Kill mibstore pretend backwards compatibility Adam Jackson was kind enough to point out that mibstore had been a no-op since xorg-1.4 and the earliest we claim to support is xorg-1.6. Signed-off-by: Chris Wilson commit 367f57faf2c422da026a25c099426dc750cf3759 Author: Chris Wilson Date: Wed Sep 26 14:06:18 2012 +0100 sna/video: Force integer promotion to avoid u16 overflow for frame size Signed-off-by: Chris Wilson commit 34772cb0e724a880a8c6db936c0bb5dd129398f2 Author: Chris Wilson Date: Wed Sep 26 12:09:39 2012 +0100 sna/dri: Handle pending flips when drawable disappears. The async case needs special handling as the current front buffer may be pending a flip onto the scanout which needs to be completed even if the Drawable is destroyed. Fixes regression from commit ad877abdc70b842afd202b2f97892d97a0b8d151 Author: Chris Wilson Date: Sat Jul 7 17:48:52 2012 +0100 sna/dri: Attach the chain of frame events to the Window, not pixmap Signed-off-by: Chris Wilson commit addfa653cf51fa45c7eae49be1fd722aaadf97d6 Author: Chris Wilson Date: Wed Sep 26 08:59:18 2012 +0100 uxa: Prevent attempting to call intel_mode_fini() before PreInit runs In the Ubuntu xserver it is apparently possible for the Screen to be deleted prior to being fully initialised. Make sure we don't crash in that situation! Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55346 Signed-off-by: Chris Wilson commit b880d8e0a9a32acd50333f72d695268c4819c29e Author: Chris Wilson Date: Tue Sep 25 16:59:20 2012 +0100 sna/gen5: Add WA reference Signed-off-by: Chris Wilson commit 74f930fd80c3f97a1b6213e9e79e02f8f51c64b9 Author: Chris Wilson Date: Tue Sep 25 15:41:51 2012 +0100 sna: Make sure the batch is flushed before setting DPMS Just in case on the off-chance that the kernel modifies the pipe and we had a pipe-dependent operation in the batch. Signed-off-by: Chris Wilson commit 0a00c3765c3ba464bb4e9cde12e964c979421ed6 Author: Chris Wilson Date: Tue Sep 25 13:50:33 2012 +0100 sna: Accommodate changes to the Damage extension not unwrapping the GC commit 4dc2a76740d921c824a4d8193f39dd373475f02a Author: Adam Jackson Date: Thu Sep 20 17:57:00 2012 -0400 miext/damage: Only wrap into the GC ops chain if there's a listener (v3) leaves the GC->funcs still pointing to the damage layer as it calls into the driver functions, so we need to be careful to restore the correct function table after we manipulate the GC for fallbacks. Signed-off-by: Chris Wilson commit 8f5a6d107bb30fa218ff3c1d07532ca834b82edc Author: Chris Wilson Date: Tue Sep 25 13:36:12 2012 +0100 Add miInitializeBackingStore() to compat-api.h Signed-off-by: Chris Wilson commit dee133a722fb137767a78b244a312620f52a51b1 Author: Chris Wilson Date: Tue Sep 25 13:22:40 2012 +0100 sna: Remove unused reference to mibstore.h, removed in future Xorg Signed-off-by: Chris Wilson commit f0dbfc20aaa9b47329438ce42150770d4d7934c3 Author: Chris Wilson Date: Tue Sep 25 13:22:21 2012 +0100 sna/dri: Fixup a DBG statement not to print gibberish Signed-off-by: Chris Wilson commit c6008068372709c73034163eddc902b47bf87d24 Author: Chris Wilson Date: Sun Sep 23 21:42:31 2012 +0100 sna: Check against op->dst.bo rather than priv->cpu_bo for composite upload In this case, we may also be handling an unattached pixmap, so avoid the deferences of the sna_pixmap unless we are sure it will exist. Signed-off-by: Chris Wilson commit 5ed840881c26e90eb8e00521b6d77b0ea514de5e Author: Chris Wilson Date: Sun Sep 23 20:29:21 2012 +0100 sna: Check that the CPU bo exists before declaring it is busy along upload Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55251 Signed-off-by: Chris Wilson commit a858afc66c1fa2eec65a7041e991f2266f82deca Author: Chris Wilson Date: Sun Sep 23 13:49:40 2012 +0100 Silence a couple of potential compiler warnings Signed-off-by: Chris Wilson commit 9326acc2917109f06dda809107c8fa5a2273c3d2 Author: Chris Wilson Date: Sun Sep 23 13:40:04 2012 +0100 Allow compilation of a separate i810 driver Allow --enable-ums-only as a counter-option to --enable-kms-only in case the distribution wishes to enable a non-root KMS driver but also offer a separate UMS driver for i81x. On the second pass, use "--enable-ums-only --disable-uxa --disable-sna" to get the trimmed down unaccelerated i810 support. Signed-off-by: Chris Wilson commit 0afb7efe8c48b5fc839e2137e870bea0f5fb3c9c Author: Chris Wilson Date: Sun Sep 23 09:41:59 2012 +0100 sna: Avoid overflows when translating the box16 extents during a copy Signed-off-by: Chris Wilson commit 0fc6e5820e4543d52bcf8d0285ca6f69b5213831 Author: Chris Wilson Date: Sun Sep 23 09:32:49 2012 +0100 sna: Add missing protection against int16 overflow when copying Signed-off-by: Chris Wilson commit c2ca1c5abca687adb08d6e137bc3b70bdecb083d Author: Chris Wilson Date: Sun Sep 23 08:56:50 2012 +0100 Reorder INTEL_XVMC define so that we pick up Option "XvMC" Otherwise it will remain disabled. Reported-by: Giacomo Comes Signed-off-by: Chris Wilson commit 92dbedc6138b923aa473935013ecb0346280c4d3 Author: Chris Wilson Date: Sat Sep 22 08:56:49 2012 +0100 sna: Force the stall before trying to upload into a busy CPU bo Under the circumstances where we can not instead stream the write into the GPU bo, we need to use the busy CPU bo. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54978 Signed-off-by: Chris Wilson commit 3e1be265cf950976b5929b14a9dad0664deaa2c1 Author: Chris Wilson Date: Fri Sep 21 17:37:58 2012 +0100 sna: Force an inplace upload if already wedged Signed-off-by: Chris Wilson commit b8967aff382c1b6bef2335dea51c979a3f0800c7 Author: Chris Wilson Date: Fri Sep 21 12:16:27 2012 +0100 sna: Do not query for the NULL edid property If the EDID blob id is set to 0, that means that it does not exist and so we can safely skip it. References: https://bugs.freedesktop.org/show_bug.cgi?id=55193 Signed-off-by: Chris Wilson commit 0be1d964713ca407f029278a8256d02d925dc9da Author: Chris Wilson Date: Tue Sep 11 21:48:24 2012 +0100 sna: Use inplace X tiling for LLC uploads Based on a suggestion by Chad Versace (taken from a patch for mesa). This allows for a faster upload of pixel data through a ShmImage, or for complete replacement of a GPU bo. Using a modified version of x11perf to upload to a pixmap rather than scanout on an IVB i7-3720qm: Before: 40000000 trep @ 0.0007 msec (1410000.0/sec): ShmPutImage 10x10 square 4000000 trep @ 0.0110 msec ( 90700.0/sec): ShmPutImage 100x100 square 160000 trep @ 0.1689 msec ( 5920.0/sec): ShmPutImage 500x500 square After: 40000000 trep @ 0.0007 msec (1450000.0/sec): ShmPutImage 10x10 square 6000000 trep @ 0.0061 msec ( 164000.0/sec): ShmPutImage 100x100 square 400000 trep @ 0.1126 msec ( 8880.0/sec): ShmPutImage 500x500 square However, the real takeaway from this is that the overheads for ShmPutImage are substantial, only hitting around 70% expected efficiency, and overshadowed by PutImage, which for reference is 60000000 trep @ 0.0006 msec (1800000.0/sec): PutImage 10x10 square Signed-off-by: Chris Wilson commit d853064e7eebc5719645c12605782f995131a6fe Author: Chris Wilson Date: Thu Sep 20 22:43:26 2012 +0100 sna/gen3+: Trim the target extents to the CompositeClip When computing the active region with of a composite operation with unknown extents we try to simply use the whole Drawable. However, this needs to be clipped otherwise it may trigger assertion failure with an offscreen pixmap. References: https://bugs.freedesktop.org/show_bug.cgi?id=55164 Signed-off-by: Chris Wilson commit 853beff4306d5a168e310af82d0ddf9db319fdce Author: Chris Wilson Date: Wed Sep 19 08:30:59 2012 +0100 sna/dri:Add some DBG and assertions to validate names passed back to clients Signed-off-by: Chris Wilson commit 8c1e60a15431b193b57d94a27b8e8de7c5a7e683 Author: Chris Wilson Date: Tue Sep 18 15:01:52 2012 +0100 sna: Add some asserts useful for tracking an xserver drawing bug Signed-off-by: Chris Wilson commit cd23ac69ffb670468b2993242ce0d596081bbca4 Author: Chris Wilson Date: Tue Sep 18 13:39:33 2012 +0100 sna/gen7: Add some ring switching sanity checks Signed-off-by: Chris Wilson commit 22c84d02ade83c39940d76fc616a9122e535606a Author: Chris Wilson Date: Tue Sep 18 12:01:49 2012 +0100 sna: Remove special casing of drawing depth=1 glyphs It turns out that they were not so special after all, and the "fast path" was missing the important handling such as validating the GC. As a result we ended up trying to render the glyphs through a planemask of 0 and similar garbage resulting in corruption. An example given by Kaus Dittrich was the use of rotated glyphs in gnuplot. Reported-by: Klaus Dittrich Signed-off-by: Chris Wilson commit 71fa350325bb06a6141f89ef14431d96f0c4956a Author: Chris Wilson Date: Mon Sep 17 13:40:24 2012 +0100 sna: Check that we do not keep a GPU bo around after flushing a ShmPixmap Signed-off-by: Chris Wilson commit 88502b32d5d2b826fc08344392c98300341410e1 Author: Chris Wilson Date: Sun Sep 16 09:02:46 2012 +0100 2.20.8 release Signed-off-by: Chris Wilson commit 395d738b5086f24659b56b4b3536bffaebad8a3e Author: Chris Wilson Date: Sat Sep 15 21:18:35 2012 +0100 sna: Improve handling of probe failure during initialisation Signed-off-by: Chris Wilson commit fe78904a45979d47bb092ce37d321854e2418344 Author: Chris Wilson Date: Thu Sep 13 22:15:14 2012 +0100 Add basic support for ValleyView Bind to the ValleyView SDV for verifying the render routines. Signed-off-by: Chris Wilson commit 3b9259f7cedc179617bc24b3912d2d8d75c5a824 Author: Ian Stakenvicius Date: Fri Sep 14 21:04:42 2012 +0100 configure: Make udev dependency optional In order to support buildbots where the udev headers may exist on the build system but not the target, we need explicit control over optional dependencies. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54942 commit d995705fb01842652a79076cbecee4392f653bfe Author: Chris Wilson Date: Fri Sep 14 19:32:06 2012 +0100 sna: Only discard CPU damage if we completely overwrite its extents If we are performing a clipped copy, then we must be careful not to completely discard the CPU damage as it may not be entirely replaced. Reported-by: Roman Jarosz Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54937 Signed-off-by: Chris Wilson commit deacab87e4ba7fe09f0f16568dc2848949f2707a Author: Chris Wilson Date: Fri Sep 14 17:01:13 2012 +0100 sna/dri: flip->next_front.bo is owned by the DRI drawable not by the flip Be careful not to delete the reference we presume we hold as it is borrowed from the DRI drawable. Signed-off-by: Chris Wilson commit 5369408d596a0611d4f76333dac2e1c93e63e8b6 Author: Chris Wilson Date: Fri Sep 14 16:12:00 2012 +0100 sna/dri: And the hunt for the use-after-free continues... Signed-off-by: Chris Wilson commit 8e9d64203428b344433ac8f86dc1f06d2c32529a Author: Chris Wilson Date: Fri Sep 14 14:06:38 2012 +0100 sna: Fix typo for sna_render_picture_flatten() Pass it the correct value for the pixmap height, and not its 'y' coordinate! Reported-by: Eugene Rosenzweig Bugzilla: https://bugs.gentoo.org/show_bug.cgi?id=434860 Signed-off-by: Chris Wilson commit 6ee9969f0e99975a24e113af9591909d9023d8af Author: Chris Wilson Date: Fri Sep 14 13:04:50 2012 +0100 sna/dri: Update AsyncSwap for recent changes in tracking DRI drawables Signed-off-by: Chris Wilson commit d2dbb991384f2b820a714eba262691ed97a6a22a Author: Chris Wilson Date: Fri Sep 14 13:04:14 2012 +0100 sna/dri: Add a bunch of assertions for hunting a use-after-free Signed-off-by: Chris Wilson commit d8756091474aeb44e5d8eed6b20b65c23b6574e7 Author: Chris Wilson Date: Fri Sep 14 12:17:28 2012 +0100 sna/dri: Version bump for prototypical AsyncSwap Signed-off-by: Chris Wilson commit 0fc3d020f3086b5c6ecef1fed1aaedef7aa52b01 Author: Chris Wilson Date: Fri Sep 14 09:49:00 2012 +0100 Skip adding unknown special modes In light of the discussion on how to add stereo modes it was revealed that we would happily add modes with unknown and unhandled strange flags. Adam Jackson mentioned that he has plans to fix that with some upcoming work, but as a first step we can simply eradicate them whilst pondering how to support the stereoscopic vision of the future. Signed-off-by: Chris Wilson commit 35d81d97b38d05f132a4823e57fff4012e966504 Author: Chris Wilson Date: Thu Sep 13 22:32:47 2012 +0100 sna/gen7: Defensively program default GT values Signed-off-by: Chris Wilson commit 3fbdedbf9d509c5ca58ae0a01fe8e54dcc990cf8 Author: Chris Wilson Date: Thu Sep 13 22:10:08 2012 +0100 sna: Fix analysis of source extents for BLT composite After we have computed the source offset vector for the transformed source bo, we need to use that with respect to the destination rectangle to verify that the source sample is wholly within bounds. Signed-off-by: Chris Wilson commit 6c7f998f7bda4f4f1286e079412d5d3a5b2eb073 Author: Chris Wilson Date: Thu Sep 13 20:41:23 2012 +0100 sna: Fix BLT composite offset Reported-by: Jiri Slaby Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54868 Signed-off-by: Chris Wilson commit 2575cd0d236b4e1694e3185a487ebfd8bfe6499f Author: Chris Wilson Date: Thu Sep 13 20:08:37 2012 +0100 sna: Avoid corrupting the CPU bo when compositing fallback data through BLT Signed-off-by: Chris Wilson commit 5781de2e5ab0d7e8e62965bda714789879555ed4 Author: Chris Wilson Date: Thu Sep 13 19:53:29 2012 +0100 sna/gen3: Tidy NO_COMPOSITE debugging Signed-off-by: Chris Wilson commit dd1432b2c00c94ac75af4740b66b5cd1a573e261 Author: Chris Wilson Date: Thu Sep 13 19:47:29 2012 +0100 sna: Do not attempt to change tiling if wedged The caller will just have to live with the current tiling, which should be fine in most cases, in other cases the gpu is wedged... Signed-off-by: Chris Wilson commit 8c7dd2219fa777bf6354c4e0ef38a2f09fe09675 Author: Chris Wilson Date: Thu Sep 13 17:14:23 2012 +0100 sna: Prefer to use indirect uploads for very small updates Signed-off-by: Chris Wilson commit d87a56ed1789e0c5058b302bb930d7e952ff3e5e Author: Chris Wilson Date: Thu Sep 13 09:57:30 2012 +0100 sna: Teach sna_replace__xor() how to tile large uploads This path is hit using eog+cairo-1.10 and a large image, e.g. http://marsrovers.jpl.nasa.gov/gallery/press/opportunity/20120705a/PIA15689_Greeley_Pan_wDeck_L257F.jpg Reported-by: Michael Laß Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54808 Signed-off-by: Chris Wilson commit 58a96f0f684fe0d7d1a7890c630539ef8b065d1e Author: Chris Wilson Date: Wed Sep 12 23:53:44 2012 +0100 sna: Also remove bogus assertion from sna_replace__xor() Like the assertion in sna_replace() this is now incorrect due to the automatic flagging of large bo with flush. Signed-off-by: Chris Wilson commit 6069e78238ca57e3433d528680fe04766a1e5307 Author: Chris Wilson Date: Wed Sep 12 16:22:49 2012 +0100 sna: Keep a very small, short-lived cache of large buffers As we now regularly retire and so discard the temporary large buffers, we find them in short supply and ourselves wasting lots of time creating and destroying the transient buffers. Signed-off-by: Chris Wilson commit 1e2a03ad19664e846ece4110e53d26babb5a2159 Author: Chris Wilson Date: Wed Sep 12 16:15:22 2012 +0100 sna: Propagate busyness when creating the proxy Signed-off-by: Chris Wilson commit e28f5a2537821dc170a7ef78bf33ecbd3c032da0 Author: Chris Wilson Date: Wed Sep 12 15:21:27 2012 +0100 sna: Add some DBG as to why kgem_check_bo() flushes Signed-off-by: Chris Wilson commit 1af1dae1feefd6567017170bab241570fb528e5c Author: Chris Wilson Date: Wed Sep 12 15:20:49 2012 +0100 sna: Ensure tiling upload buffers are trimmed to fit in the cache Signed-off-by: Chris Wilson commit e5f137807c318588f546960668345eef34159e26 Author: Chris Wilson Date: Wed Sep 12 14:11:43 2012 +0100 sna: Avoid fallbacks to shadow pixels if FORCE_GPU is in effect References: https://bugs.freedesktop.org/show_bug.cgi?id=54808 Signed-off-by: Chris Wilson commit 0d17208a66a7e54d4106f8a4034d3a928e28bb62 Author: Chris Wilson Date: Wed Sep 12 14:04:50 2012 +0100 sna: Avoid readback inplace if the target is unmappable We have to use the tiling indirect path, or else fail. References: https://bugs.freedesktop.org/show_bug.cgi?id=54808 Signed-off-by: Chris Wilson commit 4b4abdaae94d164d5d0b2755907e76b9cbe0c988 Author: Chris Wilson Date: Wed Sep 12 13:47:26 2012 +0100 sna: Flush after operating on large buffers As we know that such operations are likely to be slow and consume precious GTT space, mark them as candidates for flushing. Signed-off-by: Chris Wilson commit 15911f533d6a7ef40e42ba5921fac7c62b290f8b Author: Chris Wilson Date: Tue Sep 11 14:59:53 2012 +0100 prime: Reorder probe messages so that the driver prefix is correct Signed-off-by: Chris Wilson commit bc73acbd4da53bc50752c0413adcd0ce876e0a03 Author: Chris Wilson Date: Mon Sep 10 18:55:04 2012 +0100 uxa: Error out after failing to become DRM master during PreInit As we set the card fd to -1 upon failure, the error here is non-recoverable, so make sure it gets propagated and we abort loading the intel ddx. Bugzilla: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/962892 Signed-off-by: Chris Wilson commit b0d14071f7b60729c223af925935227393fbd3ee Author: Chris Wilson Date: Mon Sep 10 13:53:45 2012 +0100 sna: Workaround issue with global glyph privates and shared ZaphodHeads Under ZaphodHeads we end up with multple screens accessing the common sna_glyph_key and so cause conflicting updates and erroneous references into the screen-local texture atlases. Two approaches can be tried here. Transition to a screen-specific private key introduced with xorg-1.13, or to move the glyph cache (and the rest of the gpu state tracker) down into the device private rather than screen private. This is neither of those, but a workaround to avoid reusing the incorrect entries from shared screens. Reported-by: Stephen Liang Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54707 Signed-off-by: Chris Wilson commit cfa0c6162bbd2f6edab1301b6aed886af7e8ab38 Author: Chris Wilson Date: Mon Sep 10 10:50:03 2012 +0100 sna: Fixup possible_crtcs for ZaphodHeads As the possible_crtcs is a bitmask of the available crtcs exposed to the Xserver, we need to adjust it for the limited view given by Zaphod mode. Signed-off-by: Chris Wilson commit 95c71d1626dfbd97733e7e7e03fdde3af833446e Author: Chris Wilson Date: Sat Sep 8 20:12:32 2012 +0100 NEWS: Fix version reference 'bugs since 2.20.6'! Signed-off-by: Chris Wilson commit 0659711dfa9c9fc52adeaba766049887d5b6f040 Author: Chris Wilson Date: Sat Sep 8 20:01:32 2012 +0100 2.20.7 release Signed-off-by: Chris Wilson commit 2c8e48f8149499040342836491333402fd05b762 Author: Chris Wilson Date: Fri Sep 7 23:00:47 2012 +0100 sna: Remember that LineDoubleDash involves updating the GC between segments Signed-off-by: Chris Wilson commit ed7b8db24921dc0bb6ea59dacf35ea41a61d59bf Author: Chris Wilson Date: Fri Sep 7 22:12:17 2012 +0100 sna: Pass the hints down to the PolyPoint fallback Signed-off-by: Chris Wilson commit 6c9df48fa8f1ac93103ce771ae28f47faa321619 Author: Chris Wilson Date: Fri Sep 7 22:05:19 2012 +0100 sna: Fix PolyPoint fallback through a complex clip Signed-off-by: Chris Wilson commit 7d76eb155ed78343abfa15b7c8af87fc3d4211e4 Author: Chris Wilson Date: Fri Sep 7 18:06:25 2012 +0100 sna: First scan for a backlight associated with the device The goal is to find the right interface in a multi-GPU system. Signed-off-by: Chris Wilson commit 2de7d2d8150d648815feb74a3ae2239b908b971e Author: Chris Wilson Date: Fri Sep 7 16:56:33 2012 +0100 sna/prime: Fixup attaching the randr14 scanout pixmap Signed-off-by: Chris Wilson commit 448808d8a4e78afa7fe90e318d6d7afdbff66190 Author: Chris Wilson Date: Fri Sep 7 14:31:52 2012 +0100 prime: Allocate a GPU_SCREEN when not primary to fulfil the output sink role Signed-off-by: Chris Wilson commit 3dce661ef3df5558fe9c9d85324ab3395a37b7ea Author: Chris Wilson Date: Fri Sep 7 13:48:12 2012 +0100 sna: And remember to update the ScreenPixmap after resize Fixes the regression from the previous commit Signed-off-by: Chris Wilson commit 0c39c363d7235d03f27337c731c0ffe3369bf713 Author: Chris Wilson Date: Fri Sep 7 13:07:52 2012 +0100 sna: Protect sna_crtc_resize() against early invocation with no RootWindow If the outputs are resized very early on, in CreateScreenResources, then we will not yet have created a RootWindow and so trying to change its pixmap is a futile effort. Signed-off-by: Chris Wilson commit d7879a5939cabcd8b804e19fc422d2022ab7e3a4 Author: Chris Wilson Date: Fri Sep 7 12:18:08 2012 +0100 sna/dri: Fix the double-buffer pageflipping path Notably, we need to remember to exchange the front/back buffers after flipping! Signed-off-by: Chris Wilson commit 2630c81937115602faa352fca369d89fbb926b33 Author: Chris Wilson Date: Fri Sep 7 11:23:02 2012 +0100 sna/dri: Make sure we set the SCANOUT flag when creating bo for pageflips In case we should be creating an untiled framebuffer, we need to make sure we honour the scanout alignment. Signed-off-by: Chris Wilson commit aeef03520f1b2a4b8638ff1ed4fbf19c9b35e7c8 Author: Chris Wilson Date: Thu Sep 6 11:46:53 2012 +0100 sna/prime: Correct the pinning flag when importing the prime bo Signed-off-by: Chris Wilson commit 9bdf46d6c1a29b7b9e3d35e39a2e789334c0e1be Author: Chris Wilson Date: Thu Sep 6 11:38:08 2012 +0100 sna/prime: Distinguish forms of pinned pixmap So that we can allow prime to replace the backing bo of currently exported pixmaps through DRI2. Signed-off-by: Chris Wilson commit 39c4be2b0be66ed83bf059d3007f4143325e340d Author: Chris Wilson Date: Thu Sep 6 11:28:49 2012 +0100 uxa: Distinguish forms of pinning This allows us to replace the prime bo to meet sharing requirements, but still maintain the integrity with other users. Signed-off-by: Chris Wilson commit 031fa1c1c1129e486a02ffb5b3b9071f03b60048 Author: Chris Wilson Date: Thu Sep 6 11:08:23 2012 +0100 uxa/prime: Mark the pixmap as pinned once exported (similar to flink) Signed-off-by: Chris Wilson commit e5d3f3818660cb875ce2ef4721c51bc95317b78d Author: Chris Wilson Date: Thu Sep 6 10:54:57 2012 +0100 sna: Add is-pinned checks to changing the pitch on an existing bo Signed-off-by: Chris Wilson commit f2affe403baea78b9c94e3d726d1b9d8a0004f35 Author: Chris Wilson Date: Thu Sep 6 10:24:04 2012 +0100 sna: Apply the minimum 256 pitch to CREATE_USAGE_SHARED pixmaps as well Signed-off-by: Chris Wilson commit fbe4080816191ee97347128444f56bcc34858481 Author: Chris Wilson Date: Thu Sep 6 08:57:34 2012 +0100 prime: Align pitch of shared untiled buffers to 256 bytes In order for nvidia to handle the buffers we are currently generating, they need to have a pitch alignment of 256 bytes. Make it so. Signed-off-by: Chris Wilson commit c5fb025267bf1019907dfcc60f0efef2691be436 Author: Chris Wilson Date: Thu Sep 6 08:55:46 2012 +0100 sna: Correct assertions for adding damage that bypasses the region Signed-off-by: Chris Wilson commit 1f5d5a37e57e63fa6e5b336a4847ce4422b89044 Author: Chris Wilson Date: Wed Sep 5 17:17:58 2012 +0100 Add PlatformProbe to handle sharing of device entities Reported-by: Nick Bowler Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54561 Signed-off-by: Chris Wilson commit ddde40afc081f47a3b6b694aeb21682c240c9562 Author: Chris Wilson Date: Wed Sep 5 16:48:01 2012 +0100 sna: Harden against initialisation failures References: https://bugs.freedesktop.org/show_bug.cgi?id=54561 Signed-off-by: Chris Wilson commit 4627dc36fa56ab971cd90088bd29a597dd12dc02 Author: Chris Wilson Date: Wed Sep 5 16:17:17 2012 +0100 Revert "sna: Flush the batch before a render operation if the GPU is idle" This reverts commit ad57ac07a273bf376b74884de47d8ee1e7129fb8. These checks end up being too frequent and not allowing us to batch sufficient commands to offset the overhead of batch submission. Hmm. commit 886f3cef4101a674bb62656cf7f3046643cae4b1 Author: Chris Wilson Date: Wed Sep 5 15:27:21 2012 +0100 uxa: Fix cut'n'paste error in Option "Backlight" Signed-off-by: Chris Wilson commit 6211d0914e2ed1cdd4cc2e170ee16129fc9f0d73 Author: Chris Wilson Date: Wed Sep 5 12:33:08 2012 +0100 uxa: Flush batch immediately after queuing a WAIT_FOR_EVENT We still encounter hangs with kernel-3.5 with the culprit being a wait on a disabled pipe. As we thoroughly check before that the pipe is still disabled and flush before a modeset, the only possibility that remains is that DPMS is disabling the pipe before we submit. Close that race by always submitting the batch immediately after a WAIT_FOR_EVENT. Signed-off-by: Chris Wilson commit 6a1423547769c24d15f6475628981d14a76a4bb2 Author: Chris Wilson Date: Wed Sep 5 12:15:52 2012 +0100 sna: Initialise the empty boxes field of the region before subtracting it Fixes regression from commit 38fb77af757318e5fb6f605b37306ce4585b11a5 Author: Chris Wilson Date: Wed Sep 5 08:23:34 2012 +0100 sna: Don't upload ignored cpu damage Signed-off-by: Chris Wilson commit dff25e5ec4071a0404f82760e8deec3f99f4a0a9 Author: Chris Wilson Date: Wed Sep 5 11:05:28 2012 +0100 sna: Drop master after discarding framebuffers As Imre Deak pointed out in the previous patch, drmModeRmFB only works when we hold the DRM master, therefore to prevent a leak of the framebuffer across server reset we need to defer dropping master until after we release our scanouts and modes. Signed-off-by: Chris Wilson commit 46981d01700c1159bfb6bc0aebc938ff1d447a0f Author: Imre Deak Date: Fri Aug 31 14:31:51 2012 +0300 uxa: fix leakage of FB when closing the screen Calling drmModeRmFB is only allowed in DRM master mode. Since leaving the VT also drops master mode we need to remove the FB before calling I830LeaveVT. This is only a real leak in case of a server reset, otherwise the server process will exit anyway and the kernel will clean up the FB. Signed-off-by: Imre Deak Signed-off-by: Chris Wilson commit 55cef974a5dad3fda1922648fa27bcf5bb32ea03 Author: Chris Wilson Date: Wed Sep 5 09:38:47 2012 +0100 sna: Review validity of damage when discarding CPU bo Signed-off-by: Chris Wilson commit 587499bbf55b7eb0e1848822a792d535a8a3db1b Author: Chris Wilson Date: Wed Sep 5 10:56:18 2012 +0100 sna/video: Use the scanout flag and FB id for sprite framebuffers So that we can use the same teardown path as normal scanouts. Signed-off-by: Chris Wilson commit 913adacc5400f94bc754f22375447e6f90ccd510 Author: Chris Wilson Date: Wed Sep 5 09:20:55 2012 +0100 sna/gen7: Always emit a stall when flushing the texture cache Reported-by: Reinhard Karcher References: https://bugs.freedesktop.org/show_bug.cgi?id=54488 Signed-off-by: Chris Wilson commit bdfedb46bd5bb3b96543a3ff553abd8ed0774fad Author: Chris Wilson Date: Wed Sep 5 09:15:12 2012 +0100 sna: Use async upload only if the last render was not CPU Signed-off-by: Chris Wilson commit 67b75ae6f66e2d8e97b0aaf70d66b6de30f9e67e Author: Chris Wilson Date: Wed Sep 5 08:46:39 2012 +0100 sna: Tweak upload path to remove redundant code Signed-off-by: Chris Wilson commit 38fb77af757318e5fb6f605b37306ce4585b11a5 Author: Chris Wilson Date: Wed Sep 5 08:23:34 2012 +0100 sna: Don't upload ignored cpu damage Signed-off-by: Chris Wilson commit 2eb32f711ddd65a55c3c21f813b41aca49635a22 Author: Chris Wilson Date: Wed Sep 5 00:22:04 2012 +0100 sna: Add DBG control to disable CPU bo Signed-off-by: Chris Wilson commit 19e170aa6fd3652cc2e983a291f82350dca27e52 Author: Chris Wilson Date: Tue Sep 4 23:52:42 2012 +0100 sna: Fix comparison of memcpy overlap to include x-offsets Signed-off-by: Chris Wilson commit 7895416eabfc5bd361d0d0ff92c570e2f6af47e0 Author: Chris Wilson Date: Tue Sep 4 20:30:53 2012 +0100 sna: Rearrange use_cpu() tests for composite ops to avoid syncing CPU bo Signed-off-by: Chris Wilson commit ad57ac07a273bf376b74884de47d8ee1e7129fb8 Author: Chris Wilson Date: Tue Sep 4 20:06:49 2012 +0100 sna: Flush the batch before a render operation if the GPU is idle Experiment with pushing those first commands earlier. Signed-off-by: Chris Wilson commit 6325d7e29a8bacede321362c20bf3acd33c48651 Author: Chris Wilson Date: Tue Sep 4 14:51:54 2012 +0100 NEWS: fixup missing links to bugs referenced in 2.20.6 release notes Reported-by: Adam Jackson Signed-off-by: Chris Wilson commit bc309bad39035649ee377c1d5af9099491a8ea43 Author: Chris Wilson Date: Tue Sep 4 14:22:15 2012 +0100 sna: And remember to flush the damage to the output slaves... Signed-off-by: Chris Wilson commit 68d207588a177afa4e999260bfddb4d6dba1029b Author: Chris Wilson Date: Tue Sep 4 13:20:21 2012 +0100 sna: Improve handling of output offloading In particular, don't forget to flush when we only have offload slaves and no native pixmaps. Signed-off-by: Chris Wilson commit 8cf7ac776b9b47dabd5ab141e5a5385c44d3f309 Author: Chris Wilson Date: Tue Sep 4 13:55:34 2012 +0100 sna: Fix a typo in an error message s/achieve/retrieve/ otherwise it is nonsense. Signed-off-by: Chris Wilson commit be83b98d299c74101357e7bf170b1451aec2e1f2 Author: Chris Wilson Date: Tue Sep 4 11:23:58 2012 +0100 sna: Add some DBG to the prime paths Signed-off-by: Chris Wilson commit d0684d55e3a62c231c92a68403cf100c9e18351b Author: Chris Wilson Date: Tue Sep 4 11:05:43 2012 +0100 sna: Fixup CREATE_USAGE_SHARED The DRI2 code tries to create pixmaps with non-zero width/height, whoops. Signed-off-by: Chris Wilson commit b6ced7a2267f9429ad38949dda39c843e2119e99 Author: Chris Wilson Date: Tue Sep 4 10:05:46 2012 +0100 sna: Fix typo for detecting prime ioctls Signed-off-by: Chris Wilson commit df68723baae71498de95924c72d4f23fa7fc7fdf Author: Chris Wilson Date: Mon Sep 3 20:43:48 2012 +0100 sna: Port prime interfacing Preliminary prime support. Signed-off-by: Chris Wilson commit e3ad18036b4dca29744ecfa4694006f01dd3fd18 Author: Chris Wilson Date: Mon Sep 3 20:53:30 2012 +0100 sna: Enable platform probing Completing commit 0768ac4d195214825137152893deb74a87fcd11e Author: Dave Airlie Date: Wed Jul 25 16:11:23 2012 +1000 intel: add platform probing support. Signed-off-by: Chris Wilson commit 3dc644b2a959fc559e1138b332ed42d7235de42f Author: Сковорода Никита Андреевич Date: Mon Sep 3 13:44:21 2012 +0400 uxa: Fix Backlight option support. Signed-off-by: Сковорода Никита Андреевич commit a972affe0c78b23a119d5dd14bb4446c89122af1 Author: Chris Wilson Date: Sun Sep 2 17:43:33 2012 +0100 sna/gen6+: Redirect fills if the destination is too large for 3D Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54134 Signed-off-by: Chris Wilson commit 18d26076c778c20eb589b638fc47fa847793f149 Author: Dave Airlie Date: Tue Sep 4 07:19:12 2012 +1000 intel: fix return value for no pEnt Probably never gets hit but shuold return FALSE, pointed out on irc by Lekensteyn Signed-off-by: Dave Airlie commit d14ff42f2a205542df2ef723c6151d18db2bea8b Author: Dave Airlie Date: Thu Jul 26 10:43:29 2012 +1000 intel: query kernel for caps to setup scrn capabilities. This queries the kernel for prime support before advertising the capabilities. Signed-off-by: Dave Airlie commit 6705d8237aca90964449e4dbee97b4f62b87c28b Author: Dave Airlie Date: Thu Jul 26 10:37:04 2012 +1000 intel: add pixmap tracking and scanout support. (v2) This adds support for pixmap tracking and scanout of alternate pixmaps. v2: do dirty updates after uxa block handler, check if kernel can flush vmap for us so we don't have to. Signed-off-by: Dave Airlie commit 69827126abdfa289417b55fe7db8ae0535037185 Author: Dave Airlie Date: Wed Jul 25 16:22:57 2012 +1000 intel/uxa: add pixmap sharing support. Signed-off-by: Dave Airlie commit 0768ac4d195214825137152893deb74a87fcd11e Author: Dave Airlie Date: Wed Jul 25 16:11:23 2012 +1000 intel: add platform probing support. This allows the driver to be loaded by the platform loading code. Signed-off-by: Dave Airlie commit 164ae7a4f9fd3e538f1b506031d297088fc0c659 Author: Chris Wilson Date: Sun Sep 2 17:19:12 2012 +0100 man: Malformed "Backlight" section Reported-by: Matthew Monaco Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54397 Signed-off-by: Chris Wilson commit 67b44104168650473a24e69056707e0b35f1713e Author: Chris Wilson Date: Sun Sep 2 12:47:36 2012 +0100 sna: Cache the temporary upload buffer when copying boxes Signed-off-by: Chris Wilson commit 4b558281e04e4d4febfc361632a90f8a45080c49 Author: Chris Wilson Date: Sun Sep 2 12:23:59 2012 +0100 sna: Don't retire when searching for a snoopable buffer without hw support If the hw/kernel doesn't support snoopable buffers, then it makes little sense to search for one, and force a retire in the certainty of not finding any. Signed-off-by: Chris Wilson commit d933f3a7965c9aa70f70477be3bfe94d5ded948b Author: Chris Wilson Date: Sun Sep 2 10:29:49 2012 +0100 2.20.6 release Signed-off-by: Chris Wilson commit 62e7e69d88f7e5c8b0f0931fe887ee0c3bddb43f Author: Chris Wilson Date: Sun Sep 2 09:55:07 2012 +0100 sna: Tweak placement rules if we already have a CPU bo If we have a CPU bo, consider if it may be quicker to render to it then create a GPU bo. Signed-off-by: Chris Wilson commit f837807cc257fbedd35fa6101dc0b4a57e5ad78d Author: Chris Wilson Date: Sat Sep 1 10:36:09 2012 +0100 sna/dri: Remove busy-wait spin from vblank chaining of swapbuffers The issue being that, due to the delay, the chained swap would miss its intended vblank and so cause an unwanted reduction in frame throughput and increase output latency even further. Since both client and server have other rate-limiting processes in place, we can forgo the stall here and still keep the clients in check. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54274 Signed-off-by: Chris Wilson commit b56e8c5105c858452ca4eabf15b298fc06dfd3c8 Author: Chris Wilson Date: Sat Sep 1 12:15:47 2012 +0100 sna: Nullify gpu_bo after free in case final release is deferred in destroy As we may defer the actual release of the pixmap until after completion of the last shm operation, we need to make sure in that case we mark the GPU bo as released to prevent a use-after-free. Signed-off-by: Chris Wilson commit c4d994014160d8c946af731196a908991c77d9f9 Author: Chris Wilson Date: Sat Sep 1 10:44:24 2012 +0100 sna/gen2+: Add the missing assertions in case the drawrect is invalid Only the later gen had these useful assertions, add them to the rest just in case. Signed-off-by: Chris Wilson commit 25c1b478a957ceb0474e7121840b7eed1ba2e140 Author: Chris Wilson Date: Sat Sep 1 09:49:34 2012 +0100 sna: Discard cached CPU upload early when clearing with a solid fill Otherwise we end up considering the GPU bo as a real target, causing confusion and failed asserts. Signed-off-by: Chris Wilson commit 530b1d1516595cf14c5112d8833b870cd50eca46 Author: Chris Wilson Date: Thu Aug 30 22:21:57 2012 +0100 sna/dri: Use common routine for testing busyness after flush Signed-off-by: Chris Wilson commit 00d8c776b3607dbdab32c1126f91a7a38b8065f6 Author: Chris Wilson Date: Thu Aug 30 21:47:17 2012 +0100 sna/dri: Hold a reference to the vsync blit Fixes regression from commit 96a921487ef00db03a12bec7b0821410d6b74c31 Author: Chris Wilson Date: Mon Aug 27 21:50:32 2012 +0100 sna: Track outstanding requests per-ring Signed-off-by: Chris Wilson commit b2a6c74c2159c9968c19400d61a11f4773724b4a Author: Chris Wilson Date: Wed Aug 29 14:51:40 2012 +0100 sna: Remove inconsistent assertion As we now may not prefer to use the GPU even if all-damaged and clear, asserting that if we choose to use the CPU if clear is now bogus. Signed-off-by: Chris Wilson commit deaa1cac269be03f4ec44092f70349ff466d59de Author: Chris Wilson Date: Tue Aug 28 22:23:22 2012 +0100 sna: Align active upload buffers to the next page for reuse If we write to the same page as it already active on the GPU then despite the invalidation performed at the beginning of each batch, we do not seem to correctly sample the new data. References: https://bugs.freedesktop.org/show_bug.cgi?id=51422 References: https://bugs.freedesktop.org/show_bug.cgi?id=52299 Signed-off-by: Chris Wilson commit 9e0305a3928f253ded6c8e141a4dd42be3952eb9 Author: Chris Wilson Date: Tue Aug 28 20:49:20 2012 +0100 sna: Discard inplace flag if we create a CPU map for the upload buffer Signed-off-by: Chris Wilson commit f3e122554e88da0351bfb9a7a722f0715553689b Author: Chris Wilson Date: Tue Aug 28 21:27:04 2012 +0100 sna: Propagate the request to flush rather than directly submit the batch The subtly is that we need to reset the mode correctly after submitting the batch which was not handled by kgem_flush(). If we fail to set the appropriate mode then the next operation will be on a random ring, which can prove fatal with SandyBridge+. Reported-by: Reinis Danne Signed-off-by: Chris Wilson commit b5c8efe4309248e62d94d80b37a70775284ae985 Author: Chris Wilson Date: Tue Aug 28 08:33:00 2012 +0100 sna: Make sure we reset the domain tracking when exporting DRI bo Signed-off-by: Chris Wilson commit 96a921487ef00db03a12bec7b0821410d6b74c31 Author: Chris Wilson Date: Mon Aug 27 21:50:32 2012 +0100 sna: Track outstanding requests per-ring In order to properly track when the GPU is idle, we need to account for the completion order that may differ on architectures like SandyBridge with multiple mostly independent rings. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54127 Signed-off-by: Chris Wilson commit 26c731efc2048663b6a19a7ed7db0e94243ab30f Author: Chris Wilson Date: Mon Aug 27 20:50:08 2012 +0100 sna: Ensure that we create a GTT mapping for the inplace upload buffer As the code will optimistically convert a request for a GTT mapping into a CPU mapping if the object is still in the CPU domain, we need to overrule that in this case where we explicitly want to write directly into the GTT and furthermore keep the buffer around in an upload cache. References: https://bugs.freedesktop.org/show_bug.cgi?id=51422 References: https://bugs.freedesktop.org/show_bug.cgi?id=52299 Signed-off-by: Chris Wilson commit 2cbf88980ede50370b97f32e565dea33db16ac44 Author: Chris Wilson Date: Mon Aug 27 20:07:17 2012 +0100 sna: Force the stall if using a busy ShmPixmap bo with PutImage As we will stall in the near future to serialise access with the ShmPixmap, we may as well stall first and do a simple copy using the CPU in this highly unlikely scenario. Signed-off-by: Chris Wilson commit 705103d77e6b80d796a4535cade96cb6e9ebece3 Author: Chris Wilson Date: Mon Aug 27 20:07:19 2012 +0100 sna: Add a modicum of DBG to kgem_is_idle() Print out the handle of the bo we just checked. Signed-off-by: Chris Wilson commit 8218e5da2b177ca9cd0e2b1e7dbe114e5ef2ebf0 Author: Chris Wilson Date: Mon Aug 27 19:36:03 2012 +0100 sna: Fix crash with broken DBG missing one of its arguments Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54127 Signed-off-by: Chris Wilson commit 593f549b45fbb1528472feed51207d84901d142e Author: Chris Wilson Date: Mon Aug 27 16:43:19 2012 +0100 sna: A little more DBG to try and identify ratelimiting ops Signed-off-by: Chris Wilson commit 71ac12e9b6ed00c28993637aafd5186a2ba26256 Author: Chris Wilson Date: Mon Aug 27 15:49:59 2012 +0100 sna: Assert that the bo is marked as 'flush' when exported to DRI clients Signed-off-by: Chris Wilson commit cf64c8ce758cfa5d3bcd1b7626ff94cce7a84636 Author: Chris Wilson Date: Mon Aug 27 13:59:56 2012 +0100 sna: Upload PutImage inplace to a fresh GPU bo if the device doesn't snoop Signed-off-by: Chris Wilson commit 3c6758fc4a50ecfce9ed317fec669cc48addedcf Author: Chris Wilson Date: Mon Aug 27 11:58:05 2012 +0100 sna: Flush the batch if it references a ShmPixmap and the GPU is idle This helps minimise the stall when syncing with the GPU before sending the next reply to the Client. Signed-off-by: Chris Wilson commit f21079bad6e8316baf5d0295d6e7a809041bce06 Author: Chris Wilson Date: Mon Aug 27 11:25:28 2012 +0100 Revert "sna: Cleanup composite redirection after substituting the BLT" This reverts commit 5a5212117e7a73ce3fffb87c60a505a849e38c36. The clean up is in effect too early, as this is during preparation and the actual work is already being correctly done at the end. commit 414e87255cdee6eb556703ddefd194af71b985ed Author: Chris Wilson Date: Sun Aug 26 19:47:31 2012 +0100 Add Option "Backlight" to override the probed backlight control interface The automatic selection may not correspond with the correct backlight (such as in a multi-gpu, multi-panel device) or the user may simply prefer another control interface. This allows them to override the chosen interface using Option "Backlight" "my-backlight" to specify '/sys/class/backlight/my-backlight' as the interface to use instead. Suggested-by: Alon Levy References: https://bugs.freedesktop.org/show_bug.cgi?id=29273 Signed-off-by: Chris Wilson commit d4f7c58186849374cd929e20fa49ea2e93939a69 Author: Chris Wilson Date: Sun Aug 26 17:01:55 2012 +0100 sna/gen5+: Use the common methods for choosing the render targets This should afford us much more flexibility in where we render. Signed-off-by: Chris Wilson commit 3c88b5f693c29b990d69f96508b121ce97a7209e Author: Chris Wilson Date: Sat Aug 25 11:49:30 2012 +0100 sna: Use a fast span emitter for mono trapezoids without damage or clipping Signed-off-by: Chris Wilson commit 8e10a5b348a37feadcf935ec7694e46cc0802bdf Author: Chris Wilson Date: Sun Aug 26 14:53:12 2012 +0100 sna/gen6+: Do not call sna_blt_composite() after prepping the composite op As sna_blt_composite() will overwrite parts of the composite op as it checks whether or not it can execute that operation, it will lead to a crash as the normal render path finds the op corrupt. (The BLT conversion functions cater for the cases where we may wish to switch pipelines after choosing src/dst bo.) Reported-by: rei4dan@gmail.com Signed-off-by: Chris Wilson commit cbbe7727e766a5ee8767673feb6c8cdec38a7051 Author: Chris Wilson Date: Sun Aug 26 14:43:47 2012 +0100 sna/gen6+: Simplify prefer_blt_bo As we already check the tiling state, so all we need to then check is that the pitch is within the BLT constraint. Signed-off-by: Chris Wilson commit 0c15824a8143a288716d2eacf03252cc54eb9466 Author: Chris Wilson Date: Sun Aug 26 14:29:41 2012 +0100 sna: Add some DBG to kgem_is_idle() Signed-off-by: Chris Wilson commit d432983421286d343f7c487c12c7244b711f5a66 Author: Chris Wilson Date: Sun Aug 26 13:43:35 2012 +0100 sna: Add some DBG to BLT composite substitute to show if redirection is used Signed-off-by: Chris Wilson commit 5a5212117e7a73ce3fffb87c60a505a849e38c36 Author: Chris Wilson Date: Sun Aug 26 13:38:23 2012 +0100 sna: Cleanup composite redirection after substituting the BLT Signed-off-by: Chris Wilson commit 335821d588460c253b2ba2c8616a7c46e5ad0150 Author: Chris Wilson Date: Sun Aug 26 13:29:04 2012 +0100 sna/gen4+: Check for allocation failure for the clear solid bo Signed-off-by: Chris Wilson commit b5c77a6aaf520e331e82409b2592911cb1bb3100 Author: Chris Wilson Date: Sun Aug 26 12:48:18 2012 +0100 2.20.5 release Signed-off-by: Chris Wilson commit 454cc8453af1852758c3396dbe303c13c5c1be27 Author: Chris Wilson Date: Fri Aug 24 08:48:12 2012 +0100 sna: Submit the partial batch before throttling Signed-off-by: Chris Wilson commit 0e1e83ed4952f620e9422e58f955a5aea406e300 Author: Chris Wilson Date: Fri Aug 24 00:59:31 2012 +0100 sna: Allow the batch to be flushed if the GPU is idle upon a context switch Submit early, submit often in order to keep the GPU busy. As always we trade off CPU overhead versus concurrency. Signed-off-by: Chris Wilson commit 5059db0697c5516f1538f7062937664baf7b1c2e Author: Chris Wilson Date: Fri Aug 24 00:21:07 2012 +0100 sna: Correct a pair of DBG messages Signed-off-by: Chris Wilson commit c5b46e411a2c738c5ae55bffb9b3d460249f5c24 Author: Chris Wilson Date: Thu Aug 23 17:36:10 2012 +0100 sna: Tidy up users of __kgem_bo_is_busy() A lot of callsites had not been converted to the common function. Signed-off-by: Chris Wilson commit 82e91327d57e03d2117638165f298a50b946fcaa Author: Chris Wilson Date: Thu Aug 23 15:59:00 2012 +0100 sna: Use a temporary userptr mapping for a large upload into a busy target Signed-off-by: Chris Wilson commit b286ffa6beccb8fe341c464a4fb9f2af98541263 Author: Eric S. Raymond Date: Thu Aug 23 12:42:18 2012 -0400 Fix seriously malformed list syntax on intel(4). Signed-off-by: Eric S. Raymond commit fc6b7f564df88ca773ae245b1b4e278b47dffd59 Author: Chris Wilson Date: Thu Aug 23 15:13:14 2012 +0100 sna: Flush the batch if it contains any DRI pixmaps This fixes a regression from commit 02963f489b177d0085006753e91e240545933387 Author: Chris Wilson Date: Sun Aug 19 15:45:35 2012 +0100 sna: Only submit the batch if flushing a DRI client bo which made the presumption that we called sna_add_flush_pixmap() for every DRI pixmap that we used. However, that is only called for the dirty pixmaps, any native exported pixmap only marks the batch as requiring a flush. So in those cases we always need to submit the batch if it contains an exported DRI pixmap. Reported-by: chr.ohm@gmx.net Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=53967 Signed-off-by: Chris Wilson commit fd38f45ec5421802b426867a050c978a2feec937 Author: Chris Wilson Date: Thu Aug 23 12:11:48 2012 +0100 sna: Mark the CPU damage as needing flushing for DRI buffers Signed-off-by: Chris Wilson commit ac6cb667546a82b865c959a1be32f52b2da7bf7d Author: Chris Wilson Date: Wed Aug 22 12:00:49 2012 +0100 sna: Flush before adding any SHM pixmap into the batch Signed-off-by: Chris Wilson commit 785c1046b94fd9ca6f22b2a6d73639408a411cea Author: Chris Wilson Date: Wed Aug 22 10:56:12 2012 +0100 sna: Only use the GPU for an active CPU bo unless forced Signed-off-by: Chris Wilson commit 8032f51859ce1928922edf6892f493a84d9c39f0 Author: Chris Wilson Date: Wed Aug 22 10:40:53 2012 +0100 sna: If we cannot use the CPU bo along a render pathway, promote to GPU Signed-off-by: Chris Wilson commit cc967507bbf357e1d5ec3cd0f3c0c5ecfa8b867a Author: Chris Wilson Date: Wed Aug 22 10:36:04 2012 +0100 sna: Convert to using IGNORE_CPU flag rather than complicating the CPU damage Signed-off-by: Chris Wilson commit 5aa59ce7c012eb309c4f5a362ccf531c065bd7ff Author: Chris Wilson Date: Wed Aug 22 10:10:33 2012 +0100 sna: Assert that the CPU bo is not used if the GPU is clear Signed-off-by: Chris Wilson commit 273765033223024ff6a662195e0e4b96f8960463 Author: Chris Wilson Date: Wed Aug 22 00:48:08 2012 +0100 sna: Make sure the opposite damage is destroyed after reducing to all Signed-off-by: Chris Wilson commit 31e398c9dbd8539e6fd2cc7398e97ee1df2f7f23 Author: Chris Wilson Date: Wed Aug 22 00:38:17 2012 +0100 sna: Discard a no-longer-used GPU bo after moving to the CPU domain Signed-off-by: Chris Wilson commit 8be15c37df9d9b34e0f21700673212bd19c772b2 Author: Chris Wilson Date: Wed Aug 22 00:27:00 2012 +0100 sna: Balance CPU bo accounting for SHM pixmaps Signed-off-by: Chris Wilson commit 262d57a5aaac46508d8e29860c7a567bcd5bc4d8 Author: Chris Wilson Date: Tue Aug 21 23:36:12 2012 +0100 sna: Display still resident memory in inactive/snoop caches under DEBUG_MEMORY Signed-off-by: Chris Wilson commit 304581cc02adeb865a24edde934b5b9ceba68d96 Author: Chris Wilson Date: Tue Aug 21 23:29:12 2012 +0100 sna: Add a DBG to log pixmap destruction Signed-off-by: Chris Wilson commit 946d54238d90f5e31772f0df336ac9a5e7f2d62b Author: Chris Wilson Date: Tue Aug 21 21:02:00 2012 +0100 sna: Fix the assertion for tracking proxies in the batch Signed-off-by: Chris Wilson commit 4ee2e227ddf61c87f08f55d4922d2562b563ca87 Author: Chris Wilson Date: Tue Aug 21 00:28:40 2012 +0100 sna: Mark all levels of a proxy as dirty So that if we write to a surface through one view, we make sure that the sample cache is invalidated for all future views. Signed-off-by: Chris Wilson commit eee181e9d71273d94fe735805ed84e8f7b7b2180 Author: Chris Wilson Date: Tue Aug 21 00:19:10 2012 +0100 sna/gen6+: Only mark the dst as dirty again if it already is in the batch Signed-off-by: Chris Wilson commit c86df17c1455a53cb52f33a25c8c362e5331621e Author: Chris Wilson Date: Mon Aug 20 22:54:06 2012 +0100 sna/gen3: Fix assertion to check the freshly allocated vertex bo Signed-off-by: Chris Wilson commit 6aabe90587f4916a01a1cd2bbc577a1e7fa20eca Author: Chris Wilson Date: Mon Aug 20 22:09:54 2012 +0100 sna: Allow target bo promotion to GPU even on old architectures Signed-off-by: Chris Wilson commit 1a4b6fea7b1516de35e6800efa5b85f8401a5b2a Author: Chris Wilson Date: Mon Aug 20 22:08:23 2012 +0100 sna: Assign a unique id to snoopable CPU bo Signed-off-by: Chris Wilson commit 892b1a1e431e8f27133825f8a27dde4955da0054 Author: Chris Wilson Date: Mon Aug 20 22:07:05 2012 +0100 sna/gen3: Convert to sna_drawable_use_bo() Signed-off-by: Chris Wilson commit 3ca1bfb51ba522454433d58131e7dab7fcbe7e34 Author: Chris Wilson Date: Mon Aug 20 21:28:47 2012 +0100 sna: Trim a parameter from kgem_bo_mark_dirty() and add some assertions Signed-off-by: Chris Wilson commit 16f3d3a9ae145a3af51d2c0c42c6c585d676a863 Author: Chris Wilson Date: Mon Aug 20 19:42:22 2012 +0100 sna: Keep a stash of the most recently allocated requests Signed-off-by: Chris Wilson commit fb349ced91e15ecaa025321bd37d1fe3cfdd2f44 Author: Chris Wilson Date: Mon Aug 20 17:35:44 2012 +0100 sna: A few more buffer cache management assertions Signed-off-by: Chris Wilson commit ae6b3c7508d570af94f172d6a91d62d2987d654b Author: Chris Wilson Date: Mon Aug 20 17:03:16 2012 +0100 Check that the module that indeed i915 before using custom ioctls Thanks to Adam Jackson for pointing me towards drmGetVersion() and Julien Cristau for saying "Yuck!" Signed-off-by: Chris Wilson commit 3c611087e8ae09fc3fe4271f16d912ac8ae89f6b Author: Chris Wilson Date: Mon Aug 20 16:09:25 2012 +0100 Only open the matching BusID and not the first named If you pass a name to drmOpen() it will attempt to open any device corresponding to that name if it first fails to open the device corresponding to the BusID. Obviously we want the failure from opening the specified device in order to prevent wrongly opening the first found device multiple times. Signed-off-by: Chris Wilson commit dd6c67b32f726b6ad7f12f3b83f6d8c868ff4dc1 Author: Chris Wilson Date: Mon Aug 20 15:58:03 2012 +0100 sna: Add a couple of buffer cache management assertions Signed-off-by: Chris Wilson commit 01ebdb4d7a8bb751167153554f9122d996e4ea91 Author: Chris Wilson Date: Mon Aug 20 15:33:26 2012 +0100 sna: Remove confusing is_cpu() The only real user now has its own heuristics, so convert the remaining users over to !is_gpu(). Signed-off-by: Chris Wilson commit 410316d20299b9ed3447d1d897f904af786ed097 Author: Chris Wilson Date: Mon Aug 20 14:56:12 2012 +0100 sna: Refine decision making for maybe-inplace trapezoids In particular, we want to avoid preferentially taking the CPU paths when it may force any migration (including clear). Signed-off-by: Chris Wilson commit 91f1bf971f9cdc6498f513a5ddec1ad7a4e24b3d Author: Chris Wilson Date: Mon Aug 20 15:27:27 2012 +0100 sna: Correct ordering of calls to memcpy for BLT cpu composite paths Signed-off-by: Chris Wilson commit 027569bf83fc6d67dca7cfd65fdfa37ef6b47204 Author: Chris Wilson Date: Mon Aug 20 12:55:20 2012 +0100 Missing includes for b5b76ad849b The warnings of implicit function declarations were lost amongst the noise. Signed-off-by: Chris Wilson commit f1b259f627814c765ea93d8c839ee7533249974f Author: Chris Wilson Date: Mon Aug 20 12:51:51 2012 +0100 sna: Add a modicum of DBG for PolyFillRect Signed-off-by: Chris Wilson commit b5b76ad849bfda1e75192d1cb3c6c0fcc623bb91 Author: Chris Wilson Date: Mon Aug 20 12:01:39 2012 +0100 Sanity check that the driver is an i915.ko GEM device before claiming it This fixes an issue with us claiming Poulsbo and friends even though we do not speak their language. Signed-off-by: Chris Wilson commit 74f998136bac441d782faf779616af6f7bdcadad Author: Chris Wilson Date: Mon Aug 20 11:07:55 2012 +0100 sna: Avoid migrating the BLT composite src to the GPU if the dst is not Signed-off-by: Chris Wilson commit a05643eb670e91ab102465df799301def88faaf9 Author: Chris Wilson Date: Mon Aug 20 10:46:14 2012 +0100 sna: Remove unneeded source bo unref after __sna_render_pixmap_bo() As __sna_render_pixmap_bo() deliberately does not reference its returned bo, we need to avoid unreferencing it else we cause explosions later. Fixes regression from commit a13781d19defc97af6a279c11a85e33ef825020e Author: Chris Wilson Date: Sun Aug 19 09:45:12 2012 +0100 sna: Enable BLT composite functions to target CPU buffers Signed-off-by: Chris Wilson commit c80db9bc2ef0a37a4abb78c9ef667c8b36ab6fba Author: Chris Wilson Date: Mon Aug 20 10:45:36 2012 +0100 sna: Flush the batch before preparing for a FlushCallback As we end up submitting and maybe synchronising upon a batch within FlushCallback (or our client will) it is important that we start that serialized batch as early as possible to minimise the impending stalls. Signed-off-by: Chris Wilson commit c39fe0253847f5a86e16b47ba420c8ba819c9110 Author: Chris Wilson Date: Sun Aug 19 20:42:55 2012 +0100 sna: Do not use the GPU to migrate to the CPU whilst wedged! Signed-off-by: Chris Wilson commit fe05268d70088c8cad5f4b5ef756e1ffe2069fca Author: Chris Wilson Date: Sun Aug 19 19:28:42 2012 +0100 sna: Experiment with flushing the batch prior to rendering to a ShmPixmap Signed-off-by: Chris Wilson commit bbd7a825810cc9772e6d613df449cb5ecb0be3f6 Author: Chris Wilson Date: Sun Aug 19 19:28:20 2012 +0100 sna: Don't promote a ShmPixmap to GPU for a CopyArea Signed-off-by: Chris Wilson commit 7bf7a5ad1057f1aeb5b261da6dc501323c022287 Author: Chris Wilson Date: Sun Aug 19 19:17:35 2012 +0100 sna/gen3: Tidy vbo discard Signed-off-by: Chris Wilson commit d1b808fd72b477bde96f7c6737a993bd1a20baf2 Author: Chris Wilson Date: Sun Aug 19 17:53:24 2012 +0100 sna: Tweak is_cpu/is_gpu heuristics Signed-off-by: Chris Wilson commit c682c1a37692021cbd9bfc1e3f1ccf2b648c73f9 Author: Chris Wilson Date: Sun Aug 19 17:28:25 2012 +0100 sna: Discard GPU (and damage) after applying clear on migration to CPU Signed-off-by: Chris Wilson commit dc83ef49f9e8ff94ab77aa9d54fe29e32b9d7a3d Author: Chris Wilson Date: Sun Aug 19 17:20:45 2012 +0100 sna/trapezoids: Accept more operators for maybe-inplace Signed-off-by: Chris Wilson commit 47d948f5df0cd0e975cfe5183b6ce79cd3fd27ee Author: Chris Wilson Date: Sun Aug 19 17:09:21 2012 +0100 sna: Update maybe_inplace to recognise more types of handled pixel formats Signed-off-by: Chris Wilson commit 02963f489b177d0085006753e91e240545933387 Author: Chris Wilson Date: Sun Aug 19 15:45:35 2012 +0100 sna: Only submit the batch if flushing a DRI client bo Signed-off-by: Chris Wilson commit 13d1a105159222518800d3c5ad5660725864ec6b Author: Chris Wilson Date: Sun Aug 19 09:48:05 2012 +0100 sna: compare the correct trailing dword when skipping identical bitmap lines Fixes regression in 2.20.4 from commit 85192f00e345830541e3715e211b1f98154bbef4 Author: Chris Wilson Date: Wed Aug 8 12:11:50 2012 +0100 sna: Ignore trailing bits when comparing lines inside the bitmap Reported-by: Edward Sheldrake Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=53699 Signed-off-by: Chris Wilson commit a13781d19defc97af6a279c11a85e33ef825020e Author: Chris Wilson Date: Sun Aug 19 09:45:12 2012 +0100 sna: Enable BLT composite functions to target CPU buffers Signed-off-by: Chris Wilson commit be940856c74fbedd27997dd61e2a85959b321193 Author: Chris Wilson Date: Sat Aug 18 21:23:03 2012 +0100 sna: Consider sample wraparound in each direction independently Signed-off-by: Chris Wilson commit 110c7ef7f6c31929affa038918e6ce087bccddc6 Author: Chris Wilson Date: Sat Aug 18 17:42:51 2012 +0100 sna/damage: Replace the damage with a larger box if subsumed Signed-off-by: Chris Wilson commit 75a2fab766d8aed180ef795919e503db22c0e0fd Author: Chris Wilson Date: Sat Aug 18 17:28:08 2012 +0100 sna: Avoid forcing an upload for an unblittable bo unless on a fallback path Signed-off-by: Chris Wilson commit 110d5519f3523b1e2c50db637cdc4c5bc44c960a Author: Chris Wilson Date: Sat Aug 18 16:39:52 2012 +0100 sna: Reduce subtracted damage earlier Signed-off-by: Chris Wilson commit 8812e8b6e89e6432a6a768a0566ce4c153e9b256 Author: Chris Wilson Date: Sat Aug 18 16:21:13 2012 +0100 sna: Reduce damage after a large composite operation Signed-off-by: Chris Wilson commit e361627b90ea6bf2f9a8c46cf8debe562fdf4f09 Author: Chris Wilson Date: Sat Aug 18 16:16:17 2012 +0100 sna/damage: Add some more sanity checks for creating empty regions Signed-off-by: Chris Wilson commit 10f334872e9dd190e18c768219e60815acabe4d3 Author: Chris Wilson Date: Sat Aug 18 15:49:02 2012 +0100 sna: Add damage for the whole unaligned trapezoid not per component Signed-off-by: Chris Wilson commit 5ff6198c9346d84717bac28980329b048f4406e8 Author: Chris Wilson Date: Sat Aug 18 11:59:56 2012 +0100 2.20.4 release Signed-off-by: Chris Wilson commit bc77a466531fdbdd21e9354af567e5215a66edf9 Author: Chris Wilson Date: Sat Aug 18 11:09:23 2012 +0100 sna: Silence a couple of recent compiler warnings sna_accel.c: In function 'sna_poly_fill_rect_extents': sna_accel.c:11438:11: warning: unused variable 'v' [-Wunused-variable] sna_blt.c: In function 'sna_blt_composite__convert': sna_blt.c:1712:3: warning: format '%s' expects a matching 'char *' argument [-Wformat] sna_blt.c:1738:3: warning: format '%x' expects argument of type 'unsigned int', but argument 4 has type 'CARD32' [-Wformat] Signed-off-by: Chris Wilson commit d1da2c71461798bb027a260ff8480b71ebffde26 Author: Chris Wilson Date: Thu Aug 16 21:15:54 2012 +0100 sna: Correct assertion for StoreColors Fixing the assertion reveals that it was bogus! Signed-off-by: Chris Wilson commit 31ddaa0cfbde083aef42d9fdb5ed13bb4edc36c8 Author: Chris Wilson Date: Thu Aug 16 21:13:52 2012 +0100 sna: Fixup DBG after 7f5a9e3 Signed-off-by: Chris Wilson commit b8bc56c12c6fdc559354fbbf96b6e043a491aa8f Author: Chris Wilson Date: Thu Aug 16 21:00:57 2012 +0100 uxa/gen3: Simply simple fill shader generation Suggested-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 182581edde356a21a0681de526f83884e1a0de5d Author: Chris Wilson Date: Thu Aug 16 20:59:09 2012 +0100 uxa/glamor: Check for failed prepares Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit c289b607d1526d8d3625fa84c093552a9f3ea168 Author: Chris Wilson Date: Thu Aug 16 20:54:40 2012 +0100 uxa: Check for failed mmapping of the scanout Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 32ec8b979bc8cf7d8ce351ca752b806d42bc1c0f Author: Chris Wilson Date: Thu Aug 16 20:53:08 2012 +0100 test/dri2: Discard error returns from _XReply This is only test code, so keep the static analyser quiet Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 5675c36e0dca92b8d24235d82cd0e87f82fdc7b0 Author: Chris Wilson Date: Thu Aug 16 20:44:32 2012 +0100 sna: Check for failure to pin the front buffer Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit c9db2c74ab3b482489bec63b11df80bb35e21aca Author: Chris Wilson Date: Thu Aug 16 20:42:50 2012 +0100 sna/gen2: Add break rather than fallthrough The fall-through in this instance is irrelevant, and the static analysers complain for not commenting on the fall-through. Silence the analyser by removing the fall-through. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit e0abbc3ccafb51d6b2114e1c7d6dc2eda3a8c1f4 Author: Chris Wilson Date: Thu Aug 16 20:40:17 2012 +0100 i810: Query PortPriv for the surface properties Avoid the NULL deferences. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit c09518f98d5c533860b084506d9ee2c819cca52f Author: Chris Wilson Date: Thu Aug 16 20:32:53 2012 +0100 sna: Fix assignment inside assertion (should be an equality check) Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 7d3e4328d4f4421178377933ca98fb80daa85c76 Author: Chris Wilson Date: Thu Aug 16 20:32:11 2012 +0100 sna/glyphs: Fix potential leak of glyph extents (for many lists of many formats) Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 9b00c82df78cb8820c0e32fbecc7c933eed81d52 Author: Chris Wilson Date: Thu Aug 16 20:30:52 2012 +0100 sna: Remember alpha-fixup for BLT upload Still dubious whether this is truly beneficial... Signed-off-by: Chris Wilson commit 94871944a0e1351273d6029df7bf0300f31a8571 Author: Chris Wilson Date: Thu Aug 16 00:07:58 2012 +0100 sna: Avoid conflating tearfree and randr shadow modes For ordinary TearFree we do not want to adjust the crtc offsets. Signed-off-by: Chris Wilson commit 1a389842347bab9d91240444e161589071eb9a47 Author: Chris Wilson Date: Wed Aug 15 23:07:16 2012 +0100 sna: Avoid queuing a pageflip on a DPMS off pipe If the pipe is not running, attempting to queue a pageflip will result in an error and us disabling the output in retaliation. Signed-off-by: Chris Wilson commit 615739556dd1cc4565eb1c47f93fe8abd697802f Author: Chris Wilson Date: Wed Aug 15 20:10:52 2012 +0100 sna: Only continue to update the shadow buffer if the flips succeeded If the flip fail, we disable the crtc and may end up disabling the output, removing the shadow (and old scanout bo) in the process. Signed-off-by: Chris Wilson commit 2554d0d76e0fcae6e324938c28bb50deeb8814dc Author: Chris Wilson Date: Wed Aug 15 18:04:18 2012 +0100 sna: Submit the batch upon destroying a ShmPixmap The midlayer has already detached the pixmap from the segment (possibly destroying that segment in the process, thanks midlayer!) so we need to submit the batch asap before the segment disappears. Signed-off-by: Chris Wilson commit 4e604d721be5a1c0f8eaf91e3a5cf8c01a609a69 Author: Chris Wilson Date: Wed Aug 15 16:53:20 2012 +0100 sna: Flush shm pixmaps after upload Signed-off-by: Chris Wilson commit 7f5a9e30c953c174a4d6bc83e0b0a5e068cbda52 Author: Chris Wilson Date: Wed Aug 15 09:34:24 2012 +0100 sna: Remove zero-size rectangles fro PolyFillRectangle Some applications like to waste precious bw by attempting to fill zero-sized rectangles; remove them upfront rather than process them. Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 0c882f4fa991ef379e416628eb41cd2178818ad1 Author: Chris Wilson Date: Tue Aug 14 16:02:38 2012 +0100 sna/gen2,3: Prevent dereferencing a NULL bo with solid render fills Signed-off-by: Chris Wilson commit 48f436d3705059711097a222aac65f862b4a5cfd Author: Chris Wilson Date: Tue Aug 14 15:05:33 2012 +0100 sna: Defer the release of a busy SHM pixmap until the next flush callback Signed-off-by: Chris Wilson commit 9f07d1fc0181f22da36837337241f228626e0d88 Author: Chris Wilson Date: Tue Aug 14 12:03:24 2012 +0100 sna: Enable use of shm pixmap for readback A little too overzealous in the degradation of writing to the CPU bo if we need to read from a GPU bo. Signed-off-by: Chris Wilson commit 9b016d2e466f342cc6649504fa48ab6a810c7e94 Author: Chris Wilson Date: Sun Aug 12 21:36:30 2012 +0100 sna: do not bother attaching to a miniscule ShmPixmap If the pixmap is less than a page in size we are unlikely to recover the cost of tracking it via the GPU. Signed-off-by: Chris Wilson commit 2f4de90709264ad19a3e3f5f0f79f4bba78a760a Author: Chris Wilson Date: Sun Aug 12 13:01:24 2012 +0100 uxa: Only emit a warning if we fail to tile the frontbuffer and the hw allows Signed-off-by: Chris Wilson commit 4d3cafc99ab49da6744046bc2004981144cb064d Author: Chris Wilson Date: Sun Aug 12 13:02:43 2012 +0100 sna: Fix compilation without pixman glyphs Signed-off-by: Chris Wilson commit dc18eaa585c36c8e5f5b4ec405a976a835fd2ac3 Author: Chris Wilson Date: Sun Aug 12 10:34:10 2012 +0100 sna: Make the failure to create render caches non-fatal Signed-off-by: Chris Wilson commit b580abdfa68108f3e63ee1f897b6ea83b9c47935 Author: Dave Airlie Date: Sun Aug 12 20:52:12 2012 +1000 intel: fix video xvPipe range check Pointed out by Russell King on irc. Signed-off-by: Dave Airlie commit 44f848f9b2f2a2dcd9087210ea46bc4fdb63c057 Author: Chris Wilson Date: Sat Aug 11 19:44:15 2012 +0100 sna: Fix typo in computation of texel offsets for tiled 8x8 blts Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=53353 Signed-off-by: Chris Wilson commit b335d369bbcbfaef878795a5c8e72c2d40d3a536 Author: Chris Wilson Date: Sat Aug 11 12:59:49 2012 +0100 sna: Discard the GPU bo after syncing to CPU bo for writing Signed-off-by: Chris Wilson commit 029934662e0bad6cf47baf4d7202656dd3e0ba08 Author: Chris Wilson Date: Sat Aug 11 10:47:44 2012 +0100 sna: Add a little DBG for promotion of CPU source bo Signed-off-by: Chris Wilson commit f20f8556bfc8b4df6ba5050cca656a4ff0fdf18f Author: Chris Wilson Date: Sat Aug 11 10:46:40 2012 +0100 sna: Treat ShmPixmap as ordinary In theory the code handles the automatic promotion and demotion of the GPU buffers. Signed-off-by: Chris Wilson commit 16c751a06c503b85c5ab6588bae277af4348487d Author: Chris Wilson Date: Sat Aug 11 10:44:08 2012 +0100 sna: Only mark userptr bo as snooped on non-LLC architectures Signed-off-by: Chris Wilson commit 59359dba6496229eff7c60dd04536c5add69cec8 Author: Chris Wilson Date: Sat Aug 11 10:25:40 2012 +0100 sna: Mark SHM userptr bo as unreusable Signed-off-by: Chris Wilson commit 64488010504a6e76008bb3b3c1e61caeb025913c Author: Chris Wilson Date: Sat Aug 11 10:21:39 2012 +0100 sna/gen2-5: Substitute an equivalent BLT composite operation Signed-off-by: Chris Wilson commit f464d508c870293699616626d64bd64f16051467 Author: Chris Wilson Date: Sat Aug 11 10:10:32 2012 +0100 sna/gen6+: Try to use the BLT to avoid TLB misses Signed-off-by: Chris Wilson commit e9c0e54e69a5ce93bb4e79d56da4a83fddf49c4e Author: Chris Wilson Date: Sat Aug 11 08:58:14 2012 +0100 sna: Handle userptr failures more gracefully Signed-off-by: Chris Wilson commit 5d6d9231cd2003fda1c6f2dd3174014317a45704 Author: Chris Wilson Date: Fri Aug 10 23:07:07 2012 +0100 sna: Reset BLT state after copy-boxes Signed-off-by: Chris Wilson commit 50bfc172f1f89c023bb3a47418b845e31eba7126 Author: Chris Wilson Date: Fri Aug 10 22:49:26 2012 +0100 sna: Improve a DBG statement Signed-off-by: Chris Wilson commit b33f6754a99f6d11e423d6a03739fa2c04eeed88 Author: Chris Wilson Date: Fri Aug 10 21:59:36 2012 +0100 sna: Add assertions to 8x8 tiled BLTs and reset BLT state afterwards Reported-by: Clemens Eisserer References: https://bugs.freedesktop.org/show_bug.cgi?id=53353 Signed-off-by: Chris Wilson commit ff54d4af39edcddd55f597c404ab6dd548a24aa9 Author: Chris Wilson Date: Fri Aug 10 21:41:38 2012 +0100 sna: Add a couple of debug options to control use of BLT patterns References: https://bugs.freedesktop.org/show_bug.cgi?id=53353 Signed-off-by: Chris Wilson commit 22306144030b5d37df6d46321555bced6e33c50c Author: Chris Wilson Date: Wed Aug 8 19:45:09 2012 +0100 sna: Use the HAS_SEMAPHORES parameter when available Signed-off-by: Chris Wilson commit 465ae1840a883c50d5a5aab7380b30527f00e17b Author: Chris Wilson Date: Wed Aug 8 12:23:25 2012 +0100 sna: Use a compile time constant for determine the zeroth bit Signed-off-by: Chris Wilson commit 85192f00e345830541e3715e211b1f98154bbef4 Author: Chris Wilson Date: Wed Aug 8 12:11:50 2012 +0100 sna: Ignore trailing bits when comparing lines inside the bitmap References: https://bugs.freedesktop.org/show_bug.cgi?id=51422 Signed-off-by: Chris Wilson commit edc1427f3dcddb73acdb5b5e03756ecb30cb3797 Author: Chris Wilson Date: Wed Aug 8 10:53:40 2012 +0100 sna/gen5: Add DBG for context switching Signed-off-by: Chris Wilson commit fec78ddc60b1f03097b334672546c2c7ef21ab4f Author: Chris Wilson Date: Wed Aug 8 10:08:52 2012 +0100 sna: Add DBG option to disable discarding of glyph masks Signed-off-by: Chris Wilson commit b0cd430082d219500729d20c4740440e58773892 Author: Chris Wilson Date: Wed Aug 8 10:07:06 2012 +0100 sna/gen5: Fix use of uninitialised rgba value for DBG Signed-off-by: Chris Wilson commit 9ad923f95c59cdf42363f95c7dd52188c0987296 Author: Chris Wilson Date: Wed Aug 8 09:59:50 2012 +0100 sna: Minor DBG for fallback glyphs and discarding the mask Signed-off-by: Chris Wilson commit 1a0590d133ea6991e0939d1f170f9c10df6856a0 Author: Chris Wilson Date: Tue Aug 7 17:16:35 2012 +0100 sna: Check the composite extents against the clip When computing the composite extents (as opposed to the composite region) also check if the resultant box overlaps the destination clip region (we know it already fits into the extents). This helps in cases with small roi against clipped drawables, such as drawing text onto expose events. References: https://bugs.freedesktop.org/show_bug.cgi?id=51422 Signed-off-by: Chris Wilson commit 5f5a10ef04a8c01b22da2284583851d84273dc2a Author: Paulo Zanoni Date: Mon Aug 6 18:48:09 2012 -0300 Add Haswell PCI IDs Signed-off-by: Paulo Zanoni Reviewed-by: Rodrigo Vivi Signed-off-by: Chris Wilson commit d8f7f933bc2d30e529730521d5628c10df0361de Author: Chris Wilson Date: Sun Aug 5 17:18:54 2012 +0100 sna: Format markup to suppress compiler warning Signed-off-by: Chris Wilson commit 4d0a259988f67f5c569c5d8000c010a7b662efd3 Author: Chris Wilson Date: Sat Aug 4 15:54:19 2012 +0100 sna/gen6: Compile fix for DBG Signed-off-by: Chris Wilson commit d917a276bff4e933590e78404526b929b6b3b497 Author: Chris Wilson Date: Sat Aug 4 18:05:03 2012 +0100 sna/gen4: Restore the w/a flush for fill/copy as well So far, it looks like that the only one we can indeed drop is the composite with mask. Signed-off-by: Chris Wilson commit d754bc3435889b5d19085588c9c888dc08f9522e Author: Cyril Brulebois Date: Mon Nov 7 23:29:29 2011 +0100 Fix build failure when passing --with-builderstring. If --with-builderstring="foo" is passed to configure (and sna enabled): CC sna_driver.lo src/sna/sna_driver.c: In function 'sna_init_scrn': src/sna/sna_driver.c:1023:7: error: token ""foo"" is not valid in preprocessor expressions Fix the missing defined() around the BUILDER_DESCRIPTION variable. Signed-off-by: Cyril Brulebois commit 4bad76b4510d8cf4ccf3f64bcd13ebb80fae4d26 Author: Chris Wilson Date: Sat Aug 4 14:28:41 2012 +0100 sna/dri: Don't force ring selection if we have semaphores Signed-off-by: Chris Wilson commit fb7b584ec9d0ea96175ba3a621699378c6724940 Author: Chris Wilson Date: Sat Aug 4 14:27:46 2012 +0100 sna/dri: Correct ring selection for a busy bo Confused the RENDER ring with the BLT, limiting swap performance. Signed-off-by: Chris Wilson commit 20e58077fe071fd03d44a97194aa163376dc46a4 Author: Chris Wilson Date: Sat Aug 4 11:21:10 2012 +0100 sna/gen6: Reduce ring switching for overlapping copy boxes Signed-off-by: Chris Wilson commit 2345227663f443baa3c30bdf8b8c33827b7f4947 Author: Chris Wilson Date: Sat Aug 4 11:12:12 2012 +0100 sna/gen7: Use a temporary to avoid switching rings for overlapping copies Signed-off-by: Chris Wilson commit 61ec2999afd5dad22580e024421f13afe7c82f3a Author: Chris Wilson Date: Sat Aug 4 10:33:05 2012 +0100 sna/gen4: Restore w/a flush for video One flush removal too many, keep those fingers crossed that the others do not make an unwanted return. Reported-by: Roman Jarosz Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=53119 Signed-off-by: Chris Wilson commit 5833ef173a01afb710acf10e806b83c5ca6efc09 Author: Chris Wilson Date: Sat Aug 4 09:31:41 2012 +0100 2.20.3 release Signed-off-by: Chris Wilson commit 036b90f099af21e60fb4c3684616daf1927f705e Author: Chris Wilson Date: Fri Aug 3 21:41:59 2012 +0100 sna/gen7: Correct number of texture coordinates used for video Fixes regresion from commit 33c028f8be829caa4fdb9416ff177dc71f24b68e Author: Chris Wilson Date: Wed Aug 1 01:17:50 2012 +0100 sna/gen6+: Reduce floats-per-vertex for spans Signed-off-by: Chris Wilson commit 05dcc5f1699ba90fc14c50882e8d4be89bc4a4f9 Author: Chris Wilson Date: Fri Aug 3 15:08:45 2012 +0100 Pass the chipset info through driverPrivate rather than a global pointer Signed-off-by: Chris Wilson commit 2b3f4ca33a00440a7005fef69099f8dbaddbbad1 Author: Chris Wilson Date: Fri Aug 3 14:27:51 2012 +0100 Unexport intel_chipsets Only used by the core module code, so make it static. Signed-off-by: Chris Wilson commit 5ff749727d3590368806508ac0d0fa8efd1d1d51 Author: Chris Wilson Date: Wed Jul 25 22:21:29 2012 +0100 sna/gen7: Add constant variations and hookup a basic GT descriptor for Haswell Signed-off-by: Chris Wilson commit cd028cad3dc9b059a3d83b818d581f86e16ec317 Author: Chris Wilson Date: Thu Jul 26 13:17:11 2012 +0100 sna: Limit the batch size on all gen7 variants Seems the limit on the surface state size is common across the family Signed-off-by: Chris Wilson commit 4cd9ec9d404d934268952a1058afa07741b09efe Author: Gwenole Beauchesne Date: Fri May 4 18:26:46 2012 +0200 uxa: fix 3DSTATE_PS to fill in number of samples for Haswell The sample mask value must match what is set for 3DSTATE_SAMPLE_MASK, through gen6_upload_invariant_states(). Signed-off-by: Gwenole Beauchesne Reviewed-by: Kenneth Graunke commit 412668464cf9505629eac20001701af3402dc6e8 Author: Gwenole Beauchesne Date: Fri May 4 17:55:10 2012 +0200 uxa: set "Shader Channel Select" fields in surface state for Haswell For normal behaviour, each Shader Channel Select should be set to the value indicating that same channel. i.e. Shader Channel Select Red is set to SCS_RED, Shader Channel Select Green is set to SCS_GREEN, etc. Signed-off-by: Gwenole Beauchesne Reviewed-by: Kenneth Graunke commit a47ba68996f117fabcb601d35bcc5f99cbcd6122 Author: Gwenole Beauchesne Date: Fri May 4 17:17:22 2012 +0200 uxa: fix max PS threads shift value for Haswell The maximum number of threads is now a 9-bit value. Thus, one more bit towards LSB was re-used. i.e. bit position is now 23 instead of 24 on Ivy Bridge. Signed-off-by: Gwenole Beauchesne Reviewed-by: Kenneth Graunke commit ce4421e175ceb9259208c7c223af8d66282c3db3 Author: Gwenole Beauchesne Date: Fri May 4 17:09:19 2012 +0200 uxa: use at least 64 URB entries for Haswell Signed-off-by: Gwenole Beauchesne Reviewed-by: Kenneth Graunke commit 8c880aa34c522b0d67cbb932771f00c947d00dec Author: Gwenole Beauchesne Date: Fri May 4 17:43:19 2012 +0200 uxa: add IS_HSW() macro to distinguish Haswell from Ivybridge Signed-off-by: Gwenole Beauchesne commit 0c0d1d956a8ba37d9e6f4a5e4f52018c8ce498e5 Author: Gwenole Beauchesne Date: Fri Aug 3 12:03:00 2012 +0100 Introduce a chipset identifier for Haswell (Ivybridge successor) Signed-off-by: Gwenole Beauchesne commit 146959dd5ef28384a3db4fce4bf7840f2b3ec58c Author: Chris Wilson Date: Wed Aug 1 23:43:15 2012 +0100 sna: Drop the clear flag as we discard the GPU damage Hopefully only to keep the sanity checks happy... Signed-off-by: Chris Wilson commit 7404e3085b2ee36fa24f77a02d156b4b1d2dff60 Author: Chris Wilson Date: Wed Aug 1 23:37:35 2012 +0100 sna: Ensure we only mark a clear for a fill on the GPU bo Signed-off-by: Chris Wilson commit ca46d1c7a18596ea9fe2b0577ccf1d110e3e42ac Author: Chris Wilson Date: Wed Aug 1 20:20:29 2012 +0100 sna/gen7: Prefer the BLT for self-copies Looking at the test results for a third time, gives the edge to the BLT again. commit e4a3cd3d16447b5d83d1c8c63c342f1240935267 Author: Chris Wilson Date: Wed Aug 1 17:37:33 2012 +0100 sna: Add validation of the clear flag to pixmap debugging Signed-off-by: Chris Wilson commit eaeda34bef711cc566f51dee092a19a3c4ac1a16 Author: Chris Wilson Date: Wed Aug 1 16:23:05 2012 +0100 sna: Fix computation of st values for SIMD8 dispatch Fixes regression with enabling 8-pixels. Reported-by: Mehran Kholdi Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=53044 Signed-off-by: Chris Wilson commit 55231eca818c82620c0146384b19b5d5659f6cd6 Author: Chris Wilson Date: Wed Aug 1 16:22:23 2012 +0100 sna/gen6: Install a fallback 16-pixel shader In case the DBG options leave no shader compiled, make sure we always supply one. Signed-off-by: Chris Wilson commit 4e79c1fef064ce68914eb644edd7f588be3d7300 Author: Chris Wilson Date: Wed Aug 1 15:58:43 2012 +0100 Revert "sna/gen7: Prefer the BLT for self-copies" This reverts commit 89e75dbcb6749bde7587ecc08abed276c255e7f9. Having removed the forced stall for a RENDER self-copy there is no longer a need to encourage ring switching. Signed-off-by: Chris Wilson commit 85cef78a40c6a7a0254f8fba685f224eac6038e1 Author: Chris Wilson Date: Wed Aug 1 13:39:36 2012 +0100 sna/gen7: Simplify the force-stall detection After reducing the number of conditions where we think we need to force the stall on the results, we can then simplify the code to detect that remaining case. Signed-off-by: Chris Wilson commit 9391a2c71f020541a2a62ae68eadd486216a38df Author: Chris Wilson Date: Wed Aug 1 12:50:32 2012 +0100 sna/gen7: Only force a stall for a dirty target if also used as a blend source Signed-off-by: Chris Wilson commit 0a4bb8663b9fa9b39d13bfb49aea30f2aaecce78 Author: Chris Wilson Date: Wed Aug 1 11:19:27 2012 +0100 sna/gen4: Flush not required between fill vertices, only nomaskcomposite A small breakthrough... Still need to flush the primitive between state changes though. Signed-off-by: Chris Wilson commit 33c028f8be829caa4fdb9416ff177dc71f24b68e Author: Chris Wilson Date: Wed Aug 1 01:17:50 2012 +0100 sna/gen6+: Reduce floats-per-vertex for spans Signed-off-by: Chris Wilson commit 9b2873d3d97b6780d878bd9b821fba0302470f9f Author: Chris Wilson Date: Wed Aug 1 00:01:15 2012 +0100 sna/gen4+: Implement an opacity shader Avoid the cumbersome lookup through the alpha gradient texture and simply multiply the incoming opacity value. The next step will be to reduce the number of floats required per vertex. Now that we have removed the primary user of the alpha solid cache, it may be time to retire that as well. Signed-off-by: Chris Wilson commit fd3a1236051265fab700aad689a171de47d7338f Author: Chris Wilson Date: Tue Jul 31 10:20:51 2012 +0100 sna/gen6: Enable 8 pixel dispatch This gives a small performance increase when operating with rectangles, which is reasonably frequent. Signed-off-by: Chris Wilson commit 8922b804bc9ed27957c81f7cda4812ab4ecbd4de Author: Chris Wilson Date: Tue Jul 31 10:20:51 2012 +0100 sna/gen7: Enable 8 pixel dispatch This gives a small performance increase when operating with rectangles, which is reasonably frequent. Signed-off-by: Chris Wilson commit 492093d04b1486dd34aafe2f109a77ddeb836f18 Author: Chris Wilson Date: Tue Jul 31 18:39:17 2012 +0100 sna: Generate shaders for SNB+ 8-pixel dispatch Not ideal yet, sampling an alpha-only surface using SIMD8 only seems to ever return 0... Signed-off-by: Chris Wilson commit 6a5ed88f9fab654c9c11c566b841d42150d26c5d Author: Chris Wilson Date: Mon Jul 30 17:08:20 2012 +0100 sna/gen4: Tidy debugging code Cluster the ifdefs together in the initialisation code. Signed-off-by: Chris Wilson commit 46ec9b0ed55d0fcade40f92206e59c02e402d870 Author: Chris Wilson Date: Tue Jul 31 17:41:34 2012 +0100 sna: Update DPMS mode on CRTC after forcing the outputs on If we forcibly update the outputs to be on, then the core will not issue its on DPMS event and we miss out on updating the CRTC bookkeeping in sna_crtc_dpms(). So we need to update the flag on the CRTC as we manipulate the outputs during modesetting. References: https://bugs.freedesktop.org/show_bug.cgi?id=52142 Signed-off-by: Chris Wilson commit 8f166d26b8a93592939068c5a8d160981c724cfd Author: Chris Wilson Date: Tue Jul 31 11:58:24 2012 +0100 sna: Be more careful with damage reduction during CompositeRectangles We actually need to force DAMAGE_ALL in case we are promoting the GPU pixmap. Signed-off-by: Chris Wilson commit e6cb5d93eaa01e7f4763f797bba341f3cc481d98 Author: Chris Wilson Date: Mon Jul 30 11:14:58 2012 +0100 sna: Avoid overlapping gpu/cpu damage with IGNORE_CPU We cannot simply ignore the presence of CPU damage with IGNORE_CPU but must remember to discard it. Signed-off-by: Chris Wilson commit c9805ba98775bb1e969ff59c7044fe1a49673ca8 Author: Chris Wilson Date: Sun Jul 29 18:19:37 2012 +0100 sna: Export sna_drawable_use_bo() to select target for FillRectangles Signed-off-by: Chris Wilson commit 89e75dbcb6749bde7587ecc08abed276c255e7f9 Author: Chris Wilson Date: Sun Jul 29 15:36:44 2012 +0100 sna/gen7: Prefer the BLT for self-copies If we are copying to ourselves, we have to regularly flush the render cache at which point the RENDER pipeline is slower than the BLT pipeline. Signed-off-by: Chris Wilson commit 33d6afda6cec124494f49b74152768da8a3fbdb5 Author: Chris Wilson Date: Sun Jul 29 09:51:16 2012 +0100 sna/gen7: Compile basic kernels at runtime commit eba8d3b3e14a5a16cea6cb8a89f12d3feb8f3d99 Author: Chris Wilson Date: Sun Jul 29 09:51:16 2012 +0100 sna/gen6: Compile basic kernels at runtime commit 8515ec90405912b3d776defcd6e81b1b5f699f1e Author: Chris Wilson Date: Sun Jul 29 09:51:16 2012 +0100 sna/gen5: Compile basic kernels at runtime commit 00c08b1842c9493ca918a868202946b2e7150de0 Author: Chris Wilson Date: Sun Jul 29 09:51:16 2012 +0100 sna/gen4: Compile basic kernels at runtime commit 7c9dbc980b760e0053d83ca2d7cb147613285680 Author: Chris Wilson Date: Sun Jul 29 09:50:39 2012 +0100 sna: Assemble SF and WM kernels using brw Signed-off-by: Chris Wilson commit 8ebafa0493c0fa08ab9d80eeb1191b7560dc0863 Author: Chris Wilson Date: Wed Sep 21 19:06:07 2011 +0100 sna: Add the brw assembler In order to construct programs on the fly to cater for the combinatorial number of possible shaders, we need an assembler, whilst also taking the opportunity to remove some of the inefficiencies and mistakes from the current shaders. Signed-off-by: Chris Wilson commit ca9d9c02a260bf7930e04bf64e93cc051893c04e Author: Chris Wilson Date: Sat Jul 28 23:08:07 2012 +0100 sna: Prefer not to create a GPU bo without RENDER acceleration Unless that bo happens to be used on a render chain to the scanout. Signed-off-by: Chris Wilson commit eefbe5b6038424566faf7333bb09764b050dd6b4 Author: Chris Wilson Date: Mon Jul 30 10:56:29 2012 +0100 sna: Debug option to test migration of inactive pixmaps Signed-off-by: Chris Wilson commit d3499cacb59f19b5a3439a630ffbc3e105a27b75 Author: Chris Wilson Date: Sun Jul 29 23:00:15 2012 +0100 sna: Assert that we never attempt to submit a batch whilst wedged We should be asserting at the point that we insert the invalid operation into the batch, but asserting upon submitting the batch is a useful failsafe. Signed-off-by: Chris Wilson commit cb4d789f98e6e05ec29e5242887018c7450ddf10 Author: Chris Wilson Date: Sun Jul 29 22:55:44 2012 +0100 sna: Disable the warning for a hung GPU is we manually set wedged Only warn about a hung GPU if we encounter an EIO during operation, but don't warn if we set wedged during initialisation based on unsupported hw or user request. Signed-off-by: Chris Wilson commit 0fd680ff52f7ce0a101c617dfb8997c4e228e3ad Author: Chris Wilson Date: Sun Jul 29 22:51:26 2012 +0100 Don't disable acceleration on 830/845g by default Run the risk of a GPU hang (it shouldn't endanger the entire machine normally) and let the user elect to disable it through Option "NoAccel" "true" Signed-off-by: Chris Wilson commit 3d45f0affe263985f440e144203ed7cbb3803696 Author: Chris Wilson Date: Sat Jul 28 18:21:08 2012 +0100 sna: Honour the Option "DRI" References: https://bugs.freedesktop.org/show_bug.cgi?id=52624 Signed-off-by: Chris Wilson commit fb385745a2347f8966765567e78229d67ddc8d60 Author: Chris Wilson Date: Fri Jul 27 13:24:04 2012 +0100 sna/gen4: Move the common vertex_offset==0 check into the flush() Signed-off-by: Chris Wilson commit 73ddd8b0decee444a57f10a11f05deebba686649 Author: Chris Wilson Date: Fri Jul 27 12:43:00 2012 +0100 sna/gen4: Further refinement to the GT allocation Still hunting for why gen4 fails utterly. Signed-off-by: Chris Wilson commit 717823a9f77cfedc50e03c5c31c1da44a396b561 Author: Zhigang Gong Date: Fri Jul 27 18:12:26 2012 +0800 uxa/dri (glamor): Use exchange buffer in glamor fixup. The previous implementation is to create a new textured pixmap based on the newly created pixmap's buffer object. This is not efficient, as we already created it when we call CreatePixmap. We can just exchange the underlying texture/image buffers by calling intel_glamor_exchange_buffers(). And this commit seems also fix a weird rendering problem when working with compiz/mutter. Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit a8ee1406244d8b8399bf933d6b61bfd14374b5f9 Author: Chris Wilson Date: Fri Jul 27 09:07:16 2012 +0100 2.20.2 release Signed-off-by: Chris Wilson commit bef73cd9279be3438e467981db39c67bc13104f5 Author: Chris Wilson Date: Thu Jul 26 21:54:33 2012 +0100 sna/dri: Select the engine before emitting the wait So that if we have a flexible WAIT_FOR_EVENT that can go on either pipeline, we can choose our preferred pipeline for DRI. Signed-off-by: Chris Wilson commit 1ced4f1ddcf30b518e1760c7aa4a5ed4f934b9f5 Author: Chris Wilson Date: Thu Jul 26 10:50:31 2012 +0100 Reduce maximum thread count for IVB GT1 to avoid spontaneous combustion Somewhere along the way it seems that IVB GT1 was reduced to only allow a maximum of 48 threads, as revealed in the lastest bspecs. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=52473 Signed-off-by: Chris Wilson commit 8f8f8759111f791ee99adfd87296443fb0e6acad Author: Chris Wilson Date: Thu Jul 26 17:39:34 2012 +0100 sna/gen4: Tweak heuristics for render/blt usage Signed-off-by: Chris Wilson commit c9dd1401615f0ed9492a0c0f547fb37150e013d1 Author: Chris Wilson Date: Thu Jul 26 16:31:16 2012 +0100 sna/gen4: Bump thread counts Signed-off-by: Chris Wilson commit 7f3fdef98c1ab2fa27439c3be9810b7a934017ce Author: Chris Wilson Date: Thu Jul 26 15:39:05 2012 +0100 sna/gen7: IVB requires a complete pipeline stall when changing blend modes Similar to how SandyBridge behaves, I had hoped that with IvyBridge they would have made the pipelined operation actually pipelined, but alas. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=52473 Signed-off-by: Chris Wilson commit 0938b3df8c25178c8ea0012e1ead1061d03a4e7c Author: Chris Wilson Date: Thu Jul 26 15:21:59 2012 +0100 sna/dri: Add an explanatory assertion Signed-off-by: Chris Wilson commit c621183466aa55a5938027b702069e792df2272d Author: Chris Wilson Date: Thu Jul 26 15:18:56 2012 +0100 sna/dri: Tidy fallback/normal CopyRegion Signed-off-by: Chris Wilson commit 52d2491a1bafb979d79bb970027c55788f199acb Author: Chris Wilson Date: Thu Jul 26 15:19:14 2012 +0100 sna/video: Protect against attempting to use TexturedVideo whilst wedged Signed-off-by: Chris Wilson commit 8c0e3bbb0c301d0fa4652aa38edd84a9fd6b555e Author: Chris Wilson Date: Thu Jul 26 14:58:04 2012 +0100 sna: Force the damage on the DRI pixmap to be flushed to the GPU Otherwise nothing will happen if we consider ourselves wedged. Signed-off-by: Chris Wilson commit f50d7b03b5d0b81b24f0acb4ae555545bbdaa179 Author: Chris Wilson Date: Thu Jul 26 14:47:44 2012 +0100 sna: Only recommend not to create bo for a pixmap whilst wedged This allows us to continue to map a GPU bo and operate inplace if we are force to create a GPU bo for a compositor. Signed-off-by: Chris Wilson commit 49403ddd1bd9a23e1b32a10e7d0757ae2897a579 Author: Chris Wilson Date: Thu Jul 26 14:44:14 2012 +0100 sna/dri: Mark a flush required for any new DRI already on exec/dirty lists Signed-off-by: Chris Wilson commit efe3c8ff48738f2a274e1d4514d32499fc4aadbc Author: Chris Wilson Date: Thu Jul 26 14:43:33 2012 +0100 sna: Allow DRI to force allocation of a GPU bo and backing pixmap Signed-off-by: Chris Wilson commit 2069384f9f06d3ef9dbb0c3f2c64cac4b24e10fc Author: Chris Wilson Date: Thu Jul 26 14:15:45 2012 +0100 sna/dri: Implement fallback handling for CopyRegion whilst wedged Signed-off-by: Chris Wilson commit b18143de47e060b67a46d9c68590a2d35df9fca6 Author: Paul Menzel Date: Thu Jul 26 12:51:57 2012 +0200 NEWS: Correct release version: s/2.12.0/2.20.0/ Signed-off-by: Paul Menzel Signed-off-by: Chris Wilson commit c262d02fb5defe9100df54cf83cc00e11e335745 Author: Chris Wilson Date: Thu Jul 26 01:12:11 2012 +0100 Limit PCI matching to VGA devices Fixes X -configure Signed-off-by: Chris Wilson commit ad6355311b8b80777bc0fec95f6bf6cd1b4969d9 Author: Chris Wilson Date: Thu Jul 26 01:02:53 2012 +0100 sna: Compile against xorg-1.10 Signed-off-by: Chris Wilson commit b6c7c490d76c683b7b5c89d20f902603b85bd3bc Author: Chris Wilson Date: Thu Jul 26 00:48:59 2012 +0100 sna: Check for a NULL scanout after DPMS off with shadow enabled We may mark the scanout as detached when all outputs are turned off (for example during rotation) and so in the subsequent block handler we need to be careful in case we are handling a detached shadow. Reported-by: chr.ohm@gmx.net Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=52514 Signed-off-by: Chris Wilson commit f4acc01bb09f68edbad4c6cc7e04c271c92661c1 Author: Chris Wilson Date: Wed Jul 25 22:43:32 2012 +0100 sna: Restore inplace upload for DRI compositors With a DRI compositor we have to flush the output after every request, which leads to major inefficiencies. This can be mitigated slightly if we know we will have to upload shortly, which we track using the pixmap flush flag. In particular PutImage is meant to upload inplace to an active DRI buffer, however this was accidentally dropped in commit a253c95ec63b2b075e66ae7380fed6a73469eba5 Author: Chris Wilson Date: Sun Jul 15 13:32:35 2012 +0100 sna: Prefer uploads to be staged in snoopable bo Performace of putimage500 on ivb i7-3720qm: bare: 4610 gnome-shell: 3000 patched: 3390 Reported-by: Michael Larabel Signed-off-by: Chris Wilson commit 5c969a05ef815b261e157fe8d1172aebfd7f5841 Author: Chris Wilson Date: Wed Jul 25 20:28:41 2012 +0100 sna/gen7: Move the PS thread count definition into the constant struct Signed-off-by: Chris Wilson commit 901bb618215d65747eb0a8d481c77c1201f69362 Author: Chris Wilson Date: Wed Jul 25 20:13:15 2012 +0100 sna/gen7: Remove duplicated constants Signed-off-by: Chris Wilson commit 8c5077e4ed055a97bf9deda59c0e9a45e42317ca Author: Chris Wilson Date: Wed Jul 25 16:59:11 2012 +0100 Assume all unknown chipsets are future gen I think the likelihood of a new product being launched based on a 8xx design is remote enough not to worry about. Signed-off-by: Chris Wilson commit 954d9c6aca792098cb16a46b4ec8d5fc104b15a9 Author: Chris Wilson Date: Wed Jul 25 15:41:29 2012 +0100 sna: Initialise single-shot tile offsets before use As noted by the compiler amidst all the noise. Signed-off-by: Chris Wilson commit 484b072072c8297a87940c9e32097923f0a77c8f Author: Chris Wilson Date: Wed Jul 25 12:22:47 2012 +0100 sna: Add 'gmux_backlight' to list of known devices for backwards compatability Reported-by: Austin Lund References: https://bugs.freedesktop.org/show_bug.cgi?id=52423 Signed-off-by: Chris Wilson commit 0ab6d7a50d37cf4454577cb8c333d4b8683aa054 Author: Chris Wilson Date: Wed Jul 25 12:20:36 2012 +0100 sna: Prefer platform backlight devices over firmware This is in contrast to libbacklight but closer to our original code as we prefer a known custom backlight controller over the ACPI interfaces. As only the ACPI interfaces are marked as firmware, and the custom backlight controllers as platform, we therefore need to prefer platform backlight devices. Reported-by: Austin Lund References: https://bugs.freedesktop.org/show_bug.cgi?id=52423 Signed-off-by: Chris Wilson commit b6d82ab07661aba98963f239f9501b50c3910962 Author: Chris Wilson Date: Wed Jul 25 10:40:07 2012 +0100 sna: Reduce ping-pong for Composite with render disabled Signed-off-by: Chris Wilson commit de707b7dc853a3b315ead9789d5ca541412c99bc Author: Chris Wilson Date: Wed Jul 25 09:05:46 2012 +0100 uxa: Add Apple's gmux to the list of known preferred backlights Reported-by: Austin Lund Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=52423 Signed-off-by: Chris Wilson commit 3d254e4010d0753f433cfe62c6a7546b02482847 Author: Chris Wilson Date: Mon Jul 23 23:45:33 2012 +0100 sna: Use SETUP_BLT to reduce overheads for tiled BLT Signed-off-by: Chris Wilson commit 2ed44149eaa27b9632ec83a776f3ee67b0eec7b0 Author: Chris Wilson Date: Mon Jul 23 23:45:33 2012 +0100 sna: Use SETUP_BLT to reduce overheads for stippled BLT Signed-off-by: Chris Wilson commit c1785aaca2c9347114d28f114ee59ef8206d829b Author: Chris Wilson Date: Mon Jul 23 23:00:25 2012 +0100 sna: Remove a surplus function Signed-off-by: Chris Wilson commit 554fce8a65b2518cae032f1eadf58830559299c8 Author: Chris Wilson Date: Mon Jul 23 22:59:16 2012 +0100 uxa: Pass the correct parameters to intel_detect_chipset() Signed-off-by: Chris Wilson commit 40d90dfd8674c255a45b46bbdc09d497af5b3f50 Author: Chris Wilson Date: Mon Jul 23 21:55:46 2012 +0100 intel: Refactor the common chipset detection/override Reduce the duplicate messages for which type of chip we by amalgamating the common code. Signed-off-by: Chris Wilson commit e3f6c48d18c316899c71b6fc34971039c6f9e5f8 Author: Chris Wilson Date: Mon Jul 23 14:43:23 2012 +0100 sna: Refactor PutImage to avoid calling drawable_gc_flags() too early drawable_gc_flags() asserts that the gc has been moved to the CPU prior to its calls so that it can read the reduced raster operation. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 1af26ea4228a9d7768b475b4f9164d2c7620d4fd Author: Chris Wilson Date: Mon Jul 23 14:39:44 2012 +0100 sna: Flesh out tiled operations using the BLT Before enabling the RENDER pipeline for this operation, let's just see what is required to fully use the BLT pipeline as well. Signed-off-by: Chris Wilson commit ac182a006732525a921a9c539e5ebfb537ad3b52 Author: Chris Wilson Date: Mon Jul 23 11:05:05 2012 +0100 sna: Hold a reference to the full stipple pattern for repeated tiles Signed-off-by: Chris Wilson commit 83f683b47063eab8cfb5037d02133dd977c3fc25 Author: Chris Wilson Date: Sun Jul 22 23:20:23 2012 +0100 2.20.1 release A good brown paper bag bug release for SNA. Signed-off-by: Chris Wilson commit 9402bdcc13f7e96dfe527ff4a3da8d13a7870a02 Author: Chris Wilson Date: Sun Jul 22 22:14:52 2012 +0100 sna/glyphs: Also discard the glyph mask for bitmaps with an opaque source Though I expect all such glyphs to be caught by the non-overlapping checks... Signed-off-by: Chris Wilson commit b315e0ebb75d8391ebef7ebe53741a5e33c968bb Author: Chris Wilson Date: Sun Jul 22 20:07:38 2012 +0100 sna: Tweak the fallback hints for XYPixmap PutImage As the fallback uses a multiple-pass algorithm updating one plane at a time, we wish to prepare the fallback surface for reads. Signed-off-by: Chris Wilson commit 8acaf2693e176a92993a498683f121cfe0343fd4 Author: Chris Wilson Date: Sun Jul 22 12:28:34 2012 +0100 sna: Promote tiled operations to the GPU if the tile is already on the GPU Signed-off-by: Chris Wilson commit 7d4a3e371beea65bf66e54ae13789d6d5ca91f8b Author: Chris Wilson Date: Sun Jul 22 11:19:13 2012 +0100 sna: Use an upload buffer for large stipples Signed-off-by: Chris Wilson commit 40e0cf32a25e43e16184b2af87a1e1abeb8e4052 Author: Chris Wilson Date: Sat Jul 21 19:43:23 2012 +0100 sna/dri: We fail at predicting the flip frame Simply report the values from the kernel, and transfer the blame... Signed-off-by: Chris Wilson commit c6e316eeba3008b351f2cd63829154f4672c5417 Author: Chris Wilson Date: Sat Jul 21 18:18:32 2012 +0100 sna: Adjust hints to prefer rendering convex polygon with the GPU Keep the general polygons as only using the GPU if necessary, until the cost of the routines is analysed. Signed-off-by: Chris Wilson commit f1e7248cb353d634f27d297059911168ce1a0762 Author: Chris Wilson Date: Sat Jul 21 16:08:31 2012 +0100 sna: Expand the heuristic for predicting when to use CPU bo for readback For tiny transfers, the cost of setting up the GPU operation outweighs the actual savings through increased throughput. So we try to guess when it will be preferrable to simply read from the GPU bo directly. Signed-off-by: Chris Wilson commit 06db69c2c7023f702f9773be90144fdf7a1159e4 Author: Chris Wilson Date: Sat Jul 21 15:28:10 2012 +0100 sna: Update assertion for cached io buffers As kgem_buffers may be reused and repurposed through the snoop cache it is no longer true that only proxies will have the io flag set. Signed-off-by: Chris Wilson commit d715e1e01437049e167462281d51b5e214594361 Author: Chris Wilson Date: Sat Jul 21 15:04:31 2012 +0100 sna: Also discard the last-was-cpu flag when overwriting cpu damage We interpret a FillRect that erradicates the existing damage as a clear-event and an opportunity to see if it is worth migrating the render commands to the GPU. This is undermined if we leave the 'prefer-cpu' flag intact. Signed-off-by: Chris Wilson commit e95825d17ce65ad8173a5e6518a98969e236a4f8 Author: Chris Wilson Date: Sat Jul 21 14:38:17 2012 +0100 sna: Enable snooping on the reused linear buffer This explains why suddenly I was seeing clflush again on a couple of machines... Signed-off-by: Chris Wilson commit c5e6b5874f334b9124a17f017c6eb175cf88f115 Author: Chris Wilson Date: Sat Jul 21 14:26:52 2012 +0100 sna: Fix the reversed not SHM assertion Should be double checking that we are not about to free a CPU bo pinned to a SHM pixmap. Signed-off-by: Chris Wilson commit f36b656ab2bc16ec8849cadb0afb574bb742c3a3 Author: Chris Wilson Date: Sat Jul 21 13:18:40 2012 +0100 sna: Avoid marking io/snoop buffers as unreusable unnecessarily As they are kept in special caches, we can reserve the unreusable flags for exceptional buffers. Signed-off-by: Chris Wilson commit 37dfdb0e9e86effc3ca8b590c98aa2382e8f0cea Author: Chris Wilson Date: Sat Jul 21 13:01:09 2012 +0100 sna: Correct assertion for __kgem_bo_size() Only proxies are measured in bytes not pages. Signed-off-by: Chris Wilson commit 83ad661bc73e9d0094b669c5203e25afc3937bb7 Author: Chris Wilson Date: Sat Jul 21 12:51:41 2012 +0100 sna: Change the vmap interface name to userptr This is in common with the other drivers and avoids the conflict with 'vmalloc/vmap' used by the kernel for allocation of contiguous virtual mappings. Signed-off-by: Chris Wilson commit 8dcccd308222bcf1b96f2ee15842b4558ea5f29e Author: Chris Wilson Date: Sat Jul 21 12:47:21 2012 +0100 sna: s/vmap/snoop/ since we use the flag more generically Signed-off-by: Chris Wilson commit 6acc9e6a6e1de2a11597c810e02f793774cef2dd Author: Chris Wilson Date: Sat Jul 21 12:07:46 2012 +0100 sna: Fix role reversal of __kgem_bo_size() and kgem_bo_size()! Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 286b0e1a48cab85191dfbb112c8dd14aeaa70956 Author: Chris Wilson Date: Fri Jul 20 16:04:37 2012 +0100 sna: Refresh experimental userptr vmap support Bring the code uptodate with both kernel interface changes and internal adjustments following the creation of CPU buffers with set-cacheing. Signed-off-by: Chris Wilson commit 93c794eb3f80bef64f1619986a7c950229dc7a47 Author: Chris Wilson Date: Fri Jul 20 20:34:53 2012 +0100 sna: Micro-optimise copying boxes with the blitter Signed-off-by: Chris Wilson commit a0d95a9c2d3a27eafbe459e2aefe772c006e596f Author: Chris Wilson Date: Fri Jul 20 20:34:23 2012 +0100 sna: Only update a buffer when it becomes dirty Signed-off-by: Chris Wilson commit c52d265b83b033fb2a275fcc9a8a8d146e3afdf6 Author: Chris Wilson Date: Fri Jul 20 19:38:38 2012 +0100 sna: Tweak CPU bo promotion rules for CopyArea Signed-off-by: Chris Wilson commit f92a64dd9162731210b14368b6ee408356d7fefc Author: Chris Wilson Date: Fri Jul 20 15:54:35 2012 +0100 sna: Only set the vmap flag after we make the bo snoopable Otherwise if we fail then we incorrectly add the handle to the vmap cache. Signed-off-by: Chris Wilson commit 8b4cf24f1403bf3d929cc0725de66b3d0e08ebaf Author: Chris Wilson Date: Fri Jul 20 14:46:32 2012 +0100 sna: Also check whether the first upload box can use the BLT No point checking boxes 1..n if box 0 is the troublemaker! Signed-off-by: Chris Wilson commit df14b285be44f0c40a718bb8ae09a9558b1eb2c7 Author: Chris Wilson Date: Fri Jul 20 14:35:28 2012 +0100 sna/gen6: Prefer the more flexible render ring for large surfaces Signed-off-by: Chris Wilson commit 578ff11c3753ede2c81afc47302991e3d3b316f2 Author: Chris Wilson Date: Fri Jul 20 14:24:06 2012 +0100 sna: Just use composite.box() when we only have one box Signed-off-by: Chris Wilson commit fb7987fc0b51cf3b83dcf78bcefe65ec3af32ccf Author: Chris Wilson Date: Fri Jul 20 13:12:27 2012 +0100 sna/dri: Cleanup ring selection for SNB+ CopyRegion Signed-off-by: Chris Wilson commit 3b56588fbaa2c4ccdfb2f2a8f5656d2cda9dacd7 Author: Chris Wilson Date: Fri Jul 20 10:19:25 2012 +0100 sna: Update WIP userptr example usage Signed-off-by: Chris Wilson commit 473a1dfb683ed576d86b37aba36aaa0e379f4606 Author: Chris Wilson Date: Fri Jul 20 09:56:13 2012 +0100 sna: Rename kgem_partial_bo to kgem_buffer Signed-off-by: Chris Wilson commit 8e6e8a2fa8adda9ae9be8a88fbb14851e9d2df2e Author: Chris Wilson Date: Fri Jul 20 09:51:46 2012 +0100 sna: Allow the snoopable upload buffer to take pages from the CPU vma cache Signed-off-by: Chris Wilson commit 979035bb9ce04db5fe30efa4f6daab0a40f6af57 Author: Chris Wilson Date: Thu Jul 19 20:06:28 2012 +0100 sna: Remove topmost unused 'flush' attribute Signed-off-by: Chris Wilson commit b83011909aaf185f05fc2df743882c2410eff46d Author: Chris Wilson Date: Thu Jul 19 19:51:46 2012 +0100 sna: Replace 'sync' flag with equivalent 'flush' The only difference is in semantics. Currently 'sync' was only used on CPU buffers for shared memory segments with 2D clients, and 'flush' on GPU buffers shared with DRI clients. Signed-off-by: Chris Wilson commit 88bee3caeaacbbb1b4d789ea3db9a3802a62b59d Author: Chris Wilson Date: Thu Jul 19 19:40:34 2012 +0100 sna: Remove unused scanout-is-dirty? flag Signed-off-by: Chris Wilson commit 6f60f89588caa70e7d8ed53ba453bbe8c2094a95 Author: Chris Wilson Date: Thu Jul 19 16:58:34 2012 +0100 sna/gen6: Bump the WM thread count to 80 Note that we should only do this when "WiZ Hashing" is disabled. So we should be checking the GT_MODE register (bring on i915_read!) to be sure that is safe to do so. However, it gives a big boost to performance of render copies... It also causes perf benchmarks to hit thermal limits much quicker. Signed-off-by: Chris Wilson commit fc39d4b5cb105d269c5349e479daf112f5d93580 Author: Chris Wilson Date: Thu Jul 19 16:25:54 2012 +0100 sna/gen6: Add a simple DBG option to limit usage of either BLT/RENDER We can force the code to either select only BLT or RENDER operations - for those that we have a choice for at least! Signed-off-by: Chris Wilson commit 15d3eea7004822e5cbd48d676692e1b6a2b26d3e Author: Chris Wilson Date: Thu Jul 19 16:22:20 2012 +0100 sna: Handle mixed bo/buffers in assertions Signed-off-by: Chris Wilson commit e4fce3b7801038e4f64d848a0995f4b441b4d2aa Author: Chris Wilson Date: Thu Jul 19 10:50:09 2012 +0100 sna/gen4: Hookup composite spans Due to the unresolved flushing bug it is no faster (so only enable when we definitely can't do the operation inplace), however it does eliminate a chunk of CPU overhead. Signed-off-by: Chris Wilson commit 5f138176bf15682324d2e8cfa9fac3e49604bf8f Author: Chris Wilson Date: Thu Jul 19 10:32:33 2012 +0100 sna: Tweak order of screen re-initialisation Signed-off-by: Chris Wilson commit 9bd0f8f3e7783d7a6bab707fc08ec96830cd5809 Author: Chris Wilson Date: Thu Jul 19 09:40:07 2012 +0100 i810: Correct the double negative and enable XAA when available Signed-off-by: Chris Wilson commit d145d0e1459f578eea621e6944814642e5dd431f Author: Chris Wilson Date: Thu Jul 19 09:23:10 2012 +0100 i810: Handle initialisation without the XAA module present at runtime Signed-off-by: Chris Wilson commit 7a3b98e05b706548527e73b2008600391c601a62 Author: Chris Wilson Date: Thu Jul 19 08:42:05 2012 +0100 sna: Re-register the SHM funcs every server generation As the SHM layer hooks into the CloseScreen chain to free its privates, we then need to call the registration function again on the next generation to ensure that the private is reallocated before use. Reported-by: Pawel Sikora Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=52255 Signed-off-by: Chris Wilson commit 4bcab83bbddf8a698aa83f5038f9ab019a404bd5 Author: Chris Wilson Date: Wed Jul 18 22:19:45 2012 +0100 i810: DRI is not dependent upon XAA The blit routines is uses are independent of the XAA driver interface and can be used separately. Signed-off-by: Chris Wilson commit 558c8251299b786cab1ac83dbd35f077224b5950 Author: Chris Wilson Date: Wed Jul 18 22:00:30 2012 +0100 sna/gen4+: Drop unsupported source formats Once again I've confused existence of the enum with the ability of the sampler to read that format. Signed-off-by: Chris Wilson commit 9f3b3098c9f870d303a9de2b9c0db119eff5a865 Author: Chris Wilson Date: Wed Jul 18 21:39:47 2012 +0100 sna/dri: Allow DRI2 to be loaded even if we are wedged Just because the GPU is spitting EIO at us does not necessarily imply that a DRI client will also suffer. Spit out a warning for later bug reporting and let them find out for themselves! Signed-off-by: Chris Wilson commit 15b7191fd363e9e6083844a218e25419695d55f1 Author: Chris Wilson Date: Wed Jul 18 12:59:41 2012 +0100 sna/gen6: Micro-optimise render copy emission Backport of the changes made for IVB. Signed-off-by: Chris Wilson commit 4eea9ac0035dd72f3c637adc39eeaeda46472e9e Author: Chris Wilson Date: Wed Jul 18 12:59:41 2012 +0100 sna/gen7: Micro-optimise render copy emission The goal is bring the overhead down to that of using the blitter. Tricky given the number of steps to using the 3D pipeline compared to the BLT... A stretch goal would be to make IVB GPU bound for -copywinpix10! Signed-off-by: Chris Wilson commit 267429bbb146449ee4d3b88fa8e23c5b1d53470a Author: Chris Wilson Date: Wed Jul 18 19:24:44 2012 +0100 sna: Enable runtime detection of set-cacheing ioctl Signed-off-by: Chris Wilson commit c0b3674d042ff55d64ad1fd0d64926e1967be323 Author: Chris Wilson Date: Wed Jul 18 20:48:27 2012 +0100 sna/trapezoids: Only reduce bounded operators to a single pass Only for a few operators can we replace the opacity mask by premultiplying into the source. Signed-off-by: Chris Wilson commit bb0303677c38076db14dfbceec3636197a971e8c Author: Chris Wilson Date: Wed Jul 18 10:40:50 2012 +0100 sna/trapezoids: Use pixman from within the spans to reduce two-pass operations Reduce the two pass CompositeTrapezoids if we can perform the operation inplace by calling pixman_image_composite from the span. This step enables this for xrgb32. Signed-off-by: Chris Wilson commit bee1a14618797b3d3a1c1a20eb72644fa907c048 Author: Chris Wilson Date: Wed Jul 18 09:38:32 2012 +0100 sna: Fix processing of the last fallback box The evil typo caused us to misalign the clip boxes and run over a garbage array on 64-bit builds. Reported-by: Edward Sheldrake Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=52163 Signed-off-by: Chris Wilson commit 88cb1968b6dbf3edfa885da9503e91124af46007 Author: Chris Wilson Date: Wed Jul 18 09:38:03 2012 +0100 sna: Add more DBG for fallback processing Hunting the lost box... Signed-off-by: Chris Wilson commit 36f2e46619598e9bca4fe1207aa2f157bfa1ecf4 Author: Chris Wilson Date: Wed Jul 18 00:45:54 2012 +0100 sna: Reuse the snoopable cache more frequently for upload buffers Now that we are keeping a small cache of snoopable buffers, experiment with using them for uploads more frequently. Signed-off-by: Chris Wilson commit 73f07abbd2d78418e5a66262f293b5ed80b7ccb4 Author: Chris Wilson Date: Wed Jul 18 00:19:49 2012 +0100 sna: Maintain a short-lived cache of snoopable CPU bo for older gen Once again, we find that frequent buffer creation and manipulation of the GTT is a painful experience leading to noticeable and frequent application stalls. So mitigate the need for fresh pages by keeping a small stash of recently freed and inactive bo. Signed-off-by: Chris Wilson commit 77520641a332a622c0b5378bd254ed5cb46a5f0a Author: Chris Wilson Date: Tue Jul 17 22:09:33 2012 +0100 i810: Replace XAAGet.*ROP() with local tables The XAAGetPatternROP() and XAAGetCopyROP() functions were removed along with the rest of XAA so we need to implement those tables locally. Reported-by: Knut Petersen Signed-off-by: Chris Wilson commit caef63e0268e59e439b030a9a338e81d5cf8e311 Author: Chris Wilson Date: Tue Jul 17 21:22:57 2012 +0100 i810: Split xaa routines from common acceleration methods Some of the routines in i810_accel.c are specific to XAA whilst others are used elsewhere, for example in i810_dri.c. Therefore we have to be selective over which ones we compile out without xaa. Reported-by: Knut Petersen Signed-off-by: Chris Wilson commit 53ff19f45a3cc4863845c23e8d3c2c2b95e03fd9 Author: Chris Wilson Date: Tue Jul 17 19:40:16 2012 +0100 sna: Allow wedged CopyPlane to operate inplace on the destination Signed-off-by: Chris Wilson commit d4fa4d5494db45b227c9ae7f7a90cd5dfd940027 Author: Chris Wilson Date: Tue Jul 17 19:29:32 2012 +0100 sna: Allow inplace copies for wedged CopyArea Signed-off-by: Chris Wilson commit 217eeadf81a8cbb43e495e1e937acdd95c703377 Author: Chris Wilson Date: Tue Jul 17 19:35:06 2012 +0100 sna: Allow operation inplace to scanout whilst wedged Signed-off-by: Chris Wilson commit 40ff29480a0dbf458adf1a1b0d3275ad1361530e Author: Chris Wilson Date: Tue Jul 17 18:38:49 2012 +0100 sna: Tweak fast blt path Signed-off-by: Chris Wilson commit fce69c79c4840e7863d7c382da0d22be90a9f19a Author: Chris Wilson Date: Tue Jul 17 18:28:24 2012 +0100 sna: prefer fbBlt over pixman_blt It is currently much better optimised through memcpy. Signed-off-by: Chris Wilson commit c29f96d50839388377ad57c6366f9bc7ad8b9d0a Author: Chris Wilson Date: Tue Jul 17 13:02:51 2012 +0100 sna/gen7: Bump the number of pixel shader threads for IVB GT2 Spotted-by: Kilarski, Bernard R" Signed-off-by: Chris Wilson commit 799bae9e8ff53fb1b5c74f3278d530a58d66de9a Author: Chris Wilson Date: Tue Jul 17 17:07:10 2012 +0100 sna/dri: Do not allow an exchange to take place on invalid buffers If the SwapBuffers is called after we have resized a Window but before the client has processed the Invalidate notification, then the SwapBuffers will be referring to a pair of stale buffers. As the buffers are no longer attached to the Pixmap, we can not simply exchange them. Signed-off-by: Chris Wilson commit 067aeaddb8047f01ae3a20b26ba0acf5ba2d035f Author: Chris Wilson Date: Tue Jul 17 15:33:19 2012 +0100 sna: Rebalance choice of GPU vs CPU bo Signed-off-by: Chris Wilson commit 7ebeea3f5c71959773478de44b08a967fe5acc8b Author: Chris Wilson Date: Tue Jul 17 15:10:43 2012 +0100 sna: Avoid the CPU bo readback for render paths As we exclude using the CPU bo if there is overlapping GPU damage, we can forgo the call to keep the transfer the damage. Signed-off-by: Chris Wilson commit ed8c729ed02705fd03be1ab22a94b5aae13567c8 Author: Chris Wilson Date: Tue Jul 17 11:21:30 2012 +0100 sna: Catch the short-circuit path for clearing clear on move-to-gpu as well I thought the short-circuit path was only taken when already clear, I was wrong. Signed-off-by: Chris Wilson commit 359b9cc82de13b0ac89692896ac6104ff3be308b Author: Chris Wilson Date: Tue Jul 17 10:26:27 2012 +0100 sna: Limit the use of snoopable buffers to read/write uploads Signed-off-by: Chris Wilson commit 4f21dba6ee505217d63edd84611622e05aeb4593 Author: Chris Wilson Date: Tue Jul 17 09:26:46 2012 +0100 sna: Only drop the clear flag when writing to the GPU pixmap Signed-off-by: Chris Wilson commit fbfbbee8288aba1e4754fd2dbc02e71f5e118cda Author: Chris Wilson Date: Tue Jul 17 09:20:21 2012 +0100 sna: Fix glyph DBG to include clip extents and actual glyph origin Signed-off-by: Chris Wilson commit f0ed0ca234a4bed986824845ff70e8554c0e579f Author: Chris Wilson Date: Tue Jul 17 08:35:20 2012 +0100 sna: Promote an undamaged pixmap to use the full GPU Signed-off-by: Chris Wilson commit 1f79e877fb6602bd0f9dd14ac9c3511f3b7044fb Author: Chris Wilson Date: Mon Jul 16 21:18:24 2012 +0100 sna: Share the pixmap migration decision with the BLT composite routines Signed-off-by: Chris Wilson commit d141a2d59007866c9eaad020c744be446e70c346 Author: Chris Wilson Date: Mon Jul 16 23:20:58 2012 +0100 sna: Disable snoopable bo for gen4 Further inspection reveals that whilst it may not hang the GPU, the results are not pleasant or complete. Signed-off-by: Chris Wilson commit 107feed2a4ca044313c70f83a62909187ff1f905 Author: Chris Wilson Date: Mon Jul 16 18:58:30 2012 +0100 sna: Disable snoopable uplaod buffers for gen4 The sampler really does not like using snoopable buffers... Signed-off-by: Chris Wilson commit 818c21165c746b7b410a6e6e23b1675d88db685d Author: Chris Wilson Date: Mon Jul 16 16:28:00 2012 +0100 sna: Fixup pixmap validation for sna_copy_area() Remember to offset the box by the drawable deltas in order to compensate for compositing. Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=52142 Signed-off-by: Chris Wilson commit 623d84bed7c47ac39348775ce35eec54196f6dac Author: Chris Wilson Date: Mon Jul 16 16:07:37 2012 +0100 Wrap defines to avoid redefinition warnings Currently this only catches out ARRAY_SIZE, but wrap the other common defines for consistency. Signed-off-by: Chris Wilson commit 907a2a7c97514d3f7610648ed87c7042a857f786 Author: Chris Wilson Date: Mon Jul 16 14:47:03 2012 +0100 sna/trapezoids: Fix inplace unaligned fills (on gen4) Reported-by: Sergio Callegari Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=52150 Signed-off-by: Chris Wilson commit 6ce2f40249231f57cf464361ea5329cee1932ccf Author: Chris Wilson Date: Mon Jul 16 14:46:39 2012 +0100 sna/trapezoids: Add some DBG to unaligned fills Signed-off-by: Chris Wilson commit 2721214868685123c973970a8ce0d93346ae0ee2 Author: Chris Wilson Date: Mon Jul 16 13:52:39 2012 +0100 sna: Move the disabling of CPU bo for gen4 to the render unit They appear to work fine with the BLT and only seem to cause issues when used with the sammpler. So enable them for accelerated uploads and downloads. Signed-off-by: Chris Wilson commit 0777b146bf1a63c99e4d4af141e676a47b1f2dc9 Author: Chris Wilson Date: Mon Jul 16 13:11:07 2012 +0100 sna: Use set-cache-level to allocate snoopable upload buffers Signed-off-by: Chris Wilson commit 33443f7ee48fa54b6f4d09c93cddac0e32314b9c Author: Chris Wilson Date: Mon Jul 16 12:51:54 2012 +0100 sna: Add a couple of DBG options to control accelerated up/downloads Signed-off-by: Chris Wilson commit 924060293826a1cc0d9d7bc26e913e46c6b2d054 Author: Chris Wilson Date: Mon Jul 16 12:03:47 2012 +0100 sna: Discard and recreate the CPU buffer when busy during move-to-cpu Signed-off-by: Chris Wilson commit 7024ef771ff170e61e788b5216c86b46e0f8ae6a Author: Chris Wilson Date: Mon Jul 16 12:15:54 2012 +0100 sna: Add a few DBG to show when CPU bos are being used for xfer Signed-off-by: Chris Wilson commit c564414157e27417f0de1c0542dafd9b47e01eda Author: Chris Wilson Date: Mon Jul 16 09:34:30 2012 +0100 sna: Disable the scanout flush when switch off via DPMS Signed-off-by: Chris Wilson commit 536e7ab756d6821db79e4cd79a250af1c0f7d5a3 Author: Chris Wilson Date: Mon Jul 16 12:16:26 2012 +0100 intel: Don't use stdbool without declaring it Reported-by: Fabio Pedretti Signed-off-by: Chris Wilson commit 0c32be15b06ad63c1fc1371de879f2d879080f6b Author: Chris Wilson Date: Thu Jul 5 19:26:48 2012 +0100 uxa: Remove Shadow hack This was an incomplete hack so deprecate in favour of Shadow-on-Steriods, SNA. References: https://bugs.freedesktop.org/show_bug.cgi?id=47324 Signed-off-by: Chris Wilson commit 6a18a0936eafc45ab920ab0eecf2fc2a601c41a7 Author: Chris Wilson Date: Sun Jul 15 20:26:00 2012 +0100 configure: version bump for 2.20.0 release The day SNA hits mainstream... Signed-off-by: Chris Wilson commit 6fa059330decd1437eef4928d732ec91fd4e17e7 Author: Chris Wilson Date: Sun Jul 15 14:04:48 2012 +0100 sna: Avoid creating a CPU bo for uploads if we already have a large GPU bo And vice-versa if already have a large CPU bo. Signed-off-by: Chris Wilson commit a253c95ec63b2b075e66ae7380fed6a73469eba5 Author: Chris Wilson Date: Sun Jul 15 13:32:35 2012 +0100 sna: Prefer uploads to be staged in snoopable bo Signed-off-by: Chris Wilson commit ef6d94a8444927941db108811e1a26357dc3f18e Author: Chris Wilson Date: Sun Jul 15 11:46:53 2012 +0100 sna: Simply reverse all the boxes if dx <= 0 and dy <= 0 In this fairly common case, avoid both the double pass and use a simpler algorithm as we can simply reverse the order of the boxes. Signed-off-by: Chris Wilson commit 6601a943ff968ac39ba198351c50dc883cb4232e Author: Chris Wilson Date: Sun Jul 15 11:39:56 2012 +0100 sna: Keep track of the base pointer for the reordered boxes So that we avoid freeing an invalid pointer. Signed-off-by: Chris Wilson commit ef34d5cf415ad7459ab44b0ec2e70b14150735fc Author: Chris Wilson Date: Sat Jul 14 23:25:17 2012 +0100 sna: Make sure we check for a busy CPU bo before declaring is-cpu Even if the pixmap is entirely damaged on the CPU, we still may be in the process of transferring it and so cause an unwanted stall. Signed-off-by: Chris Wilson commit 0e397e4a1dc23ed07089c967612d705584f3b376 Author: Chris Wilson Date: Sat Jul 14 21:35:23 2012 +0100 sna/glyphs: Perform the fallback mask reduce before moving the glyph pointers Signed-off-by: Chris Wilson commit db1ee13a53b0c1348b7566ee60ee1b7b384ef59a Author: Chris Wilson Date: Sat Jul 14 21:26:01 2012 +0100 sna/gen7: Check for gradient allocation failure within spans Signed-off-by: Chris Wilson commit 86479e97460da798a3804cbb4ae39e62de881af1 Author: Chris Wilson Date: Sat Jul 14 21:24:04 2012 +0100 sna/gen7: Uses EXTEND_PAD for its alpha-gradient with spans Fixes regression from 2b94f9a043. Signed-off-by: Chris Wilson commit f17037275c05198c3c3f456964fd42032f9085b6 Author: Chris Wilson Date: Sat Jul 14 20:11:17 2012 +0100 sna: Reorder overlapping boxes for CopyArea/Window Signed-off-by: Chris Wilson commit 86e09d14bd00344d378b86a19ebb44f7d946926c Author: Chris Wilson Date: Sat Jul 14 17:51:31 2012 +0100 sna: Tidy sna_copy_boxes So there appears to be a bug hidden here. But only when we scroll upwards in a GTK+ application. Hmm. Signed-off-by: Chris Wilson commit 924f596463555db27214fd8227218c2e21ecddc8 Author: Chris Wilson Date: Sat Jul 14 19:03:08 2012 +0100 sna: Avoid BLT to snoopable bo on older gen Signed-off-by: Chris Wilson commit 44e226b1d9fca8cb95b0864adf8708b03ee8472c Author: Chris Wilson Date: Sat Jul 14 18:36:29 2012 +0100 sna/glyphs: A repeat of the earlier typo for pixman glyphs Signed-off-by: Chris Wilson commit 797ebf937fcfcc87502727c70e6b52f89fecc799 Author: Chris Wilson Date: Sat Jul 14 16:58:18 2012 +0100 sna/glyphs: Fix typo and render glyphs to the small mask Signed-off-by: Chris Wilson commit 2b94f9a043372ffede01339eea99377a71169fbc Author: Chris Wilson Date: Sat Jul 14 16:38:21 2012 +0100 sna/gen7: Preselect sampler Signed-off-by: Chris Wilson commit 50d61c38bdb720c17ba5bfcaf97032338e466d9b Author: Chris Wilson Date: Sat Jul 14 16:07:32 2012 +0100 sna/gen4: Fix typo introduced in checking dst format Signed-off-by: Chris Wilson commit fc91b1f157f9e6e4097e662e2890574e530344ac Author: Chris Wilson Date: Sat Jul 14 15:48:50 2012 +0100 sna/gen7: Only emit the render flushes between operations Signed-off-by: Chris Wilson commit d30dc59651f156e9d4572141e2c54af89d007a37 Author: Chris Wilson Date: Sat Jul 14 15:42:19 2012 +0100 sna/gen7: Only fallback to BLT for a redirected target As we can always use a partial surface for the source. Signed-off-by: Chris Wilson commit 9f05b0c03b1fdab474b436431c430028fca1937e Author: Chris Wilson Date: Sat Jul 14 15:03:32 2012 +0100 sna/gen4+: Fix up card/dest format confusion in previous commit That would have been much more successful had I not supplied the wrong opaque formats to the sampler. Signed-off-by: Chris Wilson commit fffbc34e4621898eee9b80bf8b6b3699bcade52a Author: Chris Wilson Date: Sat Jul 14 14:54:27 2012 +0100 sna/gen4+: Translate to card format using a switch GCC produces faster code than a walk over the format tables. Signed-off-by: Chris Wilson commit 95fdd9af5c8a8360d02debc400e75869c36f05ca Author: Chris Wilson Date: Sat Jul 14 13:05:27 2012 +0100 sna: Enable pixman_glyphs if available Signed-off-by: Chris Wilson commit 3f764ee4c50567cfb831495d42cb6c2bb94055ad Author: Chris Wilson Date: Sat Jul 14 12:29:47 2012 +0100 sna/dri: Presume an unknown chipset will want i965_dri.so Signed-off-by: Chris Wilson commit 39845280b5a9d0ed60c6a158c3d0df9fb8756d40 Author: Chris Wilson Date: Sat Jul 14 12:14:28 2012 +0100 sna: Allow BLT support on future unknown hardware Assume that if the kernel recognises the chipset and declares it has a BLT, we can use it. Signed-off-by: Chris Wilson commit b260ca44b3aaba2c8cd25640ad7ac9ca6478b0f2 Author: Chris Wilson Date: Sat Jul 14 09:59:07 2012 +0100 Drop some unused includes Signed-off-by: Chris Wilson commit f517bdb12b909ef9d0897166bdabc537551a065b Author: Chris Wilson Date: Sat Jul 14 08:56:21 2012 +0100 sna: Do not perform batch compaction on old kernels As they will reject the delta pointing outside of the target batch. Signed-off-by: Chris Wilson commit ce69a1e8686889f1eebb4cb3a39c41f473e58b93 Author: Chris Wilson Date: Fri Jul 13 23:25:02 2012 +0100 sna: Add some DBG for selecting glyph path Signed-off-by: Chris Wilson commit b68b2d90915f7a97e43a39e3117927544886fba0 Author: Chris Wilson Date: Fri Jul 13 23:24:32 2012 +0100 sna: Add some DBG for short-circuiting choice of source bo Signed-off-by: Chris Wilson commit ad4786b285074b5bd70b3ad0bea6ec1b77ad6740 Author: Chris Wilson Date: Fri Jul 13 22:24:51 2012 +0100 sna: Aim for consistency and use stdbool except for core X APIs Signed-off-by: Chris Wilson commit a05c3547bba52288bae872ea672ffe2f4dab2ffa Author: Chris Wilson Date: Fri Jul 13 15:06:15 2012 +0100 sna/gen4: Simplify comparing the pipeline-pointers against the previous Signed-off-by: Chris Wilson commit ea9ec18505645dfec85ab96bbbbbc1793830c737 Author: Chris Wilson Date: Fri Jul 13 11:39:01 2012 +0100 sna: Split CloseScreen into two phases In order to get the ordering correct we need to free the xf86_cursors before calling the miPointerCloseScreen. This requires us to insert a hook at the top of the CloseScreen chain. However we still require the final CloseScreen hook in order to do the fundamental clean up, hence split the CloseScreen callback into two phases. Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 0f03f7b4359fcbcde651bc1554ddff4fe10bc53b Author: Chris Wilson Date: Fri Jul 13 10:35:27 2012 +0100 sna: Only try to enable DRI2 if the module is available at runtime Blatantly cribbed from Michel Danzer's patch for ati... Signed-off-by: Chris Wilson commit b5d6a57f12025aef9850c7d9baa6905f776be971 Author: Chris Wilson Date: Thu Jul 12 22:16:27 2012 +0100 Enable compilation of SNA by default But only if we meet the required versions of Xorg and leave UXA as the default AccelMethod for the time being. Signed-off-by: Chris Wilson commit 6c2975ab2943478b3a246b5fb231f9f3df2d8475 Author: Chris Wilson Date: Thu Jul 12 22:14:39 2012 +0100 Fix the reporting of whether SNA is configured Signed-off-by: Chris Wilson commit fd15ce65ab8ce5cf571e37daa7db1ee245616cd4 Author: Chris Wilson Date: Thu Jul 12 20:32:14 2012 +0100 sna: Fix build without DRI2 Signed-off-by: Chris Wilson commit 22be9988b933f33fc5247a9abc3b00a7f2e4a202 Author: Chris Wilson Date: Thu Jul 12 19:49:34 2012 +0100 sna: Check for failure to initialize the sprite pointers Signed-off-by: Chris Wilson commit 32e7f4ee64867779b2def6fcd882708d7b0e2cf5 Author: Chris Wilson Date: Thu Jul 12 19:39:10 2012 +0100 sna/glyphs: Fix array allocation for list_extents Originally I intended to skip assigning the box on the last list. However, loop simplicity failed and now we run the risk of writing beyond the end of stack_extents, and overwriting the list_extents pointer. Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 0477b5fb6f040f3bad86bb314a24df1bcd660aed Author: Chris Wilson Date: Thu Jul 12 19:37:02 2012 +0100 sna/glyphs: Apply mask reduction along fallback paths as well Signed-off-by: Chris Wilson commit 16aaa51b5d326f44974489f5b29716c7ff5ab48e Author: Chris Wilson Date: Thu Jul 12 15:26:54 2012 +0100 sna: Rearrange the tests for dropping the glyph mask Signed-off-by: Chris Wilson commit 600746f923b046187a66b60b5e5d2b187475730e Author: Chris Wilson Date: Thu Jul 12 14:16:31 2012 +0100 i810: Remove vestigial reference to xaa.h i810_hwmc.c can pull it in via i810.h like everybody else. As for xaalocal.h, I have no idea what that is... Both appear to be cut'n'paste includes. Signed-off-by: Chris Wilson commit 1d9ab2e7101167075112a472ee82530dc0365183 Author: Zhigang Gong Date: Thu Jul 12 18:47:50 2012 +0800 uxa/dri: Fix a buffer leak on pageflipping when enabling glamor. We need to put current front_buffer to back buffer thus we don't need to create a new back buffer next time. This behaviou should be the same with or without glamor. Previous code incorrectly discard the previous front_buffer and cause a big buffer leak problem. Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit fa89e0614563cbe0cc45e6d7e7584cb333244c27 Author: Zhigang Gong Date: Thu Jul 12 18:47:51 2012 +0800 uxa/glyphs: Bypass uxa glyphs operations if using glamor. glamor_glyphs will never fallback. We don't need to keep a uxa glyphs cache picture here. Thus simply bypass the corresponding operations. Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit d25000e1f31e78aff2ab43adb12aec0aac36f56f Author: Chris Wilson Date: Wed Jul 11 22:37:25 2012 +0100 i810: Tidy configure detection for XAA/DGA Signed-off-by: Chris Wilson commit 44a1528c15eec9b3f93651e779013137864d4d2b Author: Chris Wilson Date: Wed Jul 11 15:05:32 2012 +0100 sna: Remove unused cached partials list Signed-off-by: Chris Wilson commit 94d489ae43a2c4d4d9ddc9ce30ff1a9142b77d4a Author: Chris Wilson Date: Wed Jul 11 14:25:20 2012 +0100 sna: Minor tweak to upload in place if the CPU bo is busy Since we have to pay the price of the stall anyway... Signed-off-by: Chris Wilson commit b5db90aa52f10897ad2d7795df94c0e3d2878aea Author: Chris Wilson Date: Wed Jul 11 12:52:24 2012 +0100 sna: Discard the mask for "non-overlapping" glyphs If we can acheive the same rasterisation results without the mask, rendering the glyphs-to-dst is so much faster that it outweighs the cost of checking for overlapping glyphs. The penalty is then for code that correctly declared that it required a mask, who now have an extra ~10% overhead in the processing of their glyphs. Signed-off-by: Chris Wilson commit 665c9cbdcf2c493cac29d316eaffa2abe197a183 Author: Chris Wilson Date: Tue Jul 10 19:01:12 2012 +0100 sna: Disable periodic scanout refresh if all outputs are disconnected Signed-off-by: Chris Wilson commit 85fdc3143b157fd7ba3453efc86da8238d9de316 Author: Chris Wilson Date: Tue Jul 10 12:43:46 2012 +0100 sna: Avoid the expensive recomputation of the region when copying Signed-off-by: Chris Wilson commit 747ed0ca28bb46fc45ff20acd1f82212fa2b46bd Author: Chris Wilson Date: Tue Jul 10 17:05:20 2012 +0100 sna: Only promote to full GPU if we already have a GPU bo Signed-off-by: Chris Wilson commit 232217eef8f99f5678d65bf9aa5b898ef6d3b3c6 Author: Daniel Stone Date: Tue Jul 10 00:32:13 2012 +0100 i810: Make DGA optional Don't build DGA when it's not available, or when we don't want it. Signed-off-by: Chris Wilson commit 78dc0c04745ad4485b994f67833f4a155749f01d Author: Daniel Stone Date: Tue Jul 10 00:32:13 2012 +0100 i810: Make XAA optional Don't build XAA when it's not available, or when we don't want it. Signed-off-by: Daniel Stone commit 45ab003a5860fd4290df24739d2520fddfe27a8f Author: Chris Wilson Date: Mon Jul 9 21:28:40 2012 +0100 sna: Check for overlapping glyphs within each list, then overlapping lists Spotted by Zhigang Gong is this optimisation to avoid the problem with multiple lines passed in a single request (using multiple lists). As the start of line will overlap with the previous line when we use the simple bbox comparison, we always declare those runs as overlapping and so we cannot substitute a glyph mask. However, we can reduce the problem to only checking for overlapping glyphs within a list and then checking for overlapping lists. Very, very clever. Signed-off-by: Chris Wilson commit 8066bc33d78e78ce7c13833b08a7daaea2f3ed22 Author: Chris Wilson Date: Mon Jul 9 11:25:11 2012 +0100 sna: Fix reversal of inside logic for BitmapToRegion Signed-off-by: Chris Wilson commit 8a9a17cd092d156272e8953974119f527b2b77eb Author: Chris Wilson Date: Mon Jul 9 11:12:12 2012 +0100 sna: Improve the check for assertions Signed-off-by: Chris Wilson commit b11bc37684181390fc1400afb44054785104ca15 Author: Chris Wilson Date: Mon Jul 9 11:09:16 2012 +0100 sna: Simplify the DBG incarnation It was only ever used in conjunction with HAS_DEBUG_FULL. For debug purposes it is as easy to redefine DBG locally. By simplifying the DBG macro we can create it consistently and so reduce the number of compiler warnings. Long term, this has to be dynamic. Sigh. Signed-off-by: Chris Wilson commit 21798a88676e91049917fafd3196dd4374b94226 Author: Chris Wilson Date: Mon Jul 9 09:23:20 2012 +0100 sna: Promote large operations to use the whole GPU Signed-off-by: Chris Wilson commit 1b6ad7a6ae6820c8f66d1c80613885ee84e7d316 Author: Chris Wilson Date: Mon Jul 9 08:54:11 2012 +0100 sna: Only consider large clears as candidates for GPU migration If we only operating on a small region of the pixmap and have require damage migration in the past, we are likely to require migration again at some point. So keep track of small damage areas. Signed-off-by: Chris Wilson commit d8a75538ea1d2a79d6282b0e2dfd73cfdea1a480 Author: Chris Wilson Date: Mon Jul 9 08:38:39 2012 +0100 sna: PadPixmap only writes to the out-of-bounds bits So we only need to delcare it as reading the source pixmap and not mark it as damaged. Signed-off-by: Chris Wilson commit eafb454edf188e7dada1ddf886d1e46f0151968d Author: Chris Wilson Date: Mon Jul 9 08:38:08 2012 +0100 sna: Rename conflicting symbols with uxa Reported-by: Christoph Reiter Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=51887 Signed-off-by: Chris Wilson commit 0af29175a087cc2e509962f8828790c8f7232611 Author: Chris Wilson Date: Mon Jul 9 01:24:23 2012 +0100 sna: Just use a linear scan to find the terminating clip box Signed-off-by: Chris Wilson commit 2941a5fe15626730869a48a63bb088e8ae2c0549 Author: Chris Wilson Date: Mon Jul 9 00:20:57 2012 +0100 sna: Remove the consideration of CPU damage for overwriting FillRect We consideer a singular FillRect to be a sequence point in the rendering commands, that is it is usually used to clear the background as the first operation in a drawing sequence. So it is useful to ask if we can move the sequence to the GPU at that point. Signed-off-by: Chris Wilson commit 8be00b6d4767ffc09328d246d02ee75312ad5842 Author: Chris Wilson Date: Sun Jul 8 20:48:40 2012 +0100 sna: Substitute the reduce clip region for fallback ops Signed-off-by: Chris Wilson commit 0457935a70f6c1ae06f61f9ad0dd1bdc825465fa Author: Chris Wilson Date: Sun Jul 8 20:38:03 2012 +0100 sna: Review placement hints for fallback operations Look for those operations that maybe better via the GTT and those that are preferred to be in CPU cache. The wonders of multiple layers of heuristics. Signed-off-by: Chris Wilson commit b7f0b0e7e3f66165b87c46f897de536cf74a9daf Author: Chris Wilson Date: Sun Jul 8 19:59:34 2012 +0100 sna: Remove function for force-to-gpu This is now enitrely done in the core move-to-gpu as a special case. Signed-off-by: Chris Wilson commit bb8770158c08394c2de79e0ca1c1b3112e17dd23 Author: Chris Wilson Date: Sun Jul 8 18:53:25 2012 +0100 sna: Apply the clear color when resizing the front buffer If the existing front buffer is clear, just apply the clear color to then new buffer rather than copy the old one across. Signed-off-by: Chris Wilson commit 8dd14855d75240501aa2d089bcdfa46e7badbf18 Author: Chris Wilson Date: Sun Jul 8 15:46:41 2012 +0100 sna/dri: Review stale comments Signed-off-by: Chris Wilson commit 3bb7a530e77a5c6e8bae0fb14c570feafe37c2bd Author: Chris Wilson Date: Sat Jul 7 20:53:27 2012 +0100 sna/dri: Fix cross-chaining of pageflip vs vblank And double-check that the drawable is still flippable before completing the delay exchange. Signed-off-by: Chris Wilson commit 1454df8caa51fff9810a9b792d17aa82b247a4ae Author: Chris Wilson Date: Sat Jul 7 19:59:50 2012 +0100 sna/dri: Use draw ref directly As we hook into the DestroyWindow notification, we can reliably use the original Drawable reference and avoid the secondary object lookups. Signed-off-by: Chris Wilson commit 9abb6c09bdb9831db5f0626eed70663a2dc39fc6 Author: Chris Wilson Date: Sat Jul 7 18:43:06 2012 +0100 sna/dri: Remove dead code for 'old_fb' The member still exists but is never set and is unused. Signed-off-by: Chris Wilson commit ad877abdc70b842afd202b2f97892d97a0b8d151 Author: Chris Wilson Date: Sat Jul 7 17:48:52 2012 +0100 sna/dri: Attach the chain of frame events to the Window, not pixmap So that we can have multiple clients swapping in separate windows concurrently. Signed-off-by: Chris Wilson commit 81cd9aa80091b9bb08b50062f117d678a3bc7a91 Author: Chris Wilson Date: Sat Jul 7 16:12:49 2012 +0100 sna: Tweak start/stop of the deferred flush As we now emit work whenever we wakeup and find the GPU idle, we rarely actually have pending work in the deferred flush queue, so try to avoid installing a timer if we are not accumulating work. Signed-off-by: Chris Wilson commit 6cb0c631e4eafc09f1677c73906de9108d735de4 Author: Chris Wilson Date: Sat Jul 7 15:10:18 2012 +0100 sna/dri: Clarify the message for one failure case Signed-off-by: Chris Wilson commit 9a314d18cef1e08b23f9dca861ad2ba396bb1080 Author: Chris Wilson Date: Sat Jul 7 14:56:05 2012 +0100 sna/dri: Add a couple of missing VG_CLEAR on vblanks Signed-off-by: Chris Wilson commit 66a53c15cb5ee729fb43ea9713fd8538a3f982ad Author: Chris Wilson Date: Sat Jul 7 14:46:08 2012 +0100 sna/dri: Couple the frame events into DestroyWindow Signed-off-by: Chris Wilson commit 975a566bed72ddc79853b329307ed72a82df24b0 Author: Chris Wilson Date: Sat Jul 7 14:12:51 2012 +0100 sna/dri: Replace the DRI2 drawable type with a devPrivate Signed-off-by: Chris Wilson commit 0da1c98f660269806408af5fd08c1ab5e538082e Author: Chris Wilson Date: Sat Jul 7 12:15:28 2012 +0100 test: Add missing header for distcheck Signed-off-by: Chris Wilson commit e3e58123d36924c760ab6f58a7155a040422e91d Author: Chris Wilson Date: Fri Jul 6 15:22:26 2012 +0100 sna: Fixup fb wrapper To accommodate changes in the Xserver and avoid breakage; would have been much easier had the fb been exported in the first place. commit 5d2f88fd9972c62c87098ddc7fee7b6f0cea0fdb Author: Chris Wilson Date: Sat Jul 7 10:07:51 2012 +0100 sna: Add a couple of DBG flags to control upload buffers Signed-off-by: Chris Wilson commit 18b3090381ac0fb61df96a7c6b2f12aacf180ee8 Author: Chris Wilson Date: Sat Jul 7 00:06:36 2012 +0100 sna: Fixup USE_INPLACE debug hint After we enabled inplace readback, we can just apply the INPLACE hint to all fallbacks, but instead apply it to the automatic testing for a potential inplace operation. Signed-off-by: Chris Wilson commit 24b59a8955eb15522ad334a541530aca937fcf07 Author: Chris Wilson Date: Fri Jul 6 14:16:36 2012 +0100 sna: Refactor a common function: is_clipped() Had I done this earlier, I would not have the bug in the open-coded version. Le sigh. Signed-off-by: Chris Wilson commit 68b357454af705f1ed3a9599435a402f2611a180 Author: Chris Wilson Date: Fri Jul 6 10:58:06 2012 +0100 sna: Clear garbage from the new front buffer when resizing Avoid displaying a buffer filled with random junk when resizing (and thereby creating a new) framebuffer. Signed-off-by: Chris Wilson commit 5784e0f21dc91f33c99a507105a0695cc53d6574 Author: Chris Wilson Date: Fri Jul 6 09:42:37 2012 +0100 Allow matching against any device supported by drm/i915 However we cannot enable acceleration if we do not recognise its hardware layout or instruction set. Signed-off-by: Chris Wilson commit 1ee10cc3b2aa0888753eeb25c7fde7296a3c92eb Author: Chris Wilson Date: Fri Jul 6 00:01:58 2012 +0100 Make the detection of broken pre-production hardware verbose These SDV should have been returned to the manufacturer long ago and replaced with real systems. So if they are still in use, add a gentle reminder. Signed-off-by: Chris Wilson commit cd2dd3016e0834d1636aa96511608022a4cdbcd1 Author: Chris Wilson Date: Thu Jul 5 19:58:54 2012 +0100 sna: Fix clipping of glyphs-to-dst for partially obscurred windows Signed-off-by: Chris Wilson commit 0f086acb259d7732560c5d0d642308de028a4445 Author: Chris Wilson Date: Wed Jul 4 12:26:18 2012 +0100 sna: Fallback to attaching a shadow fb if we fail to setup the crtc As we've chosen to fix the kernel to handle CRTC offsets > 4096, drop the automatic workaround. However, allow the user to force creation of PerCrtcPixmaps for the purpose of debugging (and to workaround the bug in older kernels) and to fallback to trying a shadow fb if the setcrtc fails with the composite fb. Signed-off-by: Chris Wilson commit 7e8060f837475c85cc061ba4a5388140cd227613 Author: Chris Wilson Date: Thu Jul 5 03:31:12 2012 +0100 sna: Do not force GPU allocation if CPU bo is already busy Signed-off-by: Chris Wilson commit c32bb286dc9a489232030f6abe9076411fbcecfd Author: Chris Wilson Date: Thu Jul 5 03:18:12 2012 +0100 sna: Make sure damage is flushed to the CPU bo before use Signed-off-by: Chris Wilson commit d46cc00b3cd903bfaf37ad7d4a60676c4b346983 Author: Chris Wilson Date: Thu Jul 5 02:19:26 2012 +0100 sna/dri: Assert that our pixmaps sizes are invariant Signed-off-by: Chris Wilson commit 3c1f58fe45e8f4f18c45733f862ca1f290dde84d Author: Chris Wilson Date: Wed Jul 4 23:03:44 2012 +0100 sna: drop an unused ret var Signed-off-by: Chris Wilson commit a69a47f94037698a360f8b8611f6f3e58a20ca87 Author: Chris Wilson Date: Wed Jul 4 22:15:45 2012 +0100 sna: Fix detection of EIO through throttle When using drmIoctl, one needs to check the errno for the actual error code. Signed-off-by: Chris Wilson commit f2fdb0da83d6208c884ffde2bb30d0a67fc75217 Author: Chris Wilson Date: Wed Jul 4 21:43:22 2012 +0100 sna: Use throttle() to confirm a GPU hang This has two effects. The first is that we confirm that the EIO was indeed a GPU hang and not some other form of I/O failure (for example swapin). And the second is that it means we also print the warning to look for an i915_error_state when we detect a GPU hang. Signed-off-by: Chris Wilson commit b8c2b34e39c7f2ee2597f8727b9b98c6aac93fb1 Author: Chris Wilson Date: Wed Jul 4 19:53:27 2012 +0100 sna: Simplify timer execution Signed-off-by: Chris Wilson commit 844ab84c64fde38db315787e16a9f7f674d0f88c Author: Chris Wilson Date: Wed Jul 4 17:58:27 2012 +0100 sna: Fix leak around ENOMEM error path in sna_copy_boxes Signed-off-by: Chris Wilson commit 54ee41bf05e328ed2f2cd2a64c21afe3efe2556a Author: Chris Wilson Date: Wed Jul 4 17:40:58 2012 +0100 sna: Pass a hint to the backends when using copy_boxes for readback If we expect to only emit this set of copy_boxes() and then submit the batch, we would prefer to use the BLT for its lower latency. Signed-off-by: Chris Wilson commit d36623aaebdc454bd281f34332edcad5d94f34c0 Author: Chris Wilson Date: Wed Jul 4 13:55:25 2012 +0100 sna/dri: Remove broken code for buffer xchg from AsyncSwap Signed-off-by: Chris Wilson commit 0bdb4d0d3693df007a6cfc9a75bb0deddd812d53 Author: Chris Wilson Date: Wed Jul 4 13:12:43 2012 +0100 sna/dri: Fixup blit fallback path to use xchg when possible Signed-off-by: Chris Wilson commit aeeed323f9081875a804ae19fb356bee2a61e7b0 Author: Chris Wilson Date: Wed Jul 4 11:52:27 2012 +0100 sna/dri: Select the appropriate copy engine based on the current src ring Signed-off-by: Chris Wilson commit 5026b6a147f411582af24ffd17c87a81da5eea0b Author: Chris Wilson Date: Wed Jul 4 09:39:59 2012 +0100 uxa: Install the drm_wakeup_handler for vblank events in !use_page_flipping Even if page-flipping itself is disabled, we still want to allow the client to schedule wakeups for some future vblank which requires listening to the kernel vblank notifications. Reported-by: Eric Anholt Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=51699 Signed-off-by: Chris Wilson commit 79309dd55f20098e12ead5427c811f237d5592fa Author: Chris Wilson Date: Tue Jul 3 22:27:39 2012 +0100 sna: check for failure to change cache level on the bo Signed-off-by: Chris Wilson commit 9e2ac8a413c05aafd1c69775c5b0fa5f1f37ac23 Author: Chris Wilson Date: Tue Jul 3 19:33:55 2012 +0100 sna: Use set-cache-level ioctl to create CPU bo As an alternative to vmap, we can use the kernel for all memory management through bo, which is much preferred for its simplicity (i.e. avoiding introducing even more vm complexity). Signed-off-by: Chris Wilson commit f26163e916e21f6b64a19c147165f2f2c9c5466d Author: Chris Wilson Date: Tue Jul 3 18:51:46 2012 +0100 sna: Guess when the client is attempting to read back the whole pixmap Signed-off-by: Chris Wilson commit c597e6cd4cb4406878eae07c04ad420d8185d99e Author: Chris Wilson Date: Tue Jul 3 13:26:33 2012 +0100 sna: Fix iterator typo Increment the iterator, not the end-stop. Signed-off-by: Chris Wilson commit 2d087eadd9cd3aa2e2ccd73a568286d21702d29a Author: Chris Wilson Date: Tue Jul 3 11:54:02 2012 +0100 sna: Also prefer to use the dst GPU bo if CopyArea reduces to an inplace upload Signed-off-by: Chris Wilson commit dbe3a5ca1e6558bcee29d893aba8cc352220a36d Author: Chris Wilson Date: Tue Jul 3 11:48:15 2012 +0100 sna: Attempt to reduce all-damage on the CPU As we now treat CPU all-damaged as a special case for deciding when and where to migrate, look out for that condition after adding damage. Signed-off-by: Chris Wilson commit 3a41248195e8b327a5d970726450bd2077cdaf0f Author: Chris Wilson Date: Tue Jul 3 11:22:54 2012 +0100 sna: Operate on the original boxes for CopyArea Be consistent and avoid the confusion when mixing operations on the region boxes and the original boxes. Signed-off-by: Chris Wilson commit e3bc91842bf186ec149c852f48993235ef7ad27f Author: Chris Wilson Date: Tue Jul 3 10:21:06 2012 +0100 sna: Allow booting on older kernels by disabling HW acceleration Signed-off-by: Chris Wilson commit affdebcb2d09cd8fdc5aadb2d8df3193587e4a06 Author: Chris Wilson Date: Mon Jul 2 18:50:26 2012 +0100 sna: And free the DIR after use Signed-off-by: Chris Wilson commit e7b31b6d0a32f76db4a8aef64c77d4afe808fb6c Author: Chris Wilson Date: Mon Jul 2 14:01:36 2012 +0100 sna: Consolidate CopyArea with the aim of reducing migration ping-pong Signed-off-by: Chris Wilson commit ce27a81ac508368d54f1237893a9b1214cf3e3d0 Author: Chris Wilson Date: Mon Jul 2 17:52:24 2012 +0100 sna: Clear cpu flag after deciding to use gpu bo Signed-off-by: Chris Wilson commit 2f1b7e8a23ac3086dda0025ecf09dd1feac94837 Author: Chris Wilson Date: Mon Jul 2 17:53:01 2012 +0100 sna: Check for non-existent /sys/class/backlight directory Signed-off-by: Chris Wilson commit d12d50d107d403c3cf4dfe24bb63ce1006d0e025 Author: Chris Wilson Date: Mon Jul 2 14:09:21 2012 +0100 sna: Skip hidden and special entries inside /sys/class/backlight Just to avoid the warnings after fallback to comparing the d_name to the list of known interfaces. Signed-off-by: Chris Wilson commit e80f9c4670a0e84521907b1baa059322784b1558 Author: Chris Wilson Date: Mon Jul 2 10:21:42 2012 +0100 sna: Prefer backlight iface based on /sys/class/backlight/*/type Signed-off-by: Chris Wilson commit 61e16dc5673a1ac96b2ecee072cc3e80971be5d9 Author: Tom Hughes Date: Mon Jul 2 10:23:56 2012 +0100 Add asus-nb-wmi backlight control Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=51593 commit 543816011d05f0cf40b05ec58f87a5954397a224 Author: Chris Wilson Date: Sun Jul 1 18:36:55 2012 +0100 sna/gen7: Correct two minor misues of equivalent GEN6 constants Signed-off-by: Chris Wilson commit 839fea7f779ca63c32817ee4ba695dfce3344980 Author: Chris Wilson Date: Sun Jul 1 18:23:38 2012 +0100 sna/gen6: Boost VS entries to maximum Signed-off-by: Chris Wilson commit fbd114507d9bf2e2b1d1e52c5e42dc6cdbd8c9a0 Author: Chris Wilson Date: Sun Jul 1 14:03:31 2012 +0100 sna/dri: Assert that the replacement bo is large enough for the pixmap Just another paranoid sanity check. Signed-off-by: Chris Wilson commit 675cbd5fade91fd6a6bf533a31b0a211237af6e8 Author: Chris Wilson Date: Sun Jul 1 12:39:13 2012 +0100 sna/trapezoids: Skip the division when converting coverage to floating point Signed-off-by: Chris Wilson commit 182c3637cc5d3a6ce52127087aa2f19ca2b42719 Author: Chris Wilson Date: Sun Jul 1 13:23:44 2012 +0100 sna: If the pixmap is marked for flushing, prefer to use the GPU Again, to avoid the forced ping-pong as we upload the damage after nearly every operation, simply prefer to use the GPU in such cases. Signed-off-by: Chris Wilson commit 35b1ac138002c206a6d6b866d49a0d73705dd3ac Author: Chris Wilson Date: Sun Jul 1 10:53:07 2012 +0100 sna: After an operation on the CPU, prefer not to use the GPU A nasty habit of applications is to fill an area, only to read it back with GetImage, render locally and replace with PutImage. This causes a readback of an active bo everytime, so let's try to mitigate that by preferring not to use the GPU after a forced readback. Signed-off-by: Chris Wilson commit e625c02e6266403fcd8a72ccce2c6c6291e2e5fc Author: Chris Wilson Date: Sat Jun 30 09:34:21 2012 +0100 sna/damage: Early check for contains-box? if subtract and box outside region Signed-off-by: Chris Wilson commit abd7be1cee6f2f494a11cd9d2e7888c3043ffc02 Author: Chris Wilson Date: Fri Jun 29 15:32:10 2012 +0100 sna/dri: Prefer GPU rendering if no more CPU damage on a DRI bo Signed-off-by: Chris Wilson commit 67b87e4f7cf6c3ab9cfccc9fe43a824bfe84f393 Author: Chris Wilson Date: Fri Jun 29 15:03:12 2012 +0100 sna/dri: Optimise clip reduction with copy-to-front to an unclipped Window Signed-off-by: Chris Wilson commit eae5e1275cd11703de2bca67dacd1d57129b561a Author: Chris Wilson Date: Fri Jun 29 13:12:52 2012 +0100 sna: Install the ModeSet handler as the base handler This way we can safely ignore it across server regen. Signed-off-by: Chris Wilson commit 15a0761cad862a5d73bbc2af81bc5267e66c307e Author: Chris Wilson Date: Fri Jun 29 12:22:55 2012 +0100 sna: Only consider the request list when deciding whether the GPU is busy Micro-optimisation to overhead extra checks and to make sure an unflushed bo doesn't prevent us from submitting more work before sleeping. Signed-off-by: Chris Wilson commit 4061f05dd61a7200d91d30b9b4a2ab0c61306870 Author: Chris Wilson Date: Fri Jun 29 11:40:18 2012 +0100 sna/trapezoids: Write unaligned fallback boxes inplace As this is a pure write operation (though we will write the edge pixels twice) we can perform this operation inplace and incur a slightly slower trap creation at the benefit of avoiding the later copy. Signed-off-by: Chris Wilson commit 44e41536b7728c03a4899a97242960ae7ed35624 Author: Chris Wilson Date: Fri Jun 29 11:39:47 2012 +0100 sna/trapezoids: Render the partial left-edge of fallback unaligned boxes Signed-off-by: Chris Wilson commit e6f9bfe1e20c4c76822d77598f4f1c7e10754929 Author: Chris Wilson Date: Fri Jun 29 11:06:33 2012 +0100 sna: Use currentTime rather than GetTimeInMillis() The overhead of reading the hpet() on every block handler (more or less) is exorbitant, so trust that we update currentTime frequently enough to be a good approximation - the side effect is that we will wakeup slightly to earlier from using an old value for the current time. Signed-off-by: Chris Wilson commit c6c4f30e194eb3b8a2421134bb005f1542b3ef9f Author: Chris Wilson Date: Fri Jun 29 10:14:52 2012 +0100 sna: Add assertions to check that we do install the timers Signed-off-by: Chris Wilson commit 87c8f5a47e3abb7ff887aab45b6389cf9b8d8008 Author: Chris Wilson Date: Fri Jun 29 09:33:09 2012 +0100 sna: Make the post-flip delay explictit As the kernel is inconsistent in enforcing this across generations, handle the synchronisation of the pageflip explicity. Ultimately this should be replaced with a tripple buffer mechanism. Signed-off-by: Chris Wilson commit 31caa43a21a68174386682fd558c911c7c03d76f Author: Chris Wilson Date: Thu Jun 28 17:05:46 2012 +0100 sna/gen5: Check harder for need_upload() fallbacks Signed-off-by: Chris Wilson commit 7c3eb1fda9e567a300a8138acdac9ec11e010b11 Author: Chris Wilson Date: Thu Jun 28 17:04:26 2012 +0100 sna: Correct inverted logic for checking xrgb drawables Reported-by: Christoph Reiter Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=51472 Signed-off-by: Chris Wilson commit c3e2c1332d8d5a3944df99cc11aa66c586add3e8 Author: Chris Wilson Date: Thu Jun 28 12:34:36 2012 +0100 sna: Fix the application of the crtc offset for posting damage The damage boxes are in framebuffer (source) space, so we need to apply the offset for the boxes in crtc (destination) space. Signed-off-by: Chris Wilson commit 47e6bfa4f40cf7efcfe7eee24d2512d737fd7e89 Author: Chris Wilson Date: Thu Jun 28 12:24:27 2012 +0100 sna: Force use of per-crtc scanout if the offset is too large On gen4+, the scanout offset into a tiled surface is specified through the DSPTILEOFF register and limited to 12bits of precision. So if we have a CRTC positioned in that nether-region, we need to allocate a separate per-crtc pixmap for it and perform shadowing. Signed-off-by: Chris Wilson commit 93e77ee019248fe77483e83f2210d584bb5d1be2 Author: Chris Wilson Date: Thu Jun 28 12:02:32 2012 +0100 sna: Quieten kernel debug complaints when disabling crtc Even if we are obviously turning the crtc off, it still complains if the number of connectors is non-zero. So make it so. Signed-off-by: Chris Wilson commit 85e4f48a87ddbc227af8f4af5ea46ae17902b111 Author: Chris Wilson Date: Thu Jun 28 11:44:45 2012 +0100 sna: Add a DBG to the periodic flush mechanism Signed-off-by: Chris Wilson commit 87dd6408a5c29e4808283df78a981de0a3c0a79c Author: Chris Wilson Date: Thu Jun 28 10:53:59 2012 +0100 sna: Correct the reversal of the periodic flushing semantics Regression from 1e9319d (sna: extend RandR to support super sized monitor configurations) which tried to take into account the need to flush the shadow CRTC bo in addition to the normal scanout bo. In the refactoring of the need_flush(), the double negative was missed. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 05f486f64bc7ea4a8a71f5d792fa586ac0843414 Author: Chris Wilson Date: Thu Jun 28 10:42:21 2012 +0100 sna: Flush the per-crtc render caches for rotated scanouts We need to manually flush the render cache in order for results to be visible on the scanout. Signed-off-by: Chris Wilson commit db7979981032d5647069f213a063dcbf78e57890 Author: Chris Wilson Date: Wed Jun 27 09:36:58 2012 +0100 sna: s/width/height/ cut'n'paste typo Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit fcbbe1664ae62ae275068cc040f385d2320d5213 Author: Chris Wilson Date: Wed Jun 27 09:30:44 2012 +0100 sna: Remove a trailing ';' The unwanted ';' caused is_cpu() to always return false if a GPU bo was attached. Not necessary a bad thing, just misses the potential optimisation where having chosen to prefer to use the CPU path we then have to migrate to the GPU even though the bo is undamaged or idle. Spotted-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit a072ab506569ecff5b4c57fa90f7a417db69f33b Author: Chris Wilson Date: Tue Jun 26 17:24:37 2012 +0100 test: Add client side copy to FakeFront for emulating CopyBuffer correctly The server manages FakeFront following a flip, but it the client optimises a swap by replacing it with a CopyRegion, it is expected to also update the FakeFront itself. Replicate that behaviour so that the timings for the test case are consistent with mesa. Signed-off-by: Chris Wilson commit 96804c74f8f0d19fc1b9db467cf050f4934fe83b Author: Chris Wilson Date: Tue Jun 26 17:09:45 2012 +0100 test: FakeFront rules Oh my, I just once again rediscovered the copy on every flip due to the requirement for keeping FakeFront uptodate for reads after a SwapBuffers. Signed-off-by: Chris Wilson commit f306cd557ef263ff5057e413c335cc75bc0f7895 Author: Chris Wilson Date: Tue Jun 26 14:59:37 2012 +0100 sna/dri: Hold a reference to the cached DRI2 buffer on the front buffer Signed-off-by: Chris Wilson commit a87f2b9325bfad2bb3f93226706b6f9a09598945 Author: Chris Wilson Date: Sun Jun 24 13:48:38 2012 +0100 sna/gen4: Check for peculiar initial values for the surface offset Signed-off-by: Chris Wilson commit 8f4221a2520b7ed1f67b41185c2e4842e83e53ef Author: Chris Wilson Date: Tue Jun 26 15:56:50 2012 +0100 test: Add a simple exercise for DRI2 swap paths Signed-off-by: Chris Wilson commit a505015a254d6c6e24f0542bc141cde873dc6f34 Author: Chris Wilson Date: Mon Jun 25 22:50:47 2012 +0100 sna: Force DPMS to be on following a modeset Similarly to UXA, this papers over inconsistent behaviour in the kernel in handling the DPMS upon a modeswitch. Signed-off-by: Chris Wilson commit b7a8c94cdb9cf42a31f8ce128d70e23458ba2042 Author: Chris Wilson Date: Mon Jun 25 20:49:59 2012 +0100 sna: remove the assert(0)s along error paths This were there as a debugging aide to see if we ever hit unreachable code paths - mainly along corruption inducing GPU wedged recovery paths. They are superfluous and just scare the reader. Signed-off-by: Chris Wilson commit 15c0ee445f603033c82f357fedfc7737d198d7b3 Author: Chris Wilson Date: Sun Jun 24 19:28:03 2012 +0100 sna/gen5: Tweak thread allocations Bump the alloted number of threads to their max. Using more threads than cores helps hide the stalls due to sampler fetch, math functions and urb write. Specifying too many threads seems to not incur a performance regression, suggesting that the hardware scheduler is sane enough not to overpopulate the EU. A small but significant boost, peak x11perf -aa10text on an i3-330m is raised from 1.93Mglyphs/s to 2.35Mglyphs/s. Signed-off-by: Chris Wilson commit fa10005ce31483827547b7f71eae066899f0026c Author: Chris Wilson Date: Sat Jun 23 17:28:00 2012 +0100 sna/dri: Perform an exchange for a composited windowed SwapBuffers If the front buffer is not attached to the scanout and has not been reparented, we can simply exchange the underlying bo between the front/back attachments and inform the compositor of the damage. Signed-off-by: Chris Wilson commit 53d735ddb16b0204662b8584aa22998ba53deec1 Author: Chris Wilson Date: Sat Jun 23 10:16:44 2012 +0100 sna/dri: Queue windowed swaps Implement "tripple-buffering" for windowed SwapBuffers by allowing the client to submit one extra frame before throttling. That is we emit the vsync'ed blit and immediately unblock the client so that it renders to the GPU (which is guaranteed to be executed after the blit so that its Front/Back buffers are still correct) and requests another SwapBuffers. The subsequent swapbuffers are appended to the vsync chain with the blit/unblock then executed on the vblank following the original blit. That is both the client and xserver render concurrently. Signed-off-by: Chris Wilson commit 1e9319d5f56583be99f573f208cebb0ee3b5cc26 Author: Chris Wilson Date: Fri Jun 22 11:22:16 2012 +0100 sna: extend RandR to support super sized monitor configurations With the introduction of the third pipe on IvyBridge it is possible to encounter situations where the combination of the three monitors exceed the limits of the scanout engine and so prevent them being used at their native resolutions. (It is conceivable to hit similar issues on earlier generation, especially gen2/3.) One workaround, this patch, is to extend the RandR shadow support to break the extended framebuffer into per-crtc pixmaps. Signed-off-by: Chris Wilson commit e8b090902e788257610374deae659f01a91888f3 Author: Chris Wilson Date: Fri Jun 22 22:01:37 2012 +0100 sna/gen3+: Remove stale assertions for cached vbo Following the previous commit, we reset the vbo when it becomes idle rather than discard it. As such, the assertions to check that we are discarding the vbo are now bogus. Signed-off-by: Chris Wilson commit 565297e6bd3457a150036af9c62fe0dc67b794ac Author: Chris Wilson Date: Thu Jun 21 13:53:02 2012 +0100 sna/gen3+: Keep vbo cached Once we switch to using a vbo, keep it cached (resetting everytime it is idle) until we expire our caches. Signed-off-by: Chris Wilson commit d806973e21cd46e605b3cd405323ae7a64c12798 Author: Chris Wilson Date: Thu Jun 21 12:57:13 2012 +0100 sna: Micro-optimise search_inactive_cache Discard the unneeded next parameter to drop a memory reference in a hot path, and don't wait for a retirement if we are looking in a larger bucket than suits. Signed-off-by: Chris Wilson commit d39fef0a7f3daf5c07686b44e4dea01c0f06c77a Author: Chris Wilson Date: Thu Jun 21 12:25:35 2012 +0100 sna: Tiles are only 128 bytes wide on gen2 Signed-off-by: Chris Wilson commit 4f2dde1fa3b04b27bae8fc0bca9c824bd362d23b Author: Chris Wilson Date: Thu Jun 21 10:31:24 2012 +0100 sna/gen7: Eliminate the pipeline stall after a non-pipelined operation Signed-off-by: Chris Wilson commit 3ef05a8d0833203e265aff392f225a11a11c2d01 Author: Chris Wilson Date: Thu Jun 21 09:36:42 2012 +0100 sna/gen7: Do not emit a pipeline stall after a non-pipelined command Signed-off-by: Chris Wilson commit 4501e131e6b737cb8f2581c8b1f7ea9d29a8e912 Author: Chris Wilson Date: Thu Jun 21 16:38:32 2012 +0100 sna/gen7: prefer using RENDER copy Further testing and the balance of doubt swings in favour of using the 3D pipeline for copies. For small copies the BLT unit is faster, 2.14M/sec vs 1.71M/sec for comppixwin10 And for large copies the RENDER pipeline is faster, 13000/sec vs 8000/sec for comppixwin500 I think the implication is that we are not efficiently utilising the EU for small primitives - i.e. something that we might be able to improve. Signed-off-by: Chris Wilson commit 3da56c48b7820ec77d704c5a16670eb86a6f673f Author: Chris Wilson Date: Wed Jun 20 15:14:23 2012 +0100 sna/gen7: Prefer using BLT rather than redirect for copies Signed-off-by: Chris Wilson commit b1f8386db6e9b3eea1bdbf8cde90f33792640ce8 Author: Chris Wilson Date: Wed Jun 20 23:28:14 2012 +0100 sna/gen7: Emit a pipeline flush after every render operation For whatever reason, this produces a 30% improvement with the fish-demo (500 -> 660 fps on i7-3730qm at 1024x768). However, it does cause about a 5% regression in aa10text. We can appear to alleviate that by only doing the flush when the composite op != PictOpSrc. Signed-off-by: Chris Wilson commit d02e6d81420a114c9622bbdaf90fc3ae5d4b15a7 Author: Chris Wilson Date: Thu Jun 21 16:10:02 2012 +0100 Encode the third pipe using the HIGH_CRTC shift for vblanks The original vblank interface only understood 2 pipes (primary and secondary) and so selecting the third pipe (introduced with IvyBridge) requires use of the HIGH_CRTC. Using the second pipe where we meant the third pipe could result in some spurious timings when waiting on the vblank. Reported-by: Adam Jackson Signed-off-by: Chris Wilson commit f8b67be8d3b5d5cbe730ba5dba3edd2d30a99d9f Author: Chris Wilson Date: Wed Jun 20 12:39:19 2012 +0100 sna: Don't clear the needs_flush flag after emitting a flush on the busy bo We use that flag to check whether we need to check whether the bo is still busy upon destruction, so only clear it if the bo is marked as idle. Signed-off-by: Chris Wilson commit 5419bbb483af595d7021e49debef7700c599217e Author: Chris Wilson Date: Wed Jun 20 11:21:52 2012 +0100 sna/gen7: Prefer BLT for copies It's faster for where the cost of the extra batches and ring switching do not dominate... Signed-off-by: Chris Wilson commit 1c0bb8c4c93a43e2932429a9f6c23d91f1be2060 Author: Chris Wilson Date: Wed Jun 20 10:57:40 2012 +0100 sna/gen7: Keep using RENDER paths for large pixmaps As the 3D pipeline is quite versatile and we only need to force BLT if we cannot extract the subregion. Signed-off-by: Chris Wilson commit b238f64e8a53883cfe7c568e37ef18bbee77996e Author: Chris Wilson Date: Wed Jun 20 10:46:59 2012 +0100 sna/gen[67]: Prefer to not force BLT paths for large pixmaps The sampler can in fact handler subregions of large pixmaps quite well, and so we prefer to keep using the 3D pipeline so long as the operation fits in. If not, then switch to the BLT in order to avoid the temporary surface dance. Signed-off-by: Chris Wilson commit 38f06a351fdd35fa5c6136e9a25e034a812a94c2 Author: Chris Wilson Date: Tue Jun 19 22:22:12 2012 +0100 uxa: Fix second regression in glyph fallback from 64a4bc To complete my show of incompetence for the evening, not only do we have to restore the original source when compositing the mask onto the destination, we also need to restore the original dst (rather than composite the mask onto the mask!). Signed-off-by: Chris Wilson commit fda9faee755cb35906ca1179a568332ef2de35a5 Author: Chris Wilson Date: Tue Jun 19 21:01:47 2012 +0100 uxa: Use the original src for fallback glyph compositing In 64a4bcb8cefff, I introduced a WHITE source for the purposes of accumulating the glyph mask correctly. Unfortunately I neglected to restore the original source picture for compositing the glyph mask on the destination, resulting in a use-after-free and then corruption. Reported-by: Maarten Lankhorst Signed-off-by: Chris Wilson commit 8141e290b1ac7c4d1524bb389a84a8f375df4634 Author: Chris Wilson Date: Tue Jun 19 20:24:33 2012 +0100 sna: Explain why we ignore the busy status result during kgem_bo_flush() Signed-off-by: Chris Wilson commit eb1d07624e5aeb7b5db7ceb46975091e92185d63 Author: Chris Wilson Date: Tue Jun 19 16:00:13 2012 +0100 sna: Ensure extents is initialised if short-circuit use-cpu-bo As we may attempt to end up using the GPU bo is the CPU bo is busy, we need to make sure we have initialised the damage extents first. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 9f216e159bd05ddd269eb8ddf3ca7a407c2901e2 Author: Chris Wilson Date: Tue Jun 19 15:57:31 2012 +0100 sna: Assert expected return values Keep the semantic analyser happy by consuming the expected return value with an assert. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 2dc93b2a6c832ce8b972de90d09080e860dcd40f Author: Chris Wilson Date: Tue Jun 19 15:34:09 2012 +0100 sna: Check results from syscalls Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 06634604abf15fdd27dd007fcf81595da994146b Author: Chris Wilson Date: Tue Jun 19 15:28:43 2012 +0100 Initialise adaptors to 0 in case xf86XVListGenericAdaptors does not Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 8bfea58dbc634cadc399d3132030c591e086880c Author: Chris Wilson Date: Tue Jun 19 15:26:18 2012 +0100 sna: Minor cleanups from sematic analyser in DBG Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 0a43d425670b883b04565296c0510e7ba03ba6de Author: Chris Wilson Date: Tue Jun 19 14:46:58 2012 +0100 uxa: Implement glyphs-to-dst to avoid fallbacks An earlier version was buggy and introduced corruption as it failed to fallback gracefully with ComponentAlpha glpyhs. This is a much simpler implementation that composites each glyph individually, leaving it to the backend to optimise away state changes. It should still be many times faster than incurring the fallback... Reported-by: Oleksandr Natalenko Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=50508 Signed-off-by: Chris Wilson commit 64a4bcb8ceffff8e4ee448d8a467620596703509 Author: Chris Wilson Date: Tue Jun 19 13:42:12 2012 +0100 uxa: Use (white IN glyph) ADD mask to compose the glyph mask As pointed out by Soren Sandmann and Behdad Esfahbod, it is essential to use white IN glyph when adding to the mask so that the channel expansion is correctly performed when adding to an incompatible mask format. For example, loading alpha as the source results in the value 000a being added to the rgba glyph mask (for mixed subpixel rendering with grayscale glyphs), whereas the desired value is aaaa. Signed-off-by: Chris Wilson commit 99845dcb3ba862269b29aec782e2bcef31c0403e Author: Chris Wilson Date: Tue Jun 19 10:37:10 2012 +0100 Post Damage on the Screen Pixmap after a pageflip This issue was raised by Dave Airlie as he is trying to integrate multiple GPUs into the xserver, and a particular setup has a slave rendering device that copies the contents from the GPU over a DisplayLink USB adaptor. As such the slave device is listening for Damage on the Screen Pixmap and needs the update following pageflips. Since we already are posting damage for all the SwapBuffers paths other than pageflip, for consistency we should post damage along the pageflip path as well. Reported-by: Dave Airlie Signed-off-by: Chris Wilson commit 4acf727941a37651f672db3774281a437940c540 Author: Chris Wilson Date: Tue Jun 19 10:20:54 2012 +0100 sna: Initialize the color value for fallback unaligned boxes Reported-by:Zdenek Kabelac Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=5047 Signed-off-by: Chris Wilson commit b0b2d3c9663c29e9844aef1608416ae40c194b55 Author: Chris Wilson Date: Tue Jun 19 00:41:35 2012 +0100 sna: Avoid copying unintialised data during source picture upload If we have never written to a pixmap, then there will be neither a GPU or shadow pointer and we would attempt to copy a NULL pointer. In this case as the user is expecting to copy unintialised data we are at liberty to replace those undefined values with the clear color. Signed-off-by: Chris Wilson commit 38472fcc53c5dceb98b96458183e6729b8311a43 Author: Chris Wilson Date: Tue Jun 19 00:37:12 2012 +0100 sna: Double check that the source is busy before performing indirect reads Signed-off-by: Chris Wilson commit 8cdfb8c24c8b49c88451714d80293c66d63e8c01 Author: Chris Wilson Date: Tue Jun 19 00:36:41 2012 +0100 sna: Fix up the shadow pointer on the source when copying Signed-off-by: Chris Wilson commit 17f3a83fdc8c0ef5c12fb4be34d86021c0c865e5 Author: Chris Wilson Date: Mon Jun 18 23:48:16 2012 +0100 sna: Review sna_copy_boxes A couple of ordering issue and more assertions. Signed-off-by: Chris Wilson commit a9045699b9cd66d0b0d96bfc964458c96845f97f Author: Chris Wilson Date: Mon Jun 18 23:47:21 2012 +0100 sna: Reset region after transferring to cpu If we adjust the region for the pixmap offset, be sure that we reset it before returning it back to the caller. Signed-off-by: Chris Wilson commit 9f51311a7d95bf4bc23926b8e6bf8ee52afd180c Author: Chris Wilson Date: Mon Jun 18 23:45:50 2012 +0100 sna: Check if the busy is truly busy before commiting to an indirect upload Signed-off-by: Chris Wilson commit 291b3c4367b455bfc5a772f8caaeee73f36d826a Author: Chris Wilson Date: Mon Jun 18 23:43:00 2012 +0100 sna: Align upload buffers to 128 This seems to be a restriction (observed on 965gm at least) that we have incoherent sampler cache if we write within 128 bytes of a busy buffer. This is either due to a restriction on neighbouring cachelines (like the earlier BLT limitations) or an effect of sampler prefetch. Reported-by: Zdenek Kabelac References: https://bugs.freedesktop.org/show_bug.cgi?id=50477 Signed-off-by: Chris Wilson commit 39e5c7491535999643c1761bb1602ad757ab486c Author: Chris Wilson Date: Mon Jun 18 21:58:27 2012 +0100 sna: Assert damage is valid after every addition Even more paranoia than just checking upon migration. Signed-off-by: Chris Wilson commit 92e1693e5fb3a1dd89fca5e5ecc660e2de78f9cd Author: Chris Wilson Date: Mon Jun 18 21:26:58 2012 +0100 sna: Validate cpu/gpu damage never overlaps References: https://bugs.freedesktop.org/show_bug.cgi?id=50477 Signed-off-by: Chris Wilson commit d2312c8f958002e54ddcb834f37916f4b46ac291 Author: Chris Wilson Date: Mon Jun 18 21:29:29 2012 +0100 sna: Fixup tracking of vmap upload buffers Signed-off-by: Chris Wilson commit 75e9eeca7e998b1ee3f8b0df780adae1b9e5b408 Author: Chris Wilson Date: Mon Jun 18 16:39:20 2012 +0100 sna: Remove overlapping CPU damage when operating inplace on the GPU Otherwise we gradually introduce garbage into the picture. Reported-by: Zdenek Kabelac Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=50477 Signed-off-by: Chris Wilson commit a936466dd41c4bb8b0956c4da5a5838f991b332a Author: Chris Wilson Date: Mon Jun 18 11:36:53 2012 +0100 sna: Prefer to attempt a Composite operation rather than use pixman composite As pixman composite performance is atrocious for anything other than solids, prefer to upload the mask and attempt a composite operation on the GPU unless we are forcing the fallback. Signed-off-by: Chris Wilson commit 4b325d6e2b99ec3f63a7c5175f6f4746f6773550 Author: Chris Wilson Date: Mon Jun 18 11:24:55 2012 +0100 sna: Fix rendering of unaligned boxes through pixman Not only do we need to make sure the source is available to the CPU, we need to actually check the right conditions for clipping the box. Signed-off-by: Chris Wilson commit caef27492b5a70bd46a09cd5094b2c93e38e39ea Author: Chris Wilson Date: Sun Jun 17 20:56:46 2012 +0100 sna: convert another instance of applying the clear to the CPU pixmap Signed-off-by: Chris Wilson commit 8695c4c77666cd07eab51efcbc7c4f11c85250fd Author: Chris Wilson Date: Sun Jun 17 16:53:53 2012 +0100 sna: Fix the blt composite op with no-ops When returning early because the operation is a no-op, we still need to fill in the function pointers to prevent a later NULL dereference. Signed-off-by: Chris Wilson commit 7905ddae1dbc8805d0fadbd6d21c7a5df7e715fc Author: Chris Wilson Date: Sun Jun 17 17:01:12 2012 +0100 sna: Further refine choice of placement when uploading source data. The goal is cheaply spot a simple copy operation that can be performed on the CPU without having to load both parties onto the GPU. Signed-off-by: Chris Wilson commit 5a675b61f27273f7ef344d4056dbba1f8cd97737 Author: Chris Wilson Date: Sun Jun 17 15:42:17 2012 +0100 sna: Correct typo forcing everything to be clear to 0! Signed-off-by: Chris Wilson commit b55bf1abbe71281e3d9ebde1c4005d40902d5e7f Author: Chris Wilson Date: Sun Jun 17 14:14:44 2012 +0100 sna: Fix cut'n'paste errors in tiling debug Rename for different variables Signed-off-by: Chris Wilson commit 9756c60b4ad15281d025b9c27f19d19e8a630958 Author: Chris Wilson Date: Sun Jun 17 11:39:33 2012 +0100 sna/gen7: Enable non-rectilinear spans Seems we have enough GPU power to overcome the clumsy shaders. Just imagine the possibilities when we have a true shader for spans... Signed-off-by: Chris Wilson commit 41aff56a1f452e409c7a49512a1d2824b74f3838 Author: Chris Wilson Date: Sun Jun 17 10:50:31 2012 +0100 sna: Add tiling for spans Semmingly only advisable when already committed to using the GPU. This first pass is still a little naive as it makes no attempt to avoid empty tiles, nor aims to be efficient. Signed-off-by: Chris Wilson commit 222e6ff43ef683e82101fb360911fc01fbe00597 Author: Chris Wilson Date: Sat Jun 16 23:40:30 2012 +0100 sna: Read inplace for fallback copies Signed-off-by: Chris Wilson commit 79d468925bb012806e097337e4e5930818c6ab46 Author: Chris Wilson Date: Sat Jun 16 17:54:21 2012 +0100 sna: Decrease latency for 1x1 GetImage by using an inplace mapping Signed-off-by: Chris Wilson commit 2c2a8d3780f1d8de3f13bee8e068fdaf608ff9e9 Author: Chris Wilson Date: Sat Jun 16 23:19:49 2012 +0100 sna: Allow reads to be performed inplace If we can guess that we will only readback the data once, then we can skip the copy into the shadow. Signed-off-by: Chris Wilson commit bc6997f6f751d3ba352dfc20c6717ec12b8fac47 Author: Chris Wilson Date: Sat Jun 16 17:54:15 2012 +0100 sna: Cleanup damage processing after operating inplace Signed-off-by: Chris Wilson commit 937ca8a5d8a0f70a0724db1519bb7b5fc0857425 Author: Chris Wilson Date: Sat Jun 16 17:53:58 2012 +0100 sna: Use memset for simple clears Signed-off-by: Chris Wilson commit de4572b0b52e2fcfcad04660ee2f81ee88d500a5 Author: Chris Wilson Date: Sat Jun 16 13:40:26 2012 +0100 sna: Inspect CPU damaged state when deciding upon Composite placement Signed-off-by: Chris Wilson commit b689cd924c500373e1e293dd9eb54a238e400381 Author: Chris Wilson Date: Fri Jun 15 16:25:51 2012 +0100 sna: Composite traps inplace if the CPU is already all-damaged One outcome is that inspecting the usage patterns afterwards indicated that we were missing an opportunity to reduce unaligned boxes to an inplace operation. Signed-off-by: Chris Wilson commit ae3c0963790cfb6f984ed4ad3ecbaae492775e1b Author: Chris Wilson Date: Fri Jun 15 16:25:51 2012 +0100 sna: Composite glyphs inplace if the CPU is already all-damaged Signed-off-by: Chris Wilson commit eaed58b2baf30eaea37be06cfc1d9d81059aba27 Author: Chris Wilson Date: Sat Jun 16 12:55:54 2012 +0100 sna: Tweak placement of operations Take in account busyness of the damaged GPU bo for considering placement of the subsequent operations. In particular, note that is_cpu is only used for when we feel like the following operation would be better on the CPU and just want to confirm that doing so will not stall. Signed-off-by: Chris Wilson commit 8eac098962891a5deb7c53d36c6dec57c7f2b972 Author: Chris Wilson Date: Sun Jun 17 09:51:12 2012 +0100 sna/gen3: Add another DBG for dropping vbo Signed-off-by: Chris Wilson commit 515c8b19d638d4a811b159ef0dc7cf4059e30217 Author: Chris Wilson Date: Fri Jun 15 15:41:14 2012 +0100 sna: Prefer to operate inplace if already mapped of the GPU is wholly dirty Signed-off-by: Chris Wilson commit d1713941e9db3e7a6d83466be1b253978fb4bf01 Author: Chris Wilson Date: Fri Jun 15 15:29:59 2012 +0100 sna: Tweaks for DBG missing glyphs through fallbacks Signed-off-by: Chris Wilson commit 2b23605efba009fb340ec10b37d54caae159b9b1 Author: Chris Wilson Date: Fri Jun 15 15:28:31 2012 +0100 sna: Don't trim prepare for glyphs_via_mask If we pass the expected width/height without passing the per-glyph offset into the preparation function, we make the erroneous mistake of analysing the glyph cache only for the mask extents and so will miss glyphs that we need to upload for the operation. Signed-off-by: Chris Wilson commit 9f66b27114fcc457fa5cb2d5889e875384f89e75 Author: Chris Wilson Date: Fri Jun 15 13:47:33 2012 +0100 sna: Remove mark-as-cpu after gem_pread The kernel no longer moves the read bo into the CPU domain, so remove the last vestiges of that tracking. Signed-off-by: Chris Wilson commit bfd88d5429528b0162eafcc9496bfd1c708b60cd Author: Chris Wilson Date: Fri Jun 15 11:21:29 2012 +0100 sna: Silence a few compiler warnings Nothing harmful, just noise. Signed-off-by: Chris Wilson commit b3659beec8c4126de5da4df5cb2cc077b7af6c0b Author: Chris Wilson Date: Fri Jun 15 11:08:18 2012 +0100 sna: Check for zero-sized damage regions after trimming Windows to their pixmaps Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 293a867ea55d3004f5be21b1d0ad765a89c28a5a Author: Chris Wilson Date: Thu Jun 14 23:02:10 2012 +0100 sna: Specialise the self-copy blitter to handle vertically overlapping copies Signed-off-by: Chris Wilson commit 3b76ab995bb70da5f76afead7c696cb7fd72b158 Author: Chris Wilson Date: Thu Jun 14 22:33:11 2012 +0100 sna: Add a little bit more DBG to migration IO paths Signed-off-by: Chris Wilson commit 49543522e789a0a8213c164b37a18c0642e60ac4 Author: Chris Wilson Date: Thu Jun 14 20:25:21 2012 +0100 sna: Fix reversed assertion Darn the double negative! After adding damage we want to assert that the region is not empty, not !not empty. Signed-off-by: Chris Wilson commit 4e50467b4f50b522a07bedae7ff68b9c37362af8 Author: Chris Wilson Date: Thu Jun 14 17:40:09 2012 +0100 uxa: Force the outputs to off for consistency with xf86DisableUnusedFunctions() Upon a VT switch, we set the desired modes and turn off the DPMS on any unused output. Make this explicit so that we always maintain consistency between the kernel and X's list of enabled CRTCs. References: https://bugs.freedesktop.org/show_bug.cgi?id=50772 Signed-off-by: Chris Wilson commit 3a56a0b10eb3ab07efd28efdc79d0fd3a6fa20ff Author: Chris Wilson Date: Thu Jun 14 19:26:42 2012 +0100 sna: Make the disable-unused after vt switch distinct from DPMS off So that we do not lose control over dpms on/off! Signed-off-by: Chris Wilson commit b415ca05c2e1c4f09f85d8769d39e5369661ed3a Author: Chris Wilson Date: Thu Jun 14 17:06:19 2012 +0100 sna: Reset damage extents after reduction goes to zero If the reduction of the damage clears all of the boxes, we need to reset the -infinite extents so that we continue to accumulate further damage. Reported-by: Zdenek Kabelac References: https://bugs.freedesktop.org/show_bug.cgi?id=50744 Signed-off-by: Chris Wilson commit 0df7c488640d3590d2a88dc353b72167b6644eaa Author: Chris Wilson Date: Thu Jun 14 16:09:38 2012 +0100 sna: Supporting overlapping copies for fallback blits Reported-by: Zdenek Kabelac References: https://bugs.freedesktop.org/show_bug.cgi?id=50393 Signed-off-by: Chris Wilson commit d5b273b20b7d917d08af120815c28b1618d3342c Author: Chris Wilson Date: Thu Jun 14 15:35:24 2012 +0100 sna: Explicitly disable the CRTC so that xserver/kernel agree after VT switch References: https://bugs.freedesktop.org/show_bug.cgi?id=50772 Signed-off-by: Chris Wilson commit 221534abe2dc04fae8b8fc332104bca275d4863b Author: Chris Wilson Date: Wed Jun 13 22:48:43 2012 +0100 sna: Only reduce damage addition to a region operation if clean Some paths bypass operating upon the region as they do not have an YX-banded set of boxes and so prefer to defer the costly construction of the region till later. As a result, we have to be careful not to overwrite any existing information if we do operate on the region after setting the dirty boxes. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=50744 Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 0db789e180b6b01fb8aff547879387058f52a0b9 Author: Zdenek Kabelac Date: Wed Jun 13 14:26:37 2012 +0200 sna: Constification Adding preserving const modifiers to decrease amount of const warnings Signed-off-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 33998a7080aa7f50ba922c764c6a93fe951c5b64 Author: Zdenek Kabelac Date: Wed Jun 13 14:26:36 2012 +0200 sna: Fix typo and compare y1 to y2 It seems 'y1' was meant to be compared here Signed-off-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 0ade32fbad4014efcc8aa8ed4cd0f8a39f9ab107 Author: Chris Wilson Date: Wed Jun 13 14:28:43 2012 +0100 sna: Fix operator preference around a | b & c | d Tell the compiler what we really mean is a | (b & (c | d)) Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 20f3114ccd63052b4fd06ef1d87efaaabbbac7ac Author: Chris Wilson Date: Wed Jun 13 14:27:10 2012 +0100 sna: Use long for simplicity when printing size_t values for debugging Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 6f75b8da6a468188dd0c00015395424598b3f502 Author: Zdenek Kabelac Date: Wed Jun 13 14:26:33 2012 +0200 sna: Valgrind updates Add some mising initialization for unknown ioctl Signed-off-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 70cf637ca47c9e67758e09c7e1d10eef51db1f98 Author: Zdenek Kabelac Date: Wed Jun 13 14:26:32 2012 +0200 legacy/i810: Fix declaration to match definition Signed-off-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 5fea2478cd5ab4156c182210d28c5e27e5f67403 Author: Chris Wilson Date: Wed Jun 13 14:10:01 2012 +0100 Lots of trivial semantic fixes Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit e499f207c161d1b3cd75f065dc89021ff5f40b63 Author: Chris Wilson Date: Wed Jun 13 13:11:17 2012 +0100 sna: Fix memleak from sna_crtc_find_plane() Signed-off-by: Chris Wilson commit 49da55da518348fc6b88e09d5132dd1b1d751304 Author: Chris Wilson Date: Wed Jun 13 11:20:41 2012 +0100 sna: Free the reversed glyph bits along with the font Signed-off-by: Chris Wilson commit 9f68c7fe6aa34ab17d82489fca5f63a3ce335444 Author: Chris Wilson Date: Wed Jun 13 11:15:42 2012 +0100 sna: Free clip rectangles after intersection with PolyRectangle extents Signed-off-by: Chris Wilson commit ab3b7fe31b5a9d7924e959f21d29c4f7352ec8cb Author: Chris Wilson Date: Wed Jun 13 10:35:14 2012 +0100 sna: Only reuse a write buffer if all external references have been dropped This avoids the unhappy situation of overwriting an upload buffer that we intend to use for a fallback. Signed-off-by: Chris Wilson commit e8cb656dc4bab1489df9dfb4cb64704b2a9fa34b Author: Chris Wilson Date: Wed Jun 13 08:10:10 2012 +0100 sna: Free clip intersection with line extents after use Signed-off-by: Chris Wilson commit 572745eb243a96819c8fa248f4ffdd703ed13b0f Author: Chris Wilson Date: Tue Jun 12 22:49:05 2012 +0100 sna/damage: Remove unused return value Signed-off-by: Chris Wilson commit aefc0417dde8b2137a787459a69d91c14902f22b Author: Chris Wilson Date: Tue Jun 12 22:47:41 2012 +0100 sna/damage: Simplify initialisation of damage extents Just use the already computed region->extents when adding the first box. Signed-off-by: Chris Wilson commit 78a96e812b912c5b25a02670f603f455f93e9d00 Author: Chris Wilson Date: Tue Jun 12 21:43:31 2012 +0100 sna: Refresh the damage self-tests They had bitrotted with the revamp and needed some massaging to update to the new interfaces. Signed-off-by: Chris Wilson commit 6138f7434a6fb014299a7e9a8392ef962c8cba30 Author: Cyril Brulebois Date: Tue Jun 12 21:19:14 2012 +0100 Fix up braindamage in previous commit. ickle: Fixing up my idiotic change, obviously too much birthday cake. commit 224d631a232d3536e0f1f2780bfbe1218ee7590f Author: Cyril Brulebois Date: Tue Jun 12 21:14:53 2012 +0100 Avoid calling xf86nameCompare() with a NULL string Device sections without a Driver property would lead to a server segfault because of a NULL pointer's being passed as the second argument of xf86nameCompare(). Debian bug #677206 Signed-off-by: Cyril Brulebois commit 3b9b64c7c9b5b0bfaafb97c9a9fe5849bbb412da Author: Dave Airlie Date: Tue Jun 12 10:26:34 2012 +0100 uxa: do copy fb at startup. Copy the current framebuffer for smooth wayland->gdm handoff. This has been hanging around in Fedora for too long now, and we've dropped the feature a few times, and yes I know the Simpsons did it^W^W^W SNA does it. I've updated the code to have some of the better fixes from nouveau. I've no idea who wrote this code either, krh or ajax. [ickle: The earliest version I've found had krh's fingerprints on it, though it may still have been a joint effort.] Signed-off-by: Dave Airlie [ickle: improve error handling, only copy the fb during initial takeover] commit b6525702b9ffd21beb8ea6bb10a8ad5ce7f9de14 Author: Chris Wilson Date: Tue Jun 12 11:59:31 2012 +0100 sna: Hook up AdjustFrame() Looks to be unused by the core, just a solitary invocation in an obscure extension it seems. However the implementation looks trivial so incorporate it until it is finally removed, just in case. Signed-off-by: Chris Wilson commit 7b281d1300127d65392aaadbbe2299fa9e1749a9 Author: Chris Wilson Date: Tue Jun 12 10:37:04 2012 +0100 sna: Remove a pair of empty stub functions The export wrappers for the glyph cache constructor/destructor existed in case there was a need to add more routines. Since that never happened, remove the extra step of indirection. Signed-off-by: Chris Wilson commit 392e33a62d729c64c57699505220b4029e015470 Author: Chris Wilson Date: Mon Jun 11 09:51:41 2012 +0100 sna: Instrument memory/bo allocations for monitoring over time Hide it behind --enable-debug=memory to avoid incurring the cost for everybody. Signed-off-by: Chris Wilson commit 4e984b79cce038ba5026d8bdd93dec9b5a6ced8a Author: Chris Wilson Date: Mon Jun 11 09:02:03 2012 +0100 configure: Correct the help string for --with-default-accel Prior to finalizing the defaults I changed my mind and realised that the default had to reflect the current behaviour of someone enabling SNA for the first time, and not the previous behaviour of --enable-sna to override UXA. This is so that distro's could offer an SNA enabled DDX for the brave whilst not affecting their typical no-xorg.conf users. Signed-off-by: Chris Wilson commit 7614a541886dd9eb5c9d441e2618785e9cc53a65 Author: Chris Wilson Date: Sun Jun 10 09:36:38 2012 +0100 configure: Harden --with-default-accel against simple mistakes If the user specifies no options, assume automatic selection. Then double check we found a valid backend and so avoid later breaking the build. Signed-off-by: Chris Wilson commit ea0209fbc12fca282bd8ef72223791bf163801ce Author: Chris Wilson Date: Fri Jun 8 22:44:23 2012 +0100 sna: Remove option to disable vmap It works everywhere, so when it is finally enabled, let it be. Signed-off-by: Chris Wilson commit 18baa1428b1c4c0a49927d4c32db091db9e4e97b Author: Chris Wilson Date: Fri Jun 8 22:43:01 2012 +0100 Make the enum list of options consistent with the actual table. Signed-off-by: Chris Wilson commit adc872a9654dc18d778323ca0721704878ad3851 Author: Chris Wilson Date: Fri Jun 8 06:44:20 2012 +0100 sna: Add DBG spew to flink() Signed-off-by: Chris Wilson commit a62ad4e80722db187766c83a16fa84ec236cc5eb Author: Chris Wilson Date: Thu Jun 7 23:37:59 2012 +0100 sna: Skip flushing the active queue if there is not a suitable bo pending Signed-off-by: Chris Wilson commit 318982566bbc7145847bd03601087150eef7a8d8 Author: Chris Wilson Date: Thu Jun 7 21:56:04 2012 +0100 sna/dri: Disable experimental code by default Since these require non-upstream patches to other components, we don't want it enabled by default and randomly breaking builds. Signed-off-by: Chris Wilson commit ebf84b8e572b5cb1a509000d412dfa5be3d1aca3 Author: Chris Wilson Date: Thu Jun 7 20:15:53 2012 +0100 sna/trapezoids: Micro-optimise cell allocation The pool is a fixed size so we can delete the generic code to handle variable sizes. Signed-off-by: Chris Wilson commit 0a25fc68c5cd82cad4b99b0f2357f430c8783c3f Author: Chris Wilson Date: Thu Jun 7 19:13:09 2012 +0100 sna: Actually create Y-tiled source pixmaps An inconsistency highlighted by 7c51cabaecac revealed that we had a mismatch between the check in move_to_gpu() and how we created the pixmap. This mismatch resulted in us creating and uploading tiled pixmaps for single shot textures, and the increase aperture pressure was causing a regression in firefox-fishbowl on pnv, for example. Signed-off-by: Chris Wilson commit c58d137d3eeb0e97bfd53e68404e04d9012b5697 Author: Chris Wilson Date: Thu Jun 7 11:18:27 2012 +0100 sna: Suppress DBG prints along potential SIGIO paths As ErrorF/fprintf is not re-entrant due to its mutex we can not use DBG from code that could be called by a signal handler. X's SIGIO handler attempts to move the cursor from within the handler (eek!) and so we need to be careful not to take any locks, such in as the aforementioned fprintf, along the cursor paths. Reported-by: Zdenek Kabelac References: https://bugs.freedesktop.org/show_bug.cgi?id=50744 Signed-off-by: Chris Wilson commit 3f5b94f3d1625b06840c6441a0b175604ee3d2f9 Author: Chris Wilson Date: Wed Jun 6 17:24:07 2012 +0100 sna: Check against integer overflows when computing cache size Even with a 1nm process, I doubt we will see 4+GiB cache sizes ;-) Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 902391bd798775e1a7a53503d4dd1756162f737f Author: Chris Wilson Date: Wed Jun 6 17:20:39 2012 +0100 uxa: Remove dead-code for SourcePictures All SourcePictures are now converted into Drawables, which had been assumed by the driver backend. However, the code still existed to attempt to pass procedural pictures onwards and so set pSrcPix to NULL which was being flagged by the static analyser as a potential NULL dereference. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 3d8a1f7176877975a31ad7a6548fa8309065f617 Author: Chris Wilson Date: Wed Jun 6 17:15:42 2012 +0100 sna/gen4+: Add missing "fall through" comments Reported-by: Signed-off-by: Chris Wilson commit 8ae4407c43e6a8d26784508f61b416138f908132 Author: Chris Wilson Date: Wed Jun 6 17:13:05 2012 +0100 sna: Silence static analyser complaining about potential NULL pointer Add an assert to prove that is not. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 08010b23a3e2bc37d202251923fac814b18fa501 Author: Chris Wilson Date: Wed Jun 6 17:10:35 2012 +0100 uxa/dri: Silence static analyser for potential NULL intel_pixmap If the intel_pixmap was NULL we should have failed to create the DRI2 buffer, so we can safely assert here to keep the analyser quiet. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 06b1b875ba13227ddaf7f28dbdcdaa3eb49f0857 Author: Chris Wilson Date: Wed Jun 6 17:05:13 2012 +0100 uxa/i915: check for failure to allocate temporary destination If the target drawable is too large for the render pipeline, we need to create a temporary surface. This may fail, so abort if it does. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit c553dcae2dd714cac413ffc7c7779cd78c9a3e61 Author: Chris Wilson Date: Wed Jun 6 17:04:01 2012 +0100 sna: Silence a few unused function warnings Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 536033b4edb40db18210e5272eb8ce9703e5e364 Author: Chris Wilson Date: Wed Jun 6 16:02:47 2012 +0100 legacy/i810/dri: Check for malloc failure for BusIdString Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit b2da80c1d1058dca5d3d85693918e53d9fa61dd3 Author: Chris Wilson Date: Wed Jun 6 15:59:54 2012 +0100 sna: Add some missing 'fall through' comments Static analysers are dumb and presume readers are too. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit c433fb4521e2fb65a555489bd09ba6bb9448b72f Author: Chris Wilson Date: Wed Jun 6 15:17:30 2012 +0100 uxa/dri: Fix up typo from f2513cb0f s/true/FALSE/? Wrong. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 1675f441049d559d891fe378e437c12db21df9e7 Author: Chris Wilson Date: Wed Jun 6 12:55:13 2012 +0100 uxa/dri: Transfer reference of the new_back buffer to the flip_chain In order to prevent a leak of the bo when the chain is eventually torn-down when the client exits. Reported-by: Andreas Lampersperger Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=50670 Signed-off-by: Chris Wilson commit 18726a4975ab2ddf85eaa6eb1602dcbe599217f7 Author: Chris Wilson Date: Wed Jun 6 10:58:49 2012 +0100 sna: Perform CopyArea directly onto a CPU bo if available Signed-off-by: Chris Wilson commit 57d7d5de78bcf01d75d7a7de03fe50a2a9bd1b7e Author: Chris Wilson Date: Wed Jun 6 00:08:17 2012 +0100 sna: Use GPU for readback onto CPU bo Time to blt from GTT to LLC 16384 bytes: 125.000µs (snb) Time to blt from GTT to LLC 16384 bytes: 71.000µs (ivb) Time to blt from GTT to LLC 1048576 bytes: 1400.000µs (snb) Time to blt from GTT to LLC 1048576 bytes: 938.000µs (ivb) Time to copy from GTT to LLC 16384 bytes: 118.000µs (snb) Time to copy from GTT to LLC 16384 bytes: 134.000µs (ivb) Time to copy from GTT to LLC 1048576 bytes: 6723.000µs (snb) Time to copy from GTT to LLC 1048576 bytes: 7424.000µs (ivb) And conversely, Time to blt from LLC to GTT 16384 bytes: 10.000µs (snb) Time to blt from LLC to GTT 16384 bytes: 8.000µs (ivb) Time to blt from LLC to GTT 1048576 bytes: 217.000µs (snb) Time to blt from LLC to GTT 1048576 bytes: 135.000µs (ivb) Time to copy from LLC to GTT 16384 bytes: 4.000µs (snb) Time to copy from LLC to GTT 16384 bytes: 4.000µs (ivb) Time to copy from LLC to GTT 1048576 bytes: 270.000µs (snb) Time to copy from LLC to GTT 1048576 bytes: 179.500µs (ivb) It seems clear then that even with the extra synchronisation cost copying from the GTT is much preferable with the GPU than using the uncached reads by the CPU. Streaming write-combines from the CPU into the GTT seem about as efficient as we can manage, so continue to use the mapping unless busy. Signed-off-by: Chris Wilson commit f2513cb0fdb0d1214854fd4e4dcd477ba8583862 Author: Chris Wilson Date: Wed Jun 6 10:41:35 2012 +0100 uxa/dri: Do not use undeclared stdbool features The header isn't pulled in, so stop using the undefined values of true/false. Signed-off-by: Chris Wilson commit 1dafb4777f8378c87f34feae667582498220204c Author: Chris Wilson Date: Tue Jun 5 21:25:40 2012 +0100 sna/damage: Add some assertions to validate that each damage box is non-empty References: https://bugs.freedesktop.org/show_bug.cgi?id=50744 Signed-off-by: Chris Wilson commit c4eb5528a456b65c673f7c984d14a622ac67cdca Author: Chris Wilson Date: Tue Jun 5 16:04:16 2012 +0100 uxa: Check for DPMS off before scheduling a WAIT_ON_EVENT Regression from commit 3f3bde4f0c72f6f31aae322bcdc20b95eade6631 Author: Chris Wilson Date: Thu May 24 11:58:46 2012 +0100 uxa: Only consider an output valid if the kernel reports it attached When backporting from SNA, a key difference that UXA does not track DPMS state in its enabled flag and that a DPMS off CRTC is still bound to the fb. So we do need to rescan the outputs and check that we have a connector enabled *and* the pipe is running prior to emitting a scanline wait. References: https://bugs.freedesktop.org/show_bug.cgi?id=50668 Signed-off-by: Chris Wilson commit 7c51cabaecac52348766c622e80ed14b9854e54d Author: Chris Wilson Date: Tue Jun 5 12:59:46 2012 +0100 sna: Try to create Y-tiled pixmaps for initial source bo Signed-off-by: Chris Wilson commit a26c05cc8936cab28d83c6beeff906a910353338 Author: Chris Wilson Date: Tue Jun 5 12:48:33 2012 +0100 sna/gen2+: Tweak placement of operations for CPU-bound large pixmaps Try to avoid uncessary migration to the GPU of large pixmaps that are wholly bound to the CPU. Signed-off-by: Chris Wilson commit a21bdbe3e312b40b936c5c68c84f5c1bc0f9fb88 Author: Dave Airlie Date: Tue May 22 15:44:28 2012 +0100 sna: port to compat api This ports SNA to the new compat-api. Signed-off-by: Dave Airlie commit 8f5001493e18b6b3b97ca5f0c923678a51975bd0 Author: Dave Airlie Date: Tue May 22 15:35:01 2012 +0100 intel: port legacy and uxa to new API. This ports the legacy and uxa driver to the new server API. Signed-off-by: Dave Airlie commit c9824827422f8ec4f46c9cba42d871c98dc54761 Author: Dave Airlie Date: Tue Jun 5 10:38:21 2012 +0100 sna: drop using block/wakeup data. These went away in the new server API, and really if this made any measurable difference, I'd be impressed. Signed-off-by: Dave Airlie commit d9850a05003008d86d0b226bcd775aebdd43bccd Author: Dave Airlie Date: Tue Jun 5 10:04:48 2012 +0100 uxa: drop enable disable access hook This looks to be unused. Signed-off-by: Dave Airlie commit 99df720216787c52d16a350dd25469e2dcea2c47 Author: Chris Wilson Date: Tue Jun 5 10:24:34 2012 +0100 sna: Add inline keyword in conjunction with attribute(always_inline) Signed-off-by: Chris Wilson commit 47d7da5a986a9f133cd3dc8314ff243001586a36 Author: Chris Wilson Date: Tue Jun 5 10:18:40 2012 +0100 sna: Maintain the operation on the CPU if not completely replacing all-damaged We may as well continue the operation on the CPU if already bound to try to reduce needless migrations (where the cost of the upload will outweigh further use). Signed-off-by: Chris Wilson commit 81f09347f2ab59cf0a3eaca7be83ded555655e93 Author: Chris Wilson Date: Mon Jun 4 16:45:05 2012 +0100 uxa/i965: Silence static analyser by asserting the bo exists for the video This is already checked at the beginning of PutImageTextured, so this check upon the return value of intel_get_pixmap_bo() should only be required to keep static analysers happy. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 1f43de322b400dcd64eb4545a978ad9b1c7de185 Author: Chris Wilson Date: Mon Jun 4 15:39:39 2012 +0100 sna: Exclude consideration of tiling flags from overwriting BLT commands Signed-off-by: Chris Wilson commit ebb1c9d5f82e8822f7400ff11a887ab047a0d78e Author: Chris Wilson Date: Mon Jun 4 15:28:31 2012 +0100 sna: Make the bo-is-busy DBG more useful by saying which bo it is Signed-off-by: Chris Wilson commit 6dc5ddfe807b307ca10af971c4f84498b2fb82a2 Author: Chris Wilson Date: Mon Jun 4 13:49:35 2012 +0100 sna: Remove bogus check for color == 0 This was written when the RGBA color value was being passed around and not the pointer to the xRenderColor. As such, the NULL deref check doesn't gain much and the check for rgba==0 irrelevant in this scenario. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 9b3937228c692598899fb39bfe448a9b457315db Author: Chris Wilson Date: Mon Jun 4 13:47:58 2012 +0100 sna: Simplify selecting default tiling for framebuffers Signed-off-by: Chris Wilson commit ff91fd4a2749f1ea8722189aaec938e51bce4222 Author: Chris Wilson Date: Mon Jun 4 12:41:05 2012 +0100 uxa: Check for failure from drmModeGetConnector() Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 4712a4008aaff7b3c57a8dd0a5e639992c9b6d30 Author: Chris Wilson Date: Mon Jun 4 12:40:56 2012 +0100 shadow commit 63a55e9db97b647cbf7385403a0f30ee4d24dc3b Author: Chris Wilson Date: Mon Jun 4 12:36:11 2012 +0100 uxa/shadow: Free the buffer along the unexpected failure to attach to the pixmap Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit c9bd4ed7d7d5c9ba6924e911652e03abc9267865 Author: Chris Wilson Date: Mon Jun 4 12:29:58 2012 +0100 legacy/i810: Silence a compiler warning for missing 'const' Signed-off-by: Chris Wilson commit 6f8b411b8a833cca6fce8888ce515db6413d57ca Author: Chris Wilson Date: Mon Jun 4 12:21:27 2012 +0100 uxa: Preferred order for calloc is (count, size) And assert that the allocation succeeds for good measure. A great big dollop of graceful failure is missing... Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 72a29aa74a1488269c2184300e573d060c49e232 Author: Chris Wilson Date: Mon Jun 4 12:21:27 2012 +0100 sna: Check that the buffer was allocated prior to caching it Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 9bff89b1892cee90a80909255876240126b700b2 Author: Chris Wilson Date: Mon Jun 4 12:21:27 2012 +0100 legacy/i810: Be paranoid and guard against xf86GetPciInfoForEntity Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit ed6004f7ad09329bd4097331aaff38edc83107b6 Author: Chris Wilson Date: Mon Jun 4 12:21:27 2012 +0100 uxa: Be paranoid and guard against xf86GetPciInfoForEntity Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit abb8d893742f4ba9567330da9706adda4e636e75 Author: Chris Wilson Date: Mon Jun 4 12:21:27 2012 +0100 sna: Be paranoid and guard against xf86GetPciInfoForEntity Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 782cd6eafa9917a5bdb181d66b7f4c05392950ba Author: Chris Wilson Date: Mon Jun 4 12:20:17 2012 +0100 uxa: check for failure to allocate drmModeCrtc Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit a99bf0125f8f166ff2d65449ecd376c99c3024b8 Author: Chris Wilson Date: Mon Jun 4 12:15:51 2012 +0100 uxa/i965: Drop superfluous 'state' variable Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 15a00ba047fdb7d388a322f2e33894d20787af2f Author: Chris Wilson Date: Mon Jun 4 12:09:52 2012 +0100 sna/gen2: Remove double 'const' So constant, we const'ed it twice. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 99129c369c16d9ff7588a0e59f9285cc339b7eb3 Author: Chris Wilson Date: Mon Jun 4 12:08:11 2012 +0100 sna/gen4+: Use an explicit invalid value for alu Since the static analyser also doesn't like comparing a uint8_t against ~0. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit fae9c054d47203b68a6c09647945d23074ea4df9 Author: Chris Wilson Date: Mon Jun 4 12:04:30 2012 +0100 legacy/i810: Remove unused variable In order to reduce the volumes of output from static analysers. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 880ec2c9a56d3de33c7fbbfa4c043082fbcf7a78 Author: Chris Wilson Date: Mon Jun 4 12:04:30 2012 +0100 legacy/i810: Reorder DRI teardown code to avoid potential NULL derefs Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 7424ea7dd08e304baa4efa09e887b833737ac9f6 Author: Chris Wilson Date: Mon Jun 4 11:59:24 2012 +0100 uxa/dri: Free wait_info along DRI2WaitMSC error paths Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit ebd6dea009479e612c67d78416b8680bb23dba94 Author: Chris Wilson Date: Mon Jun 4 11:56:11 2012 +0100 uxa/dri: Make sure is_glamor_pixmap is always initialised The code paths are too twisty to be sure otherwise. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 33f0b4b96662d953052232b1c0477ced0c326a62 Author: Chris Wilson Date: Mon Jun 4 11:54:28 2012 +0100 legacy/i810: Free offscreen image info struct on failure Impossible with the current code, the server aborts on failure. However, this looks to be the simple answer to keep static analysers quiet. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit d24340747389db971c04349d1ee517f195c2b28e Author: Chris Wilson Date: Mon Jun 4 11:53:26 2012 +0100 sna: Free clip boxes (if allocated) along error path Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 1215abc5c30cc3e183bde59b1523c09b59c484e3 Author: Chris Wilson Date: Mon Jun 4 11:49:15 2012 +0100 sna: Make the ignored return value explicit The return from __kgem_throttle_retire() is just a hint as to whether any forward progress was made. In the case of the error path, though it is a last ditch effort before aborting, so we do not really care whether or not it succeeds, we must try again. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit e230b460515a043a1b7353d153e864754ca5e064 Author: Chris Wilson Date: Mon Jun 4 11:48:23 2012 +0100 ux/i965: create_sampler_state_bo() expects enums, so feed it the right types Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 072d7a8b42f04600c9d8054f3648642a1aaff57a Author: Chris Wilson Date: Mon Jun 4 11:40:01 2012 +0100 sna/debug: Assert the non-existence of the relocation handle at source This should help the static analyzer pinpoint the blame and make it quieter. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 4a3c355e056339aed68b70470556633dea899b1b Author: Chris Wilson Date: Mon Jun 4 11:32:42 2012 +0100 sna: Silence a compiler warning for loss of 'const' qualifier Signed-off-by: Chris Wilson commit 791029cc16bca36b8dec82297ff7e07a972c51ab Author: Chris Wilson Date: Sat Jun 2 13:28:43 2012 +0100 sna/trapezoids: Implement trapezoidal opaque fills inplace Signed-off-by: Chris Wilson commit 1f78a934a423911e18d340f0585e31941f6e8663 Author: Chris Wilson Date: Sat Jun 2 13:27:52 2012 +0100 sna: Clear the counters prior to querying the property Signed-off-by: Chris Wilson commit a1953f1d4abc6e158a5e3ca53d3207548842254c Author: Chris Wilson Date: Sat Jun 2 12:02:55 2012 +0100 sna: Prefer to use memset() for extremely large clears Signed-off-by: Chris Wilson commit ce85cd1a36e31795a966ea8983c2d6f803a4eccd Author: Chris Wilson Date: Sat Jun 2 11:46:06 2012 +0100 sna: Add some DBG to retreiving EDID Signed-off-by: Chris Wilson commit e8eb273bd6153c232a9ffc558e3b7fd4beaab01b Author: Chris Wilson Date: Sat Jun 2 10:43:03 2012 +0100 sna/gen7: Add DBG for ring switching Signed-off-by: Chris Wilson commit 722afa6bc910a2ccfbb0442872a878d1b6b78315 Author: Chris Wilson Date: Sat Jun 2 10:04:02 2012 +0100 sna: Remove some unused members from the KMS state tracking Signed-off-by: Chris Wilson commit 26e7bb3f25f0c83d39ff505fa7e05bfcb976e39c Author: Chris Wilson Date: Sat Jun 2 09:16:51 2012 +0100 sna: Add a DBG message to indicate flushing for GPU idle Signed-off-by: Chris Wilson commit 58fc03b8c36688e9fa7925aa82b83d36c9decb7e Author: Chris Wilson Date: Sat Jun 2 08:53:24 2012 +0100 sna: Prevent NULL deref with early termination and DBG enabled Signed-off-by: Chris Wilson commit 1be2afb81df887ed664ac31d0641da808cfb8888 Author: Chris Wilson Date: Sat Jun 2 08:04:02 2012 +0100 legacy/i810/video: Release memory after creating adaptors Signed-off-by: Chris Wilson commit 261d086265fff6c9b28a67d2fcceed8f107d5cb0 Author: Chris Wilson Date: Fri Jun 1 23:22:31 2012 +0100 legacy/i810/dri: Propagate failure from allocating texture memory Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 2ea12b56d42686f75deb33fc126d09acee7e2899 Author: Chris Wilson Date: Fri Jun 1 23:22:31 2012 +0100 legacy/i810: Fix memset(sizeof(*ptr)) rather than memset(sizeof(ptr)) Clear the entire structure and not the first 4 bytes... Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit d3e15d1460faad192781d74fdc2b5092e9c148ad Author: Chris Wilson Date: Sat Jun 2 08:14:23 2012 +0100 uxa/i965: Make the unhandled allocation failures explicit Add assertions to the point of allocation and write failures. Signed-off-by: Chris Wilson commit 46fbb3d31bf73df4bd0259a146e9929f62a9c488 Author: Chris Wilson Date: Fri Jun 1 23:22:31 2012 +0100 uxa: NameForAtom may return NULL Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 8451d5dd3c992bcfad9f71aa890b6b225884b978 Author: Chris Wilson Date: Fri Jun 1 23:22:31 2012 +0100 uxa: Use unsigned bitfields for single bits Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit d66ca06865973833398560a04186e0067d9cb206 Author: Chris Wilson Date: Fri Jun 1 23:22:31 2012 +0100 uxa: Intialize is_glamor_pixmap It is not clear whether is_glamor_pixmap is always initialized prior to use, so set it to a safe^W likely value. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 25a17b2ca5e3945150151c4308761891c0e20bab Author: Chris Wilson Date: Fri Jun 1 23:22:31 2012 +0100 uxa: Remove 'render_source_is_solid' dead code Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit ba0eb230836fa5f94a2f50da5880fdd3b9dfd72e Author: Chris Wilson Date: Fri Jun 1 23:22:31 2012 +0100 uxa: Fix reallocation of XVAdaptors array Prevent the leak and remove some unsightly code in the process. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit affb1f82c1401bac3f54654a342e3b16b8bd374b Author: Chris Wilson Date: Sat Jun 2 08:27:59 2012 +0100 sna/gradient: Reuse old gradient bo if allocation of new fails Prefer a stall to a crash. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit e51f984a4c15de42c5bee3aeacee553e2dd4eca0 Author: Chris Wilson Date: Sat Jun 2 08:23:35 2012 +0100 sna: Check gradient allocations during init Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 2c5647a18e7856b604bce70270463fc128c2cfcd Author: Chris Wilson Date: Fri Jun 1 23:22:31 2012 +0100 sna: NameForAtom may return NULL Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 7867bff00caeff9e1fbe569ca3d37b94f704a4cd Author: Chris Wilson Date: Fri Jun 1 23:22:31 2012 +0100 sna: Replace invalid signed value with unsigned ~0 for uint8_t Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit ca72b372237287af0f4475801a9b64efca975832 Author: Chris Wilson Date: Fri Jun 1 23:22:31 2012 +0100 sna: Drop return value from 3D point transform And just fixup the computed coordinates in the face of an invalid matrix. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit b83e2a79f2a082630e0d3147abec8d3599f3f4b4 Author: Chris Wilson Date: Fri Jun 1 17:17:49 2012 +0100 uxa: Static analysis warning fixes A smattering of bugs and confusing code. Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 67e8ac365c4b7dd9c4eb2abe46aecc8cc04d4278 Author: Chris Wilson Date: Fri Jun 1 17:20:51 2012 +0100 sna/gen3+: Guard against a kgem_bo_destroy(NULL) Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit e91b27ae14a5366175a4fe9c87ee15a0a25b5a42 Author: Chris Wilson Date: Fri Jun 1 17:12:36 2012 +0100 sna: Fix potential deference of NULL cpu_bo inside assertion Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 048161e69612324d6c077ec487a2a64aaaa7c1f5 Author: Chris Wilson Date: Fri Jun 1 17:09:21 2012 +0100 sna/trapezoids: Check for malloc failure Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 2896aa7b46931d3b95567121ccb07844725ab45c Author: Chris Wilson Date: Fri Jun 1 17:08:01 2012 +0100 sna/io: Free the heap clip boxes on error paths Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 1ef00ff720226c211444806d9fdf67ce2046685b Author: Chris Wilson Date: Fri Jun 1 17:05:55 2012 +0100 sna/dri: Fix typo, check for NULL after allocations to spot failure! Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit dc0fe44bebfdee63bcda4d185b85fe9e1cb16a71 Author: Chris Wilson Date: Fri Jun 1 16:54:41 2012 +0100 sna: Fix a smattering of static analysis warnings Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 8eed569fb386a9af48a8beb28666d72c6678e48c Author: Chris Wilson Date: Fri Jun 1 13:02:50 2012 +0100 sna/trapezoids: Correct extents declaration for fallback Reported-by: Zdenek Kabelac Signed-off-by: Chris Wilson commit 91419576eef562378cccf90968c4f0277139b03d Author: Chris Wilson Date: Fri Jun 1 12:09:19 2012 +0100 sna: Tiny DBG message tweak Signed-off-by: Chris Wilson commit 4bdecc5b07a184ba136129e75a7fef914ac3b8d2 Author: Chris Wilson Date: Fri Jun 1 11:06:04 2012 +0100 test: Add a very basic test to exercise BLT text drawing Signed-off-by: Chris Wilson commit ea0f326d80f21f89c3cff3595a6e1afc42b52efd Author: Chris Wilson Date: Fri Jun 1 09:04:23 2012 +0100 sna: Debug XDrawRectangle Signed-off-by: Chris Wilson commit 6a7efa991a705bdb5a3b9d5300d824a1ce8759af Author: Chris Wilson Date: Fri Jun 1 08:47:20 2012 +0100 test: Exercise basic rectangle drawing code Signed-off-by: Chris Wilson commit bc4323558bebd53e474fbc5404e1c41ab16d02e6 Author: Chris Wilson Date: Thu May 31 21:19:26 2012 +0100 sna: Handle negative values when computing the stipple modulus Signed-off-by: Chris Wilson commit aca994e03e6e0e16f55841418b0061b175e91e5d Author: Chris Wilson Date: Thu May 31 21:17:24 2012 +0100 sna: Fill early break for clip process of spans When on the same Y-band as the span, as soon as the clip boxes are too far to the right, we can stop searching for more intersections. Signed-off-by: Chris Wilson commit 0fe150f898120ba9a00e1e6b9d66bec10d7e8a29 Author: Chris Wilson Date: Thu May 31 20:02:47 2012 +0100 sna: Fix computation of box for clipped stippled rectangles Signed-off-by: Chris Wilson commit a65c3b7b45df46ba993af6755b2460aa80fde930 Author: Chris Wilson Date: Thu May 31 19:15:05 2012 +0100 sna: Add some more DBG for stippled blts Signed-off-by: Chris Wilson commit b2cc64253217a90cf2ade9a8d6f1ed21b23fe875 Author: Chris Wilson Date: Thu May 31 19:45:56 2012 +0100 sna: And combine the two migrate to CPU clauses Just a small tidy up to remove a duplicated predicate. Signed-off-by: Chris Wilson commit bd1dd4cf663ca230724dc6beb329b7eb270ea9d5 Author: Chris Wilson Date: Thu May 31 17:57:56 2012 +0100 sna: If the CPU damage already contains the migration region, we are done Signed-off-by: Chris Wilson commit a2d92ade3911e62f950c85b5650fb13da1279ced Author: Chris Wilson Date: Thu May 31 17:49:01 2012 +0100 sna: Prevent readback of 1x1 if it lies inside CPU damage If the pixel exists only in the CPU damage, attempting to read it back from the GPU only results in garbage. Signed-off-by: Chris Wilson commit a34a06c4ba36b21e386b1f6371a4bd3f03ff35a1 Author: Chris Wilson Date: Thu May 31 20:14:38 2012 +0100 sna/glyphs: Clip the damage to the drawable for unclipped glyphs-to-dst Signed-off-by: Chris Wilson commit dcc7ba8ccf95db1c265bd5492f041f94ce368b7a Author: Chris Wilson Date: Thu May 31 17:48:40 2012 +0100 sna/composite: Add debugging modes to disable acceleration Signed-off-by: Chris Wilson commit 984400b95a879c7c020de14545ac2e1d9e759e05 Author: Chris Wilson Date: Thu May 31 16:56:21 2012 +0100 sna/glyphs: Don't apply drawable offsets twice The offsets from image_from_pict() already include the drawable offset so we need not apply them twice. Signed-off-by: Chris Wilson commit 7bbd445c8cbeb7bcea2c56c4deb9bf6ccf6f2497 Author: Chris Wilson Date: Thu May 31 16:45:03 2012 +0100 sna/trapezoids: Apply the drawable offset for the CPU migration Signed-off-by: Chris Wilson commit c5313620a2f66bc8daf211fc77a9c480a79ea271 Author: Chris Wilson Date: Thu May 31 15:06:16 2012 +0100 sna: Suppress flushes once we have cleared the kernel dirtiness Signed-off-by: Chris Wilson commit 0347c8abec9740d706d87dd2f719ac68f3d1c32e Author: Chris Wilson Date: Thu May 31 13:50:31 2012 +0100 sna/glyphs: Fix upconverting of glyphs via gpu When using the mask instead of the src for adding the glyphs we need to pass the glyph coordinates through the mask coordinates rather than the source. Signed-off-by: Chris Wilson commit 90ae4f853222ee33206134f4efdc4accfb2f2c38 Author: Chris Wilson Date: Thu May 31 14:17:40 2012 +0100 sna: Avoid mixing signed/unsigned int/int16 arithmetric Life becomes unpleasant with sign extension. Signed-off-by: Chris Wilson commit 660c89e9742bac5ce7cbd480e08b4667e37dee8c Author: Chris Wilson Date: Thu May 31 13:18:21 2012 +0100 sna: Use full 16-bit unsigned values for absolute differences Beware the overflow implicit in: adx = x2 >= x1 ? x2 - x1 : x1 - x2; when both x2 and x1 may be large signed 16-bit values Reported-by: Zdenek Kabelac Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=50532 Signed-off-by: Chris Wilson commit 035c2953751f58225bd6b1fcb4c1275ccb5526cb Author: Chris Wilson Date: Thu May 31 09:29:40 2012 +0100 sna/glyphs: use add(WHITE, glyph, mask) for channel expansion If the glyph format does not match the mask format we can not simply add the two together, but must first perform a channel expansion (or contraction) by multiplying the glyph against a WHITE source. Normally the glyph and the mask are equivalent formats and so we hit the fast path. Signed-off-by: Chris Wilson commit cf5b3e2ebf4ee0330f5421b9377bb512a94ec284 Author: Chris Wilson Date: Wed May 30 14:44:59 2012 +0100 sna: Don't attempt to flush scanout if unattached This was lost in the midst of the OSTimer overhaul. Reported-by: Zdenek Kabelac References: https://bugs.freedesktop.org/show_bug.cgi?id=50393 Signed-off-by: Chris Wilson commit 3a64ba0895035f6567e9a15357396b06c0f2d6dd Author: Chris Wilson Date: Wed May 30 13:51:07 2012 +0100 sna: Convert remaining drmCommands to drmIoctl This helps to silence valgrind Signed-off-by: Chris Wilson commit 47e2db0ba8ae3fbcdf58cba26c58ec0932b4c90b Author: Chris Wilson Date: Wed May 30 10:37:22 2012 +0100 sna: Replace FREELIKE_BLOCK with MAKE_MEM_NOACCESS And similarly for MALLOCLIKE_BLOCK. The issue being that valgrind may overwrite such blocks with alloc-fill and free-fill values, but when in fact they are defined and still in use by the GPU. This can lead to the GPU processing garbage, and GPU hangs. Reported-by: Zdenek Kabelac Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=50393 Signed-off-by: Chris Wilson commit 08a630dc5ef87e551865e558fe4fc45ea66457b4 Author: Chris Wilson Date: Wed May 30 08:50:44 2012 +0100 sna: Ensure we flush scanout even when otherwise idle. Reported-by: Zdenek Kabelac References: https://bugs.freedesktop.org/show_bug.cgi?id=50477 Signed-off-by: Chris Wilson commit 10ca765ad346931087be146f3e6d1f4069c95f26 Author: Chris Wilson Date: Wed May 30 08:32:34 2012 +0100 Mark a few more options as being UXA specific Signed-off-by: Chris Wilson commit 740368c4c6eb547adad247ff529d16e594d6459b Author: Nick Bowler Date: Mon May 28 18:28:28 2012 -0400 Include config.h in intel_options.c Commit 8a9a585341e2 ("Only create a single instance of the intel_options array") moved the definition of intel_options into a separate .c file. Several of the defined options are #ifdef'd based on the configuration, but since config.h is never included, the macros being tested are never defined. Therefore, none of the configuration-specific options will ever be available at runtime, even if they should be. Add an inclusion of config.h so that such configuration-specific options work again. Signed-off-by: Nick Bowler Signed-off-by: Chris Wilson commit e1b720c40cf511b4ed6d387891d715522d0d54b6 Author: Chris Wilson Date: Wed May 30 07:47:53 2012 +0100 sna/sprite: Restore another xf86drm.h Signed-off-by: Chris Wilson commit e986b06162abf286b3f8f97b791fd66dcf8f9aec Author: Chris Wilson Date: Tue May 29 22:21:05 2012 +0100 sna/overlay: Handle return from GETPARAM query correctly drmIoctl() returns 0 on success, not a positive value. Signed-off-by: Chris Wilson commit 68874e2d694cc68f5f051ba5fe4a79a71fdb2dbc Author: Chris Wilson Date: Tue May 29 21:28:39 2012 +0100 sna: Fix chunking for large stipples Reported-and-tested-by: Matti Hamalainen Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=49510 Signed-off-by: Chris Wilson commit 12af82581118470f59817f626d61df655950e243 Author: Chris Wilson Date: Tue May 29 21:45:32 2012 +0100 sna/overlay: Replace drmCommand with direct invocations of drmIoctl Signed-off-by: Chris Wilson commit 3a1b6e57fbbbb4c8e5ac9588537840f4ec176ae6 Author: Chris Wilson Date: Tue May 29 21:37:16 2012 +0100 sna/overlay: Restore one xf86drm.h include This was removed in haste, when it is required for drmIoctl and friends. Signed-off-by: Chris Wilson commit 264d1b1e1fb368f7845a8112bfa2ad2224acabd9 Author: Chris Wilson Date: Tue May 29 13:03:56 2012 +0100 sna: Also check for a free exec slot for the upload buffers As we check before allocating the upload buffer, we can not be certain that we will allocate an already attached bo or that we have a free exec slot. So always check that we have an extra exec slot available - the false positive rate is going to be negligible. Reported-by: Zdenek Kabelac Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=50457 Signed-off-by: Chris Wilson commit c7bcca9a3fb6c669d8b093e6399914693770b0b3 Author: Chris Wilson Date: Tue May 29 12:49:12 2012 +0100 sna: Make finding DPMS property valgrind clean Signed-off-by: Chris Wilson commit 3c074797094dc7675bb6ae549630fdf061ab5258 Author: Chris Wilson Date: Tue May 29 12:45:20 2012 +0100 sna: Make GetEDID valgrind clean. Signed-off-by: Chris Wilson commit 560e50c0a5afa0824086557581ca6aa5696b44a2 Author: Chris Wilson Date: Tue May 29 12:36:31 2012 +0100 sna: Make GetEncoder() valgrind clean Signed-off-by: Chris Wilson commit b4b6fa795201379694a03d537064dbca446cfdfd Author: Chris Wilson Date: Tue May 29 11:32:28 2012 +0100 sna: Trim the set of includes Signed-off-by: Chris Wilson commit 9d1a9e666763a3e1270c8f000b1b37a6b926441f Author: Chris Wilson Date: Mon May 28 20:06:35 2012 +0100 sna: Bump experimental ioctl number for vmap Gazumped! Signed-off-by: Chris Wilson commit 55e6f5f220401318529e81f7c96fe0af3b893a0c Author: Chris Wilson Date: Mon May 28 16:21:09 2012 +0100 sna: Use the local function for turning the cursor off prior to release Signed-off-by: Chris Wilson commit 77dd429222922aa1ba7f283553e11e60e4d5c496 Author: Chris Wilson Date: Mon May 28 14:40:23 2012 +0100 sna/dri: Make WAIT_VBLANK valgrind clean Signed-off-by: Chris Wilson commit 8ebe84818ab70e662a9bb3f232b4664ff40ac375 Author: Chris Wilson Date: Mon May 28 14:27:46 2012 +0100 sna: Make sna_copy_fbcon() valgrind clean Signed-off-by: Chris Wilson commit f37dae734ebac5c0ed2d6f50fc92bc107f4e4a33 Author: Chris Wilson Date: Mon May 28 14:20:54 2012 +0100 sna: Just use the CRTC id from the array Signed-off-by: Chris Wilson commit b1f24a0eae4bb0081ff7469a2aee63a1f32140f7 Author: Chris Wilson Date: Mon May 28 14:16:03 2012 +0100 sna: Make sna_crtc_apply() valgrind clean Signed-off-by: Chris Wilson commit 0ebfcdbb80245f98c0c7ef9f090a2be6900fb457 Author: Chris Wilson Date: Mon May 28 14:10:04 2012 +0100 sna: Make AddFB valgrind clean Signed-off-by: Chris Wilson commit 32ca3c7b65516b0bd48b0d0979b086ee1580167a Author: Chris Wilson Date: Mon May 28 14:02:11 2012 +0100 sna: Make sna_crtc_is_bound() valgrind clean Signed-off-by: Chris Wilson commit 7973f6751c9bf565dee4c89aa3e1badbcc45018f Author: Chris Wilson Date: Mon May 28 14:00:55 2012 +0100 sna: Add a little bit more verbosity to cursor routines for valgrind Signed-off-by: Chris Wilson commit 19c463e52df919fc75de7e420fd3565f0e9a0576 Author: Chris Wilson Date: Mon May 28 12:54:35 2012 +0100 sna: Only wait if there is a suitable active buffer There is not point repeating the search after retiring if we know that there is no outstanding suitable active buffer. Signed-off-by: Chris Wilson commit 5b99c7cd340f782d3057d4257865c5feb96b71f0 Author: Chris Wilson Date: Mon May 28 10:55:49 2012 +0100 uxa/dri: Enable vblank scheduling even with pageflipping disabled Signed-off-by: Chris Wilson commit 810357ad65d551ec5d35dbf228f1b62fe235801f Author: Chris Wilson Date: Mon May 28 10:12:38 2012 +0100 uxa/dri: Fix ordering of setting vs using swap members Trivial readibility fix, as the actual ordering is serialised through there being only a single thread. Signed-off-by: Chris Wilson commit 2b56a188ef2c48f82572eca2201a3a0ecf6b4f45 Author: Chris Wilson Date: Mon May 28 10:01:09 2012 +0100 Mark another couple of options as being UXA only. Signed-off-by: Chris Wilson commit b025f1a604dbbbd456a23f330ceed5f97ecdffcf Author: Chris Wilson Date: Mon May 28 09:48:49 2012 +0100 sna/dri: Tweak vblank_mode=n swapbuffers to account for throttle delay Signed-off-by: Chris Wilson commit 45148a714f1461fff898aeb393ad225f081b9bd5 Author: Chris Wilson Date: Mon May 28 09:39:51 2012 +0100 sna/dri: Enable handling for Option "SwapBuffersWait" Signed-off-by: Chris Wilson commit c709f2447dfc6dc36c50ff741d5d9bbdc7c03b58 Author: Chris Wilson Date: Mon May 28 08:17:21 2012 +0100 sna/dri: Requeue vblank throttling until the vsync'ed copy completes If the GPU is busy, then we may not actually schedule our copy for several vblanks, resulting in us falsely reporting that the work completed too early and allowing the client to continue scheduling more work and racing ahead of the queued copies. Signed-off-by: Chris Wilson commit 7e73fa02ed361a9c0c08f61d00421671bf10ce9f Author: Chris Wilson Date: Mon May 28 08:16:43 2012 +0100 sna: Add some debugging to show count of outstanding requests during retire Signed-off-by: Chris Wilson commit 62b557065edc0555f2bf83b0eed9169329a2f2ba Author: Chris Wilson Date: Sun May 27 16:52:51 2012 +0100 sna: Use magic upload buffers for video textures So that we may benefit from the caching of buffers and the automatic selection of the preferred upload method. Signed-off-by: Chris Wilson commit 35291d2db813f75fedcdca9920a40592acd3cca3 Author: Chris Wilson Date: Sun May 27 16:54:29 2012 +0100 sna: Search the inactive bo cache for a mappable upload buffer See if we have a bo that we can cheaply map to an inplace upload, rather than rely on an existing GTT map. Signed-off-by: Chris Wilson commit 36d53ff52b055b2b5fc03aff7b2cab83037d9f42 Author: Chris Wilson Date: Sun May 27 16:46:42 2012 +0100 sna: Validate all CRTCs after updating one Updating one CRTC may cause the kernel to turn off another, so be paranoid and run the check in a loop after applying the CRTC set. Signed-off-by: Chris Wilson commit 7fdd8bd6d2ba7bf113d4109e6bb6f750f909565f Author: Johannes Obermayr Date: Sat May 26 23:52:28 2012 +0200 glamor: Fix misspelled xf86GetOptValString and OPTION_ACCEL_METHOD introduced by commit e456291. Signed-off-by: Chris Wilson commit 8ea4ba081de0206351394481f54dcbe6922a085b Author: Chris Wilson Date: Sat May 26 21:48:16 2012 +0100 sna: Fix typo for debug compilation s/ctrc/crtc/ Signed-off-by: Chris Wilson commit 317bf05196086eb5dedb6436c07f253f01c9bf63 Author: Chris Wilson Date: Fri May 25 14:34:16 2012 +0100 sna: Perform the selftest after probing We do no want to slow down the detection phase by performing our self-tests, so only running those before initialising the driver. Signed-off-by: Chris Wilson commit a5fe863e3f1a6750edaffb518c4e768e9bc11521 Author: Chris Wilson Date: Fri May 25 14:28:56 2012 +0100 sna: Check the bus type before declaring probe success This should never fail, but still better to fail during detection rather than pretend it works. Signed-off-by: Chris Wilson commit 4094826aee76ef24dad13bc5a8a723bfe4a69162 Author: Chris Wilson Date: Fri May 25 08:55:13 2012 +0100 sna: Trust the crtc-is-bound determination after modeset and hotplug As these should be the only time that they change and we now have the checks in place, we can drop the workaround of doing the check just before emitting the wait. Signed-off-by: Chris Wilson commit 8a9a585341e2dd43c649204fcf6d92a867671ba3 Author: Chris Wilson Date: Fri May 25 08:50:33 2012 +0100 Only create a single instance of the intel_options array Signed-off-by: Chris Wilson commit e45629135065d0cc73c285f8df35ab4e1d07c6dc Author: Chris Wilson Date: Thu May 24 18:22:45 2012 +0100 Allow runtime switching of AccelMethod between uxa/sna and even glamor Section "Device" Option "AccelMethod" "uxa/glamor/sna" EndSection The appropriate backend must also be enabled at compile time for the runtime option to be available (i.e. --enable-uxa (default) --enable-sna --enable-glamor) Demanded-by: Adam Jackson Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=50290 Signed-off-by: Chris Wilson commit df6ab02c3690eea8393ecc8c113e2f2891856cc6 Author: Eugeni Dodonov Date: Mon Oct 31 14:43:22 2011 -0200 Unify options handling between UXA and SNA Unifies available options for both UXA and SNA drivers, and moves them into a common header file, intel_opts.h. Signed-off-by: Eugeni Dodonov commit ae2be7e25bda46551381c19a673b321b4382e1f9 Author: Chris Wilson Date: Thu May 24 16:33:27 2012 +0100 sna/trapezoids: Correct rounding for downsampling onto sample grid Reported-by: S. Christian Collins Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=49446 Signed-off-by: Chris Wilson commit 0ab226e27e7920bdb9f7eb62c5174cd097ac7f7f Author: Chris Wilson Date: Thu May 24 12:51:46 2012 +0100 sna: Query CRTC states following a hotplug event Signed-off-by: Chris Wilson commit 3f3bde4f0c72f6f31aae322bcdc20b95eade6631 Author: Chris Wilson Date: Thu May 24 11:58:46 2012 +0100 uxa: Only consider an output valid if the kernel reports it attached Reported-by: Kyle Hill Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=50078 Signed-off-by: Chris Wilson commit 11db66fedf96f158cbbac8011a8ba0b29a20ba3a Author: Chris Wilson Date: Thu May 24 11:07:19 2012 +0100 sna: Add some more DBG tracepoints around modesetting Signed-off-by: Chris Wilson commit fb9da4cb9e12ca733c31879b8b37906f361a8c35 Author: Chris Wilson Date: Wed May 23 13:53:37 2012 +0100 sna: Only mark an unattached output as inactive So that a latter attempt to set the DesiredMode may succeed and we do not modify the configuration without notifying clients. Signed-off-by: Chris Wilson commit 5ae032e22b127c7c95753197e0914a8028a3b22e Author: Chris Wilson Date: Wed May 23 13:27:07 2012 +0100 sna: Suppress modesetting errors on return from VT switch If we presume that userspace will set the correct mode shortly afterwards, we can ignore the failure of the automatic restore. Signed-off-by: Chris Wilson commit 34882a979d9817d33bd6a8ae73a9f7083556578c Author: Chris Wilson Date: Wed May 23 10:27:44 2012 +0100 sna: Keep checking retirement for flushing list Even after all outstanding requests have been completed we may still have buffers on the flushing list that need to become idle. Once such consequence would be to prevent the vblank flush from going idle, causing spurious wakeups every vrefresh when otherwise idle. References: https://bugs.freedesktop.org/show_bug.cgi?id=50078 Signed-off-by: Chris Wilson commit 54fde5bf05d43f26e7f1893584af736a138d245f Author: Dave Airlie Date: Wed May 16 15:34:08 2012 +0100 intel: convert sna/uxa to using new glyph picture accessors. The compat-api.h takes care of old/new servers. Signed-off-by: Dave Airlie commit 43a34186d13d29c671431832469ca5301751b3cf Author: Dave Airlie Date: Wed May 16 15:26:55 2012 +0100 intel: convert to new screen conversion APIs The compat header takes care of the old server vs new server. this commit was autogenerated from util/modular/x-driver-screen-scrn-conv.sh Signed-off-by: Dave Airlie commit 9aabc04d76fa976abb4a74f4e9676c7556e88df3 Author: Dave Airlie Date: Wed May 16 15:14:32 2012 +0100 intel: add compat header file for conversion fns Signed-off-by: Dave Airlie commit c9ce6ae8a6fc7b1a91aef1e59640c39024d0426d Author: Dave Airlie Date: Wed May 23 08:33:20 2012 +0100 legacy/i810: rename pScreen->screen For consistency before moving to new APIs. This just changes the files where the API changes will touch. Signed-off-by: Dave Airlie commit 70ee0f84f5b9c6c13a0338f8b2757e587c849c14 Author: Dave Airlie Date: Wed May 23 08:31:52 2012 +0100 legacy/i810: rename pScrn->scrn This is need to make the compat api stuff easier to implement. Signed-off-by: Dave Airlie commit 3410db008c6c20dc6590a89c05b40a0c6de6744a Author: Chris Wilson Date: Tue May 22 15:54:52 2012 +0100 sna/dri: Avoid confusion of attempting to exchange pixmaps for DPMS-off flips Not only do we confuse ourselves, but we end up confusing the damage tracking on the root window whenever we update the Screen Pixmap. So for the time being, don't. Signed-off-by: Chris Wilson commit 28c4fb0b70e098972af972d30a660d167a8973ad Author: Chris Wilson Date: Tue May 22 09:49:53 2012 +0100 sna: Swallow disconnection event upon mode restoration So that we can VT switch back to X even though the output configuration has now changed. Signed-off-by: Chris Wilson commit b17c44c8d234cb2c651e88c7892bcd5b33139c1c Author: Chris Wilson Date: Tue May 22 09:43:51 2012 +0100 sna: Remove incorrect assertion for mode restoration Signed-off-by: Chris Wilson commit 49b7742999ee11b0c36754ea10bc5609ebe8c609 Author: Chris Wilson Date: Tue May 22 09:40:25 2012 +0100 sna: Force config restoration after VT switch Signed-off-by: Chris Wilson commit 952e3dcba3a651d22d44dbb039ad97e111d9bc63 Author: Chris Wilson Date: Mon May 21 19:48:52 2012 +0100 sna: Only override active on DPMSModeOff Along the enable path we then only want to change the value if we actually call sna_crtc_apply. Signed-off-by: Chris Wilson commit afdaf184594bfe3633305969eb1166c28e1006bf Author: Chris Wilson Date: Mon May 21 14:32:39 2012 +0100 sna: Add a log message for the change of CRTC mode Signed-off-by: Chris Wilson commit 9fb18462ec1428a486f998585e7b4caf4dd08f53 Author: Chris Wilson Date: Mon May 21 14:12:23 2012 +0100 sna: Confirm that the modeset takes If we attempt to change to a mode with a disabled connector the kernel will silently switch off that connector (and crtc) and report that the modeswitch is successful. Reported-by: Kyle Hill References: https://bugs.freedesktop.org/show_bug.cgi?id=50078 Signed-off-by: Chris Wilson commit cd3b48854403ee63723ef570c1a7f9f8447e7105 Author: Chris Wilson Date: Sat May 19 12:56:45 2012 +0100 sna/gen7: Trim a dead assignment Signed-off-by: Chris Wilson commit cf79cd3e27b7077f67f202b16ed39eb1cc70ac65 Author: Chris Wilson Date: Sat May 19 12:44:11 2012 +0100 sna: Emit a new batch if we need to clip after PolyText without space Signed-off-by: Chris Wilson commit f91dcc44dcc15850f82666b1bcdd27182400e7dc Author: Chris Wilson Date: Fri May 18 20:09:41 2012 +0100 sna: Give the proxy a unique name So that if we cache the current destination bo (for example, gen3) then a new proxy (or even just a new batchbuffer) will indeed cause the destination buffer to be updated. Reported-and-tested-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=48636 Signed-off-by: Chris Wilson commit ee073d613bba38f90951405d5ecddfcf3ac5e043 Author: Chris Wilson Date: Fri May 18 15:19:00 2012 +0100 sna/traps: Fix processing of full-steps A missing factor of Y-height caused the computation of coverage for the spans to be completely wrong. This affects the vertical segments of rounded rectangles, for instance. Signed-off-by: Chris Wilson commit 8ba800c63906fb29d34f40b9437092a665bffb14 Author: Chris Wilson Date: Thu May 17 22:46:18 2012 +0100 sna: Don't consider uploading inplace if the dst bo is unmappable Handle (and take advantage of) the fallback at the high level rather than masquerading an inplace write. Signed-off-by: Chris Wilson commit dad24721a13ce3a357e8ddae3c2dea61045f6fc2 Author: Chris Wilson Date: Thu May 17 21:44:22 2012 +0100 Revert "sna: Always try to operate inplace if we an LLC gpu bo" This reverts commit 10b4a9bb5f46ab9d9c8b165084ce4174b54a8d39 as it causes a regression for pixel data uploads to active buffers. commit 681c6e72412fff96b203a09be6ac8d393f3489a5 Author: Chris Wilson Date: Thu May 17 19:43:02 2012 +0100 sna: Track flushing status of live bo Currently we only move a bo with an outstanding kernel flush onto the flushing list if it is no longer in use. This leaves us potentially stalling on a flush if we try then to write to the object believing it to be retired and idle. Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit d99502a33d5bdbad010b7a036c1aee989fe29947 Author: Chris Wilson Date: Thu May 17 17:56:43 2012 +0100 sna/glyphs: Pass the extents to the backend for preparing to composite glyphs This information should not be required, but it might come in handy, so pass it along. Signed-off-by: Chris Wilson commit fb21c2df1a10d66f115a5978b0db508058fdc412 Author: Chris Wilson Date: Thu May 17 16:30:31 2012 +0100 sna/io: Assert that we do not replace a bo->flush object These should be pinned by the higher layers and so we should never be attempting to replace them. If we do replace a bo->flush, then we will end up miscounting outstanding flush bo. Signed-off-by: Chris Wilson commit 596c0a68709a93bb376647c1b566e9df4f23b35d Author: Chris Wilson Date: Tue May 15 18:53:39 2012 +0100 sna: Assign GCops after checking for fallback to pass sanity checks We assert that prior to installing the fallback GCops the current ops are the default set. This is broken if we point GCops to our GPU ops, but then fallback. So check for the fallback first. Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 3c9759ef2ad755bbe720d4aa031ec67dbc3b7734 Author: Chris Wilson Date: Mon May 14 14:12:06 2012 +0100 sna/gen3: Fix pre-multiplication of opacity masks The return type for the multipliation was only 8-bits wide dating back to the time when the function just computed a single channel and didn't try to blend the output back into the argb pixel value. Inlining the shift into the function means that we need the output to be 32-bits wide instead. Reported-by: Jiri Slaby Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=49887 Signed-off-by: Chris Wilson commit 053bd5bf2425aed44e1c2bb981d98ee5171211a0 Author: Chris Wilson Date: Mon May 14 12:08:09 2012 +0100 sna: Use the correct storage for box_from_seg Fixes regression from 3aa98289e3a2 with clipped segments. As we clipped the empty box rather than the segment, we never drew any outlines. Signed-off-by: Chris Wilson commit b654b8794db7b00666ce5c59535a9302932c483b Author: Chris Wilson Date: Mon May 14 11:36:15 2012 +0100 sna/trapezoids: Fix picture creation for fallback trapezoids Being a little lax in not updating the format after upconversion to PICT_a8, meant we were trying to composite with a depth 1, 8 bpp a8 image and thoroughly confusing everybody when creating the upload trapezoid mask. Signed-off-by: Chris Wilson commit 80567f61afe77a003e663b17c1fc6b6c3ed04042 Author: Chris Wilson Date: Mon May 14 10:40:38 2012 +0100 sna/trapezoids: Do not reduce SRC to a clear pixmap to unbounded As we instruct the migration code to drop the clear when copying from the GPU to the CPU, we then need to emit the zeros during the span writing. Fixes some occassional corruption behind complex clip masks. Signed-off-by: Chris Wilson commit 067e9375809ea6cfa0c0b5e2159b359535c3b362 Author: Chris Wilson Date: Mon May 14 10:00:00 2012 +0100 sna: Trim unused partial buffer uploads Signed-off-by: Chris Wilson commit effb0b9ec579624e38f0007cd7096e7da04048ca Author: Chris Wilson Date: Sun May 13 19:29:44 2012 +0100 sna: Tweak usage of kgem_bo_can_map vs kgem_bo_mapped More often than not we only want to consider whether we can map the bo and decide whether doing so is the better option. Whether the bo is already mapped is not such an issue any more with the throttling. Signed-off-by: Chris Wilson commit 37aa1e1291127ff4d27407fac936af64d0e353a8 Author: Chris Wilson Date: Sun May 13 16:59:30 2012 +0100 sna: Simplify partial buffer tracking As we only want to track partial buffers with asynchronous reuse, exclude all overs from the active buffers list. Secondly, keep the list in most-recently-used order rather than by size. Signed-off-by: Chris Wilson commit 7048a71fb6ff5e7c37e21a794f7b42d9a1da3473 Author: Chris Wilson Date: Sun May 13 15:42:28 2012 +0100 sna: Flush batch if GPU is idle upon wakeup Signed-off-by: Chris Wilson commit 12e340c4b8efa70f3305f489af30461b7c0e5edd Author: Chris Wilson Date: Sun May 13 11:01:35 2012 +0100 sna: Optimise kgem_clear_dirty() by keeping dirty buffers at the front Signed-off-by: Chris Wilson commit b0dfd8223adee87e28b391a019effa7bb3630533 Author: Chris Wilson Date: Sun May 13 08:33:23 2012 +0100 sna: Flush the batch before blocking if the GPU is idle If we have work pending for the GPU and we believe it is idle, just submit the batch in order to improve GPU/CPU concurrency and reduce latency. Signed-off-by: Chris Wilson commit 6924fc525d6bc82901cfed769c176b44c0bce024 Author: Chris Wilson Date: Fri May 11 14:04:09 2012 +0100 sna: Fix off-by-one in computation of width for inplace trapezoids This lead to the dropping of the last pixel for small area trapezoids, such as the right hand outline of buttons under ClearLooks. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=48320 Signed-off-by: Chris Wilson commit a58a543e84ed15c41f4fa7644be3ba7865d31b92 Author: Chris Wilson Date: Thu May 10 12:59:55 2012 +0100 sna: handle vmap creation failures gracefully Signed-off-by: Chris Wilson commit 9af651fdf93f6ccdbcb147cb16989f84c2d0ebdc Author: Chris Wilson Date: Thu May 10 10:57:56 2012 +0100 sna: Include some DBG for segment drawing Signed-off-by: Chris Wilson commit 3aa98289e3a2f59633ac5f1a5f64db4ee55609d9 Author: Chris Wilson Date: Thu May 10 10:31:03 2012 +0100 sna: Correct adjustment of LineCap for rectilinear segments Signed-off-by: Chris Wilson commit ad69316ab1054c5e4b56c19b8eec87b9f86939e5 Author: Chris Wilson Date: Thu May 10 10:03:49 2012 +0100 sna: Include a small amount of timer-slack to avoid very short sleeps Signed-off-by: Chris Wilson commit 29d1d409b259ba0c85947ef5eef903ea527aa118 Author: Chris Wilson Date: Thu May 10 09:54:53 2012 +0100 sna: Avoid doubling the final step of poly lines Signed-off-by: Chris Wilson commit a83d90ee61be44a2a36b56ad24bbc6544320448f Author: Chris Wilson Date: Wed May 9 20:34:52 2012 +0100 sna: Avoid randomly changing domains of active bo After attaching the bo to the scanout, mark it as retired in order to update its domains so that the assertion during retirement later is correct. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=49526 Signed-off-by: Chris Wilson commit d5200510b8f2a0ccb2e0c45dfebc659f8360780c Author: Chris Wilson Date: Wed May 9 17:58:58 2012 +0100 sna: Respect the error terms when reversing the PolyLine DDA Signed-off-by: Chris Wilson commit a3d37fb29f8dffb0e370ad95783994aaa7eccfaf Author: Chris Wilson Date: Tue May 8 15:47:14 2012 +0100 sna: Force remapping for IO transfer Should fix regression from fcccc5528 (sna: Improve handling of inplace IO for large transfers) whereby it was aborting the transfer it we need to remap the buffer for the upload. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=49546 Signed-off-by: Chris Wilson commit 2a9a93e4484e0c616724610f4c8019fcbaa7ad53 Author: Chris Wilson Date: Tue May 8 15:12:51 2012 +0100 sna: Only avoid ring switching for indirect uploads Signed-off-by: Chris Wilson commit 613902b60e0f2ca2a916e68306a1a37bc236d00d Author: Chris Wilson Date: Tue May 8 14:17:46 2012 +0100 sna: Fix off-by-one in checking available execbuffer slots Signed-off-by: Chris Wilson commit 663e387b35c314c4c2bee8137d6b70d27fa9f729 Author: Chris Wilson Date: Tue May 8 11:45:34 2012 +0100 sna: Only submit a batch to the scanout if it is not already busy Signed-off-by: Chris Wilson commit cd7a56b7313233190a4c4a735d4a141e99c9b688 Author: Chris Wilson Date: Tue May 8 08:53:41 2012 +0100 sna: Throttle independently of scanout updates As we are now throttling to relieve GTT pressure, it is a benefit to consistently throttle before blocking. Signed-off-by: Chris Wilson commit 2372176f73b7c945f56b7673eab5eccb86366416 Author: Chris Wilson Date: Mon May 7 12:57:35 2012 +0100 sna: Throttle execution when searching for inactive buffers If we have some active buffers that we may reuse and old outstanding requests, throttling before retiring should prevent the CPU from running away from the GPU and hogging the entire GTT and RAM. Signed-off-by: Chris Wilson commit 4df228749729dd540b639368400fa20118cdf412 Author: Chris Wilson Date: Mon May 7 10:42:30 2012 +0100 sna: Rate-limit and shrink bo usage if we hit system resource limits Signed-off-by: Chris Wilson commit ca4d2296e6e42e837627756790b262cae0fd3b6c Author: Chris Wilson Date: Mon May 7 10:23:19 2012 +0100 sna: Expand batch buffers As batch buffers are compacted to fit into the smallest bo, the only cost is the larger static array allocation (and presumably cache misses). Signed-off-by: Chris Wilson commit 9281b80644ce76ad9e0f3f8f812cbae97c10814a Author: Chris Wilson Date: Mon May 7 23:25:22 2012 +0100 sna/gen[345]: Clear used vertices when discarding unmappable vbo Signed-off-by: Chris Wilson commit a6ee376e93517659391905e6c9018b3bb735135d Author: Chris Wilson Date: Mon May 7 10:24:21 2012 +0100 sna: Use the correct invocation of kgem_bo_destroy() for sanity-checks Signed-off-by: Chris Wilson commit c89c7e9a04314e40cee5514a182a8364c4f99374 Author: Chris Wilson Date: Mon May 7 11:54:50 2012 +0100 sna: Update select timeout when installing a timer in the block handler The block handler is run after the timers are queried for their expiration and so if we install a timer in the block hander, we must set the timeout ourselves. Signed-off-by: Chris Wilson commit 4ba366ead3a133136554579fe9a604da1fc1da68 Author: Chris Wilson Date: Mon May 7 08:55:35 2012 +0100 sna: Manually execute the timer as TimerForce does not run an inactive timer Signed-off-by: Chris Wilson commit 2c801c45dbad58d18f2673e17723825dda0c4e55 Author: Chris Wilson Date: Sun May 6 17:23:11 2012 +0100 sna: Replace timerfd with OsTimer As timerfd is linux-specific, and OsTimer an OS-agnostic abraction, replace the former with the later. Arguably this has slightly better performance characteristics in select-bound loops. Signed-off-by: Chris Wilson commit 74d2707de4a0f94d2143f1e9c60762e4167b1ea6 Author: Chris Wilson Date: Sun May 6 21:51:27 2012 +0100 sna: Remove short-circuiting for large font fallbacks Unlike the fallback for an unhandled depth, we need to ensure that the pixmaps are mapped to the CPU before calling the fb routines. Reported-by: Toralf Förster Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=49558 Signed-off-by: Chris Wilson commit c5b6741d3729c6867702ab64a6c59cb8052c0ef3 Author: Chris Wilson Date: Sun May 6 12:40:54 2012 +0100 sna/gen2+: Fix typo for computing redirected extents for render copy Signed-off-by: Chris Wilson commit 771090f25db702d25ebbd3f2b44429cf0acfe8fd Author: Chris Wilson Date: Fri May 4 20:56:37 2012 +0100 sna: Add a pair of asserts to track down a NULL pointer dereference Looks like the assumption for the location of the data is invalid, allocation failure, perhaps? References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit d376a960df7a081a5d449f77b81ae13223b98929 Author: Chris Wilson Date: Fri May 4 10:09:46 2012 +0100 sna/dri: Only track a single pending flip across all pipes Signed-off-by: Chris Wilson commit 450592f989efd0d3bc9ef2de245fce0a180e91a2 Author: Chris Wilson Date: Fri May 4 09:50:19 2012 +0100 sna: Cache the framebuffer id Also fixup a weakness of only tracking scanout with a single bit, as we used to clear it forcibly after every flip. Signed-off-by: Chris Wilson commit 29d035279b2fe98d5ba9cf01125faea34d36fb76 Author: Chris Wilson Date: Fri May 4 09:11:31 2012 +0100 sna/dri: pageflip unref debugging Signed-off-by: Chris Wilson commit 079b491ced2c9c3c73d938ef6025d040016ad3a7 Author: Chris Wilson Date: Thu May 3 23:52:17 2012 +0100 sna: Ensure drawables are clipped against the pixmap before migration Signed-off-by: Chris Wilson commit 1a5f8599b150064339d62a97c58026e62b49ff27 Author: Chris Wilson Date: Thu May 3 23:21:48 2012 +0100 sna: Compile fix for fresh assertion Signed-off-by: Chris Wilson commit 61cac5c265279d45677262216a0ba56f548cd898 Author: Chris Wilson Date: Thu May 3 22:33:59 2012 +0100 sna: Maintain a reference to the chain of proxies Rather than attempt to flatten the chain to the last link, we may need to hold a reference to the intermediate links in case of batch buffer submission. Fixes http://tnsp.org/~ccr/intel-gfx/test.html Reported-by: Matti Hamalainen Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=49436 Signed-off-by: Chris Wilson commit dea5d429f7a52dfc945b17a57ef79744cc796b0e Author: Chris Wilson Date: Thu May 3 17:35:24 2012 +0100 sna: Remove extraneous SCANOUT flags These are now fixed by obeying a minimum alignment restriction. Signed-off-by: Chris Wilson commit 19fd24a4db994bb5c5ce4a73f06d9394a758ea91 Author: Chris Wilson Date: Thu May 3 17:35:10 2012 +0100 sna: Fix offset for combining damage Signed-off-by: Chris Wilson commit 1376c81dbf3b789e04e6804df1b1fd32bcb2bd1d Author: Chris Wilson Date: Thu May 3 17:34:28 2012 +0100 sna: Debug option to force particular upload/download paths Signed-off-by: Chris Wilson commit f0d464d6b18855616fc43d9a25aa6853f86c8e2b Author: Chris Wilson Date: Thu May 3 16:17:35 2012 +0100 sna/dri: Balance flush counting Signed-off-by: Chris Wilson commit d47e98dd64c0b9fe2979db42622c5ee8168e8b35 Author: Chris Wilson Date: Thu May 3 14:54:17 2012 +0100 sna: Minor glyph fallback fixes Signed-off-by: Chris Wilson commit a1f08b8850616952fb0babe2275eb36b13a380ec Author: Chris Wilson Date: Thu May 3 14:53:29 2012 +0100 sna: Don't discard GPU buffer if we only want to read back for the operation Signed-off-by: Chris Wilson commit fcccc5528b8696fb4f9b3f9f528673b95d98a907 Author: Chris Wilson Date: Thu May 3 11:27:44 2012 +0100 sna: Improve handling of inplace IO for large transfers If the transfer is large enough to obliterate the caches, then it is preferrable to do it inplace rather than upload a proxy texture and queue a blit. This helps prevent an inconsistency where one layer believes the operation should be done inplace only for the IO layer to perform an indirect upload. Testing show no significant impact upon the cairo-traces, but it does prevent x11perf -shmput from exploding. Signed-off-by: Chris Wilson commit 53568e8e49559094ce5b24b8709669f1f76fe2bf Author: Chris Wilson Date: Mon Apr 30 09:24:06 2012 +0100 sna/gen7: Fix debug printing of primitives Signed-off-by: Chris Wilson commit 01c26a44fdce761781908be11102e7a6a3db523c Author: Chris Wilson Date: Wed May 2 21:09:27 2012 +0100 sna: Avoid reducing damage for synchronisation Signed-off-by: Chris Wilson commit f4c34e9ab32f31669896b8f626195827a85af337 Author: Chris Wilson Date: Wed May 2 19:09:10 2012 +0100 sna: Bring back the minimum alignment for G33 The underlying cause is still not fixed. It should be possible to use the much laxer alignment for single-stream linear. Still no idea how I fail to convince the GPU to drop the depth buffer. Reported-by: Matti Hamalainen References: https://bugs.freedesktop.org/show_bug.cgi?id=49391 Signed-off-by: Chris Wilson commit 10b4a9bb5f46ab9d9c8b165084ce4174b54a8d39 Author: Chris Wilson Date: Wed May 2 12:45:09 2012 +0100 sna: Always try to operate inplace if we an LLC gpu bo commit dd80fb00bf7acf37dc3b9125431a12b67d7e92d3 Author: Chris Wilson Date: Wed May 2 11:31:42 2012 +0100 sna: Fallback for glyphs too large for XY_TEXT_IMMEDIATE Signed-off-by: Chris Wilson commit 7e09babb3e9e23882db30ee7d0c22c503962faa9 Author: Chris Wilson Date: Tue May 1 12:37:45 2012 +0100 sna: Only attempt to reuse exported scanout buffers Yet more mesa w/a. Signed-off-by: Chris Wilson commit b4b32e7a0172a74372f800e9c74d639a23c5ff34 Author: Chris Wilson Date: Tue May 1 11:15:00 2012 +0100 sna: Fast-path unclipped glyphs Avoid the redundant computation of the glyph intersection with the drawable bounding box. Signed-off-by: Chris Wilson commit ffdf9aca12adcfa1ec7ab7a1706873105a5f0d4a Author: Chris Wilson Date: Sun Apr 29 21:59:52 2012 +0100 legacy/i810: hwmc additionally depends upon building DRI Signed-off-by: Chris Wilson commit 444da84c47266bcbbdf5121507901de8eb36f11b Author: Chris Wilson Date: Sun Apr 29 21:49:04 2012 +0100 configure: Version bump for 2.19.0 release commit a206a1eee1fb799d567d68db564d663cfc66f6cc Author: Chris Wilson Date: Sat Apr 28 01:54:43 2012 +0100 sna: Tune relocation array size Signed-off-by: Chris Wilson commit 4f1908c651ef9e2af33d8831466a605234978c46 Author: Chris Wilson Date: Fri Apr 27 23:19:56 2012 +0100 sna: PolyPoint only uses the gc->fgPixel Signed-off-by: Chris Wilson commit 8453034c7dd893f1d4c32ee87724f3a13137595d Author: Chris Wilson Date: Fri Apr 27 22:08:51 2012 +0100 sna/gen6: Allow ring switching at the start of a batch Signed-off-by: Chris Wilson commit 93ad7793894787600c5074917c753fa7c6816134 Author: Chris Wilson Date: Fri Apr 27 22:03:55 2012 +0100 sna: Tweak semaphores-enabled heuristic The kernel module now defaults to -1, confusing the test. Signed-off-by: Chris Wilson commit 986dbdda3bf8dcf208e55543d8a3393c4b53f10b Author: Chris Wilson Date: Fri Apr 27 19:21:44 2012 +0100 sna: Tweak placement choice for high-overhead operations Some operations cost more to setup than to transfer data back and forth! Signed-off-by: Chris Wilson commit daac9a1d036d80ccce83438b49115a236a16bfb6 Author: Chris Wilson Date: Fri Apr 27 16:50:28 2012 +0100 sna: Micro-optimise common case of checking a single fenced bo Signed-off-by: Chris Wilson commit 5967d76ca09a257ec9db66ea664158e1dfd083ba Author: Chris Wilson Date: Wed Apr 25 17:15:37 2012 +0100 sna: Fixup broken assertion It is valid for the cpu_bo to be NULL, as we may be choosing to free the large shadow pixel buffer instead. Signed-off-by: Chris Wilson commit 860d3859b586939cd52e45b944cb6abd2a2ca71b Author: Chris Wilson Date: Wed Apr 25 16:04:33 2012 +0100 sna/gen7: Add CS stall before changing WM binding table Signed-off-by: Chris Wilson commit c219283460c0f2dfdb823e0cb139d05075c6afce Author: Chris Wilson Date: Wed Apr 25 15:04:01 2012 +0100 sna/gen7: Apply more recent improvements from SNB perf tuning Signed-off-by: Chris Wilson commit a3371613c9bf577a69cdf811ca1bebaea46bbe95 Author: Chris Wilson Date: Wed Apr 25 11:09:35 2012 +0100 sna: Do not automagically convert GTT mappings on untiled scanout to CPU The likelihood of an untiled mapping of the scanout is slim, except for gen3 with large desktops, and there it should never be in the CPU domain... The issue is that we may perform an operation "inplace", yet incoherent with the display engine, and never flush the CPU cache, resulting in render corruption. In theory at least! Signed-off-by: Chris Wilson commit 1abd92cd012ee46d44ed4873a5e750d56ae6668f Author: Chris Wilson Date: Wed Apr 25 11:16:30 2012 +0100 sna: Clear the domain tracking after attaching the bo to scanout This is basically to make sure we don't continue treating it as CPU coherent. Signed-off-by: Chris Wilson commit 8c58c840b1ba579a5601804fc710c58e1e00213f Author: Chris Wilson Date: Mon Apr 23 11:09:37 2012 +0100 sna/dri: Always clear the scanout when destroying dri2 buffers As we may end up holding onto and releasing the Screen pixmap last, we may also be responsible for flushing the last reference to the scanout. Signed-off-by: Chris Wilson commit caf9144271a10f90ea580c246b2df3f69a10b7a0 Author: Chris Wilson Date: Fri Apr 20 17:15:37 2012 +0100 sna: Remove the assertions that the cached upload buffers are active These were added to track down some corruption, but the assertions themselves are incorrect, just very rare. The upload buffer may genuinely be cached if we abort the render operation after uploading the source data, leaving the proxy not coupled to any request. References: https://bugs.freedesktop.org/show_bug.cgi?id=48400 Signed-off-by: Chris Wilson commit aff3614efd5c12e658fa5723934e5bd50a83a316 Author: Chris Wilson Date: Fri Apr 20 13:21:40 2012 +0100 sna: Always clear the mmapped domains when reusing partial upload buffers As we need to make sure that we do invalidate the caches appropriately on reuse. Mildly paranoid, but strictly required by the spec. Signed-off-by: Chris Wilson commit cb6a3dc2edf3cd612f833bc9a4656166735ee856 Author: Chris Wilson Date: Thu Apr 19 10:34:23 2012 +0100 sna: Discard proxy upload buffer if we choose to render to it Even if we try to avoid treating an upload buffer as a real GPU target, we may still choose to migrate the buffer to the GPU in order to keep other buffers on the GPU. In that case, we do want to create a real GPU bo. Signed-off-by: Chris Wilson commit 4cf74d409ca63c6a479c1ee2187908c04f3b830b Author: Chris Wilson Date: Thu Apr 19 09:09:32 2012 +0100 sna: Don't consider upload proxies as being on the GPU for render targets The upload proxy is a fake buffer that we do not want to render to as then the damage tracking become extremely confused and the buffer it self is not optimised for persistent rendering. We assert that we do not use it as a render target, and this patch adds the check so that we avoid treating the proxy as a valid target when choosing the render path. Signed-off-by: Chris Wilson commit fd81408b978c9b57c046ee43d2d32e1370e83a7d Author: Chris Wilson Date: Wed Apr 18 11:39:43 2012 +0100 sna: Increase the render target alignment to 4 pixels on gen4+ as well Repoerted-and-tested-by: Toralf Förster commit 11599e52b842b5db76798879b0fbb57762fe6002 Author: Chris Wilson Date: Tue Apr 17 20:08:14 2012 +0100 sna/dri: Decouple the frame event info after attaching along error paths Signed-off-by: Chris Wilson commit b817200371bfe16f44b879a793cf4a75ad17bc5c Author: Chris Wilson Date: Tue Apr 17 17:54:58 2012 +0100 Don't issue a scanline wait while VT switched Be paranoid and check that we own the VT before emitting a scanline wait. If we attempt to wait on a fb/pipe that we do not own, we may issue an illegal command and cause a lockup. Signed-off-by: Chris Wilson commit af4a6e8cb52ace594934446e6d8a7aaa1945a9b0 Author: Chris Wilson Date: Mon Apr 16 20:57:03 2012 +0100 sna: Don't assert exported buffers are not busy As we do not fully control these buffers, we cannot truly say when they are idle, we can only trust that the split between us and the compositor doesn't lead to much corruption. Signed-off-by: Chris Wilson commit a16616209bb2dcb7aaa859b38e154f0a10faa82b Author: Chris Wilson Date: Sat Apr 14 19:03:25 2012 +0100 uxa: Fix leak of glyph mask for unhandled glyph composition ==1401== 7,344 bytes in 34 blocks are possibly lost in loss record 570 of 587 ==1401== at 0x4027034: calloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==1401== by 0x8BE5150: drm_intel_gem_bo_alloc_internal (intel_bufmgr_gem.c:689) ==1401== by 0x899FC04: intel_uxa_create_pixmap (intel_uxa.c:1077) ==1401== by 0x89C2C41: uxa_glyphs (uxa-glyphs.c:254) ==1401== by 0x21F05E: damageGlyphs (damage.c:647) ==1401== by 0x218E06: ProcRenderCompositeGlyphs (render.c:1434) ==1401== by 0x15AA40: Dispatch (dispatch.c:439) ==1401== by 0x1499E9: main (main.c:287) Signed-off-by: Chris Wilson commit ae145c21e9fd3a12164f8b4720d059f9c158249e Author: Chris Wilson Date: Sat Apr 14 18:42:23 2012 +0100 sna: Avoid leaking the plane resources when determining sprite planes Fixes the tiny, one-off leak: ==1407== 8 bytes in 1 blocks are definitely lost in loss record 48 of 527 ==1407== at 0x402894D: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==1407== by 0x8580BE8: drmMalloc (xf86drm.c:147) ==1407== by 0x8583D54: drmAllocCpy (xf86drmMode.c:73) ==1407== by 0x8585265: drmModeGetPlaneResources (xf86drmMode.c:955) ==1407== by 0x8A1BCE9: sna_video_sprite_setup (sna_video_sprite.c:367) ==1407== by 0x8A1A0A3: sna_video_init (sna_video.c:523) ==1407== by 0x89FD4E0: sna_screen_init (sna_driver.c:935) ==1407== by 0x15AD80: AddScreen (dispatch.c:3909) ==1407== by 0x19A2DB: InitOutput (xf86Init.c:817) ==1407== by 0x14981C: main (main.c:204) Signed-off-by: Chris Wilson commit 69a7737abeded6ee923643bd8a80a5a84e6a979c Author: Chris Wilson Date: Sat Apr 14 12:06:51 2012 +0100 sna: Align texture subsurfaces to 2x2 texture samples Signed-off-by: Chris Wilson commit 1d2a46e0902d82b43a5e12af36521a6a7fd6ba39 Author: Chris Wilson Date: Sat Apr 14 12:04:23 2012 +0100 sna: Align redirect subsurfaces to 2x2 or 4x4 render spans Signed-off-by: Chris Wilson commit 1ce2b65d622797000e0a4db7dc851d5b1da04f85 Author: Chris Wilson Date: Sat Apr 14 11:59:31 2012 +0100 sna: Align render target sizes on gen2/3 to 4x4 render spans Signed-off-by: Chris Wilson commit 89f2b09b1e5be9842747998ea4fe32a6f1ede4cc Author: Chris Wilson Date: Fri Apr 13 16:37:43 2012 +0100 sna: Avoid using TILING_Y for large objects on gen2/3 References: https://bugs.freedesktop.org/show_bug.cgi?id=48636 Signed-off-by: Chris Wilson commit eaadbce122059066353743f1653aa16e9d9b747f Author: Chris Wilson Date: Fri Apr 13 15:12:36 2012 +0100 sna: Relax bogus assertion The bo may be considered unmappable due to being bound to outside the mappable region, which we are attempting to rectify through mapping into the GTT domain. Signed-off-by: Chris Wilson commit b478420740d05fa87ddbd92042b1f7f2d002f73e Author: Chris Wilson Date: Fri Apr 13 15:07:13 2012 +0100 sna: Limit the buffer reuse for mappable uploads to only those with mmaps Signed-off-by: Chris Wilson commit 90e2740e7e459c56205fa65bab1ae3dbfd5d3945 Author: Chris Wilson Date: Fri Apr 13 13:51:57 2012 +0100 sna: Remove the conflicting assertion during GTT map Reported-by: Clemens Eisserer References: https://bugs.freedesktop.org/show_bug.cgi?id=48636 Signed-off-by: Chris Wilson commit 9e6d55a8d63f10ca6a2b10e44d00c84b07724485 Author: Chris Wilson Date: Fri Apr 13 09:56:39 2012 +0100 sna: Don't use miSpan code for wide-spans by default, too expensive Only use the fall-forward miSpans code when it prevents a readback. Signed-off-by: Chris Wilson commit 9becfbbf89f2b170e50f705cabfc7bbf1dcf9846 Author: Chris Wilson Date: Fri Apr 13 09:38:57 2012 +0100 sna: Draw dashed PolyLines twice, once for the fgPixel, once for the bgPixel As the fast paths only setup state upfront, we were missing the state changes required between dash-on/off. Take advantage of that each pixel is only drawn once to batch the state changes and run the miZeroDashLines twice. A future task would be to use a custom line drawing routine... Fixes regression from ec1267df746512c2e262ef0bd9e9527bc5efe6f4. Signed-off-by: Chris Wilson commit e269ed5d4e1b9d758aeb9a85ed0fa631f0aff0b1 Author: Chris Wilson Date: Thu Apr 12 22:46:22 2012 +0100 sna: Restore CPU domain for vmapped buffers when reusing For a vmapped upload buffer, we need to notify the kernel (and thereby the GPU) to invalidate the sampler and flush its caches when we reuse an idle buffer. Signed-off-by: Chris Wilson commit d29b8650c40c673e6ddddaf52db9247e9836cba8 Author: Chris Wilson Date: Thu Apr 12 22:23:12 2012 +0100 sna: Revert use of mmap64() As this just causes mayhem on a 64-bit platform. Doomed if you, doomed if you don't. :( Signed-off-by: Chris Wilson commit a78b1d71a39ae29d5f85bd82c09202ebec3e6539 Author: Chris Wilson Date: Thu Apr 12 15:37:25 2012 +0100 sna: Declare AC_SYS_LARGEFILE for mmap64 In order to use the full 32-bits of mmap address space on small platforms we need to use mmap64(). Signed-off-by: Chris Wilson commit 09deba927daa96be6230b1c3e1b425622512d8a2 Author: Chris Wilson Date: Tue Apr 10 20:59:36 2012 +0100 sna: Check ioctl return from set-domain Let's not assume it succeeds and so avoid altering our bookkeeping along failure paths. Signed-off-by: Chris Wilson commit 0b12f1d8e4d0a4fafac9553f144535efc4ebe0be Author: Chris Wilson Date: Tue Apr 10 15:23:54 2012 +0100 sna: Fix shadowed variable sna_accel.c: In function 'sna_pixmap_move_area_to_gpu': sna_accel.c:1751:12: warning: declaration of 'flags' shadows a parameter [-Wshadow] sna_accel.c:1731:72: warning: shadowed declaration is here [-Wshadow] Signed-off-by: Chris Wilson commit 755a7107aed268d87c5cc0feb1ba388b0cb7fc59 Author: Chris Wilson Date: Tue Apr 10 15:19:19 2012 +0100 sna: Fix typo and use the right pointer for kgem_bo_destroy Useless warnings in xorg headers ftl. References: https://bugs.freedesktop.org/show_bug.cgi?id=48400 Signed-off-by: Chris Wilson commit c8502e350cb18f6f5d821d237ffcee453f347eba Author: Chris Wilson Date: Tue Apr 10 14:43:28 2012 +0100 sna: Add missing alloc failure check for creating tile source Signed-off-by: Chris Wilson commit 9cc6f7ccc55cc11f47b3b7d626c9f5a7c1327d57 Author: Chris Wilson Date: Tue Apr 10 12:06:56 2012 +0100 sna: Release the freed bo cache upon expire Signed-off-by: Chris Wilson commit 102d11906a672140bac099e7bd1b35345d13a2fc Author: Chris Wilson Date: Mon Apr 9 22:54:51 2012 +0100 sna: Check for an inactive partial buffer to reuse after retiring requests Signed-off-by: Chris Wilson commit e2fb2421d8f5c07925d1699673aa4b1dd6c6b22c Author: Chris Wilson Date: Mon Apr 9 22:54:15 2012 +0100 sna: Release partial buffers during cache expiration Signed-off-by: Chris Wilson commit 333fdcad8677675a4758223c1a980c90d970ee42 Author: Chris Wilson Date: Mon Apr 9 22:53:31 2012 +0100 sna: Repeat expire whilst there remaining outstanding requests Do not allow the cache expiration to finish if we are still running requests. Signed-off-by: Chris Wilson commit 778232e3d2fb5340a3092014801dc00a56c56d42 Author: Chris Wilson Date: Tue Apr 10 11:14:50 2012 +0100 sna: Only move the bo into the read domain for readback And mark it as currently in no domain afterwards, so that if we reuse it, it will be appropriately moved later. Signed-off-by: Chris Wilson commit dd093eafb9b94b8e4cd8853d74078c3aa7e72f57 Author: Chris Wilson Date: Mon Apr 9 14:09:42 2012 +0100 sna: Add assertions around proxy list handling References: https://bugs.freedesktop.org/show_bug.cgi?id=48400 Signed-off-by: Chris Wilson commit 7f0bede3e7e3f92a637d1c886304b16afc0e34f2 Author: Chris Wilson Date: Mon Apr 9 10:48:08 2012 +0100 sna/traps: Use a temporary variable for the write pointer To avoid accumulating the write offset for wide spans, we need to reset the destination pointer between spans. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=48332 Signed-off-by: Chris Wilson commit 2e4da00e3e03b873f5cad0cc5b1f6cc791852ca5 Author: Chris Wilson Date: Mon Apr 9 10:42:18 2012 +0100 sna/traps: Assert that the inplace row is contained before writing References: https://bugs.freedesktop.org/show_bug.cgi?id=48332 Signed-off-by: Chris Wilson commit 0464e93a088a9e8bc29ad8b36b6e12c3dda32ec6 Author: Chris Wilson Date: Sun Apr 8 17:16:03 2012 +0100 sna: Add some assertions for misuse of proxies References: https://bugs.freedesktop.org/show_bug.cgi?id=48400 Signed-off-by: Chris Wilson commit 479cb6ba71038fe44f66fb31fad90d0d454fea7a Author: Chris Wilson Date: Sun Apr 8 14:25:49 2012 +0100 sna: Compress adjoining spans during FillSpans Signed-off-by: Chris Wilson commit c5c01c13badeb7c2ead0c848b746d8d474277a77 Author: Chris Wilson Date: Sun Apr 8 13:51:13 2012 +0100 sna: Remove the duplicated check for use-bo? in PolySegments Signed-off-by: Chris Wilson commit f5deea4f60433ee2b0c2d02fba682fff8b7829e8 Author: Chris Wilson Date: Sun Apr 8 13:39:47 2012 +0100 sna: Correct partial-write flag for PolySegments fallback Signed-off-by: Chris Wilson commit c3d7f4c1cf7d052163b7c4e74bb202c618f0eb76 Author: Chris Wilson Date: Sun Apr 8 13:18:56 2012 +0100 sna/gen3: Reset accumulated constants for each composite In particular the glyph routines require the composite setup to reinitialise state between glyph runs. This affects anything trying to use glyphs without a mask with a gradient source, for example. Signed-off-by: Chris Wilson commit 701473d20485a0557b4fb36efcbfbb8656e2f619 Author: Chris Wilson Date: Sun Apr 8 10:09:42 2012 +0100 sna: Release cached upload buffers when reusing a write buffer for readback References: https://bugs.freedesktop.org/show_bug.cgi?id=48400 Signed-off-by: Chris Wilson commit 1ecf17b2507f95e1fefea15833fa9f57ec256a2e Author: Chris Wilson Date: Sat Apr 7 10:01:01 2012 +0100 sna/gradient: Compute the absolute delta between color stops Otherwise we do not detect gradients that start from white! Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=48407 Signed-off-by: Chris Wilson commit 4356fae72db3a33935b575edf95c84fbb48072a7 Author: Chris Wilson Date: Fri Apr 6 21:10:50 2012 +0100 sna/video: Only wait upon the scanout pixmap Caught by the addition of the assertion. Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit b790ba2ec9ead51227d85fc8630bc7505eb7d7b3 Author: Chris Wilson Date: Fri Apr 6 15:38:02 2012 +0100 sna: Correct the damage offset for redirected rendering Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=48385 Signed-off-by: Chris Wilson commit 0b81bafb802bb86454739ed46cf45571bccef735 Author: Chris Wilson Date: Fri Apr 6 15:14:45 2012 +0100 sna/glyphs: Prefer a temporary upload mask for large glyph masks If the required temporary mask is larger than the 3D pipeline can handle, just render to a CPU buffer rather than redirect every glyph composition. Signed-off-by: Chris Wilson commit 42a84613e34522af885b4b50d6c68ef77e81ffc3 Author: Chris Wilson Date: Fri Apr 6 14:27:15 2012 +0100 sna: Relase the upload cache when overwriting with PutImage Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=48359 Signed-off-by: Chris Wilson commit 999aa210ff87919945c673bdd34bae76ac097681 Author: Chris Wilson Date: Fri Apr 6 09:24:36 2012 +0100 sna: Use a sentinel value to prevent accessing beyond the end of the y_buckets Signed-off-by: Chris Wilson commit 1ae6328c57eb496072f0d0e27440f5d0901633b0 Author: Chris Wilson Date: Fri Apr 6 09:12:08 2012 +0100 sna: Remove redundant check from tor_inplace() We only execute full-steps for vertical edges so we do not need the second check. Signed-off-by: Chris Wilson commit 51b9202d27db3d98c6d82ba224bd8eb218533dd9 Author: Chris Wilson Date: Wed Apr 4 11:13:27 2012 +0100 sna: Only engage the GPU detiler for multiple rows Signed-off-by: Chris Wilson commit 98ad4c3cd8647ba3ec90fb45157773c8e85e886c Author: Chris Wilson Date: Tue Apr 3 19:47:15 2012 +0100 sna/gen3: Don't force use of the render pipeline just for vmap Signed-off-by: Chris Wilson commit 0915d414f55a1bff4171981feb87bae212f29f23 Author: Chris Wilson Date: Tue Apr 3 19:04:29 2012 +0100 sna/gen3: Fix pre-multiplication of mask value Signed-off-by: Chris Wilson commit 04851e4210d2d71542359c14d4b68d0851b36326 Author: Chris Wilson Date: Tue Apr 3 12:34:24 2012 +0100 sna/gen3: Convert the clear-color from picture->format to a8r8g8b8 The shaders treat colours as an argb value, however the clear color is stored in the pixmap's native format (a8, r5g6b5, x8r8g8b8 etc). So before using the value of the clear color as a solid we need to convert it into the a8r8g8b8 format. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=48204 Reported-by: Paul Neumann Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47308 Signed-off-by: Chris Wilson commit 87a672dafd9d6f47f31b77b406b7f0fb2b4030ac Author: Chris Wilson Date: Tue Apr 3 09:24:06 2012 +0100 sna: Apply CoordMode when computing point extents Reported-by: Patrick Truebe Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=48220 Signed-off-by: Chris Wilson commit 0a0ee491ea18dc59748ff4419ae73bd1a369ae79 Author: Chris Wilson Date: Tue Apr 3 09:23:49 2012 +0100 sna: Debugging flil spans and their clipping Signed-off-by: Chris Wilson commit ec1267df746512c2e262ef0bd9e9527bc5efe6f4 Author: Chris Wilson Date: Mon Apr 2 16:16:24 2012 +0100 sna: Use the solid spans fast paths for dashed zero-width lines as well Signed-off-by: Chris Wilson commit 2d1f3cb198f2fe9602356a334a076abf3c68a9c7 Author: Chris Wilson Date: Mon Apr 2 13:51:37 2012 +0100 sna/gen4: Remove the accidental debugging hack from the last commit Signed-off-by: Chris Wilson commit 5c4dc9c5db7b2a5b936bdbc15536c5cf0f7f5f23 Author: Chris Wilson Date: Mon Apr 2 13:40:22 2012 +0100 sna/gen3+: Fix sampling of borders around gradients Incurs a slight loss of precision for the internal gradient, but much more preferable to the artefacts around the borders with RepeatNone. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=45016 Signed-off-by: Chris Wilson commit 0b2651dc04cef8f9692b2557684f044b4980700f Author: Chris Wilson Date: Sun Apr 1 19:43:15 2012 +0100 sna: Apply composite offset to damage for spans fast paths Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 4ea9ab9303d21a62683055b75eaed66c97a5f289 Author: Chris Wilson Date: Sun Apr 1 16:36:16 2012 +0100 sna: Fix assertion to look at bbox of all boxes/points Signed-off-by: Chris Wilson commit 932743bb333e35d5f6529a701137aad4d7490555 Author: Chris Wilson Date: Sun Apr 1 09:54:43 2012 +0100 sna: Assert that drawing boxes are within bounds More sanity checks required. Signed-off-by: Chris Wilson commit cc20c45aa0ca15720510668d6918bf3c99104626 Author: Chris Wilson Date: Fri Mar 30 22:51:21 2012 +0100 sna: Minimise the risk of hotplug hangs by checking fb before vsync Everytime we issue a MI_WAIT_FOR_EVENT on a scan-line from userspace we run the risk of that pipe being disable before we submit a batch. As the pipe is then disabled or configured differently, we encounter an indefinite wait and trigger a GPU hang. To minimise the risk of a hotplug event being detected and submitting a vsynced batch prior to noticing the removal of the pipe, perform an explicit query of the current CRTC and delete the wait if we spot that our framebuffer is no longer attached. This is about as good as we can achieve without extra help from the kernel. Reported-by: Francis Leblanc Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=45413 (and others) Signed-off-by: Chris Wilson commit 305734ebdf3d51c084cfbee8804b6c60b1f03a98 Author: Chris Wilson Date: Fri Mar 30 21:27:29 2012 +0100 sna: Separate out scanline waiting for gen4 So that we do not set a gen4 bit on gen2 and apply the old workaround of trimming y2 instead. Signed-off-by: Chris Wilson commit 6f2814db6f7b89e94e54b8d73c7e176ab7d1c469 Author: Chris Wilson Date: Fri Mar 30 20:45:55 2012 +0100 sna/traps: Align the pointer+index It's the location of the pixels within the row that matter for alignment! Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47418 Signed-off-by: Chris Wilson Tested-by: Magnus Kessler commit ee075ced844350785685a0f93f88f1dc310bcc73 Author: Chris Wilson Date: Fri Mar 30 19:09:30 2012 +0100 sna/traps: Align the pointer not the indices Magnus found that inplace_row was still crashing on his setup when it tried to perform an 8-byte aligned write to an unaligned pointer. This time it looks like the row pointer itself was not 8-byte aligned, so instead of assuming that and fixing up the indices, ensure that the (index+row) results in an 8-byte aligned value. Reported-by: Magnus Kessler Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47418 Signed-off-by: Chris Wilson commit fde8a010b3d9406c2f65ee99978360a6ca54e006 Author: Chris Wilson Date: Fri Mar 30 12:47:21 2012 +0100 uxa: Remove broken render glyphs-to-dst Reported-by: Vincent Untz Reported-by: Robert Bradford Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=48045 Signed-off-by: Chris Wilson commit 451489b49916cf5a9d27844196f9656e590d9124 Author: Chris Wilson Date: Fri Mar 30 10:21:26 2012 +0100 sna/gen7: Allow per-device specific maxima As the maximum thread count and urb size differs between different incarnations of the GT units, be a little more flexible in programming those maximums. Signed-off-by: Chris Wilson commit ea36f2c4a3fa9afa8184eeaf944af9924c080368 Author: Eugeni Dodonov Date: Thu Mar 29 21:08:29 2012 -0300 Add support for Ivy Bridge GT2 Server chipset Sometimes known as Bromlow. Signed-off-by: Eugeni Dodonov Signed-off-by: Chris Wilson commit 6142232fa0feeb39412cda85ca727cc770eaa042 Author: Chris Wilson Date: Wed Mar 28 18:59:26 2012 +0100 sna: Add video sprite support for ILK+ Based on the work by Jesse Barnes. Signed-off-by: Chris Wilson commit ae8aa172a7330439a8e6dda41f5e33eb257a139b Author: Chris Wilson Date: Wed Mar 28 22:07:10 2012 +0100 sna: Fix up 32-bit overflow for maximum object size calculation Signed-off-by: Chris Wilson commit 2e7b5f7eafbf452c781e50eba7dc8323260af59e Author: Chris Wilson Date: Tue Mar 27 10:42:59 2012 +0100 sna/traps: Prefer to try mono spans on the GPU before trying inplace CPU Signed-off-by: Chris Wilson commit 522b41b393b979fdccf1ad62194cde8129f98c8c Author: Chris Wilson Date: Tue Mar 27 10:42:21 2012 +0100 sna/traps: Fix the width of the left-hand edge of an unaligned box Signed-off-by: Chris Wilson commit 008ad39b72eb86dc2f89789427be269c148feaf4 Author: Chris Wilson Date: Tue Mar 27 10:41:37 2012 +0100 sna/gen6: Reduce opaque solid OVER to SRC for render composite Signed-off-by: Chris Wilson commit a9e9041934b2486e5661ba6c550e6670009f69f3 Author: Chris Wilson Date: Mon Mar 26 16:03:42 2012 +0100 sna: Remove bogus assertion of no references to in-flight upload buffers As we may hold a cached reference to an upload buffer whilst it is in-flight, the assertion that there are no such references to a buffer being reused is no longer true. Those cached references will be released as soon as we retire the buffer during the readback and so we are free to reuse such an upload buffer for immediate readback. Reported-by: Jiri Slaby References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit 89091dd8a3fba6520a651e3e135c87546b81df9b Author: Paulo Zanoni Date: Tue Mar 20 11:53:21 2012 -0300 Avoid duplicated code with intel_output_create_ranged_atom Same change for intel_display.c and sna_display.c. Signed-off-by: Paulo Zanoni Signed-off-by: Chris Wilson commit 62f9833298ea936eaefe973eca04b1dde858fbd8 Author: Chris Wilson Date: Sun Mar 25 21:25:15 2012 +0100 sna/gen2+: Approximate expensive gradients when using imprecise rendering If we lack the ability to use a shader to compute the gradients per-pixel, we need to use pixman to render a fallback texture. We can reduce the size of this texture and upsample to reduce the cost with hopefully imperceptible loss of quality. Signed-off-by: Chris Wilson commit 25807f472d051163ed96556a409110fa405c24d1 Author: Chris Wilson Date: Fri Mar 23 14:56:06 2012 +0000 uxa: Remove hook for CompositeRectangles It was broken and not flushing damage correctly. With the improvements made to the kernel, it is no longer a significant advantage per se and not worth its additional complexity. Reported-by: Tilman Sauerbeck Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=32547 Signed-off-by: Chris Wilson commit 4460c6d0d30cf20b76854d47fd9e3aecf1839f15 Author: Chris Wilson Date: Fri Mar 23 11:05:55 2012 +0000 configure: Stop the debug build erroring out if it cannot find valgrind Another case where I passed an empty string believing that would be sufficient to replace the error path... Signed-off-by: Chris Wilson commit 168ecd96f23756e185af627f814f81b8a4c0c529 Author: Chris Wilson Date: Thu Mar 22 23:47:56 2012 +0000 sna: Make the fallback debugging messages more consistent Signed-off-by: Chris Wilson commit 8693005b5427f256bcd7c2e7cfc48d56f5b8c5ea Author: Chris Wilson Date: Thu Mar 22 23:43:04 2012 +0000 sna: Update the target bo after changing from Y tiling Signed-off-by: Chris Wilson commit 83f98d6e5c303e52c1e0fb95b6237ebf62a8edfe Author: Chris Wilson Date: Thu Mar 22 21:57:14 2012 +0000 sna: Correctly test for clear glyphs when searching for ones to skip With xterm, it is quite common for it to redraw itself by using lots of spaces and so it is efficient for us if we ellide those clear glyphs and only draw the backing boxes. However, we were only checking the first 8 pixels in each line because of a missing pointer increment. Fixes absent '=' characters when using a compositor and ImageText. Reported-by: Jiri Slaby commit 86121a3af9a9fc9a2c76d7ac9f3ec17105d20d80 Author: Chris Wilson Date: Thu Mar 22 11:21:43 2012 +0000 sna: Adjust the damage region for the composite offset References: https://bugs.freedesktop.org/show_bug.cgi?id=47597 Signed-off-by: Chris Wilson commit a6b48dd7f1eeb1a8e3841b8f3326c60b300ee9e9 Author: Chris Wilson Date: Thu Mar 22 09:22:52 2012 +0000 sna: Force fallbacks if the destination is unattached Since the removal of the ability to create a backing pixmap after the creation of its parent, it no longer becomes practical to attempt rendering with the GPU to unattached pixmaps. So having made the decision never to render to that pixmap, perform the test explicitly along the render paths. This fixes a segmentation fault introduced in 8a303f195 (sna: Remove existing damage before overwriting with a composite op) which assumed the existence of a backing pixmap along a render path. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47700 Signed-off-by: Chris Wilson commit 52f39ae1697bef86471b7c5eef8553661f255b67 Author: Chris Wilson Date: Wed Mar 21 13:31:03 2012 +0000 sna: Assert that the bo created is large enough Double check that the maximum access size computed from the bo parameters is within the allocated size for the bo. Signed-off-by: Chris Wilson commit 2f23c660f58e0db2107255f32a3503dae62fe304 Author: Chris Wilson Date: Wed Mar 21 13:30:45 2012 +0000 sna: Assert that the tiled blt is correctly clipped Signed-off-by: Chris Wilson commit e0cd13e2af3db237c171f8ca1c70ddc67b1fbce4 Author: Chris Wilson Date: Wed Mar 21 09:06:10 2012 +0000 sna: Fallback to inplace upload if forced to tile the indirect replacement Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47629 Signed-off-by: Chris Wilson commit 1db1eced5462723811e535a53cb41b108fc6c5a3 Author: Chris Wilson Date: Tue Mar 20 22:57:50 2012 +0000 sna: Tidy an assertion when handling tiled copies Signed-off-by: Chris Wilson commit 38b0cc24b45afd4ed57187fb066896bf1e467f45 Author: Chris Wilson Date: Wed Mar 21 01:01:47 2012 +0000 sna: Remove the short-circuiting of move-to-cpu for read if no damage The danger now is that we may have either discarded the shadow pixels or have replaced them with a GTT mapping, either way undesirable and so we should reconstruct the shadow mapping. Signed-off-by: Chris Wilson commit f49a6b1fd7e23252b2a306f8bdb9baed5c1e8d4b Author: Chris Wilson Date: Tue Mar 20 21:36:26 2012 +0000 sna: Decouple the private data after UnrealizeFont As the font is kept around and reused after UnrealizeFont, we need to nullify the pointer to our private data in order to prevent the later use-after-free. Reported-by: Peter Jordan Signed-off-by: Chris Wilson commit 1bdd6461de18a6812d5af0283f4222cbb80ca2f8 Author: Chris Wilson Date: Tue Mar 20 11:08:44 2012 +0000 sna/traps: Remove bogus assertion As we only enter that path for singular unbounded boxes, we are guaranteed to fill the entire trapezoid extents and so do not need the unbounded fixup the assertion was fretting about. Signed-off-by: Chris Wilson commit 1c2932e9cb283942567c3dd2695d03b8045da27f Author: Chris Wilson Date: Mon Mar 19 15:51:43 2012 +0000 uxa: Defer the call to EnterVT till after outputs are initialised We need to do this apparently or else we never perform the VT switch. However, we can not do it too early, especially not before we have finished intialising the outputs. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47395 Signed-off-by: Chris Wilson commit 972e1108a5606399a679f97af0815ec9730ab5be Author: Chris Wilson Date: Mon Mar 19 14:38:28 2012 +0000 sna: Declare videoRam correctly on gen2 devices Signed-off-by: Chris Wilson commit c0bab1df509d1f7fe36173715846c13b931e9be1 Author: Chris Wilson Date: Mon Mar 19 08:25:22 2012 +0000 sna/dri: Make the drawable as damaged for the off-screen immediate exchange In some cases off-screen is stil visible, for example under a rotation. As such xrandr -o left; glxgears -fullscreen was broken. Reported-by: Phillip Haddad Signed-off-by: Chris Wilson commit 17c19ea8e21e1b20eee446045573dfd94ce6f537 Author: Chris Wilson Date: Sun Mar 18 13:23:26 2012 +0000 sna/traps: Remove separate edge->vertical flag Mark vertical edges with dy==0 to reduce structure size and reduce memory load during edge walking. Signed-off-by: Chris Wilson commit 82023397cd05c05cfcb8939829a95e287d438875 Author: Chris Wilson Date: Sun Mar 18 11:10:14 2012 +0000 sna/gen3: Improve clear-to-solid reduction Signed-off-by: Chris Wilson commit 7ce5f4a11389b2fe9f27a6f09c4848ac71424d5d Author: Chris Wilson Date: Sun Mar 18 10:55:17 2012 +0000 sna/gen3: Do not force tiling for large pixmaps As the extraction routine is now smarter and can construction subsurfaces without copying we do not need to force tiling. Signed-off-by: Chris Wilson commit 342dda3fe361c8be2f3af5af1516cdc6a5fdcaa9 Author: Chris Wilson Date: Sun Mar 18 10:55:06 2012 +0000 sna/gen3: Prevent copy-fallback if we cannot blit Signed-off-by: Chris Wilson commit 28c089781fdf74a9bac2e138e65383748558fcc1 Author: Chris Wilson Date: Sun Mar 18 10:55:27 2012 +0000 sna: Fixup the cpu shadow mappings before uploading the box On the off-chance we arrive here with a pointer to the GTT mapping. Signed-off-by: Chris Wilson commit 79258fe560d08dd3dd22b4782360e1597e54948c Author: Chris Wilson Date: Sun Mar 18 09:45:27 2012 +0000 sna/traps: Apply some more operator and unbounded reductions Signed-off-by: Chris Wilson commit fe8866d6112c3e187d6682e9e4610325668427a0 Author: Chris Wilson Date: Sat Mar 17 21:49:56 2012 +0000 sna/gen[345]: Convert CPU mappings to GTT for vertices on submit Signed-off-by: Chris Wilson commit 97cd0c7da51024400e8900e46f51620a5f7ad402 Author: Chris Wilson Date: Sat Mar 17 23:57:46 2012 +0000 sna/traps: Upon reducing an ADD to a SRC, we need to apply the pending clear Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47444 Signed-off-by: Chris Wilson commit e31d9dacafe060dc86de801114b475fdd0142eb6 Author: Chris Wilson Date: Sat Mar 17 09:21:00 2012 +0000 sna/traps: Align indices for unrolled memset in row_inplace() The compiler presumes that the uint64_t write is naturally aligned and so may emit code that crashes with an unaligned moved. To workaround this, make sure the write is so aligned. References: https://bugs.freedesktop.org/show_bug.cgi?id=47418 Signed-off-by: Chris Wilson commit 2b4e11923d9f683f43acf8053bcec1701df25c1f Author: Chris Wilson Date: Sat Mar 17 00:05:47 2012 +0000 sna/traps: Tune inplace_end_subrows() Signed-off-by: Chris Wilson commit d887f209f1210a294afc2c3883a97f7ac9f42bad Author: Chris Wilson Date: Fri Mar 16 17:19:11 2012 +0000 sna/dri: Mesa expects the 8-bit stencil buffer to have 2-bytes per pixel The seperate stencil buffer is full of lies, why worry about one more? Signed-off-by: Chris Wilson commit 63c0d10faee3c7cca050505c2e81c416119e57e9 Author: Chris Wilson Date: Fri Mar 16 14:51:53 2012 +0000 sna/dri: Improve handling of non-front attachments for CopyRegion Confusion reigns between using the backing pixmap for the drawable for the front buffer, and a fake pixmap for the auxiliary buffers. Signed-off-by: Chris Wilson commit 324a1dffdc7dd896224fab265bd6a9cf99d7587f Author: Chris Wilson Date: Fri Mar 16 13:07:00 2012 +0000 sna/gen3: Micro-optimise solid span emission Signed-off-by: Chris Wilson commit 47c47b85f6b7ef5cbc2192da9ef0cee91f6744ca Author: Chris Wilson Date: Fri Mar 16 12:59:13 2012 +0000 sna/traps: Micro-optimise span emission Signed-off-by: Chris Wilson commit 3a26437f195bf3e5d5913f5552a0b838cafb9b2c Author: Chris Wilson Date: Fri Mar 16 12:37:25 2012 +0000 sna/traps: Tune nonzero_row Signed-off-by: Chris Wilson commit cb04cf9f4395c258987faead80de5c3a2c93082e Author: Chris Wilson Date: Fri Mar 16 09:28:24 2012 +0000 sna/traps: Make the inline u8 arithmetic more robust Signed-off-by: Chris Wilson commit d55bbdf42c75eb0fdb634e963ce81953583964b9 Author: Chris Wilson Date: Fri Mar 16 09:18:29 2012 +0000 sna/traps: Remove the old paths for mono inplace traps Dead code elimination. Signed-off-by: Chris Wilson commit fe4ad66aadcfa0c907ba6460a88c49e65225602c Author: Chris Wilson Date: Thu Mar 15 21:33:00 2012 +0000 sna: Correct the order of screen private teardown So that we do not keep a stale references of the last cached pixmap across server regeneration (or shutdown). Reported-by: Thierry Reding Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47357 Signed-off-by: Chris Wilson commit 0a8218a535babb5969a58c3a7da0215912f6fef8 Author: Chris Wilson Date: Thu Mar 15 13:20:35 2012 +0000 sna: Prefer the CPU bo for uploads if last access was not through the shadow Signed-off-by: Chris Wilson commit 77c9c03b102e6fac2279f0085034557bd68c5112 Author: Chris Wilson Date: Thu Mar 15 13:18:43 2012 +0000 sna: Fixup the shadow pixmap pointer for move-to-gpu If we choose not to use the CPU bo for the upload (because we fear the subsequent synchronisation cost), we need to fixup the shadow pointer before dereferencing it. On the move-to-cpu side, the fixup is already performed as we will need to access the shadow pixels for the subsequent drawing operation. Signed-off-by: Chris Wilson commit 0e2fbb6a1c1ad6b0971d07a91a790b0b3236dad3 Merge: 64ccc66 84d7a82 Author: Chris Wilson Date: Thu Mar 15 12:43:47 2012 +0000 uxa: Merge 'remove complicated second level caching' Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47345 Signed-off-by: Chris Wilson commit 84d7a82122530a62ab9a2f4bba3044c23a0db9b0 Author: Chris Wilson Date: Sat Mar 10 10:03:23 2012 +0000 uxa: Simplify allocation of backing pixmap Signed-off-by: Chris Wilson commit 1b1016624ad16cf5063e88a1d422e28865a9e87f Author: Chris Wilson Date: Sat Mar 10 09:31:24 2012 +0000 uxa/i915: Remove broken CA pass, fallback to magic 2-pass composite helper The backend failed to handle all the corner cases, so remove the complication. Signed-off-by: Chris Wilson commit 895a56a553e1386323a593063e5b7fbb7a410abe Author: Chris Wilson Date: Sat Mar 10 09:28:27 2012 +0000 uxa: Remove unused render_mask_solid members Signed-off-by: Chris Wilson commit 9c6f79440e8af9f264e3f6270384064dc1f44721 Author: Chris Wilson Date: Sat Mar 10 09:27:36 2012 +0000 uxa: Remove unused tracking of the current render target Signed-off-by: Chris Wilson commit 219467ac8bfab98bca82108b22eae8af3fc0bf36 Author: Chris Wilson Date: Sat Mar 10 09:26:10 2012 +0000 uxa: Simplify flush tracking Signed-off-by: Chris Wilson commit bd8fafe0c48df7f138459f590a0e9e8d0c3267b7 Author: Chris Wilson Date: Sat Mar 10 09:10:16 2012 +0000 uxa: Kill the complicated in-flight tracking Reference leak hunting. Signed-off-by: Chris Wilson commit 64ccc6698def517fc37e9ba3c41715626df0e3ca Author: Thierry Reding Date: Thu Mar 15 13:10:20 2012 +0100 configure: Keep passed-in CFLAGS for DRI tests When the user passes extra CFLAGS and CPPFLAGS to the configure script, they should be kept when performing subsequent checks with additional flags. This is required to properly build in cross-compilation setups where the user may pass in flags like --sysroot in order to pick up the cross-built dependencies. Signed-off-by: Thierry Reding commit c5c61ef378cad3c8259631138ea1e03eb02036af Author: Chris Wilson Date: Thu Mar 15 12:38:22 2012 +0000 sna/traps: dst IN WHITE does not reduce to SRC! I was getting too carried with my reductions. However, IN over a clear surface is a no-op, though unlikely! Signed-off-by: Chris Wilson commit fba49e1bb8e5b6b0e3ceace2dbddb5796ece954e Author: Chris Wilson Date: Thu Mar 15 10:31:51 2012 +0000 sna/traps: Fix off-by-one for filling vertical segments in tor_inplace If the last solid portion was exactly 4-pixels wide, we would miss filling in the mask. Signed-off-by: Chris Wilson commit 27f8f121b3d7433c6c3470b048428ca088acd2c9 Author: Chris Wilson Date: Wed Mar 14 23:43:20 2012 +0000 sna/damage: Handle a reduced damage rather than assert As we may reduce a damage to empty along the migration paths and not detect that reduced damage till later, handle those scenarios rather asserting. References: https://bugs.freedesktop.org/show_bug.cgi?42426 Signed-off-by: Chris Wilson commit abd104600252f0ebd1eb77eb5419693e1c17deda Author: Chris Wilson Date: Wed Mar 14 23:08:31 2012 +0000 sna: Treat unmapped but CPU-mappable bo as available for mapping Signed-off-by: Chris Wilson commit 2c492c0064e23457bffdf4f9ab1a7c9f1be93a87 Author: Chris Wilson Date: Wed Mar 14 23:05:58 2012 +0000 sna: Disable tiling for single row pixmaps (unless required for hw limits) Signed-off-by: Chris Wilson commit 6890592cd2b2d6f0d06c530f5e770fdc98577d4f Author: Chris Wilson Date: Wed Mar 14 21:30:13 2012 +0000 sna/traps: Explicitly create an unattach pixmap for fallback References: https://bugs.freedesktop.org/show_bug.cgi?id=42426 Signed-off-by: Chris Wilson commit 23b5640a0f8d8a944275e574f4c72f2380f7ed7a Author: Chris Wilson Date: Wed Mar 14 20:48:56 2012 +0000 sna: Avoid using kgem_bo_reference() internally So that we can keep the assertion to track the refcnt elsewhere. Signed-off-by: Chris Wilson commit 03731b4e37b0cbd3a1cb8c8ec08fddf0f670a918 Author: Chris Wilson Date: Wed Mar 14 20:46:59 2012 +0000 sna: Ellide no-op image glyphs Signed-off-by: Chris Wilson commit d0e05b4294b2f150a41dd95d52c2e6ee8479283d Author: Chris Wilson Date: Wed Mar 14 20:19:30 2012 +0000 sna: Don't mark cached upload buffers for inactivity expiration As these do not follow the normal rules of damage tracking, we have to be careful not to force migration. References: https://bugs.freedesktop.org/show_bug.cgi?id=42426 Signed-off-by: Chris Wilson commit 96ead26c4dddab637875c74a9afca6fd6662352b Author: Chris Wilson Date: Wed Mar 14 19:56:42 2012 +0000 sna: Add a couple of asserts for inactive_partial reference counting Signed-off-by: Chris Wilson commit 6a4ac18e37521d54dc36b215cbd27c56d20b2922 Author: Chris Wilson Date: Wed Mar 14 11:51:39 2012 +0000 sna/traps: Use a more direct fallback path for the CPU Signed-off-by: Chris Wilson commit 8580677ee6725144acf63f5cc2f8097376352d6e Author: Chris Wilson Date: Wed Mar 14 11:46:18 2012 +0000 sna/traps: Rasterise using pixman inplace where appropriate Signed-off-by: Chris Wilson commit 8a303f195b53a78dcbb2b6c84e7e691bedc86d25 Author: Chris Wilson Date: Wed Mar 14 11:26:03 2012 +0000 sna: Remove existing damage before overwriting with a composite op Signed-off-by: Chris Wilson commit 63eeb98463a77adadc26d9de1e640b75eefebf1a Author: Chris Wilson Date: Wed Mar 14 10:30:47 2012 +0000 sna/gen3: Look harder to see if we can indeed use the BLT for composite Signed-off-by: Chris Wilson commit d23ee0380b61e0dfd3ed56b8b4a15fd0b7956491 Author: Chris Wilson Date: Tue Mar 13 22:00:25 2012 +0000 sna: Reuse the cached upload as a source GPU bo References: https://bugs.freedesktop.org/show_bug.cgi?id=42426 Signed-off-by: Chris Wilson commit 0acec1685328b1dd9dffa804a27a122fc6e4225a Author: Chris Wilson Date: Tue Mar 13 21:29:57 2012 +0000 sna: Defer the release of the upload buffer cache till retirement Signed-off-by: Chris Wilson commit 06b28d541bdf2607edc2eb476919b28e747885d8 Author: Chris Wilson Date: Tue Mar 13 20:01:53 2012 +0000 sna: Destroy the cached upload buffer if before copying into it As we discard the buffer after the next batch, we will lose the contents of the pixmap. Instead discard the cache, and treat it as a normal shadow again. References: https://bugs.freedesktop.org/show_bug.cgi?id=42426 Signed-off-by: Chris Wilson commit 015db20287ec90f81ddb7eb2e7ce23b90aacfb2b Author: Chris Wilson Date: Tue Mar 13 16:29:41 2012 +0000 sna: Refactor source upload-to-gpu This was originally split between into functions for flexibility that nothing used, so merge the duplicated code into a common function. Signed-off-by: Chris Wilson commit 6dfe28f978b59ab2fc44cdd8ddf3743d6507588b Author: Chris Wilson Date: Tue Mar 13 15:47:27 2012 +0000 sna/gen6: Remove the double application of the render offset Cut'n'paste error from an older generation. Signed-off-by: Chris Wilson commit c751242e245eb168b92616bdcea3421be4e420bc Author: Chris Wilson Date: Tue Mar 13 12:04:04 2012 +0000 sna: Only use the cpu bo for xfer between CPU and GPU if either is busy The synchronisation costs overwhelm any benefit from offloading the copy, unless we are currently streaming the updates anyway. Signed-off-by: Chris Wilson commit c9668a772f15ab36d643a9b0616b309856146c89 Author: Chris Wilson Date: Mon Mar 12 23:28:51 2012 +0000 sna: Reduce OVER with a clear pixmap to a BLT Signed-off-by: Chris Wilson commit 2ea58256f5ce6e1c0f48d366ff7a1fbf300600de Author: Chris Wilson Date: Mon Mar 12 23:05:17 2012 +0000 sna: Reuse the same upload buffer for the duration of the batch Signed-off-by: Chris Wilson commit 8b9abe2be1f54bd8e8593ed155cc4725ac97627a Author: Chris Wilson Date: Mon Mar 12 21:09:19 2012 +0000 sna: Prefer to render very thin trapezoids inplace Signed-off-by: Chris Wilson commit 1e2d6ee31a21267ba27e4bebb883aaab08a12f30 Author: Chris Wilson Date: Mon Mar 12 20:34:42 2012 +0000 sna: Always reset the source counter after rendering to with the CPU The goal is to avoid moving to the GPU too early for a frequently modified CPU buffer. Signed-off-by: Chris Wilson commit 7bde1f55a1075eefab0f083833508265fc0e91e7 Author: Chris Wilson Date: Mon Mar 12 20:25:50 2012 +0000 sna: After move-to-gpu signals yes, force the GPU bo creation Signed-off-by: Chris Wilson commit 2ee7de1f1711d1058e5812c6aa0d94ecf9af7d8d Author: Chris Wilson Date: Mon Mar 12 20:09:05 2012 +0000 sna/trapezoids: Reduce mono ADD/OVER against a clear background to a SRC Signed-off-by: Chris Wilson commit 99c239e380b9f5134afc75bf55bf3f69e5113e38 Author: Chris Wilson Date: Mon Mar 12 19:49:30 2012 +0000 sna: Treat backing pixmaps no differently from their forward facing cousins Another fix for the large buffers overhaul. Signed-off-by: Chris Wilson commit 28792be1f31190171644d911d7b5573186fe2df2 Author: Chris Wilson Date: Mon Mar 12 19:45:35 2012 +0000 sna/display: Only flush pending output when installing a new scanout Signed-off-by: Chris Wilson commit 278c329b9e2aa9a94bd0d0ee73939814ac514e3d Author: Chris Wilson Date: Mon Mar 12 11:50:54 2012 +0000 sna/trapezoids: Further improve the clipping criteria for inplace traps Not only must we defend against the span starting too far to the right, we must also defend against the span terminating too far to the left. Signed-off-by: Chris Wilson commit 7ae45584327a10b05f7aee99bcb71e9d990a3e9b Author: Chris Wilson Date: Mon Mar 12 10:49:46 2012 +0000 sna/trapezoids: Add paranoia to ensure that the span starts within the clip Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47226 Signed-off-by: Chris Wilson commit 8136bc5e113ae06c30def3c91b1615e5fab8af44 Author: Chris Wilson Date: Sun Mar 11 19:45:55 2012 +0000 sna: Make the maximum BLT pitch assertions consistent The maximum permissibly BLT pitch value is 32767, so make the assertions match... Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47206 Signed-off-by: Chris Wilson commit 989615493608525fc252e4e94ac7259cba0741f5 Author: Chris Wilson Date: Fri Mar 9 22:44:16 2012 +0000 sna: Feed fallback mono trapezoids through the mono rasteriser Signed-off-by: Chris Wilson commit 552e4fbd2c25eb5ab0ae77e11f5f8ba2fdb29daa Author: Chris Wilson Date: Fri Mar 9 20:02:44 2012 +0000 sna/traps: Add a fast path for narrow masks Signed-off-by: Chris Wilson commit 494edfaaacaae13adfa5e727c66a83cb2294d330 Author: Chris Wilson Date: Fri Mar 9 12:19:33 2012 +0000 sna: Handle partial reads with a pending clear Skip the filling of the whole pixmap if we have a small read and we know the GPU bo is clear. Also choose to operate inplace on the GPU bo if we meet the usual criteria. Signed-off-by: Chris Wilson commit bd62dc73dcdbab34aa5c83382e46c7315d554a1a Author: Chris Wilson Date: Fri Mar 9 10:37:34 2012 +0000 sna/traps: Apply somes simple but common operator reduction for clipmasks Signed-off-by: Chris Wilson commit c25a3f7f46010660f441070ab7b9d5d1bc39ed0d Author: Chris Wilson Date: Fri Mar 9 09:43:46 2012 +0000 sna/dri: Only delivered a delayed flip if the drawable is still on the root Signed-off-by: Chris Wilson commit 93846b468e778440549ef0cae171c7fe9678ed9a Author: Chris Wilson Date: Fri Mar 9 09:43:24 2012 +0000 sna/traps: Remove some dead code This function was never used in this implementation, remove it. Signed-off-by: Chris Wilson commit 90c995736555ce14b08b69a42832d9774ba58304 Author: Chris Wilson Date: Fri Mar 9 00:37:32 2012 +0000 sna: Emit a INFO when compiled with debugging enabled It is useful to know and to receive confirmation that you have successfully compiled and executed the driver with debugging enabled. Signed-off-by: Chris Wilson commit 2e194f33db0437ea2f25c22efdad9552aefcab2f Author: Chris Wilson Date: Thu Mar 8 18:07:22 2012 +0000 sna/traps: Fix the initialisation of the error term for vertical mono edges Signed-off-by: Chris Wilson commit 635c604b787625f93763001951f8bdf66482c682 Author: Chris Wilson Date: Thu Mar 8 18:06:51 2012 +0000 sna/traps: Unroll insertion sort As the compiler cannot know the loop is bounded by a sentinel, manually unroll it. Signed-off-by: Chris Wilson commit a087430ad99c06e79249d2cdd019cb8bf7f955d3 Author: Chris Wilson Date: Thu Mar 8 17:13:39 2012 +0000 sna/gen6: Replace the memset with explict initialisation The profiles told me to kill it... Signed-off-by: Chris Wilson commit 61226cd41faf320f79ee4bd72dc77163079da853 Author: Chris Wilson Date: Thu Mar 8 13:41:58 2012 +0000 sna: Fix handling of large glyphs following large and shared buffer work Part of the large buffer handling was to move the decision making about whether to create GPU bo for a pixmap to creation time. The single instance where we change our minds later is involving large glyphs which we choose not to cache. Signed-off-by: Chris Wilson commit aa7a2bd71275e069a2e4383a26355854b0b8044c Author: Chris Wilson Date: Thu Mar 8 12:27:23 2012 +0000 sna: Fix reversed logic for CREATE_NO_RETIRE If the flag is set, we cannot retire, not the other way around! Signed-off-by: Chris Wilson commit f8d520950edb5cec878999a09d3f7e6f8b15bf5e Author: Chris Wilson Date: Thu Mar 8 12:13:36 2012 +0000 sna: Avoid NULL deference in DBG Only print out the details of the allocated CPU bo, if we actually allocate it. Signed-off-by: Chris Wilson commit 33af42e6284f104ed3c4cdba4bf3b1c29322ce9c Author: Chris Wilson Date: Thu Mar 8 12:10:24 2012 +0000 sna: Force the creation of a backing pixmap for scanout Ordinarily if the GPU is wedged, we just want to create a shadow buffer. Except that we must ensure that we do allow a bo to be created for attaching to the scanout. Signed-off-by: Chris Wilson commit 6abfa4c5662e5294edc501953dbfa7aabf0b2f57 Author: Chris Wilson Date: Thu Mar 8 11:59:02 2012 +0000 sna: Mark the pixmap for writing when creating the screen resources Signed-off-by: Chris Wilson commit 5c5ebd6a3ecf3bd83747003bc272a736b7c333c0 Author: Chris Wilson Date: Thu Mar 8 11:13:34 2012 +0000 intel: Fix typo s/asert/assert/ The joy of conditional compiles masked this compilation failure when testing. Reported-by: Reinhard Karcher Signed-off-by: Chris Wilson commit f6474883d325cff443da9ceaa99ec734e6cdc1d6 Author: Chris Wilson Date: Thu Mar 8 11:04:05 2012 +0000 sna/gen2+: Use the reduced operator from CompositeRectangles Do not attempt to further reduce the operator locally in each backend as the reduction is already performed in the upper layer. References: https://bugs.freedesktop.org/show_bug.cgi?id=42606 Signed-off-by: Chris Wilson commit 2ae3cd36ae98fa43716ef482e70364e7563cf1ea Author: Chris Wilson Date: Thu Mar 8 10:51:05 2012 +0000 sna: Tidy marking pixmap->clear for CompositeRectangles Reduce the two unsightly checks into one. Signed-off-by: Chris Wilson commit 8b21659ef2f9cec799b8215a2e5481667bc744bf Author: Chris Wilson Date: Thu Mar 8 10:50:32 2012 +0000 sna: Add some assertions around pixmap creation for render operations Signed-off-by: Chris Wilson commit 7d74300647dd815ae5b4399f84ea339045d12563 Author: Chris Wilson Date: Thu Mar 8 09:54:16 2012 +0000 sna: Discard unbound partial buffers Instead of keeping a virgin partial buffer around on its inactive list, just transfer it to the global bo cache (in actuality destroy it since it is just a kmalloc with no pages bound). Signed-off-by: Chris Wilson commit 9eb8ba4df2a99342dbb3569c043d84aa82506aba Author: Chris Wilson Date: Thu Mar 8 09:42:58 2012 +0000 sna: Preserve the offset alignment when trimming unused rows from partials Signed-off-by: Chris Wilson commit 8e6166a0e87a00248d0d4925953c923850f5b840 Author: Chris Wilson Date: Thu Mar 8 02:09:25 2012 +0000 sna: Becareful not to reduce operators for superluminal colors wine-1.4 is such an example of a crazy application. References: https://bugs.freedesktop.org/show_bug.cgi?id=42606 Signed-off-by: Chris Wilson commit 3f73cc706ff39cd4c10433791f12b5f829f62e6d Author: Chris Wilson Date: Thu Mar 8 08:54:24 2012 +0000 sna/dri: Use a counter for the number of DRI drawables attached to a pixmap The root pixmap, for instance, may have unique DRI2Drawables for each inferior window. We only want to clear the flush flag on the last release, so we need to keep a count of how many DRI drawables remain attached rather than a solitary flag. Signed-off-by: Chris Wilson commit 49a80ce1ff336fb2fa7d214bd3fddbce5a62b77a Author: Chris Wilson Date: Wed Mar 7 15:52:41 2012 +0000 sna/gen2+: Prefer not to fallback if the source is busy As if we try to perform the operation with outstanding operations on the source pixmaps, we will stall waiting for them to complete. Signed-off-by: Chris Wilson commit 4899740f6f0f5b1a5b2b3490502ccdeb7b092877 Author: Chris Wilson Date: Thu Mar 8 00:57:12 2012 +0000 sna: Do not reset partial buffers if they are not attached to the current batch As we may be holding on to them as an active mapping whilst they are executing; reseting the used counter back to zero in this case can cause corruption. Signed-off-by: Chris Wilson commit 72a7538d4e7bcf0bd7455d9e67d8751e17739e6c Author: Chris Wilson Date: Wed Mar 7 15:45:21 2012 +0000 sna: Convolution filter fixes A couple of typos made the convolution filter explode rather than convolve. Signed-off-by: Chris Wilson commit 34fe3cbb316c36c7022735cf9b03d8b655e04434 Author: Chris Wilson Date: Wed Mar 7 17:49:01 2012 +0000 sna: Avoid recursive calls to kgem_retire_partials() Whilst iterating the partial list and uploading the buffers, we need to avoid trigger a recursive call into retire should we attempt to shrink a buffer. Such a recursive call will modify the list beneath us so that we chase a stale pointer and wreak havoc with memory corruption. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47061 Signed-off-by: Chris Wilson commit 46c79e4d59ec4f90a1fa97b24a3e7058fdbfa6ba Author: Chris Wilson Date: Wed Mar 7 13:17:21 2012 +0000 sna: Restore checking for all-clipped-out for CompositeRectangles In the refactoring to avoid repeatedly applying the singular pCompositeClip, the check for the all-clipped state was lost. Signed-off-by: Chris Wilson commit 9944f1834a9c53091e4415f928f917e9247f17c5 Author: Chris Wilson Date: Wed Mar 7 10:47:42 2012 +0000 sna/gen2: Fix transformation of linear gradients Signed-off-by: Chris Wilson commit df25495eaab5bcd5baf86047a2dd0149eea00d1e Author: Chris Wilson Date: Wed Mar 7 10:43:24 2012 +0000 sna/gen4: Hook in the poor-man's linear gradient Signed-off-by: Chris Wilson commit 69d6a0387b8eec8396631714a0564b7e28d187ab Author: Chris Wilson Date: Wed Mar 7 10:43:24 2012 +0000 sna/gen5: Hook in the poor-man's linear gradient Signed-off-by: Chris Wilson commit b0d3c4f661bace84b78b194eaeee7bfa009a32cf Author: Chris Wilson Date: Wed Mar 7 10:43:24 2012 +0000 sna/gen7: Hook in the poor-man's linear gradient Signed-off-by: Chris Wilson commit dcc364a7b190ce9cec2ff35aa22c10529f84e4c3 Author: Chris Wilson Date: Wed Mar 7 10:40:50 2012 +0000 sna/gen6: Add poor-man's linear implementation Still no JIT, in the meantime we can at least cache the gradient ramps. Signed-off-by: Chris Wilson commit 232972c0e5bd833c1d0b33432fa8092c601fd0e5 Author: Chris Wilson Date: Tue Mar 6 12:17:03 2012 +0000 sna: Remove the 2-step damage flush The idea was to reduce the number of unnecessary flushes by checking for outgoing damage (could be refined further by inspecting the reply/event callback for a XDamageNotifyEvent). However, it does not flush sufficiently for the compositors' liking. As it doesn't appear to restore performance to near uncomposited levels anyway, remove the complication. Signed-off-by: Chris Wilson commit eb10ade0fc8e21cd01c0d8acba7891fd7f2aa222 Author: Chris Wilson Date: Mon Mar 5 22:55:57 2012 +0000 sna: Defer the FlushCallback removal until after the next flush Try to reduce the amount of Add/Delete ping-pong, in particular around the recreation of the DRI2 attachment to the scanout after pageflipping. Signed-off-by: Chris Wilson commit 60dacdb127599606db13ad111af8ce26c1141da5 Author: Chris Wilson Date: Mon Mar 5 22:46:20 2012 +0000 sna: Only install the flush callback for the duration of the foriegn buffer After we are no longer sharing the bo with foreign clients, we no longer need to keep flushing before every X_Reply and so we can remove the callbacks to remove the overhead of having to check every time. Signed-off-by: Chris Wilson commit b39d9f9166547effe066acfc3327dd88a019d273 Author: Chris Wilson Date: Mon Mar 5 22:29:38 2012 +0000 sna: Check for flush at the start of every WriteToClient The goal is to simply avoid the flush before going to sleep when we have no pending events. That is we only want to flush when we know there will be at least on X_Reply sent to a Client. (Preferably, it would a Damage reply!) We can safe assume that every WriteToClient marks the beginning of a new reply added to the Client output queue and thus know that upon the next flush event we will emitting a Reply and so need to submit our batches. Second attempt to fix a438e4ac. Signed-off-by: Chris Wilson commit f30b0beea4f5657a60eb5b286f41105298fa451a Author: Chris Wilson Date: Sun Mar 4 22:23:39 2012 +0000 sna/trapezoids: Ellide empty cells Signed-off-by: Chris Wilson commit b69c9dfae128ed69a397a066b8fbe62012742bf1 Author: Chris Wilson Date: Mon Mar 5 21:05:34 2012 +0000 sna/composite: Skip clipping the rectangle region against the singular clip As we will already have taken it into account when constructing the region from the rectangles. Signed-off-by: Chris Wilson commit f4846168a60feca7437f0039612ba9986c5f8d77 Author: Chris Wilson Date: Mon Mar 5 21:04:25 2012 +0000 sna: Flush dirty CPU damage before notifying the compositor Fixes regression from a438e4ac (sna: Revamp vmap support) Signed-off-by: Chris Wilson commit d7600e4e7726bb8bd6c7256ad2baf195d4427c60 Author: Chris Wilson Date: Sun Mar 4 19:12:29 2012 +0000 sna: Add some assertions to partial buffer list tracking Signed-off-by: Chris Wilson commit 3b5d556a93ac0afebf6c1dd02e0c92f6eb73f633 Author: Chris Wilson Date: Sun Mar 4 15:48:33 2012 +0000 sna: Fix assertion for checking inactive shadow buffers We may have an ordinary malloc with no CPU bo attached so check before dereferencing. Signed-off-by: Chris Wilson commit aaed9e9722aa30a3d6dc9a3f07309655de65b6bd Author: Chris Wilson Date: Fri Mar 2 23:31:24 2012 +0000 sna: Encourage promotion of snooped CPU bo to real GPU bo This fixes the regression in performance of fishietank on gen2. As the texture atlas is too large to be tiled, one might presume that it has the same performance characteristics as the snooped linear CPU buffer. It does not. Therefore if we attempt to reuse a vmap bo, promote it to a full GPU bo. This hopefully gains the benefit of avoiding the copy for single shot sources, but still gives us the benefit of avoiding the clflushes. On the plus side, it does prove that gen2 handles snoopable memory from both the blitter and the sampler! Signed-off-by: Chris Wilson commit 599cd0e8ef3080fc735860bef4e47107c1c05f9a Author: Chris Wilson Date: Fri Mar 2 20:18:32 2012 +0000 sna: Align allocations with partial buffers to 64 bytes. A magic number required for so many functions of the GPU. In this particular case it is likely to be that the offset of a texture in the GTT has to have a minimum alignment of 64 bytes. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=46415 Signed-off-by: Chris Wilson commit 4918e309dfadaf14da6997468f08db03ac0a6327 Author: Chris Wilson Date: Fri Mar 2 18:18:48 2012 +0000 sna: Silence an assertion failure during shutdown Clear the scanout flag on the front buffer during teardown to silence the debugger. Signed-off-by: Chris Wilson commit f890fc25c6c2ca358323be5a0d636c3f2ab4b298 Author: Chris Wilson Date: Fri Mar 2 18:11:56 2012 +0000 sna: And fix compilation for last commit I skipped a GCC warning about the implicit function declaration, which of course results in a runtime silent death. Oops. Signed-off-by: Chris Wilson commit 4f853acfeccb92885f154b03125d5716591a87bd Author: Chris Wilson Date: Fri Mar 2 17:36:50 2012 +0000 sna: Prevent backing pixmaps being created later We used to allow the backing pixmap to be created later in order to accommodate ShmPixmaps and ShmPutImage. However, they are now correctly handled upfront if we choose to accelerate those paths, and so all choice over whether to attach to a pixmap are made during creation and are invariant. Signed-off-by: Chris Wilson commit 866a61a2590f0c5ae6592a13d4e3de3e68f5e373 Author: Chris Wilson Date: Fri Mar 2 16:06:59 2012 +0000 sna: Disable vmap on 965gm The sampler just dies if it encounters a snoopable page, for no apparent reason. Whilst I encountered the bug on Crestline, disable it for the rest of gen4 just to be safe. Signed-off-by: Chris Wilson commit 1c653786895fc30be0e88455ce5f9caf9adc835d Author: Chris Wilson Date: Fri Mar 2 14:34:23 2012 +0000 sna: Pass usage hint for creating linear buffers As we wish to immediate map the vertices buffers, it is beneficial to search the linear cache for an existing mapping to reuse first. Signed-off-by: Chris Wilson commit 29ec36ff063472e0744af99aa81ed5ad8e291a36 Author: Chris Wilson Date: Fri Mar 2 10:01:07 2012 +0000 sna: Only discard the inplace flag for LLC partial buffers KGEM_BUFFER_WRITE_INPLACE is WRITE | INPLACE and so the typo prevented uploading of partial data through the pwrite paths. Signed-off-by: Chris Wilson commit f039ccf9587eb07528034c3247a6e700c87a5500 Author: Chris Wilson Date: Fri Mar 2 09:47:10 2012 +0000 sna: Be careful not to discard the clear operation for move-region-to-cpu When moving only a region to the CPU and we detect a pending clear, we transform the operation into a move whole pixmap. In such situations, we only have a partial damage area and so need to or in MOVE_READ to prevent the pending clear of the whole pixmap from being discarded. References: https://bugs.freedesktop.org/show_bug.cgi?id=46792 Signed-off-by: Chris Wilson commit 392593e61dac3ac65ee8e32de492c4439413ee85 Author: Chris Wilson Date: Thu Mar 1 17:54:51 2012 +0000 sna/gen5: Help the compiler avoid an uncached read Debug builds are excruitatingly slow as the compiler doesn't store the temporary in a register but uses an uncached readback instead. Maybe this will help... Signed-off-by: Chris Wilson commit 9c0c04cac245db046ef17ff24c32e6ab93535f48 Author: Chris Wilson Date: Thu Mar 1 14:52:39 2012 +0000 sna: Split storage of inactive partials As we now attempt to keep retain partial buffers after execution, we can end up will lots of inactive buffers sitting on the partial buffer list. In any one batch, we wish to minimise the number of buffers used, so keep all the inactive buffers on a seperate list and only pull from them as required. Signed-off-by: Chris Wilson commit a438e4ac9ba162e870fb22bc54024d35daa2121e Author: Chris Wilson Date: Tue Feb 28 19:15:34 2012 +0000 sna: Revamp vmap support Dust off the kernel patches and update to reflect the changes made to support LLC CPU bo, in particular to support the unsynchronized shadow buffers. However, due to the forced synchronisation required for strict client coherency we prefer not to use the vmap for shared pixmaps unless we are already busy (i.e. sync afterwards rather than before in the hope that we can squash a few operations into one). Being able to block the reply to the client until the request is actually complete and so avoid the sync remains a dream. Signed-off-by: Chris Wilson commit 272f5d9f8407d8084846b429c1722bddb3e861e9 Author: Chris Wilson Date: Thu Mar 1 11:10:03 2012 +0000 sna: Discard use of inplace GTT uploads on LLC architectures As the buffer is cache-coherent, we can read as well as write to any partial buffer so the distinction is irrelevant. Signed-off-by: Chris Wilson commit 43b1a717bae047c7ebbf99e6fa4c03b7a67896b8 Author: Chris Wilson Date: Tue Feb 28 13:37:14 2012 +0000 sna: Sort the partial buffers after stealing a write buffer It will be decoupled and not used again, but this keeps the sanity checks happy. Signed-off-by: Chris Wilson commit 8198e5872c3771e2aefabe1e3e93afa94d2ea0ec Author: Chris Wilson Date: Tue Feb 28 10:42:19 2012 +0000 sna/gen3: Tweak glyph rendering fast paths Signed-off-by: Chris Wilson commit 3c4f29820bca336af2c997bafc7ef288b455813c Author: Chris Wilson Date: Mon Feb 27 16:29:38 2012 +0000 uxa/gen3: Remove special casing of solid pictures Fixes use of alpha-groups and opacity masks in cairo. Signed-off-by: Chris Wilson commit 8f3066f0c70654f04f4acadf140e3c5c8dda051d Author: Chris Wilson Date: Mon Feb 27 13:58:58 2012 +0000 sna/gen2; Initialise channel.is-opaque for fills Signed-off-by: Chris Wilson commit 3640a0d4cb9e0f115fda9ea36212670f6ccafb22 Author: Chris Wilson Date: Mon Feb 27 12:28:22 2012 +0000 Revert "meh" This reverts commit 4adb6967a84af8a04769c2d936a41f4a49ed1428. Oops, this debugging commit was not intended to be pushed along with the bugfix. :( Signed-off-by: Chris Wilson commit 6fd8d74a6aedb6484e53e704d9f44e0bf83ae9ab Author: Chris Wilson Date: Mon Feb 27 12:06:50 2012 +0000 sna: Upload the ordinary partial buffers! Signed-off-by: Chris Wilson commit 4adb6967a84af8a04769c2d936a41f4a49ed1428 Author: Chris Wilson Date: Mon Feb 27 11:36:35 2012 +0000 meh commit 4fbb0baff525115037f7e5a1689541880a6cdafb Author: Chris Wilson Date: Sun Feb 26 22:34:30 2012 +0000 sna: Avoid reusing mmapped partial write buffers for readback An artefact of retaining the mmapped partial buffers is that it magnified the effect of stealing those for readback, causing extra writes on non-llc platforms. Signed-off-by: Chris Wilson commit a3c398a6731874ba47e0a46bbd42bf9378e12ab8 Author: Chris Wilson Date: Sat Feb 25 10:59:14 2012 +0000 sna: Retain unfinished partial buffers between batches Signed-off-by: Chris Wilson commit 8d773b88f45594f45174dc6f1a264d968690ce84 Author: Chris Wilson Date: Sat Feb 25 09:32:20 2012 +0000 sna/gen3+: Keep the vertex buffer resident between batches Signed-off-by: Chris Wilson commit 8cb773e7c809e1de23cd64d3db862d1f8e7e955a Author: Chris Wilson Date: Sat Feb 25 11:07:16 2012 +0000 sna: Ensure we trigger a retire for search_linear_cache Bo used for batch buffers are handled differently and not tracked through the active cache, so we failed to notice when we might be able to run retire and recover a suitable buffer for reuse. So simply always run retire when we might need to create a new linear buffer. Signed-off-by: Chris Wilson commit b1b4db8942e69d47aabfad3751165dc2252fa448 Author: Chris Wilson Date: Sat Feb 25 00:43:30 2012 +0000 sna: Skip a tiled bo when searching the cache for a linear mmap If we change tiling on a bo, we are effectively discarding the cached mmap so it is preferable to look for another. Signed-off-by: Chris Wilson commit 85e48d2e5eb029d8c17714cb5e7db39ea06a2455 Author: Chris Wilson Date: Fri Feb 24 21:40:44 2012 +0000 legacy: Rename XF86DRI to HAVE_DRI1 to avoid conflicts with xorg-server.h We use the XF86DRI as a user configurable option to control whether to build DRI support for i810, but it is also used internally within xorg and there exists a public define in xorg-server.h which overrides our configure option. So rename our define to HAVE_DRI1 to avoid the conflict. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=46590 Signed-off-by: Chris Wilson commit 96db90e819a1990c2d139725e522055e92def959 Author: Chris Wilson Date: Fri Feb 24 21:36:30 2012 +0000 legacy: Delete unused XF86DRI_DEVEL #define References: https://bugs.freedesktop.org/show_bug.cgi?id=46590 Signed-off-by: Chris Wilson commit b870a3e5cd922ce37050ece73e2469802dd21da2 Author: Chris Wilson Date: Fri Feb 24 11:21:49 2012 +0000 configure, NEWS: Bump version to 2.18.0 for release Another quarter, a bit late as I was debugging a few regressions, another release. Signed-off-by: Chris Wilson commit 5b5cd6780ef7cae8f49d71d7c8532597291402d8 Author: Chris Wilson Date: Fri Feb 24 11:14:26 2012 +0000 uxa: Add a option to disable the bo cache If you are suffering from regular X crashes and rendering corruption with a flood of ENOSPC or even EFILE reported in the Xorg.log, try adding this snippet to your xorg.conf: Section "Driver" Option "BufferCache" "False" EndSection References: https://bugs.freedesktop.org/show_bug.cgi?id=39552 Signed-off-by: Chris Wilson commit f8ca50818cd4f82e0ad8eeb245e8ee838b083056 Author: Gaetan Nadon Date: Sat Feb 18 13:49:02 2012 -0500 Revert "Update autotools configuration" This reverts commit 9184af921bc2f332fcb6c9b47001414378eab8e2. All X.Org modules must be able to be configured with autoconf 2.60. In addition, version 2.63 has GPL licensing issues which prevents some vendor to release software based on it. The AM_SILENT_RULES are already handled by XORG_DEFAULT_OPTIONS. All X.Org modules must be able to be configured with libtool 1.5. AM_MAINTAINER_MODE default value is "enabled" already. We use the same autogen script for all x.org modules. There are proposals for changes which should be reviewed and eventually applied to all modules together. The lt*.m4 patterns are already included in the root .gitignore file. This can be proposed as a change to all modules, but it invloves changing the topvel .gitignore, the m4/.gitignore, the ACLOCAL_AMFLAGS and the AC_CONFIG_MACRO_DIR together. For more information on project wide configuration guidelines, consult http://www.x.org/wiki/ModularDevelopersGuide and http://www.x.org/wiki/NewModuleGuidelines. Acked-by: Matthieu Herrb Signed-off-by: Gaetan Nadon commit a647aff5124674b638ef52152a5c16c27466eed7 Author: Chris Wilson Date: Thu Feb 23 12:04:09 2012 +0000 sna/gen3: Silence the compiler complaining with DBG enabled Signed-off-by: Chris Wilson commit cd3a618f583a546bafbe0c171a267774adc4d72b Author: Chris Wilson Date: Thu Feb 23 10:40:57 2012 +0000 sna/gen4 Refactor get_rectangles() to re-emit state after a flush Condense the work performed by each caller into the callee. Signed-off-by: Chris Wilson commit 6a3fa4d1b6849b666a9232b017ce82329494621b Author: Chris Wilson Date: Thu Feb 23 10:40:57 2012 +0000 sna/gen7 Refactor get_rectangles() to re-emit state after a flush Condense the work performed by each caller into the callee. Signed-off-by: Chris Wilson commit fe914eaca4b9178ab3d14053a703772f93295895 Author: Chris Wilson Date: Thu Feb 23 10:40:57 2012 +0000 sna/gen5 Refactor get_rectangles() to re-emit state after a flush Signed-off-by: Chris Wilson commit 4ecf882c838c1f044b4ed3add486b6579ea2c431 Author: Chris Wilson Date: Thu Feb 23 10:40:57 2012 +0000 sna/gen6: Refactor get_rectangles() to re-emit state after a flush Signed-off-by: Chris Wilson commit dfa21713c27eb9135cb5e1745807324e49422fde Author: Chris Wilson Date: Thu Feb 23 10:29:38 2012 +0000 sna/gen3: Refactor get_rectangles() to emit composite state and retry As gen3 only uses the single state emission block, and uniformly calls get_rectangles(), we can move that caller protocol into the callee. Signed-off-by: Chris Wilson commit a48e6e0db970ad07cd7452ba24c362d0c2fcf1bf Author: Chris Wilson Date: Thu Feb 23 10:17:34 2012 +0000 sna/gen3+: Force a batch flush when run out of CA vbo As we prematurely end the batch if we bail on extending the vbo for CA glyphs, we need to force the flush. Signed-off-by: Chris Wilson commit 57c19b10db2b512c52a593fad98b5ac5db4f1497 Author: Chris Wilson Date: Thu Feb 23 00:33:16 2012 +0000 sna: Use a CPU mapping if the bo is already in the CPU domain The heuristic of using the mapping only before the first use in an execbuffer was suboptimal and broken by the change in bo initialisation. Signed-off-by: Chris Wilson commit 510767e213c2f44563f5c438ad1234113567be90 Author: Chris Wilson Date: Wed Feb 22 18:33:09 2012 +0000 sna/gen4: Fix vertex flushing across batch flushing Due to the w/a for its buggy shaders, gen4 is significantly different that backporting the simple patch from gen5 was prone to failure. We need to check that the vertices have not already been flushed prior to flushing again. Signed-off-by: Chris Wilson commit f6392048e3e761b35644268ef161045524cfa369 Author: Chris Wilson Date: Wed Feb 22 17:00:38 2012 +0000 sna/blt: Avoid clobbering the composite state if we fail to setup the BLT Signed-off-by: Chris Wilson commit 4e86da64be947fe4b8b5f285d26d200827847a8c Author: Chris Wilson Date: Wed Feb 22 14:36:35 2012 +0000 sna/gen3+: Flush the vertices during vertex-finish But only when finishing the vbo, which is the tricky part of the recent CA glyph bugs. Signed-off-by: Chris Wilson commit 71512b2438e06b2bd256aff3657797152fbef38e Author: Chris Wilson Date: Wed Feb 22 13:44:24 2012 +0000 sna: Clear the flush flag on release of scanout/dri Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=46445 Signed-off-by: Chris Wilson commit 025d3d46dc28e967b4f42e5748d13c7fb3ad283b Author: Chris Wilson Date: Wed Feb 22 13:17:01 2012 +0000 sna/trapezoids: Only the inplace PictOpIn is unbounded Signed-off-by: Chris Wilson commit b02f866d67d60538368619849f2acda4c1706476 Author: Chris Wilson Date: Wed Feb 22 09:15:25 2012 +0000 sna: Ensure we restore the shadow pixels before uploading CPU data Reported-by: Joe Nahmias Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=46425 Signed-off-by: Chris Wilson commit be5df7b5ab487a1282c96dda92f3799e35cd19af Author: Chris Wilson Date: Tue Feb 21 23:27:33 2012 +0000 sna/gen5: Remove CA glyph workaround The root cause has been found and destroyed, so the w/a is now redundant. Signed-off-by: Chris Wilson commit 6038cede83e7f360428b4625d288411794f9d052 Author: Chris Wilson Date: Tue Feb 21 21:26:29 2012 +0000 sna/gen3+: Re-emit composite state after flushing CA vertices Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42891 Signed-off-by: Chris Wilson commit fe6602cbbc4eed1b88ac731a30b46cc970ea444f Author: Chris Wilson Date: Tue Feb 21 21:26:29 2012 +0000 sna/gen3+: Only flush the vertices after checking for end-of-batch Or upon actually closing the vertex buffer. However, the underlying issue remains. That is we are failing to re-emit the first-pass for CA text after flushing the vertex buffer (and so emitting the second-pass for the flushed vertices). Reported-by: lemens Eisserer References: https://bugs.freedesktop.org/show_bug.cgi?id=42891 Signed-off-by: Chris Wilson commit 168c87a340119e65b1d7ccbbf59da820044ca936 Author: Chris Wilson Date: Tue Feb 21 13:16:43 2012 +0000 sna: Clear the scanout flag after releasing the scanout pixmap In the future, this will be a good place to restore the cache level of the bo as well. Signed-off-by: Chris Wilson commit d7415742a5f78958489216f450411603b1eff9a7 Author: Chris Wilson Date: Tue Feb 21 16:33:26 2012 +0000 sna/dri: Queue a flush on the back DRI2 when enqueing a flip As we may wait upon the bo having finished rendering before we can execute the flip, flushing the render cache as early as possible is beneficial Signed-off-by: Chris Wilson commit 2715a455f7bfbecd7a6977184dc6180a09d06e1f Author: Chris Wilson Date: Tue Feb 21 14:49:30 2012 +0000 uxa: Prevent laggy applications by throttling after rendering Before blocking and waiting for further input, we need to make sure that we have not developed too large a queue of outstanding rendering. As we rendering to the front-buffer with no natural throttling and allow X clients to render as fast as they wish, it is entirely possible for a large queue of outstanding rendering to develop. For such an example, watch firefox rendering the fishietank demo and notice the delay that can build up before the tooltips appear. Signed-off-by: Chris Wilson commit 26721893cb41cef66db7ef626881d1eba8a5bdea Author: Chris Wilson Date: Tue Feb 21 13:31:16 2012 +0000 sna: Mark the pixmap as active for the force-to-gpu short-circuit Signed-off-by: Chris Wilson commit 1541f1afca028453379083e564bc82debf0ba39a Author: Chris Wilson Date: Tue Feb 21 13:17:35 2012 +0000 sna: Skip the CPU synchronization when marking a pixmap as inactive Signed-off-by: Chris Wilson commit 74fd55a96bd6c63e9c49a4b7fee7ef91f6b00a2c Author: Chris Wilson Date: Tue Feb 21 13:11:32 2012 +0000 sna/dri: Ensure that we reattach to the DRI2 front buffer after modeswitch If we change the Screen pixmap due to a change of mode, we lose the flag that we've attached a DRI2 buffer to it. So the next time we try to copy from/to it, reassert its DRI2 status. Signed-off-by: Chris Wilson commit d051793b9194060f5408503b1fac56958c6e58e4 Author: Chris Wilson Date: Tue Feb 21 11:42:31 2012 +0000 sna/dri: Improve error handling of failing to create a DRI2 pixmap Signed-off-by: Chris Wilson commit 580ae520cad749fb86a1bddd0fa2bcadfd60abb6 Author: Chris Wilson Date: Tue Feb 21 10:55:46 2012 +0000 sna: Short-circuit repeated calls to force-to-gpu Signed-off-by: Chris Wilson commit f2aafb98026b5c476b7f84aa2dc4c1f9ba2e573d Author: Chris Wilson Date: Tue Feb 21 10:43:11 2012 +0000 uxa: Silence compiler warning for const arguments i965_video.c: In function 'gen6_create_cc_state': i965_video.c:1374:12: warning: passing argument 4 of 'intel_bo_alloc_for_data' discards 'const' qualifier from pointer target type [enabled by default] Repeated ad nauseam. Signed-off-by: Chris Wilson commit 507f72d6d44963cae5d8d4b9da68165f73c6fd36 Author: Chris Wilson Date: Tue Feb 21 10:39:48 2012 +0000 uxa: Remove DPRINTF stubs It wasn't being used for anything non-trivial and was throwing compiler warnings, so remove it. Signed-off-by: Chris Wilson commit c72a67390ea243bf43e2ee4efe237ab88a4615b7 Author: Chris Wilson Date: Tue Feb 21 10:21:56 2012 +0000 sna/dri: Update for AsyncSwap interface changes We now need to return TRUE/FALSE depending on whether we need to invalidate the drawable. Signed-off-by: Chris Wilson commit 27bc2acf0e6a0e5e071e0d187bdf71577e821af8 Author: Chris Wilson Date: Tue Feb 21 09:29:41 2012 +0000 sna: Fix use of RegionInit() for singular regions For a singular region, we want to use a value for nboxes of 0 not 1, fortunately if you pass in a box, it ignores the value of nboxes. RegionInit() is a most peculiar API! Signed-off-by: Chris Wilson commit d98b41ce4323b43d2359349a04a4a56559e341a2 Author: Zhigang Gong Date: Fri Feb 17 19:50:52 2012 +0800 uxa/glamor/dri: Should fixup the drawable pixmap. Two fixes in this commit, first we only need to check the front left buffer, for other attachment we don't need to check them. The second is, we should fixup the pixmap's drawable not the original drawable. Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit 1e0d702c3a77f6db3dfd55b8cafc5fca4d778751 Author: Zhigang Gong Date: Fri Feb 17 19:50:51 2012 +0800 uxa/glamor/dri: Enable the pageflip support on glamor. To support easy buffer exchange at glamor layer, glamor added a new API glamor_egl_exchange_buffers() to exchange two pixmaps' EGL image and fbos and textures without recreating any of them. But this simple method's requirement is that there are two pixmaps. A exceptional case is: If we are using triple buffer when do page flipping, we will have an extra back_buffer which doesn't have a pixmap attached to it. Then each time we set that buffer to a pixmap, we will have to call the create_egl_textured_pixmap to create the corresponding EGL image and fbo and texture for it. This is not efficient. To fix this issue, this commit introduces a new back_pixmap to intel structure to hold the back buffer and corresponding glamor resources. Then we will just need to do the light weight buffer exchanging at both DDX and glamor layer. As the new back pixmap is similar to the screen pixmap and need to be handled carefully when close screen. As the glamor data structure is a per screen data, and will be released at its close screen method. The glamor's close screen method must cleanup the screen pixmap and back pixmap's glamor resources. screen pixmap is easy to get, but there is no good way to store the back pixmap. So the glamor add a new API glamor_egl_create_textured_screen_ext function to pass the back pixmap's pointer to glamor layer. This commit make us depend on glamor commit: 4e58c4f. And we increased the required glamor version from 0.3.0 to 0.3.1 Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit ce7a57994d662f340b9457a2750e4385e7d669cd Author: Zhigang Gong Date: Fri Feb 17 19:50:50 2012 +0800 uxa/dri: Refine the pageflip processing. Add a new element back_name to intel structure to track the back bo's name then avoid flink every time. And at function I830DRI2ExchangeBuffers, after finish the BO exchange between info's front and back pixmap, it set the new front bo to the screen pixmap. But the screen pixmap should be the same as front's pixmap, so this is a duplicate operation and can be removed. Signed-off-by: Zhigang Gong commit 1a65e2b8a2ebfb4d736efb7631515babad75faf2 Author: Chris Wilson Date: Tue Feb 21 08:09:52 2012 +0000 sna: Split up/down edge walking in order to handle endpoint clipping In order to prevent walking upwards off the top of the pixmap when rendering a clipped vertical edge, we need to tweak the boundary conditions for the vertical edge walker. Reported-by: Clemens Eisserer References: https://bugs.freedesktop.org/show_bug.cgi?id=46261 Signed-off-by: Chris Wilson commit 7ea44997553ffdf57b346dc9d83742c511c9e5a4 Author: Chris Wilson Date: Mon Feb 20 23:50:42 2012 +0000 sna: Restore the shadow pixels when reducing CPU damage to all Reported-by: Joe Nahmias References: https://bugs.freedesktop.org/show_bug.cgi?id=46346 Signed-off-by: Chris Wilson commit dbe0580e207ad85cb6a659f86c5746a7ecbcd036 Author: Chris Wilson Date: Mon Feb 20 16:30:53 2012 +0000 sna: gen4+ suffer no penalty for changing tiling On gen4, the tiling/fence constraints are fairly lax, only requiring page alignment of the object and its size, and so we can switch tiling modes without incurring a GPU stall on active bo. Signed-off-by: Chris Wilson commit b68b76cf54a322e80685f1ec93538cd6c5813ea4 Author: Chris Wilson Date: Mon Feb 20 13:38:11 2012 +0000 sna: Move sync'ing of CPU bo after allocation to first write The idea was that we could afford to allocate an active CPU bo for copying to from using the GPU and later sync just before we need to write to the shadow pixels. Having the sync inside the allocation function potentially causes an unwanted stall. Signed-off-by: Chris Wilson commit bbb6794a3b97b1fcf72c8712ab0ec591683b128b Author: Chris Wilson Date: Mon Feb 20 12:25:31 2012 +0000 sna: Trim clipped lines to end within bounds References: https://bugs.freedesktop.org/show_bug.cgi?id=46261 References: https://bugs.freedesktop.org/show_bug.cgi?id=45673 Signed-off-by: Chris Wilson commit 805bc3310cd0a13eab8e48e7615bdd42638cfa33 Author: Chris Wilson Date: Mon Feb 20 12:09:19 2012 +0000 sna: When reversing line-drawing direction, use the clipped endpoint Make sure we take the clipping into account if we choose to reverse the draw direction (to maintain left-to-right inside the box emission). References: https://bugs.freedesktop.org/show_bug.cgi?id=46261 References: https://bugs.freedesktop.org/show_bug.cgi?id=45673 Signed-off-by: Chris Wilson commit 7f0a4a5f7224003a0cd226137de5a068949a41b4 Author: Chris Wilson Date: Mon Feb 20 09:52:37 2012 +0000 sna/dri: Ensure the domain tracking is reset when releasing bo used for swaps Signed-off-by: Chris Wilson commit 7d147c7462ba183f5316e7e8dafeca62add6f97c Author: Chris Wilson Date: Sun Feb 19 17:50:56 2012 +0000 sna: Correct tile sizes for Y-tiling on i915g 128-byte Y-tiling wasn't introduced until the 945. Signed-off-by: Chris Wilson commit 95391b7312147760d8da01fce68b8398aa8e4e2f Author: Chris Wilson Date: Thu Feb 16 22:01:41 2012 +0000 sna/trapezoids: Presume that Precise mono rasterisation adheres to the spec References: https://bugs.freedesktop.org/show_bug.cgi?id=46156 Signed-off-by: Chris Wilson commit 52b11f63d7922032caef0f0a5979b080dbddcbfc Author: Chris Wilson Date: Thu Feb 16 11:22:23 2012 +0000 sna: Upconvert fallback trapezoids to a8 Since the hardware only handles a8 without tricky emulation and pixman insists on using a1 for sharp trapezoids we need to ensure that we convert the a1 to a8 for our trapezoidal mask. More worryingly, this path should never be hit... References: https://bugs.freedesktop.org/show_bug.cgi?id=46156 Signed-off-by: Chris Wilson commit 8050ced6204f5aca12e6c57f86308b6ad1b98209 Author: Chris Wilson Date: Wed Feb 15 17:01:18 2012 +0000 sna/dri: Mark bo as reusable after completion of a flip-event After the flip chain is completed, any residual buffers are no longer in use and so available for reuse. Signed-off-by: Chris Wilson commit fc046aabde76142fce130773d78d797d7d750ab7 Author: Chris Wilson Date: Wed Feb 15 16:08:23 2012 +0000 sna/dri: Don't attempt to change tiling if it is a no-op Signed-off-by: Chris Wilson commit 66cc9c69657ac2703f2c7fc3c2c50f06bf5daa99 Author: Chris Wilson Date: Wed Feb 15 11:58:42 2012 +0000 Be paranoid about the definition of container_of Replace any existing definition with a correct version, since there are broken container_of macros floating around the xorg includes. Signed-off-by: Chris Wilson commit c0376b7f7b083ab2e87edc36e56fd8eb99c3cd05 Author: Chris Wilson Date: Mon Feb 13 00:48:15 2012 +0000 Add a missing macro for old xorg/list.h list_last_entry() needs to be defined if we are including the xorg list.h as opposed to our standalone variant. Signed-off-by: Chris Wilson commit 87bed52180cd2abd80ef6b58384f9fd9c9968f7b Author: Chris Wilson Date: Sat Feb 11 20:54:18 2012 +0000 Include a local copy of list.h In 1.11.903, the list.h was renamed to xorg-list.h with a corresponding change to all structures. As we carried local fixes to list.h and extended functionality, just create our own list.h with a bit of handwaving to protect us for the brief existence of xorg/include/list.h. Reported-by: Armin K Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=45938 Signed-off-by: Chris Wilson commit c64ebee5fdccf313cbd3c73850e02e6fa7dd2a65 Author: Chris Wilson Date: Sat Feb 11 13:31:12 2012 +0000 sna/gen6: Prefer the render ring for copies Slower for fills, but on the current stack faster for copies, both large and small. Hopefully, when we write some good shaders for SNB, we will not only improve performance for copies but also make fills faster on the render ring than the blt? As the BLT copy routine is GPU bound for copywinpix10, and the RENDER copy routine is CPU bound and faster, I believe that we have reached the potential of the BLT ring and not yet saturated the GPU using the render copy. Note that we still do not casually switch rings, so the actual routine chosen will still be selected by the preceeding operations, so is unlikely to have any effect in practice during, for example, cairo-traces. Signed-off-by: Chris Wilson commit 6a9b50177408c919ed5c6c2463f687476af2c698 Author: Chris Wilson Date: Sat Feb 11 11:02:53 2012 +0000 sna/gen6: Suppress the CS stall for the first command in the batch The batch emission serves as a full stall, so we do not need to incur a second before our first rendering. Signed-off-by: Chris Wilson commit cbe8bed83f0b4097958c4541ad7809a05e6c6f43 Author: Chris Wilson Date: Sat Feb 11 10:58:05 2012 +0000 sna/gen7: Mention the depth-stall required before changing VS state Because one day we may actually start using VS! Copied from the addition of the w/a to Mesa by Kenneth Graunke. Signed-off-by: Chris Wilson commit 6193f2f00fa7205f9d736340318c66d116dca53e Author: Chris Wilson Date: Thu Feb 9 14:16:17 2012 +0000 sna: Fix retire after readback Upon reading, we encounter a serialisation point and so can retire all requests. However, kgem_bo_retire() wasn't correctly detecting that barrier and so we continued to using GPU detiling thinking the target was still busy. Signed-off-by: Chris Wilson commit 4d8369f8e60fd4f5a0ef49f3e9866ea5ecb21927 Author: Chris Wilson Date: Wed Feb 8 13:15:46 2012 +0000 sna/gen2+: Force upload rather than perform source transformations on the CPU If both the source and destination is on the CPU, then the thinking was it would be quicker to operate on those on the CPU rather than copy both to the GPU and then perform the operation. This turns out to be a false assumption if transformation is involved -- something to be reconsidered if pixman should ever be improved. Signed-off-by: Chris Wilson commit 8634d461bd9e5a3d3f75b0efc11db87b8d3e1245 Author: Chris Wilson Date: Wed Feb 8 09:13:27 2012 +0000 sna: Limit max CPU bo size to prevent aperture thrashing on upload Copying between two objects that consume more than the available GATT space is a painful experience due to the forced use of an intermediatory and eviction on every batch. The tiled upload paths are in comparison remarkably efficient, so favour their use when handling extremely large buffers. This reverses the previous idea in that we now prefer large GPU bo rather than large CPU bo, as the render pipeline is far more flexible for handling those than the blitter is for handling the CPU bo (at least for gen4+). Signed-off-by: Chris Wilson commit 5b16972d7850b2347efc084311d664e14263cba1 Author: Chris Wilson Date: Tue Feb 7 23:45:37 2012 +0000 sna: Check that we successfully retired an active linear buffer If we go to the trouble of running retire before searching, we may as well check that we retired something before proceeding to check all the inactive lists. Signed-off-by: Chris Wilson commit 207b4d4482a6af4a39472ec20ff04fa0c9322d73 Author: Chris Wilson Date: Tue Feb 7 21:56:29 2012 +0000 sna: Relax must-be-blittable rules for gen4+ The render pipeline is actually more flexible than the blitter for dealing with large surfaces and so the BLT is no longer the limiting factor on gen4+. Signed-off-by: Chris Wilson commit 13c960db9ef876ee99991d97dfc34fef184c0341 Author: Zhigang Gong Date: Thu Feb 2 11:30:57 2012 +0800 uxa/glamor: Use a macro to specify module name. This depends upon glamor commit b5f8d, just after the 0.3.0 tag. Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit 70092bfbc51ddc5a51c9cae21c6b2852c216a6fc Author: Zhigang Gong Date: Wed Feb 1 19:47:28 2012 +0800 uxa/glamor: Refine CloseScreen and InitScreen process. The previous version calls glamor_egl_close_screen and glamor_egl_free_screen manually which is not align with standard process. Now glamor change the way to follow standard method: glamor layer and glamor egl layer both have their internal CloseScreens. The correct sequence is after the I830CloseScreen is registered, then register glamor_egl_close_screen and the last one is glamor_close_screen. So we move out the intel_glamor_init from the intel_uxa_init to I830ScreenInit and just after the registration of I830CloseScreen. As the glamor interfaces changed, we need to check the glamor version when load the glamor egl module to make sure we are loading the right glamor module. If failed, it will switch back to UXA path. This depends upon glamor commit 1bc8bf tagged with version 0.3.0. Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit 798aad6c95a1a95fd587430dc7a6d59497a10ce1 Author: Chris Wilson Date: Tue Feb 7 20:16:48 2012 +0000 sna/gen[4-7]: Fix erroneous scale factor for partial large bo render copies Signed-off-by: Chris Wilson commit ea6588726107f1ab0ef5a8f69d420b5bff819a76 Author: Chris Wilson Date: Tue Feb 7 15:26:50 2012 +0000 sna: Apply offsets correctly for partial src/dst in large copy boxes Signed-off-by: Chris Wilson commit 14c91e108464a305fb23b3313b842bfaeb9420fc Author: Chris Wilson Date: Tue Feb 7 15:25:48 2012 +0000 sna/tiling: Request Y-tiles if we know we cannot BLT to either the src or dst Signed-off-by: Chris Wilson commit 3131217e3ecbdf6cd40b044bd31e8228ff5c6bff Author: Chris Wilson Date: Tue Feb 7 14:35:59 2012 +0000 sna: Mark up the temporary allocations Signed-off-by: Chris Wilson commit ec1ccb6bf6a984cbe3317e636d7ff73887348a46 Author: Chris Wilson Date: Tue Feb 7 13:37:52 2012 +0000 sna: Set the damage for render->copy_boxes to NULL before use Signed-off-by: Chris Wilson commit 58f634b792c6d109c5f2374a3e24d759e744abb4 Author: Chris Wilson Date: Tue Feb 7 13:32:20 2012 +0000 sna: Handle tile alignment for untiled large bo more carefully We ended up trying to align the upper bound to zero as the integer divsion of the tile width by pixel was zero. Signed-off-by: Chris Wilson commit bf3518ea910a2a3c5b4500dcba7c829576fbfd47 Author: Zhigang Gong Date: Tue Feb 7 09:40:54 2012 +0800 uxa/glamor/dri: Fix a typo bug when fixup glamor pixmap. Should modify the old pixmap's header not the new one which was already destroyed. Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit 1467a4ba1a327877026cc76b3eabeb51d1415509 Author: Chris Wilson Date: Mon Feb 6 20:41:53 2012 +0000 sna: Use the proper sna_picture_is_solid() test Rather than the specialised routines that assumed pDrawable was non-NULL, which was no longer true after f30be6f743. Signed-off-by: Chris Wilson commit ef335a65a9d9f7eb96873201b546c25fd03d090c Author: Chris Wilson Date: Mon Feb 6 20:37:54 2012 +0000 sna: Search all active buckets for a temporary allocation Reduce the need for creating a new object if we only need the allocation for a single operation. Signed-off-by: Chris Wilson commit b7e3aaf773f05ce82405e135c0f99b40b3c2f434 Author: Chris Wilson Date: Mon Feb 6 18:08:19 2012 +0000 sna: Use the clipped end-point for recomputing segment length after clipping References: https://bugs.freedesktop.org/show_bug.cgi?id=45673 Signed-off-by: Chris Wilson commit f30be6f74392f5687ffe8bc9dd2c6dc024ae06c8 Author: Chris Wilson Date: Mon Feb 6 15:59:21 2012 +0000 sna/gen2+: Exclude solids from being classed as requiring an upload We treat any pixmap that is not attached to either a CPU or GPU bo as requiring the pixel data to be uploaded to the GPU before we can composite. Normally this is true, except for the solid cache. References: https://bugs.freedesktop.org/show_bug.cgi?id=45672 Signed-off-by: Chris Wilson commit f009386de87acccadb1803567e4c494e5e80a2a9 Author: Chris Wilson Date: Mon Feb 6 09:50:03 2012 +0000 sna: If we have a CPU bo, do not assert we have shadow pixels When transferring damage to the GPU, on SNB it is not necessarily true that we have a shadow pixmap, we may instead have drawn onto an unmapped CPU bo and now simply need to copy from that bo onto the GPU. Move the assertion onto the path where it truly matters. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=45672 Signed-off-by: Chris Wilson commit 22e452ebe01c32a08599411743cf18f9ad0545a7 Author: Chris Wilson Date: Mon Feb 6 09:19:56 2012 +0000 sna: Disable use of xvmc for SNB+ Not yet implemented, so don't bother setting it to fail. References: https://bugs.freedesktop.org/show_bug.cgi?id=44874 Signed-off-by: Chris Wilson commit a8ed1a02ada1f8d2f910dfefb150d26c840bf9ea Author: Chris Wilson Date: Sat Feb 4 20:13:07 2012 +0000 sna: Discard the redundant clear of the unbounded area if already clear Signed-off-by: Chris Wilson commit b899a4b69696141ca8b897a7abf52649b09f7b3b Author: Chris Wilson Date: Sat Feb 4 20:07:49 2012 +0000 sna: Always pass the clear colour for PictOpClear Having made that optimisation for Composite, and then made the assumption that it is always true in the backends, we failed to clear the unbounded area outside of a trapezoid since we passed in the original colour and the operation was optimised as a continuation. Signed-off-by: Chris Wilson commit c107b90a44abb45c837ff8924939872be5b490eb Author: Chris Wilson Date: Sat Feb 4 16:33:34 2012 +0000 sna/gen6: Reduce PictOpClear to PictOpSrc (with blending disabled) The advantage of PictOpSrc is that it writes its results directly to memory bypassing the blend unit. Signed-off-by: Chris Wilson commit 4baa2806bc0f51e7576b769ca6750deb3821c4d3 Author: Chris Wilson Date: Sat Feb 4 12:06:22 2012 +0000 sna: Check if the damage reduces to all before performing the migration An assert exposed a situation where we had accumulated an unreduced damage-all and so we were taking the slow path only to discover later that it was a damage-all and that we had performed needless checks. Signed-off-by: Chris Wilson commit 2653524dffc1fe0dbff7d74bfc9be535d9ececb1 Author: Chris Wilson Date: Fri Feb 3 20:06:43 2012 +0000 sna: Reduce the downsample tile size to accommodate alignment If we need to enlarge the sampled tile due to tiling alignments, the resulting sample can become larger than we can accommodate through the 3D pipeline, resulting in FAIL. Signed-off-by: Chris Wilson commit 93a0b10f163ee79b6a6a7ea46b0a33b622b1f86e Author: Chris Wilson Date: Fri Feb 3 19:30:24 2012 +0000 sna: Apply redirection for the render copy into large pixmaps If the pixmap is larger than the pipeline, but the operation extents fit within the pipeline, we may be able to create a proxy target to transform the operation into one that fits within the constraints of the render pipeline. This fixes the infinite recursion hit with partially displayed extremely large images. Signed-off-by: Chris Wilson commit 4774c6b8331831e0c9f3b24f5f6e1b6ea399f628 Author: Chris Wilson Date: Thu Feb 2 15:23:03 2012 +0000 sna: Add a couple of sanity checks that the CPU drawable is on the CPU Signed-off-by: Chris Wilson commit 418cd98db7c4a2886c9e310f3691eb6c77421dd7 Author: Chris Wilson Date: Wed Feb 1 19:10:41 2012 +0000 sna/gen6: Ring switching outweighs the benefits for cairo-traces At the moment, the jury is still out on whether freely switching rings for fills is a Good Idea. So make it easier to turn it on and off for testing. Signed-off-by: Chris Wilson commit 2d0e7c7ecd7371ac7b5fe3f382fc5d04792f7019 Author: Chris Wilson Date: Wed Feb 1 14:52:56 2012 +0000 sna: Search again for a just-large-enough mapping for inplace uploads Signed-off-by: Chris Wilson commit 55c7088f54655609fbb00106679a566b46ee8dba Author: Chris Wilson Date: Wed Feb 1 01:33:52 2012 +0000 sna: Add debugging code to verify damage extents of fallback paths After using the CPU, upload the damage and read back the pixels from the GPU bo and verify that the two are equivalent. Signed-off-by: Chris Wilson commit c8fc2cde53ef7aa011ec7c47e7fc5486de0651f5 Author: Chris Wilson Date: Wed Feb 1 01:27:43 2012 +0000 sna: Fill extents for ImageGlyphs The spec says to fill the characters boxes, which is what the hardware does. The implementation fills the extents instead. rxvt expects the former, emacs the latter. Overdraw is a nuisance, but less than leaving glyphs behind... Reported-by: walch.martin@web.de Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=45438 Signed-off-by: Chris Wilson commit 13508ab5ea136caca90c846ff1026c0c1acd2ad5 Author: Chris Wilson Date: Tue Jan 31 22:38:46 2012 +0000 sna: PolyGlyph supports all of fill/tile/stipple rules The hw routines only directly supports solid fill so fallback for the interesting cases. An alternative would be to investigate using the miPolyGlyph routine to convert the weird fills into spans in order to fallback. Sounds cheaper to fallback, so wait for an actual use case. Signed-off-by: Chris Wilson commit df4e1059a4e09998334dde6aa1c8ccfe76e442c0 Author: Chris Wilson Date: Tue Jan 31 19:19:13 2012 +0000 sna/gen6: Prefer to do fills using the BLT Using the BLT is substantially faster than the current shaders for solid fill. The downside is that it invokes more ring switching. Signed-off-by: Chris Wilson commit 8b012de0a1af4ec97c3197af3f1efdcc67bc2118 Author: Chris Wilson Date: Tue Jan 31 20:29:58 2012 +0000 sna/gen5: Always prefer to emit solid fills using the BLT As the BLT is far, far faster than using a shader. Improves cairo-demos/chart from 6 to 13 fps. Reported-by: Michael Larabel Signed-off-by: Chris Wilson commit 0a748fc49d60dc2bc9494f95c4934592b111831a Author: Chris Wilson Date: Tue Jan 31 10:51:02 2012 +0000 sna: Split the tiling limits between upload and copying The kernel has a bug that prevents pwriting buffers large than the aperture. Whilst waiting for the fix, limit the upload where possible to fit within that constraint. Signed-off-by: Chris Wilson commit 9c1f8a768ca1f762c722f63bab2747e4ff1fd773 Author: Chris Wilson Date: Tue Jan 31 10:29:02 2012 +0000 sna: Avoid converting requested Y to X tiling for large pitches on gen4+ The only strong requirement is that to utilize large pitches, the object must be tiled. Having it as X tiling is a pure convenience to facilitate use of the blitter. A DRI client may want to keep using Y tiling instead. Signed-off-by: Chris Wilson commit e872c1011fc7b67683703fd891234f07dd7acd04 Author: Chris Wilson Date: Tue Jan 31 10:21:05 2012 +0000 sna/dri: We need to reduce tiling on older gen if we cannot fence Only apply the architectural limits to enable bo creation for DRI buffers. Reported-by: Alban Browaeys Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=45414 Signed-off-by: Chris Wilson commit a4caf67d8da37d04f8915d96b10411ba7267937e Author: Chris Wilson Date: Tue Jan 31 00:35:42 2012 +0000 sna: Trim tile sizes to fit into bo cache Signed-off-by: Chris Wilson commit 3f7c1646c78d8854c88b214d3699e51839ba9711 Author: Chris Wilson Date: Tue Jan 31 00:09:42 2012 +0000 sna: Check that the intermediate IO buffer can also be used for blitting Signed-off-by: Chris Wilson commit e504fab6c5354ae9d18ccefb10bd586fa49b924c Author: Chris Wilson Date: Mon Jan 30 23:49:18 2012 +0000 sna: Discard the cleared GPU buffer upon PutImage to the CPU buffer Signed-off-by: Chris Wilson commit ed1c1a7468d78e99cb4f9a4a8b8a6b00c3257a75 Author: Chris Wilson Date: Mon Jan 30 13:43:59 2012 +0000 sna: Track large objects and limit prefer-gpu hint to small objects As the GATT is irrespective of actual RAM size, we need to be careful not to be too generous when allocating GPU bo and their shadows. So first of all we limit default render targets to those small enough to fit comfortably in RAM alongside others, and secondly we try to only keep a single copy of large objects in memory. Signed-off-by: Chris Wilson commit d53d93ffa6e133f46c39595294ecf8e2182b5a68 Author: Chris Wilson Date: Mon Jan 30 15:11:34 2012 +0000 sna: Update the partial buffer allocation size when reusing an old mapping Whilst the old mapping is guaranteed to be larger than the requested allocation size, keep track of the actual size allows for better packing of future buffers. And the code also performs a sanity check that the buffer is the size we claim it to be... Signed-off-by: Chris Wilson commit 6f99555b6b64a0e1baad1853569f7bf521c327c3 Author: Chris Wilson Date: Mon Jan 30 11:48:59 2012 +0000 sna: Allow the creation of render targets larger than the maximum bo cache Given that we now handle uploads to and from bo that are larger than the aperture and that usage of such large bo is rare and so unlikely to benefit from caching, allow them to be created as render targets and destroy as soon as they become inactive. In principle, this finally enables GPU acceleration of ocitysmap on gen4+, but due to the large cost of creating and destroying large bo it is disabled on systems that require clflushing. It is, however, a pre-requisite for exploiting the enhanced capabilities of IvyBridge. Signed-off-by: Chris Wilson commit c65ec096e79aa6bda7b2b3ef235e3fd9698b4da7 Author: Chris Wilson Date: Mon Jan 30 11:41:07 2012 +0000 sna: Decrease tiling step size in case we need to enlarge the box later We can juggle rendering into large bo on gen4 by redirecting the rendering through a proxy that is tile aligned, and so the render target may be slightly larger than the tiling step size. As that is then larger than the maximum 3D pipeline, the trick fails and we need to resort to a temporary render target with copies in and out. In this case, check that the tile is aligned to the most pessimistic tiling width and reduce the step size to accomodate the enlargement. Signed-off-by: Chris Wilson commit 95f3734dd69b82e007095a599cc21f4c63d6ec00 Author: Chris Wilson Date: Mon Jan 30 11:40:02 2012 +0000 sna: Allow creation of proxies to proxies Just update the offset of the new bo by the offset of the existing proxy. Signed-off-by: Chris Wilson commit 488937edb67a60389380b405f8f8a548f51e64c7 Author: Chris Wilson Date: Mon Jan 30 11:38:36 2012 +0000 sna: Base prefer-gpu hint on default tiling choice As on gen4+, tiling increases the maximum usable pitch we can accommodate wider pixmaps on the GPU. Signed-off-by: Chris Wilson commit ca252e5b51d7b2f5a7b2c2e0d8fdb024b08096db Author: Chris Wilson Date: Sun Jan 29 19:36:55 2012 +0000 sna: Detect batch overflow and fallback rather an risk an ENOSPC Having noticed that eog was failing to perform a 8k x 8k copy with compiz running on a 965gm, it was time the checks for batch overflow were implemented. Signed-off-by: Chris Wilson commit 3aee521bf236994628c4d103a2b8f391a4be2aa7 Author: Chris Wilson Date: Sun Jan 29 19:21:54 2012 +0000 sna: Add a tiled fallback for large BLT copies If we are attempting to copy between two large bo, larger than we can fit into the aperture, break the copy into smaller steps and use an intermediatory. Signed-off-by: Chris Wilson commit df148c962108a7f3efead0b80ab4fe77f3f79c8b Author: Chris Wilson Date: Sun Jan 29 18:07:14 2012 +0000 sna: Limit the tile size for uploading into large pixmaps As we may have a constrained aperture, we need to be careful not to exceed our resources limits when uploading the pixel data. (For example, fitting two of the maximum bo into a single batch may fail due to fragmentation of the GATT.) So be cautious and use more tiles to reduce the size of each individual batch. Signed-off-by: Chris Wilson commit e1e67e8f394480eb4fef1238ccfd49cc36e4b6f2 Author: Chris Wilson Date: Sun Jan 29 15:43:42 2012 +0000 sna: Fix the "trivial" fix to improve error handling The logic was just backwards and we tried to upload a shadowless GPU pixmap. Signed-off-by: Chris Wilson commit d3fb1e1e89ccf5cefe6add66de4f960ef07cac60 Author: Chris Wilson Date: Sun Jan 29 14:20:33 2012 +0000 sna: Handle GPU creation failure when uploading subtexture Signed-off-by: Chris Wilson commit 518a99ea34b26aa094f29a4cc1ea5419f63a0e56 Author: Chris Wilson Date: Sun Jan 29 14:09:46 2012 +0000 sna: Always create a GPU bo for copying from an existent source GPU bo Make sure we prevent the readback of an active source GPU bo by always prefering to do the copy on the GPU if the data is already resisent. This fixes the second regression from e583af9cc, (sna: Experiment with creating large objects as CPU bo). Signed-off-by: Chris Wilson commit 624d9843abda9ca6bd1b004d70a6fdc082ba9653 Author: Chris Wilson Date: Sun Jan 29 13:55:20 2012 +0000 sna: Ignore map status and pick the first inactive bo for reuse This fixes the performance regression introduced with e583af9cca, (sna: Experiment with creating large objects as CPU bo), as we ended up creating fresh bo and incurring setup and thrashing overhead, when we already had plenty cached. Signed-off-by: Chris Wilson commit 5c6255ba2f12f04938fd586ca02562ee3cae05af Author: Chris Wilson Date: Sun Jan 29 11:02:38 2012 +0000 sna: Determine whether to use a partial proxy based on the pitch On gen4+ devices the maximum render pitch is much larger than is simply required for the maximum coordinates. This makes it possible to use proxy textures as a subimage into the oversized texture without having to blit into a temporary copy for virtually every single bo we use. Signed-off-by: Chris Wilson commit 65466f86263b3788b438fe021a12ade371190b01 Author: Chris Wilson Date: Sat Jan 28 01:54:47 2012 +0000 sna: Allow ridiculously large bo, up to half the total GATT Such large bo place extreme stress on the system, for example trying to mmap a 1GiB into the CPU domain currently fails due to a kernel bug. :( So if you can avoid the swap thrashing during the upload, the ddx can now handle 16k x 16k images on gen4+ on the GPU. That is fine until you want two such images... The real complication comes in uploading (and downloading) from such large textures as they are too large for a single operation with automatic detiling via either the BLT or the RENDER ring. We could do manual tiling/switching or, as this patch does, tile the transfer in chunks small enough to fit into either pipeline. Signed-off-by: Chris Wilson commit 03211f4b0b7e32b6d7dc28e60be72db204b8c8d4 Author: Chris Wilson Date: Fri Jan 27 23:18:05 2012 +0000 sna: Guard against the upload buffer growing past the maximum bo size Signed-off-by: Chris Wilson commit 2afd49a28429cdeb36583cfc31cc9b1742c1fb83 Author: Chris Wilson Date: Fri Jan 27 22:18:30 2012 +0000 sna: Limit inplace upload buffers to maximum mappable size References: https://bugs.freedesktop.org/show_bug.cgi?id=45323 Signed-off-by: Chris Wilson commit 8f4bae01e39392faa8978090db4cbe28fa00b013 Author: Chris Wilson Date: Fri Jan 27 20:28:44 2012 +0000 sna/video: Ensure the video pixmap is on the GPU The presumption that the pixmap is the scanout and so will always be pinned is false if there is a shadow or under a compositor. In those cases, the pixmap may be idle and so the GPU bo reaped. This was compounded by that the video path did not mark the pixmap as busy. So whilst watching a video under xfce4 with compositing enabled (has to be a non-GL compositor) the video would suddenly stall. Reported-by: Paul Neumann Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=45279 Signed-off-by: Chris Wilson commit d02bd80b2f9f8ee3840be5d8d6b8d389192c57f5 Author: Chris Wilson Date: Fri Jan 27 20:12:49 2012 +0000 sna: Use a proxy rather than a temporary bo for too-tall but thin targets If the render target is thin enough to fit within the 3D pipeline, but is too tall, we can fudge the address of the origin and coordinates to fit within the constaints of the pipeline. Signed-off-by: Chris Wilson commit ea433995a37f0a9d1579f74029418f22a63a2bc0 Author: Chris Wilson Date: Fri Jan 27 19:34:39 2012 +0000 sna: Experiment with a partial source If the source is thin enough such that the pitch is within the sampler's constraints and the sample size is small enough, just fudge the origin of the bo such that it can be sampled. This avoids having to create a temporary bo and use the BLT to extract it and helps, for example, firefox-asteroids which uses an 64x11200 texture atlas. Signed-off-by: Chris Wilson commit ad910949beb0c42e2d7b864f030b055ca40adacd Author: Chris Wilson Date: Fri Jan 27 18:37:39 2012 +0000 sna: Mark diagonal lines as partial write Signed-off-by: Chris Wilson commit b9c83e0b2c42f7657b1854a7cdcbc0407957c382 Author: Chris Wilson Date: Fri Jan 27 18:24:01 2012 +0000 sna/video: Add some DBG messages to track the error paths References: https://bugs.freedesktop.org/show_bug.cgi?id=45279 Signed-off-by: Chris Wilson commit 45d831c8b11d6f9f38c0cb9b5877541a971f2c67 Author: Chris Wilson Date: Fri Jan 27 15:45:17 2012 +0000 sna: Consolidate routines to choice destination bo Combine the two very similar routines that decided if we should render into the GPU bo, CPU bo or shadow pixmap into a single function. Signed-off-by: Chris Wilson commit 6402e7f1195f60f96dc269a701532c3270d8920f Author: Chris Wilson Date: Fri Jan 27 18:05:51 2012 +0000 sna: Ensure that we have a source bo for tiled fills Signed-off-by: Chris Wilson commit 6c5fb84f4de346b06e5a538e683c5a118f2579bc Author: Chris Wilson Date: Fri Jan 27 14:08:57 2012 +0000 sna/glyphs: Check that we attached to the cache pixmaps upon creation If the hw is wedged, then the pixmap creation routines will return an ordinary unattached pixmap. The code presumed that it would only return a pixmap with an attached bo, and so would segfault as it chased the invalid pointer after a GPU hang and the server was restarted. Considering that we already checked that the GPU wasn't wedged before we started, this is just mild paranoia, but on a run-once piece of code. Signed-off-by: Chris Wilson commit 86f1ae9164a94323c08e1dc6cb301e5bc1126b10 Author: Chris Wilson Date: Fri Jan 27 10:26:11 2012 +0000 sna/video: Add some more DBG breadcrumbs to the textured PutImage Signed-off-by: Chris Wilson commit ce1cae7f4715fc8e14327c5b705d9f2cc45a3741 Author: Chris Wilson Date: Fri Jan 27 00:37:51 2012 +0000 sna/video: Simplify the gen2/915gm check And make the later check in put image match. Signed-off-by: Chris Wilson commit 541908524f9ee754db3bc45d2e1681d34479c1cc Author: Chris Wilson Date: Thu Jan 26 23:14:14 2012 +0000 sna: Remove extraneous clipping from GetImage The spec says that they must wholly contained with the valid BorderClip for a Window or within the Pixmap or else a BadMatch is thrown. Rely on this behaviour and not perform the clipping ourselves. Signed-off-by: Chris Wilson commit 7ff40b572ec5cd860d7c7ff23beca0388f37c31c Author: Chris Wilson Date: Thu Jan 26 22:55:04 2012 +0000 sna: Avoid fbBlt for the easy GetImage cases From (i5-2520m): 60000 trep @ 0.6145 msec ( 1630.0/sec): GetImage 500x500 square To: 60000 trep @ 0.4949 msec ( 2020.0/sec): GetImage 500x500 square Signed-off-by: Chris Wilson commit adb1320bba15a3a3b4fa8e7d0fd0360fa696721d Author: Chris Wilson Date: Thu Jan 26 16:05:48 2012 +0000 sna/gen2+: Include being unattached in the list of source fallbacks If the source is not attached to a buffer (be it a GPU bo or a CPU bo), a temporary upload buffer would be required and so it is not worth forcing the target to the destination in that case (should the target not be on the GPU already). Signed-off-by: Chris Wilson commit b1f9415bf355e9fbd6fc3ce8fa1af59083ca2943 Author: Chris Wilson Date: Thu Jan 26 13:59:24 2012 +0000 sna: Allow gen4+ to use larger GPU bo As the blitter on gen4+ does not require fence registers, it is not restricted to operating on large objects within the mappable aperture. As we do not need to operate on such large GPU bo in place, we can relax the restriction on the maximum bo size for gen4+ to allocate for use with the GPU. Signed-off-by: Chris Wilson commit d35b6955dbb5d652d8685d2c1ea82c5e08de55ea Author: Chris Wilson Date: Thu Jan 26 14:34:36 2012 +0000 sna: Prevent mapping through the GTT for large bo If the bo is larger than a quarter of the aperture, it is unlikely that we will be able to evict enough contiguous space in the GATT to accommodate that buffer. So don't attempt to map them and use the indirect access instead. Signed-off-by: Chris Wilson commit 7c81bcd0c425cc0f7ddf2ad8289bb739c8d44289 Author: Chris Wilson Date: Thu Jan 26 13:35:15 2012 +0000 sna: Add FORCE_FALLBACK debugging hook for PutImage Signed-off-by: Chris Wilson commit 35c0ef586bf508c577642d772f18eae0b64cfd44 Author: Chris Wilson Date: Thu Jan 26 11:20:03 2012 +0000 sna/gen3: Use cpu bo if already in use Signed-off-by: Chris Wilson commit b76a6da3fa0148ef32600dd9505e22b90de037df Author: Chris Wilson Date: Thu Jan 26 10:47:01 2012 +0000 sna: Search the buckets above the desired size in the bo cache It is preferrable to reuse a slightly larger bo, than it is to create a fresh one and map it into the aperture. So search the bucket above us as well. Signed-off-by: Chris Wilson commit e2b8b1c145932e2254a705905c60f18c200cf2e8 Author: Chris Wilson Date: Thu Jan 26 12:42:12 2012 +0000 sna: Apply any previous transformation when downsampling In order to handle rotations and fractional offsets produced by the act of downsampling, we need to compute the full affine transformation and apply it to the vertices rather than attempt to fudge it with an integer offset. References: https://bugs.freedesktop.org/show_bug.cgi?id=45086 Signed-off-by: Chris Wilson commit 352828ee59164a9e81093d88dfdd45bc21f0c739 Author: Chris Wilson Date: Thu Jan 26 12:41:34 2012 +0000 sna: Tweak aperture thresholds for batch flushing In order to more easily accommodate operations on large source CPU bo. Signed-off-by: Chris Wilson commit cff6a1a2e4648eb211a1789ae9f711e2f16e9d4d Author: Chris Wilson Date: Wed Jan 25 23:21:36 2012 +0000 sna: Use the cpu bo where possible as the source for texture extraction Signed-off-by: Chris Wilson commit e583af9cca4ad2e5643317447c6b065d3ee7d11e Author: Chris Wilson Date: Wed Jan 25 23:04:50 2012 +0000 sna: Experiment with creating large objects as CPU bo Even on non-LLC systems if we can prevent the migration of such objects, we can still benefit immensely from being able to map them into the GTT as required. Signed-off-by: Chris Wilson commit 55569272f7d4232ef50f7b964dda82f85a190b99 Author: Chris Wilson Date: Wed Jan 25 20:13:27 2012 +0000 sna: Apply the same migration flags for the dst alphamap as for the dst pixmap Signed-off-by: Chris Wilson commit 4a132ddbf06e5ffc364c25002a1e46ad8bf0e45a Author: Chris Wilson Date: Wed Jan 25 20:12:55 2012 +0000 sna: Correct offset for moving drawable regions to the CPU Signed-off-by: Chris Wilson commit 65164d90b7b17ec7eea1e24d4b02ec037b55b1ff Author: Chris Wilson Date: Wed Jan 25 20:12:27 2012 +0000 sna/gen2+: Do not force use of GPU if the target is simply cleared Signed-off-by: Chris Wilson commit 307f493d76580687a3cf56106bf296475f1f53e5 Author: Chris Wilson Date: Wed Jan 25 20:11:21 2012 +0000 sna: Map freshly created, unbound bo through the CPU Take advantage that we know we will have to clflush the unbound bo before use by the GPU and populate it inplace. Signed-off-by: Chris Wilson commit d785bb7df054a1f15d59db69b089deb743bbdb40 Author: Chris Wilson Date: Wed Jan 25 16:45:43 2012 +0000 sna: GetImage is allowed to read a window's border We need to adjust the clip to include the border pixels when migrating damage from the backing pixmap. This also requires relaxing the constraint that a read must be within the drawable. Signed-off-by: Chris Wilson commit 36425ba49ecbd87b1e3bf4340ca2496d8de24a7f Author: Chris Wilson Date: Wed Jan 25 12:49:24 2012 +0000 sna: Round up buffer allocations when downsampling The pathological case being nx1 or 1xm resulting in an illegal allocation request of 0 bytes. One such example is wolframalpha.com: x = (200 + x) / 100 which generates an approximately 8500x1 image and so needs downscaling to fit in the render pipeline on all but IvyBridge. Bring on Ivy! Signed-off-by: Chris Wilson commit a2e83c6dcba1e911f42a3004b3d0782049e243e2 Author: Chris Wilson Date: Wed Jan 25 11:22:38 2012 +0000 sna: Silence compiler warning for a potential uninitialised return on error sna_accel.c: In function 'sna_copy_plane': sna_accel.c:5022:21: warning: 'ret' may be used uninitialized in this function [-Wuninitialized] Signed-off-by: Chris Wilson commit 8d22a76506133e0f76424159c0944d29bdf39da9 Author: Chris Wilson Date: Wed Jan 25 09:36:20 2012 +0000 sna: Run the miHandleExposures for no-op CopyPlane Signed-off-by: Chris Wilson commit 338941eda3c7591a85b83000eafae0407d0d7cd0 Author: Chris Wilson Date: Wed Jan 25 02:42:25 2012 +0000 sna: Handle self-copies for CopyPlane Prepare the source first as this has the dual benefit of letting us decide how best to proceed with the op (on the CPU or GPU) and prevents modification of the damage after we have choosen our preferred path. Signed-off-by: Chris Wilson commit 2e8b398ca383f5292adab8b351b8837dde3e131a Author: Chris Wilson Date: Wed Jan 25 01:42:56 2012 +0000 sna: Only shrink partial buffers that are being written to Ignore inactive and mmapped buffers. Signed-off-by: Chris Wilson commit b79252efaafe2ebc998d6cf6176a425dd897e66f Author: Chris Wilson Date: Wed Jan 25 01:36:27 2012 +0000 sna: Apply source clipping to sna_copy_plane() Ensure that the migration region is within bounds for both the source and destination pixmaps. Signed-off-by: Chris Wilson commit 46252bc7bcc7e08e47d00cdc87d6c1ed93830fcc Author: Chris Wilson Date: Wed Jan 25 01:31:34 2012 +0000 sna: Set the source clip for CopyArea fallback correctly The source window is (src->x, src->y)x(src->width, src->height) in pixmap space. However, we then need to use this to clip against the desination region, and so we need to translate from the source coordinate to the destination coordinate. Signed-off-by: Chris Wilson commit ae6d3a311783d7e063de0347363331f14bd74d74 Author: Chris Wilson Date: Wed Jan 25 01:18:11 2012 +0000 sna: Print source and destination regions for CopyArea fallback for DBG Signed-off-by: Chris Wilson commit dd5e90adfc73870cebcb215ad9fb9b5aedd38673 Author: Chris Wilson Date: Wed Jan 25 01:17:49 2012 +0000 sna: Clip GetImage to drawable so that damage migration is within bounds Signed-off-by: Chris Wilson commit b1fba5e8534da7fe253e21a3501854c04d82a108 Author: Chris Wilson Date: Wed Jan 25 01:16:34 2012 +0000 sna: Clear GPU damage first when moving a clear pixmap to the CPU This allows us to discard any busy GPU or CPU bo when we know we are going to clear the shadow pixmap afterwards. Signed-off-by: Chris Wilson commit 5ad95d66665802bce25e127ae0d06f3e0a9b0e62 Author: Chris Wilson Date: Tue Jan 24 22:11:20 2012 +0000 sna: Reduce number of reads required to inspect timers By using the information provided by select at wakeup. Signed-off-by: Chris Wilson commit aae19cbc5d0ddcf247451d06e063b2550a7ff16f Author: Chris Wilson Date: Tue Jan 24 19:58:29 2012 +0000 sna: Only reset devPrivate.ptr if owned by the CPU bo when freeing If the pixmap is mapped to the GPU bo, we should continue to use the current mapping rather than revoke it. Otherwise if we write to the GPU bo inplace, thereby discarding the CPU bo, we set the pointer we are about to copy to, to NULL. Signed-off-by: Chris Wilson commit 5312ee90ad80eda1c6ff72fbef60f77f03094972 Author: Chris Wilson Date: Tue Jan 24 19:02:56 2012 +0000 sna: mark the pixmap as no longer clear after rendering video Signed-off-by: Chris Wilson commit 69d3fc91f4a397f704b9971f97a905626e86eb3c Author: Chris Wilson Date: Tue Jan 24 18:51:32 2012 +0000 sna: Set up GC for general FillArc to FillSpans callback Signed-off-by: Chris Wilson commit 855c45c65a69d34ad32a97a5fc54dd09d40d3e26 Author: Chris Wilson Date: Tue Jan 24 18:48:17 2012 +0000 sna: Enable clipping for mi FillSpans gpu callback The mi routines do not ensure that their output is suitably constrained to the clip extents, so we must run it through the clipper. Signed-off-by: Chris Wilson commit c3fe42cde70a06a9a9a6190934df5cddb41d2f5f Author: Chris Wilson Date: Tue Jan 24 18:36:08 2012 +0000 sna: Don't attempt to create pixmaps for dpeths smaller than 8 Signed-off-by: Chris Wilson commit 22bab3cd7d0b2bcedfdd86f198d1ed9a21dd82aa Author: Chris Wilson Date: Tue Jan 24 18:33:14 2012 +0000 sna: Handle malloc failures when downsampling Signed-off-by: Chris Wilson commit 4aef9e6f50cb96c57b7ba345a18e55c97d788192 Author: Chris Wilson Date: Tue Jan 24 17:23:43 2012 +0000 sna: Add more damage-is-contained asserts Signed-off-by: Chris Wilson commit 16462e551d575bd5a370a8ac8b450ba72fea7d47 Author: Chris Wilson Date: Tue Jan 24 17:47:22 2012 +0000 sna: Install CPU handler for PolyLine for miPolySegment callback Signed-off-by: Chris Wilson commit b8b49ae70b3aadf4423bd6830c446c040f87a2f1 Author: Chris Wilson Date: Tue Jan 24 16:32:40 2012 +0000 sna: Mutate a temporary copy of the GC->ops so we don't alter new GCs In order to avoid using the wrong function for a scratch GC created during the course of a MI function whilst we have a specialised GC in use, we need to avoid modifying the original function table. Signed-off-by: Chris Wilson commit 1a116a15a5d978a1edf413a439f6ef6d3205a7aa Author: Chris Wilson Date: Tue Jan 24 16:53:21 2012 +0000 sna: Reduce the assertion that we would have hinted to operate inplace As the drawable_gc_flags() operate on lower level information than the hint, it is able to spot more oportunities to reduce the READ flags and so the assertion was overly optimistic. Signed-off-by: Chris Wilson commit 3d1bba033bc29fdf498dc082f3542c520a5ed39a Author: Chris Wilson Date: Tue Jan 24 18:22:12 2012 +0000 sna/gen3: Apply damage to video pixmap Reported-by: Paul Neumann References: https://bugs.freedesktop.org/show_bug.cgi?id=44504 Signed-off-by: Chris Wilson commit ff4b8a03a40f7eb1d71b96aebb5daa1d8df83d2a Author: Chris Wilson Date: Tue Jan 24 17:51:35 2012 +0000 sna: Use RepeatPad to avoid sampling the border color when downsampling References: https://bugs.freedesktop.org/show_bug.cgi?id=45086 Signed-off-by: Chris Wilson commit 09bddddaf94b848f417e1c9449cfe4b21d2554dd Author: Chris Wilson Date: Tue Jan 24 15:50:02 2012 +0000 sna: handle negative modulus for tiled blits Signed-off-by: Chris Wilson commit f3b11ebc28114062581c39089781114323946654 Author: Chris Wilson Date: Tue Jan 24 16:05:09 2012 +0000 sna: Add more assertions for copy errors Signed-off-by: Chris Wilson commit b56971986d2d495aa38f015beb659fb12f0d433a Author: Chris Wilson Date: Tue Jan 24 17:23:59 2012 +0000 sna: Trim the clipped Segment to the right length Signed-off-by: Chris Wilson commit 358a0861e642c26d925a69656039b4c95adba237 Author: Chris Wilson Date: Tue Jan 24 13:30:19 2012 +0000 sna: Do not discard CPU damage for a partial copy If we are copying a region that does not fill its extents (i.e. is not singular) then we must be care not to discard the CPU damage that is not overwritten by the copy. Fixes regression from 77ee92248582d65a03619d1bb1d93a74468eea00 (sna: Use full usage flags for moving the dst pixmap for a copy). Signed-off-by: Chris Wilson commit 672c6ad8170572c0f2562d5a8381ab2d17fd1e1b Author: Chris Wilson Date: Tue Jan 24 12:56:59 2012 +0000 sna: Silence the compiler with a few DBG format warnings The size of the integers being printed changed causing warnings on 32/64 bit. Signed-off-by: Chris Wilson commit 238f3d9bc563950475ee823f844197cf15a909e2 Author: Chris Wilson Date: Tue Jan 24 12:19:38 2012 +0000 sna: A deferred attachment will never have a CPU bo So we can remove that check in sna_copy_boxes. Signed-off-by: Chris Wilson commit c353a8cfde838b070f3e71d3946f3b40d1d5113a Author: Chris Wilson Date: Tue Jan 24 10:52:27 2012 +0000 sna: Revise use of the inplace hint to favour small cacheable operations The goal of the heuristic is to reduce readbacks and damage tracking on active GPU bo whilst simultaneously offering the best performance for small operations which would prefer to be performed on the shadow rather than in place. This restores ShmPutImage performance. Signed-off-by: Chris Wilson commit 29547421f392a9b589ba069cbfbba01dbe535d93 Author: Chris Wilson Date: Tue Jan 24 10:23:46 2012 +0000 sna: Use the reduced ROP from the fbGC to accurately determine readback Signed-off-by: Chris Wilson commit ec794f7594ae5e2c223b7f7255ff8979d9fb06c1 Author: Chris Wilson Date: Mon Jan 23 22:14:15 2012 +0000 sna: Track a proxy's dirty status upon itself rather than the target As proxy's are short-lived and are not used outside of the operation for which they are created, dirtied or flushed, we can safely copy the dirty status onto the proxy object itself. Signed-off-by: Chris Wilson commit 2574a04c940e032fdcf1cd39a0950515e8652471 Author: Chris Wilson Date: Mon Jan 23 21:45:29 2012 +0000 sna: Split the active bo cache into per-tiling caches Signed-off-by: Chris Wilson commit 566cbf1ef53b1f970289fcd5b3b389a74beb7e78 Author: Chris Wilson Date: Mon Jan 23 21:15:10 2012 +0000 sna: Abort search for a linear bo if we encounter a purged buffer Given the rarity of encountering a purged buffer versus the frequency of scanning the list and the then likely result of allocation a new buffer, simply abort the search on the first purged bo. Signed-off-by: Chris Wilson commit b8d3353624be51f2b618467c899a9ce3cbf3cbfe Author: Chris Wilson Date: Mon Jan 23 19:13:22 2012 +0000 sna: Refactor tests for a solid gc to use the new function This way all paths can test to see if they might be able to reduce the tiled fill or the opaque fill into a solid fill. Signed-off-by: Chris Wilson commit 78238b3a37c1c8a152bd3920e73aa6034c0f663f Author: Chris Wilson Date: Mon Jan 23 18:00:05 2012 +0000 sna: Prefer mi span filling functions to call through to the GPU Having removed the double analysis for the fast paths, at least, the span filling code on the GPU is now faster than doing the same operations in cache memory for the majority of cases. So allow the driver to prefer to use those functions when it has a GPU bo. Signed-off-by: Chris Wilson commit c5c77d04d79666962fd7c2d86bc6e7aef716084b Author: Chris Wilson Date: Mon Jan 23 18:24:21 2012 +0000 sna: Reduce GC to solid if possible along general spans path Again, we should no longer be hitting this code if at all possible, but for completeness... Signed-off-by: Chris Wilson commit 03611f662db854dc3b14f32c0315bf45bc415bbf Author: Chris Wilson Date: Mon Jan 23 19:59:06 2012 +0000 sna: Guard against being passed n==0 to FillSpans Apparently this does happen, notably by miFillPolyArc. Signed-off-by: Chris Wilson commit 7ac13a4d5aca1627b3a5fc9e7261d5dfafba970b Author: Chris Wilson Date: Tue Jan 24 00:21:12 2012 +0000 sna: Provide a fast path for miZeroLine for PolySegment Signed-off-by: Chris Wilson commit e7817a2206bd0b1cc4e4458686c328f7b41ea32c Author: Chris Wilson Date: Mon Jan 23 17:43:55 2012 +0000 sna: Override sna_fill_spans for PolySegment Signed-off-by: Chris Wilson commit 5b8db54d0fa9eab2610feed6947e917f2e400e1a Author: Chris Wilson Date: Tue Jan 24 00:21:12 2012 +0000 sna: Provide a fast path for miZeroLine for PolyLine Signed-off-by: Chris Wilson commit 280a9165db5382b93ab449158106bdd5aa7b2329 Author: Chris Wilson Date: Mon Jan 23 17:43:55 2012 +0000 sna: Override sna_fill_spans for PolyLine Signed-off-by: Chris Wilson commit 72fc3122b85fe1a1a16c7f7401c300089273fd88 Author: Chris Wilson Date: Tue Jan 24 01:27:08 2012 +0000 sna: Prefer to always use span generation for PolyArc Signed-off-by: Chris Wilson commit 98879e3786b7f6241378a32a10f2d46eb01baf85 Author: Chris Wilson Date: Mon Jan 23 23:22:00 2012 +0000 sna: Specialise PolyPoint for callback from miZeroPolyArc miZeroPolyArc may use either FillSpans for PolyPoint to generate its curves, so also provide custom point filling routines. Signed-off-by: Chris Wilson commit 5b10cb4a5ce84bd677b0a67a17d8674defb19f31 Author: Chris Wilson Date: Mon Jan 23 17:02:46 2012 +0000 sna: Override sna_fill_spans for PolyArc This is the missing like that allows the GPU to outperform fbPolyArc. Signed-off-by: Chris Wilson commit 6390a44a69d7f13a3ba16e8b48c640dc39dd227f Author: Chris Wilson Date: Mon Jan 23 15:57:53 2012 +0000 sna: Pass the reduce clip region to sna_fill_spans__cpu Since we compute it for the pixmap migration, we may as well use to perform the clipping within FillSpans as well. Signed-off-by: Chris Wilson commit 40e3745debe9738362661c7d7f498f3489c28876 Author: Chris Wilson Date: Mon Jan 23 15:42:18 2012 +0000 sna: Wrap gc->ops->PolyFillArc The goal is to avoid the overhead of performing multiple region analysis when calling sna_fill_spans by doing it once at the top level and then choose the most appropriate drawing method. Signed-off-by: Chris Wilson commit 55c2211143b3be2827fd4e299c3c211f3bd7935b Author: Chris Wilson Date: Mon Jan 23 18:15:36 2012 +0000 sna: Use a streamlined sna_fill_spans for when we know we want to use the gpu Signed-off-by: Chris Wilson commit b20ae331566e92f1a995f49fd0760f020db25035 Author: Chris Wilson Date: Mon Jan 23 19:49:18 2012 +0000 sna: Switch to using spans based on use_wide_spans() This just gives us a hook with which to conveniently enable or disable the new code. Signed-off-by: Chris Wilson commit 41739ff9517ca158d9f7b7ac82e5bacef62a3d68 Author: Chris Wilson Date: Mon Jan 23 15:42:18 2012 +0000 sna: Wrap gc->ops->FillPolygon The goal is to avoid the overhead of performing multiple region analysis when calling sna_fill_spans by doing it once at the top level and then choose the most appropriate drawing method. Signed-off-by: Chris Wilson commit aef56b194d4de9bd71ab71e3d93d81e1750262c5 Author: Chris Wilson Date: Mon Jan 23 15:38:30 2012 +0000 sna: Mark GPU bo as busy when drawing to an all-damaged pixmap for core The fast path to decide whether to use the GPU bo for the core drawaing operations forgot to update the active status of the pixmap. This included forgetting to clear the is-cleared flag. Signed-off-by: Chris Wilson commit c67e6a05ff729c0ddb26775e1c8533be1144d85a Author: Chris Wilson Date: Mon Jan 23 11:40:13 2012 +0000 sna: Align the origin of downsampling subregion to an even pixel In order for us to produce stable downsampled images across multiple frames, we need to sample the same pairs of pixels every time. This requires us to align the origin of the sample region to an even pixel. Reported-by: Clemens Eisserer References: https://bugs.freedesktop.org/show_bug.cgi?id=45086 Signed-off-by: Chris Wilson commit 9267da93969d1727cb95956c3f63643f0039c60d Author: Chris Wilson Date: Mon Jan 23 11:20:12 2012 +0000 sna: Apply composite offset to copy-clear GPU path Signed-off-by: Chris Wilson commit 803f6558c665c16ba2d20415e778b5bd1d308c2f Author: Chris Wilson Date: Mon Jan 23 11:05:50 2012 +0000 sna: Only subtract the fillrect from the CPU damage if it covers the drawable The idea behind that optimisation is for the inactive pixmap to be refreshed and allowed to be transfered back to the GPU when it is entirely redrawn. As such performing the subtraction when it does not completely remove it only incurs additional overhead. Signed-off-by: Chris Wilson commit 267493f39137704bd52df65b7605be30d2a062b2 Author: Chris Wilson Date: Mon Jan 23 10:54:13 2012 +0000 sna: Assert that the subtract operation does reduce an all-damaged Somewhere somewhen it appears that I am discarding the all-damaged flag on the pointer. The only possibility I can see is for a no-op subtraction, so put an assert there just in case the impossible is happening. Signed-off-by: Chris Wilson commit 2908fe2df7b8f870a55fc5856f94ea541a7c7b75 Author: Chris Wilson Date: Sun Jan 22 18:46:57 2012 +0000 sna: Remove assertion that all flushing list members are in the GPU domain This may not be true for external buffers that are put on the flushing list because they have foreign requests pending. Signed-off-by: Chris Wilson commit a0bc17c0fa34c9f9243817d1674d1ff235dc99d6 Author: Chris Wilson Date: Sun Jan 22 18:27:34 2012 +0000 sna/gen6: stall before blending dirty pixels from the render cache Another restriction on the render pipeline, it turns out, is that before the blend unit can read back the dst pixels in a subsequent primitive, we must stall the pipeline for the completion of that earlier primitive. This is demonstrated by cacomposite. Signed-off-by: Chris Wilson commit 5aa494305e90eed5df211427741a75380add8e96 Author: Chris Wilson Date: Sun Jan 22 16:10:39 2012 +0000 sna: Correct the offsets for performing partial downsampling on the GPU Reported-by: Clemens Eisserer References: https://bugs.freedesktop.org/show_bug.cgi?id=45086 Signed-off-by: Chris Wilson commit a7b24bb4342eefb2b23fa40d7c7216c38f2edc09 Author: Chris Wilson Date: Sun Jan 22 15:50:47 2012 +0000 sna: Correct image buffer size when uploading a downsampled image Reported-by: Clemens Eisserer References: https://bugs.freedesktop.org/show_bug.cgi?id=45086 Signed-off-by: Chris Wilson commit 238fc67aa945d5669eccd2a8fe8e98402b9a6223 Author: Chris Wilson Date: Sun Jan 22 16:04:13 2012 +0000 sna: Incude prefer-gpu hint in decision to upload CPU textures Signed-off-by: Chris Wilson commit aeaffb102190a8846c6341e72a4338b41639d679 Author: Chris Wilson Date: Sun Jan 22 15:52:41 2012 +0000 sna: Add some more DBG along the downsampling source extraction path Signed-off-by: Chris Wilson commit 92badc9c9c2636284af8bc56b098d24ff222f167 Author: Chris Wilson Date: Sun Jan 22 12:41:31 2012 +0000 sna: Avoid a common readback after clearing It is surprisingly common for a pixmap to be created, cleared and then used as an upload target or, even worse, as a source for a ShmGetImage. In order to prevent this folly, we can trivially track when we clear an entire pixmap and its GPU bo and avoid the readback in such cases. Signed-off-by: Chris Wilson commit 67a166d1247a1197fc851308c33c4970868f6a21 Author: Chris Wilson Date: Sun Jan 22 13:46:05 2012 +0000 sna: Pass inplace hint for sna_copy_boxes fallback In the unlikely event that this makes a difference, provide the hint as to when we do not read back from the destination and so a streaming copy would be preferable. Signed-off-by: Chris Wilson commit 77ee92248582d65a03619d1bb1d93a74468eea00 Author: Chris Wilson Date: Sun Jan 22 13:42:57 2012 +0000 sna: Use full usage flags for moving the dst pixmap for a copy Now that the migration code can decide for itself when to not move damage, we can pass the hints to the code rather than perform the optimisation in sna_copy_boxes. Signed-off-by: Chris Wilson commit dd079dd1f64e7832c6343c24958aa9eeb2f58789 Author: Chris Wilson Date: Sat Jan 21 19:38:40 2012 +0000 sna: Check the blt composite op is within the source bounds If we attempt to read from a drawable that is partially off its backing pixmap (such as a partially offscreen composite window) we need to fixup the read from the out-of-bounds regions to return clear. Since we don't, the easier answer is to just to switch to the render pipeline for such an operation. Signed-off-by: Chris Wilson commit 8a91c7d85740a5adc25d2a9b1972c367780ce714 Author: Chris Wilson Date: Sat Jan 21 01:18:17 2012 +0000 sna: Give kgem_create_buffer_2d a unique id The gen3, among others, backend uses the unique id of a buffer to track the currently attached buffer and uses 0 as the invalid value. Linear buffers as created by kgem_create_buffer_2d were not being assigned a unique id causing mayhem when they were then being passed to the backends as render targets and sources. In particular, gen3 did not notice the switch in render target and did not emit commands to change the GPU target nor attach the buffer to the batch, causing the sna_read_boxes to fail and for us to trigger an assertion for an uncomsumed read buffer. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42718 Signed-off-by: Chris Wilson commit d21e80433f2574fe96036ad1d563515406a1946f Author: Chris Wilson Date: Sat Jan 21 01:17:39 2012 +0000 sna/gen3: Add DBG markup for tracking target state In particular for when we switch to a new destination and mark it dirty. Signed-off-by: Chris Wilson commit 20006a9ae7f993f02c45ccbc1e414d74bfee3acd Author: Chris Wilson Date: Fri Jan 20 22:17:44 2012 +0000 sna: Cleanup PolyText and ImageText fallback debugging Signed-off-by: Chris Wilson commit 3fbd7cafb4394e0641565dc794b82bef615b4aac Author: Chris Wilson Date: Fri Jan 20 22:16:53 2012 +0000 sna: Copy the glyphs metrics before checking for an empty glyph The empty glyph still needs the correct advance, and copying it too late left it as zero and so we were collapsing spaces in PolyText8 and friends. Signed-off-by: Chris Wilson commit f0d8f5d377460bb7da82546e8304252312e97147 Author: Chris Wilson Date: Fri Jan 20 14:34:28 2012 +0000 sna/gen3: Set bo to NULL for render_fill_one It had to happen, I missed one! From fc9531fc2df. Signed-off-by: Chris Wilson commit 0286ddbd67d3b2585ffa3deccb7eeb674cd7a3e0 Author: Chris Wilson Date: Fri Jan 20 11:33:05 2012 +0000 sna/gen6: Cleanup non-pipelined w/a Move the workaround CS stall into the emit drawrect which is the only non-pipelined op we emit. This removes the split between deciding whether we will emit a drawrect and actual emission. Signed-off-by: Chris Wilson commit a0c4024e32ccab70e7f7893896d30c80ad9f4240 Author: Chris Wilson Date: Fri Jan 20 09:42:18 2012 +0000 sna/gen6: Do not enable blending on video sources Signed-off-by: Chris Wilson commit bbbda4c408c2576927582097f7cd4f606b1761fa Author: Chris Wilson Date: Fri Jan 20 09:40:04 2012 +0000 sna/gen6: Reduce the sampler stall if already stalling for drawrect If we are force to emit a stall for the non-pipelined workaround, we do not then need to emit a stall for switching samplers. Signed-off-by: Chris Wilson commit e91601278a057f70df76b74539ff4f80d203bf00 Author: Chris Wilson Date: Fri Jan 20 01:37:29 2012 +0000 sna/gen6: Force a CS stall for changing samplers On closer inspection, we still need the workaround of forcing a pipeline stall if we update the samplers. Signed-off-by: Chris Wilson commit fc9531fc2df12b896e6d9e8b3f0d7248e55bfcdd Author: Chris Wilson Date: Thu Jan 19 21:15:35 2012 +0000 sna: Move the flush to the backends This allows us to implement backend specific workarounds and use the more appropriate device specific flushing. Signed-off-by: Chris Wilson commit 2e0a534a888c59b921fd4cad69b980c6eda72212 Author: Chris Wilson Date: Thu Jan 19 18:52:35 2012 +0000 sna/gen7: Forward port recent changes from gen6 Fixes for resubmitting batches after running out of space for vertex buffers and also a couple of trivial spans functions. Signed-off-by: Chris Wilson commit 5caf806d42a1909ff3a7716d8bc592fa68a02377 Author: Chris Wilson Date: Thu Jan 19 17:33:25 2012 +0000 sna: BLT use dword pitch only for tiled surfaces The gen4+ spec is a little misleading as states that all BLT pitches for the XY commands are in dwords. Apparently not, as the upload/download functions were already demonstrating. This only became apparent when accelerating core text routines to offscreen pixmaps, such as composited windows. Signed-off-by: Chris Wilson commit dbc75532d5f6a10e9fc7cfe3634d5a0db7304a99 Author: Chris Wilson Date: Thu Jan 19 14:36:32 2012 +0000 sna: Tweak move-to-cpu to ignore inplace hint if its already on the CPU If we test the area to be drawn against the existing CPU damage and find it is already on the CPU, we may as well continue to utilize that damaged region. Signed-off-by: Chris Wilson commit 7ad4a0c9423ee7f4041173c428b07ac4af312fe1 Author: Chris Wilson Date: Thu Jan 19 11:15:38 2012 +0000 sna: Only use the blitter to emit wide spans if we cannot stream the updates If either the region is busy on the gpu or if we need to read the destination then we would incur penalties for trying to perform the operation through the GTT. However, if we are simply streaming pixels to an unbusy bo then we can do so inplace faster than computing the corresponding GPU commands and uploading them. Note: currently it is universally slower to use the GPU here (the computation of the spans is too slow). However that is only according to micro-benchmarks, avoiding the readback is likely to be more efficient in practice. Signed-off-by: Chris Wilson commit 9db6b9fad808175184849f3030862ea115ef0708 Author: Chris Wilson Date: Thu Jan 19 11:54:14 2012 +0000 sna: Also check for the inplace hint when migrating the whole pixmap Signed-off-by: Chris Wilson commit d3f7d5d614748306cc376d1929da2bbc14494a6e Author: Chris Wilson Date: Thu Jan 19 11:15:38 2012 +0000 sna: Only use the blitter to emit spans if we cannot stream the updates If either the region is busy on the gpu or if we need to read the destination then we would incur penalties for trying to perform the operation through the GTT. However, if we are simply streaming pixels to an unbusy bo then we can do so inplace faster than computing the corresponding GPU commands and uploading them. Signed-off-by: Chris Wilson commit ff2eb116ef85182eea9ed06daaa1e9a4f7bdbad3 Author: Chris Wilson Date: Wed Jan 18 00:56:07 2012 +0000 sna: Micro-optimise line extents for zero line width Handling zero line widths is the common case, so avoid the extra work. Signed-off-by: Chris Wilson commit 3c010745076204ed3b66e947c9eab6de84f670fe Author: Chris Wilson Date: Thu Jan 19 09:46:52 2012 +0000 sna: filter out degenerate segments whilst drawing unclipped PolySegment The damage layer was detecting that we were asking it to accumulate a degenerate box emanating from PolySegment, as the unclipped paths made the fatal assumption that it would not need to filter out degenerate boxes. However, a degenerate line becomes a point, does the same apply to a degenerate segment? Signed-off-by: Chris Wilson commit 35f81005f91d294e61bb4ced7cbddd1a76ccb324 Author: Chris Wilson Date: Thu Jan 19 00:41:08 2012 +0000 sna/damage: Always mark the damage as dirty when recording new boxes A few of the create_elts() routines missed marking the damage as dirty so that if only part of the emebbed box was used (i.e. the damage contained less than 8 rectangles that needed to included in the damage region) then those were being ignored during migration and testing. Reported-by: Clemens Eisserer References: https://bugs.freedesktop.org/show_bug.cgi?id=44682 Signed-off-by: Chris Wilson commit 36e691ea900d7979950c19714e15c5887d3ea039 Author: Chris Wilson Date: Thu Jan 19 00:39:28 2012 +0000 sna: Demote MOVE_READ if the GC operation covers the clip If the write operation fills the entire clip, then we can demote and possible avoid having to read back the clip from the GPU provided that we do not need the destination data due to arithmetic operation or mask. Signed-off-by: Chris Wilson commit 17efdbc48c964d9d7b0af630c13a6ceb62fb1787 Author: Chris Wilson Date: Thu Jan 19 00:36:52 2012 +0000 sna: Clip damage area with source extents for fallback The damage tracking code asserts that it only handles clip regions. However, sna_copy_area() was failing to ensure that its damage region was being clipped by the source drawable, leading to out of bounds reads during forced fallback. Signed-off-by: Chris Wilson commit fb07243c9a08b24e2dc03b86dc16d3cd72cbfcba Author: Chris Wilson Date: Wed Jan 18 21:30:34 2012 +0000 sna: Fine grained fallback debugging for core drawing routines Signed-off-by: Chris Wilson commit 05f9764a8860cbdd7ea2ce7f04c221ec8481e0bc Author: Chris Wilson Date: Wed Jan 18 21:29:45 2012 +0000 sna/damage: Fast path singular regions Mainly for consistency, so that we treat it like the other damage addition functions. Signed-off-by: Chris Wilson commit 96529e345d0718d253e4abfa0a02cd1f9f7c0406 Author: Chris Wilson Date: Wed Jan 18 21:05:40 2012 +0000 sna: Make sure we create a mappable GPU bo when streaming writes If we decide to do the CPU fallback inplace on the GPU bo through a WC mapping (because it is a large write-only operation), make sure that the new GPU bo we create is not active and so will not^W^W is less likely to cause a stall when mapped. Signed-off-by: Chris Wilson commit efce896e1dbab6177a773a7cf1c8eaf2dab0cdee Author: Chris Wilson Date: Wed Jan 18 20:53:55 2012 +0000 sna: Check number of boxes to migrate during move-to-cpu When reducing the damage we may find that it is actually empty and so sna_damage_get_boxes() returns 0, be prepared. Signed-off-by: Chris Wilson commit 334f3f70a8b4372cb089773ecded8339d2c7d984 Author: Chris Wilson Date: Wed Jan 18 20:09:26 2012 +0000 sna/gen3: Set the batch mode for emitting video state The lack of kgem_set_mode() here is causing some recently added assertions to fail. Signed-off-by: Chris Wilson commit 76203b70708b8750f52348ea8b9135191e650ff7 Author: Chris Wilson Date: Wed Jan 18 18:36:43 2012 +0000 sna: Almagamate writes based on the total number of bytes written Cachelines will only be dirtied for the bytes accessed so a better metric would based on the total number of pages brought into the TLB and the total number of cachelines used. Base the decision on whether to try and amalgamate the upload with others on the number of bytes copied rather than the overall extents. Signed-off-by: Chris Wilson commit 470741e84c813a257897ee69f14500207a2674dc Author: Chris Wilson Date: Wed Jan 18 18:35:59 2012 +0000 sna: Debug uploads All of the asserts and debug options that lead me to believe that the tiling was completely screwy for some writes. Signed-off-by: Chris Wilson commit ab387a89cf64ace9f7d7fa52b7743e78b625f47d Author: Chris Wilson Date: Wed Jan 18 18:32:51 2012 +0000 sna: Update bo->tiling during search_linear_cache search_linear_cache() was updated to track the first good match whilst it continued to search for a better match. This resulted in the first good bo being modified and a record of those modifications lost, in particular the change in tiling. Signed-off-by: Chris Wilson commit 4b893ab0812452238305b29b1beea224219f641d Author: Chris Wilson Date: Wed Jan 18 16:34:17 2012 +0000 sna: Remove defunct debugging option FORCE_GPU_ONLY now has no effect except for marking the initial pixmap as all-damaged on the GPU, and so not testing the paths for which it was originally introduction. Signed-off-by: Chris Wilson commit 965586544a8559b6b6b98c20e14c1266e46bf20d Author: Chris Wilson Date: Wed Jan 18 16:11:50 2012 +0000 sna/gen6: Don't assume that a batch mode implies a non-empty batch Just in case we set a mode then fail to emit any dwords. Sounds inefficient and woe betide the culprit when I find it... Signed-off-by: Chris Wilson commit d2e05750360fb19575f3becd87816184615855bf Author: Chris Wilson Date: Wed Jan 18 16:08:51 2012 +0000 sna: Fix some tracking of to-be-flushed dri pixmaps Signed-off-by: Chris Wilson commit 1ad5320fd40eaaf0bffcd209bf1eabd9a20e4bf8 Author: Chris Wilson Date: Wed Jan 18 15:39:35 2012 +0000 sna: Add valgrind markup for tracking CPU mmaps Signed-off-by: Chris Wilson commit f3da610eadd1f560812603ab1a72c2e173679689 Author: Chris Wilson Date: Wed Jan 18 12:59:48 2012 +0000 sna: Prevent switching rings with render disabled We fudge forced used of the BLT ring unless we install a render backend and so we must also prevent the ring from being reset when the GPU is idle. Therefore we make handing the ring status a backend function. Signed-off-by: Chris Wilson commit 6d31cb2d943be456f4ebaa663275c6a939013d9e Author: Chris Wilson Date: Wed Jan 18 11:25:30 2012 +0000 sna: Restore use of shadow pixmaps by default without RENDER support If we do not have access to an accelerated render backend, only create GPU buffers for the scanout and use an accelerated blitter for upload/download and operating inplace on the scanout. Signed-off-by: Chris Wilson commit 15a150579c3907054e186c68382873efdbadca75 Author: Chris Wilson Date: Wed Jan 18 10:26:33 2012 +0000 intel: Trivially remove a piece of XAA dependency for shadow The wolves are gathering at the door baying for the removal of XAA from Xorg. Signed-off-by: Chris Wilson commit 850495f956c811b1eb617d2e704e6bb7b5a86369 Author: Chris Wilson Date: Tue Jan 17 22:51:29 2012 +0000 sna: Fix increment of damage boxes after updating for rectangles Found by valgrind: ==13639== Conditional jump or move depends on uninitialised value(s) ==13639== at 0x5520B1E: pixman_region_init_rects (in /usr/lib/x86_64-linux-gnu/libpixman-1.so.0.24.0) ==13639== by 0x89E6ED7: __sna_damage_reduce (sna_damage.c:489) ==13639== by 0x89E7FEC: _sna_damage_contains_box (sna_damage.c:1161) ==13639== by 0x89CFCD9: sna_drawable_use_gpu_bo (sna_damage.h:175) ==13639== by 0x89D52DA: sna_poly_segment (sna_accel.c:6130) ==13639== by 0x21F87E: damagePolySegment (damage.c:1096) ==13639== by 0x1565A2: ProcPolySegment (dispatch.c:1771) ==13639== by 0x159FB0: Dispatch (dispatch.c:437) ==13639== by 0x1491D9: main (main.c:287) ==13639== Uninitialised value was created by a heap allocation ==13639== at 0x4028693: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==13639== by 0x89E6BFB: _sna_damage_create_boxes (sna_damage.c:205) ==13639== by 0x89E78F0: _sna_damage_add_rectangles (sna_damage.c:327) ==13639== by 0x89CD32D: sna_poly_fill_rect_blt.isra.65 (sna_damage.h:68) ==13639== by 0x89DE23F: sna_poly_fill_rect (sna_accel.c:8366) ==13639== by 0x21E9C8: damagePolyFillRect (damage.c:1309) ==13639== by 0x26DD3F: miPaintWindow (miexpose.c:674) ==13639== by 0x18370A: ChangeWindowAttributes (window.c:1553) ==13639== by 0x154500: ProcChangeWindowAttributes (dispatch.c:696) ==13639== by 0x159FB0: Dispatch (dispatch.c:437) ==13639== by 0x1491D9: main (main.c:287) ==13639== Use 'count' everywhere for consistency. Signed-off-by: Chris Wilson commit 4b5c9affd4968fd5b3d85551cf5a0b4d600a7892 Author: Chris Wilson Date: Tue Jan 17 20:23:26 2012 +0000 sna: Restore orginal shadow pointer before uploading CPU damage Detected by valgrind: ==22012== Source and destination overlap in memcpy(0xd101000, 0xd101000, 783360) ==22012== at 0x402A180: memcpy (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==22012== by 0x89BD4ED: memcpy_blt (blt.c:209) ==22012== by 0x89F2921: sna_write_boxes (sna_io.c:364) ==22012== by 0x89CFABF: sna_pixmap_move_to_gpu (sna_accel.c:1900) ==22012== by 0x89F49B0: sna_render_pixmap_bo (sna_render.c:571) ==22012== by 0x8A268CE: gen5_composite_picture (gen5_render.c:1908) ==22012== by 0x8A29B8A: gen5_render_composite (gen5_render.c:2252) ==22012== by 0x89E6762: sna_composite (sna_composite.c:485) ==22012== by 0x21D3C3: damageComposite (damage.c:569) ==22012== by 0x215963: ProcRenderComposite (render.c:728) ==22012== by 0x159FB0: Dispatch (dispatch.c:437) ==22012== by 0x1491D9: main (main.c:287) ==22012== Signed-off-by: Chris Wilson commit bbd6c8123635899e89911104bf84e1b7d11d66a1 Author: Eugeni Dodonov Date: Tue Jan 17 16:16:37 2012 -0200 sna: check for LLC support Instead of checking for CPU generation, use the libdrm-provided I915_PARAM_HAS_LLC instead. v2: use a define check to verify if we have I915_PARAM_HAS_LLC. Signed-off-by: Eugeni Dodonov Signed-off-by: Chris Wilson commit e4efde920bcab980451e79df7d645d1814e5f78d Author: Chris Wilson Date: Tue Jan 17 13:44:12 2012 +0000 sna: Track whether damage is a complete representation of the dirt The previous commit undoes a premature optimisation that assumed that the current damage captured all pixels written. However, it happens to be a useful optimisation along that path (tracking upload of partial images), so add the necessary booking that watches for when the union of cpu and gpu damage is no longer the complete set of all pixels written, that is if we either migrate from one pixmap to the other, the undamaged region goes untracked. We also take advantage of whenever we damage the whole pixel to restore knowledge that our tracking of all pixels written is complete. Signed-off-by: Chris Wilson commit a9b705f9a7bbd37a0543ce5fbd4c60c912d1017d Author: Chris Wilson Date: Tue Jan 17 12:58:43 2012 +0000 sna: Mark GPU as all-damaged discarding the CPU bo to prevent stalls If we discard the CPU bo, we lose knowledge of whatever regions had been initialised but no longer dirty on the GPU, but instead must assume that the entirety of the GPU bo is dirty. Signed-off-by: Chris Wilson commit 9d631e26d758eb4c2032a502604f07c5403b25ff Author: Chris Wilson Date: Tue Jan 17 12:58:03 2012 +0000 sna: Mark the freshly allocated CPU bo as in the CPU domain As we immediately use it after creation, we need to inform GEM of the domain transfer. Signed-off-by: Chris Wilson commit dfbf02b8778719b86e0f5d78ee833a19436d43a0 Author: Chris Wilson Date: Tue Jan 17 10:55:29 2012 +0000 sna: Add some DBG breadcrumbs to put_image upload paths Signed-off-by: Chris Wilson commit da90afc32f19e9168cbaa4e0cdcd941d12f2c142 Author: Chris Wilson Date: Tue Jan 17 10:24:06 2012 +0000 sna: Add DBG breadcrumbs to gradient initialisation Put some markers into the debug log as those functions create many proxies causing a lot of debug noise. Signed-off-by: Chris Wilson commit d14341cb22e37f52070cd92b707fec5e08038e96 Author: Chris Wilson Date: Tue Jan 17 00:31:43 2012 +0000 sna: Add a render ring detiling read path For SNB, in case you really, really want to use GPU detiling and not incur the ring switch. Tweaking when to just mmap the target seems to gain most anyway... The ulterior motive is that this provides fallback paths for avoiding the use of TILING_Y with GTT mmaps which is broken on 855gm. Signed-off-by: Chris Wilson commit 3620f9ca459f7bbbb01786c95b5583023b6d870c Author: Chris Wilson Date: Tue Jan 17 00:23:04 2012 +0000 sna: Cap pwrite buffer alignment to 64k We only want to create huge pwrite buffers when populating the inactive cache for mmapped uploads. In the absence of using mmap for upload, be more conservative with the alignment value so as not to simply waste valuable aperture and memory. Signed-off-by: Chris Wilson commit b9f59b1099eeba3fb8b25693d4f37dd13267d8a3 Author: Chris Wilson Date: Tue Jan 17 00:20:51 2012 +0000 sna: correct adjust of a stolen 2d read buffer If we steal a write buffer for creating a pixmap for read back, then we need to be careful as we will have set the used amount to 0 and then try to incorrectly decrease by the last row. Fortunately, we do not yet have any code that attempts to create a 2d buffer for reading. Signed-off-by: Chris Wilson commit 6fc4cdafeba4517b94202ff292f05fdd90e52b2e Author: Chris Wilson Date: Mon Jan 16 21:34:43 2012 +0000 sna: Correct assertion for a partial read buffer The batch may legitimately be submitted prior to the attachment of the read buffer, if, for example, we need to switch rings. Therefore update the assertion to only check that the bo remains in existence via either a reference from the exec or from the user Signed-off-by: Chris Wilson commit 377f5e16cd14bdf42e67b0c8f2e13b75526f3a85 Author: Chris Wilson Date: Mon Jan 16 15:51:33 2012 +0000 sna/gen[45]: clear the state tracker before setting the formats When backporting the patches from gen6, I didn't notice the memset that came later, and this wasn't along the paths checked by rendercheck. Signed-off-by: Chris Wilson commit 6387f2fb8a40f32cc4a3e4228983dcf0604ac4dc Author: Chris Wilson Date: Mon Jan 16 14:12:36 2012 +0000 sna/gen[4567]: x1r5g5b5 is only a render target, not sampler Whilst we can render to and blend with an depth 15 target, we cannot use it as a texture with the sampling engine. Signed-off-by: Chris Wilson commit 8b2bb666662305ab88aad8198ad69b1c98407d75 Author: Chris Wilson Date: Mon Jan 16 13:35:05 2012 +0000 sna/gen6: Restore the non-pipelined op after every WM binding table update The hw wants it as demonstrated by the '>' in KDE's menus. Why is it always KDE that demonstrates coherency problems... Signed-off-by: Chris Wilson commit a11b22d172898ebfc657339a9ca56102ae92592f Author: Chris Wilson Date: Mon Jan 16 13:06:58 2012 +0000 sna/gen[23]: Remark the destination bo as dirty after flushing One of the side-effects of emitting the composite state is that it tags the destination surface as dirty as a result of the *forthcoming* operation. So emitting the flush after emitting the composite state clears that tag, so we need to restore it for future coherency. Signed-off-by: Chris Wilson commit 2f09363a6e2c77fb23bf0e4c12c4507521fa631b Author: Zhigang Gong Date: Thu Jan 12 15:30:15 2012 +0800 uxa/glamor: Create glamor pixmap by default. When creating native glamor pixmaps we will get much better performance than using the textured-drm pixmap, this commit is to make that the default behaviour when configured to use glamor. Another advantage of this commit is that we reduce the risk of encountering the "incompatible region exists for this name" and the associated render corruption. And since we now never intentionally allocate a reusable pixmap we could just make all (intel_glamor) allocations non-reusable without incurring too great an overhead. A side effect is that those glamor pixmaps do not have a valid BO attached to them and thus it fails to get a DRI drawable. This commit also fixes that problem by adjusting the fixup_shadow mechanism to recreate a textured-drm pixmap from the native glamor pixmap. I tested this with mutter, and it works fine. The performance gain to apply this patch is about 10% to 20% with different workload. Signed-off-by: Zhigang Gong Reviewed-by: Chris Wilson commit fd4c139a3959df90c7f078817fc6e2f3db715bf8 Author: Chris Wilson Date: Mon Jan 16 00:32:12 2012 +0000 sna: On LLC systems quietly replace all linear mmappings using the CPU If the GPU and CPU caches are shared and coherent, we can use a cached mapping for linear bo in the CPU domain with no penalty and so avoid the penalty of using WC/UC mappings through the GTT (and any aperture pressure). We presume that the bo for such mappings are indeed LLC cached... Signed-off-by: Chris Wilson commit c20a729d0a57fd5a782d9114535d9474f39a8950 Author: Chris Wilson Date: Sun Jan 15 22:19:22 2012 +0000 sna/gen6: Force a batch submission after allocation failure during composite Signed-off-by: Chris Wilson commit 380a2fca3cce4c99c5026ab800f7885a1959b16d Author: Chris Wilson Date: Sun Jan 15 22:13:20 2012 +0000 sna: Optimise call to composite with single box Signed-off-by: Chris Wilson commit 9f89250de1dc134f54cb68012fa5eb996c61d57e Author: Chris Wilson Date: Sun Jan 15 21:54:13 2012 +0000 sna: Use the prefer-GPU hint for forcing allocation for core drawing Similar to the render paths and simpler than the current look up tiling method. Signed-off-by: Chris Wilson commit 8652bf7a196f53842db70f5c70aded31b470b0ab Author: Chris Wilson Date: Sun Jan 15 19:04:01 2012 +0000 sna: Don't track an unmatching tiled bo when searching the linear cache Signed-off-by: Chris Wilson commit cc4b616990fa71e96d38512d8cdb4c3e2abb21c0 Author: Chris Wilson Date: Sun Jan 15 19:55:50 2012 +0000 sna/video: Increase the level of paranoia In how many different ways can we check that the scanout is allocated before we start decoding video? Signed-off-by: Chris Wilson commit 7f480ba02c66fcc02bad483731c4c0cf6c746c0d Author: Chris Wilson Date: Sun Jan 15 18:45:06 2012 +0000 sna: Tidy search through active bo cache Perform the assertions upon cache consistency upfront, and tidy the indentation. Signed-off-by: Chris Wilson commit 6f7bc35d7f956f7c2507eabc874ead1a83c85ddb Author: Chris Wilson Date: Sun Jan 15 18:14:24 2012 +0000 sna: Use indirect uploads rather than teardown existing CPU maps Allow the snoopable CPU mapping to be used inplace of the GTT map for untiled bo. Signed-off-by: Chris Wilson commit 475fa67ed320f94df37ec86fe6c5dda886923751 Author: Chris Wilson Date: Sun Jan 15 17:30:00 2012 +0000 sna: Fast path move-area-to-cpu when the pixmap is already on the cpu Signed-off-by: Chris Wilson commit 37ced44a53008debaf869ec9ef4ba2e5d6982e76 Author: Chris Wilson Date: Sun Jan 15 15:35:57 2012 +0000 sna: Be a little more lenient wrt damage migration if we have CPU bo The idea being that they facilitate copying to and from the CPU, but also we want to avoid stalling on any pixels help by the CPU bo. Signed-off-by: Chris Wilson commit e3732a6f7f61a959521be9a668bba045591e633c Author: Chris Wilson Date: Sun Jan 15 11:06:59 2012 +0000 sna: Defer ring switching until after a period of idleness Similar to the desire to flush the next batch after an overflow, we do not want to incur any lag in the midst of drawing, even if that lag is mitigated by GPU semaphores. Signed-off-by: Chris Wilson commit 5df7147b0934dadc63459c2c0f33e0663692425a Author: Chris Wilson Date: Sun Jan 15 10:28:00 2012 +0000 sna: Restore the kgem_create_map() symbol As the stub is exported to the driver even in the absence of vmapping. Signed-off-by: Chris Wilson commit be53740c6f0bb182e3d88f0597f127732c839316 Author: Chris Wilson Date: Sun Jan 15 10:16:13 2012 +0000 sna: Various DBG typos Fix some mispellings inside the DBG messages Signed-off-by: Chris Wilson commit 349e9a7b94199e759acaaccac3abf5e28f3c246f Author: Chris Wilson Date: Sun Jan 15 10:04:11 2012 +0000 sna: Prefer read-boxes inplace again Using the gpu to do the detiling just incurs extra latency and an extra copy, so go back to using a fence and GTT mapping for the common path. Signed-off-by: Chris Wilson commit 09dc8b1b358aa33836d511b75f92e8d096bc7e59 Author: Chris Wilson Date: Sat Jan 14 23:21:29 2012 +0000 sna/gen7: Check reused source for validity Be sure the mask picture has a valid format even though it points to the same pixels as the valid source. And also be wary if the source was converted to a solid, but the mask is not. Signed-off-by: Chris Wilson commit d9871f01d8c4584c1f39eb09b51d5461863934e9 Author: Chris Wilson Date: Sat Jan 14 23:21:29 2012 +0000 sna/gen6: Check reused source for validity Be sure the mask picture has a valid format even though it points to the same pixels as the valid source. And also be wary if the source was converted to a solid, but the mask is not. Signed-off-by: Chris Wilson commit 1d6030322e2c3bae87a0173a32fb8d341dea560c Author: Chris Wilson Date: Sat Jan 14 23:21:29 2012 +0000 sna/gen5: Check reused source for validity Be sure the mask picture has a valid format even though it points to the same pixels as the valid source. And also be wary if the source was converted to a solid, but the mask is not. Signed-off-by: Chris Wilson commit 0e4a24ef6c186909c99a501cb606994b5c10a813 Author: Chris Wilson Date: Sat Jan 14 23:21:29 2012 +0000 sna/gen4: Check reused source for validity Be sure the mask picture has a valid format even though it points to the same pixels as the valid source. And also be wary if the source was converted to a solid, but the mask is not. Signed-off-by: Chris Wilson commit ea299f2523d5adeca84274bb7003cde8ec743a44 Author: Chris Wilson Date: Sat Jan 14 23:21:29 2012 +0000 sna/gen3: Check reused source for validity Be sure the mask picture has a valid format even though it points to the same pixels as the valid source. And also be wary if the source was converted to a solid, but the mask is not. Signed-off-by: Chris Wilson commit 007da2f97894814eaded4d24e0481f950ca7bd00 Author: Chris Wilson Date: Sat Jan 14 23:21:29 2012 +0000 sna/gen2: Check reused source for validity Be sure the mask picture has a valid format even though it points to the same pixels as the valid source. And also be wary if the source was converted to a solid, but the mask is not. Signed-off-by: Chris Wilson commit 1d55b42fbd77e85ab8930b78bb5a9e3bccbca49e Author: Chris Wilson Date: Sat Jan 14 23:44:23 2012 +0000 sna: Fix read back of partial mmapped buffers Do not move a GTT mapped buffer into the CPU domain, it causes untold pain for no benefit! Signed-off-by: Chris Wilson commit 046e945173397e5fee36a231303b68a6bc9d809d Author: Chris Wilson Date: Sat Jan 14 23:25:52 2012 +0000 sna: Discard read buffers after use Rather than pollute the cache with bo that are known not to be in the GTT and are no longer useful, drop the bo after we read from it. Signed-off-by: Chris Wilson commit 421ee0bb53321e9fb2f17c850c51b80cc385bf57 Author: Chris Wilson Date: Sun Jan 15 00:46:47 2012 +0000 sna: Do not assume that the mappable aperture size is a power of two And instead derive a power-of-two alignment value for partial buffer sizes from the mappable aperture size and use that during kgem_create_buffer() Reported-by: Clemens Eisserer References: https://bugs.freedesktop.org/show_bug.cgi?id=44682 Signed-off-by: Chris Wilson commit 5d5da35c9fab4e7154921861ecbb83befb8840e5 Author: Chris Wilson Date: Sat Jan 14 18:28:24 2012 +0000 sna/gen[23]: Check for room in the batch before emitting pipeline flushes Use a single idiom and reuse the check built into the state emission, for both spans/boxes paths. Signed-off-by: Chris Wilson commit f7e4799687d5010ea0b56aeb6fbbb2e854b1b363 Author: Chris Wilson Date: Sat Jan 14 16:53:38 2012 +0000 sna/gen6: Allow greater use of BLT Typically we will be bound to the RENDER ring as once engaged we try not to switch. However, with semaphores enabled we may switch more freely and there it is advantageous to use as much of the faster BLT as is feasible. The most contentious point here is the choice of whether to use BLT for copies by default. microbenchmarks (compwinwin) benefit from the coallescing performed in the render batch, but the more complex traces seem to prefer utilizing the blitter. The debate will continue... Signed-off-by: Chris Wilson commit c1ce34d4509a3f3e963d82ac0569a21706892f8e Author: Chris Wilson Date: Sat Jan 14 15:08:18 2012 +0000 sna/gen6: Tidy markup for when using the BLT is truly preferrable Signed-off-by: Chris Wilson commit b64751dbdb1b88b91ad97aaf995b4261876cf860 Author: Chris Wilson Date: Sat Jan 14 13:50:50 2012 +0000 sna: Be more lenient wrt switching rings if the kernel supports semaphores If the kernel uses GPU semaphores for its coherency mechanism between rings rather than CPU waits, allow the ring to be chosen on the basis of the subsequent operation following a submission of batch. (However, since batches are likely to be submitted in the middle of a draw, then the likelihood is for ddx to remain on one ring until forced to switch for an operation or idle, which is the same situation as before and so the difference is miniscule.) Signed-off-by: Chris Wilson commit 295a22d2709b2442b5254968437f897dac22a0ec Author: Chris Wilson Date: Sat Jan 14 14:12:25 2012 +0000 sna: Ensure that the batch mode is always declared before emitting dwords Initially, the batch->mode was only set upon an actual mode switch, batch submission would not reset the mode. However, to facilitate fast ring switching with semaphores, reseting the mode upon batch submission is desired which means that if we submit the batch in the middle of an operation we must redeclare its mode before continuing. Signed-off-by: Chris Wilson commit 0d2a50772200d868d094f90dc1d30c1b4d7930ba Author: Chris Wilson Date: Sat Jan 14 11:03:55 2012 +0000 sna/glyphs: Cache the glyph image on the fallback path as well The glyph cache grew to accommodate the fallback pixman image for mask generation, and is equally applicable along the full fallback path. Signed-off-by: Chris Wilson commit f3e0ba4f65db2b85f89aa3868d153434bc41c811 Author: Chris Wilson Date: Sat Jan 14 10:17:27 2012 +0000 sna/gen5: Disable render glyphs_to_dst Processing more than a single rectangle using the CA path on ILK is extremely hit-or-miss, often resulting in the absence of the second primitive (ie. the glyphs are cleared but not added.) This is reminiscent of the complete breakage of the BRW shaders, none of which can handle more than a single rectangle. Signed-off-by: Chris Wilson commit fb92818ba4dc81ce62d58a87b5af6cb1e3a96708 Author: Chris Wilson Date: Sat Jan 14 01:48:11 2012 +0000 sna: Pass render operation to flush and avoid the implicit flush-on-batch-end Signed-off-by: Chris Wilson commit a62429a1f79b8fa4a5ddaf61b2bc80fc8dbe576c Author: Chris Wilson Date: Fri Jan 13 19:00:01 2012 +0000 sna: Upload continuation vertices into mmapped buffers In the common case, we expect a very small number of vertices which will fit into the batch along with the commands. However, in full flow we overflow the on-stack buffer and likely several continuation buffers. Streaming those straight into the GTT seems like a good idea, with the usual caveats over aperture pressure. (Since these are linear we could use snoopable bo for the architectures that support such for vertex buffers and if we had kernel support.) Signed-off-by: Chris Wilson commit 24df8ab9742f771cfeb6d30bd8a61a17a9e22ca7 Author: Chris Wilson Date: Fri Jan 13 16:15:21 2012 +0000 sna: Reverse the chronological sort order of inactive vma entries The goal is to reuse the most recently bound GTT mapping in the hope that is still mappable at the time of reuse. Signed-off-by: Chris Wilson commit 2f26bbe3dd55bfe26b6f93fc1f9e9813fc11f7c4 Author: Chris Wilson Date: Fri Jan 13 15:59:51 2012 +0000 sna: Remove the short-circuiting of all-damage in move_to_cpu To allow a replacement of the complete pixmap to be performed in place. Signed-off-by: Chris Wilson commit c81dba18e6a308d716a11b90da6e53ca7f5f8a02 Author: Chris Wilson Date: Fri Jan 13 14:21:13 2012 +0000 sna: Hint whether we prefer to use the GPU for a pixmap This includes the condition where the pixmap is too large, as well as being too small, to be allocatable on the GPU. It is only a hint set during creation, and may be overridden if required. This fixes the regression in ocitysmap which decided to render glyphs into a GPU mask for a destination that does not fit into the aperture. Signed-off-by: Chris Wilson commit 2bd942d55314426d7f429d7ccc0b62a622a79009 Author: Chris Wilson Date: Fri Jan 13 14:08:34 2012 +0000 sna/trapezoids: Quieten the debugging of the gory details of the rasteriser Hide the noise under another level of debugging so that hopefully the reason why it chose a particular path become clear. Signed-off-by: Chris Wilson commit 5dbcfc2ee3af64846298dbcb20db27c93b3d57f2 Author: Chris Wilson Date: Fri Jan 13 13:52:10 2012 +0000 sna: Be more lenient in not forcing to the GPU if the sources have CPU damage Signed-off-by: Chris Wilson commit 20ff4a1d73cc10e1f53050b19b8799ccbb5c7d04 Author: Chris Wilson Date: Fri Jan 13 13:30:13 2012 +0000 sna: Use top_srcdir to detect .git rather than top_builddir For srcdir != builddir builds, we need to be searching the source tree for the git id. Signed-off-by: Chris Wilson commit a4d5d725996b94e314ae7697c7a597ed2f60e8cd Author: Chris Wilson Date: Fri Jan 13 10:41:44 2012 +0000 sna: Experiment with GTT mmapped upload buffers In a few places, we can stream the source into the GTT and so upload in place through the WC mapping. Notably, in many other places we want to rasterise on a partial in cacheable memory. So we need to notify the backend of the intended usage for the buffer and when we think it is appropriate we can allocate a GTT mapped pointer for zero-copy upload. The biggest improvement tends to be in the PutComposite style of microbenchmark, yet throughput for trapezoid masks seems to suffer (e.g. swfdec-giant-steps on i3 and gen2 in general). As expected, the culprit of the regression is the aperture pressure causing eviction stalls, which the pwrite paths sidesteps by doing a cached copy when there is no GTT space. This could be alleviated with an is-mappable ioctl predicting when use of the buffer would block and so falling back in those cases to pwrite. However, I suspect that this will improve dispatch latency in the common idle case for which I have no good metric. Signed-off-by: Chris Wilson commit 252f3818250ecf4776f20afa8111d7f1f6f29c18 Author: Chris Wilson Date: Fri Jan 13 11:30:32 2012 +0000 sna: Relinquish the GTT mmap on inactive buffers if moved out of the aperture Signed-off-by: Chris Wilson commit 9c73dd91e90810116c215060bac434c5e8174f5f Author: Chris Wilson Date: Sat Jan 14 17:00:01 2012 +0000 Include to repair build intel_module.c:41:48: error: missing binary operator before token "(" commit b213f6e876f29cbcaccffeeb733726be4c91529e Author: Stefan Dirsch Date: Sat Jan 14 05:43:33 2012 +0100 Make driver backwards compatible for server 1.6.x. Signed-off-by: Stefan Dirsch commit 94217a4dd908f1368dfdef90797ce74a081663fb Author: Chris Wilson Date: Fri Jan 13 10:03:28 2012 +0000 sna: Decouple dirty pixmaps from list if we fail to upload them Rather than iterate endlessly trying to upload the same pixmap when failing to flush dirty CPU damage, try again on the next flush. Signed-off-by: Chris Wilson commit 65ef369c733ab45945a7d5fe4b76fe5c7167b51b Author: Chris Wilson Date: Fri Jan 13 09:58:40 2012 +0000 sna: Decouple from CPU dirty list after removing all CPU damage In the paths where we discard CPU damage, we also need to remove it from the dirty list so that we do not iterate over it during flush. Signed-off-by: Chris Wilson commit 0845fcef9e664eb100b55ad1cc20743844deee58 Author: Chris Wilson Date: Thu Jan 12 23:45:03 2012 +0000 sna: Correct iteration counter for stippled blits ==7215== Invalid read of size 2 ==7215== at 0x51A72F3: sna_poly_fill_rect_stippled_8x8_blt (sna_accel.c:7340) ==7215== by 0x51A9CDF: sna_poly_fill_rect_stippled_blt (sna_accel.c:8163) ==7215== by 0x51A3878: sna_poly_segment (sna_accel.c:6090) ==7215== by 0x216C02: damagePolySegment (damage.c:1096) ==7215== by 0x13F6E8: ProcPolySegment (dispatch.c:1771) ==7215== by 0x1436B4: Dispatch (dispatch.c:437) ==7215== by 0x131279: main (main.c:287) ==7215== Address 0x6f851e8 is 0 bytes after a block of size 32 alloc'd ==7215== at 0x4825DEC: malloc (vg_replace_malloc.c:261) ==7215== by 0x51A3558: sna_poly_segment (sna_accel.c:6049) ==7215== by 0x216C02: damagePolySegment (damage.c:1096) ==7215== by 0x13F6E8: ProcPolySegment (dispatch.c:1771) ==7215== by 0x1436B4: Dispatch (dispatch.c:437) ==7215== by 0x131279: main (main.c:287) An example being the stippled outline in gimp, the yellow marching ants, would randomly walk over the entire image. Signed-off-by: Chris Wilson commit 5c2c6474efd4dbc5a0fc9c68ef4e5b5e5cfee415 Author: Chris Wilson Date: Thu Jan 12 23:11:05 2012 +0000 sna/dri: Hook up a compile option to switch colour buffers to Y-tiling Signed-off-by: Chris Wilson commit 59b79e5952fdc40f8c50f74a72a19363cb426140 Author: Chris Wilson Date: Thu Jan 12 21:05:39 2012 +0000 sna: Reorder composite-done to destroy mask bo before source bo Just in the unlikely event that we hit the delete-partial-upload path which prefers destroying the last bo first. Signed-off-by: Chris Wilson commit 983b755313df8a0d256c59c32ec4106e35f237aa Author: Chris Wilson Date: Thu Jan 12 21:05:16 2012 +0000 sna/damage: Fix union of extents with dirty damage but no region By failing to account for certain paths which would create a damage elt without fully initialisating the damage region (only the damage extents), we would later overwrite the damage extents with only the extents for this operation (rather than the union of this operation with the current damage). This fixes a regression from 098592ca5d, (sna: Remove the independent tracking of elts from boxes). Include the associated damage migration debugging code of the callers. Signed-off-by: Chris Wilson commit 8d2f1eefe142b65db7d8821ba0f80fdb0902b2d5 Author: Chris Wilson Date: Thu Jan 12 21:08:41 2012 +0000 sna: Pass a hint that we may like to perform the fallback in place If we do not read back from the destination, we may prefer to utilize a GTT mapping and perform the fallback inplace. For the rare event that we wish to fallback and do not already have a shadow... Signed-off-by: Chris Wilson commit 48ab72754d0069a3247c5fee8c353a6b593eaed9 Author: Chris Wilson Date: Thu Jan 12 21:07:15 2012 +0000 sna: Use the GPU bo if it is all damaged By marking the scratch upload pixmap as damaged in both domains, we confused the texture upload path and made it upload the pixmap a second time. If either bo is all-damaged, use it! Signed-off-by: Chris Wilson commit 20a4d7181983c7fed289844be49ccaf3f56965c6 Author: Chris Wilson Date: Thu Jan 12 21:05:51 2012 +0000 sna: Dump batch contents for debugging before modification We need to dump the batch contents before the maps are made by the construction of the batch itself. Signed-off-by: Chris Wilson commit 7932a2a259069bb7d19af8566f7b7704f6c2eade Author: Chris Wilson Date: Thu Jan 12 14:06:22 2012 +0000 sna: Update for removal of backwards compatible miWideDash miWideDash() no longer calls miZeroLineDash() when called with gc->lineWidth==0, we need to do so ourselves. Signed-off-by: Chris Wilson commit b7cefddd467bfd0c100219969712a0133475004c Author: Chris Wilson Date: Thu Jan 12 13:17:43 2012 +0000 sna: Re-enable min-alignment workaround on pre-SNB hw Confirmed as still being required for both gen3 and gen4. One day I will get single-stream mode working, just not today apparently. Signed-off-by: Chris Wilson commit 978e1aeceab3c1a524f7d7a070fe04f37530c8d9 Author: Chris Wilson Date: Thu Jan 12 11:38:56 2012 +0000 sna: Only shrink a partial buffer if it is no longer used. The condition on being able to shrink a buffer is more severe than just whether we are reading from the buffer, but also we cannot swap the handles if the existing handle remains exposed via a proxy. Signed-off-by: Chris Wilson commit d3169154d18600e0d41db5f833fad52970e17b55 Author: Chris Wilson Date: Thu Jan 12 11:36:05 2012 +0000 sna: Improve a DBG message Signed-off-by: Chris Wilson commit 2a22990968aebd53a7bfeaf19e58ee609b980e5f Author: Chris Wilson Date: Thu Jan 12 11:22:06 2012 +0000 sna: Prevent 60Hz wakeups if the client stops in mid-render Only continue to wake up if the scanout remains active. Signed-off-by: Chris Wilson commit 1c0e9916caef6594511009c8cb79071ac0d82090 Author: Chris Wilson Date: Thu Jan 12 11:12:56 2012 +0000 sna: Align the partial buffer contents to cachelines To enable Daniel's faster pwrite paths. Only one step removed from using whole page alignment... Signed-off-by: Chris Wilson commit 1e4080318f58fd51f37792eefd367e2e4c813ea3 Author: Chris Wilson Date: Thu Jan 12 10:47:19 2012 +0000 sna: Replace the open-coded bubble sort of the partial list Signed-off-by: Chris Wilson commit 7290ced5791f9860b00901fa9a4545ab5a067fae Author: Chris Wilson Date: Thu Jan 12 02:59:43 2012 +0000 sna/video: Fix for changes in damage api We can avoid both calls to modify the damage with one simple check. Signed-off-by: Chris Wilson commit 87e6dcb3b06c274ecba5de56d4e86fec73e46dbb Author: Chris Wilson Date: Thu Jan 12 02:09:22 2012 +0000 sna: Don't call RegionIntersect for the trivial PutImage Signed-off-by: Chris Wilson commit 1bd6665093574372248a0743e5630e68aac653ed Author: Chris Wilson Date: Thu Jan 12 01:37:08 2012 +0000 sna: Disable the min alignment workaround Allow all generations to use the minimum alignment of 4 bytes again as it appears to be working for me... Or at least what remains broken seems to be irrespective of this alignment. Signed-off-by: Chris Wilson commit 112b8959261712aaa82f92af0aca4b97fa7c7f03 Author: Chris Wilson Date: Thu Jan 12 01:38:18 2012 +0000 sna: Prevent shrinking a partial buffer stolen for a read If we reuse a partial buffer for a read, we cannot shrink it during upload to the device as we do not track how many bytes we actually need for the read operation. Signed-off-by: Chris Wilson commit b09ae4c20313fea7af11f77cd673039635d4dcc7 Author: Chris Wilson Date: Thu Jan 12 00:58:08 2012 +0000 sna: Don't drop expired partial bo immediately, wait until dispatch As the partial bo may be coupled into the execlist, we may as well hang onto the memory to service the next partial buffer request until it expires in the next dispatch. Signed-off-by: Chris Wilson commit a3c42565a8f557b2e7f7ff7bfa45b13b606f2968 Author: Chris Wilson Date: Wed Jan 11 16:39:45 2012 +0000 sna: Store damage-all in the low bit of the damage pointer Avoid the function call overhead by inspecting the low bit to see if it is all-damaged already. Signed-off-by: Chris Wilson commit c64a9d0683e047a7eb041df78db746f6dd387b5e Author: Chris Wilson Date: Wed Jan 11 19:51:42 2012 +0000 sna: Choose a stride for the indirect replacement Don't blithely assume that the incoming bytes are appropriately aligned for the destination buffer. Indeed we may be replacing the destination bo with the shadow bytes out of another,larger, pixmap, in which case we do need to create a stride that is appropriate for the upload an perform the 2D copy. Signed-off-by: Chris Wilson commit b82851e74d5010ee08938ee42fa44c29fed633b1 Author: Chris Wilson Date: Wed Jan 11 15:42:00 2012 +0000 sna: Mark upload pixmaps as being wholly GPU damaged So that subsequent code resists performing CPU operations with them (after they have been populated.) Marking both sides as wholly damaged breaks the rules, but should work out so long as we check whether we can perform the operation within the target damage first. Signed-off-by: Chris Wilson commit 2a5ab05f1690484c230e8f876a3f7aefb371af71 Author: Chris Wilson Date: Wed Jan 11 15:24:28 2012 +0000 sna: Use a minimum alignment of 64 We should be able to reduce this by disabling dual-stream mode of the GPU (which we want to achieve any way for 2D performance). Artefacts in small uploads demonstrate that we fail to do. References: https://bugs.freedesktop.org/show_bug.cgi?id=44150 Signed-off-by: Chris Wilson commit e94807759eb6cfb10bd2d372fa71cc64a730bc7c Author: Chris Wilson Date: Wed Jan 11 12:13:18 2012 +0000 sna/gen6: Special case spans with no transform As the no transform is a special case of affine, we were attempting to deference the NULL transform in order to determine if it was a simple no-rotation matrix. As the operation is extremely simple, add a special case vertex program to speed it up. Signed-off-by: Chris Wilson commit 0a5313900ec9a7c499eb5051f3a5f078a9b0bbde Author: Chris Wilson Date: Wed Jan 11 12:09:13 2012 +0000 sna: Explicitly retire the bo following a serialisation point This is to keep the sanity checks in order, but conceptually should be useful as well. Signed-off-by: Chris Wilson commit 2add5991a7d186d8271cb79ae4f62dad7b4ca243 Author: Chris Wilson Date: Wed Jan 11 11:33:19 2012 +0000 sna: Mark the bo as no longer in the GPU domain after clearing needs_flush Signed-off-by: Chris Wilson commit fec70985719eec75b7fa799bddc3115fb329ab86 Author: Chris Wilson Date: Wed Jan 11 10:57:57 2012 +0000 sna: Add assertions to track requests Signed-off-by: Chris Wilson commit a93c93be76f6d5d2b481971349aabd15f282c3e8 Author: Chris Wilson Date: Wed Jan 11 00:05:20 2012 +0000 sna/gen6: Add a vertex program for a simple (affine, no rotation) spans I long for the day when this code is obsolete... Until then, this gives a nice boost in the fishtank. Signed-off-by: Chris Wilson commit 3cf5da1090ac777044912ec24619d349d1f6b521 Author: Chris Wilson Date: Tue Jan 10 23:32:29 2012 +0000 sna: Amalgamate small replacements into upload buffers Similar for the standard io paths, try to reuse an upload buffer for a small replacement pixmap. Signed-off-by: Chris Wilson commit f0e3f6b5bebf7471d3e3e84bd9b2d8469eb64093 Author: Chris Wilson Date: Tue Jan 10 22:12:41 2012 +0000 sna: Check needs-flush status immediately upon destroy Signed-off-by: Chris Wilson commit b4ae6dbaedd774f250b610b1e7323897c4abe23a Author: Chris Wilson Date: Tue Jan 10 21:25:28 2012 +0000 sna: Align the small upload buffers to 2 texels, and the pitch to dwords References: https://bugs.freedesktop.org/show_bug.cgi?id=44150 Signed-off-by: Chris Wilson commit 46f6c6917effbc43305367581cc1bbaa962c0251 Author: Chris Wilson Date: Tue Jan 10 20:34:09 2012 +0000 sna: A partial read buffer is allowed to be in the GPU domain As we can create the read buffer from an active cached bo, it may already be in the GPU domain by the time we first finish it, so fix the broken assertion. Signed-off-by: Chris Wilson commit 3c26055639f789b211f7db982e65ec0b86b23d4a Author: Chris Wilson Date: Tue Jan 10 20:06:51 2012 +0000 sna: Shrink the partial upload buffers before compacting the batch So that the relocation entries point into the contiguous surface/batch and can be trivially fixed up. Signed-off-by: Chris Wilson commit 7b077a4d3d549cde79d5b876289203788294493c Author: Chris Wilson Date: Tue Jan 10 18:40:53 2012 +0000 sna: Make the check for a single fill-rectangle clearer before modifying damage Signed-off-by: Chris Wilson commit ca2a07adc45273dc1abeb4b3ba7f88461aaf9c00 Author: Chris Wilson Date: Tue Jan 10 17:13:38 2012 +0000 sna: Release the stale GTT mapping after recreating the bo with new tiling Signed-off-by: Chris Wilson commit 8dd913fd3a093b54fc024fea2b31a3db2518db8d Author: Chris Wilson Date: Tue Jan 10 10:38:43 2012 +0000 sna: Add reminder about possible future tiling optimisations Signed-off-by: Chris Wilson commit 21948578d0d87f38447ef1ae44fada406949bca8 Author: Chris Wilson Date: Tue Jan 10 03:19:16 2012 +0000 sna: Disable the inline xRGB to ARGB upload conversion As we have to upload the dirty data anyway, setting the alpha-channel to 0xff should be free. Not so for firefox-asteroids on Atom at least. Signed-off-by: Chris Wilson commit 87f73b043426c47efa7670fb65bdcc7dfcf71fc3 Author: Chris Wilson Date: Tue Jan 10 01:42:37 2012 +0000 sna/gen[23]: Tile render fill to oversized bo If we are forced to perform a render operation to a bo too large to fit in the pipeline, copy to an intermediate and split the operation into tiles rather than fallback. Signed-off-by: Chris Wilson commit 2ccb31c5a4120efb940168c5e52154add01b0586 Author: Chris Wilson Date: Tue Jan 10 01:41:09 2012 +0000 sna: Shrink upoads buffers If we do not fill the whole upload buffer, we may be able to reuse a smaller buffer that is currently bound in the GTT. Ideally, this will keep our RSS trim. Signed-off-by: Chris Wilson commit 572cc76be5c78102166f2b3640640e21acdfa129 Author: Chris Wilson Date: Tue Jan 10 09:13:34 2012 +0000 sna: Destroy the counter damage after a reduction-to-all If, for instance, we reduce the GPU damage to all we know that there can be no CPU damage even though it may still have a region with a list of subtractions. Take advantage of this knowledge and cheaply discard that damage without having to evaluate it. This should prevent a paranoid assertion that there is no cpu damage when discarding the CPU bo for an active pixmap. Signed-off-by: Chris Wilson commit 4a255e13164440b797e5ac11dcbf72f0827a3094 Author: Chris Wilson Date: Tue Jan 10 16:34:23 2012 +0000 sna: Replace the free-inactive-gpu-bo with the generic code The function was semantically equivalent to moving the pixmap to the CPU for writing, so replace it with a call to the generic function. Signed-off-by: Chris Wilson commit c1d403266a611a68081690d19f6debb8e343095f Author: Chris Wilson Date: Tue Jan 10 00:41:05 2012 +0000 sna: Allow for xRGB uploads to replace their target bo Signed-off-by: Chris Wilson commit 406776cd955c3c384d4a537300e21eebe4413666 Author: Chris Wilson Date: Tue Jan 10 00:25:14 2012 +0000 sna: Rearrange buffer allocation to prefer llc bo over vmaps If we can create snoopable bo, we prefer to use those as creating a vmap forces a new bo creation increasing GTT pressure. Signed-off-by: Chris Wilson commit b76865fa3deff2f44a1158914a124b9c81d67eca Author: Chris Wilson Date: Mon Jan 9 22:11:21 2012 +0000 sna/gen2: Try to avoid creating a bo for solid colours As we try to use the diffuse/specular and only resort to using a texture operation for convenience in the rare case of a solid mask. Signed-off-by: Chris Wilson commit 981aae104a96b41db88cc381cc7592818f3e1298 Author: Chris Wilson Date: Mon Jan 9 22:02:40 2012 +0000 sna/gen2: Eliminate some switching between logic op and blend If the new mode can be done either using a logic op or with the blend unit, prefer the currently enabled unit. Signed-off-by: Chris Wilson commit d65b7f9cf46a48e3bfb37c0b75df55aa1e7bff41 Author: Chris Wilson Date: Mon Jan 9 21:58:03 2012 +0000 sna/blt: Rearrange to reduce a out-of-bounds copy to a clear If we asked to use the BLT, try to avoid trigging a context switch for a trivial case where we sample outside of a NONE source and so can reduce the operation to a clear. Signed-off-by: Chris Wilson commit 09e54c553680cbc2f3b4319cdab0f3d1af1574a8 Author: Chris Wilson Date: Mon Jan 9 14:04:41 2012 +0000 sna/gen2: Add poor-man's linear gradient support Convert the linear gradient to a texture ramp and compute the texture coordinates in the standard manner. Signed-off-by: Chris Wilson commit 6c70558ae7298db94724c931d88a730ef0151608 Author: Chris Wilson Date: Sun Jan 8 20:48:48 2012 +0000 sna: mark the cpu bo used for the upload buffer as in CPU domain For correctness we need to inform GEM of the change of domain for the buffer so that it knows to invalidate any caches when it is next used by the GPU. Signed-off-by: Chris Wilson commit 9ec31af02922bb016d0dfba07bc60cdca35b36f8 Author: Chris Wilson Date: Sun Jan 8 15:52:17 2012 +0000 sna/io: Combine small uploads into single writes For a small update, try and amalgamate the upload buffer. Signed-off-by: Chris Wilson commit 4db1bb3fd81b51e74b7f3e90078627d9d96fbefe Author: Chris Wilson Date: Sun Jan 8 14:20:40 2012 +0000 Removed deprecated xf86PciInfo.h includes The driver should and does provide its own PCI-IDs. Signed-off-by: Chris Wilson commit 54232d1a5da51d0f68fe099436bb0a1a2e249954 Author: Chris Wilson Date: Sun Jan 8 15:22:58 2012 +0000 sna: Add ricer stripes to memcpy_xor Signed-off-by: Chris Wilson commit c037b4f542a7b21cbaecedec259da3589db10039 Author: Chris Wilson Date: Sun Jan 8 08:58:41 2012 +0000 sna: Tune cache size for cpu bo cache This helps SNB on cairo-traces that utilize lots of temporary uploads (rasterised sources and masks for instance), but comes at a cost of regressing others... In order to counter the regression from increasing the GTT cache size, the CPU/GTT vma cache are split and accounted separately. Signed-off-by: Chris Wilson commit 26042b2660d87044e1920a1267d9984c00c9566a Author: Chris Wilson Date: Sun Jan 8 11:45:11 2012 +0000 sna: Bubble sort the partial buffer list back into order after trimming padding After reducing the used size in the partial buffer, we need to resort the list to maintain the list in decreasing amount of available space. Signed-off-by: Chris Wilson commit 3f7ea44bf19a03ee81b683885c9c2416092254a3 Author: Chris Wilson Date: Sun Jan 8 02:27:28 2012 +0000 sna/gen[67]: Hook into the clear operation for glyph masks Allow SandyBridge to specialise its clear routine to reduce the number of ring switches. It may be interesting to specialise the clear routines even further and use the special render clear commands... Signed-off-by: Chris Wilson commit 803ac5c6b992cb5448c67b11345b87a5d2b9c60d Author: Chris Wilson Date: Sun Jan 8 01:38:09 2012 +0000 sna/trapezoids: Don't risk stalling for inplace SRC trapezoids Optimistically we would replace the GPU damage with the new set of trapezoids. However, if any partial damage remains then the next operation which is often to composite another layer of trapezoids (for complex clipmasks) using IN will then stall. This fixes a regression in firefox-fishbowl (and lesser regressions throughout the cairo-traces). Signed-off-by: Chris Wilson commit 0229841c0d705138d6317a22ba80f4fa10d26acc Author: Chris Wilson Date: Sun Jan 8 01:46:24 2012 +0000 sna: Do not upload an untiled GPU bo if we already have a CPU bo Continuing the tuning for sna_copy_boxes. Signed-off-by: Chris Wilson commit 42eb9b7c4b55b620d9652ad3a54c8ab0a76fd7c2 Author: Chris Wilson Date: Sun Jan 8 00:21:26 2012 +0000 sna: Trim usage of vmapping The first, and likely only, goal is to support SHMPixmap efficiently (and without compromising SHMImage!) which we want to preserve as vmaps and never create a GPU bo. For all other use cases, we will want to create snoopable CPU bo ala the LLC buffers on SandyBridge. Signed-off-by: Chris Wilson commit e52f0204939924181480feec6d1511259169c8f1 Author: Chris Wilson Date: Sun Jan 8 00:20:11 2012 +0000 sna: Do not move-to-gpu for sna_copy_boxes if we have a cpu bo We trade-off the extra copy in the hope that as we haven't used the GPU bo before then, we won't need it again. Signed-off-by: Chris Wilson commit c05e90aa99c093431a74a94ceba27d7dce49d05b Author: Chris Wilson Date: Sat Jan 7 18:23:43 2012 +0000 sna: Missing chunks from last commit And update the check for reusing the blit! Signed-off-by: Chris Wilson commit 292aebfcdc042b08d3a5a21f839e6aec1175359e Author: Chris Wilson Date: Sat Jan 7 18:09:47 2012 +0000 sna: Prevent reuse of scanlines after the buffer is destroyed Once the buffer is destroyed, it may be reallocated with a new pitch. We could track handle and pitch, but it is easier to simply restart the blit after the buffer is freed. References: https://bugs.freedesktop.org/show_bug.cgi?id=44277 References: https://bugs.freedesktop.org/show_bug.cgi?id=44555 Signed-off-by: Chris Wilson commit d7d07d1df3ffc889ccc32c2ed44ce374e37f851d Author: Chris Wilson Date: Sat Jan 7 10:37:01 2012 +0000 sna: Pad upload buffers to ensure there are an even number of rows One restriction common to all generations is that samplers access pairs of rows and so we need to pad the buffer to accommodate access to that second row. Do so unconditionally along paths that may be used by the render pipeline. Signed-off-by: Chris Wilson commit e2ad0f6272c776939499c6202fb60bdcc2e955fa Author: Chris Wilson Date: Fri Jan 6 17:11:00 2012 +0000 sna/blt: Amalgamate many PolyFillRect of single boxes Signed-off-by: Chris Wilson commit c085de905c30cb5bae6b339841badf1e105c6ee6 Author: Chris Wilson Date: Fri Jan 6 15:47:47 2012 +0000 sna: Also mark a bo created by force-to-gpu as being all-damaged Similar to the action taken into move-to-gpu so that we forgo the overhead of damage tracking when the initial act of creation is on the render paths. Signed-off-by: Chris Wilson commit 9f1935bb4e894264053d94e53c99d5ad607700fb Author: Chris Wilson Date: Fri Jan 6 15:26:11 2012 +0000 sna: Support performing alpha-fixup on the source By inlining the swizzling of the alpha-channel we can support BLT copies from an alpha-less pixmap to an alpha-destination. Signed-off-by: Chris Wilson commit 141001df6c9c3485c500ed531a214c09b46c1d3b Author: Chris Wilson Date: Fri Jan 6 13:35:27 2012 +0000 sna: always skip active search when requested for find an inactive bo References: https://bugs.freedesktop.org/show_bug.cgi?id=44504 Signed-off-by: Chris Wilson commit eac0d9652b2399f8c36ba0288db6fe347ed78dc9 Author: Chris Wilson Date: Fri Jan 6 12:39:49 2012 +0000 sna: Optimise sna_poly_segment() for the frequent no-op case Strange as it may seem... But the principle of doing less work with greater locality should help everywhere, just not as noticeable when real work is performed. Signed-off-by: Chris Wilson commit a3699fff5ada85e4dea739aade25ebbb728e18f4 Author: Chris Wilson Date: Fri Jan 6 12:15:46 2012 +0000 sna: Only force a pipeline flush for a change of destination, not sources Signed-off-by: Chris Wilson commit 2fefee6015ed2df52c8513ae180ca83c01ff83c0 Author: Chris Wilson Date: Fri Jan 6 12:10:27 2012 +0000 sna/gen6: Reuse current no-blending setup for PictOpClear Signed-off-by: Chris Wilson commit 800ca0b4d1194544fe2461f91cbdc632c4d3dd7a Author: Chris Wilson Date: Fri Jan 6 12:00:43 2012 +0000 sna/gen6: Tidy emission of CC state (blending) Signed-off-by: Chris Wilson commit b9c9e9970cdb542173e3ed0da2bef614abedd3f7 Author: Chris Wilson Date: Fri Jan 6 11:38:31 2012 +0000 sna/trapezoids: Add debug markers for move-to-cpu Signed-off-by: Chris Wilson commit 2841c5fee79c42bca3e098ec620755d341b6888f Author: Chris Wilson Date: Fri Jan 6 11:32:54 2012 +0000 sna: fast path move-to-cpu of an all-damaged CPU bo When the bo is already completely damaged on the CPU, all we need to do is to sync with the CPU bo. Signed-off-by: Chris Wilson commit 07f1724ba5759772043fc4c5d1163ddd640004c2 Author: Chris Wilson Date: Fri Jan 6 11:00:38 2012 +0000 sna: Pass the real destination pixmap to sna_write_boxes() Faking it for the render upload simply isn't good enough, since we need the correct drawrect. Signed-off-by: Chris Wilson commit 5c2e9a98715eeddd8dfdff9fe3871bde659484cc Author: Chris Wilson Date: Fri Jan 6 02:26:39 2012 +0000 sna/trapezoids: Use a slightly faster miTrapezoidBounds() Signed-off-by: Chris Wilson commit 75f61678249018c84872c61b4609ba9d3d0d73dc Author: Chris Wilson Date: Fri Jan 6 01:38:42 2012 +0000 sna: Only need to remove the pixmap from the dirty CPU list when clearing damage Signed-off-by: Chris Wilson commit 86cc2cd3343c16b12e35e2048c113481f49196cf Author: Chris Wilson Date: Fri Jan 6 01:31:15 2012 +0000 sna: The condition for automatically using the GPU source is no CPU damage With no CPU damage to upload, we know that there is no reason not to use the GPU bo. Signed-off-by: Chris Wilson commit 69c556c310da544ddb66c2441f5d06dca29fb9f3 Author: Chris Wilson Date: Fri Jan 6 00:21:29 2012 +0000 sna: Rearrange use-cpu-bo to avoid reducing GPU damage If we spot that the region is wholly contained within the CPU damage initially, we can conclude that is not in the GPU damage without reduction. Signed-off-by: Chris Wilson commit 34c4dbe825a45315b05f0bd49539cbcf5de276f1 Author: Chris Wilson Date: Thu Jan 5 23:18:37 2012 +0000 sna/gen6: Only force BLT if the src and dst overlaps for self-copy Signed-off-by: Chris Wilson commit fc79af7a4abf1f0ceccacf23e8467ee2872eec09 Author: Chris Wilson Date: Thu Jan 5 23:06:17 2012 +0000 sna: Only force a batch continuation if the scanout is written to Signed-off-by: Chris Wilson commit c16c16d2553aa9a8bf9b0b0b367715b59020a357 Author: Chris Wilson Date: Thu Jan 5 22:51:55 2012 +0000 sna/gen6: Enable reuse of source Picture Check if the source and mask are identical pictures and just copy the source channel to the mask in that case. Signed-off-by: Chris Wilson commit f642b08870c6d7686368e053962075c153f76f86 Author: Chris Wilson Date: Thu Jan 5 22:26:09 2012 +0000 sna: Quickly handle the common case of using a CPU source pixmap For the common case (at least with llc bo) where we are immediately using an uploaded image from its linear buffer, check upfront before computing the sampled region for transfer to the GPU. Signed-off-by: Chris Wilson commit 8a9baa59e9d0c0761b82f23734514f6847a9bcc1 Author: Chris Wilson Date: Thu Jan 5 22:16:31 2012 +0000 sna: Quickly handle the common case of using a GPU source pixmap For the common case of glyphs, the pixmap is entirely on the GPU which can be quickly tested before performing the more complex transformations to determine how much pixel data we need to upload. Signed-off-by: Chris Wilson commit d4dad6e3723d70d78247f29f86079862af175266 Author: Chris Wilson Date: Thu Jan 5 21:39:24 2012 +0000 sna: Check that the GTT is no longer mapped before ensuring the shadow pointer Signed-off-by: Chris Wilson commit 352f0a3ffb8ae8503bb2fab779ea84f4e6d76317 Author: Chris Wilson Date: Thu Jan 5 21:33:26 2012 +0000 sna: Only mark the pixmap as GTT mapped if the private owns the pixels Signed-off-by: Chris Wilson commit 3061382f98c0b284c250c1a36f7a4fee22046664 Author: Chris Wilson Date: Thu Jan 5 21:05:52 2012 +0000 sna: Make sure that we do not attach to SHM pixmaps in ignorance Without xserver support for notification of when scratch pixmaps are reused, we simply cannot attach our privates to them lest we cause corruption with SHM pixmaps. This is a recent regression back unto an old, old xserver issue. Reported-by: Paul Neumann Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=44503 Signed-off-by: Chris Wilson commit f7431726338e72d7a989e838f48aecdab15ab366 Author: Chris Wilson Date: Thu Jan 5 20:26:30 2012 +0000 sna: Add more debugging to use-gpu-bo To aide debugging in conjunction with compositors and their crazy offsets. Signed-off-by: Chris Wilson commit 53c943b551b86402067ea96cd9fcca2848e84387 Author: Chris Wilson Date: Thu Jan 5 18:41:26 2012 +0000 sna: Check for usable pixmap before deference its private Beware the NULL pointer and early deference. Signed-off-by: Chris Wilson commit 307276d7da99ceb84c175e412452e5d86d615cf4 Author: Chris Wilson Date: Thu Jan 5 14:26:05 2012 +0000 sna: Enable using the render pipeline for uploads to Y-tiled pixmaps Signed-off-by: Chris Wilson commit f9451aaa39e1d77b08d2201f5c4a44ab834f2237 Author: Chris Wilson Date: Thu Jan 5 14:22:21 2012 +0000 sna: Mark the upload as having acquired the GTT mapping In order for the entire PutImage to be performed inplace, we need to maintain the tendency to keep doing inplace operations. This hint is provided by tracking whether or not the last operation used the GTT mapping. However, that hint was not being provided by zpixmap_blt. Signed-off-by: Chris Wilson commit ce93c00a33a395986ffc0d33aae85d2c23aabc5d Author: Chris Wilson Date: Thu Jan 5 14:20:55 2012 +0000 sna: Avoid operating inplace if we have pending CPU damage If the operation does not replace existing CPU damage, we are likely to want to reuse the pixmap again on the CPU, so avoid mixing CPU/GPU operations. Signed-off-by: Chris Wilson commit f5d02d90f3349f33341954fc55824d12a7829e6c Author: Chris Wilson Date: Thu Jan 5 12:56:21 2012 +0000 sna: Avoid switching rings for uploading data On systems that incur painful overhead for ring switches, it is usually better to create a large buffer and perform a sparse copy on the same ring than create a compact buffer and use the BLT. Signed-off-by: Chris Wilson commit a42111ff94911d5fad34e6dbac8aa4a11b2d8032 Author: Chris Wilson Date: Thu Jan 5 12:12:22 2012 +0000 sna/trapezoids: Try creating the trapezoids mask inplace if not using spans Signed-off-by: Chris Wilson commit 4c6ccbd46a1681c73382b79a0246f7d8cb097606 Author: Chris Wilson Date: Thu Jan 5 11:09:01 2012 +0000 sna: Throttle, throttle, throttle. This reverts 281425551bdab7eb38ae167a3205b14ae3599c49 as it was causing insufferable lag in firefox. Signed-off-by: Chris Wilson commit dbb495c3aeef34eab2aabfdfaafb81facdea4100 Author: Chris Wilson Date: Thu Jan 5 10:34:08 2012 +0000 sna: Attach to the pixmap for source counting when deferring uploads If we decide to defer the upload for this instance of the source pixmap, mark it so. Then if we do use it again we will upload it to a GPU bo and hopefully reuse those pixels. Signed-off-by: Chris Wilson commit 292097590ba0f81b7505aedb2a4fbd18021641a5 Author: Chris Wilson Date: Thu Jan 5 10:31:15 2012 +0000 sna: Try doing a cheap is-contained check before reducing damage If the pixmap is entirely within the current CPU damage, we can forgo reducing either the GPU or CPU damage when checking whether we need to upload dirty pixels for a source texture. Signed-off-by: Chris Wilson commit 5ddae8bf21067e7d2a031a6b9020c02ab977cfed Author: Chris Wilson Date: Thu Jan 5 10:27:37 2012 +0000 sna: Check the composite-rectangles will be on the GPU before forcibly attaching If we think that the operation is better performed on the CPU, avoid the overhead of manipulating our privates. Signed-off-by: Chris Wilson commit 6ffd3605640bacaf0d744075622463aacf78d797 Author: Chris Wilson Date: Thu Jan 5 10:25:12 2012 +0000 sna: Deferred attachment to a pixmap needs to mark the pixmap as entirely dirty As demonstrated with oversized glyphs and a chain of catastrophy, when attaching our private to a pixmap after creation we need to mark the entire CPU pixmap as dirty as we never tracked exactly which bits were dirtied. Signed-off-by: Chris Wilson commit a09ebe0b6cc66d08e52b57851d9b8a6a1f71df2b Author: Chris Wilson Date: Thu Jan 5 10:23:25 2012 +0000 sna: Immediately upload oversized glyphs Glyphs, even large ones, we suspect will be reused and so the deferred upload is counterproductive. Upload them immediately and mark them as special creatures for later debugging. Signed-off-by: Chris Wilson commit 797b27365dc09a3f84349a33cffbfdc67cba0baa Author: Chris Wilson Date: Thu Jan 5 10:30:47 2012 +0000 sna: Don't perform a deferred attachment for vmapping if not supported Signed-off-by: Chris Wilson commit 23e3959e1f603c3ba0bd4f01128290241cf93edf Author: Chris Wilson Date: Thu Jan 5 04:00:32 2012 +0000 sna: Use the created cpu-bo for upload and download As we explicitly create CPU bo when wanted, we no longer desire to spontaneously create vmaps for simply uploading to the GPU bo. Signed-off-by: Chris Wilson commit 4530b87e445aea30ccbe751699963118ad5672d4 Author: Chris Wilson Date: Thu Jan 5 00:01:27 2012 +0000 sna: Fix assertion as we may want to create active bo with cpu mappings Signed-off-by: Chris Wilson commit 4119e68fb157fc612bce5e9c5669112ce35b4ca1 Author: Chris Wilson Date: Wed Jan 4 19:35:30 2012 +0000 sna/damage: Fix reduction to copy the boxes correctly We need to be carefully to copy the boxes in a strict lifo order so as to avoid overwritting the last boxes when reusing the array allocations. Signed-off-by: Chris Wilson commit 71b0924b586d9a60397e92e941e3d0cfa636ee61 Author: Chris Wilson Date: Wed Jan 4 19:34:41 2012 +0000 sna: Mark tiny CPU pixmaps as completely dirty Avoid the overhead of tracking damage on small pixmaps when using CPU rasterisation; the extra cost of sending the whole pixmap compared to the damage is negligble should it ever be required on the GPU. Signed-off-by: Chris Wilson commit fc14e63256a9b756c7c77ffe73f2f13784396c5a Author: Chris Wilson Date: Wed Jan 4 19:33:30 2012 +0000 sna: Only request page size allocations for the replacement buffer A precondition on bo creation is that the size must be page aligned. Signed-off-by: Chris Wilson commit 89739b711f42c3dbed7d3f4e6da0cdd61a5205ae Author: Chris Wilson Date: Wed Jan 4 17:41:16 2012 +0000 sna: Expand small stipples into 8x8 patterns Signed-off-by: Chris Wilson commit 878fbfe509da0a25dff8515d6d3b38ca3826466e Author: Chris Wilson Date: Wed Jan 4 16:56:55 2012 +0000 sna: Align tiled stipple uploads with the stipple pixmap correctly Signed-off-by: Chris Wilson commit 28a222a66bb450c7f6d61aef16fe73332854c8ce Author: Chris Wilson Date: Wed Jan 4 12:48:25 2012 +0000 sna: Always prefer to use cacheable operations after the GPU is wedged As rasterisation will be performed upon the CPU we need to avoid the readbacks form uncached memory and so we should restrict ourselves to only create further damage within the CPU pixmap. Signed-off-by: Chris Wilson commit 99ff83a3bebf1aa25c44a87f2c344307d20bc062 Author: Chris Wilson Date: Wed Jan 4 12:35:12 2012 +0000 sna: Prevent creation of bo purely for GTT mapping when wedged Signed-off-by: Chris Wilson commit a7c35fa4c3155d10ad75ac33aefb03dafa010aba Author: Chris Wilson Date: Wed Jan 4 12:33:09 2012 +0000 sna: Only allow mappable pixmaps to be mapped If we did not allocate the pixel data, such as for wedged pixmaps or scratch buffers, then we cannot perform the pointer dance nor do we want to create the GPU buffer. Signed-off-by: Chris Wilson commit 2cbc97d28aae138566e49df05e16d274b533ea86 Author: Chris Wilson Date: Wed Jan 4 12:32:06 2012 +0000 sna: Force creation of ordinary pixmaps when wedged. If the pixmap was intended for scanout, then the GPU bo will be created upon attachment to the fb. Signed-off-by: Chris Wilson commit f6cc73ba66f34d190b7c17421a62aa1256698431 Author: Chris Wilson Date: Wed Jan 4 12:56:31 2012 +0000 sna: The batch is counted in dwords, not bytes! Yikes, I choose the wrong units for the max_batch_size. Signed-off-by: Chris Wilson commit 227fbb90c4884bbc58c6c0cfff9663ec9ca54171 Author: Chris Wilson Date: Wed Jan 4 11:47:32 2012 +0000 sna: Carefully free the freed_pixmap upon exit As the contents of the pixmap are now rubbish, we need to manually destroy it rather than pass it to the normal sna_pixmap_destroy() routines. Signed-off-by: Chris Wilson commit 0ed758cd2176ee4f34e03d05d05130d52d75e577 Author: Chris Wilson Date: Wed Jan 4 12:18:20 2012 +0000 sna: Limit batch to a single page on 865g Verified on real hw, this undocumented (at least in the bspec before me) bug truly exists. Signed-off-by: Chris Wilson commit 3449f1cbe184a618e662d2a662167f05362b82e5 Author: Chris Wilson Date: Wed Jan 4 02:38:16 2012 +0000 sna: Guard against NULL damage deref The fast version of damage checking assumes that the damage has already been determined to be non-NULL, so make sure it is. Signed-off-by: Chris Wilson commit 04a6260016fb5ebefc1056ed6acf5001ec535d5c Author: Chris Wilson Date: Wed Jan 4 01:14:19 2012 +0000 sna: Fix typo during partial list deletion And keep the asserts that lead to its discovery. Signed-off-by: Chris Wilson commit f1dc1eadd84097fc691e85c636535ceeeb601a18 Author: Chris Wilson Date: Wed Jan 4 00:09:30 2012 +0000 sna/gen3: Remove incorrect premultiplication of solid component-alpha mask Signed-off-by: Chris Wilson commit 8cb9b8d7d7a1eb62eb3b20e6a50d3f1c9bde40c1 Author: Chris Wilson Date: Tue Jan 3 23:13:24 2012 +0000 sna: Discard mask and source for PictOpClear Signed-off-by: Chris Wilson commit feb75c55d6105a9ecc5f5ec7830276d1654918b9 Author: Zhigang Gong Date: Sat Dec 31 21:18:15 2011 +0800 uxa/glamor: Route some missing drawing function to glamor We have to route all the drawing function to glamor first, when glamor is enabled. This adds a few more functions that were previously just falling back to swrast and passes them to glamor instead. Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit f4ea732d7181d4ef7f97fb2585e87cdcaef99d46 Author: Zhigang Gong Date: Tue Dec 27 17:09:17 2011 +0800 uxa/glamor: Let glamor do the GC validation If we are using GLAMOR, then a tile pixmap or stipple pixmap may be pure glamor pixmap and thus UXA will not know how to render to them, and we need to let glamor do the validation. Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit e8aa9cedbfca09e22d46ec8882b9bebb3eb83022 Author: Zhigang Gong Date: Tue Dec 27 17:09:16 2011 +0800 uxa/glamor: Remove extraneous flush When glamor is enabled, a pixmap will not be accessed by UXA's accelerated functions. Only unaccelerated functions may access those pixmaps, and before each unaccelerated rendering, it calls uxa_prepare_access which will do a glFlush. Combined with a flush before sending to DRI clients, we no longer need to flush after every operation. Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit 4ca94ed91fab7a9de5eed1053b891e8420cdf43d Author: Zhigang Gong Date: Tue Dec 27 17:09:15 2011 +0800 uxa/glamor: Remove dead code. Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit 5576afc7d12ec0f028e416f4b4a076fc6d3c29d9 Author: Chris Wilson Date: Tue Jan 3 16:48:52 2012 +0000 sna: Inline checks for over-sized buffers during pixmap creation Make the tests for acceptable GPU pixmaps explicit and upfront. Signed-off-by: Chris Wilson commit b2dc17678fa3a548be61e1055d4d321dbd6494c3 Author: Chris Wilson Date: Tue Jan 3 15:50:21 2012 +0000 sna: Always search for an exact active match first from the bo-cache And accept second-best only if permitted by flags. Signed-off-by: Chris Wilson commit caf41ae23d263c7baed6b23a80693690c4dfe563 Author: Chris Wilson Date: Tue Jan 3 15:11:13 2012 +0000 sna: Force tiling on medium pitches if not reduced for other reasons Signed-off-by: Chris Wilson commit 1b688ca0cfc1e623f896e60a6a6abf724d57efb4 Author: Chris Wilson Date: Tue Jan 3 14:18:05 2012 +0000 sna: Delete the last buffer if destroyed without being used on the GPU Signed-off-by: Chris Wilson commit 4a68e7277104d7dccbdc4739eb5651902b9ccc20 Author: Chris Wilson Date: Tue Jan 3 12:19:24 2012 +0000 sna/gen3: Prefer to use the blitter for solid fills Signed-off-by: Chris Wilson commit 199e357dc693cfe232464e4faa82d9223a8325d6 Author: Chris Wilson Date: Tue Jan 3 12:16:48 2012 +0000 sna/trapezoids: Pass the sna pointer to the callees Avoid the repeated multiple indirect dereferences. Signed-off-by: Chris Wilson commit bd7aa1754dcf9d58d7deb92d99da9b78c91a0eb2 Author: Chris Wilson Date: Tue Jan 3 12:46:40 2012 +0000 sna: Avoid unnecessary call overhead for read-only move-to-cpu Signed-off-by: Chris Wilson commit 294c8791c435acf5933109fbcab85623a55b2c6c Author: Chris Wilson Date: Tue Jan 3 12:09:51 2012 +0000 sna: Use a cheaper no-reduction damage check for simply discarding further damage Signed-off-by: Chris Wilson commit 7ff42e9c83cdb17660a64901497704b38032af8c Author: Chris Wilson Date: Tue Jan 3 19:15:07 2012 +0000 sna/damage: Mark the box as packed so that the embedded_box is aligned correctly valgrind was complaining about an overlapping memcpy on a 64-bit platform as gcc padded the sna_damage_box to 28 bytes... Signed-off-by: Chris Wilson commit 4652a20de5fee0d5d1a8baadc4a93b6a93525564 Author: Chris Wilson Date: Tue Jan 3 21:20:09 2012 +0000 sna/gen3: Check for a non-NULL bo before incrementing its ref in reuse_source() For gen3, we may reduce a source into a constant operator and so dispense with keeping a bo. When duplicated into the mask channel, we then need to be careful not to dereference the NULL pointer. Signed-off-by: Chris Wilson commit 777fcc9b11e8cbfe942aa6bf898749f552acb3cf Author: Chris Wilson Date: Sun Jan 1 16:03:42 2012 +0000 sna: Prefer not to force the creation of gpu bo for pixmap uploads As this causes a significant regression when benchmarking firefox on SNB with firefox-planet-gnome if we already have CPU buffers. Signed-off-by: Chris Wilson commit 1a6c16a24cb74f82df9757c034c50f2cf141523f Author: Chris Wilson Date: Sun Jan 1 14:32:48 2012 +0000 sna: Keep a freelist of buffers As reallocation of bo is the most frequent cause of malloc/free. Signed-off-by: Chris Wilson commit f008a2a90c632d843f07846426f6d14af728c13b Author: Chris Wilson Date: Sun Jan 1 10:24:49 2012 +0000 sna: Tune blt routine Signed-off-by: Chris Wilson commit b57c7248acc2ad6a51b7eac74449d155defc998f Author: Chris Wilson Date: Sun Jan 1 14:10:20 2012 +0000 sna: Delete some redundant code Signed-off-by: Chris Wilson commit 83d16455b70e2f0348e51f95128d4fcae3736e7f Author: Chris Wilson Date: Sun Jan 1 13:58:44 2012 +0000 sna: Eliminate one union pass when reducing damage-add Signed-off-by: Chris Wilson commit 2f5b769e7144c7b85010429e6d8aebf9c9fc1628 Author: Chris Wilson Date: Sun Jan 1 10:00:37 2012 +0000 sna: Move some checks against programming errors to asserts Signed-off-by: Chris Wilson commit 24bf59c43fbedd9d85c913ac3c7f0e752c232eb4 Author: Chris Wilson Date: Sat Dec 31 18:57:57 2011 +0000 Revert "sna: Increase the minimum alignment constraint for g33" This reverts commit 2934e778f01cdf1307732b248b11a31c0e79e866. The actual cause of the bug I was seeing on my PNV box turned out to be a1f585a3d0a, so time to reinvestigate the alignment issues. commit 770a953ff03bb8328c3f29e274d225528840f30c Author: Chris Wilson Date: Sat Dec 31 17:19:08 2011 +0000 uxa/video: Clear all state structures before uploading A regression from eb859f644633e left some of the state uninitialised before uploading to the GPU leading to undefined behaviour. Reported-by: Alexey Shumitsky Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=44338 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=44252 Signed-off-by: Chris Wilson commit f1995c684d2e074ae04c9e42727e59cc77e16104 Author: Chris Wilson Date: Sat Dec 31 15:57:32 2011 +0000 uxa: the video destination should be in the render write domain Signed-off-by: Chris Wilson commit 511146d2e40e267f74e18c0dd06b65f5d6a69f5d Author: Chris Wilson Date: Sat Dec 31 13:27:13 2011 +0000 sna: Implement inplace trapezoidal addition (for clipmasks) Signed-off-by: Chris Wilson commit a1f585a3d0ae72e896684f62960a9ae96dc259f5 Author: Chris Wilson Date: Sat Dec 31 12:43:04 2011 +0000 sna: Update pixmap->devKind for GTT mappings Signed-off-by: Chris Wilson commit 343de19ef98eea016601dbbe2d1f601dbff9de20 Author: Chris Wilson Date: Sat Dec 31 11:19:44 2011 +0000 sna: Implement tiled uploads for small stipples larger than patterns Signed-off-by: Chris Wilson commit fb15cf2294da20bfc375ae62443b822770850316 Author: Chris Wilson Date: Sat Dec 31 10:24:58 2011 +0000 sna: Make sure the shadow pixmap is suitable for scanout Signed-off-by: Chris Wilson commit 281425551bdab7eb38ae167a3205b14ae3599c49 Author: Chris Wilson Date: Sat Dec 31 10:10:00 2011 +0000 sna: Tweak throttling to be run before going idle The goal is to avoid introducing extra latency in the middle of a command sequence and to delay it until we have to wait for the clients. Signed-off-by: Chris Wilson commit 2f5448f1329f344702e4e29b2509d7190677bbf0 Author: Chris Wilson Date: Fri Dec 30 22:13:05 2011 +0000 sna: compare tiling against minimum width for relevancy Signed-off-by: Chris Wilson commit e82b3d8069629ae5776e039d6ea1f3c901a3a094 Author: Chris Wilson Date: Fri Dec 30 09:16:00 2011 +0000 sna: Defer the initial source upload if the GPU bo will be untiled Signed-off-by: Chris Wilson commit 72624b0b5ba6549cb8d96f992d2656a28ec6078c Author: Chris Wilson Date: Thu Dec 29 22:43:35 2011 +0000 sna: Remove redundant use-gpu-bo clause Signed-off-by: Chris Wilson commit 6dc9bbc0161f071bbaf5c7e379974158a5ad666f Author: Chris Wilson Date: Thu Dec 29 22:43:44 2011 +0000 sna: Fix DBG for no intersection with damage Signed-off-by: Chris Wilson commit 0dc5c0651cb691fb8811cdf3075b3d322f9d37f8 Author: Chris Wilson Date: Thu Dec 29 13:49:23 2011 +0000 sna: Tune disarming of the scanout flush timer Signed-off-by: Chris Wilson commit ccd895c61d9f409a36c70eaf02d7ab43cc6c03c2 Author: Chris Wilson Date: Thu Dec 29 12:09:04 2011 +0000 sna: Allow uploading inplace to a freshly created GPU bo If the operation is favoured to be performed using a WC upload, presume that we will use the uploaded pixmap on the GPU and so prefer to create a GPU buffer to hold the fresh data. Signed-off-by: Chris Wilson commit f85a853455249fd782e907ce1d78d11f75362def Author: Chris Wilson Date: Thu Dec 29 11:59:57 2011 +0000 sna: Only update the damage region as necesary for contains-box Signed-off-by: Chris Wilson commit f53936407bd68f3d36ab525b4e0531fad58098e3 Author: Chris Wilson Date: Thu Dec 29 11:59:37 2011 +0000 sna: Reset the source counter upon dirtying the shadow pixmap Signed-off-by: Chris Wilson commit c9f7f10bc528ac783f4e89bdb36fc56ccdcd0088 Author: Chris Wilson Date: Thu Dec 29 09:03:21 2011 +0000 sna: Be paranoid and check for errors from mmapping the bo Now that the error propagation is actually in place, we may as well use it. Signed-off-by: Chris Wilson commit 98e053503aed292beb2006043e8508d59262c7c2 Author: Chris Wilson Date: Thu Dec 29 08:50:53 2011 +0000 uxa: Depend upon DRI2 not DRI The symbols required for building intel_dri.c are checked during configure under the DRI2 defines. Signed-off-by: Chris Wilson commit 58cee9d0def82d19f03c840325b6e84c20e77ace Author: Chris Wilson Date: Wed Dec 28 22:52:58 2011 +0000 sna/dri: Update set_bo() for change of bo ownership transfer during page-flip Signed-off-by: Chris Wilson commit f9ee02be269d1658d3a8a5166e9bd2a49ab6dea0 Author: Chris Wilson Date: Wed Dec 28 22:52:44 2011 +0000 sna: Fix debug message for read-boxes Signed-off-by: Chris Wilson commit 698bd43e0f7a59bcce6601b0c96df85935e54bb3 Author: Chris Wilson Date: Wed Dec 28 21:45:12 2011 +0000 sna: Do not create glyph caches if the gpu is wedged As these will only be created in normal memory and never used. Signed-off-by: Chris Wilson commit 78425a15597079c0549de0fa1bbbdbebcf9eaf28 Author: Chris Wilson Date: Wed Dec 28 15:26:14 2011 +0000 sna: Refactor common code for testing gpu busyness of a pixmap Signed-off-by: Chris Wilson commit 9105b7e03ab7e3602c7200ebdc44e89f873afe1f Author: Chris Wilson Date: Wed Dec 28 13:46:18 2011 +0000 sna: Perform clip mask compositing inplace Avoid the extra composite-in pass for simple clipmask construction. Signed-off-by: Chris Wilson commit a5583165da0d2ae5eb1e5a2e11ee6e245d4b5aa4 Author: Chris Wilson Date: Wed Dec 28 13:57:09 2011 +0000 uxa: Remove concept of global FB ownership made obsolete by GEM and KMS With the introduction of GEM, we can continue to submit batch buffers irrespective of ownership of the console, so do so. Signed-off-by: Chris Wilson commit 337635ab97bbfc9b4455eadb63214783bb90bb2b Author: Chris Wilson Date: Wed Dec 28 10:56:57 2011 +0000 sna: Add some more debugging to find the source of a read-boxes error Signed-off-by: Chris Wilson commit 3bf81241cbc97763e5f3cbc82772437ccbff5a6c Author: Chris Wilson Date: Tue Dec 27 17:09:59 2011 +0000 sna: Skip zero-sized glyphs Similar to the empty glyphs make sure that we just advance by the character width without drawing anything. Signed-off-by: Chris Wilson commit 8fc21328a0bdf87fde35d68d2b27834011acde7b Author: Chris Wilson Date: Tue Dec 27 15:26:46 2011 +0000 sna: Don't discard empty glyphs, just skip them A space is encoded as a 1x1 blank glyph, but we still need to advance by its character width and so we cannot simply discard the glyph. References: https://bugs.freedesktop.org/show_bug.cgi?id=44091id=44091 Signed-off-by: Chris Wilson commit 726290ea2a54716d3d8fb998fe7d2ab252421de7 Author: Chris Wilson Date: Tue Dec 27 13:16:34 2011 +0000 sna: Rearrange damage for tail-call optimisations Signed-off-by: Chris Wilson commit 04396b6e5039c0cd1ac655a8a9afe88bb8a1ebde Author: Chris Wilson Date: Tue Dec 27 09:58:40 2011 +0000 sna: Don't force the creation of a GPU bo if it will be untiled Signed-off-by: Chris Wilson commit 655a96cd5f12ff7fe593e7da5f34c7c92ef9b87d Author: Chris Wilson Date: Mon Dec 26 14:37:37 2011 +0000 sna: Mark the glyph cache as being unpageable In order to prevent it falling foul of the inactive pixmap reaper, we need to mark the pixmap as pinned. Signed-off-by: Chris Wilson commit 2934e778f01cdf1307732b248b11a31c0e79e866 Author: Chris Wilson Date: Mon Dec 26 11:30:39 2011 +0000 sna: Increase the minimum alignment constraint for g33 Still not getting the single-stream mode that I am aiming for! Reported-by: Matti Hamalainen Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=44150 Signed-off-by: Chris Wilson commit f7593a995a829978ce81397dde5c9ea6f9bb7681 Author: Chris Wilson Date: Sun Dec 25 09:40:46 2011 +0000 sna: Move the is-mappable check into the callers of region_inplace() As they slightly differ in their requirements. Signed-off-by: Chris Wilson commit aff32e3e08fce3c7b8dab3a25c96d69c409471fc Author: Chris Wilson Date: Sun Dec 25 00:36:33 2011 +0000 sna: Tweak damage not to reduce if it will not affect the outcome of reducing to all Signed-off-by: Chris Wilson commit be21a09fef0a3499153f2c123b65f1d5c872d437 Author: Chris Wilson Date: Sun Dec 25 00:28:55 2011 +0000 sna: Only call move-to-gpu on scanout flush if we have cpu damage to move Reduce the number of redundant calls. Signed-off-by: Chris Wilson commit 7b69dc88094ee2bb017364cd4356f3ae69cbc1f1 Author: Chris Wilson Date: Sun Dec 25 00:21:21 2011 +0000 sna: Inline the common portion of sna_get_pixel_from_rgba() The function overhead completely dominates for the common case. Signed-off-by: Chris Wilson commit b117f65520919f4ba36010cfe913a8c53166bf23 Author: Chris Wilson Date: Sat Dec 24 21:03:01 2011 +0000 sna: Jump straight to the fallback copy routines if the dst is not attached Marginally simplify the convoluted logic for choosing the most appropriate path and help prevent further errors. Signed-off-by: Chris Wilson commit 72217790ee2c080d618274456360b481d015e898 Author: Chris Wilson Date: Sat Dec 24 19:14:09 2011 +0000 sna: Use shadow if the GPU is busy or not immediately mappable Signed-off-by: Chris Wilson commit 0be136c21f0373d1eb2259b83c598655f4eb841e Author: Chris Wilson Date: Sat Dec 24 18:44:15 2011 +0000 sna: use indirect uploads if the bo was last known to be unmappable Signed-off-by: Chris Wilson commit e764a52ee8f0c552e218b3110318be9ba06634ae Author: Chris Wilson Date: Sat Dec 24 15:17:06 2011 +0000 sna: Encourage large operations to be migrated to the GPU Signed-off-by: Chris Wilson commit f9f8535db6a9e3affba9ba2c2a9314dfe12ab270 Author: Chris Wilson Date: Sat Dec 24 21:01:39 2011 +0000 sna: Fix damage reduction by adding new boxes to the tail of the box list Signed-off-by: Chris Wilson commit f8c91f229a9ca18fc58b0cabbea595d26627db8b Author: Chris Wilson Date: Sat Dec 24 12:17:53 2011 +0000 sna: We can always compare the box to be subtracted against the dirty extents Signed-off-by: Chris Wilson commit cc21d3fa045209861564cf240a9a082e0bdb63f8 Author: Chris Wilson Date: Sat Dec 24 12:17:43 2011 +0000 sna/damage: Fix the is-contained test It was a non-overlapping test which is almost the reverse of what was intended. Signed-off-by: Chris Wilson commit b86e4f59299f935d5a0ea8375da97e6fc57571f9 Author: Chris Wilson Date: Sat Dec 24 11:45:27 2011 +0000 sna: Check that the copy dst is attached before replacing damage Signed-off-by: Chris Wilson commit 3bcc9f08e1309725c18dfa5ca0e905bb88ebf4ea Author: Chris Wilson Date: Sat Dec 24 09:59:24 2011 +0000 sna: Create a GPU for a streaming write If we are about to do a write-only drawing operation that will exceed our cache, allocate a GPU bo and perform the operation inplace. Signed-off-by: Chris Wilson commit efc8d04fc114e9966e5ca00600f9663ecf03a5ca Author: Chris Wilson Date: Sat Dec 24 09:32:18 2011 +0000 sna: Promote an unattached pixmap to the GPU for inplace copies If we would prefer to perform the copy on the GPU and if the pixmap is virgin, create a GPU bo for the operation. Signed-off-by: Chris Wilson commit 8c0e7ff5ac996a5a5c0790966084d1c9e1b1f859 Author: Chris Wilson Date: Sat Dec 24 09:19:18 2011 +0000 sna: Clear the CPU damage for other overwriting rectangular blits as well And make the accessing of the sna_pixmap safe -- it may not even exist! Signed-off-by: Chris Wilson commit 7f9ed3ca56466f319d9ec06aebb7055552fb1fe1 Author: Chris Wilson Date: Sat Dec 24 08:57:49 2011 +0000 sna: reset the pixmap on the initial clear after being swapped out Pixmaps are swapped out into the CPU after a period of inactivity. This then prevents the core rendering routines from migrating the pixmap back to the GPU until it gets used again on a Render path. However, we can clear that CPU damage and enable migration before a number of key steps in the expose process. Signed-off-by: Chris Wilson commit 8c987b6ccfa83b99a316a7da49a6d7e30fcdf295 Author: Chris Wilson Date: Sat Dec 24 09:09:58 2011 +0000 sna: Cheaply reduce damage when subtracting and removing the entire damage Signed-off-by: Chris Wilson commit 098592ca5d79af2e4bdcd82ee598c4b2ba08df6b Author: Chris Wilson Date: Sat Dec 24 02:49:24 2011 +0000 sna: Remove the independent tracking of elts from boxes Following the switch to a global mode for damage, the elts array became redundant and all that is required is the list of boxes. Signed-off-by: Chris Wilson commit 73df0c7ab7c3a9edf0be2439c7e7ab07c0d75ecf Author: Chris Wilson Date: Sat Dec 24 02:48:18 2011 +0000 sna: Tune region upload inplace threshold Signed-off-by: Chris Wilson commit ef66c5b5ad221211948ec795ade031591c3f0ac7 Author: Chris Wilson Date: Sat Dec 24 00:15:33 2011 +0000 sna: Search the inactive VMA cache first for a linear mapping Signed-off-by: Chris Wilson commit 1cc07fa2d24b10ac95c7a84908290ec06539d447 Author: Chris Wilson Date: Fri Dec 23 20:31:21 2011 +0000 sna: Avoid forced creation of GPU bo for tiny operations and dirty pixmaps Signed-off-by: Chris Wilson commit e5bfea5826117bdd29664d83b3d478ff9db29ca3 Author: Chris Wilson Date: Fri Dec 23 20:30:58 2011 +0000 sna: Discard any GPU damage when overwriting with trapezoids Signed-off-by: Chris Wilson commit 819dc93e52533afc9b9005db1472672c3d42229f Author: Chris Wilson Date: Fri Dec 23 20:29:59 2011 +0000 sna/glyphs: Create GPU bo for large enough destination surfaces with glyphs Signed-off-by: Chris Wilson commit 9580ae8490a4119dac2fefe0085326db350209d7 Author: Chris Wilson Date: Fri Dec 23 19:11:43 2011 +0000 sna: Pass usage-hint to move-to-gpu When simply creating a source GPU bo it is preferrable not to mark it as all-damaged. Signed-off-by: Chris Wilson commit 98f15fc61361b7f1e01969f8d4237c13e93e3fb0 Author: Chris Wilson Date: Fri Dec 23 17:25:42 2011 +0000 sna: Don't align pwrite to cachelines for doing discontiguous copies The batch compaction breaks the 1:1 mapping between the cpu buffer and the bo, so we can no longer safely align the transfer to whole cachelines. References: https://bugs.freedesktop.org/show_bug.cgi?id=44091 Signed-off-by: Chris Wilson commit 3850f4ad48986691e1fb98038ae921deb6c25423 Author: Chris Wilson Date: Fri Dec 23 12:42:02 2011 +0000 sna: Silence unsigned comparison against 0 The good news is if this ever failed, the kernel is far too broken... Signed-off-by: Chris Wilson commit d9ca113a83177562cde3537bf955164c6950cbd0 Author: Chris Wilson Date: Fri Dec 23 10:06:58 2011 +0000 sna: Trim the unused pages from the batch between the last command and surface Signed-off-by: Chris Wilson commit 013eda17e91ff421cd1ef70470e3575f1183fcbd Author: Chris Wilson Date: Fri Dec 23 10:40:58 2011 +0000 sna: Free the additional bindings on proxies commit 84d97bdba02b909369b54de21425ffc9f6ad581a Author: Chris Wilson Date: Fri Dec 23 00:01:50 2011 +0000 sna/gen2+: Reuse source channel for mask where possible GTK+ has a clever trick for premultiplying its images by loading the same pixel data into both the source and mask, and then performing the composite. This causes us to upload the same pixel data twice! Signed-off-by: Chris Wilson commit 281f620573917faef52d9226b12737ce1e2dffdc Author: Chris Wilson Date: Thu Dec 22 22:23:41 2011 +0000 sna: Age active wholly damaged GPU buffers more slowly Signed-off-by: Chris Wilson commit f8575b8bd150ae11147fbf6b59171cbbb072dfec Author: Chris Wilson Date: Thu Dec 22 22:15:42 2011 +0000 sna: Cancel the immediate batch submit once handled in the block handler Signed-off-by: Chris Wilson commit 609e1d9da282323ab7abc1baad4d3f24ac55588f Author: Chris Wilson Date: Thu Dec 22 15:54:39 2011 +0000 sna: Bump the inactivity timeout With the goal of removing inactive pixmap froms the GPU after 5 minutes of idleness. Signed-off-by: Chris Wilson commit 62602209e54c3568795122ee1c902c3b81985303 Author: Chris Wilson Date: Thu Dec 22 14:20:20 2011 +0000 sna: discard damage-all even for width|height==0 operations Even if we don't know the extents of the render operation, if the entire pixmap is damaged we can still reduce the damage tracking. Signed-off-by: Chris Wilson commit f68a99a55e1e63d7e76a13a7b77b4bb3f7802b9e Author: Chris Wilson Date: Thu Dec 22 15:04:22 2011 +0000 sna: Mark fresh GPU only bo as being all damaged Presume that we will not fallback and so treat a GPU only bo (one that is initially created on the GPU) as being all-damaged. This makes future operations cheaper as the damage tracking overhead is much reduced, and the cost of the first readback will mainly be in the synchronisation overhead. Signed-off-by: Chris Wilson commit 1f2cd536bc712843233de3f5c50802c388d63c72 Author: Chris Wilson Date: Thu Dec 22 14:30:04 2011 +0000 sna: Reduce damage after complete solid fills RenderFillRectangles is often used to initially clear a pixmap after creation by flooding it with a solid colour, as is PolyRect. We can reduce further damage operations by checking for this condition. Signed-off-by: Chris Wilson commit 7a6a2c5d4cc98da76c73152cb7eafcbdb0c088ec Author: Chris Wilson Date: Thu Dec 22 13:24:07 2011 +0000 sna: Setup the pixmap correctly in order to attach our private During creation of sna_pixmap we validate that we can use a GPU bo with the target pixmap. This fails if we pass in a raw pixmap header, so make sure the scratch pixmap is fully initialised first. Signed-off-by: Chris Wilson commit f621b3de841f6037d387ca1439a0abe12ef29811 Author: Chris Wilson Date: Thu Dec 22 11:19:45 2011 +0000 sna: flatten source alphamaps Replace the source picture+alpha with a bo that contains the RGB channels from source and A from the alpha map. Signed-off-by: Chris Wilson commit 6c08eb4d6f8789e692ef018e007d1ae97a57c25f Author: Chris Wilson Date: Thu Dec 22 11:42:39 2011 +0000 sna/gen2+: Prefer to use the CPU if we have a source alphamap and CPU pictures Signed-off-by: Chris Wilson commit 354dc3c65b9cb86885b6927ee2b80cd2ee6a82ff Author: Chris Wilson Date: Wed Dec 21 21:35:06 2011 +0000 sna: Avoid fallbacks for convolutions by rendering the convolved texture If we have no shader support for generic convolutions, we currently create the convolved texture using pixman. A multipass accumulation algorithm can be implemented on top of CompositePicture, so try it! Signed-off-by: Chris Wilson commit 34cfa15e522ba486d591d37f37d494e399f70354 Author: Chris Wilson Date: Wed Dec 21 00:47:47 2011 +0000 sna: Drop suggested tiling parameter when creating bo The only place where we did anything other than use the default was when creating a new bo for CopyArea. In that case, basing the choice on the src GPU bo was not only wrong but a potential segfault. Signed-off-by: Chris Wilson commit 4d348425592251f49fd31b9259651e8b05e92d1e Author: Chris Wilson Date: Tue Dec 20 21:35:58 2011 +0000 sna: Intiialise pixmap size before use After 0c12f7cb0 we were setting the width/height of the pixmap *after* trying to use them to determine if the pixmap could be created on the GPU. Normally this would be corrected when we attempt to render, except for the core drawing protocol. Signed-off-by: Chris Wilson commit e71058c9c81067cd72b1193242acbd85272ab73a Author: Chris Wilson Date: Tue Dec 20 21:20:48 2011 +0000 sna: Always readback untiled bo in place We are going to gain no advantage from the kernel detiling the busy buffer. Signed-off-by: Chris Wilson commit 6690ff51ec939367051985172a00bf8ed65ac4e3 Author: Chris Wilson Date: Tue Dec 20 21:09:01 2011 +0000 sna: Treat all exported bo as potentially active Don't assume that a read/write will clear the active flag if the bo has been exported to another DRI client. Signed-off-by: Chris Wilson commit 743084bbc4c411694190f60cf5161086ada6454f Author: Chris Wilson Date: Tue Dec 20 20:42:08 2011 +0000 sna: Only upload tiny pixmaps on the second use Most small pixmaps appear to be single shot, so amalgamate them into one buffer and trim our memory usage. Signed-off-by: Chris Wilson commit ca24ffe8b9acef034a5610376c73ebbdce87d41c Author: Chris Wilson Date: Tue Dec 20 20:11:12 2011 +0000 sna: Improve a debug message Give the units when saying the pixmap is too small to bother tiling. Signed-off-by: Chris Wilson commit f837b9bcc7d384aeb37c2e9ebdac95571deaedc6 Author: Chris Wilson Date: Tue Dec 20 20:06:25 2011 +0000 sna/render: If the pixmap already has a GPU bo, use it for the source The usage hint not to create a GPU bo is obviously superceded if we already have a GPU bo. Signed-off-by: Chris Wilson commit dfafe76a57655784c313bc2078d0b28a7ac2675b Author: Chris Wilson Date: Tue Dec 20 18:39:20 2011 +0000 sna: Do not use Y-tiling for composite back pixmaps These are treated by the core drawing routines as replacements for the front-buffer attached to Windows, and so expect the usual BLT accelerations are available, for example overlapping blits to handle scrolling. If we create these pixmaps with Y-tiling and then they are pinned by the external compositor we are forced to perform a double copy through the 3D pipeline as it does not handle overlapping blits and the BLT does not handle Y-tiling. Signed-off-by: Chris Wilson commit 0cda7b4fa82870d57fcd0036a3b734cc73e2bf28 Author: Chris Wilson Date: Tue Dec 20 17:58:33 2011 +0000 sna: Implement extended fallback handling for src == dst copies Only marginally better than falling all the way back to using the CPU, is to perform a double copy to workaround the overlapping copy. Signed-off-by: Chris Wilson commit d257a967396c517146cfb12bbec5cd28418752a3 Author: Chris Wilson Date: Tue Dec 20 12:53:42 2011 +0000 sna: Explicitly handle depth==1 scratch pixmaps Short-cut the determination of whether it can be tiled and accelerated -- we know it can't! This is mainly to cleanup the debug logs. Signed-off-by: Chris Wilson commit 0c12f7cb01e75de3bf9c2af8ac6d5b4152566457 Author: Chris Wilson Date: Tue Dec 20 10:19:01 2011 +0000 sna: Tidy up some recent valgrind complaints with reuse of scratch pixmaps Signed-off-by: Chris Wilson commit d3a4f5db149888239b78df1b9fcc5379b1603005 Author: Chris Wilson Date: Tue Dec 20 12:12:39 2011 +0000 sna: Fixup the refcnt to avoid an assert Signed-off-by: Chris Wilson commit ac52a1fcd11a3ab10e1843ed2d19f9cf818de4c1 Author: Chris Wilson Date: Tue Dec 20 02:20:48 2011 +0000 sna: Don't immediately check for region intersection after subtract In the READ==0 case we know that the region does not intersect damage because we have just subtracted, and checking the intersection causes us to immediately apply the subtraction operation defeating the optimisation and forcing the expensive operation each time. Signed-off-by: Chris Wilson commit 4071dca0ef8b2b1605f16bed5c42991885a35efd Author: Chris Wilson Date: Tue Dec 20 00:25:29 2011 +0000 sna: Don't mark mapping as synchronous by default Only those that point into scratch memory need to synchronized before control is handed back to the client. Signed-off-by: Chris Wilson commit 4c2a97e9d2b1073f0e4b3f7b6670939e1b1c5121 Author: Chris Wilson Date: Mon Dec 19 23:10:45 2011 +0000 sna: Always pass the damage to sna_drawable_use_gpu_bo() As it now assumes that the damage is always writable. Signed-off-by: Chris Wilson commit 06a1792f91d3ae2ebb17e4ad55caec10c5407aa6 Author: Chris Wilson Date: Mon Dec 19 21:04:02 2011 +0000 sna: Avoid the GPU readback with READ==0 move_to_cpu Signed-off-by: Chris Wilson commit 0b5fec3f80c8d2e4d3b504af5bb02fb76be6b3a6 Author: Chris Wilson Date: Mon Dec 19 20:13:05 2011 +0000 sna: Drop the is-mapped flag after operating via the GPU Mark the end of a sequence of CPU operations and force the decision to map again to be based on the current upload operation. Signed-off-by: Chris Wilson commit 351c8f16336e757b8a4d218b3afda0698781d143 Author: Chris Wilson Date: Mon Dec 19 19:51:12 2011 +0000 sna: Discard all damage when replacing pixmap contents Signed-off-by: Chris Wilson commit addf66dda7cecaee8e58f53bb4e8d725dfdd2717 Author: Chris Wilson Date: Mon Dec 19 18:37:08 2011 +0000 sna: Tweak the rendering priorities If the last operation was on the GPU, continue on the GPU if this operation overlaps any GPU damage or does not overlap CPU damage. Otherwise, if the last operation was on the CPU only switch to the GPU if we do not overlap any CPU damage and overlap existing GPU damage. Signed-off-by: Chris Wilson commit 9b6ade12346efa5b9cc095ad44c7d71880a19ab0 Author: Chris Wilson Date: Mon Dec 19 17:21:26 2011 +0000 sna: Create a GPU bo for accelerated core drawing As we now can accelerate most of the common core drawing operations, we can create GPU bo for accelerated drawing on first use without undue fear of readbacks. This benefits Qt especially which heavily uses core the drawing operations. Signed-off-by: Chris Wilson commit eeb81dd6b4651c6b3e8676007d37480888726efa Author: Chris Wilson Date: Mon Dec 19 16:48:02 2011 +0000 sna: Remove the forced inplace upload Make the decision per-operation, with a tendency to remain mapped. Signed-off-by: Chris Wilson commit c3a8d77a2b6806a69de87fbaadea8612ef5f91dd Author: Chris Wilson Date: Mon Dec 19 16:51:29 2011 +0000 sna: Tune the inplace cross-over point to be half-cache size The theory being that we will also require cache space to copy from when uploading into the shadow. Signed-off-by: Chris Wilson commit d53b1b28951fd7558e9913a2973c17983e0bcecf Author: Chris Wilson Date: Mon Dec 19 15:33:39 2011 +0000 configure: Bump the required pixman version UXA now also uses pixman_triangle_t in order for its fallback, so we need to bump the required pixman version for UXA as well as SNA. Reported-by: Fabio Pedretti Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43946 Signed-off-by: Chris Wilson commit 1fa5721f064a8d1f34e4032b52f24597f4015313 Author: Chris Wilson Date: Mon Dec 19 00:37:03 2011 +0000 sna: Reset the GTT mapping flag when freeing the shadow pointers Signed-off-by: Chris Wilson commit 7326d3098662688b8040c6e1261064caed1f5d06 Author: Chris Wilson Date: Mon Dec 19 00:35:14 2011 +0000 sna: Restore CPU shadow after a GTT memory When mixing operations and switching between GTT and CPU mappings we need to restore the original CPU shadow rather than accidentally overwrite. Signed-off-by: Chris Wilson commit ae32aaf4b20dafef138dc9c28dbddbfe49f24b83 Author: Chris Wilson Date: Mon Dec 19 00:34:12 2011 +0000 sna/gen[23]: We need to check the batch before doing an inline flush A missing check before emitting a dword into the batch opened up the possibility of overflowing the batch and corrupting our state. Signed-off-by: Chris Wilson commit e32ad646762ccc7f22f938454e222d43abfb38ed Author: Chris Wilson Date: Sun Dec 18 23:42:07 2011 +0000 sna: Continue searching the linear lists for CPU mappings Prefer to reuse an available CPU mapping which are considered precious and reaped if we keep too many unused entries availabled. Signed-off-by: Chris Wilson commit 15a769a66fa1afbcffc642ef980387cffefc6bef Author: Chris Wilson Date: Sun Dec 18 22:18:02 2011 +0000 sna: Distinguish between GTT and CPU maps when searching for VMA Similarly try to avoid borrowing any vma when all we intend to do is pwrite. Signed-off-by: Chris Wilson commit d0ee695ef091671e2cc69b773f517030ebe961b2 Author: Chris Wilson Date: Sun Dec 18 20:28:18 2011 +0000 sna: the active cache is not marked as purgeable, so skip checking it Otherwise we do a lot of list walking for no-ops. Signed-off-by: Chris Wilson commit 8df9653135c6390ed699ba9f8cbf374b2cc84396 Author: Chris Wilson Date: Sun Dec 18 19:53:14 2011 +0000 sna: clear the request list when reusing a flushing bo That the rq is NULL when on the flushing list is no longer true, but now it points to the static request instead. Signed-off-by: Chris Wilson commit b51e3de66292e4b29483324e1dbfeb2495817351 Author: Chris Wilson Date: Sun Dec 18 19:51:17 2011 +0000 sna: When freeing vma, first see if the kernel evicted any Signed-off-by: Chris Wilson commit fed8d145c148bfa8a8a29f4088902377f9a10440 Author: Chris Wilson Date: Sun Dec 18 19:26:38 2011 +0000 sna: Use a safe iterator whilst searching for inactive linear bo As we may free a purged bo whilst iterating, we need to keep the next bo as a local member. Include the debugging that led to this find. Signed-off-by: Chris Wilson commit 2a98dabcabf25067abcda60f233656e19e83493a Author: Chris Wilson Date: Sun Dec 18 18:48:30 2011 +0000 sna: Purge all evicted bo After we find a single bo that has been evicted from our cache, the kernel is likely to have evicted many more so check our caches for any more bo to reap. Signed-off-by: Chris Wilson commit 8ae105b2c7a06fc81757f06ec1aaa2447b53498a Author: Chris Wilson Date: Sun Dec 18 18:20:11 2011 +0000 sna: Only retire for the VMA search if there are cached VMA If there are no VMA that might become inactive, there is no point scanning the inactive lists if we are searching for VMA. This prevents the regression in firefox-fishbowl whilst maintaining most of the improvement with PutComposite. Signed-off-by: Chris Wilson commit a0c0a3765ca348c74096fb157885da5b1258ee08 Author: Chris Wilson Date: Sun Dec 18 16:59:15 2011 +0000 sna: Retire if the inactive vma list is empty Try to recycle vma by first trying to populate the inactive list before scanning for a vma bo to harvest. Signed-off-by: Chris Wilson commit 34efb7314612cedcda4f866bc33f3ad5b6929ae2 Author: Chris Wilson Date: Sun Dec 18 16:58:04 2011 +0000 sna: Hint likely usage of CPU bo If we are going to transfer GPU damage back to the CPU bo, then we can reuse an active buffer and so improve the recycling. Signed-off-by: Chris Wilson commit 301896743867de0cbd22063c4bf2e1a8cf491656 Author: Chris Wilson Date: Sun Dec 18 16:17:04 2011 +0000 sna: Only upload to the source GPU bo if we need tiling to avoid TLB misses Signed-off-by: Chris Wilson commit b7f5d75aa54db7d0542b09f4b3081f7692ede602 Author: Chris Wilson Date: Sun Dec 18 16:11:34 2011 +0000 Silence uxa-only compilation Kill the stray warning for the undeclared extern used by the module loader. Signed-off-by: Chris Wilson commit a73cc4bf1e554806f403c6704d1cf98491f4d444 Author: Chris Wilson Date: Sun Dec 18 14:54:12 2011 +0000 sna/gen5: Tidy checking against hardcoded maximum 3D size Signed-off-by: Chris Wilson commit b43548af39f8773283b744e979ee575032183cdc Author: Chris Wilson Date: Sun Dec 18 12:11:01 2011 +0000 sna: Explicitly handle errors from madv In order to avoid conflating whether a bo was marked purgeable with its retained state, we need to carefully handle the errors from madv. Signed-off-by: Chris Wilson commit 954cf5129d462f7f4ff5d4ff44b256e0f091667c Author: Chris Wilson Date: Sun Dec 18 10:40:35 2011 +0000 sna/gen[67]: check for context switch after preparing source If we used the BLT to prepare the source, see if we can continue the operation on the BLT. Signed-off-by: Chris Wilson commit 90a432431c8fc293db611289c3369669c2975706 Author: Chris Wilson Date: Sun Dec 18 10:23:11 2011 +0000 sna/gen[23]: Try BLT if the source/target do no fit in the 3D pipeline Signed-off-by: Chris Wilson commit eeb9741981be528d1fba3021a30ab24084a5ce8c Author: Chris Wilson Date: Sun Dec 18 10:04:08 2011 +0000 sna/gen3: Tidy checks against hardcoded maximum 3D pipeline size Signed-off-by: Chris Wilson commit dcfcf438a511ee43e5969d01e41b3040a08cac3a Author: Chris Wilson Date: Sun Dec 18 09:58:08 2011 +0000 sna/gen2+: If we use the BLT to prepare the target, try using BLT for op If we incurred a context switch to the BLT in order to prepare the target (uploading damage for instance), we should recheck whether we can continue the operation on the BLT rather than force a switch back to RENDER. Signed-off-by: Chris Wilson commit 507debe8015ee10734a0b8a6ddf4aa251b7dddc5 Author: Chris Wilson Date: Sun Dec 18 01:45:47 2011 +0000 sna/gen5: If we need to flush the composite op, check to see if we can blit If we need to halt the 3D engine in order to flush the pipeline for a dirty source, we may as well re-evaluate whether we can use the BLT instead. Signed-off-by: Chris Wilson commit de530f89a36a80b652ed01001e6f0d4c8b1dc11a Author: Chris Wilson Date: Sun Dec 18 01:41:03 2011 +0000 sna/gen5+: First try a blt composite if the source/dest are too large If we will need to extract either the source or the destination, we should see if we can do the entire operation on the BLT. Signed-off-by: Chris Wilson commit 7b88f87945f23b3f80ae44b2aa19f00dff1904c8 Author: Chris Wilson Date: Sun Dec 18 00:44:00 2011 +0000 sna: Upload images in place from CopyArea As for PutImage, if the damage will be immediately flushed out to the GPU bo, we may as well do the write directly to the GPU bo and not staged via the shadow. Signed-off-by: Chris Wilson commit 1418e4f3156a17ff6c5cd0f653918160e30b7c0e Author: Chris Wilson Date: Sat Dec 17 23:11:03 2011 +0000 sna: Tune the default pixmap upload paths One issue with the heuristic is that it is based on total pixmap size whereas the goal is to pick the placement for the next series of operations. The next step in refinement is to combine an overall placement to avoid frequent migrations along with a per-operation override. Signed-off-by: Chris Wilson commit 25c353503a25d20e7db5acdc63d83564804efdf4 Author: Chris Wilson Date: Sat Dec 17 17:16:07 2011 +0000 sna: Simplify write domain tracking Replace the growing bitfield with an enum marking where it was last used. Signed-off-by: Chris Wilson commit d20d167a753d8e4fe581950e1bc49f29e0ec9f1f Author: Chris Wilson Date: Sat Dec 17 16:28:04 2011 +0000 sna: Upload to large pixmaps inplace When the pixmap is large, larger than L2 cache size, we are unlikely to benefit from first copying the data to a shadow buffer -- as that shadow buffer itself will mostly reside in main memory. In such circumstances we may as perform the write to the GTT mapping of the GPU bo. As such, it is a fragile heuristic that may require further tuning. Avoiding that extra copy gives a 30% boost to putimage500/shmput500 at ~10% cost to putimage10/shmput10 on Atom (945gm/PineView), without any noticeable impact upon cairo. Reported-by: Michael Larabel Signed-off-by: Chris Wilson commit dd8fd6c90612ada39eb32b98adc5acc97e7902aa Author: Chris Wilson Date: Sat Dec 17 12:38:09 2011 +0000 sna: Search through the inactive VMA cache for potential upload bo Signed-off-by: Chris Wilson commit 8ef5d8c1955e2e2ee19c64730f600639ac42de55 Author: Chris Wilson Date: Thu Dec 15 17:52:20 2011 +0000 sna: Map the upload buffer using an LLC bo In order to avoid having to perform a copy of the cacheable buffer into GPU space, we can map a bo as cacheable and write directly to its contents. This is only a win on systems that can avoid the clflush, and also we have to go to greater measures to avoid unnecessary serialisation upon that CPU bo. Sadly, we do not yet go to enough length to avoid negatively impacting ShmPutImage, but that does not appear to be a artefact of stalling upon a CPU buffer. Note, LLC is a SandyBridge feature enabled by default in kernel 3.1 and later. In time, we should be able to expose similar support for snoopable buffers for other generations. Signed-off-by: Chris Wilson commit 6e47f283711d122d96384a1a82854c11644e6d68 Author: Chris Wilson Date: Fri Dec 16 23:49:18 2011 +0000 sna/gen3: Enforce a minimum width of 2 elements for the render target Signed-off-by: Chris Wilson commit 2ff0826f94ca16e95cd662385f7091be750dec30 Author: Chris Wilson Date: Fri Dec 16 23:33:25 2011 +0000 sna: Discard GPU damage first before choosing where to fill_boxes() Signed-off-by: Chris Wilson commit 55520bab578865f878965aa362ec4933f4b26050 Author: Chris Wilson Date: Fri Dec 16 23:22:38 2011 +0000 sna/gen3: Initialise missing value of need ca pass for fill_boxeS() Signed-off-by: Chris Wilson commit e56d5081ea2da930bc82036ac7c4c78b433117ed Author: Chris Wilson Date: Fri Dec 16 23:18:02 2011 +0000 sna: Wrap I915_GEM_GET_PARAM with valgrind markup Signed-off-by: Chris Wilson commit e0399ec1619fe4f87f0578791e697a7e2a8c86dc Author: Chris Wilson Date: Fri Dec 16 23:07:37 2011 +0000 sna: Suppress an overwritten XY_SRC_COPY Signed-off-by: Chris Wilson commit 1684ed6a5e8c26ecb48cc1a5025107466526fe94 Author: Chris Wilson Date: Fri Dec 16 19:35:35 2011 +0000 sna: Clean up compiler warnings for shadowed variables No outright bug, just plenty of unwanted noise. Signed-off-by: Chris Wilson commit 300586b229ae941ac31850af00d120d8c441c583 Author: Chris Wilson Date: Fri Dec 16 17:56:53 2011 +0000 sna/gen4+: disable the blend unit for PictOpSrc Signed-off-by: Chris Wilson commit 0de7604d8ecfc73ef9e92059340d00ce5cfcdd75 Author: Chris Wilson Date: Fri Dec 16 17:40:01 2011 +0000 src/gen4+: Add support for depth 15 render copies/fills Signed-off-by: Chris Wilson commit a8fe50ab6503d9cb63931771318dc3e84d002092 Author: Chris Wilson Date: Sat Dec 17 11:15:36 2011 +0000 uxa: Explicitly check for libdrm_intel in configure And remove the excess dependencies from the common files. Signed-off-by: Chris Wilson commit 3320b459d6fe90d1145e3874c840b95279fb16c8 Author: Chris Wilson Date: Sat Dec 17 12:08:24 2011 +0000 sna: Clean up caches stored upon the batch bo Until the advent of the VMA cache, we were safe to reap the batch bo by hand. However, as we continue to add additional data to the bo, it is wise to use the common free function. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43899 Signed-off-by: Chris Wilson commit e2c8bac972cf46982bf64bd786ecc001009081f0 Author: Zhigang Gong Date: Fri Dec 16 15:11:16 2011 +0800 uxa/glamor: Fallback to new glamor pixmap if failed to create textured pixmap. If we failed to create textured pixmap from BO's handle, we turn to create a new glamor pixmap by call glamor_create_pixmap rather than fallback to in-memory pixmap. Have to introduce a new wrapper function intel_glamor_create_pixmap. Signed-off-by: Zhigang Gong commit 19c184b7e4f8de747ed6fb1f6f910238193cf2a1 Author: Chris Wilson Date: Thu Dec 15 18:18:19 2011 +0000 sna/gen3: Check for upload failure of video bo And propagate that failure back to the client. Reported-by: Paul Neumann References: https://bugs.freedesktop.org/show_bug.cgi?id=43716 Signed-off-by: Chris Wilson commit e68a8748f3418179eb41e8d81ba1050b883b7e47 Author: Chris Wilson Date: Thu Dec 15 09:44:20 2011 +0000 uxa: Move the region creation beyond the call into glamor So that we avoid leaking the region if hooking into glamor. Signed-off-by: Chris Wilson commit e3153a779ed220ad71958ce56be179d458794964 Author: Chris Wilson Date: Thu Dec 15 09:30:28 2011 +0000 sna: cap the number of times we attempt to resubmit the batch upon EBUSY Just in case the kernel gets stuck in a loop and we fail to make any progress at all. Signed-off-by: Chris Wilson commit 1cc43dc97bd6438d7fc808cb5ee449889a966bdf Author: Chris Wilson Date: Thu Dec 15 00:00:56 2011 +0000 sna: More missing move-to-cpu allocation checks Signed-off-by: Chris Wilson commit 12448b56060fb9c14da6b4ab02c95c7834fd3874 Author: Chris Wilson Date: Thu Dec 15 00:01:37 2011 +0000 sna: silence warning for unused 'priv' Signed-off-by: Chris Wilson commit eb8e979b4de5d854f15ae4c86d3db29371b386c1 Author: Chris Wilson Date: Wed Dec 14 23:02:28 2011 +0000 uxa/glamor: Allocate a fbPixmap with storage for fallbacks When we try to create a glamor pixmap and fail we need to create a real pixmap along with its pixel allocation, instead of detaching ourselves and returning the fake pixmap header. Signed-off-by: Chris Wilson commit 232fa93a4bf1760e16e913e8fb4d5159c4be9295 Author: Chris Wilson Date: Wed Dec 14 20:11:39 2011 +0000 uxa/glamor: Always notify glamor that the glyph has been unrealized The danger of the early return when UXA is not using glyphs is evident in the eventual crash when glamor begins evicting and reusing its glyph cache slots. Signed-off-by: Chris Wilson commit 95cceb5ae5503af0ac50a923fa47e134f0da8743 Author: Chris Wilson Date: Wed Dec 14 19:27:53 2011 +0000 sna: Fix DBG crash whilst pruning inactive GPU buffers Don't attempt to dereference the NULL gpu_bo after having just freed it. Here in lies the folly of trying to blindly silence the compiler. Instead we should heed the error return as it means that we didn't decouple the pixmap from the inactive list and so we choose to place it back on the active list to purge again in the near future. Reported-by: Paul Neumann commit e7f4b7fd91a41cac77e5eb1cb4f185141b09a09e Author: Chris Wilson Date: Wed Dec 14 18:21:22 2011 +0000 sna: Add some DBG() around Y-to-X fallbacks Signed-off-by: Chris Wilson commit 32bb2c89b8cab92f16af32e7a084bfefa406661d Author: Chris Wilson Date: Wed Dec 14 16:28:54 2011 +0000 sna: Check allocation of pixman_image_t And just fail to perform the copy, clearing the dst instead. Signed-off-by: Chris Wilson commit 37c525a11c0fab4e279e80ad24e5cf791239e005 Author: Chris Wilson Date: Wed Dec 14 16:26:46 2011 +0000 sna: Skip glyphs if we fail to allocate pixel data for them Signed-off-by: Chris Wilson commit 6a8188bb4dd0eba58e29803566023ad5c3d4d58a Author: Chris Wilson Date: Wed Dec 14 16:24:54 2011 +0000 sna: Close any handles after bo allocation failures Signed-off-by: Chris Wilson commit 43a22743124a83310379122d509c35840b583b2e Author: Chris Wilson Date: Wed Dec 14 13:26:05 2011 +0000 sna: Use the provided bo for blitting rather than assume priv->gpu_bo Reported-by: nkalkhof@web.de Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43802 Signed-off-by: Chris Wilson commit 30f5ee11f8ec3688807bbaded92561e96f9a439b Author: Chris Wilson Date: Wed Dec 14 12:30:40 2011 +0000 sna: Use a static request and synchronous rendering in case of malloc failure Signed-off-by: Chris Wilson commit 23fb2cebbe1d4d7df7403a64635339efa08a4dfe Author: Chris Wilson Date: Wed Dec 14 12:34:03 2011 +0000 sna/blt: Add a missing allocation check upon the source CPU pixmap Signed-off-by: Chris Wilson commit 5b0f3ff9a83b7cc932e96400999dc5e4a49369cb Author: Chris Wilson Date: Wed Dec 14 12:23:04 2011 +0000 sna/damage: Guard against malloc failures In the event of failure, we choose to loose track of the damage and choose rendering corruption over crashing. Signed-off-by: Chris Wilson commit f8b9a2a5473efd0fefb3d2878e70c24c320245ca Author: Zhigang Gong Date: Tue Dec 13 21:46:14 2011 +0800 uxa/glamor: Enable the rest of the glamor rendering routines This commit hooks up all the remaining rendering routines to call into glamor; the takeover is nearly complete! When tested with the latest glamor master branch, it passes rendercheck. One thing need to be pointed out is the picture's handling. Pictures support many different color formats, but glamor's texture only support a few color formats. And the most common scenario is that we create a pixmap with a color depth and then attach it to a picture which has a specific color format with the same color depth. But there is no way to change a texture's internal format after the texture was allocated. If you do that, the OpenGL will allocate a new texture. And then the glamor side and UXA side will be inconsitent. So for all the picture related operations, we can't fallback to UXA path directly, even it is rather a straight forward operation. So for the get_image, Addtraps.., we have to add wrappers function for them to jump into glamor firstly. Signed-off-by: Zhigang Gong [ickle: prefer access; ok = glamor(); finish; if (!ok) goto fallback; return; ] Signed-off-by: Chris Wilson commit d5456e40d9bebc9d6a121abdb124a90a2a8a6958 Author: Chris Wilson Date: Wed Dec 14 10:41:33 2011 +0000 uxa/glamor: Silence a compiler warning for some unused code intel_glamor.c: In function 'intel_glamor_create_screen_image': intel_glamor.c:192:12: warning: variable 'pixmap' set but not used [-Wunused-but-set-variable] Signed-off-by: Chris Wilson commit 4f1a99a70e76ea5637c5ee8226b2e52a464f5948 Author: Chris Wilson Date: Tue Dec 13 19:44:15 2011 +0000 sna: Protect against deferred malloc failures for pixel data As we now defer the allocation of pixel data until first use, it can fail in the middle of a rendering routine. In order to prevent chasing us passing a NULL pointer into the fallback routines, we need to propagate the failure from the malloc and suppress the failure, discarding the operation, which is less than ideal. Signed-off-by: Chris Wilson commit d2c6d950ed2c5882e7d501b6974e72be4d6da8a8 Author: Chris Wilson Date: Tue Dec 13 15:04:10 2011 +0000 sna: Mark upload buffers as unaccessible upon submission Use valgrind to catch use-after-finish bugs. Signed-off-by: Chris Wilson commit e39ea29bcc5d1d76dd5cb3db555516ef848647a5 Author: Chris Wilson Date: Tue Dec 13 13:18:56 2011 +0000 sna: Allow the debugger to map bo from the batch during kgem_submit() Signed-off-by: Chris Wilson commit 2fabb5068d36ce97242844066b5121fcf6d9c24e Author: Chris Wilson Date: Tue Dec 13 13:15:51 2011 +0000 sna: Debug fixup for non-LLC systems The cpu bo is only allocated on LLC systems, so do avoid the NULL deref on debugging for others. Signed-off-by: Chris Wilson commit e037379c8e57f8e42956863c10a4e2b18057a6d5 Author: Chris Wilson Date: Tue Dec 13 13:14:15 2011 +0000 sna: Fix a debugging assert The bo is allowed to be NULL, so defer the assert until after it is known to be non-NULL. Signed-off-by: Chris Wilson commit 5d5b2b8ee203ae2274fc7d13ed38d2945facca9e Author: Chris Wilson Date: Tue Dec 13 11:30:05 2011 +0000 uxa: Cap the maximum number of VMA cached Since we can not keep an unlimited number of vma cached due to the hard per-process limits on the number of mappings and recreating mappings is slow due to excruciatingly slow GTT pagefaults, we need to compromise and keep a small MRU cache of inactive mmaps. This uses the new API in libdrm-2.4.29 to specify the limit upon the VMA cache maintained by libdrm. Signed-off-by: Chris Wilson commit 1128825efb3179a5a5d951fa24db6b769ee41219 Author: Chris Wilson Date: Tue Dec 13 11:20:25 2011 +0000 uxa: Wakeup 3s after the last rendering to reap the bo-cache libdrm expires its bo 2s after entry into the cache, but we need to free a buffer to trigger the reaper. So schedule a timer event to trigger 3s after the last rendering is submitted to free any resident bo during long periods of idleness. Signed-off-by: Chris Wilson commit db7c9e8561afcc1ca7ab16b3bf2d5b49938e26d9 Author: Chris Wilson Date: Tue Dec 13 10:05:25 2011 +0000 configure: Link the extra valgrind debugging to --enable-debug Signed-off-by: Chris Wilson commit d02dc0fd84194b6acc607d17e4396dd1fdc681f8 Author: Chris Wilson Date: Tue Dec 13 01:18:59 2011 +0000 sna: Set the refcnt on the replacement bo The paranoia wasn't in vain. Signed-off-by: Chris Wilson commit 7472db8c8c9994798ea36de1ce2f51725d8b387a Author: Chris Wilson Date: Mon Dec 12 21:55:04 2011 +0000 sna: Double-check that the submitted buffers were not purged More paranoia is good for the soul. Signed-off-by: Chris Wilson commit 0bbd6a08fe485c80bd5c9a1b7027618a03a26f84 Author: Chris Wilson Date: Mon Dec 12 21:52:56 2011 +0000 sna/gen2: Tidy checking against too large pixmaps for the 3D pipeline Signed-off-by: Chris Wilson commit b392474f3a052fefc410383725b64f0fdbe7788e Author: Chris Wilson Date: Mon Dec 12 21:28:09 2011 +0000 sna: Force a suitable minimum stride for 3D temporaries Signed-off-by: Chris Wilson commit 3c22baaba97a89d24e507e3d832d53d764902040 Author: Chris Wilson Date: Mon Dec 12 21:16:56 2011 +0000 sna/gen2: Check for unhandled pitches in the render pipeline Signed-off-by: Chris Wilson commit f6a30df8dc75852d1f6daeabe4511b275594e52e Author: Chris Wilson Date: Mon Dec 12 20:40:29 2011 +0000 sna: Enable memcpy uploads to SHM pixmaps Signed-off-by: Chris Wilson commit 3c163d105e964a1084d665500ef917254d8f2179 Author: Chris Wilson Date: Mon Dec 12 16:26:13 2011 +0000 sna: Use the CPU bo as a render source if compatible and no GPU bo This is principally to catch the cases of compositing after a fresh PutImage. Signed-off-by: Chris Wilson commit c481bec356b2e40e66a000dbaaf261bf7aae930d Author: Chris Wilson Date: Mon Dec 12 14:10:57 2011 +0000 sna: Experiment with creating the CPU pixmap using an LLC BO A poor cousin to vmap is to instead allocate snooped bo and use a CPU mapping for zero-copy uploads into GPU resident memory. For maximum performance, we still need tiled GPU buffers so CPU bo are only useful in situations where we are frequently migrating data. Signed-off-by: Chris Wilson commit 6c9aa6f9cf8e59ca6aa1866b83690a1de8cfb757 Author: Chris Wilson Date: Mon Dec 12 11:47:25 2011 +0000 sna: Defer allocation of memory for larger pixmap until first use In the happy scenario where the pixmap only resides upon the GPU we can forgo the CPU allocation entirely. The goal is to reduce the number of needless mmaps performed by the system memory allocator and reduce overall memory consumption. Signed-off-by: Chris Wilson commit 4b48d28f6e85f345730cb7242f2152ac115b267a Author: Chris Wilson Date: Mon Dec 12 10:52:34 2011 +0000 sna: Fix a typo, end statements with semi-colons Signed-off-by: Chris Wilson commit 4d20798c7871ffe8581e2cf509b6aa2e40b3ae5f Author: Chris Wilson Date: Mon Dec 12 10:10:16 2011 +0000 sna: We need to remap the gpu_only mmap prior to every use Since the VMA may be reaped at any time whilst the mapping is idle. Signed-off-by: Chris Wilson commit 2682308c10c9622247409dee9c6bb6a862b73c3f Author: Chris Wilson Date: Mon Dec 12 09:24:11 2011 +0000 sna: Remove bo transference for whole XCopyArea In benchmarking firefox this performs whose - it would appear the sources are indeed used more often than not. Signed-off-by: Chris Wilson commit 7703424222f8bd08450c1b109ea64c6b0ee901d8 Author: Chris Wilson Date: Mon Dec 12 00:15:10 2011 +0000 sna/gen6: Only use CPU bo for a render target if untiled For large render targets, we prefer to use tiled bo in order to avoid severe performance degradation. However, if we don't have a GPU bo but do have a CPU bo and the operation would be untiled, then simply use the CPU bo. Signed-off-by: Chris Wilson commit a92a41ba3291a4304948ed1961d1242ca7d3981e Author: Chris Wilson Date: Sun Dec 11 19:03:54 2011 +0000 sna/gen6: Tidy the usage of the max pipeline size Signed-off-by: Chris Wilson commit e9e6d6f7c8af300716c64b041f96c09d58b3eaa2 Author: Chris Wilson Date: Sun Dec 11 17:25:46 2011 +0000 sna/gen3: Move the video dst_bo to make the conditional clearer Signed-off-by: Chris Wilson commit 118ef0781c5064aa0696af59a5dac67091b10046 Author: Chris Wilson Date: Sun Dec 11 16:56:53 2011 +0000 sna/composite: Make the check for a no-op earlier and clearer Signed-off-by: Chris Wilson commit 2674ef864c93034d728ee11ee0b7ab96cd4cba57 Author: Chris Wilson Date: Sun Dec 11 16:23:13 2011 +0000 sna: Enable hooking up of valgrind during debugging Signed-off-by: Chris Wilson commit c83fd4e24de187700bc125037cd9ae311514213e Author: Chris Wilson Date: Sun Dec 11 16:14:38 2011 +0000 sna: Add some more debug messages for VMA caching Signed-off-by: Chris Wilson commit 3ae7fb918a2ca47d3944521a1ae30143dcb14eeb Author: Chris Wilson Date: Sun Dec 11 13:52:42 2011 +0000 sna: Restrict pitch alignment on 945gm to 64 bytes In theory we should be able to disable dual-stream mode and so be subject to much looser restrictions (such as the pitch need only be dword aligned). However, achieving single-stream mode seems quite difficult! Reported-by: Paul Neumann References: https://bugs.freedesktop.org/show_bug.cgi?id=43706 Signed-off-by: Chris Wilson commit 2f35d77cd07a4974051c922f4809ce44939b0556 Author: Chris Wilson Date: Sun Dec 11 13:37:18 2011 +0000 sna: Update computation of untiled pitch to cater for CREATE_SCANOUT Signed-off-by: Chris Wilson commit 5a0139487f5a5caada56121395aa166fc9f08663 Author: Chris Wilson Date: Sun Dec 11 12:34:40 2011 +0000 sna/gen3: Ensure that depth read/writes are disabled before first use Our goal is to achieve "single-stream" rendering where the entire RenderCache is allocated to the colour buffer (rather than split between colour and depth). In theory all that is required is for the pipeline not to reference the depth buffer at all, however it is not made clear when that evaluation is made. Signed-off-by: Chris Wilson commit a02bbd8700690c98c6cf3ae98dd7ee3da9887b0a Author: Chris Wilson Date: Sun Dec 11 10:34:37 2011 +0000 sna: Only transfer the bo if the src/dst are of matching size If the src replaces the dst, it could just be a much larger pixmap! Signed-off-by: Chris Wilson commit 43a99648635fa76e0d6d71acaeb6ba919d24eb26 Author: Chris Wilson Date: Sun Dec 11 10:30:48 2011 +0000 sna: Only transfer unpinned buffers Signed-off-by: Chris Wilson commit eb859f644633ee716083d253a5b7ff95163380e5 Author: Chris Wilson Date: Sun Dec 11 01:34:05 2011 +0000 uxa/video: Correct the offset of the binding table in the surface buffer The binding table is intended to be after all the surface descriptions, so make sure we write it with the appropriate offset into the buffer. Fixes regression from 699888a64 (uxa/video: Use the common bo allocations and upload) Reported-by: Cyril Brulebois Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43704 Signed-off-by: Chris Wilson commit 051a18063df075536cb1ac0dc4dfc3c1306ab74e Author: Chris Wilson Date: Sat Dec 10 22:45:25 2011 +0000 sna: Implement a VMA cache A VMA cache appears unavoidable thanks to compiz and an excrutiatingly slow GTT pagefault, though it does look like it will be ineffectual during everyday usage. Compiz (and presumably other compositing managers) appears to be undoing all the pagefault minimisation as demonstrated on gen5 with large XPutImage. It also appears the CPU to memory bandwidth ratio plays a crucial role in determining whether going straight to GTT or through the CPU cache is a win - so no trivial heuristic. x11perf -putimage10 -putimage500 on i5-2467m: Before: bare: 1150,000 2,410 compiz: 438,000 2,670 After: bare: 1190,000 2,730 compiz: 437,000 2,690 UXA: bare: 658,000 2,670 compiz: 389,000 2,520 On i3-330m Before: bare: 537,000 1,080 compiz: 263,000 398 After: bare: 606,000 1,360 compiz: 203,000 985 UXA: bare: 294,000 1,070 compiz: 197,000 821 On pnv: Before: bare: 179,000 213 compiz: 106,000 123 After: bare: 181,000 246 compiz: 103,000 197 UXA: bare: 114,000 312 compiz: 75,700 191 Reported-by: Michael Larabel Signed-off-by: Chris Wilson commit 735a15208dd600eefa3090f344186df9cac0462d Author: Chris Wilson Date: Sat Dec 10 23:45:56 2011 +0000 sna/gen5: Remove a redundant format check Signed-off-by: Chris Wilson commit c5584252c34b792313578cc31d56248d8990b571 Author: Chris Wilson Date: Sat Dec 10 23:34:16 2011 +0000 sna: Remember to assign a new unique id for the replaced bo Missed from the previous patch. Signed-off-by: Chris Wilson commit 9c764dc13be40fc14238a4c130f7d8b44dbdf7db Author: Chris Wilson Date: Sat Dec 10 22:38:57 2011 +0000 sna: Be more pessimistic with CPU sources Try to avoid a few more unnecessary context switches. Signed-off-by: Chris Wilson commit 358aaef6dbff0737f026046b0c9a2e21d8cfdf2a Author: Chris Wilson Date: Sat Dec 10 22:38:33 2011 +0000 sna/dri: Prefer using the BLT for DRICopyRegion on pre-SNB Signed-off-by: Chris Wilson commit c295ad8da91e39c8fffa540901097651df5d24b2 Author: Chris Wilson Date: Sat Dec 10 22:37:31 2011 +0000 sna: Transfer the whole bo for a replacement XCopyArea If we are copying over the entire source onto the destination,just copy across the GPU bo. This is often used for caching images as pixmaps. Signed-off-by: Chris Wilson commit ece7fc8afeb8eefcf0ad1a054f02e7fac8db6327 Author: Chris Wilson Date: Sat Dec 10 21:04:08 2011 +0000 sna: Only use the 64-byte pitch alignment for scanout Signed-off-by: Chris Wilson commit b3816cf3a99d23d0c3ab4cd716b24ea544a07283 Author: Chris Wilson Date: Sat Dec 10 20:32:56 2011 +0000 sna: Remove assertions that external bo are not busy We have to be careful to assume bo via exposed are under our full control, in particular not to assert their state. :( Signed-off-by: Chris Wilson commit b5a6bc9e33a2797c926969e1f98b4a9a796248c5 Author: Chris Wilson Date: Sat Dec 10 13:14:45 2011 +0000 sna/gen[23]: Fixup render targets with pitches below hw minimum gen2/3 have a restriction that the 3D pipeline cannot render to a pixmap with a pitch less than 8/16 respectively. Rather than mandating all pixmaps to be created with a stride greater than 16, fixup the bo for the rare occasions when it is necessary. Reported-by: Paul Neumann Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43688 Signed-off-by: Chris Wilson commit c0dab7b1cf17fe751c86ad2b3fabce682eb50366 Author: Chris Wilson Date: Sat Dec 10 12:46:46 2011 +0000 sna/trapezoids: Try to render traps onto a8 destinations in place Signed-off-by: Chris Wilson commit c73b14cabb059075af0f1727847074a89941c9df Author: Chris Wilson Date: Sat Dec 10 11:41:18 2011 +0000 sna/trapezoids: First try the scan converter for fallbacks Signed-off-by: Chris Wilson commit 22d9bc0bc1ccd869ebcc93f9a57efc8a85840cb8 Author: Chris Wilson Date: Fri Dec 9 23:48:11 2011 +0000 sna: Use a single definition for the inactive cache timeout And share it between the timer and the expiration function, just to simplify the code. Signed-off-by: Chris Wilson commit eb3e04d960cd749adf696ba15c501a17cd1a184e Author: Chris Wilson Date: Fri Dec 9 23:36:56 2011 +0000 sna: Fallback to ordinary monotonic clock if coarse is not supported Signed-off-by: Chris Wilson commit 1c202cc0746541b541d0af22eab5c9798d229a69 Author: Chris Wilson Date: Fri Dec 9 17:25:19 2011 +0000 sna: s/MONOTONICE/MONOTONIC/ A late addition to be flexible for compiling on different systems heralded its doom. Signed-off-by: Chris Wilson commit c51626ccb660c777da8d84bb425de0d590864448 Author: Chris Wilson Date: Fri Dec 9 15:45:29 2011 +0000 sna: Use the coarse monotonic clock to coalesce wakeup events For the long interval events (such as expiring the caches), we do not need precise timing and so can use a coarse timer to allow the system to coalesce and reduce wakeup events. Signed-off-by: Chris Wilson commit c22197f25bc0419d9f2abfcc978df5ef439feb47 Author: Chris Wilson Date: Fri Dec 9 15:12:03 2011 +0000 sna: Discard bo for idle private pixmaps If a pixmap lies around for a couple of minutes not being used, it is unlikely to be used again in the near future. Reap the GPU buffers of any of those idle pixmaps (copying to a more compact buffer in system memory) in order to free up resources for use elsewhere. Any object that is exposed via DRI is obviously exempt from this reaping. Signed-off-by: Chris Wilson commit 429a36f7481b9bfd5ed137642d2916d69a713557 Author: Chris Wilson Date: Fri Dec 9 09:54:12 2011 +0000 uxa: Fix clip processing for uxa_fill_spans() Fixes regression from e0066e77e026b0dd0daa0c3765473c7d63aa6753 (uxa: Simplify Composite solid acceleration for spans by only clipping once) [2.15.901] Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43649 Signed-off-by: Chris Wilson commit 699888a6410b0699c69a7f8a8d82dc4fde6fcc7f Author: Chris Wilson Date: Thu Dec 8 16:55:20 2011 +0000 uxa/video: Use the common bo allocations and upload In order to avoid inconsistent usage of coherency domains and to avoid completely unnecessary clflushing during video playback, use the same buffer allocation and upload functions as the rest of the driver. Reported-by: Christophe Roland Bugzilla: http://bugs.debian.org/cgi-bin/bugreport.cgi?msg=60;bug=651316 Signed-off-by: Chris Wilson commit 706d3a97bd3049e60c08d121a3d4d81c03029a87 Author: Chris Wilson Date: Thu Dec 8 18:05:14 2011 +0000 sna/trapezoids: Fix detection of rectilinearity after projection A typo confused left and right, rejecting true vertical edges, and worse might have incurred false positives. Signed-off-by: Chris Wilson commit bc081420a5d6d28c4e6fd5527a70caaa3a8361c4 Author: Simon Que Date: Wed Dec 7 16:20:41 2011 -0800 xf86-video-intel: change order of DPMS operations The operations when setting dpms on should be in the order opposite of what's done when setting dpms off. This is because of potentially conflicting effects: ~ drmModeConnectoSetProperty() enables/disables the backlight driver. Some backlight drivers such as intel_backlight set the backlight to 0 when disabled and to max when enabled. ~ intel_output_dpms_backlight() saves the backlight value when turning DPMS off and restores it when turning DPMS on. Here's the current order of operations: xset dpms force off (backlight is nonzero) drmModeConnectoSetProperty(DPMSModeOff) kernel: disable backlight, backlight=0 intel_output_dpms_backlight(DPMSModeOff) save backlight value (0) <-- it has been set to 0 by kernel set backlight to 0 xset dpms force on drmModeConnectoSetProperty(DPMSModeOn) kernel: enable backlight, backlight=max intel_output_dpms_backlight(DPMSModeOn) set backlight to saved value (0) The correct way to do this would be to reverse the operations during xset dpms force off: intel_output_dpms_backlight(DPMSModeOff) save backlight value (nonzero) set backlight to 0 drmModeConnectoSetProperty(DPMSModeOff) kernel: enable backlight, backlight=0 This restores the saved nonzero backlight value during the force on. Signed-off-by: Simon Que Signed-off-by: Chris Wilson commit 84aaf1537cbd29e163346d03debc39f4623c69eb Author: Chris Wilson Date: Thu Dec 8 12:36:08 2011 +0000 sna/gen7: Reduce dst readbacks for unsupported sources Signed-off-by: Chris Wilson commit 440ac68ec074e82818713773f3e2cb5d363862aa Author: Chris Wilson Date: Thu Dec 8 12:35:54 2011 +0000 sna/gen6: Reduce dst readbacks for unsupported sources Signed-off-by: Chris Wilson commit bc68211d18356c4a0e011ac360665d600de9fa30 Author: Chris Wilson Date: Thu Dec 8 12:35:39 2011 +0000 sna/gen5: Reduce dst readbacks for unsupported sources Signed-off-by: Chris Wilson commit a5df7c28e458c2f2b173fe5d745f447087eb8c6f Author: Chris Wilson Date: Thu Dec 8 12:35:22 2011 +0000 sna/gen4: Reduce dst readbacks for unsupported sources Signed-off-by: Chris Wilson commit cc8cab649c0a4788803d2c62f186142ec7152b89 Author: Chris Wilson Date: Thu Dec 8 12:34:58 2011 +0000 sna/gen3: Reduce readbacks on dst for unsupported sources Signed-off-by: Chris Wilson commit e5bc0c823b8fd46b3f851c5e9a4b4de39cda9a91 Author: Chris Wilson Date: Thu Dec 8 12:34:34 2011 +0000 sna/gen2: Avoid readbacks for unsupported sources Signed-off-by: Chris Wilson commit 874c722c86b06ac0aa10e6943d3faf76fcb751be Author: Chris Wilson Date: Thu Dec 8 12:07:28 2011 +0000 sna: Beware flushing partial buffers before they are written A partial buffer is considered finished upon the next batch submission, so one needs to be careful that it is completely written to before such an event is triggered. move-to-cpu is such a trigger as demonstrated by the picture fixup routine for handling convolution filters. Reported-by: Victor Machado Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43607 Signed-off-by: Chris Wilson commit 6ccb114a7e685c69fb388ebd119393455b315c36 Author: Chris Wilson Date: Wed Dec 7 21:09:31 2011 +0000 sna: Prefer to use our pixmap upload paths Signed-off-by: Chris Wilson commit 101942d41df7efaa6103e31e738775fafdb63159 Author: Chris Wilson Date: Mon Dec 5 10:11:04 2011 +0000 uxa: Unmap the buffer after swrast Otherwise we may exhaust the per-process vma limit and cause applications to stop rendering and eventually crash the X server. Will only work in conjunction with a new libdrm (2.4.28) and commit c549a77c (intel: Unmap buffers during drm_intel_gem_bo_unmap) in particular. Reported-by: nobled@dreamwidth.org References: https://bugs.freedesktop.org/show_bug.cgi?id=43075 References: https://bugs.freedesktop.org/show_bug.cgi?id=40066 Signed-off-by: Chris Wilson commit b424b10e771b1d3d041efdd2b77f576251364744 Author: Chris Wilson Date: Sun Dec 4 13:01:10 2011 +0000 sna: use tight pitches for a8 As we never use these with a depth nor attach them to scanout, we can safely relax the multiple-of-64 byte pitch restriction. In the unlikely event that we do need A8 surfaces with depthbuffers, this is broken... Signed-off-by: Chris Wilson commit 46c7df803881fa0e733c0d0fdd37567ebdccd6ac Author: Chris Wilson Date: Sat Dec 3 18:02:00 2011 +0000 sna: Remove one redundant retire There is no need to retire immediately after a batch and no indication that it will be useful. Signed-off-by: Chris Wilson commit b99c6b13ebba9521333b8dd5982ac37b6e244b54 Author: Chris Wilson Date: Sat Dec 3 18:01:11 2011 +0000 sna: Pass the current value of the batch offset to the kernel relocator Signed-off-by: Chris Wilson commit 735219cd59e6184a6622d3d429a704ca3f58b9cd Author: Chris Wilson Date: Fri Dec 2 10:42:00 2011 +0000 uxa: Ensure that we can fallback with all of (src, mask, dst) as GTT mappings Signed-off-by: Chris Wilson commit f6c82c73b673ec3c9cce432fe38d5e0076234efd Author: Chris Wilson Date: Fri Dec 2 10:34:10 2011 +0000 uxa: Fix runtime linking of previous commit So much for relying on compiler warnings. Signed-off-by: Chris Wilson commit 85d3dc5910a2eea3a10b822e01443e11eaae9291 Author: Chris Wilson Date: Fri Dec 2 10:22:51 2011 +0000 uxa: Reset size limits based on AGP size The basis for the constraints are what we can map into the aperture for direct writing with the CPU, so use the size of the mappable region as opposed to the size of the total GTT. Signed-off-by: Chris Wilson commit e55198746102afb7427f577bd5bfc76667438da9 Author: Chris Wilson Date: Thu Dec 1 13:49:03 2011 +0000 sna: Reuse the full size of an old handle for io Signed-off-by: Chris Wilson commit c5632369cbd6473304c06e4230347abbe46513ec Author: Chris Wilson Date: Thu Dec 1 13:23:56 2011 +0000 sna: Move the preservation of the io handle into the common destroy path In order to capture and reuse all io buffers. Signed-off-by: Chris Wilson commit 95f4da647a4055545b09cae0834df0fa2127a458 Author: Chris Wilson Date: Wed Nov 30 11:59:31 2011 +0000 sna: Align pwrite to transfer whole cachelines Daniel claims that this is will be faster, or will be once he has completed rewriting pwrite! Signed-off-by: Chris Wilson commit ecd6cca617ac29cf2b1b2a4d33fca19b84fea2a9 Author: Chris Wilson Date: Tue Nov 29 19:27:46 2011 +0000 sna/gen5: Handle cpu-bo for render targets Signed-off-by: Chris Wilson commit d8f2e87473eae81e07922271b5f608fdceb4ae5e Author: Chris Wilson Date: Tue Nov 29 10:59:48 2011 +0000 sna/render: Fix check for "migrate whole pixmap" The whole pixmap means the sample covers the full width and height, not just either! Signed-off-by: Chris Wilson commit 20e57914089d6959529ad222e04853f5ffcb364f Author: Chris Wilson Date: Tue Nov 29 10:59:20 2011 +0000 sna: Fix assertion around flushing of mmap(PROT_READ) Signed-off-by: Chris Wilson commit 56155c91afb5a0dcafe4dbc22ade3b7c0c977674 Author: Chris Wilson Date: Tue Nov 29 10:27:18 2011 +0000 sna/gen6: Set the batch mode prior to checking limits and flushing If we change contexts, then we will submit the batch obsoleting the earlier resource checks. Signed-off-by: Chris Wilson commit 5b1e9e15738b9001346ab6e0166f861ce308008e Author: Chris Wilson Date: Mon Nov 28 22:01:00 2011 +0000 sna: Always reduce tiling for thin pixmaps Benchmarking on the current code base, says this is now a win. A reversal of older benchmarks, so expect further tuning. Signed-off-by: Chris Wilson commit dacb301c05dd1686e788d6ed8101887c27968389 Author: Chris Wilson Date: Mon Nov 28 18:59:10 2011 +0000 sna: Use the correct pitch when creating an untiled active buffer Signed-off-by: Chris Wilson commit 8657128fa7e758a2dde93340d6e58928d5f11255 Author: Chris Wilson Date: Mon Nov 28 18:22:01 2011 +0000 sna: Pass the pixmap to sna_replace() Signed-off-by: Chris Wilson commit 4e38d22105da2bd97db005dc505e75dcd22291d3 Author: Chris Wilson Date: Mon Nov 28 18:21:19 2011 +0000 sna: Tidy kgem_choose_tiling() Reduce the calls to compute the surface size down to one. Signed-off-by: Chris Wilson commit 04f47e52da06cc20c2a266da92ba3f8e961433cb Author: Chris Wilson Date: Sun Nov 27 17:45:05 2011 +0000 sna: Don't try to guess when the kernel believes a buffer to be flushed Signed-off-by: Chris Wilson commit d87c332653001db4413c7dd81116ee03830db491 Author: Chris Wilson Date: Sun Nov 27 10:18:11 2011 +0000 sna: Eliminate empty glyphs during format conversion Signed-off-by: Chris Wilson commit 413776c79ee7c55b3a48b1ed29af63f8cea39337 Author: Chris Wilson Date: Sun Nov 27 10:17:40 2011 +0000 sna: Use 64-bit mov (when available) for bitmap glyphs Signed-off-by: Chris Wilson commit 507f99eba2c72bc109c9267c698f3d5f9fb6f539 Author: Chris Wilson Date: Sat Nov 26 17:57:57 2011 +0000 sna/gen6+: Only use BLT if the untiled bo will cause per-pixel TLB misses i.e. only force the BLT if using the sampler is going to be incredibly slow. Signed-off-by: Chris Wilson commit f11e9f189111f53ace20381a0bd4f559ccd3605f Author: Chris Wilson Date: Sat Nov 26 17:12:03 2011 +0000 sna: Only check for all-damaged CPU bo if we may discard the GPU bo The goal of the optimisation is to discard the GPU bo early, so we can skip the extra damage reduction if there is no gpu bo. Signed-off-by: Chris Wilson commit ccee730e45d8c2490e578d0b85d9e76f3a44f76f Author: Chris Wilson Date: Sat Nov 26 16:47:04 2011 +0000 sna: Discard the gpu-bo is the entire pixmap is written to by the CPU. Signed-off-by: Chris Wilson commit df68b20a17e28c0898226f970315bc8e515e9f56 Author: Chris Wilson Date: Sat Nov 26 16:45:22 2011 +0000 sna: Disabling tiling if the bo would be smaller than a page Once again experiment with untiled smalled buffers. Signed-off-by: Chris Wilson commit d88ad2e60ac3d9f5608db1653a7af789f7f1b446 Author: Chris Wilson Date: Sun Nov 27 09:01:09 2011 +0000 sna: Round up stride to alignment for TLB miss checking We want to avoid the condition of reducing the tiling mode (when reusing an active untiled buffer in preference to creating a new) for a wide buffer when doing will force a TLB miss on each sample. Signed-off-by: Chris Wilson commit 16f5e224dcfd97012b38ee1af6c72dbe3c0f3304 Author: Chris Wilson Date: Fri Nov 25 16:21:41 2011 +0000 sna: Avoid the double application of drawable offsets for tiled spans As we handle tiled spans indirectly, we need to avoid applying the drawable offsets twice (once in the mi layer generating the spans, and then once more in the tiled rect renderer). Reported-by: Ulrich Müller Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43245 Signed-off-by: Chris Wilson commit e06789871b5cdf0c5535d46622b8ec3f85eb4386 Author: Chris Wilson Date: Fri Nov 25 15:24:29 2011 +0000 sna/gen5+: Prefer using the BLT when either src or dst is untiled The cost of the TLB miss on every sample far outweighs the impact of the context (and ring) switch. Signed-off-by: Chris Wilson commit d0cce0da71fae275a0f1be33b2bebad9d41212c9 Author: Chris Wilson Date: Fri Nov 25 13:16:37 2011 +0000 sna/gen5: Use the same prefer_blt_fill() switch for fill-one Signed-off-by: Chris Wilson commit 7ad2d3b695115439f685a12d43ab240cf5e18ef7 Author: Chris Wilson Date: Fri Nov 25 09:44:21 2011 +0000 sna/gen2: Wire up NO_COMPOSITE for debugging Signed-off-by: Chris Wilson commit 394ad51c421f41e4bc6545fcee10b1757cb5f518 Author: Chris Wilson Date: Thu Nov 24 17:35:36 2011 +0000 sna/video: Check for Xv presence in the server before initialising Signed-off-by: Chris Wilson commit a2fab9560ec5dd6efc352917fe7626e59bed69e3 Author: Chris Wilson Date: Thu Nov 24 17:14:08 2011 +0000 sna: Downgrade Y-tiling if we must use the BLT There are many operations, usually the core drawing acceleration, where the BLT is much more preferable than using the CPU. However, the BLT is limited to only using X-tiling, so if we encounter a Y-tiled pixmap target we need to recreate it as X-tiling before proceeding. Hopefully, the pixmap is then kept around and rendered multiple times to amoritize the cost of the copy. Signed-off-by: Chris Wilson commit e0a4492c8b51d6ae09cde0b895da6177284e95e0 Author: Chris Wilson Date: Thu Nov 24 17:00:18 2011 +0000 sna: Use Y-tiling for source pixmaps Y-tiling is slightly faster with RENDER operations, so attempt to allocate source-only pixmaps using this tiling mode. Actually using Y-tiling is a delicate balance because it then prevents the use of the BLT. For instance, enabling Y-tiling by default gives a 30% performance improvement on the fish-demo (compositing benchmark) at 2560x1440 on Ironlake but regresses tiger-demo by 2x (spans benchmark). So experiment with this compromise and allow for changing the default tiling. Signed-off-by: Chris Wilson commit 7e4a1b7ed25a881113e197acde62b07f26f2d6ef Author: Chris Wilson Date: Thu Nov 24 11:34:51 2011 +0000 src/sna/gen5: Replace the precompiled shaders Take advantage of a couple of new instructions introduced with Cantiga to reduce the instruction count inside the shaders and improve performance by around 10% in the fish-demo. Signed-off-by: Chris Wilson commit 67ea8808d7f0ce47733a66d8708dcfa0dbec14c1 Author: Chris Wilson Date: Thu Nov 24 11:45:38 2011 +0000 sna/gen5: Remove the unused SIP Signed-off-by: Chris Wilson commit 499ed74e979d99bbe1670ad54875fba6055fce5d Author: Chris Wilson Date: Thu Nov 24 11:02:01 2011 +0000 sna: Tweak cancellation of deferred flush The goal is to keep running until the tick after every stops, irrespective of forced flushes. Signed-off-by: Chris Wilson commit 35239d207c38e1ae0cb843ed7a72621711cb001e Author: Chris Wilson Date: Thu Nov 24 02:51:51 2011 +0000 sna: Always call retire following a mmap Signed-off-by: Chris Wilson commit 11e0b456de93dae9f7ba4298d8a32b6b60edc503 Author: Chris Wilson Date: Thu Nov 24 03:04:09 2011 +0000 sna/gen5: Use the BLT rather than flush when copying Signed-off-by: Chris Wilson commit a44663af9335136cba59b12078c6717992283b62 Author: Chris Wilson Date: Thu Nov 24 02:54:43 2011 +0000 sna/gen4,gen5: Fix typo and only emit a flush if the video source is dirty Signed-off-by: Chris Wilson commit 4a27dd287cdb85d80ceaf6b54e7de0c17e01c00d Author: Chris Wilson Date: Mon Nov 21 12:56:32 2011 +0000 uxa: Make the glamor/uxa transition more verbose And so hopefully make it clearer. In the process we restore the flushing behaviour for UXA back to before the glamor intervention. Signed-off-by: Chris Wilson commit b0d64a9567d9cb749568448e4ce090b187308028 Author: Chris Wilson Date: Mon Nov 21 11:49:49 2011 +0000 sna: Correct end-point adjustment for stippling PolySegment Reported-by: Roman Jarosz Reported-by: da_fox@mad.scientist.com Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43134 Signed-off-by: Chris Wilson commit a5353a7c1d1f02b887f61f543c8e4bd4e8bd104f Author: Chris Wilson Date: Mon Nov 21 10:58:17 2011 +0000 sna/gen4: Use the special composite rect emission for fill_one() In order to workaround a bug in the shaders on gen4, we need to flush the pipeline after every rectangle. The recently introduced fill-one mechanism for gen4, missed this vital step triggering a random hang with an otherwise sane batchbuffer (the missing flush is hard to spot!). Fixes regression from 86f99379ee5 (sna/gen4: Add fill-one). Reported-by: Albert Damen Reported-by: Fryderyk Dziarmagowski Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43083 Signed-off-by: Chris Wilson commit 3b9479dc39d32fd97f80c1e5e0fac67d36ee5e40 Author: Chris Wilson Date: Sat Nov 19 19:12:06 2011 +0000 sna: Micro-optimise sna_push_pixels_solid_blt() Unroll the byte reversal as we know the row length is word aligned. Signed-off-by: Chris Wilson commit 297f7680cfd73ab6eb099adca7aefe00e61c9e85 Author: Chris Wilson Date: Sat Nov 19 16:11:42 2011 +0000 sna: Avoid the NULL damage deref Fixes x11perf -wdcircle100 -time 1 -repeat 1 -rop GXxor Reported-by: Fryderyk Dziarmagowski Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43084 Signed-off-by: Chris Wilson commit 4a86ec5ddbaf119fd9a32ecd963f7238152d7d28 Author: Chris Wilson Date: Sat Nov 19 00:33:28 2011 +0000 sna/gen7: Apply recent SNB tuning as applicable Signed-off-by: Chris Wilson commit fe944d61d53c275a168da2e9515d1e42be81630b Author: Chris Wilson Date: Sat Nov 19 00:02:17 2011 +0000 sna/gen7: Correct shifts for surface state Signed-off-by: Chris Wilson commit 70042400fd95f613b9547d0a72d6288abcc2357d Author: Chris Wilson Date: Fri Nov 18 23:41:33 2011 +0000 sna/gen7: minor tidy of redundant defines Signed-off-by: Chris Wilson commit f938cd15990d1a851b9d92414545dda37f00c5db Author: Chris Wilson Date: Thu Nov 17 13:43:37 2011 +0000 glamor: Include glamor cflags and libs for atypical builds Signed-off-by: Chris Wilson commit eb6148a36a318eb0b4aad865c34b50fedf4b709e Author: Zhigang Gong Date: Wed Nov 16 15:04:37 2011 +0800 uxa/glamor: Turn on glamor for fill spans and rects This commit only enables two glamor functions for uxa_fill_spans and uxa_poly_fill_rects. Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit fbabe60f48006ee664c983082498a863e8beec71 Author: Zhigang Gong Date: Wed Nov 16 15:04:36 2011 +0800 glamor: Initial commit to introduce glamor acceleration. Added one configuration option --enable-glamor to control whether use glamor. Added one new file intel_glamor.c to wrap glamor egl API for intel driver's usage. This commit doesn't really change the driver's control path. It just adds necessary files for glamor and change some configuration. Reviewed-by: Eugeni Dodonov Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit c4c2eb1fae2518578a56a278afeaec66d544439d Author: Zhigang Gong Date: Wed Nov 16 15:04:37 2011 +0800 uxa/glamor: Introduce additional access modes to wrap glamor acceleration Integrate glamor acceleration into UXA framework. Add necessary flushing at the following points: 1. Flush UXA batch buffer before call into glamor. 2. Flush GL operations after return from a glamor function. 3. The point we need to flush UXA batch buffer, we also need to flush GL operations, for example, in intel_flush_callback and couple of places in intel_display.c. Signed-off-by: Zhigang Gong Signed-off-by: Chris Wilson commit ac7df2919b92c9fe47c6745975e1848932f989e8 Author: Chris Wilson Date: Wed Nov 16 23:25:16 2011 +0000 Revert "Disable adding normal RTF modes for an eDP" This reverts commit 212fa9868767637e8f430485eeb522c99e63fd16. The underlying register programming for eDP is now believed to be fixed as of linux-3.1. References: https://bugs.freedesktop.org/show_bug.cgi?id=38012 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=41070 Signed-off-by: Chris Wilson commit 24ece4a87e34c27115ffb55f91294bce2bdf1502 Author: Chris Wilson Date: Wed Nov 16 23:14:25 2011 +0000 configure: Version bump for 2.17.0 release Signed-off-by: Chris Wilson commit 14b3176962bd535933623d8ed77ba286b077d6f0 Author: Chris Wilson Date: Wed Nov 16 23:13:44 2011 +0000 NEWS: Release notes for 2.17.0 Signed-off-by: Chris Wilson commit c259144e3fc52d078b0a78107c38f0f3d3a2bbc1 Author: Chris Wilson Date: Wed Nov 16 10:28:23 2011 +0000 sna: The block handler is passed an indirect pointer to the timeval Signed-off-by: Chris Wilson commit bfd2bb40274d1242001d295a4010211fd51b0fc3 Author: Chris Wilson Date: Tue Nov 15 10:38:09 2011 +0000 sna: Correct dependencies for DRI2 Signed-off-by: Chris Wilson commit 3771387ad11b5842a83e58a4b373c2acdd827bd2 Author: Chris Wilson Date: Tue Nov 15 10:32:34 2011 +0000 Compile out UXA if so desired Signed-off-by: Chris Wilson commit edbeab8c4edf9e0e89d85add485fe659795b6350 Author: Chris Wilson Date: Tue Nov 15 10:17:06 2011 +0000 sna: Reduce and clarify dependencies Signed-off-by: Chris Wilson commit 78d4e99fc916e6477edb01c6f24b69ad73adc552 Author: Chris Wilson Date: Tue Nov 15 11:06:04 2011 +0000 sna: And keep unity happy Rewrite the DRI layer to avoid the various bugs and shortcomings of the Xserver and interfacing with mesa. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=38732 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=39044 Signed-off-by: Chris Wilson commit aac022cbb3342e7027301a3f8c9add8929edbad6 Author: Chris Wilson Date: Mon Nov 14 19:49:29 2011 +0000 sna: Optimise single pixel transfers Surprisingly frequent. Signed-off-by: Chris Wilson commit 5647e2981d81fd2b5cdb22d544f83d2d0f9445c9 Author: Chris Wilson Date: Mon Nov 14 14:11:56 2011 +0000 sna/damage: Always reduce damage for testing PIXMAN_REGION_OUT Reported-by: Clemens Eisserer References: https://bugs.freedesktop.org/show_bug.cgi?id=42414 Signed-off-by: Chris Wilson commit 405b015fe255c75f91b0f3fe93f06bd14faee5a1 Author: Chris Wilson Date: Mon Nov 14 14:05:10 2011 +0000 sna: Be explicit in all sna_damage_contains_box() tests Clarify the exact in/out/maybe expected result when testing. Signed-off-by: Chris Wilson commit 5d23149647b1645057dbd89087691e873494f763 Author: Chris Wilson Date: Mon Nov 14 13:53:37 2011 +0000 sna: Remove redundant 'can_fill_spans()' Spans are almost always accelerated now... Signed-off-by: Chris Wilson commit ae9de984adb275e028e7fbcb8b74a3ce11920eed Author: Chris Wilson Date: Mon Nov 14 13:39:34 2011 +0000 sna: Convert non-FillSolid rectilinear lines to boxes And render using the tiled/stippled rect routines instead. Signed-off-by: Chris Wilson commit e861e816f53989ce3ab3ec89dd1d9a35155f999a Author: Chris Wilson Date: Mon Nov 14 13:39:34 2011 +0000 sna: Convert non-FillSolid rectilinear segments to boxes And render using the tiled/stippled rect routines instead. Signed-off-by: Chris Wilson commit 6fc2928f578ae1e278c968d5c778d99d56bd667a Author: Chris Wilson Date: Mon Nov 14 13:05:00 2011 +0000 sna: tidy assignment of composite damage Make sure that the damage is always set, even if only to NULL, so that we are safe if in future the operation state is not initially cleared. Signed-off-by: Chris Wilson commit 99338ebe61917a07611cbc81cfaf70c0ec905014 Author: Chris Wilson Date: Mon Nov 14 12:04:49 2011 +0000 sna: Check whether damage can be reduced to all-damage on moving to GPU Signed-off-by: Chris Wilson commit 00d1c539e4b69636bfc35c0c47a6f1db3a33e731 Author: Chris Wilson Date: Mon Nov 14 11:19:42 2011 +0000 sna/damage: Add a little more verbosity to debugging Signed-off-by: Chris Wilson commit 60c0fc101adbc457a7c1927602175f8079175e10 Author: Chris Wilson Date: Mon Nov 14 11:12:47 2011 +0000 sna: Refactor some common code into the common destroy-gpu-bo function Signed-off-by: Chris Wilson commit 571457e9a4b0941ec956ef98391e96fe90fb4abc Author: Chris Wilson Date: Mon Nov 14 10:31:21 2011 +0000 sna: Be more thorough discarding fences for large objects Signed-off-by: Chris Wilson commit e33dfbe3c00e634328adc7319261e5501f4c7a7a Author: Chris Wilson Date: Mon Nov 14 09:41:53 2011 +0000 sna/damage: Verify that all-damage covers the entire pixmap after reduction Signed-off-by: Chris Wilson commit f033172d2f383c7e17362a10f56c4958fb9a3a6b Author: Chris Wilson Date: Mon Nov 14 09:32:53 2011 +0000 sna: Fix debug compilation Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42887 Signed-off-by: Chris Wilson commit 0c405595fb8739b22569f81f400100b5dfa472cf Author: Chris Wilson Date: Mon Nov 14 09:27:49 2011 +0000 sna: Downgrade tiling in order to fit within fence constraints Fixes a later assert that the fenced size is valid for kgem_create_2d() Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42888 Signed-off-by: Chris Wilson commit 1baff9f8cf79fdb4a5649eabcc66cf73487c873e Author: Chris Wilson Date: Mon Nov 14 08:44:35 2011 +0000 uxa/gen4+: Re-emit composite invariant after video Reported-by: Ben Hutchings Bugzilla: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=635953 Signed-off-by: Chris Wilson commit 7ad3b377a394c9342c3ab2f003d48b7cb3c0891d Author: Chris Wilson Date: Sun Nov 13 17:53:11 2011 +0000 sna/video: Use the right pointer for unmapping Don't just deference any old random pointer, use the one we actually mapped in the first place! Reported-by: Matti Hamalainen Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42880 Signed-off-by: Chris Wilson commit 8da8809d03734d92c3bcf478fe051077ce3caa29 Author: Chris Wilson Date: Sun Nov 13 17:49:23 2011 +0000 sna/video: Simplify check for 915G[M] which is simply gen==30 Signed-off-by: Chris Wilson commit cac335b979a95e31e201f9766d19d737361e09b2 Author: Chris Wilson Date: Sun Nov 13 13:13:03 2011 +0000 sna/video: Constify a couple of attribute arrays Signed-off-by: Chris Wilson commit e592eb93dbd7df177993986f01ba6f65c4e20d4a Author: Chris Wilson Date: Sun Nov 13 11:16:45 2011 +0000 configure: Version bump for 2.16.902 snapshot Signed-off-by: Chris Wilson commit 56707b97dc746f7b46dd6cf94ba6c1db85ed619c Author: Chris Wilson Date: Sun Nov 13 11:16:12 2011 +0000 NEWS: Details for 2.16.902 Signed-off-by: Chris Wilson commit f0acc6c0f15875fd51f6df6c8f13d589f9149771 Author: Chris Wilson Date: Sun Nov 13 09:39:47 2011 +0000 sna/composite: Attempt to reduce the damage is the operation is contained Signed-off-by: Chris Wilson commit 87147b63e905c3566ac1c12e0712f068e90d5e2e Author: Chris Wilson Date: Sat Nov 12 19:35:43 2011 +0000 sna/damage: Reduce the damage for evaluating sna_damage_is_all Signed-off-by: Chris Wilson commit e8799cdea461df5102d421fda26fecceae79b929 Author: Chris Wilson Date: Sat Nov 12 12:19:31 2011 +0000 sna: Be stricter and disallow allocation of large fenced objects When allocating objects, we need to check the size of the full fenced regions against the mappable limits in order to be able to mmap the object later. References: https://bugs.freedesktop.org/show_bug.cgi?id=42813 Signed-off-by: Chris Wilson commit 0269ec5533ecf7bec0f01c682e085861a3d2ab00 Author: Chris Wilson Date: Sat Nov 12 11:49:32 2011 +0000 sna: Handle incremental uploads from PutImage PutImage rarely uploads the entire image inside a single request, instead breaking up into scanline segments to fit within the protocol limits. A few optimisations are based on detecting when we can discard the GPU bo for an all-dirty CPU pixmap, which are useful in this case so check for an entirely dirty pixmap following an incremental PutImage. Signed-off-by: Chris Wilson commit ed68f1b00ae7c0fb3be1f0e758b3683f3c26675b Author: Chris Wilson Date: Sat Nov 12 11:13:59 2011 +0000 sna/gen7: Fix PRIMITIVE command The topology is now an extra dword rather than an embedded field in the command. Signed-off-by: Chris Wilson commit 991ffcb60a15f08ebb396d4d4ec86d5aff0ded31 Author: Chris Wilson Date: Fri Nov 11 23:20:36 2011 +0000 sna: Reset the source counter if we completely dirty the CPU pixmap For a long lived pixmap which we are repeatedly using for upload and copying to other pixmaps, we don't want to keep to maintain a GPU buffer. So instead, reset the source counter if we discard the GPU bo and treat the pixmap as purely CPU. Signed-off-by: Chris Wilson commit 403a4dd353eef67decfba965e40623e8137e1944 Author: Chris Wilson Date: Fri Nov 11 14:59:03 2011 +0000 sna: Defer source migration for a solitary upload via CopyArea Try to avoid allocating a GPU pixmap for PutImage followed by CopyArea. Signed-off-by: Chris Wilson commit 11acfaa62362660ff303f3afe90d1afa36b9ce73 Author: Chris Wilson Date: Fri Nov 11 14:46:33 2011 +0000 sna/glyph: Tweak the small mask threshold Signed-off-by: Chris Wilson commit c6e6ae1829e06ee8fe8eb063f2433cce603c9f96 Author: Chris Wilson Date: Fri Nov 11 14:22:38 2011 +0000 sna/glyphs: Cache the glyph pixman_image_t wrapper Signed-off-by: Chris Wilson commit 8f50950f467eb2440009a807081f3ba2c9db209b Author: Chris Wilson Date: Thu Nov 10 15:58:36 2011 +0000 sna/gen7: Remove stray no-op from GEN7_3DSTATE_SBE Signed-off-by: Chris Wilson commit 6a8338fc08a1b7e5e2c6722baa87cff3915b6ef9 Author: Chris Wilson Date: Thu Nov 10 15:11:13 2011 +0000 sna: Begin debugging gen7 This is the stub of the decoder, sufficient to give details of the ops within the batch and to keep the debugger happy. Signed-off-by: Chris Wilson commit 2309f19638f8b2c35eb60fb44fa988aa5aaab57f Author: Chris Wilson Date: Thu Nov 10 11:57:21 2011 +0000 sna/dri: Apply the damage for fullscreen async blits Otherwise gnome-shell forgets to update. Eventually, I'll get the async pageflipping bits merged into the Xserver and this path and its extra bw wastage will be history! But still I'll be undermined by the compositor, grrr. Signed-off-by: Chris Wilson commit e65c2c54f5efcec16e97085dbc41042007c39151 Author: Chris Wilson Date: Thu Nov 10 11:37:46 2011 +0000 sna: Store the sna pointer on the pixmap for cheap lookups A large part of the function preamble overhead is the multi-indirection lookup for retrieving the sna pointer. We can eliminate most of these by storing a pointer on the Pixmap as well, which we often need to lookup anyway. Signed-off-by: Chris Wilson commit bddac63de2d26c6ad4ade2f0a038d756f7a41424 Author: Chris Wilson Date: Thu Nov 10 10:55:13 2011 +0000 sna: Defer purging a target buffer This is to keep the sanity checks upon the caches happy by keeping a potential flush out of the inactive cache. Signed-off-by: Chris Wilson commit 9083f7edcb05b1deb7a77c9c5a8b54fe73176648 Author: Chris Wilson Date: Thu Nov 10 10:51:32 2011 +0000 sna/gen3: Fixup some false asserts Signed-off-by: Chris Wilson commit ea01f7681901e3181204d3117fc0a23655de631f Author: Chris Wilson Date: Thu Nov 10 10:51:01 2011 +0000 configure: Force full-debug builds to disable compiler optimisations commit a72563464f53f24246379afcf1365206fae9aad3 Author: Chris Wilson Date: Thu Nov 10 00:55:55 2011 +0000 sna/gen3: Clear the video shader flag upon batch reset Fixes regression from 3252b46e1. Signed-off-by: Chris Wilson commit 3252b46e115e6cc46abbcc4f502feeb9edc31ad6 Author: Chris Wilson Date: Wed Nov 9 15:16:04 2011 +0000 sna/gen3: Pack video frames If you're going to benchmark, you may as well do so favourably. Signed-off-by: Chris Wilson commit 8e926759528a12076bd7565672acd82a37aa3eb1 Author: Chris Wilson Date: Wed Nov 9 14:44:54 2011 +0000 sna/video: Pass texture video limits to the client Signed-off-by: Chris Wilson commit bca58ff6f02b20ccf88937df8e2688f18cabac80 Author: Chris Wilson Date: Wed Nov 9 14:18:57 2011 +0000 sna: Clear kgem->busy when all requests have been processed Signed-off-by: Chris Wilson commit 8fdbd01c94c0f6f56247feed8f225901ba39e18e Author: Chris Wilson Date: Wed Nov 9 13:14:04 2011 +0000 sna/video: Use the normal bo cache for texture video streams Signed-off-by: Chris Wilson commit 436a78684507129ca1245e065ff68870eb27624b Author: Chris Wilson Date: Wed Nov 9 12:48:27 2011 +0000 sna: Track purgeability separately from reusability Signed-off-by: Chris Wilson commit 2699c01a396b644a35a9acb969ab556454be3517 Author: Chris Wilson Date: Wed Nov 9 12:27:52 2011 +0000 sna: Handle non-reusable bo more correctly during retire Signed-off-by: Chris Wilson commit fcc89d26600a57317007035fa720915e8a2ade81 Author: Chris Wilson Date: Wed Nov 9 12:25:43 2011 +0000 sna: Safety first; make all flinked bo unreusable Signed-off-by: Chris Wilson commit f62b9f94f64f402e731463205635d386386e274e Author: Chris Wilson Date: Wed Nov 9 10:15:11 2011 +0000 sna/video: Pass cropped source dimensions along with frame data So pack all the relevant details into the same structure. Signed-off-by: Chris Wilson commit 5809b1ad934d6e19dbcf00f1b59e156761ec16b6 Author: Chris Wilson Date: Wed Nov 9 09:41:53 2011 +0000 sna/video: Tidy up copy routines Signed-off-by: Chris Wilson commit 1cd06100185a37ee33209fb18362da89f9646e6b Author: Chris Wilson Date: Tue Nov 8 17:37:44 2011 +0000 sna: Protect against NULL deference of damage after reduction Signed-off-by: Chris Wilson commit 4b100b099a4d5877c79989e8b3b4ffcd3aa2306f Author: Chris Wilson Date: Tue Nov 8 16:30:22 2011 +0000 sna: Convert stippled spans to rects Signed-off-by: Chris Wilson commit 7b95f87b26675af3a1923fef824c45e087098d61 Author: Chris Wilson Date: Tue Nov 8 16:18:58 2011 +0000 sna: Use the blitter for overlapping lines if the alu doesn't read dst Signed-off-by: Chris Wilson commit 9a8d3a9dfff8be65b4ab28312ec49dbca36019e0 Author: Chris Wilson Date: Tue Nov 8 13:51:00 2011 +0000 sna: Avoid the penalty of only writing partial channels with glyphs Yes, writing 3 channels is slower than writing 4. But it's okay we simply ignore the alpha value anyway. Signed-off-by: Chris Wilson commit 576c5169efd068c364fb7d266fbb6b1dd2ad3f4d Author: Chris Wilson Date: Tue Nov 8 13:27:05 2011 +0000 sna: Tidy sna_copy_bitmap_blt Signed-off-by: Chris Wilson commit 32bbeefb8563ae2ab3a105d0b1c9b55d9587b788 Author: Chris Wilson Date: Tue Nov 8 13:17:20 2011 +0000 sna: Pack clipped stippled uploads This are even more likely to fit inside the immediate payload. Signed-off-by: Chris Wilson commit 3620cd2d157f3d81ccb76bce2ab813bd1e058878 Author: Chris Wilson Date: Tue Nov 8 11:25:14 2011 +0000 sna: Begin hooking up valgrind/memcheck Signed-off-by: Chris Wilson commit aeaadb16093f83c9b830958c5946d4d048ffd26b Author: Chris Wilson Date: Tue Nov 8 11:03:19 2011 +0000 sna: Handle creation of large Screen pixmap when the GPU is wedged Reported-by: Clemens Eisserer References: https://bugs.freedesktop.org/show_bug.cgi?id=42606 Signed-off-by: Chris Wilson commit 7577d6ea45794b83d2f84b175d453d8a460b0e9b Author: Chris Wilson Date: Tue Nov 8 10:24:34 2011 +0000 sna: Add some error messages to explain why we failed to create the screen Signed-off-by: Chris Wilson commit 33256af40b3ce2cf8a899ced1fcbf40e316772e4 Author: Chris Wilson Date: Tue Nov 8 09:55:07 2011 +0000 sna: Fixes for DBG_NO_HW (i.e. simulated GPU hanges); A couple of the recent GPU paths were failing to check for !wedged. Signed-off-by: Chris Wilson commit 5f0886dae29429f498fb10a12d5dc8de6bd798fc Author: Chris Wilson Date: Tue Nov 8 00:02:18 2011 +0000 sna/trapezoids: Use ints for the offsets to accommodate multiplication Although the original precison need only 16-bits to store the offsets, after projecting on to the sample grid we need a few more bits of precision and so need a larger integer type to avoid overflow and render glitches. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42680 Signed-off-by: Chris Wilson commit 33351d5c3dd912534c54e64ccfc7adc4c6f6ecae Author: Chris Wilson Date: Mon Nov 7 22:57:44 2011 +0000 sna/glyphs: Bypass masks for single glyphs Signed-off-by: Chris Wilson commit e4872225ace7e3b129292f5822aa75b84c52beff Author: Chris Wilson Date: Mon Nov 7 22:57:05 2011 +0000 sna/io: Minor tidy of setting command flags Signed-off-by: Chris Wilson commit 5ba8ba7421e36bcdf83e84423595326939dbe7cd Author: Chris Wilson Date: Mon Nov 7 22:55:10 2011 +0000 sna: Avoid overwriting an upload buffer during readback Reported-by: Clemens Eisserer Buzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42677 Signed-off-by: Chris Wilson commit 8f7a8a80237db77452f02273bd8ade68dfba575f Author: Chris Wilson Date: Mon Nov 7 20:08:25 2011 +0000 sna/composite: Minor fixes in operator and colour reduction Signed-off-by: Chris Wilson commit b1234f3d3a27f326b8048e3d6b476021a26e9101 Author: Chris Wilson Date: Mon Nov 7 20:07:37 2011 +0000 sna: Expand multiplies of two 16-bit values to a full 32-bit range Signed-off-by: Chris Wilson commit afdf931e61821985b31b339d1f346ddd7c4e9e3c Author: Chris Wilson Date: Mon Nov 7 20:06:35 2011 +0000 sna: Add some more debugging in the hunt for overflows Signed-off-by: Chris Wilson commit 8f3c845782fdb2fa8bdf751bdd7dd83ca02c42ac Author: Chris Wilson Date: Mon Nov 7 20:05:41 2011 +0000 sna/blt: Small cleanups Whilst perusing for overflows, remove some redundant conditionals. Signed-off-by: Chris Wilson commit 8e775cecccebe543d344721d45b2d43ee9f122b3 Author: Chris Wilson Date: Mon Nov 7 20:01:33 2011 +0000 sna/gen3: Fix false reduction of ComponentAlpha with white source The principle behind the opertator reduction of WHITE * maskca is valid, except that we failed to account for the src/mask transposition when emitting the vertices - garbage ensued. Given that we agressively reduce the shader required for WHITE * maskca, it does not seem worthwhile to special case the primitive emitter as well. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42676 Signed-off-by: Chris Wilson commit 65a440543b13e3e605a4a2d6209a460fbbe55736 Author: Chris Wilson Date: Mon Nov 7 20:00:20 2011 +0000 sna: Fix 16-bit overflow of rowlength for memcpy Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42619 Signed-off-by: Chris Wilson commit d4edbd480445bc6aadd2c9f17262bd4b3eefbca6 Author: Chris Wilson Date: Mon Nov 7 11:16:06 2011 +0000 sna/glyph: Fallback if we fail to clear the scratch pixmap for the glyph mask Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42663 Reported-by: Clemens Eisserer Signed-off-by: Chris Wilson commit 05f3e96a69af7750d2d62cf5ba991b89f39bc88a Author: Chris Wilson Date: Mon Nov 7 09:28:19 2011 +0000 test: Compile fix for make check Reported-by: Jeremy Huddleston Signed-off-by: Chris Wilson commit cd3d86037746f3a03c4004c1e3ba6e8281344745 Author: Chris Wilson Date: Sun Nov 6 11:13:27 2011 +0000 sna: Beware unsigned promotion of int16_t to uint32_t Mmakes for an unhappy mempy! Signed-off-by: Chris Wilson commit 0b9408d972050cb02c1024926c406cd45508a158 Author: Chris Wilson Date: Sun Nov 6 09:55:09 2011 +0000 sna: Self-intersection of wide PolyLine are only drawn once We need to process the union of the PolyLine command if lineWidth!=0 so we cannot generally feed lineWidth==1 into our special case handler. Proving the lines do not intersect is as difficult as finding the intersections and thereby finding the union of the path - so there is no advantage in adding a check whether a wide line could be special cased. Signed-off-by: Chris Wilson commit c76714c29d29687f941a9112e80223c817185d53 Author: Chris Wilson Date: Sun Nov 6 09:43:47 2011 +0000 test: Add a basic line tester Starting with exercising drawing of a single segment. Signed-off-by: Chris Wilson commit c1e1e20fe7875262530a4d522e970985642c3f13 Author: Chris Wilson Date: Sat Nov 5 20:59:17 2011 +0000 sna: Add the pixman version to the debug output Signed-off-by: Chris Wilson commit e2165f0e6b0620e3d788546924a2174506fbbde5 Author: Chris Wilson Date: Sat Nov 5 19:46:13 2011 +0000 sna: For a 32k max window size, we need to handle up to 128k strides Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42619 Signed-off-by: Chris Wilson commit c5d94b21d5c65dcae3e294412a8f20bad347a6d8 Author: Chris Wilson Date: Sat Nov 5 14:32:16 2011 +0000 sna: Extend the zero-line segments by one Fixes misrendering of the wine tabs Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42606 Signed-off-by: Chris Wilson commit 0cf29afda9b061455bc438a22bbbfb1d6db6528e Author: Chris Wilson Date: Sat Nov 5 12:36:53 2011 +0000 sna/glyphs: Fix clip detection for small masks -ENOCOFFEE. Signed-off-by: Chris Wilson commit d21c30d0b8b08ec85c3b5d9a02a6969f7c8b9315 Author: Chris Wilson Date: Sat Nov 5 11:56:17 2011 +0000 sna: Constrain GPU pixmaps to always fit within the blitter Otherwise we end up always doing expensive readbacks where we would obviously prefer it if we simply used cached memory for the CPU operation and then upload. Signed-off-by: Chris Wilson commit e309cea3ae5c22c58a918008a52e3d8d694c9ff2 Author: Chris Wilson Date: Sat Nov 5 11:28:06 2011 +0000 sna: Do the supported PictOp check first There is no point even attempting a BLT operation if we know that it is an unusual render operation. Signed-off-by: Chris Wilson commit ab3d6d878fa27515eb760107d963d522cc87cd77 Author: Chris Wilson Date: Sat Nov 5 10:08:02 2011 +0000 sna: Render small glyph masks on the CPU Signed-off-by: Chris Wilson commit c25cf434d81ac1b4da29c9c75ca1d7c28be71d04 Author: Chris Wilson Date: Sat Nov 5 01:04:54 2011 +0000 sna/gen3: Initiailse the op for fill_boxes() After removing the memset(0), a couple of fields where left uninitialised, causing potential rendering glitches. Signed-off-by: Chris Wilson commit 96e760ea7589cbe953bd0a6f2f41980a7f1c4795 Author: Chris Wilson Date: Sat Nov 5 01:04:24 2011 +0000 sna: s/flush/vblank/ fixes for DBG() Signed-off-by: Chris Wilson commit 676cb4e38dc381b2ef4fb092b66db80687aa5013 Author: Chris Wilson Date: Fri Nov 4 23:30:09 2011 +0000 sna: Run the deferred flush at vrefresh This helps to reduce the perceived jerkiness of the redraw. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42413 Signed-off-by: Chris Wilson commit 8052c3904a1ac0308846ac13534c4c76bb96d5dd Author: Chris Wilson Date: Fri Nov 4 21:14:23 2011 +0000 sna: Drain the delayed timer after forcing a flush The goal is to remove a spurious wakeup when we have no work outstanding afterwards and so would end up prematurely disabling the timer. Signed-off-by: Chris Wilson commit 112c61f6557c2337d5f1b25eb71d0dc9ada6ec89 Author: Chris Wilson Date: Fri Nov 4 21:13:55 2011 +0000 sna: Convert GXcopy with -1 to GXset Signed-off-by: Chris Wilson commit 573b23c6b618af808d3236395bbf4f4993b406e5 Author: Chris Wilson Date: Fri Nov 4 15:57:24 2011 +0000 sna: Add some asserts to detect buffer overflow. Signed-off-by: Chris Wilson commit 4ba55c3d110ba51400a72da7d4f4c62efcea1335 Author: Chris Wilson Date: Fri Nov 4 15:55:15 2011 +0000 sna/gen5: Prefer using the BLT for many solid fills Even if it means incurring a context switch, the BLT unit is significantly faster so long as we do enough fills. And there is the catch ;-) Signed-off-by: Chris Wilson commit 703cf8abdbcda783e2bb0ceb423909ee8ca98176 Author: Chris Wilson Date: Fri Nov 4 15:24:57 2011 +0000 sna: Add earlier guards against to wedged to sna_glyph_blt Signed-off-by: Chris Wilson commit fc415ec3d3e982d89d7a25db8d5fd18e0412a99b Author: Chris Wilson Date: Fri Nov 4 13:53:17 2011 +0000 sna/gen5: Also experiment with lazy spans here Signed-off-by: Chris Wilson commit 567cfa508f788b4dd5de9aafbe2f148bb6a8dbc6 Author: Chris Wilson Date: Fri Nov 4 13:43:03 2011 +0000 sna/gen6: Enable spans interface for boxes Signed-off-by: Chris Wilson commit 465515a1446fd05fdc4691522803c94515187b4c Author: Chris Wilson Date: Fri Nov 4 13:40:13 2011 +0000 sna: Silly compile fix, escaped when testing uxa Signed-off-by: Chris Wilson commit a26c5d44ccaa7e9ff931948032b073f0a550343a Author: Chris Wilson Date: Fri Nov 4 12:36:40 2011 +0000 uxa: Ensure that alphaMaps are mapped into the CPU for fallbacks Reported-by: Hans-Peter Budek Signed-off-by: Chris Wilson commit 34758895cdd93bd7671a78464e79b3891bca113d Author: Chris Wilson Date: Fri Nov 4 11:57:09 2011 +0000 sna: Ensure operations on a ShmPixmap are synchronous with clients If we are rendering to or from a ShmPixmap, we need to be sure that the operation is complete prior to sending an XSync response to client in order to preserve mixed rendering coherency. Signed-off-by: Chris Wilson commit 5525691eb024f2a04b486652c24e0e34176fdd5e Author: Chris Wilson Date: Fri Nov 4 10:49:48 2011 +0000 sna/gen6: Poor man's spans layered on top of the exisiting composite Performance of this lazy interface looks inconclusive: Speedups ======== xlib swfdec-giant-steps 1063.56 -> 710.68: 1.50x speedup xlib firefox-asteroids 3612.55 -> 3012.58: 1.20x speedup xlib firefox-canvas-alpha 15837.62 -> 13442.98: 1.18x speedup xlib ocitysmap 1106.35 -> 970.66: 1.14x speedup xlib firefox-canvas 33140.27) -> 30616.08: 1.08x speedup xlib poppler 629.97 -> 585.95: 1.08x speedup xlib firefox-talos-gfx 2754.37 -> 2562.00: 1.08x speedup Slowdowns ========= xlib gvim 1363.16 -> 1439.64: 1.06x slowdown xlib midori-zoomed 758.48 -> 904.37: 1.19x slowdown xlib firefox-fishbowl 22068.29 -> 26547.84: 1.20x slowdown xlib firefox-planet-gnome 2995.96 -> 4231.44: 1.41x slowdown It remains off and a curiosity for the time being. Signed-off-by: Chris Wilson commit 5dba7028cd5024282b601c41ddfc9081eaa1d189 Author: Chris Wilson Date: Fri Nov 4 11:24:03 2011 +0000 sna: Make sure the alpha map is accessible by the CPU on fallbacks Signed-off-by: Chris Wilson commit 70c9e70f35af21de2b5b56d72f2a9e7fe9ed49d6 Author: Chris Wilson Date: Fri Nov 4 10:19:21 2011 +0000 sna: Don't reuse partial vmapped bo A fun use after free. Signed-off-by: Chris Wilson commit 7e8c9a5b8b1625fdfe885740c36da3f4c1373ee6 Author: Chris Wilson Date: Wed Nov 2 14:02:56 2011 +0000 sna: Submit the batch on the next blockhander if operation overflows If an operation overflows from one batch into another, we submit the complete batch and begin a new. That new batch will not be submitted unless it is filled or on the next delayed flush update. This can cause a flicker as a large operation is broken up, such as performing a CopyArea through a Clipmask. So if we submit a full batch during a flush interval, immediately flush any partial batch at the next blockhandler. This stops rude Santa flashing Rudolf in xsnow! Signed-off-by: Chris Wilson commit 5b21838471651a59d3e425c05a389c3f9196e21d Author: Chris Wilson Date: Thu Nov 3 22:51:24 2011 +0000 sna: Compute the correct extents for the PolyRectangle Otherwise we may leave one behind... A regression from the introduction of sna_poly_rectangles: 40af32a0e9ed971a1f2c2a45266f32016bda7ed3 (sna: Execute blits directly for PolyRectangle) Reported-by: Matti Hamalainen Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42568 Signed-off-by: Chris Wilson commit 2174f840158aa9cfa370ade38be28f8dc8e4b526 Author: Chris Wilson Date: Thu Nov 3 20:41:31 2011 +0000 uxa: Remove caching of surface binding location If the pixmap were to be used multiple times within a batch with mulitple formats, the cache would only return the initial location with the incorrect format and so cause rendering glitches. For instance, GTK+ uses the same pixmap as an xrgb source and as an argb mask in order to premultiply and composite in a single pass. Rather than introduce an overly complication caching (handle, format) mechanism, kiss and remove the invalid implementation. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=40926 Signed-off-by: Chris Wilson commit a1b40a20bbba4b70990a8983a2916d3d5850d828 Author: Chris Wilson Date: Thu Nov 3 19:27:23 2011 +0000 sna: Support binding of a bo for multiple formats Applications may use the same pixmap with multiple formats within the same operation. For instance, you can premultiply and composite a normal pixmap in this manner. However, as we reused the sampler binding locations of the source (without an alpha channel) for the mask, we failed to read and multiply by the alpha channel causing it to remain black instead of transparent. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=40926 Signed-off-by: Chris Wilson commit 31c5eb8e906bf8e59743372edb2d703b50cd311e Author: Chris Wilson Date: Thu Nov 3 17:07:48 2011 +0000 sna: Clean up the fallback code for glyphs Signed-off-by: Chris Wilson commit fa0fefd638d148d48760c95fe05520cddd96e5a4 Author: Chris Wilson Date: Thu Nov 3 15:33:25 2011 +0000 sna: Restore the lowlevel glyph routines for the sake of Damage Damage bypasses the Text interface, preventing the backend from hooking into the font and storing private glyph representations, and calls directly into the Glyph routines. So to prevent a segfault we have to restore them. Signed-off-by: Chris Wilson commit 1677b273af67e2b690bd21a1b43a7d9f9d5a70c1 Author: Chris Wilson Date: Thu Nov 3 14:31:53 2011 +0000 sna: Skip encoding zero sized glyphs Signed-off-by: Chris Wilson commit 239cfb99f9697392d7c25328093e6662ad04bdc9 Author: Chris Wilson Date: Thu Nov 3 14:25:54 2011 +0000 sna: Unroll the quadword upload of the glyph data We know that the length is nicely aligned and so can avoid a relatively expensive call into memcpy. Signed-off-by: Chris Wilson commit e2542bad88573e590ef4cbeed6b5884166b56adb Author: Chris Wilson Date: Thu Nov 3 14:17:05 2011 +0000 sna: Add the missing returns to prevent fbImageGlyphBlt fallbacks Signed-off-by: Chris Wilson commit bc032c9be1bab7477fbf4b2a64fb7aca6d75ac18 Author: Chris Wilson Date: Thu Nov 3 11:47:18 2011 +0000 sna: Coalesce reduction of cpu damage Signed-off-by: Chris Wilson commit c92671b33ef30a8eb723eb366692ae3a5878b8f9 Author: Chris Wilson Date: Thu Nov 3 11:47:18 2011 +0000 sna: Coalesce reduction of gpu damage Signed-off-by: Chris Wilson commit 3b4a508eb022e9ade5929d3ebc97e0f9216a80b1 Author: Chris Wilson Date: Thu Nov 3 00:32:29 2011 +0000 sna: Translate glyphs into MSBFirst upon initial load Signed-off-by: Chris Wilson commit 8a259e34d3295af3be864e4fae96e6c3a9a632c7 Author: Chris Wilson Date: Thu Nov 3 01:05:52 2011 +0000 sna: gc->miTranslate is always 1 So we can perform some constant folding and eliminate dead code. Signed-off-by: Chris Wilson commit 8f68f9e5f8341e20c4e1e46044f79806a44ecd03 Author: Chris Wilson Date: Thu Nov 3 00:59:46 2011 +0000 sna: Trimming is redundant given that we always have a CompositeClip Signed-off-by: Chris Wilson commit f4bdd84b846e6e778b7c496e7ed0d51e77310f36 Author: Chris Wilson Date: Thu Nov 3 00:48:10 2011 +0000 sna: Simplify the uncommon check for gpu-only damage by using damage-all The use of a gpu-only scratch bo is uncommon with the core acceleration routines, and we can eliminate the check for not incrementing the damage by allocating a damage-all and using the common optimisation of reduce_damage(). Signed-off-by: Chris Wilson commit 353fa4218cd904857fe95ef140e7f38aefe07578 Author: Chris Wilson Date: Wed Nov 2 22:53:51 2011 +0000 sna: Don't call into retire unless there are outstanding requests Signed-off-by: Chris Wilson commit 6fdd0f254d2e58d425d6d23023e9db3d2924c6be Author: Chris Wilson Date: Wed Nov 2 22:23:16 2011 +0000 sna: Use an integer value for the sentinel clang complains otherwise. Signed-off-by: Chris Wilson commit 1073c78f6cebfd6380b53dd891b7a72e50f398d6 Author: Chris Wilson Date: Wed Nov 2 22:14:40 2011 +0000 sna: Pack small 1-bpp uploads into immediate buffers Signed-off-by: Chris Wilson commit 73b2ef5a7de8d733fa1821e5010480ede305e25e Author: Chris Wilson Date: Wed Nov 2 22:02:51 2011 +0000 sna: gc->pCompositeClip always exists after validate Signed-off-by: Chris Wilson commit 6fd0754e9c81164ca0fdcf96ba2bfb419568187f Author: Chris Wilson Date: Wed Nov 2 18:03:07 2011 +0000 sna: defer calling fbValidateGC until we fallback A secondary effect is that this prevents needless migration of the tiling pixmap which we want to optimistically keep on the GPU. Signed-off-by: Chris Wilson commit 6f8875e216194474e818ba6e743e7c22b817f96e Author: Chris Wilson Date: Wed Nov 2 15:41:40 2011 +0000 sna: Wait an extra interval before switching off periodic updates Signed-off-by: Chris Wilson commit def6b5e733e623333abf596bbdbabf8dd0ac8b00 Author: Chris Wilson Date: Wed Nov 2 14:35:45 2011 +0000 sna: Only execute the accel block handler if we are about to block Signed-off-by: Chris Wilson commit 088b875affb8655010bc678ca4a3e87703c92daf Author: Chris Wilson Date: Wed Nov 2 13:11:04 2011 +0000 sna: Use the packed 8x8 stipple for non-repeating small rectangles Signed-off-by: Chris Wilson commit 0c373db73d1d2b7cf844653307c60a3891a3cc10 Author: Chris Wilson Date: Wed Nov 2 12:43:55 2011 +0000 sna: Fix the 8x8 stippled pattern origin Signed-off-by: Chris Wilson commit 9cdf8be55029f30b47cfc39ddd2754d40602cf2c Author: Chris Wilson Date: Wed Nov 2 12:17:31 2011 +0000 sna: Add clipped stippled upload support For xsnow! Signed-off-by: Chris Wilson commit 6553c9e1cbfcecf781ba106de99f9e70c6d373ba Author: Chris Wilson Date: Wed Nov 2 10:03:45 2011 +0000 sna: Quieten a fewer compiler sign compare warnings Signed-off-by: Chris Wilson commit 42e2036cda6b76f2b64e28a81bdaa661e4cbffa5 Author: Chris Wilson Date: Tue Nov 1 20:56:48 2011 +0000 sna: Use bo for scratch pixmaps allocated by the dix/mi routines Now that we have the rudiments of accelerated deep-plane copies, we can begin to benefit from using BO for the core dix/mi routines like ShmPutImage. Signed-off-by: Chris Wilson commit 45cc952994e169590ac4300f4ccf54810377f7f2 Author: Chris Wilson Date: Tue Nov 1 20:20:35 2011 +0000 sna: Fix debugging assertions for Composite When comparing drawable clip extents against pixmap boundaries we need to include the pixmap screen offset on a Composited desktop. Signed-off-by: Chris Wilson commit ebc9f3966d28343aebeba6886500abf511631785 Author: Chris Wilson Date: Tue Nov 1 19:47:51 2011 +0000 sna: Accelerate deep-plane copy Signed-off-by: Chris Wilson commit e0fd07bc251296784bf70f02877765171a053cc6 Author: Chris Wilson Date: Tue Nov 1 17:02:17 2011 +0000 sna: Accelerate XYPixmap upload when using GXcopy Mostly for the lols. Signed-off-by: Chris Wilson commit 22c43efe6b9b5f669593aa9f3af6ee437426c5d2 Author: Chris Wilson Date: Tue Nov 1 13:43:23 2011 +0000 sna: Implement 8x8 stippled rect fills Signed-off-by: Chris Wilson commit c58b7643e9b86599a41ede516a778bd9869adc8e Author: Chris Wilson Date: Tue Nov 1 10:13:20 2011 +0000 sna: Accelerate 1bpp uploads These still get used (see Wine and Swing) by applications which like to do "crisp" 1-bit rendering on the client side and then put onto the scanout. So avoid the readbacks, and push them through the BLT instead. It turns out to be faster than using fb too, bonus! Signed-off-by: Chris Wilson commit 8939ddbc27efedbb3e446dd26425a6bed39577f2 Author: Chris Wilson Date: Mon Oct 31 17:45:02 2011 +0000 sna: Improve rendering of thin segments Signed-off-by: Chris Wilson commit 98b830a975e04685455b244a8414a293857cbd00 Author: Chris Wilson Date: Mon Oct 31 17:41:27 2011 +0000 sna: Improve rendering of thin lines Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42443 Signed-off-by: Chris Wilson commit 6a5834ae560ef3b05c42b58ee0654316a66eaed0 Author: Chris Wilson Date: Mon Oct 31 16:28:59 2011 +0000 sna/gen3: Fix invalid assert SHADER_CONSTANT is expected here, the other IMMEDIATES however should have already been handled. Signed-off-by: Chris Wilson commit e12375d05bfb32933897b62829d82c6f1aa81d70 Author: Chris Wilson Date: Mon Oct 31 14:24:19 2011 +0000 sna: Fix a few minor cut'n'paste errors Use the constant value [last_box-box] where applicable. Signed-off-by: Chris Wilson commit 20a4f533958991557a3b05af00f2856657579d68 Author: Chris Wilson Date: Mon Oct 31 14:14:40 2011 +0000 sna: Apply the GPU damage for clipped PolyFillRectangles Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42425 Signed-off-by: Chris Wilson commit 7172f28c2b36aaa8c3a74ae4a7f699d3d4af10eb Author: Chris Wilson Date: Mon Oct 31 12:53:35 2011 +0000 sna: Fix placement of clipped wide PolyRectangle Signed-off-by: Chris Wilson commit 59535d0e1c587cfe9249ab1baf9acf8470b49b8b Author: Chris Wilson Date: Mon Oct 31 12:10:52 2011 +0000 sna: Set the flush interval based on output vrefresh Rather than a blank 25Hz, use twice the vblank interval to hopefully avoid bad values. Signed-off-by: Chris Wilson commit 9a2e59bfba6748325918f895b3e58351c7eb4baa Author: Chris Wilson Date: Mon Oct 31 11:56:53 2011 +0000 sna: Fix reversal of zero/wide for PolyRectangle Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42414 Signed-off-by: Chris Wilson commit b1287e49347e0fff67f43d157fa87758d5cbe588 Author: Chris Wilson Date: Mon Oct 31 11:27:36 2011 +0000 sna/video: Fix copy region for cropped video Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42412 Signed-off-by: Chris Wilson commit b4ab412cb3eb8d37a33f5eeae5c613090ad2df0f Author: Chris Wilson Date: Mon Oct 31 10:37:43 2011 +0000 sna/blt: Optimise fill with GXcopy and pixel==0 to a GXclear Signed-off-by: Chris Wilson commit 088728ad26c7efc2d65b97ff22d526c65a0fa442 Author: Chris Wilson Date: Mon Oct 31 10:04:35 2011 +0000 sna/blt: Don't idiotically override alu for pixel==0 and track alu Another classic and utterly stupid mistake. At least I was consistent in my error. Reported-by: Clemens Eisserer Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42414 Signed-off-by: Chris Wilson commit 9c399c74adc5e5bc7d425382a74d84315d2cef02 Author: Chris Wilson Date: Mon Oct 31 09:20:54 2011 +0000 sna: Add an option to disable the delayed flush Section "Device" Option "DelayedFlush" "False" EndSection Signed-off-by: Chris Wilson commit c8860ce26eaa8f9e876af2354d61c35338076ac8 Author: Chris Wilson Date: Sun Oct 30 23:25:48 2011 +0000 sna: Report whether anything was retired to the inactive list So that we can skip a fresh search of the inactive list if nothing changed. Signed-off-by: Chris Wilson commit a7adcc8e47ba550e7c36a6543a6e0d99040c7354 Author: Chris Wilson Date: Sun Oct 30 21:19:35 2011 +0000 sna: Stash the last scratch pixmap Signed-off-by: Chris Wilson commit c8d4f5d32af94f712cc81ff495b19907261759ea Author: Chris Wilson Date: Sun Oct 30 21:21:25 2011 +0000 sna: Correct PolyRectangle verticals to be the end-point not the length Fixes an off-by-one in the drawing of the rectangles caused by a too-literal translation of the rectangles into boxes. Signed-off-by: Chris Wilson commit 5c45622b889ba66b6476241601e1458d219c45ba Author: Chris Wilson Date: Sun Oct 30 17:49:19 2011 +0000 uxa/glyphs: Fallback rather than fail to render with a non-solid Source Reported-by: Uli Schlachter Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=31819 Signed-off-by: Chris Wilson commit e6eb803cb44b01e300fb4b08304227a430912beb Author: Chris Wilson Date: Sun Oct 30 17:44:26 2011 +0000 sna: Loop over all clip rects for glyph blt rather than fallback Signed-off-by: Chris Wilson commit 1eefb75c91fb769551b485f1b2c6b59023c69906 Author: Chris Wilson Date: Sun Oct 30 17:17:36 2011 +0000 sna: Don't mess with NDEBUG This is set in configure and redefining it later inside the C files just leads to trouble and broken compilation. Signed-off-by: Chris Wilson commit 7237374a03d1e0c2e6c0b2d0eb42afc8df25df0d Author: Chris Wilson Date: Sun Oct 30 16:46:10 2011 +0000 configure: Bump version to 2.16.901 commit 0a78a75e75225d89e465804c482a667840a2c8de Author: Chris Wilson Date: Sun Oct 30 16:45:46 2011 +0000 NEWS: Add release notes for 2.16.901 commit c7552cb819ae8c90975193c159236e0070628dd9 Author: Daniel Vetter Date: Sun Oct 30 16:35:14 2011 +0100 i810: kill mmio vga access helpers Signed-Off-by: Daniel Vetter commit e8b8adeab3b2e69a8684217271604a5f658f617d Author: Daniel Vetter Date: Sun Oct 30 16:34:05 2011 +0100 i810: use vgahw functions instead of roll-your-own Signed-off-by: Daniel Vetter commit 4aeea489ed3189d7ebcae4199153c1ae736ebf68 Author: Daniel Vetter Date: Sun Oct 30 16:07:45 2011 +0100 i810: rip out PIO access functions All the io done at init time is already converted to libpciaccess Signed-Off-by: Daniel Vetter commit e735636194671e221eca3e86b9bd06fd0d4b6c5a Author: Daniel Vetter Date: Sun Oct 30 15:51:49 2011 +0100 i810: kill dead i830 defines Signed-Off-by: Daniel Vetter commit bb840ee09fa620648065884306fbe3e0da37ad77 Author: Chris Wilson Date: Sun Oct 30 16:13:04 2011 +0000 sna: Convert PolyRectangles to use Box32 For the same overflow protection. Signed-off-by: Chris Wilson commit af52e234a78bbde964a93ef8cce21f41a5c959f3 Author: Chris Wilson Date: Sun Oct 30 15:34:34 2011 +0000 sna: Fix compilation Remember to drink coffee and compile "obvious" fixes before pushing. Signed-off-by: Chris Wilson commit 23b314729c43101e515e2625db2a2401c72c2596 Author: Chris Wilson Date: Sun Oct 30 15:24:21 2011 +0000 sna: Rewrite box32_add_rect() to enable int32_t promotion Signed-off-by: Chris Wilson commit 427622aca31b7463eb2dcd52e352dc8997e84baa Author: Chris Wilson Date: Sun Oct 30 12:02:42 2011 +0000 sna: Push overflow detection to the end of extents Removes a couple of conditionals from the middle of the hotpath and on Intel we are not realising the benefit of only utilising 16-bit values. Signed-off-by: Chris Wilson commit 2d3aba8518ae790325c6123dacc3ce73b6e91cbd Author: Chris Wilson Date: Sun Oct 30 11:22:26 2011 +0000 Revert "sna: Don't flush the render caches if in the process of writing again" This reverts commit 15266e1b9500f6b348661c60d1982bde911f2d0e. KDE relies upon the ability to render into a sampler and then render upon itself. Not the first sign of madness... Will have to find another way of winning back the compwinwin performance. commit d4062705178856cffc83030fa40f758e9d884566 Author: Chris Wilson Date: Sun Oct 30 10:40:32 2011 +0000 sna/composite: Fix incorrect operator reduction for RenderFillRectangles As exemplified by KDE (using Kate) on gen3, it would attempt to render a large set of boxes using OVER and a transparent colour. As gen3 copied across some of the BLT assumptions, it was incorrectly reducing that to a CLEAR and thus rendering incorrectly. Signed-off-by: Chris Wilson commit da3f8656744f7c801608c7c89a859c0f7dc9e6e0 Author: Jeremy Huddleston Date: Sat Oct 29 20:15:11 2011 -0700 Use malloc/calloc/realloc/free directly Signed-off-by: Jeremy Huddleston commit 17ba07a7befc7f442ae1c917a087ac026516d886 Author: Chris Wilson Date: Sat Oct 29 20:03:37 2011 +0100 sna/gen3: Set the src flags for fill A regression from c2040fb8e62076d87f7a5b407505ee3f3482945e (sna: Remove the memset(0) of the fill op) left the flags for the source channel uninitalised leaving the shader construction undefined and causing visual glitches. Reported-by: Paul Neumann Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42367 Signed-off-by: Chris Wilson commit 4835fe5b2e19ac456773565399f334845f7442c0 Author: Chris Wilson Date: Sat Oct 29 18:56:29 2011 +0100 sna/genX: Improve reduction of Render operator to BLT alu This appeared to introduce a visual gitch into the xfce4 selection box on gen6 at least. References: https://bugs.freedesktop.org/show_bug.cgi?id=42367 Reported-by: Paul Neumann Signed-off-by: Chris Wilson commit 33383fde55609fb4bd372d2d41a6254799499225 Author: Chris Wilson Date: Sat Oct 29 12:53:15 2011 +0100 sna: Tweak PolyRectangle blitting A box of height/width 1 and lineWidth 0 can also be drawn with a single box as the far edge is contiguous with far side of the near edge. Signed-off-by: Chris Wilson commit 64a57c8dbec81f6420ff032985237eb16cebfb87 Author: Chris Wilson Date: Sat Oct 29 11:01:51 2011 +0100 sna: Further clipping bugs Make sure the line is clipped to the current clip box and discard if the clip is reduced to nothing. References: https://bugs.freedesktop.org/show_bug.cgi?id=42361 Signed-off-by: Chris Wilson commit 719626ef7918d5cd69d13f983b8b615aefe215d6 Author: Chris Wilson Date: Sat Oct 29 10:26:08 2011 +0100 sna: Increment clip box when short-circuiting the iteration Reported-by: nkalkhof@web.de Bugzilla: https://bugs.freedesktop.org/show_bug.cgi Signed-off-by: Chris Wilson commit 32a9e0352a1e28509580d84bfc5377d8fae913c4 Author: Chris Wilson Date: Fri Oct 28 23:14:24 2011 +0100 sna: Retire and search inactive again before creating new buffer Attempt to retire an active buffer into the inactive cache before giving up and creating a new buffer. Signed-off-by: Chris Wilson commit 15266e1b9500f6b348661c60d1982bde911f2d0e Author: Chris Wilson Date: Fri Oct 28 18:00:56 2011 +0100 sna: Don't flush the render caches if in the process of writing again Signed-off-by: Chris Wilson commit d3c7ee921172d4898ebdc6e944812a02426008e9 Author: Chris Wilson Date: Fri Oct 28 16:44:16 2011 +0100 sna: Faster unclipped rectilinear segments Signed-off-by: Chris Wilson commit 3cd909cfe9e7d7e97a1d51513a049e312fd9ddae Author: Chris Wilson Date: Fri Oct 28 15:48:49 2011 +0100 sna: Call miZeroArcLine where appropriate for GPU paths Signed-off-by: Chris Wilson commit 9921c98df070032e082570f0663f6d3ad0cccd56 Author: Chris Wilson Date: Fri Oct 28 14:09:11 2011 +0100 sna: Faster unclipped PolyFillRect Signed-off-by: Chris Wilson commit c1718f96f520874c1fb2af0eb975c7f548762d34 Author: Chris Wilson Date: Fri Oct 28 14:01:24 2011 +0100 sna: Faster unclipped PolyPoint Signed-off-by: Chris Wilson commit a4762ed891a50238abddf6c762a053db5c34b87f Author: Chris Wilson Date: Fri Oct 28 13:39:17 2011 +0100 sna: Call directly into tiled rects for tiled spans. As we already know the extents and that this is a candidate for GPU-acceleration, we can skip over those steps and emit the tiled rects. Signed-off-by: Chris Wilson commit a02069df2ddc259d65ad468834f7968d283b713f Author: Chris Wilson Date: Fri Oct 28 11:30:37 2011 +0100 sna: Faster unclipped spans Signed-off-by: Chris Wilson commit 48eff1676c4066c8b8e069fd579a1b6fc2329960 Author: Chris Wilson Date: Thu Oct 27 23:30:06 2011 +0100 sna/gen2: Reset BLT registers across 2D/3D context switches It appears the hardware trashes the BLT registers after a 3D context switch, so we need to reload. Signed-off-by: Chris Wilson commit f006fac72a733e047b9468704e9c9206f62f96f5 Author: Chris Wilson Date: Thu Oct 27 21:50:25 2011 +0100 sna: Fast path typically unclipped tiled rects Signed-off-by: Chris Wilson commit d6f543011bf3d03b9298b5ceeddfc1fc20fc534b Author: Chris Wilson Date: Thu Oct 27 20:16:35 2011 +0100 sna: Use private identifier for internal scratch pixmaps This was meant to be a part of the previous commit. These are the scratch pixmaps for which we truly do want to allocate GPU-only. Signed-off-by: Chris Wilson commit 14a065a267523f9a31be07daaab19b67e9f02290 Author: Chris Wilson Date: Thu Oct 27 18:07:53 2011 +0100 sna: Create scratch pixmap on the CPU Contrary to earlier belief, the only usage of CREATE_PIXMAP_USAGE_SCRATCH in the dix/mi are either along paths unused by us or more importantly for operations which are detrimental to create GPU-only buffers. So create CPU pixmaps instead. Signed-off-by: Chris Wilson commit 5d52d5009375618bb703909fa0fad04e6d54a2ba Author: Chris Wilson Date: Thu Oct 27 17:06:51 2011 +0100 sna: Tweak zero line spans to include missing last segment Signed-off-by: Chris Wilson commit 72e00096351137bd7eb0ae91474657a597836252 Author: Chris Wilson Date: Thu Oct 27 16:36:39 2011 +0100 sna: Fix advance through clip boxes for fill->boxes Signed-off-by: Chris Wilson commit c7bd89fb54686c14b687c007e42c34c23423b0fd Author: Chris Wilson Date: Thu Oct 27 16:23:35 2011 +0100 sna: Fix PolyPoint not to always send 512 boxes Just emit the remaining boxes! Signed-off-by: Chris Wilson commit cfd5b7f6ce80af911bfc8b71bba776ea814e307a Author: Chris Wilson Date: Thu Oct 27 15:28:11 2011 +0100 sna: Use fill->boxes to pack PolyFillRectangle Signed-off-by: Chris Wilson commit 8d304f47e35be344550f389158eac3d4c21d9440 Author: Chris Wilson Date: Thu Oct 27 14:43:46 2011 +0100 sna: Use fill->boxes for rectilinear blits Signed-off-by: Chris Wilson commit 5d51ff0cf901b808c3cd5aea761d51e6d15b73bf Author: Chris Wilson Date: Thu Oct 27 13:57:59 2011 +0100 sna: Clip PolyLines to the current clip box Signed-off-by: Chris Wilson commit 990043f6b8cf8e580aad2e52f8d236a6aec838b5 Author: Chris Wilson Date: Thu Oct 27 12:33:55 2011 +0100 sna: Handle singular clip regions for PolyRectangles Otherwise X will crash! Signed-off-by: Chris Wilson commit 991582f6678581e3d92c0497e10ecee67536efc2 Author: Chris Wilson Date: Thu Oct 27 09:27:41 2011 +0100 sna/io: Update batch mode after submitting partial requests Signed-off-by: Chris Wilson commit 91745de113340c8a281e8cb724b4a7e0b7a0711a Author: Chris Wilson Date: Thu Oct 27 09:17:50 2011 +0100 sna: Further improve use of fill->boxes for PolyRectangle Signed-off-by: Chris Wilson commit ffe1bdea82c07a2738743855dfa83494531b41aa Author: Chris Wilson Date: Wed Oct 26 18:37:29 2011 +0100 sna/blt: After submitting the batch, it will be empty Signed-off-by: Chris Wilson commit b34e7454574f36796560955fb9d3d33ca00924d6 Author: Chris Wilson Date: Wed Oct 26 18:11:25 2011 +0100 sna/blt: Convert Copy with color 0 into a Clear Signed-off-by: Chris Wilson commit 389241aa6469b2e64a58cc0527e3310fdc69ffb7 Author: Chris Wilson Date: Wed Oct 26 17:50:38 2011 +0100 sna: Suppress an overwritten fill Signed-off-by: Chris Wilson commit baf78fcddb87eba987163485ad8fd227fb244f29 Author: Chris Wilson Date: Wed Oct 26 12:54:13 2011 +0100 sna/gen2: Cache the last solid colour for spans Signed-off-by: Chris Wilson commit ff4715a36999a43cee20cf3474b4ae0af646342c Author: Chris Wilson Date: Wed Oct 26 12:40:46 2011 +0100 sna/composite: Discard opaque masks Signed-off-by: Chris Wilson commit 0079579ebd548fc1b490a42928373507a93229bd Author: Chris Wilson Date: Wed Oct 26 11:49:53 2011 +0100 sna: Fix debug compilation, again. commit 5988422455ebb2a3b1cafadc1ff302c35a7a5039 Author: Chris Wilson Date: Wed Oct 26 11:47:45 2011 +0100 sna: Handle degenerate copy area whilst wedged We want to return early to avoid handing the empty box as the lower code makes the assumption that it has work to do. Signed-off-by: Chris Wilson commit 3fc7cd2dc8e2b9c0d4143a9f6657f84050ae249a Author: Chris Wilson Date: Sun Oct 23 20:59:50 2011 +0100 sna/blt: Rename the composite paths Whilst re-reading, I found the old naming scheme slightly muddled and not nearly as descriptive as it should be. Hopefully this slight tweak reads better in 6 months time. Signed-off-by: Chris Wilson commit 280325900bf3c17c3177f6e3f52236d6b452e0f3 Author: Chris Wilson Date: Sun Oct 23 20:58:48 2011 +0100 sna/blt: Share the common fill->boxes between blt and render paths Signed-off-by: Chris Wilson commit 56f65e65204f1ecbad598419e525e69f6179f9c7 Author: Chris Wilson Date: Sun Oct 23 20:40:21 2011 +0100 sna: Inline box_intersect() Most of this branchy function is spent in manipulating the stack for the function call, so inline it. Signed-off-by: Chris Wilson commit f65e73f8a3a013e79b753d625f79994408e717f7 Author: Chris Wilson Date: Sun Oct 23 19:55:40 2011 +0100 sna: Use fill->boxes to unroll FillSpans Signed-off-by: Chris Wilson commit 94f9cd40dcdd444d3b5a5598deb0f3d04026df3a Author: Chris Wilson Date: Sun Oct 23 19:11:35 2011 +0100 sna: Use fill->boxes to unroll PolyPoint Signed-off-by: Chris Wilson commit 5e08ca75a5eb852c09b9acf6004668bc63f1fe6c Author: Chris Wilson Date: Sun Oct 23 17:53:54 2011 +0100 sna: Use the new fill->boxes for PolyRectangle Signed-off-by: Chris Wilson commit efdc7985e2f619ca94cf6ad77ead4f6867ceedc5 Author: Chris Wilson Date: Wed Oct 26 14:57:28 2011 +0100 self-copy-damage commit 0ed3426a810336b666604d34c10f996f318ebf82 Author: Chris Wilson Date: Fri Oct 21 23:45:28 2011 +0100 sna: Convert diagonal zero-width lines into blits This is slower than falling back to swrast for x11perf (up to 4x slower on SNB), it is still faster than doing that rasterisation through a WC-mapping and much faster in ordinary usage due to avoiding the readback hit. Signed-off-by: Chris Wilson commit c12371d9e7b3bbff7f318186a0933d6108db0bc8 Author: Chris Wilson Date: Fri Oct 21 22:28:56 2011 +0100 sna: Flatten the branching for fill-spans Signed-off-by: Chris Wilson commit e7f19d85fead62df21eedc87409b5abacfec4a17 Author: Chris Wilson Date: Fri Oct 21 22:15:09 2011 +0100 sna/blt: Upload the box using a single 64-bit instruction So long as we have 64-bit instructions of course! Signed-off-by: Chris Wilson commit 40af32a0e9ed971a1f2c2a45266f32016bda7ed3 Author: Chris Wilson Date: Fri Oct 21 20:10:02 2011 +0100 sna: Execute blits directly for PolyRectangle By constructing the batch buffer directly for PolyRectangle, rather than via miPolyRectangle->(PolyFillRectangle/PolyLine), we dramatically reduce the CPU overhead and can saturate the GPU. Signed-off-by: Chris Wilson commit a18f559961135fa288dda3b94207abb0b6d4d302 Author: Chris Wilson Date: Fri Oct 21 16:37:09 2011 +0100 sna: Fix debug compilation Signed-off-by: Chris Wilson commit a3a075201c8e47cfad6ec296b7baf549deb44cb3 Author: Chris Wilson Date: Fri Oct 21 15:30:52 2011 +0100 sna: Treat a bo marked as being last seen on the GPU as requiring a flush We are hitting this path upon retiring the old scanout buffer following a page-flip. We want to treat this as being hot and available for reuse, so mark it as such until the next retirement pass. Signed-off-by: Chris Wilson commit 999d13f32162ce11fe4042825b5b849073944ca8 Author: Chris Wilson Date: Fri Oct 21 12:43:02 2011 +0100 sna: Fast path unclipped points Signed-off-by: Chris Wilson commit eb80013f4ea50ad4f54b4d1d90c5e9da27e1c34c Author: Chris Wilson Date: Fri Oct 21 11:49:55 2011 +0100 sna: Fast path for unclipped rectangles Signed-off-by: Chris Wilson commit d6ae86a51cc49dbba579838edd24a94e8f68294c Author: Chris Wilson Date: Fri Oct 21 03:06:20 2011 +0100 sna: Optimise reduce_damage() to handle all-damaged pixmaps Signed-off-by: Chris Wilson commit 5b945b85ee938688412df6adfea488766f4656af Author: Chris Wilson Date: Fri Oct 21 02:45:50 2011 +0100 sna: Fast-path unclipped fill spans Signed-off-by: Chris Wilson commit e9a193e9803bb69fc4a7e712f33a36ba395b3c89 Author: Chris Wilson Date: Fri Oct 21 02:20:22 2011 +0100 sna: Short-circuit GetWindowPixmap() to speed-up pixmap retrieval Signed-off-by: Chris Wilson commit 662402e41e5c7976002a71a7f961ca97c771b1f3 Author: Chris Wilson Date: Fri Oct 21 02:11:19 2011 +0100 sna: Micro-optimise PolyLine blt Signed-off-by: Chris Wilson commit c2040fb8e62076d87f7a5b407505ee3f3482945e Author: Chris Wilson Date: Fri Oct 21 02:00:44 2011 +0100 sna: Remove the memset(0) of the fill op The backends are all expected to initialise the state required. Signed-off-by: Chris Wilson commit 7e7759db4c1dbf8b782fab9b9ca95cfa1db6cf59 Author: Chris Wilson Date: Fri Oct 21 01:41:56 2011 +0100 sna: Pass a BoxRec to the fill op For many of the core drawing routines, passing a BoxRec for the fill is more convenient since they already have one generated by the clip intersection. Signed-off-by: Chris Wilson commit 98bca2dbb9c33c548b1d9789ddb633ae0389c8f1 Author: Chris Wilson Date: Fri Oct 21 01:07:52 2011 +0100 sna: Check whether we can blt whilst determining the PolyLine extents Signed-off-by: Chris Wilson commit 91f49549750aac99d38ae8f3a132c955a2c8d8f5 Author: Chris Wilson Date: Fri Oct 21 00:42:43 2011 +0100 sna: Speed-up blitting of unclipped lines Signed-off-by: Chris Wilson commit 06d7574d7da047a6314be5af27f4ac47ccd5a350 Author: Chris Wilson Date: Thu Oct 20 21:38:30 2011 +0100 sna/damage: Stash the last freed damage for quick reallocation Signed-off-by: Chris Wilson commit 1471ef82b595a207dc8873c45176ec31f277a43f Author: Chris Wilson Date: Thu Oct 20 21:32:14 2011 +0100 sna/damage: Only track the mode globally As damage accumulation is handled modally, we do not need to track the mode per elt and so attempt to simplify the code. Signed-off-by: Chris Wilson commit 1229bf6a690e7f315ab22957da23481f58515ff9 Author: Chris Wilson Date: Thu Oct 20 17:37:41 2011 +0100 sna: Actually apply the composite offset for the self-copy I translated the region to copy by the composite pixmap offset, only failed to use the translated region for the actual copy command (using instead the original boxes). Fix that mistake by avoiding the temporary region entirely and applying the translation inplace. We also have to be careful in the case of copying between two composited windows that have different offsets into the same screen pixmap. This fixes the regression introduced with a3466c8b69af (sna/accel: Implement a simpler path for CopyArea between the same pixmaps). Signed-off-by: Chris Wilson commit 3526d83e460ce6410f23f59d1315793ff9607253 Author: Chris Wilson Date: Thu Oct 20 00:29:52 2011 +0100 sna/dri: Perform a little dance to restore the gpu status on release of the bo As the bo is outside of our control whilst it is under the influence of an external renderer, we try to maintain it on the gpu so as to avoid unnecessary ping-pong. But once it is wholly back under our control, we want to stop paying the penalty for sharing it. Signed-off-by: Chris Wilson commit 50b980f12e02401cdd4bc21b970d92e7bd1e6459 Author: Chris Wilson Date: Wed Oct 19 21:09:01 2011 +0100 sna: Reuse any partial write buffer for readback Take advantage of any available temporary buffer that we reuse for readback knowing that it is the last operation in the batch. Signed-off-by: Chris Wilson commit 972989276dd3f84c1cedca0494d04f907875f8f3 Author: Chris Wilson Date: Wed Oct 19 20:41:14 2011 +0100 sna: Add some debug to discern the nature of the Cr fallback Signed-off-by: Chris Wilson commit fc224e2f8c0200c114f8dbddfc4f5395a3ef0fad Author: Chris Wilson Date: Wed Oct 19 20:20:55 2011 +0100 sna: Clean up some debug messages for 64bit Signed-off-by: Chris Wilson commit f4346e5d255f419ee6148f7d69f02560732dd4de Author: Chris Wilson Date: Wed Oct 19 19:26:19 2011 +0100 sna/blt: Use SCANLINE_BLT for multiple fill boxes Signed-off-by: Chris Wilson commit d7fb98efdcc1bcc2cca6deb54b16d425b0350196 Author: Chris Wilson Date: Wed Oct 19 17:52:56 2011 +0100 sna: The initial aperture check for a set of bo is unlikely to fail Signed-off-by: Chris Wilson commit 3c8e1c5c4cbe703781f2845926f76a9537a971ea Author: Chris Wilson Date: Wed Oct 19 16:35:08 2011 +0100 sna: Reset the ring flag upon idling We track the last ring used when active so as to avoid stalling between batches. Once the GPU has retired all the pending requests, we can use whichever ring is preferrable for the next operation without any danger of stalling upon submission. Signed-off-by: Chris Wilson commit 594c5f86bb2e1cd0390b360c926ead58ba49979f Author: Chris Wilson Date: Wed Oct 19 16:25:42 2011 +0100 sna: Don't rewrite the solid colour cache if it hasn't changed Signed-off-by: Chris Wilson commit 5515f75647bb148d9e720dcc4713a93b59ffbd49 Author: Chris Wilson Date: Wed Oct 19 13:49:55 2011 +0100 sna: Enlarge the minimum pixmap size to migrate for Render This is to workaround a ping-pong issue involving small icons. The horrible sequence of operations appears to use a tiled FillRect to copy from the scanout onto to a temporary pixmap, which causes us to readback from the scanout. We are destined to hit the fallback path there anyway until we implement stippling... References: https://bugs.freedesktop.org/show_bug.cgi?id=41718 Signed-off-by: Chris Wilson commit 87ba33bc0bcaf3c5160edfc61cff3ba0d28d3fd8 Author: Chris Wilson Date: Wed Oct 19 13:09:19 2011 +0100 sna/damage: Avoid reducing the damage for simple tests Signed-off-by: Chris Wilson commit 63aa84ef57ff1e55d2df6d680d6bbf88be0619a6 Author: Chris Wilson Date: Wed Oct 19 13:08:50 2011 +0100 sna/gen6: Apply the unknown blitter death workaround Signed-off-by: Chris Wilson commit 25f2d181946738d1ac2deb724eb48453c8692ed1 Author: Chris Wilson Date: Tue Oct 18 23:03:10 2011 +0100 sna: Compute region extents in place to eliminate redundant stack space Signed-off-by: Chris Wilson commit 1a43b2a46a67fdea215719c4446224c9d2e26a20 Author: Chris Wilson Date: Tue Oct 18 17:52:56 2011 +0100 sna: Use the unlikely wedged() throughout Signed-off-by: Chris Wilson commit 7a9c76e1812d106fedf160c959e6e502998e4ce8 Author: Chris Wilson Date: Tue Oct 18 15:51:33 2011 +0100 sna: Micro-optimise fill-spans Signed-off-by: Chris Wilson commit 0b83abfb2bc3f65447205048ae0af1a94fa9ef15 Author: Chris Wilson Date: Tue Oct 18 14:43:37 2011 +0100 sna/gen6: Don't modify composite arguments whilst checking for BLT Signed-off-by: Chris Wilson commit ae7bd2bf70b42b8c12e5e5ae074891595827a1c0 Author: Chris Wilson Date: Tue Oct 18 11:54:05 2011 +0100 sna/blt: Move the conditional out of the fill-many hotpath Signed-off-by: Chris Wilson commit 86f99379ee55b157634174c4c8aad61a30b4f57f Author: Chris Wilson Date: Tue Oct 18 11:35:12 2011 +0100 sna/gen4: Add fill-one Signed-off-by: Chris Wilson commit 26c082dbd3f89b987168d321f72e78db8c5ddc2c Author: Chris Wilson Date: Tue Oct 18 10:50:53 2011 +0100 sna/gen6: Precompute floats_per_rect Signed-off-by: Chris Wilson commit c69a7989b3c9cedc961e29b1a569b174207964ae Author: Chris Wilson Date: Tue Oct 18 10:48:37 2011 +0100 sna/gen5: Add fill-one implementation Signed-off-by: Chris Wilson commit 26a1918290b2d23de72a67805fe5e766da0c5309 Author: Chris Wilson Date: Tue Oct 18 10:42:48 2011 +0100 sna/gen6: Try continuing with the BLT if the last batch was also BLT In the vain hope of reducing switching between rings and introducing stalls between batches. Signed-off-by: Chris Wilson commit 4b1398f26d4de44167096c6548ff7384c24d5e06 Author: Chris Wilson Date: Tue Oct 18 00:29:18 2011 +0100 sna/gen6: Micro-optimise gen6_rectangle_begin We can only emit state between primitives, ergo we need only check for state updates if we've finished the vbo or are starting a new operation. Signed-off-by: Chris Wilson commit b2d842df5e3198dab401fbd450d4af83d44917dd Author: Chris Wilson Date: Mon Oct 17 23:34:19 2011 +0100 sna/gen2: Precompute floats-per-rect Signed-off-by: Chris Wilson commit 729f1ec7869167ff91fa969d06dbdfbf0e8b4126 Author: Chris Wilson Date: Mon Oct 17 23:20:20 2011 +0100 sna/gen2: Prefer to use the BLT Signed-off-by: Chris Wilson commit adef3a56cde1210554dc8327baa37c36fc5f18e3 Author: Chris Wilson Date: Mon Oct 17 23:08:32 2011 +0100 sna/gen3: Prefer to use the BLT where possible for composite ops Signed-off-by: Chris Wilson commit f40e6ab50b264ef7eb7750f72ab7418c0d3dc9c1 Author: Chris Wilson Date: Mon Oct 17 23:06:30 2011 +0100 sna/gen3: Prefer to use the BLT to clear the scratch glyph pixmaps Signed-off-by: Chris Wilson commit 941e3504e049be8af50ca9df2d0aafe9eb477983 Author: Chris Wilson Date: Mon Oct 17 19:51:33 2011 +0100 sna: Some more debug output for request/bo retirement Signed-off-by: Chris Wilson commit 1d82fe3e528578d11d2a258b586b67c807eb7cb4 Author: Chris Wilson Date: Mon Oct 17 19:47:03 2011 +0100 sna: Only retire the flush handler if nothing was submitted within the period Signed-off-by: Chris Wilson commit 7758e333409409393c4c974adb2831a7b5b18fe4 Author: Chris Wilson Date: Mon Oct 17 19:32:44 2011 +0100 sna/gen3: Use immediates for black/white solid sources Signed-off-by: Chris Wilson commit 29ca1a3922cb0e6f3d7b71857a252e5de81941b5 Author: Chris Wilson Date: Mon Oct 17 17:16:16 2011 +0100 sna/gen3: More removal of memset(0) Signed-off-by: Chris Wilson commit 88bc10878830dffefda6ed905c1f598d7a7d67ef Author: Chris Wilson Date: Sun Oct 16 20:05:49 2011 +0100 sna/gen5: Rearrange the BLT->RENDER workaround So that we can simply use the pending DRAWRECT as the non-pipelined flush required following use of a BLT command. Signed-off-by: Chris Wilson commit 18aed47ba37ccfbcdd72ed251fc97222741c66a7 Author: Chris Wilson Date: Mon Oct 17 16:41:14 2011 +0100 sna: Fast path common colour conversions Signed-off-by: Chris Wilson commit a3466c8b69afeda95180fcdc97b56d31f7a1d1ad Author: Chris Wilson Date: Sun Oct 16 20:52:47 2011 +0100 sna/accel: Implement a simpler path for CopyArea between the same pixmaps Signed-off-by: Chris Wilson commit 8afa05191a9a337ce6e02b1064de2876f5635435 Author: Chris Wilson Date: Mon Oct 17 00:55:51 2011 +0100 sna: Improve debug output for mi/fb fallbacks Signed-off-by: Chris Wilson commit 2b118658d5b9cfa56132f91238f2ea61fba8b357 Author: Chris Wilson Date: Mon Oct 17 15:00:01 2011 +0100 sna/gen3: Remove memset(0) from fill_boxes And initialise only the state used for the operation. Signed-off-by: Chris Wilson commit f1bb4ebfd8991f2f9eb9c38b9259792c11e7c86a Author: Chris Wilson Date: Mon Oct 17 14:22:58 2011 +0100 sna/damage: Clear damage->all when subtracting I missed clearing the all-damaged flag on the subtract fast paths, causing us to disregard further GPU damage and losing track of render coherency. Reported-by: Roman Jarosz Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=41769 Signed-off-by: Chris Wilson commit 46f97127c22ea42bc8fdae59d2a133e4b8b6c997 Author: Chris Wilson Date: Sun Oct 16 21:40:15 2011 +0100 snb,ivb: Workaround unknown blitter death The first workaround was a performance killing MI_FLUSH_DW after every op. This workaround appears to be a stable compromise instead, only requiring a redundant command after every BLT command with little impact on throughput. Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=27892 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=39524 Tested-by: Daniel Vetter Signed-off-by: Chris Wilson commit 71bf291e563ec5224777b4907a5347a3fbfe64c5 Author: Chris Wilson Date: Mon Oct 17 12:53:19 2011 +0100 sna/gen3: Micro-optimise gen3_rectangle_begin() We only need to emit the vbo description either at the beginning of a new op (when the state may have changed) or after finishing a full vbo. Signed-off-by: Chris Wilson commit a032feb2531439b9945d7ae1e0c2e3ab95a960e1 Author: Chris Wilson Date: Mon Oct 17 12:41:21 2011 +0100 sna/gen3: Store floats_per_rect alongside floats_per_vertex Moves a multiply out of the hot path. Signed-off-by: Chris Wilson commit 2cbe79b85b46796d561105afa980e7a52f4f1889 Author: Chris Wilson Date: Mon Oct 17 11:57:07 2011 +0100 sna/gen3: Improve reduction of render operator to blt Signed-off-by: Chris Wilson commit 606e18bd360d904a8aed01aef43644fb4a51d929 Author: Chris Wilson Date: Mon Oct 17 10:50:33 2011 +0100 configure: Add a check for pixman version As SNA requires a fairly recent release and calls directly into pixman. Signed-off-by: Chris Wilson commit 17acaf8e232a8bad9bc2c653a9fceb06f0f65f6b Author: Chris Wilson Date: Mon Oct 17 00:55:51 2011 +0100 sna: Defer CPU fallback if fb calls into mi If the fb routine does not access the pixel data directly, but instead calls into an mi routine, we can defer the readback and possibly avoid it. Signed-off-by: Chris Wilson commit c348b69f66cbe1cafc88396523da9768e1006889 Author: Chris Wilson Date: Sun Oct 16 18:23:33 2011 +0100 sna: Prefer to accelerate non-RENDER operations if already using the GPU Signed-off-by: Chris Wilson commit 6fdc9e3fd3f8defb7ad62de11f8cb069a10e5736 Author: Chris Wilson Date: Sun Oct 16 17:57:55 2011 +0100 sna: Simplify busy tracking by trusting the bo->gpu flag Signed-off-by: Chris Wilson commit 30dff81cd62b5fc5a28d5175a08c5a3ee09667bd Author: Chris Wilson Date: Sun Oct 16 17:41:34 2011 +0100 sna/accel: Upload source to GPU for !copy CopyAreas Signed-off-by: Chris Wilson commit eefa925e3d3e009b21ecc0b428d93a6c732bfa14 Author: Chris Wilson Date: Sun Oct 16 17:39:06 2011 +0100 sna/accel: Fall-forward for handling a non-copy CopyArea to a dst gpu bo Signed-off-by: Chris Wilson commit 01f41e624b31af88d433a25eaefce557e7ff3d62 Author: Chris Wilson Date: Sun Oct 16 17:38:26 2011 +0100 sna/gen2: Eliminate redundant diffuse and rectlist emission Signed-off-by: Chris Wilson commit d9e3dbffcb7e6cff35e10100d81544936813095a Author: Chris Wilson Date: Sun Oct 16 12:48:23 2011 +0100 sna/gen2: Fix fill-one-box Lets only have one special gen2 value for the source channel pixel colour and so remove the confusion and misrendering. Signed-off-by: Chris Wilson commit b89592b16e373da1fe259ad067f65ba95021dc49 Author: Chris Wilson Date: Sun Oct 16 19:39:14 2011 +0100 sna/gen7: Add missing tail to 3DSTATE_SBE I was a few NULs short of the declared amount. Signed-off-by: Chris Wilson commit 33f17b996dd0fcbb8e07875770e11571380d6509 Author: Chris Wilson Date: Fri Oct 14 12:01:34 2011 +0100 Remove a couple of trivial compile warnings for unused variables Signed-off-by: Chris Wilson commit e5e27528ffeb74f404823c678e04e847cf2109aa Author: Chris Wilson Date: Fri Oct 14 11:51:44 2011 +0100 sna/blt: Add a pair of operator reductions for opaque fills Extend the coverage of the BLT for a couple more composite operations, though they be relatively infrequent. Signed-off-by: Chris Wilson commit 3f5289b29962c3e5d94deb30782a156292d20053 Author: Chris Wilson Date: Fri Oct 14 11:31:46 2011 +0100 sna/trapezoids: Fallback if the composite operation is not supported Signed-off-by: Chris Wilson commit 3d79264ef8a8be225d9b13efd14f37a7478e6b1b Author: Chris Wilson Date: Fri Oct 14 11:31:00 2011 +0100 sna/tiling: Hook up composite_box Now used for spans and so triggering death by NULL function pointer on older chipsets. Signed-off-by: Chris Wilson commit f5f3135e3cc1f4740508773cfefca0954aed2fd3 Author: Chris Wilson Date: Fri Oct 14 10:40:39 2011 +0100 sna: Ensure that the scanout is flushed before we sleep So that we do not see any delayed rendering on i965+. Signed-off-by: Chris Wilson commit 020d84d2d09c71a8209dd774a0735a3c11cb148d Author: Chris Wilson Date: Thu Oct 13 22:32:00 2011 +0100 sna/accel: Fix sense of nothing-to-do due to clipping A missing not due to a hasty conversion from an assert. Signed-off-by: Chris Wilson commit b7fd6906c41e328649b97e16c42848a39f6e48f8 Author: Chris Wilson Date: Thu Oct 13 17:46:48 2011 +0100 sna/accel: Actually apply the clip to the glyph extents References: https://bugs.freedesktop.org/show_bug.cgi?id=41718 Signed-off-by: Chris Wilson commit eb5f86b78181b174664aaa9bc689db5421b7e7bd Author: Chris Wilson Date: Thu Oct 13 17:32:54 2011 +0100 sna/accel: Correctly offset the damage intersection for glyph runs Signed-off-by: Chris Wilson commit a4766026ad0aa14283e35598d33f13d311cc029b Author: Chris Wilson Date: Thu Oct 13 17:32:30 2011 +0100 sna/accel: Check for reloc overflow when adding a new glyph run Signed-off-by: Chris Wilson commit b10850af4125b8dbc36d13fa83f1dcce13ac6684 Author: Chris Wilson Date: Thu Oct 13 17:09:25 2011 +0100 sna/blt: Check for reloc overflow when initialising solid fills Signed-off-by: Chris Wilson commit d5242565ef17cb4265177338df5b564b960e3692 Author: Chris Wilson Date: Wed Oct 12 21:14:43 2011 +0100 sna/gen2: Trim some surplus state commands Signed-off-by: Chris Wilson commit 7c5020532335c07d9b0f6cecb0b004de4af6d273 Author: Chris Wilson Date: Wed Oct 12 20:36:53 2011 +0100 Remove vestigial includes from DRI1 Signed-off-by: Chris Wilson commit b6b3bb9b5e56949cbd9189d9857ffa1ff46377fd Author: Chris Wilson Date: Tue Sep 20 19:41:03 2011 +0100 sna: compare against the right array of cache buckets Fortunately harmless today as there are the same number of inactive/active buckets. Signed-off-by: Chris Wilson commit 1f70095837a30d6a88b9e313d3583d38ea55221d Author: Chris Wilson Date: Tue Oct 11 16:06:31 2011 +0100 sna: And free the clip after creation for PolySegments Signed-off-by: Chris Wilson commit 92f4d978c8b1f5e9a6cf3affa15d90bfb3d4e7b9 Author: Chris Wilson Date: Tue Oct 11 15:52:13 2011 +0100 sna: More micro-optimisation of messing around with clip regions Signed-off-by: Chris Wilson commit 57151f654752ed64716358ecb5e9217b59784da8 Author: Chris Wilson Date: Tue Oct 11 15:23:17 2011 +0100 sna: Micro-optimise checking for singular clip boxes Signed-off-by: Chris Wilson commit 823a4272c50247482428a16cb08741bf87a302ea Author: Chris Wilson Date: Tue Oct 11 13:51:41 2011 +0100 sna/gen3: Avoid RENDER/BLT context switch for fill boxes Signed-off-by: Chris Wilson commit 887361de173e082841453902089427447c5a158a Author: Chris Wilson Date: Tue Oct 11 13:51:25 2011 +0100 sna: Enable single fill fast path for PolySegment Signed-off-by: Chris Wilson commit 721cf30e9eeb0102c77f337332ad8259463b7fef Author: Chris Wilson Date: Tue Oct 11 13:28:45 2011 +0100 sna/accel: If the data is already on the GPU, use it for the source Fixes regression from 1ec6a0e2775 (sna: Move the source to the GPU if it is reused). Signed-off-by: Chris Wilson commit 15a4410cec111c3b3bd15ec33f29bc1f4cb709c1 Author: Chris Wilson Date: Tue Oct 11 13:28:30 2011 +0100 sna: use correct insertion point for sorting partials Signed-off-by: Chris Wilson commit a9b53c48774078d404149e86267c39eb82234c3a Author: Chris Wilson Date: Tue Oct 11 10:37:25 2011 +0100 sna: Mark the spans render functions as fastcall This reduces the amount of dancing required to call into the span functions as we can pass the arguments in both the integer and floating point registers. Signed-off-by: Chris Wilson commit 208fa8e6b88daf8446239a1a001a68ceb7d02e76 Author: Chris Wilson Date: Tue Oct 11 10:08:42 2011 +0100 sna/trapezoid: Perform the NULL check for damage in the caller Save the function call overhead in the common case. Signed-off-by: Chris Wilson commit 5050fead0ef81785a2a9be735852921463377e04 Author: Chris Wilson Date: Tue Oct 11 09:59:13 2011 +0100 sna/gen3: avoid applying zero offset to common spans Signed-off-by: Chris Wilson commit d0184b59095d5b8fab1a65ceba075d29189130d4 Author: Daniel Vetter Date: Sun Oct 9 18:43:14 2011 +0200 snb: implement PIPE_CONTROL workaround Sandybdrige requires an elaborate dance to flush caches without hanging the gpu. See public docs Vol2Part1 1.7.4.1 PIPE_CONTROL or the corrensponding code in mesa/kernel. This (together with the corresponding patch for the kernel) seems to fix the hangs in cairo-perf-traces I'm seeing on my snb machine. v2: Incorporate review from Chris Wilson. For paranoia keep all three PIPE_CONTROL cmds in the same batchbuffer to avoid upsetting the gpu. Signed-off-by: Daniel Vetter commit 4a2e833ab17e4facf0f90166b82a5696a1deef91 Author: Chris Wilson Date: Tue Oct 11 08:45:47 2011 +0100 sna/gen7: Add render support for fill one To prevent the RENDER to BLT transition and potential stall. Signed-off-by: Chris Wilson commit 41f525fab5a82c24adafc7e8c8409417d16a0e9a Author: Chris Wilson Date: Tue Oct 11 00:22:11 2011 +0100 sna/gen6: Add render support for fill-one-box Signed-off-by: Chris Wilson commit 5b6575bdded4b24ec1c9515203f44798225c10b6 Author: Chris Wilson Date: Mon Oct 10 16:39:02 2011 +0100 sna: Support a fast composite hook for solitary boxes Signed-off-by: Chris Wilson commit c5414ec992d935e10156a2b513d5ec2dded2f689 Author: Chris Wilson Date: Sun Oct 2 12:02:41 2011 +0100 sna: Use BLT operations to avoid fallbacks in core glyph rendering Signed-off-by: Chris Wilson commit 6b62b9d7c4301457a024f840383fb4515704ebdb Author: Chris Wilson Date: Tue Oct 4 19:30:06 2011 +0100 sna: Increase reserved space in batch to accommodate gen5 workaround Signed-off-by: Chris Wilson commit 6351d8eb823dbfffb4792af8592d1ce565211e23 Author: Chris Wilson Date: Tue Oct 4 19:29:46 2011 +0100 sna/gen[23]: Fix compilation with debugging enabled Signed-off-by: Chris Wilson commit 1ec6a0e27753027a68beeaff0276ae65c658d468 Author: Chris Wilson Date: Mon Oct 3 16:14:18 2011 +0100 sna: Move the source to the GPU if it is reused We attempt to skip upload a source pixmap to the GPU in the event it is used only once (for example during image upload by firefox). However, if we continue to use the CPU source pixmap then it obviously was worth uploading to the GPU. So if we use the CPU pixmap a second time, do the upload and then blit. Signed-off-by: Chris Wilson commit 413c9f7111c605a3842d4d0a85f8cf01470e5faa Author: Chris Wilson Date: Mon Oct 3 00:24:50 2011 +0100 sna/blt: Suppress repeated SETUP_BLT Signed-off-by: Chris Wilson commit 1067335305a6b8123217518085f70a77b7f0334f Author: Chris Wilson Date: Mon Oct 3 00:11:55 2011 +0100 sna/blt: SETUP_BLT needs 9 dwords of batch space, not 3! Signed-off-by: Chris Wilson commit d8c96a6a1d50cd3dd40fd4b78958c8f78337a2fb Author: Chris Wilson Date: Sun Oct 2 12:57:53 2011 +0100 sna/blt: Use SETUP_MONO to reduce the number of fill relocations Signed-off-by: Chris Wilson commit 4d227d43f0dfca1fa1822207d68539635aac37b3 Author: Chris Wilson Date: Sun Oct 2 12:03:21 2011 +0100 sna/accel: Correct syntax for constifying BoxPtr Signed-off-by: Chris Wilson commit 3dd80524160d6b6cf0fb668afcc766f2bc6f8383 Author: Chris Wilson Date: Sun Oct 2 09:59:03 2011 +0100 sna/accel: Only throttle after flushing Signed-off-by: Chris Wilson commit 04b8f0a5a16ff71c5e55e0281dbcaa32554703ec Author: Chris Wilson Date: Sun Oct 2 09:36:38 2011 +0100 sna/accel: Add a compile option to force flushing on every blockhandler Signed-off-by: Chris Wilson commit 32cef71efe231692f7b24f66913a8ae57bc00aa8 Author: Chris Wilson Date: Sun Oct 2 08:55:55 2011 +0100 sna/accel: Add an compile option to disable use of spans Using spans has a tremendous effect (~100x) on x11perf, some good but mostly bad. However, in reality operations are mixed and so preventing migration on alternate opertaions is a win. In the x11perf slowdowns, it appears to be CPU bound and so it seems like there should be plenty of scope for recovering the lost performance. However, for the time being, just go back to the old fallbacks. Signed-off-by: Chris Wilson commit dc1ec0dd1a5f66114ee4249309d3a283ad84d584 Author: Chris Wilson Date: Sat Oct 1 20:11:47 2011 +0100 sna/accel: Only disable the flush migitration if nothing changed Previously we ignored updating the scanout in place, and so we were not amoritizing the shadow cost of common core rendering operations. Signed-off-by: Chris Wilson commit c6acf1325833b8679ef09ab74f0cb0fd82a8cd92 Author: Chris Wilson Date: Sat Oct 1 11:26:54 2011 +0100 sna/accel: Micro-optimise sna_fill_spans_blt Signed-off-by: Chris Wilson commit 8029765515399b130bee18db0a2830eb83f47a07 Author: Chris Wilson Date: Sat Oct 1 11:19:49 2011 +0100 sna/accel: Don't attempt converting to spans if we will only fallback As the span code does not yet handle plane masks or stippling, it is disadvantageous to convert to spans only to fallback. Signed-off-by: Chris Wilson commit cd11bd69f4c1b961593886945946c7d7c9269900 Author: Chris Wilson Date: Sat Oct 1 10:45:18 2011 +0100 sna/accel: Use miPolyArc to convert arcs into spans for gpu bo This is actually tricker than it looks since miPolyArc() sometimes uses an intermediate bitmap which performs worse than the fbPolyArc() fallback. Signed-off-by: Chris Wilson commit d07256cc3360cf94d131d9a1b24d9f1c6a70ee04 Author: Chris Wilson Date: Sat Oct 1 10:42:45 2011 +0100 sna/accel: Convert segments into spans similarly to PolyLine Signed-off-by: Chris Wilson commit d09a229e32468c810f5056bfd0043f2fb68038d3 Author: Chris Wilson Date: Sat Oct 1 10:01:21 2011 +0100 sna/accel: Use the mi*Line routines to convert the line into spans for gpu bo Signed-off-by: Chris Wilson commit e7a662b92e3012735f7a3d72d4f7860ca61a3810 Author: Chris Wilson Date: Sat Oct 1 09:26:52 2011 +0100 sna: Sort partials by remaining space Signed-off-by: Chris Wilson commit 13b9b5d8d6b6f6db59fe2418270ac93d9b74436c Author: Chris Wilson Date: Sat Oct 1 09:23:41 2011 +0100 sna/io: Only mark the buffer as LAST if we know we will flush the IO Otherwise we can continue to batch up the data upload into larger buffers. Signed-off-by: Chris Wilson commit 7ecc6993b8c8205e5896ee87c14d6a76310c9aa7 Author: Chris Wilson Date: Fri Sep 30 16:11:05 2011 +0100 sna/gen6: Fix offset of Scan-Line-Compare register Reported-by: Frank Mariak Signed-off-by: Chris Wilson commit d8fe941bc245e24c83c417ccff5c57e83baac3f7 Author: Chris Wilson Date: Wed Sep 28 23:04:03 2011 +0100 sna: Check for request retires after every batch In the beginning, I did perform a retire after ever batch. Then I decided that it was too much CPU overhead for too little gain. On reflection, i.e. further benchmarking, we do see a performance improvement for recycling active buffers faster. Signed-off-by: Chris Wilson commit e74a39b45446e2c43df5419efc4a2e0e73275c45 Author: Chris Wilson Date: Wed Sep 28 14:47:33 2011 +0100 sna/gen7: Confusion reigns as trying to fix errors found by an outdated checker Signed-off-by: Chris Wilson commit 6395894ada6b9c14deb62814ccf55848eaa80527 Author: Chris Wilson Date: Tue Sep 27 23:26:33 2011 +0100 sna/gen7: Fix up a couple instances of my inability to count Signed-off-by: Chris Wilson commit a53538659d5e18cb5a7a7a7dbd460f35985183fd Author: Chris Wilson Date: Sun Sep 25 12:40:01 2011 +0100 sna/accel: Fix s/x/y/ typo in computing relative drawing coordinates Reported-by:Roman Jarosz Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=41165 Signed-off-by: Chris Wilson commit 960688d1684dd062c345aa8df1ff4f105015f467 Author: Chris Wilson Date: Sun Sep 25 12:23:10 2011 +0100 sna/accel: Debug option to force CPU fallbacks Signed-off-by: Chris Wilson commit 4fd46b8bb7e7a104a0afa0e5dee92993e043ef57 Author: Chris Wilson Date: Sun Sep 25 10:02:02 2011 +0100 sna/glyphs: Add glyphs directly onto a client temporary buffer Signed-off-by: Chris Wilson commit 5154e281ed66334a8ff6342236c1203a1e4ac5a3 Author: Chris Wilson Date: Sun Sep 25 09:56:16 2011 +0100 sna/glyph: Avoid useless attempt at GPU glyph rendering to a1 destinations The actual bug is a little involved as we don't damage the temporary glyph mask correctly presuming that we only hit GPU paths. However, should we fail to prepare the composite operation that paints the mask on to the destination, things fail horribly. One particular example is that wine like to create its own temporary a1 buffer for the glyphs (which we render to via another temporary mask...) which triggers the delayed fallback and then sw compositing with a random buffer. Reported-by: Roman Jarosz Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=41165 Signed-off-by: Chris Wilson commit 46fedf0cf1a413381109c1de34619e84ad614b8a Author: Chris Wilson Date: Sun Sep 25 09:38:43 2011 +0100 sna/kgem: Check all operation bo in a single amalgamation Signed-off-by: Chris Wilson commit ccf6547a8f1b77ddf4e79bdefe2a0878c0e9b256 Author: Chris Wilson Date: Sat Sep 24 23:16:57 2011 +0100 sna: Paranoid debug flush after every op (as well as before) Signed-off-by: Chris Wilson commit 02337600345951c0ea629db11f0484403a260991 Author: Chris Wilson Date: Sat Sep 24 23:03:52 2011 +0100 sna/gen5: Debug option to disable state caching Signed-off-by: Chris Wilson commit af4d3853ae8f834fbf2fc55fb2b397b3f7f17656 Author: Chris Wilson Date: Sat Sep 24 19:26:46 2011 +0100 sna/glyphs: Convert all sub-8bpp masks to a8 Signed-off-by: Chris Wilson commit c79e90da71ae58cb3804f8abaf0541350e47eb14 Author: Chris Wilson Date: Sat Sep 24 19:26:31 2011 +0100 sna: Add a debug option to disable caching Signed-off-by: Chris Wilson commit 6bbb88af096e054877409a54d0e0a4ccf5ee317e Author: Kenneth Graunke Date: Sat Sep 24 09:27:33 2011 +0100 Fix incorrect maximum PS thread count on IvyBridge I mistakenly set GEN7_PS_MAX_THREAD_SHIFT to 23; it's actually 24 on Ivybridge. Not only did this halve our thread count, it caused us to write 1 into a bit 23, which is marked as MBZ (must be zero). Furthermore, it made us write an even number into this field, which is apparently not allowed. Apparently we were just lucky it worked. Signed-off-by: Kenneth Graunke Signed-off-by: Chris Wilson commit 7f7f95abbf57e6e71f6a30d917f97c2f2bd6cea9 Author: Chris Wilson Date: Fri Sep 23 11:38:18 2011 +0100 sna/accel: Use the PolyFillRect to handle tiled spans Would be preferrable to duplicate the tiling logic. Leave the task of reimplementing XAA to another day! Signed-off-by: Chris Wilson commit 964c96b181e6b55f4de6a4011a5beb6cc2cac33d Author: Chris Wilson Date: Fri Sep 23 11:37:48 2011 +0100 sna/accel: Always subtract the enlarged region from the outstanding GPU damage Signed-off-by: Chris Wilson commit c68856f34653ac3e7af900dfbba41a108ffe119e Author: Chris Wilson Date: Fri Sep 23 11:35:50 2011 +0100 sna/accel: Only skip undamaging the GPU for reads Introduced with ac1b83240e5 (sna/accel: Simplify single pixel read-back) Signed-off-by: Chris Wilson commit 5913c90967091124e7c7b262782f0e99cf400eab Author: Chris Wilson Date: Thu Sep 22 17:13:20 2011 +0100 sna/accel: fix assert to include the offset of copy Signed-off-by: Chris Wilson commit 31857068026138fe5edd8656381fc1ae59d3058b Author: Chris Wilson Date: Thu Sep 22 17:12:53 2011 +0100 sna: move an assert Signed-off-by: Chris Wilson commit 61764af13aa3c770d19d51c8ad198cab8a5866f1 Author: Chris Wilson Date: Thu Sep 22 12:59:32 2011 +0100 sna/dri: Bump DRI2INFOREC_VERSION An intervening dri2 patch landed in the sever. Signed-off-by: Chris Wilson commit 12070b8eb1e98cb158802fff3e59f906b1547877 Author: Chris Wilson Date: Thu Sep 22 12:18:26 2011 +0100 sna/dri: Play with marking bo reusable Since we have no recycling of wc pages in the kernel, we try hard to recycle buffers in userspace to avoid GTT thrashing. This requires co-operation between DRI clients and X, which is sadly lacking and so we need to discard any buffer given out to a client after it is finished. We cheat slightly for page-flips and access to the scanout. A further compromise. References: https://bugs.freedesktop.org/show_bug.cgi?id=38732 Signed-off-by: Chris Wilson commit 902a0234191d77816259affe331339caaa9131d6 Author: Chris Wilson Date: Mon Sep 19 20:34:15 2011 +0100 sna: Use pot buckets for active cache On x11perf one of the major hotspots is the search through the active list for an object large enough to reuse as the target surface. We can eliminate that overhead by keeping those active objects in pre-sorted lists by size. Signed-off-by: Chris Wilson commit 7bb2c5c4d55560f7901fd8036cab164460f81edc Author: Chris Wilson Date: Mon Sep 19 19:12:46 2011 +0100 sna: Move an assertion to the point of import All the guesswork is so that when we require an inactive bo, we do actually get a buffer that is not currently on a GPU active list. For some unresolved reason, this assertion was firing when putting the buffer onto the inactive list - so just workaround the worrisome issue by delaying the check until use. Signed-off-by: Chris Wilson commit 686a5ec52c9ae475fac22c36fff18814372def71 Author: Chris Wilson Date: Mon Sep 19 18:14:05 2011 +0100 sna/trapezoids: Fix overflow during sorting of mono edge step We were tracking the 32bit value of the prev_x using only a 16bit variable, and so failing to sort the edges after advancing to the next scanline. Fixes cairo a1-clip-fill-rule. Signed-off-by: Chris Wilson commit 0390105bc239bf2ac22189f39fccc9d98bae4992 Author: Chris Wilson Date: Mon Sep 19 12:48:38 2011 +0100 sna: compile fixes for debugging Update the DBG messages to reflect changes in function parameters. Signed-off-by: Chris Wilson commit 7f0cba3a255eb9e76d56201b26d15f2d49134b58 Author: Paulo Zanoni Date: Fri Sep 16 18:56:00 2011 -0300 Remove useless assertion Removes 17 instances of: warning: comparison of unsigned expression >= 0 is always true Signed-off-by: Paulo Zanoni Signed-off-by: Chris Wilson commit 2d9effca142b1d82cd1a7ad3b7d2a61c4567801b Author: Paulo Zanoni Date: Fri Sep 16 18:53:01 2011 -0300 Fix "always false" conditionals Enums are unsigned by default in gcc and we can't rely on any specific signedess for the other compilers. i965_render.c: In function ‘i965_prepare_composite’: i965_render.c:2018:2: warning: comparison of unsigned expression < 0 is always false i965_render.c:2025:2: warning: comparison of unsigned expression < 0 is always false i965_render.c:2050:3: warning: comparison of unsigned expression < 0 is always false i965_render.c:2057:3: warning: comparison of unsigned expression < 0 is always false Signed-off-by: Paulo Zanoni [ickle: take advantage and rename the enum values] Signed-off-by: Chris Wilson commit 7bcd1d2b75d280cecfc23877bb13426a28c07a1f Author: Chris Wilson Date: Sat Sep 17 21:38:46 2011 +0100 sna: perform a warnings reduction pass Didn't spot anything that might have led to a genuine bug, but this should help improve the signal-to-noise ratio of warnings in the future. Signed-off-by: Chris Wilson commit a5b5b4ebb14cab19258dd1cbc12721aaa9df226d Author: Chris Wilson Date: Sat Sep 17 00:28:30 2011 +0100 sna: Apply the screen offset for direct upload of a composited pixmap Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=38733 Signed-off-by: Chris Wilson commit daeb9291e22d94cd2f81b3b56418f0f879531da5 Author: Chris Wilson Date: Sat Sep 17 00:27:55 2011 +0100 sna/trapezoids: Ensure we do not overrun the vertical edges for mono Signed-off-by: Chris Wilson commit 481df2618af3c10877db0610c4cadf6fc509e72f Author: Chris Wilson Date: Sat Sep 17 00:27:32 2011 +0100 sna: Add debug option to force allocaiton of gpu-only pixmaps Signed-off-by: Chris Wilson commit 52f93c95465151d16f588d648a58b1f979e946a3 Author: Chris Wilson Date: Fri Sep 16 23:24:16 2011 +0100 sna/glyphs: Disable the glyph cache for debugging Signed-off-by: Chris Wilson commit c8a144d6f84a00a26785fd518cb890ad77edd45f Author: Chris Wilson Date: Fri Sep 16 22:56:36 2011 +0100 sna/glyphs: Correct offset of fallback-via-mask glyphs Under a compositing manager where we have fun values for both drawable->x/y and pixmap->screen.x/y, we were not drawing either the glyphs into the mask correctly and then failed to composite the mask in the right position on top of the pixmap. Signed-off-by: Chris Wilson commit 1eea2c4e60af535ac6500b1b62367f9f63f2a784 Author: Chris Wilson Date: Fri Sep 16 20:35:03 2011 +0100 sna: Can't free bo from the active list As the active bo is still referenced in the request list, we can not simply free it but need to wait for it to be purged on expiration. Signed-off-by: Chris Wilson commit d743c456db11bd3c0d21948dc3a2eb4b33e46342 Author: Chris Wilson Date: Fri Sep 16 20:34:33 2011 +0100 sna: Debug option to disable tiling Signed-off-by: Chris Wilson commit 8c35df4493c910ef62c66a411c14e5be3b26bcf0 Author: Chris Wilson Date: Fri Sep 16 20:33:57 2011 +0100 sna: Remove bad assert The caller intentionally violates this assert, so it is useless. Signed-off-by: Chris Wilson commit 72ca1fdf2fea80ff4cf003bb025424d15d8bac58 Author: Chris Wilson Date: Thu Sep 15 17:52:45 2011 +0100 configure: Introduce --with-builderstring Julien Cristau disliked my broadcasting of the git tree used to build his distribution package as it bore little relevance to his users. As it is only useful for people installing their own drivers (as a means of sanity checking that they are running the right driver), we introduce the --with-builderstring idiom borrowed from the xserver. This allows the builder to override the use of `git describe` and either leave it blank or to fill it with something useful for their own purposes. Signed-off-by: Chris Wilson commit 544a4167c156eac7b630f90ddef30a6f00829835 Author: Chris Wilson Date: Wed Sep 14 18:39:37 2011 +0100 sna/trapezoids: Amalgamate adjacent mono spans Signed-off-by: Chris Wilson commit 3565c48c4bb77c836d817de75d098791dbb529d3 Author: Chris Wilson Date: Wed Sep 14 17:45:41 2011 +0100 sna: Yet another s/x/y/ typo Every time I do a transformation into pixmap space I like to include one of these copy'n'paste errors. Reported-by: Paul Neumann Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=40850 Signed-off-by: Chris Wilson commit 4d4aca21fd32a367d78dfaf1c23b4faeabe36428 Author: Chris Wilson Date: Wed Sep 14 13:06:26 2011 +0100 sna: assert that the memcpy upload path points to valid regions Signed-off-by: Chris Wilson commit 83961051789e418f6022096b1ed2302cb2f05e5b Author: Chris Wilson Date: Tue Sep 13 18:37:01 2011 +0100 sna/gen3: Reset the shader after playing video Another piece of state we zap without marking as dirty when playing video. Reported-by: Paul Neumann Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=40842 Signed-off-by: Chris Wilson commit 34c03f22158d8fc1ec2fc56cb5e87f74e42d9a8f Author: Chris Wilson Date: Tue Sep 13 17:24:14 2011 +0100 sna: Clear the list of buffers upon server regen Or else we may try to clear the new framebuffer with an invalid batch, because it will reuse the same bo as last time and that bo may still think it is part of the old batch. Signed-off-by: Chris Wilson commit 831cdb837189e92aee80030dec553f28e53ba0bc Author: Chris Wilson Date: Tue Sep 13 17:09:38 2011 +0100 sna/debug: Assert that BLT commands on gen2/3 are properly fenced Signed-off-by: Chris Wilson commit a9242a77f0952d029be57d75615dd107af660de1 Author: Chris Wilson Date: Tue Sep 13 16:20:40 2011 +0100 sna: Add a couple of debug statements for why we may not flush the scanout Signed-off-by: Chris Wilson commit 0056bfd11e9930c1bcad6c186cd80fbab3694a8d Author: Chris Wilson Date: Tue Sep 13 14:44:48 2011 +0100 video: Stop advertising unsupported Xv attributes We have yet to implement a yuv-shader that applies contrast/brightness/saturation and so should not advertise such features, potentially allowing the client to fallback and perform the adjustments itself. Signed-off-by: Chris Wilson commit 84a7c11a8134dfd040d2f90bb1e0670aa2c89962 Author: Chris Wilson Date: Tue Sep 13 14:42:33 2011 +0100 sna/video: Stop advertising unsupported Xv attributes Signed-off-by: Chris Wilson commit a0d533ea8d098dbbd86af166de9c07ce67050c35 Author: Chris Wilson Date: Tue Sep 13 12:23:10 2011 +0100 sna/trapezoids: Implement a mono-rasteriser to skip opacity spans For mono-rasteriser we can simply emit the composite spans without requiring an opacity shader. For single trapezoids, it will be more interesting to emit triangles directly. However, we still need to compute the union of many trapezoids, and this builds upon existing code. Signed-off-by: Chris Wilson commit 343b99d2a85c520c81d1408534b2531fcbedfaae Author: Chris Wilson Date: Tue Sep 13 11:39:39 2011 +0100 sna: Reuse the smallest bo that fits the request When scanning the active list, pick the smallest that we can reuse. Signed-off-by: Chris Wilson commit 90071454d24e7230d00c49f5a897296298fc1f22 Author: Chris Wilson Date: Tue Sep 13 11:37:17 2011 +0100 sna: Don't clear the kernel-needs-flush flag if we emit a MI_FLUSH We need to mirror the kernel active lists in order to predict when it will stall upon an access to a bo, and so we cannot clear the needs_flush for our own MI_FLUSH. Signed-off-by: Chris Wilson commit 2deafe2d95d07b574f5a92471f7d7f13aebc790f Author: Chris Wilson Date: Tue Sep 13 11:34:43 2011 +0100 sna/accel: Use correct enum for requesting a write-buffer Signed-off-by: Chris Wilson commit 544b1a678d5bf49559d8422750866685c36cb152 Author: Chris Wilson Date: Tue Sep 13 11:34:25 2011 +0100 sna/gen5: Fix debug compilation Signed-off-by: Chris Wilson commit 6b1ed58d63e9ac80d7d028fa3036633436154816 Author: Chris Wilson Date: Mon Sep 12 19:07:25 2011 +0100 sna/gen5: Avoid bitfields for simple assignments commit 03a7fc16f9ef00ca5591655337621ec67bc37cba Author: Chris Wilson Date: Mon Sep 12 12:20:24 2011 +0100 sna: Avoid the call overhead for the trivial clip case Profile guided micro-optimisation. /o\ Signed-off-by: Chris Wilson commit 2540c877d4811b318e1c2cd707745e5b06c9fa4b Author: Chris Wilson Date: Mon Sep 12 11:54:46 2011 +0100 sna/gen6: Prefer RENDER for copies as it compacts better Signed-off-by: Chris Wilson commit f2a617dcc5fbeaf503db4777f0ec7b3d862085b9 Author: Chris Wilson Date: Mon Sep 12 11:54:02 2011 +0100 sna: Use single precision for computing the texcoord scale factor Signed-off-by: Chris Wilson commit c2e19987c333de31b258a3af37a2ed1f63ef949a Author: Chris Wilson Date: Mon Sep 12 10:37:24 2011 +0100 sna: Provide a direct lookup cache of alpha values To reflect recent changes in cairo. Signed-off-by: Chris Wilson commit 04c5a3df02f6f40a904ff4edb927ae6ff0ce6408 Author: Chris Wilson Date: Mon Sep 12 09:20:58 2011 +0100 sna: Fix compilation without timerfd() Based on the patch by Konstantin Belousov. Reported-by: Konstantin Belousov Signed-off-by: Chris Wilson commit ae887c9d1cb2a06c515e054b36464bcf65440f9f Author: Chris Wilson Date: Sun Sep 11 13:59:08 2011 +0100 sna/trapezoids: Prevent a memleak if feed lots of degenerate boxes Signed-off-by: Chris Wilson commit 72fafdfd374eb165e703b791c70592c47e5772fc Author: Chris Wilson Date: Fri Sep 9 23:15:58 2011 +0100 gitignore: add git_version.h Signed-off-by: Chris Wilson commit ac4d4cdbc1336c86a16a2ec3da2ef6e9842c1079 Author: Chris Wilson Date: Fri Sep 9 23:12:59 2011 +0100 sna: Mark the bo as reusable after extracting the handle from the buffer The whole purpose for that little dance was so that we could reuse the bo. However, we left it marked as non-reusable in order for us not to tie up memory with too many buffers and so defeated the purpose of trying to place it into the inactive cache. Signed-off-by: Chris Wilson commit b3429cf12d6e3e8dd89372f8f70c64205d835b0c Author: Chris Wilson Date: Fri Sep 9 18:09:55 2011 +0100 sna/gen3: Use a clear pattern for ill-defined radial gradients Signed-off-by: Chris Wilson commit b6837c21b459a99cb51b0d41906d82604e515dc0 Author: Chris Wilson Date: Fri Sep 9 14:57:09 2011 +0100 sna/gen5: Prefer BLT for solids And Ironlake also fails to live up to the promise that its GPU is fast enough to run simple programs at memory rates. x11perf -trap300 5x fold improvement. No obvious improvement elsewhere yet. Signed-off-by: Chris Wilson commit 27e42b4e1216099f4a6aa7fddeb1e83b87124b6e Author: Chris Wilson Date: Fri Sep 9 14:06:53 2011 +0100 sna: Prefer memcpy_blt over fbBlt We know we have compatible formats since we have a gpu_bo attached to the pixmap, so we can use the simpler direct memcpy rather than calling fbPutZImage/fbBlt. On my i3-330m, this improves putimage500 from 730 to 1100 ops/s. Reported-by: Michael Larabel Signed-off-by: Chris Wilson commit 2e1bf7e1b44db16d0c322f17535fc6a6fa07353b Author: Chris Wilson Date: Fri Sep 9 12:36:32 2011 +0100 sna: Record git-tree used for compilation Hopefully, I have all the dependencies correct for auto-updating and should continue to work with tarballs... The next step is to perhaps include it in the usual version number, perhaps as patch level? Signed-off-by: Chris Wilson commit f73cd955e760064f57f414dce5f39e43022e4ea6 Author: Chris Wilson Date: Fri Sep 9 11:37:32 2011 +0100 sna/trapezoids: Hook up Imprecise AddTraps in lieu of spans Signed-off-by: Chris Wilson commit adde6eab5d0e532633c864830f8edfa63b36ef4e Author: Chris Wilson Date: Thu Sep 8 23:35:33 2011 +0100 sna/trapezoids: Fast upload path for gpu busy bo Signed-off-by: Chris Wilson commit e9ca05331db9849fba39fbd0583db7052fc80a31 Author: Chris Wilson Date: Thu Sep 8 22:54:22 2011 +0100 sna/traps: Use the trapezoid path for AddTraps Usually this will be to CPU-only pixmap, but just on the off-chance that we are stalling for a GPU pixmap just the faster path developed for Trapezoids. Signed-off-by: Chris Wilson commit afdb8aa89a88c62ccd5e26a3ed9adb70fec310a8 Author: Chris Wilson Date: Wed Sep 7 20:10:02 2011 +0100 sna/gen3: Do not assume video updates are always vsync'ed In case the video is running async, then there may be subsequent instructions within the batch and so we do need to mark the clobbered state as dirty when setting up the video frame. Reported-by: Paul Neumann Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=40693 Signed-off-by: Chris Wilson commit 6aee152cb8444bf63c0009a75c6537e9886b6070 Author: Chris Wilson Date: Sun Sep 4 12:57:17 2011 +0100 sna/gen2: Flush the batch when we run out of vertex space Unlike the later gen, we do not yet use a separate vertex buffer and so when can no longer fit a rectangle (and its CA ghost) we must flush the batch. Due to the duplication required for the CA pass, the normal checks to see whether we had sufficient space to add the new command were passing as they failed to take into account the need to submit the whole primitive again. Signed-off-by: Chris Wilson commit 48bfe4e6ded746f32080b83079235a7576f50d7a Author: Chris Wilson Date: Sun Sep 4 12:46:32 2011 +0100 sna/gen2: Improve batch decoder. Signed-off-by: Chris Wilson commit 2cda0aaf397de1a0ca049508c6fa76f2dd4e61e8 Author: Chris Wilson Date: Sat Sep 3 14:46:57 2011 +0100 sna/trapezoids: Check for alignment after projection If after projection onto the Imprecise fast sample grid, the trapezoid becomes a pixel-aligned box, treat it as such and send it down the fast paths. Signed-off-by: Chris Wilson commit db0fb368c135d4fef4ae993df67ed4610a80fd52 Author: Chris Wilson Date: Fri Sep 2 13:13:38 2011 +0100 sna: Add missing implementation for Triangles Feed both into spans and as a mask fallback. Signed-off-by: Chris Wilson commit 695e7115efab96c4467dc926be07dc2609011720 Author: Chris Wilson Date: Thu Sep 1 15:50:29 2011 +0100 sna/trapezoids: Edges may lie out of bounds We cannot assume that the edge lies completely within the target, so we must make sure that the initial prev_x is truly less than any possible value whilst sorting intersections. Signed-off-by: Chris Wilson commit 9a563ea03b6ad87d41bc091c5819e6c462100450 Author: Chris Wilson Date: Wed Aug 31 23:29:00 2011 +0100 sna: Use the shadow buffer for PutImage This is optimising for the x11perf putimage benchmark, but nevertheless, uploading the PutImage directly into the uncached scanout is between 2-20x slower than making a temporary copy in the shaodw buffer and doing a deferred update. Most of the overhead is in the kernel, and should be addressed there (rather than worked around) and a portion is due to the overdraw in the benchmark (which is not likely to be realistic, but then again neither should PutImage be!). The argument for uploading inplace when possible is that given that the buffer already exists on the GPU implies that is likely to be used again in future by the GPU and so we will be uploading it at some point. Deferring that upload incurs an extra copy. The putimage benchmark does not actually use the pixel data and so that extra cost is not being measured. Reported-by: Michael Larabel Signed-off-by: Chris Wilson commit 32fc0c896e0dfd06617c12beda1ccacedf69fb4a Author: Chris Wilson Date: Wed Aug 31 23:21:54 2011 +0100 sna/gen6: Prefer the BLT ring, except for copies on behalf of DRI As demonstrated by the all-important trap300, using the BLT is 2x faster than the RENDER ring for the simple case of solid fills. (Though note that performing the relocations costs 3x as much CPU for 2x GPU performance.) One case that may regress from this change is copywinpix which should benefit from the batching in the RENDER commands, and might warrant revisiting in the future (with realistic and synthetic benchmarks in hand!) However, due to the forced stall when switching rings, we still want to perform RENDER copies on behalf of DRI clients and before page-flips. Checking against cairo-perf-trace indicated no major impact -- I had worried that setting the BLT flag for some clears might have had a knock-on effect causing too many operations that could be pipelined on the RENDER ring to be sent to the BLT ring instead. Reported-by: Michael Larabel Signed-off-by: Chris Wilson commit 5586dd729b153e37f942a285d328a07f3fe7ae16 Author: Chris Wilson Date: Wed Aug 31 09:25:48 2011 +0100 sna/trapezoids: Refactor to project the trapezoid only once And doing so means that we can go back to using the common validity check. Signed-off-by: Chris Wilson commit 3507437cdba62c28772822cce8c4b656448546db Author: Chris Wilson Date: Tue Aug 30 22:13:24 2011 +0100 sna/trapezoids: Reject invalid traps after projecting onto the sample grid If either of the edges are degenerate on the sample grid, then the trap has zero height and must be skipped. (Otherwise if just one edge becomes degenerate than the polygon becomes unbalanced and the rasteriser will implode.) Signed-off-by: Chris Wilson commit 150a0612d545b7fd77bf393efdef1cc92ed91e61 Author: Chris Wilson Date: Tue Aug 30 19:45:22 2011 +0100 sna/trapezoids: Allocate sufficient space for a8 mask for mono traps Oops, a silly cut'n'paste from caused us to allocate an A1 pixmap for mono traps instead of the A8 pixmap that we tried to write to; mayhem ensued. Reported-by: Eugeni Dodonov Signed-off-by: Chris Wilson commit aafe03d3d1cfd5cbdf345cd51436de268ac27b6f Author: Chris Wilson Date: Mon Aug 29 16:47:48 2011 +0100 sna: Retain the GTT space used for an upload buffer In order to retain the GTT space without keeping hold of the memory used for the upload buffer, we have to create a new bo and copy the relevant details across. Signed-off-by: Chris Wilson commit 28c8c5ca14c5b4eab7e2dc58c87050c1019037bb Author: Chris Wilson Date: Mon Aug 29 16:27:35 2011 +0100 sna: Free the buffers immediately upon release They do not appear to have been leaked per-se, but we end up accumulating the unused buffers. A more complicated solution would be to reallocate the handle for retained buffers so that the GTT region can be reused. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=39184 Signed-off-by: Chris Wilson commit 4f2fc00944272243af343a8b99f6eb8e0dfda006 Author: Chris Wilson Date: Mon Aug 29 15:14:41 2011 +0100 sna: Cleanup up the cache upon close To help with leak-chasing under valgrind. Signed-off-by: Chris Wilson commit 0ac4b974b90a5614d1114d5e211cb4b986a89454 Author: Chris Wilson Date: Mon Aug 29 10:46:46 2011 +0100 sna/video: Defend against PutImage to a broken screen Similar to the previous commit, check that the Screen Pixmap is bound to a bo before proceeding. [Note that in this case, the absence of the bo would have been picked up much later after doing all of the setup...] Signed-off-by: Chris Wilson commit 0a74cd77a3f462d5d2707bf1f35663279bcb9be5 Author: Chris Wilson Date: Mon Aug 29 10:41:26 2011 +0100 video: check that the pixmap exists before use Now, the pixmap being used is meant to the Screen pixmap and by rights that has to exists in a GPU buffer! Evidence contrary to the above exists and so we had better check that we have a bo before using... Reported-by: Toralf Förster Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=40439 Signed-off-by: Chris Wilson commit 8216c92d5c49d4b8de5c16b615bca0acece49a70 Author: Chris Wilson Date: Sun Aug 28 22:32:09 2011 +0100 sna/trapezoids: Use the tor scan converter to compute the low precision mask Take of the advantage of the faster mask computation available using the imprecise tor scan converter for chipsets non yet supporting spans. In doing so, limit the ability to full step only for vertical only rows as the small sample grid reduces the benefits of the computationally more expensive full-step. Signed-off-by: Chris Wilson commit aeee6db798e9ded746b399bc08716cff0bd16ef8 Author: Chris Wilson Date: Sat Aug 27 11:43:48 2011 +0100 sna/trapezoids: Reduce imprecise sampling to 4x4 Note this also revealed a subtle bug in the handling of degenerate trapezoids after shrinking to the raster grid. Signed-off-by: Chris Wilson commit ac1b83240e5d4dc5d5bacad3ed49446143cc5589 Author: Chris Wilson Date: Sat Aug 27 11:42:19 2011 +0100 sna/accel: Simplify single pixel read-back The single pixel case is usually assocated with synchronisation of perf clients and so we do not want to incur extra complication along that path. Also the cost of tracking a single pixel of non-damage outweighs its benefit. Signed-off-by: Chris Wilson commit 786a770f528a0daee2971494352672cb89f48384 Author: Chris Wilson Date: Thu Aug 25 19:52:13 2011 +0100 sna/video: Flush the video state at the end of the operation Or in the case where a second command is received prior to the batch being flushed, the vertex data is not flushed and leads to the a miscompution of the number of vertices emitted. Reported-by: Elias Probst Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=40332 Signed-off-by: Chris Wilson commit bd98001a49be061f0f2cb008b515f5505c8a63a4 Author: Chris Wilson Date: Thu Aug 25 14:49:30 2011 +0100 sna: Clear structures across server reset Signed-off-by: Chris Wilson commit 0865acb3ad03239ffdf8cbf3c9fc204b6c97121b Author: Chris Wilson Date: Thu Aug 25 12:47:27 2011 +0100 sna/dri2: Add some debug around the use of the Resource database Signed-off-by: Chris Wilson commit 98b67457caea583de50bff1b11a3c5a2bd694e26 Author: Chris Wilson Date: Thu Aug 25 12:45:43 2011 +0100 sna/display: Destroy shadow data Under certain circumstances the shadow can be destroy after being allocated but before being created. The pixmap is a NULL pointer at that time, but we know that its value should be data, so just use the data pointer instead. Signed-off-by: Chris Wilson commit b9ae4e7e71527f1cf2b6dd6890420da22a22fad9 Author: Chris Wilson Date: Wed Aug 24 21:38:28 2011 +0100 sna/gen3: reset blend state after applying CA pass Otherwise we use the stale value when rendering CA glyphs directly to the front-buffer and subsequent rendering have a tendency to become invisible. (Rendering via a temporary glyph mask has a fortunate side-effect of reseting sufficient state to force the re-emission of the blend state.) Signed-off-by: Chris Wilson commit ef52f6c8c3421a31552d5965c40b4d9cf68f6a5c Author: Chris Wilson Date: Wed Aug 24 11:04:28 2011 +0100 sna/render: allow CLAMP_TO_EDGE for outside samples of extract regions as well When clipping the sample region to the edge of the texture we can also allow the GPU to use CLAMP_TO_EDGE (as well as CLAMP_TO_BORDER) to emulate the RepeatPad mode of the parent texture. (Only the RepeatNormal, RepeatReflect need special treatment with regard to tiling that is not yet handled.) This fixes the recent performance regression due to a slight change in the fish benchmark that caused it to sample outside of the texture atlas for one of its little fish. Signed-off-by: Chris Wilson commit d330f3751e484d28293fc593a20375a31b068bc7 Author: Stefan Dirsch Date: Fri Jul 22 12:21:01 2011 +0200 Fix array size calculation for intel_pci_probe(). commit ccddff087df0c567c28416941b175be81190a1d3 Author: Chris Wilson Date: Mon Aug 8 00:42:21 2011 +0100 sna/trapezoids: Speedup tor rasteriser Faster sorts for the win. Signed-off-by: Chris Wilson commit bfbe36cfea76337689dd8a101ec03469f6d3553d Author: Chris Wilson Date: Thu Aug 11 01:08:49 2011 +0100 sna/gradient: Use a high-precision ramp for a color step rather than fallback Slightly less precise, but the difference should not be observable in practice... Signed-off-by: Chris Wilson commit 0e61e235bf7a926fd4e5b1f5a05b72dce4c450f3 Author: Chris Wilson Date: Wed Jul 13 22:55:31 2011 +0100 sna/damage: Take advantage of marking all-damaged Return early from adding new damage regions if we know that we have already marked it as all-damaged. Signed-off-by: Chris Wilson commit 3a81bb6bafdbd37802dab96b8f05173ec6701d7f Author: Chris Wilson Date: Tue Aug 9 09:42:40 2011 +0100 NEWS: 2.16.0 release Signed-off-by: Chris Wilson commit f4bbbd1dfe59031f864c7d2bbf7bb0252a3aa6a7 Author: Edward Sheldrake Date: Mon Aug 1 14:46:08 2011 +0100 Fix man page formatting Two option sections were not starting at the beginning of a new line. commit 63518c42234001ec96f638af5732c09079a3d682 Author: Chris Wilson Date: Mon Aug 1 13:36:28 2011 +0100 dri: Build fix for xserver-1.7.7 Back in the olden days before the introduction of dixRegisterPrivate(). Signed-off-by: Chris Wilson commit 7976f5144d42a03ccd027908252a600db2631054 Author: Chris Wilson Date: Sat Jul 30 09:21:36 2011 +0100 NEWS: 2.15.901 snapshot Signed-off-by: Chris Wilson commit 2cfb703bbe7ec043d443452f4fa94f06b1ff7266 Author: Chris Wilson Date: Sat Jul 30 09:23:22 2011 +0100 Fix typos for distcheck Signed-off-by: Chris Wilson commit 6f919264da69ed8bce6f2573629d89962a670783 Author: Chris Wilson Date: Sat Jul 30 09:00:06 2011 +0100 sna: Mark the stencil as untiled In reality, Mesa will be treating it as W-tiling, only we have no way of communicating that requirement to the kernel (as not only does the kernel not understand W-tiling, but also the GTT is incapable of fencing a W-tiled region.). Ported from Chad Versace's 3e55f3e88. Signed-off-by: Chris Wilson commit 326a84e832c43cc200a32e091b4e60b8b6a4a7c8 Author: Chris Wilson Date: Fri Jul 22 11:10:26 2011 +0100 sna: Port IVB acceleration code (Xrender + Xv) Based on the superlative work by Kenneth Graunke and Xiang, Haihao. Signed-off-by: Chris Wilson commit 10790921570fea21d6b2bcc07a0204ef315c9182 Author: Chris Wilson Date: Sat Jul 30 08:46:19 2011 +0100 sna: Include the pixmap size in the debug info for moving to cpu Signed-off-by: Chris Wilson commit 5691c8cdec2a5264b8552f096b4a34b97afe4ae5 Author: Kenneth Graunke Date: Thu Jul 14 10:14:53 2011 -0700 render: Enable RENDER acceleration on Ivybridge. Signed-off-by: Kenneth Graunke Acked-by: Eric Anholt Acked-by: Chris Wilson commit 0d92612b2a2782f80196a08eb9a17af906169f18 Author: Kenneth Graunke Date: Thu Jul 14 00:30:09 2011 -0700 render: Update pixel shader state for Ivybridge. Signed-off-by: Kenneth Graunke Acked-by: Eric Anholt Acked-by: Chris Wilson commit 7460ee73d1fd22e6b02ce125f11ac38efff743ce Author: Kenneth Graunke Date: Thu Jul 14 00:16:54 2011 -0700 render: Use Ivybridge variants for 3D pipeline setup. Signed-off-by: Kenneth Graunke Acked-by: Eric Anholt Acked-by: Chris Wilson commit e3a09608716c1dee2554e8fc26d6f77aeec684bd Author: Kenneth Graunke Date: Thu Jul 14 00:04:31 2011 -0700 render: Refactor to use newly shared pipeline setup code in i965_3d.c. Slightly generalize the shared SF and CC code to accomodate both. Signed-off-by: Kenneth Graunke Acked-by: Eric Anholt Acked-by: Chris Wilson commit 682a690bfeeabae710b1392282163eab35b58eed Author: Kenneth Graunke Date: Wed Jul 13 22:52:52 2011 -0700 Xv: Refactor out pipeline setup functions for future reuse in render. While we're at it, make the functions simply take an intel_screen_private pointer directly instead of having to fetch it from ScrnInfoPtr. Also coalesce some gen6/gen7 functions that were 98% identical. Signed-off-by: Kenneth Graunke Acked-by: Eric Anholt Acked-by: Chris Wilson commit 54b3222658a285d26b7800bdc5f8343c918a804e Author: Kenneth Graunke Date: Wed Jul 13 22:27:37 2011 -0700 render: Update 3DPRIMITIVE for Ivybridge. Signed-off-by: Kenneth Graunke Acked-by: Eric Anholt Acked-by: Chris Wilson commit 4e491a1f6d7a15009904cb79e2c0b13cf2d9d6ec Author: Kenneth Graunke Date: Wed Jul 13 22:19:21 2011 -0700 render: Set Address Modify Enable in 3DSTATE_VERTEX_BUFFERS on Gen7. Signed-off-by: Kenneth Graunke Acked-by: Eric Anholt Acked-by: Chris Wilson commit 2787cd66f0d2907110f774392370537df63a96fa Author: Kenneth Graunke Date: Wed Jul 13 22:10:17 2011 -0700 render: Update SAMPLER_STATE for Ivybridge. Signed-off-by: Kenneth Graunke Acked-by: Eric Anholt Acked-by: Chris Wilson commit 1a7e541d0d33727953fa69f1e29fa4eda665a0d9 Author: Kenneth Graunke Date: Wed Jul 13 21:53:27 2011 -0700 render: Update SURFACE_STATE for Ivybridge. Signed-off-by: Kenneth Graunke Acked-by: Eric Anholt Acked-by: Chris Wilson commit 07cc488bcf3f7653cd54928e3cedb1f4f102c5e0 Author: Kenneth Graunke Date: Sat Jul 9 01:36:52 2011 -0700 render: New Ivybridge assembly programs for render acceleration. These are exactly the same as the ones for Sandybridge, but with message registers translated (hopefully) in the same way as Haihao's new programs (m1 == g65). Signed-off-by: Kenneth Graunke Acked-by: Eric Anholt Acked-by: Chris Wilson commit 1b9e82b4b59dd50c1b057f58d71d5c04dcc648a7 Author: Chris Wilson Date: Tue Jul 26 08:29:36 2011 +0100 sna: Revert enabling scan-line wait on SNB Hanging the machine does indeed prevent video tearing. Just not quite what the user expected... Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=39497 Signed-off-by: Chris Wilson commit 6dbbb74bde034f5f00aee0396ccd1e03a6625fbd Author: Chris Wilson Date: Sat Jul 16 14:05:54 2011 +0100 sna: Enable gen6 scan-line waiting The code was ready and waiting, just forgot to turn it on. Signed-off-by: Chris Wilson commit 3e55f3e88b40471706d5cd45c4df4010f8675c75 Author: Chad Versace Date: Mon Jul 18 00:38:00 2011 -0700 dri: Do not tile stencil buffer Until now, the stencil buffer was allocated as a Y tiled buffer, because in several locations the PRM states that it is. However, it is actually W tiled. From the PRM, 2011 Sandy Bridge, Volume 1, Part 2, Section 4.5.2.1 W-Major Format: W-Major Tile Format is used for separate stencil. The GTT is incapable of W fencing, so we allocate the stencil buffer with I915_TILING_NONE and decode the tile's layout in software. This commit mutually depends on the mesa commit: intel: Fix stencil buffer to be W tiled Author: Chad Versace Date: Mon Jul 18 00:37:45 2011 -0700 Signed-off-by: Chad Versace Reviewed-by: Ian Romanick Acked-by: Kenneth Graunke commit 212fa9868767637e8f430485eeb522c99e63fd16 Author: Chris Wilson Date: Wed Jul 13 21:11:14 2011 +0100 Disable adding normal RTF modes for an eDP This is causing a hard hang with 2.6.39+, we don't know why so play safe and disable for the time being. References: https://bugs.freedesktop.org/show_bug.cgi?id=38012 Signed-off-by: Chris Wilson commit 7a695c9f6b919eba8785d62094526762ba2a0449 Author: Chris Wilson Date: Wed Jul 13 10:57:13 2011 +0100 sna: Fast-path single span boxes These are very common when compositing unclipped trapezoids, and the majority of the overhead is in handling the arbitrary number of boxes and misses out on the constant folding the compiler can do if it is known we have just one box. Signed-off-by: Chris Wilson commit 0190964906ad878d469d6021c10cde4e7f5799c3 Author: Chris Wilson Date: Wed Jul 13 10:26:44 2011 +0100 sna/damage: Avoid testing against a completey damaged region Signed-off-by: Chris Wilson commit b929717c89352d4b47dc2b9ce58e26fbbf327201 Author: Chris Wilson Date: Wed Jul 13 09:49:09 2011 +0100 sna/gen3: Tune emit_spans_primitive_constant Signed-off-by: Chris Wilson commit fbdbfaf38d4da5204750d91cf5a3f43307a6ac33 Author: Chris Wilson Date: Wed Jul 13 17:38:41 2011 +0100 sna/glyphs: Discard GLYPH_PICTURE hint if the glyph doesn't fit into the cache If the glyph is too big to fit into the cache, than ideally we do want to keep an associated GPU bo around for future use. As it is too large to fit into the cache, it of reasonable size and there is little wastage in allocating indiviual GPU bo for each oversized glyph. Signed-off-by: Chris Wilson commit 12f52530db1727e2f835e55dc922e5efaf4d3a16 Author: Chris Wilson Date: Wed Jul 13 17:38:22 2011 +0100 sna: Add some extra debugging to the texture upload fallback paths Signed-off-by: Chris Wilson commit a861094c23b5f0df7b889bcc03904e32c4c2c6be Author: Chris Wilson Date: Tue Jul 12 20:17:00 2011 +0100 sna/dri: Fix a couple of typos Somehow these were lost in the rebasing. Signed-off-by: Chris Wilson commit c221d0356db57c10f5d29ca17fdf592724b35385 Author: Chris Wilson Date: Tue Jul 12 15:39:42 2011 +0100 sna/dri: Remove the unused id/type members for Resource tracking ...and reduce it to a simple list. Signed-off-by: Chris Wilson commit 644b1a903301531e1fb59f27952c15e87888c4c6 Author: Chris Wilson Date: Tue Jul 12 11:50:24 2011 +0100 dri: Always initialise resource members of DRI2FrameEvent As we now attempt to always decouple the lists upon freeing the frame event, we need to initialise them along all code paths. Signed-off-by: Chris Wilson commit 32f42358140ee812984149ae52b0df3dfd1778c3 Author: Chris Wilson Date: Mon Jul 11 20:29:53 2011 +0100 sna/dri: Add some simple debugging Signed-off-by: Chris Wilson commit a46598220ebf5d4e629e1e0a7baf47ce144ed2c8 Author: Chris Wilson Date: Mon Jul 11 14:40:30 2011 +0100 sna/dri: Refactor common code for assigning a pixmap to the DRI2 buffer Signed-off-by: Chris Wilson commit 7538be3315b8683b05e8f6b22023baadcc0bc4da Author: Chris Wilson Date: Mon Jul 11 10:50:36 2011 +0100 dri: Enable triple-bufferred pageflips By popular demand. Triple-buffering trade-offs output latency versus jitter. By having a pre-rendered frame ready to swap in following a pageflip, we avoid the scenario where the latency between receiving the flip complete signal from the kernel, waking up the vsynced application, it render the new frame and then for the server to process the swap request is greater than the frame interval, causing us to miss the vblank. The result is that application can become frame-locked to 30fps. Instead, we report to the application that the first frame swap is immediately completed, supply a new back buffer (or else the rendering would be blocked on waiting for the front-buffer to be swapped away from the scanout) and let them proceed to render the second frame. The second frame is added to the swap queue, and the client throttled to vrefresh. (If the client missed the vblank, the swap queue is empty and the client is immediately woken again, whilst the pageflip is pending.) Note, for practical reasons this only applies to page-flipping, for example, calls to glXSwapBuffer() on fullscreen applications. Signed-off-by: Chris Wilson commit 2608a367acba7247e50754c3daeed09ba2e97d05 Author: Chris Wilson Date: Mon Jul 11 16:28:15 2011 +0100 dri: Prevent abuse of the Resource database The Resource database is only designed to store a single value for a particular type associated with an XID. Due to the asynchronous nature of the vblank/flip requests, we would often associate multiple frame events with a particular drawable/client. Upon freeing the resource, we would not necessarily decouple the right value, leaving a stale pointer behind. Later when the client disappeared, we would write through that stale pointer upsetting valgrind and causing memory corruption. MDK. Instead, we need to implement an extra layer for tracking multiple frames within a single Resource. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=37700 Signed-off-by: Chris Wilson commit ab1000821ae881a301fb0e1f2210493ec383e681 Author: Chris Wilson Date: Sat Jul 9 19:44:26 2011 +0100 dri: Remove the shadow copy of attachment Signed-off-by: Chris Wilson commit 9f22ea7ca440c788f8104647356393c96c62d155 Author: Chris Wilson Date: Sat Jul 9 14:54:33 2011 +0100 sna: Clamp results for computing BoxRec coords from xRectangle As the width/height in the rectangle is specified as uint16_t, the result may be larger than is storagable in the int16_t of the box. Of course it would take a really inane client to do attempt to draw something much larger than the largest possible surface... Is it strange that first example I've found to do so is a Java application? Reported-by: Nicolas Kalkhof Signed-off-by: Chris Wilson commit f4c5dc88171435aa04a5d99372913120cdafe92b Author: Chris Wilson Date: Sat Jul 9 14:17:24 2011 +0100 sna/accel: Fix fallback for depth=1 copy A little carelessness with passing down the offsets caused us to incorrectly copy depth=1 bitmaps, as exemplified by gkrellm. Reported-by: Nicolas Kalkhof Signed-off-by: Chris Wilson commit 649ebcef0938c7cde4f3172a639d39488454999a Author: Chris Wilson Date: Fri Jul 8 17:44:58 2011 +0100 sna: A buffer only needs a flush if it remains dirty at the end of the batch Signed-off-by: Chris Wilson commit 625e37b317e4d237bdc9f80d7ab86b5d2badddc7 Author: Chris Wilson Date: Fri Jul 8 18:58:32 2011 +0100 sna/gen3: So we also need to ensure stippling is cleared... My theory that we used nothing that invoked polygon stippling proved baseless. Fixes regression from 3b5971bd2359383cb8326702d80e03bc15d34c69 Signed-off-by: Chris Wilson commit 1e2cae0ab339e477929e429f8ff31b11fcaf2cbf Author: Chris Wilson Date: Fri Jul 8 09:54:43 2011 +0100 sna/gen3: Restore disabling the use of stencil/fog in the invariant One cleanup too far causing spurious results after rebooting. We also need to ensure that the writemask is fully enabled (ie not disabled) as well. Signed-off-by: Chris Wilson commit ec3dd64e739c4c2bbd8141de5275fe8a90f24a57 Author: Chris Wilson Date: Fri Jul 8 00:17:27 2011 +0100 sna/dri: Enable chaining of page-flips Trade off extra frames of latency for extra frames of anti-jitter buffering and loss of completion information; compiz users rejoice. Signed-off-by: Chris Wilson commit a32694b0f07e35c22dd83bcfb828196d5d86628e Author: Chris Wilson Date: Thu Jul 7 23:11:39 2011 +0100 sna/dri: Remove redundant NULL check in reference The buffer has already been dereferenced by this point... Signed-off-by: Chris Wilson commit d180c5f5f781f4fa7f747dd7bb8073a20771cd79 Author: Chris Wilson Date: Wed Jul 6 20:19:00 2011 +0100 sna: Take advantage of the needs_flush tracking on the front buffer Signed-off-by: Chris Wilson commit 0be47f459ba576ed3f93367de18732fb2367357a Author: Chris Wilson Date: Wed Jul 6 13:02:53 2011 +0100 sna: Check against an execbuffer reference before discarding partials Signed-off-by: Chris Wilson commit f6573fe7577e00a1d2706dbc39179ab658c021d2 Author: Chris Wilson Date: Wed Jul 6 12:44:08 2011 +0100 sna: Compute aligned tiled heights for gen2 correctly We were underestimating the height of X-tiled surfaces (and less harmfully overestimating the height of Y-tiled surfaces.) Signed-off-by: Chris Wilson commit d6afd66461ebcdc2e8dcd94b3f46f374d8acf469 Author: Chris Wilson Date: Tue Jul 5 23:12:59 2011 +0100 sna: Reset unused partial buffers Whilst searching for available space on the active partial buffer list, if we discover an unreferenced one, reset its used counter to zero. Signed-off-by: Chris Wilson commit 6e7a0c86419bf6c928837f592784333c25d8b27b Author: Chris Wilson Date: Tue Jul 5 23:05:37 2011 +0100 sna: Discard unused partial buffers If we allocate a partial buffer and then fallback for the operation, the buffer would remain on the partial list waiting for another user. Discard any unused partials at the next batch submission or expiration point. Signed-off-by: Chris Wilson commit 3b5971bd2359383cb8326702d80e03bc15d34c69 Author: Chris Wilson Date: Tue Jul 5 22:20:28 2011 +0100 sna/gen2: Restore invariant ENABLES One deletion too many, unnoticed until the next reboot. Besides the failure to disable logic op and enable colour buffer blending which causes a hang if you subsequently try to enable both, you also need to request texture caching... Signed-off-by: Chris Wilson commit 5fa3e73f2c03bcadb87c1760f76f853c22fdba18 Author: Chris Wilson Date: Tue Jul 5 21:38:08 2011 +0100 sna/gen[23]: Do as the comments suggest and prefer the BLT Signed-off-by: Chris Wilson commit f749ed618e21b56a36a2feb9b4333ec797ec2ae5 Author: Chris Wilson Date: Tue Jul 5 21:37:02 2011 +0100 sna: Reduce tiling if pitch is less than a tile_width/height only on pre-G33 (Note this only applies to 2D pixmaps.) The rationale, borne out by experimentation with cairo-perf-trace, is that on the pre-G33 devices we always need a fence region region for tiled surfaces, i.e. at least .5/1MiB in size, and that combined with the smaller GTT on those devices, we loose the benefit of tiling to the excessive GTT thrashing. Signed-off-by: Chris Wilson commit b9de6a98d33db660ffad3f186c2b0ab2aea71ac0 Author: Chris Wilson Date: Tue Jul 5 17:48:11 2011 +0100 sna: Remove unused aperture_size member Signed-off-by: Chris Wilson commit fd3bc2af471a3f9a73064cb28818fdd3485cffc7 Author: Chris Wilson Date: Mon Jul 4 16:31:20 2011 +0100 sna: Clamp object size to the min of 1/4 of the whole GTT or 1/2 the mappable ... for those pesky early devices whose GTT was no larger than the AGP aperture. Signed-off-by: Chris Wilson commit d294e41a6a49ddaa6f8d7d5cda266168095fe1f5 Author: Chris Wilson Date: Mon Jul 4 13:45:06 2011 +0100 sna: Update flush/retirement lists after a implicit flush for mmap Signed-off-by: Chris Wilson commit 3e53b0f3a365c031bfef8d087bcee0ae1c2d2988 Author: Chris Wilson Date: Mon Jul 4 13:44:28 2011 +0100 sna: Enable relaxed-fencing for gen2 devices (Just as dependent upon non-buggy kernels as gen3...) Signed-off-by: Chris Wilson commit 33ddaf54299979e98b196d479f1b2060d9551cc5 Author: Chris Wilson Date: Mon Jul 4 13:43:48 2011 +0100 sna: Fix gen2 tiled surface sizes Actually use the gen2 path for gen2 devices! Signed-off-by: Chris Wilson commit 9eceddf69f78fc79c02ca75ed10b000beeff2033 Author: Chris Wilson Date: Mon Jul 4 13:42:58 2011 +0100 sna/gen2: fix batch buffer acounting Signed-off-by: Chris Wilson commit 3f80f7edb8c511b94e66159cd59ba9362c3f69a2 Author: Chris Wilson Date: Sat Jul 2 23:16:04 2011 +0100 sna: Manually set to the GTT domain for mmap ...since the kernel no longer does strict coherency. Signed-off-by: Chris Wilson commit f91ee24b2dabb48288d6e81dcdd82191f158e312 Author: Chris Wilson Date: Sat Jul 2 23:05:04 2011 +0100 sna: Trim number of downsample passes If we can fit the entire width or the entire height into the pipeline when downsampling, do so. Signed-off-by: Chris Wilson commit 6db93720a73f59a9857a3c5ab260fab8b957813e Author: Chris Wilson Date: Sat Jul 2 23:04:14 2011 +0100 sna: Don't change tiling modes on replace This was trying to workaround a kernel bug, and instead causes a performance cliff for textures that *need* to be tiled. Signed-off-by: Chris Wilson commit 430c905ef306ece8a4cb19091711a9feae74b00b Author: Chris Wilson Date: Sat Jul 2 18:24:40 2011 +0100 sna: Force tiled modes for large pitches If the surface is so big that the 2x2 texel sampling will cause a TLB miss everytime, i.e. the row pitch exceeeds 4096, then we need to encourage tiling to prevent attrocious performance. For example, try downscaling a 2560x1600 background image on a gen3 device using I915_TILING_NONE... Using slideshow-demo /usr/share/backgrounds/cosmos/whirlpool.jpg, on a PineView netbook, fps goes from under 4 to over 40. Signed-off-by: Chris Wilson commit ae567b783e5af53f17f49dbf58b9be17fcb53737 Author: Chris Wilson Date: Sat Jul 2 00:11:21 2011 +0100 sna: Finer-grained debugging for trapezoids Signed-off-by: Chris Wilson commit 98f2e3855d70c02b05e2721a70ebce0c17e44283 Author: Chris Wilson Date: Sat Jul 2 09:53:11 2011 +0100 sna/video: Downgrade severity of "overlay not found" message We don't need to warn the user that their hardware does not support the video overlay plane (but Jesse is working on it!), but merely inform them that its presence is lacking. Signed-off-by: Chris Wilson commit 01c258718e981f4da949a4e48595a0660d55d66e Author: Chris Wilson Date: Fri Jul 1 23:17:54 2011 +0100 sna/gen2: Add missing stub debug files Signed-off-by: Chris Wilson commit 5c8a108d2c99f542372efb1734c96ddd2af6bc76 Author: Chris Wilson Date: Fri Jul 1 21:24:43 2011 +0100 sna/gen2: Recompute blend pipeline for component-alpha pass Signed-off-by: Chris Wilson commit 121511d3bd63da5eca28edf971aae3d062aed46e Author: Chris Wilson Date: Fri Jul 1 16:54:28 2011 +0100 sna/gen2: Pack solid sources into the default diffuse component Signed-off-by: Chris Wilson commit a303f85c163979276ff13e00b7c56f55b5fd1d07 Author: Chris Wilson Date: Fri Jul 1 16:27:11 2011 +0100 sna/gen2: Remove unused state from invariant setup ... and also some state that gets clobbered when we install the composite pipelines. Signed-off-by: Chris Wilson commit 120c98ac10435c8e848a8337c1f544f81a05cd3a Author: Chris Wilson Date: Fri Jul 1 14:55:44 2011 +0100 sna: Downsample sources 2x too large to fit in the 3D pipeline This is quite trivial to hit given the 2k limits on gen2/gen3. We compromise on image quality by pre-downscaling the source by a fixed factor to make it fit into the pipeline in preference to performing the entire operation on the CPU. Signed-off-by: Chris Wilson commit f6c8c3bb6fd75bca6c7704b7d5869a5d44ce3832 Author: Chris Wilson Date: Fri Jul 1 11:17:32 2011 +0100 sna/gen2: Use specular component for solid spans Signed-off-by: Chris Wilson commit de14e3c8595f9e315dc3ce23ad15b04e118499f4 Author: Chris Wilson Date: Fri Jul 1 09:58:27 2011 +0100 sna/gen2: Add missing render fallbacks for blt ops Signed-off-by: Chris Wilson commit ecbf6bbd27b1205dcf76cfe34ae2a7a3f5ec195a Author: Chris Wilson Date: Fri Jul 1 08:50:58 2011 +0100 sna/gen2: Implement composite-spans Signed-off-by: Chris Wilson commit c89b37d7b43c9e588097b7fadcba3bc13a03f8bc Author: Chris Wilson Date: Thu Jun 30 17:53:49 2011 +0100 sna: Mappable aperture is region 0 on gen2 Signed-off-by: Chris Wilson commit c0434ab49035bf278dad6f5f84a541ea58536fb4 Author: Chris Wilson Date: Thu Jun 30 16:31:28 2011 +0100 sna: Distinguish 830/845 vs 855/865 using the generation id Remove the PCI ID device checks by using the simpler check on the generation id for errata pertaining to 830/845. Signed-off-by: Chris Wilson commit 18d08e49d270b7a05f14a309759c9315e5ab9679 Author: Chris Wilson Date: Tue Jun 28 20:25:46 2011 +0100 uxa/glyphs: Fallback instead of crashing on large strings Not ideal, but being slow is a major improvement over losing data. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=36860 Signed-off-by: Chris Wilson commit 95866bd6bdaeb612732fd43ec1b93e088763086e Author: Chris Wilson Date: Mon Jun 27 21:07:22 2011 +0100 sna/video: Use EXTEND_PAD to avoid mixing in the border color ...which is 0 and appears green around an unaligned YUV-video. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=38723 Signed-off-by: Chris Wilson commit 7446b39f04306ae2ffccdfe8dd6c8bda64a27043 Author: Chris Wilson Date: Mon Jun 27 15:26:49 2011 +0100 sna: Disable XVideo using the TexturedAdapter if the GPU is wedged Signed-off-by: Chris Wilson commit 44cd6ebf43aa5b7f23c03629a11cdf83b6e96e3a Author: Chris Wilson Date: Sun Jun 26 18:41:20 2011 +0100 sna: Typo in debug assertion. Signed-off-by: Chris Wilson commit 5cd7f5b59857f221bd3aa9965b43b938ff7824ed Author: Chris Wilson Date: Sun Jun 26 18:36:05 2011 +0100 sna/dri: Mark the GPU bo as damaged upon a page-flip ...so that CPU pixmap is correctly invalidated for the next readback. For instance, if you were to take a screenshot on a composited destkop. Reported-by: Sitosfe Wheeler Signed-off-by: Chris Wilson commit c4f7959266571f14c43c0f0cea69134bb8556875 Author: Chris Wilson Date: Sun Jun 26 14:09:11 2011 +0100 sna: Only create bo up to half the size of the mappable aperture As we use GTT mappings if writing directly into the tiled buffer and the available aperture is reported by the kernel as the total GTT and not limited to the fenceable/mappable region, we need to manually probe this value and ensure that our creation and fenced routines observe this distinct limit. Signed-off-by: Chris Wilson commit 130350111720a0e4ae825bb9d21a8c481cd2b548 Author: Chris Wilson Date: Sat Jun 25 23:21:20 2011 +0100 sna/dri: Reorganise schedule_swap to not confuse immediate and wait paths Eeek, the wait-for-target-msc was using the immediate swap path, meaning that for copy-swaps the copy was submitting immediately but the client throttled waiting upon the target vblank. What is actually intended is for the presentation to be delayed until the target_msc. Signed-off-by: Chris Wilson commit b460b9645451af84136c5daebbc00c7545de67f4 Author: Chris Wilson Date: Sat Jun 25 16:34:15 2011 +0100 sna/dri: Fix composited copy-swaps The secret is not to cheat and render directly to the front buffer, but remember to mark the Window as damaged. Signed-off-by: Chris Wilson commit 28fffbd1d07890c8b4d697369159f6a30f267675 Author: Chris Wilson Date: Sat Jun 25 16:33:31 2011 +0100 sna/display: Protect against drmModeGetCrtc returning NULL Signed-off-by: Chris Wilson commit 2c73b4651a04390668c3ed2aa10699663e44fd8a Author: Chris Wilson Date: Sat Jun 25 16:32:30 2011 +0100 sna/gen4+: Use the drawable rectangle offset for copy boxes Saves a little bit of work whilst emitting the rectangles. Signed-off-by: Chris Wilson commit 3833ff967766b0b99f1d636c6453de1783a90586 Author: Chris Wilson Date: Sat Jun 25 14:02:50 2011 +0100 sna: Correct typo in computing damage of PolyPoint Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=38614 Signed-off-by: Chris Wilson commit 471115a9806162bf0cafe27947f49d94fdd63d87 Author: Chris Wilson Date: Fri Jun 24 13:32:45 2011 +0100 sna: Also allow BLT copies to discard the alpha channel Signed-off-by: Chris Wilson commit c3b1a0d7046a83b6daec03e5a562116e3adf3c71 Author: Chris Wilson Date: Fri Jun 24 12:19:09 2011 +0100 sna/dri: Copy to real front upon swap The front-buffer of a DRI2 drawable, may not in fact be pointing to the scanout pixmap. So override the destination for swapbuffers to update the scanout. Signed-off-by: Chris Wilson commit 669378cd8db8af89d12f1b95a07fc1ead0b1569c Author: Chris Wilson Date: Fri Jun 24 11:53:17 2011 +0100 sna: Wrap the fbcon in a scratch pixmap for render-copy across depth changes Signed-off-by: Chris Wilson commit 33b2ea0392fe944b210ef744398aa22989bfdf33 Author: Chris Wilson Date: Fri Jun 24 11:43:07 2011 +0100 sna: Avoid using the BLT to copy between mismatching depths We either conflated bpp (which fails given a mixture of depth-24 and depth-30 pixmaps) or neglected to check at all. Signed-off-by: Chris Wilson commit 28f6dc24d3c50aa399dd098a9ba6a18a95837800 Author: Chris Wilson Date: Fri Jun 24 11:24:59 2011 +0100 sna: Support depth-30 and some more logging to show the depth Signed-off-by: Chris Wilson commit 312460ea69792594fe9acca8591ead55baea7d99 Author: Xiang, Haihao Date: Thu Jun 23 00:20:03 2011 +0800 Xv: set up pipeline for Xv on Ivybridge The configuration is same as that on Sandybridge, but many state commands are changed Signed-off-by: Xiang, Haihao commit 90a0800588f83d78254136b94bca656c4c24215d Author: Xiang, Haihao Date: Tue Jun 21 14:14:38 2011 +0800 Xv: upload new shaders to GEM objects for Xv on Ivybridge Signed-off-by: Xiang, Haihao commit 00614f712b243a7724cf919f94d204a123d2789a Author: Xiang, Haihao Date: Thu Jun 23 00:09:21 2011 +0800 Xv: update SURFACE_STATE & SAMPLER_STATE for Xv on Ivybridge SURFACE_STATE & SAMPLER_STATE are changed since Ivybridge Signed-off-by: Xiang, Haihao commit 70f884772a3adc5dbf56572499c5fb2e080aa84a Author: Xiang, Haihao Date: Thu Jun 23 00:02:38 2011 +0800 Xv: New shaders for Xv on Ivybridge Redefine some M4 macros, also update the check for intel-gen4asm to support Ivybridge Signed-off-by: Xiang, Haihao commit 3cf423bd3a9483181e59ee87376a9487fa1f923d Author: Xiang, Haihao Date: Wed Jun 22 23:33:08 2011 +0800 Xv: separate fragments from M4 macros It is to prepare for Xv on Ivybridge. The difference from Sandybridge is that all message payload must be in GRF registers instead of MRF registers on Ivybridge. We will only redefine some M4 macros for Ivybridge Signed-off-by: Xiang, Haihao commit 58d7a89b93ba4022f45465e479d2799b8903137a Author: Chris Wilson Date: Fri Jun 24 00:35:30 2011 +0100 sna: Don't render to invalid surfaces Fixes a regression from d0362a. In bypassing the is_wedged checked, we also ended up bypassing the checks that we could indeed render to the target bo. With the result that we were creating GPU buffers for SHM surfaces, something that requires Xserver fixes before we can actually enable... Signed-off-by: Chris Wilson commit 3c1d5bc35e02dfd8234c029dcab5661e1656ca8e Author: Chris Wilson Date: Fri Jun 24 00:34:11 2011 +0100 sna: Clip the fbcon to the frontbuffer ...both to correct the placement of the fbcon into the smaller scanout and to ensure that we correctly clip the boxes to be copied. Signed-off-by: Chris Wilson commit e3b4c10f239f69e5282bde1d3b0c3685b24cf0aa Author: Chris Wilson Date: Thu Jun 23 20:54:28 2011 +0100 sna: Relent and workaround the lack of O_NONBLOCK on /dev/dri/card0 So in order not to block server shutdown, check that the fd is readable before attempting to read from it. Signed-off-by: Chris Wilson commit 33d3077266942b23c5bc05ef4a9a6cebe357272a Author: Chris Wilson Date: Thu Jun 23 15:53:45 2011 +0100 sna: Debug compile fix, and some extra comments Signed-off-by: Chris Wilson commit 03e6aee8fd2cc7486fbd2ec55a537c712786486b Author: Chris Wilson Date: Thu Jun 23 15:51:59 2011 +0100 sna: For an unbounded op, we need to convert the whole surface Otherwise if we fail to check, then we create a 0x0 surface to sample with the operation -- net result is that we end up using a clear source instead of the desired mask. Signed-off-by: Chris Wilson commit 356ded1308e8fe5f8dbec217b219228a991356d5 Author: Adam Jackson Date: Wed Jun 22 17:17:15 2011 -0400 Don't include xf86Priv.h I'd like it to not be in the SDK anymore, and we're not using anything from it. Signed-off-by: Adam Jackson Signed-off-by: Chris Wilson commit 8fb98e22a63a4354b049ace7870abb6406541c28 Author: Jian Zhao Date: Wed Jun 22 08:54:59 2011 +0800 configure: Define required_xorg_xserver_version later so it can make successfully. commit ea71133da78632d4cfee5b0b4c96e8dddd6cdf44 Author: Chris Wilson Date: Tue Jun 21 12:28:13 2011 +0100 sna/video: Use pwrite for upload of unclipped, unrotated frames Signed-off-by: Chris Wilson commit 2f6afb5b1f02cc448da1b342627108ceddda4f0d Author: Chris Wilson Date: Mon Jun 20 17:29:17 2011 +0100 sna: Only reduce gpu damage after migration Otherwise we end up doing work for no reason, ala the regression in grads-heat-map. However, it is important that the damage is reduced at some point or else it may grow unbounded. Hopefully normal usage will never hit the pathologocial case... Signed-off-by: Chris Wilson commit 88d840b53d64e67cfee14cc4cef8da46d716c401 Author: Chris Wilson Date: Mon Jun 20 15:22:57 2011 +0100 sna/gen4: Check for sufficient batch space before emitting pipelined pointers For the broken shader workaround, we need to reset the pipeline state before every blt as the state is changed by the magic CA pass. So we need to check that we actually have sufficient space to emit the pipelined pointers before doing so. Signed-off-by: Chris Wilson commit 84bd6e51199bf8b49e6f6e0c29a1d71c89f6996f Author: Chris Wilson Date: Mon Jun 20 15:22:00 2011 +0100 sna/gen4: Correct the number of dwods required to emit the ca pass We need 7 for pipelined pointers, 6 for the primitive and 1 for the flush. Signed-off-by: Chris Wilson commit 22255335e428cb1f349fe4198d36e0ab3fe06948 Author: Chris Wilson Date: Mon Jun 20 13:51:12 2011 +0100 sna: Don't perform a GPU copy of the scanout if it is wedged. Signed-off-by: Chris Wilson commit e2684498a4905b4ac157a1c752c1c4f75301e7ad Author: Chris Wilson Date: Mon Jun 20 13:50:43 2011 +0100 sna: Add some debugging asserts for handling wedged GPUs Signed-off-by: Chris Wilson commit d0362a20bb3ead2fab9401b63209f917bdd72d55 Author: Chris Wilson Date: Sun Jun 19 19:33:01 2011 +0100 sna: Bypass the gpu-is-wedged? check for forced bo creation In order to attach a scanout (and use swrast) on a wedged GPU, we need to bypass the safety checks for normal pixmap creation. Signed-off-by: Chris Wilson commit 585667c2f9f88554ed89ff21ae38600f761d964c Author: Chris Wilson Date: Sat Jun 18 15:52:22 2011 +0100 sna: Bump the required xserver version to 1.10 SNA requires some pending bug fixes to the xserver so it makes little sense to conditionalise the code and deliberately cause broken behaviour. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=3843 Signed-off-by: Chris Wilson commit de0286ed9f7814f903d56e488c7a4b3eb0ca56b0 Author: Chris Wilson Date: Fri Jun 17 09:40:53 2011 +0100 sna/display: After copying the fbcon, tell the server that we have a background ... so that the core knows to skip the clear. Signed-off-by: Chris Wilson commit f3869d60cc3bc349d1efba642baa011378f1c09b Author: Chris Wilson Date: Fri Jun 17 09:37:15 2011 +0100 sna/display: Apply damage for the fbcon copy ... so that any immediate shadow usage will read back the fbcon contents. Signed-off-by: Chris Wilson commit 39f0b0ae17d00402971bb62555d3807bdf39ee93 Author: Chris Wilson Date: Thu Jun 16 23:49:06 2011 +0100 sna: Copy the fbcon contents onto the front buffer upon X startup This patch has been carried by the distributions every since they started doing graphical boot splashes. Time to integrate it and give it some TLC. Signed-off-by: Chris Wilson commit 37ba33f502af0878f6cb75f890c9fe288d0f7e41 Author: Chris Wilson Date: Fri Jun 17 07:03:25 2011 +0100 sna: Fix kgem_create_from_name error paths and to mark the bo as non-reusable Signed-off-by: Chris Wilson commit 665bc1735748f266fad5255f82c40edc59ef98e5 Author: Chris Wilson Date: Fri Jun 17 07:00:54 2011 +0100 sna/display: Remove the caching of the drmModeCrtc We only use it for the id. Everything else stored on it, like the buffer_id, is not permanent and we need to query the current status as required. Signed-off-by: Chris Wilson commit ad6235cfb11a5dfdd27106ac45106aaacba73bde Author: Chris Wilson Date: Thu Jun 16 19:02:36 2011 +0100 sna/accel: convert BOX_ADD_RECT to an inline function Signed-off-by: Chris Wilson commit a86f43988c4ded10cc3b1b964668300c9aa3af59 Author: Chris Wilson Date: Thu Jun 16 19:00:09 2011 +0100 sna/accel: convert BOX_ADD_PT to an inline function Signed-off-by: Chris Wilson commit f67a26590aadac8b9b9870ae3b03d23eb7eda1a2 Author: Chris Wilson Date: Thu Jun 16 18:58:38 2011 +0100 sna/accel: convert TRIM_AND_TRANSLATE_BOX to an inline function Signed-off-by: Chris Wilson commit df1b117dedfdda425fbc3e4ee394169902df4716 Author: Chris Wilson Date: Thu Jun 16 18:57:38 2011 +0100 sna/accel: convert TRANSLATE_BOX to an inline function Signed-off-by: Chris Wilson commit 76ff3c715cf6059316d08217e7e7783ca0388393 Author: Chris Wilson Date: Thu Jun 16 18:56:24 2011 +0100 sna/accel: convert TRIM_BOX to an inline Signed-off-by: Chris Wilson commit e4a66368e71df288323fce66b27237f1d9f93c3a Author: Chris Wilson Date: Thu Jun 16 18:55:03 2011 +0100 sna/accel: convert CLIP_BOX to an inline function Signed-off-by: Chris Wilson commit d7b51cc5d6e624a72b2f2fb92b4488e49a1c749c Author: Chris Wilson Date: Thu Jun 16 18:52:56 2011 +0100 sna/accel: convert BOX_EMPTY macro to an inline function Signed-off-by: Chris Wilson commit 1c3c08e80f08adb12a03277a60c4a56115281743 Author: Chris Wilson Date: Thu Jun 16 18:37:43 2011 +0100 sna: Fix the typo in BOX_EMPTY BOX_EMPTY(&box) surprisingly compiles and generates interesting code that evaluates to if (0). Oops. Signed-off-by: Chris Wilson commit 17bf0019b6962c06c87a08e4b41799481eaccd26 Author: Chris Wilson Date: Thu Jun 16 14:26:06 2011 +0100 sna: Only retire requests after IO if the bo was active We only want to process retirements if we know we caused a stall waiting for the GPU to release the bo before performing IO. Signed-off-by: Chris Wilson commit b707f18d17564653a2b45adf46a4e8eafcc6ab2c Author: Chris Wilson Date: Thu Jun 16 14:16:06 2011 +0100 sna: Purge the bo cache only if we detect aperture thrashing Signed-off-by: Chris Wilson commit 855ced5a4eb67e287001ae22cb6489639c548cfa Author: Chris Wilson Date: Thu Jun 16 13:24:14 2011 +0100 uxa: Remove the attempt to use the 3D pipeline for core rendering The attempt was still ridden with bugs, such as http://bugs.freedesktop.org/show_bug.cgi?id=28768 http://bugs.freedesktop.org/show_bug.cgi?id=28798 http://bugs.freedesktop.org/show_bug.cgi?id=28908 http://bugs.freedesktop.org/show_bug.cgi?id=29401 A fresh approach was taken with SNA, but in the mean time before that can be enabled downstream, restore correct behaviour. Signed-off-by: Chris Wilson commit dd4ad11e0d432f3a8eff4574a241c16f8463278d Author: Chris Wilson Date: Thu Jun 16 12:53:22 2011 +0100 sna: compile fix. A pox on all those who write trivial patches, attempt to compile on a system not having dri2proto and just assume that all is good. Signed-off-by: Chris Wilson commit 2ff36af7e1841a82fd43a9317d07b3229a309df5 Author: Chris Wilson Date: Thu Jun 16 11:23:16 2011 +0100 sna/accel: Prevent initialising a region with an invalid box pixman complains if you attempt to use a region created with an invalid box, so don't do that. Reported-by: Reinhard Karcher Signed-off-by: Chris Wilson commit 3c02110052ecf69d43577ed8537a8850227da560 Author: Chris Wilson Date: Thu Jun 16 10:30:58 2011 +0100 sna/dri: Add some comments to explain the choice of flags upon the bo Signed-off-by: Chris Wilson commit 881f961a1c78e554e566ef624d6edcf28fee8fdf Author: Chris Wilson Date: Wed Jun 15 19:11:05 2011 +0100 sna: Clear the gpu flag once is not busy and no flush is pending Signed-off-by: Chris Wilson commit a1ee4b930846d4ba9274028c08800b882fc926f1 Author: Chris Wilson Date: Tue Jun 14 23:46:15 2011 +0100 sna/dri: And fix the blit swap paths Following on the recent successes with the pageflip paths. Signed-off-by: Chris Wilson commit 2f675cf402a6d158448a9300779829a10ca991fd Author: Chris Wilson Date: Tue Jun 14 08:13:58 2011 +0100 sna: Split zaphod across the crtcs Since we have no global resource allocator for zaphod mode, that's what RandR-1.4 solves, we have to further constrain zaphod mode to only use one crtc per screen. This also means that you must match the output restrictions within the Screen definitions, noting that the crtc pipe id corresponds with the screen number. Reportede-by: Phillp Haddad Signed-off-by: Chris Wilson commit 49f265915a4bca52724c345f04cddfeaca972655 Author: Chris Wilson Date: Tue Jun 14 07:59:50 2011 +0100 sna: Compile fix for debugging enabled Signed-off-by: Chris Wilson commit 12b8a62652e219de255fdd8b3c801c23274130e9 Author: Chris Wilson Date: Mon Jun 13 20:51:56 2011 +0100 test: Exclude the long running stress tests from make distcheck tinderbox complains when it is left running for days on end... Reported-by: Jeremy Huddleston Signed-off-by: Chris Wilson commit 4173ca344bb7d3fb7c03ebdce0e31aa82dba1192 Author: Chris Wilson Date: Mon Jun 13 22:31:55 2011 +0100 sna: Invalidate the mode if the front pixmap was swapped whilst blanked Signed-off-by: Chris Wilson commit 2158534421f26bcd1eca9ad74ab7a30f27e47a75 Author: Chris Wilson Date: Mon Jun 13 15:27:23 2011 +0100 sna/dri: Accurately track front and pending front for async flips By not tracking the front buffer correctly, i.e. performing the exchange on every swap, GL_FRONT was no longer pointing at the updated buffer and neither was the root pixmap. So both X and GL would read the wrong buffer was the flip was pending. The other issue was that we would feed the old front buffer back to the application as a future back buffer (due to buffer caching) and so the kernel would duly insert a WAIT_EVENT for the pending flip to complete before allowing rendering to affect it. Signed-off-by: Chris Wilson commit 2facaa910798c0e7adb34345e6d6913849b7d8a3 Author: Chris Wilson Date: Mon Jun 13 11:46:04 2011 +0100 sna/dri: Complete the pending async flip even after DrawableGone If the client exits before the async page-flip completes then we still need to decrement the pending flip count in order to correctly flip the scanout next time. Signed-off-by: Chris Wilson commit 86888723901030c5cb9da16cbb03508d4b13332f Author: Chris Wilson Date: Sun Jun 12 22:38:01 2011 +0100 sna: Just do a pointer exchange when flipping with no scanout Signed-off-by: Chris Wilson commit b881d1b16a8beec024ac18eca8308418e34cb05d Author: Chris Wilson Date: Sun Jun 12 20:13:26 2011 +0100 sna: Check that the scanout is still attached before waiting for scanline Signed-off-by: Chris Wilson commit bb08e73c557c379ef131225ffea5aa729ca781d6 Author: Chris Wilson Date: Sun Jun 12 20:07:20 2011 +0100 sna: Prevent tracking scratch pixmaps on older xservers Before the xserver is patched to send notification of release of the scratch pixmaps, we cannot associated GPU resources with them. In practice, the only advantage for doing so is for SHM pixmaps... Signed-off-by: Chris Wilson commit ff262aca9c46b7616c59d8e8f04d33a5bbc6e324 Author: Chris Wilson Date: Sat Jun 11 10:15:14 2011 +0100 sna/dri: Fix tripple-buffering for vblank_mode=0 Signed-off-by: Chris Wilson commit b199bc2b415fd0f06e456e0b1f44dd25ba8dbe02 Author: Chris Wilson Date: Sat Jun 11 09:57:46 2011 +0100 sna/dri: Skip querying current msc for immediate swaps Signed-off-by: Chris Wilson commit 5f4dbcc8bde3690668bde739ac61178037612475 Author: Chris Wilson Date: Sat Jun 11 09:06:50 2011 +0100 sna/dri: Queue page-flips immediately for next vblank Signed-off-by: Chris Wilson commit 00e8b4bec6dc5bb46e6f073af9c1fa34f4a2c2c9 Author: Chris Wilson Date: Sat Jun 11 09:00:16 2011 +0100 sna/dri: For offscreen, swaps just exchange the buffers Signed-off-by: Chris Wilson commit e9397b28eff5be4cd5e8a2ceca80333424f24391 Author: Chris Wilson Date: Sat Jun 11 08:54:04 2011 +0100 sna: Process dri wakeups first The goal is to minimise the latency in receiving the event from the kernel and acting upon it. Signed-off-by: Chris Wilson commit bcad5b21fef4573f3144608364cf5b1ea6d241e3 Author: Chris Wilson Date: Fri Jun 10 21:27:38 2011 +0100 sna: Unbreak configure after last commit I went a step too far... I still need some define in order to switch between uxa/sna at compile time. Signed-off-by: Chris Wilson commit d0d65940b4f4d4993eccb8c66ec7e8b633b331da Author: Chris Wilson Date: Fri Jun 10 19:51:12 2011 +0100 sna: Remove the ability to disable chipset specific code This was a fun little, but pointless, exercise. Signed-off-by: Chris Wilson commit 6f035e80b98e478ff4c59427e50e57ecd710da62 Author: Chris Wilson Date: Fri Jun 10 18:41:56 2011 +0100 sna/dri: Keep reference on swap buffers until the end of the pageflip Signed-off-by: Chris Wilson commit cf6c7901825bb831443615ef21555c5843184bdb Author: Chris Wilson Date: Fri Jun 10 17:49:13 2011 +0100 sna: Use the ScreenPixmap->serialNumber as a generation count DRI2 presumes that the pixmap->serialNumber can be used as unique id. If it changes revokes *all* the buffers, it presumes a new pixmap has been attached to the window, for example after a reconfiguration event (resizing of a window, or a mode switch). However, as we updated the root pixmap upon a pageflip, we were triggering revocations everytime, causing further revocations and massive aperture thrashing. Signed-off-by: Chris Wilson commit 3f19270d6795738b58d9ce1721896b3067b38c9f Author: Chris Wilson Date: Fri Jun 10 15:29:57 2011 +0100 sna/dri: Actually restrict pageflipping to properly aligned fs windows Signed-off-by: Chris Wilson commit 9e9c628708ae838e5df3df89331243f84b297e8e Author: Chris Wilson Date: Fri Jun 10 15:17:18 2011 +0100 sna/dri: Create pixmap suitable for the framebuffer for fullscreen buffers Signed-off-by: Chris Wilson commit b22de0b01453884d5018e7e0faef439421fa63a0 Author: Chris Wilson Date: Fri Jun 10 13:30:54 2011 +0100 sna/blt: Suppress overwriting boxes through fill_boxes() As the introduction of the unrolled fill_boxes() was bypassing the effective optimisation in fill_one(). Signed-off-by: Chris Wilson commit ebeebc2044610e6e31d9cef746c768d2e435f32c Author: Chris Wilson Date: Fri Jun 10 13:07:13 2011 +0100 sna: Use temporary for storing the current crtc box when computing best crtc ... as the caller may be reusing an input parameter for the result. Signed-off-by: Chris Wilson commit b3949d0d7424242703c9e60e2e33908499aa6a43 Author: Chris Wilson Date: Fri Jun 10 12:45:24 2011 +0100 sna/kgem: remove erroneous assert during submit This was only valid for gen6+ with the multiple rings. Signed-off-by: Chris Wilson commit 33abb68da2c49081f28898f93cc4e14702b40d70 Author: Chris Wilson Date: Fri Jun 10 12:44:56 2011 +0100 sna/kgem: remove redundant clears upon initialisation Signed-off-by: Chris Wilson commit cfde920a2e449efaa387850173e46f87d928023b Author: Chris Wilson Date: Fri Jun 10 11:27:04 2011 +0100 sna: Remember to decouple the fb on closing ... so that we actually attach a new one after regen! Signed-off-by: Chris Wilson commit b6192ce95af8aa8be6d9a68a500393f2a73c2f4c Author: Chris Wilson Date: Fri Jun 10 11:05:41 2011 +0100 sna/gen4: Tweak calculation of required batch space for CA Signed-off-by: Chris Wilson commit c9b89fc87f1d43a715cd9c5915bfed957175986d Author: Chris Wilson Date: Fri Jun 10 10:53:29 2011 +0100 sna: Drain the event queues before CloseScreen Signed-off-by: Chris Wilson commit ea560e25dfcce3750339b477a1a3c05f504cf91e Author: Chris Wilson Date: Fri Jun 10 10:44:27 2011 +0100 sna: Workaround the random free of the ScreenPixmap in fbCloseScreen Signed-off-by: Chris Wilson commit 1f364c6d241e8971e8fc1447871c718240e4563e Author: Chris Wilson Date: Fri Jun 10 10:03:01 2011 +0100 sna: Reset the kgem state on server regen Signed-off-by: Chris Wilson commit 95ae39a93467655b2107b82bd7970f732c8de066 Author: Chris Wilson Date: Fri Jun 10 09:57:22 2011 +0100 sna: Add some debug statements to the core DDX interface Signed-off-by: Chris Wilson commit 0594724dc4e9e459240b8b290ddf42ed711be22b Author: Chris Wilson Date: Fri Jun 10 08:18:07 2011 +0100 sna/gen4: Restore normal state after CA pass and FLUSH_EVERY_VERTEX By working around the broken shaders, we emitted the CA rectangle in the middle of a sequence of glyphs and left the state setup for CA. So we need to reset the pipeline state at the start of every composite blt. Signed-off-by: Chris Wilson commit 2dff681efe1b89ffe59e5d33979c6e059b72a786 Author: Chris Wilson Date: Fri Jun 10 08:23:19 2011 +0100 sna/blt: Remove a couple of unused members from its op-state Signed-off-by: Chris Wilson commit fbbd1c9dde3b2fafd8aaebb88daf3c546d86e5a4 Author: Chris Wilson Date: Fri Jun 10 08:22:00 2011 +0100 sna/glyphs: Fix glyphs to dst fallback path commit 3f46c34d534e389f541b5c7987b78981c556d868 Author: Chris Wilson Date: Fri Jun 10 08:20:12 2011 +0100 sna/composite: Tweak setup for the minimal composite extents Clip the operation extents before passing to the backend to setup the operation, so that we only need to upload the minimum amount of data necessary, or avoid tiling etc. Signed-off-by: Chris Wilson commit a34caf1e3ded03dca53bbea3432dcd45101460fd Author: Chris Wilson Date: Thu Jun 9 20:45:18 2011 +0100 sna/gradient: Check solid busy status not current domain We can write into the bo if it has no outstanding requests, whereas we were checking to see it was last on the gpu instead. Signed-off-by: Chris Wilson commit 1638a0a6c0b335dd31def93a9199f7407ae51170 Author: Chris Wilson Date: Thu Jun 9 16:00:30 2011 +0100 sna/dri: Add a comment to warn about a potential race between DRI clients Signed-off-by: Chris Wilson commit 4b54c89fbddc7aa72e5f2c5eded44e84bce70b77 Author: Chris Wilson Date: Thu Jun 9 15:53:11 2011 +0100 sna: Place the pixmap on the DRI dirty list after damaging the shadow copy In order that we remember to flush the contents back to the GPU before we wake up the DRI clients, we need to add those dirty pixmaps to the flushing list. I caught the obvious place, the central move-to-cpu, but I missed a couple of paths were we optimise the copy onto the shadow. Signed-off-by: Chris Wilson commit 190a6786563aadd69f63bb7b509f3edf712e81ef Author: Chris Wilson Date: Thu Jun 9 15:13:17 2011 +0100 sna: Periodically reduce the gpu damage (after migrating any cpu damage) This will prevent the gpu damage list from growing unbounded. Signed-off-by: Chris Wilson commit 38d084ee6241527193ba08c6600b6734e1833118 Author: Chris Wilson Date: Thu Jun 9 14:53:34 2011 +0100 sna: Move !kgem->wedged check down into kgem_can_create_2d Signed-off-by: Chris Wilson commit d1a3bb1467429310f5d17bf6afbf2eeb043ae64d Author: Chris Wilson Date: Thu Jun 9 14:51:49 2011 +0100 sna: Prefer inactive buffers when creating a bo for a dirty pixmap As we will immediately attempt to replace it with an inactive when moving the data to the GPU, short-circuit that replacement. Signed-off-by: Chris Wilson commit 8ee53cc7449de7df294f49d089ff1d6b2e944323 Author: Chris Wilson Date: Thu Jun 9 13:05:19 2011 +0100 sna: Debugging compile fix Still not enough coffee this morning. Signed-off-by: Chris Wilson commit 6f59d7d19053fed4e9d6402a5f03bc8a62cb56af Author: Chris Wilson Date: Thu Jun 9 12:59:59 2011 +0100 sna: Use --enable-debug=full to enable extremely verbose debugging Signed-off-by: Chris Wilson commit 021209d5d3add8b28143611cfad4c5481a2945a3 Author: Chris Wilson Date: Thu Jun 9 10:27:12 2011 +0100 sna: Remove the stubs from sna_render.c These only existed to work around an include order problem, when kgem was intended to be entirely separable from sna. Moving the function pointer into kgem simplifies matters. Signed-off-by: Chris Wilson commit 6141b1aea159759e7e9dcf2561deb9d8c02bd0a2 Author: Chris Wilson Date: Thu Jun 9 10:06:08 2011 +0100 sna: Warn after detecting a hung GPU ...and include the instructional error message from uxa. Signed-off-by: Chris Wilson commit 1786d2ee03a190ebe242ac7b58bed7cc31d66b16 Author: Chris Wilson Date: Thu Jun 9 09:48:57 2011 +0100 sna/accel: Add debug messages for falling back due to a wedged GPU Signed-off-by: Chris Wilson commit 4d509d501b09f565fea232947e6f53d54a08749f Author: Chris Wilson Date: Thu Jun 9 01:03:16 2011 +0100 sna: Tweak retiring old bo As we check for retirement everytime we wakeup, it is seldom useful to check again until we know we have invoked an operation that may block. But when we do check, we do not want to scan the entire active list looking for flushing candidates, so track those on a separate list. Signed-off-by: Chris Wilson commit 17be5e2eb41b2ada94954b87c855961003c2fc7c Author: Chris Wilson Date: Thu Jun 9 00:38:29 2011 +0100 sna: Reduce the frequency of the timer interrupts Signed-off-by: Chris Wilson commit d0b09f10242e09adfddbbe6ee1ea39dbf5d1b32d Author: Chris Wilson Date: Wed Jun 8 20:34:57 2011 +0100 sna: Make the device fd non-blocking So that we can simply query it from each of the Zaphod instances without blocking. Requires a fixed kernel... Signed-off-by: Chris Wilson commit 67ed14c5369cfd967f90e5a3e6053c95a48dd16e Author: Chris Wilson Date: Wed Jun 8 19:42:02 2011 +0100 sna/dri: Set bo=NULL to actually detect unknown dri attachment points. Signed-off-by: Chris Wilson commit 68755bc5ff830dc5ed6a34b18a1217b7d1003ce4 Author: Chad Versace Date: Wed Jun 8 18:34:02 2011 +0100 sna/dri: Add support for DRI2BufferStencil and DRI2BufferHiz Based on Chad's commits 2662db5b and 3daba7e6: Signed-off-by: Chris Wilson commit 0030c3aab6172cdb83ca8f1b7b07c7db5915990d Author: Chris Wilson Date: Wed Jun 8 18:31:12 2011 +0100 sna/dri: Drop support for old dri2 versions with incompatible ABI Signed-off-by: Chris Wilson commit 715d466ad44e82b740f5454c41db944863420596 Author: Chris Wilson Date: Wed Jun 8 13:32:59 2011 +0100 sna/dri: valgrindify Lots of scary warnings found by valgrind. Signed-off-by: Chris Wilson commit a62db5b050dee10246c02c72385358acb5e72b56 Author: Chad Versace Date: Mon May 23 14:23:54 2011 -0700 dri: Add support for DRI2BufferStencil and DRI2BufferHiz And bump configure.ac to require dri2proto >= 2.6, because DRI2BufferStencil and DRI2BufferHiz were introduced in that version. When a client requests DRI2BufferHiz or DRI2BufferStencil, I830DRI2CreateBuffer() now returns a Y-tiled buffer. The stencil buffer is handled as a special case due its quirky pitch requirements. CC: Eric Anholt CC: Ian Romanick CC: Kristian Høgsberg Signed-off-by: Chad Versace commit 263daba7e6afb37bd471fdc8dd8e4458da0a45ff Author: Chad Versace Date: Sun Jun 5 19:48:19 2011 -0700 dri: Do not create DRI2 buffers for unrecognized DRI2 buffer tokens Before this commit, if a client were to request an unrecognized DRI2 buffer, such as DRI2BufferStencil, then I830DRI2CreateBuffer() allocated and returned an X-tiled buffer by accident. The problem was that unrecognized tokens were caught by the default case of a switch statement. Now, when given unrecognized DRI2 tokens, I830DRI2CreateBuffers() returns null. This shouldn't break older Mesa versions, because they never query (via DRI2GetBuffersWithFormat) for the drawable's DRI2BufferStencil. CC: Eric Anholt CC: Ian Romanick CC: Kenneth Graunke CC: Kristian Høgsberg commit 038f29c259c1463c0bff12f5f8f0b228c02e45b3 Author: Chris Wilson Date: Tue Jun 7 19:16:01 2011 +0100 sna/blt: Flush batch before overwrite surface data A left-over from before the surface was embedded into the tail of the batch, we were only checking for room against the total size of the batch buffer. So under the wrong set of circumstances we ended up overwriting surface data with batch and triggering a GPU hang on gen4+. Signed-off-by: Chris Wilson commit 265d94e0aa46b30a3198893544dd3619cc9145de Author: Chris Wilson Date: Tue Jun 7 14:23:51 2011 +0100 sna: Add zaphod support Zaphod support is a rudimentary method for creating an Xserver with multiple screens from a single device. The Device is instantiated, with a duplication of its resources, as many as required up to a maximum of the number of its outputs, and each instance is attached to a Screen and added to the ServerLayout. A Device can be bound to a selection of outputs using a comma separated list of RandR names. Note: in general, this is not the preferred solution! And will be superseded by per-crtc-pixmaps in RandR-1.4. For example, the following xorg.conf fragment creates an XServer with two screens, one attached to the LVDS panel on the laptop, and the other to any external output: Section "Device" Identifier "Intel0" Driver "intel" BusID "PCI:0:2:0" Option "ZaphodHeads" "LVDS1" Screen 0 EndSection Section "Device" Identifier "Intel1" Driver "intel" BusID "PCI:0:2:0" Option "ZaphodHeads" "DVI1,VGA1" Screen 1 EndSection Section "Screen" Identifier "Screen0" Device "Intel0" EndSection Section "Screen" Identifier "Screen1" Device "Intel1" EndSection Section "ServerLayout" Identifier "default" Screen "Screen0" Screen "Screen1" EndSection Based on a patch by Ben Skegs Signed-off-by: Chris Wilson commit ad5ead8257b7dda84a886461873d87831d43e663 Author: Chris Wilson Date: Tue Jun 7 11:06:52 2011 +0100 sna/gen2: Support covered xrgb sources on 830/845 830/845 cannot directly sample from an x8r8g8b8 source, but if we know that we are only sampling from within the confines of the source then we force the alpha channel to one. (Outside of the source we require the sampler to return a==0.) Signed-off-by: Chris Wilson commit 790f90a277ff2a6fbb615b8cd3c06faa0d549c41 Author: Chris Wilson Date: Tue Jun 7 08:18:30 2011 +0100 sna/gen6: Initialise a couple more composite op members for copy_boxes Valgrind detected that I missed initialised a couple of fields for use with the generic state emission paths: ==28683== Conditional jump or move depends on uninitialised value(s) ==28683== at 0x83BE646: gen6_get_blend (gen6_render.c:251) ==28683== by 0x83BF769: gen6_emit_state (gen6_render.c:818) ==28683== by 0x83C38ED: gen6_emit_copy_state (gen6_render.c:2280) ==28683== by 0x83C3C89: gen6_render_copy_boxes (gen6_render.c:2356) ==28683== Conditional jump or move depends on uninitialised value(s) ==28683== at 0x83C15C3: gen6_rectangle_begin (gen6_render.c:1458) ==28683== by 0x83C177D: gen6_get_rectangles (gen6_render.c:1502) ==28683== by 0x83C3D16: gen6_render_copy_boxes (gen6_render.c:2363) Signed-off-by: Chris Wilson commit 4e443cbef589e016d085e8023677f47c1f01a858 Author: Chris Wilson Date: Tue Jun 7 00:24:17 2011 +0100 sna: Add some more debug commentary to render picture source migration Signed-off-by: Chris Wilson commit 367298c5109b47fa4961a60fa3cb454c800f02c3 Author: Chris Wilson Date: Mon Jun 6 23:21:09 2011 +0100 sna: Subtract the Solid RenderFillRectangles from CPU damage ... and so avoid having to move it the GPU, as seen in the wild. It looks like I will actually need to handle mixed Render/Core operations on the frontbuffer. Signed-off-by: Chris Wilson commit 62e4266b4dc9b5c3d165aca980738fa777a0f259 Author: Chris Wilson Date: Mon Jun 6 16:20:38 2011 +0100 sna/gen5: Only emit the non-pipelined op after BLT commands We were always terminating the batch with the non-pipelined op, and not just at the end of a BLT sequence. Signed-off-by: Chris Wilson commit 8f97157d2e807d6c9acb521ec59757f419e6518b Author: Chris Wilson Date: Mon Jun 6 10:21:17 2011 +0100 sna/gen2: Replicate alpha for non-CA masks Signed-off-by: Chris Wilson commit c8a2fa3a2e3f22e12f363e8ea060c70585156190 Author: Chris Wilson Date: Mon Jun 6 08:43:34 2011 +0100 sna/gen2: Correct command length for CA LOAD_IMMEDIATE_STATE_1 Signed-off-by: Chris Wilson commit a89fc7181b6ccec8ccca2cd4ccab4f180f663656 Author: Chris Wilson Date: Mon Jun 6 08:39:17 2011 +0100 sna/gen2: Only emit the mask texcoord if there is a mask Signed-off-by: Chris Wilson commit 4fb7784e1e1b2b999c0c2420583e9c48eb1127ac Author: Chris Wilson Date: Mon Jun 6 08:27:17 2011 +0100 sna/gen3: Non-affine texcoords require space for 4 floats not 3. Signed-off-by: Chris Wilson commit d9344ab8d04fe4795a3a532c5c0e85dc11ecf968 Author: Chris Wilson Date: Mon Jun 6 08:25:53 2011 +0100 sna/gen2: Set op->floats_per_vertex Signed-off-by: Chris Wilson commit 741c1101f17bfa379a15178e4760211737fa18cf Author: Chris Wilson Date: Mon Jun 6 08:22:45 2011 +0100 sna/gen2: Enable selection of gen2 only Signed-off-by: Chris Wilson commit c76ec69660bf40de052159f60c16de8fd87341a8 Author: Chris Wilson Date: Mon Jun 6 08:02:09 2011 +0100 sna/gen2: The inline primitive takes a length, not a vertex count Signed-off-by: Chris Wilson commit 91424d49373d8117e71441fa3221bd1801171692 Author: Eric Anholt Date: Tue May 31 23:29:43 2011 -0700 uxa: Simplify uxa_poly_fill_rect by only clipping once. Reviewed-by: Keith Packard commit e0066e77e026b0dd0daa0c3765473c7d63aa6753 Author: Eric Anholt Date: Tue May 31 23:17:16 2011 -0700 uxa: Simplify Composite solid acceleration for spans by only clipping once. Unlike the previous commit removing this style of code, the code in this one was originally wrong, and would fail to clip in the second pass of clipping when y was > pbox->y2. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=37233 Reviewed-by: Keith Packard commit ace324e4aa27effdd621156eec03f3f87b610732 Author: Eric Anholt Date: Tue May 31 23:13:18 2011 -0700 uxa: Simplify BLT solid acceleration for spans filling by only clipping once. We were clipping each span against the bounds of the clip, throwing out the span early if it was all clipped, and then walked the clip box clipping against each of the cliprects. We would expect spans to typically be clipped against one box, and not thrown out, so we were not saving any work there. For multiple cliprects, we were adding work. Only for many spans clipped entirely out of a complicated clip region would it have saved work, and it clearly didn't save bugs as evidenced by the many fix attempts here. Reviewed-by: Keith Packard commit bdb396a44bf64f823a44aafee19b5ec692f1bf5f Author: Chris Wilson Date: Sun Jun 5 19:41:25 2011 +0100 sna: PutImage: Copy straight to GTT if the bo is idle This saves a copy in the typical PutImage to frontbuffer favoured by flash. And we also happen to fix a bug if we should be requested to PutImage outside of the clip region... Signed-off-by: Chris Wilson commit 407257570fea1c6456fda47c9fab36c39f169c2b Author: Chris Wilson Date: Sun Jun 5 15:33:27 2011 +0100 sna/gen6: Flush the pipeline before effecting a change of blend modes ... also make sure that we flush if we change the blend mode for the CA pass. Reported-by: Ivan Bulatovic Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=37946 Signed-off-by: Chris Wilson commit 73167711226f430cc9ef4c27f267655d0edf4622 Author: Chris Wilson Date: Sun Jun 5 14:56:32 2011 +0100 sna: 915gm does not have 128-byte Y-tiling Signed-off-by: Chris Wilson commit 0260c4ce32fa4d7ae8c6ee79d5488b0114e96251 Author: Chris Wilson Date: Sun Jun 5 14:39:34 2011 +0100 sna: Fallback if presented with mask under NO_COMPOSITE Just making sure that the debug paths actually work... Signed-off-by: Chris Wilson commit fcbe2d9ee79c936cad3ee86836286dcb280f0047 Author: Chris Wilson Date: Sat Jun 4 18:11:54 2011 +0100 sna/gen4: Flush every vertex for the magic CA pass gen4 dies hard if it has two rectangles in the pipeline, and despite the stringent and crippling efforts to prevent us from efficiently using the GPU, I missed a flush before submitting the CA rectangle. Reported-and-tested-by: Fryderyk Dziarmagowski References: https://bugs.freedesktop.org/show_bug.cgi?id=28768 Signed-off-by: Chris Wilson commit bcef98af561939aa48d9236b2dfa2c5626adf4cb Author: Chris Wilson Date: Fri Apr 8 07:17:14 2011 +0100 sna: Introduce a new acceleration model. The premise is that switching between rings (i.e. the BLT and RENDER rings) on SandyBridge imposes a large latency overhead whilst rendering. The cause is that in order to switch rings, we need to split the batch earlier than is desired and to add serialisation between the rings. Both of which incur large overhead. By switching to using a pure 3D blit engine (ok, not so pure as the BLT engine still has uses for the core drawing model which can not be easily represented without a combinatorial explosion of shaders) we can take advantage of additional efficiencies, such as relative relocations, that have been incorporated into recent hardware advances. However, even older hardware performs better from avoiding the implicit context switches and from the batching efficiency of the 3D pipeline... But this is X, and PolyGlyphBlt still exists and remains in use. So for the operations that are not worth accelerating in hardware, we introduce a shadow buffer mechanism through out and reintroduce pixmap migration. Doing this efficiently is the cornerstone of ensuring that we do exploit the increased potential of recent hardware for running old applications and environments (i.e. so that the latest and greatest chip is actually faster than gen2!) For the curious, sna is SandyBridge's New Acceleration. If you are running older chipsets and welcome the performance increase offered by this patch, then you may choose to call it Snazzy instead. Speedups ======== gen3 firefox-fishtank 1203584.56 (1203842.75 0.01%) -> 85561.71 (125146.44 14.87%): 14.07x speedup gen5 grads-heat-map 3385.42 (3489.73 1.44%) -> 350.29 (350.75 0.18%): 9.66x speedup gen3 xfce4-terminal-a1 4179.02 (4180.09 0.06%) -> 503.90 (531.88 4.48%): 8.29x speedup gen4 grads-heat-map 2458.66 (2826.34 4.64%) -> 348.82 (349.20 0.29%): 7.05x speedup gen3 grads-heat-map 1443.33 (1445.32 0.09%) -> 298.55 (298.76 0.05%): 4.83x speedup gen3 swfdec-youtube 3836.14 (3894.14 0.95%) -> 889.84 (979.56 5.99%): 4.31x speedup gen6 grads-heat-map 742.11 (744.44 0.15%) -> 172.51 (172.93 0.20%): 4.30x speedup gen3 firefox-talos-svg 71740.44 (72370.13 0.59%) -> 21959.29 (21995.09 0.68%): 3.27x speedup gen5 gvim 8045.51 (8071.47 0.17%) -> 2589.38 (3246.78 10.74%): 3.11x speedup gen6 poppler 3800.78 (3817.92 0.24%) -> 1227.36 (1230.12 0.30%): 3.10x speedup gen6 gnome-terminal-vim 9106.84 (9111.56 0.03%) -> 3459.49 (3478.52 0.25%): 2.63x speedup gen5 midori-zoomed 9564.53 (9586.58 0.17%) -> 3677.73 (3837.02 2.02%): 2.60x speedup gen5 gnome-terminal-vim 38167.25 (38215.82 0.08%) -> 14901.09 (14902.28 0.01%): 2.56x speedup gen5 poppler 13575.66 (13605.04 0.16%) -> 5554.27 (5555.84 0.01%): 2.44x speedup gen5 swfdec-giant-steps 8941.61 (8988.72 0.52%) -> 3851.98 (3871.01 0.93%): 2.32x speedup gen5 xfce4-terminal-a1 18956.60 (18986.90 0.07%) -> 8362.75 (8365.70 0.01%): 2.27x speedup gen5 firefox-fishtank 88750.31 (88858.23 0.14%) -> 39164.57 (39835.54 0.80%): 2.27x speedup gen3 midori-zoomed 2392.13 (2397.82 0.14%) -> 1109.96 (1303.10 30.35%): 2.16x speedup gen6 gvim 2510.34 (2513.34 0.20%) -> 1200.76 (1204.30 0.22%): 2.09x speedup gen5 firefox-planet-gnome 40478.16 (40565.68 0.09%) -> 19606.22 (19648.79 0.16%): 2.06x speedup gen5 gnome-system-monitor 10344.47 (10385.62 0.29%) -> 5136.69 (5256.85 1.15%): 2.01x speedup gen3 poppler 2595.23 (2603.10 0.17%) -> 1297.56 (1302.42 0.61%): 2.00x speedup gen6 firefox-talos-gfx 7184.03 (7194.97 0.13%) -> 3806.31 (3811.66 0.06%): 1.89x speedup gen5 evolution 8739.25 (8766.12 0.27%) -> 4817.54 (5050.96 1.54%): 1.81x speedup gen3 evolution 1684.06 (1696.88 0.35%) -> 1004.99 (1008.55 0.85%): 1.68x speedup gen3 gnome-terminal-vim 4285.13 (4287.68 0.04%) -> 2715.97 (3202.17 13.52%): 1.58x speedup gen5 swfdec-youtube 5843.94 (5951.07 0.91%) -> 3810.86 (3826.04 1.32%): 1.53x speedup gen4 poppler 7496.72 (7558.83 0.58%) -> 5125.08 (5247.65 1.44%): 1.46x speedup gen4 gnome-terminal-vim 21126.24 (21292.08 0.85%) -> 14590.25 (15066.33 1.80%): 1.45x speedup gen5 firefox-talos-svg 99873.69 (100300.95 0.37%) -> 70745.66 (70818.86 0.05%): 1.41x speedup gen4 firefox-planet-gnome 28205.10 (28304.45 0.27%) -> 19996.11 (20081.44 0.56%): 1.41x speedup gen5 firefox-talos-gfx 93070.85 (93194.72 0.10%) -> 67687.93 (70374.37 1.30%): 1.37x speedup gen4 evolution 6696.25 (6854.14 0.85%) -> 4958.62 (5027.73 0.85%): 1.35x speedup gen3 swfdec-giant-steps 2538.03 (2539.30 0.04%) -> 1895.71 (2050.62 62.43%): 1.34x speedup gen4 gvim 4356.18 (4422.78 0.70%) -> 3276.31 (3281.69 0.13%): 1.33x speedup gen6 evolution 1242.13 (1245.44 0.72%) -> 953.76 (954.54 0.07%): 1.30x speedup gen6 firefox-planet-gnome 4554.23 (4560.69 0.08%) -> 3758.76 (3768.97 0.28%): 1.21x speedup gen3 firefox-talos-gfx 6264.13 (6284.65 0.30%) -> 5261.56 (5370.87 1.28%): 1.19x speedup gen4 midori-zoomed 4771.13 (4809.90 0.73%) -> 4037.03 (4118.93 0.85%): 1.18x speedup gen6 swfdec-giant-steps 1557.06 (1560.13 0.12%) -> 1336.34 (1341.29 0.32%): 1.17x speedup gen4 firefox-talos-gfx 80767.28 (80986.31 0.17%) -> 69629.08 (69721.71 0.06%): 1.16x speedup gen6 midori-zoomed 1463.70 (1463.76 0.08%) -> 1331.45 (1336.56 0.22%): 1.10x speedup Slowdowns ========= gen6 xfce4-terminal-a1 2030.25 (2036.23 0.25%) -> 2144.60 (2240.31 4.29%): 1.06x slowdown gen4 swfdec-youtube 3580.00 (3597.23 3.92%) -> 3826.90 (3862.24 0.91%): 1.07x slowdown gen4 firefox-talos-svg 66112.25 (66256.51 0.11%) -> 71433.40 (71584.31 0.14%): 1.08x slowdown gen4 gnome-system-monitor 5691.60 (5724.03 0.56%) -> 6707.56 (6747.83 0.33%): 1.18x slowdown gen3 ocitysmap 3494.05 (3502.44 0.20%) -> 4321.99 (4524.42 2.78%): 1.24x slowdown gen4 ocitysmap 3628.42 (3641.66 9.37%) -> 5177.16 (5828.74 8.38%): 1.43x slowdown gen5 ocitysmap 4027.77 (4068.11 0.80%) -> 5748.26 (6282.25 7.38%): 1.43x slowdown gen6 ocitysmap 1401.61 (1402.24 0.40%) -> 2365.74 (2379.14 4.12%): 1.69x slowdown [Note the performance regression for ocitysmap comes from that we now attempt to support rendering to and (more importantly) from large surfaces. By enabling such operations is the only way to one day be faster than purely using the CPU, in the meantime we suffer regression due to the increased migration and aperture thrashing. The other couple of regressions will be eliminated with improved span and shader support, now that the framework for such is in place.] The performance increase for Cairo completely overlooks the other critical aspects of the architecture: World of Padman: gen3 (800x600): 57.5 -> 96.2 gen4 (800x600): 47.8 -> 74.6 gen6 (1366x768): 100.4 -> 140.3 [F15] 144.3 -> 146.4 [drm-intel-next] x11perf (gen6); aa10text: 3.47 -> 14.3 Mglyphs/s [unthrottled!] copywinwin10: 1.66 -> 1.99 Mops/s copywinpix10: 2.28 -> 2.98 Mops/s And we do not have a good measure for how much improvement the reworking of the fallback paths give, except that xterm is now over 4x faster... PS: This depends upon the Xorg patchset "Remove the cacheing of the last scratch PixmapRec" for correct invalidations of scratch Pixmaps (used by the dix to implement SHM operations, used by chromium and gtk+ pixbufs. PPS: ./configure --enable-sna Signed-off-by: Chris Wilson commit 340cfb7f5271fd1df4c8948e5c9336f5b69a6e6c Author: Diego Elio Pettenò Date: Sun May 15 02:17:32 2011 +0200 build: do not use AC_CHECK_FILE to find the header files. Using AC_CHECK_FILE will cause cross-builds to fail picking the right file; instead use compile/preprocessor checks properly, and check for xf86driproto earlier. Reviewed-by: Rémi Cardona Signed-off-by: Chris Wilson commit 9d6e02a135efdea1d169d1938359ab2b553e941c Author: Adam Jackson Date: Tue May 10 15:27:45 2011 -0400 Remove the memory of Option "AccelMethod" Signed-off-by: Adam Jackson Reviewed-by: Eric Anholt Signed-off-by: Chris Wilson commit 895a46e8ff70195c1a4bdccbeb652e330376f64a Author: Chris Wilson Date: Tue May 10 20:38:25 2011 +0100 dri: Flush the batch after a DRI swap/copy event To minimise lag in those every so critical games, we want to ensure that the copy happens as soon as it is received, so we need to flush the batch after processing a swap event and before we go to sleep. References: https://bugs.freedesktop.org/show_bug.cgi?id=37068 Signed-off-by: Chris Wilson commit 0b4ca9313cc7eb4845cf7f4e87c869c0c6d6ff0d Author: Chris Wilson Date: Tue May 10 20:36:10 2011 +0100 video: Flush the batch on the next blockhandler after queuing In order to avoid video lag and jerky playback we need to ensure that any queued video is flushed before we go to sleep. Fixes regression from 6f104189bb. Reported-and-tested-by: Edward Sheldrake Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=37068 Signed-off-by: Chris Wilson commit bb8bf2a28bcb5e838bf4bb7e80ce220d28540719 Author: Chris Wilson Date: Tue May 10 14:35:02 2011 +0100 Correct chipset detection for Q33, Q35, B43_G1 Everytime we update these tables we trip over this bit of marketing genius. Signed-off-by: Chris Wilson commit fd1ebd44fb72e7bdf57d00f8941cd6110a529cac Author: Chris Wilson Date: Tue May 10 07:30:58 2011 +0100 module: Adopt IVB's more detailed naming convention for SNB This should fix the seven-fold repetition of "SandyBridge" in the list of supported chipsets during start-up... And be more useful in bug reports! Signed-off-by: Chris Wilson commit e9811bb777dfc51af19836175645400489f7d991 Author: Chris Wilson Date: Tue May 10 07:28:23 2011 +0100 Whitespacing cleanup for intel_module.c Bring intel_module.c into line with the kernel whitespacing rules abided by everywhere else in the tree. Signed-off-by: Chris Wilson commit 79e59fb2a047b1e733a7b0dee608db3311391725 Author: Eric Anholt Date: Fri May 6 12:40:12 2011 -0700 Add support for Ivybridge chipset. This gets display and 2D blit acceleration up and running. No Render acceleration is provided yet. commit 792738adfc5164d30358f045875dfc9b199a46da Author: Eric Anholt Date: Fri May 6 13:14:25 2011 -0700 Remove the static list of PciChipset and construct it from SymTabRec instead. This is one less place the new hardware enabler has to spam the chipset in. The PciChipset is just a match structure from PciId to the SymTabRec entry token, and our SymTabRec entry tokens are just the PciId, so it's trivial to construct. Acked-by: Kenneth Graunke commit 583e80dfa12d6c73fc677c81cb605a07b2768979 Author: Eric Anholt Date: Fri May 6 13:00:53 2011 -0700 Use the existing deviceID -> name mapping in SymTabRec instead of duping it. We need to have this array anyway for the xf86 interfaces, apparently, so just store the name in one location. This drops the i852/i855 subdevice distinction in the name printed, but I haven't seen us ever care about that. Acked-by: Kenneth Graunke commit adf7bbd3a8758de6cdecbace42e399dd27188497 Author: Eric Anholt Date: Fri May 6 12:56:53 2011 -0700 Store the chipset info struct in the PCI match struct, instead of a switch(). Acked-by: Kenneth Graunke commit 3145530feed879082bcfab11ffc8e7fd0911c920 Author: Chris Wilson Date: Sat May 7 19:51:04 2011 +0100 Ensure that the partial batch is flushed upon the blockhandler Currently, we require that a batch containing a dirty bo be submitted before we mark the device as requiring a flush. So if we never submit a batch between block handlers, we can end up sleeping without ever flushing either the partial batch or the rendering to the scanout. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=36776 Tested-by: Vasily Khoruzhick Signed-off-by: Chris Wilson commit 67e5a74e997f199327f9115c7ba867df3c49da8d Author: Paul Menzel Date: Fri Apr 22 15:09:12 2011 +0200 NEWS: fix typo (s/2.14/2.15/) to match corresponding release Signed-off-by: Paul Menzel Signed-off-by: Chris Wilson commit 0944e2d5749f5122cd9802c4a2421106fa829ea4 Author: Jesse Barnes Date: Wed Apr 20 11:39:55 2011 -0700 Add basic 30 bit depth support Still need to handle video and gamma correction, but this gets the display up and running at 30 bit depth if the kernel and display support it. Signed-off-by: Jesse Barnes commit 1d102cc6ed21d1c4afa47800eecd24b9d663f689 Author: Chris Wilson Date: Tue Apr 12 13:48:38 2011 +0100 Use SwapbuffersWait config option to control waiting on fullscreen swaps As fullscreen swaps were going via a different path to the swapping of ordinary windows, we were no longer honouring the xorg.conf option to disable swapbuffer waiting. This changes the code to only use pageflipping if the Option "SwapbuffersWait" is set to "TRUE" (default). Jesse's comment was that this should be superseded by actually supporting asynchronous page flips. As we are missing kernel and dix level support for that, in the meantime honour the config option. Signed-off-by: Chris Wilson Cc: Jesse Barnes Cc: Kristian Høgsberg commit c9fb69cb2502917dfb2828c90802de7766072899 Author: Chris Wilson Date: Sun Apr 17 10:42:05 2011 +0100 i965/video: We need 150 dwords of space for video state emission (Actually around 131, with additional 10% just for safety.) Reported-by: Modestas Vainius Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=36319 Signed-off-by: Chris Wilson commit a51cd83d25f2f9f2107219d5671194f931601244 Author: Chris Wilson Date: Sun Apr 17 10:36:26 2011 +0100 intel: Beware the unsigned promotion when checking for batch overflows Reported-by: Modestas Vainius References: https://bugs.freedesktop.org/show_bug.cgi?id=36319 Signed-off-by: Chris Wilson commit 030aa3d13643a0acb7126fa1fa6c5a8eac134639 Author: Chris Wilson Date: Thu Apr 14 10:25:21 2011 +0100 NEWS: typo. Spotted too late... Signed-off-by: Chris Wilson commit 0e425b30e12b11dbf51eea1d3366e3372f8b93af Author: Chris Wilson Date: Thu Apr 14 10:16:03 2011 +0100 configure,NEWS: 2.15.0 release Signed-off-by: Chris Wilson commit 686018f283f1d131073ef5917213e6a8ac013f26 Author: Chris Wilson Date: Tue Apr 12 08:23:04 2011 +0100 Turn relaxed-fencing off by default for older (pre-G33) chipsets There are still too many unresolved bugs, typically GPU hangs, that are related to using relaxed fencing (i.e. only allocating the minimal amount of memory required for a buffer) on older hardware, so turn off the feature by default for the release. Reported-and-tested-by: Knut Petersen Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=36147 Signed-off-by: Chris Wilson Acked-by: Daniel Vetter commit 3d2b79098ccb3fed86be66f619f4ed338741d454 Author: Chris Wilson Date: Mon Apr 11 15:20:16 2011 +0100 dri: Rearrange code to compile against xorg-server-1.7 Signed-off-by: Chris Wilson commit 4fa35dd5e13a58070220f787fc0678f1c679808d Author: Chris Wilson Date: Mon Apr 11 10:44:55 2011 +0100 NEWS: version bump for 2.14.903 snapshot Signed-off-by: Chris Wilson commit 97e9557619e58ef769eb7cbf1a03fbd52be7f2ed Author: Chris Wilson Date: Fri Apr 8 13:38:48 2011 +0100 intel: Restore manual flush for old kernels Daniel Vetter pointed out that the automagic flush by the kernel for the busy-ioctl was only introduced upstream in 2.6.37. So we still need to manually emit a flush on old kernels. Signed-off-by: Chris Wilson commit fb40bf2b33a6d26f0e6a4e5798d10c905faa8aad Author: Daniel Vetter Date: Fri Apr 8 12:37:35 2011 +0200 Tell users to grab i915_error_state on gpu hangs Acked-by: Chris Wilson Signed-off-by: Daniel Vetter commit 59ed6b05db99d7e5144f4f8499c67cee9b833673 Author: Chris Wilson Date: Thu Apr 7 15:28:46 2011 +0100 Revert "i965: Convert to relative relocations for state" This reverts commit d2106384be6f9df498392127c3ff64d0a2b17457. Breaks compiz (but not mutter/gnome-shell) on gen6. Not sure if this is not seem deep interaction issue with multiple clients sharing the GPU or just with compiz, but for now we have to revert and suffer the inane performance hit. It looks suspiciously like another deferred damage issue... Bugzilla: 51a27e88b073cff229fff4362cb6ac22835c4044 Signed-off-by: Chris Wilson commit 25521900df11bc71020ee80db2223f979bec5ec6 Author: Chris Wilson Date: Thu Apr 7 15:09:30 2011 +0100 gen6: Invalidate texture cache Signed-off-by: Chris Wilson commit ad22003033eb502474ae538a97e3b42cf8f83880 Author: Chris Wilson Date: Thu Apr 7 10:29:40 2011 +0100 i965: Avoid transform overheads for vertex emit where possible Minor improvement as the bottlenecks lie elsewhere. But it was annoying me. Signed-off-by: Chris Wilson commit 007c2f86cbb386861a1f711786523657f92e12cb Author: Chris Wilson Date: Thu Apr 7 09:53:34 2011 +0100 i965: Refactor to use constant sampler_state offsets Signed-off-by: Chris Wilson commit 8dc99b305a514dcd42c4260698e685a66dc95518 Author: Chris Wilson Date: Mon Apr 4 22:30:29 2011 +0100 i965: Reset vertex_id after every batch So that we always remember to re-emit the initial vertex elements state. Signed-off-by: Chris Wilson commit 5982ed4da12a964351029c6f786348a11e33dc81 Author: Chris Wilson Date: Mon Apr 4 19:32:43 2011 +0100 i965: Always update last_floats_per_vertex Signed-off-by: Chris Wilson commit 6f104189bb9439ab0e05f04d4be020813eb04bf9 Author: Chris Wilson Date: Mon Apr 4 19:21:10 2011 +0100 Take advantage of the kernel flush for dirty bo in the busy ioctl Rather than just creating and submitting a batch that simply contains a flush in order to periodically ensure that rendering reaches the scanout, we can simply ask the kernel whether the scanout is busy. The kernel will then submit a flush on our behalf if it is dirty, which takes advantage of the kernel's dirty state tracking. Signed-off-by: Chris Wilson commit 314439860e595c473d168c0cafb2b5b49b7fef30 Author: Chris Wilson Date: Mon Apr 4 17:20:04 2011 +0100 Remove unused function 'intel_bo_alloc_for_data' Signed-off-by: Chris Wilson commit ced747cefb5e697e6caa65296dff728904f52b93 Author: Chris Wilson Date: Mon Apr 4 17:13:50 2011 +0100 Remove the unnecessary MI_FLUSH from the flush handler The kernel will emit any required flushes between the dri client and the ddx, and we do not rely on the MI_FLUSH here for scanout. Signed-off-by: Chris Wilson commit 79444291a39c42039192a5baa3a71d52300cf4ee Author: Chris Wilson Date: Mon Apr 4 16:34:58 2011 +0100 i965: segregate each vertex element into its own buffer Reduce the number of relocations emitted by only emitting one relocation per vertex element per vertex buffer. References: https://bugs.freedesktop.org/show_bug.cgi?id=35733 Signed-off-by: Chris Wilson commit d2106384be6f9df498392127c3ff64d0a2b17457 Author: Chris Wilson Date: Mon Apr 4 12:33:04 2011 +0100 i965: Convert to relative relocations for state References: https://bugs.freedesktop.org/show_bug.cgi?id=35733 Signed-off-by: Chris Wilson commit 972569f6fd1e14519f46e9f50d2509faf1d0aa55 Author: Chris Wilson Date: Fri Mar 25 10:46:14 2011 +0000 MI_LOAD_SCAN_LINES_INCL are inclusive and range [0, display height-1] We have seen GPU hangs with: batchbuffer at 0x0f9b4000: 0x0f9b4000: 0x09000000: MI_LOAD_SCAN_LINES_INCL 0x0f9b4004: 0x00000300: dword 1 0x0f9b4008: 0x09000000: MI_LOAD_SCAN_LINES_INCL 0x0f9b400c: 0x00000300: dword 1 0x0f9b4010: 0x01820000: MI_WAIT_FOR_EVENT 0x0f9b4014: HEAD 0x02000006: MI_FLUSH on a 1366x768 display. That according to the specs an invalid command for the pipe. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=35576 Signed-off-by: Chris Wilson commit 537a836dd6db384d53b52eb457a7d257c440217f Author: Chris Wilson Date: Fri Apr 1 07:10:25 2011 +0100 946GZ is a 965G! Sales & Marketing score another victory in confusing me. Bugzila: https://bugs.freedesktop.org/show_bug.cgi?id=35854 Signed-off-by: Chris Wilson commit 630d77bf10ba6234bb9c04538636f7d8aa319aea Author: Chris Wilson Date: Wed Mar 30 07:24:42 2011 +0100 Add a DRI2SwapEventPtr typedef for retro xservers Because people are still trying to build upon our solid historical foundations. Signed-off-by: Chris Wilson commit f660df2cb44b310740ed850037ff37891b671f28 Author: Daniel Vetter Date: Tue Mar 29 18:02:24 2011 +0200 Cleanup gen2 tiling confusion A tile on gen2 has a size of 2kb, stride of 128 bytes and 16 rows. Reviewed-by: Chris Wilson Signed-off-by: Daniel Vetter commit a2c8442fcb29ef2d0df41518952a2d8df1671070 Author: Chris Wilson Date: Tue Mar 29 11:38:45 2011 +0100 configure: version bump for 2.14.902 snapshot Signed-off-by: Chris Wilson commit 2f2680793952876c89796174bed53c824fe6d1cb Author: Chris Wilson Date: Tue Mar 29 11:38:18 2011 +0100 NEWS: Entry for 2.14.902 snapshot Signed-off-by: Chris Wilson commit 7ccbec801e9ee32fc110db730dfec674a94dea21 Author: Keith Packard Date: Thu Mar 24 11:06:57 2011 -0700 Recover from i830_dri2_add_frame_event out-of-memory condition If adding either the frame or client resources fails, we need to clean up afterwards properly. First, add_frame_event needs to internally clean up after itself by undoing any partial execution. Second, the callers need to look at the return value and free the swap/flip info structure when necessary. Signed-off-by: Keith Packard Reviewed-by: Julien Cristau commit ec133abc4bd8caba15ed54e18621b816afb06981 Author: Chris Wilson Date: Thu Mar 24 15:32:50 2011 +0000 Use the per-generation batch context switch for atomic sequences A modest boost to throughput and reduction in CPU overhead from not flushing the batch on every transition from BLT to RENDER. Signed-off-by: Chris Wilson commit 86f23f21ab57fcbc031bcd2b8f432a08ff4cc320 Author: Keith Packard Date: Wed Mar 23 17:07:49 2011 -0700 Skip client and drawable resource delete calls when deleting frame event As the frame_event is about to be freed, there's no point in cleaning up references to the drawable and client. Signed-off-by: Keith Packard commit e1ff5182304e00c0d392092069422cae7626cf8d Author: Keith Packard Date: Wed Mar 9 17:00:41 2011 -0800 Handle drawable/client destruction in pending swaps/flips A pending swap or flip holds references to drawables and clients which become invalid when destroyed. Add suitable resources to the database to track those lifetimes and clean up the pending data structure then. Later, when the pending swap or flip occurs, handle a missing drawable by just discarding the flip or swap. Handle a missing client by not sending an event or reply. Signed-off-by: Keith Packard commit aa0328f342177e33f9c863531712eaa4577b5f2b Author: Keith Packard Date: Wed Mar 9 16:23:22 2011 -0800 dri2: Make DRI2FrameEvent public and use instead of void * Instead of using void * for all of the flip_info and swap_info pointers, just make the underlying structure a public data type and use that. Signed-off-by: Keith Packard commit ee740778f5d5355c04f6fc4564f598993b106d62 Author: Chris Wilson Date: Wed Mar 16 08:37:24 2011 +0000 i915/video: Clip indirect Xv output In the case of a singular clip box and rendering onto the front buffer larger than 2048, the clip extents were being ignored. Here we can simply reduce the size of the indirect pixmap to the clip extents. Reported-by: Michael Chang Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=35346 Signed-off-by: Chris Wilson commit 014fc7abb7b2cc2110e3ab9a0bd6f7cff2c64c05 Author: Chris Wilson Date: Wed Mar 9 17:10:50 2011 +0000 dri: Disable page-flip between a tiled buffer and a linear scanout Keith Packard pointed out a loophole that could cause the DDX to end up with a tiled scanout even if the user required a linear framebuffer; that is by using page-flipping we could replace the scanout pixmap with another of our choosing, and not necessarily tiled. Close that loophole by only allowing an exchange of buffers between identical tiling modes. For the common case, this is fine since they will indeed be allocated with the same tiling. For the linear framebuffer case with mesa using a tiled pixmap, we force it to blit onto the scanout instead. Reported-by: Keith Packard Signed-off-by: Chris Wilson commit 049ce4397ddf7fd088ce364cbb53cacf5133176f Author: Chris Wilson Date: Wed Mar 9 08:27:02 2011 +0000 Give each user of tiling separate xorg.conf options So that you can indeed allocate a linear framebuffer if you so desire without breaking mesa. Adds: Section "Driver" Option "LinearFramebuffer" "False|True" # default false EndSection to xorg.conf Signed-off-by: Chris Wilson commit 0bb1a5f19e09dc553761ddd90bf6319eab94a597 Author: Chris Wilson Date: Wed Mar 9 08:26:44 2011 +0000 Update priv->stride after bo reallocation Signed-off-by: Chris Wilson commit 34f9a3335f96b16ef9df6213eb0a586c94f8f05b Author: Chris Wilson Date: Wed Mar 2 17:46:19 2011 +0000 configure: Bump for 2.14.901 snapshot Signed-off-by: Chris Wilson commit 057783937d7a42a766d0e45174c4c2dc0ffacd40 Author: Chris Wilson Date: Wed Mar 2 17:45:43 2011 +0000 NEWS: Add entry for 2.14.901 snapshot Signed-off-by: Chris Wilson commit e889d3a709b55a0731ab098b17a3364b9bf39387 Author: Chris Wilson Date: Sun Feb 27 10:51:50 2011 +0000 dri: Protect against destroying a foreign DRI drawable I have no clue as to how such an alien drawable reached us, but we have the evidence of a segfault to say it can happen. Reported-by: Bernie Innocenti Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=34787 Signed-off-by: Chris Wilson commit d21d781466785c317131a8a57606925867265dc8 Author: Daniel Vetter Date: Tue Feb 22 18:31:44 2011 +0100 Fix relaxed tiling on gen2 Signed-off-by: Daniel Vetter commit 9599fde65a0d8b7e7c85199346f7b620bdd8388d Author: Bryce Harrington Date: Tue Feb 15 22:30:18 2011 -0800 Quell excessively verbose vblank counter failed error messages Certain error situations can result in the following printed to Xorg.0.log at a high enough rate to make log file size a problem. (WW) intel(0): I830DRI2GetMSC:1062 get vblank counter failed: Invalid argument (WW) intel(0): I830DRI2ScheduleWaitMSC:1118 get vblank counter failed: Invalid argument Following in the tradition of commit 0ad6d6e1, limit the warnings to be output 5 times, then quell the remainder. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=34322 Ref.: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/710594 Signed-off-by: Bryce Harrington Signed-off-by: Chris Wilson commit 0ca595e9d533019b241666d29b421c7b36f9647a Author: Adam Jackson Date: Thu Feb 17 15:26:35 2011 -0500 Fix IGD and IGDNG constants to be comprehensible Since, with GPU-on-package, it's hard to talk about a model number for a specific chipset like 855GM, just use the platform names. Signed-off-by: Adam Jackson Signed-off-by: Chris Wilson commit 4c66b28870b050493ad96f7b0fe2d70d7ee539c7 Author: Chris Wilson Date: Thu Feb 17 16:48:24 2011 +0000 uxa: Fallback if the temporary is too large If the render operation requires a temporary source Picture and the operation is large, larger than the maximum permitted bo, then we will fail to allocate the bo. In this case, we need to fallback and perform the operation on the CPU rather than dereference a NULL bo. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=34399 Signed-off-by: Chris Wilson commit 23f9b14df7c102c1036134835dd5d1a508059858 Author: Chris Wilson Date: Sat Feb 12 10:42:34 2011 +0000 i965: Remove broken maximum base addresses from video WRONG. The hardware was never limited to 0x1000000 and the kernel can quite rightly place objects above that limit. Specifying such had no relation to reality, so why did we do it? TWICE! Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=34017 Signed-off-by: Chris Wilson commit 9184af921bc2f332fcb6c9b47001414378eab8e2 Author: Javier Jardón Date: Wed Feb 9 00:36:37 2011 +0000 Update autotools configuration Use new libtool syntax and silent-rules to silent the build output a bit (linux-like) commit 6e721e098b9181e8e77e314f966729d28e705582 Author: Bryce Harrington Date: Fri Feb 4 00:15:13 2011 -0800 Check return value of uxa_acquire_solid() since it can return NULL uxa_acquire_solid returns NULL under OOM. Thus the value of solid must be checked before dereferencing it in the uxa_get_offscreen() call. Signed-off-by: Bryce Harrington Signed-off-by: Chris Wilson commit da990536eca09c6de74627541cd56ecfad925eda Author: Chris Wilson Date: Thu Feb 3 09:41:48 2011 +0000 uxa: Undo damage translation before appending The region is used to paint onto the backing pixmap (and thus translated) prior to being passed to the damage layer (wrt to the drawable). So the local translation needs to be undone first. Identified by Christopher James Halse Rogers. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=33650 Signed-off-by: Chris Wilson commit ce1e0969058f8c70b624bc85bb8d6698a35794d3 Author: Chris Wilson Date: Mon Jan 31 20:27:26 2011 +0000 Rename 'intel' backlight to match upstream name in 2.6.38 (Just waiting for libbacklight to bring sanity...) Signed-off-by: Chris Wilson commit 80f59c8a4f51ccc27653a2de2dee2563b1e911f1 Author: Cyril Brulebois Date: Mon Jan 31 06:22:39 2011 +0100 xvmc: Stop using uninitialized variable. The actual code was removed in 5f64122551, where the get_surface_status driver callback was dropped. Just return Success if there were no issues with the parameter. Get rid of the following with CFLAGS="-Wall -Werror": | CC intel_xvmc.lo | cc1: warnings being treated as errors | intel_xvmc.c: In function ‘XvMCSyncSurface’: | intel_xvmc.c:677: error: ‘ret’ may be used uninitialized in this function | intel_xvmc.c:672: note: ‘ret’ was declared here Signed-off-by: Cyril Brulebois Signed-off-by: Chris Wilson commit 86284abafbe1364f82509d2656c82ba11cdce6cc Author: Cyril Brulebois Date: Mon Jan 31 06:22:38 2011 +0100 xvmc: Silence gcc, adding a cast to work around signedness issues. Both exported XvMCPutSlice/XvMCPutSlice2 functions have a prototype with a parameter whose signedness differs from the one in the put_slice/put_slice2 functions they call. Make it unsigned through a cast. Get rid of this accordingly with CFLAGS="-Wall -Werror": | CC intel_xvmc.lo | cc1: warnings being treated as errors | intel_xvmc.c: In function ‘XvMCSyncSurface’: | intel_xvmc.c:677: error: ‘ret’ may be used uninitialized in this function | intel_xvmc.c:672: note: ‘ret’ was declared here Signed-off-by: Cyril Brulebois Signed-off-by: Chris Wilson commit fc2eb64d7a184770ccd708a5b2337a538ca2b9db Author: Cyril Brulebois Date: Mon Jan 31 09:59:04 2011 +0000 xvmc: Silence gcc, add parentheses around arithmetic. Get rid of those accordingly with CFLAGS="-Wall -Werror": | CC i915_xvmc.lo | cc1: warnings being treated as errors | i915_xvmc.c: In function ‘i915_mc_one_time_state_emit’: | i915_xvmc.c:369: error: suggest parentheses around arithmetic in operand of ‘|’ | i915_xvmc.c:374: error: suggest parentheses around arithmetic in operand of ‘|’ | i915_xvmc.c:379: error: suggest parentheses around arithmetic in operand of ‘|’ | i915_xvmc.c: In function ‘i915_mc_load_indirect_render_emit’: | i915_xvmc.c:708: error: suggest parentheses around arithmetic in operand of ‘|’ | i915_xvmc.c:713: error: suggest parentheses around arithmetic in operand of ‘|’ Signed-off-by: Cyril Brulebois Signed-off-by: Chris Wilson commit 1b9349bceb9172d9bf01638b2df10b9b79e883b4 Author: Cyril Brulebois Date: Mon Jan 31 06:22:36 2011 +0100 xvmc: Silence gcc, add parentheses around assignment. Get rid of the following with CFLAGS="-Wall -Werror": | CC intel_xvmc_dump.lo | cc1: warnings being treated as errors | intel_xvmc_dump.c: In function ‘intel_xvmc_dump_open’: | intel_xvmc_dump.c:41: error: suggest parentheses around assignment used as truth value Signed-off-by: Cyril Brulebois Signed-off-by: Chris Wilson commit ae8877e30707cae586565b4deee4c7fd9ac7d3ed Author: Cyril Brulebois Date: Mon Jan 31 06:22:35 2011 +0100 xvmc: Get rid of unused variables. Signed-off-by: Cyril Brulebois Signed-off-by: Chris Wilson commit 8bd6a7149fcb5f9ed6580dc2aa0180874938a1ea Author: Chris Wilson Date: Sat Jan 29 12:04:09 2011 +0000 Enable conditional compilation of asserts With '--disable-debug' we define NDEBUG and so disable the asserts, that is we continue to default to compiling asserts into the code. Signed-off-by: Chris Wilson commit a5a1ab7bbc58ec884cd0b90fe7139f402291b92f Author: Chris Wilson Date: Fri Jan 28 17:20:06 2011 +0000 i915: Remove unused 'w' and 'h' Signed-off-by: Chris Wilson commit 2c9b3225d8c7b53e2d921705dbb069710abcea5d Author: Chris Wilson Date: Fri Jan 28 17:19:39 2011 +0000 i915: Remove unused 'num_floats' variable Signed-off-by: Chris Wilson commit c6dc27562abbc8ca9e873ad502ca49ae010461d2 Author: Chris Wilson Date: Mon Jan 24 20:25:27 2011 +0000 uxa: Only recreate the glyph cache on *generational* updates The screen resources are recreated when the screen is rotated as well, without being finalized. In this case, we do not need to reconstuct the cache (or if we did, we would need to tear it down first). Reported-by: Till Matthiesen Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=33412 Signed-off-by: Chris Wilson commit 5baa63c634990810a66c3150b4f2b76fcee2df38 Author: Xiang, Haihao Date: Fri Jan 21 08:45:52 2011 +0800 Fix an error in 4c4ad555564a80311df1a4b762eb1e119c6d95fb. See the original patch at https://bugs.freedesktop.org/show_bug.cgi?id=24767 Signed-off-by: Xiang, Haihao commit 3e28a0c0b43823d3104fe8fc50b5994b41fc0cc1 Author: Chris Wilson Date: Thu Jan 20 15:07:26 2011 +0000 Create the UXA generational resources during screen create Signed-off-by: Chris Wilson commit 4c4ad555564a80311df1a4b762eb1e119c6d95fb Author: Xiang, Haihao Date: Wed Jan 19 15:25:06 2011 +0000 Correct offset of planes within clipped window Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=24767 Reported-and-tested-by: Sitsofe Wheeler Signed-off-by: Chris Wilson commit 5a22bc999de8555f26c639eaffbe73b680746174 Author: Eric Anholt Date: Mon Jan 17 11:32:37 2011 -0800 Quiet compiler warning about is_affine_src same way we do is_affine_mask. commit 3a2a4b0784f0fc96a5457b18931471f15ad745fc Author: Simon Farnsworth Date: Mon Jan 17 17:38:23 2011 +0000 Fix textured video when destination is larger than screen In our application, the screen is never rotated from the point of view of the driver; instead, the compositor applies a suitable rotation as it composites the display. This works fine on 945, but on 965, videos are limited in height to the actual height of the screen. Change various bits of code so that we use the width and height of the destination pixmap instead of the width and height of the virtual screen. This works correctly both for XVideo to offscreen storage (CompositeRedirect) and for XVideo to the screen (no compositor). commit fd9235ebe03a01982238cdd6e8b55f613e14b6af Author: Chris Wilson Date: Fri Dec 17 00:58:36 2010 +0000 Delete unused memory allocation flags. Signed-off-by: Chris Wilson commit 03248a798409ca4dd1caf5e25a3d5f5a80173260 Author: Chris Wilson Date: Fri Dec 17 00:57:01 2010 +0000 Cache the fixed crtc<->pipe relationship Signed-off-by: Chris Wilson commit 00a2aee38dc5cbbd4907a593fa8d617f5a20292e Author: Chris Wilson Date: Fri Dec 17 00:50:33 2010 +0000 Remove unused I830Output Signed-off-by: Chris Wilson commit a851879695c48e88d421b053a78f867d9902273f Author: Chris Wilson Date: Fri Dec 17 00:48:26 2010 +0000 Remove bitrotted, but fortunately unused, I830CrtcPrivate Signed-off-by: Chris Wilson commit f8a5d7737e014e99943b947d088a4f9f1b465d46 Author: Chris Wilson Date: Fri Dec 17 00:37:47 2010 +0000 Delete some dead code Signed-off-by: Chris Wilson commit 968151898b807b06953b2f4170ac02889f16c8dc Author: Chris Wilson Date: Fri Dec 17 00:18:25 2010 +0000 Remove unused GTT/Map sizes and addresses These have been made obsolete by KMS and GEM. Signed-off-by: Chris Wilson commit 19b5817a54d4a17161340ada9bf20027644ff074 Author: Chris Wilson Date: Fri Dec 17 00:15:31 2010 +0000 Remove unused struct _intel_memory definition Signed-off-by: Chris Wilson commit 40a7bde9140c0ae6412b8f916b1455d44e2af6cf Author: Chris Wilson Date: Fri Jan 7 23:46:07 2011 +0000 configure: version bump for 2.14.0 commit 8ecbba2aca657da3894eb77e856ac2d4e9efc0cc Author: Chris Wilson Date: Fri Jan 7 23:45:06 2011 +0000 NEWS: Release notes for 2.14.0 commit 22d7b61791c382088a6c0df5dce3a15405d6c495 Author: Chris Wilson Date: Tue Jan 4 12:32:01 2011 +0000 i965: Fix off-by-one in assert Signed-off-by: Chris Wilson commit a7c7a9108f76aa312f3d5efa466052b914c81484 Author: Chris Wilson Date: Tue Jan 4 10:53:30 2011 +0000 configure: version bump for 2.13.903 snapshot Signed-off-by: Chris Wilson commit 3fbecf34630c3588618900afdfbd9a4266856178 Author: Chris Wilson Date: Tue Jan 4 10:52:44 2011 +0000 NEWS: Add entry for 2.13.903 Signed-off-by: Chris Wilson commit 90538ecfff9c324d61419f451d41d26ab026f232 Author: U. Artie Eoff Date: Tue Dec 21 17:42:38 2010 -0700 configure: suggest upstream to find macros in case they're missing. added upstream suggestion to error message in case xorg macros or xserver macros are missing. removed package manager references in suggestion since these are 'user' and/or 'distribution' preferences. Signed-off-by: Chris Wilson commit 6d50f5a140dbefdf86eb25d18d08473aa7e42acc Author: Chris Wilson Date: Mon Jan 3 23:06:14 2011 +0000 If the crtc is not enabled, then it can't be on Signed-off-by: Chris Wilson commit 145c42a101c3150fb1e2cae8a1780b26e8b56f42 Author: Adam Jackson Date: Mon Jan 3 17:52:27 2011 -0500 dri2: Fix interlace computation scrn->currentMode is a hack for xf86vidmode and in general is wrong for RANDRful drivers. Use the mode on the associated CRTC instead. Signed-off-by: Adam Jackson [ickle: crtc->mode is a ModeRec not Ptr] Signed-off-by: Chris Wilson commit b0d10ff04a24a951256a79710ba22b325b549ba4 Author: Adam Jackson Date: Mon Jan 3 17:52:26 2011 -0500 xv: Fix interlace computation scrn->currentMode is a hack for xf86vidmode and in general is wrong for RANDRful drivers. Use the mode on the associated CRTC instead. Signed-off-by: Adam Jackson [ickle: crtc->mode is a ModeRec] Signed-off-by: Chris Wilson commit 0ad6d6e1a678de7d8c2d4cd7c734887e1deb2f9b Author: Chris Wilson Date: Mon Jan 3 16:52:07 2011 +0000 dri: Only issue a warning for an "impossible" flip return 5 times As it appears that some kernels do indeed return the "wrong" value, issuing a warning 60 times a second is a cruel and unusual punishment. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=32680 Signed-off-by: Chris Wilson commit d729ef02f2955f7476df4c65403bc1f8e705b780 Author: Chris Wilson Date: Sun Jan 2 09:11:10 2011 +0000 dri: Don't wait upon a NULL current mode There is a race condition between the dri swapbuffers code and hotplugging whereby we might attempt to execute a wait upon a non-existent output. This causes a NULL dereference and a loud crash. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=32770 Signed-off-by: Chris Wilson commit 537fa55ed2449e91f3dd1e04abc720c6818d7227 Author: Chris Wilson Date: Sun Jan 2 09:06:28 2011 +0000 dri: Fix the use of the uninitialised bo for flink Reported-by: Jeff Chua Signed-off-by: Chris Wilson commit dbc542b9d38ff9781efcdf1d5308e97f04499138 Author: Chris Wilson Date: Fri Dec 31 23:29:05 2010 +0000 dri: Protect against using dri with an non-gem pixmap Reported-by: Jeff Chua Signed-off-by: Chris Wilson commit 53fbc9f1760ee481cba1f6dceb9e7c97282a2976 Author: Chris Wilson Date: Thu Dec 30 15:32:40 2010 +0000 Don't replace the scanout bo through PutImage As the bo may be pinned for either use by the scanout or through sharing with another application, under those circumstances we cannot replace the bo itself but must force the blit for PutImage. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=31367 Reported-and-tested-by: Bjørn Mork Signed-off-by: Chris Wilson commit 7667ad8432c032aec3a2aa004fc4dfc1877971b3 Author: Chris Wilson Date: Fri Dec 24 09:42:29 2010 +0000 dri: Differentiate identical "get vblank failed" messages with line no Signed-off-by: Chris Wilson commit 875d4828352c501ec5d0169735fb078807e76f2e Author: Chris Wilson Date: Thu Dec 23 19:36:12 2010 +0000 i830: amalgamate consecutive composites into a single primitive Improve aa10text on i845 from 218kglyphs/s to 234kglyphs/s Signed-off-by: Chris Wilson commit 58d9912e447cf6e29b20747b5ed6ee595687ce8d Author: Chris Wilson Date: Wed Dec 22 13:50:53 2010 +0000 Remove the deprecated function 'XNFprintf' As we know the maximum length of the string, we can replace our single usage of XNFprintf with snprintf. Signed-off-by: Chris Wilson commit e8db77e019e19ace48592b2f6d96af6baa9a228e Author: Chris Wilson Date: Tue Dec 21 23:03:21 2010 +0000 Revert "Suggest where to find xorg-macros in case it's missing" I failed my reading comprehension and didn't spot that the help strings pointed to two separate downstream packages. Thanks to Carl Worth for pointing out my mistake. This reverts commit ebb7aca66785730737bc739624006e55d43dfe1b. commit ebb7aca66785730737bc739624006e55d43dfe1b Author: Chris Wilson Date: Tue Dec 21 22:23:24 2010 +0000 Suggest where to find xorg-macros in case it's missing [Modified version of U. Artie Eoff's commit to remove the duplicated string. The paraphrased commit message is repeated below for clarity.] xorg-macros is often package under an unusual name, so provide some clues as to what name to try on the user's distribution. Installing build-deps would be just too easy... Also include a check for the existence of the XORG_DRIVER_CHECK_EXT macro, otherwise configure might continue in spite of the missing macro and cause obtuse syntax errors. Signed-off-by: Chris Wilson commit 90576f95f35e528fa2c667df162c5dc7ec98975d Author: U. Artie Eoff Date: Tue Dec 21 15:10:41 2010 -0700 configure: updated m4 macro check in configure.ac Added m4 check for XORG_DRIVER_CHECK_EXT macro definition. Updated m4 fatal messages to give better hint on how to resolve error when xorg macros are missing. Previously, configure would continue in spite of the missing macros and the build would fail for syntax errors. commit 1ac2e04023f84dbf1f3db2ecad1cadd159aa614d Author: Chris Wilson Date: Mon Dec 20 10:55:11 2010 +0000 Undo: Disable BLT for i830 and 845G Reported-by: György Balló Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=32482 Signed-off-by: Chris Wilson commit 6f21405454487adf7623cf22f0fdd9f127099afd Author: Chris Wilson Date: Fri Dec 17 16:37:55 2010 +0000 G35 is gen4 and not gen3 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=32478 Reported-and-tested-by: Michal Marek Signed-off-by: Chris Wilson commit 5743c223877fbff710cdd5114cff6d3ee3108309 Author: Mario Kleiner Date: Thu Dec 9 03:27:59 2010 +0100 Check consistency of pageflip completion vblank count. Implements a consistency check on returned vblank count values of pageflip completion. Impossible values are detected, a x-warning is logged and returned (msc,ust) values are marked invalid, so clients could perform error handling. Such a warning would indicate bugs in the pageflip completion routine of future kms drivers or the ddx and thereby aid driver debugging. Signed-off-by: Mario Kleiner Signed-off-by: Chris Wilson commit 2177e6032146f4720f244e98f7c0d6df1b925784 Author: Mario Kleiner Date: Thu Dec 9 03:12:35 2010 +0100 Fix reporting of pageflip completion events on multi-head. When a drawable is page-flipped on multiple crtc's (fullscreen drawable on mirror-mode or multi-head x-screen), only one pageflip event is finally delivered, after the last participating crtc signals flip completion, this to avoid visual corruption. Old code returned vblank count and timestamps of flip completion of this last crtc, instead of the values of the "master crtc", the one that was used for initially scheduling/triggering the pagflip via vblank events. (master = I830DRI2DrawablePipe(drawable)) This patch makes sure that the pageflip completion values of the "master" crtc are returned, otherwise client applications will get confused by the random (msc, ust) values returned by whichever crtc was the last to complete its flip. Without this, the returned values change randomly and jump forward and backward in time and count. Signed-off-by: Mario Kleiner Signed-off-by: Chris Wilson commit 71af40a75fbdd1054b1111e8cbe67ad1f97e6613 Author: Chris Wilson Date: Fri Dec 10 15:55:09 2010 +0000 NEWS: 2.14, I meant the upcoming 2.14 release! Signed-off-by: Chris Wilson commit 556331a22101d08babbd276236f74ffcc380171c Author: Chris Wilson Date: Fri Dec 10 15:28:37 2010 +0000 configure: version bump for 2.13.902 This is actually the first release candidate for 2.14. Signed-off-by: Chris Wilson commit e01ea4a45c1cf8953a6d5ff6c600d76dbbe4760d Author: Chris Wilson Date: Fri Dec 10 15:28:00 2010 +0000 NEWS: Add entry for the 2.13.902 snapshot Signed-off-by: Chris Wilson commit 8df6f0a6082ad762ec69be7789a961db0c74154a Author: Chris Wilson Date: Fri Dec 10 12:26:16 2010 +0000 configure: Bump required libdrm to 2.4.23 To pull in the required headers for BLT and relaxed fencing support. Signed-off-by: Chris Wilson commit 9b967807c2d240488a715509649663aac3583532 Author: Chris Wilson Date: Tue Dec 7 20:45:16 2010 +0000 Revert "i965: The RenderCache flush after every glyph is required for compiz" This reverts commit 03e8351179b1c25d219842ef3e01ee8e176f594f. * sigh. This was only meant to be a temporary debugging hack, not for public consumption (or embarrassment). Signed-off-by: Chris Wilson commit 27e33928f87ecb806d1f8c459d4a4ada76b0e30e Author: Chris Wilson Date: Tue Dec 7 20:42:24 2010 +0000 i965: Mark sure we mark reused render targets as dirty ... or else we may forget to flush them again. Signed-off-by: Chris Wilson commit 03e8351179b1c25d219842ef3e01ee8e176f594f Author: Chris Wilson Date: Tue Dec 7 19:27:57 2010 +0000 i965: The RenderCache flush after every glyph is required for compiz ... now who can explain why. Signed-off-by: Chris Wilson commit d90c8f4d0bb6b79ecdcaa82f87291169c18c64b9 Author: Chris Wilson Date: Tue Dec 7 19:22:30 2010 +0000 i965: Invalidate pixmap binding location on reuse. Signed-off-by: Chris Wilson commit 4a186a612376bdd6f86c026e8b8b442108868a0a Author: Chris Wilson Date: Tue Dec 7 16:56:57 2010 +0000 Always flush the batch before blocking for new X requests This should prevent any lag when waiting upon user input, for example whilst logging in with gdm. Signed-off-by: Chris Wilson commit 1ba983034b3a70fb541dc359189c020ee497c634 Author: Chris Wilson Date: Tue Dec 7 12:27:29 2010 +0000 uxa: Emit the damage after the render for the workaround in uxa_solid_rects Signed-off-by: Chris Wilson commit 00c204a7b2df7a6b501db9b57603534a45ecafa9 Author: Chris Wilson Date: Mon Dec 6 16:34:52 2010 +0000 snb: Only emit CC and DepthStencil bos once per batch Signed-off-by: Chris Wilson commit 084ae43110240c582ae7cd6452966dd9f22350fa Author: Chris Wilson Date: Mon Dec 6 15:57:19 2010 +0000 snb: Restore drawrect, we need the implicit flush Something is wrong, we should be tracking when to invalidate the caches as appropriate, yet I can not finding the missing flush to replace the implicit one of DRAW_RECTANGLE. Fixes cacomposite. Signed-off-by: Chris Wilson commit f3a47d7f235d18e4529e3898a48673c7c3cbd489 Author: Chris Wilson Date: Mon Dec 6 14:11:05 2010 +0000 snb: Cache pixmap binding locations Signed-off-by: Chris Wilson commit 4d48fed9aa0c3c7b84e74b4f72f298f580a8973c Author: Chris Wilson Date: Mon Dec 6 13:50:43 2010 +0000 snb: Cache state between composite ops Signed-off-by: Chris Wilson commit a58e5a1bdf10be3b96ecaa3d5e3ee288eab1063f Author: Chris Wilson Date: Mon Dec 6 13:34:53 2010 +0000 snb: Emit more invariants only once Signed-off-by: Chris Wilson commit e8f41c371024e2f3845c7652b22af8f5900200f2 Author: Chris Wilson Date: Sun Dec 5 21:37:23 2010 +0000 uxa: Prevent reading past the last byte on upload/download Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=29752 Reported-by: Sergey Samokhin Signed-off-by: Chris Wilson commit c2fac6ca108734dd9566570b15f01cc476fa53dc Author: Matthias Hopf Date: Fri Nov 19 04:29:57 2010 +0100 Don't use hardware acceleration on Sandybridge rev 07 hardware or earlier. This is known to lock up the GPU even with the workaround in place. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=31773 Signed-off-by: Matthias Hopf Signed-off-by: Chris Wilson commit 8abade8a2139b500251488be5d6390f149666fb2 Author: Chris Wilson Date: Sun Dec 5 09:55:43 2010 +0000 display: Flush any pending batches before changing modes. Signed-off-by: Chris Wilson commit 8450aba6c731b20e5cd7b55f1d747f00a6523291 Author: Chris Wilson Date: Sun Dec 5 01:08:51 2010 +0000 i965: Also flush the vertex buffer when restarting the array. As a corollary to filling one vertex array and beginning a new one is remembering to emit the old one before overwriting... Signed-off-by: Chris Wilson commit f40f8bcf07bd7f496bfd3069eddfcbeb3d1bf8be Author: Chris Wilson Date: Sun Dec 5 00:26:22 2010 +0000 i965: Check for potential vertex array overflow every time There was a reason why we need to check at the start of every composite operation to see if we have enough space in the array to fit the vertices, which I promptly forgot when moving the code around to make it look pretty. * sigh. Signed-off-by: Chris Wilson commit 55c5f1876e2329a938955967f5d45c814e50beb5 Author: Chris Wilson Date: Fri Dec 3 02:05:05 2010 +0000 Wait on the current buffer to complete when running synchronously. And remove the vestigal wait upon changing crtc as this is more properly done in the kernel. Signed-off-by: Chris Wilson commit 3cc74044ce3546cc7dc2e918cbabbb41a77f4026 Author: Chris Wilson Date: Thu Dec 2 12:07:44 2010 +0000 i965: Amalgamate surface binding tables Signed-off-by: Chris Wilson commit a1fa0dbfdafea32139d4457d81b9d722df955eaf Author: Chris Wilson Date: Thu Dec 2 11:25:05 2010 +0000 i965: Upload an entire vbo in a single pwrite, rather than per-rectangle Signed-off-by: Chris Wilson commit 23437fe6769322d48cc3d264660717475b5d0d74 Author: Chris Wilson Date: Fri Dec 3 01:16:16 2010 +0000 i965: Use reciprocal scale factors to avoid the divide per-vertex-element Signed-off-by: Chris Wilson commit 0bb135c40e5ac1bf7593ec1d68d2815cbf47aa25 Author: Chris Wilson Date: Tue Nov 23 22:27:38 2010 +0000 Disable BLT for i830 and 845G This pair of chipsets seem broken beyond repair, specifically the erratum that causes the wrong PTE entry to be invalidated, so disable our incorrect attempts to use the BLT on those devices. Signed-off-by: Chris Wilson commit 33c08882c0d551afb28baef643279901dcc65fa9 Author: Keith Packard Date: Wed Nov 17 16:37:53 2010 +0800 Mark outputs as DPMSModeOn and restore backlight at mode set The kernel always turns monitors on when doing mode setting, and so no further DPMS action is required. Note this in the mode setting code by marking the updated DPMS mode and restoring any saved backlight level. Signed-off-by: Keith Packard Reviewed-by: Eric Anholt commit cc930a37612341a1f2457adb339523c215879d82 Author: Chris Wilson Date: Sun Nov 14 19:47:00 2010 +0000 uxa: Relax fencing some more for gen3 Allow fenced allocations even for small pixmaps if the kernel supports relaxing fencing (where only the used pages are allocated). Signed-off-by: Chris Wilson commit 3c5b1399e29ef577b8b91655b5e1c215d1b6dfbb Author: Chris Wilson Date: Tue Nov 9 20:20:06 2010 +0000 i915: Disable maximum state addresses As the kernel controls the relocation of state buffers, we should not hard code the maximum permissible value for them. Fixes an eventual hang with full-gtt. Reported-by: Peter Clifton Signed-off-by: Chris Wilson commit 81d355a8dcdfe3d0924bb843bcc0e6546f695391 Author: Chris Wilson Date: Tue Nov 9 09:28:54 2010 +0000 uxa: Fix crash after allocation failure Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=31487 Reported-by: Thomas Fjellstrom Signed-off-by: Chris Wilson commit 861ee17ae662ffec2269292f92da7833df508b1a Author: Carl Worth Date: Fri Nov 5 12:55:02 2010 -0700 Update version number to 2.13.901 In preparation for a snapshot. commit aaa436b437839a80d340cc5ad598ec4b47ed2eff Author: Carl Worth Date: Fri Nov 5 12:54:16 2010 -0700 NEWS: Add release notes for 2.13.901 snapshot. An intermediate snapshot to capture recent developments. commit 6b66b45b1ff207b80b81973745f42a776448f764 Merge: a44a63d 9f232ad Author: Carl Worth Date: Fri Nov 5 12:51:55 2010 -0700 Merge commit '2.13.0' This changes the version number and adds the 2.13.0 release notes, (which were otherwise missing from the master branch). commit a44a63d2ff6c01c3dc61de6f736dd441ddd25e52 Author: Chris Wilson Date: Fri Nov 5 09:58:45 2010 +0000 Wait for any pending rendering before switching modes. A perennial problem we have is the accursed WAIT_FOR_EVENT hangs, which occur when we switch the framebuffer before the WAIT_FOR_EVENT completes and upsets the GPU. We have tried more subtle approaches to detected these and fix them up in the kernel, to no avail. What we need to do is to delay the framebuffer flip until the WAIT completes, which is quite tricky in the kernel without new ioctls and round-trips. Instead, apply the big hammer from userspace and synchronise all rendering before changing the framebuffer. I expect this not to cause noticeable latency on switching modes (far less than the actual modeswitch) and should stop these hangs once and for all. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=31401 (...) Signed-off-by: Chris Wilson commit 52b32436b9e14a3e13818f80102150ff5bc3c002 Author: Chris Wilson Date: Wed Nov 3 19:42:26 2010 +0000 Downgrade tiling allocation failure to a warning We emitted this message as an error even though we fallback and attempt to allocate a non-tiled framebuffer before failing (with an appropriate error message). Signed-off-by: Chris Wilson commit 18839aaec505f8bbdb0690fe694162bf09a87d5c Author: Chris Wilson Date: Wed Nov 3 18:14:29 2010 +0000 Fallback to shadow for Sandybridge if we don't have access to the BLT If we attempt to emit BLT batches without kernel support, we just end up with EINVAL and no rendering. Prevent this, and avoid uncached rendering, by restoring the shadow fallback paths if there is no BLT support. Signed-off-by: Chris Wilson commit 8ff37667bf864b771d16a58fc5041cb48408b6a8 Author: Eric Anholt Date: Tue Nov 2 10:36:03 2010 -0700 Remove the intermittent GEM_THROTTLE call. This is a holdover from early GEM work when we weren't syncing on the DRI client side. It would keep clients from getting too far ahead and killing their interactivity, by bringing everyone to a halt when anyone was too far ahead. Now, GL clients throttle themselves to avoid the problem, and it turns out that in the case that they don't (long rendering to buffers with no swap), this actually reduces X Server interactivity: instead of lagging of X rendering behind input, you get no response for seconds at a time, then a burst of rendering, then nothing again. Reported by ajax. Tested with moving a window while running cairo-perf-trace on the GL backend (improvement) and X backend (no significant change in responsiveness). commit 540c5742186c26c3aeccb7b5d3ff0f374722a20c Author: Xiang, Haihao Date: Tue Nov 2 11:05:32 2010 +0800 render: use headerless render target write It is weird that some rendercheck cases only work fine with headerless write. Need to update intel-gen4asm to support headerless write Signed-off-by: Xiang, Haihao commit 7a3109312e43ce4c2f600a65fe6bc821a8f0ebde Author: Xiang, Haihao Date: Wed Oct 27 16:51:28 2010 +0800 render: acceleration for composite on Sandybridge Signed-off-by: Xiang, Haihao commit 0ab2c05a291738ee763eb518c46e6dcfe29249a2 Author: Xiang, Haihao Date: Wed Oct 27 16:49:26 2010 +0800 render: fragments for composite on Sandybridge Signed-off-by: Xiang, Haihao commit 21c86548dc2c80632c49ec6ec4da2935a0c99476 Author: Xiang, Haihao Date: Wed Oct 27 16:47:36 2010 +0800 render: fix send instruction used in sampling fragments To prepare for composite on Sandybridge Signed-off-by: Xiang, Haihao commit 8d008ca89d44aa10aeb032d631971aaebb388675 Author: Xiang, Haihao Date: Mon Oct 25 15:39:39 2010 +0800 render: set the surface state base address It is the same as commit 73d4c7d7 Signed-off-by: Xiang, Haihao commit b5ae596a0746e5a37e86f5b0980f9265e0061142 Author: Chris Wilson Date: Mon Nov 1 11:11:09 2010 +0000 Flush BLT batches before starting an atomic RENDER batch Signed-off-by: Chris Wilson commit 42363134bd38fb406d90cdb9e5b227836a793912 Author: Zou Nan hai Date: Mon Nov 1 13:23:36 2010 +0800 Support BLT acceleration on gen6 Signed-off-by: Zou Nan hai Signed-off-by: Chris Wilson commit 5bed685f765671d63642b24e17abd70579c40ddf Author: Zou Nan hai Date: Mon Nov 1 13:23:35 2010 +0800 add BLT ring support gen6+ platform has a BLT engine with seperate command streamer to support BLT commands. Signed-off-by: Zou Nan hai [ickle: merge trivial conflict] Signed-off-by: Chris Wilson commit 6ec3ff134baeac33d9d5821695d5f80843a96c91 Author: Xiang, Haihao Date: Fri Oct 15 11:21:09 2010 +0800 Xv: don't call intel_wait_for_scanline on Sandybridge MI_LOAD_SCAN_LINE_INCL command is not available on sandybridge. Signed-off-by: Xiang, Haihao commit 9e73cbd7eea61f2fe69692056ca446a4d95e17c2 Author: Xiang, Haihao Date: Fri Oct 15 11:20:22 2010 +0800 Xv: enable TextureAdaptor for Sandybridge Signed-off-by: Xiang, Haihao commit 3213849907bf1b1468872668d10764054c1034fe Author: Xiang, Haihao Date: Wed Oct 27 09:54:05 2010 +0800 Xv: setup pipeline for Xv on Sandybridge Signed-off-by: Xiang, Haihao commit 9e4dd27aa848acae1b74b77bbbc0a5bb9f6a502a Author: Xiang, Haihao Date: Thu Oct 21 16:10:14 2010 +0800 Xv: fragments for xv on Sandybridge. Need to update intel-gen4asm to build these fragments Signed--off-by: Xiang, Haihao commit e34b3f6ef5811d5e8d156dd6582df8f442e9792e Author: Xiang, Haihao Date: Thu Oct 21 13:15:45 2010 +0800 Xv: Send instruction doesn't use implied move when sampling YUV surface The two fragments will be reused for sampling YUV surface and send doesn't have implied move on Sandybridge Signed-off-by: Xiang, Haihao commit 73d4c7d7b8efc920ac09885dcace5a536f93b70f Author: Xiang, Haihao Date: Wed Oct 13 09:13:50 2010 +0800 Xv: set the surface state base address To prepare for Xv on Sandybridge. It is easy to fill the binding table without relocation and make sure that the pointer to binding table only uses bits[15:0]. Signed-off-by: Xiang, Haihao commit 5afc7472b155e8e940f12a38baf80c298dc3b364 Author: Chris Wilson Date: Fri Oct 29 12:23:59 2010 +0100 intel: don't pass a dangling pointer to GET_PARAM I fail at cut'n'paste. Signed-off-by: Chris Wilson commit 6b3ce2e8701e7c3fcd30ca07fc13a2cdcf62560d Author: Chris Wilson Date: Fri Oct 29 11:07:50 2010 +0100 uxa: Enable reduced fence sizes for i915 Depends on libdrm 362457715faacd3101929e5f0d8ae250d0ad09df (for HAS_RELAXED_FENCING define). Signed-off-by: Chris Wilson commit b066ddda31e40afa652ae51ffc35025ac3ef6f2a Author: Adam Jackson Date: Sun Oct 3 14:42:56 2010 -0700 intel: Listen for hotplug uevents (V3) This connects the kernel uevent indicating monitor hotplugging to the RandR notification events so that X applications can be notified automatically when monitors are connected or disconnected. This also adds a configuration option to disable hotplug events. V2: missed a #ifdef HAVE_UDEV around some udev-specific declarations V3: document Hotplug option in man page Signed-off-by: Keith Packard commit a1c54f69643671ce296c57d132852e9846cc41d3 Author: Eric Anholt Date: Mon Oct 18 13:28:44 2010 -0700 Fix violation of gen6 requirements for depthbuffer tiling. In general, demoting of tiling of DRI2 buffers seems dubious, as we've got various bits of functionality that won't all work together unless buffers are tiled as expected. This just covers one instance of the problem, caught by assertions in Mesa. Fixes: fbo-1d fbo-d24s8. glean/readPixSanity glean/rgbTriStrip glean/scissor commit 33133a1d38a4dce60d2e135663b545895fc85abc Author: Chris Wilson Date: Sun Oct 17 09:21:06 2010 +0100 display: do not report failure for setting unrecognised properties We need to accept any changes to properties not handled by ourselves -- we can't validate the changes ourselves. Denying those changes breaks EDID reporting, for example. Reported-by: Elvis Pranskevichus Signed-off-by: Chris Wilson commit 8507ef50ba08061ac158848a7c4c136e1755d87e Author: Chris Wilson Date: Wed Oct 13 09:45:59 2010 +0100 Fix driverName regression for i830 from 4083197a Reported-by: Stefan Glasenhardt Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=30808 Signed-off-by: Chris Wilson commit 15aab1ef982fb67eb2ea463089f070afd9a2cfe5 Author: Chris Wilson Date: Wed Oct 13 09:50:25 2010 +0100 Move EDID_COMPLETE_RAWDATA define to intel.h to avoid redifinition warning The compiler was simply warning that we defined the name prior to including the original definition, so reorder. Signed-off-by: Chris Wilson commit d2bc2212f8be746e697a65555d6d83bc66d9ce43 Author: Stefan Dirsch Date: Fri Oct 8 17:48:38 2010 +0200 Make driver compile for 1.6 Xserver series again. Part deux. Signed-off-by: Matthias Hopf Acked-by: Chris Wilson commit 5472359d6860af655a3c286d30558540376c9fdb Author: Chris Wilson Date: Fri Oct 8 15:23:13 2010 +0100 dri: Check for pixmap privates before dereferencing them It is still possible for the pixmap allocator to return a software only pixmap (i.e. without an associated GEM buffer or intel_pixmap), so check before dereferencing. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=30707 Reported-by: Matthias Hopf Signed-off-by: Chris Wilson commit 4083197a44d1a1a05d33654b3c7d6e96d7472fe7 Author: Chris Wilson Date: Thu Oct 7 12:45:02 2010 +0100 Include a chipset generation number to clarify device specific paths. Signed-off-by: Chris Wilson commit 1444ea35706992a1f094fe029e6a7d9eec0f93c6 Author: Chris Wilson Date: Wed Oct 6 12:36:30 2010 +0100 dri+shadow: Only tweak the acceleration of CopyRegion if using shadow. Signed-off-by: Chris Wilson commit f286891d1415fc607f0c1b574e6bce4b3759493e Author: Chris Wilson Date: Wed Oct 6 11:38:48 2010 +0100 dri: Reattach the fake pixmap for the shadow scanout to the drawable. Signed-off-by: Chris Wilson commit 23ee926bcd8ddd7a2785786efdf119e4fae75bfd Author: Chris Wilson Date: Wed Oct 6 10:20:49 2010 +0100 uxa: Skip a pixmap lookup if there is no driver finish access function Signed-off-by: Chris Wilson commit 707901bf98073fb56179a0b61c806f85ef8f413c Author: Chris Wilson Date: Wed Oct 6 08:47:20 2010 +0100 uxa: Re-enable acceleration. A side-effect of discriminating offscreen based on the devPrivate.ptr was that it broke uxa_finish_access and so after any fallback to s/w on a Pixmap, it remained in software for the reminder of its life. Introduce an explicit boolean to mark whether or not hardware acceleration is enabled for a pixmap (with a GEM buffer). Signed-off-by: Chris Wilson commit 54f545e0631a287f421ef6a6f20831624b212cf2 Author: Chris Wilson Date: Wed Oct 6 08:50:13 2010 +0100 Revert "Clear pixmap->devPrivate.ptr [regression in 7c7294e]" This reverts commit 48b4e224297fa807be0e2bc7a67bf7e94579e8de. The better fix is to manually mark the pixmap when acceleration is and is not permitted. Whilst the devPrivate.ptr are invalid upon creation, it is not worth carring code that serves no purpose. commit d1925deedd70c6c4374e42de5aa44989fe328b58 Author: Chris Wilson Date: Wed Oct 6 00:17:03 2010 +0100 Clear pixmap->devPrivate.ptr [regression from 7c7294e] ModifyPixmapHeader(pixdata = NULL) does not clear the pixmap->devPrivate.ptr, instead the NULL value is interpreted as meaning to keep the current value. (This is similar to the interpretation of the other invalid values like depth=-1 which also mean not to change the current property). However pixadata=NULL is indeed a valid value, and in 7c7294e, devPrivate.ptr == NULL was used to differentiate a bo pixmap from a system pixmap. Except that we never did clear the ptr as intended, and so X would immediately crash. Signed-off-by: Chris Wilson commit 4bf2182bf9ee64795e2bf524fedf7b6c36f3cd03 Author: Chris Wilson Date: Tue Oct 5 09:59:04 2010 +0100 shadow: Use a cacheable shadow for all generations Always avoid direct rendering to the uncached scanout buffer, redirecting all 2D access to the shadow instead. Then for the couple of platforms where either the BLT is not ready or cannot be trusted (i8xx) perform the front buffer fixup using the uncached writes from the CPU. Signed-off-by: Chris Wilson commit b27a521214728052b8b10342aaacbaf1622b8e2c Author: Chris Wilson Date: Mon Oct 4 20:31:44 2010 +0100 shadow: Enable shadow by default on SandyBridge SandyBridge 2D support is far from complete, so instead of permanently falling back and always using uncached GTT mapping for rendering, use the shadow buffer instead. Signed-off-by: Chris Wilson commit 3e641459f8997f45288cc3cb314e50f2b5a629fd Author: Chris Wilson Date: Mon Oct 4 20:31:01 2010 +0100 shadow: Disable BLT for SandyBridge The blitting code is incorrect for SandyBridge so disable until the BLT ring is ready. Signed-off-by: Chris Wilson commit 7c7294ec00d6c3a454a17a1b9983d14d0655162c Author: Chris Wilson Date: Mon Oct 4 11:56:27 2010 +0100 shadow+dri2: Allow dri2 to be independently enabled with shadow To enable DRI we create GEM buffers for the client to render into with hardware acceleration. In order to maintain coherency between any 2D render operations with the independent 3D clients (this includes the reading of 2D rasterisation by the direct rendering client, e.g. compiz using texture_from_pixmap) we need to replace the shadow pixmap with the GTT mapping. Therefore 2D rendering to a DRI buffer will be to uncached memory and thus penalised -- but the direct rendering clients will have full hardware acceleration. Signed-off-by: Chris Wilson commit 16a5d0ee3ca4d41af6a6876a8baf81e3f715781d Author: Chris Wilson Date: Mon Oct 4 11:21:17 2010 +0100 shadow: Map the scanout directly on i8xx Even with the minimal use of the BLT to copy from the GTT shadow to the GTT scanout, i830 was still hanging. Just write to the scanout directly. Signed-off-by: Chris Wilson commit 516d235c5b3bd51e9533925140d95d3b4533129c Author: Chris Wilson Date: Mon Oct 4 11:10:57 2010 +0100 Split shadow handling routines to their own file. This is about to get messy, so separate out the shadow from the normal code. Signed-off-by: Chris Wilson commit 455f2939a661764ebb8d1747d44e16a0a8937808 Author: Chris Wilson Date: Sat Oct 2 13:32:58 2010 +0100 Do not claim the PCI device if !KMS By returning FALSE whilst probing if we can't find a KMS driver, we allow X to fallback to trying the VESA driver -- rather than failing. The initial idea for this was by Julien Cristau. Reported-by: Julien Cristau Signed-off-by: Chris Wilson commit 97fea2fd0c36d78f4e940b124a7886c4a3f3dfc7 Author: Chris Wilson Date: Sat Oct 2 10:05:59 2010 +0100 legacy/i810: Remove pVisualConfigs Completely unused pulling in an undesired glxproto dependency. Reported-by: Cyril Brulebois Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=30554 Signed-off-by: Chris Wilson commit 8784c4f5a1524fb979b00c7ce7981cbc1dcf0ec0 Author: Chris Wilson Date: Fri Oct 1 21:40:36 2010 +0100 video: Disable TextureAdaptor for Sandybridge The acceleration code does not yet exist, so we try to use gen4 shaders which hangs the chip and makes everybody upset. Disable it until such a time as support is written for Sandybridge XVideo. Reported-by: Matthias Hopf Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=30500 Signed-off-by: Chris Wilson commit 9f232ad67c72b7869dcd112207bc41dc01d7d104 Author: Carl Worth Date: Thu Sep 30 08:45:49 2010 -0700 Increment version to 2.13.0 For the new, major release. commit 99ec9bb0f1eaf2e5c04614957a51624a928526e5 Author: Carl Worth Date: Thu Sep 30 08:45:15 2010 -0700 Add release notes for the 2.13.0 release. Which is functionalliy identical to 2.12.902. commit 345c963e443ec325f1ff530512a356ddb318ff70 Author: Zhenyu Wang Date: Thu Sep 30 10:58:49 2010 +0800 Always use tiling on Sandybridge Sandybridge requires kind of buffer must be tiling, like depth. And we would or have all tiling cases handled fine. So not allow user to turn off tiling on Sandybridge+ may be fine. Signed-off-by: Zhenyu Wang commit d4c013cb9d92d2959c9a98eca20e43ab14ae22e3 Author: Carl Worth Date: Tue Sep 28 20:49:52 2010 -0700 Increment version to 2.12.902 Just before making the 2.12.902 (2.13-rc2) snapshot. commit 35e7da7df0b286b5168a85d540cc9fdf55c1cae6 Author: Carl Worth Date: Tue Sep 28 20:49:10 2010 -0700 Add release notes for the 2.12.902 snapshot. Summarizing the 5 recent changes to the driver. commit 5704955538f59a2b7a5eff67cc23953d2f27d523 Author: Carl Worth Date: Tue Sep 28 20:44:40 2010 -0700 Fix to depend on the (just-released) libdrm >= 2.4.22 With libdrm < 2.4.22, the compilation of xf86-video-intel would fail due to an undefined DRM_MODE_CONNECTOR_eDP. We now ensure that a sufficiently new libdrm is available before compiling. commit 40c9d180bf43af84ae05d88e993deedf76e263e6 Author: Chris Wilson Date: Sun Sep 26 23:09:04 2010 +0100 Retry framebuffer allocation if first attempt fails. If we are tiled, we may fail the allocation due to an EIO and so not being able to set tiling. Try again with an untiled request in this case. Signed-off-by: Chris Wilson commit 537e73f3f935b917f2f5f9b51499cb29d65e3889 Author: Chris Wilson Date: Fri Sep 24 17:37:41 2010 +0100 Disable dri2 after forcing fallbacks If we force fallbacks, then we will only create pixmaps in system memory, preventing DRI2 from passing valid bo names to the clients. In this case, they will just fallback to swrast. If we disable DRI2 after forcing fallbacks (e.g. regenerating after a GPU hang or explicitly disabled with the shadow buffer) then it is simpler just to disable the extension and allow mesa to use pure swrast. Reported-by: Julien Cristau Signed-off-by: Chris Wilson commit b84925b9c0842ba4dfa3481c09d3a80f84db4838 Author: Matthias Hopf Date: Tue Aug 17 18:05:02 2010 +0200 Make driver compile for 1.6 Xserver series again. Signed-off-by: Matthias Hopf commit 99f8d68d3dbc54fdd7de4a6df546cfa89422a0c2 Author: Chris Wilson Date: Wed Sep 22 12:34:00 2010 +0100 display: Refactor is_panel() Signed-off-by: Chris Wilson commit 4ba27e1f2198fc10a1dd268ed9f137bac6e80086 Author: Carl Worth Date: Tue Sep 21 17:43:01 2010 -0700 Bump version to 2.12.901 Just prior to pushing out this new snapshot. commit 334265fa5ef18405bd8c897343fff49ad53f5fce Author: Carl Worth Date: Tue Sep 21 17:42:21 2010 -0700 NEWS: Add notes for the 2.12.901 snapshot (Otherwise known as rc1 for 2.13) commit bfba6d6148b938588ffbefdf04f4e71683c66dfd Author: Krzysztof Halasa Date: Tue Sep 21 22:56:26 2010 +0100 Allow interlaced modes. This may not the best method, but it should be a good base on which to build... Signed-off-by: Chris Wilson commit f63c7df6866fed61551d39209b1a262e2a0c4b28 Author: Zhenyu Wang Date: Fri Sep 17 16:39:12 2010 +0800 Disable swap buffer wait on Sandybridge MI_LOAD_SCAN_LINE_INCL command is not available on sandybridge. I haven't got reply on any substitution for it, so turn it off for now. Signed-off-by: Zhenyu Wang commit 55b5fe888002258dd863d41977bfbfcf1a922179 Author: Chris Wilson Date: Thu Sep 16 16:50:13 2010 +0100 Add alternate pci-id for B43 Confirmed by http://en.wikipedia.org/wiki/Intel_GMA Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=30221 Signed-off-by: Chris Wilson commit 08c2caca48323d6d5701dcef3486f850619d7905 Author: Chris Wilson Date: Sun Sep 12 12:34:51 2010 +0100 uxa: Apply source clipping to blits Yes, this should be done in the higher layers. Yes, I have written code to that. No, it is not ready, hence add the sanity check to the SRC_COPY_BLT. This isn't the first report that I've seen, but will be the last. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=30120 Reported-by: rezbit.hex@gmail.com Signed-off-by: Chris Wilson commit ae160d7fbfc79e78dad8702efcc55d9c0c25ff67 Author: Chris Wilson Date: Fri Sep 10 13:19:12 2010 +0100 shadow: Simply modify the Screen pixmap header This is a slightly less risky strategy than having to remember to update all pointers to the old Screen pixmap. Signed-off-by: Chris Wilson commit 0515256490d5bcd55f85af83b84918d1bfe7f8f8 Author: Chris Wilson Date: Fri Sep 10 00:08:58 2010 +0100 display: Free the EDID blob after we copy it to the output, not before. Signed-off-by: Chris Wilson commit 2b96c18165d713cd6781dbf217ec33e11cc961bc Author: Chris Wilson Date: Wed Aug 25 12:56:43 2010 +0100 Enable a shadow buffer and disable GPU acceleration. An attempt to workaround the incoherency in gen2 chipsets, we avoid using dynamic reallocation as much as possible. The first step is to disable allocation of pixmaps using GEM and simply create them in system memory without a backing buffer object. This forces all rendering to use S/W fallbacks. The second step is to allocate a shadow front buffer and assign that to the Screen pixmap. This ensure that the front buffer remains in the GTT and pinned for scanout. The shadow buffer will be rendered to in the normal fashion via the Screen pixmap, and be marked dirty. In the block handler, the dirty shadow buffer is then blitted (using the GPU) over the front buffer. This should completely avoid having to move pages around in the GTT and avoid incurring the wrath of those early chipsets. Secondly, performance should be reasonable as we avoid the ping-pong caused by the small aperture and weak GPU forcing software fallbacks. Signed-off-by: Chris Wilson commit 0fa4321a765126228170ecb9536f32c134886d51 Author: Jesse Barnes Date: Tue Sep 7 13:56:16 2010 -0700 KMS: initialize backlight support for eDP panels too Since being able to control the backlight is handy. Signed-off-by: Jesse Barnes commit f56c839f30ec39ea22bbf6a7fda4cefd7cbfbfc2 Author: Jesse Barnes Date: Tue Sep 7 13:53:47 2010 -0700 KMS: rename LVDS fields to reflect actual usage These are used for panels, not just LVDS connections, so name them as such. Signed-off-by: Jesse Barnes commit 0d26d950fdada1f59dc6cb31fe2f03004825f773 Author: Jesse Barnes Date: Tue Sep 7 13:46:58 2010 -0700 KMS: add fake EDID on eDP too This gives us a few more standard modes on eDP panels with just a simple fixed timing in the VBT, just like on older, LVDS attached panels. Fixes FDO bug https://bugs.freedesktop.org/show_bug.cgi?id=30069. Reviewed-by: Chris Wilson Tested-by: Manoj Iyer Signed-off-by: Jesse Barnes commit 273d34fbc43e54a8539fbe7efc1dca66bbb4e6dc Author: Chris Wilson Date: Tue Sep 7 12:57:04 2010 +0100 display: Query current level after finding max value. The current backlight value is clamped to the valid range [0, max] and so as we queried the value before setting the max, we forced the current backlight to 0 and so set it to be zero on initialising the display. Fixes: Bug 30063 - start X will modify brightness value to zero https://bugs.freedesktop.org/show_bug.cgi?id=30063 which is a regression due to 38f940dfea494d3093236f065392c431be06ae6e. Signed-off-by: Chris Wilson commit 53767cc0d0a58d36cd445da3a31c65b349eebbba Author: Zhenyu Wang Date: Tue Sep 7 14:17:05 2010 +0800 Add more sandybridge graphics device ids New ids for GT2 and GT2+ on desktop and mobile sandybridge, and server sandybridge device ids. commit 00f6af2c8ec899258ffab60dc2d9d80c4d9daec8 Author: Chris Wilson Date: Sat Sep 4 18:45:11 2010 +0100 display: Set MONITOR_EDID_COMPLETE_RAWDATA for large EDIDs Quoting Adam Jackson: "But the X driver looks like it never sets MONITOR_EDID_COMPLETE_RAWDATA, which means the X core doesn't know that any sections beyond the first are present, so it won't ever hand back more than 128 bytes to clients. Boo." This patch is based on his. Signed-off-by: Chris Wilson commit 501e78b0095b00a635dc1dc182fc132f14f08f22 Author: Chris Wilson Date: Sat Sep 4 12:37:39 2010 +0100 Force use of GTT and fence registers for mapping tiled objects If the buffer object is tiled, we need to use the fence registers to perform the appropriate untiling for CPU access. Ensure that we always take this path for tiled objects, regardless of their size. Signed-off-by: Chris Wilson commit b7a8087fbc5cbd944f8345b8a098270c1f63ebab Author: Chris Wilson Date: Wed Aug 25 15:44:20 2010 +0100 Revert "Leave adjustment of backlight to the driver." This reverts commit 9c3e34703dcdc2155faeb2c1886930def19d9989. This commit is not ready, as first the driver needs to handle all controllers, especially those that ignore the BLC and require their own interface. Fortunately, by moving that discovery into the kernel - where it just means finding which ACPI device is attached to the video and has a backlight interface - the userspace code should become much more sane, and work even with multi-gpu, multi-lid systems. But that is for tomorrow. commit 68a5ad497b557ab9e8bff301098ea78c84d385a2 Author: Chris Wilson Date: Wed Aug 25 14:59:04 2010 +0100 uxa: Fallback if faced with large A1 glyphs. Rather than assert, we should fixup the use of large A1 glyphs. However, the simplest approach is to simply fallback to s/w. Fixes: Bug 29430 - [UXA] Crash due assert (uxa_pixmap_is_offscreen(src_pixmap)); https://bugs.freedesktop.org/show_bug.cgi?id=29430 Signed-off-by: Chris Wilson commit 9c3e34703dcdc2155faeb2c1886930def19d9989 Author: Chris Wilson Date: Sun Aug 22 21:27:45 2010 +0100 Leave adjustment of backlight to the driver. commit 104cd0554bde1d109a54db7a93700d5edfabd914 Author: Zhenyu Wang Date: Mon Aug 23 09:48:22 2010 +0800 Add sandybridge D0 support Signed-off-by: Zhenyu Wang commit 271dda84be2f81b32dbe5947e64f5558a89e385e Author: Chris Wilson Date: Sun Aug 22 19:57:24 2010 +0100 display: Use the native intel backlight controller If the i915 driver exposes a native ACPI interface to modify the panel backlight use it in preference to the generic interfaces. On multi-GPU systems, the panel backlight is meant to be connected via the IGP and this ensures that we always find the right interface. Fixes: Bug 29273 - XORG Intel driver chooses wrong acpi_video to control brightness in multi-GPU system https://bugs.freedesktop.org/show_bug.cgi?id=29273 Signed-off-by: Chris Wilson commit 42312bbd8c9cbadf0adc4556b76987857d105bd9 Author: Chris Wilson Date: Sun Aug 22 09:54:18 2010 +0100 Remove accel_pitch_alignment This has to be 64 on all generations currently, so replace the variable with a constant. Signed-off-by: Chris Wilson commit 7f86e5b5da88201e5443e26619c35a1a9745cf68 Author: Matt Turner Date: Sat Aug 21 17:32:25 2010 -0400 Replace ROUND_* macros with ALIGN. Signed-off-by: Matt Turner Signed-off-by: Chris Wilson commit b611bced15c30f7bcd03106ce90668b684c1ada6 Author: Matt Turner Date: Sat Aug 21 17:32:24 2010 -0400 Use ALIGN macro instead of open coding it. Signed-off-by: Matt Turner Signed-off-by: Chris Wilson commit 8b04b350a983b89eb2d741f55baa297a933ac6ea Author: Chris Wilson Date: Thu Aug 19 20:05:08 2010 +0100 Open-code DRICreatePCIBusID() During -configure we would attempt to query the availablility of KMS before the DRI module was loaded, thus we were unable to create a valid bus identifier and so the query failed and we disowned the device. Fixes: Bug 29611 - Xorg -configure fails https://bugs.freedesktop.org/show_bug.cgi?id=29611 Reported-by: Sergey Samokhin Signed-off-by: Chris Wilson commit c882f6a22a862c1664c375e05e5e6fc4bdb04edb Author: Chris Wilson Date: Wed Aug 18 10:21:22 2010 +0100 Move registration of vsync fd from pre-init to screen-init Marty Jack reported an issue he found where the page-flipping handler was being lost on server reset. This results in the swap completion notification being lost, with the sporadic hang of full screen applications like Compiz, flash and even glxgears! Fixes: Bug 29584 - Server in compute loop https://bugs.freedesktop.org/show_bug.cgi?id=29584 There are also several possibly related bugs with similar symptoms, i.e. OpenGL applications hanging on missed swap notifications. Reported-by: Marty Jack Signed-off-by: Chris Wilson Cc: Keith Packard commit 19c48d3b3f33582baa87a9b3a9189e320e4cea45 Author: Chris Wilson Date: Mon Aug 9 10:13:58 2010 +0100 display: outputs are enabled automatically by KMS When an output is attached to a crtc and that crtc is enabled, the output is automatically enabled so we can remove the redundant manual dpms on. Signed-off-by: Chris Wilson commit 6304cb048c745be81dae13f1d936996e04eaa530 Author: Chris Wilson Date: Thu Aug 5 18:13:55 2010 +0100 display: Minor cleanup for adding extra LVDS modes Signed-off-by: Chris Wilson commit 41ae9564350cba8ac9f040f22bad8740e08e14c4 Author: Chris Wilson Date: Thu Aug 5 18:13:30 2010 +0100 display: Refactor EDID attachment to output. Signed-off-by: Chris Wilson commit a8919ab2967a8425e1cdc04b73702ce534f6abf0 Author: Chris Wilson Date: Thu Aug 5 09:59:22 2010 +0100 Revert "display: Cache whether we have probed for an EDID" Dave Airlie advised that hotplug detection can be unreliable and that mode caching, in general, should be done in the kernel in any case. This reverts commit 622e600069ab0efd22586c7a71eecbd4baf21c40. commit 622e600069ab0efd22586c7a71eecbd4baf21c40 Author: Chris Wilson Date: Wed Aug 4 09:57:12 2010 +0100 display: Cache whether we have probed for an EDID Remember for the detection cycle whether we have already probed for the EDID -- as this can be slow. Signed-off-by: Chris Wilson commit a6a707ca13097b85b319283b3a174b1986056ab8 Author: Chris Wilson Date: Wed Aug 4 09:54:28 2010 +0100 display: Embed the lvds size into the connector Remove one very common allocation. Signed-off-by: Chris Wilson commit 6c7d105ccae32fba49a07a03b726cc3fe2de3d27 Author: Chris Wilson Date: Wed Aug 4 09:51:34 2010 +0100 display: Handle cursor error paths. Check that the cursor was allocated before freeing. Signed-off-by: Chris Wilson commit 38f940dfea494d3093236f065392c431be06ae6e Author: Chris Wilson Date: Wed Aug 4 09:50:14 2010 +0100 display: Tidy backlight initialisation Mostly whitespace and a single error-code fix. Signed-off-by: Chris Wilson commit 2b7263b771d94401cb4ea6cbf4dc7a295eeda7c0 Author: Chris Wilson Date: Wed Aug 4 09:46:01 2010 +0100 display: Check for buffer overrun in output name lookup. The kernel may know about more types than we do, so protect ourselves from reading from beyond the end of the string array. Signed-off-by: Chris Wilson commit 56629166915155628e70e6ec8052c0d220a8bb68 Author: Dave Airlie Date: Wed Aug 4 15:35:41 2010 +1000 intel: add output names for later additions to kernel commit fe7dee7fe1cf12665bc5ba3923cd1a8f74cbbe57 Author: Chris Wilson Date: Sun Aug 1 14:15:09 2010 +0100 Remove the final references to the drmmode prefix In particular fix the compile regression for intel_do_pageflip(). Signed-off-by: Chris Wilson commit 4f8b279f32cfda0f3b8fa1542b3fe00051520012 Author: Chris Wilson Date: Sun Aug 1 12:35:57 2010 +0100 intel_display: Miscellaneous tidy A mixture of whitespace and closing of leaks on error paths. Signed-off-by: Chris Wilson commit db7cd7b9f0ce8e435c121071eaadf8e6dcebf9e8 Author: Chris Wilson Date: Sun Aug 1 11:25:42 2010 +0100 Rename drmmode_display to intel_display And fixup all the drmmode_* functions to have an intel prefix and categorise those into intel_mode, intel_crtc, intel_output and intel_property so that the functions are a little more self-descriptive and, more importantly, are consistent. Signed-off-by: Chris Wilson commit 0be3e95c844247746742805830860ace9f546d99 Author: Kristian Høgsberg Date: Fri Jul 30 09:39:58 2010 -0400 Remove explicit batchbuffer submit in DRI2 copyregion Now that we submit from the flush callback chain, we know we'll always submit before the client receives the reply or event that blocks it from rendering the next frame. commit 69d65f9184006eac790efcff78a0e425160e95aa Author: Kristian Høgsberg Date: Thu Jul 29 18:31:48 2010 -0400 Submit batch buffers from flush callback chain There are a few cases where the server will flush client output buffers but our block handler only catches the most common (before going into select). If the server flushes client buffers before we submit our batch buffer, the client may receive a damage event for rendering that hasn't happened yet. Instead, we can hook into the flush callback chain, which the server will invoke just before flushing output. This lets us submit batch buffers before sending out events, preserving ordering. Fixes 28438: [bisected] incorrect character in gnome-terminal under compiz https://bugs.freedesktop.org/show_bug.cgi?id=28438 Signed-off-by: Kristian Høgsberg commit b68d4fcab5d671946c187947f5d00aedff939852 Author: Chris Wilson Date: Wed Jul 28 13:41:41 2010 +0100 drmmode: Only treat a backlight as connected if it has a non-zero max Optimistically might help https://bugs.freedesktop.org/show_bug.cgi?id=29273 Signed-off-by: Chris Wilson commit 938ef4eaec0e406e198105d866f6d2a1be5e0ef8 Author: Kristian Høgsberg Date: Wed Jul 28 07:57:42 2010 -0400 legacy: Remove long gone use of GlxSetVisualConfigs() This removes the last dependeny on anything GL/GLX in the driver. commit fba6651a92af76d7a8d034e1acfdad5023cd311c Author: Kristian Høgsberg Date: Tue Jul 27 12:59:39 2010 -0400 Drop use of GL types in the driver Still used in i810 for building the glx visuals. commit 34b9d45be6cb0c42f031677a0bda8a5013472ae8 Author: Jesse Barnes Date: Tue Jul 27 09:10:04 2010 -0700 Revert "configure.ac: add xi and gl requirements" This reverts commit d580fa82a1cf339d2d1fd1055d137c0b23cd04f3. Oops, these should be server or other package dependencies, we don't actually need them for xf86-video-intel. commit d580fa82a1cf339d2d1fd1055d137c0b23cd04f3 Author: Jesse Barnes Date: Mon Jul 26 16:44:19 2010 -0700 configure.ac: add xi and gl requirements Could probably be earlier versions, but having these here makes packaging easier. commit f46a8dfce59013ce593c15f7166598d0fff9198f Author: Daniel Vetter Date: Mon Jul 26 23:02:02 2010 +0200 video: kill do { ... } while (ret != -EINTR) loops Chris Wilson likes to sprinkle these all over, but in this case it's just misleading. libdrm already does this for us. Signed-off-by: Daniel Vetter commit 1cb69b9a77b7afbb4358757556065e10a6b15ea8 Author: Daniel Vetter Date: Mon Jul 26 23:00:20 2010 +0200 video: kernel overlay needs triple buffering The kernel overlay code does asynchronous overlay flips. So keep onto two old buffers, for otherwise the rendering of the next frame might overwrite the contents of the currently still displaying one. With ~25fps videos and ~50 Hz screens that's rather unlikely, still, fix it. Signed-off-by: Daniel Vetter commit 68df6b2790891683ee2e58daaad34ef17ae344f5 Author: Gaetan Nadon Date: Fri Jul 23 17:39:37 2010 -0400 simplify Makefile as per-target compilation flags are not needed Per-target compilation flags (libIntelXvMC_la_CFLAGS) are required when multiple targets which require different compiler flags, are build in the same makefile. Automake issues a command with -c and -o flags which not all compilers support. The object fles are prefixed with libIntelXvMC_la. The macro AM_PROG_CC_C_O must then be used to provide this feature on compilers that do not have it. If not, a warning is issued at make time. This macros checks for compiler support and if missing, uses a "compile" script it generates in the package root directory. Currently the driver uses per-target flags but the macro is missing. Rather than adding the macro, this patch stops using per-target flags by using the AM_CFLAGS variable for all targets in the makefile, as there is only one. Acked-by: Chris Wilson Signed-off-by: Gaetan Nadon commit 0ae30b6d1f4f6c2f2db9443f1027ea41849ad087 Author: Gaetan Nadon Date: Wed Jul 21 16:49:04 2010 -0400 config: add comments for main statements commit b12b2219677d94933774955b50545efceff57279 Author: Gaetan Nadon Date: Wed Jul 21 16:07:00 2010 -0400 config: replace deprecated use of AC_OUTPUT with AC_CONFIG_FILES Signed-off-by: Gaetan Nadon commit 8b2d65da41461735b1138fdd6e0efc96419cd815 Author: Gaetan Nadon Date: Wed Jul 21 14:37:41 2010 -0400 config: replace deprecated AC_HELP_STRING with AS_HELP_STRING Signed-off-by: Gaetan Nadon commit 3feeef5294b1a2df05a25f4f4eb6400be6cc4ee1 Author: Gaetan Nadon Date: Wed Jul 21 14:05:22 2010 -0400 config: replace deprecated AM_CONFIG_HEADER with AC_CONFIG_HEADERS Signed-off-by: Gaetan Nadon commit 74a828e286c46473b8edf6de32c1989c052dd82c Author: Gaetan Nadon Date: Wed Jul 21 09:27:42 2010 -0400 config: complete AC_INIT m4 quoting Signed-off-by: Gaetan Nadon commit 33ec66d7006e2a7e287fd77c53a31db49220591a Author: Gaetan Nadon Date: Tue Jul 20 21:54:11 2010 -0400 config: remove unrequired AC_SUBST([DRI_CFLAGS]) This macro is called by PKG_CHECK_MODULES Signed-off-by: Gaetan Nadon commit b20b8ac937505310fbe68f07e42cafe4ba88da76 Author: Gaetan Nadon Date: Tue Jul 20 20:24:42 2010 -0400 config: remove unrequired AC_HEADER_STDC Autoconf says: "This macro is obsolescent, as current systems have conforming header files. New programs need not use this macro". Signed-off-by: Gaetan Nadon commit 9fe5169fffeee10b1262f160a15c11c283e2faa7 Author: Gaetan Nadon Date: Tue Jul 20 19:41:31 2010 -0400 config: remove AC_PROG_CC as it overrides AC_PROG_C_C99 XORG_STRICT_OPTION from XORG_DEFAULT_OPTIONS calls AC_PROG_C_C99. This sets gcc with -std=gnu99. If AC_PROG_CC macro is called afterwards, it resets CC to gcc. Signed-off-by: Gaetan Nadon commit e6b8811e5bb30e2cd4e35b1fe84d69f18a73ee0e Author: Gaetan Nadon Date: Tue Jul 20 18:45:19 2010 -0400 config: update AC_PREREQ statement to 2.60 Unrelated to the previous patches, the new value simply reflects the reality that the minimum level for autoconf to configure all x.org modules is 2.60 dated June 2006. ftp://ftp.gnu.org/gnu/autoconf/autoconf-2.60.tar.gz Signed-off-by: Gaetan Nadon commit 2d85efaa36791bd441f069b612ee86387353d1ed Author: Gaetan Nadon Date: Tue Jul 20 16:15:29 2010 -0400 config: upgrade to util-macros 1.8 for additional man page support Use MAN_SUBST now supplied in XORG_MANPAGE_SECTIONS The value of MAN_SUBST is the same for all X.Org packages. commit 142ffa287236adbc3c5f6ead46c63c4c541ace0c Author: Chris Wilson Date: Thu Jul 22 16:42:26 2010 +0100 video/i915: ValidateGC after setting clip. Order is important. And ensure that the scratch GC is performing clip by children. Fixes: Bug 29213 - video artifacts if used dualscreen mode https://bugs.freedesktop.org/show_bug.cgi?id=29213 Signed-off-by: Chris Wilson commit c57840b272ba88fddf22484929608431879b0fab Author: Chris Wilson Date: Wed Jul 21 09:06:06 2010 +0100 uxa: Check for failed pixmap allocation Fixes: Bug 29187 - crash in intel_drv https://bugs.freedesktop.org/show_bug.cgi?id=29187 Backtrace: 0: /usr/bin/X (xorg_backtrace+0x28) [0x466808] 1: /usr/bin/X (0x400000+0x67c79) [0x467c79] 2: /lib/libpthread.so.0 (0x7ff19b297000+0xef60) [0x7ff19b2a5f60] 3: /usr/lib/xorg/modules/drivers/intel_drv.so (0x7ff197986000+0x34684) => uxa/uxa-render.c:841 Signed-off-by: Chris Wilson commit 7a4bfaf42412f373bbc8ebe1536018dcacde7436 Author: Dave Airlie Date: Tue Jul 20 11:30:44 2010 +1000 intel: respect tiling disable. For testing purposes its nice to know tiling isn't being used anywhere. Signed-off-by: Dave Airlie commit d48d584a822ae5adae64f6d5f09faa4ac9b8de73 Author: Chris Wilson Date: Thu Jul 15 14:37:59 2010 +0100 video: Free the buffers immediately after turning off. Signed-off-by: Chris Wilson commit 24bdfe0d5eb4e890e9c63bbb4617efaa0768ab7f Author: Chris Wilson Date: Thu Jul 15 13:54:04 2010 +0100 video: Reuse the old buffers. After passing the new buffer to the kernel, the old buffer is unpinned and becomes available for re-use. So keep hold of the old buffer and swap after a PutImage. This greatly reduces the amount of CPU time consumed by the kernel on behalf of the video overlay -- by only allocating two buffers for an entire sequence, we avoid clflushing and page allocation on every frame. Signed-off-by: Chris Wilson commit 2267e5928bd90a6997970584462bd97df447e819 Author: Chris Wilson Date: Tue Jul 13 10:36:34 2010 +0100 Workaround a broken container_of define in list.h commit 798c3a5fc6e3d44c9b94e6bc4e72385a10939f52 Author: Chris Wilson Date: Tue Jul 13 10:30:33 2010 +0100 Teardown the bufmgr on shutdown as well. Signed-off-by: Chris Wilson commit b2e98227d11a60e90b3ea518f14962f2764ffa68 Author: Chris Wilson Date: Tue Jul 13 10:15:34 2010 +0100 Remove the duplicate drmmode prototypes. Signed-off-by: Chris Wilson commit 5de1b74d64f807b59c730871d4cb171484db9780 Author: Chris Wilson Date: Tue Jul 13 09:34:53 2010 +0100 modes: There may be more than one crtc and output... DESTROY THEM ALL! In order to cleanup all CRTCs and outputs on shutdown, we need to keep a list of the individual structures and iterate over that list on shutdown. Also, the output and crtcs are configured just once and not for each screen generation so move the shutdown to the termination and not on CloseScreen. Oops. Signed-off-by: Chris Wilson commit 3a7c25ff8ddd45c9d9eca5cc2228552847ca9e7d Author: Chris Wilson Date: Mon Jul 12 19:47:46 2010 +0100 video: Apply overlay stride errata for i830 and i845 Due to an erratum on these chipsets, the overlay stride must be a multiple of 256 bytes. Signed-off-by: Chris Wilson commit 56e5816252afbc7dd555c308ca678fce7ee584cc Author: Chris Wilson Date: Mon Jul 12 17:26:37 2010 +0100 video: Copy DummyEncoding into each adapter. As we use the static DummyEncoding and may attempt to modify it for each adaptor (on each device), we should use copies instead. Signed-off-by: Chris Wilson commit e30f0338fbf894fa3fdab20f4f9a97495cd341da Author: Keith Packard Date: Thu Jul 1 09:56:40 2010 -0400 Destroy screen pixmap on screen close. This avoids a memory leak on server reset. Signed-off-by: Keith Packard [ickle: Added comments from Keith that explain the necessity of destroying the pixmap ourselves and why chaining up in this instance is not the correct approach.] Reviewed-by: Chris Wilson commit 6e08b0f48f0d6ddaca963de6f4fd3b56f32512c6 Author: Fernando Carrijo Date: Thu Jul 1 07:15:22 2010 -0300 Purge macro NEED_EVENTS Signed-off-by: Fernando Carrijo Acked-by: Tiago Vignatti Reviewed-by: Alan Coopersmith commit 5ce3f536b7169a1ee9eadf830e61bab230dd7239 Author: Chris Wilson Date: Fri Jul 9 14:06:43 2010 +0100 drmmode: Destroy the output on shutdown Signed-off-by: Chris Wilson commit 6ff369cd2602497ee11273b50f76d1a7bf25991d Author: Chris Wilson Date: Fri Jul 9 14:00:31 2010 +0100 drmmode: Destroy Crtc on screen shutdown Should fix: Bug 26946 - CRTC cursor BO leak in 2D https://bugs.freedesktop.org/show_bug.cgi?id=26946 Signed-off-by: Chris Wilson commit 6fba8c449f61341a98a5ae8d97effa6fd0610fd4 Author: Chris Wilson Date: Fri Jul 9 12:12:13 2010 +0100 Add support for I854. I spotted that the kernel knew of the I854, but the pci-id was never added to the ddx. Signed-off-by: Chris Wilson commit 141e88c8730a099a6ca5eab1350c2e53a680cb0d Author: Chris Wilson Date: Fri Jul 9 12:11:07 2010 +0100 video: forgotten amendment to previous commit. An extra sanity check to skip the wait if all clipped... Signed-off-by: Chris Wilson commit 272d1c14a39c32ade39b5a8b080a891f2b3d6e8e Author: Chris Wilson Date: Fri Jul 9 10:41:19 2010 +0100 video: apply the crtc box checks from dri. The dri code is much more careful in ensuring that the scan lines that is waits for are valid. Copy this code to video, with a bit of work this can be refactored, and perhaps even teach dri how to handle rotated front buffers. References: Bug 28964 - [i965gm] GPU infinite MI_WAIT_FOR_EVENT while watching video in Totem https://bugs.freedesktop.org/show_bug.cgi?id=28964 Signed-off-by: Chris Wilson commit 75850e824bd66b961b9ce21696d9090169f7b28f Author: Evan McClain Date: Thu Feb 25 01:51:42 2010 -0500 Add mbp_backlight support. Acked-by: Julien Cristau commit a2aa4c23f65b03c389c2ba8eac0298dfb205e67d Author: Dave Airlie Date: Mon Jul 5 14:02:42 2010 +1000 uxa: oops typo in previous commit commit feff2ec80eeeba5074d19efcddf4867adf639b4f Author: Dave Airlie Date: Mon Jul 5 09:07:08 2010 +1000 uxa: don't compare planemask with FB_ALLONES. planemask is an unsigned long initialised to ~0, on 64-bit this is not equal to an (unsigned int)-1. Use the macro provided to do this. Signed-off-by: Dave Airlie commit 7ad9c0bebf2c71db574e0302264859ce53cecd37 Author: Jesse Barnes Date: Fri Jul 2 12:39:27 2010 -0700 add cscope files to gitignore commit afcd41820d7481912b49093670e9d7a4a6a2338f Author: Chris Wilson Date: Thu Jul 1 22:14:03 2010 +0100 Reduce front buffer stride prior to rejection If we reject the front buffer because it has too large a stride, repeat the allocation using untiled for the cases where we can utilize laxer hardware restrictions. Signed-off-by: Chris Wilson commit f8778b66a98d68e622cdf5b2d1fd959c4f7a2b2b Author: Chris Wilson Date: Thu Jul 1 21:58:43 2010 +0100 drmmode: Add missing newlines at the end of log messages. Signed-off-by: Chris Wilson commit 690fbd1a64c1e1fd07f7f8a5e39e05a33e4d0721 Author: Chris Wilson Date: Thu Jul 1 21:55:04 2010 +0100 drmmode: Use a copy of the converted mode on resize Avoid a potential use-after-free of the copied mode string by reusing the converted kernel mode on resize. ==19897== Invalid read of size 8 ==19897== at 0x661C330: ??? (strcpy.S:1308) ==19897== by 0x8618AE7: drmmode_set_mode_major (drmmode_display.c:293) ==19897== by 0x8618E6F: drmmode_xf86crtc_resize (drmmode_display.c:1299) ==19897== by 0x529A77: xf86RandR12ScreenSetSize (xf86RandR12.c:708) ==19897== by 0x4BD528: ProcRRSetScreenSize (rrscreen.c:301) ==19897== by 0x42B820: Dispatch (dispatch.c:432) ==19897== by 0x4254C9: main (main.c:289) ==19897== Address 0x72e91e0 is 0 bytes inside a block of size 9 free'd ==19897== at 0x4C23DBC: free (vg_replace_malloc.c:325) ==19897== by 0x48424F: xf86DeleteMode (xf86Mode.c:1921) ==19897== by 0x4942B7: xf86ProbeOutputModes (xf86Crtc.c:1572) ==19897== by 0x5290BB: xf86RandR12GetInfo12 (xf86RandR12.c:1551) ==19897== by 0x5313AE: RRGetInfo (rrinfo.c:202) ==19897== by 0x4BCCAA: rrGetScreenResources (rrscreen.c:337) ==19897== by 0x42B820: Dispatch (dispatch.c:432) ==19897== by 0x4254C9: main (main.c:289) Signed-off-by: Chris Wilson commit 772f8236d50725f0b330508616b4f2a9a910662a Author: Chris Wilson Date: Wed Jun 30 13:58:05 2010 +0100 dri: Handle errors during GetBuffers() gracefully. Unwind the array of Pixmaps already allocated and report failure for the old dri GetBuffers() path. Signed-off-by: Chris Wilson commit 17884af4eda01c05e7429ed6a195f3e4ceeff892 Author: Chris Wilson Date: Mon Jun 28 17:44:45 2010 +0100 Repair the damage to 'make distcheck' after splitting out i810 Signed-off-by: Chris Wilson commit 28c0ca676c47e7e38fabdd9ef24a70bd26701f33 Author: Chris Wilson Date: Fri Jun 25 13:40:22 2010 +0100 Remove unused inclusion of Signed-off-by: Chris Wilson commit 2306e675c48bc00a8ca1f4545bb88c97d6ef4fa6 Author: Chris Wilson Date: Fri Jun 25 13:32:02 2010 +0100 configure: Remove check for unused function, mprotect Signed-off-by: Chris Wilson commit 6695296ca28da7c4f551b0e80a8d06b557d59a5e Author: Chris Wilson Date: Fri Jun 25 13:31:11 2010 +0100 configure: Remove unused checks for xext Signed-off-by: Chris Wilson commit 5c663ce8442bd3b69c96e29061c57cc565fca91d Author: Chris Wilson Date: Sun Jun 20 15:01:16 2010 +0100 Rename common infrastructure to the intel namespace. After splitting out the i810 driver into its own legacy directory, we can identify the common routines not as i830 but as intel. This clarifies the code which *is* i830 specific. Signed-off-by: Chris Wilson commit 797d173a9af12055ba2609293182b616dd673ef4 Author: Chris Wilson Date: Sun Jun 20 12:52:10 2010 +0100 i810: Move into a legacy directory. The driver is still built but is no longer under active development so move it and supporting files to a new directory. Signed-off-by: Chris Wilson commit cd61531a78b9a2cbdd6019199c43258f83e6e2ce Author: Chris Wilson Date: Sun Jun 20 21:03:44 2010 +0100 Remove unused configure option: --enable-video-debug Signed-off-by: Chris Wilson commit ca16e3133658504b68a6b5c83bb8712899016544 Author: Carl Worth Date: Thu Jun 24 14:07:15 2010 -0700 Update version to 2.12.0 For the imminent 2.12.0 release. commit 8caf1564a75d65636d9b7a4da03219bb302bb37b Author: Carl Worth Date: Thu Jun 24 14:06:43 2010 -0700 NEWS: Add notes for the 2.12.0 release. Let it ship! commit b58a6a39c1568800938eb0e3ebc7664683b61200 Author: Chris Wilson Date: Wed Jun 23 23:56:44 2010 +0100 uxa: Fallback to pixman if source is out-of-bounds If the source is outside the drawable, then CopyArea will fail to initialise the source correctly. The simplest fix in this case is to fallback to pixman to generate the source texture. Fixes: Bug 28497 - Graphics corruption after opening a specific website https://bugs.freedesktop.org/show_bug.cgi?id=28497 Signed-off-by: Chris Wilson commit 6d33e578de4e23336ac69cc3c5d0935a65d4dda1 Author: Chris Wilson Date: Wed Jun 23 21:28:50 2010 +0100 Limit maximum tiled stride to 8k and untiled to 32k. Tiling on gen 2/3 hardware is only supported for pitches up to 8192 bytes, so above this limit the surface will be untiled and we will no longer have to comply with the power-of-two pitch alignment. So disabling tiling for these too wide surface should ~halve the memory requirement for the full surface. Also the absolute limit for the 2D blitter is 32,768 bytes. The documentation says "up to 32,768 bytes" and my PineView box was malfunction with a surface stride of 32,768 so set the limit to be 32,767. References: Bug 28497 - Graphics corruption after opening a specific website https://bugs.freedesktop.org/show_bug.cgi?id=28497 Signed-off-by: Chris Wilson commit 5bf470bd38b1f6a7a540585186a54c9dbbca98f9 Author: Chris Wilson Date: Mon Jun 21 22:28:58 2010 +0100 i965: Compile fix. Oops, I spent more time discussing these flushing bugs than I spent paying attention to what I was actually doing. Signed-off-by: Chris Wilson commit 0203cf91b54ee8a7cea8560e559288ee9b6e8554 Author: Chris Wilson Date: Mon Jun 21 22:25:08 2010 +0100 Do not clear need_mi_flush within the batch. This is a situation that should not be possible, need_mi_flush being true but the list of pending flush pixmaps being clear. However, an earlier bug in doing just that revealed this minor bug. So for correctness, be careful not to clear need_mi_flush without emitting a MI_FLUSH. Signed-off-by: Chris Wilson commit 5107b6fa26ecfdbdd60b869a86765c9c484db3a2 Author: Chris Wilson Date: Mon Jun 21 22:21:58 2010 +0100 i965: Mark the render target as dirty within composite_setup() The key difference between i965 and earlier, is that the surfaces passed to the samplers through an indirect table and so the batch and render target was not being marked dirty by the relocation (since the relocation only happens within prepare_composite() which may have been in another batch.) Simply call intel_pixmap_mark_dirty() when binding the sampler table into the batch to ensure that the dirty is tracked appropriately. Signed-off-by: Chris Wilson commit bebd64d82121d2b61893e1d0adebda03f4faea61 Author: Chris Wilson Date: Mon Jun 21 22:05:19 2010 +0100 Also submit any pending flush for *this* batch in the BlockHander. We still need to submit an additional flush if we have further writes since the last flush. Signed-off-by: Chris Wilson commit c4d200517755e6b0f48d6c6a6ba8dd0ddc883d8b Author: Chris Wilson Date: Mon Jun 21 21:57:21 2010 +0100 Only append the pixmap to the flushing list if we are writing to it. Signed-off-by: Chris Wilson commit c942585098ac84ae461821cbb8f52dedce7a0da1 Author: Chris Wilson Date: Mon Jun 21 21:45:04 2010 +0100 Emit the flush after a potential draw from the BlockHandler. As the batch submit may not trigger further drawing through flushing the vertices, pass the requirement to emit the flush down to the submission routine so that the flush can be appended after the final commands. Signed-off-by: Chris Wilson commit e8783869ad55d337601b6f6a51c02f6576c64f38 Author: Chris Wilson Date: Sun Jun 20 00:09:20 2010 +0100 uxa: Apply the source offsets to the pixmap source, not target. A slight confusion in computing the correction image location resulted in the application of the source offsets to the pixel location in the target and not in the source as intended. Fixes the visual corruption of the scrollbar in Chromium, and hopefully the crash reported by Robert Hooker when starting gdm after plymouth. Signed-off-by: Chris Wilson commit 4b7142baa0b3bf6f38843d06aadc579d8624cefc Author: Chris Wilson Date: Sat Jun 19 12:30:23 2010 +0100 uxa: Enable SHM pixmaps Now with streaming uploads and downloads for composite operations in place, shared memory pixmaps are no longer that dire performance wise. With careful use these can in fact be the most efficient means of transfer between a wholly software renderer in the client and a backing store. For instance, Chromium renders internally to an ARGB32 image buffer and uses a shared pixmap to composite dirty regions into the backing store. Thereby using the GPU to either perform the blit or the format conversion. Enabling shared pixmaps, reduces our CPU overhead whilst scrolling by a factor of 5 or so. And this is achieved simply by deleting obsolete code! Signed-off-by: Chris Wilson commit d748f8e6fcebdf5b5b00c90b6c3d519560ee2928 Author: Chris Wilson Date: Thu Jun 17 18:29:55 2010 +0100 uxa: Use accelerated get_image for copying to !offscreen Signed-off-by: Chris Wilson commit 78ee25f005fac9e9cc85425aa3fca8c396d015ea Author: Chris Wilson Date: Thu Jun 17 18:37:32 2010 +0100 uxa: Match depth 30 to format. Signed-off-by: Chris Wilson commit af5c4fc96db56a14bf6347774d8e8f23b38c5967 Author: Chris Wilson Date: Sat Jun 19 13:36:20 2010 +0100 uxa: Check for allocation failure. Check for the NULL Picture prior to passing it to the backends for inspection. Signed-off-by: Chris Wilson commit be55066c6481b4c5e2cd39ef1c0f3be88cae0c93 Author: Chris Wilson Date: Thu Jun 17 14:23:21 2010 +0100 i830: Remove domain tracking from pixmaps. The 4 integers can be reduced to a single boolean value, so do so. Signed-off-by: Chris Wilson commit c187da9a24e98cbdf86f0f0dfbdcd9025c8a4c76 Author: Chris Wilson Date: Wed Jun 16 13:16:28 2010 +0100 i830: GetImage acceleration. The presumption is that we wish to keep the target hot, so copy to a new bo and move that to the CPU in preference to causing ping-pong of the original. Also the gpu is much faster at detiling. Before (PineView): 400000 trep @ 0.1128 msec ( 8860.0/sec): GetImage 10x10 square 18000 trep @ 1.3839 msec ( 723.0/sec): GetImage 100x100 square 800 trep @ 30.0987 msec ( 33.2/sec): GetImage 500x500 square After: (PineView) 180000 trep @ 0.1478 msec ( 6770.0/sec): GetImage 10x10 square 60000 trep @ 0.4545 msec ( 2200.0/sec): GetImage 100x100 square 4000 trep @ 8.0739 msec ( 124.0/sec): GetImage 500x500 square Signed-off-by: Chris Wilson commit 0e0101758438debf98f989d815989b45e78cf5f6 Author: Chris Wilson Date: Tue Jun 15 12:48:57 2010 +0100 i830: Tidy i830_uxa_put_image() Use a single code path to upload the image data after selecting the right bo, and take advantage of pwrite() when possible. Fixes: Bug 28569 - [i965] IGN's flash-based video player crashes X https://bugs.freedesktop.org/show_bug.cgi?id=28569 Bug 28573 - [i965] Fullscreen flash and windowed SDL games fail to update the screen https://bugs.freedesktop.org/show_bug.cgi?id=28573 Reported-and-tested-by: Brian Rogers Signed-off-by: Chris Wilson commit 2ff7a2fc9d939554502e9adec89cae5597a9e82c Author: Chris Wilson Date: Thu Jun 17 13:56:30 2010 +0100 i915: Force the emission of BUF_INFO on every composite_setup We should be able to eliminate these as the drawable remains unchanged. However, the implicit flush of BUF_INFO fixes the rendering in KDE. Alternatively, we need an MI_FLUSH | INHIBIT_RENDER_CACHE_FLUSH between composites. (Note that it is not stale cache data causing the rendering corruption and that a pipelined flush is not sufficient either.) Also, having tried varies points at which to flush, the only place where the flush is effective seems to be between composite operations - that is a flush after 2D is not sufficient. Reported-by: Vasily Khoruzhick Reported-by: Clemens Eisserer Signed-off-by: Chris Wilson commit a25573d5c47ebea34c076075e1993233d7db2b4f Author: Chris Wilson Date: Tue Jun 15 20:26:19 2010 +0100 drmmode: Use the tiled stride for the rotated pixmap. After d41684d5459 we now allocate all framebuffers as tiled bo, and so we must be careful to use the appropriate stride as returned from the allocation, instead of assuming that it is just an aligned width. Fixes: Bug 28461 - screen rotation results in corrupted output. https://bugs.freedesktop.org/show_bug.cgi?id=28461 Signed-off-by: Chris Wilson Reported-by: Till Matthiesen commit 23ddcf45344936da2931a3bf25bd187a9ea28344 Merge: 254bcc4 ffc6897 Author: Julien Cristau Date: Tue Jun 15 17:46:52 2010 +0100 Merge commit '2.11.901' commit 254bcc40fca6ed9aa0b71f2115c654667f16aec6 Author: Chris Wilson Date: Tue Jun 15 10:24:40 2010 +0100 configure: Bump drm requirement to 2.4.21 We depend upon the recent introduction of drm_intel_bo_is_reusable(). Signed-off-by: Chris Wilson Reported-by: Julien Cristau commit ffc689767fa56f2c7710e6f396441c3d8a24ea18 Author: Carl Worth Date: Mon Jun 14 17:35:20 2010 -0700 Update version to 2.11.901 For the 2.12 RC1 snapshot we're making now. commit 68c8b3f12482489f2e2e8040fd6728f1a2b9e420 Author: Carl Worth Date: Mon Jun 14 17:34:38 2010 -0700 NEWS: Add notes for 2.11.901 This is the 2.12 RC1 snapshot we're making now. commit 8899cd03e2e7ee673e8b3b2a9d2873839cdc1bf8 Author: Carl Worth Date: Thu Jun 10 15:38:54 2010 -0700 Fix to require libdrm 2.4.19 or newer. This is necessary for the drm_intel_bo_emit_reloc_fence function which did not exist in 2.4.16 which is all we were requiring earlier. commit 995a4b2b1d09c31672d9258a8ac732dcf9a8fe9f Author: Chris Wilson Date: Mon Jun 14 12:15:34 2010 +0100 i965: Sanity check ComponentAlpha status in prepare_composite Fixes: Bug 28446 - Garbled Font with Mathematica 7 https://bugs.freedesktop.org/show_bug.cgi?id=28446 Rewriting the glyphs to render to the destination directly and removing the more expensive multiple invocations of CompositePicture per picture was a great performance boost -- except that it needs special handling in the backend in order to not fallback. Having done so for i915, I neglected to ensure the sanity checking in i965_prepare_composite() was sufficient. As it turns out, it was not and so we misrendered CA-glyphs when rendering directly to the destination. This causes us to fallback properly, but is a performance regression as we no longer try the 2-pass magic helper before resorting to s/w. At the moment, I'd rather live with the temporary regression and fix i965 to do the same magic as i915, as it critical to fixing the severe performance issues currently crippling i965, as I believe that this regression only affects the minority of applications (incorrect, as it turns out, as the glyphs are overlapping) rendering directly to the destination. Signed-off-by: Chris Wilson commit 84d65bace521ad80b9bede4ae9e3e2a8d0668b90 Author: Chris Wilson Date: Mon Jun 14 11:31:40 2010 +0100 Compile fix for alternate list.h from xserver-1.9 Signed-off-by: Chris Wilson commit 00a54eee8f29ced610ef9d65b077e64bc65655e9 Author: Gaetan Nadon Date: Sat Jun 12 20:24:10 2010 -0400 COPYING: replace stub file with actual Copyright notices. Signed-off-by: Gaetan Nadon commit 94217ed5f5a3726db3fad6cac2132d99ed45c154 Author: Chris Wilson Date: Sat Jun 12 17:21:44 2010 +0100 uxa: Always clip glyphs to destination. Even if there is only a single clip rect, since the clip may be smaller than the drawing rectangle on the destination we need to actually compute the clipped glyph rectangle. Signed-off-by: Chris Wilson commit 5a0a8a1cf6d9b0616d6a097e783f2aa318b45736 Author: Chris Wilson Date: Thu Jun 10 22:56:34 2010 +0100 i830: Limit disabling acceleration following EIO to !i965 Following a conversation with Owain G. Ainsworth, it was decided that the second best approach to handling a wedged GPU was to hope that the kernel could successfully reset it, which currently is only possible for i965 and later chipsets. The best approach is of course to prevent such hangs from ever occurring in the first place. Signed-off-by: Chris Wilson commit 3bf4ca2cdc2493e209bbb9c597bfa17ed7fcf5dc Author: Chris Wilson Date: Wed Jun 9 12:02:21 2010 +0100 i830: Only emit the disabling GPU error message once. But emit the warning about rendering corruption every time for the transient errors like out-of-memory. Signed-off-by: Chris Wilson commit 35a12f029005d8f432755fb78f10c9d48ea2f347 Author: Chris Wilson Date: Wed Jun 9 11:47:37 2010 +0100 Fallback implementation for trapezoids for hung GPUs. Signed-off-by: Chris Wilson commit 8c1a8d22978f143044cc3cd0ff2ade5736ee9d48 Author: Chris Wilson Date: Wed Jun 9 10:03:08 2010 +0100 Revert "xp:trapezoids" This reverts commit f429fb9d872950705e11171d0e7407fb7673c786. An experimental patch I forgot was on my main branch as I was bugfixing. ARGH! commit 994aa1ef571ac2ab05f6588f9068d33b536c7a37 Author: Chris Wilson Date: Wed Jun 9 09:59:36 2010 +0100 uxa: Handle all-clipped out case with destination glyphs. Fixes the crash reported in: Bug 28446 - Garbled Font with Mathematica 7 https://bugs.freedesktop.org/show_bug.cgi?id=28446 pDst=0x3d663c0, src_x=0, src_y=0, xDst=142, yDst=112, nlist=0, list=0x7fffea026580, glyphs=0x7fffea025d88, extents=0x0) at uxa-glyphs.c:809 dx = 0 y1 = 101 x2 = 150 x1 = 142 dy = 0 y2 = 112 rects = 0x5491000 this_atlas = 0x2456d00 mask_y = 128 glyph = 0x35933a0 mask_x = 736 priv = 0x39309e0 screen = 0x8d2cc0 uxa_screen = 0x2443eb0 src_pixmap = 0x37c29e0 dst_pixmap = 0x45ddbf0 localSrc = 0x361a450 glyph_atlas = 0x2456d00 x = 142 y = 112 n = 18 nrect = -9975128 box = {x1 = 23152, y1 = -5630, x2 = 32767, y2 = 0} __PRETTY_FUNCTION__ = "uxa_glyphs_to_dst" Though the meat of that bug regarding the incorrect remains unsolved. Signed-off-by: Chris Wilson commit f429fb9d872950705e11171d0e7407fb7673c786 Author: Chris Wilson Date: Tue Jun 8 10:24:51 2010 +0100 xp:trapezoids commit 0776a42b70f2de7b7d7d8804046c79442da1cb8a Author: Chris Wilson Date: Tue Jun 8 10:43:24 2010 +0100 implicit-flush commit d41684d54592cf93554a4d6534e7ea74562b1798 Author: Eric Anholt Date: Mon Jun 7 11:18:09 2010 -0700 Allocate rotate shadow buffers using the usual framebuffer allocator. This means we can get tiling on them, which should significantly boost performance, and also allow for FBC. commit b5c9de10ba3a811f312e622b97ee5e601bd3b74f Author: Eric Anholt Date: Sun Jun 6 23:41:23 2010 -0700 Allocate a correctly sized framebuffer when tiling by using libdrm's support. When I made libdrm stop overallocating so much memory for the purpose of bo caching, things started scribbling on the bottom of my frontbuffer (and vice versa, leading to GPU hangs). We had the usual mistake of size = tiled_pitch * height instead of size = tiled_pitch * tile_aligned_height. commit e6acbc763229fd5b5b2cc1d65136404d02ac4655 Author: Chris Wilson Date: Mon Jun 7 18:04:51 2010 +0100 uxa: Setup acceleration functions prior to the damage layer We need to install the acceleration functions so that they are wrapped by the Damage layer. This fixes the corruption under a compositing WM introduced in commit 8700673157fdd3a87ad5150f2f30823261fec519. Signed-off-by: Chris Wilson Reported-and-tested-by: Arkadiusz Miśkiewicz commit 1788b16eb26580624dd3df96ed84f68d40e78ed7 Author: Chris Wilson Date: Mon Jun 7 15:31:47 2010 +0100 i915: Fix typo from previous commit. A trivial change, I thought, having tested it before rebasing, unworthy even of a perfunctory compile test. How wrong I was. Signed-off-by: Chris Wilson commit d9bc36ae03e3c56ab165d24fe4bdc6688fd62b6d Author: Chris Wilson Date: Mon Jun 7 15:28:18 2010 +0100 i915: Remove screen size limit from video setup. The i915 textured video routine know how to handle drawing on an output larger than the 3D pipe, so allow them to do so. Signed-off-by: Chris Wilson commit 6555ef5fd16c9e4fe733caf9e831f29f97462278 Author: Chris Wilson Date: Mon Jun 7 13:16:34 2010 +0100 i915: Replace structure passing with macros for shader generation. gcc is horribly bad at collapsing the constants: text data bss dec hex filename 282336 8720 256 291312 471f0 intel_drv.so.old 269280 8720 256 278256 43ef0 intel_drv.so Signed-off-by: Chris Wilson commit d56ea7a852d7090360fe080acec268de55ee908d Author: Chris Wilson Date: Mon Jun 7 00:20:35 2010 +0100 Use the direct dixGevPrivate() API when available This is quicker and smaller than the old indirect function call to dixLookupPrivate(). Signed-off-by: Chris Wilson commit 8700673157fdd3a87ad5150f2f30823261fec519 Author: Chris Wilson Date: Mon Jun 7 00:16:18 2010 +0100 Adapt glyphs for changes in devPrivates API Signed-off-by: Chris Wilson commit 42ddc39430a10513c49a415ddf1a0dc5fa52089e Author: Keith Packard Date: Fri Apr 30 13:33:54 2010 -0700 Adapt to DevPrivate API changes This allows the driver to be built against either the old or new DevPrivate API. Signed-off-by: Keith Packard commit 2c1fda08e889cad07acb452230da06f9c383d21c Author: Eric Anholt Date: Fri Jun 4 16:04:37 2010 -0700 Use libc instead of deprecated libc wrappers for malloc/calloc/free. commit b586624d4f2908d2a998ba87fe0ae31c10f46b91 Author: Chris Wilson Date: Sun Jun 6 17:11:40 2010 +0100 uxa: Force fallback for copies. All but uxa_copy_window() perform the preliminary checks for whether acceleration is available. The simplest method for adding the fallback for uxa_copy_window() seems to be to add it in the core copy function, so be it. This allows X to survive a little longer once we encounter a GPU hang. Signed-off-by: Chris Wilson commit 6db1e5231b7a0e79611f771d4efea686f7849e04 Author: Chris Wilson Date: Wed Jun 2 20:43:49 2010 +0100 dri: Protect against NULL dereference following GPU hang. References: Bug 28361 - "glresize" causes server segfault with single buffering. https://bugs.freedesktop.org/show_bug.cgi?id=28361 [ 14528.767] (EE) intel(0): Failed to submit batch buffer, expect rendering corruption or even a frozen display: Input/output error. [ 14528.767] (EE) intel(0): Disabling acceleration. [ 14528.788] Backtrace: [ 14528.858] 0: /usr/bin/X (xorg_backtrace+0x28) [0x491818] [ 14528.858] 1: /usr/bin/X (0x400000+0x65ca9) [0x465ca9] [ 14528.858] 2: /lib/libpthread.so.0 (0x7f9df2dc9000+0xedf0) [0x7f9df2dd7df0] [ 14528.858] 3: /usr/local/lib/libdrm_intel.so.1 (drm_intel_bo_flink+0x0) [0x7f9defd60c60] [ 14528.858] 4: /usr/local/lib/xorg/modules/drivers/intel_drv.so (0x7f9deff6a000+0x2fdfd) [0x7f9deff99dfd] [ 14528.858] 5: /usr/lib/xorg/modules/extensions/libdri2.so (0x7f9df01b8000+0x19e7) [0x7f9df01b99e7] [ 14528.858] 6: /usr/lib/xorg/modules/extensions/libdri2.so (0x7f9df01b8000+0x1fdb) [0x7f9df01b9fdb] [ 14528.858] 7: /usr/lib/xorg/modules/extensions/libdri2.so (DRI2GetBuffersWithFormat+0x10) [0x7f9df01ba250] [ 14528.858] 8: /usr/lib/xorg/modules/extensions/libdri2.so (0x7f9df01b8000+0x3834) [0x7f9df01bb834] [ 14528.858] 9: /usr/bin/X (0x400000+0x2fc2c) [0x42fc2c] [ 14528.858] 10: /usr/bin/X (0x400000+0x24da5) [0x424da5] [ 14528.858] 11: /lib/libc.so.6 (__libc_start_main+0xe6) [0x7f9df1d60a26] [ 14528.858] 12: /usr/bin/X (0x400000+0x24959) [0x424959] [ 14528.858] Segmentation fault at address 0x20 [ 14528.858] Fatal server error: [ 14528.858] Caught signal 11 (Segmentation fault). Server aborting Signed-off-by: Chris Wilson commit 2989f51caf3134460c2551de597e7e54fe74ee92 Author: Chris Wilson Date: Tue Jun 1 23:08:02 2010 +0100 i830: Remove unused coord-adjust. Signed-off-by: Chris Wilson commit dc402334f4e9b0de624bc89cd77eae4ec7cf1708 Author: Chris Wilson Date: Tue Jun 1 23:03:19 2010 +0100 i915: Centre sampling. Use centre sampling of textures to match pixman, and remove numerous off-by-one and visual artefacts when rendering. The classic example for this is cairo/text/xcomposite-projection where the edge of the rotated rectangle is jaggy due to the incorrect sample position. Fixes: Bug 16917 - [i915] Blur on y-axis also when only x-axis is scaled billiear https://bugs.freedesktop.org/show_bug.cgi?id=16917 And about 15 tests from the Cairo test suite. Signed-off-by: Chris Wilson commit f74b3f82bab11463a0f4de9f614fc6aa1492ef24 Author: Chris Wilson Date: Tue Jun 1 22:31:35 2010 +0100 i915; Avoid the implicit flush on changing BUF_INFO 3DSTATE_BUF_INFO is an implicit flush of the piepline, so avoid emitting that and associated state unless the destination pixmap has actually changed. This is a win of around 3-5% for cairo-perf-trace, notably for firefox. Signed-off-by: Chris Wilson commit f2272402035574c206a0e3383c55373c440fd928 Author: Jesse Barnes Date: Tue Jun 1 13:46:15 2010 -0700 DRI2: fix new buffer exchange check Chris's new buffer exchange check is a good one, but we don't want to hit the immediate blit fallback path if it fails. We still want to schedule a blit for sometime in the future, and we need to use it wherever an exchange might occur (like the secondary flip check or the currently disabled CanExchange check). Fixes https://bugs.freedesktop.org/show_bug.cgi?id=28252. Signed-off-by: Jesse Barnes commit a386a003e7b0ff7100434dccd51fedd822454a6c Author: Chris Wilson Date: Mon May 31 22:18:43 2010 +0100 uxa: Spans, try again to get the early break correct. Trigger happy bug fixing. The sign *was* right, the endpoint was wrong. Signed-off-by: Chris Wilson commit 1672ee0421a99b745d65d72a3818de01d8100b6b Author: Chris Wilson Date: Mon May 31 22:08:43 2010 +0100 uxa: Sign reversal on early break from spans passing the YXband Introduced with e5c971e7639095d38da3518a5dc404b708d45cfb. Signed-off-by: Chris Wilson commit cd38b705be84901a38380dc2a1b90889b553d565 Author: Chris Wilson Date: Mon May 31 17:23:36 2010 +0100 Disable acceleration if we detect a hardware error. This is wildly optimistic, but it should work in a surprising number of error situations and some output in those cases will be hopefully be better than none... If we submit a batchbuffer and the kernel reports the GPU is hung (which will be caused by an earlier execbuffer, and so the kernel should have had enough time to determine whether or not it could reset the GPU) then disable any further attempt to accelerate gfx and force fallbacks to map the buffers and use the CPU. We cannot normally map any more buffers if the GPU is hung, so only those already mapped prior to the hang can be written to, or those allocated in system memory. However, we can expect that the framebuffer is already mapped, and so have a reasonable expectation to continue to see the display update. Signed-off-by: Chris Wilson commit 5fff430046db2030f89e49beb66c6476dc3fcd4c Author: Chris Wilson Date: Sat May 29 17:35:00 2010 +0100 uxa: Mega-Glyphs! Rewrite glyph rendering to avoid the intermediate buffer, accumulating the glyph rectangles directly in the backend composite routines. And modify the glyph cache routines to fully utilise the allocated size of the tiled buffer on older hardware. To do this we alias all glyph sizes into the same texture using a technique suggested by Keith Packard. PineView: 885/856-> 1150/1110 kglyph/s (aa/rgb) Signed-off-by: Chris Wilson commit d31abccd41c417338aac7c681e8bc6bd187b1843 Author: Chris Wilson Date: Mon May 31 12:14:23 2010 +0100 i915: Support textured video on an extended desktop. Handle rendering textured video onto an extended desktop (>2048) by using a temporary pixmap. Note that we still cannot handle rendering to a greater than 2048 destination region, for that we will need to tile. Hmm, time to request a 2560x1600, 10bpc monitor... Signed-off-by: Chris Wilson commit 2cfd5bc134f0dd86ea714594d61f6d5eb29019ce Author: Chris Wilson Date: Sat May 29 17:55:19 2010 +0100 dri: Compilation fix. 17:53 < arekm> ickle: i830_dri.c:630:28: error: ‘DrawableRec’ has no member named ‘bpp’ 17:53 < arekm> ickle: i830_dri.c:630:57: error: ‘DrawableRec’ has no member named ‘bpp’ * sigh. I need to fix this machine to have the right version of the * headers. Signed-off-by: Chris Wilson commit e2615cdeef078dbd2e834b68c437f098a92b941d Author: Chris Wilson Date: Sat May 29 16:37:12 2010 +0100 dri: Only flip if the front and back pixmaps match. An unredirected window (thanks Michel for the reminder) is backed by the Screen pixmap, and so uses a reference of that as its front buffer. The back buffer is a pixmap appropriately sized for the drawable. When the application requests to swap its buffers, obviously we cannot simply exchange the front and back buffer as they do not match, but need to copy the appropriate region from the back to the front. Signed-off-by: Chris Wilson commit 8b2039187fc23f5e9dddf4081f8b1ab6f4c01416 Author: Chris Wilson Date: Sat May 29 12:14:55 2010 +0100 Revert "dri: Use size from backing pixmap when creating buffers." This reverts commit 44d45d3fa56f121ce89ffe5b28beb48be01a95df. Michel Dänzer pointed out the flaw in using the pixmap size instead of the drawable size: Using the backing pixmap dimensions for this is not desirable. In particular, it means that the DRI2 buffers of non-redirected windows always have the same size as the screen. But even for redirected windows it wastes some graphics memory with a re-parenting window manager, that is if it doesn't break in various ways due to the top left corner of the DRI2 buffers no longer corresponding to the top left corner of the window. commit 44d45d3fa56f121ce89ffe5b28beb48be01a95df Author: Chris Wilson Date: Sat May 29 10:39:28 2010 +0100 dri: Use size from backing pixmap when creating buffers. This avoid using the garbage values stored in the Screen drawable, instead of the true values which are only maintained in its backing pixmap. The consequence of using the wrong size was to hand a 1x1 pixmap to metacity/mutter and have it believe it was a full screen drawable; GPU hangs ensued if using page flipping. Signed-off-by: Chris Wilson commit 90c74a43149ec72bef64fb7e64b812294bda69a5 Author: Chris Wilson Date: Fri May 28 21:49:38 2010 +0100 i915: Don't re-emit vertex size unless it has changed. Signed-off-by: Chris Wilson commit a94ae175d6a69243a4465de0be0a8a3368e0ab28 Author: Eric Anholt Date: Fri May 28 12:40:02 2010 -0700 uxa: Fix prepare_solid being called without check_solid first. Fixes GPU hang on gen6. commit 66c90158e45e890c6f655195c6707d216c7ac4c5 Author: Chris Wilson Date: Fri May 28 17:13:30 2010 +0100 uxa: Skip the redundant miComputeCompositeRects() when adding to the mask As we are in full control of the destination (the temporary glyph mask) and the source (the glyph cache) we know that there are no clip regions on either and so can skip computing the composite rectangles. (We trust the device clipping to prevent compositing outside the target.) x11perf on PineView: 701/686 -> 881/856 kglyphs/s [aa/rgb] Signed-off-by: Chris Wilson commit 5b2254838eb89bf77eeb893c73eecb8c737822f4 Author: Chris Wilson Date: Fri May 28 12:47:26 2010 +0100 uxa: Make the glyph caches' fixed size explicit. Until we actual resize the glyph cache dynamically, make it obvious to the reader and the compiler that the size is fixed. Signed-off-by: Chris Wilson commit 11581dda99cb2e4ae78fc73be4b02185b3be58ed Author: Chris Wilson Date: Fri May 28 11:38:05 2010 +0100 uxa: Use a glyph private rather than a hash table. Store the cache position directly on the glyph using a devPrivate rather than an through auxiliary hash table. x11perf on PineView: 650/638 kglyphs/s -> 701/686 kglyphs/s [aa/rgb] Signed-off-by: Chris Wilson commit 73111cf2a212ee5cc2e03af1c600867df0c55b39 Author: Chris Wilson Date: Wed May 26 21:05:27 2010 +0100 Decouple non-reusuable pixmaps from batch lists on unref. ==7596== Invalid write of size 4 ==7596== at 0x491ACA8: intel_batch_teardown (i830_batchbuffer.c:118) ==7596== by 0x491C9D6: I830CloseScreen (i830_driver.c:1419) ==7596== by 0x8103A9C: RRCloseScreen (randr.c:105) ==7596== by 0x80DE794: xf86CrtcCloseScreen (xf86Crtc.c:759) ==7596== by 0x80BEBA3: DGACloseScreen (xf86DGA.c:268) ==7596== by 0x80D044B: DPMSClose (xf86DPMS.c:134) ==7596== by 0x488B050: XvCloseScreen (xvmain.c:320) ==7596== by 0x81841B1: VidModeClose (xf86VidMode.c:110) ==7596== by 0x80EB12F: CursorCloseScreen (cursor.c:191) ==7596== by 0x810CA17: AnimCurCloseScreen (animcur.c:108) ==7596== by 0x816937E: compCloseScreen (compinit.c:86) ==7596== by 0x48D39B9: glxCloseScreen (glxscreens.c:221) ==7596== Address 0x49c1a50 is 24 bytes inside a block of size 52 free'd ==7596== at 0x4024866: free (vg_replace_malloc.c:325) ==7596== by 0x80B023C: Xfree (utils.c:1096) ==7596== by 0x4927CFD: i830_set_pixmap_bo (i830_uxa.c:647) ==7596== by 0x491C9B4: I830CloseScreen (i830_driver.c:1413) ==7596== by 0x8103A9C: RRCloseScreen (randr.c:105) ==7596== by 0x80DE794: xf86CrtcCloseScreen (xf86Crtc.c:759) ==7596== by 0x80BEBA3: DGACloseScreen (xf86DGA.c:268) ==7596== by 0x80D044B: DPMSClose (xf86DPMS.c:134) ==7596== by 0x488B050: XvCloseScreen (xvmain.c:320) ==7596== by 0x81841B1: VidModeClose (xf86VidMode.c:110) ==7596== by 0x80EB12F: CursorCloseScreen (cursor.c:191) ==7596== by 0x810CA17: AnimCurCloseScreen (animcur.c:108) Signed-off-by: Chris Wilson commit a6fb6aa5f9c1c81b8b81b9d1138a526ffbbd5ad3 Author: Chris Wilson Date: Wed May 26 20:31:45 2010 +0100 Add vertex bo to the list of buffers to be torn down. Signed-off-by: Chris Wilson commit 5dce69002dd48dd6dce4f0e7ad9ecf2c4c17cfe4 Author: Chris Wilson Date: Wed May 26 20:26:42 2010 +0100 i965: Remove ATOMIC_BATCH. This paranoid check is deceased; pining for the fjords. Signed-off-by: Chris Wilson commit 06ebb55d308defbf8e01e30e4f994c2dd1071c40 Author: Eric Anholt Date: Sun May 23 20:33:47 2010 -0700 Add a workaround for Ironlake errata relating to disabling the clipper. commit 158a158dad26b6888dd2ebaf13e6020894475fcb Author: Eric Anholt Date: Sun May 23 20:28:24 2010 -0700 Add a workaround for Ironlake errata regarding blits and other engines. commit 3461f8f4bc2a49fd41c408d225c592623404be91 Author: Eric Anholt Date: Sun Apr 18 19:14:13 2010 -0700 Remove remaining REG_DUMPER build stuff. commit 03bbb4c896ef3cd275312b413a2c85d9f499c032 Author: Chris Wilson Date: Wed May 26 10:16:36 2010 +0100 uxa: Perform manual damage for CompositeRects [xserver-1.8] The damage layer doesn't wrap CompositeRects, so we need to manually append the damaged region ourselves. This works for miCompsiteRects since that translates the call into multiple invocations of either PolyFillRectangle or Composite, which themselves cause damage. Fixes: Bug 28120 - Tint2's tooltip borders end up at 0,0 and do not disappear https://bugs.freedesktop.org/show_bug.cgi?id=28120 Signed-off-by: Chris Wilson commit b9ada52a3081f54e0ed094ac31188bb240866f81 Author: Chris Wilson Date: Wed May 26 10:14:52 2010 +0100 uxa: Force the alpha value to 0xffff when treating Over as Src Since we have at most 8 bits of alpha, we treat >= 0xff00 as opaque. However, being paranoid we should set the alpha value to 0xfff in case something unexpected happens when converting from the xRenderColor to the pixel value. Signed-off-by: Chris Wilson commit 3055d40164590147d35b5e7059ebe5f5858c85fa Author: Chris Wilson Date: Wed May 26 10:12:57 2010 +0100 uxa: Use Composite rather than solid blitter for PolyRect Due to the relocation overhead, using a single composite with many rectangles outperforms many solid blits. Signed-off-by: Chris Wilson commit ec2437f958ec4f5ac5222b37cba4cd403b5c8855 Author: Chris Wilson Date: Wed May 26 10:12:18 2010 +0100 uxa: Add PICT format mapping for depth 4 pixmaps. Signed-off-by: Chris Wilson commit 309bd3a29943ef7502d40c67be49c19a59b6d1a2 Author: Chris Wilson Date: Wed May 26 10:11:32 2010 +0100 i830: Skip an empty fill. In the extremely unlikely event that the higher layer erroneous gave us an empty fill, skip it. Signed-off-by: Chris Wilson commit b645ec83e0d86f2247b8338ceab60b9502516e70 Author: Chris Wilson Date: Tue May 25 09:49:20 2010 +0100 uxa: Apply the drawable offset to the solid rects Fixes: Bug 28120 - Tint2's tooltip borders end up at 0,0 and do not disappear https://bugs.freedesktop.org/show_bug.cgi?id=28120 Signed-off-by: Chris Wilson commit 9d8ac271404ff219c3d40ae8f829e8b76ac7c359 Merge: 509df27 ea07535 Author: Chris Wilson Date: Fri May 21 11:01:30 2010 +0100 Merge branch 'glyphs' Tweak glyphs to improve x11perf on i915 by about 33%. PineView, aa10text: 460 -> 617 kglyphs/s. PineView, rgb10text: 434 -> 610 kglyphs/s. Speedups ======== xcb poppler 18.636 -> 13.958: 1.34x speedup xlib firefox-talos-gfx 71.905 -> 56.232: 1.28x speedup xcb firefox-talos-gfx 72.882 -> 57.969: 1.26x speedup xlib gnome-terminal-vim 38.126 -> 34.472: 1.11x speedup xcb gnome-terminal-vim 35.164 -> 32.573: 1.08x speedup xlib poppler 19.634 -> 18.246: 1.08x speedup Note the lack of significant improvement for firefox-planet-gnome. commit ea07535240dafc4c6ef55b4b7a2eeaa595febe86 Author: Chris Wilson Date: Fri May 21 14:33:18 2010 +0100 i915: Emit CA over using OutReverse + Add passes On PineView: 578/621 -> 610/617 kglyphs/sec [rgb/aa] commit 80a9e64f50aeda6004e3aba1fbfdda50bb1f1c82 Author: Chris Wilson Date: Fri May 21 09:55:55 2010 +0100 uxa: Use temporary dest when target is too large for compositor If the destination cannot fit into the 3D pipeline when we need to composite, we fallback to doing the operation on the CPU. This is very slow, and quite easy to trigger on i915 by plugging in an external display. An alternative is to extract the extents of the operation from the destination using the blitter which can usually handle much larger operations. This gives us a temporary target that can fit into the 3D pipeline and thus be accelerated, before copying back into the larger real destination. For x11perf this boosts glyph rendering on PineView, from 38kglyphs/s to 480kglyphs/s. Just a little shy of the native performance of 601kglyphs/s Signed-off-by: Chris Wilson commit 91f560034fc2695680d1208a78fc56d814b0da79 Author: Chris Wilson Date: Thu May 20 10:26:59 2010 +0100 uxa: Composite glyphs directly onto dst when possible. Without using a mask and compositing directly onto the destination, takes us from 580 kglyphs/s to 850 kglyphs/s on i945 [x11perf -aa10text]. However, the extra intersection check almost entirely cancels out the speed up and we discover that the glyphs in x11perf are always overlapping. Nothing is ever easy. Signed-off-by: Chris Wilson commit e3ece83f577d3664962edeec6ab5bdc41c5d77cf Author: Chris Wilson Date: Wed May 19 16:52:11 2010 +0100 i915: compute normalized texcoords using a scale factor. 500 -> 580kglyphs/s on i945. Signed-off-by: Chris Wilson commit c2abf8d659b8b161a4f9df100b614ee3c8f8e458 Author: Chris Wilson Date: Wed May 19 16:18:40 2010 +0100 uxa: translate the region in line for composites When compositing, we need to convert the box into a rect and so the advantages of using REGION_TRANSLATE are lost. Signed-off-by: Chris Wilson commit 2adf823b80b7b1f6df1bf3422a1219e93321a8fb Author: Chris Wilson Date: Wed May 19 15:02:58 2010 +0100 i915: Add special case primitive emitters for glyphs. Signed-off-by: Chris Wilson commit f64ab9e0d97dd9c654b4ae1924e62ef6813d9bb0 Author: Chris Wilson Date: Wed May 19 14:50:12 2010 +0100 i915: Move vertices into a vertex buffer object. In theory this should allow us to pack far more operations into a single batch buffer, and reduce our overheads. Signed-off-by: Chris Wilson commit 2b050f330f78d02e7f476e55be29d760271ac61c Author: Chris Wilson Date: Wed May 19 10:57:46 2010 +0100 Use pwrite to upload the batch buffer By using pwrite() instead of dri_bo_map() we can write to the batch buffer through the GTT and not be forced to map it back into the CPU domain and out again, eliminating a double clflush. Measing x11perf text performance on PineView: Before: 16000000 trep @ 0.0020 msec (511000.0/sec): Char in 80-char aa line (Charter 10) 16000000 trep @ 0.0021 msec (480000.0/sec): Char in 80-char rgb line (Charter 10) After: 16000000 trep @ 0.0019 msec (532000.0/sec): Char in 80-char aa line (Charter 10) 16000000 trep @ 0.0020 msec (496000.0/sec): Char in 80-char rgb line (Charter 10) Signed-off-by: Chris Wilson commit dcef703a7cdcf360f12312a338361697acffc3e9 Author: Chris Wilson Date: Wed May 19 09:30:02 2010 +0100 Kill paranoid assertions on every write into the batchbuffer. On my PineView box these represent ~5% overhead on x11perf text: Before: 16000000 trep @ 0.0020 msec (495000.0/sec): Char in 80-char aa line (Charter 10) 12000000 trep @ 0.0022 msec (461000.0/sec): Char in 80-char rgb line (Charter 10) After: 16000000 trep @ 0.0020 msec (511000.0/sec): Char in 80-char aa line (Charter 10) 16000000 trep @ 0.0021 msec (480000.0/sec): Char in 80-char rgb line (Charter 10) Signed-off-by: Chris Wilson commit bc41f84e01f18548b05c670e1fd0d641adc28d0f Author: Chris Wilson Date: Tue May 18 23:54:13 2010 +0100 i915: Emit composite primitive with specialised functions. Signed-off-by: Chris Wilson commit 4a3476ea094e84887fefb558e0bba023fee34151 Author: Chris Wilson Date: Tue May 18 23:25:59 2010 +0100 i915: amalgamate composite into a single primitive list Combine all the calls to composite between prepare_composite and done_composite into a single primitive list, rather than a primitive call per composite(). Signed-off-by: Chris Wilson commit e5c971e7639095d38da3518a5dc404b708d45cfb Author: Chris Wilson Date: Tue May 18 22:16:17 2010 +0100 uxa: Spans! OMG! Use composite rather than solid blits in order to bring performance on a par with the CPU when using GEM and relocations. Signed-off-by: Chris Wilson commit 509df27c7401e96d3062890da73f6af6629adef4 Author: Kristian Høgsberg Date: Tue May 18 10:01:52 2010 -0400 dri: Clean up DRI2 API #ifdefs a bit Signed-off-by: Kristian Høgsberg commit 5e04a81369dfc76bbc1ebbd4db00c23e71eb9e1c Author: Chris Wilson Date: Mon May 17 15:16:25 2010 +0100 i830: Remove vestigal debugging ALWAYS_FLUSH and ALWAYS_SYNC These are now debugging options exposed in Xorg.conf, and now unused int the source code. Signed-off-by: Chris Wilson commit 723cc45b2707aa160eac6d5b56a20e19cc42c8a3 Author: Chris Wilson Date: Mon May 17 15:14:55 2010 +0100 dri: Check error code from GetScratchGC() It may fail so be prepared, and do use the right drawable! Signed-off-by: Chris Wilson commit 2c00297bc3b4969f28e2eb19bce36ece50ce4ccb Author: Chris Wilson Date: Sun May 16 19:15:37 2010 +0100 uxa: Replace solid planemask [0xffffffff] with FB_ALLONES Signed-off-by: Chris Wilson commit 2c69709d8afa6e9c0990efc463df0061536585e1 Author: Chris Wilson Date: Sun May 16 18:41:52 2010 +0100 i830: Encode surface bpp into format References: Bug 28135 - [855GM] Slowdown/High CPU-Usage after Git-Commit 926fbc7d90ac1d0d49d154f136f9c9ed613c98c2 https://bugs.freedesktop.org/show_bug.cgi?id=28135 The simple answer is that I had assumed that 0 was a reserved value. However, without the bbp encoded into the format 0 was used for a8r8g8b8 and r5g6b5, which are very common formats! The other possibility for the slowdown is that gtkperf is using of the now verboten xrgb formats -- but would in fact be valid if the source covers the clip and we could fixup the alpha value in the fixed function combine. Signed-off-by: Chris Wilson commit 21b5fd427f5ed6c99276349a7ae128eea580dd72 Author: Chris Wilson Date: Sun May 16 13:51:41 2010 +0100 uxa: Tidy uxa_solid_rects() Move the operator reduction after a few fallbacks, closer to its use. Signed-off-by: Chris Wilson commit 61835701fd411d3bb550ceee3365e30639e46861 Author: Chris Wilson Date: Sun May 16 13:26:07 2010 +0100 uxa: Patterns are acquired at 0,0 Set the correct offset for the gradients patterns after rendering to a local Picture. Fixes cairo/test/huge-radial and friends Signed-off-by: Chris Wilson commit 89f43f69a9c854fb2b82538a8938002675ca19e4 Author: Chris Wilson Date: Sat May 15 18:34:54 2010 +0100 uxa: Force an alpha channel when rendering source fallbacks As the source may not cover the extents, we need to represent those areas as transparent in the fallback picture, ergo we need an alpha channel. We could be smarter and force a format conversion when necessary, and we could let the backend choose the most appropriate format. Signed-off-by: Chris Wilson commit 524fd2dd0da8f69d4c3f5aebe3a345fe34af609d Author: Chris Wilson Date: Sat May 15 18:23:21 2010 +0100 uxa: Apply clip for solid rectangles. References: Bug 28120 - Tint2's tooltip borders end up at 0,0 and do not disappear https://bugs.freedesktop.org/show_bug.cgi?id=28120 Signed-off-by: Chris Wilson commit 58b089febceca1e915607bb723ee658aaa9dbed3 Author: Chris Wilson Date: Sat May 15 09:11:46 2010 +0100 uxa: Avoid using blits when with PictFilterConvolution References: Bug 28098 Compiz renders shadows wrong, garbage line of pixels along left and top edge of windows https://bugs.freedesktop.org/show_bug.cgi?id=28098 Signed-off-by: Chris Wilson commit ef95899f5b21453daeabf81a3015b22456d21fec Author: Chris Wilson Date: Sat May 15 09:02:07 2010 +0100 uxa: Check the w-scaling component is 1 for an translation matrix Signed-off-by: Chris Wilson commit 9c3da71349bcfeabae08f1572cf602c357bf7641 Author: Chris Wilson Date: Sat May 15 01:05:24 2010 +0100 i830: Remove xrgb conversion to argb, no longer required. All textures are now properly declared so that the alpha swizzling occurs in the sampler or not at all. The downside is that for quite a few composite operations we have to fallback to software on older hardware. There is scope for more performing the alpha expansion in shaders or combiners when we know the picture covers the clip - which is almost all of the time for normal operations especially those constructed by Cairo. Signed-off-by: Chris Wilson commit 926fbc7d90ac1d0d49d154f136f9c9ed613c98c2 Author: Chris Wilson Date: Sat May 15 01:01:20 2010 +0100 i830: Remove incorrectly mapped tex formats. We no longer workaround the lack of alpha expansion for xrgb textures as this interferes with EXTEND_NONE, though we could if we know the source covers the clip... Signed-off-by: Chris Wilson commit 95654cffa8643f61f153218f27e800df75b9bdee Author: Chris Wilson Date: Sat May 15 00:50:42 2010 +0100 uxa: Fix order of conditionals to only run fill_region for SRC or opaque Signed-off-by: Chris Wilson commit f67b45965b527699794ee21174809c5d71c03f22 Author: Chris Wilson Date: Sat May 15 00:50:20 2010 +0100 uxa: Expand the range of compatible formats to cover all bpp. Signed-off-by: Chris Wilson commit 82d07fdf10cac2211af74ebf7d519daacd7084c0 Author: Chris Wilson Date: Sat May 15 00:49:39 2010 +0100 uxa: Only use 1x1R as a solid with an opaque format or SRC Signed-off-by: Chris Wilson commit 3bca186a7ead84d4f0e71dfd83847ef77a4e390b Author: Chris Wilson Date: Sat May 15 00:48:31 2010 +0100 uxa: Call check_solid before running the solid blitter. Signed-off-by: Chris Wilson commit 213816c30b5ca6146e26ded85b2aa31ca16ca9bf Author: Chris Wilson Date: Sat May 15 00:47:55 2010 +0100 i915: Load texture into directly into OC when possible. Signed-off-by: Chris Wilson commit 737de9a779d05fe6fb1efe1904fda2cc452ff137 Author: Chris Wilson Date: Fri May 14 23:48:27 2010 +0100 uxa: Disable compatible src xrgb and dst argb I'm seeing garbage alpha for rendercheck blend: x8r8g8b8a 10x10 SRC ar8g8b8a so disable blitting until I work out if we can fast-path it. Signed-off-by: Chris Wilson commit 271240fd477b11885947da32868bc77151681559 Author: Chris Wilson Date: Fri May 14 23:38:02 2010 +0100 i915: Remove a couple of unsupported 16bpp no-alpha tex formats Signed-off-by: Chris Wilson commit a7c318d21c9035f6cab48c16e734b9ecb6f4b238 Author: Chris Wilson Date: Fri May 14 23:30:21 2010 +0100 uxa: Parse BGRA pixel formats. Signed-off-by: Chris Wilson commit f7bbcc492a05f0801c64ee884798cf1a7ebb71c2 Author: Chris Wilson Date: Fri May 14 23:29:13 2010 +0100 Split the prepare blitter functions into check + prepare. Allow us to check whether we can handle the operation using the blitter prior to doing any work. Signed-off-by: Chris Wilson commit 4be8d7eb89e61ffb2ceb19f1f84260e581187692 Author: Chris Wilson Date: Fri May 14 21:13:21 2010 +0100 i915: Don't force alpha=1 for RGB drawables in the shader. I was blindly fixing rendercheck without thinking. We need to force the alpha value to be in the blend unit and not before -- otherwise we generate the incorrect result whilst blending. D'oh. commit b9a5e36f957e42eaf0387ad15f2298d3e9538b05 Author: Chris Wilson Date: Fri May 14 20:52:48 2010 +0100 uxa: enable solid rects for backends that require pixmaps Convert the color into a (cached) pixmap if the backend cannot handle the SolidFill natively. Signed-off-by: Chris Wilson commit a21297d7cc5945704d8cbd28ad0a2a3f6096fa47 Author: Chris Wilson Date: Fri May 14 17:51:42 2010 +0100 drm: Remove pin(); unpin() sync GEM handles serialisation of the new front buffer with respect to page flipping and rendering and reports back when the flip is complete. Adding a sync point here is then redundant. Signed-off-by: Chris Wilson commit 7ee73d2c6f373f3d6829cb1876b1940db62be452 Author: Chris Wilson Date: Fri May 14 17:51:20 2010 +0100 drm: Remove unused old_front parameter from drmmode_do_pageflip. Signed-off-by: Chris Wilson commit 030d56279bf14d9ddd42d8fdbeaa66ef3f557b4d Author: Chris Wilson Date: Fri May 14 16:53:40 2010 +0100 drm: don't overwrite the old intel->front_buffer It's now handled in the common ExchangeBuffers() path. Signed-off-by: Chris Wilson commit 5bd022739511104e4978bdfd2f4a1f60b9206723 Author: Chris Wilson Date: Fri May 14 15:47:09 2010 +0100 i830: Teardown batch entries on reset. By not cleaning up the batch entries when resetting the X server, we left the pointers in an inconsistent state and caused X to crash. commit 0d2392d44aae95d6b571d98f7ec323cf672a687f Author: Chris Wilson Date: Fri May 14 10:32:12 2010 +0100 dri: Hold reference to buffers across swap As we schedule swaps for some time in the future and may process a detachment prior to receiving the vblank notification from the kernel, we need to hold a reference to the buffers for our swap event handler. Fixes: Bug 28080 - "glresize" causes X server segfault with indirect rendering. https://bugs.freedesktop.org/show_bug.cgi?id=28080 Signed-off-by: Chris Wilson commit 8de09a0707ee1be1b919b979843711728618ef27 Author: Chris Wilson Date: Thu May 13 17:15:28 2010 +0100 uxa: Convert 1x1R back to solid_fill In the change to prevent blitting between incompatible sources, we also prevented 1x1R pixmaps from being used for solid fills. Reorder the sequence of conditions to enable this fast path again. commit 92e9cf8af784b13c28030e38b15c8decf29e6a32 Author: Chris Wilson Date: Thu May 13 15:12:51 2010 +0100 uxa: Only use solid_fill for SRC. commit d1bd14e8b6754ba9f797642e1b33bf689e19417b Author: Chris Wilson Date: Thu May 13 15:11:16 2010 +0100 uxa: Replace source for CLEAR with a transparent solid This means that we will hit the faster try_solid_fill path instead. commit cdab72c405434ecbf7a79e402ff2d65d6a728179 Author: Chris Wilson Date: Thu May 13 15:12:32 2010 +0100 uxa: Fallback early if compositing with alphaMaps commit 25811dc7b7d1ad3fb01c31197d1ae1fe5b498975 Author: Chris Wilson Date: Thu May 13 15:09:28 2010 +0100 i915: Force output alpha to 1. if dst has no alpha channel. Ensure that garbage is not stored in the unused alpha channel so that we can rely on it being currently initialiased when used as a source or returning via GetImage. Partial fix for rendercheck -t blend commit 0e726b85ca6013ae9dc51391aaa309203352b61e Author: Chris Wilson Date: Thu May 13 09:41:39 2010 +0100 i915: Add a2r10g10b10 format and friends Signed-off-by: Chris Wilson commit 9f54107f866a25cf670f81f7c52b8c108728c6a5 Author: Chris Wilson Date: Tue May 11 14:55:16 2010 +0100 dri2: Handle reference counting across page flipping 1. Instead of swapping bos, swap the entire private structure. 2. If we update the pixmap bo for the Screen, make sure we update the reference inside intel->front_buffer so that xrandr still functions. Fixes: Bug 27922 - i965: Rapidly resizing OpenGL window causes GPU to hang. https://bugs.freedesktop.org/show_bug.cgi?id=27922 Signed-off-by: Chris Wilson commit 6c27f6e4f76b97df71094acf25083b2922966b42 Author: Chris Wilson Date: Wed May 12 12:01:46 2010 +0100 uxa: Avoid glyph ping-pong with !offscreen destination Signed-off-by: Chris Wilson commit d5383c2073e181b9ec352461c05d6202ad37f9d6 Author: Chris Wilson Date: Wed May 12 10:53:14 2010 +0100 uxa: Avoid ping-pong with !offscreen destination and traps If we are destined to target an !offscreen drawable, then uploading the trapezoid mask to a bo is the last thing we actually want to do... Signed-off-by: Chris Wilson commit 00664b8f9d3da8d0d6aa53471ed3a8a8f6391660 Author: Chris Wilson Date: Wed May 12 12:41:51 2010 +0100 uxa: Fallback when compositing to a !offscreen destination Signed-off-by: Chris Wilson commit 0c6372a77fd8c051534c1b420a02a8737aa1dd01 Author: Chris Wilson Date: Wed May 12 09:57:27 2010 +0100 i830: Prevent allocation of bo larger than half the aperture We need to prevent overcommitting the aperture, and in particular if we allocate a buffer larger than available space we will fail to mmap it in and rendering will fail. Trying to allocate multiple large buffers in the aperture, often the case when falling back, causes thrashes and eviction of useful buffers. So from the outset simply do not allocate a bo if the the required size is more than half the available aperture space. Fixes allocation failure in ocitymap.trace for instance. Signed-off-by: Chris Wilson commit 244b7cbfffdcda4761948eaa37ed2a30ca81f107 Author: Chris Wilson Date: Wed May 12 09:18:30 2010 +0100 uxa: Use accelerated PutImage for uploading pixman images. Short-circuits the current use of PutImage from CopyArea, bypassing all the temporary allocations. commit cb887cfc670bf63993bd313ff33927afb8198eae Author: Chris Wilson Date: Fri Mar 26 09:59:51 2010 +0000 uxa: solid rects The cost of performing relocations outweigh the advantages of using the blitter for solids with lots of rectangles. References: Bug 22127 - [UXA] 50% performance regression for XRenderFillRectangles https://bugs.freedesktop.org/show_bug.cgi?id=22127 By using the 3D pipeline we improve our performance by around 4x on i945, measured by the jxbench microbenchmark, and a factor of 10x by short-cutting to the 3D pipeline for blended rectangles. Before, on a i945GME: 19982.412060 Ops/s; rects (!); 15x15 9599.131693 Ops/s; rects (!); 75x75 3803.654743 Ops/s; rects (!); 250x250 6836.743772 Ops/s; rects blended; 15x15 1443.750000 Ops/s; rects blended; 75x75 495.335821 Ops/s; rects blended; 250x250 23247.933884 Ops/s; rects composition (!); 15x15 10993.073048 Ops/s; rects composition (!); 75x75 3595.905172 Ops/s; rects composition (!); 250x250 After: 87271.145975 Ops/s; rects (!); 15x15 32347.744361 Ops/s; rects (!); 75x75 5884.177215 Ops/s; rects (!); 250x250 73500.000000 Ops/s; rects blended; 15x15 33580.882353 Ops/s; rects blended; 75x75 5858.811749 Ops/s; rects blended; 250x250 25582.317073 Ops/s; rects composition (!); 15x15 6664.728682 Ops/s; rects composition (!); 75x75 14965.909091 Ops/s; rects composition (!); 250x250 [suspicious] This has no impact on Cairo, but I have a suspicion from watching xtrace that Qt likes to blit thousands of 1x1 rectangles with the same colour. However, we are still around 2-3x slower than the reported figures for EXA! Signed-off-by: Chris Wilson commit c8e10f7791fecf0fed7ad606db062d87d8ff263c Author: Chris Wilson Date: Wed May 12 12:41:00 2010 +0100 debug: Add names for operators Most useful for confirming my worst fears: unwarranted use of OutReverse + Add. Signed-off-by: Chris Wilson commit 6ea8ce640fbc7e5054b521a6f70b303945d0e849 Author: Chris Wilson Date: Tue May 11 19:35:59 2010 +0100 xvmc: Build fix with -pedantic Fixes: Bug 27352 - RPMLINT error causes build breakage https://bugs.freedesktop.org/show_bug.cgi?id=27352 Reported-by: Johannes Obermayr Signed-off-by: Chris Wilson commit e1b7e8bf1d475b3d00adcb96e6002c20c9e70d1c Author: Chris Wilson Date: Tue May 11 15:50:23 2010 +0100 drmmode: Reorder i830_set_pixmap_bo() so that the correct stride is used. The pitch needs to be set on the pixmap prior to the private intel_pixmap structure being created so that it can record the correct value from the pixmap. Signed-off-by: Chris Wilson commit dfbaf9aab80cf33d47148f087fa747eacd508b05 Author: Chris Wilson Date: Tue May 11 14:54:15 2010 +0100 i830: Never create a bo for depth=1 pixmaps. As we can not accelerate these either as a destination or a source, don't bother allocating a buffer object for them. Signed-off-by: Chris Wilson commit 5b7efe375ae951ffcdb2c81e2ad8a2ed86ffbdf2 Author: Chris Wilson Date: Tue May 11 14:53:29 2010 +0100 i830: Use set_pixmap_bo() instead of open-coding. The advantage is that this enables in-flight reuse of the old pixmap if possible. Signed-off-by: Chris Wilson commit ad8af95dd35921118f45d6cff355c80cf0906ff0 Author: Chris Wilson Date: Tue May 11 14:52:54 2010 +0100 i830: Do not cache in-flight non-reusable buffers. Signed-off-by: Chris Wilson commit f1048e14d5ef34970cb717e2a37d1c6bb4ea3a34 Author: Chris Wilson Date: Tue May 11 11:39:46 2010 +0100 i965: Add texformats mapping for additional pixman formats Signed-off-by: Chris Wilson commit a35afd4a2df8bab543700d874274228eddccae5b Author: Chris Wilson Date: Tue May 11 11:26:33 2010 +0100 uxa: Recheck texture after acquiring pattern. As the first step to handling unsupported texture formats, double check that the converted pattern can be used as a texture by the card. Fixes: rendercheck -t repeat Signed-off-by: Chris Wilson commit d745cab6c45b9fce5e46eacbdd04ceae911fddbb Author: Keith Packard Date: Mon May 10 22:58:49 2010 -0700 Must call ValidateGC in i830_uxa_put_image for scratch GC Always need to call ValidateGC or the scratch GC will not get the right composite clip. Signed-off-by: Keith Packard commit 3eded4202ea00d42edf5edad831484ebc6c0c826 Author: Chris Wilson Date: Mon May 10 23:38:17 2010 +0100 i915: Fix pixmap based masks. Signed-off-by: Chris Wilson commit 1ecd89be039b642affb2f9933f4341d186cb4dea Author: Chris Wilson Date: Mon May 10 23:33:52 2010 +0100 uxa: Protect against valid SourcePict in uxa_acquire_mask() Signed-off-by: Chris Wilson commit a8761585ef0b37736d302299d6b03e0957ad3e16 Author: Chris Wilson Date: Mon May 10 19:38:24 2010 +0100 i830: Minor cleanup Remove some extraneous prototypes and unused variables. Signed-off-by: Chris Wilson commit 9e9b0d85da801f1f95c7bc7e99566fe7056f5813 Author: Chris Wilson Date: Mon May 10 18:37:02 2010 +0100 i830: Update stride when swapping bo for PutImage Signed-off-by: Chris Wilson commit 0d4dd00aeafdd625ffe902e26e666b63a0842adc Author: Chris Wilson Date: Mon May 10 12:10:09 2010 +0100 uxa,i915: Handle SourcePict through uxa_composite() Signed-off-by: Chris Wilson commit 21c1c3c7f6eb2b5070d2153b15a8fb1afe938bbb Author: Chris Wilson Date: Mon May 10 10:19:28 2010 +0100 i915: Use 1x1R pixmap for solid drawables x11perf has a regression https://bugs.freedesktop.org/show_bug.cgi?id=25068 caused by commit e581ceb7381e29ecc1a172597d258824f6a1d2d3 i915: Use the color channels to pass along solid sources and masks. Do not convert 1x1R pixmaps into a solid color as the readback from the bo negates all the performances advantages of using a smaller vertex buffer and fewer samplers. Before (PineView): aa=66800 glyph/s, rgb=28800 glyphs/s Now: aa=96800 glyphs/s, rgb=48500 glyphs/s Signed-off-by: Chris Wilson commit f52b6e832292c02c0010b19882e38e1097beeda0 Author: Chris Wilson Date: Mon May 10 09:39:44 2010 +0100 uxa: Rearrange checking and preparing of composite textures. x11perf regression caused by 2D driver https://bugs.freedesktop.org/show_bug.cgi?id=28047 caused by commit a7b800513fcc94e063dfd68d2f63b6bab7fae47d uxa: Extract sub-region from in-memory buffers. The issue is that as we extract the region prior to checking whether the composite can in fact be accelerated, we perform expensive surplus operations. This is particularly noticeable for ComponentAlpha text, such as rgb10text. The solution here is to rearrange the check_composite() prior to acquiring the sources, and only extracting the subregion if the render path can not actually handle the texture. Performance (on PineView): a7b800513^: aa=68600 glyphs/s, rgb=29900 glyphs/s a7b800513: aa=65700 glyphs/s, rgb=13200 glyphs/s now: aa=66800 glyph/s, rgb=28800 glyphs/s The residual lossage seems to be from the extra function call and dixPrivate lookups. Hmm. More warning is the extremely low performance, however the results are consistent so the improvement looks real... Signed-off-by: Chris Wilson commit 848ab66384508c3ad3e5fb4884e4527f3ebd3bde Author: Chris Wilson Date: Sat May 8 18:36:55 2010 +0100 uxa: Transform composites with a simple translation into a blit We can also convert a composite with an integer translation into a blit, so long as the sample extents remains within the source. Signed-off-by: Chris Wilson commit a7b800513fcc94e063dfd68d2f63b6bab7fae47d Author: Chris Wilson Date: Wed Apr 14 21:14:34 2010 +0100 uxa: Extract sub-region from in-memory buffers. If the buffer is too large or not suitable for a GPU operation, we currently fallback and perform the composite on the CPU. An alternative is too extract the small region out of the source (as usually the sample extents are much smaller than the actual surface size) and try the composite with the new surface. The effect is particularly noticeable on pathological websites that use very large background images. For example, http://www.woodtv.com/ uses a 1299x15000 pattern that is obscured by another opaque pattern. Signed-off-by: Chris Wilson commit 8562b7bc6740eef2602af76b8685388efd2d4d37 Author: Chris Wilson Date: Tue Apr 27 10:27:44 2010 +0100 i830: prepare the uxa pixmap for fbCopyArea. Complete the prepare access for the PutImage fallback via fbCopyArea(), by remembering to set the private pointer to the GTT mapping. Signed-off-by: Chris Wilson commit 9a5cd65b593ea82e56e0c403f7ddcc2420a64dc3 Author: Chris Wilson Date: Mon Apr 26 09:07:07 2010 +0100 i830: if pixman_blt() fails fallback to fbCopyArea() On older versions of pixman, pixman_blt() can return false if the images are <= 8bpp. If we are being called from CopyArea, then we cannot return FALSE here as that will trigger an infinite recursion. Instead we must manually perform the fallback using fbCopyArea(). Reported-by: Peter Clifton Signed-off-by: Chris Wilson commit 86d349aa7b0a596fbb9530e896c19349405bf3ba Author: Chris Wilson Date: Mon Apr 26 09:12:54 2010 +0100 i830: tidy in flight bo reuse. A left-over cleanup patch for c374c94. *sigh* Signed-off-by: Chris Wilson commit 72fd7d191c33c8d0b75b06ab0717d7ca0d4e019a Author: Daniel Vetter Date: Mon Apr 19 21:56:57 2010 +0200 Fix "make dist" This is some fallout from my xvmc cleanup. Original-Patch-by: Rico Tzschichholz Signed-off-by: Daniel Vetter commit 9494f4e91f8c8c7a0f8d61c6883c9bfceb2cec46 Author: Daniel Vetter Date: Fri Apr 16 22:24:01 2010 +0200 i810: adjust the pitch for DRI rendering Current code forgot to adjust the pitch of the frontbuffer. Signed-off-by: Daniel Vetter Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=16729 commit c374c94e41d6e7d677334171e3255778d77cbe18 Author: Chris Wilson Date: Wed Mar 31 11:50:27 2010 +0100 uxa: Reuse in-flight bo When we need to allocate a new bo for use as a gpu target, first check if we can reuse a pixmap that has already been relocated into the aperture as a temporary target, for instance a glyph mask or a clip mask. Before: backend test min(s) median(s) stddev. xlib firefox-planet-gnome 50.568 50.873 0.30% xcb firefox-planet-gnome 49.686 53.003 3.92% xlib evolution 40.115 40.131 0.86% xcb evolution 28.241 28.285 0.18% After: backend test min(s) median(s) stddev. xlib firefox-planet-gnome 47.759 48.233 0.80% xcb firefox-planet-gnome 48.611 48.657 0.87% xlib evolution 38.954 38.991 0.05% xcb evolution 26.561 26.654 0.19% And even more dramatic improvements when using a font size larger than the maximum size of the glyph cache: xcb firefox-36-20090611: 1.79x speedup xlib firefox-36-20090611: 1.74x speedup xcb firefox-36-20090609: 1.62x speedup xlib firefox-36-20090609: 1.59x speedup Signed-off-by: Chris Wilson commit 96aa7a236ac0605324a94f7b7d10413cb219f071 Author: Chris Wilson Date: Wed Apr 14 15:04:53 2010 +0100 i830: Allocate bo's for glyphs larger than 32x32. As we only use the glyph cache for small glyphs, those large than 32x32 will first be copied to a bo and used as a mask in a composite operation. We can avoid the allocation and upload per use by allocating a bo for the over-sized glyph from the start. As the glyph is large anyway, the excess memory allocation is less significant. Using normal font sizes, firefox shows no change - as expected. However, using the 36 font size traces, we see around a 10% improvement on g45. Before: xcb firefox-36-20090609 127.333 127.897 0.22% xcb firefox-36-20090611 87.456 88.624 0.66% xcb firefox-20090601 19.522 20.194 1.69% xlib firefox-36-20090609 201.054 201.780 0.18% xlib firefox-36-20090611 133.468 133.717 0.09% xlib firefox-20090601 23.740 23.975 0.49% With large glyphs in bo: xcb firefox-36-20090609 117.256 118.254 0.42% xcb firefox-36-20090611 79.462 79.962 0.31% xcb firefox-20090601 19.658 20.024 0.92% xlib firefox-36-20090609 185.645 188.202 0.68% xlib firefox-36-20090611 123.592 124.940 0.54% xlib firefox-20090601 23.917 24.098 0.38% Thanks to Owain G. Ainsworth for the suggestion! Signed-off-by: Chris Wilson commit 2d17bd50af367bead84985c22fdd43d264a5f072 Author: Chris Wilson Date: Wed Apr 14 14:54:34 2010 +0100 Revert "Revert "uxa: Try using put_image when copying from a memory buffer."" This reverts commit 6d50553e8f70d8f2142efdfd6c90bc27a599d0bc. Now we have taught the fallback path not to infinitely recurse, re-enable the accelerated path for ShmPutImage and friends. commit 1cc2c2c44ac72460cf1c4e6bdc13c612235809c9 Author: Chris Wilson Date: Wed Apr 14 11:18:25 2010 +0100 i830: Use pixman_blt directly for performing the in-memory copy In order to avoid an infinite recursion after enabling CopyArea to use the put_image acceleration to either stream a blit or to copy in-place, we cannot call CopyArea from put_image for the fallback path. Instead, we can simply call pixman_blt directly, which coincidentally is a tiny bit faster. Signed-off-by: Chris Wilson commit 324a2810da3fbae35637ba9080f31f9383db0868 Author: Daniel Vetter Date: Tue Apr 13 08:39:43 2010 +0200 i830 render: check aperture space requirements No point not doing this. Signed-off-by: Daniel Vetter commit 804263c10df2fedb2f0debab6cdfaa6a89184a42 Author: Daniel Vetter Date: Sun Apr 11 12:56:24 2010 +0200 render: tell the kernel explicitly when fences are needed This slighlty improves xrender performance on fence reg starved i8xx hw. I've also changed a few function calls to the new names from the compat ones while looking at the code. The i915 textured video path is not converted because atm the xv code does not use tiled surfaces. Signed-off-by: Daniel Vetter commit a619a7831228dc52f0fef7d92c92f701e5aeaa94 Author: Daniel Vetter Date: Thu Apr 8 18:49:01 2010 +0200 i915 render: use tiling bits where possible This is in preparation to explicit fence allocation with execbuf2. Signed-off-by: Daniel Vetter commit 55cd36046e61e8d51b5cb754a81cdb54e3eab166 Author: Daniel Vetter Date: Thu Apr 8 14:48:35 2010 +0200 i830 render: use tiling bits where possible This is in preparation to explicit fence allocation with execbuf2. Signed-off-by: Daniel Vetter commit 6d50553e8f70d8f2142efdfd6c90bc27a599d0bc Author: Eric Anholt Date: Mon Apr 12 13:44:31 2010 -0700 Revert "uxa: Try using put_image when copying from a memory buffer." This reverts commit 27195d7dba0f3ff08b92f3fd916cdf5113cbef58. put_image often calls copy_area. Which calls put_image. Exhausting of the stack follows. commit 28024f6c5f351ce4cb3434998d5dbfbeded7c606 Author: Chris Wilson Date: Mon Apr 12 13:44:01 2010 +0100 Revert "uxa: Add fallback warnings for PutImage." This reverts commit 299b0338d0811192dc4f8eae5d79453e9882c5d1. A debugging patch, it was never intended to go into master commit 27195d7dba0f3ff08b92f3fd916cdf5113cbef58 Author: Chris Wilson Date: Wed Mar 31 20:30:27 2010 +0100 uxa: Try using put_image when copying from a memory buffer. Often, for example in the fallback for ShmPutImage, we will attempt to use uxa_copy_area() copying to a normal pixmap from a memory buffer. This triggers a fallback, and maps the destination pixmap back into the GTT. The accelerated put_image path will attempt to stream a blit to the destination pixmap if it is currently active, avoiding the stall. commit 385563417d10b5e8a005bed6ae4de9a8fac1b407 Author: Chris Wilson Date: Tue Mar 30 21:10:50 2010 +0100 Review i830_pad_drawable_width() We appear to have a confusion of stride in terms of pixels, pitch in terms of bytes and the actual width of the surface. i830_pad_drawable_width() appears to be operating aligning *pixels* to a 64 pixel boundary and has never used the chars-per-pixel causing considerable confusion in its callers. Remove the parameter and ensure that the callers are expecting a value in pixels returned, multiplying by cpp where necessary to get the pitch. Signed-off-by: Chris Wilson commit 299b0338d0811192dc4f8eae5d79453e9882c5d1 Author: Chris Wilson Date: Thu Mar 25 13:43:26 2010 +0000 uxa: Add fallback warnings for PutImage. commit 80f52482c7cde000a76b91fe3d8b6c16baf2141f Author: Daniel Vetter Date: Thu Apr 8 22:30:33 2010 +0200 XvMC: fix memory overflow Caught by a malloc library assert. Note to self: Don't just copy&paste codelines around :( Signed-off-by: Daniel Vetter Buzilla: https://bugs.freedesktop.org/show_bug.cgi?id=27540 Tested-by: Nick Bowler Tested-by: Calvin Walton commit 3d7723526b181686d62f09f81fba3ccdbf7d2b1a Author: Daniel Vetter Date: Thu Apr 8 22:21:01 2010 +0200 Xv: fixup overlay stride confusion For some reason I've made a mess out of the overlay stride constrains. Fix it up. Signed-off-by: Daniel Vetter Tested-by: Calvin Walton Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=27453 commit bdf0ee509c26ab35e60f3ed02549efa435be032f Author: Daniel Vetter Date: Thu Apr 8 13:29:04 2010 +0200 i965 Xv: fix chroma pitch In my recent fix for the chroma pitch for i915 xvmc I've forgotten about i965 class hw. For videos with a non-even sized stride (measured in dwords) the chroma pitch was internally incosistent and one dword off. Fix this by using pitch2 for the chroma pitch in i965 textured video like everywhere else. Signed-off-by: Daniel Vetter Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=27417 Tested-by: Nick Bowler Tested-by: Sven Arvidsson commit f6cb28e909c8685474e69e8bd8600f146d3a74e9 Author: Daniel Vetter Date: Thu Mar 11 15:37:54 2010 +0100 libXvMC: rip out debug stuff Almost totatlly unused, but surely totally useless. Signed-off-by: Daniel Vetter commit 676028ec0cbd4fadded92feefc8371e075abcc12 Author: Daniel Vetter Date: Wed Mar 10 19:23:06 2010 +0100 libXvMC: unify CreateSurface Simply store the desired bo size in intel_xvmc_context and initialize it in the driver's create_context function. Signed-off-by: Daniel Vetter commit 3f590a4577f554313e7577205e471757a39129b6 Author: Daniel Vetter Date: Wed Mar 10 19:04:51 2010 +0100 libXvMC i915: kill unused context private fields Signed-off-by: Daniel Vetter commit 8a31dacb067f813a5b0eafc16d265434e48ec5fc Author: Daniel Vetter Date: Wed Mar 10 18:59:29 2010 +0100 libXvMC: derive driver context from struct intel_xvmc_context ... by putting struct intel_xvmc_surface at the beginning. Also kill the common context handling code and simply keep a pointer in the surface private to the context. Signed-off-by: Daniel Vetter commit 16e5edde4d97e1818e0dbfbc165bd8199f697336 Author: Daniel Vetter Date: Wed Mar 10 16:51:10 2010 +0100 libXvMC: unify DestroySurface Surface bo's can be freed by the common code, then there's nothing driver specific left. Signed-off-by: Daniel Vetter commit 5f64122551f2d8d41d745351b2e4e68208aa6084 Author: Daniel Vetter Date: Wed Mar 10 16:25:57 2010 +0100 libXvMC: drop get_surface_status driver callback It's unused. Also drop all related generic code that tries to do clever stuff with this callback. These are all remnants from a pre-gem world. Signed-off-by: Daniel Vetter commit 6c44ce9e127e531224bac825da7918800a3e6a1a Author: Daniel Vetter Date: Wed Mar 10 16:20:03 2010 +0100 libXvMC i915: s/i915XvMCSurface/struct intel_xvmc_surface/ The private surface struct now adds nothing. Drop it. Signed-off-by: Daniel Vetter commit 583ec1f2f7ed32bb42c98baa1ca0fc3735b69b11 Author: Daniel Vetter Date: Wed Mar 10 15:55:44 2010 +0100 libXvMC i915: rip out unneeded surface fields All of these are also stored in the context. Also kill the context reference counting. Doesn't serve a purpose besides occupying a pointer to the context in the private surface struct. Signed-off-by: Daniel Vetter commit c9aaaed829731cda22d057d6ac6424c587770973 Author: Daniel Vetter Date: Wed Mar 10 15:27:02 2010 +0100 libXvMC: unify PutSurface Now that the drm bo is in the common structure, do the PutImage handling in common code, too. Signed-off-by: Daniel Vetter commit 4edff6c972311df2370a8323e346eeabd49ced35 Author: Daniel Vetter Date: Wed Mar 10 15:13:41 2010 +0100 libXvMC: move drm_intel_bo to common intel_xvmc_surface i965 lost the last field in it's private surface struct. Kill it and any associated headers. Signed-off-by: Daniel Vetter commit 9eeabf876d1f70f5525bbac438414557dd2d4335 Author: Daniel Vetter Date: Wed Mar 10 15:00:58 2010 +0100 libXvMC: derive driver surfaces from struct intel_xvmc_surface ... by putting struct intel_xvmc_surface at the beginning. This will allow to consolidate surface and bo handling. Also kill some now dead code used to handle the common surface structure. Signed-off-by: Daniel Vetter commit a0058369bbdfb49f8df5ee1e50671115927d3565 Author: Daniel Vetter Date: Wed Mar 10 14:31:40 2010 +0100 XvMC: kill intel_xvmc_command We only passed around and actually used the gem handle. Don't need a struct for one field alone ... Signed-off-by: Daniel Vetter commit 79fe4caafd21f6ca262901eac4a66ce5a4eaf548 Author: Daniel Vetter Date: Tue Mar 9 16:20:33 2010 +0100 XvMC: dynamically allocate adaptor And kill all the static structures. This way it's clearer what's common and what's specific. And the code is shorter too. Also clean up src/i830_hwmc.c - kill the nonstandard surface types for i915 and the associated code. Signed-off-by: Daniel Vetter commit 00b63ecb192b7eafea9482bf5c5c5d442ce2c6fc Author: Daniel Vetter Date: Tue Mar 9 15:29:15 2010 +0100 XvMC: directly set driver type in context creation Instead of threading it through a global variable. Signed-off-by: Daniel Vetter commit 411cdcb6fcb60085191cf712aba1e0977f145256 Author: Daniel Vetter Date: Tue Mar 9 15:23:02 2010 +0100 XvMC: init driver in one function call No point splitting this up. Gets rid of two #ifdef INTEL_XVMC blocks in i830_video.c. Signed-off-by: Daniel Vetter commit b666e3e29f22894f43a8e9f7f3d1b950d431c42d Author: Daniel Vetter Date: Tue Mar 9 14:56:44 2010 +0100 XvMC: move i965 specific stuff to common context Like for the i915. With this create_context is now also shared. Signed-off-by: Daniel Vetter commit e0068b1e63bc914940dbeb52de46657fcf5da1ea Author: Daniel Vetter Date: Tue Mar 9 14:44:53 2010 +0100 XvMC: move i915 specific stuff to common context Doing the same with the i965 code will allow us to share the create_context function. src/i915_hwmc.h is now almost empty. Move the last #defines to src/xvmv/i915_xvmc.c where they are actually used and delete the file. Also rename the ddx context struct to something sane. Signed-off-by: Daniel Vetter commit 8cab7b84da6d06c20ad9dbbcd8f3f3d971b438e7 Author: Daniel Vetter Date: Tue Mar 9 14:02:23 2010 +0100 XvMC: unify destroy context function Signed-off-by: Daniel Vetter commit 323464b8e2669423df573a18fd1137cb3cad6700 Author: Daniel Vetter Date: Mon Mar 8 23:46:25 2010 +0100 XvMC: clean up i830_hwmc.c - Drop unnecessary macros - Denote i965 specific stuff Signed-off-by: Daniel Vetter commit cfa907adf800e91a3c1f4698ac701b9b83fa600b Author: Daniel Vetter Date: Mon Mar 8 23:41:56 2010 +0100 XvMC: unify surface functions Like for the subpicture stuff, share the "do-nothing" functions ... And fix function name spelling, too. Signed-off-by: Daniel Vetter commit 036cd4bb82131c08b79f596667f00b19a0c8ab79 Author: Daniel Vetter Date: Mon Mar 8 17:33:18 2010 +0100 i965 XvMC: don't create any surface state in the ddx Like for i915. Also drop that now totally superflous limit on the available surfaces. Move the surface struct into the userspace library header now that the ddx doesn't use it anymore. Signed-off-by: Daniel Vetter commit 69218cc1b51acfee91b471bb973a4419b79cdd85 Author: Daniel Vetter Date: Mon Mar 8 17:05:25 2010 +0100 XvMC: unify subpicture functions The XvMC driver api in the server is insane. Even for optional stuff like subpicture support it doesn't check for NULL-pointers. So we have to retain some dummy functions. Wonder how many copies of these things exist on fdo ... Signed-off-by: Daniel Vetter commit 5d104e9c5cd5cd2d34c9035dd38794fa9b5578f3 Author: Daniel Vetter Date: Mon Mar 8 17:03:36 2010 +0100 XvMC: unify PutImage handling Both xvmc are handing in the bo in the exact same way. So move the code to src/i830_video.c and kill this great oeuvre of spaghetti-code. The xvmc driver ini and fini also lost their last use, kill them, too. Signed-off-by: Daniel Vetter commit ebc0ffba79cc4bd564a8bd97327f24a1d15b3d1a Author: Daniel Vetter Date: Mon Mar 8 16:36:24 2010 +0100 XvMC: merge *_hwmc.c into i830_hwmc.c After unifying i915 and i965, not much will be left of these files. Therefore merge them to make the following changes easier. This creates some warnings about some redefined macros, but when this is all cleaned up they'll all be gone. Signed-off-by: Daniel Vetter commit 8199f8f191f8385ef3a228e0ec014a8bbd44002b Author: Daniel Vetter Date: Sun Mar 7 23:17:49 2010 +0100 i915 XvMC: kill more subpict support remnants This things just won't die! Signed-off-by: Daniel Vetter commit 29ba8a84f7cf5c29a5f38688a1ac0ccf41d8e4ec Author: Daniel Vetter Date: Sun Mar 7 23:16:28 2010 +0100 XvMC: everyone's using execbuffer! XvMC was switched over from batchbuffers a few years ago ... Signed-off-by: Daniel Vetter commit 440b4d207b730112169827d0b90b46596659b504 Author: Carl Worth Date: Mon Mar 29 11:20:20 2010 -0700 Increase version to 2.11.0 In preparation for release. commit 1119c4523889ddedecf1722f0e9a2e4a7ad326b2 Author: Carl Worth Date: Mon Mar 29 11:19:31 2010 -0700 NEWS: Add release notes for the 2.11.0 release. Hurrah! We made it. commit df3b26847914557eb3c9c70732d91169b1936d77 Author: Jesse Barnes Date: Fri Mar 26 10:44:55 2010 -0700 DRI2: release our private front buffer ref when buffer swapping Pauli pointed out that we take a ref on the front buffer when exchanging but forget to release it. The ref is necessary since the set functions will drop refs as necessary, but once we set the front buffer to point at the back pixmap, we ned to release our private ref again, or we'll leak buffers. Reported-by: Pauli Nieminen Signed-off-by: Jesse Barnes commit 362a49e71fc41541b6dc121660d98e29da4b14e8 Author: Gaetan Nadon Date: Thu Mar 25 10:07:41 2010 -0400 uxa make: remove unused XORG_INCS and DIX_CFLAGS variables Most likely copied from xserver makefile. Acked-by: Dan Nicholson Signed-off-by: Gaetan Nadon commit 0d1ac4da5288a621d0493cb109fcd1e040e6c5a7 Author: Chris Wilson Date: Thu Mar 25 09:12:09 2010 +0000 uxa: Perform the xrgb -> argb conversion not inplace After reports of segmentation faults caused by d6b7f96fde1add92fd11f5a75869ae6fc688bf77 and vmware, the most obvious cause would be illegally writing to the src data when performing the alpha fill inline. So force the image upload to go via a fresh buffer whenever we need to modify the incoming data. Signed-off-by: Chris Wilson Reported-and-tested-by: Jeff Chua commit 0c47195ca805881e3fbd5b9224be5c930feeeb8c Author: Chris Wilson Date: Wed Mar 24 17:37:39 2010 +0000 i830: Clip solid fills to surface. There is a reasonable surfeit of evidence to support this error, for instance: http://bugs.freedesktop.org/attachment.cgi?id=34417 Signed-off-by: Chris Wilson commit 5537079c29a56133446f1874d24d9e6516825edb Author: Chris Wilson Date: Wed Mar 24 14:59:20 2010 +0000 uxa: After filling the alpha channel xrgb src is compatible with argb dst. Signed-off-by: Chris Wilson commit 90a971c60769781f53827b469a9be3aab14cf71c Author: Chris Wilson Date: Wed Mar 24 14:50:45 2010 +0000 uxa: Only reduce a composite to a BLT if it is wholly contained Signed-off-by: Chris Wilson commit 2eec53d0b9232970fe3d03ce6c8940ebeea44bee Author: Chris Wilson Date: Tue Mar 23 17:28:22 2010 +0000 uxa: Default to using TILING_X for pixmaps. On memory constrained hardware, tiling is vital for good performance as it minimizes cache misses. The downside is that for older hardware (which often suffers from the lack of bandwidth) requires the use of fences for many operations, which are in short supply and so may cause shorter batchbuffers. However our batch buffers are typically short and so this is unlikely to be a concern and not affect the performance wins. A quick bit of testing suggests the effect is inconclusive on firefox/i945: linear tiled xcb 205.470 206.219 xcb-render-0.0 404.704 388.413 xlib 166.410 170.805 A secondary effect of the patch is to workaround a G31 specific hang when attempting to use linear 2048x2048 surfaces. Bonus! Fixes: Bug 25375 - Performance issue using texture from pixmap (tfp) glx extension on 945 http://bugs.freedesktop.org/show_bug.cgi?id=25375 Bug 27100 - GPU Hung copying a 2048x1152 pixmap http://bugs.freedesktop.org/show_bug.cgi?id=27100 Signed-off-by: Chris Wilson Tested-by: John commit 9c037f61a490c96f9095f7ff3fecbf41f5efe9f7 Author: Carl Worth Date: Mon Mar 22 15:23:04 2010 -0700 Update version to 2.10.903 For today's snapshot. commit 11142cffe4bc81e1ce465ea725fcaef024130082 Author: Carl Worth Date: Mon Mar 22 15:22:28 2010 -0700 NEWS: Add notes for 2.10.903 For today's snapshot. commit f206816154fd44e61b0bdeaab1aa4c5359c3b70a Author: Carl Worth Date: Mon Mar 22 14:14:46 2010 -0700 man: Update list of available 'sclaing mode' property values. Include the names from the current kernel driver along with accurate descriptions of each. Indicate how to use the values with: xrandr --output output --set property value and point the user to "xrandr --prop" for an accurate list of currently available values. Closes bug: xf86-video-intel manpage needs update for KMS xrandr properties http://bugs.freedesktop.org/show_bug.cgi?id=25606 commit d1dfab6b1649214177435629d8e85239dbe44d89 Author: Matthias Hopf Date: Wed Mar 17 15:13:51 2010 +0100 Clear drmmode_output->mode_output in drmmode_output_destroy(). commit 10cd04a84bcb6313903fc23b2d7791658ebc6b8e Author: Li Peng Date: Sat Mar 20 00:21:48 2010 +0800 Initialize flip_count before using it Otherwise it would be a random value and drmmode_page_flip_handler() won't have a chance to call I830DRI2FlipEventHandler() and indicate a full page flip is complete. Signed-off-by: Li Peng commit 3d4b3f257fbbb69c6f236d9803abe54a90d7d434 Author: Dave Airlie Date: Thu Mar 18 12:48:39 2010 +1000 intel: free bus id in error path after printing it out. the error message prints out a freed string, spotted during code reappropriation to radeon driver. Signed-off-by: Dave Airlie commit 31d5f84bb4416ef92abd97264d52cdab7a184687 Author: Chris Wilson Date: Wed Mar 17 09:11:05 2010 +0000 i915: Correct preamble for emit_composite Fixes: http://bugs.freedesktop.org/show_bug.cgi?id=27123 Fatal server error: i915_emit_composite_setup: ADVANCE_BATCH: under-used allocation 100/104 Introduced with commit d6b7f96fde1add92fd11f5a75869ae6fc688bf77. Signed-off-by: Chris Wilson commit d6b7f96fde1add92fd11f5a75869ae6fc688bf77 Author: Chris Wilson Date: Sat Mar 6 15:49:04 2010 +0000 Fill alpha on xrgb images. Do not try to fixup the alpha in the ff/shaders as this has the side-effect of overriding the alpha value of the border color, causing images to be padded with black rather than transparent. This can generate large and obnoxious visual artefacts. Fixes: Bug 17933 - x8r8g8b8 doesn't sample alpha=0 outside surface bounds http://bugs.freedesktop.org/show_bug.cgi?id=17933 and many related cairo test suite failures. Signed-off-by: Chris Wilson commit 910fd171a00227025abc8bcc286a740f5bae895b Author: Chris Wilson Date: Fri Mar 5 14:32:13 2010 +0000 i830: Remove coord-adjust for nearest centre-sampling. Fixes a number of cairo test suite failures. Also affects: Bug 16917 - Blur on y-axis also when only x-axis is scaled bilinear http://bugs.freedesktop.org/show_bug.cgi?id=16917 Signed-off-by: Chris Wilson commit 753914acc31947df8e9f2fa3b9c39de745098709 Author: Carl Worth Date: Mon Mar 15 17:32:57 2010 -0700 Bump version to 2.10.902 For the imminent 2.10.902 snapshot release. commit 185196cb010cb8426ca929a9dc3f028ec9b383db Author: Carl Worth Date: Mon Mar 15 17:32:20 2010 -0700 NEWS: Add notes for the 2.10.902 snapshot. Which I'm about to push out. commit 318aa9ed799197810e2039dbe3ec51559dcc888c Author: Daniel Vetter Date: Mon Mar 8 23:22:53 2010 +0100 i915 XvMC: fixup colors My cleanup accidently created a inconsistency in the YUV plane ordering. I think we can safely assume that I'm colorblind ;) As Carl Worth rightly pointed out, this change deserves a more elaborate explanation: For Xv planar formats, the three planes are stored consecutively in memory, ordered Y U V. Now for some totally odd reason (= none at all), i915 xvmc stored it in Y V U order. Right after the release of 2.10, with commit "Xv: consolidate xmvc passthrough handling" I've inadvertently broken xvmc support (which started this whole odyssey into xvmc). When fixing stuff up, I neglected this special plane ordering and simply assumed it to be the same as Xv and dropped that special case for i915 in src/i830_video.c. This patch completes the change to standard YUV plane ordering by making the corresponding change in src/xvmc/i915_xvmc.c. Signed-off-by: Daniel Vetter commit 68629b63740fb6e62c258b8a46669e77660fa4e4 Author: Jesse Barnes Date: Mon Mar 8 15:42:39 2010 -0800 DRI2: make WaitMSC error handling more consistent with ScheduleSwap Just make it mirror ScheduleSwap: complete the wait on any error condition so as not to crash the client if the kernel is misbehaving. Signed-off-by: Jesse Barnes commit 6df74e61afb7831ebf3fbab8782f46ddccbe262b Author: Jesse Barnes Date: Mon Mar 8 15:33:20 2010 -0800 DRI2: truncate OML values to 32 bits We can only handle 32 bit values unless we totally virtualize the count, since the kernel only handles 32 bits itself. Rather than adding all that overhead, just tolerate the occasional missed event everytime the counter runs over. Reported-by: Mario Kleiner Signed-off-by: Jesse Barnes commit c66d57080dc034aa7877f47612065e388bbc38a2 Author: Jesse Barnes Date: Mon Mar 8 15:26:24 2010 -0800 DRI2: more WaitMSC fixes A couple more niggles: make sure we return a target_msc that at least matches the current count; this is a little more friendly to clients that missed an event. Also check for >= when calculating the remainder so we'll catch the *next* vblank event when the calculation is satisfied, rather than the current one as might happen at times. Reported-by: Mario Kleiner Signed-off-by: Jesse Barnes commit 7845c6ade82085488192bd76729d92fb7b534cc0 Author: Jesse Barnes Date: Mon Mar 8 15:19:14 2010 -0800 DRI2: fixup stray curly brace Got left in the wrong column after the last cleanup. Signed-off-by: Jesse Barnes commit 9656d329e4df28cfe138b657d04e5136bc794ce2 Author: Eric Anholt Date: Mon Mar 8 14:34:04 2010 -0800 Put back the pitch alignment for new framebuffers. I confused a dead assignment with dead code, because one of the args to the function was an outvalue. Fixes corruption under compiz. Bug #26814. commit b71ca26a026a356763c51c763bcdd8024fc4a783 Author: Jesse Barnes Date: Mon Mar 8 11:27:42 2010 -0800 DRI2: fixup ScheduleWaitMSC similarly to ScheduleSwap My merge of Mario's patch for this was botched. Fix it up so that OML waits work correctly, and remove a bogus warning from ScheduleSwap. Reported-by: Mario Kleiner Signed-off-by: Jesse Barnes commit 06b54e089eb21736e6d4f6462bbfca987be0c5cc Author: Matt Turner Date: Sun Mar 7 14:27:27 2010 -0500 Don't check for Xinerama. It doesn't seem to be used anywhere, so don't require it. CC: Eric Anholt CC: Jesse Barnes Signed-off-by: Matt Turner commit 8ece6cf5afa1bb0d8d9328696422f42f3c3adbd6 Author: Robert Hooker Date: Sat Mar 6 14:09:12 2010 -0500 Fix build against xserver 1.6 branch. Signed-off-by: Robert Hooker Reviewed-by: Chris Wilson commit 1cd556420277f103c47ade422f3ec8f8efb2d282 Author: Mario Kleiner Date: Fri Mar 5 12:32:18 2010 -0800 DRI2: handle target_msc, divisor and remainder properly in DRI2ScheduleSwap The current code in I830DRI2ScheduleSwap() only schedules the correct vblank events for the case divisor == 0, i.e., the simple glXSwapBuffers() case. In a glXSwapBuffersMscOML() request, divisor can be > 0, which would go wrong. This modified code should handle target_msc, divisor, remainder and the different cases defined in the OML_sync_control extension correctly for the divisor > 0 case. It also tries to make sure that the effective framecount of swap satisfies all constraints, taking the 1 frame delay in pageflipping mode and possible delays in blitting/exchange mode due to DRM_VBLANK_NEXTONMISS into account. The swap_interval logic in the X-Servers DRI2SwapBuffers() call expects the returned swap_target from the DDX to be reasonably accurate, otherwise implementation of swap_interval for the glXSwapBuffers() as defined in the SGI_swap_interval extension may become unreliable. For non-pageflipped mode, the returned swap_target is always correct due to the adjustments done by drmWaitVBlank(), as DRM_VBLANK_NEXTONMISS is set. In pageflipped mode, DRM_VBLANK_NEXTONMISS can't be used without severe impact on performance, so the code in I830DRI2ScheduleSwap() must make manual adjustments to the returned vbl.reply.sequence number. This patch adds the needed adjustments. Signed-off-by: Mario Kleiner commit 13119ffc034a3e9d6c76339d4fedc62bb3b41257 Author: Mario Kleiner Date: Fri Mar 5 11:33:45 2010 -0800 DRI2: make MSC waits handle specific target_mscs and divisor/remainders Previous code only handled divisor == 0 case correctly. This should honor a given target_msc for the divisor > 0 case and handle the (msc % divisor) == remainder constraint correctly. Signed-off-by: Mario Kleiner commit b6e0b92f398823629ba8a1ea8f5e62fbf959e725 Author: Jesse Barnes Date: Thu Mar 4 10:07:26 2010 -0800 DRI2: handle offscreen drawables better at swap time If a drawable isn't visible due to DPMS or redirection, we'll just blit it rather than schedule a swap event. However, we didn't reset the target_msc, so the swap target we receive from the server could get out of sync with the vblank count of the drawable's display. So at DPMS on time, the swap target would be the last good vblank count plus some large number (since the swaps won't have been throttled). Solve this by zeroing out the swap target like we should when we fall back to a blit. Also make the kernel error cases more friendly by making them fall back to blits too. Signed-off-by: Jesse Barnes commit 54ac4e2df987b72529a523ffbde357bec27e3658 Author: Chris Wilson Date: Thu Mar 4 21:34:52 2010 +0000 Rate limit batch buffer error. Once we hit this error it's unlikely that we're coming back - so don't flood the logs with redundant information. Signed-off-by: Chris Wilson commit 066d9b64ee243e0d255d7e12c2134951eef4ade7 Author: Daniel Vetter Date: Thu Mar 4 20:50:20 2010 +0100 i915 XvMC: kill dead code This kills one wip remnant from my i830_memory cleanup and the last remainings of the subpicture support. Signed-off-by: Daniel Vetter commit 7b7c724da97f358f5d2d4d4dae75cfc85bde2c62 Author: Eric Anholt Date: Thu Mar 4 10:29:09 2010 -0800 Remove remaining fbOffset setting. In the long long ago, fbOffset was used for DGA. The server now has only one reference to fbOffset, a leftover setting of it in fbdevhw. We can safely ignore it now, which is good since we weren't updating it in other places where the front buffer offset could change. commit 15026d64d3c0d8ad96e0a73fcae3103d1c41fd46 Author: Eric Anholt Date: Thu Mar 4 10:28:09 2010 -0800 Remove a piece of fbOffset cruft from non-DRM support. commit e8e615289213ba1704e402b923c51ce9d196f06f Author: Eric Anholt Date: Thu Mar 4 10:27:51 2010 -0800 Remove i830_allocate_2d_memory() now that it only called one function. commit e37b562083aa3293e0c009171724a3f122d8a32d Author: Eric Anholt Date: Thu Mar 4 10:23:12 2010 -0800 Init CRTC cursors with CRTC setup instead of i830_memory. commit a36bdaba6136054ae7c67943ca0215cfd177bc5b Author: Eric Anholt Date: Thu Mar 4 09:47:10 2010 -0800 Remove intel_sync() at teardown time. The kernel's still running after we're gone. This didn't matter. commit 4ada6d7a856a941b834871ff8a7c5505ff26ae23 Author: Eric Anholt Date: Thu Mar 4 09:32:42 2010 -0800 Remove 3D state clobber on EnterVT, and always clobber at batch start. We know that it's clobbered at each batchbuffer, anyway. And even if this server isn't running DRI2, it can still be clobbered at batch start in the KMS world. commit d92d42303e5ca9f4208ed97823ad8f691121370f Author: Eric Anholt Date: Thu Mar 4 09:31:56 2010 -0800 Remove pre-2.6.29 error message handling since we require KMS. commit faecd155c49229499e29815eb6e79662ed33ddd5 Author: Eric Anholt Date: Thu Mar 4 09:31:15 2010 -0800 Move batch and 965 render state setup/teardown to screen init/close. Whether we're VT switched or not shouldn't impact rendering. commit 74e2b69a317cc728b00e675c18d2976987407aec Author: Carl Worth Date: Thu Mar 4 09:51:03 2010 -0800 i915_hwmc: Remove dead code. Daniel recently identified this code as unneeded (with an #if 0). Here we take the next step and remove it entirely. commit 1cc35a8ba403ad95b67caff46b803db1edea5ad3 Author: Carl Worth Date: Thu Mar 4 09:46:33 2010 -0800 uxa: Fix type mismatch to avoid compiler warning. The code was using uint32_t where an XID (currently an unsigned long) was specified in the prototype. Use XID to avoid both the warning and any potential problem. commit 1d6537ec57535183638c62e3099bc79786c682ca Author: Carl Worth Date: Thu Mar 4 09:39:28 2010 -0800 i830_video: Remove unused variable. Certainly just a little leftover from the recent rewrites. commit 57c0043b9a519da5c9af29019ef632716782bda2 Author: Daniel Vetter Date: Tue Mar 2 09:22:33 2010 +0100 Xv: fixup relocation in i965_video.c The previous code made no sense, (multiplying an offset by 4 is meaningless). It could have onlt worked with the offset being fortuitously 0. Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit e606be463f6359fd017791a8d4d85059831a5d41 Author: Daniel Vetter Date: Tue Mar 2 21:53:28 2010 +0100 i830_memory: rip out the remainings of the old allocator Yeah! Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit 371be65fb74789250dbb1e332e46416d931da321 Author: Daniel Vetter Date: Tue Mar 2 21:53:27 2010 +0100 XvMC: kill the pinned batchbuffer in the ddx code It's been unused for quite a while. Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit d5a20c81ab76129a1e0bcefecd9c4a10b5040af4 Author: Daniel Vetter Date: Tue Mar 2 21:53:26 2010 +0100 i915 XvMC: kill pinned surface buffer in the ddx code Like with the per context stuff, also drop the now artificial limit on surfaces. Again, with that gone, a lot of code can be deleted. Reviewed-by: Carl Worth commit e6eb257ad32b393fb67c506943d8ece98b9a8061 Author: Daniel Vetter Date: Tue Mar 2 21:53:25 2010 +0100 i915 XvMC: kill pinned per-context buffers in the ddx code There's now not a reason anymore to limit the number of active contexts. So kill this accounting, too. With that all gone, per-context state in the ddx is nil, so rip out all associated code. Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit 47ae1181f6cfd2ab64cd5c8535289fdaa2a9fd9c Author: Daniel Vetter Date: Tue Mar 2 21:53:24 2010 +0100 i915 XvMC: don't stall the cpu anymore Proper bo management ensures that the cpu doesn't step on buffers used by the gpu. Drop the now unnecessary synchronization. Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit 9bba123c5f0f081562306c152e70221fc5041a3b Author: Daniel Vetter Date: Tue Mar 2 21:53:23 2010 +0100 i915 XvMC: drop superflous MI_FLUSH Cache coherency is now fully under the control of gem. For lack of hw documentation, I had to find out the correct cache placements by trial and error: Backward and forward surfaces: I915_GEM_DOMAIN_RENDER Correlation data: I915_GEM_DOMAIN_SAMPLER Changing any of them leads to visual corruptions, so I think these are the correct ones. Reviewed-by: Carl Worth commit b11623f20e303ae1d90d4a6bf0d5d73970b4e9bf Author: Daniel Vetter Date: Tue Mar 2 21:53:22 2010 +0100 i915 XvMC: switch surfaces to drm_intel_bo Now the last user of the fixed buffers provided by the ddx is gone! Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit b7f79bfd40db3ed17d7a79750d007b57d7ed157a Author: Daniel Vetter Date: Tue Mar 2 21:53:21 2010 +0100 i915 XvMC: switch corrdata buffer to drm_intel_bo It works! v2: Correlation data needs to be in the render cache! Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit 62846d88d41f1bbbd16a6e700ca611656fa20f61 Author: Daniel Vetter Date: Tue Mar 2 21:53:20 2010 +0100 i915 XvMC: switch msb to drm_intel_bo Like for the static indirect state buffer. Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit d27955c697e1768559957da8d61087644befd285 Author: Daniel Vetter Date: Tue Mar 2 21:53:19 2010 +0100 i915 XvMC: switch sis to drm_intel_bo I've decided to allocate a new buffer for every render command, to prevent stalling for the gpu. libdrm bo reuse should take care of not wasting memory in case the buffer is not busy. Also always emit the full state, it's not worth it to complicate the code over a few stores to wc memory. Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit 3203c66fbf04c55adba760e97c9fa1feb9e55f8d Author: Daniel Vetter Date: Tue Mar 2 21:53:18 2010 +0100 i915 XvMC: switch load_indirect_render_emit to batchbuffer macros Like with one_time_state_emit, this preps for relocatable bo's. Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit d30decae6a9db585f2d2589ae2c3d9e1f9acd33d Author: Daniel Vetter Date: Tue Mar 2 21:53:17 2010 +0100 i915 XvMC: switch psc to drm_intel_bo Like with the sampler state buffer. Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit b543c355f4fdcb64deb198765d9b8b86f7c234c2 Author: Daniel Vetter Date: Tue Mar 2 21:53:16 2010 +0100 i915 XvMC: switch psp to drm_intel_bo Like with the sampler state buffer. Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit 18c364084ce5495ffdfea38c17c1297d555c6c54 Author: Daniel Vetter Date: Tue Mar 2 21:53:15 2010 +0100 i915 XvMC: switch ssb to drm_intel_bo This also starts to kill the last remnants of the support for physical addresses for the indirect state buffers. With gem this would need kernel support (in the form of a new reloc type in execbuf2). This does not change the ABI between ddx and client libIntelXvMC. I've decided to do this in one swoop when all the buffer rework is done. Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit fc9e44f01910eb9fd43def9b2ac531b0742d8b4e Author: Daniel Vetter Date: Tue Mar 2 21:53:14 2010 +0100 i915 XvMC: kill last_flip and last_render Seems to be a remnant from i810 XvMC support. last_flip is always 0, so serves no real purpose anymore. Kill it and the associated code. With last_flip gone, last_render also lost its purpose. Kill it, too. Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit 04aa38a639fefcd00acb485f36ab18c2083ed1bb Author: Daniel Vetter Date: Tue Mar 2 21:53:13 2010 +0100 i915 XvMC: switch one_time_state_emit to batchbuffer macros This is in preparation for real relocatable drm_bo's instead of memory at a fixed address. By switching to the batchbuffer macros (like i965 xvmc) we can use the nice OUT_RELOC macro. Also align the code more with coding-style elsewhere, i.e. bitops instead of bitfield structures. The bitfield structures are quite a mess to work with the batchbuffer macros, so they were getting in the way, anyway. Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit 24d787335a0c319b8d9243ea1f2726575cf73b2b Author: Daniel Vetter Date: Tue Mar 2 21:53:12 2010 +0100 XvMC: kill dead code in i915_xvmc.c WIP code that hasn't changed for over two years is unlikely to suddenly start progressing. Drop it. After all, git can easily resurect it in cases it's needed. Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit 13266b152ae35ed54984844e3282cbdf20dc2e32 Author: Daniel Vetter Date: Tue Mar 2 21:53:11 2010 +0100 XvMC: kill i830_memory in 965 class xvmc Yes, this breaks binary compat of the struct passed around between X ddx and the client libXvMC. But we always ship both, so they should not get out of sync. Signed-off-by: Daniel Vetter Reviewed-by: Carl Worth commit d39d822cf887a861b37cee92c0b59533370ded2f Author: Daniel Vetter Date: Tue Mar 2 09:22:44 2010 +0100 i830_memory: hide as much of the old memory allocator as possible The only user left of this stuff is the xvmc support. Signed-off-by: Daniel Vetter Reviewed-by: Eric Anholt commit 65267d4bfbf19942beab72858333c6ee3c719223 Author: Daniel Vetter Date: Tue Mar 2 09:22:43 2010 +0100 i830_memory: switch frontbuffer to drm_intel_bo Yet another user of i830_memory gone for good. Signed-off-by: Daniel Vetter Reviewed-by: Eric Anholt commit 06f147dc04629a8a1534703be570e7f25e41cdd9 Author: Daniel Vetter Date: Tue Mar 2 09:22:42 2010 +0100 i830_memory: switch cursors to drm_intel_bo Minus one user of i830_memory, some more to go. Signed-off-by: Daniel Vetter Reviewed-by: Eric Anholt commit 23d12e3b088f38d25e83d6501d553d98be442d49 Author: Daniel Vetter Date: Tue Mar 2 09:22:41 2010 +0100 i830_memory: kill field "pitch" Totally unused. Signed-off-by: Daniel Vetter Reviewed-by: Eric Anholt commit 2fb8feeb56fae364380ffd91749eeb69fb461cff Author: Daniel Vetter Date: Tue Mar 2 09:22:40 2010 +0100 i830_memory: rip out field "size" Use the one in the drm bo instead. Signed-off-by: Daniel Vetter Reviewed-by: Eric Anholt commit 4d4d763b3d96e2cac99a0b7b03ad9bcaf4465c61 Author: Daniel Vetter Date: Tue Mar 2 09:22:39 2010 +0100 i830_memory: kill field "tiling_mode" Totally unused. Signed-off-by: Daniel Vetter Reviewed-by: Eric Anholt commit d0800d098739d5e8ea94be9c5ed2f7a8a86c06dc Author: Daniel Vetter Date: Tue Mar 2 09:22:38 2010 +0100 i830_memory: no memory allocations without a bo! Kill the corresponding !bo path in i830_free_memory. Also kill another remnant of the pre-kms era in the same file, while I was looking at the code. Signed-off-by: Daniel Vetter Reviewed-by: Eric Anholt commit 086c0e25cac1d3dd0a37def8b5cb82c1c6279edf Author: Daniel Vetter Date: Tue Mar 2 09:22:37 2010 +0100 i830_memory: rename i830_bind_all_memory to reflect code reality It doesn't bind anything anymore, but does a few random things. Give it a hopefully vague enough name to cover all cases ;) Signed-off-by: Daniel Vetter Reviewed-by: Eric Anholt commit 2a989aa057cee74154419fd0a4911ba1e95582cf Author: Daniel Vetter Date: Tue Mar 2 09:22:36 2010 +0100 i830_memory: rip out the old video memory allocator Besides the debug stuff the went away in the previous patch, this stuff was totally unused ... Signed-off-by: Daniel Vetter Reviewed-by: Eric Anholt commit bf83b9a10254966cb73b24e08954154d4296dac3 Author: Daniel Vetter Date: Tue Mar 2 09:22:35 2010 +0100 i830_memory: kill i830_desribe_allocations Totally useless debug function from the pre-gem era. No point to occasionally spam Xorg.log with a bogus "No memory allocations" message. Signed-off-by: Daniel Vetter Reviewed-by: Eric Anholt commit ff8a1e1cf728eeae848f6284bb4272193dfa918b Author: Daniel Vetter Date: Tue Mar 2 09:22:34 2010 +0100 i830_memory: rip out field "offset" Use the one in the drm bo instead. Signed-off-by: Daniel Vetter Reviewed-by: Eric Anholt commit e18ffceb142a9e18968900dd526a2d50fab72900 Author: Daniel Vetter Date: Tue Mar 2 09:22:32 2010 +0100 i830_memory: rip out field "end" It's a left-over from the non-gem era and no longer used at all. Signed-off-by: Daniel Vetter Reviewed-by: Eric Anholt commit 5018fd3097d77a5f31af4cb27e39daa37557b64e Author: Daniel Vetter Date: Mon Mar 1 22:57:40 2010 +0100 libIntelXvMC: kill ums leftovers On i965 class hw, kernel_exec_fencing was 1 always, anyway. And on i945, this patch kills a memory leak (dunno how, but it does). Signed-off-by: Daniel Vetter commit cb06aa32d433f54affe87da2cf964f0308d3c258 Author: Daniel Vetter Date: Mon Mar 1 22:57:39 2010 +0100 Xv: fixup XvMC on i915 I've accidentally broken i915 xvmc due to alignment constrains that break my assumption that Y-pitch == UV-pitch*2. Fix this up by consistenly using dstPitch2 for the Y-pitch. This also unifies the dst pitch computation slightly, now that the i915 xvmc special case is gone. Bugzilla: http://bugs.freedesktop.org/show_bug.cgi?id=25949 Signed-off-by: Daniel Vetter (Minor edit to support compilation without INTEL_XVMC defined by Carl Worth ) commit 80f2a1f3b5e477a12899ea2251aebcfa62689943 Author: Daniel Vetter Date: Mon Mar 1 22:57:38 2010 +0100 Xv: fixup YUV plane offset for xvmc case In my previous cleanup I've inadvertedly dropped the offset adjustment code for the xvmc passthrough case. Fix this up. Also reimplement that ugly hack I've accidently killed to keep i915 class xvmc a tad bit longer on life support. Signed-off-by: Daniel Vetter Tested-by: xunx.fang@intel.com commit a0ee9c3d9c72962c8d513ec8c43dd4a21e316947 Author: Carl Worth Date: Fri Feb 26 17:05:45 2010 -0800 Update version number to 2.10.901 In order to make a new snapshot. commit 63b8e890a134e5fc8242b655115491a3905ad0ff Author: Carl Worth Date: Fri Feb 26 17:05:06 2010 -0800 NEWS: Add notes for the 2.10.901 snapshot. From skimming the git lot since 2.10.0. commit d5409303128f5c9ba940ab564ff407a82eabc597 Author: Carl Worth Date: Fri Feb 26 16:12:50 2010 -0800 Remove dead code: fill_detailed_lvds_block All callers of this function were recently removed, so it can join them in the bit bucket. commit 7d0e6ff4dadcf243b1006ce6f85bd06c5f4e4908 Author: Adam Jackson Date: Wed Dec 2 12:05:52 2009 -0500 kms: Fix LVDS mode list construction. Rather than mangle the EDID block and hope the server does the right thing, just build a sensible mode list up front. Do this for LVDS where there is no EDID or where it does not claim to be continuous-frequency (since in the latter case, the server will add reasonable modes for us). Signed-off-by: Adam Jackson commit 1730af4437530ab7f6fe6ca74b16efb4a843b3ae Author: Carl Worth Date: Thu Feb 25 17:12:49 2010 -0800 NEWS: Fix typo (we just released 2.10 *not* 1.10) Sigh. Every version number that must be typed manually will eventually be typed incorrectly at some release. (I think that's Owne Taylor's Law) commit 529bf185fbcb9f7705b315a5106054ee25c1c77f Author: Eric Anholt Date: Wed Feb 24 17:54:13 2010 -0800 In frame event handling, track drawable id instead of drawable pointer. Windows aren't refcounted, so if the event came in after the window was destroyed, we'd dereference garbage and segfault. commit 633c7033170b0e9b468dbee444b94922f6c30940 Author: Eric Anholt Date: Wed Feb 24 17:40:30 2010 -0800 Fix up a bunch of inconsistent 4-space indentation in i830_dri.c commit 9291828a569a01ed4a6ef71f530b93f8a54c84aa Author: Eric Anholt Date: Fri Jan 15 12:51:50 2010 -0800 Add new mobile Sandybridge PCI IDs. commit 3c71f98b9e5262675e61fafb317d0c35e62a873f Author: Eric Anholt Date: Thu Oct 22 16:55:02 2009 -0700 Add initial defines and probing for Sandybridge commit c2c670ef18755cf5c878edf8a6b7d1617f54fe73 Author: Eric Anholt Date: Fri Feb 19 12:54:48 2010 -0500 Remove more DRI1 dead code. commit b320449f75adf912e06832dacc0b861ce26353bd Author: Eric Anholt Date: Fri Feb 19 12:44:11 2010 -0500 Remove dead i830_ring.h. commit 6d48b26098691d772a8de001c89a05b2442ef3b5 Author: Eric Anholt Date: Fri Feb 19 12:27:55 2010 -0500 Remove dead i830_display.h. commit 4126f73051aa73defcf342c81aa0fc9062f83f90 Author: Eric Anholt Date: Fri Feb 19 12:25:47 2010 -0500 Remove dead i830_bios.h. commit ca173c1eaf0e28f7fa138b1965f5e768b3add895 Author: Eric Anholt Date: Fri Feb 19 12:24:00 2010 -0500 Remove the last tool now that it's been moved to gpu_tools. commit 6da7cda5830817b0c977ed4767a9432748f12b09 Author: Eric Anholt Date: Fri Feb 19 12:21:47 2010 -0500 Remove swf_dumper. We have intel_mmio_read in gpu_tools to do this. commit 71c613c8475909e58fc491131bf9c10f4b047b16 Author: Eric Anholt Date: Fri Feb 19 12:20:33 2010 -0500 Remove bios_reader now that it's moved to gpu_tools. commit 7ada4eb3836f58a72f1c5b6b8139030fab9b7d63 Author: Eric Anholt Date: Fri Feb 19 12:18:23 2010 -0500 Remove intel_hotplug tool now that it won't work with KMS. We'd want to put something like it in sysfs or something if we wanted this tool to exist again. commit 3284aacfe095f6260eee5ed126685d0638626a29 Author: Eric Anholt Date: Fri Feb 19 12:14:12 2010 -0500 Remove dead i830_bios.c. Things that used it are now either in the kernel or stashed off in gpu_tools. commit 761d386a940081688b9b599c7a5b0270abfc5138 Author: Eric Anholt Date: Fri Feb 19 12:13:25 2010 -0500 Remove intel_gtt and intel_lid now that they're in gpu_tools. commit 6199af00cb47df6347ecd72ff7f6c86a564cec57 Author: Eric Anholt Date: Fri Feb 19 12:06:18 2010 -0500 Remove xprintf.c from the old "I can't believe it's not an X Server" tools. commit cba8e3136a100ba274f48c78df5b23fa8960fa16 Author: Eric Anholt Date: Fri Feb 19 12:04:51 2010 -0500 Remove intel_statuspage, now that we have /debug/dri/0/i915_gem_hws commit 40f5f72e30003993bfe298cd634d4f3f56148d4a Author: Eric Anholt Date: Fri Feb 19 12:01:30 2010 -0500 Remove dead main.c from reg_dumper/ commit 0c5e4a65f5d0f8d41c6bcb4c6978054f40064494 Author: Eric Anholt Date: Fri Feb 19 12:00:56 2010 -0500 Delete the audio dumping tool now that it lives in gpu_tools. commit 6bdab841766c8d9febaa23cb01a7bbc9edfae471 Author: Eric Anholt Date: Fri Feb 19 11:21:40 2010 -0500 uxa: Skip adjusting mask coordinates when no mask is present. Quiets clang warnings about garbage variable usage. commit a86869e6c3131b83a2ad529bc313270a9f45f5bd Author: Eric Anholt Date: Fri Feb 19 08:15:30 2010 -0800 Fix an unused variable warning for !INTEL_XVMC. commit ec5deb2bcbf828c3c47488eb77461c26cab38ed6 Author: Eric Anholt Date: Fri Feb 19 08:11:00 2010 -0800 Remove dead assignments noticed by clang. commit a4180eabfa00c256494ed09aa336a17605c49a8a Author: Eric Anholt Date: Fri Feb 19 08:10:17 2010 -0800 Remove stale junk in VideoRam setup, noticed by clang. commit 03657d4a698e0a25ab3863a15ad4c8b9d64dbb9a Author: Eric Anholt Date: Fri Feb 19 08:03:44 2010 -0800 Remove dead UMS CRTC resize code. commit 1c3aaad09d6ef207fba748ad4ef4575a26ab2e5c Author: Chris Wilson Date: Wed Feb 17 17:48:32 2010 +0000 dri2: Silence the compiler for an unused function with proto < 4 Move the unused function into the #if DRI2INFOREC_VERSION >= 4 block. Signed-off-by: Chris Wilson commit 00e7312dc45e54cd4547a943897a524639cb0b38 Author: Jesse Barnes Date: Thu Feb 11 09:46:12 2010 -0800 DRI2: handle full height blits without tearing in CopyRegion On 965 and up, if we detect a full height blit, we should just wait for vblank, rather than try to do a scanline wait for the whole display. On pre-965, doing a scanline wait followed by a blit works, but in the full height case we need to give the blitter time to start up, so we wait until the bottom line of the blit minus 2 padding scanlines to accommodate. Fixes FDO bug #22475. Signed-off-by: Jesse Barnes commit f0d760bfd71e2b2b40d2b250cd84b4626492ba47 Author: Jesse Barnes Date: Wed Feb 10 11:24:46 2010 -0800 Disable bo reuse on shadow framebuffer This keeps us from trying to set tiling on it while pinned, which also keeps us from trying to unpin it in the kernel, causing an error. Signed-off-by: Jesse Barnes commit 41784e15d39c29af5a87fe8d0982bd1b0be3562d Author: Gaetan Nadon Date: Fri Jan 29 15:54:16 2010 -0500 config: remove dead LINUXDOC macro usage This module does not generate LINUXDOC documentation. Signed-off-by: Gaetan Nadon commit 6610bcbac51c9ac970128012f0d4566d8cfba000 Author: Jesse Barnes Date: Fri Feb 5 17:18:50 2010 -0800 DRI2: only use version 4 APIs if kernel support exists Check for page flipping support before enabling flip and vblank event support needed for the new DRI2 APIs. Signed-off-by: Jesse Barnes commit 1a76fa5574e8e8f88ac3518a4e4494e1af301dc1 Author: Keith Packard Date: Fri Jan 29 23:28:46 2010 -0800 Initialize DRI2 info rec version 4 list of driver names With DRI2 supporting multiple subsystems, the video driver must initialize the list of driver names instead of just passing the single driver name used by Mesa. Without this, the X server will fail to initialize DRI2 as the numDrivers field in this structure will be uninitialized. Signed-off-by: Keith Packard commit 918151a7955c26174db80b775205f6ffb4f44ab6 Author: Chris Wilson Date: Wed Jan 27 20:33:08 2010 +0000 uxa: Fix compatible_formats() for OVER In separating the boolean logic out into a separate function, dc6522dd, I reversed the sense of one particular test: src->format == dst->format The OVER optimisation is only valid if the src and dst formats match, but not always. Signed-off-by: Chris Wilson commit 197cb08a2d54cabbfe97454d7db85cfe1f5f27ba Author: Chris Wilson Date: Mon Jan 25 15:15:04 2010 +0000 Extract pixel value for all formats to avoid hitting fallbacks. On failing to extract the pixel value for an alpha-only solid we actually triggered a fallback. Since this path is commonly hitting whilst fading in images, for example cairo_paint_with_alpha(), the fallback was detected during the Moblin boot sequence where it was adding a second to the overall boot time. See fallback intel: Moblin startup is hitting a composite fallback, costing a ton of performance https://bugs.freedesktop.org/show_bug.cgi?id=26189 Based on the initial patch by Arjan van de Van. Signed-off-by: Chris Wilson commit 5f93d019dc6311dd16b6792ffb60dbfc45ef3d08 Author: Chris Wilson Date: Mon Jan 25 15:47:11 2010 +0000 uxa: Adjust uxa_get_color_for_pixmap to match prototype The prototype says this function returns a Bool and not just an int, so be pedantic and return TRUE/FALSE. Signed-off-by: Chris Wilson commit dc6522dd491831f3843606e3ed57cc01fe2c804c Author: Chris Wilson Date: Sun Jan 24 09:37:23 2010 +0000 uxa: Protect against a potential NULL src->Drawable reference One of the convoluted if branches dereferenced Drawable when it is potentially NULL. Avoid this by explicitly handling the NULL Drawable cases earlier, and enabling solid fills for solid sources. Signed-off-by: Chris Wilson commit 31bbd7f919f4f1f545dd49861c15e60561c075cb Author: Chris Wilson Date: Sun Jan 24 09:04:16 2010 +0000 uxa/uxa-render: Always remove useless repeats during composite. I added a jump if there was no src or mask Drawable, but we do actually need to check for useless src repeats even if we have a source-only mask. Signed-off-by: Chris Wilson commit 326fe00df4160fbc3513ddbedfec90bdb2d7101b Author: Chris Wilson Date: Sun Jan 24 09:02:05 2010 +0000 uxa: Increase amount of composite fallback verbage The fallback log for http://bugs.freedesktop.org/show_bug.cgi?id=26189 does not actually state the reason why we actually fallback. This is possibly because we need to fallback for reasons other than the operation cannot be performed in hardware -- such as using an alpha map or the screen is swapped out, so add this information to the fallback log. Signed-off-by: Chris Wilson commit 93cd943d41c646c794b8cb5a960d8f0805e15395 Author: Eric Anholt Date: Fri Jan 15 12:53:47 2010 -0800 intel: Use the compositing-aware colorkey filler instead of homebrew fail. Of course, it's still fail since you can't correctly composite colorkey overlay, but at least this doesn't spam colorkey to the root window. Tested-by: Daniel Vetter commit 6e61de896b42cffe1af6b0384cb69265d3d121a1 Author: Alan Coopersmith Date: Fri Jan 15 15:51:28 2010 -0800 Update Sun license notices to current X.Org standard form Signed-off-by: Alan Coopersmith commit 96f45c66eed2631eba98ae416c3afdf540fa5c34 Author: Jesse Barnes Date: Wed Jan 13 13:20:43 2010 -0500 DRI2: if the swap condition is satisfied, complete it immediately If we get to the point where we check the divisor/remainder equation and it's satisfied, we should complete the swap immediately. Signed-off-by: Jesse Barnes commit 51c75906329a4727e37c8d1f64f257ea9602caa2 Author: Jesse Barnes Date: Fri May 1 14:52:26 2009 -0700 DRI2: support new DRI2 APIs The new interfaces allow for improved buffer swap, and support for the SGI_swap_control, SGI_video_sync and OML_sync_control GLX extensions. The Intel implementation allows page flipping to occur for swaps that are full screen and not rotated. Signed-off-by: Jesse Barnes commit 4902f546be19e3d5bb47f6c75e2199dc4856c0f4 Author: Chris Wilson Date: Sun Dec 13 10:44:12 2009 +0000 i965: Ensure that URB_FENCE is aligned to 64-bytes The PRM (Vol 1, p32) specifies that the URB_FENCE command must not cross a cache-line boundary (64-bytes) in order to workaround a silicon issue. Ensure that it does not by inserting an alignment point before the atomic section. This is a slightly too large hammer, but the easiest method to work with the current BEGIN_BATCH/ADVANCE_BATCH protections. Signed-off-by: Chris Wilson commit 83626aba357ffb4dd7931daaf163c1dd1d08f9d3 Author: Chris Wilson Date: Sun Nov 29 21:39:41 2009 +0000 uxa-glyphs: Enable TILING_X on glyph caches. Signed-off-by: Chris Wilson commit 50e07da8094c8c8c593b6eb8c41fc42444851d04 Author: Chris Wilson Date: Sun Dec 13 09:35:36 2009 +0000 i830: Do not use vtSema when chosing mapping type. The mapping type to use is determined by the tiling of the underlying object, not by whether or not not we control the vt. This was a left-over wart that was intended to mean that we had GEM and so could use GTT mappings. Signed-off-by: Chris Wilson commit 7a2b7cfab5cdef277f0feb838683422d9fcb0db3 Author: Chris Wilson Date: Sun Dec 13 09:33:45 2009 +0000 Consolidate determining maximum sizes for use with GEM Add a small wrapper function so that the callsites need only call the single function when checking the available aperture size for determining the maximum viable size for operations. This will allow us to easily extend this set in the future by only needing to adding the check to a single location. Signed-off-by: Chris Wilson commit 229d23fb18d696fb7ad476ce335be14ec9811bd3 Author: Daniel Vetter Date: Tue Dec 1 14:32:34 2009 +0100 Xv: don't enable XVMC port on unsupported configs This just makes it _really_ clear, what's supported. No other changes. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit ce7ba18f577cc9aedddaff303dbc9662a276b4cf Author: Daniel Vetter Date: Tue Dec 1 14:32:33 2009 +0100 Xv: consolidate xmvc passthrough handling It's now all in I830PutImageTextured. Also kill some leftovers from XVMC-on-overlay support and ums-XVMC-on-i915 support. Plus a small comment as a reminder for where to add i915 xvmc support back in. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 1ac7c94083a3266e2d5dd932709118436074dd00 Author: Daniel Vetter Date: Tue Dec 1 14:32:32 2009 +0100 Xv: hide ugly semantics in i830_clip_video_helper I'm still curious as to why fixed-point semantics are necessary for this generic XV helper function that's been causing all this. Can modern X really run on hw without floating-point support? Anyway, the ugliness is now all nicely under the carpet (in i830_clip_video_helper). Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 31f13fa8a0d4cd28067de37d0d31d23e8200d0ff Author: Daniel Vetter Date: Tue Dec 1 14:32:31 2009 +0100 Xv: move users of x1, x2, y1, y2 to PutImage After this there are no other external users of these strange variables, so we can nicely hide them somewhere in the next changeset. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit ce6526b9b4df304ccd83a0a02a95621300dbaed3 Author: Daniel Vetter Date: Tue Dec 1 14:32:30 2009 +0100 Xv: kill unnecessary parameters for hw PutImage functions This is the first part of my small crusade to rip out x1, x2, y1, y2 from I830PutImage*. These variables have strange semantics (they change from simple integers to fixed-point values somewhere in the middle) and don't really seem to be what we actually need. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit cf74caaa91667457bba3583b216dfc157d09f39e Author: Daniel Vetter Date: Tue Dec 1 14:32:29 2009 +0100 Xv: kill an unnecessary if We always pass a non-null pointer for crtc_ret, no point to check for this. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit d8353c737b5b8077f499461b1e6c78f09544d226 Author: Daniel Vetter Date: Tue Dec 1 14:32:28 2009 +0100 Xv: split up I830PutImage into textured and !textured case This wasn't making much sense anymore, and further cleanups will make this even more apparent. This change just makes two copies of I830PutImage and kills the not-applicable if-clauses in both versions. There is one small functional change in here: The textured video path doesn't munch around with adaptor_priv->videoStatus anymore, which is only used by the overlay. This could prevent the overlay from being switched off if someone would use textured video at the same time. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 091035146463bf1aa6674bff6947d04fc620c18f Author: Carl Worth Date: Mon Jan 4 14:20:11 2010 -0800 configure.ac: Bump version to 2.10.0. In preparation for the 2.10.0 release. commit 01f7d0307b3f5f085e383ff85026757de9cda359 Author: Carl Worth Date: Mon Jan 4 14:19:17 2010 -0800 NEWS: Add final release notes for the 2.10.0 release. Mentioning that it's functionally identical to our most recent release candidate. commit e966bca3a9143cff37beb54359bea13c9f693b44 Author: Zhenyu Wang Date: Tue Dec 29 21:49:49 2009 -0800 Take note for Pineview support in README and manpage Signed-off-by: Zhenyu Wang commit 6137791e9f1a4778afabbbc06de3da68648ae258 Author: Carl Worth Date: Mon Jan 4 14:08:24 2010 -0800 man page: Remove section describing BACKLIGHT_CONTROL property. This apparently no longer exists in a KMS world, so remove it from the documentation. commit 7f36a439249cc2101d8985de5e95af652e5b984b Author: Carl Worth Date: Mon Jan 4 14:07:24 2010 -0800 man page: Rename PANEL_FITTING to "scaling mode". The old UMS name was PANEL_FITTING while the new KMS name is "scaling mode". Fixes bug #25606. commit a6fb71e6e26fb31f4cd10fad2d3a87d9114a649a Author: Carl Worth Date: Mon Jan 4 14:05:27 2010 -0800 man page: Add additional indentation for some output configuration options. The BACKLIGHT_CONTROL and PANEL_FITTING options appear in a list, and then each contain a sub-list of sub-options. Use indentation to make this structure more apparent to the reader. commit 25a6c8dfae240143309b14cc32ebac6008c3a378 Author: Gaetan Nadon Date: Tue Dec 15 21:46:48 2009 -0500 configure.ac: use backticks rather than $() for cmd subs Use "$PKG_CONFIG" rather than hard coded "pkg-config" Acked-by: Dan Nicholson Acked-by: Daniel Stone Signed-off-by: Gaetan Nadon commit 2c142e421e859406b5aff16ba18624150269fc06 Author: Carl Worth Date: Thu Dec 10 15:25:44 2009 -0800 Update version to 2.9.99.902. For the second release-candidate snapshot in preparation for 2.10. commit 8ecf70ea553083cbc26928dc3973c8f6f8b3d9d0 Author: Carl Worth Date: Thu Dec 10 15:17:57 2009 -0800 NEWS: Add notes for 2.9.99.902 (and preliminary notes for 2.10) This comes from my cursory glance over the commit log from 2.9.1 to master. commit 88b2209a7526a7c7f3b3a2d491d48bd0d5d56e04 Author: Carl Worth Date: Thu Dec 10 14:52:46 2009 -0800 Document the DebugFlushBatches, DebugFlushCaches, and DebugWait options. These were added in 3c0815abf28744e215bea286e71d935cd486955a . The documentation added here comes straight from that commit message. commit 8d6faf7dc546bda64e11bf6dea7f3c997f07f887 Author: Tobias Doerffel Date: Wed Dec 9 09:57:20 2009 +0100 Fix compiler warning in i830_copy_video_data() if XvMC is disabled The variable "intel" is unused when building i830_video.c without XvMC support which results in a compiler warning: i830_video.c: In function 'i830_copy_video_data': i830_video.c:1443: warning: unused variable `intel' Trivial fix via #ifdef. commit bd81734465912d79d6320a6fb021ce43d258b906 Author: Daniel Vetter Date: Tue Dec 8 15:27:41 2009 +0100 Xv: enable drmmode overlay Now that libdrm 2.4.16 is released (and already required) we can unconditionally enable this. Please add something like this to the release-notes/NEWS file: * Overlay support for kernel modesetting. This needs at least kernel v2.6.33 to work. A backport to 2.6.32 is available at: http://gitorious.org/daniel-s-linux-stuff/linux-kernel/commits/intel-kms-overlay-for-2.6.32 Signed-off-by: Daniel Vetter commit 37f631d669c165c4fb56ccd7a6fc0a432f453b52 Author: Chris Wilson Date: Thu Dec 10 08:50:16 2009 +0000 Revert "uxa-glyphs: Enable TILING_X on glyph caches." This reverts commit 3f11bbec420080151406c203af292e55177e77d1. For unknown reasons, enabling tiling for the glyph cache is causing glyph corruption both across suspend and resume and VT switching, on a wide range of chipsets (reports include both i8xx and gm45) This strongly suggests that we are handling tiling, or updates to tiled buffers, incorrectly across i915_gem_idle(). However, until we can find the root cause, we want to fix this regression before the next stable release, so simply revert this patch. :( Fixes: [Bug 25406] fonts garbled after resuming from suspend since 6729b508 http://bugs.freedesktop.org/show_bug.cgi?id=25406 Signed-off-by: Chris Wilson commit 093bb9ebe69760975a3fcf2322db950312e6c2d7 Author: Chris Wilson Date: Tue Dec 8 23:45:28 2009 +0000 i965: Only use the affine kernels if both src and mask are affine Signed-off-by: Chris Wilson commit 0cf04ea4d736f7d7848f33b772d88a0f1b6678b1 Author: Chris Wilson Date: Tue Dec 8 23:44:45 2009 +0000 i965: Set src_filter before testing. Signed-off-by: Chris Wilson commit 417ad2712edcbca635eb5eeff0d11cdb16c069ed Author: Chris Wilson Date: Tue Dec 8 23:43:57 2009 +0000 Assert that we only call OUT_BATCH() inside a BATCH Signed-off-by: Chris Wilson commit 47416b1eea09b238a997636d35998d71e0d18161 Author: Chris Wilson Date: Tue Dec 8 13:47:07 2009 +0000 i965: Maximum number of vertices per composite is 24, not 18 Beware the potential buffer overflow. Signed-off-by: Chris Wilson commit c1afc831c8fe4cbececee7dfa23506a6746c2425 Author: Chris Wilson Date: Mon Dec 7 21:27:56 2009 +0000 uxa: Cache solid fills. Maintain a small cache of pixmaps to hold SolidFill pictures. Currently we create a pixmap the size of the damaged region and fill that using pixman before downloading it to the GPU and compositing. Needless to say this is extremely expensive compared to simply emitting the solid colour. To mitigate this cost, we maintain a small cache of 1x1R pictures which is recognised by the driver as being a solid, but at the very least is maintained as a GPU ready pixmap. This gives a good boost to cairo-xcb (which uses solid fills) on a gm45: Before: gnome-terminal-vim: 41.9s After: gnome-terminal-vim: 31.7s Compare with using a cache of 1x1R pixmaps in cairo-xcb: gnome-terminal-vim: 31.6s Signed-off-by: Chris Wilson commit cd475bad23c02130d11c49882c11261c9f0d4ef1 Author: Chris Wilson Date: Mon Dec 7 11:09:14 2009 +0000 batch: Ensure we send a MI_FLUSH in the block handler for TFP This should restore the previous level of synchronisation between textures and pixmaps, but *does not* guarantee that a texture will be flushed before use. tfp should be fixed so that the ddx can submit the batch if required to flush the pixmap. A side-effect of this patch is to rename intel_batch_flush() to intel_batch_submit() to reduce the confusion of executing a batch buffer with that of emitting a MI_FLUSH. Should fix the remaining rendering corruption involving tfp [inc compiz]: Bug 25431 [i915 bisected] piglit/texturing_tfp regressed http://bugs.freedesktop.org/show_bug.cgi?id=25431 Bug 25481 Wrong cursor format and cursor blink rate with compiz enabled http://bugs.freedesktop.org/show_bug.cgi?id=25481 Signed-off-by: Chris Wilson commit 415aab474edd1425034981306718afd8506445f1 Author: Chris Wilson Date: Fri Dec 4 09:02:36 2009 +0000 intel: And remember to flush the batch... In commit 98e11210 Remove flush parameter from intel_batch_flush() Maxi spotted that I had broken screen updating. It appears in my haste to eliminate the extra parameter I removed a call to intel_batch_flush() when throttling, i.e. when pushing the updates to the screen before idling. Should fix: Bug 25409 [bisected] rendering corruption since a938673e https://bugs.freedesktop.org/show_bug.cgi?id=25409 Signed-off-by: Chris Wilson commit 8438372dcc9d15c7b708332cfcb3a1c16a2c5cf2 Author: Kristian Høgsberg Date: Thu Dec 3 14:39:43 2009 -0500 Require libdrm 2.4.16 Needed for drmGetDeviceNameFromFd(). commit c439207ec0cc16d7d9f523598fcdebf7cec893e2 Author: Jesse Barnes Date: Wed Dec 2 14:43:17 2009 -0800 Update man page to reflect currently available options Many have been removed or are obsolete now that UMS is gone. And some are only available on i810/i815 or i830+, so move them to the appropriate section. Signed-off-by: Jesse Barnes commit a938673ee84d51ef655c37dfa7bbc5c35334cd28 Author: Chris Wilson Date: Wed Dec 2 20:51:53 2009 +0000 batch: Downgrade batch submission from a FatalError. If we wedge the GPU then we will return -EIO for the current batch and then attempt to reset the GPU. Meanwhile the X server detects the error, throws a FatalError and to all intents and purposes appears to crash to the user - whereas before it often just appeared to momentarily freeze. Of course, on older hardware the server remains frozen until we can find a way to reset those GPUs at runtime. Signed-off-by: Chris Wilson commit 98e11210367c950e3f932419d2a4722cf971885d Author: Chris Wilson Date: Wed Dec 2 20:48:37 2009 +0000 Remove flush parameter from intel_batch_flush() There is only a single caller that wishes to forcibly append a flush into the batch: intel_sync(). So move the logic there. Signed-off-by: Chris Wilson commit 57336c26f1fb90d43851ddcf78539585b67d86d9 Author: Chris Wilson Date: Wed Dec 2 20:42:41 2009 +0000 Rename I830Sync() to intel_sync() Signed-off-by: Chris Wilson commit 370157f4932cf9067ba81c8bd5a311aff610882b Author: Chris Wilson Date: Wed Dec 2 20:28:49 2009 +0000 batch: Avoid flushing a NULL batch During shutdown from a FatalError during batchbuffer submission, it is possible for the batch_ptr to be NULL, so we must be careful not to append a flush on this error path. Signed-off-by: Chris Wilson commit ad68881b670aabf8dbfd3b954e8796f91260579d Author: Chris Wilson Date: Wed Dec 2 14:14:39 2009 +0000 uxa_check_composite: Minor whitespace. Signed-off-by: Chris Wilson commit 49d2ccab2a82083110fe796636f3f91ba8c31237 Author: Chris Wilson Date: Wed Dec 2 14:13:43 2009 +0000 uxa_prepare_access() don't force a flush. Only the kernel knows whether the mapping requires a flush, so do not preempt it. Signed-off-by: Chris Wilson commit b68d3646f1fdfe012c16741958c7a62136a9b5aa Author: Chris Wilson Date: Wed Dec 2 14:12:19 2009 +0000 Review use of errno after libdrm call Since drm may not actually set the appropriate errno after a failure, we must use the return code instead when determining the cause of failure. Signed-off-by: Chris Wilson commit 0ff4d42a42b9e537b083343ee7dcc41cb41ae7cf Author: Chris Wilson Date: Wed Dec 2 12:12:07 2009 +0000 uxa: Review uxa_prepare_access() to remove potential nesting Around a call to uxa_put_image() it is possible to mix both accelerated and fallback paths, with the fallback code making the presumed optimisation of only trying to call uxa_prepare_access() once. This fails if the accelerated path also uses prepare/finish access on the same drawable and then later fallback to the fallback path. This can happen currently if an error is reported whilst attempting to accelerate PutImage. #0 memcpy () at ../sysdeps/x86_64/memcpy.S:162 #1 0x00007ffff43ce4bd in fbBlt (srcLine=, srcStride=40, srcX=, dstLine=0xffffffffffffffff, dstStride=64, dstX=0, width=, height=8, alu=3, pm=4294967295, bpp=8, reverse=0, upsidedown=0) at fbblt.c:93 #2 0x00007ffff43ce740 in fbBltStip (src=0xffffffffffffffff, srcStride=156555204, srcX=34, dst=0xfffffffc, dstStride=64, dstX=40, width=304, height=8, alu=3, pm=4294967295, bpp=8) at fbblt.c:944 #3 0x00007ffff4c32c53 in uxa_do_put_image (pDrawable=0x246aa410, pGC=0x2c0a4f0, depth=8, x=0, y=0, w=38, h=8, leftPad=0, format=2, bits=0x954d7c4 "") at uxa-accel.c:196 #4 uxa_do_shm_put_image (pDrawable=0x246aa410, pGC=0x2c0a4f0, depth=8, x=0, y=0, w=38, h=8, leftPad=0, format=2, bits=0x954d7c4 "") at uxa-accel.c:223 #5 uxa_put_image (pDrawable=0x246aa410, pGC=0x2c0a4f0, depth=8, x=0, y=0, w=38, h=8, leftPad=0, format=2, bits=0x954d7c4 "") at uxa-accel.c:289 #6 0x00000000004d574f in damagePutImage (pDrawable=0x246aa410, pGC=0x2c0a4f0, depth=8, x=0, y=0, w=38, h=8, leftPad=0, format=2, pImage=0x954d7c4 "") at damage.c:905 #7 0x00000000004287db in ProcPutImage (client=0x47ca72d0) at dispatch.c:2073 #8 0x000000000042bd94 in Dispatch () at dispatch.c:445 #9 0x000000000042513a in main (argc=4, argv=0x7fffffffe2a8, envp=) at main.c:285 Signed-off-by: Chris Wilson commit 6be26cae8379f973d9ca27e0d5371d16618e4f7b Author: Chris Wilson Date: Tue Dec 1 16:13:25 2009 +0000 i830: Simplify prepare_access / finish_access Reduce the 3 conditions into the 2 distinct cases. This has the secondary benefit of also distinguishing between the reported errors. Signed-off-by: Chris Wilson commit 637f003b047e426901cab6b1fe3a0924bcb9a38a Author: Chris Wilson Date: Tue Dec 1 13:20:20 2009 +0000 uxa: Don't treat prepare_access as a flush synchronisation point. The kernel will only emit a flush iff the buffer is currently owned by the GPU. Instead of presuming that the kernel must emit a flush, it is safer to assume that it does not and so cannot mapping the buffer on to the CPU as a synchronisation point. The most obvious counter-example is when we map the same buffer twice without using it in a batch. Signed-off-by: Chris Wilson commit cd5a9568ce0a541f030c27cdae529fe18e5f0437 Author: Carl Worth Date: Mon Nov 30 20:17:04 2009 -0800 Add i830_bios.h and i830_display.h to EXTRA_DIST. These files have been dropped from the generated tar file since the removal of UMS support. However, the bios_reader code still includes these, so "make distcheck" fails unless these are distributed. There's probably a cleaner fix possible, but this at least fixes the build so that the snapshot can be pushed out. commit 5bdac72a5c3f68ab6d9f7f2f1dfa6e5b78f04e57 Author: Carl Worth Date: Mon Nov 30 20:16:19 2009 -0800 NEWS: Note that the driver now has a hard dependency on KMS. Since the UMS code has all been removed. commit 813a910d7e4cc1c3deec0dcb1b536f8af337993a Author: Carl Worth Date: Mon Nov 30 19:48:05 2009 -0800 configure.ac: Bump version to 2.9.99.901. In preparation for a new snapshot. commit b62c72913a85895a60268a49eed1c136f63f160d Author: Carl Worth Date: Mon Nov 30 19:46:59 2009 -0800 NEWS: Add generic notes for the 2.9.99.901 snapshot. We plan to collect real release notes before the 2.10.0 release. commit 00aa4f7a45a318af5b651f9f3928e9da4443233a Author: Chris Wilson Date: Mon Nov 30 20:50:31 2009 +0000 uxa: Limit maximum size of tiled objects On older chipsets (i.e. pre-i965) tiling is very restrictive and imposes severe size and alignment constraints. Combine that with relatively small apertures and it is very easy to create a batch buffer that cannot be mapped into the aperture (but would otherwise fit based purely on total object size). To prevent this we need to not use tiling for large buffers (the very same buffers where tiling would be of most benefit!). Signed-off-by: Chris Wilson commit 8dd1c9eca02fb8da0c51f6fa4a38eb5e5ff41855 Author: Gaetan Nadon Date: Sun Nov 22 17:13:02 2009 -0500 video-intel: remove i2c_vid.h from src/Makefile.am make dist failed due to missing i2c_vid.h Commit b9b159c49854d8d9d2207946bb583537bb0d48d6 Remove UMS support. The above commit did not remove this header file from the makefile. Signed-off-by: Gaetan Nadon commit 6729b508c44bfca0b0dbef238a8732adbed6d4c9 Author: Chris Wilson Date: Mon Nov 30 16:52:10 2009 +0000 uxa: Initialise lists for private pixmap structure. When updating a buffer object for the framebuffer, we may need to allocate a fresh pixmap private structure, for example if the pixmap is replaced due to resize. When doing so it is then imperative to initialise the circularly linked lists correctly. Should fix the fault: #0 i830_set_pixmap_bo (pixmap=0x24ab380, bo=0x24ab780) at i830_uxa.c:524 #1 0x00007f8615c629fd in drmmode_xf86crtc_resize (scrn=0x247a320, width=1280, height=800) at drmmode_display.c:1345 #2 0x000000000051246c in xf86RandR12ScreenSetSize (pScreen=0x24824f0, width=, height=, mmWidth=, mmHeight=) at xf86RandR12.c:709 #3 0x0000000000512aa8 in xf86RandR12CreateScreenResources (pScreen=) at xf86RandR12.c:839 #4 0x0000000000514ec0 in xf86CrtcCreateScreenResources (screen=0x24824f0) at xf86Crtc.c:727 #5 0x0000000000424fb3 in main (argc=, argv=, envp=) at main.c:215 as reported by 'buscher'. Signed-off-by: Chris Wilson commit 85fe41126e83b35954dc2066eb103e0555e207a4 Author: Chris Wilson Date: Mon Nov 30 15:57:42 2009 +0000 uxa: Remove cache flush for copy As the copy uses the 2D blitter, it uses the render cache so the source should not require flushing if it has previously been used as a destination. Signed-off-by: Chris Wilson commit 2d434eed09936328a52cb942450a95aedc3eac63 Author: Kristian Høgsberg Date: Mon Nov 30 10:27:42 2009 -0500 Use new drmGetDeviceNameFromFd() for mapping to drm device name Should have been in libdrm to begin with. commit cfcabc45140d19bfbfa4737c0a11cdbb042d11eb Author: Chris Wilson Date: Mon Nov 30 13:58:30 2009 +0000 i915: Disable centre-point sampling. I still have no idea how this is triggering failures, but it is. So revert until the problem is solved. Should fix once again: Bug 23803 [bisected i915] gnome characters disappear http://bugs.freedesktop.org/show_bug.cgi?id=23803 Signed-off-by: Chris Wilson commit 8f8b6bd03d275379918777eaf7f63c0157b7ed9d Author: Chris Wilson Date: Mon Nov 30 14:03:40 2009 +0000 i915: Whitespace Signed-off-by: Chris Wilson commit 47916ea9d351f0ce6dc34713d6c164b0032f9830 Author: Chris Wilson Date: Mon Nov 30 13:58:06 2009 +0000 debug: Enable dumping of batchbuffer [compile-time only] Signed-off-by: Chris Wilson commit b118a52cd1a006321571967bd5f6c2a9e674de3a Author: Chris Wilson Date: Mon Nov 30 11:03:32 2009 +0000 i915: Remove routing of alpha channel to green. This modification is redundant since the routing is done in the blend unit anyway. Signed-off-by: Chris Wilson commit 5e04ded2bce4c135b57d391f5f4e24e030103e61 Author: Chris Wilson Date: Mon Nov 30 10:57:04 2009 +0000 i915: Fix missing texture offset for mask. In commit e581ceb, I modified the shader generation to accommodate mixed textures and solids but missed applying the new computed sampler for the mask. References: Bug 23803 [bisected i915] gnome characters disappear http://bugs.freedesktop.org/show_bug.cgi?id=23803 Bug 25031 rendering and color corruption since 14109abf http://bugs.freedesktop.org/show_bug.cgi?id=25031 Bug 25047 [945GM bisected] rendercheck/repeat/triangles regressed http://bugs.freedesktop.org/show_bug.cgi?id=25047 Signed-off-by: Chris Wilson commit a8ea20100de5be45699e71eb6ba67bef546ad0cd Author: Chris Wilson Date: Mon Nov 30 10:13:18 2009 +0000 debug: Don't always flush the batch when emitting a debugging flush I incorrectly changed the logic in 285f286 and caused the batch to always be flushed when debugging, instead of merely inserting a MI_FLUSH between operations. Signed-off-by: Chris Wilson commit c10850c63f9f88fbf08135bc4dcef1e5a2c71ee6 Author: Chris Wilson Date: Mon Nov 30 09:07:57 2009 +0000 overlay: Fix build after 646b4a9483 The compile cleanup was not without fault... Apparently I don't have XVMC enabled anymore and so missed that this variable is actually used. Signed-off-by: Chris Wilson commit 3f11bbec420080151406c203af292e55177e77d1 Author: Chris Wilson Date: Sun Nov 29 21:39:41 2009 +0000 uxa-glyphs: Enable TILING_X on glyph caches. Signed-off-by: Chris Wilson commit 19d8c0cf50e98909c533ebfce3a0dd3f72b755c1 Author: Chris Wilson Date: Sun Nov 29 21:16:49 2009 +0000 uxa: PutImage acceleration Avoid waiting on dirty buffer object by streaming the upload to a fresh, non-GPU hot buffer and blitting to the destination. This should help to redress the regression reported in bug 18075: [UXA] XPutImage performance regression https://bugs.freedesktop.org/show_bug.cgi?id=18075 Using the particular synthetic benchmark in question on a g45: Before: 9542.910448 Ops/s; put composition (!); 15x15 5623.271889 Ops/s; put composition (!); 75x75 1685.520362 Ops/s; put composition (!); 250x250 After: 40173.865300 Ops/s; put composition (!); 15x15 28670.280612 Ops/s; put composition (!); 75x75 4794.368601 Ops/s; put composition (!); 250x250 which while not stellar performance is at least an improvement. As anticipated this has little impact on the non-fallback RENDER paths, for instance the current cairo-xlib backend is unaffected by this change. Signed-off-by: Chris Wilson commit f7540f06090753cba1190aa9e8cdea05a9512077 Author: Chris Wilson Date: Sun Nov 29 21:12:07 2009 +0000 Only flush batch during prepare access if it may modify the pixmap. As we track when a pixmap is active inside a batch buffer, we can avoid unnecessary flushes of the batch when mapping a pixmap back to the CPU. Signed-off-by: Chris Wilson commit 9a2c18fb92659d57741bfdcacbe4f69aab361532 Author: Chris Wilson Date: Sun Nov 29 21:07:45 2009 +0000 batch: Emit a 'pipelined' flush when using a dirty source. Ensure that the render caches and texture caches are appropriately flushed when switching a pixmap from a target to a source. This should fix bug 24315, [855GM] Rendering corruption in text (usually) https://bugs.freedesktop.org/show_bug.cgi?id=24315 Signed-off-by: Chris Wilson commit 285f286597df5af13ac3f3d366f2fc9d0468dafa Author: Chris Wilson Date: Sun Nov 29 22:42:03 2009 +0000 batch: Track pixmap domains. In order to detect when we require cache flushes we need to track which domains the pixmap currently belongs to. So to do so we create a device private structure to hold the extra information and hook it up. Signed-off-by: Chris Wilson commit 2c3aee2b570dadd9270a08d8ff675d07ac405e33 Author: Chris Wilson Date: Sun Nov 29 20:53:35 2009 +0000 uxa-glyphs: Stream uploads via temporary bo Avoid mapping the glyph cache back to the cpu by allocating temporary buffer objects to store the glyph pixmap and blit to the cache. Signed-off-by: Chris Wilson commit 646b4a9483c01509a7324cc05eaadb72bc940c6d Author: Chris Wilson Date: Sun Nov 29 10:53:36 2009 +0000 Cleanup a few compiler warnings. Simple warnings for unused variables and C99-style declarations. Signed-off-by: Chris Wilson commit 917f9bb2435ba36bb99ef4d4f7b380d7f265e862 Author: Gaetan Nadon Date: Mon Nov 23 09:25:05 2009 -0500 Makefile.am: add ChangeLog and INSTALL on MAINTAINERCLEANFILES Now that the INSTALL file is generated. Allows running make maintainer-clean. commit eda2bb26783f7e9ef1ae659265930308378206e1 Author: Daniel Kahn Gillmor Date: Fri Nov 20 01:52:05 2009 +0100 Makefile.am: Add missing .g4i to be included in the tar file. Apparently Debian packages were having to manually add these files back in. Distribute them in the first place like we meant to. commit c5e86453c3ae3709933779a9dd609bbaebe21e8e Author: Gaetan Nadon Date: Wed Oct 28 14:41:41 2009 -0400 INSTALL, NEWS, README or AUTHORS files are missing/incorrect #24206 Automake 'foreign' option is specified in configure.ac. Remove from Makefile.am commit 12c4a22dd0b4de92a9c3d9e0baab029292f118f0 Author: Gaetan Nadon Date: Wed Oct 28 14:09:09 2009 -0400 INSTALL, NEWS, README or AUTHORS files are missing/incorrect #24206 Add missing INSTALL file. Use standard GNU file on building tarball README may have been updated Remove AUTHORS file as it is empty and no content available yet. Remove NEWS file as it is empty and no content available yet. commit ad4030932490f848c8ac21ba5a7a8d734994ed15 Author: Gaetan Nadon Date: Mon Oct 26 12:54:21 2009 -0400 Several driver modules do not have a ChangeLog target in Makefile.am #23814 The git generated ChangeLog replaces the hand written one. Update configure.ac to xorg-macros level 1.3. Use XORG_DEFAULT_OPTIONS which replaces four XORG_* macros Update Makefile.am to add ChangeLog target if missing Remove ChangeLog from EXTRA_DIST or *CLEAN variables This is a pre-req for the INSTALL_CMD commit bb994ce018ae6c3e3c5548ae7439cfc5e9aa4554 Author: Gaetan Nadon Date: Thu Oct 22 13:02:04 2009 -0400 .gitignore: use common defaults with custom section # 24239 Using common defaults will reduce errors and maintenance. Only the very small or inexistent custom section need periodic maintenance when the structure of the component changes. Do not edit defaults. commit 67bbda0bddae3e0d39bc8deb7378f467aedfd0e3 Author: Gaetan Nadon Date: Thu Oct 22 12:34:17 2009 -0400 .gitignore: use common defaults with custom section # 24239 Using common defaults will reduce errors and maintenance. Only the very small or inexistent custom section need periodic maintenance when the structure of the component changes. Do not edit defaults. commit c180baf43b8a0e407448018f3a7e42491cf974ae Author: Chris Wilson Date: Fri Nov 13 19:35:23 2009 +0000 i915: Derive the correct target color from the pixmap by checking its format Particularly noting to route alpha to the green channel when blending with a8 destinations. Fixes: rendercheck/repeat/triangles regressed http://bugs.freedesktop.org/show_bug.cgi?id=25047 introduced with commit 14109a. Signed-off-by: Chris Wilson commit e9064eacb0ad8867e320597453facbb3c376522c Author: Chris Wilson Date: Fri Nov 13 18:35:44 2009 +0000 uxa: Do not remove repeat from solids for 1x1 composites. Or else we hit the buggy 1x1 source path and trigger: rendercheck/mcoords regressed http://bugs.freedesktop.org/show_bug.cgi?id=25046 caused by the recent commit e581ceb. commit 14109abf285866ad4cd99d0cd16b0954a0a73a62 Author: Chris Wilson Date: Tue Nov 10 11:17:23 2009 +0000 i915: Fix texture sampling coordinates. RENDER specifies that texels should sampled from the pixel centre. This corrects a number of failures in the cairo test suite and a few off-by-one bug reports. Grey border around images https://bugs.freedesktop.org/show_bug.cgi?id=21523 Note that the earlier attempt to fix this was subverted by the buggy use of 1x1R textures for solid sources -- which caused the majority of text to disappear. Signed-off-by: Chris Wilson commit e581ceb7381e29ecc1a172597d258824f6a1d2d3 Author: Chris Wilson Date: Tue Nov 10 11:14:23 2009 +0000 i915: Use the color channels to pass along solid sources and masks. Instead of allocating and utilising the texture samplers for 1x1R solid sources and masks we can simply use the default diffuse and specular colour channels and adjust the fragment shader appropriately. The big advantage is the reduction in size of batches which should give a good boost to glyph performance, irrespective of the additional boost from using simpler shaders. However, the motivating factor behind the switch is that our use of 1x1 textures turns out to be buggy... Signed-off-by: Chris Wilson commit 33cabbfca6acb5149e26f87a538a7cb79f00cad2 Author: Chris Wilson Date: Tue Nov 10 11:09:52 2009 +0000 i915: Check for overflow before overflowing. As the immediate victim of the overflow would be to overwrite the maximum permissible value, the test was optimistic. Signed-off-by: Chris Wilson commit 67af5a99253b1295f8dc09b28863eb7dc8b59e1d Author: Chris Wilson Date: Tue Nov 10 11:05:20 2009 +0000 Check that batch buffers are atomic. Since batch buffers are rarely emitted by themselves but as part of a sequence of state and vertices, the whole sequence is emitted atomically. Here we just enforce that batches are marked as being part of an atomic sequence as appropriate. Signed-off-by: Chris Wilson commit 998d6b3d8c549086fbc8a9f0e309694b23398d8d Author: Chris Wilson Date: Tue Nov 10 11:00:43 2009 +0000 uxa: Force alpha bits to fill remaining bits In the case of x8r8g8b8 and similar where the alpha channel is ignored, but should be interpreted as being 1, then it is convenient if those bits are set appropriately in the colour. In order to do so for these formats, where PIXMAN_FORMAT_A() returns 0 we need to compute the alpha channel width as the remaining bits instead. Signed-off-by: Chris Wilson commit dbb68168dc909ab2ec1d935322c3fd8581e666f1 Author: Eric Anholt Date: Thu Nov 5 15:40:20 2009 -0800 Revert "configure: make --disable-dri work even if the server supports DRI" This reverts commit a851139c2141f6da370186148f2836e18b2acf83. It broke the build, and I don't see why we should be supporting this anyway. Conflicts: configure.ac src/Makefile.am commit 4c8e783d84d2c14c8a1638b6a12307c0164d3e31 Author: Eric Anholt Date: Wed Oct 21 13:30:38 2009 -0700 Fix "Remove flow-control macros for fallbacks in the 2D driver." I guess this is the sort of failure due to rebase-happiness that makes Linus yell at us for rebasing. commit d0e08fe611681fcc840d1b9ee8d49acdf50f7e58 Author: Daniel Vetter Date: Mon Oct 26 13:15:24 2009 +0000 Kill some more #defines only needed for User-Modesetting Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 8ff2a6496413e6b12fad9352f4bd9b6736bda56c Author: Eric Anholt Date: Wed Oct 21 13:30:38 2009 -0700 Remove flow-control macros for fallbacks in the 2D driver. It's poor style, and has confused new developers. commit c87585229b36790f883b9b8954ed061e00624df6 Author: Jesse Barnes Date: Thu Nov 5 13:33:55 2009 -0800 Increase stride limit for IGDNG It can go up to 32k. Upping this lets me use my 2560x1600 and 1920x1200 monitors in an extended desktop configuration. Signed-off-by: Jesse Barnes commit cbcfb711747e3637f20145aed00ac3d6256d0d1d Author: Rémi Cardona Date: Wed Oct 28 13:12:46 2009 +0100 configure: drop some more cruft DEBUGFLAGS wasn't used anywhere and the configure switch is useless. "CFLAGS=-g ./configure" does the same thing and actually works. PKG_CHECK_MODULES already calls AC_SUBST, no need to do it twice. Signed-off-by: Rémi Cardona Acked-by: Chris Wilson commit 3c0a43b24cc7540cc139beca0ec2e4c202ff5593 Author: Rémi Cardona Date: Wed Oct 28 13:01:32 2009 +0100 configure: use CWARNFLAGS from xorg-macros.m4 Signed-off-by: Rémi Cardona Acked-by: Chris Wilson commit a851139c2141f6da370186148f2836e18b2acf83 Author: Rémi Cardona Date: Wed Oct 28 12:37:38 2009 +0100 configure: make --disable-dri work even if the server supports DRI XF86DRI is defined by the SDK so not defining it here just breaks the build. Define HAVE_DRI instead to avoid collisions. Note: DRI2 is still enabled/disabled entirely by SDK defines. Signed-off-by: Rémi Cardona Acked-by: Chris Wilson commit 07e0b2cff6107d6c86096c7da4e31b0c976794d5 Author: Rémi Cardona Date: Wed Oct 28 11:45:55 2009 +0100 configure: group system header checks at the top Signed-off-by: Rémi Cardona Acked-by: Chris Wilson commit 0401a4c3143c5b94875cb4c66c94a0d247c32da5 Author: Rémi Cardona Date: Wed Oct 28 11:42:41 2009 +0100 configure: check for libdrm and libpciaccess earlier Signed-off-by: Rémi Cardona Acked-by: Chris Wilson commit 10946118dd3a63f1375a1bfde0b2f0542a93c1c2 Author: Albert Damen Date: Sun Oct 25 18:02:07 2009 +0100 Fix crash in uxa_acquire_pattern when pDst is NULL This avoids a crash when an XRenderComposite call is made with a -1 value for width/height, (which apparently compiz's gtk-window- decorator likes to do). Fixes bug: X crashes in uxa_acquire_pattern when logging in (gdm) http://bugs.freedesktop.org/show_bug.cgi?id=24724 Signed-off-by: Albert Damen Reviewed-by: Carl Worth commit 751e0a3e4576bbf4bffa56bbd6d4de28f10db98e Author: Thomas Arnhold Date: Mon Oct 19 11:35:30 2009 -0700 Fix 64-bit compiler warnings in intel_xvmc_dump_render(). Bug #24396. commit 3e8f2eae3a586aa29be4858698e666e0ec778cea Author: Eric Anholt Date: Thu Oct 15 13:48:56 2009 -0700 XVMC: Use XCB DRI2 instead of cargo-culting our own copy of Xlib stuff. (v2) v2: Incorporate comments from Jamey on device name handling and extension detection. commit 38ab403d7a1c461c8ac65a056bee2dd5c7f2f58e Author: Eric Anholt Date: Thu Oct 15 11:39:32 2009 -0700 Enable XVMC by default on gen4. commit f171069608bf174d920921fa102b8619794ed272 Author: Zhenyu Wang Date: Mon Sep 28 18:35:57 2009 +0800 i965 XvMC cleanup Remove bo pin for surface buffer access, and remove access attempt for possible unmapped framebuffer. Using xv buffer pointer to pass current xvmc surface bo handler, which is assigned to src image bo and handle that the same way as in Xv. Signed-off-by: Zhenyu Wang [anholt: Fixed up for conflict against the XV rework. Not tested, because both mplayer and xine segfault with XVMC currently.] Signed-off-by: Eric Anholt commit aaedeffe00d9414bb03723dbc30b4938a07ce5fa Author: Daniel Vetter Date: Wed Oct 14 18:09:08 2009 +0200 Xv overlay: fix planar YUV copy for right rotated crtcs While copying and rotating the buffer, array access was out of bounds when rotated to the right (RR_Rotate_270). My buffer handling changes probably made this bug much more likely to actually result in a SIGSEGV. I've checked the logs and the bug exists since rotation has been supported, i.e. this looks like a candidate for cherry-picking for all supported releases. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 703e3326bb41528cc57c0d25003707df209fc714 Author: Daniel Vetter Date: Wed Oct 14 15:56:55 2009 +0200 Xv overlay: further cleanups Kill some unnecessary stuff. Small code changes, but no functional ones. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 1c2aedfce9e6a7561347e873d125b6889dee7941 Author: Daniel Vetter Date: Wed Oct 14 15:56:54 2009 +0200 Xv: fixup the disabled drmmode overlay code This code didn't survive the global renaming of vars to saner names. Fix it up. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 909990f40a437cbd7026a10e32af1ea120f4c2a7 Author: Daniel Vetter Date: Wed Oct 14 15:56:53 2009 +0200 Xv overlay: remove some more dead stuff from ums overlay support Mostly unused definitions and variables, but also some strange ums debug code. Also kill some now obsolete comments. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 86bc23ab5da34137c82250395c68aa92ecd88a24 Author: Chris Wilson Date: Wed Oct 14 12:04:40 2009 +0100 debug: Enable cache flushing after every operation If DEBUG_FLUSH_CACHES is enabled then emit a MI_FLUSH after every rendering operation. This is intended to 'fix' cases where we are missing a required flush in the middle of a sequence of operations, such as switching between 2D to 3D and render to sampler. Signed-off-by: Chris Wilson commit 3c0815abf28744e215bea286e71d935cd486955a Author: Chris Wilson Date: Wed Oct 14 11:29:21 2009 +0100 conf: Add debugging flush options Make the following options available via xorg.conf: Section "Driver" Option "DebugFlushBatches" "1" # Flush the batch buffer after every # single operation; Option "DebugFlushCaches" "1" # Include a MI_FLUSH at the end of every # batch buffer to force data to be # flushed out of cache and into memory # before the completion of the batch. Option "DebugWait" "1" # Wait for the completion of every batch buffer # before continuing, i.e. perform synchronous # rendering. EndSection Signed-off-by: Chris Wilson commit 2f134b84445d2d0bfe3d81276bc7f6648df062ee Author: Zhao Yakui Date: Wed Sep 30 07:31:57 2009 +0000 Return failure when the enum-type output property values are not found Failure to do so causes xrandr to report incorrect property values. Signed-off-by: Zhao Yakui Signed-off-by: Eric Anholt commit fcc2ee48b866b81c79315ff10189b56fc201539d Author: Albert Damen Date: Wed Oct 7 21:55:42 2009 +0200 Drop frontbuffer from crtc in I830CloseScreen By dropping the frontbuffer from the crtc, the new frontbuffer can be properly added to the crtc when the xserver is reset. Signed-off-by: Albert Damen commit 8a77877f9c2c6a8a1308bc1a3be9e7ad88bc7f49 Author: Dave Airlie Date: Fri Oct 9 14:16:06 2009 +1000 drmmode: with 1.7 server, set mode major doesn't get gamma setup. Noticed this on Fedora, where 1.7 server does gamma via the randr codepaths however kms doesn't have this call which happens in the non set_mode_major path. probably should be backported to released drivers. Signed-off-by: Dave Airlie commit d8c7678ddadce89ca7fc0edd1b4d9ed3d959e687 Author: Jesse Barnes Date: Mon Oct 12 14:02:12 2009 -0700 DRI2 compat build fix: it's drawable->pScreen not drawable->screen Fallout from the conversion; DRI2 compat path was broken. Signed-off-by: Jesse Barnes commit b37ac9d317ae537d993922976f87072040b04d04 Author: Chris Wilson Date: Mon Oct 12 14:31:06 2009 +0100 uxa: Refactor create Picture for pixman format Pull the common methods for creating a Picture given a pixman format into its own method, and tidy the surrounding code. The benefit is that we can now composite directly to the Picture and so save an intermediate copy when creating patterns for gradients. Signed-off-by: Chris Wilson commit 7e8f32d0a7279dce1976f87612833d9092554cfe Author: Chris Wilson Date: Mon Oct 12 14:33:08 2009 +0100 uxa: Free the ScratchPixmapHeader after its associated Picture Fixes: http://bugs.freedesktop.org/show_bug.cgi?id=24459 Intel Driver > 2.8: Cairo rendering bug, triggered in QtCurve GTK engine Signed-off-by: Chris Wilson commit 1556c62e0336ea2fef866722ee44d2d188e318f3 Author: Eric Anholt Date: Thu Oct 8 17:34:13 2009 -0700 Replace dolt and shave with relying on current autotools. commit 8b2d2ff0d026eea445a071102a62d782f3bbec78 Author: Eric Anholt Date: Wed Oct 7 16:12:25 2009 -0700 Clean up more i830_memory.c madness. It was cooking up insane alignment values for buffers that new libdrm was justifiably complaining about, but it turns out we don't need the alignment values anywhere because the only case they're needed, they're computed entirely by the kernel. Also, the XVMC code was passing a completely unused flag in. commit d525a0e993a59e118ab1e8519b3d73465f8f0169 Author: Eric Anholt Date: Tue Oct 6 19:05:24 2009 -0700 Rename pScreen to screen. commit b6262dcd2754f7e87f27979dcad145133ee3b945 Author: Eric Anholt Date: Tue Oct 6 19:00:41 2009 -0700 Rename pGC to gc. commit fdcfeb822d15cbfe6fccceab002b75458dca3264 Author: Eric Anholt Date: Tue Oct 6 18:57:22 2009 -0700 Rename pDraw to plain old drawable. commit 7bbf4ac7137ed0e4ec0a75806a396c12f102e779 Author: Eric Anholt Date: Tue Oct 6 18:50:54 2009 -0700 Rename the xv pPriv to adaptor_priv to reflect whose private it is. commit f309d475241260cf60567100511d5f7c6c487a29 Author: Eric Anholt Date: Tue Oct 6 18:38:53 2009 -0700 Call pPixmaps plain old pixmaps. commit da0f6616ad63f1581cf91a98104e5287aa44e7ce Author: Eric Anholt Date: Tue Oct 6 18:37:05 2009 -0700 de-pCamelHungarian the Render pictures and pixmaps. commit 050a141b7bc94b459061615124b7686a9c331e01 Author: Eric Anholt Date: Tue Oct 6 18:30:57 2009 -0700 Share several render fields between render implementations. Also, start settling on the cairo naming for things: source, mask, and dest. commit af27a3a0a5645c6f41f583611bd0f2559dc7cb2f Author: Eric Anholt Date: Tue Oct 6 18:24:50 2009 -0700 Rename the xf86 screen private from pScrn to scrn. commit cc5d3ba3c331c3b1becf2d19277b24144bf34cfa Author: Eric Anholt Date: Tue Oct 6 18:02:38 2009 -0700 Rename the screen private from I830Ptr pI830 to intel_screen_private *intel. This is the beginning of the campaign to remove some of the absurd use of Hungarian in the driver. Not that I don't like Hungarian, but I don't need to know that pI830 is a pPointer. commit 03e8e64f8669263e3cecb79ea57d5a26c0eaee3f Author: Chris Wilson Date: Wed Oct 7 22:15:37 2009 +0100 Complete the removal of UMS build support Eric missed removing the now dead references to the deleted subdirectories in configure.ac with commit b9b159c Signed-off-by: Chris Wilson commit 5e44a0fa4209523bee56f513a2029869f96a710a Author: Eric Anholt Date: Tue Oct 6 18:08:51 2009 -0700 Remove more dead UMS code. commit 6a716a25abc84f1c7f766fd3bfa1c2aac91e6ae7 Author: Eric Anholt Date: Tue Oct 6 17:58:32 2009 -0700 Remove the reg_dumper code, now that it's been moved to intel_gpu_tools. commit e9aff787a55d684e3b9c6d5ea8ac2926be952280 Author: Eric Anholt Date: Tue Oct 6 17:53:59 2009 -0700 Remove the stepping debug app, replaced by version in intel_gpu_tools. commit 71b9cdaa86bb9d3deb44f32f47771c33898253c2 Author: Eric Anholt Date: Tue Oct 6 17:53:21 2009 -0700 Trim down the hotplug debug app. commit e95eb483e210def0814bc61fc36d672145da7dff Author: Eric Anholt Date: Tue Oct 6 17:51:09 2009 -0700 Remove intel_idle, replaced by intel_gpu_top in intel_gpu_tools. commit 8ae0e44e42db645abe6d385f561260d2ae4a1960 Author: Eric Anholt Date: Tue Oct 6 16:30:08 2009 -0700 Move to kernel coding style. We've talked about doing this since the start of the project, putting it off until "some convenient time". Just after removing a third of the driver seems like a convenient time, when backporting's probably not happening much anyway. commit b9b159c49854d8d9d2207946bb583537bb0d48d6 Author: Eric Anholt Date: Wed Sep 30 16:29:53 2009 -0700 Remove UMS support. At this point, the only remaining feature regressions should be the lack of overlay support (about to land), and the need to update the XVMC code to work in the presence of KMS. Acked-by: Keith Packard (in principle) Acked-by: Carl Worth (in principle) commit d26f4c493721dbb2e2cbf4efcf8d37228d1b1f3b Author: Eric Anholt Date: Mon Oct 5 13:31:31 2009 -0700 Remove UMS overlay support. The replacement code is now landed, and the rest of UMS is about to disappear. commit 33c488e83676d26e69145ea615f75ca52768f400 Author: Eric Anholt Date: Mon Oct 5 10:38:05 2009 -0700 Remove error state dumping code. This is replaced by intel_gpu_dump, and would no longer be used once UMS is gone. commit 2370af32fe31bed8e5715639023635bdb3b83b1b Author: Daniel Vetter Date: Tue Aug 11 16:06:47 2009 +0200 Implement drmmode overlay This does not restore the overlay on EnterVT/disable it on LeaveVT. Does not look like this is necessary. Signed-off-by: Daniel Vetter [anholt: Hacked in avoiding the actual kernel calls with Signed-off-by: Eric Anholt commit 99230864050ea2f26c2412c4c87c50947fc3cbe1 Author: Daniel Vetter Date: Tue Aug 11 16:06:46 2009 +0200 Xv: introduce an overlay hal This is the last preparatory step for overlay support with drmmode. Safe two (specially marked) function calls in the setup code, all hw accessing code goes now through these three new functions with the ums_overlay prefix. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit bb04e184fdc55325a362ebf291e16a7c112a5f1d Author: Daniel Vetter Date: Tue Aug 11 16:06:45 2009 +0200 Xv: rework overlay buffer management The basic idea is to only pin the buffer into the gtt when the overlay hw is actually using it. This results in a few changes: - Unify data copied/buffer handling with textured video. Now offsets are always buffer relative and we just use drm_bo_map to access a buffer. - Implement double buffering using two bo's. This is necessary because we can't pin the same buffer to the gtt and map it as normal memory. - Kill XV_DOUBLE_BUFFER. With the above changes, overlay video is always doubel buffered. There is still the XvMC passthrough case, which makes the code slightly ugly. Unfortunately we can't get at the bo behind this buffer. Changes since the last review-round: - Don't overallocate by a factor of 2. - Prevent possible use-after-free issue. Signed-off-by: Eric Anholt commit 60462eb5b5c1931beecc1ed2d32f91090f64174c Author: Daniel Vetter Date: Tue Aug 11 16:06:44 2009 +0200 Xv: create xvmc_passthrough helper This way all thes strange special cases make much more sense. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 71276dff9415322f75792a46c1a1cc0b900588b9 Author: Daniel Vetter Date: Tue Aug 11 16:06:43 2009 +0200 Xv: scrap overlay offscreen pixmap support The code looks like it's been bitrotting since being copied over from the i810 driver. Furthermore painting rgb pixmaps with the overlay engine is in these days of modern compositing X an absolute no-go. And textured video doesn't support it neither, so its likely never ever used by applications. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 99afdf985fa9f763fda4bc49ccd8111960a4ae0f Author: Daniel Vetter Date: Tue Aug 11 16:06:42 2009 +0200 Xv: small cleanups in I830PutImage - scrap unused variable overlay - scrap an superflous if and attach the code to the preceeding else - tiny layout fix. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit d598456f1fe013c250530730c57ad720d38cde3a Author: Daniel Vetter Date: Tue Aug 11 16:06:41 2009 +0200 Xv I830PutImage splitup: extract i830_wait_for scanline Also scrap the unecessary variable sync in I830PutImage and the accompanying obfuscated logic. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 2ba03fa78197d1cb10d6401ad02cfe9edd2623b3 Author: Daniel Vetter Date: Tue Aug 11 16:06:40 2009 +0200 Xv I830PutImage splitup: extract i830_copy_video_data Just moves the code and passes back allocation failures. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 81fc74cc4ce81b9be8b9ba3bcacf1284c47ce80c Author: Daniel Vetter Date: Tue Aug 11 16:06:39 2009 +0200 Xv I830PutImage splitup: extract i830_setup_video_buffer Just move the code and pass back allocation failures. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 0bf4cc5130482ee0f3924f34e37fa5fc988119b3 Author: Daniel Vetter Date: Tue Aug 11 16:06:38 2009 +0200 Xv I830PutImage splitup: extract i830_dst_pitch_and_size Just moves the code. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit c20ef591931b5efc5745d00c06f296c89cf32745 Author: Daniel Vetter Date: Tue Aug 11 16:06:37 2009 +0200 Xv: kill destId in I830PutImage It's only used to remember that XvMC has ỲV12 as output. is_planar_fourcc already takes care of that in all necessary cases. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 5ce944994d57176d0ddf6f6d7c7779fdfcf39cb8 Author: Daniel Vetter Date: Tue Aug 11 16:06:36 2009 +0200 Xv: kill hw double buffering logic The idea for the hw double buffering support is to program two fixed buffers and then only switch buffers in the OCMD register. But the driver as-is always programs the new buffer address (in both register sets when double buffered). Therefore we gain nothing by using this hw capability. Scrap the software support for it. When double buffered, we now allocate just a buffer of size 2*size and switch between the two parts purely in software. To make reviewing this easier, I'll shortly explain the differences of how double-buffering (i.e. tear-free video) is achieved before and after this change: - When double buffer, allocate a buffer twice the size (unchanged). - Depending upon the currently shown buffer-half, copy the new frame into the other buffer-half. In the old code this is done by using the right set of buffer offsets, either *Buf0Offset or *Buf1Offset. The new code simply programs the offset for the right buffer-half into the single set of offsets. The end-result is unchanged. Now the big difference in hw-programming: Old: Programm new buffer offset into both sets of _hw_ buffer offset registers. Depending upon the current _sw_ buffer, select the _hw_ buffer and program this into the OCMD register. This just complicates matters unnecessarly. New: Just always use the hw buffer 0. And then it's again the same story in both old and new code: - Execute an overlay flip (MI_OVERLAY_FLIP) to read in the contents of the hw registers into the shadow hw registers (which are actually being used by the overlay, not the ones we write stuff into). This is synchronized with the respective crtc vblank by the hw. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 232418d72ee967a81e39481b9f2ee379cb685458 Author: Daniel Vetter Date: Tue Aug 11 16:06:35 2009 +0200 Xv: use is_planar_fourcc helper some more Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 219b14310efe49aff5d3d9023d2ba440c9f702c1 Author: Daniel Vetter Date: Tue Aug 11 16:06:34 2009 +0200 Xv: introduce planar memcpy helper Reduced 3 copies of the same code to one. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit b90171f16652d1145fb80a63919f76a804a94cef Author: Daniel Vetter Date: Tue Aug 11 16:06:33 2009 +0200 Xv: rename i830_display_video to i830_display_overlay This function only programs the overlay and is never called for textured video. Make this obvious. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 857e40a6e00db72401717da3ed79edbf65db31d4 Author: Daniel Vetter Date: Tue Aug 11 16:06:32 2009 +0200 Xv i830_display_video splitup: extract i830_overlay_cmd This slightly moves around (and simplifies) the OSTRIDE reg programming, too. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 50d70e644ea0574752d0fe160ce8d873653c913b Author: Daniel Vetter Date: Tue Aug 11 16:06:31 2009 +0200 Xv i830_display_video splitup: extract i830_calc_src_regs Also introduce an is_planar_fourcc helper. I'll use that one later. In i830_display_video this changeset moves the XVMC case (previously obscured as the default case) around. I've figured this default case does not make sense, here's why: XvMC is everywhere else handled as a planar format (e.g. in the register programming a few lines down). Furthermore the id variable gets mapped to FOURCC_YV12 if IS_I915(pI830) is true in I830PutImage. There's a second caller in the offscreen overlay support code. But I think that code is bitrotten and not reliable as an information source. So we have a different behaviour only for id=FOURCC_XVMC and i965 class hw (i830 class doesn't have xvmc). I've crawled through various sources/intel documentations. Finally in the textured video implemention for i965 class hw (src/i965_video.c) I've found a switch statement that puts XVMC into the same case as I420 and YV12. So also in i965 class hw xvmc uses a planar format. In conclusion I claim that this code was bogus and XvMC on i965 class hw over Xv overlay was most likely broken. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt commit 2841a4cd8c6c64fdf4b2203ab73b57b38a50f651 Author: Carl Worth Date: Mon Sep 28 20:03:11 2009 -0700 Increment version to 2.9.0 And add a reminder to RELEASING that the incremented version number needs to be committed, (since I forgot to do this with 2.8.99.902). commit a790aff4fef7d0c86b7b9c0da4afdc9c0aba6636 Author: Carl Worth Date: Mon Sep 28 20:02:28 2009 -0700 NEWS: Add notes for the 2.9.0 release. Just renamed notes as from 2.8.99.902. commit a92bbcc94904684e7709b3ddaad82bc04607af26 Author: Carl Worth Date: Tue Sep 22 17:08:23 2009 -0700 Increment version to 2.8.99.902 commit ae00a0923717caec6d0dad94527da26397bd12ce Author: Carl Worth Date: Tue Sep 22 15:14:38 2009 -0700 NEWS: Add notes for 2.8.99.902 commit 7e7db7ac530b5282b0841585959597b54fcc633b Author: Matthias Hopf Date: Mon Aug 17 19:24:03 2009 +0200 Add new backlight driver "samsung". commit 7ae1d0dde6cef3437b67dbc21384cb179616a6c0 Author: Zhao Yakui Date: Mon Aug 31 13:51:01 2009 +0800 Skip setting tv format property if output crtc is NULL When TV is not connected and X start, after plugging TV cable again, system will crash because output crtc is NULL. This patch will return, do not handle crtc immediately, meanwhile set value will be effective until user really enable output by xrandr command. Signed-off-by: Ma Ling Signed-off-by: Zhao Yakui commit 02fe9be695f7e209944bd0f7b67950f93619feee Author: Zhao Yakui Date: Tue Sep 22 10:58:38 2009 +0800 Check whether the DVI-I/D is connected or disconnected based on EDID When the monitor is digital type for SDVO-DVI D, there should exist the EDID. If there is no EDID, it should be detected as disconnected. Signe-off-by: Zhao Yakui commit 762e406d138ac80854b6a23b1078b52f6581f0d8 Author: Chris Wilson Date: Tue Sep 22 01:34:37 2009 +0100 Revert "8xx: Fallback for any non-affine transformation." This reverts commit 505025053d66d415e1c23ac858b9238fa8541d37. In theory, the non-affine paths work -- at least for the stated test case, so re-enable them and avoid the slow work-around. Signed-off-by: Chris Wilson commit 2cc1f3cb6034dddd65b3781b0cde7dff4ac1e803 Author: Keith Packard Date: Sat Sep 19 17:30:57 2009 -0700 i8xx: Format projective texture coordinates correctly. Projective texture coordinates must be delivered as TEXCOORDFMT_3D using TEXCOORDTYPE_HOMOGENOUS. This meant selecting the correct type in i830_texture_setup, the correct format in i830_emit_composite_state and sending only 3 coordinates in i830_emit_composite_primitive. Signed-off-by: Keith Packard [ickle: tweaked to fix up a couple of use-before-initialised] Signed-off-by: Chris Wilson commit 00e8de212b46a243f243b437b7eb866315ab89b3 Author: Chris Wilson Date: Mon Sep 21 22:14:21 2009 +0100 Check the correct Picture for error during creation. commit bd817e2d733dfdb1140874b06595ccd1ef39159b Author: Keith Packard Date: Mon Sep 21 17:21:17 2009 -0700 Split i915/i830 composite_emit_primitive into two functions. The i915 and i830 take similar but different data when emitting the primitives, instead of trying to share code here, just split this apart and avoid potentially breaking things later on. Signed-off-by: Keith Packard commit 5e80297d088e8cdbf66d765f7d252dab66c8df86 Author: Keith Packard Date: Fri Sep 18 21:05:23 2009 -0700 Remove DGA support from the driver. The xf86DiDGA code required that the scanout buffer always be mappable, stay be at a fixed address in the aperture and have a constant size. With frame buffer resizing, the latter two are no longer true, and with KMS, we'd really prefer to not allow the former. The only option available to the driver is to completely disable DGA as the modes code has internal calls to the xf86DiDGA code when fetching new modes from the hardware. A fix for the DiDGA code will be added to the X server which will automatically initialize DGA for mode switching and input, but not frame buffer access, and not require any driver cooperation. Thus, the correct solution is for the driver to not call xf86DiDGAInit at all. For old servers, this eliminates a potential catastrophic problem where random memory is written by the X server. New servers will get the DIX-based behaviour automatically. Signed-off-by: Keith Packard commit 4758311842a16600287c8f9f77ce0af1a31b9264 Author: Keith Packard Date: Fri Sep 18 20:59:52 2009 -0700 Remove vestigial internal rotation which broke KMS DGA/VidMode modesetting. Pre-2.0, the driver supported rotation internally, rather than relying on the X server rotation support. The last piece of this dealt with rotating the mouse coordinates and also tried to preserve rotation across DGA/VidModeExtension modesetting requests. That latter bit of code broke under KMS as the rotation value was never initialized, and when set to zero would create an invalid configuration. This would confuse xrandr which would bail before making any changes, leaving the user without a way to recover. Signed-off-by: Keith Packard commit 505025053d66d415e1c23ac858b9238fa8541d37 Author: Carl Worth Date: Mon Sep 21 13:50:09 2009 -0700 8xx: Fallback for any non-affine transformation. There are definitely bugs in the 8xx code dealing with non-affine transformations. Disable that code for now to get things working. Fixes bug #22947 ([855GM, xf86-video-intel-2.8.0] "Freeze" when RENDER extension is being used) commit b4d29452b929a3ef224d3625e4bc66b787c5edb7 Author: Zhenyu Wang Date: Mon Sep 21 15:28:09 2009 +0800 More dumps for Arrandale LVDS Signed-off-by: Zhenyu Wang commit 57fc09cef28bad2e3e8455b93ef2927118f8a3a3 Author: Chris Wilson Date: Sun Sep 20 01:02:39 2009 +0100 Avoid fallbacks for a1 src/mask Carl Worth did the hard work in identifying that the regression in cairo between X.org 1.6 and 1.7 was caused by cairo sending an a1 mask to the server in 1.7 whereas in 1.6 cairo used local fallbacks (as the source was using RepeatPad, which triggers cairo's 'buggy_pad_reflect' fallback for X.org 1.6). This was causing the driver to do a fallback to handle the a1 mask instead, which due to the GPU pipeline stall is much more expensive than the equivalent fallback in cairo. Reference: cairo's performance downgrades 4X with server master than server-1.6. https://bugs.freedesktop.org/show_bug.cgi?id=23184 The fix is a relatively simple extension of the current uxa_picture_from_pixman_image() to use CompositePicture() instead of CopyArea() when we need to convert to a new format. Signed-off-by: Chris Wilson commit 33f98e4056706f4c30bb4327677ac49e82058231 Author: Keith Packard Date: Thu Sep 17 18:16:02 2009 -0700 Don't destroy bufmgr at CloseScreen time Under KMS, the bufmgr is not initialized at InitOutput time and so it won't be re-initialized during server regen. Thus we must leave the bufmgr running during regen and cannot destroy it in CloseScreen. Under UMS, each place the bufmgr is initialized, it checks to see if it has already happened. Hence, we can safely leave the bufmgr running across server regen for UMS too. Signed-off-by: Keith Packard commit ee9ad853574e3ee1a210db6207b84ecd466e08ee Author: Keith Packard Date: Thu Sep 17 17:57:40 2009 -0700 Make sure DRM output properties are freed after we're done with them. drmmode_output_get_modes was fetching output properties but only freeing some of them. Signed-off-by: Keith Packard commit b8c5c996e888485c3a16d645c8490592534a7882 Author: Jesse Barnes Date: Tue Sep 15 19:45:47 2009 -0700 Eliminate cursor flicker We can update the cursor without hiding and showing it. In fact, doing the hide/show causes noticable flicker when running in KMS mode. Signed-off-by: Jesse Barnes commit c2abfa8e54acab61250dba1e435760e3b1499c8c Author: Chris Wilson Date: Fri Sep 4 17:10:25 2009 +0100 Avoid fallbacks for compositing gradient patterns Currently when asked to composite using a gradient source or mask, we fallback to using fbComposite(). This has the side-effect of causing a readback on the destination surface, stalling the GPU pipeline. Instead, like uxa_trapezoids(), we can use pixman to fill a scratch pixmap and then copy that to an offscreen pixmap for use with uxa_composite(). Speedups on i915: firefox-talos-svg: 710378.14 -> 549262.96: 1.29x speedup No slowdowns. Thanks to Søeren Sandmann Pedersen for spotting the missing ValidatePicture(). Signed-off-by: Chris Wilson commit efbcf29dd1a1ca058b7a2a93f0685102c06c9369 Author: Carl Worth Date: Wed Sep 9 09:32:48 2009 -0700 Update version to 2.8.99.901 for snapshot. commit 54fa27fcd97a6735d8eb6073f5ec7d0d40a7327f Author: Carl Worth Date: Wed Sep 9 09:27:01 2009 -0700 NEWS: Add notes for the 2.8.99.901 snapshot commit 1a77ca74bc829e019a06fe9ad559f013054c27ff Author: Chris Wilson Date: Wed Sep 9 12:09:05 2009 +0100 i915: Restore nearest sampling My recent commit [94fc93] to use the pixel centre for sampling with the i830 broke the i915. This restores the previous sampling coordinates for the i915 whilst preserving the correct coordinates for i830. Fixes: gnome characters disappear http://bugs.freedesktop.org/show_bug.cgi?id=23803 Signed-off-by: Chris Wilson commit ce10b5b6fca086eb4af45c1db28352e06ee4ce0b Author: Zhenyu Wang Date: Tue Sep 8 00:56:34 2009 +0800 Add B43 chipset support Signed-off-by: Zhenyu Wang commit 94fc93d4e2b88565dca17f72903d8991213c9ee8 Author: Chris Wilson Date: Sat Sep 5 09:35:10 2009 +0100 i830/i915: Set the sample position to the pixel center. And in particular we apply the nearest sample bias separately for src/mask. Fixes cairo/test: device-offset-scale finer-grained-fallbacks mask-transformed-{similar,image} meta-surface-pattern pixman-rotate surface-pattern-big-scale-down text-transform Signed-off-by: Chris Wilson commit ced0cc8bb28106c18473777bb184872786325e99 Author: Chris Wilson Date: Sat Sep 5 09:05:36 2009 +0100 i830: Update comments i830_composite() is no longer shared with i915 but i830_emit_composite_primitive() is. Signed-off-by: Chris Wilson commit 8863706e25a73f68993d946e2f1c1040bae0f9c5 Author: Chris Wilson Date: Sat Sep 5 09:03:56 2009 +0100 i830: Trim composite setup Remove a couple of redundant NOOPs from the setup and correct the required space checking for atomic batch operation. commit a9b12111f9787950ad6b8f295a70a872c5933c93 Author: Chris Wilson Date: Sat Sep 5 00:57:39 2009 +0100 i830: remove padding NOOPs from composite Bumps aa10text up from 249k to 260k! These NOOPs have existed uncommented since 04d1584737fd0d14e99608a97281fd7b1549ae0e. Signed-off-by: Chris Wilson commit 9c1bf6d01ca307b7a9b91e181ad7f341862e5e1c Author: Chris Wilson Date: Fri Sep 4 23:31:44 2009 +0100 i830: do not use stale mask transform Not only were incorrectly falling back if we had non-affine transformations, but we made the decision based on a stale transformation matrix. Related bug 22877: batch_start_atomic horribly breaks performance after a while https://bugs.freedesktop.org/show_bug.cgi?id=22877 Signed-off-by: Chris Wilson Tested-by: Maximilian Grothusmann commit e903b3ebad29d943c05ff550940034c3a43e0235 Author: Matthias Hopf Date: Thu Sep 3 16:34:08 2009 +0200 Don't set backlight level if going on->on. Otherwise changed backlight will be reset upon DPMS switch off, because first this function is called with DPMSModeOn. commit 91f26937c1a3a4f33da86aa6b62ab9b288a54e15 Author: Matthias Hopf Date: Tue Aug 18 18:44:38 2009 +0200 Use "Backlight" as backlight brightness property. BACKLIGHT is now deprecated, though still usable. commit dbc8944ee9e6755ab11dfc7375c3394f531a49ce Author: Matthias Hopf Date: Tue Aug 18 18:34:45 2009 +0200 Add get_property support. So far only BACKLIGHT is changed. commit c18fc7955dc2aec878dd3bea1d51aaecf3f08858 Author: Matthias Hopf Date: Tue Aug 18 18:05:45 2009 +0200 Don't ignore backlight level change to the same level. If set externally to a different level, this would result in a no-op. OTOH if the display is switched off (DPMS) you do not want the change to take place immediately, but rather to be saved and set later when the display is active again. commit fe7693c94a396b9d17bb66a000178750a432ce3a Author: Zhenyu Wang Date: Wed Sep 2 22:57:33 2009 +0800 Make DGA optional As DGA is optional in xserver, we should check this too instead of always trying to init DGA. Found when update xserver to 6fffcd5825454a7fe58ffbcfb219f007cf38e731, but not update xf86dgaproto, which caused X fails to start. Signed-off-by: Zhenyu Wang commit 5812531e08147576de776b2dd64e7f94c08eb851 Author: Carl Worth Date: Fri Jul 31 11:20:23 2009 -0700 debug: i830_valid_command: Return invalid for subopcodes with no name Previously the code would always return the count, before ever looking into the _3d_cmds table to see if there was actually a valid command. Thanks to Alan Coopersmith who reported that the code was confusing parfait: https://bugs.freedesktop.org/show_bug.cgi?id=21666 commit 7c48c21b22bf5862c5a35bda1635753cc5a7197c Author: Zou Nan hai Date: Thu Aug 27 09:47:12 2009 +0800 set correct value for indirect access check bound commit 6361c3b9af39265df9222b1f3b6fb9c4197087c1 Author: Keith Packard Date: Tue Aug 25 19:23:27 2009 -0700 Fix SHM functions to work with server after 1.6.0 Signed-off-by: Keith Packard commit 2786a66719a6dbb735eb7c551c412475c30ffa51 Author: Keith Packard Date: Tue Aug 25 18:35:54 2009 -0700 KMS: allocate one bo per crtc for cursor The KMS API doesn't provide for sharing a single bo for multiple cursor images, so allocate one bo for each crtc to hold the cursor image. KMS also only supports ARGB cursors, so don't bother to allocate buffers for two color cursors. Signed-off-by: Keith Packard commit e51126c57132492c664f86981c55b166dbb54c79 Author: Keith Packard Date: Tue Aug 25 18:34:25 2009 -0700 Add cursor registers to debug dump output. Signed-off-by: Keith Packard commit 5fa8d04d9c86f343802c05bd3e11c6e733f01b63 Author: Keith Packard Date: Tue Aug 25 18:33:04 2009 -0700 Reload cursors as needed when setting new modes. Cursor images may need rotation, or positions updated when new modes are set. The server provides a convenience function, xf86_reload_cursors for precisely this purpose. Just call it after the new mode is set. Signed-off-by: Keith Packard commit 1fc3f467ab3edd405adc569ac7f629077e6ffb9d Author: Jesse Barnes Date: Tue Aug 25 09:46:10 2009 -0700 Add KMS only build flag Rather than refactoring all our init code only to have it go away when we remove UMS, this patch adds a build time flag to allow the driver to assume KMS support. With this flag active, the driver will not request that I/O or MEM be enabled at probe time, which can allow the server (if other drivers also cooperate) to run as a non-root user. Signed-off-by: Jesse Barnes commit 5dccd1be3ab80b642ef2022446f5bdc1656ed943 Author: Wu Fengguang Date: Fri Aug 21 14:57:11 2009 +0800 Add HDMI audio registers Dump some of the audio registers at server startup time. (II) intel(0): AUD_CONFIG: 0x00000004 (II) intel(0): AUD_HDMIW_STATUS: 0x00000000 (II) intel(0): AUD_CONV_CHCNT: 0x00000000 (II) intel(0): VIDEO_DIP_CTL: 0x20000600 (II) intel(0): AUD_PINW_CNTR: 0x00000040 (II) intel(0): AUD_CNTL_ST: 0x00002000 (II) intel(0): AUD_PIN_CAP: 0x00000094 (II) intel(0): AUD_PINW_CAP: 0x004073bd (II) intel(0): AUD_PINW_UNSOLRESP: 0x80000008 (II) intel(0): AUD_OUT_DIG_CNVT: 0x00000001 (II) intel(0): AUD_OUT_CWCAP: 0x00006211 (II) intel(0): AUD_GRP_CAP: 0x00000004 Signed-off-by: Wu Fengguang commit 38e97d2366738e83e76c72353b65edb13a9d2c7f Author: Wu Fengguang Date: Fri Aug 21 11:31:13 2009 +0800 Add intel_audio reg dumping program It can dump HDMI audio registers for G45. Signed-off-by: "Wang, Zhenyu Z" Signed-off-by: Wu Fengguang commit 6955fc7a74edf6034a292c31a304577c35e925e6 Author: Kristian Høgsberg Date: Thu Aug 20 16:48:58 2009 -0400 kms: Don't use fb offset when using shadow buffer commit 465a4ab416b2e5ad53b96702720331a44fffa2fe Author: Eric Anholt Date: Wed Aug 12 19:29:31 2009 -0700 Align the height of untiled pixmaps to 2 lines as well. The 965 docs note, and it's probably the case on 915 as well, that the 2x2 subspans are read as a unit, even if the bottom row isn't used. If the address in that bottom row extended beyond the end of the GTT, a fault could occur. Thanks to Chris Wilson for pointing out the problem. commit a3962e6f74ddd954ae1390d150a347745d7bdb24 Author: Jesse Barnes Date: Tue Aug 18 10:56:50 2009 -0700 Print block length of backlight table For debugging VBIOS dumps commit 320f21669900f99a7daf8f2294f37be9ad71d05f Author: Jesse Barnes Date: Mon Aug 17 15:19:41 2009 -0700 Dump LVDS backlight info from bios_reader Add LVDS backlight and power VBT structures and dump from the BIOS reader. commit a50916530426a662f5ed262892080b860a445da3 Author: Matthias Hopf Date: Mon Aug 17 15:53:15 2009 +0200 Add BACKLIGHT property support in KMS case. commit 376397c21eb9a7e4ea79d349af41da81c1af861f Author: Zhenyu Wang Date: Tue Aug 18 10:01:12 2009 +0800 Fix VGA plane disabling Only apply on G4X with SR01 bit5 workaround for VGA plane disable, and restore behavior back for other chips to make sure other modes got disabled too. For bug #17235, #19715, #21064, #23178 Signed-off-by: Zhenyu Wang commit 926c7e7d30458078f8185d4e0b9b32b40102b3d5 Author: Matthias Hopf Date: Thu Aug 13 13:59:55 2009 +0200 Add HP Mini 5101 to quirks list. commit 713820197755ea53003b36a920922c3c525eeeea Author: Adam Jackson Date: Tue Aug 11 14:50:03 2009 -0400 Fix the chip names printed in the log to be less obnoxious. Names taken from pci.ids. Pineview appears to be a platform not a GMCH, so use the G/GM convention to distinguish. commit e8f0763d405a8152c74c28792c52fe12c1d41dd5 Author: Eric Anholt Date: Fri Aug 7 18:24:44 2009 -0700 Fix math in the tiling alignment fix. commit 222b52ef16895823fbf3a0fc0be4eb23b930ed1b Author: Eric Anholt Date: Fri Aug 7 18:05:29 2009 -0700 Align tiled pixmap height so we don't address beyond the end of our buffers. commit 62494407e529cfa68529b7267155a12d75418f21 Author: Zhenyu Wang Date: Thu Aug 6 13:52:54 2009 +0800 Fix typo in bios_reader for invalid pointer cast Fixed locally for af45482a52999b52bf41468c458808e30c100e35, but pushed wrong commit. commit 79b6851148574419389ac8055b0c31b8bdac3ab3 Author: Eric Anholt Date: Wed Aug 5 12:45:16 2009 -0700 Fix sampler indexes on i965 planar video. We only set up one sampler, because all of our sampling is the same. By using a non-zero index for the other two samplers, we'd dereference (likely) zeroed data, resulting in using NEAREST filtering. This was a regression in 40671132cb3732728703c6444f4577467fa9223f which incidentally switched from having 6 samplers to 1. Bug #22895, #19856 commit f4e4c1a8544e264c5a1da02f4e7990a1beecf71e Author: Zou Nan hai Date: Wed Aug 5 15:00:37 2009 +0800 It seems that indirect data upper bound check in STATE_BASE_ADDRESS is not acting like what bspec told on 965gm. G45+ follow bspec, but we have to set it to a large value for 965gm. commit af45482a52999b52bf41468c458808e30c100e35 Author: Zhao Yakui Date: Fri Jul 24 10:44:20 2009 +0800 Calculate the DVO relative offset in LVDS data entry to get the DVO timing Now the DVO timing in LVDS data entry is obtained by using the following step: a. get the entry size for every LVDS panel data b. Get the LVDS fp entry for the preferred panel type c. get the DVO timing by using entry->dvo_timing In our driver the entry->dvo_timing is related with the size of lvds_fp_timing. For example: the size is 46. But it seems that the size of lvds_fp_timing varies on the differnt platform. In such case we will get the incorrect DVO timing because of the incorrect DVO offset in LVDS panel data entry. Calculate the DVO timing offset in LVDS data entry to get the DVO timing a. get the DVO timing offset in the LVDS fp data entry by using the pointer definition in LVDS data ptr b. get the LVDS data entry c. get the DVO timing by adding the DVO timing offset to data entry https://bugs.freedesktop.org/show_bug.cgi?id=22787 Signed-off-by: Zhao Yakui commit 50e2a6734de43a135aa91cd6e6fb5147e15ce315 Author: Dave Airlie Date: Tue Jul 28 18:26:25 2009 +1000 intel: drop RES_SHARED_VGA not needed anymore commit f3387310f312a4a9e1d691974834c3d290c2fa32 Author: Dave Airlie Date: Tue Jul 28 18:27:10 2009 +1000 xserver: fix up for stable build reported by Arkadiusz Miskiewicz commit 9bc0096f9de4b85ca6d6a5db109e49c4364bcee7 Author: Dave Airlie Date: Tue Jul 28 13:55:39 2009 +1000 intel: since driver depends on newer server don't need to wrap this drop resource/RAC interactions commit 9a3b568d62a0b48f4a42ea5377740b2df1af432a Author: Dave Airlie Date: Tue Jul 28 13:32:30 2009 +1000 intel: update for resources/RAC API removal commit 3418c6c16b108e45f67f3c868d28932266f7a0bc Author: Krzysztof Halasa Date: Tue Jul 28 10:47:44 2009 +0800 h/v bias in 3DSTATE_DEST_BUFFER_VARIABLES is 4-bits wide Fixes bug #22370 commit 378445738494663f2ee1615598d4c77b870745af Author: Dave Airlie Date: Tue Jul 28 10:10:13 2009 +1000 intel: remove unneeded includes none of these need the resource includes commit 8084f76d86f048ca5b82da089fffa9665dbbcdd5 Author: Keith Packard Date: Sun Jul 26 13:14:05 2009 -0700 Allow DRM mode setting to include transformations This removes the explicit transform disabling code in drm_set_mode_major. Without a fixed X server, transforms will still be broken, but even a fixed X server can't work around this driver bug. Signed-off-by: Keith Packard commit 9a45ace207199eb40e95dc6d2670b3096e66ecd9 Author: Xiang, Haihao Date: Thu Jul 23 11:09:53 2009 +0800 XvMC: enable XvMC/XvMC-VLD on IGDNG commit 043b4a866ab51acffc52d2d71db3475007747571 Author: Xiang, Haihao Date: Thu Jul 23 11:07:13 2009 +0800 add compiled shader programs for XvMC/XvMC-VLD on IGDNG commit 7684adaa370e8b0bd22e6e7dd1d5fa05f6142bab Author: Xiang, Haihao Date: Thu Jul 23 10:57:22 2009 +0800 Check the version of intel-gen4asm tool in configure.ac commit 7dc95b4f1dfecbeb7e9f4a0c35fda32d020be8fe Author: Xiang, Haihao Date: Fri Jul 24 10:39:05 2009 +0800 XvMC: pin XvMC buffers under KMS. Under KMS, the buffer allocated by i830_allocate_memory isn't pinned anymore. However currently 915 XvMC needs static offsets. Fixes bug #22872 commit 12c5aeca7a3db92d3522d00f5daf338d522e2176 Author: Eric Anholt Date: Thu Jul 16 13:34:09 2009 -0700 8xx render: Add limited support for a8 dests. This improves aa10text performance from 74k to 569k on my 855 laptop. This also causes my 865 to hang on aa10text like it does on rgb10text, thanks to actually hitting render accel. commit 6b7728491c3b771bcba2c7ffd75330c0a0b37f44 Author: Eric Anholt Date: Wed Jul 15 16:38:07 2009 -0700 Only align DRI2 tiled pixmaps to the DRI2 tiled pixmap alignment requirement. This should save significant amounts of memory for glyph and other small pixmap storage. Bug #21387 commit 22f7cbc32b70a89d55c79bbea39fb10c50a310ec Author: Eric Anholt Date: Thu Jul 9 23:56:22 2009 -0700 uxa: Tell the driver when we're just going to immediately map the pixmap. This lets the driver allocate a nice idle buffer object instead of a busy one, reducing runtime of firefox-20090601 on my G45 from 50.7 (+/- .41%) to 48.4 (+/- 1.1%). commit 5ef3db45e059df136162584d00d4b0b511456a33 Author: Eric Anholt Date: Thu Jul 9 19:24:38 2009 -0700 uxa: Skip fill of temporary alpha picture that just gets copied over. This was needed when we were doing the mask computations in this pixmap, but now they're done in a temporary and then uploaded later. This reduces runtime of firefox-20090601 from 52.6 (+/- .96%) to 50.7 (+/- .41%) seconds on my G45. commit 6f3fc6b20f3daedab02e31f49678d4d2ff0fa7a3 Author: Keith Packard Date: Tue Jul 21 12:32:10 2009 -0700 drmmode_output_get_modes: Replace existing EDID property blob with new one This synchronizes the X EDID data with the kernel EDID data each time the kernel data may have changed. Otherwise, X ends up stuck with the first EDID data it sees, failing to accomodate to different monitors. Signed-off-by: Keith Packard commit 840a787a191b31ece5068e10daed04bee70bb1a2 Merge: 0a4c4c5 5d50a94 Author: Carl Worth Date: Mon Jul 20 23:00:06 2009 -0700 Merge branch '2.8' commit 5d50a949b3c5d0ad2bc4cf48ab25da1f707a4f6f Author: Carl Worth Date: Mon Jul 20 22:59:37 2009 -0700 Increment version number to 2.8.0 for release. commit b12220bd81f3a0509a3746dac3258e53f3879b23 Author: Carl Worth Date: Mon Jul 20 22:59:02 2009 -0700 NEWS: Add notes for 2.8.0 release Many thanks to Gordon for his notes from http://intellinuxgraphics.org/2009Q2.html commit 0a4c4c5fe8ebad2dd13f5770bd90a194eebb2890 Author: Peter Hutterer Date: Thu Jul 16 11:40:15 2009 +1000 Update to xextproto 7.1 support. DPMS header was split into dpms.h (client) and dpmsconst.h (server). Drivers need to include dpmsconst.h if xextproto 7.1 is available. SHM is now shm.h instead of shmstr. Requires definition of ShmFuncs that's not exported by the server. Signed-off-by: Peter Hutterer commit 57c7cbade9556e7b21867e61353f0928fd553616 Author: Owain Ainsworth Date: Thu Jul 16 20:38:43 2009 +0100 accessing a pixmap if prepare_access fails is verboten. Don't do it, treat this the same as every other prepare access call in uxa. Reviewed-by: Keith Packard Signed-off-by: Owain Ainsworth commit bb3007384298cb57625ec0b3868dff9b23568f3e Author: Eric Anholt Date: Thu Jul 16 12:56:07 2009 -0700 Really fix i915 render. Fail at commit --amend. commit 8dd7ccf37e2de6d80b556e6d18af244cefc1e417 Author: Eric Anholt Date: Thu Jul 16 11:40:51 2009 -0700 Fix 915-class Render after the 8xx-class Render fix. The two shared i830_composite.c, so giving i830 atomic batch support triggered anger about starting i830's atomic area while in i915's atomic area. Instead, split the emit-a-primitive stuff from the state emission. commit a1e6abb5ca89d699144d10fdc4309b3b78f2f7a9 Author: Eric Anholt Date: Wed Jul 15 14:15:10 2009 -0700 Use batch_start_atomic to fix batchbuffer wrapping problems with 8xx render. Bug #22483. commit e386e7b14b139f15205e14b173e8222bf38d9e18 Author: Keith Packard Date: Wed Jul 15 09:43:04 2009 -0700 Reset framebuffer offset when rebinding aperture (22760). scrn->fbOffset may be changed when binding objects to the aperture during server initialization or VT enter. This was accidentally removed when the NoAlloc option was eliminated. Signed-off-by: Keith Packard commit b74bf3f9a65af9e72921d4e9028d9d4d023f8bc6 Author: Barry Scott Date: Mon Jul 13 16:34:20 2009 -0700 Fix XV scan line calculation when rotated. commit 82905c7c0b871a97ec435a765c2ca407903ba595 Author: Carl Worth Date: Mon Jul 13 05:29:49 2009 -0700 Increment version to 2.7.99.902 commit 925bc6cbd430a00928fac2ef58724dd37c3bc349 Author: Carl Worth Date: Mon Jul 13 05:27:40 2009 -0700 RELEASING: Fix typo in instructions commit 67c0afc7b7446a7b98aa7c65043ddba4c7c72b82 Author: Carl Worth Date: Mon Jul 13 05:27:06 2009 -0700 NEWS: Add notes for 2.7.99.902 commit 34c674dd45879b8ba8395b93b16c8a9e7b848f1f Author: Keith Packard Date: Sat Jul 11 22:53:42 2009 -0700 Remove vestiges of NoAccel options from i830_driver.c The enum and OptionInfoRec weren't removed in the initial patch Signed-off-by: Keith Packard commit 33d6e7a2355dfb8ad324c4fa28ce61c7e051b435 Author: Keith Packard Date: Sat Jul 11 22:53:11 2009 -0700 intel.man: Mark NoAccel option as i810/i815 only The NoAccel option is not valid for other chips. Signed-off-by: Keith Packard commit ed8a9a94e1a670ca35311c9ed83d0c479530d41a Author: Keith Packard Date: Fri Jul 10 17:13:14 2009 -0700 i830_uxa_prepare_access: Flush and wait for idle for non-bo pixmaps Without kernel support and explicit knowledge about where in the ring the last rendering operation for a specific pixmap was, we must synchronize with any outstanding rendering before accessing a pixmap which does not have a buffer object. Signed-off-by: Keith Packard commit cb19ac207b784d814f6f389110fd1b21a0f34e8b Author: Keith Packard Date: Fri Jul 10 14:01:02 2009 -0700 KMS: Keep screen pixmap devPrivate.ptr NULL during init and resize The frame buffer only has a valid address between prepare_access and finish_access calls, so remove all other attempts to compute an address from the driver. Signed-off-by: Keith Packard commit 704b88dd50a7e7e3f362264b86d0401bee8603aa Author: Keith Packard Date: Wed Jul 8 13:06:47 2009 -0700 i830_bind_memory: Under UMS: Bind GEM bos with dri_bo_pin, else through the GART We only need to get static offsets for objects when not running KMS, otherwise the kernel will manage those as needed for us. Binding objects is done in one of two ways. For GEM buffer objects, we use dri_bo_pin. For GART allocated memory, we bind that to the GART. commit 7b273732f70e91df8b41d5c48e1379271557dd8e Author: Keith Packard Date: Wed Jul 8 11:53:13 2009 -0700 Allocate GTT space for GEM only under UMS GEM requires GTT space to map objects. Under KMS, the kernel driver has already provided all available GTT space to GEM, so the X server need not do anything. Signed-off-by: Keith Packard commit 56bfee8705f5d7d965227013b205dbc4c93e220c Author: Keith Packard Date: Fri Jul 10 14:49:20 2009 -0700 Always set screen pixmap data pointer at init and resize times For non-DRM environments, the screen pixmap will be GART allocated memory and not a libdrm buffer object and so uxa will only use devPrivate.ptr to find the associated memory. Make sure devPrivate.ptr is set each time the framebuffer is allocated so that uxa will be able to draw to it. Signed-off-by: Keith Packard commit 98087a0b966d5dc69faf72719153a2c878ba3de1 Author: Keith Packard Date: Wed Jul 8 11:47:25 2009 -0700 Make xorg.conf DRI option work under KMS. Fix name of I830AccelMethodInit KMS mode does not call I830AccelMethodInit as that does the user modesetting initialization (yes, it was misnamed), but that means that the DRI option was ignored. Create a new i830_check_dri_option function to do the option detection, then remove that from I830AccelMethodInit, which is renamed i830_user_modesetting_init to reflect what it actually does. Signed-off-by: Keith Packard commit d655a3ff423e69c19a5dc07140cbf3caaa32cb86 Author: Keith Packard Date: Wed Jul 8 18:06:40 2009 -0700 Remove NoAccel support This removes yet another 'debugging' option that hasn't seen real use in a long time, and wasn't supported under KMS in any case. Signed-off-by: Keith Packard commit 4e4b947f0b13f4a62606ccfd5729d5eb26ca0e92 Author: Eric Anholt Date: Thu Jul 9 15:52:16 2009 -0700 Remove bad comment about 3DSTATE_DRAWING_RECTANGLE size. commit 9155cfca75a207bce0fad945f32f0cb33eab8c4e Author: Eric Anholt Date: Thu Jul 9 14:16:07 2009 -0700 Fix lols in trying to figure out whether this is a 64-bit build. Noticed by: Michel Dänzer commit 40e7c9505265823786cf730214db84812a5e494e Author: Eric Anholt Date: Mon Jul 6 11:54:50 2009 -0700 Refuse to allocate giant BOs on 32-bit systems. The overcommit of address space combined with these buffers hitting SW fallbacks all the time means that we're probably better off telling the application "no" instead of likely silently failing later. Bug #22601. commit 6337cd23e692cae789d07f429442c425c18e1d4f Author: Eric Anholt Date: Wed Jul 1 15:54:38 2009 -0700 Initialize the other argument to drmSetInterfaceVersion. The kernel ignores it if dd_major is -1, anyway. commit 216d939858abc924f2e32c95518f937f29ea018e Author: ling.ma@intel.com Date: Tue Jul 7 14:26:02 2009 +0800 enable sdvo lvds scaling function Currently we implemented basic sdvo lvds function, But except for sdvo lvds fixed mode, we can not switch to other modes, otherwise display get black. The patch intends to work for all modes whose HDisplay and VDisplay are lower than fixed mode. Signed-off-by: Ma Ling commit 0402f4f331148084552bd3963dbcb3fb900be8ea Author: Simon Farnsworth Date: Thu Jun 18 12:33:47 2009 +0100 Raise XV limit to 2048x2048 to match hardware limits. The bigrequests limit isn't present in current X servers (tested using textured video on a 965 with both image and window at 2048x2048 on a 1920x1200 display, and image at 2048x2048, window at 1024x1024). Remove the artificial limit, enabling full-screen HD video when rotated. commit a66357832388ba9db21a4b3bf8311d9d1f1ab308 Author: Keith Packard Date: Tue Jul 7 14:13:57 2009 -0700 non-DRI FB resize failed to assign the screen pixmap devPrivate.ptr (22328) When not using DRI, the screen pixmap is not in a bo, and so the usual enable/disable access functions don't adjust the pixmap devPrivate field, leaving it to the frame buffer allocation code to assign this correctly. During mode setting and fb resizing, FB access is disabled, and the screen pixmap devPrivate is stashed away by xf86EnableDisableFBAccess, to be restored when FB access is turned back on. This means that we have to set the pixmap devPrivate.ptr (in case xf86EnableDisableFBAccess doesn't do this), along with storing the address in the scrn->pixmapPrivate field. Signed-off-by: Keith Packard Reviewed-by: Jesse Barnes commit c889b34e432198d5410a068eff3089ff5314ac27 Author: Chris Wilson Date: Tue Jul 7 18:05:01 2009 +0100 Check for a valid I830Ptr before closing master. After failing to become DRM master, the X server dies attempting to close the master fd during free: (EE) intel(0): [drm] failed to set drm interface version. (EE) intel(0): Failed to become DRM master. (EE) intel(0): failed to get resources: Bad file descriptor (EE) intel(0): Kernel modesetting setup failed Backtrace: 0: X(xorg_backtrace+0x3b) [0x8133a3b] 1: X(xf86SigHandler+0x55) [0x80c7945] 2: [0xb805d400] 3: /usr/lib/xorg/modules/drivers//intel_drv.so [0xb7b4bfcc] 4: X(xf86DeleteScreen+0x6b) [0x80d465b] 5: X(InitOutput+0x548) [0x80b0158] 6: X(main+0x1cb) [0x807220b] 7: /lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xe5) [0xb7d107a5] 8: X [0x8071881] Saw signal 11. Server aborting. ddxSigGiveUp: Closing log ddxSigGiveUp: re-raising 11 Segmentation fault Signed-off-by: Chris Wilson commit 705042f497b7b3843c2dcc5c160fb8dfeac1472a Author: Keith Packard Date: Mon Jul 6 13:49:31 2009 -0700 Handle DRI2INFOREC version 3 This DRI2 version does not support the old CreateBuffers/DestroyBuffers interface anymore. Signed-off-by: Keith Packard commit 2ebc7d32e47b5edd5b776c39f936ed4e053caac2 Author: Keith Packard Date: Thu Jul 2 13:13:14 2009 -0700 Update to multi-API DRI2 interface The DRI2 interface was changed to support both old and new drivers in an API/ABI compatible fashion. This change syncs the intel driver with the new version of the DRI2 API. Signed-off-by: Keith Packard commit f6f79eb629184366b1355743d601129a526da90c Author: Rémi Cardona Date: Mon Jul 6 11:01:31 2009 +0200 remove unused shader program This file is not even referenced by any Makefile.am Acked-by: Zhenyu Wang commit 7e722ada533777c5e9ddf44bb4d770bacf8e13bf Author: Zhenyu Wang Date: Mon Jul 6 16:25:13 2009 +0800 Disable FBC on IGDNG Don't make FBC count for memory allocation. Signed-off-by: Zhenyu Wang commit 74227141923a2f5049592219ab80e8733062a5d9 Author: Barry Scott Date: Tue Jun 23 14:14:50 2009 +0100 Fix segv for clipped movie window When playing a movie that is clipped on its left and right edges the Xorg server will SEGV sometimes. This is because the intel driver ignores the clipping info when it copies the planes out of the XV data. The check for the optimised copy was wrong to ignore the width required. Which leads to too much data being copied by the memcpy. It the source buffer happens to end exactly on a page boundary the server will SEGV. As we reviewed the code we checked the calculation of src1, src2 and src3. The patch includes additional comments to make it clear what the elements of the calculation are. This bug exists in git head and we also see it in 2.4.1. Barry Reviewed-by: Ian Romanick commit c1755599db1d9a20954b84ccc07afd892bb6ac9e Author: Jesse Barnes Date: Thu Jul 2 13:04:51 2009 -0700 Clear the bo on the rotate scratch pixmap Since the scratch pixmap header will be re-used after allocation, we need to clear its bo attachment when we stop using it, otherwise a later user will use a bogus bo. Reviewed-by: Keith Packard Signed-off-by: Jesse Barnes commit 324b4686204feb3a7370eeecaff8ba44635f73ca Author: Daniel Vetter Date: Thu Jul 2 14:15:37 2009 +0200 Xv i830_display_video splitup: extract i830_update_scaling_factors Just moved the code, no other changes. Signed-off-by: Daniel Vetter commit c2410addbfb99fcd7069591d9f387c35ed760522 Author: Daniel Vetter Date: Thu Jul 2 14:15:36 2009 +0200 Xv i830_display_video splitup: extract i830_update_polyphase_coeffs To slightly clean up the implementation of i830_update_polyphase_coeffs, introduce the two small helper functions i830_limit_coeff and i830_store coeffs_in_overlay_regs. Signed-off-by: Daniel Vetter Reviewed-by: Ian Romanick commit 795c11c49cf10525f02127a3629d35378d802fa7 Author: Daniel Vetter Date: Thu Jul 2 14:15:34 2009 +0200 Xv i830_display_video splitup: extract i830_update_dst_box_to_crtc_coords Just moved the code ouf of line. Signed-off-by: Daniel Vetter commit 4100abdf5d208bbcbb4ceabad0572c04221443c9 Author: Daniel Vetter Date: Tue Jun 30 13:12:45 2009 +0200 Xv: kill !textured condition This is in the overlay path and therefore always true. Signed-off-by: Daniel Vetter commit b0df0fe91e2b800ed096f369850aa1af4be2f157 Author: Daniel Vetter Date: Tue Jun 30 13:12:44 2009 +0200 Xv overlay: implement GAMMA5 errata - also ensure that the most significant byte is zero - while I was looking at the code, add the Overlay suffix to SetPortAttribute like in the textured case. Signed-off-by: Daniel Vetter commit 5ef4d3cde1335350d82469ebbaed1b547a59552c Author: Daniel Vetter Date: Tue Jun 30 13:12:43 2009 +0200 Xv: kill unneeded indirection overlay and textured video have the exact same QueryImageAttributes function. Signed-off-by: Daniel Vetter commit 1e4784bf26e3c154f5673f7b5add3ef7af3b1474 Author: Eric Anholt Date: Tue Jun 30 19:52:36 2009 -0700 uxa: Fix segfault on source-only picture usage with FallbackDebug. Bug #22107. commit 7e79fc8aa93df4df37c25cf37ee0ec6c7caca1d9 Author: Zhenyu Wang Date: Tue Jun 30 15:40:34 2009 +0800 Fix 945GM VT switch in UMS Bug #19578. We should set private intel_crtc state according to current, as fail to do so pipe A needs active won't be taken care of. Also make sure pipe swap operation always set during VT switch. Signed-off-by: Zhenyu Wang commit 50392ac8234d643c8a99e1753bdb196c0062a891 Author: Zhenyu Wang Date: Mon May 18 13:53:33 2009 +0800 Load fbcon too if kernel mode setting is checked on If i915 module has already been loaded and kms check is true, it would be nice to load fbcon module too. Signed-off-by: Zhenyu Wang commit 058ad9e6ab4a00ff66046d94c2129056011ebee9 Author: Jesse Barnes Date: Tue Jun 30 16:02:19 2009 -0700 Don't try to pin buffers in KMS mode The only things we try to pin in KMS mode are the cursor objects and front buffer, and those are taken care of by the kernel anyway, so we shouldn't even bother trying to pin them (well, not entirely true, XvMC tries to pin as well, but it needs work w/KMS anyway). Signed-off-by: Jesse Barnes commit cec9fc6f6cffce186606f39982d0d78ff7c63bbf Author: Jesse Barnes Date: Tue Jun 30 12:27:59 2009 -0700 Make KMS set_resource function return TRUE This is what's expected by the server, and allows the EDID for example to be exported in the KMS case. Signed-off-by: Jesse Barnes commit 53e248af4365232416d2143a791a07c6751f8319 Author: Jesse Barnes Date: Tue Jun 30 12:00:28 2009 -0700 Use DVO timing block instead of fp_timing when parsing LFP data The KMS side was correct, but the UMS patch was broken. We need to use the DVO timing block of the LFP data to get the timing, not the fp_timing block. Fixes fdo bug #22529. Signed-off-by: Jesse Barnes commit 362883c2f9a3442f7678e6e815f41b21baaa3f53 Author: Jesse Barnes Date: Mon Jun 29 16:26:15 2009 -0700 Add a few error messages for DRM initialization Makes it easier to see where things go wrong. Signed-off-by: Jesse Barnes commit afac333bef4a0ac934f0e4d933dc5053d81ca88c Author: Zhenyu Wang Date: Tue Jun 30 11:10:52 2009 +0800 Remove unused packed yuv sampler shader programs Signed-off-by: Zhenyu Wang commit 150c9adebc0b502f19c970783f411928e5a5c3b1 Author: Zhenyu Wang Date: Tue Jun 30 10:42:41 2009 +0800 Xv: fix domain usage for binding table on i965+ chips Signed-off-by: Zhenyu Wang commit 7770958e42d60966d95702e75c704e20ae2d90dd Author: Zhenyu Wang Date: Tue Jun 30 10:41:26 2009 +0800 Add XV support on IGDNG This brings necessary change for IGDNG for texture video support from 2D render code. Signed-off-by: Zhenyu Wang commit f806fe7d675b966680a63406167ce37e4f8a1ae8 Author: Xiang Haihao Date: Thu Jun 25 15:39:32 2009 +0800 Enable 2D composite on IGDNG This patch enables 2D composite on IGDNG. IGDNG requires new compiled shader programs for Gen5 and some command changes. The most notable is the layout of vertex element has changed, but we tried to keep it as origin to not change shader programs. Also vertex buffer state requires end address of vertex buffer instead of origin max index. Signed-off-by: Zhenyu Wang commit 9fb34012f667e37f480085696ef9c2632d6eb7e1 Author: Zhenyu Wang Date: Thu Jun 25 15:30:04 2009 +0800 Add new compiled shader program for IGDNG Also check intel-gen4asm tool here for new -g option, which is required to compile new programs. Signed-off-by: Zhenyu Wang commit 488acc4595bb7f40130afcb8bcb05656ff3ae82c Author: Zhenyu Wang Date: Thu Jun 25 14:05:40 2009 +0800 Move shader programs under its own subdirectory Signed-off-by: Zhenyu Wang commit 170cae0c8d58fc141de1d8a2f17a4328d39c1263 Author: Jesse Barnes Date: Mon Jun 29 12:55:43 2009 -0700 Only get the VBIOS in non-KMS mode In KMS mode, the kernel takes care of this for us, so don't bother. commit 6511c082459789cf279e7a4528775a1c821ad8cc Author: Jesse Barnes Date: Mon Jun 29 12:53:51 2009 -0700 Output error info if we fail to get DRM resources Useful for debugging. commit 00eb73286c3512a362dce00efdeae740772d0dcd Author: Jesse Barnes Date: Mon Jun 29 09:12:28 2009 -0700 Use swapbuffers_wait control Commit 1eec83a203c48822400742a1fb184b2cb52c62f7, which added the new SwapbuffersWait option, didn't actually include the code which used it. So add a test to DRI2's CopyRegion call, only emitting the scanline wait command if the swapbuffers_wait option is set. Signed-off-by: Jesse Barnes commit f53b3239dbc0ed723774e386e07ac9d8ce96bb89 Author: Zhenyu Wang Date: Thu Jun 25 10:22:23 2009 +0800 Disable XvMC on 915G/GM in KMS These chips require physical address for XvMC surface, which is not available in KMS case. Instead of crashing X, disable it now. Signed-off-by: Zhenyu Wang commit f0270bbb47baed78a0ff6189ae20d3ac322ec02b Author: Jesse Barnes Date: Wed Jun 24 14:42:08 2009 -0700 Fix i830_crtc_on to only check outputs associated with the given CRTC Otherwise we may end up returning a false positive if some other output & crtc are on, but not the one in question, again leading to hangs. Reported-by: Eric Anholt Signed-off-by: Jesse Barnes commit 5d80e24b5fc6d6028028da6ded35389c08bfce29 Author: Jesse Barnes Date: Wed Jun 24 13:12:25 2009 -0700 Treat disabled CRTCs as "not covering" for scanline wait purposes Now that swapbuffers does a scanline wait to avoid tearing, it's important to take into account the CRTC status to avoid hangs. If we do a scanline wait when the CRTC is off (due to DPMS for example) we'll hang the GPU. So add some code to check the CRTC DPMS status to the i830_covering_crtc function, returning NULL if none of the covering CRTCs are actually active. KMS vs UMS logic is hidden in new i830* functions, cleaning up both DRI2 & video paths a bit. Fixes fdo bug #22383. Signed-off-by: Jesse Barnes commit 6d025e679a99778496576af9a6a6fa8c043ae811 Author: Alan Coopersmith Date: Tue Jun 23 09:53:14 2009 -0700 Harden i830 render in case check_composite didn't throw out bad formats. Fixes a warning in a static analysis program, and the code's a little clearer. Bug #21667 commit 1eec83a203c48822400742a1fb184b2cb52c62f7 Author: Jesse Barnes Date: Tue Jun 23 15:05:03 2009 -0700 Add option to control swapbuffers behavior Until we get triple buffering, we'll want this so users can avoid taking a performance hit on apps that render slower than the refresh rate. Fixes fdo bug #22234. Signed-off-by: Jesse Barnes commit bfeeac6de096256fca82244338bb45d53ee53cbc Author: Zou Nan hai Date: Tue Jun 23 11:31:37 2009 +0800 i915 xvmc, fix fd.o bug #22103 commit 15af8ea6ab6998bbab9f4eeda227565c409da229 Author: Jesse Barnes Date: Mon Jun 22 11:11:06 2009 -0700 Fix LFP data block fetch Apparently the proper way to do this is to use the LFP data pointer block to figure out the LFP data block entry size, then use that plus the panel index to calculate an offset into the LFP data block array. Fixes fdo bug #19450. Signed-off-by: Jesse Barnes commit 534e73ad4f234a04755917f2bf17ba821c27eb52 Author: Zhao Yakui Date: Thu Jun 18 09:46:32 2009 +0800 Don't change the blank/sync width when calculating scaled modes Don't the change the blank/vsync width while doing LVDS scaled modes. And use the border instead of border minus one. At the same time, make sure the horizontal border and hsync are even for the LVDS that works in dual-channel mode. So both horizontal border and hsync start are also changed to be even, even for the LVDS in single-channel mode. https://bugs.freedesktop.org/show_bug.cgi?id=20951 Signed-off-by: Jesse Barnes Signed-off-by: Zhao Yakui commit b5cd2130f97591f4a387db1b98c940c30bc6404c Author: ling.ma@intel.com Date: Tue Jun 16 16:52:25 2009 +0800 TV: Set correct voltage level override values We detect TV connect status by setting DAC voltage level override values as 0.7 voltage for DAC_A/B/C. The corresponding 2-bits shold be 0x2, In order correctly to set last bit as 0, at first we must clean it. It fixed freedesktop.org bug #21204 Signed-off-by: Ma Ling commit a6cdcd9fee0164f79075063fd163d142a55fcbc5 Author: ling.ma@intel.com Date: Tue Jun 16 16:47:52 2009 +0800 Set hot plug interrupt to detect HDMI output We detect HDMI output connection status by writing to HOT Plug Interrupt Detect Enable bit in PORT_HOTPLUG_EN. The behavior will generate an specified interrupt, which is caught by audio driver, but during one detection driver set all Detect Enable bits of HDMIB, HDMIC and HDMID, which generate wrong interrupt signals for current output, according to the signals audio driver misunderstand device status. The patch intends to handle corresponding output precisely. It fixed fredesktop bug #21371 Signed-off-by: Ma Ling commit 6b93afc564a5e74b0eaaa46c95f557449951b3b9 Author: Bryce Harrington Date: Wed May 27 03:40:52 2009 -0700 add pipe a force quirk for Dell mini Add quirk to solve issue with black screen and hang occuring after closing the lid with attached external monitor, on Dell Mini. Fixes fdo bug #21960. Signed-off-by: Bryce Harrington commit eb09014ce10428bbcab04e155186382975545f0a Author: Li Peng Date: Tue Jun 16 15:29:57 2009 -0700 don't wait for vblank on rotated displays We may hang or wait for the wrong line if the display is rotated, so just skip the wait in that case. Fixes fdo bug #22196. commit 6c56521bdc0443c0656271caaa795feb13bc1d6b Author: Bryce Harrington Date: Wed May 27 05:18:53 2009 -0700 pipe-a quirk for thinkpad x30 Fixes freeze when closing lid on ThinkPad X30. Fixes FDO bug # 21976. Signed-off-by: Bryce Harrington commit 246cec965958e94babf5377e6f221522b05fb458 Author: Zhao Yakui Date: Wed Jun 10 11:17:28 2009 +0800 Fix EDID for LVDS output device to add the default modes Fix the EDID for the LVDS output device to add the default modes.This is similar to what we have done in UMS mode. a. When there exists the EDID, either find the DS_RANGES block or replace a DS_VENDOR block, smashing it into a DS_RANGES block with open refresh to match all the defaults modes. b. When there is no EDID, we will construct a bogus EDID and add a DS_RANGES block with the open refresh to match all the default modes. http://bugs.freedesktop.org/show_bug.cgi?id=20801 http://bugs.freedesktop.org/show_bug.cgi?id=21094 http://bugs.freedesktop.org/show_bug.cgi?id=21346 http://bugs.freedesktop.org/show_bug.cgi?id=21417 http://bugs.freedesktop.org/show_bug.cgi?id=21671 Signed-off-by: Zhao Yakui commit 5d1dc7677004d445a7a781decd8c1ef9747c14fb Author: Zhao Yakui Date: Wed Jun 10 11:17:27 2009 +0800 Get the LVDS panel limit and check whether the given modeline is valid When the connector type is LVDS, it will traverse the mode list returned by KMS kernel to get the LVDS panel limit. Then it will use the panel limit to check whether the given modeline is valid. If the given modeline exceeds the LVDS panel limit, it will be invalid. http://bugs.freedesktop.org/show_bug.cgi?id=20801 http://bugs.freedesktop.org/show_bug.cgi?id=21094 http://bugs.freedesktop.org/show_bug.cgi?id=21346 http://bugs.freedesktop.org/show_bug.cgi?id=21417 http://bugs.freedesktop.org/show_bug.cgi?id=21671 Signed-off-by: Zhao Yakui commit 115e28639fbf6a1eba636dafac02fadd83036c75 Author: Zhao Yakui Date: Wed Jun 10 11:17:26 2009 +0800 Add the private data for the LVDS connector Add the private data when the connector type is LVDS. We can use the private_data to store the LVDS panel limit. For example: Hdisplay, Vdisplay. http://bugs.freedesktop.org/show_bug.cgi?id=20801 http://bugs.freedesktop.org/show_bug.cgi?id=21094 http://bugs.freedesktop.org/show_bug.cgi?id=21346 http://bugs.freedesktop.org/show_bug.cgi?id=21417 http://bugs.freedesktop.org/show_bug.cgi?id=21671 Signed-off-by: Zhao Yakui commit 3da549f5b350dd42516e5cb3576f7fefe012d95e Author: Zhao Yakui Date: Wed Jun 10 11:37:03 2009 +0800 GFX: Initialize the SDVO device based on the valid slave address When the slave address is found for the SDVO port, the SDVO device will be initialzied. When the slave address is not found for the SDVO port, it will return the slave address by using the following flowchart: a. If the SDVO device info is found for another SDVO port, it will return the slave address that is not used. For example: if 0x70 is used, then 0x72 is returned. b. If no SDVO device info is found for another SDVO port, it will return 0x70 for SDVOB and 0x72 for SDVOC. http://bugs.freedesktop.org/show_bug.cgi?id=20429 Signed-off-by: Zhao Yakui commit 51b87b9913ba164d4d5de79e558233915b37a0a5 Author: Zhao Yakui Date: Wed Jun 10 11:37:02 2009 +0800 parse general definition block to get the SDVO device info The general definition block contains the child device tables, which include the child device info. For example: device slave address, device dvo port, device type. We will get the info of SDVO device by parsing the general definition blocks. Only when a valid slave address is found, it is regarded as the SDVO device. And the info of DVO port and slave address is recorded. http://bugs.freedesktop.org/show_bug.cgi?id=20429 Signed-off-by: Zhao Yakui commit 6ba148bbd78783f59eed3d898638c39b950dcd89 Author: Zhao Yakui Date: Wed Jun 10 11:37:01 2009 +0800 Dynamically get the number of child device in general definition block The size of general definition block varies on different platform/machines. In such case the number of child device is also different. And it will be better to get the number of child device in general definition block dynamically. The number of child device can be calculated by the following formula: (block_size - block_header_size) / sizeof( struct child_device_config) http://bugs.freedesktop.org/show_bug.cgi?id=20429 Signed-off-by: Zhao Yakui commit cdbf84f20295c8a78624318aa6fdfff3f5c8ce27 Author: Keith Packard Date: Thu Jun 11 13:57:09 2009 -0700 intel_batch_init: test have_gem, not directRenderingType Under KMS, directRenderingType will get set to DRI_NONE during driver initialization. When the first batch buffer is allocated, as directRenderingType is DRI_NONE, the GEM bufmgr would get trashed as intel_batch_init called a fake-bufmgr specific function. Signed-off-by: Keith Packard commit 6d062e9e9fc4ac92a5dc2941b4ed0251a1dc6e11 Author: Julien Cristau Date: Thu Jun 11 17:15:04 2009 +0200 Fix xvmc build If we don't find xext.pc, disable xvmc instead of failing configure Also add dependencies on xfixes and dri2proto (src/xvmc/dri2.h includes and ). commit 9d3c3b056c1d4da6ef7afd3f71a5994fad810632 Author: Carl Worth Date: Wed Jun 10 15:30:45 2009 -0700 Increment version to 2.7.99.901 This is the first release candidate for 2.8.0 commit e2622045155f5aa79bacbec8f81c1231194f3033 Author: Carl Worth Date: Wed Jun 10 15:30:09 2009 -0700 NEWS: Summarize new developments for 2.7.99.901 commit e5bfa2702a31566fa94fa75f7289d7cbe9825420 Author: Lukasz Kurylo Date: Wed Jun 10 07:55:31 2009 -0700 DRI2 copyregion: don't wait for scanlines that won't happen In some configurations, it's possible to wait for a scanline outside of a given CRTC range. Make sure that can't happen to fix multihead cases with dead space. Fixes fdo bug #22203. Signed-off-by: Lukasz Kurylo commit cda5561716891484aa6f1c4a234be8d88f17f659 Author: Zou Nan hai Date: Wed Jun 10 16:43:30 2009 +0800 add missing files for previous commit commit 6d500be3cdecd7617dc6119b58ae273eea478af7 Author: Zou Nan hai Date: Wed Jun 10 15:53:24 2009 +0800 enable MC on g45 for player without vld patch commit 88da9b48737229b305da4423e9aa43a09ccdcfac Author: Zhenyu Wang Date: Wed Jun 10 10:45:51 2009 +0800 Add new chipset register dumps Signed-off-by: Zhenyu Wang commit 0d56ef94be0592aec9aa83e4a5e4ff13348640c2 Author: Zhenyu Wang Date: Fri Jun 5 12:59:07 2009 +0800 Disable composite on new chipset for now This depends on updated intel-gen4asm be ready for new chip. Signed-off-by: Zhenyu Wang commit 440ccc44f51d3a5d6f46c28cfcc576cad155fbbc Author: Zhenyu Wang Date: Fri Jun 5 12:56:04 2009 +0800 Add new register definitions Signed-off-by: Zhenyu Wang commit 0d8a9e2c6f58115b9b8449de52e795699ed032af Author: Zhenyu Wang Date: Fri Jun 5 12:51:51 2009 +0800 Update intel_gtt utility for new chipset GTT check Signed-off-by: Zhenyu Wang commit fb524caa3e2f1f516717669642bb4b2244f9e7e4 Author: Zhenyu Wang Date: Mon Jun 8 10:22:14 2009 +0800 Remove fixed MMIO size Use pci resource size instead, which will get the correct MMIO range. New chipset uses obviously larger MMIO range. Signed-off-by: Zhenyu Wang commit 4f40b33ef4b069b18a6a18406da83a23ca6e1127 Author: Zhenyu Wang Date: Fri Jun 5 11:57:57 2009 +0800 Add new chipsets PCI ids Desktop and mobile version of new chipsets are added. Also do memory config like Intel 4 series chipset. Signed-off-by: Zhenyu Wang commit accdbd23676d812d2345f86d8e3ee62f108841ff Author: Carl Worth Date: Fri May 29 15:34:20 2009 -0700 UXA: Rasterize trapezoids to system memory, not a pixmap Since we're only doing software rasterization right now, anyway, it makes more sense to just rasterize to system memory and then upload to a pixmap once complete. This avoids expensive read-modify-write cycles. This results in a 2.4x speedup for a real-world test case that's heavy on trapezoids, which is swfdec running on the following file: http://michalevy.com/wp-content/uploads/Giant%20Steps%202007.swf Many thanks to Chris Wilson for his cairo-traces repository and cairo-perf-trace tool which makes it so easy to measure things like this. commit b5e32c9cf896a0b93d193d797a8e83b4aa4691fb Author: Carl Worth Date: Fri May 1 13:51:39 2009 -0700 Rename i830_exa.c to i830_uxa.c It was just confusing otherwise, (since the EXA code has all been removed now). commit ee539e58c3bf39766c560d625f6e4158d419e64e Author: Kristian Høgsberg Date: Mon Jun 8 21:14:14 2009 -0400 Fall back to fb pixmaps for backing glyph pictures GEM pads pixmaps to 512 byte stride and backs them with a kernel side buffer objects. We typically don't render out of glyph pictures, so we're incurring a lot of overhead per glyph by allocating a GEM pixmap per glyph. By looking at the usage hint, we can fall back to fbCreatePixmap for pixmaps backing glyph pictures, which gives us a nice tight malloced pixmap. The fast path for text rendering is compositing from the glyph cache pixmap to the destination, which shouldn't be significantly affected. Quick bit of testing: (firefox-20090601) xlib-rgba-before 384512.49: 1.01x xlib-rgba-after 389633.94: 1.00x The difference being within the margin of error for the benchmark. Signed-off-by: Eric Anholt Tested-by: Chris Wilson commit 4698b3bd79452ae2066a3d195cf58dd5e30c93f3 Author: Xiang, Haihao Date: Fri Jun 5 10:43:50 2009 +0800 xvmc: don't include xf86dri.h commit 4d9f3a21ab9b2737fe7a5a088c2da1b0c2be7f04 Merge: b8e360b 52054b6 Author: root Date: Thu Jun 4 10:27:41 2009 +0800 Merge branch 'xvmc-vld' commit b8e360bf2b77d28559d15a7c0f9c766848eb6ced Author: Eric Anholt Date: Wed Jun 3 10:12:25 2009 +0000 Fix segfault in DRI2 vblank syncing if the region isn't onscreen. Also, fix some weirdness in the checking for whether the target was the screen. commit 5901a67fc85ac80fabfa98b78202a388445275c3 Author: Eric Anholt Date: Wed Jun 3 10:00:50 2009 +0000 Fix some drawable abuse in i830_dri.c We were casting pixmaps to drawables even if they weren't pixmaps. They did happen to work out since we only used the drawable record out of them. commit 5e48146777921b693c828af9566a77422cc4f85f Author: Eric Anholt Date: Wed Jun 3 09:57:10 2009 +0000 Replace some pixmap-related idioms with inline functions to do so. commit ea0b00e675281b2914450992501566122f9affe0 Author: Ma Ling Date: Mon Jun 1 17:13:53 2009 +0800 Provide mode line from VBT to xorg. If sdvo get modes function can't find modes from EDID, transmit sdvo lvds modes from VBT to xorg. Signed-off-by: Ma Ling commit 9e7ee50bea5f65003f1d70cc06d8c1ace1282548 Author: Ma Ling Date: Mon Jun 1 17:13:22 2009 +0800 Fetch mode line from VBT, then keep it. Parse SDVO LVDS option section, then according to panel type fetch fixed mode line from SDVO LVDS DTDS section . Signed-off-by: Ma Ling commit 88f766be008008d76c150e3ac16f09d4ecbb6d53 Author: Ma Ling Date: Fri May 15 15:22:11 2009 +0800 Wait doubled regis to be stable for load pipe detection We have two approaches for VGA detections: hot plug detection for 945G onwards and load pipe detection for Pre-945G. load pipe detection will get one free pipe ,and set border color as red and blue, then check CRT status by swf register. Because pipe registers in hires mode are double buffered, once set force border bit in pipeconf register, we have to wait for a vblank until it is effective, otherwise result is unstable. It fixed freedesktop bug #20463 Signed-off-by: Ma Ling commit 6a635be5ea2b2df9d6b9a8006325244394038474 Author: Ma Ling Date: Fri May 15 14:01:18 2009 +0800 Set activation periods as 64 cdclk sync up with KMS We use force CRT detect trigger bit(1 << 3) to detect VGA in hot plug mode, which triggers a CRT hotplug/unplug detection cycle independent of the interrupt enable bit(1 << 9), so keep bit 9. And although spec says CRT_HOTPLUG_ACTIVATION_PERIOD_64(1 << 8) is only useful for mobile platform, it is also required to detect vga on G4x platform correctly. Tested the patch on G45/G43/Q45 platforms with no regressions It fixed freedesktop.org bug #21120 and part of bug #21210. Signed-off-by: Ma Ling commit ec2fde7c8250fdc30984f16c8a1d3587d70b0144 Author: Jesse Barnes Date: Tue Jun 2 16:42:56 2009 +0100 Sync DRI2 CopyRegion to vertical retrace Akin to the Xv code, wait for the scanline to be outside the range to be copied by the DRI2 CopyRegion hook. Fixes fdo bug #20664. Reviewed-by: Eric Anholt Signed-off-by: Jesse Barnes commit 704771f1c595df481d244a6f074e9f5620d4a720 Author: Adam Jackson Date: Thu May 28 15:03:45 2009 -0400 Remove useless loader symbol lists. commit 8e942b70cb9a784b3f1311affd6fc74c4bcf68bb Author: Carl Worth Date: Thu May 21 13:12:52 2009 -0700 Revert "Rely on BO pixmaps being present in acceleration paths." This reverts commit 4653a7db622ad54a3182d93c81331765d930db34. Eric was getting a little too ambitious about our brave, new world. We do still want the driver to work with old, non-GEM kernels after all. commit 1a039f4371bec455cad43f0fb7b329f2ee09a974 Author: Eric Anholt Date: Mon Apr 27 17:45:02 2009 -0700 Fold GEM detection into DRM master open. We don't have anything to do with the DRM unless it's GEM-enabled, unless we were to support GEM-but-not-DRI2, which doesn't seem useful. Compilation fixes by Carl Worth commit a04a51c9bb6066454e0fda3c7897f97dab436358 Author: Eric Anholt Date: Mon Apr 27 17:29:36 2009 -0700 Open the DRM and keep the handle throughout server startup to finish. This will let us configure the server from start to finish with the most pertinent information available (KMS vs UMS, DRI2 vs non-DRI). Also, we now close the DRI2 fd at terminate, which we didn't before. This duplicates some code from DRI1 for getting a master FD like I'd done in DRI2, but given that we weren't loading DRI1 ourselves, this is also a bogosity cleanup, and avoids allocating the extra DRI1 private. commit c3bf8b980134a2761701e4bc18235695a1cb07a4 Author: Eric Anholt Date: Thu May 21 11:05:01 2009 -0700 Fix backwards logic on whether to sync to vblank or not. Thanks to Michel Dänzer for catching it. commit ad2128825ba28551cfef203da017151e2eac32ef Author: Kristian Høgsberg Date: Wed May 20 12:32:10 2009 -0400 Only return FALSE when dri_bo_map() fails Small typo in the previous commit. commit 09beee378cecd1079e7a9fa6eee8f084d680d37e Author: Eric Anholt Date: Mon May 18 18:01:05 2009 -0700 Don't do GTT maps on objects bigger than half the available aperture size. The basic problem is that software fallbacks will do single instructions that copy from one GTT-mapped BO into another GTT-mapped BO. If we can't get both of them bound simultanously, we fault one in, retry the instruction, fault the other in (kicking out #1), retry the instruction, fault #1 back in (kicking out #2), etc. Note that we'll still get into a nasty spot if you do a composite operation with a mask where all 3 are big-but-less-than-half-available-aperture, where you'll thrash. It at least means you'll make progress, though, since each instruction will only be operating on two BOs at at time, and the situation seems unlikely. Bug #20152 (3/3) commit 52054b6a4c1ca5117c9750361f71aedd91220c39 Author: Li Shao Hua Date: Tue May 19 16:27:32 2009 +0800 switch XvMC to gem commit 34660fd2df5d61b77ed7041d32ac29053fc94f5a Author: Eric Anholt Date: Fri May 15 23:21:05 2009 -0700 Only sync XV to vblank when drawing to the frontbuffer. This fixes emitting syncs to random pipes with boxes bigger than that pipe's vertical, leading to GPU hangs. Bug #21738 commit b622860429e00d6ab4407980232659c283a8fe8d Author: Xiang Hai hao Date: Mon May 18 17:12:30 2009 +0800 DRI2 for XvMC commit 50c10ededba15dd0c118f1b65756362061491090 Merge: 13c730e 87332a7 Author: Zou Nan hai Date: Mon May 18 17:11:28 2009 +0800 Merge branch 'master' into xvmc-vld commit 87332a7cc16af82aa47e07fbf90da3635b071dbf Author: Keith Packard Date: Mon May 11 13:52:00 2009 -0700 Use drm_intel_bo_disable_reuse API to flag scanout and cursor buffers Buffers referenced by the kernel for scanout or cursor display should not be reused by the driver. Use the new drm API to disable reuse of these buffers. Signed-off-by: Keith Packard commit ebe05200df381c0e6ee636f0f83440bfedea9bcb Author: Jesse Barnes Date: Fri May 15 08:50:17 2009 -0700 Add --enable-debug flag to configure Defaults to enabled. commit 04772b6c09a88f0483c2a7efc48029967c77b9bc Author: Keith Packard Date: Thu May 14 16:57:11 2009 -0700 If DRM can't figure out which pipe to sync on, then don't sync at all. Syncing to the wrong pipe can wedge the hardware if the exclusion area is larger than the vtotal Signed-off-by: Keith Packard commit 128c1c3b7d57b157604788f82bf9fd389839068f Author: Carl Worth Date: Wed Apr 29 14:43:56 2009 -0700 Use libdrm to lookup pipe for tear-free sync of XV Previously, the code was trying to examine a driver_private field, but those fields are only set by the userland-modesetting code so would fail in the case of KMS. This fixes bug #21076: [945GME] [KMS] XV_SYNC_TO_VBLANK does not prevent tearing of xv video https://bugs.freedesktop.org/show_bug.cgi?id=21076 commit 2572fcc6196aff7a2f1095d211fd85d8668647ca Author: Wu Fengguang Date: Thu Apr 30 19:51:26 2009 +0800 README: kill an evil dot To make copy and paste a more pleasure. Signed-off-by: Wu Fengguang Signed-off-by: Eric Anholt commit b9462516d18bc57be5f33f57adb6c3e8beede5ff Author: Eric Anholt Date: Tue May 12 18:53:27 2009 -0700 Remove dead "avail" variable from XAA stuff. commit 1c68bc376a9cb3c0a010c8e28f69a776755c8f64 Author: Jesse Barnes Date: Wed May 13 13:19:53 2009 -0700 Add new have_gem flag Prior to this patch, code that wanted to check whether GEM was present would look at pI830->memory_manager. This turned out to be occasionally problematic in the KMS case, since memory_manager didn't always get set correctly. So add a new pI830->have_gem flag to make things clear in the various code paths, and set it after GEM initializes or when KMS is detected. Reviewed-by: Eric Anholt Tested-by: Magnus Kessler Signed-off-by: Jesse Barnes commit 52367847087206b92f18c40d356d36ab9ee89d39 Author: Keith Packard Date: Thu May 7 14:58:02 2009 -0700 Load i915 and fbcon when checking for KMS Signed-off-by: Keith Packard Signed-off-by: Eric Anholt commit ff7494b4c4b1bd8bb6f169402a9edbe9780787bb Author: Keith Packard Date: Thu May 7 14:58:01 2009 -0700 Add DP link and GMCH M and N registers Signed-off-by: Eric Anholt commit e54a23bff068416ccbdb75d538dc7dcd40a6c95c Author: Keith Packard Date: Thu May 7 16:35:19 2009 -0700 Fallback when VT inactive While the VT is inactive, pI830->batch_bo will be NULL, so use that as a simple check for when to not use the accelerator. The alternative is to ignore VT switch and just keep drawing, which would also be fine. Bug #21468. Signed-off-by: Keith Packard [anholt: Removed extra return FALSE -- I830FALLBACK does that.] Signed-off-by: Eric Anholt commit 660bd745c21c56e330d844fe4c8eaae862a77483 Author: Wu Fengguang Date: Thu Apr 30 18:42:18 2009 +0800 Remove the offset parameter from i830_allocator_init() offset is redundant. i830_allocator_init() is only called in one place with offset=0. Acked-by: Magnus Kessler Signed-off-by: Wu Fengguang Signed-off-by: Eric Anholt commit b1f5cc3349f6811ff4d8dc64cc291788d6726372 Author: Carl Worth Date: Tue May 12 11:45:03 2009 -0700 RELEASING: Better instructions for where to send announcements The xorg-announce list doesn't need to hear about development snapshots and release candidates. The intel-gfx list is good for that. commit f16ee218845ec48940ea457b921d34896d80a807 Author: Alan Coopersmith Date: Sun May 10 16:25:24 2009 -0700 Fix "Unkown" typo in two FatalError messages Signed-off-by: Alan Coopersmith commit 8d27247829fe4f55691ce68f9f4b14810fb34b32 Author: Stijn van Drongelen Date: Tue May 5 16:37:06 2009 +0200 Fix typo in (unused) INTEL_BIOS_32 macro Debian bug#527062 [jcristau: fix same typo in bios_reader.c] Signed-off-by: Julien Cristau commit a8a771a853478e5f45f71d0eff3c4d55bf24d0ad Author: Arkadiusz Miśkiewicz Date: Wed Apr 29 00:21:18 2009 +0200 Restore CFLAGS after tests. Reverts single line of broken commit e5133a94dc78411eaef324131ea8056aeb81779e. Signed-off-by: Eric Anholt commit 8255cca2c9092f7ecb798944aa8f03fa3efcfa6c Author: Keith Packard Date: Fri May 1 18:34:43 2009 -0700 Split i915 textured video commands to fit into batch buffers. i915 textured video commands are quite long, but must be contained in the same batch buffer as the 3D setup commands. When the number of clip rects for the video becomes too large for the associated commands to fit in the same batch buffer, this change breaks the sequence into pieces, ensuring that each batch contains the necessary setup sequence. Signed-off-by: Keith Packard commit e5e0fb846bda3e1757b89e50d5244d28457b9fe3 Author: Keith Packard Date: Fri May 1 12:26:04 2009 -0700 Call down to lower CloseScreen before shutting down DRM allocator Lower level functions will destroy objects that are managed by the DRM allocator, so make sure those are done before the allocator shuts down. Signed-off-by: Keith Packard commit cfb98f3da936907945f48dad4c2345930f7beb60 Author: Keith Packard Date: Fri May 1 12:23:32 2009 -0700 Revert "Leave allocator running until lower-level CloseScreens are done" Uh, oops -- the FD used to talk to the kernel is closed before the allocator_fini function is called now. This reverts commit 1872869e6ffcc8e6cab820c508fe5404d7e8ff9c. commit 11a853bd8e5d907fe7f5bd907453bcdac9032861 Author: Keith Packard Date: Fri May 1 11:51:13 2009 -0700 Hold reference to video binding table until all rects are painted. The optimization of unreferencing the binding table when the relocation is posted causes the object to be dereferenced for each box in the clip list, causing general chaos in the buffer manager. It's easier to just hold a reference to the object until all of the boxes are painted and then drop it. Signed-off-by: Keith Packard commit ed492131c13715b73c14d328d0668120acb58b40 Author: Keith Packard Date: Fri May 1 11:50:17 2009 -0700 3D_STATE_VERTEX_BUFFERS takes four 32-bit values, not three. The spec says this command takes an extra (mbz) 32-bit value, so let's provide it with one. Signed-off-by: Keith Packard commit 66d1536a2ecc7a3d44da4c0dbe6c85aaa4e05791 Author: Keith Packard Date: Fri May 1 11:48:51 2009 -0700 Don't bother to enable VF statistics during 965 video playback This was used while bringing up the driver to debug vertext fetches. Signed-off-by: Keith Packard commit 1872869e6ffcc8e6cab820c508fe5404d7e8ff9c Author: Keith Packard Date: Fri May 1 11:46:51 2009 -0700 Leave allocator running until lower-level CloseScreens are done The lower level close screen functions will free allocated objects, causing a crash if the allocator isn't still available. Signed-off-by: Keith Packard commit 1142353b487c155a31011923fbd08ec67e60f505 Author: Keith Packard Date: Fri May 1 11:44:13 2009 -0700 intel_batch_start_atomic: fix size passed to intel_batch_require_space (*4) intel_batch_start_atomic takes an argument in 32-bit units, and so it must multiply that by 4 before passing it to intel_batch_require_space, which takes an argument in bytes. We should figure out what units we want to use and use the same everywhere... Signed-off-by: Keith Packard commit 8232177a005127a0001af4b14c3beb766958ef11 Author: Keith Packard Date: Thu Apr 30 15:30:05 2009 -0700 Allow cursors to roam past 2048x2048 limit. There's no reason to clip cursor positions to an artificial limit; the hardware cursor limits always mirror the hardware display limits. Signed-off-by: Keith Packard commit 417f3784b7fae8de3559c7607a2de60661a6a448 Author: Carl Worth Date: Tue Apr 28 20:42:22 2009 -0700 Add doltlibtool to DISTCLEANFILES Otherwise make distcheck fails. commit bc9a688476f02233450eb05dfb40acbf337d679b Author: Carl Worth Date: Tue Apr 28 20:37:13 2009 -0700 NEWS: Add notes for 2.7.99.1 snapshot commit a90cab43209bba52d35b953f3d686a19c154fc43 Author: Kristian Høgsberg Date: Tue Apr 28 13:12:55 2009 -0400 Remove a handful of pointless comments CVS keywords, comments about how the source was once reformatted, and the ad-hoc changelog comment in in i830_driver.c commit c231b41f3f3eb3ba9c8436d5fd93a30704ed437c Author: Eric Anholt Date: Tue Apr 21 15:55:01 2009 -0700 Remove checks for xf86RotateFreeShadow, introduced in server 1.4. commit e5133a94dc78411eaef324131ea8056aeb81779e Author: Eric Anholt Date: Tue Apr 21 15:49:01 2009 -0700 unifdef LIBPCIACCESS and XSERVER_LIBPCIACCESS. Depending on new server means these are always present. commit 4653a7db622ad54a3182d93c81331765d930db34 Author: Eric Anholt Date: Mon Apr 20 22:05:54 2009 -0700 Rely on BO pixmaps being present in acceleration paths. commit 1784efc1f0c74edda8c78fb1b0cdd989f0a088ad Author: Eric Anholt Date: Mon Apr 20 18:25:59 2009 -0700 Remove XF86DRM_MODE. We now depend on server 1.6 and current libdrm, so no need to ifdef it. commit d82ea4372fb74654eb0f37d996977003473846ed Author: Eric Anholt Date: Mon Apr 20 13:20:03 2009 -0700 Replace I830Sync's irq emit/wait code with bufmgr use. This pre-GEM code was all sorts of broken -- see intel_bufmgr_fake.c for the hoops that must be jumped to use that kernel interface successfully. Yet we continued to use it even with KMS/DRI2/UXA, which may account for some hangs. commit 612c1f1f1859ce0cef41049d79cc473657d48612 Author: Eric Anholt Date: Mon Apr 20 16:16:48 2009 -0700 Remove I830_XV ifdef. The developers of the driver don't ever run or test without XV. Don't do it. commit af6892f2b873dd508994d8bbf3d67e719b535e35 Author: Eric Anholt Date: Mon Apr 20 16:07:05 2009 -0700 Replace i830WaitSync with just I830Sync(), as nearly all callers had it wrong. People were trying to BEGIN_BATCH()/ADVANCE_BATCH() then i830WaitSync on the results, which wouldn't necessarily wait and lead to various painful bugs, since only EXA called MarkSync and only for certain rendering operations. commit 393abc15b76804580bfee0760ddbd1fb77e66c02 Author: Eric Anholt Date: Mon Apr 20 15:58:35 2009 -0700 Now that video destination pixmaps are always in BOs, no more MarkSync. commit dc71573292c215f49716f4dc1ebc416c6b172995 Author: Eric Anholt Date: Fri Apr 17 15:13:34 2009 -0700 Remove can_resize, which is now always true (except when a bug interfered) commit 96df22740b2cdc4f82a4a36c0f77663105d9337e Author: Eric Anholt Date: Fri Apr 17 15:11:12 2009 -0700 unifdef I830_USE_UXA. We only have on acceleration architecture now, and you can always build it if you're building the driver. commit fa81ae1637bfd7b73b43355b78c88a9b84c0393c Author: Eric Anholt Date: Fri Apr 17 14:01:47 2009 -0700 Remove EXA support. UXA has completely replaced EXA at this point. UXA is the same rendering core as EXA, but relying on kernel memory management or a fake bufmgr instead of trying to manage memory in the X Server. commit 1b10745a2528622a32271f64c35fcdb7b7154d11 Author: Eric Anholt Date: Fri Jan 23 08:39:59 2009 -0800 Remove XAA support. While EXA/UXA aren't completely good replacements (see bugzilla for performance and stability problems), we are pretty sure at this point that it's the right way to go and that having multiple acceleration architectures is getting in the way of producing a stable codebase. commit 30c226eeb35f034c3ddb881e77454051dc210298 Author: Eric Anholt Date: Thu Jan 22 17:54:49 2009 -0800 Remove DRI1 support. This was blocked on wide distribution of X Server 1.6 (now in the current or next version of major distributions) and solutions for a couple of significant architectural problems (vblank sync and frontbuffer rendering, which we now have code or good plans for). This includes disabling XVMC which is DRI1-only currently. commit 47591334a183881704a121ae06ebc2fadebe6f73 Author: Eric Anholt Date: Fri Apr 17 13:54:47 2009 -0700 Remove pre-server-1.5 support. commit 3187c7698b93f1d0e07798e9c42ca2877d55e055 Author: Eric Anholt Date: Mon Apr 27 14:50:54 2009 -0700 Fix assert at startup with DRI initialiation failure. In this path, we'd make it to allocator_init -> init_bufmgr without GEM and without FbBase being initialized, leading to assertion failure to catch this mistake. Comedy ensued when trying to move just the MapMem up, leading to the rest of the commit. Some day (like tomorrow after I rebase intel-cleanup) I'll clean this path up. Tested with 2 X Servers on 2.6.28 (one gets DRI2, one fails successfully), 2 UMS X Servers on 2.6.30rc2 (each gets DRI2), and 2 KMS X Servers on 2.6.30rc2 (success all around). commit fbaf13c93511547c563129527797fbef8628652e Author: Eric Anholt Date: Mon Apr 27 16:05:50 2009 -0700 Don't clear the framebuffer if it isn't currently bound. This segfault can be triggered in non-KMS memory init before the EnterVT, and will happen anyway at EnterVT time. commit ab878118b9fb46c437a02f824d29f4a206e7fd80 Author: Eric Anholt Date: Mon Apr 27 16:00:05 2009 -0700 Don't initialize DRI2 if the fd we get is not master-capable. commit 385ac2197b64574f30620a203e675d8da0e1472f Author: Eric Anholt Date: Mon Apr 27 15:28:21 2009 -0700 Fix drmSetMaster/DropMaster error messages. commit 5b05a589efb23b2fc09b06e4271174d922b1ab02 Author: Pierre Willenbrock Date: Sat Apr 25 22:58:20 2009 +0200 format == 0 means "use default" in I830DRI2CreateBuffer Reviewed-by: Ian Romanick commit e55d943126cdd3eac7dfec5f40e794f89dbf038b Author: Eric Anholt Date: Mon Apr 27 11:18:07 2009 -0700 Don't try to do anything for I830Sync when VT switched. Fixes a crash with fake bufmgr and UXA. commit 9748aa5594a8357754c06dd82108b131260b1a71 Author: Shuang He Date: Sat Apr 18 11:48:08 2009 +0800 Free front buffer bo when X exit It's needed when KMS or DRI2 is enabled, or there will be memory leak. Also fixes a segfault at startup with fake bufmgr. Signed-off-by: Eric Anholt commit b8ca146b060e6f76bf048f1f6cf6ee89d38d015e Author: Alan Coopersmith Date: Fri Apr 24 16:03:27 2009 -0700 Fix UXA to build with Sun compilers (use __func__ instead of __FUNCTION__) Signed-off-by: Alan Coopersmith commit 5d6f4f6eb7a4dcbe1ce5a134d882e56f958ed2ba Author: Ian Romanick Date: Fri Apr 24 12:28:13 2009 -0700 DRI2: If the SDK supports it, use the DRI2GetBuffersWithFormat interfaces If DRI2INFOREC_VERSION is defined in the server's dri2.h and has a value greater than 1, compile to use the CreateBuffer and DestroyBuffer interfaces. The format parameter parameter to CreateBuffer is assumed to be the bits-per-pixel of the buffer. Signed-off-by: Ian Romanick Reviewed-by: Kristian Høgsberg commit 106e4b44c5af6552cbd079c4ec34def9dcfb168a Author: Zhenyu Wang Date: Fri Apr 24 10:55:09 2009 +0800 SDVO: switch bus once for i2c transfer Before sdvo try to get edid by i2c bus, it must do switch control bus to ddc active state from sdvo only state. However if current state has been ddc active state, redundant switch operation in our driver will cause error- "Unable to write to SDVOCTRL_E for SDVOB Slave 0x70". The patch will do switch control bus only one time during whole edid transmission. It has fixed bug #19937 Signed-off-by: Zhenyu Wang Tested-by: Ma Ling Signed-off-by: Ma Ling commit 8f64837e56b2de0fb8a9100d1a844fd3f18d751c Author: Shaohua Li Date: Mon Feb 23 15:19:25 2009 +0800 Disable FBC on IGD for UMS It appears the new chip doesn't support FBC currently. Signed-off-by: Shaohua Li Signed-off-by: Zhenyu Wang commit 7b01aa5cc41620da5bb48f391ff98d9e82572e52 Author: Eric Anholt Date: Wed Apr 22 09:03:21 2009 -0700 Revert "fix overflow warning on videoRam" This reverts commit 08ebde4715b87867184d42b60762cd774e151f5c, which was apparently untested and broke KMS. commit ceb3a2eaf9090d67e4dfcbed188125108ceab78d Author: Zdenek Kabelac Date: Tue Apr 21 18:58:23 2009 -0700 Remember allocated offscreenImages memory for freeing it in CloseScreen() Signed-off-by: Zdenek Kabelac [anholt: renamed the member to match the variable name] Signed-off-by: Eric Anholt commit 701c6bef14b5d464847cc42c5ccb66fdeb157f1a Author: Zdenek Kabelac Date: Tue Apr 21 18:53:07 2009 -0700 free allocated name inside sdvo_destroy Signed-off-by: Zdenek Kabelac Signed-off-by: Eric Anholt commit fe71217ca86eb7ae8081bf3e26d30f2002d27691 Author: Zdenek Kabelac Date: Tue Apr 21 18:51:02 2009 -0700 Unreference allocated bos in i965 render error paths Signed-off-by: Zdenek Kabelac Signed-off-by: Eric Anholt commit b8b6ca830b1fd6956c76a5417462ca55174cb881 Author: Zdenek Kabelac Date: Tue Apr 21 18:47:03 2009 -0700 Fix leak of some 965 render state on VT switch. Signed-off-by: Zdenek Kabelac [anholt: fixed up unneeded != NULL checks] Signed-off-by: Eric Anholt commit d7ca870e1ce251d42e3689a8e1e7d080ab1325fb Author: Albert Damen Date: Sun Apr 5 16:36:35 2009 +0200 Fix crash with XV with large virtual display If a virtual display with width > 2048 is used, the first time an XV buffer is needed will result in a BadAlloc error message, but the next time X would crash. Signed-off-by: Eric Anholt commit 04ed92e3db8d9fcf2a1089ed433ab67bf0969655 Author: Eric Anholt Date: Tue Apr 21 15:17:31 2009 -0700 Remove dead xoffset/yoffset from pre-randr. commit ab2c988767aec854d3c1702bee1e5a06bd00870c Author: Eric Anholt Date: Tue Apr 21 14:53:45 2009 -0700 Remove dead monitor detect debugger. watch xrandr works fine. commit 928a37041defcca6f57f9452ba62e67524cb4510 Author: Eric Anholt Date: Mon Apr 20 15:46:11 2009 -0700 Replace a bunch of #ifdef debug flushing/syncing with a single function. This removes it from a callsite where it would have just resulted in a fatalerror. commit 1fc93ee184ceefaea9528bb46ae82884c44d9b36 Author: Eric Anholt Date: Mon Apr 20 14:42:31 2009 -0700 Staticize a bunch of functions and variables in the driver. This cleans up findstatic.pl output for the i830+ code, which resulted in removing some code. The only odd part of this commit is the if (0) i830_sdvo_dump() in i830_sdvo.c -- it tells the compiler that the code is used, without using it since we want the code around while debugging. It's also in a likely place to ask for the dump, so I think it's OK. commit b2d7928f195f6439ecd140e6280b958b8a1f534b Author: Eric Anholt Date: Mon Apr 20 14:48:59 2009 -0700 Use a static inline to replace if (0) to an unused stub function. commit 093f51646b2f2301c907e14d0ad4b1f80ff60a16 Author: Eric Anholt Date: Mon Apr 20 14:36:52 2009 -0700 Remove unused i830_output_type_names commit c774ef282bfb0c594479bbab3527fe5fda2a8559 Author: Eric Anholt Date: Mon Apr 20 14:29:17 2009 -0700 Remove dead mono cursor load code. commit 75c531b9652b3919ddbf51f8eec804e6bdbb2331 Author: Eric Anholt Date: Fri Apr 17 14:03:39 2009 -0700 Rename EXA rendering functions to UXA, since we're keeping them post-EXA. commit 40dbba5cf710a6e392afca73965aa48fa3953d1f Author: Eric Anholt Date: Fri Apr 17 15:15:27 2009 -0700 Remove some dead i830.h struct members. commit 84f69081abaeef8b05cafb64d3102eb2abdf9a8e Author: Julien Cristau Date: Thu Apr 16 14:27:15 2009 +0200 Require xserver 1.3 I don't think anyone tests this against an old server anymore. Signed-off-by: Eric Anholt commit 08ebde4715b87867184d42b60762cd774e151f5c Author: Zhenyu Wang Date: Mon Apr 20 09:43:09 2009 +0800 fix overflow warning on videoRam Note that pScrn->videoRam is an 'int'. i830_driver.c: In function ‘I830ScreenInit’: i830_driver.c:3019: warning: overflow in implicit constant conversion Signed-off-by: Wu Fengguang commit 5a07ab502fe1e58e7e37fe554fb42d8d2c8c53ec Author: Ian Romanick Date: Fri Apr 17 20:59:04 2009 -0700 DRI2: Respect the src and dst parameters of CopyRegion. Signed-off-by: Ian Romanick commit 2e3b95ed0197971e81ab7509245c899e96859d5b Author: Carl Worth Date: Wed Apr 15 18:54:19 2009 -0700 Increment version to 2.7.99.1 This is the number we want for any development snapshots along what will eventually become 2.8.0. (Once we get to 'release candidates' we'll bump this up to 2.7.99.901.) commit c9e174b31ae3b9a138ae696d563420a841062f57 Author: Carl Worth Date: Wed Apr 15 18:52:38 2009 -0700 RELEASING: Note that --with-xserver-source is needed for make distcheck I was always forgetting this without this reminder. commit 73c3be1aa033e8c5c7ee777eb2fd43c19668fa86 Author: Carl Worth Date: Wed Apr 15 18:18:14 2009 -0700 README: Fix typos in chipset list, and point to how_to_report_bug web page Thanks to Gordon Jin for these suggestions. commit c51dddb724a79a75491369a4c3e8b7b26231e7ac Author: Carl Worth Date: Wed Apr 15 18:07:17 2009 -0700 AUTHORS: Add Robert Lowery to the authors file Rob got missed from my first scan since one commit lists his name as just 'Rob' and 3 commits don't attribute him as author: 83d304c61ad5fdc58b0a9309dbd1e5a3f6cd9b01 7552d80e367fe38bbc594fe94abd649917fe54d5 6eecef4fed8a21dfdabef42eb69fd150b96167b2 commit 4b5edde5da4b3e955eb2d77004de81e47bec7f69 Author: Robert Lowery Date: Wed Apr 15 18:03:31 2009 -0700 Fix typo in comment Thanks to Robert Lowery for the sharp eyes on this one. commit 3fd5a1ecd1d5140ae07ccc279298bcadd515e97f Author: Carl Worth Date: Wed Apr 15 16:44:11 2009 -0700 RELEASING: Update instructions to reflect some minor process improvements We've added a NEWS file now, so that needs to be updated for each release. We're also now using tag names of just rather than xf86-video-intel-. commit e1cace16a6130dcdd93965d2329a349d49200fa6 Author: Carl Worth Date: Wed Apr 15 16:33:12 2009 -0700 NEWS: Add note about broken PAT code in some kernels Hoping to cut off some false bug reports here. commit 9ffd1951d1f2fd2f53273d04ea29de050f07af55 Author: Carl Worth Date: Wed Apr 15 16:14:44 2009 -0700 Add AUTHORS and NEWS to EXTRA_DIST These new files don't do us much good if we don't distribute them in our releases. commit e4cd9de2933ada3e2a4b43552729ae3a370128bf Author: Carl Worth Date: Wed Apr 15 16:14:03 2009 -0700 Add a NEWS files with release-notes for 2.7.0 It will be nice to have release-notes under revision control, as well being able to document issues in an obviously time-sensitive file, (as opposed to README where we were documenting some of this previously). commit 506c810f8f3db89048dda9777902f142ffeb86aa Author: Carl Worth Date: Wed Apr 15 16:10:52 2009 -0700 Clarify that the default acceleration is UXA if KMS is available. Stale documentation considered harmful of course. commit b9716b836cb2b4569c90b81f344932ac668dc5bf Author: Carl Worth Date: Wed Apr 15 15:39:06 2009 -0700 Add a new AUTHORS file This is a sorted list of everyone with more than 2 commits in the git revision history. We also list some historical authors mentioned in the man page, (with code presumably pre-dating the beginning of revision history). commit 8deb3a3709a9aaa549be404566715a01246354d9 Author: Carl Worth Date: Wed Apr 15 15:38:11 2009 -0700 README: Remove almost all time-sensitive information This was all very stale, and is better convered in intel.man. We replace this with a list of pointers to where to get current information, (man page, web site, and mailing list). commit 9b615a52671aacf34666f90ecfff98651ce6afe2 Author: Li Peng Date: Fri Apr 10 14:39:35 2009 +0800 Turn on front buffer tiling in KMS. This code disabled front buffer tiling in KMS. Turn it on since kernel handles all tiling now, this improves performance of x11perf -aa10text from 97k to 286k on my 945GME. Should help with #20761, if not totally fix it. Acked-by: Jesse Barnes Signed-off-by: Li Peng commit 053432991c812146f6e7c6f13c6ace55385c825f Author: Ma Ling Date: Mon Apr 13 14:27:35 2009 +0800 update manpage for BROADCAST_RGB property commit 62ba7211fe9b6aada125ebfe34cf7161e817ad6b Author: Ma Ling Date: Mon Apr 13 14:24:57 2009 +0800 set broadcast RGB mode for integrated HDMI output. commit 69388953ce889080d5f014123d89bf3eb45f3d8d Author: Ma Ling Date: Mon Apr 13 14:23:06 2009 +0800 set broadcast RGB mode for HDMI and TMDS from SDVOX output Almost all digital TVs accept broadcast RGB values from 16 to 235 for each channel, otherwise for those uncompensated videos, when RGB values are set from 0 to 255, they will shows black and whiter clamping, which seriously degrades picture quality. The patch will enable the broadcast RGB mode for hdtv according to user's setting. It fixed bug #14486 commit 6d345c49f693cc5cffaa00b94559d2afcb3a0864 Author: Carl Worth Date: Fri Apr 10 14:07:14 2009 -0700 Add a RELEASING file documenting the release process Thanks to Jesse Barnes for the original recipe. commit 7e516b6d24d8c0c6549a9a60fcf487e3a1615020 Author: Jesse Barnes Date: Wed Apr 8 16:38:08 2009 -0700 Silence warning in i830_dmi_store_field Just add a dummy ret variable to shut up gcc. commit 620e97bbd6a811ad69b8ac94df1fe2c9edf65549 Author: Jesse Barnes Date: Wed Apr 8 15:49:00 2009 -0700 Don't enable kernel execbuf fencing w/EXA If we enable kernel execbuf fence register management, it's best if the kernel manages all fence registers. This works fine if the accel method is managing pixmaps or doesn't use offscreen pixmaps. However with EXA, pixmap accesses are done relative to the framebuffer BAR mapping (pI830->FbBase) and the Screen pixmap address. So if we try to set the screen pixmap to point at a GTT mapped (and therefore properly fenced) address, later calls to intel_get_pixmap_offset() will call into EXA, which will use the pseudo-random pixmap addr and the EXA offscreen base addr (which is really just FbBase) to calculate the offset. This will fail. So disable kernel fence reg management in the EXA case (this is easier than adding proper EXA pixmap management to xf86-video-intel, and makes more sense since we'll be removing EXA soon anyway). Fixes FDO #21027. Also happens to fix FDO #21029 (as tested by Carl Worth Date: Tue Apr 7 12:31:07 2009 -0700 Fix value for MI_WAIT_FOR_PIPEA_SCAN_LINE_WINDOW Since the change to scan-line based video sync, (rather than vblank- based), we've only been getting tear-free video on one of the two pipes. This fixes that bug by using the correct constant for waiting on PIPEA. commit 940c2aad4d174b6609bdc49f8c99a4bc37926516 Author: Carl Worth Date: Mon Apr 6 14:36:33 2009 -0700 Don't clip video to CRTC in the case of textured video Since we're not limited by a single overlay plane on a single pipe, we want to not clip at all, (so that the correct video appears on both pipes). This fixes bug #20980 which shows a video spanning two pipes being rendered incorrectly. commit 63b4b5efac936c674dedad8125a8dbac4f000908 Author: Zhenyu Wang Date: Tue Apr 7 10:53:08 2009 +0800 quirk LVDS on ibase MB890 855GM board fix bug #19529 commit 5d9d9a2e466474a0508a15b294a91507ccb3ccc1 Author: Carl Worth Date: Mon Apr 6 14:02:08 2009 -0700 Fix new video sync-to-blank code for multi-head We need to account for a non-zero Y offset for the CRTC. Without this, we don't sync to the correct region, so tearing becomes visible again. commit 3d4ee3cac1d63dfdf7b54c8ba577f3b77637499f Author: Carl Worth Date: Mon Apr 6 11:31:20 2009 -0700 Remove support for 'auto'(-1) value of XV_SYNC_TO_VBLANK We previously had a heurstic here where we would only sync to vblank for windows that covered more than 25% of the screen. We don't need this anymore since the new approach to sync, (WAIT_FOR_SCANLINE_WINDOW), is not excessively costly for small windows. commit bc3312fd7c03d09a231dfebfe390fe668ad15d1e Author: Carl Worth Date: Mon Apr 6 11:16:40 2009 -0700 Use WAIT_FOR_SCAN_LINE instead of WAIT_FOR_VBLANK Either way, the goal is tear-free video playing. But waiting for a scan-line window not only has the advantage of being cheaper for small windows, but also avoids hanging the GPU in the case of the pipe getting turned off, (by screensaver, for example), while a batch is waiting for a VBLANK that will never occur. This fixes that GPU hang. commit 6cd914ef315036ce8e91c7b6492994353e8ed2d8 Author: Jesse Barnes Date: Fri Apr 3 11:15:18 2009 -0700 Fix offset in begin_gtt_access case Don't use bo->virtual in the begin_gtt_access case, use the framebuffer mapping and bo offset instead. Reviewed-by: Eric Anholt Signed-off-by: Jesse Barnes commit 13c730e003e805e19deee0996b2af30f69e54c4d Author: Zou Nan hai Date: Wed Apr 1 15:12:55 2009 +0800 fix 1.multiple instance 2.memory leak commit fad714c40078d22fff82dc0692a344f66ddf9680 Author: Zhenyu Wang Date: Wed Apr 1 10:11:35 2009 +0800 SDVO: fix output flag dumping for unknown type Found by Hugo Jacques commit 00de1757dd5776962bdd4c8968181878c2ebf4c9 Author: Hugo Jacques Date: Wed Apr 1 10:10:05 2009 +0800 SDVO: add composite TV out support commit 087f72e1f5d7d11b8795ba80a842874f5a9bb01d Author: Jesse Barnes Date: Tue Mar 31 07:50:10 2009 -0700 Match GTT unmap with map in KMS rotation case Missed this when the GTT unmap call was added. If we don't do this we trigger an assertion in libdrm, since the buffer has never been mapped normally. Fixes bug #20943. Signed-off-by: Jesse Barnes commit 4f046af760b92c07f59664359453933fb5358e3d Author: Zhenyu Wang Date: Tue Mar 31 13:49:44 2009 +0800 Disable LVDS detect methods Both methods ACPI lid and SWF bit have issues in LVDS detect from wider testing. Fallback to origin code. commit e964d4e53af3a47de6d09c884be1cc0044d03bea Author: Albert Damen Date: Mon Mar 30 11:38:02 2009 -0700 Non-GEM allocations incorrectly force TILE_NONE (bug 20797) With -intel 2.6.3 performance was very bad when using a non gem enabled kernel (2.6.27) and EXA. For example sauerbraten ran with 4 fps and screensaver GLBlur with 1 fps. With -intel 2.6.1 performance was good using the same kernel. Git bisecting led me to commit f1ed73c1ef3e3daa9f695194dcc813167cbcb53d (in 2.6 branch) "Make i830_allocate_memory take tiling parameters" as first bad commit. Using gdb I found tiling was set exactly the same in 2.6.3 as in 2.6.1, so that was good (TILE_XMAJOR for front, back and depth buffers). Looking further I found the line mem->tiling = TILE_NONE; (line 961 in src/i830_memory.c) at the end of i830_allocate_memory suspicious, as mem->tiling now already gets set via i830_allocate_aperture and some buffers do have tiling. Removing that line indeed fixed the performance issue. Now sauerbraten runs with 30+ fps and GLBlur runs smoothly. commit 51cf8a453c2bc2e8604bfc41a649e971c1ba5026 Author: Jesse Barnes Date: Mon Mar 30 11:27:56 2009 -0700 Require libdrm 2.4.6 for GTT unmap support Need the new functions available. commit 8dabcc40747bfd478f296728741240241698f165 Author: Jesse Barnes Date: Mon Mar 30 09:53:40 2009 -0700 Tiling fixes, third set Hopefully this concludes the fixes necessary to deal with the various combinations of kernel and user level tiling. We have several cases to handle: 1) KMS (kernel handles all tiling) 2) UMS w/memory management + kexec fencing (kernel handles all tiling) 3) UMS w/memory mangement but no kexec fencing (userland handles tiling) 4) UMS w/o memory management (userland handles tiling) For cases (1) & (2) we can use GTT mapping, which will give us good performance and take care of allocating fence registers as needed. It's important *not* to have userland set up fence regs in this case, since the kernel will be using all of them. For case (3), we use the begin/end GTT map functions provided by libdrm, in combination with pinning and fence register setup in i830_memory.c to deal with tiled surfaces. This also gives us good performance and correctness. For case (4) we use the old style virtual mapping + offset for dealing with surfaces; note that UXA doesn't seem to work in this configuration regardless of these fixes. Fixes bug #20803. Reviewed-by: Eric Anholt Signed-off-by: Jesse Barnes commit 375b2e40fcb17e94538a75392950e2533c1bb031 Author: Zhenyu Wang Date: Wed Mar 25 11:13:52 2009 +0800 Disable LVDS config parsing from VBT for now As wider tests showed that this doesn't work for all VBIOS, so disable it for now and reenable it after we get reliable method. commit defa6971c6b639ab72d9d4c85c4607d5f852b2b0 Author: Zou Nan hai Date: Thu Mar 26 15:24:29 2009 +0800 enable UV half pixel commit 69c84f2c8204771b68f40ed64e64657237b54546 Author: Zhenyu Wang Date: Thu Mar 26 13:52:05 2009 +0800 gitignore intel_lid commit fe96e517699e37cc990a14dfb48b9611a02254ed Author: Zhenyu Wang Date: Thu Mar 26 10:54:02 2009 +0800 xvmc: add missed vld g4i sources in Makefile and fix distcheck Don't miss the vld shader programs for dist. commit 91886f82a1155ca24eb895f77678ea1edfccf9fe Author: Zhenyu Wang Date: Thu Mar 26 10:51:05 2009 +0800 xvmc: move 965 mc media programs to shader/mc directory Also include some missed g4i files in Makefile.am, fix shader build. commit ebcb64fdc8a45105f529d5e15c335514ecaf1a05 Author: Zou Nan hai Date: Wed Mar 25 16:11:33 2009 +0800 XvMC VLD extension support for G4X commit e6af995b24767815ab30364385768867e80605c2 Author: Zhenyu Wang Date: Mon Mar 23 19:19:58 2009 +0800 KMS: hook up output properties for randr This gets output properties from kernel, then hook them up for randr. So we can control output properties through randr like in UMS. Signed-off-by: Zhenyu Wang commit b5b377806422c9538fa9f846819ad253c1a50846 Author: Ma, Ling Date: Wed Mar 25 14:39:46 2009 +0800 disable center mode for 965GM and G4X platform When disabling VGA mode, usually we don't need to touch VGA center mode. However because of hardware reason, for Cresline, Cantiga & Eaglelake platform, we have to disable center mode as well. The patch fixed bug- TV Out strobing regression, reported by Robert Lowery in intel-gfx@lists.freedesktop.org mailing list. Signed-off-by: Ma Ling commit 74d0713e002a9781ed00fdd10eb6f75907ae796c Author: Kalev Lember Date: Fri Mar 20 02:21:41 2009 +0200 Fix VT switch with XV overlay video enabled. drm_intel_bo_unpin() was called with NULL argument. Signed-off-by: Kalev Lember commit b345b4e6adf00f8b2626e5c2563fc159e2295cc5 Author: Rémi Cardona Date: Fri Nov 14 13:55:52 2008 +0100 clean up man page generation and remove all traces of the i810 driver The i810 compatibility symlink has been broken since libpciaccess, so just let it die. commit 48b91e066878db63a1558e4cd3e6d12ff9c49197 Author: Jesse Barnes Date: Fri Mar 20 12:33:22 2009 -0700 Don't manage fences part two Don't try to clear fences that were never installed. Missed this bit in the last fix for #20265. commit 28319d60aa2d793c209043c1ce2d38b14c66a4be Author: Zhenyu Wang Date: Fri Mar 20 13:14:15 2009 +0800 Add a lid status test program "intel_lid" commit 636d252f3b1eac687f7b11952e949c383cb86ed4 Author: Jesse Barnes Date: Thu Mar 19 13:25:29 2009 -0700 Don't install fences if the kernel is managing them If execbuffer is setting up fences, it also means that the kernel is managing them at pin time, so installing one in the 2D driver in that case is an error. The fence should stick around as long as the buffer is pinned (the kernel won't steal these), though it will be freed at leavevt and re-allocated at entervt. On 965+ chips, the pin ioctl will *not* install a fence reg, but that's also ok because all 965+ operations include tiling bits, and sw fallbacks will be protected by prepare/finish access hooks, which will either access the backing store or use the GTT, which will ensure proper fencing at fault time. Fixes #20265. Acked-by: Eric Anholt commit d92e6158c86d7da873913461ef7e828bd411d8c4 Author: Carl Worth Date: Thu Mar 19 10:06:03 2009 -0700 Revert the rest of the EXA_VERSION_MAJOR bump This was mistakenly added in the unrelated change in revision fe08b81d0f5d6f96e0124e6286bd24aba6e140ad and wasn't completely reverted in the later revision 78a60e1b66fe2e8449702dd43d9b062d279af8f1 commit 85e6b528582c2397ee9eb2132cd0d05ce12eb43d Author: Ma Ling Date: Thu Mar 19 09:10:19 2009 -0700 Set SSC frequency for 8xx chips correctly All 8xx class chips have the 66/48 split, not just 855. Fixes #18358. commit bedc894a565ce3c7a50990e3f78953fc2432ad40 Author: Jesse Barnes Date: Wed Mar 18 20:19:44 2009 -0700 Re-enable disabled outputs after rotation The server may have made a DPMS call before doing rotation, so after we do the mode set with the rotated framebuffer, we need to re-enable the corresponding output(s). Fixes bug #20573. commit 6deb26ae7bd796e88a5dd90df5f6c35fbc44e798 Author: Jesse Barnes Date: Wed Mar 18 09:36:58 2009 -0700 Create known output configuration at EnterVT time Since we added the pipe A force quirk (leaving pipe A on all the time), DPMS calls to disable it have silently returned, leaving the pipe on. If another driver (like vesafb) has enabled it, we may end up with a bad configuration, leading to hangs or blank screens at VT switch time. Fixes bug #19603. commit 7c94227dd4fa2164bebb36234958053bf1d26c12 Author: Ma Ling Date: Tue Mar 17 10:33:15 2009 +0800 Use best PLL timing values for G4X platform construct function to find precise parameters from internal spreadsheet table on G4X platform. Signed-off-by: Ma Ling commit 48db5bde9298f1126dfb42f4be8a3d61166abfd8 Author: Ma Ling Date: Tue Mar 17 10:41:02 2009 +0800 Define documented PLL timing limits for G4X platform These timings on G4X platform were specified by internal spreadsheet from the chipset group. Signed-off-by: Ma Ling commit 78a60e1b66fe2e8449702dd43d9b062d279af8f1 Author: Keith Packard Date: Mon Mar 16 08:41:52 2009 -0700 Revert EXA_DRIVER_KNOWN_MAJOR bump This was mistakenly added in the unrelated change in revision fe08b81d0f5d6f96e0124e6286bd24aba6e140ad commit 1a6e70cd98abc63f24f710819961c24601afb59b Author: Zhenyu Wang Date: Mon Mar 16 10:19:39 2009 +0800 TV: fix contrast and saturation for 945G Bug #20670. commit 4e95327323e3d081b565147f7738eb49c28542bc Author: Zhenyu Wang Date: Mon Mar 16 09:30:22 2009 +0800 TV: force TV as connected with TV_Connector option In order to bypass failure in TV load detect, TV_Connector option will always force TV as connected with user specified connector type. commit d9dbdb325543bd747cd1bfb3e1142ea6daf2b637 Author: Eric Anholt Date: Fri Mar 13 15:48:40 2009 -0700 Add shave support, enabled by default. This cuts down build system noise so that warnings are more visible. The old style output can be reenabled for build system debugging using "make V=1", or --disable-shave at configure time. commit c3a82106a1a1a94c9e1e465c7dc0d828c1cbf50d Author: Eric Anholt Date: Fri Mar 13 15:29:45 2009 -0700 Move contributed m4 (dolt) to a subdirectory so we can include it with others. commit e0aba04da0323efc079ce0a2b78139c4d0d49fa5 Author: Eric Anholt Date: Fri Mar 13 13:37:35 2009 -0700 Add DCC register dumping. commit a57814cc13f4287eccaf1906963b80c9205c680c Author: Eric Anholt Date: Fri Mar 13 13:20:35 2009 -0700 Add dumping of 915 and 945 fence registers. The debug dumper functions can now return NULL to indicate no output, so we get appropriate results on 915, 945, and 965. commit 8166a7ff5c2c1e3736dd06a8453c4e4d769d8b75 Author: Eric Anholt Date: Fri Mar 13 13:12:33 2009 -0700 Add PGETBL_CTL to the debug output. It's nice to know when it gets stomped on. commit fe08b81d0f5d6f96e0124e6286bd24aba6e140ad Author: Keith Packard Date: Mon Mar 2 07:39:41 2009 -0800 Use CopyArea to load glyphs from per-glyph pixmap to cache pixmap With glyphs sitting in per-glyph pixmaps, there's no reason to use the CPU to move them to the cache pixmap, and lots of reasons to use the accelerator. Signed-off-by: Keith Packard commit 2026c57cf0a352d9e6f9d208cfb7d4d550614477 Author: Kalev Lember Date: Fri Mar 13 21:32:08 2009 +0200 Fix Xv crash with overlay video. Bug #20585. commit 3ef9d85371a97ea5baee0c47787b3bb3cdaf5135 Author: Jesse Barnes Date: Fri Mar 13 12:21:34 2009 -0700 Use UXA when KMS is active EXA doesn't support KMS, so force UXA on if KMS is detected. And warn the user if they've specified something other than UXA in their xorg.conf. Fixes bug #20620. commit dc3ff0b514b609448025680778f0e95e1980a5d8 Author: Zhenyu Wang Date: Thu Mar 12 16:32:02 2009 +0800 Revert "SDVO: Switch control bus only before DDC access" This reverts commit ddedf19f889da2ce6d69a3afce4665e2245682fa. After i2c STOP, control bus will return back to internal registers. So this brings back to origin code that we switch to DDC bus before START. But it's ideal to only issue DDC bus switch after STOP, not before every START, which might eliminate some complains from SDVO device, that will be another patch later. commit c6b0135d209bdad3dbc641d0e264596eaf6f99d3 Author: Zhenyu Wang Date: Thu Mar 12 16:31:01 2009 +0800 SDVO: reset privates before output setup For multifunction encoder, forget to reset these values will cause wrong output type in later mode setting operations. commit 28e7f0d71fa09e15a68ab4f0de169474b6235093 Author: Dan Nicholson Date: Tue Mar 10 20:16:03 2009 -0700 Fix dist of xvmc sources The XVMC AM_CONDITIONAL is only needed around the library expression. None of the other definitons will cause anything to be built without libXvMC, but they're needed for 'make dist'. Signed-off-by: Dan Nicholson commit 73db44e7ac524e84e5f0fda2d60069a9e954ad1b Author: Kristian Høgsberg Date: Wed Mar 11 11:10:57 2009 -0400 Drop Legacy3D option, only use fixed texture space with non-gem. With this change, we always expect the 3D driver to use GEM textures when the 2D driver uses GEM. When GEM is not available or disabled, we fall back to legacy fixed textures. commit 2fcf4fcccfe7cfa1425985d21a144137eca07f4e Author: Zhenyu Wang Date: Tue Mar 10 14:21:36 2009 +0800 SDVO: fix pixel multiplier setting for TV We should use preferred input timing's clock for correct pixel multiplier setting, otherwise we might get inconsistent multiplier setting on pipe and SDVO device for some modes. commit fb6e00f40f713a87c760fc7603159ed11ea9b0d5 Author: Eric Anholt Date: Mon Mar 9 20:06:30 2009 -0700 Fix XV with non-GEM kernels by allocating a larger fake bufmgr. Ideally we'd not be using the EXA offscreen memory manager and just hand all that memory to the fake bufmgr for non-GEM, but the fake bufmgr's too slow for that, at least currently. So compromise and take enough memory that it will succeed at XV allocations but hopefully not anger tiny-aperture systems too much. Bug #20563. commit dc12c4b3eb297b2f225409eacf1f3cd521453ab6 Author: Eric Anholt Date: Sat Mar 7 23:22:54 2009 -0800 Flip the update_dri_buffers test around to only run when DRI1 is active. Fixes segfaults at startup with DRI2 and load detection, or with DRI disabled entirely. commit 646e12a9783d1d48ef21841f0909287a876731a4 Author: Xiang, Haihao Date: Mon Mar 9 10:59:36 2009 +0800 typo in intel.man commit 73aa23d9150121a4e4b70a78cab910acd164abf5 Author: Eric Anholt Date: Fri Dec 5 13:06:05 2008 -0800 DRI1: Update sarea (and other information) when CRTC configuration changes. Bug #14423. Signed-off-by: Eric Anholt commit abb213d933ac0d808fc10d4f8d88d7b8cef76346 Author: Eric Anholt Date: Fri Mar 6 14:33:46 2009 -0800 Document the UXA AccelMethod. Signed-off-by: Eric Anholt commit 568297d327cc321f1186afc54b38d08db3f2914d Author: Eric Anholt Date: Thu Jan 22 17:08:19 2009 -0800 Don't allocate the render power saving context in KMS mode. That would be the kernel's job if it chooses to do it. Signed-off-by: Eric Anholt commit 4e1144108424a4525bbd97c3d5a56de06760bdd9 Author: Eric Anholt Date: Fri Mar 6 14:30:05 2009 -0800 unused variable warning fix. Signed-off-by: Eric Anholt commit 043a76a040d4576b7a8397dca805466a99bfcdd6 Author: Eric Anholt Date: Fri Mar 6 14:29:22 2009 -0800 clean up arguments to i830_allocate_framebuffer since zaphod removal. Signed-off-by: Eric Anholt commit b23f57b310b693f56af273526383221a4f8b96f5 Author: Eric Anholt Date: Thu Jan 22 17:17:09 2009 -0800 Use REGION_EQUAL in place of a local implementation. Signed-off-by: Eric Anholt commit 6b61f9945f54df7469f2b2d702b621d4d6064c3f Author: Eric Anholt Date: Thu Jan 22 17:19:04 2009 -0800 nuke unused define. Signed-off-by: Eric Anholt commit c3a747cb54acc1b037b559313e6a2113ae2ac4c7 Author: Dan Nicholson Date: Wed Nov 26 09:57:24 2008 -0800 man: Put option defaults on separate indented line The groff .IP macro is used to put the option defaults in a new indented paragraph so they are separated from the explanations. Signed-off-by: Dan Nicholson [anholt: hand-applied due to conflicts. mistakes are my own] Signed-off-by: Eric Anholt commit d2af21a66d7e1d1dd62c6aa8fb41d3fd6045bcd7 Author: Eric Anholt Date: Fri Feb 27 23:38:00 2009 -0800 Remove configurable support for disabling XV. google shows one instance of this being used a year and a half ago. commit 755757669f0cad670cfa084d33d7c3e928d27855 Author: Eric Anholt Date: Fri Feb 27 22:57:15 2009 -0800 Remove configured SW cursor support. Any time we actually need SW cursors, it gets enabled automatically. commit 917b20ead3cacf1e88314f20edde6a02b97b96d7 Author: Eric Anholt Date: Fri Feb 27 23:01:28 2009 -0800 Remove StolenOnly support. We rely on having AGPGART present to successfully allocate video memory as we configure it by default. Admit that fact, and remove support for non-AGPGART/KMS setups. commit 73b7190421132ad73179c3fb7bb0e06c427dce5c Author: Eric Anholt Date: Fri Feb 27 22:51:22 2009 -0800 intel: Nuke shared-entity support (zaphod mode). It's been broken for years now, and KMS offers a much better chance of getting this working sensibly without making a mess of the 2D driver. commit e9d6bbbe0b41e29c58a79844decd81771da85dd4 Author: Eric Anholt Date: Tue Mar 3 16:45:03 2009 -0800 remove more page flipping leftovers. commit 1cc9b1423c5df591c615ef9588b6eefd81448f80 Author: Lukáš Hejtmánek Date: Fri Mar 6 14:44:03 2009 -0500 Fix another VT switch leak The batch_bo buffer object is reallocated on enter VT, so we need to unref it on leave vt. Signed-off-by: Lukas Hejtmanek commit 67fef27f4b76490be085d232aba0ca9cbb3c5e59 Author: Xiang, Haihao Date: Fri Mar 6 09:40:07 2009 +0800 Xv: free tearing on textured video Add an Xv attribute XV_SYNC_TO_VBLANK which has three values -1(auto), 0(off) and 1(on) to control whether textured adapter synchronizes the screen update to the vblank. The default value is -1(auto). commit 0d20bbbc2005a51f427a9ae6b6a66dbbb101dbab Author: Zhenyu Wang Date: Thu Feb 19 14:24:24 2009 +0800 SDVO: handle multifunction encoder (try 2) For SDVO encoder that advertise multiple functions, we have to get attached display to determine current output, and update output's name according with current type. commit d4c64f01b9429a8fb314e43f40d1f02bb8aab30f Author: Lukas Hejtmanek Date: Wed Mar 4 17:33:27 2009 -0500 Fix serious memory leak at Enter/LeaveVT This fixes huge memory leak at each VT switch (about 600 BOs + 6MB of RSS of Xserver). commit 095a001f755201d3c19335b67a84c57b1d080a83 Author: Kristian Høgsberg Date: Wed Mar 4 17:10:24 2009 -0500 Use drmSetMaster() and drmDropMaster() in enter/leave VT This allows multiple X server to use DRI and makes it possible to run multiple X servers under KMS. This requires a 2.6.29 kernel to work. On older kernels it will just log a warning and DRI will fail to initialize for the second X server. commit 14bb61e0c2e28725a2f6167d3263649bc845be18 Author: Zhenyu Wang Date: Mon Mar 2 16:55:45 2009 +0800 SDVO: only check digital monitor when EDID exists commit 42e34e90e2e4048b38481cab61cef46f932eada7 Author: Zhenyu Wang Date: Tue Mar 3 22:55:35 2009 +0800 TV: add property control for TV attributes This is based on Jesse's origin patch for bug #12763. But export integer range to user instead of hardware float point format, and fix different real format on 965G and 945G for contrast and saturation. commit aa9da5e393c804019720503fe58bdd247fe1eabd Author: Zhenyu Wang Date: Tue Mar 3 20:26:19 2009 +0800 TV: add option to set TV connector type This can let user override non-stable driver TV load detect, and set connector type manually, e.g for s-video to component converter, this patch seems must needed to use HD modes. commit ab1ef05cc0fd5ba9768c26cc51bc2c7b52baa45f Author: Zhenyu Wang Date: Thu Feb 19 20:23:02 2009 +0800 TV: update output and crtc modes when TV format change (try 3) This is to fix bug #16566, change TV format will cause BadMatch error when crtc config apply. Everytime when we change TV format, we may generate a new list of modelines as TV clock changed. After randr get info request, new modelines will be probed and randr output's modes will be renewed too. But crtc's mode failed to be updated, as it never can find a matching mode now within new modes list. So get info will return an invalid crtc's mode, later set crtc config will pass that info, and xserver catches a bad match. This patch trys to refresh output modes and setup crtc's mode with new modelines in TV format change. So get info would be sure to turn valid crtc mode that reference in current new modelines. commit a67a911a93ff3c0b3d2e6f6fb20c0787b9d4d41d Author: Zhenyu Wang Date: Thu Feb 19 01:00:33 2009 +0800 TV: sort input mode lines commit bd360e7517835626bee54bec968cdc78908c6545 Author: Adam Jackson Date: Mon Mar 2 10:33:35 2009 -0500 KMS: Wire up output DPMS. commit 2d0aa553609a0e64fa4b2e755b9b1e244e5c3fa1 Author: Zhenyu Wang Date: Mon Mar 2 15:26:44 2009 +0800 Skip LVDS config parsing on pre-9xx chips. 855GM laptops seems pretty broken when parsing this block. commit 55359ef9ab0a2be42a55e0279835f76a191d6c74 Author: Zhenyu Wang Date: Fri Feb 27 09:02:30 2009 +0800 Update driver feature block definition for missed fields commit 38a7683561cee7fffab174c2a166bfd51b51ba27 Author: Eric Anholt Date: Fri Feb 27 22:55:44 2009 -0800 warnings cleanup commit 5bfd73cd31ba197a62f549cdbad1a1270b571027 Author: Eric Anholt Date: Fri Feb 27 19:09:49 2009 -0800 Only allocate pixmaps aligned for tiling when requested by DRI2 GetBuffers. This saves massive quantities of memory on pre-965 since the DRI2 tiling enable caused the minimum size of any pixmap to be 1MB. commit f53bdad1412f841075232455837578f00709c6ef Author: Shaohua Li Date: Mon Feb 23 15:19:23 2009 +0800 Intel video driver patch This is the intel video driver patch for a new chip, which is G33-like and has some clocking setting related register changes. This patch adds the pci id and DPLx/FPx register changes. The gtt tool should just work to me, as the chip hasn't any changes against G33 on this side. Signed-off-by: Shaohua Li Signed-off-by: Eric Anholt commit 22dc9a5580d77cc4707bfb8e19474e611a06ae9a Author: Eric Anholt Date: Tue Feb 24 18:22:20 2009 -0800 Fix UXA for server 1.4. commit cb1f7ec0876746c1b52b63cdb508544e9e4e32e3 Author: Eric Anholt Date: Tue Feb 24 21:23:59 2009 -0800 uxa: Fix composite fallback debug printing of main memory versus bo info. It was just printing whether it was a pixmap (it is), instead of whether the pixmap was offscreen. commit 635eaa511f28fb673fe306e46ed5370e78a8a534 Author: Eric Anholt Date: Tue Feb 24 21:48:11 2009 -0800 Regard the screen pixmap as suitable for acceleration. With UXA on the fake bufmgr, the screen pixmap doesn't have a BO and so no acceleration was occurring. commit 70e0261208654c6c875ad462da2734c6aa9eeb96 Author: Eric Anholt Date: Tue Feb 24 20:54:05 2009 -0800 Disable fb resizing for DRI1-only server so that DRI1 can initialize. commit d7aa330db31100b7cb54d8165f9a4b94329ece32 Author: Eric Anholt Date: Tue Feb 24 20:34:10 2009 -0800 Fix up i915 composite and common solid/copy code to use check_aperture. This could fix complaints about binding BOs and resulting failure to render. commit a625a07022ea7f6757b288fcc2ffb9e27c7f8341 Author: Xiang, Haihao Date: Wed Feb 25 16:48:22 2009 +0800 XvMC: fix broken xvmc on 965 commit f6d8ae69b0f97e696c142f06c8038f336ed024f9 Author: Zhenyu Wang Date: Wed Feb 25 09:57:00 2009 +0800 Use LVDS config in Driver feature BDB for integrated LVDS check The LVDS config bits in VBT driver feature block is used by vendor to identify the board implement of integrated LVDS/eDP or SDVO LVDS. And video bios uses these bits for LVDS enabling or not. So check these bits for integrated LVDS might eliminate more quirks. commit 8718551f14e064b461e80a583597f0ea9fb8ca9f Author: Zhenyu Wang Date: Wed Feb 25 09:54:03 2009 +0800 Update LVDS config bits definition in driver feature block ALL_LVDS is actually not defined before and include GM45. Embedded DP bit will be used for newer chips. commit 9d8e5c21a1688b915bf39261d4c3b0bf2906daef Author: Eric Anholt Date: Tue Feb 24 14:05:15 2009 -0800 Fix distcheck from drmmode_display.h deletion. commit f4e2f522a5c5b03ea530b9eb67e9d1a9a96274ce Author: Eric Anholt Date: Sat Feb 21 20:36:58 2009 -0800 Don't do AdjustFrame in KMS mode. This was hit by xv86vm's SwitchMode path, and for that the CRTC offsets get set at mode setting time anyway. commit 170f00e161931fdaa8c2812fc710649e1d6d977a Author: Kristian Høgsberg Date: Tue Feb 24 14:07:23 2009 -0500 Remove a handful of unused variable warnings. commit a6b31f38ebf470c61de0e10b0ce2af0d7ee1684b Author: Kristian Høgsberg Date: Tue Feb 24 13:58:20 2009 -0500 Update kms to work with drmModeModeInfo API update. commit 13ee9402e9822d6f57e3ebcc5ae658ce8322118e Author: Adam Jackson Date: Mon Feb 23 15:44:26 2009 -0500 Limit CRT DAC speed better. Verified against the public docs for i8xx parts, although not 9xx yet. commit 668b2352a47bcfba75fe0492a5805726222755eb Author: Keith Packard Date: Mon Feb 23 13:31:51 2009 -0800 Revert "Limit CRT DAC speed better." This reverts commit 8fd0e46571c7ba15c05f0a759113f8ca842c76a2. This doesn't even build. Please try again. commit 8fd0e46571c7ba15c05f0a759113f8ca842c76a2 Author: Adam Jackson Date: Mon Feb 23 15:44:26 2009 -0500 Limit CRT DAC speed better. Verified against the public docs for i8xx parts, although not 9xx yet. commit 73bc7f113969834d00cd92be8374dbadc62f96a9 Author: Kristian Høgsberg Date: Mon Feb 23 15:16:51 2009 -0500 KMS: Fix bug that prevented EDID data from getting propagated. commit ef952760551ad15cb9f63025d1e087645949a227 Author: Kristian Høgsberg Date: Sun Feb 15 22:10:42 2009 -0500 Use stat() and the dev_t to find the drm device filename. Simpler and more robust. Works when we haven't set the device PCI ID, ie when not going through the legacy DRI module. commit 81c652e9a666a7459bcc5217c8a5ec518b6e00da Author: Helge Bahmann Date: Sat Feb 21 10:10:04 2009 -0800 Move disable_render_standby to EnterVT instead of startup. Otherwise, with a pre-2.6.28 older kernel the disable would be lost at resume time and cause hangs. Bug #20214 commit 0621ba12a3b694720e67a49b25ca52f0e09b3802 Author: Eric Anholt Date: Wed Feb 18 13:32:44 2009 -0800 uxa: Ask for BOs ready for rendering for pixmaps. The assumption is that we're almost always accelerating our drawing to new pixmaps (fill, copy, etc.). commit 5018d0f16cb8b44c743b5b37d194fe806d955568 Author: Kristian Høgsberg Date: Fri Feb 20 15:34:29 2009 -0500 KMS: Hook up rotated shadow buffers. commit beca598bc2848093b710bd47828d622205d273df Author: Kristian Høgsberg Date: Fri Feb 20 11:08:51 2009 -0500 Access the Xv buffer through the GTT for the non-KMS case. commit e97e2571703e3d6188bf18f211b793fc50383f9c Author: Kristian Høgsberg Date: Thu Feb 19 16:41:49 2009 -0500 Fix i915 textured video to work with the i830_memory -> bo change. Forgot to update i915_video.c in 872aadc7102bd5131e1582ede081e22672911ba2. commit 96da26b6813a8c1da8a43036c375aa0d2bb70f16 Author: Kristian Høgsberg Date: Thu Feb 19 16:40:19 2009 -0500 Dont allocate overlay registers in KMS mode. commit 872aadc7102bd5131e1582ede081e22672911ba2 Author: Kristian Høgsberg Date: Wed Feb 18 17:26:06 2009 -0500 Make Xv used a buffer object instead of i830_memory. We still pin the buffer object in case of overlay, but for textured video we're now no longer using i830_memory for Xv anymore. commit 527e8177cde3abbabbcdccee0dbc0dcc0068a1be Author: Zhenyu Wang Date: Wed Feb 18 11:26:58 2009 +0800 bios_reader: parse driver feature BDB commit 62ca1c479825ffa0e9cf444b4e25080150faa45b Author: Zhenyu Wang Date: Wed Feb 18 10:49:00 2009 +0800 Fix SDVO mulitiplier setting for 945G Wrong SDVO multiplier setup has been slipped in SDVO TV patch. Thanks Michael Fu to point this out! commit 723b6065093adb56a2d7204bd990ceae41bfafc9 Author: Kristian Høgsberg Date: Tue Feb 17 13:48:04 2009 -0500 KMS: Hook up crtc::gamma_set. commit 506bbb8341c052499057834a669b82787454b426 Author: Zhenyu Wang Date: Mon Feb 16 14:09:49 2009 +0800 Safely init SDVO found variable Found by Michael Fu for my last SDVO detect fix. commit 9d464bd5b0d2724f5edb26e859888ceb6a248f9b Author: Zhenyu Wang Date: Fri Feb 13 11:15:48 2009 +0800 Fix SDVO/HDMI detect SDVOC detect bit is only valid for HDMIC. And for SDVO devices, SDVOB detect bit should be used to probe all possible SDVO outputs. commit 48445d2e939328495b4abe0fb7e579dfcef727bb Author: Zhenyu Wang Date: Fri Feb 13 11:11:18 2009 +0800 SDVO: remove ForceSDVODetect option Which is just a hack to hide our SDVO detect drawback, we will have SDVO/HDMI detect fix later. commit ddedf19f889da2ce6d69a3afce4665e2245682fa Author: Zhenyu Wang Date: Fri Feb 13 10:09:42 2009 +0800 SDVO: Switch control bus only before DDC access Instead of set control bus switch before every i2c start, just set once before doing DDC. This should eliminate some encoders returning error during that. commit f1ca56e17d0ecd4f1299061a6b3272bfd289e123 Author: Zhenyu Wang Date: Fri Feb 13 10:02:02 2009 +0800 SDVO: Fix TV support As SDVO TV uses SDVO_TVClkIn from SDVO encoder for clock reference, it needs to generate proper PLL for current input clock. This uses fixed PLL table from vbios for this. And possible sdvo mulitiplier has to be setup correctly. This makes TV output stable on my 945GCLF2 board with NTSC-M format. commit acde0ef683d6ec33d0b478923ffb11bd6785798f Author: Zhenyu Wang Date: Fri Feb 13 09:53:57 2009 +0800 SDVO: fix CREATE_PREFERRED_INPUT_TIMING command commit 824b2f0c5530c3196901c961757e6677b042caf3 Author: Zhenyu Wang Date: Fri Feb 13 09:50:45 2009 +0800 SDVO: fix usage for SET_TV_FORMAT and GET_SDTV_RESOLUTION_SUPPORT command They both needs parameters. commit 62c0c2f5549a51c5df209f7353a19ca301f221be Author: Zhenyu Wang Date: Fri Feb 13 09:48:34 2009 +0800 SDVO: fix error in modeline and DTD convert commit 37c67088a887e6380571e6eec8a8f058e3e24717 Author: Zhenyu Wang Date: Wed Feb 11 15:15:27 2009 +0800 SDVO: check EDID info for DVI-I For SDVO DVI-I, check EDID info for digital output, otherwise mark it to be disconnected as analog output is driven by VGA then. commit 38079bc0f1038da77048bbf6e5c10758f9fb8a55 Author: Zhenyu Wang Date: Wed Feb 11 14:41:58 2009 +0800 SDVO: Fix for HDMI encode and audio setup (try 5) SDVO HDMI encode and audio is not setup in detect, which fails in hotplug case for HDMI audio. Fix to check current encode type and set flag for HDMI audio enabling. Check and set HDMI encode state in get_modes. commit 3012d85cc5eb58c2447e93c05c39dc14feaae988 Author: Eric Anholt Date: Tue Feb 10 18:47:28 2009 -0800 uxa: Fix breakage from UXA_FALLBACK conversion from "do {} while (0)" construct. Thanks to keithp for post-commit review. commit 5009127de7d9527ae329d1c2fbc7283648bde2e6 Author: Eric Anholt Date: Tue Feb 10 18:23:35 2009 -0800 uxa: Fix driver against fbDoCopy -> miDoCopy change in the server. commit b53977f4c53c7c8f562f909e985b8d5a7b2526f3 Author: Eric Anholt Date: Tue Feb 10 18:22:46 2009 -0800 uxa: Fix failure to --amend in further changes in previous commit. commit 5212ec6515c6562f66b86fc16928b601bf04e49b Author: Eric Anholt Date: Tue Feb 10 15:35:20 2009 -0800 uxa: hook up the fallback debug to the driver's fallback debug option. commit 3aa8591abfbe8db0f13912910c850fdd748808df Author: Ma Ling Date: Fri Feb 6 09:14:15 2009 +0800 Don't disable vga centering bit. commit id b9f5915ce812144ffd9d2aa42e8ba856129c35e, which resolved bug #17235, but generate new regression-bug #19715. This patch intends to resolve bug #17235, and avoid regression as well. We have successfully re-tested it for bug #17235 and #19715 respectively. commit 9fe5fca3fe761a4f11857d9766138a60f471a9e6 Author: Zhenyu Wang Date: Wed Feb 4 06:00:39 2009 +0800 TV quirk for HP Compaq nx6310 commit 5c370091620b38447172ebeffbc6ed3256e86c9d Author: Kristian Høgsberg Date: Mon Feb 2 12:02:00 2009 -0500 Fix front buffer memset() for non-KMS case. Missed the pI830->FbBase condition when removing the KMS hook. commit 2013799b20599a58de48cb21a5a389e898a58af1 Author: Kristian Høgsberg Date: Mon Feb 2 11:51:47 2009 -0500 Un-revert the I915_SETPARAM_NUM_USED_FENCES commit reverted by accident. Oops, my bad. Reverted 8d4bc36fae50b09a73ba2cfab920adb32141a358 since my kernel doesn't yet have the new param, committed the revert by accident. commit 127330bfd53ac7571bdd12a551142528b972893f Author: Kristian Høgsberg Date: Mon Feb 2 11:02:59 2009 -0500 Fix last-minute "cleanup" that broke the patch. commit 0cb87ccfe97b0e016e47dcf236fd5ce78dddfc4b Author: Kristian Høgsberg Date: Fri Jan 30 17:53:03 2009 -0500 Implement front buffer resize for KMS. This adds back the resize hook so we can resize the front buffer under kernel mode setting as well. The patch also pulls the drmmode_* structs from drmmode_display.h into drmmode_display.c and eliminates the header file. commit 66bc44e8f9a0505c0b11b8042243ca74079da85f Author: Eric Anholt Date: Tue Jan 27 15:43:58 2009 -0800 dri2: Use modesetting's master fd instead of opening our own non-master. This fixes failure to auth DRI2 clients under KMS. commit 8d4bc36fae50b09a73ba2cfab920adb32141a358 Author: Jesse Barnes Date: Mon Jan 26 17:14:06 2009 -0800 Support tiled back/depth on 915-class hardware with DRI2. Set alignments, tile settings and flags correctly in the 2D driver to support tiled rendering. UXA's create pixmap function currently assumes the worst about the alignment constraints; that should probably be fixed. Some of the 1M alignment fixes could probably be done more cleanly as well. commit 6c0ca1676bf60529dd331cc739abdf68fa9e918d Author: Eric Anholt Date: Wed Jan 21 15:52:02 2009 -0800 Don't forget the new state bos in check_aperture. They're tiny so it shouldn't have been a problem, but play it safe. This is another <5% loss on top of the previously reported value, bringing the whole series to about 8%. commit 57a02b50c60c10a25ff0f3cd93af9f37fa0d1b11 Author: Jesse Barnes Date: Mon Jan 26 14:58:28 2009 -0800 Fixup bogus VBT modes when detected Several VBT modes out in the wild have H or VSyncEnd values greater than the H or VTotal value. This clearly ends up creating a bad mode, causing some panels to either ignore the timing or display some sort of corrupt image. Check for these cases and fix them up by default, making things work for several Dell and Sony machines. Fixes FDO bug #17292. commit e20e1cf76fb00ba4f933a1ed6d1a4896be346c91 Author: Bill Nottingham Date: Sat Jan 24 08:36:20 2009 +0800 Quirk MSI IM-945GSE-A LVDS, TV outputs. The IM-945GSE-A claims to have a TV output, and always claims a connected LVDS output. It has neither. commit 05ff561234cc2b93fe1ea2a35041fa2e119a7e38 Author: Vincent Mussard Date: Sat Jan 24 08:33:16 2009 +0800 quirk for AOpen MP45 commit fbf003ef2767a1a9f5e4064f04a17992030d8f5c Author: Eric Anholt Date: Fri Dec 5 17:27:13 2008 -0800 Move i965 render sampler state to BOs. This eliminates the pinned memory allocation for 965 render state. commit befd4ad8beae39377f804e06c7cbd926ec4251db Author: Eric Anholt Date: Tue Jan 20 14:43:08 2009 -0800 Move i965 render kernels to BOs. commit 5d705de5d11297f4d6bd5237fb67619e703745bc Author: Eric Anholt Date: Fri Dec 5 15:30:35 2008 -0800 Move 965 render unit state to BOs. This is a first step in a series of changes to avoid requiring a pinned object, which gets in the way of doing non-root KMS. This change appears to result in about a 2-6% loss in x11perf -aa10text, which better algorithms in libdrm could make up for (it hasn't really had to deal with code this bad before). commit 64b08ed5ad46ccd76964972e39bfed8721a920aa Author: Eric Anholt Date: Tue Jan 20 17:13:52 2009 -0800 Remove 965 render wm scratch space, which was just unused. commit f126aabdf8952177bb15f392041da7a7094eb31b Author: Eric Anholt Date: Wed Jan 21 12:52:36 2009 -0800 Fix build with server 1.4. Debian "unstable" is still stuck with this ancient version. commit 253b8db298f38676e47dc902534465054f7b58b8 Author: Jesse Barnes Date: Wed Jan 21 09:02:13 2009 -0800 Don't run in KD_TEXT mode even with KMS Leaving the VT in KD_TEXT mode keeps the kernel's blanking code active, so when a DPMS event happens, the fb console is restored rather than X's configuration. On the downside it means the kernel won't print messages in the background, which would be visible if a panic or emergency switch occurred. The proper fix here is a new kernel mode, which we can move to when ready. commit 131b414feb2ecabe31b538d65725ac4427a4387a Author: Jesse Barnes Date: Wed Jan 21 09:01:05 2009 -0800 Tear down batchbuffers unconditionally on LeaveVT Even if KMS is enabled we should do this, to avoid running batches that depend on other state we tear down in LeaveVT. commit b6f3ce32e295929f461a7bc37e61f126fb51e4aa Author: Eric Anholt Date: Tue Jan 20 10:45:23 2009 -0800 Use drm_intel_bo_subdata to put render vb data in. This improves performance by avoiding repeated map/unmap cycles, which are a bit expensive on my machine with lock debugging on in the kernel. It could do much better if we did more than 18 or so floats at a time. commit 9a8bbb1951ad0ca0a9407a97348fc7fa03127900 Author: Eric Anholt Date: Mon Jan 19 19:34:50 2009 -0800 Move i965 render vb setup to use time, and decouple state emit from it. The require_space had failed since it only checked for the space required by the batch emits in the function itself, but not in the i965_emit_composite_state() that it called (the state we were concerned about having set up for that 12 * 4 dwords to follow!). This is replaced by intel_batch_start_atomic(), which will catch such mistakes in the future. commit 3d739597c4f5817079efd9067ad5db2f4105f765 Author: Eric Anholt Date: Tue Jan 20 10:37:20 2009 -0800 Move i965 render transform setup from emit_composite_state to prepare_composite. commit e20f7278f3abb44a3a151ac91f83c45cf1a2745a Author: Eric Anholt Date: Mon Jan 19 20:31:31 2009 -0800 i965: Pull check_aperture out to a separate function and make it dtrt. Previously it wouldn't count the pixmaps that were about to be used, which is pretty much the only purpose of having the pain around. This also eliminates the check_twice confusion with emit_batch_header_for_composite(). commit 013e2adfbf955cb21450b610091542ebd54392c2 Author: Eric Anholt Date: Mon Jan 19 20:24:20 2009 -0800 Move filter computation from emit_batch_header to prepare_composite. commit a340fe5e4227ebea5493e658eb6289624b07ab0b Author: Eric Anholt Date: Mon Jan 19 19:11:41 2009 -0800 Use intel_emit_reloc from video to prettify 965 render bind_bo setup. commit aefe198ca427a5ad69717f49948eb3ede713bb28 Author: Eric Anholt Date: Mon Jan 19 18:57:01 2009 -0800 Move i965 render state bo setup back to prepare_composite. We want the objects to be created once per prepare/done both for efficiency and so we can handle aperture checking better. commit 946c7ef8170e74ac178c83b1465242d57fa86f2e Author: Eric Anholt Date: Mon Jan 19 14:43:20 2009 -0800 Do check_aperture_space and batch_start_atomic for i965 video. This increases the overhead for video in the presence of cliprects, but we were already doing nasty things in that case and don't seem to care. This could fix potential bad rendering or hangs with video, particularly with DRI2. commit 7be668179a12918918cad863f6936ced4ab78dbf Author: Eric Anholt Date: Mon Jan 19 14:29:25 2009 -0800 Move 965 video setup to a separate function so we can move it around. commit c7db3201106f07f3228c989014d6db5ace5378f6 Author: Owain G. Ainsworth Date: Tue Jan 13 18:46:41 2009 +0000 Remove the pageflipping infrastructure. It was broken on current kernels, and deprecated anyway. commit c82adfd0169317efb6c2f9de0f315651f9adbae1 Author: Owain G. Ainsworth Date: Tue Jan 13 17:09:00 2009 +0000 Remove triple-buffering support It never worked with any upstream linux kernel, and is quite heavily deprecated. A new solution based around DRI2 will probably be forthcoming. Pageflipping itself is next. commit ada44c1c0edcd3ea9e41ed6b6fdb2bf0e87c9c67 Author: Kshitij Kulshreshtha Date: Tue Jan 20 11:35:36 2009 +0800 Support sysfs backlight control for Sony laptops in xf86-video-intel The sony_laptop kernel module (since v2.6.23) supports backlight control via the sysfs interface: /sys/class/backlight/sony This patch will enable xf86-video-intel to use this backlight control method for Sony VAIO Laptops with Intel integrated video. commit d89de6d60a86105a198d904821853e6ed7de3305 Author: Eric Anholt Date: Fri Jan 16 16:59:17 2009 -0800 Protect i915 textured video against batchbuffer wrapping. commit 15780c53f5717936ea10ac87aff8a881c172f1dc Author: Eric Anholt Date: Sat Jan 17 16:23:38 2009 -0800 Fix i915 batch_start_atomic counting. commit fc46cb6403387215d13aa28d720f205688ae9d67 Author: Eric Anholt Date: Sat Jan 17 12:47:53 2009 -0800 Fix libdrm version number requirement regression that got spammed in. commit 1459f794e2a4b9032807c3794b00a33fa6392837 Author: Dave Airlie Date: Sat Jan 17 22:14:26 2009 +1000 intel: fix DRI2 should be DRI_DRI2 commit db43b7870a37ea273941302096a6f00120dfae71 Author: Eric Anholt Date: Thu Jan 15 19:05:52 2009 -0800 Re-emit i915 composite setup when the batchbuffer wraps. This also introduces tests to make sure that we asked for enough reserved space and that we don't allow wrapping at the wrong time. This fixes a hang during text rendering with DRI2 and a GL client running, but could potentially affect text rendering with GEM in general with an exceptional batchbuffer setup. commit cab5b7a7b0e17414f98b2363b0961c87f32f9c05 Author: Eric Anholt Date: Thu Jan 15 09:31:55 2009 -0800 Fix invarient state emits for DRI2 (do it per batch, since there's no lock). commit 9f306193c4b128ec72c5b4db16d07302cee50ab5 Author: Owain G. Ainsworth Date: Tue Jan 13 18:42:44 2009 +0000 Fix ioctl type. I915_EMIT_IRQ is a read/write ioctl, not a write only. Found by OpenBSd's kernel code which checks these things a long more strongly than Linux. commit 1e8588ad5087c69eb77399cfaab8e4ec15eb4da9 Author: Owain G. Ainsworth Date: Tue Jan 13 17:02:59 2009 +0000 use ifdef __linux__ where needed. since modesetting is compiled by default now, ifdef __linux__ the linux only includes and ioctls. commit db9f5915ce812144ffd9d2aa42e8ba856129c35e Author: Ma Ling Date: Wed Jan 14 14:46:52 2009 +0800 Disable VGA plane reliably This fixes #17235, VGA random hang on recent G45/43 board. From spec, SR01 bit 5 should be set before VGA plane disable through control register, otherwise we might get random crash and lockups. commit a4b023c17b9c3bd65fb9466ddb8a953f60244402 Author: Keith Packard Date: Tue Jan 13 20:01:49 2009 -0800 Assign rotation memory dri_bo to rotation pixmap. As the rotation memory and rotation pixmap are allocated separately (to make rotation at startup work), the allocate dri_bo needs to be set in the pixmap for acceleration to work. This restores the performance in rotated modes. Signed-off-by: Keith Packard commit f659cc37e61364a408355b9a6a44d39f4d759935 Author: Jesse Barnes Date: Tue Jan 13 16:49:41 2009 -0800 bios_reader: make mode timing output friendlier Print out the calculated mode line values (as i830_bios.c uses) and check for validity against known problems. commit c80f1a9c5131721eaf87d12a7a67e603cdfed63b Author: Bernhard Rosenkraenzer Date: Tue Jan 13 10:35:19 2009 -0800 UXA: Declare glyph cache picture as component-alpha when necessary. Without this, rendering component-alpha glyphs may break without a mask. Bug #19534. Ported from fix by Michel Dänzer in xserver commit 639f289dcdbe00a516820f573c01a8339e120ed4 commit 3d206f9e46b5237bda7ca3c0f92d64c45ee8bdf5 Author: Ma Ling Date: Tue Jan 13 10:26:40 2009 +0800 set continuous-frequency flag in get modes function http://bugs.freedesktop.org/show_bug.cgi?id=19247 Because latest xorg will check whether the display is continuous frequency through one flag in monitor info structure, if not it doesn't touch default modes. When laptop failed to fetch edid, We don't set the flag. In i830_lvds.c, so currently we can not get default modes except only one mode line from bios. In order to achieve default modes from xserver successfully,I set the flag and other EDID features. commit 3354e660b0744976871683ce226f17e873f26b50 Author: Keith Packard Date: Wed Nov 26 16:32:12 2008 +0800 Fix LVDS EDID to match all possible default modes If the EDID data from the LVDS doesn't indicate support for a wide range of continuous frequencies, it will not match any of the default modes although our LVDS scaler logic ignores refresh rates when programming LVDS modes. Fix this by smashing the compute EDID data to open up the sync rates. Signed-off-by: Keith Packard commit f1e9ca4e4fb3ddb083252aea79c67c5e5e71f29c Author: Alan Coopersmith Date: Fri Jan 9 16:30:15 2009 -0800 Remove xorgconfig & xorgcfg from See Also list in man page commit 91c49067abab262edce54d33af8bfc03eee20f2e Author: Ian Romanick Date: Wed Jan 7 21:45:54 2009 -0800 Prevent redefinitions of CARD8 and friends. Fixes build. commit 934008a2cbcec0c790580750c672c0367a9e4c55 Author: Eric Anholt Date: Tue Jan 6 10:55:59 2009 -0800 Always enable KMS if server's new enough, and remove option. The API should be stable at this point, and we don't want to allow mistakes. commit 342120be0956bfc12822d1ffbfbd8aaabf3e922f Author: Eric Anholt Date: Mon Jan 5 23:21:07 2009 -0800 Fix pin leakage with EXA GTT-mapping shortcut, and crash with UXA on KMS. commit 9a5082d2920c1a212fe935b5a093013e8035c321 Author: Eric Anholt Date: Mon Jan 5 23:28:50 2009 -0800 Disable DRI2 buffer tiling on non-965, as those need fence regs for 2D blits. This fixes glReadPixels failure on single-channel 915GM, as the software code for readpixels was actually the only code in the driver doing tiling against these buffers (everything else says "rely on fence registers", since the 2D blits don't have a "don't rely on fence registers" option). commit 7736b65be4fb4d5c59d7aedb1e64da976bb10ae9 Author: Keith Packard Date: Mon Dec 15 15:49:01 2008 -0800 FatalError on batchbuffer map failure Yes, it would be nice to do something other than crash here. Signed-off-by: Keith Packard commit 632f816c72cb4b48b690fd92d1cc1d5a9c9285c7 Author: Keith Packard Date: Mon Dec 15 15:43:34 2008 -0800 uxa: handle uxa_prepare_access failure uxa_prepare_access may fail to map the pixmap into user space. Recover from this without crashing. Signed-off-by: Keith Packard commit 90b28a56553d809374fa6d9b9529b7a8b583488c Author: Keith Packard Date: Mon Dec 15 15:41:28 2008 -0800 Handle drm_bo_map failure in 965 video and composite paths. These two paths allocate a number of objects directly. Signed-off-by: Keith Packard commit 8237faf8f3ca73ecdf0ef009a7d361b318726f6f Author: Keith Packard Date: Mon Dec 15 15:35:35 2008 -0800 Resize framebuffer on screen size change (requires UXA and DRI2) Signed-off-by: Keith Packard commit 21bd4e8974e4c0e83f5f95adb0fc17290444caf5 Author: Keith Packard Date: Mon Dec 15 15:19:00 2008 -0800 Make i830_allocate_memory take tiling parameters. This eliminates the separate i830_allocate_memory_tiled function which means that all memory objects will have tiling parameters set correctly. Signed-off-by: Keith Packard commit 1f61e97904dfe5f8c08bb9f284cfdfe878f7e541 Author: Zhenyu Wang Date: Wed Dec 31 22:56:57 2008 +0800 UXA: Fallback to dri_bo_map() if pin failed This fixes VT switch issue with UXA after Eric's aae4008096399a0e84abc7c016b35092caf9db25 on 2D side. commit 830bf916724afd21b7947f797c22a8c8aab7a0a4 Author: Eric Anholt Date: Mon Dec 29 13:42:44 2008 -0800 Don't touch the pipestat regs for detecting FIFO underrun. The kernel owns them. Since we don't perform any synchronization with the kernel on these regs, we could race with the kernel to write stale values and end up not having vblank interrupts enabled when somebody was waiting on one. commit d96f774d1bb39640486c72338fe8b19ee1ceaa23 Author: Eric Anholt Date: Mon Dec 29 13:57:24 2008 -0800 warning fix. commit 59b0fbb9be880d489374b141f818948a2721a2ef Author: Henry unbongo Date: Mon Dec 29 13:54:38 2008 -0800 Add support for SDVO LVDS. commit 750d8e105831718d4a44a145fdb87571fa9f9d8e Author: Eric Anholt Date: Mon Dec 29 12:44:26 2008 -0800 Fix compile failure after 45f45c73469f1bd46a1b6fb206f2e9e5e4fd66b3 commit 45f45c73469f1bd46a1b6fb206f2e9e5e4fd66b3 Author: Eric Anholt Date: Mon Dec 29 11:26:11 2008 -0800 Remove logical context setup. This should be a noop. If it wasn't a noop, it means that on pre-g33 chipsets we were spamming some data into a page of system memory because we used a virtual instead of a physical address. It was also supposed to not work when we submit it from a batchbuffer, as we have been doing for some time now. This code has existed since about the beginning of the driver's existence, with no justification. commit 3544bbe22d8cf2640289e1e4febe755a47f26631 Author: Eric Anholt Date: Mon Dec 29 10:41:02 2008 -0800 Add PCI write posting to LeaveVT path when we're about to wait on write results. commit 7b67914b23b54d4d9566190440a3430e40615aa8 Author: Eric Anholt Date: Fri Mar 28 15:51:50 2008 -0700 Add SDVO LVDS register definitions. commit 8464fc285d7b22fd45b7af616fd52aa15e16638a Author: Eric Anholt Date: Mon Dec 29 09:51:45 2008 -0800 Rely on libdrm 2.4.3 and stop checking for xf86drmMode.h. commit c1dde7ac06ce6470c74198b2560ee67d28fb0aea Author: Eric Anholt Date: Mon Dec 22 16:49:57 2008 -0800 Remove old mergedfb includes and defines, which bother spatch. commit 75799d2834be84b016391ec95b221df32380e3e7 Author: Zhenyu Wang Date: Tue Dec 30 00:55:30 2008 +0800 Bug #18004: Add Aopen 915GM LVDS quirk commit cfaaf6af777ad8e56da5a077bdc01f4f2d7bc4d6 Author: Zhenyu Wang Date: Mon Dec 29 21:17:44 2008 +0800 Don't count vertex buffer in second aperture size check With batch flush notify vertex buffer will be unreferenced, so don't count it in later aperture check. Also adding uninitialized vertex buffer check in batch flush notify. commit b710a688a7383df320f9d4e765b48331310a4d1d Author: Dave Airlie Date: Mon Dec 29 12:50:25 2008 +1000 intel/kms: disable overlay when it needs physical address We can't do phy address allocations, need kernel support for this. Signed-off-by: Dave Airlie commit 649374b88b330838133d78be1953ce8b18ddd2c6 Author: Dave Airlie Date: Mon Dec 29 12:48:11 2008 +1000 intel: don't call enter/leave VT for KMS enabled systems Signed-off-by: Dave Airlie commit 408f7139354a942f29334795b5480a0b6bb6142a Author: Dave Airlie Date: Mon Dec 29 12:47:30 2008 +1000 exa: fix map gtt call to use current API commit fcd280b730f86ea54ecd89d3e037e71f8375b5f5 Author: Zhenyu Wang Date: Mon Dec 29 10:05:17 2008 +0800 Bug #19239: Add a quirk for broken ACPI lid state For broken hardware/bios with incorrect ACPI LID state, there's machine that can not be fixed in ACPI way, customed DSDT that reprogram _LID method to read EC state. Although this is ACPI issue, this quirk can be used to work around that. commit 6b9f421b792c701e909d81ae2b6e6a47fb069b0b Author: Dave Airlie Date: Mon Dec 29 11:51:58 2008 +1000 modeset: transformPresent is a new API member commit a320541e51818833a6a445707835fbf70e9babd4 Author: Dave Airlie Date: Mon Dec 29 11:34:51 2008 +1000 modeset: fix xf86CrtcRotate API change across server versions commit e38fd84fcccc18284b649a60b4cfd8e24eaf059d Author: Ma Ling Date: Mon Dec 22 10:35:52 2008 +0800 SDVO: reset pixel repeat in avi frame For #19115, the root cause is avi_if.u.avi.PR in i830_sdvo_set_avi_infoframe() belongs to element for interlaced mode based on CEA_861B, but currently we don't support interlaced mode. So it should be set as 0. commit d8e89b26ef5ef2c15e5d34162b14d279a7f0bb1c Author: Zou Nan hai Date: Fri Dec 19 09:34:14 2008 +0800 [965-xvmc] remove the vblank wait code, drm not support mutlple client to wait on vblank now. commit aae4008096399a0e84abc7c016b35092caf9db25 Author: Eric Anholt Date: Wed Dec 17 14:25:22 2008 -0800 uxa: Do a hack to use the aperture mapping instead of bo_map in sw fallbacks. Because of how fallbacky the uxa rendering core is, and our inability (without wfb in userland or page faulting in the kernel) to tell the kernel just where we're going to fall back, the clflush overhead can become outrageous, for example with emacs and xcompmgr. Instead of using drm_intel_bo_map, pin the buffer and do the fallback to the aperture mapping. This gets us the bad old performance that fb is designed for, instead of bad new performance. commit 74bc420940de4ed02705211ddebbc4033c032242 Author: Peter Alfredsen Date: Fri Dec 19 08:10:06 2008 +0800 xvmc: fix up needed libs commit 555eea5411cf8c725df5f1b4cb80198fa6a1225b Author: Zou Nan hai Date: Wed Dec 17 13:55:07 2008 +0800 wait vblank before render to fix tearing commit ed267072db7c58ee16a458fd3dc24ce3a2d0061d Author: Robert Noland Date: Wed Dec 17 09:27:30 2008 +0800 Fix drmOpen for non-linux 965 XvMC drmOpen by name only works on linux after falling back to groping around in /proc. This doesn't work on other OS. Signed-off-by: Robert Noland commit ecdd706873c1f990f4a78fbaecf7380411edabcd Author: Eric Anholt Date: Sun Dec 14 16:09:25 2008 -0800 uxa: Correctly prepare/finishaccess of stipple in ValidateGC (and only it) This avoids prepare/finish_access_gc overhead when we're not changing things (since GCTile is already handled) and get us the RW flag for the prepare on of the stipple pixmap so thing will be synced correctly. commit e6479f96e5d8da39fcbb5376c4a66a1f924ec4e4 Author: Kristian Høgsberg Date: Mon Dec 15 14:54:20 2008 -0500 Quiet some KMS warnings. commit 30fb0ef53e19b759715f1ee14b81b11c81d79045 Author: Zhenyu Wang Date: Tue Dec 16 00:39:45 2008 +0800 Revert " [965 xvmc] update dual prime g4b files" This reverts commit ea2b6b405e4c8b1bfb4bc568d0453a39a9194a8f. Duplicate with Keith's commit. No idea what's diff target of this one. commit ea2b6b405e4c8b1bfb4bc568d0453a39a9194a8f Author: Zou Nan hai Date: Mon Dec 15 09:19:43 2008 +0800 [965 xvmc] update dual prime g4b files commit d917583c19e2eb20a559eddaa100ce71d8fbe48c Author: Pierre Willenbrock Date: Mon Dec 15 09:08:30 2008 +0800 closedir only after finishing use of any results from readdir commit 865735d8408ee6b75be8fb1a8ab32b2a53dd3adb Author: Keith Packard Date: Fri Dec 12 11:07:20 2008 -0800 Dump out fence registers by default, add fence end registers as well In debugging the frame buffer resize code, I needed to see what the server was doing to the fence registers, so I added this debug code. Seems useful enough to include it. Signed-off-by: Keith Packard commit f3e59e59b54696667b127c82327e16b14d7bab22 Author: Zou Nan hai Date: Fri Dec 12 11:18:25 2008 -0800 Update binary versions of the dual-prime kernels Signed-off-by: Keith Packard commit 80d588e9c49719ec494e705edfc14c78908bbc3f Author: Zhenyu Wang Date: Fri Dec 12 18:27:29 2008 +0800 Don't warn on ring enabled in GEM commit 60c4ee9ece8ea57e61b1590dfeb69d08555e465c Author: Zhenyu Wang Date: Fri Dec 12 18:00:09 2008 +0800 Fix directRenderingType check Don't miss classic texture memory allocation in DRI. commit edf765155497d89ecac328a5b268ecf60e2f377c Author: Zou Nan hai Date: Thu Dec 11 15:41:59 2008 +0800 [965-xvmc] fix dual prime kernel, flush issue on G4x commit 80e2d90139dd99f50beb4f9353599608624b777d Author: Zhenyu Wang Date: Thu Dec 11 13:45:17 2008 +0800 Let lid status be unknown if no acpi lid object found commit 83377b543defdca7226d7c1a7794e4ff3d8b4c84 Author: Bryce Harrington Date: Thu Dec 11 09:38:27 2008 +0800 Pipe-A quirk for HP 2730p (bug #18852) commit 6c4e134a1a30785c2e5c6d57b21fde54a2dd3413 Author: Bryce Harrington Date: Thu Dec 11 09:34:15 2008 +0800 PipeA quirk for Quanta/W251U (launchpad bug #244242) commit 1e974ff6b8446ecd64677b3c9aba60ca850923cc Author: Zhenyu Wang Date: Wed Dec 10 10:54:31 2008 +0800 Try to always probe SDVOC on 965G/965GM Detect bit of SDVOC is reserved on 965G/965GM, instead of ignore SDVOC this trys to always probe it on these chipsets. commit f5f67e1b54e67b4bfc3db3482b2693211be81d63 Author: Zhenyu Wang Date: Wed Dec 10 10:36:07 2008 +0800 bug #17395: Quirk CRT for Sony VGC-LT71DB commit d8b764fbd27dc9c8b28386093931b8d38855bd19 Author: Keith Packard Date: Tue Dec 9 21:51:14 2008 -0800 Add RandR 1.3 panning support by supporting the crtc set_origin function RandR 1.3 panning support can use the regular mode setting interface, but that's really slow. Providing set_origin makes it nice and snappy. Signed-off-by: Keith Packard commit 4d7a95959d8223aec41550eb19f60b3edd7210a1 Author: Zhenyu Wang Date: Tue Dec 9 19:48:11 2008 +0800 Try to add LVDS detect support This one trys to use lid status for LVDS detect, which works when internal panel is not used as primary display alone, or there's no internal panel at all. ACPI button driver's lid state interface is preferred, and SWF state is also checked if ACPI method failed. commit 0fe61b0b7e3bbe8ced1b0ad2be72c438d200c64b Author: Zhenyu Wang Date: Tue Dec 9 19:19:58 2008 +0800 Remove Cappuccino SlimPRO SP625F 855GM LVDS quirk It breaks bug #18462 on IBM 855GM with same subdevice ids. commit bea98cdfd93fc1181a06c51e57fcab227ff4827e Author: Carl Worth Date: Fri Dec 5 15:42:53 2008 -0800 Set vertex_buffer_bo to NULL after unreference. Which is just being tidy. We initially were looking at this code path due to a report of a crash on server shutdown which started after this unreference call was added. Setting this to NULL apparently didn't avoid the crash, but it's a good thing to do regardless. commit ce7efc2e3676c8f80206415480dda91e5e021396 Author: Carl Worth Date: Thu Dec 4 11:41:02 2008 -0800 Don't smash fixed_mode if skip_panel_detect is set. Without this change, setting LVDSFixedMode to false is not effective as i830_bios_init calls i830_parse_panel_data which in turns sets a fixed_mode. To fix this we still call parse_panel_data to set the various lvds_options but we return before setting fixed_mode. commit e8b95efbf5d9c3a5b75b2bb8b5b51844b5fcdfbc Author: Carl Worth Date: Wed Dec 3 13:49:52 2008 -0800 i965: Add batch_flush_notify hook to create new vertex-buffer bo This avoids mapping a buffer object which is being referenced by a batch that has already been flushed, (which is a terribly expensive operation). On my machine this brings the performance of x11perf -aa10text from 85k back to 150k, (where it was before a recent kernel upgrade). Also, before this patch, when I used my X server actively performance would drop as low as 15k---hopefully that bug is gone with this change. commit 768f317cf0da4cd6682af2e71e71c3e130e05182 Author: Zhenyu Wang Date: Mon Dec 8 10:33:13 2008 +0800 Fix DRI2 compiling warning commit 95596f51503bb468364719aec9083d59999e34b7 Author: Zhenyu Wang Date: Mon Dec 8 10:30:12 2008 +0800 Fix TV compiling warning commit c47b6d1def917fad5ad2b5a3e4167edfd354f2c8 Author: Kristian Høgsberg Date: Sat Dec 6 21:20:52 2008 -0500 Make sure DRI/DRI2 can initialize properly with KMS. commit 70af658d4e94cc372f9e9c831611f70b3c1cecab Author: Kristian Høgsberg Date: Sat Dec 6 19:19:21 2008 -0500 Simplify crtc preinit a bit. commit 3ad9c9a82d7b359b9b711070628e6ff07a2aa9f7 Author: Kristian Høgsberg Date: Sat Dec 6 19:17:39 2008 -0500 Fix KMS compilation. commit 2e3c098c5ed9a8451713dc754a5f086992249336 Author: Eric Anholt Date: Fri Dec 5 12:21:53 2008 -0800 uxa: Reject solid/copy to under-8bpp destinations. EXA wouldn't create pixmaps for under-8bpp, but UXA does. Fixes mis-rendering in xfwm, evolution message compose, firefox link drag'n'drop, and I'm sure more. Big thanks to Pierre Willenbrock for debugging the issue! Bug #18050 commit 261c20a479f6ec1e94c2ba801323072227cc3ade Author: Eric Anholt Date: Fri Dec 5 12:13:26 2008 -0800 uxa: Add in EnableDisableFBAccess handling like examodule.c did. This fixes assertion failures when rendering text while VT switched. commit caecd6031e416705b1f0a7051535211feaebdedd Author: Eric Anholt Date: Thu Dec 4 15:18:08 2008 -0800 Remove the extra memory allocation for 965 video state now that it's all in BOs. commit 48c113ade26d5ad64999cd5cc288495c10e02fc4 Author: Eric Anholt Date: Thu Dec 4 15:10:49 2008 -0800 Emit proper relocations to pixmaps in BOs in i965 video. commit ae2cd8b75e41393e072e27064ecefecf1aa1bc6c Author: Eric Anholt Date: Thu Dec 4 15:02:12 2008 -0800 Move i965 video surface state and binding table to BOs. commit 1b3c3c9d79305abe1785fdaef26a950dafa8890f Author: Eric Anholt Date: Thu Dec 4 14:30:56 2008 -0800 Move i965 video vertex data to BOs. This eliminates extra syncing when clipping is involved. commit 84825972597042f0aa6784594dace96be96a0234 Author: Eric Anholt Date: Thu Dec 4 14:26:09 2008 -0800 Move remaining i965 video programs to BOs. commit 40671132cb3732728703c6444f4577467fa9223f Author: Eric Anholt Date: Thu Dec 4 14:20:21 2008 -0800 Move i965 video wm and sampler state to BOs. commit 48803eb7463ad14f3109f67fcf4ccff4362baaa2 Author: Eric Anholt Date: Thu Dec 4 13:22:12 2008 -0800 Stop allocating unused scratch space for i965 video. commit f3fe46b557a4dd7e212c2790fb47142d375c1c75 Author: Eric Anholt Date: Thu Dec 4 13:18:06 2008 -0800 Move i965 video vs/sf state to BOs. commit 4b9d3eac57a972c055c4acd7a10dfe8aa918131c Author: Eric Anholt Date: Thu Dec 4 13:11:13 2008 -0800 Move i965 video cc state to BOs. commit 176e92d89fdfb199780014722feab6ac25836dcc Author: Eric Anholt Date: Thu Dec 4 12:53:37 2008 -0800 Move I965DisplayVideoTextured unit state setup to separate functions. commit 61929f4c641e2ecb145ad2b22f7092d40e31ae6d Author: Eric Anholt Date: Thu Dec 4 11:44:35 2008 -0800 Move I965DisplayVideoTextured surface/sampler setup to separate functions. commit 73d03cb93ca761ee555b87558882e26bd2f10d91 Author: Eric Anholt Date: Thu Dec 4 11:37:38 2008 -0800 Move debug code for I965DisplayVideoTextured to separate functions. commit df01d8f0791bd825c96d9d5e20313d5ffcfeb532 Author: Zhenyu Wang Date: Thu Dec 4 16:11:33 2008 +0800 Change master version to 2.6.99.1 commit b156b3165e1aae5df0353737d0335ac2e653f5fd Author: Julien Cristau Date: Wed Dec 3 04:35:38 2008 +0100 Don't unconditionally define DRI2 xorg-server.h will define it if the server has dri2 support. commit 8d7cbab267e8fbcb2fcf90b18346b60607277266 Author: Zhenyu Wang Date: Thu Dec 4 18:43:03 2008 +0800 Fix maxium backlight level restored issue The origin check for bring back max value for '0' backlight level is ok for legacy or combo control method as '0' mostly doesn't act in ideal lowest level. But it breaks in using kernel control method which should provide a reasonable backlight range. This is tested fine on T61 with thinkpad_acpi module. commit 32e7d47925b445a85847d77651b258e117aaaffe Author: Zhenyu Wang Date: Thu Dec 4 09:24:31 2008 +0800 Disable XvMC on DRI2 for now commit 293f6232c6c0ce12802a2123edfab4551d25e266 Author: Dave Airlie Date: Wed Dec 3 16:51:25 2008 -0800 uxa: don't call composite routines with no buffer. We can get a case with gnome-terminal + links, where we get two arrays of glyphs all with 0 width and 0 heights in them. If this happens we manage to get to this case without any buffer setup and segfault. (cherry picked from commit 717c7492a0f6ba3fb3eabda33515881eef314155) commit 457a680afd0d8f835131ea72be0c3c618c2892c7 Author: Eric Anholt Date: Wed Dec 3 13:17:26 2008 -0800 Enable tiling for DRI2 back/depth buffers. This results in allocation overhead for small (8x8-128x128 or so) pixmaps with DRI2, but we're interested in looking at tiling them in general in the near future, anyway. commit b662ecccb5c036fcc4aa19026642bde0a1ca2ac8 Author: Eric Anholt Date: Wed Dec 3 10:50:33 2008 -0800 Re-enable composite accel on 965 with UXA. This was blocked on getting 965 render aware of BO pixmaps, which is happily now the case. commit d426f799f541cb32fab134f2af068c6cd31923d9 Author: Paulo Cesar Pereira de Andrade Date: Wed Dec 3 14:59:21 2008 -0200 Include prior to including "edid.h" from the sdk. commit 07f5a8223187c1abc79c104d2fa5859a54cecd30 Author: Robert Lowery Date: Wed Dec 3 09:48:23 2008 +0800 TV: add support to set TV margins in xorg.conf commit 00ae7a571b6413aa2530e4f1310f8f4646631946 Author: Eric Anholt Date: Tue Dec 2 13:08:23 2008 -0800 UXA: Add support for tiled front/back/depth by cutting over to the GTT map. commit 34d54db945c67a2fee0a38cc9eafc463b8413669 Author: Eric Anholt Date: Tue Dec 2 13:01:56 2008 -0800 Remove DRI_MM defines which are always true now. commit f7540df184a75a749cabcfa2da8974d3d0714ea6 Merge: 5f34702 ee87f7b Author: Keith Packard Date: Mon Dec 1 19:06:09 2008 -0800 Merge commit 'origin/dri2' commit 5f347020a52b583e19d333361ea7bf2591255899 Author: Paulo Cesar Pereira de Andrade Date: Sat Nov 29 02:35:59 2008 -0200 Export libIntelXvMC and libI80XvMC symbols. This is a side effect of adding the visibility flags to XORG_CFLAGS. commit ec7fd3d1b477928f9bffde1a72d7f3fdea89e629 Merge: d978cd4 baec75a Author: Zou Nan hai Date: Thu Nov 27 10:48:09 2008 +0800 Merge branch '965-xvmc' Conflicts: src/common.h src/i830_hwmc.c src/i830_video.c commit d978cd4b453ea588ed2fc2f2cb4ec26856fe00d4 Author: Wu Fengguang Date: Thu Nov 20 00:47:25 2008 -0800 refresh batch_bo reference after intel_batch_flush() The call into intel_batch_flush() will invalidate the pI830->batch_bo stored in bo_table[0]. Fix it by re-read the refreshed value. Signed-off-by: Wu Fengguang Signed-off-by: Eric Anholt commit ee87f7becc2138348ec0d948829db6e86cee86da Merge: 3dd7f0f a5b1e62 Author: Kristian Høgsberg Date: Wed Nov 12 16:40:37 2008 -0500 Merge commit 'origin/master' into dri2 Conflicts: src/i830_dri.c src/i830_driver.c src/i830_memory.c commit 30c29627c859f1f6780aa652d6de4fa985b63dd0 Author: Jesse Barnes Date: Thu Nov 13 12:35:02 2008 -0800 Make sure DRM library paths are included We need @DRM_LIBS@ in order to find drm_intel if installed in a non-standard path. commit 42ec6938bdea3a5b065a4248fdca5f031415191e Author: Ma Ling Date: Fri Nov 7 14:23:41 2008 +0800 enable Intel G35 SDVO HDMI audio output Set the SDVO_AUDIO_ENABLE bit to enable SDVO HDMI audio output of the Intel G35 chipset. Signed-off-by: Ma Ling Signed-off-by: Wu Fengguang commit 42b3903c3fda95a7c536cb957d8163d8dc54a9a1 Author: Wu Fengguang Date: Fri Nov 7 14:23:40 2008 +0800 enable Intel G45 integrated HDMI audio output Enable audio output for the integrated HDMI of Intel G45 chipset by introducing the SDVO_AUDIO_ENABLE bit. Signed-off-by: Wu Fengguang commit 293f4a8b6d1ca8a47d14c4d4b5904b85a8167e12 Author: Wu Fengguang Date: Thu Nov 13 09:50:37 2008 +0800 introduce i830_hdmi_priv.has_hdmi_sink HDMI is compatible with DVI, and we've seen many boards that use HDMI port for DVI output. So Zhenyu proposed this flag: i830_hdmi_priv.has_hdmi_sink to indicate the presence of HDMI capable monitors. Signed-off-by: Wu Fengguang commit 667923559219429b0c5fec12a0164f7eba1f8f2d Author: Jesse Barnes Date: Tue Nov 11 12:55:06 2008 -0800 Default to FULL_ASPECT panel fitting Now that 8xx is fixed, we should be able to preserve aspect ratio by default. Fixes fdo bz #18033. commit 65cd0fbb018b2c18f1571dc0924c7d92eaf794ad Author: Zhenyu Wang Date: Tue Nov 11 09:36:50 2008 +0800 TV: fix contrast and saturation for 915/945G 915/945G uses exponent-mantissa format instead of fixed-point format on 965G. commit a5b1e62337d4e8840347bb186db48697f0690a19 Author: Zhenyu Wang Date: Fri Nov 7 16:06:26 2008 +0800 quirk LVDS on Asus Eee box Reported by Yang Sheng commit 9942cfa6dcc70a09ea38f738b1e73e3f005080b9 Author: Keith Packard Date: Thu Nov 6 15:04:27 2008 -0800 Use long crt hotplug activation time on GM45. The GM45 b-spec requires the use of the longer hotplug activation period, but does not require looping twice over the detection logic. With this patch, CRT detection appears solid on my GM45. Signed-off-by: Keith Packard commit 226efa7c74c0ed8a7b974a3727c5d7322c8ef815 Author: Zhenyu Wang Date: Wed Oct 29 22:41:20 2008 +0800 TV: subcarrier fix for NTSC and PAL Using subcarrier DDA values from spec give more close result to standard frequency. commit 3651341292d90b7ded4c3f013bcb0f46537a113a Author: Zhenyu Wang Date: Wed Oct 29 20:41:46 2008 +0800 TV: fix timing parameters for PAL, 480p, 1080i commit b404afb755b608b02bcf0be1f8fe8a38d3d7bc1e Author: Zhenyu Wang Date: Wed Oct 29 20:32:18 2008 +0800 TV: save serveral TV_CTL register fields in mode set Driver should keep those fields according with spec. commit 2ae91f0ffdadfb393d526b94e21914a31aa14232 Author: Zhenyu Wang Date: Wed Oct 29 20:26:44 2008 +0800 TV: fix default contrast and saturation modifier Color knobs was set with higher modifier which caused strong color on TV screen. Setting fixed point modifier to default 1.0 makes picture on TV look nicer. commit c4cab00ef7f57fc27776f53263aacec2edf6f959 Author: Zhenyu Wang Date: Wed Oct 29 20:26:03 2008 +0800 TV: white space cleanup commit b130f94f26111e1a12ce869292915504588d0369 Author: Zhenyu Wang Date: Thu Nov 6 10:14:00 2008 +0800 SDVO: fix more command definition errors commit 3ab979da5c867fb862e33a80b973a35d56d56a55 Author: Zhenyu Wang Date: Mon Nov 3 16:05:28 2008 +0800 SDVO: add command for set monitor power state SDVO encoder power state must be higher than monitor's state. This only adds command and doesn't enable monitor power state setting yet. commit f558a4972d1cca03785da025170259c61541ee63 Author: Zhenyu Wang Date: Mon Nov 3 16:01:40 2008 +0800 SDVO: add GetScaledHDTVResolutionSupport command commit cd8cbff5d7921d2f4fdac106c3d54414310af94d Author: Zhenyu Wang Date: Mon Nov 3 15:20:18 2008 +0800 SDVO: fix sdvo tv format and sdtv resolution request/reply definition commit 61672b8c458ecf449138c51aeb12cf24dbbd59d9 Author: Zhenyu Wang Date: Mon Nov 3 14:59:53 2008 +0800 SDVO: add HDMI audio encrypt change bit for GetInterruptEventSource command commit 45fb04474d0e67643a633dcc3aeb4efe7fd71ea1 Author: Zhenyu Wang Date: Mon Nov 3 14:51:14 2008 +0800 SDVO: fix wrong order of sdvo version's major/minor commit 5cbf1e72711e906b5309694045b56933e14dca06 Author: Zhenyu Wang Date: Thu Nov 6 09:46:54 2008 +0800 Make IS_GM45 into IS_G4X define Those are identical that using one define is much clear. And it can also apply fixes for GM45 too, which is missing with origin define. commit 78b6cff3b6ec9513ede3e71d81a828967e6c0068 Author: Eric Anholt Date: Wed Nov 5 14:39:19 2008 -0800 i965: Support render acceleration with pixmaps in BOs. This required reordering the relocation emits for surface/binding table so that we didn't add new relocations to things that had already been relocated at (the check_aperture requirement). commit 801f0eac4f7b0071ed21eb9c68c5a951e1cc409a Author: Eric Anholt Date: Wed Nov 5 13:29:39 2008 -0800 Make I830FALLBACK debugging a runtime instead of compile-time option. commit d828b0802c939cb12664db4dd157a4944a4db01b Author: Maxim Levitsky Date: Wed Nov 5 13:47:09 2008 -0800 Add an option to make the overlay be the first XV adaptor. commit 040d9bf9d8748d1ed8f977a6356d198def978b51 Author: Eric Anholt Date: Thu Oct 30 13:55:07 2008 -0700 Remove the CheckDevices timer. This was an early attempt to support display switch hotkey. It never really worked, as the events didn't show up when they should. It also cost a wakeup every second to check just in case, instead of taking an interrupt like we can now using ACPI. Additionally, some machines apparently get stuck with the event set and end up DDCing checking for new devices every second and smashing your video modes. commit 08914cceda6c57530023cdcdb5ad7e4024f36a6e Author: Carl Worth Date: Thu Oct 30 16:46:06 2008 -0700 Use buffer objects for binding table and surface-state objects. Instead of having a static array for these and doing an ugly sync everytime we recycle the array, we now simply allocate short-lived buffer objects for this dynamic state. The dri layer, in turn, can take care of efficiently reusing objects as necessary. On a GM965 this change was tested to improve the performance of x11perf -aa10text from roughly 120000 to 154000 glyphs/sec. commit 47cc3d79da8174ba30ca130b0fb6c7d9c871caed Author: Carl Worth Date: Thu Oct 30 16:53:57 2008 -0700 Unreference the vertex_buffer_bo in gen4_render_state_cleanup This avoids leaking one buffer object. commit 88700acf30f9eab8f96c197c7d113ce38c0af6e7 Author: Carl Worth Date: Mon Oct 27 14:23:02 2008 -0700 Rename gen4_dynamic_state to gen4_vertex_buffer We don't actually plan to put any other data in this structure, so it doesn't make sense to have a generic name, (since we'll only be using it for our vertex buffer). commit 9e95722763e2379d14a6b46c3750a44713da5135 Author: Carl Worth Date: Thu Oct 23 15:43:19 2008 -0700 965: Move composite setup to new _emit_batch_header_for_composite This function is the new name for _allocate_dynamic_state now that it also emits everything to the batch necessary for setting up a composite operation. This happens in prepare_composite() every time and in composite() whenever our vertex buffer fills up. It's not yet strictly necessary to be redoing this setup in composite() but it will be soon when the setup starts referring to buffer objects for surface state and binding table. This move prepares for that. commit fcb2a5a1253c505913e66b08107c0a9f57b07bad Author: Carl Worth Date: Wed Oct 22 17:12:47 2008 -0700 Use buffer object for vertex buffer (in new gen4_dynamic_state) This begins the process of separating the dynamic data from the static data, (still to move are the surface state and binding table objects). The new dynamic_state is stored in a buffer object, so this patch restores the buffer-object-for-vertex-buffer functionality originally in commit 1abf4d3a7a and later reverted in 5c9a62a29f. A notable difference is that this time we actually do use check_aperture_space to ensure things will fit, (assuming there's a non-empty implementation under that). commit 7e68786cf746b6b984b184080578db1947b268ed Author: Jesse Barnes Date: Thu Oct 30 13:40:23 2008 -0700 Don't modify render standby if kernel mode setting is active Since we can't bang on registers directly in that case. commit 87ea531c5dc5b39809395b277c330854aaaaf019 Author: Adam Jackson Date: Fri Oct 24 14:47:27 2008 -0400 Quirk: No LVDS on Dell Studio Hybrid commit 4d5360b1f3666bca4ed44c7f1b9217b8157c74b9 Author: Carl Worth Date: Wed Oct 22 16:17:39 2008 -0700 Use consistent idiom for obtaining static_state This doesn't make any difference, but it's cleaner to have each function follow the same idiom for obtaining these pointers. commit 36fc53d068673497e77796ba69e010d08c359447 Author: Carl Worth Date: Wed Oct 22 15:45:51 2008 -0700 Rename state_base_offset to static_state_offset in gen4_static_state_init More cleanup here, and again no functional change. commit 8fda7df0071046d057eeaf74328668b227af52a9 Author: Carl Worth Date: Wed Oct 22 15:28:24 2008 -0700 Rename gen4_state_init to gen4_static_state_init This follows naturally from the structure rename. Also we make things less muddled by having this function actually accept a pointer to a gen4_static_state_t rather than a gen4_state_t, (and then fetching the desired pointer out from that). Again, no intended change in functionality here. commit acba4238116f2ead7be88d49c4c56c7cd6d7f3b8 Author: Carl Worth Date: Wed Oct 22 15:12:57 2008 -0700 Rename gen4_state_t to gen4_static_state_t It doesn't contain only static data yet, but it will soon, so this renaming prepares for that. Also, this helps make things more clear between gen4_render_state_t and gen4_state_t which were muddled before, (particularly because the corresponding identifiers were render_state and card_state). The card_state identifier is now known as static_state which should be less confusing. This change is strictly search-and-replace with no functional changes. commit c6550fa53d92c21dc677b86ebcd528a4ac640764 Author: Carl Worth Date: Wed Oct 22 14:51:42 2008 -0700 Ignore intel_gtt binary We simply forgot to add this to .gitignore when it was created. commit ee3df629eaa388f09bec33ad04d443359d63ea7f Author: Dave Airlie Date: Mon Oct 20 18:46:49 2008 -0700 Default kernel mode setting to off, add configure flag to enable Should help avoid unpleasantness. commit 55cb65c6fdcf932d09e97abfa5374ee574676798 Author: Eric Anholt Date: Sun Oct 19 21:22:37 2008 -0700 Enable Option "Legacy3D" for 965 as well, and clarify both the docs and code. Signed-off-by: Eric Anholt Acked-by: Keith Packard commit 28bb056f8cd326770590157412835318863b4134 Author: Keith Packard Date: Sat Oct 18 19:37:09 2008 -0700 Use pipes, not planes for sarea geometry data Vblank in the kernel is far simpler if it deals with pipes instead of planes, so we're changing both user and kernel side. This is the user mode side. This fixes both i830_crtc_dpms and i830PipeSetBase, the two functions which copy geometry from the crtc to the sarea. Signed-off-by: Keith Packard commit 546e2aca5b8b5d9f486d119ea2872ccd2b056054 Author: Keith Packard Date: Tue Oct 14 14:38:05 2008 -0700 Reduce incidence of MI_FLUSH usage. This tracks whether the last command in each batch is an MI_FLUSH command and avoids appending another MI_FLUSH in the non-GEM cases. Signed-off-by: Keith Packard commit 9b91cdf7ac88982880c887ffb0811c92157a7945 Author: Keith Packard Date: Sun Oct 19 14:24:46 2008 -0700 No need to check for drmCommandWriteRead; we link against libdrm My checks for DRM enabled were a bit too extensive; drmCommandWriteRead is part of libdrm which the driver is always linked against. Only the symbols in the DRI module need to be checked here. Signed-off-by: Keith Packard commit c25437a69490b09acff335a81a38aa540f0822f5 Author: Keith Packard Date: Sun Oct 19 13:39:04 2008 -0700 Remove accidental 'return;' left at the top of I830DRIInitBuffers I was testing the behaviour of the XAA-based DRI buffer drawing code for tiled buffers and accidentally left I830DRIInitBuffers disabled. Signed-off-by: Keith Packard commit ae224be27bb4432f25dae851debea6966bf31261 Author: Keith Packard Date: Tue Oct 14 14:35:36 2008 -0700 Check for DRI module before initializing GEM mm GEM requires the DRI extension module currently, so make sure that is loaded (by checking for the DRIQueryVersion symbol) before trying to call it. This allows the server to start with the DRI extension disabled. Signed-off-by: Keith Packard commit 7ddea0447c8972104d43cd7966f5ce89b4cca20c Author: Keith Packard Date: Fri Oct 17 00:07:09 2008 -0700 Handle differently tiled front/back/depth/third in DRI window management When moving or clearing the extra buffer contents associated with DRI windows, the XAA code needs to see which buffer is being manipulated in the Setup functions to program the tiling values correctly. Calling I830SelectBuffer and not then calling I830Setup... would result in mis-tiled rendering. Signed-off-by: Keith Packard Signed-off-by: Eric Anholt commit c946383afc644ae7740e3c3146424fdd86c05285 Author: Keith Packard Date: Fri Oct 17 00:07:08 2008 -0700 XAA tiling support was mis-computing adjusted pitch (>>4 instead of >>2) This may well explain why XAA never worked well on tiled front buffers; tiled buffers require a different pitch programming on 965 than non-tiled buffers, in dwords rather than bytes. Signed-off-by: Keith Packard Signed-off-by: Eric Anholt commit 871728a0cbba12579c830e67020cc0c69c1611bb Author: Zhenyu Wang Date: Thu Oct 16 10:35:17 2008 +0800 Don't handle irq in GEM mode commit 2419bce9efbff63529074e64af5ec5c2e62e368b Author: Zhenyu Wang Date: Thu Oct 16 10:28:16 2008 +0800 Make GTT dumper work on other 9XX chips commit 986c8df79e83c369655ad786a6bf6342b53c535d Author: Adam Jackson Date: Thu Oct 16 10:24:04 2008 +0800 Fix Mac mini crash in DDC mode probe SDVO on Mac mini trys to get EDID from CRT port, which failed with recent DVI-I change. commit 6707371176147340fabc9ab6f1e3d6d5ac980662 Author: Eric Anholt Date: Wed Oct 15 08:12:11 2008 -0700 Remove gratuitous flushing in EXA after solid operations. This snuck in with the UXA rename commit. commit 3dd7f0f9423bb891bc99cd3b77dc3d57e057a7ef Author: Kristian Høgsberg Date: Tue Oct 14 23:57:10 2008 -0400 Update to DRI2 changes. Implement DRI2CopyRegion and provide drm device filename to DRI2 module. commit 282f51c3f0e5bc2cedd2f60f458ca2662290d471 Merge: 8a54e3b 4dd0068 Author: Kristian Høgsberg Date: Tue Oct 14 23:13:40 2008 -0400 Merge commit 'origin/master' into HEAD commit 4dd00681dd0f9fce8dfd4592b46418edbbd2eeb4 Author: Eric Anholt Date: Tue Oct 14 11:33:33 2008 -0700 Fix broken stolen memory counting on G4X. On the GM45 we were assuming too little stolen memory (mostly harmless, except when it wasn't, until the AGP fix), and on the G45 we were assuming too much stolen memory, which was quite harmful when we touched the page that didn't get mapped. Future stolen memory accounting should use src/reg_dumper/intel_gtt before and after enabling AGP on the chipset to confirm that only the GTT entries not mapped to stolen are replaced, and that all of the unmapped GTT entries are replaced with the constant scratch page. commit e7e49bed7e254256f8cc0d4afcdfadc6dadf19e6 Author: Eric Anholt Date: Tue Oct 14 10:38:05 2008 -0700 Add a GTT dumper for G4x debugging. commit 1c6ea4ab4772453da915306045add8d147d567f2 Author: Julien Cristau Date: Sun Oct 12 15:10:22 2008 +0200 Typo fix Reported by Tomas Carnecky on IRC commit 6cb4150160bb1e1365773561fb53294ad9248a0e Author: Zhenyu Wang Date: Sun Oct 12 08:07:44 2008 +0800 Bug #17892: Fix possible crash in CRT probe Introduced in patch for DVI-I CRT probe, if no EDID CRT monitor is connected, origin code destroys default DDC bus which causes crash in later get_modes call. Change it to setup and destroy DDC bus as needed in get_modes, so we always reprobe and get current state. commit 140451abb41fbd68d5aac4736c1cf1ec2805dd9e Author: Keith Packard Date: Thu Oct 9 18:18:34 2008 -0700 For non-DRM, add NOOPs after BATCH_BUFFER_START to verify completion In non-DRM mode, the driver waits for the hardware by checking the ring pointers; when the ring is empty, it assumes the hardware is idle. However, the hardware updates the ring pointers before executing a command, so if the MI_BATCH_BUFFER_START is the last command in the ring, the driver will think the hardware is idle while it may still be processing the contents of the batch buffer. Placing NOOPs after the BATCH_BUFFER_START allows the driver to know that the hardware has completed the batch buffer. Signed-off-by: Keith Packard commit d09d938bf26ea0c44352ff6e7a57ba43f4047fdb Author: Keith Packard Date: Thu Oct 9 18:17:13 2008 -0700 i830 nondrm batch buffer insertion was missing ADVANCE_LP_RING() call The ring commands to insert a batch buffer to the ring in i830 form were not terminated by a call to ADVANCE_LP_RING(). This surely would have caused chaos. Signed-off-by: Keith Packard commit d24010b7b3f2419beb40dc5ae1e8aeb3e04b5a93 Author: Carl Worth Date: Thu Oct 9 14:29:04 2008 -0700 Disable frame buffer compression by default for GM965. We haven't found a way to make FBC work reliably with GM965 yet, (it often fails to notice CPU writes). This appears to be a specific problem with this device, (as we haven't gotten similar bug reports for subsequent devices such as GM45). So FBC is now disabled by default for GM965 but can still be enabled with the FrameBufferCompression option for experimenting/debugging. This resolves bug #16257: [GM965 EXA] Frame-buffer compression broken for CPU writes (XPutImage) https://bugs.freedesktop.org/show_bug.cgi?id=16257 commit 4128b01ec84ffa0f03c335c36738f383376d9381 Author: Zhenyu Wang Date: Thu Oct 9 09:19:41 2008 +0800 Remove Lenovo T61 TV quirk BIOS already says no integrated TV, and this quirk blocks TV on R61i which has the same subdevice id as T61. commit 74571363539426abeb0a1af11f3bb545d91ed6c2 Author: Zhenyu Wang Date: Wed Oct 8 15:36:10 2008 +0800 Put forware VBIOS data parsing i830_bios_init() is called too late after output init, which makes bios data mostly useless, and caused all TV init fail as tv_present flag is not set. commit 111dd151dedf535464d638461f9485ec2939780e Author: Lukas Hejtmanek Date: Tue Oct 7 15:07:08 2008 +0200 Fix driver build against server master. Broken by 3a4151b69daa478ac6edf042d604ee41e8429c0d commit 3a4151b69daa478ac6edf042d604ee41e8429c0d Author: Eric Anholt Date: Mon Oct 6 17:00:08 2008 -0700 Fix driver build against server 1.4.2. This disables UXA and DRM modesetting pre-1.5, due to privates handling issues. commit 3621183cf4acef23414e8d69c34b1e587f52ec67 Author: Eric Anholt Date: Mon Oct 6 16:59:16 2008 -0700 Fix bios_reader build against old servers. The server's pci_device structure ends up conflicting with libpciaccess's. Just don't include the server structure for this non-server tool. commit 080d36648f92d3b9a559e5458687fa82afab411f Author: Eric Anholt Date: Wed Oct 1 16:13:09 2008 -0700 Add support for RepeatPad and RepeatReflect to 915 and 830-class Render accel. commit 5e04b5ad1b618655e201cbf1f5b2585aa6be0c17 Merge: 979bb10 260cbcf Author: Jesse Barnes Date: Mon Oct 6 16:08:12 2008 -0700 Merge branch 'master' of ssh://git.freedesktop.org/git/xorg/driver/xf86-video-intel commit 979bb109c6d76fe18b9e29f55afb2548f890bbfe Author: Jesse Barnes Date: Mon Oct 6 16:05:46 2008 -0700 Document more VBIOS functionality Document a few more VBT structures and comment on VBIOS communication a bit. There should be enough there now for a sufficiently motivated developer to start implementing support for hotkeys and other features on pre-IGD OpRegion machines. commit 260cbcfe61868175ba3e649ce07d43f57601f9be Author: Carl Worth Date: Mon Oct 6 13:17:59 2008 -0700 Document and use 'legacy' border color mode It's very convenient that the hardware supports this non-default mode since it's exactly what is specified by the Render extension. This provides a more efficient means of fixing bug #16820: [EXA] Composition result in black for areas outside of source-surface bo https://bugs.freedesktop.org/show_bug.cgi?id=16820 without the software fallback we had in the earlier fix, (commit 76c9ece36e6400fd10f364ee330faea470e2da64 ). commit 98ca9e2af1b0a67ed91e10cf661af70dfba44d43 Author: Carl Worth Date: Mon Oct 6 13:16:37 2008 -0700 Rename default_color to border_color This is consistent with the documentation, (and just plain makes more sense). commit 175b30e3828266a89021c2439119545099d72715 Author: Carl Worth Date: Mon Oct 6 12:48:07 2008 -0700 Revert "Fallback to software for RepeatNone with transformed RGB-only pictures." This reverts commit 76c9ece36e6400fd10f364ee330faea470e2da64. We've learned a new technique that should let us avoid this fallback to software. See following commit. commit 21cce1998a10db520c390eb5061814867564c583 Author: Shuang He Date: Mon Oct 6 09:48:21 2008 +0800 Fix a typo in G965 texture video code commit 76c9ece36e6400fd10f364ee330faea470e2da64 Author: Carl Worth Date: Thu Oct 2 20:40:14 2008 -0700 Fallback to software for RepeatNone with transformed RGB-only pictures. We wish it wouldn't, but the hardware ignores the alpha in the BorderColor we set when the source picture format has no alpha in it, (and it uses alpha of 1.0 where we want 0.0). For now, fallback for these cases. This gives a correct result, but obviously is not as fast as we would like. This fixes bug #16820: [EXA] Composition result in black for areas outside of source-surface bounds https://bugs.freedesktop.org/show_bug.cgi?id=16820 commit f1dbc266ccfe26c6b9a272e40a5bbe9afaa4f2e0 Author: Eric Anholt Date: Thu Oct 2 14:45:12 2008 -0700 Work around libpciaccess reporting a 0 rom size by guessing. I required the following patch on top of this to work around libpciaccess brokenness. libpciaccess reports 0 rom size if there's no rom resource, even if the rom file exists in sysfs. commit 8304b405e0dc2f31fd2d2fd82e150ba502ab74e2 Author: Keith Packard Date: Mon Sep 29 17:37:28 2008 -0700 Eliminate INT10 call to get BIOS contents libpciaccess (and the old X server PCI code as well) provides a function to get the ROM contents. Code to use that was already present in the driver and used if the INT10 function failed. Skip the INT10 and just use libpciaccess as that eliminates several module loads and scary use of vm86. Signed-off-by: Keith Packard commit ab2b70f831314f81a581bfb1e48d059a3a2b0b06 Author: Carl Worth Date: Wed Oct 1 16:05:16 2008 -0700 Prefer repeatType field over using both repeat and repeatType. Eric informed me that the repeat field exists only for backwards compatibility with old drivers that weren't prepared for values other than 0 or 1 here. Since we are, we can just ignore that field and examine only repeatType. So the code's a (tiny) bit simpler this way. commit 128223ee9b7880e640056475462eca9a88415492 Author: Carl Worth Date: Wed Oct 1 15:29:04 2008 -0700 Add support for RepeatPad and RepeatReflect. It's quite simple to support these modes---we simply need to turn on the support for them in the hardware. These changes have been verified with the extend-pad and extend-reflect tests in cairo's test suite. However, this currently required using a custom-modified version of cairo. The issue is that released versions of cairo, (and even cairo master so far), don't pass RepeatPad and RepeatReflect to Render, (due to various bugs and workarounds in cairo and pixman). I do plan to fix those issues in cairo, so that in a future release of cairo, (1.8.2 perhaps?), the cairo test suite will usefully test these new repeat modes in our driver. commit b7279f1be1b913c1c6ee8ebfb95c97800217a821 Author: Carl Worth Date: Wed Oct 1 15:26:39 2008 -0700 Examine picture repeatType as well as repeat field. The existing switch statement was switching on the Boolean repeat field rather than the correct repeatType field. This had not caused any problem before as only two possible repeat values were supported (RepeatNone = 0 and RepeatNormal = 1) so they were always the same as the repeat field. Soon, however, we'll be supporting more repeat types, so we'll need to switch on the correct value. commit ffcbbb071f1cde90fe0dc4887a05dd66c0e66985 Merge: 1eb46d0 bf8a3ff Author: Jesse Barnes Date: Tue Sep 30 16:59:35 2008 -0700 Merge branch 'master' into xf86-video-intel-2.5-branch commit bf8a3ffeaa4df9ecbad7abff9a44d714de2fa104 Author: Jesse Barnes Date: Tue Sep 30 16:59:02 2008 -0700 Revert "Use -Werror by default" This reverts commit f315e9d1ad92562195ce42b956d4be6b31e8a13e. The world isn't ready for a warning free build. commit 1eb46d055b147941fa529ac582c336a114f47892 Merge: 8408995 497b742 Author: Jesse Barnes Date: Tue Sep 30 12:54:27 2008 -0700 Merge branch 'master' into xf86-video-intel-2.5-branch commit 497b7420c4a76df553ed53322a390a1c754ce30c Author: Olivier Fourdan Date: Tue Sep 30 12:53:11 2008 -0700 Fix ordering of VGA vs. plane disable From the Intel 965 Programmer’s Reference Manual, volume 3, chapter 2.2.2 "Mode Switch Programming Sequence". The disable sequence should be: - Disable planes (VGA or hires) - Disable pipe - Disable VGA display in 0x71400 bit 31 This patch implements that order plus waits for a vblank at the end. Fixes bug #17756. commit a4568740ee296bc392843fd324d2a047e7862187 Author: Jesse Barnes Date: Tue Sep 30 12:48:32 2008 -0700 Be more verbose about panel data in VBIOS dumper Dump more panel data, including number of expected entries. Had to refactor things a bit, but now each function should get size information so further checking can be added more easily. commit fa2586a40f20e73ec7420466638e8f595e0da987 Author: Jesse Barnes Date: Tue Sep 30 12:46:20 2008 -0700 Use VBT LFP info pointers by default On some machines it appears that the LFP info pointers give us more accurate panel info than if we index into the LFP data table using the panel type index. Early reports indicate that using the pointers doesn't cause regressions, so switch to them by default to help 8xx machines. Fixes bug 17310 (and hopefully 17658 too). commit f315e9d1ad92562195ce42b956d4be6b31e8a13e Author: Jesse Barnes Date: Tue Sep 30 12:07:52 2008 -0700 Use -Werror by default commit f082e877d54dbafa437fd05e9c07e870cd513be9 Author: Jesse Barnes Date: Tue Sep 30 12:06:46 2008 -0700 Work around gcc uninitialized variable warnings GCC isn't smart enough to analyze the control flow and figure out that these are false positives, but initializing them shouldn't hurt, so work around it. commit 836d24f2cc3836fbd1695cf1f88c2af975b4862c Author: Keith Packard Date: Mon Sep 29 19:02:51 2008 -0700 Use uintptr_t instead of uint64_t to hold pointer value commit baec75a56d91546443a2e04812face4997bc764b Author: Zou Nan hai Date: Sun Sep 28 11:18:40 2008 +0800 [965-xvmc] 965q indirect media data workaround commit 11d304e99c0e11c28901ec28115d9c8b81a2b9cc Author: Zhenyu Wang Date: Sun Sep 28 10:08:26 2008 +0800 Bug #16631: add option for SDVO force detect Some ADD2 card doesn't get SDVO detect status setup right, which disabled outputs on those cards. This adds a new option "ForceSDVODetect" to probe all SDVO ports anyway. commit 1cc15ba454fdf54a7dea9da066e0a023a4742fab Author: Zhenyu Wang Date: Fri Sep 26 10:01:52 2008 +0800 Render register clock gating disable fix on 4 series chipset commit d6b2696f9ac14a81598e0147698209ad428fd45d Author: Zhenyu Wang Date: Fri Sep 26 09:53:03 2008 +0800 Do force CRT detect sequence twice on 4 series chipset commit d8a007b056e3d6c3e132ecac06368b21d2ac4123 Author: Zhenyu Wang Date: Fri Sep 26 09:44:59 2008 +0800 Bug #16515: Fix VT switch with DVI on G45 On DVI, centered VGA mode is used instead of native mode VGA, and PLL on pipe is used instead of VGA PLL setting. So make sure PLL settled down in restore time. commit a428892b76dba4c55e2ae28933d30e98d9fd646d Author: Jesse Barnes Date: Tue Sep 23 12:36:11 2008 -0700 Update supported hardware list commit 3c2f7d84785954eb91341f85abb175f299b2797d Author: Zou Nan hai Date: Tue Sep 23 17:15:42 2008 +0800 [965-xvmc] IGD support commit 08ab23bb87563517d5653bddf82763b5ef641bed Author: Zou Nan hai Date: Tue Sep 23 14:57:18 2008 +0800 [965-xvmc] optimize, pack some payload commit 8408995ffbf705aa0bc09ab72c58c2e31a4b70c3 Merge: d18b86a afa630b Author: Jesse Barnes Date: Mon Sep 22 18:38:54 2008 -0700 Merge branch 'master' into xf86-video-intel-2.5-branch commit afa630b448e5993850433c9f0b129758ec4d37b5 Author: Bryce Harrington Date: Mon Sep 22 18:38:35 2008 -0700 Add TV out quirk for HP Compaq nx6110 Adds a TV out quirk for HP Compaq nx6110. Fixes bz #17683. commit 204a6281aa59e29027793ba5c84a5d9af7dc9e90 Author: Jesse Barnes Date: Mon Sep 22 18:38:03 2008 -0700 Revert "Add no TV out quirk for HP Compaq nx6110" This reverts commit 6a46022df57d09109fc119fedce99c156715a4d4. It should have been attributed to Bryce. commit d18b86aa7fc9270e24ed9ea861ebbbbb54558fee Merge: 9a7da65 6a46022 Author: Jesse Barnes Date: Mon Sep 22 18:03:39 2008 -0700 Merge branch 'master' into xf86-video-intel-2.5-branch commit 6a46022df57d09109fc119fedce99c156715a4d4 Author: Jesse Barnes Date: Mon Sep 22 13:26:58 2008 -0700 Add no TV out quirk for HP Compaq nx6110 Hopefully we can fix the VBT parsing code so we don't need this, but for the time being it seems necessary. Fixes bz #17683. commit 10909d9b665864bda2b1654de009d556cd068726 Author: Jesse Barnes Date: Mon Sep 22 13:11:55 2008 -0700 Add Cappuccino SlimPRO SP625F to no LVDS quirks list Looks like this platform might contain VBTs that indicate an LFP is present even though it isn't. Intended to fix bz #11368. commit 9a7da65e624060a37eef989963a9c79f13df30d3 Merge: 043b6e7 2f93cfb Author: Jesse Barnes Date: Mon Sep 22 09:59:52 2008 -0700 Merge branch 'master' into xf86-video-intel-2.5-branch commit 2f93cfbc7e96acc32efb5e1ca49b817a81cba6e3 Author: Zhenyu Wang Date: Fri Sep 19 15:20:55 2008 +0800 Fix output detection for DVI-I For CRT this trys to probe all possible port for EDID and detects got confirmed by EDID's d/a type bit. For HDMI/DVI, also using EDID d/a type bit to ensure it should handle the connect or not. commit 1fbe4d602816c9dfc5fba917b9fdc257d8d025b0 Author: David Schleef Date: Thu Sep 18 15:37:00 2008 +0800 Bug #17277: fix upscaling limit Oh duh (i830_video.c): /* Clamp dst width & height to 7x of src (overlay limit) */ if(drw_w > (src_w * 7)) drw_w = src_w * 7; if(drw_h > (src_h * 7)) drw_h = src_h * 7; The condition I see in the documentation appears to be src_h/drw_h < 8, that is, src_h < 8*drw_h. It appears this was "fixed" incorrectly in e784e152. It seems difficult to believe that this limitation would exist at all for the texture unit. commit bc36608e321e01a2be792688b4b734bb7c0667f7 Author: Zhenyu Wang Date: Thu Sep 18 10:42:33 2008 +0800 Check display stride limit when allocate front buffer commit 043b6e71b83eb05339a6f8c4814e6941f8b9695a Author: Xiang, Haihao Date: Tue Sep 16 13:21:43 2008 -0700 Move bufmgr init earlier so it's available at I830DRIDoMappings time. Fixes a crash with non-GEM mode. Bug #17540. (cherry picked from commit 62b75df84c893bf28e20014cf88ce45064611dc9) commit 0a2d17f7c2ea9b695df1855aab4ccc519546d8e5 Author: Xiang, Haihao Date: Tue Sep 16 11:49:39 2008 -0700 Put back check for pI830->hw_status in setting hws in non-GEM mode. Fixes crashes on non-GEM systems with physical hardware status page. Bug #17540 (cherry picked from commit 188d58dac9a87b56dbc34ec219cd196928bbcf64) commit 62b75df84c893bf28e20014cf88ce45064611dc9 Author: Xiang, Haihao Date: Tue Sep 16 13:21:43 2008 -0700 Move bufmgr init earlier so it's available at I830DRIDoMappings time. Fixes a crash with non-GEM mode. Bug #17540. commit 188d58dac9a87b56dbc34ec219cd196928bbcf64 Author: Xiang, Haihao Date: Tue Sep 16 11:49:39 2008 -0700 Put back check for pI830->hw_status in setting hws in non-GEM mode. Fixes crashes on non-GEM systems with physical hardware status page. Bug #17540 commit e2743a409a02978ca8e953dbbeeb96f61bbd5ce6 Author: Jesse Barnes Date: Fri Sep 12 14:22:48 2008 -0700 Only BO map render state if kernel mode setting is active We'll probably end up doing this differently, but avoid this path for now. commit 95b3e3a65a60aa264fbbfd141df980b534005264 Merge: 94a86fa ec17c88 Author: Jesse Barnes Date: Fri Sep 12 13:42:06 2008 -0700 Merge branch 'master' into xf86-video-intel-2.5-branch Conflicts: configure.ac - resolved version number conflict commit 8a54e3be5c5057fe8e3c52c03401fdada7978c45 Author: Keith Packard Date: Fri Sep 12 12:47:13 2008 -0700 Fix mis-merge of DRI2 changes related to pI830->directRenderingType The old code had a directRendering boolean, the new one has an enum. A bunch of the merges failed to remove the '!' in front of the old name, breaking a bunch of 'we don't have DRI' tests. commit d8d95d8c71f2cd4bab277f44132ece7963714a5b Author: Eric Anholt Date: Thu Sep 11 16:11:46 2008 -0700 Fix build failures that should have been in the previous merge commit. commit ec17c88a0ed7c9cf4ad68aa52a7a891946a1c0f4 Author: Zhenyu Wang Date: Thu Sep 11 16:14:47 2008 +0800 Add support for G41 chipset G41 is another 4 series chipset like G45/43. commit f9c625e1e5ddfff06b38fdd4e596fd8eae5934d5 Author: Zhenyu Wang Date: Thu Sep 11 15:35:27 2008 +0800 Disable render standby Render standby is known to cause possible hang issue on some mobile chips, so always disable it. commit 7e8eb6e20b6b1a2a3c1ef28f694fa23c68a15c48 Merge: fba9b5a 58a3817 Author: Eric Anholt Date: Wed Sep 10 14:10:26 2008 -0700 Merge branch 'master' into dri2 Conflicts: src/i830_batchbuffer.c src/i830_display.c commit 58a3817305ef1455a11be6ead8a8521dfc5875c7 Author: Eric Anholt Date: Tue Sep 9 11:02:49 2008 -0700 Track move of exec to bufmgr, and restoration of emit/wait funcs for non-drm. commit f367334c6392a717f6cd2f4ed02200be1c6d356a Author: Eric Anholt Date: Tue Sep 9 11:01:33 2008 -0700 Track the move of irq emit/wait to fake bufmgr. commit 0b4f7b630312b148ce4e172cb7cd9f673751b2a3 Author: Eric Anholt Date: Tue Sep 9 10:57:08 2008 -0700 Track move of bufmgr functions to libdrm_intel. commit 0f804bfa1e1e972e9b4e3b7c8db61e9877c50f50 Author: Eric Anholt Date: Tue Sep 9 19:17:05 2008 -0700 Bug #17446: Don't try to manage IRQs in GEM mode. The kernel told us that it was already doing so, resulting in failure. commit da63b5adec8555cc7b3e71f33933f4c9dd6f714e Author: Eric Anholt Date: Sun Aug 31 14:27:29 2008 -0700 Add some MCHBAR registers for debugging tile swizzling issues. commit c7aaf0118baa34e583df5f1c29c9dab9a6af6eb7 Author: Dave Airlie Date: Tue Sep 9 18:13:56 2008 +1000 mode: fix missing comma commit b9ef0ed7d7b96eca6394cd0d367369ec511d1bcd Author: Daniel Stone Date: Fri Sep 5 05:02:08 2008 +0300 i830: Fix timer leak TimerCancel just cancels the timer: it still leaves the TimerRec intact and unfreed. commit fba9b5aff450d874ef98cafd3ecc0fd85f9597b5 Author: Eric Anholt Date: Wed Sep 3 16:49:07 2008 +0100 DRI2: Move pixmap pitch alignment for use with depth to pixmap create. The previous location for pitch fixup would have only worked when depth was used with the backbuffer, and no page flipping or other adventures occurred. commit 71ec627c3a65cfc7bca7353af43c60b18e73230d Author: Eric Anholt Date: Wed Sep 3 16:37:25 2008 +0100 DRI2: Emit the MI_FLUSH before flushing batch in swapbuffers. Should fix issues with swapbuffers flushing to front buffer on 965. commit daaefd173b4c98d0ebabd43352bfa3a030a62e4b Author: Eric Anholt Date: Wed Sep 3 16:26:27 2008 +0100 UXA: Re-enable non-965 render. commit 32bb5fc7db14911d0438aea1504e1551b2c9fc25 Author: Eric Anholt Date: Wed Sep 3 16:19:03 2008 +0100 Don't set up sarea or drm mappings in DRI2 mode. commit 4a5e31138115f435be7f8f5999b1b2329a18d669 Author: Kristian Høgsberg Date: Wed Sep 3 16:18:23 2008 +0100 Fix broken test for DRI1 in DRI2 conversion. commit fca7a4e9a5c73e1c129e94eeccf8fc795f729931 Author: Stefan Dirsch Date: Wed Sep 3 15:32:11 2008 +0200 Pipe A force quirk for Toshiba Satellite A30. commit 57ad9cc689724a2f3583eda862250eab0a6798d5 Author: Fabio Date: Mon Sep 1 13:33:50 2008 +0800 Man page patch to clarify meaning of VideoRam option with i810/i815 commit fd6a9ca5b59a868260eb91272091ea6169091737 Author: Kristian Høgsberg Date: Fri Aug 8 15:36:35 2008 -0400 Add DRI2 support. commit a1c802e4a0324386cf7370594a46354c89e3b646 Merge: 3733a1b 808b72f Author: Jesse Barnes Date: Fri Aug 29 09:23:35 2008 -0700 Merge branch 'master' of ssh://git.freedesktop.org/git/xorg/driver/xf86-video-intel commit 3733a1b54c95aa378f32577f9e996946e8e8e48c Author: Jesse Barnes Date: Fri Aug 29 09:12:05 2008 -0700 Fix build when using kernel DRM headers Unfortunate mismatch between kernel and DRM master headers. Kernel: typedef struct _drm_i915_batchbuffer { ... } drm_i915_batchbuffer_t; DRM master: typedef struct drm_i915_batchbuffer { ... } drm_i915_batchbuffer_t; so use the typedef rather than the morphing structure name. commit bd7ee7fcf51138d21da92a3d8cbaff49871d7188 Author: Zou Nan hai Date: Fri Aug 29 11:57:46 2008 +0800 [965-xvmc] There is some issue with uv half pixel, remove it till fix commit 973a5fb25652a97e28af03295d430a79557be756 Author: Zou Nan hai Date: Fri Aug 29 10:59:17 2008 +0800 [965-xvmc] some shader optimize commit e7379d1c48a522f137cf1a06a334748c88471946 Author: Zou Nan hai Date: Thu Aug 28 09:49:57 2008 +0800 [965-xvmc] half-pixel bug fix commit be49ae6bd683ee9eb778f7ea937aaee4d72c51f0 Author: Zou Nan hai Date: Wed Aug 27 11:29:15 2008 +0800 [965-xvmc] add missing g4a file commit e813b139b7c353a0930c3e00408700619d3949e4 Author: Zou Nan hai Date: Wed Aug 27 11:11:41 2008 +0800 [965-xvmc] support field prediction and dct type, optimize in GPU commit 808b72f81454061c815321e51a9b9f925c4bf786 Author: Eamon Walsh Date: Tue Aug 26 22:24:36 2008 -0400 Change uxa private keys to integer variables. Prepares for a devPrivates system that will store an index. commit 62ce9e8f9c8cc2014645d58f4249c496aebc36e8 Author: Eamon Walsh Date: Tue Aug 26 22:23:06 2008 -0400 Remove unused exa_pixmap_key. commit 087ade8e66cf7a34b8a96e1efe438099376cd896 Author: Robert Noland Date: Tue Aug 26 16:35:07 2008 -0400 Fix typo in last commit commit 6404ac178bf2c32a715a66e7fd6ab0184a05af36 Author: Robert Noland Date: Tue Aug 26 15:09:47 2008 -0400 Check for drm before calling modeset ioctl. commit c889198ab54336248a03050ad410254912d79bb7 Author: Zhenyu Wang Date: Tue Aug 26 10:12:21 2008 +0800 Fix X exit crash in NoAccel NoAccel should still be needed for performance evaluation, so don't crash me when exit. commit 0929a191ae36473acb0244a818098e4acf26d675 Author: Zhenyu Wang Date: Tue Aug 26 10:01:23 2008 +0800 Destroy bufmgr after allocation finish Fixed X exit crash for me, allocation reset will unreference bo which still requires bufmgr to be live. commit 94a86fa51a931f0059ea17728f857ec92cec5b16 Author: Jesse Barnes Date: Mon Aug 25 16:59:03 2008 -0700 Fix UXA build for distcheck Was missing proper path info in CFLAGS. commit 1c224f72d5f8fa8b6c4d9a6e06698f2462bb43b6 Author: Jesse Barnes Date: Mon Aug 25 16:59:03 2008 -0700 Fix UXA build for distcheck Was missing proper path info in CFLAGS. commit 09a235af0f216052d3101b153c18be1fa75abf15 Author: Jesse Barnes Date: Mon Aug 25 16:38:17 2008 -0700 Update version to 2.4.97 for first 2.5 test release commit 8e3243b8e5549033aa2f901a3bdd882418f6a69b Author: Jesse Barnes Date: Mon Aug 25 16:37:34 2008 -0700 Update version to post-2.5 commit 7a5b090abc39c380955489bad5e46d1278f38fee Author: Jesse Barnes Date: Thu Aug 21 09:40:57 2008 -0700 Hide kernel mode setting EXA code behind XF86DRM_MODE Prevents compiler warnings in the non-kms case. commit 4937b98981923c599786f5007ab0580ad08e1909 Author: Jesse Barnes Date: Thu Aug 21 09:39:33 2008 -0700 Fix compiler warnings in VBIOS utils commit 08326827fdb8abecbff20c7b051537dbad433c40 Merge: 5af5041 0de8ca3 Author: Jesse Barnes Date: Thu Aug 21 09:30:38 2008 -0700 Merge branch 'modesetting-gem' commit 5af504166f14f4401818f3b7c27ac2ccba7b9ff5 Author: Jesse Barnes Date: Wed Aug 20 15:06:36 2008 -0700 Don't disable planes in i830_update_dsparb We need to program DSPARB with only one plane enabled at most, and that's guaranteed to be the case when we're called during mode set, so just assert that case but otherwise leave the planes alone to avoid flicker on active displays during output detection for example. Fixes bug #17050. commit 7b6f4d22211d71480caf6335a3eacaacff369371 Author: Jesse Barnes Date: Wed Aug 20 14:40:29 2008 -0700 Don't allocate a pipe for hotplug detection It shouldn't be needed... commit ba4a4b7887d5c2da21cadfeeed601eaec8515550 Author: Jesse Barnes Date: Wed Aug 20 14:39:34 2008 -0700 Add more panel debugging info to register dump & vbios reader commit 0de8ca36306c506356e82d500134b4eeb96a6080 Merge: f26bcb9 5eccb5e Author: Jesse Barnes Date: Wed Aug 20 12:11:00 2008 -0700 Merge branch 'master' into modesetting-gem Conflicts: configure.ac commit 5eccb5ec7ea78b5800f98decd4216d39b9ebe5ec Author: Julien Cristau Date: Wed Aug 20 15:33:59 2008 +0800 Don't skip the checks for DRI headers with --enable-dri commit 1baebabc3af9f21a5b8084788ecd88dc66ce78de Author: Jesse Barnes Date: Tue Aug 19 16:02:55 2008 -0700 Add swf_dumper to .gitignore commit f26bcb9cb879e2e9ee2b26c95438cf46286c0208 Merge: 67ab546 7a1cc48 Author: Jesse Barnes Date: Tue Aug 19 15:53:18 2008 -0700 Merge branch 'master' into modesetting-gem Conflicts: src/i830_driver.c commit 67ab5462527c5ed94ed073421b60e85a213ea267 Author: Jesse Barnes Date: Tue Aug 19 15:46:52 2008 -0700 Use GTT mapping for EXA PrepareAccess function Makes software fallbacks *much* faster. commit 547cf774801bea2177d1224204f7de0dbf6baf3c Author: Jesse Barnes Date: Tue Aug 19 15:45:26 2008 -0700 Don't allocate EXA offscreen space if kernel mode setting is active We'll use GEM alloc & free routines in that case. commit 2ddd85f8aef5da1aebbfe9fab510858e7a41860d Author: Jesse Barnes Date: Tue Aug 19 15:41:41 2008 -0700 Bail out if kernel mode setting is active but DRI fails commit 646bbdc4b6a7b887ee8f0e51fdff5d07c4586329 Author: Jesse Barnes Date: Tue Aug 19 15:05:00 2008 -0700 Add quirk for pre-915s with working PFIT regs The regs are undocumented, but on some machines they work fine, so add this quirk to indicate it. commit 7a1cc48276f974d04e1c5ef7c92d98fe5ae9d4fa Author: Jesse Barnes Date: Tue Aug 19 10:44:04 2008 -0700 Add VBIOS software flags dumper program Still need to actually print all the flags & test on various platforms, but at least the bits are documented now. commit 385dc85881c56e218e7d17d7e1d022568aa29cfb Author: Zou Nan hai Date: Tue Aug 19 16:27:12 2008 +0800 Fix from He Shuang(shuang.he@intel.com), XvMCDestroyContext to check pointer valid commit 3147fa775c7e2a933dc74af1372a3318e77d3218 Author: Zou Nan hai Date: Tue Aug 19 16:20:16 2008 +0800 Fix from He Shuang(shuang.he@intel.com), XvMCDestroyContext to check pointer valid commit 078a59dd48ed00a6512d155883956c18752da9ce Author: Jesse Barnes Date: Mon Aug 18 12:53:33 2008 -0700 Remove unused VBIOS flag defines We just use the structure definitions now. commit 7048141254b7c1948fe77b1742b463f8723610d8 Author: Jesse Barnes Date: Mon Aug 18 12:51:31 2008 -0700 Pack bdb_general_definitions block More correct & consistent with other definitions. commit 0e6759bb6ad7293b6447bd20e3d7d8fdee46aba7 Author: Zhenyu Wang Date: Tue Aug 19 07:30:01 2008 +0800 Fix reverted LVDS bios capability dword definition Issue found by Peter Clifton. commit 76b5a7ff5c943cfaa254d2fd94b6397c91d1f8e6 Author: Shaohua Li Date: Mon Aug 18 11:13:20 2008 +0800 [PATCH] avoid duplicate mode set in lvds xf86SetDesiredModes() already sets lvds to full mode. later when xf86CrtcScreenInit() initialized randr12, i830_lvds_set_property will recall xf86CrtcSetMode and set mode to full. This patch is to remove the duplication. In my test, this can save about 0.2 - 0.4s x startup time. commit b8ee31cd80a945d1fbbfc45234ffe34a109b5717 Author: Jesse Barnes Date: Sat Aug 16 12:18:01 2008 -0700 Remove last TTM bits Makes it build again with drm#modesetting-gem. commit 6eb3e0f2f4e43e436029fc82e458ac8de1f94745 Author: Adam Jackson Date: Fri Aug 15 15:47:35 2008 -0400 Don't touch pScrn->monitor->DDC directly. commit ed7269e0a377b3135b06cacecc59119f06958feb Author: Zhenyu Wang Date: Fri Aug 15 10:10:43 2008 +0800 Always check and free driver private in LVDS destroy commit f4cb9a135dca9279af8186a9b18bf62ef7351019 Author: Zhenyu Wang Date: Fri Aug 15 10:02:44 2008 +0800 Make skip_panel_detect clear for its meaning commit 22918f62c89a4314fb5d01c58f22fee5b9a15a27 Author: Jesse Barnes Date: Thu Aug 14 16:02:37 2008 -0700 Fix pipe A force quirk Last commit introduced a logic buglet, we went from (foo & BLAH) -> (!foo & BLAH) rather than !(foo & BLAH), so fix it up to make my laptop work again. commit c1687f9ca9ce7b52043272e877e070c810e6599f Author: Jesse Barnes Date: Thu Aug 14 15:48:02 2008 -0700 Map/unmap render state only when bo is available Otherwise just use the GTT address. commit 380c80712f78b3673b64ea07746a8e25e15fba8e Author: Jesse Barnes Date: Thu Aug 14 15:39:57 2008 -0700 Fixup AccelMethod kernel mode setting code Allow UXA or EXA in the kernel mode setting case, defaulting to EXA. commit 4475dfb541c988ad19b177e60f31f333e2fb3355 Author: Jesse Barnes Date: Thu Aug 14 15:38:07 2008 -0700 Use pwrite for cursor updates Don't open code map/memcpy/unmap, let libdrm do that for us if necessary. commit cb217d4bfd941d0fa9ceae3e483dd1ca1d768e86 Author: Jesse Barnes Date: Wed Aug 13 16:55:39 2008 -0700 Make EXA & UXA share bo getting function Needed for proper acceleration & batch buffer handling. commit 1ab7239b73a5e298cadaf44e5605cf05855f08e9 Author: Jesse Barnes Date: Tue Aug 12 18:16:37 2008 -0700 Map gen4 render state buffer before initializing commit 603f48e31b021c4dc0bbf7b5efbb2e68aeb421d5 Author: Jesse Barnes Date: Tue Aug 12 18:04:08 2008 -0700 Don't set tiling (yet) if kernel mode setting is active commit f744aa8d4b22374e1de6dda4facb673c3c428d4d Author: Jesse Barnes Date: Tue Aug 12 18:02:21 2008 -0700 Don't run old accel init code Had the wrong logic. Throw in a couple of cleanups while we're there. commit 2321c865f3da9fd321910236a92b32e0af1aa458 Author: Jesse Barnes Date: Tue Aug 12 17:59:16 2008 -0700 Don't wait for ring if kernel mode setting is active We won't have the ring mapped. commit a1b334a4c8ae090744e802f8d41a36ea631d6c84 Author: Jesse Barnes Date: Tue Aug 12 17:58:29 2008 -0700 Update DRM based modesetting support Just a checkpoint, still needs a lot of work to properly handle resize, rotate & cursor handling. commit 5d78cf7cf0d23f60e6d61a92cbc2065b440027d3 Author: Jesse Barnes Date: Tue Aug 12 17:56:59 2008 -0700 Add EXA pixmap management functions for kernel mode setting Mostly pulled over from the old kernel modesetting code; a few updates were necessary. commit 9ec36e0c8bd8a4bd7c40569412fc1a21219b5af9 Author: Olivier Fourdan Date: Mon Aug 11 12:07:32 2008 -0400 Fix DPMS off in the presence of the pipe A quirk. Still turn off the VGA plane, and also handle the DRI path at the end. commit 445c2e9ef8fc1d49bbb03e07bf36e0339be16a80 Author: Olivier Fourdan Date: Mon Aug 11 11:30:51 2008 -0400 Apply pipe A quirk to 845 as well. commit 915c370964d8dbab48ff24ede2c5e78e7b39132d Author: Zou Nan hai Date: Mon Aug 11 15:45:35 2008 +0800 [965-xvmc] some fixes for earlier 965 chips commit 19ebdf1b24184c0a293a698d44c4e0571109d153 Author: root Date: Mon Aug 11 15:31:13 2008 +0800 [i965-xvmc] bypass copy when put image with xvmc commit d592eabc806b752053ade3c18e9dd8e0f39b45a3 Author: Zhenyu Wang Date: Mon Aug 11 15:16:44 2008 +0800 Fix possible spurious interrupts in hotplug detect on 4 series chip commit d0018a96064ee0adfe87c2d50c341bf7d2e45eb0 Author: Zhenyu Wang Date: Mon Aug 11 15:16:09 2008 +0800 Disable display clock gating for 4 series chips commit 173b909a71955997ed6366e70faebf63fe922a8e Author: Jesse Barnes Date: Fri Aug 8 18:11:13 2008 -0700 Make it actually build the kernel stuff if possible Fixup the kernel stuff to have a slightly better chance of working. Still need to fixup the EXA pixmap functions. commit fe90c0522f65f60c4c431787c889b7fb639ab61b Author: Eric Anholt Date: Fri Aug 8 16:35:25 2008 -0700 Set lvds_ddc_mode before use to avoid a segfault on mac mini. commit b1aef6f63c151dcb202fce869e4b80598b4b2052 Author: Jesse Barnes Date: Fri Aug 8 11:54:18 2008 -0700 Initial port of kernel modesetting from old intel-kernelmode branch Thanks airlied! commit 91e7b93c12f612e41b04a2d6106ff29b53b22a43 Author: Eric Anholt Date: Thu Aug 7 11:02:19 2008 -0700 intel_idle: Instead if #if 0, add an ignore flag for unreliable INSTDONE bits. This lets us produce a decent total value, and still show the unreliable bits in case they aren't on your platform. commit f9504eff31eb3c9c6c6b33dced9875866ff8307e Merge: f911347 5c9a62a Author: Jesse Barnes Date: Wed Aug 6 12:40:47 2008 -0700 Merge branch 'drm-gem' commit 5c9a62a29f62a9ecce37fae98cb01f8217eaba15 Author: Jesse Barnes Date: Wed Aug 6 12:39:03 2008 -0700 Revert "Switch to using a buffer object for the vertex buffer" This reverts commit 1abf4d3a7a203ff5d6e5ceda29573e7fd69ddf8e. Conflicts: src/i965_render.c - flushing was removed, keep it that way commit f91134795b545c8baebf218975b261c76a0e5873 Author: Zhenyu Wang Date: Wed Aug 6 16:19:29 2008 +0800 Fix SDVO HDMI encoding detect (#16920) Check return value of get supported encode command and current encoding mode, which could be DVI or HDMI. commit fc3e287e6b6db21b113aa40ec4d397802c067f8b Author: Keith Packard Date: Tue Aug 5 22:50:01 2008 -0700 [uxa] Remove unused pixmap size limits. All size-related rendering limits should be managed by the driver in the pixmap_is_offscreen call. There's no need for uxa to even know these values. commit 68f0872db6ec4d5dc3b524ee08ecad0aa125acd9 Author: Keith Packard Date: Tue Aug 5 22:36:03 2008 -0700 [uxa] Check xalloc returns and deal with failure Failing xalloc in a rendering function means just dropping the drawing on the floor (that's what we've always done). commit 8f10bfb127bfe73d83d58f1f306fb9a4dfd825d6 Author: Keith Packard Date: Tue Aug 5 22:34:24 2008 -0700 Use EXA by default instead of UXA until we have GTT mapping commit b2d058d80ccd08d9e02ef866ee7a95b58686f6a3 Author: Keith Packard Date: Mon Aug 4 23:43:13 2008 -0700 Rename uxa using _ instead of caps commit fc4d9c55a7fa8001786c1e4da10f005406c57ece Author: Keith Packard Date: Mon Aug 4 18:17:20 2008 -0700 Change PrepareAccess to take access mode rather than index commit c155bb3cb17a3bd3b2e90be52cd1fc90147c4e17 Author: Keith Packard Date: Mon Aug 4 04:11:13 2008 -0700 Add batch flush in i830_uxa_prepare_access commit 66706718553cd272eab6f817b5a059df3e0a4347 Author: Keith Packard Date: Mon Aug 4 04:10:21 2008 -0700 Add throttling commit 12df8f40d2fb41f5446db1b49beeb442da18bee2 Author: Keith Packard Date: Mon Aug 4 01:43:44 2008 -0700 Use dri_bo for all object allocations, including pixmaps under uxa commit 4cc20b7f6e25f4be4598f8edbe0077117126b4ee Author: Keith Packard Date: Mon Aug 4 01:42:48 2008 -0700 Don't call sync on prepare_access -- just let the driver deal with it. Let the driver do whatever sync is necessary from the prepare_access hook rather than forcing a full sync. commit b0b0998b5d52d105eb1e631f688aa8f1bd55ef39 Author: Keith Packard Date: Wed Jul 30 16:15:37 2008 -0700 Make EXA functions work for UXA as well EXA and UXA have the same acceleration interface, but UXA doesn't provide pixmap stride information as it doesn't manage pixmaps. Move all of that into the driver structure so that the acceleration functions needn't reference the EXA structure. commit 59774e9aca2d743e82d616bb644d20ff6d60d492 Author: Keith Packard Date: Tue Jul 29 22:57:09 2008 -0700 Add UXA - the unified memory acceleration architecture. This eliminates the cost of EXA migration management while providing full pixmap allocation control to the driver. The goal is to make something useful for UMA drivers. commit a21d4794b6812ce05d08f06dc47b26c4fb1c1fef Author: Jesse Barnes Date: Tue Aug 5 13:37:38 2008 -0700 Reorganize VBIOS code Make VBT parsing happen at driver init time rather than in each output init function, to save time and better separate VBIOS code into i830_bios.[ch]. The changes end up touching the output files due to field name changes, and allow us to reorder & simplify our LFP mode detection code. commit c2f0df4dc97c87539b66525a277c7d1e2c421f61 Merge: 4f1f308 750bd0b Author: Jesse Barnes Date: Tue Aug 5 13:03:46 2008 -0700 Merge branch 'drm-gem' of ssh://git.freedesktop.org/git/xorg/driver/xf86-video-intel into drm-gem commit 4f1f30854260e18b8b9c4103eebbad5dcf729ce6 Merge: e9e6b47 2aaa207 Author: Jesse Barnes Date: Tue Aug 5 13:02:47 2008 -0700 Merge branch 'master' into drm-gem Conflicts: src/i830_driver.c commit 1a59cc6b9acf312de1755d67757bf7f1967342e4 Author: Julien Cristau Date: Mon Aug 4 12:18:12 2008 +0200 Fix up the HP Pavilion ze4944ea quirk The chip is 855GM, not GM45. commit 27e9506fda86e90fd67de1715fa32d23aaa8a683 Author: Alan Coopersmith Date: Thu Jul 31 20:02:21 2008 -0700 Man page typo fixes commit da2eb83fb9a52291ea98f3285aee7bee2d55e0ca Author: Jesse Barnes Date: Thu Jul 31 19:21:36 2008 -0700 Update man page Add example dual head config, add info on bug reporting. commit 2aaa207db2cea03ba05c439b5074807689f2c5be Merge: 2049ba2 76eb8e6 Author: Jesse Barnes Date: Thu Jul 31 13:10:19 2008 -0700 Merge branch 'master' of ssh://git.freedesktop.org/git/xorg/driver/xf86-video-intel Conflicts: src/i830_display.c commit 2049ba211e7cdc383976c09f52c2b43acdd59481 Author: Jesse Barnes Date: Thu Jul 31 13:07:20 2008 -0700 Update DSPARB while planes are still off This avoids the flickering people reported in the 2.4.0 release. commit 750bd0bde09adf956c17bbb49c5a6020f12e60a4 Author: Carl Worth Date: Tue Jul 29 15:22:39 2008 -0700 Call DRM_I915_GEM_THROTTLE from I830BlockHandler This prevents the CPU from ridiculously outrunning the GPU. commit a893f176dda0b64f7dadfda6bf0331240037851e Author: Carl Worth Date: Fri Jul 25 15:56:35 2008 -0700 Add call to intel_bufmgr_gem_enable_reuse This instructs GEM to reuse buffer objects and improves the performance of my favorite 'x11perf -aa10text' from about 169k to about 188k glyphs/sec. commit b3c1a148679a4d943e556f996ef6b9004f549a41 Author: Carl Worth Date: Fri Jul 25 15:18:28 2008 -0700 Eliminate unnecessary flush from i965_composite This improves 'x11perf -aa10text' performance from ~144k to ~169k commit 1abf4d3a7a203ff5d6e5ceda29573e7fd69ddf8e Author: Carl Worth Date: Fri Jul 25 14:48:45 2008 -0700 Switch to using a buffer object for the vertex buffer commit e5fab0b7681be06a5a3be4bbd769ba5c435e2128 Author: Carl Worth Date: Fri Jul 25 13:44:29 2008 -0700 Add OUT_RELOC macro and backing intel_batch_emit_reloc function commit 76eb8e6f1f0c6962b23550564f4273f392567857 Author: Zhenyu Wang Date: Thu Jul 31 16:59:43 2008 +0800 Fix SDVO reg definition Remove wrong set tv resolution command, adding HDMI regs in dump. commit 77ed3d7600c1d92bf4a3ef4f54405cde8c232986 Author: Zhenyu Wang Date: Thu Jul 31 13:13:45 2008 +0800 Don't program dsparb on new Intel chip On new chip, DSPARB is controlled by hardware only. commit e9e6b4738573ffa3822ae68088bd53a645eedbce Author: Eric Anholt Date: Wed Jul 30 14:15:57 2008 -0700 intel-gem: Use new getparam to detect kernel GEM support. commit 42fb06f3f14fbec070350cf48361be4a0be0af04 Author: Tomas Carnecky Date: Sun Jun 15 14:27:16 2008 +0200 Reorder visuals reported by the intel driver The root window visual can not be changed. Neither at runtime nor through the configuration file. The xserver simply selects the first one that matches the class (usually TrueColor). I need a root window visual with stencil buffer because my compiz plugin uses the it for some operations. This patch reorders the visuals that the 3D driver reports and puts the one with stencil (and depth) bits as first. commit 1e90c82161789d1fd50b79db50edb7860b934639 Author: Julien Cristau Date: Sun Jul 27 12:07:21 2008 +0200 Link with -lm for sin/cos commit d77d42e543ee82e801a8563a911826521c0ba557 Author: Eric Anholt Date: Sat Jul 26 16:10:05 2008 -0700 intel-gem: Give a better error message if the kernel rejects the tiling mode. commit 6ac352f146bafe3b4bb5832340eb0ccb730565d2 Author: Zhenyu Wang Date: Wed Jul 23 15:10:07 2008 +0800 shut up gcc warning commit 075cf7fce57da4e51acafb3bfd3d61ba1a8096d5 Author: Zhenyu Wang Date: Wed Jul 23 15:07:48 2008 +0800 Fix distcheck for clean doltcompile commit 29d982c6bf873c6685bd0e395a8b0e5e3b1063e0 Author: Zhenyu Wang Date: Wed Jul 23 11:16:14 2008 +0800 Fix HDMI output number Output 33 is confusing. commit fd060ce89d86f6e8ff742d5b287abe8ecea32927 Author: Zhenyu Wang Date: Tue Jul 22 15:29:34 2008 +0800 Bug #16801: fix X crash when NoAccel on 965 commit 577c4d1a0c753fb46d43be05a9781d2e204fac45 Author: Zhenyu Wang Date: Tue Jul 22 15:14:14 2008 +0800 only check pipe underrun if vtSema is true. commit 04bbed2248e3e8f86efd64225abf2db8eb552653 Author: Zhenyu Wang Date: Tue Jul 22 15:08:57 2008 +0800 SSC is LVDS only Fix regression on my 855GM for VGA output. commit 7defa4f1f0eba82f39e74f96d2ad7ed2481b537e Author: Zhenyu Wang Date: Tue Jul 22 10:51:01 2008 +0800 Check underrun on enabled pipe commit b7765b0837af4cc80f1257ba04495140ef5d536e Author: Zhenyu Wang Date: Tue Jul 22 09:51:54 2008 +0800 Thinkpad R60e TV quirk via DMI info commit 286ff63f83db8ea16514512385e6f8c6875871ef Author: Eric Anholt Date: Thu Jul 17 17:06:28 2008 -0700 Don't set up the HWS page in GEM mode now that the kernel manages it. commit ed0fbd016b8fbcf01eddfd17fd25e745e7af2ba4 Author: Jesse Barnes Date: Mon Jul 21 11:48:38 2008 -0700 Add no LVDS quirk for Transtec Senyo 610 mini PC Fixes FDO #16757. commit 62a037b0585d6ecb43daac9d4eb0927a4618a367 Author: Eric Anholt Date: Fri Jul 18 14:50:08 2008 -0700 Fix distcheck. commit 3cbfc6c76bcd2bc9194a944092f6ce0881ff9da2 Author: Eric Anholt Date: Fri Jul 18 14:56:18 2008 -0700 Fix uninitialized-use warning in i830_debug.c ring dumping. commit 1b1c4975768f63ed0f3933ec384d24c324a383b3 Author: Eric Anholt Date: Fri Jul 18 14:53:26 2008 -0700 Get prototype for i830_bios_get_tv(). commit 978abf84e8cf3ba3e915774981fe03610e15efeb Author: Eric Anholt Date: Fri Jul 18 14:50:42 2008 -0700 Bump version number past the 2.4 stable branch. commit f4ab1f6ad47b8f9a33d45b35c1cb2ba9610b96a6 Author: Carl Worth Date: Fri Jul 18 14:31:45 2008 -0700 Add intel_statuspage to .gitignore commit c0dce28892d7a31ebcca615f87ba8d4fa6872204 Merge: a6ad504 b8ca1c7 Author: Keith Packard Date: Fri Jul 18 10:49:50 2008 -0700 Merge commit 'origin/master' into drm-gem commit b8ca1c747a679c931267363639fc0bc690cae2d6 Author: Jesse Barnes Date: Thu Jul 17 14:41:51 2008 -0700 Choose a split for DSPARB based on the configured modes for both planes. Previously, we were attempting to give both planes equal space in the FIFO to be fair. However, larger modes require more fifo space, so split it based on the relative HDisplay of the modes. This should resolve some fifo underrun issues with differently-sized displays, or single large ones. Bug #16169. commit b37a2a8ca82279468e3806dcf77d5fa7bdd0e874 Author: Jesse Barnes Date: Thu Jul 17 12:30:57 2008 -0700 Don't disable pipe A on 855 chips It needs to stay enabled or we may see hangs when trying to re-enable it (say at VT switch time). Fixes FDO bug #15168. commit a6ad50402cb879b08de95f94959bb1f59701c1ff Merge: e9916b9 c4565a9 Author: Carl Worth Date: Thu Jul 17 00:38:54 2008 -0700 Merge branch 'master' into drm-gem Conflicts: configure.ac src/reg_dumper/Makefile.am commit c4565a9811487402d899d0933cc63e27ffe1ff08 Author: Zhenyu Wang Date: Wed Jul 16 10:45:28 2008 +0800 Only initialize integrated TV encoder for mobile chips commit bca316863b79148b2f51a97d58b94e52ba3eae9f Author: Zhenyu Wang Date: Tue Jul 15 13:35:21 2008 +0800 xvmc: use vector based structure This is more clear to read MV components. commit 0c67219d0e6cfc858af2eb375c2f9473d0f6db60 Author: Zhenyu Wang Date: Tue Jul 15 13:29:33 2008 +0800 move FBC register dump out of display registers This makes it easier to read. commit a76ae7cab3001ce6639ae23fb0769c4e46b57e71 Author: lipeng Date: Tue Jul 15 10:26:33 2008 -0700 Fix GPIO pin usage for DDC on second HDMI port. commit 5054a86b1b362f3f63310afb9ef2d37c9f1cfca5 Author: Eric Anholt Date: Mon Jul 14 13:43:31 2008 -0700 Fix DSPARB setting on 845/865, which have only the AEND field and 96 entries. commit 8ac00ca97995e36514ff593fec3c0f0d316ed138 Author: Julien Cristau Date: Fri Jun 6 13:40:20 2008 +0200 Link the driver with -lpciaccess and -ldrm if needed This makes sure the driver ends up with a DT_NEEDED reference to the libraries it's using. commit d61182511b1520b04133d28480e1ffaacbaeb753 Author: Julien Cristau Date: Sun Jul 6 12:00:52 2008 +0200 Fix gen4asm rule to work with a build dir commit 757c00927a6f5760135136450b8d02d0f999ac1c Author: Carl Worth Date: Mon Jul 7 15:58:27 2008 -0700 Use up to 256 separate vertex buffers This allows us to only call i830WaitSync once every 128 calls to composite rather than on every call. However, we do need to also call MI_FLUSH to avoid the vertex cache getting in our way, (since our "separate" buffers are all allocated as one contiguous chunk). commit 0c548cd040d0c5e1812470ccdf6de86b6a2926d7 Author: Carl Worth Date: Mon Jul 7 14:01:15 2008 -0700 Allow for multiple vertex buffers (though only use one for now) Using more than one (in the future) will allow for doing less frequent calls to i830WaitSync. commit cc2249333cd462b4d99d110a12c454ca141b2be8 Author: Carl Worth Date: Mon Jul 7 13:25:42 2008 -0700 Move VERTEX_BUFFERS setup from prepare_composite to composite This is in preparation for having larger (or multiple) vertex buffers in the future. commit e9916b9d496fd0b4df717892dda33f81a2e2990a Author: Eric Anholt Date: Fri Jul 11 18:59:05 2008 -0700 Set tiling state for buffers allocated using GEM. commit 3c0f2bcc9965155fecab606edae68874427d3423 Author: Jesse Barnes Date: Thu Jul 10 12:57:25 2008 -0700 Add VBIOS based TV connector detection Now that the VBIOS code supports it, we can use the general features block to detect whether a TV connector is present on a given platform. Reviewed by Nanhai Zou. commit e00d9435609bcff1afb71aa6638a6b42a64f5178 Author: Jesse Barnes Date: Thu Jul 10 12:55:12 2008 -0700 Improve VBIOS feature detection, add SSC support Improve the VBIOS feature detection and use it to find whether the platform supports spread spectrum clocking. Use the specified reference clock, but disable SSC if multiple heads are active, since it can cause problems in cloned configurations. Reviewed by Nanhai Zou. commit 7332132a79e5b5c208d43e93dfe0c8b12eb1728d Author: Jesse Barnes Date: Mon Jul 7 10:48:56 2008 -0700 Improve FBC size checking In hindsight, this is obvious, since nowhere do we tell the FBC unit how much memory it has available. We need to make sure the compressed buffer is big enough to handle the uncompresed buffer, both in terms of vertical size and total framebuffer size, or the compressor could overwrite the memory immediately following the compressed buffer. commit f2ec3fd8f2c63b1bc0745802dcf737eda4bb42b6 Author: Stefan Dirsch Date: Thu Jul 3 23:12:13 2008 +0200 Add pipea force enable quirk for another 855GM machine. Fixes Novell Bug #406123, thanks Christoph. commit a34a4e3f6420e2b06bbdaa124fe0ccb1bc6a0bd9 Author: Zhenyu Wang Date: Thu Jul 3 01:03:14 2008 +0800 Fix official name for GM45 chipset commit 6506ac7aa008436158db4f4196802cab462f3446 Author: Zhenyu Wang Date: Wed Jul 2 11:12:54 2008 +0800 Give asus and eeepc backlight method higher priority On #16418, Evgeniy Manachkin reported that last asus and eeepc backlight patch is wrong, as acpi_video0 method will take priority and doesn't work. commit 55678c64bc6e3ed613ea6db14c105c18a0cf28ce Author: Dave Airlie Date: Tue Jul 1 18:51:18 2008 +1000 intel: fix drm check. The -mm check is broken now that 2.3.1 is out.. make it 2.4.0, to be fixed with GEM etc. commit 0597d94001afe0a9a34104285ce4dffc75b68071 Author: Roland Test-tools Bär Date: Mon Jun 30 10:32:42 2008 -0700 i810: Remove an effectively unused variable (only used in an incorrect free()) Bug #15401. commit 37661d916e8384f537d5ed8cbec66a93795c6a05 Author: Zhenyu Wang Date: Fri Jun 27 11:21:59 2008 +0800 Fix SDVOC typo commit 7834a3b118ae4e034f064257762d5c25ada5fe52 Author: Zhenyu Wang Date: Fri Jun 27 09:54:35 2008 +0800 xvmc: fix motion_type dump for frame/field picture commit 989ec9e8a69f909cb64f17e4465982613b4b054d Author: Zhenyu Wang Date: Fri Jun 27 09:39:02 2008 +0800 xvmc: Don't copy on xvmc surface in PutImage As xvmc rendering result has already been in fb, we shouldn't do extra copy on it. Although special care is required for i915 xvmc surface pitch alignment, which must be at least 1KB aligned. So video display function should take it into acount instead of always setting Y pitch to be double of U/V pitch. commit 550082070a3fdb951e3cf08974dc56276c0a739c Author: Zhenyu Wang Date: Thu Jun 26 14:22:34 2008 +0800 xvmc: enlarge batch buffer size commit d50cec6ef9e2178ea663e58d548390f0f3da7692 Author: Zhenyu Wang Date: Thu Jun 26 14:20:05 2008 +0800 xvmc: only allocate memory requiring physical address on 915G Later 945-ish chipset can use graphics address instead. commit d1c34d8c0e7ae2a1d952791343e131837fbfda99 Author: Zhenyu Wang Date: Thu Jun 26 10:05:21 2008 +0800 xvmc: init mc render load indirect command once commit 75e38fa8348198ba151afa37e10be3b0b0b468f8 Author: Zhenyu Wang Date: Thu Jun 26 09:47:28 2008 +0800 xvmc: init dest and reference buffer once Init them right after context create, and only update buffer address info later. commit b4d8ca8b38e495b56bb3b4143e5dfe91ee651f15 Author: Zhenyu Wang Date: Wed Jun 25 14:27:16 2008 +0800 xvmc: init one-time mc context once Don't setup one-time mc context everytime, as the content is always unchanged. And several structs got packed layout inside to ease static state initialization. commit 54f3f528e4804b958936ac773e7a507b2676f815 Author: Zhenyu Wang Date: Mon Jun 23 15:51:49 2008 +0800 xvmc: environ debug option support export INTEL_XVMC_DEBUG=1 commit 7a705081788bf957409e17baa5f189241589057c Author: Zhenyu Wang Date: Mon Jun 23 15:49:29 2008 +0800 xvmc: formalize surface type definition We should also carefully select surface type id. commit 0283d22f510ba47dfdac2806a58e46649c26580b Author: Jesse Barnes Date: Fri Jun 27 13:39:31 2008 -0700 Add pipea force enable quirk for HP Pavilion ze4944ea Fixes bug #16540, thanks Bryce. commit 08903abe4dc0295c7ed7d1ff1a22e0e579540c15 Author: Jesse Barnes Date: Fri Jun 27 13:37:10 2008 -0700 Add pipe a force enable quirk for Lenovo T60 Fixes bug #16494. commit 05df8c0b31721a9ccc7215fb1cda1115758367c7 Author: Hong Liu Date: Fri Jun 20 10:57:14 2008 +0800 Fix SDVO HDMI output. While some cards had enough initialized at startup to work already, others required that the driver actually initialize the required AVI info frame. commit 59f39e0ea0ad48c52d9b8f7afb6e04ce95db717e Author: Jesse Barnes Date: Tue Jun 24 11:01:46 2008 -0700 Remove ErrorF debugging from modeset ioctl Oops. commit 1056d79809a3765f81113f3988bd948c1f9c673a Author: Jesse Barnes Date: Tue Jun 24 10:44:23 2008 -0700 Fix back buffer damage handler for 965+ chips When page flipping is enabled, we need to make sure any changes to the front buffer are reflected in the back buffer(s) or corruption might occur at page flip time. So make the damage handler work on 965 by adding appropriate tiling flags and pitch adjustments. commit 65eee25d7d2ca979205f3776d620dbb36bf68a13 Author: Jesse Barnes Date: Tue Jun 24 10:41:46 2008 -0700 Add support for keeping vblank counters sane across mode setting The DRM supports disabling of vblank interrupts when not in use, but in order to function properly it must also be aware of mode setting, which will reset the frame counter to 0. Add code to call into the DRM before and after mode setting, so that it can account for any lost vblank events. commit 277632d8c37cf43c7c1178ae7d5a2e6d5705968c Author: Zou Nan hai Date: Tue Jun 24 16:43:40 2008 +0800 965 xvmc, current only frame based DCT_type support commit 9d7929436488969ca4fd1fef4d2e92be5e1a93aa Author: Keith Packard Date: Fri Jun 20 00:23:45 2008 -0700 Flush pending batch in block handler Make sure any pending rendering commands are delivered to the hardware before the server goes to sleep. commit 2f8a0aa8cd24ede824aa52102b45a295f1b6b5be Author: Keith Packard Date: Tue May 20 20:29:25 2008 -0700 Remove some unreliable regs for i915 commit 6b6be2b2a7179c1c9fa12989e16076aae105f4e2 Author: Eric Anholt Date: Thu Jun 19 15:01:00 2008 -0700 Fix a crash in i830_sdvo_init error paths by setting up dev_priv earlier. commit 65ad29d78793c7804f133a58de80ffaa0404ca28 Author: Eric Anholt Date: Thu Jun 19 14:27:30 2008 -0700 Automatically detect the presence of HDMI. Now, SDVO is only probed if the SDVO detected bit is set. If the SDVO probe fails, but the detect bit is set, assume that it's an HDMI output. commit ea0d21006ec71bc56acde7291e5f8d28e54b83fe Author: Eric Anholt Date: Thu Jun 19 14:10:29 2008 -0700 Fix hdmi POSTING_READ to use the register number instead of the register value. We won't dwell too long on how this ever worked. commit 2fa9240a3b685a0c3d0e2644fc144c927b67b6b3 Author: Eric Anholt Date: Thu Mar 20 15:14:46 2008 -0700 Get HDMI output working. HDMI, despite claiming to have pixel multiply support, actually doesn't want it, at least in the way that SDVO did. Disable it. Also disable the NULL_PACKETS_DURING_VSYNC bit. Despite the docs telling us to set it, the output doesn't like it when you do that, and the BIOS doesn't set it. Also add a posting read to SDVOx setting. Without it, half the X starts wouldn't get anything on the screen. (interestingly, it was every other startup, not just a chance). commit dc8ab9d35b261b6105a62949cfb47a3554531d0b Author: Eric Anholt Date: Thu Mar 20 13:49:47 2008 -0700 Set the sync active bits like we're supposed to, matching the BIOS. commit af13826ed063b60a1e4179101e8889b4ab7a9d0b Author: Eric Anholt Date: Mon Mar 10 11:51:05 2008 -0700 Add pixel multiplier support for HDMI commit 9a05662918223477eb1cf6f80ffac08712721c70 Author: Eric Anholt Date: Mon Mar 10 11:47:15 2008 -0700 The phase shift its are now reserved, and add HDMI clock limits. commit beba1dd3561e38573ed9f507328caf7f8fb9f84a Author: Eric Anholt Date: Thu Mar 6 16:05:17 2008 -0800 Initial HDMI work. Not currently hooked up at startup. commit da58dc3b02999f3244d0eaf77180b828d85bd609 Author: Eric Anholt Date: Thu Mar 6 14:57:57 2008 -0800 Add DisplayPort registers. commit 52a414d1158aa7512fb79560dc220e2a78456a2d Author: Eric Anholt Date: Thu Mar 6 14:00:18 2008 -0800 Remove VGA regs from debug output. commit b60bbbfd7b7dcf31726ad615a493f34f17cc6309 Author: Eric Anholt Date: Thu Mar 6 13:59:06 2008 -0800 Add little hotplug detector app. commit 64fa32e2eba21cb94aa0f8d87a6ee6b7c178a7a5 Author: Zhenyu Wang Date: Thu Jun 19 11:02:29 2008 +0800 Add backlight kernel method support on Asus and Eeepc Noted and tested by Evgeniy Manachkin for asus-laptop support, also add eeepc support. commit 1cfe769c74d1a3a392bf1aaaf5c2dcc8273daf66 Author: Zhenyu Wang Date: Tue Jun 17 12:39:28 2008 -0700 Add support for Intel 4 series chipsets. commit 289790c0467d27e96b537598a6589fc6a36da8b8 Author: Zhenyu Wang Date: Tue Jun 17 09:46:12 2008 +0800 Fix compiling with server master in LVDS backlight patch commit 66418cc99cfd360e303efcdf74a9ee31d435ff15 Author: Zhenyu Wang Date: Thu Jun 12 14:04:41 2008 +0800 Fix compiler warning when disable xvmc config (cherry picked from commit df0bbdc7cbb6ff357a81ed28d12e56c9c7d643f7) commit 078d05865884ca5784b532c0a79c8440632002ac Author: Shuang He Date: Mon May 26 09:20:53 2008 +0800 Fix typo in xvmc block destroy (cherry picked from commit 87ace420a34df7425641d089f71830e44fced098) commit 4c4ef27779aebf4df90b6233de05be2bb972de4c Author: Lukas Hejtmanek Date: Mon Jun 16 02:49:41 2008 +0800 Fix maximum backlight issue commit d775ddc64dc8349b8ef9e85b0be9e93cb1997fea Author: Eric Anholt Date: Fri Jun 13 10:03:24 2008 -0700 [gem] Catch -EINTR from blocking GEM ioctl and restart. commit 02d7141ac395446613edba0f2e2a951e750e4e9f Author: Dave Airlie Date: Fri Jun 13 11:30:58 2008 +1000 [gem] remove one more unused bit commit f07acbdaac95931d184ea2c557edb632e577eb47 Author: Eric Anholt Date: Thu Jun 12 13:47:35 2008 -0700 Require libdrm 2.4.0 always since we need the bufmgr code. This lets us remove a lot of conditional compile stuff in the DRI case, as if we're doing DRI and have 2.4.0, we can rely on GEM ioctls existing. commit 754df0aa55c4b5f0c3c415ad97d7612504410229 Author: Eric Anholt Date: Wed Jun 11 14:43:38 2008 -0700 [gem] Chase move of create ioctl from generic to device-specific. commit 339f00e340c3a9c8ac8f17147e37906494ef370b Author: Eric Anholt Date: Tue Jun 10 16:14:06 2008 -0700 [gem] Don't set up the ring in GEM mode, as that'll be handled by the kernel. commit e930c0dc5b5f98d256a4167ed70ba813c88924d2 Author: Eric Anholt Date: Sat Jun 7 13:57:59 2008 -0700 Use bufmgr_gem when available instead of the fake bufmgr. This is a bit unstable still sometimes. commit ecf19e1cda60a938d41413075ae6e00f24e0ec1a Author: Eric Anholt Date: Fri Jun 6 15:27:44 2008 -0700 Change most usage of pixmap offsets to using a reloc macro. This is based on airlied's RING->BATCH commit. The 965 code still needs to be fixed up for relocations. commit bade7d7d2505a10a8a7d24b084aff9742e2d6d64 Author: Eric Anholt Date: Fri Jun 6 14:03:25 2008 -0700 Use the DRM for submitting batchbuffers when available. There are some concerns with this, as the DRM will be setting the nonsecure flag on the batchbuffer, and the server may be submitting some secure-only commands. It appears to work on the 915GM test system currently. commit 6e94affcc2240e668bcf1aa41f3c8b19929d144b Author: Eric Anholt Date: Fri Jun 6 14:01:05 2008 -0700 Avoid needless flush emits in the blockhandler. The EmitFlush in i830_dri.c was added as a pageflipping workaround, and was noted to not even be sufficient then. There's no reason for it to be there, so it's removed. After that, we just have to not emit an MI_FLUSH if we already had, and cursor movement no longer bashes memory manager. commit b2216e7bc2f1a35f9fc1794bad83208cd5c583d1 Author: Eric Anholt Date: Wed Jun 4 16:31:16 2008 -0700 Use batchbuffers instead of ring emits for general commands. The batchbuffers are managed using libdrm and bufmgr_fake, and dispatched from the ring from userland. commit d0fda9d24c5fe4870cdbd2ba0213091165666a8e Merge: cc98d41 79f94ee Author: Eric Anholt Date: Tue Jun 10 11:32:49 2008 -0700 Merge branch 'master' into drm-gem commit 79f94ee3c4652019395b9c329d94f7779c558571 Author: Eric Anholt Date: Tue Jun 10 11:31:22 2008 -0700 Set up/restore PWRCTXA from enter/leavevt not server start/exit. This should improve behavior in the presence of VT switching, but also avoids a crash on X exit from writing the register after unmapping mmio. commit 231a302013981cc597ba09ee89b367c8ab56e8ba Author: Bryce Harrington Date: Tue Jun 10 14:10:39 2008 +0800 Two more Dell quirks fd.o #16160 commit cc98d41df63f1ee2561440fab3aa471a80a12920 Author: Eric Anholt Date: Mon Jun 9 12:00:14 2008 -0700 Add a little program to dump out the first 64 dwords of the status page. commit 1142be53eb8d2ee8a9b60ace5d49f0ba27332275 Author: Jesse Barnes Date: Mon Jun 9 08:52:59 2008 -0700 Fix TV programming: add vblank wait after TV_CTL writes Fxies FDO bug #14000; we need to wait for vblank after writing TV_CTL or followi ng "DPMS on" calls may not actually enable the output. commit 64a8f2433d7774d06119793b57cec6d3be6389c1 Author: Robert Lowery Date: Tue May 20 21:09:23 2008 +1000 Fix TV out connection type detection Make sure we wait for vblank when using the TV DAC to detect the connection type. Fixes FDO bug #14000. commit a993a8f3ce3a2079c5b5ebaca2ec2564047d7d4e Author: Tomas Carnecky Date: Fri Jun 6 10:03:23 2008 +0800 Fix compile error in LeaveVT commit 432d07fb175b385715eb3ccf97a303d2ad7bfe11 Merge: 3a967b8 bff180e Author: Eric Anholt Date: Thu Jun 5 14:57:18 2008 -0700 Merge branch 'master' into drm-gem commit bff180e6cac4452ef491c81855eb12bfa03d0bf3 Author: Eric Anholt Date: Thu Jun 5 14:55:23 2008 -0700 Move debug clock printout from ErrorF to X_INFO. commit ad459b21b7de4a79552ac155803d5930432fb84b Author: Eric Anholt Date: Thu Jun 5 14:37:59 2008 -0700 Remove SVG_WORK_CONTROL init. The bit set is now reserved -- used to be a workaround for early revisions. commit e41dc98a0aeec389b221038753056c725ecee1c5 Author: Eric Anholt Date: Thu Jun 5 14:37:26 2008 -0700 Initialize clock gating from EnterVT and save/restore it appropriately. commit 8061e5ac27a5f61f940bccc940be922999cc1d3f Author: Eric Anholt Date: Thu Jun 5 14:09:45 2008 -0700 Move BIOS scratch register setup to EnterVT instead of PreInit. We want these to always be set when our driver's in control. They are already appropriately save/restored at leave/entervt. commit b61cb9283185eb5211e84eb7d8e68beea607c2eb Author: Eric Anholt Date: Thu Jun 5 13:57:54 2008 -0700 Remove gratuitous wait_ring_idle after I830Sync. Syncing implies that. commit b65f67b74ff99aea1490d52f170ef4fe80cef6c7 Author: Eric Anholt Date: Thu Jun 5 13:40:55 2008 -0700 Move the remaining two lines of SetHWOperatingState to the caller. commit adb4f5a5e826e584ab212d23fc8d474c3e7bb8e8 Author: Eric Anholt Date: Thu Jun 5 13:39:08 2008 -0700 Remove duplicated i830_stop_ring()/SetHWOperatingState() in EnterVT(). commit c2e2fe48113667c683c6e1e9b1237635c41c61c3 Author: Eric Anholt Date: Thu Jun 5 13:21:55 2008 -0700 Move DSPARB setup into a separate function, and save/restore it at VT switch. commit 7e51384c973a96366b02ea646392c43574674111 Author: Eric Anholt Date: Thu Jun 5 11:56:42 2008 -0700 Replace a couple of wait-for-ring-idles with a single function to do so. commit 65306cdd71dad71e4ca7229764f81a0880dd70bf Author: Eric Anholt Date: Thu Jun 5 11:26:02 2008 -0700 Just remove the mprotect kludge. Besides not being #ifdef __linux__ed as requested, some linux kernels break in exciting new ways when you try to mprotect from PROT_NONE back to PROT_READ|PROT_WRITE. Yes, there are bugs in the code we're calling in a bug-exploiting bug workaround. If you want this workaround for the original bug exposed when moving to libpciaccess, it's already in libpciaccess. commit beb72ae5aa053303f5cc419e9c9d7c6db964f160 Author: Hong Liu Date: Wed Jun 4 16:58:05 2008 +0800 Set SDVO sync polarity to default on 965 Fix fd.o bug 15766 commit c7fee208fd46e143965ea173984d284e1eec2a9b Author: Hong Liu Date: Wed Jun 4 16:56:50 2008 +0800 Fix DSPARB change on 965G From the spec, only 965GM and IGD_GM have 128 FIFO entries. With DSPARB change introduced by commit bd137a, I've got PIPE B underrun when dual-headed on G35 platform. commit 64ccc8a036a1d80fc918e9b124d24ce0d26bdbbb Author: Zhenyu Wang Date: Wed Jun 4 22:30:04 2008 +0800 xvmc: add render dump function commit 1729a4f29dd52346a9fa997f818d57884047657f Author: Zhenyu Wang Date: Wed Jun 4 22:25:39 2008 +0800 xvmc: move block size lookup helper to generic commit 3cc8876db7ba82fa18e8fab4f267ac6e87964c47 Author: Zhenyu Wang Date: Tue Jun 3 16:15:36 2008 +0800 xvmc: a little cleanup Also safe check context size to not exceed surface max. commit 96955003557686469c7ae9f7d42620d2851e6fad Author: Zhenyu Wang Date: Tue Jun 3 16:13:52 2008 +0800 xvmc: remove unused dri drawable commit bd137a19dc29dd466eac030e040f729ed0807e3f Author: Jesse Barnes Date: Wed May 28 15:47:47 2008 -0700 Fixup DSPARB for 855 & 945 It turns out 855 has a different DSPARB layout than 915+. And 945+ have more FIFO entries, so we have to allocate things differently. So on 855 split the FIFO evenly again between A & B planes, and do the same on 945, where we have a larger FIFO. Fixes an issue reported by Daniel Stone with the previous default value. commit 3a967b8359bd6c05a10111076cdb02df15cbf2e8 Author: Eric Anholt Date: Mon May 12 13:51:11 2008 -0700 [gem] Note if pinning a buffer fails. commit 363ccc3dfbf8ab91608d502e0376a43653581856 Author: Jesse Barnes Date: Mon May 26 09:43:40 2008 -0700 Fixup FIFO comment What I originally checked in was a bit misleading. commit 2e1425246ccc75216247b0c2fa6fce2635db472b Author: Jesse Barnes Date: Mon May 26 09:40:10 2008 -0700 Handle display FIFOs better Add some debug code to catch FIFO underruns, which are normally bugs (unless they occur during mode setting) and remove any plane C FIFO allocations, since we don't use that plane at all. We may eventually need to be a little smarter about this on platforms that use plane C for the popup. commit 89bb53cc7a853d88fc34a0ca65ae2b6227a8dd24 Author: Jesse Barnes Date: Mon May 26 09:34:34 2008 -0700 Fixup power saving registers Update clock gating disable bits to match docs and allocate a power context memory area so that newer chips can save state and power down the render unit. commit ced4b4bb0c44b55feb5e32ddb860f1c1dc3bbde5 Author: Eric Anholt Date: Thu May 22 22:10:25 2008 -0700 [gem] Reduce console spam from debugging. commit 165c0865d849b7d280a3a119fe9ae0ad34637df0 Author: Alan Coopersmith Date: Wed May 21 14:56:27 2008 -0700 Define DEFFILEMODE for OS'es that don't have it commit 71180653825a1b141a08590e4b767d33d9b5d8c1 Author: Jesse Barnes Date: Wed May 21 11:51:55 2008 -0700 Revert "Disable FBC by default on 965GM" This reverts commit 53e3693ef13f31f3fc33bcff7286ab2b03b2d430. Conflicts: src/i830_driver.c - default FBC on for 965+ commit 740a73dad1d9a9b0ebf1cc7d6f2a056abad57aaf Author: Zhenyu Wang Date: Tue May 20 08:54:01 2008 +0800 Revert "Add FIFO watermark regs to register dumper" This reverts commit 0c00a638ef57aa9d6a3047176b0bfad733f781f0. Those FIFO watermark regs are 945-ish, and cause problem on G35. commit c81050c0058e32098259b5078515807038beb7d6 Merge: 9c9a5d0 e9532f3 Author: Keith Packard Date: Sat May 17 23:26:14 2008 -0700 Merge commit 'origin/master' into drm-gem commit 9c9a5d0e48c6e911574695fc417d2dc1a0fd1c20 Author: Keith Packard Date: Tue May 13 12:24:50 2008 -0700 Add check for GEM, use that to enable driver GEM support commit e9532f3abf1e0b62b2569d4aae60a5a63add571f Author: Rémi Cardona Date: Wed May 14 14:52:37 2008 +0800 Add glproto to DRI dependencies commit fe455a38cbac8d47a5c82b3eeb3cc6a59a6e8443 Author: Eric Anholt Date: Tue May 13 11:56:10 2008 -0700 Formatting fixes for intel_idle. commit 5c1720ecbc5528c40bb48084dc68ef233e9bb43b Author: Eric Anholt Date: Mon May 12 16:48:05 2008 -0700 Add i915 support to intel_idle. commit 089011daf3da8db3bd16d50e1d6a6457da82300e Author: Keith Packard Date: Mon May 12 12:18:19 2008 -0700 Track name changes in GEM ioctls. allocate → create unreference → close name → flink commit 10f1d835b8ac7bf3153ac39d295ad2a72c4aa076 Author: Keith Packard Date: Mon May 12 12:15:09 2008 -0700 Record alignment requirements in mem structure for use by GEM. GEM needs memory alignment requirements sent at pin time, which is a bit after the allocation itself. Store the required alignment in the memory object for later use by pin. commit 5103e5a39a4869a714b8a59da8bf51ecc8c65e4f Author: Jesse Barnes Date: Mon May 12 10:58:24 2008 -0700 Panel fitting: fix letterbox modes In full_aspect mode, we try to preserve the aspect ratio by adding either top & bottom or left & right borders. In the letterbox case (top & bottom borders) we were miscalculating the top border which led to programming a bad mode. Fix the calculation and bug #15559. commit f280c50dd5d29c5717adc6da2368363cb3d0adbb Author: Zhenyu Wang Date: Mon May 12 08:57:56 2008 +0800 Revert last HP 965GM pipe A quirk Seems not resolve the issue (fdo bug #15885). commit d91d9e6a2f2ba18b35cb6fd7bc3fe8bc617eb44f Author: Bryce Harrington Date: Sat May 10 07:58:59 2008 +0800 More Pipe A force quirks commit 548d4c0f979a72eac5d0c9c76da12f0bd05e64e1 Author: Zhenyu Wang Date: Thu May 8 12:43:30 2008 +0800 Fix i965 render getting dest format error Depend on value returned by function within assert is wrong. Fixed weird render corrupt on i965. commit 9f324860431ff8199a78d19bbaa74046e1476b89 Author: Mike Isely Date: Thu May 8 10:00:17 2008 +0800 Implement option to ignore external fixed mode settings The Intel xorg driver tries mightily to determine the native fixed panel mode settings for the LVDS output. It does this through various means, including scanning video BIOS tables, and noticing if the pipe in question has already been set up by somebody else (and adopting those timings). This strategy works well for say a laptop where the LCD panel is an integral part of the machine. But for other applications where the display is unrelated to the system's BIOS or other software, then the BIOS will likely have no clue how to configure the LVDS output. Worse still, the BIOS can simply "get it wrong", leaving the pipe misconfigured. Unfortunately the Intel driver can potentially notice this, adopt the same settings, leaving a messed up display. All of this complexity normally happens independently, behind the scenes, from the mode timings that might otherwise be specified by the user. This driver has a concept of fixed, i.e. "native" mode, and then user-specified mode. If the corresponding resolutions between those concepts don't match, then the driver in theory will arrange for scaling to take place while adhering to the actual native mode of the panel. Said another way, if the user says 800x600 but the driver mistakenly (see above) thinks the native mode is 640x480, then 640x480 is the mode set with scaling to an 800x600 frame buffer. If the driver gets the wrong native mode, then the result is a miserable mess with no way for the user to override what the driver thinks is right. This patch provides a means to override the driver. This implements a new driver option, "LVDSFixedMode" which defaults to true (the normal, probe-what-I-need behavior). However when set to false, then all the guessing is skipped and the driver will assume no fixed, i.e. "native" mode for the display device. Instead with this option set to false, the driver will directly set the timings specified by the user, providing an escape hatch for situations where the driver can't correctly figure out the right mode. Under most scenarios of course, this option should not be needed. But in situations where the Intel video BIOS is hopelessly fouled up related to the LVDS output, this option provides the escape hatch for the user to get a working display in spite of the BIOS situation. Signed-off-by: Mike Isely commit 33f033cbf346c13a687e469e8879579fcd5bb2fb Author: Jesse Barnes Date: Wed May 7 14:46:52 2008 -0700 Save/restore cursor registers There are lots of good reasons for doing this, one of them is fdo bug #11305. commit a7188b1f2dd9a69fa7daefe478d283735226f9f3 Author: Zhenyu Wang Date: Wed May 7 14:21:59 2008 +0800 Use I830FALLBACK when check pitch for debug commit 79fde3ad7a949fb7f0dc9907cb6eea6ca9a910ae Author: Zhenyu Wang Date: Sat Apr 19 08:54:38 2008 +0800 Check pitch for EXA operation 2D pitch limit applys to all chips. Pre-965 chip has 8KB pitch limit for 3D. 965 supports max pitch by current exa (128KB). (cherry picked from commit 8187a5a16f8bd8f0ba5e7f5357f355928b3b8f07) commit 36ec93300926084fb2951d69b001e4c67bc6ff79 Author: Eric Anholt Date: Tue May 6 18:48:20 2008 -0700 Bug #15807: Fix use of the ring while VT-switched, hit by fast user switching. The fix for flushing at blockhandler with no DRI on 965 was broken and would try to flush the chip even when the driver wasn't in control of the VT. Hilarity ensued. commit 0c00a638ef57aa9d6a3047176b0bfad733f781f0 Author: Jesse Barnes Date: Tue May 6 14:38:48 2008 -0700 Add FIFO watermark regs to register dumper commit c81a4687fca80bf7367d7f0e9a00a6a09737c5bb Author: Zhenyu Wang Date: Tue May 6 13:52:22 2008 +0800 Only check xvmc lib when xvmc is enabled. Don't check xvmc lib if user has already wanted to disable it. Fix fdo bug #15762. commit be746a90a87d7a9807fa4243493e7e4d48f7f1c0 Author: Bryce Harrington Date: Tue May 6 13:51:46 2008 +0800 More quirks from ubuntu/dell commit 417f86d80525d2f5038628afba8f70f8c0fecca3 Author: Keith Packard Date: Mon May 5 13:36:19 2008 -0700 Move a declaration under #ifndef HAVE_FREE_SHADOW commit 587dc5bfc4ef32942bd7511f932a3a2e2fcdccb0 Author: Keith Packard Date: Mon May 5 13:35:22 2008 -0700 Fix up ring dumping code for non-i965 commit 65877d32bb2cd71ccc8d6a66244f029fc3577f69 Merge: b9d12da fff17b9 Author: Keith Packard Date: Mon May 5 12:05:00 2008 -0700 Merge branch 'master' into drm-gem commit b9d12da91eb1afb05bd78a40e15f4333e697093d Author: Keith Packard Date: Mon May 5 10:44:09 2008 -0700 Use GEM for buffer naming now. commit 0741020f3725c32c48ad15535b9c0549e4092c23 Author: Keith Packard Date: Mon May 5 10:17:01 2008 -0700 Set pin alignment for gem on non-965 non-965 tiled frame buffers have fairly strict alignment requirements, 512K on 8xx and 1MB on 9xx, plus they must be aligned to the size of the allocation. commit 271abfd0ec29851273d8ca5241d963be5c64e43c Author: Eric Anholt Date: Fri May 2 10:06:13 2008 -0700 Add initial GEM hacks to bring the server up. commit a0ced923bb793aa22e6bfbeeec0888d3b42ce176 Author: Jesse Barnes Date: Tue Apr 29 13:19:02 2008 -0700 Add a new quirk for BIOSes that reprogram regs at lid close/open time Dell Latitude D500s seem to have this problem. At lid close/open, the DSPABASE reg gets reset to 0, so we either need to keep the framebuffer at offset 0 or make sure we reprogram the CRTCs after the lid opens again. Since we can't make sure the former is always true (buffer resize, etc.), this patch adds a quirk to reset the modes at lid open time. Fixes FDO bug #14890. commit fff17b9d1b58cb53032d153094826dd306836d59 Author: Keith Packard Date: Tue Apr 29 10:32:14 2008 -0700 Use new xf86RotateFreeShadow function to clean up shadow buffers. This simply moves code from the driver up into the X server; use it where available. commit 1d467a8038946a37844795e8860be113d43219ac Author: Keith Packard Date: Wed Apr 23 11:08:38 2008 -0700 Overlay video doesn't require that the target pixmap be in video memory. I830PutImage was checking to make sure the target pixmap resided in video memory, but this isn't necessary when using the overlay. Test commit 40e0a03af57d7b3180d7066bcb15e03dcc9ca295 Author: Keith Packard Date: Mon Apr 21 01:03:22 2008 -0700 Dump batch buffers found from main ring commit c8ae3b781f0d8e325876a74c91cd0a685d34454b Author: Keith Packard Date: Sun Apr 20 02:11:15 2008 -0700 Add a bunch of 965 ring stuff to the debug dump commit 2c135ef8ac40f8e7cd071de7414adfae019f9198 Author: Eric Anholt Date: Wed Apr 16 21:39:58 2008 -0700 Make the binding table and surface state be arrays to reduce syncing. commit c3fb62df4e60b63295f94c99b3c5de70dbf94e1c Author: Keith Packard Date: Thu Apr 17 10:04:55 2008 -0700 Add a kludge-around to fix cd/wt bits in fb ptes on linux. Mmap from /sys/devices/pci* on linux forces the cache-disable and write-through bits, which turns our write-combining map into an uncached-map, seriously impacting performance. It turns out that a bug in mprotect allows us to fix this by disabling access to those pages and then immediately re-enabling them. commit 0ae283582d21776d3317d5fc1c25751d50d562c7 Author: Hong Liu Date: Thu Apr 17 10:57:34 2008 +0800 fix possible segfault in I830FreeScreen pI830 may point to NULL if I830PreInit fails commit 7008eac855302fd467e001495483c2dccfddf68f Author: Eric Anholt Date: Wed Apr 16 12:09:41 2008 -0700 Update dolt from git, fixing fallback to libtool. commit 23d1df22d177e54bfc46304053d8115047ff85d4 Author: Eric Anholt Date: Mon Apr 14 16:46:17 2008 -0700 Abstract surface setup into a separate function. commit e27ce4238f748d024635835824c120eb298b84b2 Author: Eric Anholt Date: Mon Apr 14 23:30:21 2008 -0700 Move the vertex buffer to the state structure as well. commit 69709fe36cfdb59c140f5cba233dcb606f8dbc85 Author: Eric Anholt Date: Mon Apr 14 18:34:13 2008 -0700 Move the binding table for 965 to the state structure. commit 049d057dfd73b6c191a8ba5d73eddb77cf7748e7 Author: Eric Anholt Date: Mon Apr 14 17:09:22 2008 -0700 Move the 965 static variables local to the function they're used in. commit 092962c962fdf3e33a8d387221448fe7c8c36376 Author: Eric Anholt Date: Mon Apr 14 14:02:58 2008 -0700 Make the VS unit state static for 965 render. commit 4bbdd7096d4927fe48673006fff7df922972f116 Author: Eric Anholt Date: Mon Apr 14 13:53:36 2008 -0700 Rename some variables in i965_render.c for clarity. The gen4_render_state is now always called "render_state" (i965_render.c bookkeeping) and gen4_state_t is now always called "card_state" (the buffer for state used by the chip). commit 456bb529335c28cf60f3b6e5900b804efa5c185a Author: Carl Worth Date: Fri Nov 9 17:24:21 2007 -0800 Associate one sf_state object with each sf_kernel (cherry picked from a2b5c23184d19b386fdfd04f578a55566df60132 commit) commit 90d6b178473ba32cf66e6e654e608cb4374e4a19 Author: Zhenyu Wang Date: Mon Apr 14 16:31:03 2008 +0800 Change default panel fitting mode to origin i8xx currently only works in FULL mode. (cherry picked from commit 33ffd781bbca3d0dee8c1b47e7b90be5824b9a4f) commit c1ad0a1a344a87a537aa5f93f6c774997a3189e4 Author: Zhenyu Wang Date: Mon Apr 14 10:02:00 2008 +0800 Remove old sf/wm program from Makefile Fix distcheck (cherry picked from commit 1d8383923bebf38ab8ada6a262cb4fc8c30e074a) commit 18ef4158e5574bfc6621b268821532f13e261d13 Author: Keith Packard Date: Sun Apr 13 21:37:49 2008 -0700 Only use FOURCC_XVMC when INTEL_XVMC is defined The XVMC code uses a magic FOURCC code to signal frame updates, but that code is only defined when the XVMC code is used. commit 269809030ee9d08b56ad0178a7a9c0194e25785e Author: Keith Packard Date: Sun Apr 13 12:47:02 2008 -0700 Move wm_scratch first to meet alignment requirements commit 08d3206f3b891e611e3e16c9eb79cfca21ab8a51 Author: Keith Packard Date: Sun Apr 13 12:35:14 2008 -0700 Kludge to make planar video work - duplicate sampler state This patch duplicates all of the surface sampler state needed to read from the separate YUV surfaces and then has the planar sampler read from samplers 0, 2, 4 instead of 0, 1, 2. This appears to make things work, instead of having the samplers break at random. I do not understand why this works. commit 6366e4de540c4713cbc8402c89516464bbdc35b9 Author: Keith Packard Date: Sat Apr 12 16:39:00 2008 -0700 Skip copying on FOURCC_XVMC surfaces commit 0fec42b64ee529eb7ec15febdaa5e4986ec1ab17 Author: Keith Packard Date: Sat Apr 12 11:01:14 2008 -0700 Set some mask bits to try and get yuv conversion working commit 91d3e19786e22061f4b617cf39a3cd713139a2d9 Author: Keith Packard Date: Sat Apr 12 11:00:51 2008 -0700 Use compressed instructions for yuv conversion commit 03836067b77606c134c71b30c7078d09d77c95fa Author: Eric Anholt Date: Fri Apr 11 12:17:28 2008 -0700 Fix the offset to sampler default color, and a compiler warning. commit d3138d9ff0f821fb3adbd27684b0c22a8a910c3e Author: Carl Worth Date: Tue Nov 13 12:37:13 2007 -0800 Enumerate all possible wm_state objects We have a collection of wm_state objects for each ps kernel, (one for each combination of src and mask extend and repeat values). Thanks to Dave Airlie for noticing an errant write through a wild wm_state pointer in an early version of this commit. (cherry picked from 7763706a93d3021907273f9b330750ba110e2fc3 commit) This cherry-pick required more reformatting than most, due to the projective texturing merge. commit 2293a3677d1dcf294de6a8712bf0d2f65b50dcc3 Author: Carl Worth Date: Mon Nov 12 16:09:00 2007 -0800 Enumerate all possible src,mask sampler state pairs This will eventually allow for the elimination of sampler state updates while compositing---and initializing everything in the initialization function. (cherry picked from commit d0874697be8086cd64740c24698df8cd4d31c76f) commit da85d1b822dcb31698e9c5ab85a7bb27ad745eee Author: Keith Packard Date: Fri Apr 11 11:22:29 2008 -0700 Revert "remove '#line NUM ...' in macro process" This reverts commit f47486fab3dffcbb03e7ad89f777abba1e887299. Sorry for failing to push the #line processing changes to intel-gen4asm; those are now pushed, so this change should no longer be necessary. Having correct file name and line numbers in the error messages is awfully nice though. commit f47486fab3dffcbb03e7ad89f777abba1e887299 Author: Zhenyu Wang Date: Fri Apr 11 10:12:40 2008 +0800 remove '#line NUM ...' in macro process intel-gen4asm doesn't allow '#' line commit 5c9cde37e769287fb7bf4e08c3600a33c2e92dce Author: Hong Liu Date: Fri Apr 11 09:54:34 2008 +0800 Bug #14935: Fix i9xx reference clock for spread spectrum. commit 152a50703aa5e9ebaa9abbe448518742734a5eb7 Author: Eric Anholt Date: Thu Apr 10 16:06:41 2008 -0700 Fix compiler warning from 24-bit lvds change. commit baa55ffe3b67ca952e4d1ef7087335d76eb2f24b Author: Carl Worth Date: Tue Nov 13 12:56:13 2007 -0800 Enumerate all possible cc_state objects We need one for each possible combination of src and dst blend_factors. Again, as with recent changes, this eliminates state updates from prepare_composite and allows that function to instead simply reference an existing object initialized within gen4_state_init. Thanks to Dave Airlie (and git-bisect) for pointing out that with gnome-terminal all text was appearing as solid black with an early version of this commit. As expected the bug was an alignment issue. (cherry picked from 0c0ab52c2d100c47f38c7ef826ef585c8b9815e9 commit) Performance is approximately equivalent on text tests, but may be around +2%. commit d5a80e1e3ab5724d34b20f9ee6f830efd0f5b076 Author: Keith Packard Date: Wed Apr 9 16:27:40 2008 -0500 Single memcpy when pitches align on planar image transfer commit ac97f2b1487df5574875350a9cded958dae33afa Author: Keith Packard Date: Wed Apr 9 16:27:23 2008 -0500 Use available symbolic register name commit f270456e5612cb88933e6aabcd9a816c5c292229 Author: Keith Packard Date: Wed Apr 9 10:00:08 2008 -0500 Remove .g4b files on clean commit fb6ed8b8b59e9dd18801afef5d7c62042ad176d7 Author: Keith Packard Date: Wed Apr 9 09:58:44 2008 -0500 Update yuv->rgb conversion programs to write to src_sample regs The YUV->RGB code was written to write directly to the dataport registers, but that didn't work for the compositing functions (cause still unknown). This change makes that code write RGB values to the src_sample registers as with the other sample computation fragments. commit b01d582e23fc99e32bc47a395e9caa366731372a Author: Keith Packard Date: Wed Apr 9 09:30:35 2008 -0500 Revert "Compute pixel values directly into data port" This reverts commit 346cf57deabb4c336612df4c13650a87b5ef6775. Mixing randr transforms and video caused screen corruption for Render operations. No, I don't understand why. commit 3fc3d1a701bae257b70aa7b7654c722f30e71399 Author: Keith Packard Date: Wed Apr 9 02:02:56 2008 -0500 Remove sync after 965 video put. The hardware has been marked as needing a sync, so the next video put will block waiting for the previous one to complete. Adding a sync here just stalls the video playback for no good reason. commit 825d9e50c59450f07178a54fed2616e551dc0455 Author: Keith Packard Date: Wed Apr 9 00:09:34 2008 -0500 Add planer video decode kernel Support for planar video reduces bus bandwidth by 25% and also reduces CPU usage during planar->packed conversion. commit a03eaaa67b33c57530e92c53d28917e2563b4427 Author: Keith Packard Date: Tue Apr 8 16:24:55 2008 -0500 Use shared exa_wm code for packed yuv decode Eliminate special video sf and ps programs. commit 32ef98518394d29cb87405005c660278489396bb Author: Keith Packard Date: Tue Apr 8 16:21:55 2008 -0500 Compute pixel values directly into data port Instead of leaving pixel values in src_sample registers, compute the pixel values directl to the data port to save 8 moves. This cannot work when no computation is done as there is both no way to wait for the sampler to finish and because the sampler returns data in a different order from that required by the data port (sigh). commit b68d9f4245d0ebe3371c179401ff145f1a4d101b Author: Keith Packard Date: Tue Apr 8 16:09:00 2008 -0500 Use symbolic names for channels in YUV code commit 781be9d47289713b0a8fcd95c769a9c6241d62e9 Author: Keith Packard Date: Tue Apr 8 16:08:20 2008 -0500 Rename src/mask/data registers to indicate channel commit 2871ac8eefd0192080bb0569140c3f5d0e1d9b44 Author: Eric Anholt Date: Thu Apr 10 13:34:13 2008 -0700 Statically allocate the sampler default color, which we never change. Performance change is in the noise. Also from Carl Worth. commit 80dd784e33847e431403d4659a7b8d8425b2676f Author: Eric Anholt Date: Thu Apr 10 13:24:51 2008 -0700 Add copyright information for recent editors of this file. commit b606278db83ec84b1db562a2d65697c50561b169 Author: Eric Anholt Date: Thu Apr 10 13:17:58 2008 -0700 Keep static copies of the 965 render programs in video memory. This reduces the CPU overhead of memcpying them in every time, for a speedup in aa24text of around 30%. This is based on work by Carl Worth which is in the intel-batchbuffer branch. commit bd41d2bb858770c109d5013274a4354e8e6ae32d Author: Eric Anholt Date: Thu Apr 10 10:24:48 2008 -0700 Add dolt to speed up compile time and reduce insanity. Hopefully some day we can be rid of libtool entirely. commit 96ae9a9f0abc582e45ff4e17dd213c8739c24cb3 Author: Eric Anholt Date: Wed Apr 9 14:17:29 2008 -0700 SDVO: Rename output_dtd variable to input_dtd, which is its new job. commit 72c3995d5158286d20e97296f4552697f32f935f Author: Eric Anholt Date: Tue Apr 8 15:00:18 2008 -0700 SDVO: Set adjusted_mode in mode_fixup from the preferred input timings for TV. commit b61178524bece756dc0fd4f798c6c7743b5725db Author: Eric Anholt Date: Tue Apr 8 14:11:42 2008 -0700 SDVO: Fix lols TV mode list handling. commit 5db73f18d2badec37523850024b9aa824e4e6bd8 Author: Eric Anholt Date: Tue Apr 8 13:53:51 2008 -0700 SDVO: Use the TV reference clock for SDVO TV outputs. Based on a patch by Will Wiseman. commit e031cc02e65acfbafb48136dad414751e04425c5 Author: Mike Isely Date: Wed Apr 9 14:15:03 2008 +0800 Implement support for 24 bit pixel format The Intel driver appears to be coded to only work with displays expecting 18 bit pixels. However I have an application using a LCD display that expects pixel data in 24 bit format. The difference is only 2 bits in a single GPU register. This patch implements that change, controlled by a new driver option, "LVDS24Bit". The default value is false, which is the previous behavior. When set to true, then 24 bit panels should work (at least the one I'm testing here does). Fd.o bug #15201 Signed-off-by: Mike Isely commit 79b18980ac9f60b1978abe421352df965aed1681 Author: Zhenyu Wang Date: Wed Apr 9 14:03:45 2008 +0800 Include in i810.h for integer types definition commit 2b720262e1235f1c9da860ba3e9181f0c377aa5e Author: Jesse Barnes Date: Wed Apr 9 17:34:17 2008 +0800 Fix LVDS regression Disable panel fitting on 855GM, and fix dither setting. commit af92f4f885c8eae2211d09080b2289aa5bab9ec1 Author: David Schleef Date: Tue Apr 8 12:43:20 2008 -0700 Bug #15219: Fix default overlay CSC factors, and use BT709 if available. Using the updated factors even when BT709 conversion isn't available (non-965) should still give us better color reproduction. Tested on a 945GM, examining the +/-5% of black bars of videotestsrc. commit 37bc23660a8c346f1eaa6c93ed2c7a840828f0b0 Author: Bryce Harrington Date: Mon Apr 7 17:26:49 2008 +0800 Quirks from Ubuntu/Dell FD bug #15353. Launchpad bug ID is available for reference. (cherry picked from commit 24248097988775d62f6c416f2988e74d31c91cd0) commit dcbdc901d64d2fab489292ef1c9ca1490bdd1e94 Merge: 0147c1c 13563bd Author: Keith Packard Date: Fri Apr 4 19:19:35 2008 -0700 Merge commit 'origin/master' into projective-965 commit 0147c1c84872f7a109721a53d88a539932d9be81 Author: Keith Packard Date: Fri Apr 4 19:17:55 2008 -0700 Forgot to build exa_wm_src_sample_argb.g4b commit 13563bd65cbbbb3e697dacd814efa3517e1493dc Author: Hong Liu Date: Wed Apr 2 16:05:43 2008 -0700 SDVO: Fix DDC bus selection based on output priority list. commit 6ad2d6ba86689674876f5f4c473f11e39243ac38 Author: Eric Anholt Date: Wed Apr 2 15:16:17 2008 -0700 SDVO: Fix mixups with input and output channels. The 2-bit input_mask was actually an input count -- in0 is always there, and in1 is optional. The output flags weren't being reported in the log, so I mistakenly took controlled_output == RGB0 to mean that the device only reported an RGB0, while it actually reported RGB0|SVID0|YPRPB0|misc|other. Move SVID0 up in priority and remove the RGB-is-it-really-TV hack I had just come up with. Finally, set the input/output mapping at mode set time. We're always supposed to do this, but haven't had to so far as we've never handled devices with more than one output. commit 3a17400dc67534f0eb474ece080f01061469569c Author: Eric Anholt Date: Wed Apr 2 14:21:23 2008 -0700 SDVO: Handle RGB outputs that are really TV outputs, and select a TV format. Still doesn't light anything up. commit 3c1701797b61afaae826a78455079e2115483053 Author: Eric Anholt Date: Mon Mar 31 10:04:18 2008 -0700 Add WIP SDVO TV-out support. Doesn't include properties for selecting TV formats or picture enhancements, and totally untested. commit 9d5ba26fb5c337388920b45eadda85e43bc564fa Author: Eric Anholt Date: Fri Mar 28 14:28:28 2008 -0700 Add SDVO TV-out registers. commit 90886f9a602d58b754e9a8d0f1a9c40803d34fa2 Author: Keith Packard Date: Tue Apr 1 00:54:13 2008 -0700 Remove extra flushing commit 0836373dc6e2f8612f120074980561f7ac11f6f7 Author: Keith Packard Date: Tue Apr 1 00:16:05 2008 -0700 Add projective versions of the PS kernels commit bfd803e085e938866efb45c67a79facef78ec399 Author: Keith Packard Date: Tue Apr 1 00:06:08 2008 -0700 Shrink WM thread to 32 registers and 1024 scratch space. Saving registers means we can run more in parallel. commit a6492661ae07310128eb73c3ef037c42ce7ab184 Author: Keith Packard Date: Mon Mar 31 23:50:20 2008 -0700 Fix composite with mask using new compositing thread code Clean up register allocation to never overlap Always write 4 values for each texture vertex. commit f8081178eb6fda0e405967cbacad532561619262 Author: Keith Packard Date: Mon Mar 31 12:06:37 2008 -0700 remove old shader source files commit 879f8717b09f79156b218ee9cc2107700190d586 Author: Keith Packard Date: Mon Mar 31 12:05:56 2008 -0700 remove old monolithic shaders commit 6db8faeb754897b21af045d00f50db9640b080bb Author: Keith Packard Date: Mon Mar 31 12:04:00 2008 -0700 remove old monolithic shaders commit 6bb92213374f278387c539bbe05b773e87e11b90 Author: Keith Packard Date: Mon Mar 31 11:59:14 2008 -0700 Back to new shaders, fix urb read length commit 05710145b6fc4ed2c528128b2e6022591a53d050 Author: Keith Packard Date: Mon Mar 31 11:31:31 2008 -0700 Dont set the compr bit on 8-unit sends commit 08500507284f13ad7084eb231b43e117e9728129 Author: Keith Packard Date: Mon Mar 31 02:20:43 2008 -0700 Use m4 to clean up gen4 asm progs. Start adding projective transform support. Use macros for register names, modularize functions into separate files. commit 949d73271d7100c1f028fd60f185f4929461304e Author: Keith Packard Date: Sun Mar 30 19:19:46 2008 -0700 Remove rotation sf and wm progs commit 492ff1494f782240e6ca68919b2d0b9aa400fc53 Author: Keith Packard Date: Sun Mar 30 19:14:18 2008 -0700 Fix the sf_mask program to compute and pass corrected uvw cooefficients sf_mask is the same as sf except that it must compute both src and mask uvw cooefficients, which are conveniently adjacent in the same registers, and so need only an extended execution width commit 6304b38423f99190a5e54f1a7dcaa75adfad4f2a Author: Keith Packard Date: Sun Mar 30 00:54:51 2008 -0700 Reimplement wm program for nomask case to handle affine transforms This involves correctly computing u/v locations based on x/y vectors and line constants computed in new sf program. Also, use fewer instructions to make this go a bit faster (2X for 500x500 composite). commit 771a56b1ed0df69345c723cb62a73b6842cd8227 Author: Keith Packard Date: Sun Mar 30 00:54:07 2008 -0700 Clarify comment in exa_sf program commit 4f469189fed541549e5d470b2529275a29cc2f20 Author: Keith Packard Date: Sat Mar 29 14:28:05 2008 -0700 Compute du/dv/dw in no-mask SF prog commit 7dcb6e627449c80cea9812462ce6a3e125bd1240 Author: Keith Packard Date: Sat Mar 29 14:27:35 2008 -0700 Disable cursors while switching modes commit d50264fc7c4e0f80d9cc68dfdf322f68520a0f1e Author: Eric Anholt Date: Fri Mar 28 15:15:26 2008 -0700 Add a "total" field to intel_idle output (set when any unit is busy). commit 1450acd046d47e1739a3ffbd146c73ad2974a935 Author: Jesse Barnes Date: Fri Mar 28 13:23:02 2008 -0700 Use combo backlight control rather than pure legacy If the legacy bit is set, use both the BLC_PWM_CTL and LBB regs to control the backlight, rather than just LBB. Looks like more platforms want that than what the current code does. Note that kernel provided interfaces will always be used if available, so this shouldn't affect users with /sys/class/backlight interfaces at all. Fixes #14721. commit 49f409fef5e8b04a8595309ba8959ca36b2e41b1 Author: Matthias Hopf Date: Wed Jan 9 16:31:41 2008 +0100 Include missing stdlib.h for free() commit 3414313ac9d0faa95797ee18cc33afa231ec7581 Author: Matthias Hopf Date: Wed Jan 9 16:08:57 2008 +0100 Fix obviously wrong verification of DRI_SUPPORTS_CLIP_NOTIFY. commit c40f195f2dc3467259b4588e087aac9741003ebf Author: Matthias Hopf Date: Wed Jan 9 14:40:49 2008 +0100 Fix obviously stray '(' in non-I830_USE_EXA case. commit 7bba2c13310ed5ac22a355a3cc0ec8b7afaa79cf Author: Zhenyu Wang Date: Fri Mar 28 11:41:24 2008 +0800 Disable DRI earlier if fb width > 2048 commit b1f358ba97473b792ec2b7ed5170152faebe7262 Author: Zhenyu Wang Date: Fri Mar 28 11:29:40 2008 +0800 Disable overlay on new mobile chipset It's gone, really. commit 588371ba15afffa9343629f0ea7ce530586eef0b Merge: 52d6ced c13efdd Author: Jesse Barnes Date: Wed Mar 26 14:53:15 2008 -0700 Merge branch 'master' into pfit commit c13efdd3d6d3846078766814b57306a7519bce18 Author: Paulo Cesar Pereira de Andrade Date: Wed Mar 26 10:11:36 2008 +0800 More compile warning fixes. o Check for RANDR_GET_CRTC_INTERFACE before defining functions that are used only if it is defined. o Declare a variable before code, and rename it from ret to xvmc_status to better describe it. o if 0 some static functions not used. o Don't declare some unused variables. o Declare as static some functions that are used only in the file defining it. o Add a default/fallback return True to the Bool function src/xvmc/intel_batchbuffer.c:intelInitBatchBuffer(). o Ansify src/xvmc/xf86dri.c. o Add missing prototype to src/xvmc/xf86dri.h and follow pattern of other headers by adding "extern" before function prototype. commit 8cdbd55f8075cd18b563badde35815665d7d053e Author: Eric Anholt Date: Tue Mar 25 10:06:14 2008 -0700 Fix 965+ rendering issues with DRI disabled. The new chips no longer automatically flush the rendering cache, so if we don't flush the RC at blockhandler, the last rendering done may not appear on the screen. This was particularly noticable with a bare Xorg with some missing root weave, and terminals where the last character wouldn't appear until the cursor blinked. A flush in the DRI blockhandler path had hidden this issue for most people. commit c02ab432dd7058c700c35eecf6215daf5f262c51 Author: Eric Anholt Date: Mon Mar 24 13:48:20 2008 -0700 Revert "Use mprotect on unbound AGP memory to attempt to catch use while unbound." While I still like the idea, the mprotect calls themselves are failing on Linux and causing more trouble than they're worth. This reverts commit a1612b7728d4153499fe86b6713a13c8702cc7d9. Conflicts: src/i830_driver.c src/i830_memory.c commit 552a1b824db31a234d7c5cb71057ed0e0ce64477 Author: Eric Anholt Date: Mon Mar 24 13:25:37 2008 -0700 Disable a bunch of clock gating disables on IGD_GM, which doesn't need them. Besides our driver having fallen through to the GM965 path for RENCLK_GATE_D1, the BIOS was turning some of these on. It may be relevant for previous platforms as well to zero out the fields that should be zero in the other registers. commit c1a2bc357fc122d3718eca771530367ba2915815 Author: Julien Cristau Date: Mon Mar 24 13:29:24 2008 +0800 Don't try to build XvMC if DRI is disabled Instead of failing the build, don't even try unless --enable-xvmc was used. commit 5fcc002f77ebb66c1dc0f901a377ccf9bbc1d0bd Author: Maxim Levitsky Date: Mon Mar 24 13:24:12 2008 +0800 Fix crash triggered by dpms low power mode with hardware overlay running When overlay switch crtc, it ignores current crtc dpms mode which might lead to hang. commit 645980596450ed21c3b8927410a6bfe38a0c55d1 Author: Keith Packard Date: Fri Mar 21 03:08:24 2008 -0700 Transformed coord computed using floats. Don't move bilinear composite dst. The homogeneous coordinate computation in the core server cannot be used for many legal matrices as it overflows. Just use floats in the driver; faster and avoids troubles. When compositing with bilinear filter, don't push the dst coordinates around as that makes the output blurry when pixels are aligned. commit bf725b92cf93d74a6f414f81f319dddbc2c78525 Author: Zhenyu Wang Date: Fri Mar 21 11:50:27 2008 +0800 xvmc: fix distcheck Add i915 headers in Makefile.am commit e8da1a5418ab240ae7b1ef95bfea6bee22f1e0c1 Author: Joerg Sonnenberger Date: Fri Mar 21 11:41:49 2008 +0800 Fix DRI config for NetBSD Reported by Joerg. This does a little more clearn up. commit 224c36cc14a7552bc5df793ad582f0ea5349eb1e Merge: 4b9b7b0 03ccffd Author: Zhenyu Wang Date: Thu Mar 20 10:31:57 2008 +0800 Merge branch 'xvmc' commit 52d6ced652059989e6d9780a149488ccd16e3a22 Author: Jesse Barnes Date: Wed Mar 19 11:59:15 2008 -0700 Fixup backlight control and panel fitting property names They should be listed as lower case, since that's what you'd pass to xrandr. commit 4b9b7b007d729f94b01b0031d8ae478134b501da Author: Keith Packard Date: Tue Mar 18 14:08:19 2008 -0700 Handle projective transforms on 9xx for Composite. Projective transforms require un-normalized texture coordinates and the use of the texldp instruction. The coordinates are passed as x/y/z/w (the z is unused, but there isn't a vertext format for just x/y/w). commit f699389818f1f11f3edddcdddcd0a43be21ba4c0 Author: Keith Packard Date: Tue Mar 18 14:06:47 2008 -0700 Add i830_transform_is_affine and i830_get_transformed_coordinates_3d. These are needed to deal with projective transforms in the composite operation. commit a55974b435224fe5726fa44298d051c71e4056a5 Author: Keith Packard Date: Tue Mar 18 13:47:20 2008 -0700 Add defines for 9xx S3 word in 3D_STATE_LOAD_STATE_IMMEDIATE These aren't used at this point, but what the heck. commit 4f5500abe209b92b39ae1f2d7a1118362ac95034 Author: Keith Packard Date: Tue Mar 18 13:45:46 2008 -0700 8xx/9xx can handle textures to 2kx2k. 965 can do 8kx8k commit b1b173d03b3acd300c3b0f0ceffeddf1a8137839 Author: Julien Cristau Date: Tue Mar 18 12:02:25 2008 +0100 Add i8[13]0_ring.h to _SOURCES to fix distcheck Thanks to Bruno . commit 03ccffd5b9b8108166ea143884a1a0a6417f3236 Merge: ac76363 d72e18c Author: Zhenyu Wang Date: Tue Mar 18 16:54:15 2008 +0800 Merge branch 'master' into xvmc commit ac763634069fe070b3afc073ce437959612d39fe Author: Zhenyu Wang Date: Tue Mar 18 16:52:29 2008 +0800 Disable XvMC if DRI is not enabled. Default XvMC to disabled. commit d72e18c10995079121eded758a60ed99b5b8ce1a Author: Paulo Cesar Pereira de Andrade Date: Mon Mar 17 16:59:12 2008 +0800 Compile warning fixes. Move some declarations and don't declare an extra variable with the same name, to fix warnings about mixed declarations and code. commit 81df48d8d151907c2e9dd33a729860c39aa78175 Author: Zhenyu Wang Date: Mon Mar 17 10:05:57 2008 +0800 bug 15060: Dell Latitude X300 needs pipe A quirk commit ffb5a115cc62ed63110f77815b7f04976f720065 Merge: 05cf070 24e6eb2 Author: Jesse Barnes Date: Mon Mar 17 15:19:05 2008 -0700 Merge branch 'master' into pfit commit 24e6eb2ae683ed6614ac79c0d7811ac9fe94209f Author: Jesse Barnes Date: Mon Mar 17 14:17:27 2008 -0700 Add support for new ->get_crtc output callbacks Using the new interface allows the server to avoid some flicker at startup. commit 69fbc17441d0f894d17b058e65ae22300cd2a54c Author: Eric Anholt Date: Fri Mar 14 10:04:18 2008 -0700 Change OUT_RING and similar calls to OUT_BATCH for batchbuffer merge commit 65f92cfb7a05c9c028cf73ce1221cc0a651b50b3 Author: Eric Anholt Date: Fri Mar 14 10:03:11 2008 -0700 Remove h0h0 UploadToScreen implementation that used pattern blits. It never worked and was never that hot of an idea. Most of the motivation is gone with glyph pixmaps anyway. commit 05cf07071e25f84dec3476a9bed7235ed50cd249 Author: Jesse Barnes Date: Thu Mar 13 18:12:00 2008 -0700 Initial panel fitting changes Basic support for panel fitting. commit 402fbd36221d2ed27913850f6bfcce768b970fe8 Author: Jesse Barnes Date: Thu Mar 13 16:12:20 2008 -0700 Remove fbc enable/disable messages Tired of them filling up my logs. commit c20d78a7bc512219fa36dfc9707fe65f66ab80f1 Author: Eric Anholt Date: Mon Mar 10 17:27:45 2008 -0700 Add workarounds for 830/845's lack of a8/x8r8g8b8/x8b8g8r8 support. When we have unset channels (color for a8, or alpha for x8*), force them to the appropriate value in the texture combiner rather than relying on getting the desired results from teture mapping. Performance is the same on 865, so multiple paths aren't necessary and we can use the less friendly texture formats everywhere. commit f425181695b1b3864373e1919cf651229f19fa69 Merge: f4cfc98 1c0e495 Author: Zhenyu Wang Date: Thu Mar 13 23:58:56 2008 +0800 Merge branch 'master' into xvmc commit f4cfc98ab038af4d5d52af12e4d60ded62cf4a6e Author: Zhenyu Wang Date: Thu Mar 13 23:56:48 2008 +0800 nuke unneed error_base and event_base commit f269c69b4e7be076dc9838d0e0801c940ccdba08 Author: Zhenyu Wang Date: Thu Mar 13 23:41:22 2008 +0800 remove unused init/fini hooks for xvmc driver commit 1c0e4958564588162bc22a980567436dfc9d1e56 Author: Zhenyu Wang Date: Thu Mar 13 10:15:55 2008 +0800 Always bind memory after agp aperture allocation This fixed the crash appearing for Xv buffer memory, which appears allocation and free happening in stolen memory. We should always try to bind for stolen memory for correct protect setup, otherwise we might get fault that stolen memory protect reverted to non-r/w after previous free and unbind. Credit goes to Lukas Hejtmanek to identify the real problem of Xv crash! commit 2e2372912ed9bc0d86e8960653ef0bfce5cf99ab Author: Eric Anholt Date: Wed Mar 12 11:38:56 2008 -0700 Eliminate local variables defined in i830 BEGIN_LP_RING(). This lets us get better sanity asserts, and avoid mysterious braces when you do two BEGIN_LP_RING()s in a single function. Potential minor performance loss isn't too exciting, as ring access is about to become a compat path anyway. This change also removes the requirement for ring emits to be aligned to dwords. commit abf11a274e14535630742fe4c41cc0ae92555293 Author: Eric Anholt Date: Wed Mar 12 11:03:26 2008 -0700 Separate i810 and i830 ring macros out into separate files. I want to hack on i830 for changing it into a compat path for batchbuffer without having to worry about the i810 stuff getting broken. commit e04908853db74dc76718f6bc0dfb23b5ab3c5837 Author: Jesse Barnes Date: Wed Mar 12 15:44:58 2008 -0700 Add sysfs backlight support for Fujitsu laptops Tested by Dan Williams. commit 907399660b83b3939aab3dc7d0697197441e4ec8 Author: Jesse Barnes Date: Wed Mar 12 12:33:36 2008 -0700 Add stdint.h to DVO drivers Fixes the build for me (just following the git development guide on a new machine). commit 8563effc96195116c96f89a801c1de6ab982630a Author: Jesse Barnes Date: Wed Mar 12 11:14:17 2008 -0700 Add stdint.h to i830_bios.h Needed for new uint* usage. commit c82d8e51c6ff9522ad0deb33399d13f3d6fef19d Author: Jesse Barnes Date: Wed Mar 12 11:13:50 2008 -0700 Add pipe A force quirk for Thinkpad R31 Fixes #15036. commit a8f4108bea0966406b0fcae8d062ae1cfb4afbb3 Author: Hong Liu Date: Wed Mar 12 21:55:56 2008 +0800 Fix rotation mem with kernel memory manager in VT switch Unbind and bind a DRM BO may change the buffer offset, thus crtc may reference a wrong rotated memory after a VT switch cycle. Destroying it here will cause its reallocation when entering VT. commit cd73c6a5a78eac726b14ef242b8452c559c4c003 Author: Zhenyu Wang Date: Wed Mar 12 20:37:41 2008 +0800 Pass default visual in dri context create commit 21580add63307ca68790bf911243b8675f5f7c64 Author: Zhenyu Wang Date: Wed Mar 12 20:37:17 2008 +0800 disable memory alloc debug commit 5f272471878c896038b2c3f2bf704655d908828d Author: Eric Anholt Date: Tue Mar 11 12:14:51 2008 -0700 Nuke more CARD* types from i810 driver. Untested. commit 9a62d3b598cca3c28fa8b6313bba82983cc29ecf Author: Eric Anholt Date: Tue Mar 11 12:07:52 2008 -0700 Remove i830+ driver's use of CARD*/INT* types for great justice. Several uses are actually left, which are determined by the X Server interfaces we're implementing. commit 15f17aa682129c57d3864a355834edff06e91540 Author: Eric Anholt Date: Tue Mar 11 11:27:10 2008 -0700 Cleanse DVO drivers of CARD* types. commit 642ba6278876ec945ad687c2b9624e532cd3f629 Author: Eric Anholt Date: Tue Mar 11 09:59:16 2008 -0700 Add support for brightness control to i915 textured video output. commit ecdb5963ef6873ab82998dca6ca5186644666ec8 Author: Jesse Barnes Date: Tue Mar 11 11:08:25 2008 -0700 Add pipe A force enable quirk for Toshiba Portege R205-S209 Fixes #14944. commit 4ea67e9b853e9ed61557780b87bd184fc81fd731 Author: Kristian Høgsberg Date: Tue Mar 11 13:54:33 2008 -0400 Crack down on warnings. Fix printf formatting warnings, wrap a couple of long lines, nuke unused variables, add missing #include . commit 9fd13e6773371c82b9799a5bda7c96ffa5cafe8c Author: Kristian Høgsberg Date: Tue Mar 11 13:48:07 2008 -0400 Silence warnings about possible uninitialize use of dst_format. Whoa, gcc got a lot smarter about warnings. If iXXX_get_dest_format() doesn't support the picture format passed in it won't initialize the uint32_t pointed to by dst_format and return FALSE. What gcc now can detect is if dst_format is used without checking the return value, it might be used uninitialized. This patch makes sure we always check the return value before using dst_format. commit 31654f52b084ebb32f69854dfb1c342c9da07928 Merge: a03aca1 02c0ffb Author: Zhenyu Wang Date: Tue Mar 11 14:37:41 2008 +0800 Merge branch 'master' into xvmc Conflicts: man/intel.man src/i830_driver.c commit 02c0ffb1d8112a6fb03f9c8eae9ec1a1fca123ef Author: Jesse Barnes Date: Mon Mar 10 11:51:37 2008 -0700 Add pipe A force enable quirk for Dell Latitude D500 Fixes #14888. commit a1cd37f94cb8efbde84ba0a46839fb8735dd3595 Author: Dave Airlie Date: Mon Mar 10 14:43:24 2008 +1000 quirk: add DVO ivch quirk commit f2b9f387af7f618ad083b1ec8bee513ff2157d06 Author: Matthieu Herrb Date: Sat Mar 8 23:36:56 2008 +0100 Makefile.am: nuke RCS Id commit bf6d59ee98127ac6447848c4202c54d861a510d4 Author: Eric Anholt Date: Fri Mar 7 14:26:22 2008 -0800 Avoid setting reserved bits in SDVOx on everything 965-class. This will likely fix SDVO outputs on G35 hardware. commit 5915c75422c5277d530e7f8ecbdfe94654706efd Author: Jesse Barnes Date: Fri Mar 7 13:18:09 2008 -0800 Enable hardware overlay on 965 Based on Maxim Levitsky's patch for the same. Many thanks to Maxim for discovering that this features still works in recent hardware! commit 13ae79b415d944891cfdc2cd2d5397fec36a4e8e Author: Dave Airlie Date: Fri Mar 7 08:04:37 2008 +1000 fixup pciaccess version detect commit be08b61540e1edbb7673f0b6add0f1e32b136b3c Author: Jesse Barnes Date: Thu Mar 6 13:55:13 2008 -0800 Implement new get_property feature for backlight control This should keep the backlight value reported by xrandr --prop & xbacklight consistent with changes by other software in the system (like the hotkey driver or kernel backlight driver). commit 502e41f6dff91393cfeaf08c9ba312781d1b7267 Author: Zhenyu Wang Date: Thu Mar 6 19:32:29 2008 +0800 remove extra dri checking And directRenderingDisabled already has config check result. commit ea58d66cf778c0477c30e5bfe3142b64e8b971d5 Author: Zhenyu Wang Date: Thu Mar 6 19:31:33 2008 +0800 No ring, no sync commit c823fb00a20f523f50a2779c36aa1c014ba01535 Author: Jesse Barnes Date: Thu Mar 6 19:29:22 2008 +0800 Pull chipset detection out of PreInit into its own function commit 6b16cee0786b636dc849826193b26e898832a73c Author: Jesse Barnes Date: Thu Mar 6 19:27:42 2008 +0800 Account for FBC LL cache in memory reservation When we reserve memory (memory that won't be managed by the kernel) make sure to reserve space for the FBC LL cache as well. commit 6f04d02471f1cebc76eab637a9a78c4004fd3cc9 Author: Jesse Barnes Date: Thu Mar 6 19:27:07 2008 +0800 Move MTRR fixup to its own routine commit 750beb9232b51223d8e650878ce8dad071f0d0b3 Author: Jesse Barnes Date: Thu Mar 6 19:24:47 2008 +0800 Refactor memory allocation into a separate function This simplifies the memory allocation code and fixes a number of bugs. Prior to this change, some flags may have been set after memory allocation occurred, meaning they had no effect. It should also make the allocation logic clearer. commit d466b25d09bbe85abe84d9d7714ffb65fafe7593 Author: Zhenyu Wang Date: Wed Mar 5 13:38:58 2008 +0800 Really print adjusted_mode for mode debug commit d767eabe534be7f7bdcbf00614a8901124ac911f Author: Hong Liu Date: Thu Feb 28 13:51:18 2008 +0800 Fix DPLL programming in CRTC mode set The following patch fixes the display problem on internal development machines. The code in commit 3c22ed633be2ac96eea7bc533839e956f1f31b84 (Jesse's force pipe A enable patch) broke DPLL programming. Moving the DPLL set back up in the function solves the problem. Fix for internal bug #309. commit 4936e097028b91f4bdc2d9101dc49f6fe586e718 Author: Jesse Barnes Date: Tue Mar 4 08:52:40 2008 -0800 Change FBC idle mode back to default Use the "immediate idle" mode for FBC. Rather than waiting for the current compression pass to finish before signalling to the CPU that it's idle, this mode should stop any current compression pass and tell the CPU that the GPU is idle right away. Seems to fix #13326. commit 02a3b80ee8341968dbce5b7cb871b187db5ff805 Author: Zhenyu Wang Date: Fri Feb 29 11:13:34 2008 +0800 Bug 14591: comment out MI_DISPLAY_POWER_DOWN as for 965GM only commit ea7c7d088e0a3c3b7ab2bdd0475aa13379764bf4 Author: Peng Li Date: Wed Feb 27 11:56:55 2008 +0800 HP Compaq 6730s has no TV output commit fdc25cb0ca7d479051cee66a032f9ee1b68db98a Author: Mark Kettenis Date: Tue Feb 26 17:26:50 2008 -0800 Bug #14649: Fix build if pciaccess headers are not with other xorg headers. We failed to include the pciaccess header flags in our build. Even if the server was to include those flags for us, it would leave us broken for the tools-only case, and it's easier to just put the flags everywhere so we avoid future copy'n'paste mistakes. commit e55d86f654aa9dd4951b441636c8ca4b37786daf Author: Eric Anholt Date: Tue Feb 26 12:26:23 2008 -0800 Fix SDVO I2C access on Mac Mini in EFI mode. The GMBUS was being left in SDVO pin access mode, which blocked our bit-banging access to those pins. Thanks to Peter Jones for quick debugging turnaround in getting this fixed. commit 6935c732c351585f31e2094c4201a00e3d5529b3 Author: Julien Cristau Date: Mon Feb 25 14:47:02 2008 +0100 Look for src/{modes,parser} in $srcdir, not in objdir This makes the detection work in case configure is run from another directory. commit 66cdccb021a4748b2af41e415c36ed58ca808df6 Author: Zhenyu Wang Date: Mon Feb 25 14:29:09 2008 +0800 Fix overlay rotation Track current covered crtc rotation in port's private, and fix overlay rotation transform to be against crtc mode coordinate. commit 975d7833bc844485c7043538336bddcba7f7f90e Author: Jesse Barnes Date: Fri Feb 22 11:41:25 2008 -0800 Add quirk for DVO channel selection Some machines want DVOA, some DVOB. We can use this quirk to differentiate them until we come up with a better solution. Patch from Hong Liu. Fixes #13722. commit 33802b6ee3c298bae4d2b6f464ee5cc730a58506 Author: Jesse Barnes Date: Thu Feb 21 08:43:34 2008 -0800 Add pipe A force enable quirk for ThinkPad X40 ThinkPad X40s need this quirk or they'll crash frequently. Fixes #14521. commit 444984a578aae92ff55c06da897ea1d23679e706 Author: Jesse Barnes Date: Thu Feb 21 08:37:42 2008 -0800 Remove side effects from VGA debug code The VGA register dumping code was leaving ARX in data mode rather than index mode, which could cause problems for later software accessing AR* registers. Fix it to make sure it's in index mode when we're done. Fixes #14434. commit 28049540d8a9f79401fcfeb90784f5a528e7b34f Author: Jesse Barnes Date: Wed Feb 20 13:08:27 2008 -0800 Fix DSP*CNTR restoration When checking which pipe a given plane was associated with, we weren't properly masking the pipe selection bits. Fixes #14481 and should allow the driver to work with vesafb again. commit 293120bfc40a5b828567551954d8312639e73578 Author: Zhenyu Wang Date: Tue Feb 19 21:13:45 2008 +0800 Don't leak memory if no DMI info is provided by kernel commit cdc5cb7c3d24f36a30fb931b43ec6df8294709e2 Author: Zhenyu Wang Date: Tue Feb 19 08:51:53 2008 +0800 Fix last 8XX clock's p2 value commit My fault to change Hong's origin patch reversely. commit be0591880f20bdcbae159d5ab47113b6cc6fbffe Author: Zhenyu Wang Date: Mon Feb 18 14:35:06 2008 +0800 Fix Lenovo X60 TV quirk Z61 has same subsys ids with X60, but does have one S-video TV out. Use DMI info instead to quirk TV on X60, X60s. commit d5e564c5f24b0d9320bc4908693d37e7e6d68bc6 Author: Zhenyu Wang Date: Mon Feb 18 14:32:32 2008 +0800 Add DMI info for i830 quirks Linux kernel has CONFIG_DMIID to export dmi info through sysfs. commit 4a42b01f5ee5a673716d6959dfe0e693b037eb48 Author: Keith Packard Date: Sat Feb 16 18:16:12 2008 -0800 Decode DSPCLK_GATE, dump PIPE*STAT, MI_MODE, MI_DISPLAY_POWER_DOWN, MI_ARB_STATE, MI_RDRET_STATE, ECOSKPD commit 663313559873208c00c3d3748b99d837949ecdf2 Author: Zhenyu Wang Date: Fri Feb 15 13:20:22 2008 +0800 Bug #14440: fix stolen mem size mask on i830M For i830M stolen mem size mask should always be 0x70. Use 0xF0 for later chipsets should be ok, so behavior is identical to kernel agp. commit 8c8ca16120927fc2d3cf8bcd5212a4227a280708 Author: Eric Anholt Date: Fri Feb 8 11:34:04 2008 -0800 Add a little program to output the stepping information for several chipsets. commit f5896dcd1f5a964dce024ab8887ac8a79f7093a1 Author: Jesse Barnes Date: Thu Feb 7 11:56:28 2008 -0800 Add CACHE_MODE_0 register to dump output commit d59eaa8b1e6eeb9775c9d21c7a5fd28f25b2bc3a Author: Jesse Barnes Date: Wed Feb 6 09:35:43 2008 -0800 Only disable FBC if registers are available The call to disable FBC should only occur if the FBC feature is actually present or we may end up hanging on a read from a non-existent register. commit 83a071e2c01eb71462273f25d86d77159ac51114 Author: Jesse Barnes Date: Tue Feb 5 17:09:19 2008 -0800 Fix build warnings on 64 bit Use PRIx32 for printing CARD32 types, and PRIx64 for portably printing uint64_t types. Requires the addition of a new include, inttypes.h, to work. Hope C99 is ok with everybody... commit abf4b0cafa038e97bd7a6e5ba107b822117b9ccc Author: Jesse Barnes Date: Tue Feb 5 13:30:51 2008 -0800 Only enable FBC if one pipe is active Some chips can't support FBC if multiple pipes are active. So if more than one pipe is on or we're going from one->two pipes enabled, make sure FBC is disabled. Intended to fix 13418, 13326, 13152. commit f79a585d57c61fbe9af5684a336918e1fe3e8416 Author: liuhong Date: Tue Feb 5 11:04:34 2008 +0800 xf86-video-intel: fix ivch using wrong dvo register We use the wrong DVO output register for ivch which is used for DVO LVDS. Vbios enables DVOB and driver never touches it, so everything seems working correctly now. commit cabed67b6a0e4dd408eeb8d2083269100a65933c Author: Jesse Barnes Date: Tue Feb 5 11:36:24 2008 -0800 Program FBC fence offset register Just a partial fix for some of the FBC issues people have been seeing. The other half is to disable FBC if both pipes are running. commit 332587081fb05d284864aabe608dd9988eee3d44 Author: Zhenyu Wang Date: Tue Feb 5 10:34:55 2008 +0800 Fix last commit on i8xx debug p2 value commit 2d18a299cf39f2ecf9bd3ba8407f892566b2edcc Author: Hong Liu Date: Tue Feb 5 09:10:48 2008 +0800 Fix PLL reference clk debug dump commit 9536515d7717969795edc1b80d6e6a36820dd575 Author: Hong Liu Date: Tue Feb 5 09:06:14 2008 +0800 Allow non-strict free order for bo_list commit 11de831cb706025c003e98fef6d666d273192bcd Author: Hong Liu Date: Tue Feb 5 08:58:44 2008 +0800 Bug 10584: Mac Mini EDID data assigned to TMDS output EDID data for TMDS output got from crt should be applied to TMDS output on mac mini. commit 95327571134d8ebb9d936d80b70c2f642393aa9f Author: Hong Liu Date: Mon Feb 4 17:14:23 2008 +0800 Bug 10773: fix i8xx pll p2 value in i830_crtc_clock_get() Also fix debug dump, slightly modified to use macro instead. commit be1b568bc2379856c3eaaed365002512bebc218c Author: Paulo Cesar Pereira de Andrade Date: Thu Jan 31 22:47:18 2008 -0200 Make sure symbols used by other modules are public. These symbols must be explicitly exported, otherwise if compiled with hidden symbols, LoaderSymbol (and dlopen) will fail to find them. commit cc1031a834b81aa6f835c282ad86deddabe229a0 Author: Mark Kettenis Date: Thu Jan 31 21:02:48 2008 -0800 Bug #14246: Fix biuld on OpenBSD. commit b1e1c7cd993ab4f1935d750eb8852e74be757e53 Author: Eric Anholt Date: Wed Jan 30 23:01:04 2008 -0800 Add detail on different units to intel_idle. commit aa3ac79759581b5eb05293a8cbcf89eb5b76712c Author: Zhenyu Wang Date: Thu Jan 31 18:26:46 2008 +0800 Don't crash if SW cursor In case of device option or hw cursor allocation fails. commit 2e43bec8731ba1b172f7a0bf867bbb5c1adbda2d Author: Jesse Barnes Date: Wed Jan 30 18:59:12 2008 +0800 Frame buffer compression support on new chipset commit bf629466a46c4037ec7b7cc5ee16be947618bd68 Author: Zhenyu Wang Date: Wed Jan 30 18:55:20 2008 +0800 hardware status page initialization rework Order hardware status page setup more reasonable after all memory bound, in case new chipset requires non-stolen page and that could be bound then. Also clean up drm irq handler install function, and put first install in starting stage later than status page setup, so we won't make device cry for uninitialized status page. commit 04032dad28baab80131edbe8fe58aade8149bb71 Author: Zhenyu Wang Date: Wed Jan 30 18:52:32 2008 +0800 Wrap up chipsets which needs graphics address for status page Also add support on new chipset. commit b9c3fa79d9fd48c34536f92e6c1c70fe3a619410 Author: Keith Packard Date: Fri Jan 11 15:51:00 2008 -0800 Add intel_idle to measure GPU load ratio. Samples GPU idle register to detect when the GPU is busy; prints out the number of busy samples per 100 total samples once per second. commit 5d9e67aec3ce458d22b7febd3819542abb360534 Author: Erik Andren Date: Fri Jan 25 11:06:01 2008 +0800 Clevo M720R has no TV-out commit 6bf53eb48f40ad0c8ea9679ee634447410821b4f Author: Zhenyu Wang Date: Fri Jan 25 16:59:13 2008 +0800 Set vtSema before EnterVT Which was missing in our ScreenInit and initial EnterVT. This not only causes failure in initial rotation with TTM, as we won't bind in rotate_mem alloc in this case, and hide another bug that we call randr12 function in I830LoadPalete before we call xf86RandR12Init. commit 94a18fa1f8141837bdab32e545da7a7aed1cc396 Author: Julien Cristau Date: Thu Jan 24 15:24:40 2008 +0100 Don't build reg_dumper if we don't have pciaccess 0.10.0 The pci_device_map_range() function was added in libpciaccess 0.10.0, and is used by the reg_dumper tool. Don't try to build it if we have an older libpciaccess. Also make sure that util-macros >= 1.1.3 is available when running autoconf, because it's required for the PACKAGE_VERSION_* macros. commit d341e41c863f1212bf2c6b84782a7e472b6612a1 Author: Zhenyu Wang Date: Thu Jan 24 10:40:18 2008 +0800 Clear shadow memory after allocation commit 88a1041c5361964d37107c22d77feaa1b2160656 Author: Zhenyu Wang Date: Thu Jan 24 10:36:06 2008 +0800 Fix i830 block handler wrap which was observed in rotation crash with stack overflow. commit db0a7c569e383436a2725e1e74f35fb426da1196 Author: Andreas Stawinoga Date: Thu Jan 24 08:51:09 2008 +0800 Samsung Q45 has no TV output commit f54bba7600647c76fe3bd2d0133edc06dbea6a05 Author: Zhenyu Wang Date: Tue Jan 22 00:48:23 2008 +0800 Add missing i830M and 845G pci ids info commit 633dbd28218fd8a746c1735fd44a14f35bd07faa Author: Julien Cristau Date: Thu Jan 17 22:34:21 2008 +0800 Bug 14032: i810, set default depth to 16 X fails to start without an xorg.conf on i810, because the default depth set by the server (32) is not supported by the driver. I think the driver should default to 16 bits (it supports 24, but when I tested it that didn't seem to work very well, and dri is disabled at 24 bits depth anyway) instead of letting the server use 32 and failing later on. commit 4fec18e07f2b23306e885d63c74661ae0e4b04d9 Author: Jesse Barnes Date: Fri Jan 11 10:28:59 2008 -0800 Turn on backlight when LVDS panel is powered up In some configurations, the LVDS may be off at startup along with the backlight. So when turning the LVDS on for the first time, we may also need to set the backlight to a non-zero value. So try to use the saved value if possible, but if it's zero, make the backlight full brightness when turning on the LVDS. Note that this is slightly sub-optimal for configurations where zero is a valid backlight brightness. Fixes fdo bz #13958. commit b5cf8faf8102e6313c23de87a680cc327ac95478 Author: Jesse Barnes Date: Thu Jan 10 14:25:18 2008 -0800 Remove unnecessary quirk code in CRT probing Since we return early in the CRTC DPMS code, keeping the pipes & clocks enabled, this code is no longer necessary. As an added bonus, VGA connections should be reported correctly even when the pipe A quirk is active. Updates fix for fdo bz #11432. commit 3c22ed633be2ac96eea7bc533839e956f1f31b84 Author: Jesse Barnes Date: Wed Jan 9 09:47:38 2008 -0800 Add pipe A force enable quirk On some platforms, the firmware may read & write GPU registers on lid close, suspend/resume time or during various SMM events. If one of the graphics pipes is disabled at that time, the GPU may hang due to the programming dependencies of the various registers. This patch adds a quirk to force the driver to keep pipe A enabled if necessary, through user configuration in xorg.conf or via a platform specific quirk. Leaving the pipe enabled comes at a power cost however, so the quirk should only be enabled when strictly necessary. Fixes https://bugs.freedesktop.org/show_bug.cgi?id=11432. commit d3b41bcf231a6ffd289c128c06aa619f554a776c Author: Zhenyu Wang Date: Wed Jan 9 21:28:25 2008 +0800 Disable frame buffer compression on new chipset now. commit 2f0c0427dc4113bac64cda07c6365df1f0637697 Author: Zhenyu Wang Date: Wed Jan 9 21:27:42 2008 +0800 Update PIPELINE_SELECT instruction and surface state format for new chipset commit 139e9affdd49dcbe08e260a0a2bd001de16566e6 Author: Zhenyu Wang Date: Wed Jan 9 21:26:57 2008 +0800 GTT access change for new integrated graphics device commit 4f288ff2d1e7d28bf49167f073fbce21492a7c32 Author: Zhenyu Wang Date: Wed Jan 9 21:25:47 2008 +0800 Add new integrated graphics chipset ids commit f089942689df622b1af00569eade4883a84eb7a3 Author: Zhenyu Wang Date: Thu Jan 3 11:28:18 2008 +0800 Change origin i965G_1 to chipset market name G35. commit 71143200ed66cb9dcc1f9edf98128be80b05619c Author: Jesse Barnes Date: Fri Jan 4 15:01:25 2008 -0800 Don't modify low bit of BLC_PWM_CTL when using combo backlight control On some platforms, the low bit of BLC_PWM_CTL is wired as a 'max brightness' flag, rather than a regular part of the backlight duty cycle. So when in the combo mode, divide the total number of backlight levels available by two (tossing one bit) and adjust the programming in the set_brightness routine. Note that platforms with this behavior may need quirks added so that they work by default. commit b434bf4029d1a405216454629bcea717c779ae5a Author: hulifox008 Date: Fri Jan 4 02:46:54 2008 -0800 Fix copy'n'paste-o in unused airlied i2c code. commit f2ffc0f6e038357dda268363e52c11ada1d0b810 Author: Jesse Barnes Date: Thu Jan 3 11:16:15 2008 -0800 Fix compilation error when not using DRI This patch complements 88f8b688e2316ae4a1f7485f0010ce90de54783a which added uint64_t typed variables to avoid unsigned long overflows in 32-bit architectures but didn't include with the required definition. When XF86DRI and _XF86DRI_SERVER_ are defined this header gets indirectly included through "i830_dri.h", thanks to "i830_common.h" which masquerades this problem as released in 2.1.0 and that manifests with : In file included from i810_driver.c:88: i830.h:137: error: expected specifier-qualifier-list before 'uint64_t' i830.h:240: error: expected specifier-qualifier-list before 'uint64_t' Patch from Carlo Marcelo Arenas Belon commit a03aca101d4d51ed2279a99d1137ce3b7ef364e2 Author: Zhenyu Wang Date: Wed Jan 2 14:11:31 2008 +0800 xvmc: setup driver info for XvMCGetDRInfo request User don't need to have XvMCConfig file then, if ld.so can find libIntelXvMC.so in ld path. commit 0fd769b5882058e5ebf17ee1c1e35b4a3d46e6b0 Author: Zhenyu Wang Date: Wed Jan 2 13:57:23 2008 +0800 xvmc: add context id info in debug/trace commit 83d304c61ad5fdc58b0a9309dbd1e5a3f6cd9b01 Author: Nanhai Zou Date: Wed Jan 2 10:50:44 2008 +0800 TV: fix 576p refresh rate commit a2bb2afd24c95e056b801c85dc12daeac54b37f0 Author: Zhenyu Wang Date: Fri Dec 28 15:52:12 2007 +0800 xvmc: move drmOpen earlier remove wrong comments commit 96246c27cb836bae8ee02c46c68a1f2102efd14b Author: Joakim Date: Thu Dec 27 17:09:02 2007 +0800 Aopen Minipc 965GM LVDS quirk commit 666e01bd426a553f23eeb5040ecbc28b4abb87da Author: Zhenyu Wang Date: Wed Dec 26 09:51:09 2007 +0800 xvmc: add info in intel.man commit e75d0c20e3b02f649b62ac9da7f41bb43d0f3d80 Author: Zhenyu Wang Date: Wed Dec 26 09:37:34 2007 +0800 xvmc: make XvMC default on commit ed9ff6f30fff33fe2d60715f1b503047f8e1ef01 Author: Zhenyu Wang Date: Tue Dec 25 08:32:47 2007 +0800 xvmc: consolidate put surface interfaces Make surface list track like context list, and move generic PutSurface code out of i915. So MC driver just needs to fill specific the command params, and we can get one single type of intel xvmc command to issue, ddx driver's put_image_size got removed. Fix last commit with missing XvMCContext assignment. Fix priv_data free with apropriate X11 XFree. commit f0ced5edaeec3e7741828eab728417843326d56f Author: Zhenyu Wang Date: Mon Dec 24 10:53:57 2007 +0800 xvmc: store XvMCContext pointer instead of context_id in internal ctx list commit 216ad7398742f2f80b5477e89f827acb351b4d72 Author: Zhenyu Wang Date: Thu Dec 20 13:39:32 2007 +0800 xvmc: add disable xvmc build support Move xvmc source files under XVMC config commit a811ca30659e8c477027ea9c91fb9fc516fdabc7 Author: Zhenyu Wang Date: Thu Dec 20 13:08:07 2007 +0800 xvmc: handle surface create/destory in generic code commit 5f961f28310a1cd19967df2034b0452bf4446394 Author: Zhenyu Wang Date: Thu Dec 20 12:51:06 2007 +0800 xvmc: move dri context handling to generic code Use XvMCContext's context_id for dri context handling instead of driver private id. Remove unnecessary field for i915 private structs. commit 73827e78357ab3f30a21f159ecdfb26dd0ea889b Merge: 3cf8627 f69b48f Author: Zhenyu Wang Date: Wed Dec 19 22:47:02 2007 +0800 Merge branch 'master' into xvmc commit f69b48fe24ef94dac44b8123884ca71df675be4b Author: Jesse Barnes Date: Tue Dec 18 18:10:33 2007 -0800 Unconditionally restore pipe configuration This is a partial fix for #13196, which covers both leaving pipes disabled at server exit time and problems with restoring the pipe configuration on certain chipsets. It restores the pipe configuration unconditionally (previously we made sure the PLL was running and we weren't in VGA mode) but also adds some additional PLL settle time to the PLL register write paths. commit 4757a218d733af12d04674455fc6e1fad48a1cd0 Author: Michel Dänzer Date: Tue Dec 18 19:37:46 2007 +0100 Always set pPriv->buf to NULL after freeing the memory it pointed to. Fixes https://bugs.freedesktop.org/show_bug.cgi?id=13108 . commit 5cbb4110116f6d4187fd27660a78203caa1da46b Author: Adam Jackson Date: Thu Dec 13 15:10:43 2007 -0500 i830_sdvo_mode_valid: Fix return values to match what we actually check. commit e03bc385e924222d3d5f3c9a8d65e1ef63b858dc Author: Jesse Barnes Date: Tue Dec 11 13:28:01 2007 -0800 Add cscope files to .gitignore commit e720ae4476c3f986f623ce0f0ab9775b8b9b7e05 Author: Jesse Barnes Date: Mon Dec 10 13:00:14 2007 -0800 CRT hotplug detection improvements Patch from Hong Liu. Fixup CRT detection by making sure the pipe is enabled before CRT detection actually occurs. Fixes bugs Hong was seeing on G35 and other machines. commit d9df93578b74785c08ba860b4c9aa23b0c89c91c Author: Dave Airlie Date: Mon Dec 10 16:41:24 2007 +1000 ivch: fails on address mismatch as I seem to get this on my 865 system commit cd6f83519c69f72f146fea59afbd6a3542da783a Author: Zhenyu Wang Date: Mon Dec 10 05:49:58 2007 +0800 Fix tv quirk for Dell Latitude X1 which is actually 915GM, reported by Andreas Schildbach commit 13ec9c8141a9f794258869a04a6bab59dac5eefa Author: Zhenyu Wang Date: Sun Dec 9 00:52:13 2007 +0800 exa: fix rendering issue on some 855GM laptops This trys to initialize more 3d states on i830 like mesa code, which fixes exa rendering failure for some 855GM users. commit e2d9fd6425584119f28768d85c4b7d106ecf8742 Author: Jesse Barnes Date: Fri Dec 7 16:30:46 2007 -0800 Describe output properties in more detail Add descriptions for LVDS and TV output properties and also mention the EDID property a new output configuration section. commit 6f92200db39a790c95c1fac64baf0bd41a70e481 Author: Jesse Barnes Date: Wed Dec 5 08:10:17 2007 -0800 Fix backlight setting save/restore We need to save the current backlight value at LVDS init time, as well as when we change the DPMS setting. Also, since 0 is a valid backlight value, don't set the backlight value to maximum at startup if the value happens to be zero. These fixes should make the backlight user experience much more consistent and hopefully less frustrating. commit 4c1fa5235555c9ef3de1eebd006c4e91283362bf Author: Jesse Barnes Date: Tue Dec 4 18:27:24 2007 -0800 Add BCM_ to backlight control method enums Avoids polluting the global namespace with such generic terms. commit 30d74f1b1f30e751d2c47aa63a4a2043e71467dc Author: Jesse Barnes Date: Tue Dec 4 18:23:44 2007 -0800 Use "LEGACY" backlight method if backlight control is such This may regress the user experience a bit (fewer backlight levels) on machines where both the LBB and native registers work, but it's better that it work for everyone than work extra well for some and not at all for others. commit 0e554e6dd4da56be24d2f971ff4cd4d585936724 Author: Eric Anholt Date: Tue Dec 4 10:32:45 2007 -0800 Remove extra have_libpciaccess=no that broke tools build with old servers. commit 3cf862717688aa679b5e667cd041ddd6f5db0b58 Author: Zhenyu Wang Date: Thu Nov 29 23:12:16 2007 +0800 xvmc: add "XvMC" option, default is false. commit 6da4b7a80cdf730018435fe84485ecdad62bc860 Author: Zhenyu Wang Date: Thu Nov 29 22:56:13 2007 +0800 xvmc: move batch buffer init/fini code into generic commit 75ef3e669dac1259d282dcc8f54b197fc19f22b3 Author: Zhenyu Wang Date: Wed Nov 28 22:13:55 2007 +0800 Replace ALLOCATE_LOCAL/DEALLOCATE_LOCAL with xalloc/xfree commit df47b7ba57f3e7dc8bac1898679186d9d48639ac Merge: df718a7 7f9ceff Author: Zhenyu Wang Date: Tue Nov 27 01:13:43 2007 +0800 Merge branch 'master' into xvmc commit 7f9cefffbf44a2d07b51cc13aaf2d54fcd8f2f22 Author: Alan Hourihane Date: Fri Nov 16 15:56:08 2007 +0000 vendor is CARD8 commit df718a77c854c9110a81c2bc2a32a9aac85850f2 Author: Zhenyu Wang Date: Fri Nov 16 16:08:06 2007 +0800 xvmc: add xvmc decoder type string commit a03a758db0df615fbb44dae1ab6910bfeaeb54f9 Author: Zhenyu Wang Date: Fri Nov 16 15:49:10 2007 +0800 xvmc: move some i915 context destroy code to generic function commit 7bfbcb4dc7806ae7543f230373e67f35ba30668d Author: Zhenyu Wang Date: Fri Nov 16 15:43:04 2007 +0800 xvmc: clean batch buffer finish commit a6e8f4ef0481c66fda4f22fcc4bb0ef9c000f0cd Author: Zhenyu Wang Date: Fri Nov 16 13:51:44 2007 +0800 xvmc: remove i915 drm map type commit 4a2b0f340357c4ca58dc9586fad1337b83966362 Author: Jesse Barnes Date: Thu Nov 15 18:36:54 2007 -0800 Fix typo in 1920x1080 resolution entry commit 6c68fcc93917565712b6a9386df54d0204e4ea8f Author: Zhenyu Wang Date: Fri Nov 16 10:04:46 2007 +0800 xvmc: more i915 xvmc cleanups commit 04b60b6d3436fb57163c23acf335364f0ea2f79c Author: Jesse Barnes Date: Thu Nov 15 11:49:53 2007 -0800 Bump version to 2.2.0 commit 6eecef4fed8a21dfdabef42eb69fd150b96167b2 Author: Jesse Barnes Date: Thu Nov 15 10:10:59 2007 -0800 Adjust default TV out paramaters According to several users, a default brightness of 0 results in much better TV output. Improved control of these parameters will be provided by Randr1.3, which will standardize several output properties across various chips. commit 7552d80e367fe38bbc594fe94abd649917fe54d5 Author: Jesse Barnes Date: Thu Nov 15 10:07:34 2007 -0800 Add 1920x1080 mode to TV out Allows users to use 1080p modes on TV out, see bug #13034. commit e56c166ceb8c1ceddd5020e8de2d0b9d51f535a4 Author: Zhenyu Wang Date: Thu Nov 15 18:34:18 2007 +0800 remove unnecessary i830_reg.h includes i830_reg.h only contains 3d engine cmds for 8XX chips. commit 6f3de19159cd7a2d73ed212add909edb3aedce9c Author: Zhenyu Wang Date: Thu Nov 15 18:33:27 2007 +0800 Move fb compression reg definition into i810_reg.h where we put MMIO control reg in, and shared with intel_reg_dump program. commit 1ff97991a4954d508cc260dda82b7678010d13c9 Author: Zhenyu Wang Date: Thu Nov 15 17:01:31 2007 +0800 xvmc: intel_xvmc.c cleanups commit 1cfdf192882b856bfc93b735cfb9cb813cac3fdb Author: Zhenyu Wang Date: Thu Nov 15 16:38:21 2007 +0800 xvmc: remove unused SAREAPTR macro commit d9ea2fbdcce3c7c91428846eff43b65ac340f38d Author: Zhenyu Wang Date: Thu Nov 15 15:25:35 2007 +0800 xvmc: i915 rename files commit 06c0579df8d45f7f04b96956353d5f86cf59148f Author: Zhenyu Wang Date: Thu Nov 15 15:01:56 2007 +0800 xvmc: some code clean up commit 4983e3401390654098dbf99dcfb45bab060b068c Author: Zhenyu Wang Date: Thu Nov 15 13:56:24 2007 +0800 xvmc: move locking code into generic place commit bcd656c8dfe7309a5cf02ce1da8bc747ccd6e21c Author: Zhenyu Wang Date: Thu Nov 15 11:18:38 2007 +0800 xvmc: oops, should first map batch buffer before init commit 0464af581b0fcfc4d1c7fec8160d812262086177 Merge: 2256519 e56c166 Author: Zhenyu Wang Date: Thu Nov 15 11:05:44 2007 +0800 Merge branch 'master' into xvmc commit 22565198b360826ba10e7ca05d272c419054dc08 Merge: 619044b e64f492 Author: Zhenyu Wang Date: Thu Nov 15 18:53:22 2007 +0800 Merge branch 'master' into xvmc commit e64f4929690a57701241334fbca17e95b39ba3d4 Author: Jesse Barnes Date: Wed Nov 14 16:44:34 2007 -0800 Backlight fixes Open the "actual_brightness" file as read only, since we only read from it. Also set an initial backlight_duty_cycle at init time so we don't set the brightness to 0 at startup. commit 53e3693ef13f31f3fc33bcff7286ab2b03b2d430 Author: Jesse Barnes Date: Wed Nov 14 16:24:56 2007 -0800 Disable FBC by default on 965GM Several people have reported that they see frequent FBC related display corruption on 965GM, so disable it for now. Users wanting to enable it can use the driver option "Framebuffercompression" to override the default. commit b2726899bc6ebd108aa4a5dd66fe1d881bb778b3 Author: Jesse Barnes Date: Wed Nov 14 15:49:41 2007 -0800 Add more FBC regs Needed for the new debug code commit b7bc694c6b98dac17763426d905a22d3ae17a018 Author: Jesse Barnes Date: Wed Nov 14 15:46:57 2007 -0800 Fix kernel get_brightness function We need to look at "actual_brightness" rather than "brightness". The former contains the brightness value the kernel driver has actually set, while the latter is merely what the user requested. commit 9ee96ed02aacd953ab58288fddf0fa60f60b668e Author: Jesse Barnes Date: Wed Nov 14 15:44:56 2007 -0800 Correct FBC debug message To be consistent, it should say 'plane' rather than 'pipe'. commit 675b9471d84059496ca41d427a4f95d419f82be9 Author: Jesse Barnes Date: Wed Nov 14 15:17:05 2007 -0800 Add FBC registers to register dump output Just for completeness. commit a2a0821e74a61f53cc7f0c41ce629644ad712114 Author: Jesse Barnes Date: Wed Nov 14 11:30:20 2007 -0800 Don't stop ring before restoring hw state Some of the hw state restoration callbacks may want to use the ring for various things like stopping video playback, so leave the ring enabled until we get back from RestoreHWState. Also rename the functions so that their purpose is clearer and remove a couple of redundant lines. commit 619044b242d4abf605b42cfff538b63f445fe709 Author: Zhenyu Wang Date: Thu Nov 15 02:23:07 2007 +0800 xvmc: move batch buffer init/fini to intel_batchbuffer.c commit ae805ac379ccfdae9ee272896665e73f5bc7d2cf Author: Zhenyu Wang Date: Thu Nov 15 02:13:07 2007 +0800 xvmc: take out batch buffer handling code commit ad3bc0158d37e98fcbbe6a8e31413c142a260424 Author: Zhenyu Wang Date: Wed Nov 14 22:46:46 2007 +0800 Don't enable fbc with XAA or tiling is off. This slightly reworks my last fbc patch. We don't support tiled front buffer with XAA now, so also disable fbc on it. If tiled alloc failed, disable fbc too. commit 26194e19e1c80615697016e25640d4c8c244353f Author: Keith Packard Date: Tue Nov 13 22:26:40 2007 -0800 Review PLL spreadsheet and update register ranges. The PLL spreadsheet makes the precise register ranges allowed for each mode quite clear, and shows a few inaccuracies in the b-spec. In particular, the N register value may range from 1 to 6 instead of 3 to 8. This should close the gap we've seen in the reachable frequencies. commit 581509321e51becb8ee6ddcfaa15a2a713c9ef8e Author: Zhenyu Wang Date: Wed Nov 14 22:16:04 2007 +0800 Remove one redundant line. commit 8cea3c8cfa86600b5f2c3e1f21c19d4b2a320fd4 Author: Zhenyu Wang Date: Wed Nov 14 22:13:30 2007 +0800 Don't setup fb compression if fb is not tiling commit 539b669d6e23833a917a9b0d3a355d4283e0b76a Author: Jesse Barnes Date: Tue Nov 13 17:59:54 2007 -0800 Remove harmless error message This really isn't an error in general. If vblank pipe setup really fails for some reason, it'll be obvious enough when the user tries to use vblank events for something. Patch from Hong Liu. commit e8db8584ae4860aa991c916126335b287a972d5b Author: Zhenyu Wang Date: Wed Nov 14 00:11:20 2007 +0800 xvmc: attempt to move batch buffer code out of i915 private commit 11c0e0dbef42f1ae77a77d534870d3e9821fb017 Merge: c838f72 43fbc54 Author: Zhenyu Wang Date: Tue Nov 13 22:44:06 2007 +0800 Merge branch 'master' into xvmc commit 43fbc547786cf26514f95bce79fab58a66c291bf Author: Jesse Barnes Date: Mon Nov 12 12:23:41 2007 -0800 Improve backlight control This commit fixes backlight support for several platforms. Except on recent machines supporting the IGD OpRegion specification, backlight control is rather platform specific. In some cases, we can program the native backlight control regsiters directly without any trouble. On others, we need to use the legacy backlight control register. On still others, we need a combination of the two. And on some platforms, none of the above will work, so we go through the kernel backlight interface, which provides a platform specific driver for backlight control. commit 10988c5e6ec0f3c40d56bbf209b7976627cca706 Author: Jesse Barnes Date: Mon Nov 12 09:19:22 2007 -0800 Fix typo in PLL enable check Check against DPLL_A instead of DPLL_B before writing PIPEACONF. Thanks to Colin Guthrie for his sharp eyes. commit ab2055ebb20aa6de121fa377e488ce91913035ae Author: Jesse Barnes Date: Fri Nov 9 13:55:46 2007 -0800 Add 'ret' variable to i830_allocator_init Needed in the XF86DRI_MM case. commit 24280edbdf54e9559b26ea27c1b005eea73f51bc Author: Jesse Barnes Date: Fri Nov 9 13:53:58 2007 -0800 Fixup warnings Remove unused 'ret' variable, move pI830 under #ifdef XF86DRI_MM in i830_allocate_memory. commit 4fe507957bf826d81a71cd63af17c5547d1023a1 Author: Jesse Barnes Date: Fri Nov 9 13:51:11 2007 -0800 Remove unused 'palette_enable' variable commit 62404a9c9c0d98efb69ac96209dea17331e0489c Author: Jesse Barnes Date: Fri Nov 9 07:55:01 2007 -0800 Update man page to reflect EXA by default commit c838f7297f675ea9d82678ed37715955cb0a8a78 Merge: a9449ff 6f53568 Author: Zhenyu Wang Date: Fri Nov 9 19:37:07 2007 +0800 Merge branch 'xvmc' of git+ssh://zhen@people.freedesktop.org/~zhen/xf86-video-intel into xvmc commit a9449fff4be8e6b3cc82914376420e3a66b960f5 Merge: a470f6b 177924e Author: Zhenyu Wang Date: Fri Nov 9 19:36:25 2007 +0800 Merge branch 'master' into xvmc commit 177924e879564b7e9e70fd607141978bfd053fff Author: Jesse Barnes Date: Thu Nov 8 18:16:48 2007 -0800 Bump driver version to 2.1.99 in preparation for 2.2 release commit cb4e5796f0537ea5e0e646d473930c7b826c85d8 Author: Jesse Barnes Date: Thu Nov 8 18:13:47 2007 -0800 Default to EXA If EXA is compiled into the driver, default to using it for acceleration. Hopefully we can remove XAA entirely one day. commit 43a59ab26b09fcc24de1ed7bd770bb622f899ceb Author: Jesse Barnes Date: Thu Nov 8 16:52:41 2007 -0800 Clear current video crtc on DPMS off When calling the video DPMS off function, make sure we zero out the current crtc so that it will be properly re-set up next time video is turned on. Fix from Peter Clifton with changes by Keith Packard. commit e784e152a8e84b6e447b55a5c7019e7b47e17621 Author: Jesse Barnes Date: Thu Nov 8 16:37:53 2007 -0800 Fix overlay destination clamping The overlay width & height scaling clamp check was reversed. Fix that and update the comment. commit b8770f710729d616b3ac72544aa522161a78f819 Author: Jesse Barnes Date: Thu Nov 8 16:19:01 2007 -0800 Setup 3D state at EnterVT time In the absence of full suspend/resume support in the kernel, we have to save/restore state in Enter/LeaveVT. For 8xx chips, 3D state may be lost during suspend/resume, so re-emit the basic setup at EnterVT time. Patch from Peter Clifton. commit 5f92b4c2db9712496b829ee239468e3d14e27d2f Author: Eric Anholt Date: Thu Nov 8 13:31:51 2007 -0800 Bug #12059: Add ch7019 to the list of supported devices for ch7017. While I'm here, fix the chip description to be LVDS instead of TMDS in i2c device. commit eecd3ccedee6c4acf101591f7e60673660379e62 Author: Jesse Barnes Date: Thu Nov 8 09:31:08 2007 -0800 Check DPLL status before writing PIPEnCONF regs If the DPLL isn't enabled or is in VGA mode, writing the PIPEnCONF registers may cause a hang or crash. So ensure the DPLL is in the proper state before writing them. Another excellent fix from Peter Clifton. commit b434c1a437c407de88396b219560649c2dae82b2 Author: Jesse Barnes Date: Thu Nov 1 12:06:07 2007 -0700 Framebuffer compression fix: front buffer may not be at fence 0 Fix a long standing bug in the framebuffer compression code (thanks to Pierre Willenbrock!) that prevented FBC from working correctly if the front buffer was anywhere but fence register 0. commit 50d3693a6862028e50bc5ba8c788e7ea573c3eb8 Author: Zhenyu Wang Date: Thu Nov 1 22:22:27 2007 +0800 Fix allocation reset for really not free BO allocator commit ecd995d533d28b622afc71a20504d47c33ff5b8d Author: Alan Coopersmith Date: Tue Oct 30 18:20:49 2007 -0700 Fix builds without DRI commit 7c88b58a93fce9fda59b6344acb87af16336e287 Author: Alan Coopersmith Date: Tue Oct 30 18:20:15 2007 -0700 Clear compiler error: "void functions cannot return values" commit a470f6b2ca291e305979dd9223885a50dc0b334b Author: Zhenyu Wang Date: Fri Oct 26 22:40:47 2007 +0800 try to move some fields to xvmc driver structure Still a lot of things to move include batch buffer, lock, context, surface resource handling, etc. commit 6f53568cd2aeeec6bd8c1f1d9f9c5ecadc5c62f9 Author: Zhenyu Wang Date: Fri Oct 26 14:58:43 2007 +0800 Fix DRI context creation for XvMC commit d2c78f82c20f33fc9c22cab8a7ca161e57a34bf8 Author: Thomas Hellstrom Date: Wed Oct 24 19:34:12 2007 +0200 Adapt to DRM Lockfree and setStatus changes. commit 78aaec0ffc711742bf8ad77757ed8c15cc3f7a9f Author: Eric Anholt Date: Wed Oct 24 11:50:54 2007 -0700 Fix typo in my hand-application of rglowery's patch. commit 021265fbec9fd4ee31bdc9767c0c0453479ea22c Author: Rob Date: Wed Oct 24 11:06:53 2007 -0700 Fix a crash in TV mode handling by initializing the prev field of modes. commit ce9802edd25255f6c9c11e512b2ef905384b914d Author: Zhenyu Wang Date: Wed Oct 24 00:31:53 2007 +0800 fix xvmc return values Use values defined by xvmc lib spec commit 9c84543eb0fde4643eb7529eb6b08eacf57c12d0 Author: Zhenyu Wang Date: Tue Oct 23 22:16:55 2007 +0800 add missing license header for new files commit 87345e820d20f1a98216c52e156c2e18c15ffa44 Author: Jesse Barnes Date: Mon Oct 22 14:22:37 2007 -0700 intel_reg_dumper - dump VGA AR registers too Add a VGA AR dumping function so we can debug text mode problems too. commit 90e4eafdf59edd966ed626107c3909b24737da8d Author: Zhenyu Wang Date: Mon Oct 22 17:09:15 2007 +0800 revert xvmc memory allocation hacks commit 167c56bcf82f0a6d72151b202e6d56f6e0d627f6 Merge: fc9368c 78e251d Author: Zhenyu Wang Date: Tue Oct 23 00:12:22 2007 +0800 Merge branch 'master' into xvmc commit fc9368ce7dacd52ef222559d20188ddcf2a85df5 Author: Zhenyu Wang Date: Tue Oct 23 00:09:06 2007 +0800 split xvmc user lib chipset drivers commit 78e251db671e21bc859c9b505d391b70babee2dc Author: Eric Anholt Date: Fri Oct 19 15:04:10 2007 -0700 In the clock graph, draw the VCO as erasures in the lines representing clocks. This shows one of the reasons for the gaps: with the other settings, the VCO is too low inside the gap. However, it also points out another issue: we aren't using the high end of the VCO range due to some other limits being hit. commit 9f9b888525b274036d301d6e06351583d0415f9e Author: Eric Anholt Date: Thu Oct 18 11:25:24 2007 -0700 Warn in the log if we choose a PLL clock that's way out of line. commit 1f8bf110394cc1df66aae9acf5c818145ae19b52 Author: Eric Anholt Date: Thu Oct 18 11:17:38 2007 -0700 Add some nickle scripts for looking at PLL issues. While here, move similar nickle scripts under src/scripts/ commit 79636b8c776ae024518103c9fa137c8498c21d48 Author: Eric Anholt Date: Wed Oct 17 17:10:24 2007 -0700 Bump pciaccess version requirement for new API usage. commit c824c45e72908b5b3fa832fec91c7b47f305180c Author: Dave Airlie Date: Wed Oct 17 16:35:29 2007 +1000 intel: make block handler hook happen no matter what video is in use This shouldn't make a difference now, but for upcoming EXA/batch changes it will be more useful commit aa88d23e615e7f154cf850927a0a7bb9fbb0d2f9 Author: Rob Date: Wed Oct 17 09:19:19 2007 +0800 AOpen 965GM mini pc has no LVDS commit 38900f55e2b50ea383f78adb016e3a02c3bd17ad Author: Kyle McMartin Date: Tue Oct 16 16:38:28 2007 -0400 toshiba satellite u300 has no tv out commit 74ac5de14ebb77aeb39d698e9e8d188c9d9abd76 Author: Michel Dänzer Date: Tue Oct 16 15:08:26 2007 +0200 Adapt to libdrm buffer object API changes. commit d66e924680a3504aa1feb9c228145a95d1a16b1e Author: Eric Anholt Date: Mon Oct 15 11:31:51 2007 -0700 In update_front_offset(), set pScrn->fbOffset even at EnterVT during init. Failure to do so led to the memset() at EnterVT being done to an invalid offset. commit cc892d1ba1d87d08bdd04f857efc6b8f64620018 Merge: b52ce0b fd97edb Author: Zhenyu Wang Date: Mon Oct 15 22:09:14 2007 +0800 Merge branch 'xvmc' of git+ssh://zhen@people.freedesktop.org/~zhen/xf86-video-intel into xvmc commit c94cdfd6ddbc580523737f596e97b96a7ce100b9 Author: Jesse Barnes Date: Sat Oct 13 15:57:45 2007 -0700 Fix palette save/restore When restoring the palette, we weren't checking to make sure the associated pipe was enabled before writing the registers. In some configurations, this led to a driver crash. Add new routines to handle palette save/restore and verify that the pipes are on before we touch the registers (could easily be changed to enable/disable the pipes around the save/restore as well). commit e04333a6352040bc883655d606923c912d005981 Author: Eric Anholt Date: Fri Oct 12 12:42:45 2007 -0700 Fix failure in tiling setup on non-power-of-two allocations on pre-965. commit a1612b7728d4153499fe86b6713a13c8702cc7d9 Author: Eric Anholt Date: Fri Oct 12 12:42:06 2007 -0700 Use mprotect on unbound AGP memory to attempt to catch use while unbound. This doesn't help with the most common use-while-unbound cases, which are from the hardware side. commit 3af442ba52550a9d183e215d49cc12dac0cb9e4b Author: Eric Anholt Date: Thu Oct 11 11:56:06 2007 -0700 Don't double-free the memory manager allocation. commit 6c485ff5bd25e9aa6a3c1eb669843e6b969d94df Author: Eric Anholt Date: Fri Oct 5 15:52:56 2007 -0700 Update memory manager sizing for the current set of LIFETIME_FIXED bufffers. commit d47cf148776d74f9035863b23eefdc2b5893af08 Author: Eric Anholt Date: Fri Oct 5 15:45:51 2007 -0700 Move tiling fence register setup to bind time instead of allocate time. This allows us to allocate tiled buffers in buffer objects. In the process I removed the fence division that we had for tiled buffers on pre-965. If we resurrect that code, it should probably be managed by just dividing all the objects in roughly half and fencing those halves (to reduce the alignment requirement), instead of using giant fences until we run out of space and then trying to deal with scarce space on the last (or not) buffer. Halving our tiled objects would use 6/8 of our fence registers on that hardware. commit a6dc81a0864f9ab2f6fc1aa31002c0191f674ceb Author: Eric Anholt Date: Fri Oct 5 12:37:19 2007 -0700 Allow front/back/depth to move over the lifetime of the server. commit 132dce7565feeea1055899f8c1627766fe84c88c Author: Eric Anholt Date: Fri Oct 5 12:12:06 2007 -0700 Delay SAREA and mapping setup until EnterVT when using the memory manager. Otherwise, we would use uninitialized offsets in the early setup. commit 4ca3550fb7d488741f8dc1ba3c8722393277c3b8 Author: Eric Anholt Date: Thu Oct 4 17:02:15 2007 -0700 Rework DRI buffer mappings and sarea setup to allow for moving buffers. While this has been a desired feature for some time, to allow for reallocation of the front buffer, it was made more necessary by the desire to avoid requiring a NO_MOVE buffer type in TTM because buffer objects may not be left pinned over VT switch. This is a step towards making those buffers movable and resizable. commit f393a12d21eed668cf4771f022beded9a4c547c7 Author: Eric Anholt Date: Thu Oct 4 13:03:14 2007 -0700 Move drmMMLock to after we have unbound our (pinned) buffers. There are still issues due to the fact that we're allocating NEED_LIFETIME_FIXED memory as buffer objects, which we refuse to unpin because we have no way of pinning it back in the same location. commit 8fc4e3078210f726c7c375faa2f2fd8d05017c09 Author: Eric Anholt Date: Fri Oct 5 11:51:55 2007 -0700 Refuse to allocate LIFETIME_FIXED objects in buffer objects. We can't guarantee the offset will stay the same using the current DRM interface, but the correct solution is fixing our code to allow these objects to move. Breaks TTM mode of the DRI driver for now. commit 0de747f7d219a56434dad49a8a6d1d9d4c251ca7 Author: Eric Anholt Date: Thu Oct 11 10:36:01 2007 -0700 In i830_allocate_memory_bo, bind if we control the VT, not on lifetime-fixed. commit 0040bb95445e800af80ca9fa1b92f5db33f1b4ac Author: Eric Anholt Date: Wed Oct 10 16:48:12 2007 -0700 Fix potential use-after-free in XV overlay code on video stop. commit 56ae767180da4a8bd4b7a4c35b92e019ab67d753 Author: Zhenyu Wang Date: Thu Oct 11 10:09:29 2007 +0800 Revert "Fix G33 GTT stolen mem range" This reverts commit 2a8592f2ebcba86b1127aa889155d58a3dc186ca. This causes compat issue between ddx and kernel apggart version. Revert it as it shouldn't hurt normal people's default GTT size, but if you change it to 1MB in bios, it's supposed to be broken now. commit 5faee02e97b7543b41bb1f8e3751558c48802833 Author: Zhenyu Wang Date: Wed Oct 10 13:55:30 2007 +0800 EXA: fix tiled dest rendering on i8XX chips commit 5bbe9f09fda1238da04e6b584813d40fc646f49d Author: Keith Packard Date: Wed Oct 3 00:18:25 2007 -0700 Include xf86mm.h if using XF86DRI_MM This header file doesn't appear to be getting automatically included for some reason. commit 41dacf97ec1d4ba6f54057f4fcc7bb76033a454e Author: Keith Packard Date: Wed Oct 3 00:17:41 2007 -0700 Quirk for Samsung Q35 which has no TV output. commit b52ce0bd178c268d20334e2be8fad46037e47499 Author: Zhenyu Wang Date: Mon Oct 8 22:56:44 2007 +0800 use field name to initialize i915 xvmc driver commit e0fb228401b10ea4281eee48497bf1d584999705 Merge: 84d7b2b 9e770bf Author: Zhenyu Wang Date: Mon Oct 8 22:45:11 2007 +0800 Merge branch 'master' into xvmc Conflicts: src/i830_driver.c commit fd97edb62463e50a1860290d54684a68d5490bdc Author: Zhenyu Wang Date: Mon Oct 8 16:13:26 2007 +0800 move some definition out of i915 specific header commit 3a2b5222341b37322b788e329c2b9006dc14ecfd Author: Zhenyu Wang Date: Mon Oct 8 15:46:11 2007 +0800 Temp work around ttm allocation for xvmc, which should be fixed later. commit 9e770bf69edaa8944048049e11266dbe1ef145e5 Author: Eric Anholt Date: Thu Oct 4 14:31:37 2007 -0700 Replace setting of LIFETIME_FIXED on cursors with just updating the offsets. commit c9d6e90c2523096c45d330552f471e6bf1752704 Author: Eric Anholt Date: Thu Oct 4 14:27:07 2007 -0700 Don't set overlay registers LIFETIME_FIXED. It always uses the current offset. commit 1cc6bec4628a4ca04a68919a84148a1e6312072f Author: Eric Anholt Date: Thu Oct 4 14:23:48 2007 -0700 Revert setting of exa_965_state as LIFETIME_FIXED. The current offset is used every time in render setup. commit dfa22df8ce7713b491c4766ce6ce652db91b5f90 Author: Eric Anholt Date: Fri Oct 5 12:53:07 2007 -0700 Don't leak buffer object allocations in i830_reset_allocations(). commit 7db9d706a1bc9da7afc4823cee92b23e96ae55b8 Author: Eric Anholt Date: Fri Oct 5 12:40:48 2007 -0700 Warning cleanup. commit bf33c4e2d91172efd08690fadda7b3743dd3de62 Author: Eric Anholt Date: Fri Oct 5 12:40:37 2007 -0700 Fix formatting of error message. commit 04e936935f0b0045600241424f1d04a6721a2432 Author: Eric Anholt Date: Mon Oct 1 17:29:35 2007 -0700 Bring the CH7017 driver closer to spec. This is also closer to what my hardware is programmed with, except for some very confusing off-by-one bugs in an unexpected direction. commit 263d48ad93a0fb7100729d0c6e0443797c20c80a Author: Eric Anholt Date: Mon Oct 1 16:23:30 2007 -0700 Fix probing of the sample CH7017 device I found by allowing GPIO overrides. It may be that the LVDS chips need to be DVOA and GPIOB only on mobility devices with them, and DVOC/GPIOE on non-mobility like this 845. But until more examples are found, just make this one device probe. commit 219354af212c7b68c20df689692c55331e36a705 Author: Eric Anholt Date: Mon Oct 1 16:09:13 2007 -0700 Restore building of pciaccess-based tools even if the server doesn't use it. commit 1b83e5e589105d05c1b78032fa47eecee478cb99 Author: Eric Anholt Date: Mon Oct 1 14:32:28 2007 -0700 Remove more DVO probing noise. commit 29d3188ee9676f697f6fcb859c7f680f5df30318 Author: Eric Anholt Date: Mon Oct 1 13:45:09 2007 -0700 Suppress I2C failure error messages during DVO device detection. Otherwise, errors would be printed for every chip probed which wasn't present on the system. commit 6dd8228a386334d05836353672aae9f8c456e435 Author: Eric Anholt Date: Mon Oct 1 12:07:22 2007 -0700 Enable CH7017/7018 DVO driver by default. On hardware containing one of these chips (generally single pipe with an LVDS connected), it's probably better to try to enable it and fail in some cases than to not probe the output at all. commit 994ee6721d9f1650ad37e21198b86848352b73af Author: Eric Anholt Date: Mon Oct 1 12:01:16 2007 -0700 Disable TV/VGA output on the CH7017/7018 devices. We don't support those outputs on this chip yet, so turn them off if the BIOS had set them on. commit 0e5574e842eeaac75271ccd50027d3e54c57722d Author: Regis Prevot Date: Sat Sep 29 16:21:08 2007 +0200 Ignore detection of TV output on Panasonic CF-Y7 commit 84d7b2b955107264b67526a04028a8b79d8804e3 Author: Zhenyu Wang Date: Sat Sep 29 19:41:15 2007 +0800 rename xvmc init function names commit 1845e7c6e9caf7a9ad7f46a8fbfa321ddec69a1e Author: Jesse Barnes Date: Fri Sep 28 10:12:11 2007 -0700 Go back to using old drm_i915_flip_t field name This field shouldn't have been renamed in the first place. Go back to using the old name so that the tree is backward and forward compatible again. commit f3d92ab474de11babe507b0e3c15aca146b6cb66 Author: Kyle McMartin Date: Fri Sep 28 10:17:55 2007 -0400 Dell XPS 1330 has no TV out commit bbf69c7446aa9748dacecccfe0b63d803f865b00 Author: Zhenyu Wang Date: Thu Sep 27 17:37:19 2007 +0800 Pin cursor, overlay(no physical) and exa state buffers And cursor mem counting for dri mem manager is not relate to overlay. commit 64188246bb25e95a916ebb8a50a09298e56d91d9 Author: Zhenyu Wang Date: Wed Sep 26 15:48:19 2007 +0800 clean up old i915 xvmc function declare commit 320b7f08f457455f1d16b4c36d149bdc3f6ceead Author: Zhenyu Wang Date: Wed Sep 26 15:02:33 2007 +0800 Revert to origin xvmc setup steps We have to grab xvadaptor's PutImage earlier than xv init, otherwise no chance to do that.. commit 1e0ff38f44e3b5a8b4a5d509c7faa266270d522c Author: Hong Liu Date: Thu Sep 20 11:28:52 2007 +0800 Fix pixmap offset Use consistent interface for counting pixmap offset. commit 85e4bd109464a4d90856e17bc960872eeaebc2ad Author: Zhenyu Wang Date: Wed Sep 26 14:32:46 2007 +0800 Log xvmc driver name commit e58b76a24a29e0a502650807de81e650e5dc184d Author: Zhenyu Wang Date: Wed Sep 26 14:40:04 2007 +0800 clean up err report Still need to clean up return value for XvMC API commit b10a9668dc50d10489b2ae531d89c8ab7c408170 Author: Zhenyu Wang Date: Wed Sep 26 14:01:57 2007 +0800 fix i915 xvmc devPrivate reference commit f12eaaf385e7a0b85524588a5d865d1cc3f1a0cf Author: Zhenyu Wang Date: Wed Sep 26 13:49:55 2007 +0800 comment out i965_xvmc_driver for now commit 513e8a2f8abde1b11b8058e94a650796fd51779b Author: Zhenyu Wang Date: Wed Sep 26 13:45:42 2007 +0800 Create xvmc driver interface convert i915 to new xvmc driver interface commit ca67fa767dc762dac369e84b27a7ef15673d527c Author: Eric Anholt Date: Mon Sep 24 11:12:22 2007 -0700 Remove logic for supporting i915tex_dri.so vs. i915_dri.so. There can be only one. commit 130f79613bfe6a8cfa7f431c8cce06cbb93cc91a Merge: b73235f 1260316 Author: Eric Anholt Date: Fri Sep 21 17:14:44 2007 -0700 Merge branch 'buffer-objects' This branch changes i830_memory.c's allocator to use TTM when available to allocate memory, which also allows TTM to control almost the entire aperture. As a result, our front/back/depth buffers are created as real buffer objects, which may be used by the DRI driver instead of the fake buffer type in TTM. The updated DRM with bo_set_pin ioctl is required, to allow us to pin and unpin our buffers as needed. commit 126031655c3c1db9881ad17d712461533eb59632 Merge: c343835 bda3c69 Author: Eric Anholt Date: Thu Sep 20 08:57:58 2007 -0700 Merge remote branch 'origin/master' into buffer-objects Conflicts: src/i830.h src/i830_driver.c commit b73235f40497cfb10792ba191d4f6eac3a5df009 Author: Hong Liu Date: Thu Sep 20 11:28:52 2007 +0800 Fix pixmap offset Use consistent interface for counting pixmap offset. commit 798448e4641acf241b7cbae0d6c243ae383da9f4 Author: Zhenyu Wang Date: Wed Sep 19 13:29:31 2007 +0800 Add media state definition on i965 This adds VFE state and interface descriptor struct define. commit 36b81ba20d20e7b235cb701c5135cbf3f3eedb12 Merge: 02ad9ce bda3c69 Author: Zhenyu Wang Date: Wed Sep 19 11:44:41 2007 +0800 Merge branch 'master' into xvmc-i915 Conflicts: src/i830.h src/i830_memory.c commit bda3c69a7a044869ea63514b97f8aa9260d6b5a7 Author: Dmitry Babrovich Date: Tue Sep 18 09:34:34 2007 -0700 Bug #12457: Let the ch7xxx driver probe the 7301 Looking at the 7304, this chipset likely requires slightly different programming. However, this change allows for at least some functionality. commit 445a59d262b93bd9d2c16cee9934ce77a103776b Author: Jesse Barnes Date: Wed Sep 12 09:32:41 2007 -0700 Remove unused plane->pipe mappings from SAREA private Turns out we can get away without this, so remove it, fix a crash, and only swap planes/pipes if the DRM can support it. commit d02336290bea30de3c390b8121046c38fd6b0f62 Author: Jesse Barnes Date: Tue Sep 11 04:28:14 2007 -0700 Fix plane/pipe mapping compat code Move plane->pipe mapping adjustment to ScreenInit so we can check against the DRM driver version accurately. commit d9f2b3c0d2d08a4b7fad865dab7deb6224b57999 Author: Jesse Barnes Date: Mon Sep 10 23:49:34 2007 -0700 Fix crash in ScreenInit Use pScreen directly when getting at the SAREA private, since pScrn->pScreen may not be initialized yet. commit 4c7542ef43a5267e470ca1608a2ae57abf9783ec Author: Jesse Barnes Date: Mon Sep 10 23:30:50 2007 -0700 Only swap planes and pipes if DRM supports it We want to associate plane A with pipe B on pre-965 mobile chips, since that's the only way to get framebuffer compression on the builtin LVDS on those platforms. However, if we do this swapping and DRM isn't aware of it, we may end up requesting vblank events for the wrong pipe, or setting up SAREA buffer swap state incorrectly. This mod checks whether DRM supports the new plane->pipe swapping behavior, and only enables the swapping if so. This should fix the bugs Lukas found and debugged. Reviewed by Michel Danzer. commit 286f5df0b62f571cbb4dbf120679d3af029b8775 Author: Keith Packard Date: Thu Sep 6 14:31:55 2007 -0700 Switch to pci_device_map_range/pci_device_unmap_range APIs. With the libpciaccess change that added these new APIs, use them for all mapping. commit 2a8592f2ebcba86b1127aa889155d58a3dc186ca Author: Zhenyu Wang Date: Wed Sep 5 14:52:56 2007 +0800 Fix G33 GTT stolen mem range G33 GTT table lives in seperate stolen mem with graphics data stolen mem. commit 7fd9a98178cdebda4213796fdc452a8a265a1197 Author: Keith Packard Date: Tue Aug 28 16:00:01 2007 -0700 Don't set supported TV formats until after RandR initialized. The TV format property cannot be configured until RandR has been initialized. commit c6e637cd683dc60567b3b4f69b7f2b4c338c89ea Author: Keith Packard Date: Tue Aug 28 12:17:15 2007 -0700 Limit TV formats to those supported by current connection commit ddd6053987b9ca9bd3722ddbdfd412a3d8d252cf Author: Keith Packard Date: Tue Aug 28 12:30:46 2007 -0700 Add register defines for hw binning commit 3fbbd0afde49c53a5a8661f75c8c8c4be3020c30 Author: Michel Dänzer Date: Tue Aug 28 17:48:20 2007 +0200 Fix build against pre-pci-rework xserver. commit 0fdbf64b34e4114c2b89d696b268b9c7464f1efd Author: Zhenyu Wang Date: Tue Aug 28 21:56:21 2007 +0800 Fix i915 a8 color buffer blending From spec, i915 engine uses green channel when reading from 8bit color buffer for blending, and also writes back green channel. Fix blend factor in dest alpha case by using dest color instead. Now rendercheck can pass a8 tests. commit 3411eb0dbae470b910af3116a4ab960c821b9b20 Author: Keith Packard Date: Sun Aug 26 23:09:01 2007 -0700 i830_driver.c changes for libpciaccess. Change to use libpciaccess APIs, including computing and using BAR indices for various mapping activities. commit 387fed6daa7426e4a85d30ba7cf608b5f41d24bb Author: Keith Packard Date: Sun Aug 26 23:06:57 2007 -0700 i810_driver.c changes for libpciaccess. This includes new probe code (intel_pci_probe) and changes for i810 to use BAR indices to refer to suitable portions of the device mappings. commit 2c794192052ca55c3263e27e13d16aafe8caa92c Author: Keith Packard Date: Sun Aug 26 22:46:19 2007 -0700 Mechanical API conversions for libpciaccess. Uncomplicated API transistions for libpciaccess usage: Legacy xf86 API libpciaccess API --------------- ---------------- xf86ReadPciBIOS pci_device_read_rom pciReadWord pci_device_cfg_read_u16 pciWriteByte pci_device_cfg_write_u8 And, more use of the API-independent DEVICE_ID/SUBVENDOR_ID/SUBSYS_ID macros to pull PCI identification data from the underlying structure. commit 70e8e5957200401474967a467663ae049e9080f2 Author: Keith Packard Date: Sun Aug 26 22:40:25 2007 -0700 Change DRI interface to fill in PCI data from new libpciaccess structure. The DRI interface requires bus identification for each DRI object; pull that data from the libpciaccess structures as necessary. commit 5516cc781bd488c936af225123812a61ed5874b8 Author: Keith Packard Date: Sun Aug 26 22:37:38 2007 -0700 Add libpciaccess declarations to I810Rec and I830Rec. Using libpciaccess requires a different type for PciInfo (struct pci_device instead of pciVideoPtr) and it requires knowing which BAR each memory region needs to be mapped from. Add these definitions to the driver private record along with the includes necessary to use libpciaccess. commit daada59b5f8c2294b524a4b5920dc6b1c213642f Author: Keith Packard Date: Sun Aug 26 22:32:39 2007 -0700 Change IS_Ixxx tests to work with or without libpciaccess. libpciaccess has a new structure that holds the PCI identifier data; borrow macros from the mga driver to work with either the old xf86-specific structure or the new libpciaccess structure. commit 3d3bf493a3973f4067433d27a4d7ddfecaa18f1c Author: Keith Packard Date: Sun Aug 26 22:27:34 2007 -0700 Intel driver configuration (only) changes for X server libpciaccess usage. Detect whether the target X server uses libpciaccess, using it in the driver compilation as necessary. This change means that utilities that used to use libpciaccess will not do so unless the driver itself uses libpciaccess. Yes, that could be fixed, but it doesn't seem that important. This patch does not include any code changes necessary to actually have the driver build against an X server using libpciaccess. commit 5faf9cc6afe1c30fa88bc6446088a6fa47fc5d0b Author: Keith Packard Date: Sun Aug 26 09:34:06 2007 -0700 Sort quirk table, add Dell Latitude X1 commit a9e1d42a47cef79d8bbde2afd89d26aed964e344 Author: Keith Packard Date: Sat Aug 25 12:54:11 2007 -0700 Lenovo 201a is x60s, not x61s commit ffc2907f7f2eb039004eff0014c5563a01463fb0 Author: Keith Packard Date: Sat Aug 25 12:31:21 2007 -0700 Thinkpad X61s has no TV out commit 0c20fbabd18c19b2753cb60280f89e240ce5645f Author: Keith Packard Date: Fri Aug 17 22:21:47 2007 -0700 Make sure XV_PIPE is used whenever possible. The code was not consistently using XV_PIPE when the desired crtc contained any portion of the video output. commit e443f83dd6f110156743c93f7d793cdddb8195a1 Author: Eric Anholt Date: Fri Aug 17 18:13:49 2007 -0700 Tune acceleration architecture allocator sizes down. commit bd874b11bbfe582aebd3115771f90807e75afc31 Author: Eric Anholt Date: Fri Aug 17 17:49:21 2007 -0700 Replace AA allocator usage with i830_memory.c for RandR rotation. This requires EXA 2.2 (server 1.3) for rotated performance with EXA, because the i830_memory.c allocation may not fall within what EXA considers the offscreen area, so the PixmapIsOffscreen hook is needed. commit 9ad33dd65a79277ef75a6e95373614852725f5a9 Author: Eric Anholt Date: Fri Aug 17 16:46:48 2007 -0700 Use i830_memory.c instead of the AA's allocator for XV buffers. This should fix issues with XV being allocated into XAA's tiled pixmap cache and resulting bad rendering. Its also brings us closer to being able to shrink the size of the pixmap cache on XAA, which is of limited utility. commit 3655a1ecb62f6c387a16fa87cf6f00bf7835dce4 Merge: e5c336e 2231cdc Author: Jesse Barnes Date: Thu Aug 16 12:04:20 2007 -0700 Merge branch 'master' of ssh://git.freedesktop.org/git/xorg/driver/xf86-video-intel commit e5c336eaa32be8f9379a2c1dd51006b85bc8b270 Author: Jesse Barnes Date: Thu Aug 16 12:04:02 2007 -0700 Disambiguate plane and pipe mapping, use plane A on pipe B on pre-965 LVDS Add a new 'plane' field to the intel_crtc private structure for tracking planes separate from pipes. This allows pre-965 chips to use plane A on pipe B, enabling framebuffer compression for builtin LVDS displays. commit c3438356d6b7b63ad9d0086efb084be8e048980e Author: Eric Anholt Date: Thu Aug 16 11:34:02 2007 -0700 Add tiling information to BO layout description. commit 0c9e4aeea84e20a18e3b76d8cf8e802af004df57 Merge: 79d9a30 2231cdc Author: Eric Anholt Date: Thu Aug 16 11:30:16 2007 -0700 Merge branch 'master' into buffer-objects Conflicts: src/i830_dri.c src/i830_memory.c commit 2231cdcd8f1ee81b3e59cc5e3a325c22ee0f40e4 Author: Dave Airlie Date: Thu Aug 16 17:15:54 2007 +1000 i915: add support for render to a8 commit a69db6f7fe1703b473e5c1d1e0088ccc203f4d5a Author: Dave Airlie Date: Wed Aug 15 18:28:50 2007 +1000 intel: don't setup texOffsetStart unless using EXA commit 5126a71f82767b9e23cd590453718f3364789740 Author: Zhenyu Wang Date: Tue Aug 14 13:54:55 2007 +0800 Fix seg fault introduced in tiling patch when TV detect When TV does load detect, fb hasn't been setup, so we should check that in i830_display_tiled(). Caught by Nanhai. commit d9f89a1af7e7ff4056727060cdf2e35c15a4dcdd Author: Keith Packard Date: Fri Aug 10 17:59:33 2007 -0700 Save/restore tile-mode offset registers DSPATILEOFF and DSPBTILEOFF Now that the driver sets these registers, they must be saved and restored. commit 5bc194d3d3c87bb0128d9ac10f090f031345eb37 Author: Keith Packard Date: Fri Aug 10 17:54:32 2007 -0700 Set DSPATILEOFF/DSPBTILEOFF to handle 965 tiled frame buffers. DSPATILEOFF and DSPBTILEOFF replace DSPASURF and DSPBSURF when the frame buffer is in tiled mode. commit 64b943c79cf957a4c54482720195d7f27b7f0c0d Author: Eric Anholt Date: Fri Aug 10 15:48:05 2007 -0700 Add #if 0-ed fence debugging code. It's noisy, and of little use to most. commit ba9a503ba2099025e393f3382bb453985ef23497 Author: Eric Anholt Date: Fri Aug 10 15:06:58 2007 -0700 Don't force tiling on if it is disabled in configuration but fbc is possible. commit b7751c7d1d6bcf310824295c3bab4ff36760c791 Author: Eric Anholt Date: Fri Aug 10 17:42:09 2007 -0700 Fix stack-smashing in the last commit. commit cb36635a053d4ac3971fea05060d31dbd3d382d2 Author: Eric Anholt Date: Fri Aug 10 14:16:59 2007 -0700 Attempt to fix several front buffer tiling failure cases. Front buffer tiling is now disabled with G965 and XAA. Some of the acceleration that i830_xaa.c does can't be supported on tiled buffers. Adds a tiling field to struct i830_memory, and uses it instead of separate variables for each potential tiled buffer. commit ed1b106fabf3a18489bdb3083326f27387a9cb72 Author: Keith Packard Date: Fri Aug 10 14:31:16 2007 -0700 Clean up tv mode name allocation and copy. TV mode names used to contain the signalling standard along with the pixel size. The signalling has been moved to the TV_FORMAT property, but the allocation and initialization of the mode name was left a bit messy as a result. commit f71b9358b4157a8cfdc694ddef8ca3f98926ca91 Author: Jesse Barnes Date: Fri Aug 10 15:53:04 2007 -0400 Cleanup tiling and FBC driver output. Remove an extra "FBC enabled" message from i830_memory.c (only report errors if they occur), and don't print the "forcing FBC on" message if tiling was already enabled, as it's redundant and confusing. commit e6746d0f286ef9d9a87f748d40e5421c268f2f7d Author: Jesse Barnes Date: Fri Aug 10 15:48:15 2007 -0400 Enable tiling by default on 965. commit e0fcf645a228094620b8f7fdd580963611bdd6ef Author: Jesse Barnes Date: Fri Aug 10 15:43:06 2007 -0400 Tiling fixes for 965 This should be close to the last set of tiling fixes for 965 chipsets. Prior to this commit, the 965 composite hook didn't take tiling into account, nor did 965 textured video, which caused display corruption. However, there seems to be at least one last bug to squash--on occasion, a configuration with tiling enabled won't properly display text. This is likely another tiling related problem with the composite hook. commit 02ad9cee330fb5fb0008ad6f74e74e8d8a1fc36e Merge: 105f8a1 7b143e5 Author: Zhenyu Wang Date: Fri Aug 10 16:45:56 2007 +0800 Merge branch 'master' into xvmc-i915 commit 105f8a183eb64623d9a9e2833dbe27156a551f69 Author: Zhenyu Wang Date: Fri Aug 10 16:44:46 2007 +0800 disable subpicture initial until we really implement it, OSD can't work for now. commit 7b143e5c8397da077c0e02455c21c5a99cf50942 Author: Dave Airlie Date: Thu Aug 9 12:14:44 2007 +1000 i965: increase composite vertex buffer size and alignment to be safe commit 14691b24da5aa29d8c41ac7b7c61828e3cd9eab7 Author: Dave Airlie Date: Thu Aug 9 09:41:32 2007 +1000 i965: fix memcpy of the sf_kernel when a mask is needed commit 5e18c6af9051da654d2a6a97553ef4fe777bb61e Author: Carl Worth Date: Wed Aug 8 11:13:37 2007 -0700 Allow 965 composite acceleration to A8 destinations. Note that this is a slowdown in text rendering due to the high overhead of our compositing setup, but appears to be correct according to rendercheck. commit b0ec670cdb0b6ca6fc0f4f165fa3ee5a20d7c985 Author: Eric Anholt Date: Wed Aug 8 11:03:51 2007 -0700 Bug #11593: Remove dead struct vch_bdb_20 which was angering the sun compiler. commit 87cc72ef509b85635bf0e84c47dfed6b7242170b Author: Zhenyu Wang Date: Wed Aug 8 09:39:55 2007 +0800 check early if we have slot left for new context or surface commit fec6744b763f2b85b4e592c51326b9ee979bef5d Author: Zhenyu Wang Date: Wed Aug 8 09:26:10 2007 +0800 I missed to remove port attributes wrapper commit 76c084d41999f13497b80d4ac6799cb974adaa03 Author: Zhenyu Wang Date: Wed Aug 8 09:22:42 2007 +0800 explicit say "unsigned int" commit cfc614b1f9143ed1407733aaa81c4a980c613a63 Author: Zhenyu Wang Date: Wed Aug 8 09:06:47 2007 +0800 remove xvmc attributes We don't have extra attributes than Xv port. commit 92af2f4bbcb395cbde097776718449d99843ad67 Merge: da82a47 e0be352 Author: Eric Anholt Date: Tue Aug 7 15:18:17 2007 -0700 Merge branch 'origin' Conflicts: src/i830_exa.c commit da82a47a558597f3653e2b33bc6adbab18574b57 Author: Eric Anholt Date: Mon Aug 6 17:01:37 2007 -0700 Fix EXA rendering with tiled front buffer on pre-965. The 915 and earlier appear to respect the fence registers, while only the 965 requires the per-operation tiling setting and pitch shifting. This will also fix issues with rendering on the 965 involving multiple cliprects, where the pitch would get divided repeatedly. This removes the offset < 4096 fallback, which essentially resulted in no acceleration to tiled buffers, hiding the issues. commit e0be352f5017f0e645a4ff8a40961d9c2b98863a Author: Jesse Barnes Date: Tue Aug 7 12:37:25 2007 -0700 Fixup pitch in Prepare* functions, since actual hooks may be called many times for the same pixmap, and we don't want to keep dividing the pitch by 4. commit b29a932bec2a3cc5a38e9c09c012cfdc1b6209c1 Author: Zhenyu Wang Date: Tue Aug 7 16:47:13 2007 +0800 code cleanups and put wrap function in driver xvmc priv instead of per xv port priv commit 7431abee5fb971d1f8bc7ac4bea137f6ece9418b Author: Brice Goglin Date: Tue Aug 7 09:13:00 2007 +0200 Define INTEL_VERSION_MAJOR/MINOR/PATCH using PACKAGE_VERSION_* commit 9e1914270a0978ec4dfae757d3dd57ca7ffe17e5 Author: Jesse Barnes Date: Mon Aug 6 17:55:00 2007 -0700 Remove 4k offset checks from Copy & Solid hooks. Reading the docs too literally can cause you to hide bugs with false fixes... commit 3510d5728fa972b36d022b4f9189d46ff98d7b16 Author: Eric Anholt Date: Mon Aug 6 16:44:39 2007 -0700 Fix accumulated whitespace nits in i830_exa.c commit 5ff05dffe229e35da7619762628fdd0f125585e8 Author: Jesse Barnes Date: Mon Aug 6 16:01:10 2007 -0700 More tiled rendering fixes: - check for tiling, not just offset in PrepareSolid - combine pI830->tiling and frontbuffer checks into new exaPixmapTiled function for readability commit ba90d944329dd8c79a757c38128964fbbe4ab898 Author: Eric Anholt Date: Fri Aug 3 16:46:09 2007 -0700 Add the file mode for bios_dumper output so it doesn't have 000 permissions. commit 322a163cfbda885adc6bb09c1f976d36617ea83b Author: Eric Anholt Date: Fri Aug 3 10:46:39 2007 -0700 Quirk away the nonexistent TV connector on the Panasonic CF-Y4. commit ffbab2ee5dc227b2a8a5ffd1717ae00e8e37f956 Author: Jesse Barnes Date: Fri Aug 3 21:27:52 2007 -0700 Limit Solid & Copy offsets to 4k when rendering to tiled targets commit 019dbfda294aaafb28d8bea0fe2f5dadc2ea3e0b Merge: 3d3c0e8 15f71ed Author: Jesse Barnes Date: Fri Aug 3 20:45:14 2007 -0700 Merge branch 'master' of ssh://git.freedesktop.org/git/xorg/driver/xf86-video-intel commit 3d3c0e8c55f639a501c0756948b518abd903d7d0 Author: Jesse Barnes Date: Fri Aug 3 20:40:45 2007 -0700 Tiled rendering & fbc fixes: - actually enable tiling in DSP(A|B)CNTR if needed - add logic to EXA routines for tiled case (still needs work) - enable/disable fbc on DPMS events (meant moving functions higher in file) - fix fence register pitch programming (use correct pitch instead of kludged value) commit 79d9a309b19e22561e000a47b732c67479c2e6d4 Author: Eric Anholt Date: Thu Aug 2 15:50:42 2007 -0700 Mark DRI buffers as shareable, and pass their buffer handles through the SAREA. commit 0da4f2b0cd7203377ad10407928a367b8c6d310e Author: Jesse Barnes Date: Tue Jul 31 16:22:36 2007 -0700 Legacy backlight changes: - add support for 965GM - make sure legacy enabled systems don't reduce the range of backlight values we can present to the user commit 15f71edba37738f8ba279fa07452fda10cc65298 Author: Zhenyu Wang Date: Sat Jul 28 17:43:29 2007 +0800 Update Lenovo TV quirk info commit 857b4a7bcb69ca43b866b4283fe075abbafb1d22 Author: Eric Anholt Date: Fri Jul 27 18:03:12 2007 -0700 Pin some buffer objects at creation time, which can't be moved yet. A number of other interfaces of ours don't allow buffer offsets to be updated after screeninit. This attempts to catalog why for each one, so that they can be fixed one by one. This happens to restore the EXA offscreen allocator for now, as a fixed-offset object. commit d7b5b595ad29ae0ac3adbd3a176be3fe7ffa6474 Author: Eric Anholt Date: Fri Jul 27 12:48:08 2007 -0700 Delay the first screen pixmap update to CreateScreenResources. The return value of GetScreenPixmap before CreateScreenResources is not, in fact, a pixmap. commit 2b1ec0f51479c1d7e5e62803c68feca29375519d Author: Eric Anholt Date: Fri Jul 27 10:00:35 2007 -0700 Use the dontMapFramebuffer option available with DRIINFO 5.4. If not available, AIGLX init will fail. While here, simplify DRIINFO tests since we refuse to init with a version queried less than the version we compiled against, anyway. commit f403a50afbcef1e54f554481c72037338bd5357c Author: Wang Zhenyu Date: Fri Jul 27 09:24:24 2007 +0800 Add another Lenovo TV output quirk From issue report http://lists.freedesktop.org/archives/xorg/2007-July/026644.html commit 34c82ad7ce83394db47588693b578cf91991bf1c Author: Wang Zhenyu Date: Fri Jul 27 09:14:13 2007 +0800 Add quirk support This one trys to use a flag for possible quirks. It adds a quirk for my Lenovo T61 TV output, and ports some origin LVDS quirks to it. commit 1a585d03972394cb97dec2462937b0104bdf4de2 Author: Eric Anholt Date: Thu Jul 26 10:46:35 2007 -0700 Use new drmBOSetPin interface instead of NO_EVICT/NO_MOVE buffers. To do this, we have to deal with buffer offsets being set at EnterVT time instead of screen init time. We've wanted to move this direction for a long time, but there are repercussions. The EXA offscreen memory manager has to be disabled, because it can't be moved. That will be replaced by BO-backed pixmaps soon. Also unresolved is whether our moving front/back/depth/texture buffers will break the classic-mode DRI driver. This code doesn't actually work yet. commit 0fd3ba0518b3cde9ca0e4e2fc1854c00d8a43d5c Author: Brice Goglin Date: Wed Jul 25 20:11:32 2007 +0200 Fix typo in intel.man Reported by A. Costa" in http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=432061 commit 18c707a8c13ac5fffdd30d55e6f5926a68b367df Author: Eric Anholt Date: Wed Jul 25 10:11:16 2007 -0700 When TTM is available, use it instead of manual AGP allocations when possible. This is a step towards being able to expose buffer objects through the screen private to DRI clients, instead of having them have to use the fake buffer object type. This fails in two ways. First, the kernel memory manager is not currently suitable for doing the physical allocations we need, so we still use AGP for those. Additionally, the DRI lock can't be initialized early enough for us, so these buffer object allocations fail. This will be fixed by improving the DRM interface. commit c4deefa80a6349db968e25bd6778cdf7a514ca7f Author: Xiang, Haihao Date: Mon Jul 23 10:35:32 2007 +0800 only enable xvmc for 915/945/G33 commit 45962eed51120ff77326c29d72cf8b6cd8a934b5 Author: Zhenyu Wang Date: Mon Jul 23 09:50:17 2007 +0800 Fix a typo in i915 render Fence setting is in mapstate actually. This fixes rotation in tiled fb case, thanks Keith to report this. commit a52c53b8603495ab05e07e3da5092fe122fcb6f2 Author: Xiang, Haihao Date: Fri Jul 20 17:10:27 2007 +0800 don't initialize XvMC if xv is disabled. commit b1af2c0e01c54ef1d40fd0ca1ede29a1dd7ed97b Author: Zhenyu Wang Date: Fri Jul 20 15:18:48 2007 +0800 Fix device id info for 945GME, 965GME which do have new host bridge ids commit 6b4d43dbf4bbb86d479d8fca480c8bec39e2714b Author: Xiang, Haihao Date: Fri Jul 20 09:59:40 2007 +0800 lock display when creating context commit 42e91c7092f9cf7f5c0b98e9ae3d92a5b30821bf Author: Xiang, Haihao Date: Thu Jul 19 17:30:54 2007 +0800 move i830_allocate_xvmc_buffer out of #if/#endif commit 36ea26f2779d951fa173da785aca100564e5ecf8 Author: Xiang, Haihao Date: Thu Jul 19 17:21:05 2007 +0800 add copyright commit 9d03873423c0968b2d6700b7d8d44c7e23b8d7f3 Author: Xiang, Haihao Date: Thu Jul 19 10:03:19 2007 +0800 block some signals within critical section to avoid dead lock. commit 25658f838bd1bcd7559d7a032d127c626833fb7c Author: Xiang, Haihao Date: Mon Jul 16 13:03:58 2007 +0800 1. The pitches for y/u/v surfaces are 1024bytes aligned. 2. clean code commit 8f316a4584d0676e5dae051698bde0831f018cc4 Author: root Date: Thu Jul 12 16:24:03 2007 +0800 1. clean code 2. fix an error in map_state commit 4844254431695158287167e3b0fad25d9efb7a6c Author: root Date: Tue Jun 26 16:32:30 2007 +0800 a fix for 945G/945GM commit 8c249765ce788f8adb0325a8e3016a0eae74b13b Author: root Date: Mon Jun 25 14:07:01 2007 +0800 1. using batch buffer 2. implement macroblock_ipicture instruction 3. 16bit INTRA block commit 2b20b395541f786e4009211e4c56042257b8d114 Author: Xiang, Haihao Date: Mon Jun 25 10:43:49 2007 +0800 fix a bug related to display surface commit 6e47601ac7427390fa4cb6e228dfb566ff95db09 Author: Xiang, Haihao Date: Mon Jun 25 10:40:08 2007 +0800 fix some bugs commit d038ffdfffb846e22f360faac5d469c8e59795ad Author: Xiang, Haihao Date: Mon Jun 25 10:37:12 2007 +0800 Release resource allocated for surface/subpicture commit a9bf7e28b91692c7c7a4915cdf9603cd2ef025ea Author: Xiang, Haihao Date: Mon Jun 25 10:33:35 2007 +0800 fix some bugs commit ccac60bf035bb54661c13f306031def81e105a32 Author: Xiang, Haihao Date: Mon Jun 25 10:23:28 2007 +0800 fix some bugs commit 934da9f2b83716bb37d0719a61b8e6b179fff2cd Author: Xiang, Haihao Date: Mon Jun 25 10:17:08 2007 +0800 enable XvMC for i915 commit c7920a0e819308762fca3d6fc7ab194bd565b06a Author: Dave Airlie Date: Thu Jul 19 15:09:54 2007 +1000 strip out remainder of drmmm code in driver commit 37652b68880f1881b90bd22218cfe86eca7e5974 Author: Dave Airlie Date: Tue Jul 17 14:03:21 2007 +1000 intel: oops I commited pixman local workaround - undo it commit e40f6a4923d1323702406266e90eae3218a8a44e Author: Dave Airlie Date: Tue Jul 17 09:20:07 2007 +1000 intel: actually 2.3.1 should be good enough commit 1e169be25b2e4ab34afd4b8ae8ae0041f6069125 Author: Dave Airlie Date: Tue Jul 17 09:17:31 2007 +1000 intel: don't try and use TTM memory manager with old libdrm interface I probably need to release a libdrm with this interface in it now.. commit ff2be3995d33f9e4b7f63b380f166b6168c9b9c6 Author: Keith Packard Date: Fri Jul 13 12:47:18 2007 -0700 Remove hard-coded CRT blanking frobbing for load detection. CRT blanking needn't be adjusted to perform load detection on 9xx chips, and the 8xx load detection path now adjusts blanking just during load detection. Adjusting the blanking interval turned out to cause many monitors to fail to sync. commit 00f4587025a3879626623135b0a153fcdb906719 Author: Keith Packard Date: Fri Jul 13 10:58:06 2007 -0700 Ensure pipe/output active before doing load detection. If the pipe or output have been set to DPMSOff, then load detection will not work correctly. Also, share the load detection configuration code between crt and tv outputs. commit 6f18300aed1340348c6d395f326061b5315be643 Author: Keith Packard Date: Mon Jul 9 21:29:55 2007 -0700 Eliminate bogus (and harmful) blanking adjustment for load detect. Instead of always adding blanking to mode lines, use the FORCE_BORDER option on i9xx hardware where it works, and dynamically add a bit of border if necessary on i8xx hardware to make load detection work. This may cause flashing when a usable crtc is not otherwise idle when load detection is requested. commit 04130ac6b705aa49161fb6dae83ad0bdd76e89d9 Author: Wang Zhenyu Date: Wed Jul 11 11:42:56 2007 +0800 Fix i915 rendering for tiled buffer Make it to check fence register for dest buffer. commit 88f8b688e2316ae4a1f7485f0010ce90de54783a Author: Eric Anholt Date: Mon Jul 9 12:56:13 2007 -0700 Fix some physical address handling for >4GB addresses. The upper bits would have been inappropriately dropped on G33-class hardware, and on G965-class hardware in a 32-bit environment. The only use of physical addresses on these should be for FBC, though, and FBC requires addresses below 4GB. This is unresolved. commit bf831117b4659cc4f2774098dee938505f780a9b Author: Jesse Barnes Date: Sat Jul 7 10:15:32 2007 -0700 FBC fixes: - allow FBC and Tiling to be forced off if configured to do so - only touch FBC registers if pI830->fb_compression is true commit b426866fe1be2ad3861559beff69186379a6afad Author: Jesse Barnes Date: Fri Jul 6 20:48:40 2007 -0700 Fix manpage to reflect default behavior. commit 377c58373daa6bef5d37ead2b6f9a769a905b6fa Author: Jesse Barnes Date: Fri Jul 6 20:39:19 2007 -0700 Fix naming of FBC plane enable bits (mistakenly called them pipes earlier). commit 9c0388dc8d4c6495fae21af6da644b34e20173d1 Author: Jesse Barnes Date: Fri Jul 6 20:38:41 2007 -0700 Update man page with current behavior. commit cecbc71fdc9af832cef23427696f6f654f7d6104 Author: Jesse Barnes Date: Fri Jul 6 16:36:34 2007 -0700 Fix debug output in fbc enable/disable routines. Add logic to make sure fbc isn't enabled twice on two different pipes. commit 4359df9419d2d02a2f9d9adc7f5a49ecf07ddd30 Author: Jesse Barnes Date: Fri Jul 6 16:17:45 2007 -0700 Fix tiling and fb compression defaults for 965 (not yet fully supported). commit ca593a5219549df94a6d234ebbcf9e7c44723c9b Author: Jesse Barnes Date: Fri Jul 6 16:10:52 2007 -0700 FBC and tiling changes - change framebuffer option name to "FramebufferCompression" - add new "Tiling" option (controls all tiling, not just front buffer) - add debug message to fb compression enable/disable routines - update man page with new options commit 8798ef11321ee6957919279076758d47ad956cf3 Merge: 8919b22 3c552af Author: Jesse Barnes Date: Thu Jul 5 12:21:31 2007 -0700 Merge branch 'master' into fbc commit 8919b2292147add41a1c1c6e5e673257cb6c6c6e Author: Jesse Barnes Date: Thu Jul 5 12:21:06 2007 -0700 Re-add tiling kludge, but only for 965. commit 407b124af8f7bb42abe4eecc87476c4c3e555cd0 Author: Jesse Barnes Date: Thu Jul 5 11:31:34 2007 -0700 Remove tiling kludge. May need more fixes for 965. commit 7a87b9d2a2eb4d281dce67586756ff5653b2805a Author: Jesse Barnes Date: Thu Jul 5 11:23:06 2007 -0700 Revert discard alpha change, requires other fixes to work. commit fecf964534f5ba6d40480cb13adc89094946a51e Author: Jesse Barnes Date: Thu Jul 5 10:59:23 2007 -0700 FBC fixes: - properly check several FBC enablement constraints - don't use alpha discard if FBC is in use commit 60ee7b6a91b2b8c447130c60cd8b19eb68119777 Author: Jesse Barnes Date: Tue Jul 3 14:20:34 2007 -0700 Fixup line length buffer padding, add kludge for front buffer tile pitch. commit 3c552af65d28fafec1d09484a8914b690b961349 Author: Eric Anholt Date: Mon Jul 2 18:33:47 2007 -0700 Update documentation and bump driver version to 2.1.0. commit f02036aedcd7866c567a6adc070eda3dad872105 Author: Jesse Barnes Date: Mon Jul 2 15:42:02 2007 -0700 Framebuffer compression changes: - move FBC register definitions to i830_reg.h - add fix from Arjan for 965 depth buffer tiling - add VT switch and clear-at-server-start code for FBC registers commit 2b9961eb9ce8734565ecdb01cb11610714d7f610 Author: Eric Anholt Date: Mon Jul 2 15:16:33 2007 -0700 Fix reversed LVDS dither enabling logic on GM965. commit 3d9ee8b2991ec0da8cc21b8455ff7f00fd0335b5 Author: Eric Anholt Date: Mon Jul 2 14:38:28 2007 -0700 Bug #11365: Disable the panel fitter unless it's needed for the chosen mode. The automatic panel scaling appears to choose bad sampling on some GM965 hardware for 1:1 mapping modes, and there's no real sense in having it on if we just want 1:1. commit b384c608978dcd3d2ea6c0018179673cb4735f4c Author: Jesse Barnes Date: Mon Jul 2 09:32:28 2007 -0700 Enable framebuffer compression (use Option "FrameBufferCompression" "true" in your xorg.conf). Should save ~0.5W during typical 2D usage. commit 1e2e301348b4168aeed38b3fdc6b0e43d5678a86 Author: Keith Packard Date: Sat Jun 30 12:45:24 2007 -0700 Fix load detection to use border region instead of blanking. Make sure there is some border area to use by changing how the pipe is configured, then pick a scanline in the middle of the border for load detection. This lets the load detect code use an active pipe instead of requiring an idle one. commit 11862c2e1f23b77b56d7bd8b384579b5e3ae377b Author: Alan Coopersmith Date: Thu Jun 28 23:31:28 2007 -0700 Add *~ to .gitignore to skip emacs & patch backup files commit 6503eb45023d0db9a94cb9d1e14a26af07a6628d Author: Alan Coopersmith Date: Thu Jun 28 23:30:35 2007 -0700 Add AM_PROG_CC_C_O to configure.ac Clears automake-1.10 warning: src/bios_reader/Makefile.am:8: compiling `bios_dumper.c' with per-target flags requires `AM_PROG_CC_C_O' in `configure.ac' commit 5257e36f502676fd6a44bbb8e747d9138ed3bc5c Author: Keith Packard Date: Thu Jun 28 15:29:52 2007 -0700 Handle dual-channel LVDS on i855. Just as with i9xx LVDS, the i855 LVDS can operate in dual-channel mode with a modified P2 divisor value (7 instead of 14). Just using the existing 9xx code for 855 appears to work fine. commit 16bfcb8042519f24b4494fd621814f39949ceeb6 Author: Keith Packard Date: Thu Jun 28 15:27:56 2007 -0700 Decode PLL registers in LVDS mode a bit better in debug code. LVDS mode changes how the PLL works in fairly dramatic ways; the debug code wasn't properly accounting for those differences resulting in fairly bogus debug output. commit 9675ccb30818bf831ac4c634751ab4bfe35f7bfe Author: Wang Zhenyu Date: Wed Jun 27 09:23:33 2007 +0800 EXA: fallback mask transform on i965 It needs to fix shader programs which hasn't been done yet. commit 7a2300c88ae59f5b7c3ce89d33147e3f0ca23c18 Author: Wang Zhenyu Date: Wed Jun 27 09:19:22 2007 +0800 EXA: don't have to check offscreen size DDX will check it for EXA_OFFSCREEN_PIXMAPS flag commit fff4a3b58fa18ee2ad91f998d190e90b77c051ab Author: Carl Worth Date: Mon Jun 25 23:15:58 2007 -0700 Use local structures for vs_state, sf_state, and wm_state commit 0a8a4afd3c59011d6b1f5b39aedfb9bce0e55c48 Author: Carl Worth Date: Mon Jun 25 23:12:23 2007 -0700 Use local structure for src_sampler_state and mask_sampler_state commit 499166a60fcbf16021bd9ec233790ba55803aa44 Author: Carl Worth Date: Mon Jun 25 23:09:17 2007 -0700 Use local structure for mask_surf_state commit a418ef7316808b239884a90c3fe890220bcc0242 Author: Carl Worth Date: Mon Jun 25 23:07:19 2007 -0700 Use local structure for src_surf_state commit 0e3c0b17826b7b5a21ee2c1d789b084fc167f1ed Author: Carl Worth Date: Mon Jun 25 23:05:50 2007 -0700 Use local structure for dest_surf_state commit 41a2c0f15446d59678461648f476fa71d40d44e0 Author: Carl Worth Date: Mon Jun 25 23:03:47 2007 -0700 Use local structure for cc_state commit 59f2150caca7eb374a2db43a472ba85f50d23274 Author: Carl Worth Date: Mon Jun 25 11:28:57 2007 -0700 Remove redundant i830WaitSync from i965_prepare_composite There were two calls to i830WaitSync, and between them no state was being changed---just offsets were being computed. commit 66aa0e61e1e8d2216a9c0555be5be004ed0a3192 Author: Dave Mueller Date: Fri Jun 22 16:45:27 2007 -0700 Bug #11171: Add support for the Ti TFP410 DVO TMDS transmitter. commit f8d7cbc6e1322acad3351591336cefcfba7d9aaf Author: Eric Anholt Date: Fri Jun 22 16:29:21 2007 -0700 Move the ivch's fixed panel mode support to i830_dvo.c for other LVDS drivers. This also results in removal of the setup hook, which was being called unconditionally and breaking non-ivch dvo drivers. commit ec236c76b93aea5f2ee1e8b8509cde4625974fcb Author: Keith Packard Date: Fri Jun 22 16:32:46 2007 +0100 I830 needs to have plane/pipe/pll started in mode_set. The patch for the i855 to stop enabling plane/pipe/pll in mode_set broke the i830. Revert that just for the i830, leaving it enabled for the i855. commit d957c6b8e1dde8e11c1db3431e0ff58c5d984880 Author: Keith Packard Date: Fri Jun 22 01:32:02 2007 +0100 Increase vblank wait timeout from 20ms to 30ms. 49.6Hz < 20ms. The x40 LVDS mode has a 49.6Hz vertical refresh. Waiting for only 20ms can sometimes cause the driver to start programming the hardware before the vblank has occurred, which will lock up the i855 chipset. Extend this to 30ms (the maximum timeout used by the BIOS) to ensure this doesn't happen. Detecting actual vblank occurance using the various status registers should also be possible but isn't yet working. commit a67c2965385001bcb8987265f698ff0f5809cd11 Author: Keith Packard Date: Thu Jun 21 23:59:38 2007 +0100 Follow BIOS configuration for Legacy Backlight Brightness. The backlight control in the LVDS controller can either operate in 'normal' mode or 'legacy' mode. In legacy mode, it uses the PCI config space register 0xf4 which can range from 0 to 0xff. In normal mode, it reads the range and current value from the BLC_PWM_CTL register. commit d6e46f67ab3af1ad3bfa72acb0efd9fe79dbf1dc Author: Keith Packard Date: Thu Jun 21 20:16:36 2007 +0100 Eliminate some uninitialized variable warnings commit 9d104634cf03bea82d1467f01e577cb8d2e4b554 Author: Keith Packard Date: Thu Jun 21 01:15:39 2007 +0100 Add 3DSTATE_CLEAR_PARAMETERS bits commit 3bbf313ba541526a893915f8b6c64b1eccf325e0 Author: Wang Zhenyu Date: Tue Jun 19 09:33:50 2007 +0800 Fix left G33 issues Be sure to check G33 chip type in: - sdvo output - Y-major tile - crt detect - and xaa composite Sorry for that I should have fixed them very earlier... commit acef342c870f3b5b781e48c8bf44739aa5ee8ffa Author: Eric Anholt Date: Mon Jun 18 11:57:48 2007 -0700 Bug #11295: Disable textured video on i915 with framebuffer width too large. commit fbbb41bc5e03478cb46ee8f64ef68b23ff3fc14b Author: Keith Packard Date: Sun Jun 17 14:59:24 2007 +0100 Let DPMS functions enable plane/pipe/output on 8xx hardware. On 855, letting crtc_mode_set enable the plane and pipe will occasionally hang the chip. Instead, wait for crtc_enable to light things up. For 9xx, leave things alone. commit d5ca000ece145a35fd6df0dcf3fb3460bd2d64e3 Author: Rémi Cardona Date: Sat Jun 16 13:17:54 2007 +0100 Include stdint.h to get uint64_t commit 6b2ae93506d6795f87d6993bebfcb4e6632508ee Author: Dave Airlie Date: Fri Jun 15 23:30:04 2007 +1000 sdvo: add support for RGB outputs on SDVO This lights up my monitor VGA-1 - it doesn't look the best though commit 671ba03befebfdd7256855858987aabc28b2e8cd Author: Eric Anholt Date: Wed Jun 13 16:30:26 2007 -0700 Fix and enable the 915-class planar textured video path. commit 6c29e0bae5f1e7cee02b678418394abb971594eb Author: Eric Anholt Date: Wed Jun 13 13:40:39 2007 -0700 Improve the drm_i915_flip_t check. commit 420e41e7921d3cc07c784fd17936ec8a675f3b20 Author: Eric Anholt Date: Wed Jun 13 13:34:26 2007 -0700 Revert "Replace failure-prone configure test for fresh libdrm with a simple ifndef." This reverts commit c2b130354aecffbeb2a2d23c7371461feaf5766a. Sadly, a non-working DRM_IOCTL_I915_FLIP already existed. commit 51612e5ac3ddfb2bb172c58f2dfff9631093b69c Author: Eric Anholt Date: Tue Jun 12 16:09:54 2007 -0700 On hang, dump up to the head pointer, not just up to the tail. commit ceb6dd72443c094212b0281c42cbe92e9a29f682 Author: Eric Anholt Date: Mon Jun 4 16:37:53 2007 -0700 Fix context switching between DRI and X. Now, all 3D pipeline consumers in the driver just call IntelEmitInvariantState(), which handles basic state setup, the caching of that state setup, and notifying DRI clients. This also removes a mistaken idle wait in the Render code which was papering over the brokenness in the context switching. commit c2b130354aecffbeb2a2d23c7371461feaf5766a Author: Eric Anholt Date: Tue Jun 12 08:49:21 2007 -0700 Replace failure-prone configure test for fresh libdrm with a simple ifndef. commit 0e1deb607f94e4aa3ec4b9df8ff7a07a1c95e31d Author: Eric Anholt Date: Mon Jun 11 17:33:33 2007 -0700 Fix a typo in _3DSTATE_DEPTH_SUBRECT_DISABLE definition. This is already fixed in the definition in the 3d driver. commit 8d7a0ccd4f674659eb781def2cfdc3a6e5a219ce Author: Eric Anholt Date: Mon Jun 11 17:33:07 2007 -0700 Clean up some nits in i915_video.c setup. - The screen dimensions were used for the clipping despite drawing being done to any pixmap, not necessarily the screen. - One piece of state setup was not documented anywhere, and isn't used in other 3d hardware paths that also work. - A 3DSTATE_MODES_1 command (830-class only) was issued even though it no longer exists. commit 15caa64a497dcc0eacb0f91166d9b70206a8db35 Author: Keith Packard Date: Fri Jun 8 18:44:28 2007 -0700 Add description for how to use the frame and pixel counter registers. The 24-bit frame and pixel counters were not described in detail and will be useful for DRM. commit 404fd47573f855b0442d49a383542fc093825ad0 Author: Wang Zhenyu Date: Wed Jun 6 11:01:48 2007 +0800 Enable overlay on G33 class chipsets Which have to use gfx vm offset fot setup overlay regs. commit f4c05973d391bdb0a9b0eadb155548310baa98fd Author: Wang Zhenyu Date: Tue Jun 5 11:07:41 2007 -0700 Add support for the G33, Q33, and Q35 chipsets. These chipsets require that the hardware status page be referenced by an offset in the GTT rather than a physical memory address, so the X Server allocates it rather than the DRM. commit 36fcaeb2ef94db5399071540bba106dec3db81d5 Author: Eric Anholt Date: Mon Jun 4 15:58:31 2007 -0700 Fix misplaced merge of 1280x768 panel fixup. commit 8a19e7d57bc23dd163b45e0ab7deca4f074c934d Author: Keith Packard Date: Tue Jun 5 00:09:57 2007 -0700 Always ensure the pipe A is lit when activating overlay on pipe B. Ok, so moving video from pipe A to pipe B still requires that pipe A be active during the transition. Instead of trying to be fancy, just ensure that pipe A is running on each transition to pipe B. commit e986f6cb62f8644c5fa835bd7dfb7b014c2677c5 Author: Keith Packard Date: Mon Jun 4 23:52:23 2007 -0700 Automatically switch overlay when crtcs are reconfigured. As crtcs are disabled and enabled, make sure the automatic crtc selection mechanism drives overlay configuration at each request to display an image. commit 0984c1fc0963f1ebab31f5b8fce5ad4c387fbd2c Author: Wang Zhenyu Date: Thu May 31 10:17:26 2007 +0800 Add pci ids for 945GME commit 88ee25ebad78e54d243d728b775a69365359b5fb Author: Wang Zhenyu Date: Thu May 31 10:13:30 2007 +0800 Add pci ids for 965GME/GLE chip commit 92e4deb50e049cc83cbde4995ba0b901feceb15a Author: Keith Packard Date: Wed May 30 11:49:07 2007 -0600 Extend XV_PIPE range to include new -1 value commit 888a4a5f469bf955e3ee3e184b628808ae8a4498 Author: Wang Zhenyu Date: Wed May 30 14:42:35 2007 +0800 Fix i965 render's draw clip rectangle Use scrn's virtual size is not correct in rotation rendering. This fixes initial rotation problem on i965. commit 516fb73ffee0aea7cf892e6703d37f8ecf52b812 Author: Eric Anholt Date: Tue May 29 10:22:25 2007 -0700 Remove README statement that 830M panels are unsupported. commit 72462568da589054828b72ace83232a71636ee73 Author: Eric Anholt Date: Tue May 29 10:21:12 2007 -0700 Expand manpage description of outputs supported. commit 906b974bfeeed18d79c244ad3db4f5d30e13e4c8 Author: Eric Anholt Date: Tue May 29 09:49:08 2007 -0700 Add a fixup to LVDS panel mode detection for 1280x768 panel from text mode. Apparently some BIOSes will program a small mode with large blanking instead of using the pannel fitter. commit 4b2781291844b61b397e257a0fdb43e964e5f603 Author: Keith Packard Date: Sat May 26 10:09:11 2007 -0700 Mark IVCH as connected when detected commit 2a365eab0178c28782fba97bdd22365f30ce8963 Author: Keith Packard Date: Sun May 27 12:35:55 2007 -0700 On i830, Pipe B cannot be lit the first time unless Pipe A is running. I don't understand it, but just like the video overlay, if Pipe A is not running, Pipe B will not turn the first time it is activated. This patch restructures the code used for the video overlay to share it with the crtc commit function. commit ff8c8cb869a3c780dbd826f7c94f06e4f3fda6af Author: Keith Packard Date: Fri May 25 20:29:59 2007 -0700 Compute and clip to crtc before call to xf86XVClipVideoHelper. By clipping to the crtc ahead of time, xf86XVClipVideoHelper will correctly clip to the bounds of the crtc, eliminating the need for any custom crtc clipping. Also, replace the broken xf86XVFillKeyHelper with a private version that doesn't end up stuck with the wrong clip list when the root window changes size. commit 33f635d79fe891079558fd909d564f3cf424c482 Author: Keith Packard Date: Fri May 25 17:32:06 2007 -0700 Ensure Pipe A is active when enabling overlay the first time. The overlay on the i830 appears to be clocked by Pipe A when being enabled. If pipe A is not running, it will freeze the overlay and blank the screen. Setting a random mode on the Pipe and turning it on fixes this problem nicely. commit ff0ac8ea63dd6e55573652c5826c482881da0d62 Author: Keith Packard Date: Fri May 25 17:26:15 2007 -0700 Xv does not require directRendering commit 109e5d597b76f7414601cb39c07b133ebf1b0e61 Author: Keith Packard Date: Thu May 24 11:40:50 2007 -0700 Clean up CRTC selection. Remove more overlay blocking. Create separate CRTC selection function, use ints instead of floats for coverage measurement. Remove pipe stalls waiting for overlay update. commit dfb1ec9a07f74125cb1724d41ed4342c4714e12b Author: Keith Packard Date: Thu May 24 11:23:44 2007 -0700 Eliminate blocking for video overlay. No need to block for the overlay; just use the idle buffer. This will always work because the buffer switch occurs at vblank time, so there is always plenty of time to get the next buffer contents in place before it starts getting scanned to the screen. commit d172344599585e11388e59659dc9aaa86d7a31c1 Author: Keith Packard Date: Wed May 23 20:00:16 2007 -0700 Reformat i830_video.c to four-space indents. Yes, I can't stand it anymore. it's a huge patch, but git-diff -w shows no changes. commit 02935ced3fba598a01d908ae49ccc30cbcc765a8 Author: Keith Packard Date: Wed May 23 19:24:25 2007 -0700 Clean up overlay management. Create separate on/continue/off functions for overlay. Manage overlayOn boolean within those functions. Eliminate redundant management code in other routines. commit 08753f9b79f3f09879a18b552d90d88dbf52d4be Author: Keith Packard Date: Wed May 23 18:59:10 2007 -0700 Use FLIP_CONTINUE with ~OVERLAY_ENABLE to turn overlay off. This makes the overlay work on i830 with the modesetting driver. I don't know why the pre-modesetting driver worked without this, but it did. A more 'correct' fix would be welcome, but this does seem to do the trick. commit aa187186dc4f2d770a642060fe54f547ea8952b3 Author: Keith Packard Date: Mon May 21 10:49:08 2007 -0700 Automatically select crtc based on coverage. By default, select crtc based on which one covers more of the video output. pipe property can be used to override selection when both have partial coverage. commit f5017a06a271bba0ace3c5415b78e78bc0c96f22 Author: Keith Packard Date: Sun May 20 17:25:33 2007 -0700 Use CRTCs instead of pipe indices for video pipe selection. Replace pipe indices with crtc indices and store references to the xf86Crtc objects in the video private structure. commit 2df87256df755e972eb884bc742832038a020b2c Author: Keith Packard Date: Sun May 20 17:23:26 2007 -0700 Dump pending ring on crash. When the hardware locks up, dump the pending commands in the ring for analysis. commit 9971fac87622c93503540196e1756fded3d4869e Author: Michel Dänzer Date: Tue May 22 10:46:39 2007 +0200 i830: Provide new DRI texOffsetStart hook when available with EXA. commit 5390a2e2611950d3f48cc735df4a0c37bc5377a5 Author: Michel Dänzer Date: Fri May 18 17:51:36 2007 +0200 Update vblank pipes when a pipe gets disabled. commit 4120a20626998272424225261f2cf7960b7ec0ca Author: Wang Zhenyu Date: Fri May 18 10:10:34 2007 +0800 EXA: add render enter helper function That notify mesa rendering is smashing the state, and check last 3d operation to do sync after we're swapped in or others. commit 8db28aeaa6e908017b40bd9180f144a2972f6278 Author: Wang Zhenyu Date: Fri May 18 09:54:34 2007 +0800 Fix ring debug code Use proper unsigned type for timer variables, and try to dump 965G state. commit 16e50a91dd8b3676e8ce06052c549ab27e6843b7 Author: Wang Zhenyu Date: Fri May 18 09:53:36 2007 +0800 EXA: remove a flush cmd in i915 render code commit 12a9fcfe1b25cee850380d8ce11ef11cde9aaacb Merge: b930bb9 e89d5f2 Author: Keith Packard Date: Thu May 17 15:11:29 2007 -0700 Merge branch 'master' of git+ssh://git.freedesktop.org/git/xorg/driver/xf86-video-intel commit e89d5f275442915cc7777e75d3fcf7e7ed0f2084 Author: Keith Packard Date: Thu May 17 15:00:12 2007 -0700 Make each output control clones/crtcs. Split DVO into LVDS, TMDS, TV. Move clone/crtc config into each output where it's easier to understand (no need for a switch statement in I830PrepareOutputs. Also, split DVO into three sub-types (TMDS, LVDS, TVOUT) as those have different cloning abilities. commit a441954630c6cdabbf463bfc3404160f97a04b4f Author: Keith Packard Date: Thu May 17 14:11:49 2007 -0700 Enable panel fitter on ivch DVO. Using BIOS source code as a guide, set up the panel fitter on the ivch. This involves setting the pipe to the panel fixed mode, the DVO to the source size and assigning vertical and horizontal scaling factors in the ivch itself. commit c0daa0a982e7074af4b50653b4a45b0a6352b43d Author: Keith Packard Date: Wed May 16 14:02:00 2007 -0700 Change DVO module interface to pass more state across. Fix IVCH display. The DVO module interface reflected most of the xf86Output API to the underlying functions; finish that work given the changes that have since occurred in the xf86Output API. Move the LVDS-specific code into the IVCH module and make that work on the Thinkpad X30 (an i830-based laptop). Panel scaling does not work yet. commit b28817a87a1608e849e4a9a736dda43533a84b0c Author: Keith Packard Date: Wed May 16 13:59:36 2007 -0700 Add i830_bios_get_aim_data_block to read AIM data from BIOS Add-in modules have per-module data in the BIOS which contains configuration information which cannot be entirely discovered. commit b31bef1a8effa9acb6de7edd206b9d8c48d88144 Author: Keith Packard Date: Sat May 12 20:04:31 2007 -0700 Deal with i830 CRT load detection which cannot use FORCE_BORDER. Chips newer than the i830 can force the border color for the active period of the screen, allowing the load detection to easily see the right data. In addition, newer chips appear to have more sensible load detection hardware which either ignores inactive periods on the screen or performs some longer-term averaging. The i830 appears to provide unfiltered samples of the detected load. For the i830, then, emit a border at the bottom of the screen and, for load detection, simply turn it purple and wait for the current line to be within the border. Sample an entire scanline, counting the number of times the load detection sees a monitor. In my testing, the presence of a monitor will cause the detection to succeed every time, while the absense will cause it to fail about 75% of the time. The code here, checks for presence at least 75% of the time, which should be adequate. Also, as the new mode configuration code has already taken care to enable the CRT output, eliminate much of the load detection code which is simply duplicating functionality from the general mode setting code. This should result in faster load detection as this code will now run in no more than one frame time. It does burn the CPU the whole time though, polling the displayed scanline register. commit b930bb9d6da8c24dbe0949afb7bb2aa4bcb24687 Author: Eric Anholt Date: Thu May 3 13:44:12 2007 -0700 Disable vblank interrupts when no DRI clients are running. (cherry picked from commit 6621dd71ada839f4c1742e9e5b272e924cee21d9) commit 3b769af53e0ef6ef9b56afd679446c73a0e63ea5 Author: Eric Anholt Date: Thu May 3 13:44:12 2007 -0700 Disable vblank interrupts when no DRI clients are running. commit 775fc125aa7ecd0f054959ef210be2df4dc54345 Author: Dave Airlie Date: Thu May 3 20:58:50 2007 +1000 i810: be a bit more verbose about disabling DRI commit 34f362d099d255f8f0bb34e9de30f953ee770163 Author: Eric Anholt Date: Wed May 2 15:40:49 2007 -0700 Fix typo s/i/index/ in LoadPalette for depth 16. Reported by: Haihao Xiang commit f850d4727a2ad55c2116d0788f6684b2a0192d24 Author: Eric Anholt Date: Wed May 2 14:16:21 2007 -0700 Make up a fixed panel timing for DVO LVDS, and use DVOA for DVO LVDS. The fixed panel timing will only be available when the LVDS is already on at X startup. So far, our only mostly-working LVDS driver is for the i830, and on i830 the LVDS is always on DVOA, so use that for all LVDS chips. This may need to change if we support the ch7017 I've seen used on embedded i845, for example. commit f3168e3b0c5664a322ca6bb1c81fc94844cb30ab Author: Eric Anholt Date: Wed May 2 14:08:30 2007 -0700 Disable non-working GTT decoding on i830, and fix map/unmap of GTT. commit 1fc630f24f8ad9e304cb0761f9cacca2224203c4 Author: Eric Anholt Date: Wed May 2 13:29:21 2007 -0700 Add DVO[ABC] register debugging. commit d0ec37e9c0ceab1080700cd7be4a7cc58552d465 Author: Eric Anholt Date: Tue May 1 15:56:37 2007 -0700 Make the DVO output name LVDS if it's an LVDS chip. commit 490d05f99d2b62dd612d514d9ae0badbac9285ce Author: Eric Anholt Date: Tue May 1 15:47:01 2007 -0700 Fix typo in previous commit with s/XF86_DRI/XF86DRI/ commit c7bb34e83d7c459d932d01070cfeffbbf6c703ac Author: Dave Airlie Date: Wed May 2 14:25:39 2007 +1000 disable all outputs on EnterVT This disables all outputs on EnterVT as the SDVO output can confuse the VGA output if the BIOS has enabled it on the same pipe but X isn't going to use the SDVO. Worked out on irc with keithp commit cae0ae237b79fa7d3a82dfc8d3fb595ccb6c63e1 Author: Samuel Thibault Date: Tue May 1 12:41:18 2007 -0700 Bug #10714: Fix build without DRI. commit 6748d620fbf39dd98982856c09256bdec0fc82a1 Author: Eric Anholt Date: Mon Apr 30 17:27:23 2007 -0700 Ignore VideoRam now that its original purpose is obsolete. It had been necessary to allow more than a small amount of memory to be allocated, but now those old small allocations people had configured are getting in the way. commit a4f1a7872f6f959bb4bc6568face710bee3589de Author: Eric Anholt Date: Mon Apr 30 17:13:09 2007 -0700 Allow physical-memory allocations within stolen memory. Because stolen memory happens to be a contiguous block of high system memory, we can just read the GTT entries for it to get physical addresses for our allocations there if needed. This reduces fragmentation of the aperture space, and will often reclaim up to 7 MB of memory that had been left unused since the simplified aperture manager was put in place, but without reintroducing the complexities of the old aperture manager. commit 7d0d34cfdcc67d07e7667e13a9413743853134f8 Author: Eric Anholt Date: Mon Apr 30 10:39:54 2007 -0700 Disable some clock gating functions documented to work incorrectly. commit 138ac8f36cb4e4b3776f313955372522646acbb2 Author: Wang Zhenyu Date: Sun Apr 29 14:43:19 2007 +0800 Alloc state mem buffer on 965G for xaa rotation 965G needs state mem buffer to setup render pipeline. Thanks Barry Scrott for report this. commit 0cd524e5411e35c8483c02ecc5062625809e6fc6 Author: Kristian Høgsberg Date: Wed Apr 25 18:09:10 2007 -0400 Implement the custom I2C protocol used by the ivch DVO. The ihch DVO uses a modified I2C addressing scheme as described in section 5.2 of the data sheet. Implement this by over-riding the I2C read and write word routines. commit 880314aabe6326ed56517034940f0e10fb16e866 Author: Keith Packard Date: Tue Apr 24 11:37:08 2007 -0700 CRTC Rotation under XAA wasn't hitting accelerated path. The server rotation code is now using the root window in IncludeInferiors mode rather than using the screen pixmap. Change the XAA Composite code to check for this case now. commit b23eae55c8cdd73e0aba1bf7ced283d402ee6470 Merge: 31bf269 cebdb8b Author: Keith Packard Date: Thu Apr 19 20:38:18 2007 -0700 Merge branch 'origin' commit 31bf269afed0a830e79cbbd9d4b1ee9843af326c Author: Keith Packard Date: Thu Apr 19 20:03:41 2007 -0700 Update version to 2.0.0 commit cebdb8bfc6170a0fb441039f4422917fd0c77e70 Author: Wang Zhenyu Date: Fri Apr 20 10:54:34 2007 +0800 EXA: set enabling bits properly for i830 This was found when debug exa on a 865GV, we should set pipeline state bits properly, otherwise the engine will hang. commit 60e891915af7d0f522c9c3f966599fa07779f7aa Author: Keith Packard Date: Thu Apr 19 14:02:23 2007 -0700 Eliminate LinearAlloc option and code. With the fixes to the 2D frame buffer allocation that allows up to 65536 lines of 2D frame buffer in XAA mode, the old linear allocation hacks are no longer necessary. commit cca389769001c657435f056e1f1c26b0f52a48bd Author: Keith Packard Date: Thu Apr 19 13:15:23 2007 -0700 Clean up 1.2 xserver build support. Convert relative X server source path to absolute. Check for local copies of needed header files before building, rather than requiring server source. Remove extra duplicate -I elements in AM_CFLAGS in sub directories. commit 07797fee88d6be0dfb30394a419dd86f8a3c9095 Author: Keith Packard Date: Thu Apr 19 13:01:37 2007 -0700 Fix mismatching braces when XF86DRI_MM is not defined. A closing brace was left inside #ifdef XF86DRI_MM while the matching open brace was outside. commit 163c565527e8cda1f5a47c7fd63f04c80feaf3c7 Author: Keith Packard Date: Thu Apr 19 13:00:03 2007 -0700 Use I2C delay function instead of usleep. usleep isn't always available, and we have an existing delay mechanism available to use. commit 378ceea3d9ddbec7a08ac2f07f9a8cd9cf3cef36 Author: Wang Zhenyu Date: Thu Apr 19 17:30:28 2007 +0800 Fix mem list order and remove extra unbind call when free memory commit db4b9e18810990e8900bdf54aa3091b876ea2658 Author: Alan Hourihane Date: Wed Apr 18 13:52:08 2007 +0100 Fix return status commit 3f5111940e35989d334aa99cd1b0eb26293ebf1b Author: Alan Hourihane Date: Wed Apr 18 13:27:59 2007 +0100 Update read_response to include the try on PENDING status and remove it from get attached displays call. commit 902388fa06f85486fe8010807ab53e4926cc979a Author: Eric Anholt Date: Tue Apr 17 14:21:25 2007 -0700 Don't try to init the XAA linear region unless we allocated memory for it. Reported by JM Ibanez commit 8abecae202b609375b6754dbd5ecce3d59036daf Author: Eric Anholt Date: Tue Apr 17 12:28:43 2007 -0700 Fix i852 EXA Composite acceleration setup. Reported by JM Ibanez. commit 2dbe8d678b02b724c4f06255383f49bb4c2708b0 Author: Timo Aaltonen Date: Mon Apr 16 14:14:19 2007 -0700 Fix build against xserver 1.2. commit a089ac11beb4c801928c17780401e913bc0d5257 Author: Eric Anholt Date: Tue Apr 17 11:01:05 2007 -0700 Add all the possible ivch slave addresses (still commented out). commit 37ee68a95ca8c86ebe9abafaaf55b060dd2a2f73 Author: Eric Anholt Date: Tue Apr 17 10:50:02 2007 -0700 Bug #10438: Fix 965 XV when sourcing from less than the full image. Bob deinterlacing in MythTV, and the zoom options in totem would result in attempting to source from outside the video instead of scaling appropriately. commit ac9181c014638dbeb334b40b4029d0ccb2b7a0fc Author: Alan Hourihane Date: Tue Apr 17 16:30:17 2007 +0100 Check for the PENDING message when reading the attached displays. Ensures the command has completed before continuing. (probably need to check PENDING in other SDVO calls too) commit ab5bdee8a62c842ae32aaef57eb841ebcb644d2b Author: Wang Zhenyu Date: Tue Apr 17 16:23:46 2007 +0800 EXA: fix i830 render Fix tex blend pipeline in case that src/mask pict has no alpha. Unmask color buffer write disable bits. These make rendercheck run fine on 855GM. commit 1a29750b8dba1371d7d0802744cdf2f3bfa83c13 Author: Wang Zhenyu Date: Mon Apr 16 16:21:39 2007 +0800 EXA: fix 830/845G pict format Fallback in 830/845G when pict format is a8, x8r8g8b8 or x8b8g8r8. The hw doesn't support them. commit 3a634bbd198650c1597dec4306d99928374c30f3 Author: Wang Zhenyu Date: Mon Apr 16 15:14:49 2007 +0800 EXA: Add i830 supported pict format XRGB8888, XBGR8888 commit 3bcb9a0b4ba7f3df346b5708617a7aafcbe2490a Author: Wang Zhenyu Date: Mon Apr 16 14:27:49 2007 +0800 EXA: i830 render misc fix and cleanups Try to map texture stream when setup texture map, and use correct order in load_immediate_1 cmd, which fixed crash on 845GV. Also remove some flush cmds. commit 64c30cf896f8bde3ee74c92b970132ab91b418cd Author: Wang Zhenyu Date: Mon Apr 16 13:58:50 2007 +0800 Update intel.man with 965GM chipset support commit b67adb6de34cede0e31f02f26cd5ec7b1adfa586 Author: Wang Zhenyu Date: Fri Apr 13 22:34:52 2007 +0800 Fix crash on G965 XAA with LinearAlloc option We should alloc xaa_linear mem in LinearAlloc case, otherwise we get crash when initializing xf86 fb manager. commit b5b243e4120d6a048fb6bbe8814fe3184271e9d9 Author: Eric Anholt Date: Thu Apr 12 12:34:23 2007 -0700 Add license headers to 965 programs. commit 08cd5f9b0f086e51112008d50de48556372899f9 Author: Eric Anholt Date: Tue Apr 10 11:26:46 2007 -0700 Bug #10520: i810 manpage not installed despite users expecting it to be there. Install it as an alias to intel.4x, since we're letting people load the driver as "i810" still. commit bf9771e9711361632afe3abeeedca7ce03497005 Author: Eric Anholt Date: Mon Apr 9 15:26:05 2007 -0700 Don't use extended regexps in sed when not necessary. Some seds require a flag to enable extended regexps. commit 656fbd952542ba5ddc9b018071008ceb38b1bd19 Author: Brice Goglin Date: Mon Apr 9 13:06:46 2007 -0700 Bug #10515: Fix module version number with more than one digit per field. commit aa850d3c633a510d787ee59e538e26e5c3acbf4a Author: Eric Anholt Date: Mon Apr 9 11:44:25 2007 -0700 Turn off ALWAYS_SYNC in EXA. This slipped in as a debugging aid, and never got turned off. The driver appears to work fine without it on an i915 system, and for the non-default EXA option, we'd rather see issues found than continue running with debugging aids and hiding them behind bad performance. commit d2e75d8acde2541b85a0050d9e47182c1db7fc2c Author: Eric Anholt Date: Mon Apr 9 11:34:57 2007 -0700 Avoid a magic number and correct a (currently harmless) ifdef in ResetState(). commit 9ed446e12bd21b7c8222c6e63a03a8e2ec6b97e2 Author: Eric Anholt Date: Tue Apr 3 11:39:48 2007 -0700 It's 2007 now. Didn't I get the memo? commit d9d05b553ca1f98efc230ef13f34576122311ac1 Author: Eric Anholt Date: Tue Apr 3 00:15:02 2007 -0700 Bump version to 1.9.94 for 2.0 RC4. commit eb45315b07624cb0506d539f523728a651efb798 Author: Eric Anholt Date: Mon Apr 2 18:20:33 2007 -0700 Move README to being a plain old text file. commit 9b78208f1ef1a8ee98f2aa139956659169e234ab Author: Eric Anholt Date: Mon Apr 2 18:20:15 2007 -0700 Move a bunch of debugging verbosity under Option "ModeDebug". commit 9ea83d440d6b9e30d4627e34168226a6b4b2a841 Author: Eric Anholt Date: Mon Apr 2 11:39:27 2007 -0700 Update README source for new release notes. commit c6d0bf9fb6033339159d98dc2b9fe44759ad9de5 Author: Eric Anholt Date: Mon Apr 2 10:09:43 2007 -0700 Remove long-stale TODO file. commit c59e5895a0ea1f219c8f63a02b429a024bd1ed66 Author: Dave Airlie Date: Mon Apr 2 07:18:04 2007 +1000 make warning info only so ppl don't go reporting stupid bugs commit e119eaabb2ad1ffc1d83f18199f49c52ec71b0fe Author: Eric Anholt Date: Fri Mar 30 17:09:18 2007 -0700 Warnings cleanup in ch7xxx. commit fd98e88d1a6e0e8d35bda868e7afcb78d2d11524 Author: Eric Anholt Date: Fri Mar 30 17:07:29 2007 -0700 Finish off some function renaming that was missed before. Thanks to alanc for catching this. commit 274b99e5e750228eba9a67773725a40ebaa9bf56 Author: Eric Anholt Date: Fri Mar 30 17:06:43 2007 -0700 Don't forget the warnings CFLAGS in subdirectories. commit aa6a9abb34e93780c07563ab5b21d8c064ea0a1c Merge: 670c852 107ac12 Author: Keith Packard Date: Fri Mar 30 12:10:17 2007 -0700 Merge branch 'crestline-qa' commit 670c8521f18f01b9ea899ceb68d2a3dbb884b199 Author: Eric Anholt Date: Thu Mar 29 13:43:31 2007 -0700 Add missing file for old server compatibility build. commit 89c84f939f2dbed1d03a5f0e1daae23a229e47bc Author: Eric Anholt Date: Thu Mar 29 13:40:52 2007 -0700 Whitespace cleanup in Makefile. commit 6cd3b2c2de4596886aab85988b16cf63e8887add Author: Wang Zhenyu Date: Tue Mar 27 15:34:32 2007 +0800 EXA: i830 fix blend action i830_get_blend_cntl() has already added S8 offset. commit 35b3dd881a333569ce20ddf36b915b095757a1ba Author: Wang Zhenyu Date: Tue Mar 27 15:33:43 2007 +0800 EXA: fix i830 componentAlpha support Pick fix from i915 render, change tex blend pipeline for CA. commit df96d9a11d038ad64673022ce14c24223b3bd34c Author: Wang Zhenyu Date: Tue Mar 27 15:32:53 2007 +0800 EXA: fix i830 texture setup Use LOAD_IMM_2 helper cmd for tex setup. Enable RepeatNormal support. Fix A8 format, i830 can support it now. commit aa78e67553ffe8ca986330f0ee49b5414c71a71f Merge: 768821b 2191634 Author: Eric Anholt Date: Mon Mar 26 22:00:36 2007 -0700 Merge branch 'master' of git+ssh://git.freedesktop.org/git/xorg/driver/xf86-video-intel commit 107ac12867eda6b86212159db15c640d3490f2da Merge: 6548817 2191634 Author: Nian Wu Date: Tue Mar 27 12:51:45 2007 +0800 Merge git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit 768821bd3832aad01e44856199e60c6b7726a83b Author: Eric Anholt Date: Mon Mar 26 21:27:54 2007 -0700 Bump version to 1.9.93 for 2.0RC3. commit 2191634dd67b3219bb88f365bcf951d5a58140e8 Author: Alan Coopersmith Date: Mon Mar 26 14:58:41 2007 -0700 Add Solaris to list of OS'es with agpgart in man page commit 1c809f92467f97b098bcc89d66e60f6f88d9bade Author: Eric Anholt Date: Mon Mar 26 13:03:28 2007 -0700 Update .gitignore. commit f7befe50af4c13554d1f7aee6b05848ac312411b Author: Eric Anholt Date: Mon Mar 26 13:01:11 2007 -0700 Power on the LVDS B-channel pairs only when we've chosen dual-channel mode. It was basing off of the clock rate, but we have an override to use the existing dual channel state when we can detect it, so the two settings were conflicting. commit 827dc457a4fd0b97c95ec696ec9c3cd2f4a4becb Author: Eric Anholt Date: Mon Mar 26 12:45:23 2007 -0700 Update manpage to remove Rotate option and point at xorg.conf instead. While here, update a few other bits as well. commit 654881794ae1ad7214e85091b9015ae0fbdc5ddc Merge: 6de3edc d572771 Author: Nian Wu Date: Mon Mar 26 17:00:11 2007 +0800 Merge git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit d5727717c9141be28a69b1154ccd23c23207f8f6 Author: Keith Packard Date: Sun Mar 25 23:44:59 2007 -0700 Switch TV_FORMAT property to format XA_ATOM This allows the driver to report the set of valid formats in the property data. commit 6de3edcb52e6258f1af75e4f4bef73de1698445d Merge: d874aa3 1e6e675 Author: Nian Wu Date: Sat Mar 24 17:00:13 2007 +0800 Merge git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit 1e6e675524461ef0eb1983de89e2877426571a55 Author: Keith Packard Date: Fri Mar 23 23:42:52 2007 -0700 Eliminate calls to RRPostPendingProperty. RRPostPendingProperty has been removed in favor of RRPostPendingProperties, and that call is now managed outside of the driver. commit 28da6f4e307880326dd553f50fe3fff3b9be9f4f Author: Keith Packard Date: Fri Mar 23 14:36:42 2007 -0700 Record 3D state loss at EnterVT in last_3d value. last_3d set to LAST_3D_OTHER indicates that the 3D hardware has unknown state. commit ab19439cf4592e4607dc0bfc602aba3d56645a42 Author: Keith Packard Date: Fri Mar 23 14:34:27 2007 -0700 Don't call AdjustFrame on EnterVT. AdjustFrame is strictly for legacy compatibility; calling it on EnterVT wrecks crtc positions. commit d874aa31599da4777438cc51469afe9b66601f55 Merge: 94dbc37 26f32ef Author: Nian Wu Date: Fri Mar 23 17:00:12 2007 +0800 Merge git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit 26f32ef680a19e63af4b7c8c84141fe32263f298 Author: Keith Packard Date: Fri Mar 23 01:17:58 2007 -0700 Property size is in units, not bytes. Several places were using byte lengths instead of unit lengths for properties. commit f48dc501fbf74e0ac348f0b77750016597849ef4 Author: Keith Packard Date: Fri Mar 23 01:08:08 2007 -0700 TV output properties: TV_FORMAT, LEFT, TOP, RIGHT, BOTTOM. Remove TV format from mode name, instead use an explicit output property and split the input resolution from the tv format. Add properties to set the blank area on all four sides of the image. commit 94dbc3725358d63fe0ac8e6749489c993d24ede2 Merge: 28af380 20b2685 Author: Nian Wu Date: Thu Mar 22 17:00:15 2007 +0800 Merge git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit 20b26854abdacb6dc45cba2d81d515b2e47e25f1 Author: Wang Zhenyu Date: Thu Mar 22 10:33:15 2007 +0800 EXA: don't smash vertex header in G965 Originally we smashed vertex header to store texture coordinates, this is working as we only use sf/wm kernel and disable all other stages on pipeline. But better to not do this. This also cleans up vertex elements state and makes vertex buffer order looks "normal". commit 28af380ab133eb14d21dc650c77bdbab66576255 Merge: 300e893 e06c5f7 Author: Nian Wu Date: Wed Mar 21 17:00:04 2007 +0800 Merge branch 'crestline' of git://otc-graphics.jf.intel.com/git/xorg/driver/xf86-video-intel into crestline commit e06c5f727fb2e2de111ac9e691a877c56d2ca1b4 Merge: 0202399 3025fa0 Author: Wang Zhenyu Date: Wed Mar 21 15:04:21 2007 +0800 Merge branch 'master' into crestline commit 02023998663cc7f0735fadfb1719d93dc2e5a112 Author: Eric Anholt Date: Wed Mar 21 00:00:56 2007 -0700 Whitespace and symbolic reg names cleanup in i830_panel_fitter_pipe(). commit 3e9ec78b4f54defb9986e11e6f2ac3475755849d Author: Eric Anholt Date: Tue Mar 20 23:58:48 2007 -0700 Set the panel fitter to the right pipe on Crestline. commit 3025fa0fb2bf5ace7076796e45e2560fe8410e8d Author: Wang Zhenyu Date: Wed Mar 21 14:50:45 2007 +0800 EXA: try to enable rotation for G965 The new sf/wm should handle the texture sampling only in rotated case. Also fix possible hole in VUE slot. commit a50009604152bf9237c40bd098b3b1da3d018929 Merge: 0a612e7 2239448 Author: Eric Anholt Date: Tue Mar 20 23:32:19 2007 -0700 Merge branch 'master' into crestline Conflicts: src/i810_reg.h src/i830_display.c commit 300e893cec19dca48e00ee25014b8714dc13b278 Merge: 96e8699 0202399 Author: Nian Wu Date: Wed Mar 21 08:55:50 2007 +0800 Merge branch 'crestline' of git://otc-graphics.jf.intel.com/git/xorg/driver/xf86-video-intel into crestline commit 223944878cf38f86580df5a7d3102d86cfc061b9 Author: Eric Anholt Date: Tue Mar 20 14:33:53 2007 -0700 Attempt to fix single/dual-channel issues on i9xx LVDS panels. - Use the existing single/dual-channel state when available, as changing it doesn't appear to work out. - Set the power state of the CLKB and B0-B3 pairs according to whether choose to go dual-channel or not. - Restore the LVDS register at the appropriate point (before DPLLs are re-programmed. commit 96e86994f3d1b4938e99a751454ee99bebfe40d3 Merge: f465c23 0a612e7 Author: Nian Wu Date: Tue Mar 20 13:11:09 2007 +0800 Merge branch 'crestline' of git://otc-graphics.jf.intel.com/git/xorg/driver/xf86-video-intel into crestline commit 0a612e7115ff993bb8e9a00df13c0b0d20122fd6 Merge: 8bb6778 4c4faf2 Author: Wang Zhenyu Date: Tue Mar 20 11:34:40 2007 +0800 Merge branch 'master' of git://proxy.ims.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline Conflicts: src/i830_display.c Change LVDS output and postread like upstream. This might need to be retested on 965GM LVDS. commit 4c4faf260eb4dad1b1919c6168fa9ef477b98a39 Author: Eric Anholt Date: Mon Mar 19 13:36:37 2007 -0700 Set the CURSOR_SIZE register when present. Failure to do so gets you a lot of pretty colors. commit 64c14204453bea3f98d19861c450612e718e6c69 Author: Eric Anholt Date: Mon Mar 19 13:35:43 2007 -0700 Print the mode actually being set per pipe. commit 3ce802414a20ca8af128a00e6925a099dd90ceb4 Author: Eric Anholt Date: Mon Mar 19 11:35:11 2007 -0700 Add debug output for ADPA. commit c21b88d838fda1f00f6f6bcfe7855d32543c6f3f Author: Michel Dänzer Date: Mon Mar 19 10:45:58 2007 +0100 Fix build when DAMAGE is not defined. commit 991719c21a6cc1b5d9b7cbe30d4b333718b3e686 Author: Keith Packard Date: Sun Mar 18 23:05:33 2007 -0700 Make i830_sdvo_write_sdvox write everything twice. For some reason, certain chips don't correctly enable the SDVO hardware when this register is written only once. We're following what the BIOS code does and writing it twice now, but with extra posting reads to boot. Yes, this is cult-and-paste, but it fixes problems found on deployed hardware. commit 9118122a232d4cf7760bcb0874fe970c25251378 Author: Keith Packard Date: Sat Mar 17 21:34:03 2007 -0700 Allocate 4 separate buffers for HW Cursors on Linux. Linux cannot allocate a large fixed buffer for the HW cursors as needed for FreeBSD; instead, allocate four separate buffers. The code now prefers to allocate one buffer (less overhead) and falls back to separate buffers only when necessary. commit 62a5399d70ac3f8579441d617f8d80c94942a32a Author: Keith Packard Date: Sat Mar 17 21:32:36 2007 -0700 Elide I830DRIClipNotify for older DRI versions. I830DRIClipNotify is passed to newer versions of DRI; don't include it in the server when building against older versions. commit 05e0021147a89254182c277007236448f315231c Author: Keith Packard Date: Sat Mar 17 21:31:04 2007 -0700 Cast ARGB cursor address to CARD32 * to eliminate warning. While we're just doing a memcpy, it's nice for the two argument types to match. commit d05bb5362e986c9d27bc03c7e1a939ba28824810 Author: Keith Packard Date: Sat Mar 17 20:21:59 2007 -0700 Increase DDC I2C RiseFallTime to handle older monitors Changing this value slows the entire I2C bus down, making it far more reliable on older monitors. Note the same change has been made in the core X server code; this change is included here to ensure that older X servers work reliably with this driver. commit 44d1b544cec2e75735d2e27d66a9240317b962c2 Author: Keith Packard Date: Sat Mar 17 17:15:43 2007 -0700 Build fix for pre-XF86DRI_MM environments. Misplaced brace broke builds with older DRM libraries. commit a58befe9d243bd562cb4b2f08ec5c9f754148c20 Author: Keith Packard Date: Sat Mar 17 17:12:37 2007 -0700 Remove extra (and incorrect) I2C ByteTimeout setting. Setting the value correctly and then immediately breaking it caused many I2C transactions to timeout with slow monitors. Oops. commit 9d6d9ace4bd3180a4484321c3b96a83bc4adaf84 Author: Eric Anholt Date: Fri Mar 16 19:41:54 2007 -0700 Move vendor ID check in the utils to after pci_device_probe. Even current libpciaccess seems to require this. commit a117bc9e3eb824f2056ee6416859ea3ab1fdcc30 Author: Eric Anholt Date: Fri Mar 16 19:39:34 2007 -0700 Add the (afaict) correct ch7017 I2C slave address. commit 29446cdb4f72e12c5249b0d4b79ea56d9fe19934 Author: Eric Anholt Date: Fri Mar 16 19:18:37 2007 -0700 Fix compile failure due to needed types being in inttypes.h on Linux. commit 213394fbaaf353404cbb3aaa4c20860f48ee1079 Author: Eric Anholt Date: Thu Mar 15 19:13:28 2007 -0700 Fix crashes and other failures when a cursor allocation fails. Now, we allocate one single block of memory for cursors, and either succeed or fail once, rather than trying to support partial fallback modes that generally resulted in pain due to being untested. In particular, this fixes cursors on FreeBSD, which only allowed one large physically-contiguous allocation. commit 316ee682d342556b65cbd60409201591e916aac5 Author: Eric Anholt Date: Thu Mar 15 18:53:55 2007 -0700 Don't reload the cursors if we haven't set up the screen yet. This avoids a crash during preinit if we set a mode for load detecting. commit 8b06ab50bbd79dfaf4c90a6f76116ace64b85b77 Author: Eric Anholt Date: Thu Mar 15 10:12:14 2007 -0700 Fix sparse warnings about using 0 for NULL. commit 8ae6ad93329e2842c6f2d5b20ffeb0c14d10c0de Author: Keith Packard Date: Thu Mar 15 00:00:51 2007 -0700 Use new driver-independent CRTC-based cursor layer. This eliminates all of the cursor rotation code and other cursor management infrastructure, leaving a fairly simple hardware layer in its place. commit 9fbef2de9edbbed9c2f6a80c4074b9b245547c45 Author: Eric Anholt Date: Wed Mar 14 12:34:34 2007 -0700 Bump to 1.9.92 for RC2. commit 555b801a75cafa082808bc9bb683e700fc97d79a Author: Michel Dänzer Date: Wed Mar 14 11:41:50 2007 +0100 Make sure the legacy texture area is there when needed. This currently only matters when the DRM memory manager is not available and Option "Legacy3D" "off" is specified, but that hasn't always been the case and might change again in the future. commit 66fdb08c83d353fbe4e917900c54b555c869eb80 Author: Eric Anholt Date: Tue Mar 13 17:07:10 2007 -0700 Refine the i855 LVDS clock code. In particular, p2 is always 14. This gets correct clocks detected on most harware. The SSC is always assumed to be 66Mhz, which may not be true, but we'll fix that when we find example hardware. commit 44708bdd9ebfef0328302c9a964b80deb46e57c6 Author: Eric Anholt Date: Tue Mar 13 16:55:38 2007 -0700 Get SDVO DPMS working on the Mac Mini by writing SDVOB and SDVOC together. Also, add code for setting the encoder power state like the BIOS does, but this doesn't appear to work. We do much more than the BIOS does in powering things down, so perhaps that's interfering somehow. commit 5135b3a79f9c30ebce78c84c49846bba83607fed Author: Eric Anholt Date: Tue Mar 13 10:21:06 2007 -0700 Use a POSTING_READ(reg) macro instead of using the (void)INREG(reg) pattern. commit 578da7ca705c5a58c0bd397b0831e2f95140f8ae Author: Eric Anholt Date: Tue Mar 13 10:11:29 2007 -0700 Add PCI write posting protection to i2c putbits. commit cf33abe43bd95c9437fad8e6201a24084ff96cb8 Author: Eric Anholt Date: Tue Mar 13 10:08:57 2007 -0700 Add write posting protection for the SDVO DPMS-on path. commit 9d30f0007203157e6b82fa0ffc57324490eb2ca0 Author: Eric Anholt Date: Tue Mar 13 10:04:06 2007 -0700 Replace #if 0ed LVDS register setting with updated comment. commit 9c17c6e9c63563cad5edff837519a73fe0afe313 Author: Eric Anholt Date: Tue Mar 13 09:55:49 2007 -0700 Add write posting protections to relevant register writes in the mode-set path. commit 2824ec7ccbf44ba413a6133f735f4a548c73b3cd Author: Eric Anholt Date: Tue Mar 13 09:53:35 2007 -0700 Fix uninitialized string use in SDVO non-TMDS case. commit f465c23750adf908c0ea874f95aad98ebd2f1015 Merge: d33e8da 1ed3843 Author: Nian Wu Date: Tue Mar 13 17:00:44 2007 +0800 Merge git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit 1ed3843f73a0d8efa405daff3483ebe70bf6134f Author: Eric Anholt Date: Mon Mar 12 17:47:32 2007 -0700 Make the 965 use Y-major tiling for the depth buffer, as required by the spec. An example of the failure can be seen with the reflect demo when set to depth buffer mode. Reported by: Haihao Xiang commit 7aa257154685bd2520649ce87a3a84e55644d02c Author: Michel Dänzer Date: Mon Mar 12 19:00:14 2007 +0100 Fix build against released libdrm. commit 7c561956a28e90667fef140bc3cfa0edca464f15 Author: Michel Dänzer Date: Mon Mar 12 18:55:27 2007 +0100 Defer flipping pages back to normal until the end of the DRI block handler. Doing it earlier can result in the wrong page being visible, giving the appearance of a frozen X server. commit fe59ab9f562fd10118563d80eb3351a4d3b48b3f Author: Michel Dänzer Date: Mon Mar 12 13:03:47 2007 +0100 Disable page flipping if the DRM is older than 1.9. Older versions don't support the functionality we need. commit ba55ff15df974197bebd871e28bb96d817ae41c7 Author: Michel Dänzer Date: Mon Mar 12 13:01:00 2007 +0100 Fix attempt to flip pages back to normal when the last 3D window disappears. When this succeeds, 2D rendering does not have to be synchronized to back buffers until the next 3D window appears. commit c4a23c5ef8ce56ee0fe547fbc7c6623c021f801b Author: Michel Dänzer Date: Sat Mar 10 16:15:33 2007 +0100 Remove unused have3DWindows from pI830. commit d33e8daa6874ced978d2a1f687c48922555f2524 Merge: 58aeb87 797aa6f Author: Nian Wu Date: Mon Mar 12 09:03:52 2007 +0800 Merge git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit 797aa6fcb1231587bde1efb47bc8430c4c8d8110 Author: Dave Airlie Date: Sun Mar 11 12:58:50 2007 +1100 fixup missing assert includes commit 43a80ef9094efcb49027c83f0e726f907fecfbb2 Author: Dave Airlie Date: Sun Mar 11 12:58:26 2007 +1100 fixup brace alignment for older X.org commit 0d33fd3d03cef3a7e63d88ae441354390b37a937 Author: Dave Airlie Date: Sun Mar 11 12:58:02 2007 +1100 add XMODES flags to the i2c drivers commit 3c2d6e07bdf8daef6486b594aef0d22460eb2585 Author: Michel Dänzer Date: Fri Mar 9 23:49:46 2007 +0100 Don't crash when the SAREA pointer is NULL. commit 7518b8959ee7598f3526365a83ea7e143a5d6a4e Author: Michel Dänzer Date: Fri Mar 9 19:50:03 2007 +0100 Revert change accidentally included in previous commit. commit 1991a90ae90b388c914985d20d6f8c3637856e9a Author: Michel Dänzer Date: Fri Mar 9 19:47:13 2007 +0100 Update SAREA pipe sizes in i830_crtc_dpms instead of i830PipeSetBase. This allows setting the size to 0 when a pipe gets disabled. commit aef161853033907c6916337cebe88c8f111dd542 Merge: 1b54ee4 14ee919 Author: Michel Dänzer Date: Fri Mar 9 15:14:48 2007 +0100 Merge remote branch 'origin/master' into i830-pageflip commit 1b54ee493fb96f944aa81e56fa6f00a8067f87fc Merge: e972265 81722a2 Author: Michel Dänzer Date: Fri Mar 9 15:13:50 2007 +0100 Merge remote branch 'origin/master' into i830-pageflip Conflicts: src/i830_dri.c commit 14ee9195d203192d3f613919f230b20b900ffdba Author: Eric Anholt Date: Wed Mar 7 13:00:03 2007 -0800 Bug #10157: Fix cursor corruption on server regen. commit 58aeb87f2e335d15eba73ce9dd1982e867c52403 Merge: a4e7e81 81722a2 Author: Nian Wu Date: Wed Mar 7 16:02:03 2007 -0500 Merge git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit a5f0522b1d34236278861fe15bac2df099f0a2c7 Author: Alan Hourihane Date: Wed Mar 7 15:49:47 2007 +0000 Add some additional checks when XAA is enabled. commit 81722a21d232fa6cfb11fbe3d984abab50e89bcc Author: Keith Packard Date: Tue Mar 6 23:16:53 2007 -0800 Remove usage of 'shadow' module. Use xf86CrtcScreenInit. With the new mode setting code, rotation is handled outside of the driver, so the old usage of the 'shadow' module is no longer needed. Code to initialize the crtc structures has been moved out of the driver and into the modes code. commit 04f50961e2f1610c39e7e4b45811f2a6b517cad6 Author: Eric Anholt Date: Tue Mar 6 14:23:06 2007 -0800 Bug #9898: Fix a crash with NoAccel set. commit b07dfbba5df7728232b38211c623185116dcea5c Author: Eric Anholt Date: Tue Mar 6 13:59:14 2007 -0800 Remove leftover code that was disabling tiling after we set it up. commit 94c37f35872487c04136fb659526bffefd9c46ad Author: Eric Anholt Date: Tue Mar 6 13:57:04 2007 -0800 Make the depth buffer X tiled instead of Y, and fix a Y tiling nit on 945. The previous code claimed to set the depth buffer up as Y tiled, but due to lack of implementation in SetFence, it ended up being X tiled. Actually setting the Y tiling flag in the new version broke the depth buffer, so just switch the depth buffer to X tiling, which appears to work fine. commit a4e7e814a24dbe30a33e6ad45baeb41d190a2724 Merge: 862088b 4042b27 Author: Nian Wu Date: Tue Mar 6 16:01:40 2007 -0500 Merge git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit 30bb719ca0abc2599ffb89e59f297fa9a0a00c3c Author: Eric Anholt Date: Tue Mar 6 12:23:43 2007 -0800 Continue to allocate the legacy texture pool by default. This is a partial revert of 7358642e64ab6d13bc1dc1a44703ee66d715ff61 If we don't allocate it now, when the DRM version is too low there won't be any memory allocated and DRI will fail. Instead, waste the memory in the i915tex case for now, and leave fixing it right (check DRM version up front and decide which memory manager to set up) to later. commit e972265261c421268e4fb806e587378d0adec577 Author: Michel Dänzer Date: Tue Mar 6 16:54:16 2007 +0100 Fix build failure. Not sure how I missed this before... Thanks to Todd Merrill for reporting. commit 862088ba6a91d3e7cf8d37126b1d9f4ee03a1f73 Merge: 11e9158 d717d9d Author: Nian Wu Date: Tue Mar 6 07:43:16 2007 -0500 Merge git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit 1e1b45fa6ed1683cba4ae73ac98933e74c3ab9d6 Author: Michel Dänzer Date: Tue Mar 6 10:28:41 2007 +0100 Fix reduced DRI memory manager size. pI830->mmSize is in kB. commit 3c08bc7d6974a1a6cf5f9cb81898617032966c52 Author: Michel Dänzer Date: Tue Mar 6 10:14:47 2007 +0100 Fix some conditionals related to triple buffering. Guard code that dereferences pI830->third_buffer with tests for that instead of pI830->TripleBuffer. It could happen that we want to enable triple buffering but (temporarily) can't because the third buffer couldn't be allocated. commit c25cfafbe1eb380b58b5fc16e94f5cc6f422f0cd Author: Michel Dänzer Date: Tue Mar 6 10:00:12 2007 +0100 Unify allocation of back buffers. commit e787d7b698d320a7c45df35d58c5113413561fe2 Author: Michel Dänzer Date: Tue Mar 6 09:35:42 2007 +0100 Remove warnings about potential artifacts with page flipping and mixed 2D/3D. The artifacts only seemed to occur when EXA was falling back to software for the front buffer. commit 6c299aea8e87e72c68c96c03594706c976601ede Merge: 3308152 4042b27 Author: Michel Dänzer Date: Tue Mar 6 09:28:52 2007 +0100 Merge remote branch 'origin/master' into i830-pageflip Conflicts: src/i830_display.c src/i830_driver.c commit 4042b27f01fdb94e7fc0d4e991e054fff88479ea Author: Keith Packard Date: Mon Mar 5 22:32:52 2007 -0800 Move EnterVT mode setting code to xf86SetDesiredModes. Make the application of crtc desiredModes generic code instead of per-driver by creating xf86SetDesiredModes from the code that was in EnterVT and calling it. Also, move the frame buffer clear until just before mode setting to make sure things are mapped correctly. commit 55ee46aebbf1ec1a8ce914fbd0c8894fc857db8f Merge: 5c72014 50aa094 Author: Wang Zhenyu Date: Tue Mar 6 12:45:08 2007 +0800 Merge branch 'master' of git://proxy.ims.intel.com:9419/git/xorg/driver/xf86-video-intel commit 5c720147e2b86ca4046b7c3812c1ca6b0fb78c9d Author: Wang Zhenyu Date: Tue Mar 6 12:44:04 2007 +0800 EXA: try to always alloc exa i965 state buffer in stolen mem I think this is a safe way to work around any possible chip error. commit 11e91586169b4f4bf2b41c0e0e620a69670c2f07 Merge: a24962a 0150270 Author: Nian Wu Date: Mon Mar 5 16:00:20 2007 -0500 Merge git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit a24962af9ed39fabca0152cae1265a29fe6237d3 Merge: 35e9310 8bb6778 Author: Nian Wu Date: Mon Mar 5 16:00:11 2007 -0500 Merge branch 'crestline' of git://otc-graphics.jf.intel.com/git/xorg/driver/xf86-video-intel into crestline commit 50aa09425f54c4eeca7f8b0fae9579209b10b9c4 Merge: d717d9d c2c6255 Author: Keith Packard Date: Mon Mar 5 11:28:46 2007 -0800 Merge branch 'modesetting' commit c2c62559e702e7de1fa2ef309fa647ab13564dc3 Author: Keith Packard Date: Sat Mar 3 23:12:54 2007 -0800 Move single mode setting code to X server. Code to drive the global configuration from a single mode setting operation (from RandR 1.1, XFree86-VidModeExtension or XFree86-DGA) has been included in the X server now, so remove it from this driver. commit 35e9310ef59873877422dcaf9a65b38789fc8ad6 Merge: c8e6f0b b27fa2c Author: Nian Wu Date: Mon Mar 5 09:01:55 2007 -0500 Merge branch 'modesetting' of git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit 8bb677889d3f71cde671f17a3589939acad2c3b3 Author: Eric Anholt Date: Mon Mar 5 05:08:51 2007 -0800 Fix 965GM SDVO by not setting fields in SDVO[BC] which have new meanings. commit 330815251dee808754b328cd2fd60db39f2460eb Merge: 50ba1ff 0150270 Author: Michel Dänzer Date: Mon Mar 5 13:05:20 2007 +0100 Merge remote branch 'origin/master' into i830-pageflip Conflicts: src/i830.h src/i830_dri.c src/i830_dri.h src/i830_driver.c src/i830_memory.c src/i830_xaa.c commit d717d9d566fe3c0866b06840114e1c1990bd7be0 Author: Michel Dänzer Date: Mon Mar 5 12:57:21 2007 +0100 Fix DRM memory manager initialization. It takes the offset and size in pages, not bytes. commit a0c83af3430b6705ab2ecae59085d1c74e890c19 Merge: c0f99b4 bc20b54 Author: Eric Anholt Date: Mon Mar 5 03:37:53 2007 -0800 Merge branch 'modesetting' into crestline commit 7358642e64ab6d13bc1dc1a44703ee66d715ff61 Author: Michel Dänzer Date: Mon Mar 5 11:53:09 2007 +0100 Fix handling of new vs. old texture pools. Only one of them can be active. commit 50ba1fff886a7f51b178ac6d3a1ba79a3014b214 Author: Michel Dänzer Date: Mon Mar 5 10:22:07 2007 +0100 Be more verbose when page flipping can't be enabled for various reasons. commit 015027034e970f1e3bb6ab239f7e0119235e404f Merge: 9a51064 bc20b54 Author: Eric Anholt Date: Sun Mar 4 21:49:00 2007 -0800 Merge branch 'modesetting' Conflicts: configure.ac src/i830_driver.c src/i830_modes.c src/i830_video.c commit af565872a49a1a464ee4154c27136660b184c4c1 Author: Keith Packard Date: Sun Mar 4 21:20:33 2007 -0800 Set version number to 2.0 RC1 (1.9.91). commit bc20b54c34088356a277beaebcc90bb4a7063e19 Author: Keith Packard Date: Sun Mar 4 19:35:44 2007 -0800 Support new CRTC/Output prepare/commit hooks. New hooks replace explicit invocation of DPMS functions to manage mode setting sequencing. commit b27fa2c257ccc49c6f29a20a3e672ebaaf58e7aa Author: Eric Anholt Date: Sun Mar 4 17:15:45 2007 -0800 Fix a crash with XAA and DRI disabled after the allocation rework. commit fc7d43c2b6b98be597152fbd88024273edd931d0 Author: Eric Anholt Date: Sun Mar 4 17:14:30 2007 -0800 Use -f for ln of server source. Otherwise, changing paths didn't take effect. commit b7b6063e62927dd135e118c433f48b4d0b5fe246 Author: Eric Anholt Date: Sun Mar 4 15:53:50 2007 -0800 Add a little BIOS dumper program. This isn't actually tested yet as libpciaccess doesn't do rom access on non-linux yet. commit 6aeb855b9d4e03d458773240a9c0c9c1edd3e403 Author: Keith Packard Date: Sun Mar 4 17:11:40 2007 -0800 Shadow pixmaps use fixed offset of 0. Shadow pixmap contents are copied from the main frame buffer and contain only the bits displayed by the crtc, not the whole frame buffer. The crtc origin is always interpreted by the shadow copying code, the frame buffer itself doesn't move. commit c8e6f0bac13e0e407f9771b1bbf126f023c1511b Merge: 0ed5a2f c0f99b4 Author: Nian Wu Date: Sun Mar 4 09:00:12 2007 -0500 Merge branch 'crestline' of git://otc-graphics.jf.intel.com/git/xorg/driver/xf86-video-intel into crestline commit 7d08f720a95499e8d00dcd7174f06731de420017 Merge: dd24d2a 435d355 Author: Keith Packard Date: Sat Mar 3 23:13:03 2007 -0800 Merge branch 'modesetting-origin' into modesetting commit 435d35558d8135a2c1724dfd813fb4a9a4e14178 Author: Keith Packard Date: Sat Mar 3 22:46:31 2007 -0800 Update for Aaron's xserver changes. Add xf86CrtcConfigFuncs to xf86CrtcConfigInit Add canGrow option to xf86InitialConfiguration commit c0f99b4962553e560a5cb882a5060d95db5477a4 Author: Keith Packard Date: Sat Mar 3 22:36:46 2007 -0800 LVDS dither control moved from PFIT to LVDS register for Crestline The LVDS register now contains lots of new controls for dual-channel LVDS control along with dither enabling. The PFIT register has a lot fewer controls as a result. commit 0fa3d4f51b5ee0dba3882fd74b6ac4e7da708f8f Merge: 10655c4 fd52d63 Author: Keith Packard Date: Sat Mar 3 21:02:17 2007 -0800 Merge branch 'modesetting' into crestline commit 0ed5a2fdcfe9674175aa9af155c20695ef41cadc Merge: 7ba80cc fd52d63 Author: Nian Wu Date: Sat Mar 3 09:00:20 2007 -0500 Merge branch 'modesetting' of git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit fd52d635603b7093c5a2b7fa9c987cf59f9be27c Author: Eric Anholt Date: Fri Mar 2 13:44:57 2007 -0800 Add a WIP UploadToScreen implementation. This almost displays right. commit ca0fa875e8bb5cb778d4db7d8053ec0a5ae34ef4 Author: Eric Anholt Date: Fri Mar 2 11:46:45 2007 -0800 Move EXA offscreen memory back after framebuffer to avoid fallbacks. Without this, the EXA code decides that the framebuffer is not in video memory because it's not located between the start of video memory and the end of EXA allocator memory. We should either have EXA let the driver decide if a pointer is in video memory, or add a new field for "end of video memory" (not just end of EXA offscreen allocation), but this fixes things for now. commit 5a73624b9bcce0004e5b72c74667c9dd504b9c3e Author: Eric Anholt Date: Fri Mar 2 11:03:16 2007 -0800 Fix a crash when an ARGB cursor allocation had failed. commit 7ba80cc3a5e0f469e3ee55c7537fdc952cd85911 Merge: 9796cb7 d5df52b Author: Nian Wu Date: Fri Mar 2 09:01:46 2007 -0500 Merge branch 'modesetting' of git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit d5df52be59a7dd950e73336ce5698e73480108b0 Author: Eric Anholt Date: Thu Mar 1 08:35:13 2007 -0800 Actually add the files for reg_dumper. commit 9796cb7ac4bd74d5336986bb194ff5875b028121 Merge: 42deb1e 1f5d166 Author: Nian Wu Date: Thu Mar 1 09:02:22 2007 -0500 Merge branch 'modesetting' of git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit 1f5d1666c8386ca4597c6f2c1ec239f9d821da4c Author: Eric Anholt Date: Wed Feb 28 16:27:55 2007 -0800 Add a non-installed command line tool using libpciaccess to dump registers. This reuses the i830_debug.c code, so we can run that from the console or from the BIOS-based X server to debug some remaining issues. commit cd1d4b398ec91d551bdaaa26c769e5e6a9442df1 Author: Eric Anholt Date: Wed Feb 28 09:47:38 2007 -0800 Many fixes to mode_get, mode_set, clock limits, and register dumps on i855. This should fix a number of issues with i855s, particularly with integrated LVDS panels. commit 36cad3fcb65e3dcd88e58e301cd60adb121cb96b Author: Michel Dänzer Date: Wed Feb 28 17:52:42 2007 +0100 Update vblank pipe setup when setting a mode. commit b5316fb2623e9630cbd58020e0a7c95bf354c587 Author: Michel Dänzer Date: Wed Feb 28 17:50:56 2007 +0100 I830DRIClipNotify: Fix initialization of crtcBox[1].[xy]2. commit 42deb1e10fc445d152b74e2b225daeaee2abb3d1 Merge: 76aac38 3e8e75e Author: Nian Wu Date: Wed Feb 28 09:00:24 2007 -0500 Merge branch 'modesetting' of git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit 76aac382659bc1dab7dca078f703b9cf48b4ad74 Merge: 42fc06c 1ac83f5 Author: Nian Wu Date: Tue Feb 27 14:42:42 2007 -0500 Merge branch 'modesetting' of git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit 3e8e75e5d83a2fa7e9fc6e9a3fbb07dac548ea5a Author: Wang Zhenyu Date: Tue Feb 27 16:48:48 2007 +0800 Revert "EXA: enlarge 965 exa state buffer size" This reverts commit 1ac83f51ee46d65237eae1b5c767e106ac9e29bc. oops, current exa state + kernel program size + scratch space is actually ~36KB. commit 1ac83f51ee46d65237eae1b5c767e106ac9e29bc Author: Wang Zhenyu Date: Tue Feb 27 13:58:47 2007 +0800 EXA: enlarge 965 exa state buffer size Origin size seems break all render, be sure to assign enough mem. 80k works fine for me. commit acd9f3e28fcf5e91e09d8bdca4cc58681b1fc92a Author: Wang Zhenyu Date: Tue Feb 27 13:56:31 2007 +0800 Fix typo when assign overlay reg mem pointer commit 5b765065c0bdf671b60d8a2fe512f6307f753eab Author: Eric Anholt Date: Mon Feb 26 17:25:57 2007 -0800 Consolidate shared structure allocation to the start of ScreenInit. Previously, the structures would get allocated unnecessarily once, and wouldn't get reallocated in time at server regen. commit 6ed28c5aa34ff1d8b99a8d4721ddac7b9cac41ed Author: Eric Anholt Date: Mon Feb 26 17:12:44 2007 -0800 Fix XV issues on 965 after the allocation rework commit 36bfeb890aeb17a8ee058790dd69b4e9aba644a1 Author: Eric Anholt Date: Mon Feb 26 16:58:33 2007 -0800 Move the initial EnterVT to after we've set up acceleration. This fixes a crash with initial rotation and EXA due to EXA not being initialized in time for the shadow pixmap setup. commit 732885c08daaf17034da8f4855d0b957ec3df9d7 Author: Eric Anholt Date: Mon Feb 26 15:04:56 2007 -0800 Bug #9604: Align the sizes of allocations to page increments as well. Without this, the 965 DRI driver fell over when pitch * height wasn't page-size aligned. Since the allocator only allocates at page-aligned offsets anyway this shouldn't hurt us at all. commit 0bfaeaab2838184827236c5c0fcc17f06d9e1372 Author: Eric Anholt Date: Mon Feb 26 11:11:56 2007 -0800 Fix physical memory allocations, including description in log. commit 5732dd7221b7bb24bf21e16066f014a0c7cb6b2a Author: Eric Anholt Date: Mon Feb 26 09:54:04 2007 -0800 Fix some texture memory allocation issues. commit aeeae1b40fc012bcc01fcd2243d4993de819b7cf Author: Eric Anholt Date: Mon Feb 26 09:13:17 2007 -0800 Fix a crash in startup monitor detect. commit 10655c4674cdac8a231c50dd9afc5d43fe43b4bd Author: Zou Nan hai Date: Mon Feb 26 09:23:14 2007 +0800 for merge with external tree commit 9417af8630fce0476273b693ebac12129415e0cf Author: Zou Nan hai Date: Mon Feb 26 08:02:50 2007 +0800 fix conflict with external branch commit 42fc06c85a0c487cfa26b9d4aa396402dbc4c2fe Merge: 10655c4 3bce8bf Author: Nian Wu Date: Sun Feb 25 12:53:47 2007 -0800 Merge branch 'modesetting' of git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit 84a7429730ddd29072d696accb9fdd8fea40731a Author: Eric Anholt Date: Fri Feb 23 14:59:24 2007 -0800 Always allocate logical_context, since we use it in emitting invariants. commit 7bb725dee9ef0ccdc478ed4931dc267bcb80203f Author: Eric Anholt Date: Fri Feb 23 14:56:58 2007 -0800 Improve logging of memory allocation passes, and clean the code up a bit. commit 54198b26e87baa180208932b97f10b27d9c0fddc Author: Eric Anholt Date: Fri Feb 23 14:17:07 2007 -0800 Move memory allocation to ScreenInit rather than PreInit. This lets us do memory allocation just once rather than having several passes (as long as things succeed), avoids trouble with zaphod mode, and will let us do better automatic sizing of allocations soon. commit 828deff6453872ef0f9fa35c08c9f95f464b1785 Author: Eric Anholt Date: Fri Feb 23 13:16:02 2007 -0800 Fix fence allocations on 965. commit d3ac440e127c8a7716062852cd3b5e7943e289e4 Author: Eric Anholt Date: Thu Feb 22 10:41:09 2007 -0800 Rework the video memory allocation. The previous allocator worked in multiple passes, with (at least) one of setting up allocations, another to attempt to adjust those for tiling, and then a pass to set up the offsets and fix them in memory. The new allocator is simpler, allocating memory immediately if possible, setting up tiling up front, and choosing offsets immediately. AGP memory is only allocated to back actual memory used, saving some memory that would have been allocated for padding previous. It will also allow dynamic freeing and reallocation of memory, which will be useful for framebuffer resizing. commit a61a6b1db610a07060d0dcca54b66a4b2b3686cc Author: Eric Anholt Date: Wed Feb 21 09:43:24 2007 -0800 Reformat to 4-space indents, no trailing whitespace, and 80 columns. I'm in the process of rewriting this file, and wanted to separate my whitespace changes. commit dd24d2a082b41abc466391c37a9b3d066586dc55 Author: Keith Packard Date: Thu Feb 22 11:46:44 2007 -0800 Remove copied X server headers from driver. These two headers were copied during mode setting development; as that code is now inside the server, these files are no longer needed here. commit 681966ce6548c6f47b087eb7e369674b7ddbaade Author: Keith Packard Date: Thu Feb 22 11:45:43 2007 -0800 Disable unused functions after configuring the remaining resources. This eliminates an unsightly flash of the hardware cursor during X server startup. commit f5ab9e0ad3a65c972861dd53be6e33e1aac13191 Author: Michel Dänzer Date: Thu Feb 22 18:57:23 2007 +0100 I830DRISwapContext: Some cleanups. Also call I830EmitFlush() when entering/leaving the server context. Doesn't seem to help for artifacts with page flipping and mixed 2D/3D unfortunately. commit 8cce74f195e5ef6ce5599ecd52f35bcfcb0d7913 Author: Michel Dänzer Date: Thu Feb 22 18:27:59 2007 +0100 Do not sync DRI windows between pages when possible. commit f346549183a407a4bff6521ec2450dacb97442e9 Author: Michel Dänzer Date: Wed Feb 21 09:19:54 2007 +0100 intel.man: Fix typo. commit 3bce8bf0e95e5842399959a5d6f6413e96f03adb Author: Keith Packard Date: Tue Feb 20 23:10:17 2007 -0800 Pre-965 hardware had fixed pipe 1 for panel fitter. The pipe mode setting code needs to disable the panel fitter when using the pipe for things other than LVDS output. The driver was checking for panel fitter conflicts using bits that the 965 chipset defines for selecting which pipe the panel fitter is connected to. However, on pre-965 hardware, the panel fitter works only with pipe 1 and those bits returned 0. The result was that when pipe 1 was using the panel fitter, configuring pipe 0 would disable the panel fitter. The fix provided uses a model-specific test for the panel fitter pipe. commit 0bee64f4bc7581de7ab28ca438581d215e85c610 Author: Michel Dänzer Date: Tue Feb 20 19:09:37 2007 +0100 Add support for triple buffering using a third static buffer. Need to bump the DRI DDX version minor for the added SAREA fields. commit 2212baa8454abb4c7948c3f2e20e337f831d1b86 Author: Michel Dänzer Date: Tue Feb 20 18:39:13 2007 +0100 Don't mark page flipping as active when it really isn't. commit fac8f2eb49de22abd853ca75f3c985d4d94e3633 Author: Michel Dänzer Date: Tue Feb 20 18:38:40 2007 +0100 Document Option "PageFlip". commit 5a1a04649c62aa4b1c0617560b813642ce0c67b5 Author: Keith Packard Date: Sat Feb 17 17:01:04 2007 -0800 Make distcheck fixes. Distribute all sources needed even for older servers. Always allow (but do not require) link to server sources so that needed files can be included in the generated tar files. Add remaining .g4a files and assembly output to distributed file lists. commit 0ca926e63c7fa9122a8cba1424ce4cdee1c7b215 Author: Keith Packard Date: Fri Feb 16 02:19:10 2007 -0800 Check for presence of xf86Modes.h file. xf86Modes.h file signals the availability of the new modes API in the server; use that instead of counting on X server version numbers. Also, finish eliminating use of local copies of those header files. commit 972f1109d23c0a3f821fa573e2aa40cd17ce2795 Author: Keith Packard Date: Fri Feb 16 01:07:20 2007 -0800 Remove common code that has moved to xserver/hw/xfree86/modes commit 20e146a09e3b4907a6800588ebfa787bfc5b0d89 Author: Keith Packard Date: Fri Feb 16 00:57:44 2007 -0800 Change rotation pixmap creation API to permit start-time rotation. Start time rotation requires that the pixmap be created after the server has initialized the screens. Delay the pixmap creation until the first block handler invocation. commit 17e38e39a49b2a96e31cca4a0b922631c5f7879e Author: Keith Packard Date: Thu Feb 15 23:09:43 2007 -0800 Enable mode setting function renames from shared code. commit 8e7bc4b1f95f5d62ff18cd94f14f5e3949ef6a29 Author: Keith Packard Date: Thu Feb 15 11:17:53 2007 -0800 Use server sources for crtc/output generic code. As sources for these have been merged to the server, use them from the server source code instead of keeping copies in this driver. commit 6e33c4c44030e05d6d07ca9274cfe4d6df2c0387 Author: Eric Anholt Date: Thu Feb 15 09:54:43 2007 -0800 Print the correct meaning of bit 30 of pipeconf for 965 in debug output. commit 4541150cdce3f2f83b8530d0863aecf28c6610a9 Author: Eric Anholt Date: Thu Feb 15 09:51:04 2007 -0800 Detect core clock frequencies, to avoid double-wide mode when possible. Additionally, don't attempt to set double-wide on the 965, where there is no such thing any more (not that we'd ever see modes high enough to trigger it). commit d1be0dc4e9d443dcbd218a083cfb21a0b5d634ee Author: Wang Zhenyu Date: Thu Feb 15 11:03:12 2007 +0800 EXA: fix i965 default color (0,0,0,1.0) -> (0,0,0,0) (0,0,0,1.0) is actually black... commit c010d28f0a91a9831d366ea490878ef3d958021f Author: Wang Zhenyu Date: Thu Feb 15 11:00:06 2007 +0800 EXA: mask componentAlpha support we're getting load of multiple sf/wm programs for i965, might need to be cleanup in future. commit ec55dd16c4c5f7612d33ae22a6249b3b1c60e2b6 Author: Keith Packard Date: Wed Feb 14 14:37:02 2007 -0800 Global namespace cleanups, filename fixes. Ensure all xf86 symbols created here are protected with XF86NAME. Remove accidentally exported symbols from namespace. Make all to-be-DI files prefixed with i830_xf86. commit 22a271555a46267c40448fa926d45692498ef7c6 Author: Keith Packard Date: Wed Feb 14 14:17:14 2007 -0800 Create driver-independent DGA hooks, remove i830 DGA code. As RandR needs to poke at DGA code, and we want the RandR code to be driver-independent, it seemed easier to just make the DGA code driver-independent as well. commit a7c64d5e28f608d54c94fa7a5c92bd5b471179c7 Author: root Date: Wed Feb 14 10:20:28 2007 -0800 WIP DGA generic support. commit 55c7b017ed718107b9b11467030addf225728b84 Author: Michel Dänzer Date: Wed Feb 14 16:45:22 2007 +0100 Use new DRI ClipNotify hook to track page flipping per CRTC when available. commit 3bd8edb95fcbbb49e97a1a2bdd5c8eae4f83d3a8 Author: Michel Dänzer Date: Fri Feb 2 17:41:18 2007 +0100 Revive i830 page flipping support. Use the damage layer directly instead of via shadowfb. commit 1d22bad33b634aaf61976907cb2ceeec8a9ebe4d Author: Michel Dänzer Date: Wed Feb 14 12:57:12 2007 +0100 Update SAREA pipe fields in i830PipeSetBase. commit 0df490938af10d3a748221e2fb5467b0f244f9eb Author: Michel Dänzer Date: Thu Dec 21 11:27:28 2006 +0100 If the DRM can handle it, enable vertical blank interrupts for both pipes. commit 1623b47c4863f5362587cdac855e3e11c109b007 Author: Eric Anholt Date: Tue Feb 13 17:37:56 2007 -0800 Switch the RECTLIST vertex order to (x2, y2), (x1, y2), (x1, y1). This is the documented correct ordering, and while the previous ordering (reversed) worked on some hardware, it failed on others. Reported by: Wang Zhenyu commit 4bd4058dfdca84d6ebab82a1dfba99adf3852c9c Author: Eric Anholt Date: Tue Feb 13 14:29:52 2007 -0800 Bug #9941: ViewSonic VX2025wm sync polarity quirk to fix mode distortion. commit 0ec4ffcb4e8e5e707e68bef1b13a324677f19f8f Author: Eric Anholt Date: Tue Feb 13 13:18:15 2007 -0800 Fix double free in the case of LVDS disabled through quirks. commit 699fa88d2570de6173e2d866e11b7437f4842d74 Merge: 681b919 6641aec Author: Eric Anholt Date: Tue Feb 13 13:10:08 2007 -0800 Merge branch 'modesetting-origin' into modesetting commit 6641aec0a1cbc869fba1956c556cdd204631545a Author: Eric Anholt Date: Tue Feb 13 10:21:12 2007 -0800 Attempt to detect panel fixed mode from EDID or current programmed mode. These two sources are placed in higher priority to the BIOS data when available, since the BIOS data has proven unreliable. The BIOS data is still read, and warnings printed if it doesn't match what we probe. The BIOS data remains useful for the situation where we want to turn on LVDS but there is no EDID available and no current mode programmed (i.e. booting with VGA or TV connected). commit 991439d4c78cf5b2a8f6bb8f5b36fffbfcc4e4fc Author: Eric Anholt Date: Tue Feb 13 10:07:47 2007 -0800 Remove dead #if 0-ed structure. commit c3aed56d46baba057d83dc6ea12c6b4e705e54ba Author: Eric Anholt Date: Fri Feb 9 13:16:37 2007 -0800 Fix comments on 965 pitch alignment restrictions. commit 98318dfcc99264fb7b7ef34568712812aa6300b6 Author: Daniel Stone Date: Fri Feb 9 16:20:57 2007 +0200 add missing files to fix distcheck A bunch of files were missing from _SOURCES, and the build was also broken for missing files (from EXTRA_DIST) when building without gen4asm. commit fce4390c4f1a76c65423114bb479814787612cd7 Author: Zou Nan hai Date: Fri Feb 9 11:32:22 2007 +0800 TV Enable commit 00dadaa115951c36de9caf2a9d78909811ef34bc Author: Zou Nan hai Date: Fri Feb 9 11:14:50 2007 +0800 set DPLL before detect commit 10712c7a781e50475ee7e8ccd5bfd2d820f863e2 Author: Wang Zhenyu Date: Thu Feb 8 08:56:42 2007 +0800 EXA: i965 pixmap pitch align This seems like a typo, which causes screen artifacts. commit 44eacf2323454e26b535cc5a4f0789cb0ff0e7fb Author: Wang Zhenyu Date: Wed Feb 7 17:30:51 2007 +0800 EXA: fix render issue with i965 Fix SF kernel with corrent coeffient work, and correct VUE storage in multi texture case. commit 920aeaf8478f706ab3b1de4f20f5d351e64a0197 Author: Zou Nan hai Date: Wed Feb 7 13:15:52 2007 +0800 Make TV code same with external tree. Fallback to old filter table. 4 spaces indent reformat. Auto scaling seem to not work well on some low res interlace mode when input width > 1024, filter those modes. Fix some subcarriar number commit 6d9757e466863594300c83f3806fd4376bea0504 Author: Zou Nanhai Date: Mon Feb 5 14:59:53 2007 +0800 Fix hang when TV is not connect. commit 681b91924c364a1c00732f548539f2767929ba0e Author: Eric Anholt Date: Fri Feb 2 17:40:34 2007 -0800 Add subpixel offsets to fix accelerated rotated rendering on i915. This fixes the rendercheck "transformed src/mask coords 2" tests. Previously, the source pixels chosen would be off by one in some cases. The particular values were taken from Mesa, which uses .125 offsets (except apparently broken for y), but the signs are changed. I would be happier if I had better justification for why this worked. commit 76a4f2eea826036e2ac16b7e8c61ab946038c120 Author: Eric Anholt Date: Fri Feb 2 17:34:41 2007 -0800 Fix bad offsets in "right" rotation of the CRTC in cursor and update handling. commit 785a59ead0e8d1d681b2cb6827ee58ad2c51f8c6 Author: Keith Packard Date: Thu Feb 1 22:48:44 2007 -0800 Rename driver from i810 to intel. Permit old name for compatibility. Driver installs as intel_drv.so with symlink to i810_drv.so to ensure existing configurations continue to work. Updated manual page to reflect name change and add attributions for recent work. commit 1dde7a15a3a42b881c57ece95feceffadf412cff Merge: 16f30b2 1ba45a3 Author: Keith Packard Date: Thu Feb 1 21:31:27 2007 -0800 Merge branch 'modesetting-origin' into modesetting commit 16f30b2e6c72e228c66946ff9efadd7278379745 Author: Keith Packard Date: Thu Feb 1 21:30:19 2007 -0800 Add "Ignore" option to per-output monitor, removes output from server. Setting option "Ignore" "Yes" will cause the server to pretend as if the specified output does not exist at all. It will not be listed by the RandR1.2 extension, and the server will not attempt to detect monitors at startup time. commit b3633506dcd4fd58f0f9519cd67f6e75d6fb2042 Author: Wang Zhenyu Date: Fri Feb 2 11:35:44 2007 +0800 Add Intel 965GM chipset support commit 380129a17c61c783d77a87685ad5333e8b1c253c Author: Wang Zhenyu Date: Fri Feb 2 11:33:46 2007 +0800 Revert original crestline pci id patch commit f7489bbec150349bf03bea8d9d55451dbc92bfc2 Merge: f1285ec ae6d358 Author: Wang Zhenyu Date: Fri Feb 2 11:26:57 2007 +0800 Merge branch 'modesetting' into crestline Conflicts: src/i830_display.c src/i830_tv.c src/i830_xf86Crtc.c commit ae6d3585a4d96aa4dc05ca8563cebfa7796324ef Merge: 6d549ed 1ba45a3 Author: Wang Zhenyu Date: Fri Feb 2 11:21:59 2007 +0800 Merge branch 'modesetting' of git://proxy.ims.intel.com:9419/git/xorg/driver/xf86-video-intel into modesetting commit 9f0acf1eb01ad8320f4da4cc5e498af25c0ecc5e Author: Keith Packard Date: Thu Feb 1 15:44:18 2007 -0800 Check for rotation change in xf86RandR12CrtcSet. xf86RandRCrtcSet was ignoring changes to only rotation, so the screen would be left incorrectly rotated. commit 1ba45a3fcd40aaf4d373447fb9f252ca053f9105 Author: Eric Anholt Date: Thu Feb 1 13:26:38 2007 -0800 Improve register debugging output. This includes not reporting some fields on hardware where those bits are reserved, correcting one of the hardware error bit numbers, and reducing the severity of the debugging output warnings. commit 9bfeadbd5b37acb56cb646009239a94727ec52bc Author: Eric Anholt Date: Thu Feb 1 12:04:58 2007 -0800 Remove some more dead options. commit d6bc03c379c46842676f640b8edb13ea828bf20c Author: Eric Anholt Date: Thu Feb 1 11:28:34 2007 -0800 Bug #9819: Add two headers that are apparently necessary on older X Servers. commit 351866b0328b1d7afdc6b414c77d9ab4da4bd085 Author: Eric Anholt Date: Wed Jan 31 15:48:52 2007 -0800 Sync before overwriting Render acceleration state on the 965. commit 96783a458957ca280ebb6275fa692dd415df6931 Author: Eric Anholt Date: Wed Jan 31 15:48:02 2007 -0800 Fix copy-and-paste-o in the render code cleanup that broke XAA. commit 5238ff9c80718f9f8baa06123aa1bb6e3193fce6 Author: Eric Anholt Date: Wed Jan 31 14:01:41 2007 -0800 Naming and formatting cleanup in Render acceleration code. commit c47210d94ed17ad9e8e489a1e1886732e6dc7220 Author: Eric Anholt Date: Wed Jan 31 12:50:31 2007 -0800 Move the render code to non-EXA-specific filenames. commit 72ea0e514dcbebd009833c21f40e9bfb84074628 Author: Eric Anholt Date: Wed Jan 31 12:43:38 2007 -0800 Remove custom accelerated rotation code now that it's all done through Render. commit e62751db8b1a631c22ba0f77c932be4ab39ba741 Author: Eric Anholt Date: Wed Jan 31 12:15:38 2007 -0800 Add an accelerated path for rotation Render operations in XAA. The now-generic (kind of) EXA code will be cleaned up and moved to generic files in a later commit. commit f1285ecb72e807ac10db398fa983bf69d4154d97 Author: Zou Nan hai Date: Wed Jan 31 14:58:08 2007 +0800 tune commit 0620a7650f758652d05a0c25d159587f641aaf3e Author: Zou Nan hai Date: Wed Jan 31 11:21:11 2007 +0800 fix commit 12441e217a1968dea3a6b0646b6c5d0e2ede0e7e Author: Zou Nan hai Date: Wed Jan 31 11:20:23 2007 +0800 add 720p@50Hz, however 720p in this freq seems to not support src width, hactive > 800, so skip those inputs. commit 726c7dd89627409820ee5cf9bf657d5a926b3648 Author: Zou Nan hai Date: Wed Jan 31 10:34:39 2007 +0800 TV fixes. commit 4cd552e8f4851e029e43bf778cd8340f6c2c4881 Author: Eric Anholt Date: Tue Jan 30 16:34:50 2007 -0800 Fix accelerated Render transformations. Previously, we tried to use 2 points instead of 3 to describe the source rectangles, which mostly just worked for scaling. commit 6a628ae12b0568d656059891c5bca4415d8a735f Author: Eric Anholt Date: Tue Jan 30 14:19:05 2007 -0800 Simplify EXA acceleration transform handling. commit c8581254cb811f76aa6eae49d74489b543b3eb05 Merge: 9da1791 d329fa1 Author: Keith Packard Date: Mon Jan 29 21:25:13 2007 -0800 Merge branch 'modesetting-rotation' into modesetting This adds reasonably driver-independent rotation support to the common layer. The piece required in the driver is to allocate and redirect the crtc to a shadow frame buffer. The driver uses Render to perform the actual rotation operation (which leaves us free to do fun projective transforms at some point in the future :-). commit 9da1791eeca446cd59e2e7d0803e8a7920dbbea5 Author: Eric Anholt Date: Mon Jan 29 17:10:14 2007 -0800 Restore a few important lines from 89ace3e81fe72075b1709b073c05532f13b7f0d1 Typical results were failure to sync, and a black screen. commit 67c6a490e63f6997f8198bc3ab180a1af097fe86 Author: Eric Anholt Date: Mon Jan 29 09:53:58 2007 -0800 Remove dead code to undo the damage of xf86SetModeCrtc() blank limiting. commit 89ace3e81fe72075b1709b073c05532f13b7f0d1 Author: Eric Anholt Date: Mon Jan 29 09:43:35 2007 -0800 Bug #9680: Remove bogus blank length limiting in xf86SetModeCrtc(). Our modes typically come from EDID or default modes, and when the monitor asks for a specific mode, deciding to tweak it usually results in incorrect display. And if the user is specifying a mode by hand, tweaking it then is still pretty rude. Reviewed by: ajax commit 6eb69e737864d06dc6d4fbb4cd85a027cbb7643a Author: Dave Airlie Date: Sun Jan 28 19:40:12 2007 +1100 ch7xxx: actually power up the chip around connection detection commit d329fa121b2401cadb991c2854e077cfa68e15e8 Author: Keith Packard Date: Sat Jan 27 18:36:15 2007 -0800 Fetch from root window instead of screen pixmap during rotation. The XAA Composite wrapper doesn't check for sourcing from the root pixmap when deciding whether to sync the hardware drawing. Fix the rotation code to reference the root window in IncludeInferiors mode to avoid reading without correct synchronization. commit 20419d664c77e9a1a2de82987838b45f44774b47 Author: Keith Packard Date: Sat Jan 27 18:34:12 2007 -0800 Paint ARGB cursor data to ARGB cursor memory space in frame buffer. When converting from global cursor to per-crtc cursor, the ARGB cursor data was accidentally painted to the wrong location, overwriting some critical data and causing the hardware to lock up. (along with generating a garbage cursor image from uninitialized data). commit 00decd3b6cea0de22c88b9504dbe26d680e8ab16 Author: Krzysztof Halasa Date: Sat Jan 27 12:29:31 2007 -0800 Don't zero out mode structure after setting name field. Instead of using memset to zero the structure (at the wrong time, no less), use xcalloc to gather pre-initialized memory. commit 7a5f17087bda2833e84fa7e7ff0cb168943b89b6 Author: Keith Packard Date: Sat Jan 27 00:11:21 2007 -0800 Mark crtc region as damaged when rotation is set. When shadow frame buffer is allocated for rotation, it needs to be initialized by copying from the frame buffer. Do this by simply marking the entire screen as damaged which will force an update. commit 2d95cb6f041653f7e530b1f32cf007929c23ef3b Author: Keith Packard Date: Sat Jan 27 00:04:13 2007 -0800 Make rotated pixmap size of mode, not rotated mode. Update only damage. Rotated pixmap should match size of displayed mode, not the rotated version of that size. Take damaged region, rotate each box and display them instead of displaying the whole screen after each damage update. This exposes a bug where the screen is not correctly damaged after rotation. commit 5c1e27cdd243dc24dd2bfdeb46d757bbef6ba6af Author: Eric Anholt Date: Thu Jan 25 15:31:22 2007 -0800 Make rotated shadow buffer allocation dynamic. For EXA, this requires version 2.1 of EXA to do rotation, as the VT switching issues were too complicated otherwise. commit 5a67f3d3690903a181ca854d060cfa8b4c2aca26 Merge: 3cc583e 9811059 Author: Nian Wu Date: Thu Jan 25 13:40:50 2007 -0800 Merge branch 'crestline' of /git/xorg/driver/xf86-video-intel into crestline commit 3cc583ec8e79ec65103308220dd6047a4a7acddf Merge: 566e1d3 effe579 Author: Nian Wu Date: Thu Jan 25 13:37:05 2007 -0800 Merge branch 'modesetting' of git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into crestline commit 6d549ed280d3fcf3fe611b095d9f8adc0196bfb6 Merge: 75f4df2 effe579 Author: Nian Wu Date: Thu Jan 25 13:36:33 2007 -0800 Merge branch 'modesetting' of git://proxy01.pd.intel.com:9419/git/xorg/driver/xf86-video-intel into modesetting Conflicts: src/i830_driver.c commit 83cc4601b27d871484a2408f31154e9387064b9e Author: Eric Anholt Date: Wed Jan 24 16:52:22 2007 +0800 Fix a possible failure to misalign video allocation in XAA. commit e21332419581eff1d7651741bae0b640c84d0ecd Author: Eric Anholt Date: Wed Jan 24 12:32:38 2007 +0800 Warning fix. commit effe579e691b044e3ce59b41b5c0eaaac4368dda Author: Dave Airlie Date: Sun Jan 21 20:23:13 2007 +1100 ch7xxx: fix output detection as per datasheet, have no DVI to test this. commit c68dae2705d6e5b05dba7d40b6da112b98b926b4 Author: Dave Airlie Date: Sun Jan 21 20:22:39 2007 +1100 fix building with 7.1 server by hiding stuff inside RANDR_12_INTERFACE commit 8eb861fb61fda78e6c142ffa66c586357aaa2e44 Author: Eric Anholt Date: Mon Jan 22 15:50:14 2007 +0800 Replace the custom SaveScreen function with a generic function calling DPMS. With this, additional power saving occurs with the core screensaver, for example by turning off LVDS backlight. commit 7161e824e89a6a69f1277e510f2344ad31721f62 Author: Wang Zhenyu Date: Fri Jan 19 15:15:19 2007 +0800 Fix DRI issue when EXA enabled tell DRI we are smashing the context state. commit 74ebff6732b9bfcf8c865b52cbebfd9bf6b73eb2 Author: Eric Anholt Date: Thu Jan 18 14:26:21 2007 -0800 Allocate separate cursor memory per CRTC and rotate cursors appropriately. Also, add bind/unbind of the shadow rotate buffers, which was missed in a previous commit. commit 8759f4d89a141f7d69f1894ec3059ba6bd1cf86b Author: Eric Anholt Date: Thu Jan 18 13:13:25 2007 -0800 Partially deal with cursor rotation. The remaining issue is having a cursor image per CRTC so that the cursors can be rotated on the independently rotated CRTCs. commit b296cd9b8f63ab80e8fe46fcfcdba2e9af846468 Author: Eric Anholt Date: Thu Jan 18 13:02:26 2007 -0800 Fix transformation matrices for rotation at 90 and 270 degrees. commit d2ae2e2ccc3a8e7dc84745c235836cba150754d8 Author: Eric Anholt Date: Thu Jan 18 11:37:14 2007 -0800 Update CRTC values first so they can be used by the driver in mode setting. With this, the screen now gets rotated, though there are major issues. commit 92b0f3d2013c32d94fe57fe674da1b195288f347 Author: Eric Anholt Date: Thu Jan 18 11:23:22 2007 -0800 Disable the clipping for now, since it won't work. commit 14d265a02312e9cda7a21614d77525c7140c0ca8 Author: Eric Anholt Date: Thu Jan 18 11:21:58 2007 -0800 Store fixed values, not integers, in the transformation matrix. Also, add a bit more error checking. commit cc09dc3efe7279bf67512babb0f6cdaeba654772 Merge: 53e765d 3094100 Author: Eric Anholt Date: Thu Jan 18 10:20:36 2007 -0800 Merge branch 'modesetting' into modesetting-rotation commit 53e765d1cf5016961c25a798a27f0fe09bd1f237 Author: Eric Anholt Date: Wed Jan 17 16:56:06 2007 -0800 Fix broken last-minute change in determining whether we're rotated or not. commit f256243fff195f73bc8aa72e8dd2c6f7b33446cf Author: Eric Anholt Date: Wed Jan 17 16:01:18 2007 -0800 Allocate per-crtc rotation buffers and hook up RandR 1.2 per-CRTC rotation. It currently displays garbage. commit d17c386aee55bd223297adf78cf0fc1db43baf2d Author: Eric Anholt Date: Wed Jan 17 15:49:08 2007 -0800 Fix some segfaults: crtc->rotatedPixmap not set and dangling damage pointer. commit 67a152c43cb7231695b99030c952c52d4edd1e65 Author: Eric Anholt Date: Wed Jan 17 14:35:31 2007 -0800 Set all CRTCs' allowable rotations when xf86RandR12SetRotations() is called. commit 98110591001dfaf2b8b5ac440bea0c38456da718 Author: Zou Nan hai Date: Wed Jan 17 10:45:59 2007 +0800 enable all TV modes commit 3094100cdd04560af1b098bbe8024ae5270a10f4 Author: Gary Wong Date: Tue Jan 16 13:25:24 2007 -0800 Bug #9550: Add a quirk for too-high clock from EDID on the EN-7100e. commit 66c6427cc83e5e381932b0c10743cba4cb9ac883 Author: Eric Anholt Date: Tue Jan 16 12:54:29 2007 -0800 Add a settable backlight property for LVDS. This extends the output funcs to have a callback for when it's time to configure properties, and another for when the server has changed a property whose value isn't pending a mode set. (Pending properties are to be picked up by the driver at mode_set time). commit a52df90ffcb19e7c0a3b8018db3cd1391a3ec343 Merge: 37946c9 45696aa Author: Keith Packard Date: Tue Jan 16 12:20:26 2007 +1100 Merge branch 'modesetting' into modesetting-rotation Conflicts: src/i830_driver.c src/i830_randr.c Just a few sync changes. commit 37946c9c8a4afa4a4fc1fe9864a8eec1c4571d2c Author: Keith Packard Date: Tue Jan 16 12:14:55 2007 +1100 Interim work on rotation support with new xf86 code. Yes, there are lots of bits tied together here, and we should split this patch apart. Move I830 mode setting to xf86 mode setting. Make mode setting function also set the base address. This should fix problems where the base gets set incorrectly at times. Add driver-independent rotation support, requires driver-specific hooks for shadow pixmap allocation, otherwise it uses Render for painting. commit 9a5106401a65c90df32cb71987fca7126dc22e81 Author: Emilio Scalise Date: Mon Jan 15 17:22:27 2007 +0000 patch to control mergedfb options, bug #9664 commit bfeda3bfc59d309772398f688d29452ba97f3973 Author: Keith Packard Date: Mon Jan 15 08:05:32 2007 +1100 Convert I830PipeSetMode to xf86CrtcSetMode. Add rotation structures. Makes the mode setting logic device independent. Magic rename hooks allow multiple drivers to provide the generic code without name conflicts. Rotation code requires special pixmap creation hook, and uses Render to perform the actual rotation. commit 319be199ba657d2b82ba034edf3581aea6dcc3f0 Author: root Date: Sat Jan 13 21:34:00 2007 +0000 Tweak to bug #9639 commit 85de57947570de6b3dc9197647be3e9237480880 Author: Alan Hourihane Date: Sat Jan 13 19:18:24 2007 +0000 Fix bug #9639, i810 XvMC needs linking against libdrm. commit 45696aa29124e2852f94880642e70bb2e0cee827 Author: Eric Anholt Date: Thu Jan 4 12:34:47 2007 -0800 EXA: Use PRIM3D_RECTLIST instead of TRIFAN so we don't get diagonal tearing. A side effect is the reduction in vertex dispatch, which is nice. commit 6874a6f25ac87783d3770f77b9192e2d36d083a3 Author: Eric Anholt Date: Thu Jan 11 12:34:38 2007 -0800 Restore legacy overlay size limits, due to card hangs at larger sizes. commit c288aea40775a9cf561fda9912187c3cb5baa419 Author: Eric Anholt Date: Thu Jan 11 10:39:00 2007 -0800 Bug #8845: Unify all our XV ports to 1920x1088 maximum size. This increases the "legacy" size, but that was checked against the 845 documentation which claims support for 2048x2048 as well. It decreases the textured video size, which was running into the limits of the bigrequests extension. The new limits should fit within bigrequests while still supporting HD videos. commit f5d528f8ea27de31054e7f1843e34d8379f811ea Author: Eric Anholt Date: Thu Jan 11 10:36:33 2007 -0800 Don't limit cachelines to a vertical of 2048, and increase default allocation. The cachelines are used for two things: XAA pixmap cache and XV memory. Only XAA pixmap cache is referred to using an offset pointing at the beginning of the front buffer in rendering, and XAA only uses the 2d BLT engine, which actually has a vertical limit of 65536. So, pixmap cache is now limited to that much vertical. Additionally, the previous cachelines allocation was too small for our advertised XV limits, so video at the limits would fail with BadAlloc. Now, XAA allocates the same approximate amount of offscreen memory as EXA: 3 times the screen size, plus one packed HD video. commit d13bc016c0723f1df633ddaf5610ad73003b7c96 Author: Eric Anholt Date: Wed Jan 10 15:06:56 2007 -0800 Correct x/y/pitch limitations in several cases, and detail them in i830_exa.c. This reduces max framebuffer width and increases max framebuffer height on 965, reduces max X/Y on pre-965 EXA (could have caused mis-rendering), and increases max X/Y on 965 EXA (would have prevented acceleration). commit fa383289ac8a6dd1cb359e6f1991cc42beb6ff02 Author: Eric Anholt Date: Thu Jan 4 14:12:45 2007 -0800 EXA: Wait for sync before we set up new state in our static state buffers. commit 5857b4a1693085b8b42dd9560a7c4f5c3c82f862 Author: Wang Zhenyu Date: Wed Jan 10 15:09:47 2007 +0800 Formalize sync interface We should just call i830MarkSync/i830WaitSync in places we need, which care for both XAA and EXA. commit efb75f56053ee06f1dbd4edfaea9986b27162afb Author: Alan Hourihane Date: Tue Jan 9 14:05:43 2007 +0000 Bump to 1.7.4 commit 33c0fac6d63d9a509a24c7a5e5a46a0d6ee4dc5d Author: Alan Hourihane Date: Tue Jan 9 14:02:47 2007 +0000 Fix bug #8536, i915 BIOS fails when restarting Xserver. commit ee52c0ec4e95fa6e5f35c9cd75005a0c0003fd97 Author: Eric Anholt Date: Mon Jan 8 16:53:07 2007 -0800 Correct typo resulting in a crash with ivch. = != ==. commit 566e1d397744d0b477de4da6a206919906176f49 Merge: b13d638 da6a00f Author: Keith Packard Date: Mon Jan 8 12:37:49 2007 -0800 Merge branch 'modesetting' into crestline commit da6a00f787e4d13e6b75768c1976f1c44ae5bf72 Author: Keith Packard Date: Mon Jan 8 12:36:54 2007 -0800 Move EXA function declarations from source to header. Several new global functions were not put into the header file leading to potential mismatches between declaration and definition. commit 66546d8a8006e1c828e18ccab850214fd8d56b63 Merge: 0fd2752 40af0ee Author: Keith Packard Date: Mon Jan 8 10:14:24 2007 -0800 Merge branch 'modesetting-origin' into modesetting commit b13d6386dfb3a00bd9d21ac0695cdce9f812d1f3 Author: Keith Packard Date: Sun Jan 7 22:56:37 2007 -0800 Back out rotation changes for RandR 1.2 which won't work. RandR 1.2 requires a new rotation structure for per-CRTC rotation. commit 7473e7d5297f94164e0f35822166713fb21a11bd Merge: cdde9e7 a87801f Author: Keith Packard Date: Sun Jan 7 22:50:27 2007 -0800 Merge branch 'crestline-otc' into crestline commit cdde9e7f4a0645ab1ee3e124de54433c1a250097 Merge: 3110630 0fd2752 Author: Keith Packard Date: Sun Jan 7 22:50:12 2007 -0800 Merge branch 'modesetting' into crestline commit 0fd2752f199928f846fe03c9087f7b6d48cc28d9 Author: Wang Zhenyu Date: Thu Jan 4 14:35:55 2007 +0800 minor fix on last exa mem binding commit commit 3110630e04f37e184609c91494fa7f9f4d59b93f Merge: be9b635 736d82a Author: Keith Packard Date: Sun Jan 7 22:44:36 2007 -0800 Merge branch 'modesetting' into crestline Conflicts: src/i830_rotate.c Pull in upstream changes to crestline branch leaving only the PCI-IDs as local changes. commit 40af0ee6ba4ab7596fbc7fcc3ad04d109746ca6f Author: Keith Packard Date: Sat Jan 6 18:19:34 2007 -0800 Numerous symbol scope issues. I830EntityIndex is shared between 810 and newer driver. Move most EXA rendering state into I830 structure. Declare shared variables in shared header files rather than .c. commit 736d82a6b43f174cb95b425faacd4b0b889916fa Merge: 53b42f5 4c790f6 Author: Wang Zhenyu Date: Mon Dec 4 15:48:04 2006 +0800 Merge branch 'modesetting-origin' into modesetting commit 53b42f5bc7a58d02106436486e5bb56e56dbbfa1 Author: Zou Nan hai Date: Sat Jan 6 14:59:14 2007 -0800 support NTSC 480i M-J, PAL 576i for 640x480-1280x1024 sizes I still have problem with non-interlace mode and Hi Res mode. also I don't know how to pickup those mode in xorg.conf commit 71946bcdc3c68c220996afac944698eea1974a36 Author: Wang Zhenyu Date: Wed Jan 3 22:37:32 2007 -0800 [PATCH] Add rotation support for 965. commit 35cebed70827999812f8343ac97ad0dffda20786 Author: Eric Anholt Date: Thu Nov 16 15:12:43 2006 -0800 [PATCH] Replace broken PCI resource size detection with pciGetBaseSize() call. Signed-off-by: Keith Packard commit 33e912aca08fa11ef588eb386e16ba5f9ea13727 Author: Wang Zhenyu Date: Thu Jan 4 11:25:31 2007 +0800 [PATCH] Fix EXA mem binding We should check if EXA is really enabled. Signed-off-by: Keith Packard commit fa54a3c08301e59558ab0493b3d22324f4162496 Author: Wang Zhenyu Date: Wed Dec 6 13:24:44 2006 +0800 [PATCH] fix Makefile.am Signed-off-by: Keith Packard commit 2f2c443e971faa54ffcf751b6acb444e8e7875ce Author: Wang Zhenyu Date: Wed Dec 6 10:43:29 2006 +0800 [PATCH] Formats fixes We should use card_fmt for src/mask picture, and use dest color buffer format helper. Also fix wrong name for G965 texture formats, and pict_x1r5g5b5 isn't supported by sampler engine. Signed-off-by: Keith Packard commit 0bf04fe78a8a915310ef8a90f5c7872be7476e2e Author: Wang Zhenyu Date: Mon Dec 4 15:48:04 2006 +0800 [PATCH] set correct default border color Signed-off-by: Keith Packard commit 5c461063cde68092e778c44ac6abd9129cd8019e Author: Wang Zhenyu Date: Mon Dec 4 15:47:31 2006 +0800 [PATCH] fix typo in ps kernel fix corrupt in some subspans Signed-off-by: Keith Packard commit 89a42d489bd370b89e5ff4e01f026b4d64723cd8 Author: Wang Zhenyu Date: Wed Nov 29 17:16:46 2006 +0800 [PATCH] shut up warning Signed-off-by: Keith Packard commit 01bfa4fa6fc0ceec8581676e5d72c68dd71efa96 Author: Wang Zhenyu Date: Wed Nov 29 17:14:55 2006 +0800 [PATCH] fix alpha blending state Signed-off-by: Keith Packard commit 79018bb47c43510d59c592592f06204189bd12dc Author: Wang Zhenyu Date: Wed Nov 29 17:05:32 2006 +0800 [PATCH] Add in sf/wm program for mask picture without CA Signed-off-by: Keith Packard commit a5b9b438469f171b002fa0b99d8cab83e51ec968 Author: Wang Zhenyu Date: Wed Nov 29 16:52:44 2006 +0800 [PATCH] misc cleanup for G965 vs/sf/wm states Signed-off-by: Keith Packard commit b7c1e1656f45e43ea2f9a47f1a487050c0884c22 Author: Wang Zhenyu Date: Wed Nov 29 16:40:15 2006 +0800 [PATCH] WM kernel needs scratch space Signed-off-by: Keith Packard commit db9cfaa35adaf79ea57bc06b27c7e37935d3e1a7 Author: Wang Zhenyu Date: Wed Nov 29 16:37:06 2006 +0800 [PATCH] Setup default border color for our samplers Signed-off-by: Keith Packard commit 70276e4e9a8a5026ec436d2be5bf5eab868aa178 Author: Wang Zhenyu Date: Wed Nov 29 16:30:53 2006 +0800 [PATCH] fix vertex buffer size Signed-off-by: Keith Packard commit e8a4cbdeff4125e28d807d0a563efc0606d21a75 Author: Wang Zhenyu Date: Wed Nov 29 16:24:24 2006 +0800 [PATCH] clean up issue cmd to ring buffer Make it easy to track different part of ring state, and use rectlist primitive instead. Signed-off-by: Keith Packard commit 42534474fd2556e5987205626cca8f30e25855a8 Author: Wang Zhenyu Date: Wed Nov 29 16:04:16 2006 +0800 [PATCH] remove wrong scale_units Signed-off-by: Keith Packard commit 05e8a7989db1b8b9dfa7b688629af65d9910cc53 Author: Wang Zhenyu Date: Wed Nov 29 16:02:16 2006 +0800 [PATCH] Rename exa sf/wm program files Also fix some minors in wm program. Signed-off-by: Keith Packard commit 1329c86f2a4b3664920ded970a984a745a530da4 Author: Wang Zhenyu Date: Wed Nov 29 15:47:19 2006 +0800 [PATCH] EXA state mem for G965 Not split offscreen mem for exa, but alloc a dedicated one for G965 states. Signed-off-by: Keith Packard commit 4c790f614ecba1f6468e51779cfaf0e36b6b17ad Author: Jesse Barnes Date: Sat Jan 6 08:21:36 2007 -0800 More sparse cleanups: NULL vs. 0 make some variables static remove redundant variables There are only a few sparse warnings left now: some bitfield warnings and a few 'mixing code and declarations' warnings from the ring macros. commit a87801f73a73e53524237be7835b8cd8b3eb282c Author: Wang Zhenyu Date: Thu Jan 4 14:47:48 2007 +0800 Fix TV load detect Actually load TV, otherwise we don't have expected reg state to probe. commit 9cf759a2cb2580c4ac5f8cdc983389642f3cfce0 Author: Wang Zhenyu Date: Thu Jan 4 14:35:55 2007 +0800 minor fix on last exa mem binding commit commit f9c3ef40100acb85fca6e49c5c0e98f5c9ac97ca Author: Wang Zhenyu Date: Tue Oct 10 15:50:10 2006 +0800 [PATCH] Mark current ps kernel is experimential with little test. Signed-off-by: Keith Packard commit ef2f25e5c6ce04c202698c5122bc8627130073c0 Author: Wang Zhenyu Date: Tue Oct 10 14:11:35 2006 +0800 [PATCH] Use sf_prog.h instead Signed-off-by: Keith Packard commit ca608028c5301700444d39a1c631cc0d5648e1a2 Author: Wang Zhenyu Date: Thu Sep 28 13:55:52 2006 +0800 [PATCH] Fix picture's transform checking Signed-off-by: Keith Packard commit 848368d5d0b90e03d3ec447cb5bd39fc87aea8df Author: Wang Zhenyu Date: Thu Sep 28 11:15:33 2006 +0800 [PATCH] Fallback in mask picture for now Do it later after finish wm kernel program. Signed-off-by: Keith Packard commit df23624eebe938fa444c80cbedcd61919ec1aeda Author: Wang Zhenyu Date: Thu Sep 28 11:09:52 2006 +0800 [PATCH] Fix compile, add wm header file. Signed-off-by: Keith Packard commit fc944859b1b9605c748162bad1c93a6303c84aae Author: Wang Zhenyu Date: Thu Sep 28 10:36:00 2006 +0800 [PATCH] Add simplest wm kernel program for no mask picture composite This is a try to use new gen4asm language, and will finish composite program for mask picture with or without CA case later. Signed-off-by: Keith Packard commit acdc2da77b445e9347a4c6e53e35c81763cbb0b8 Author: Wang Zhenyu Date: Wed Sep 27 16:48:43 2006 +0800 [PATCH] Add mask sampler state Signed-off-by: Keith Packard commit 926d7fb09aaaabf050949ce7c6127c68441c8801 Author: Wang Zhenyu Date: Wed Sep 27 13:54:14 2006 +0800 [PATCH] change some src sampler states sampler for mask should also be set up, and fix default border texel. Signed-off-by: Keith Packard commit ed73bbaf5c2e9d555c884037a249cf03e7f60fa0 Author: Wang Zhenyu Date: Mon Sep 25 14:35:51 2006 +0800 [PATCH] Add file for i965 exa composite This does not include ps program, which will be added in g4a form. Signed-off-by: Keith Packard commit be9b635b82c0c77ccc0555f178f94de6e5338e27 Merge: 9e48808 5a793b0 Author: Keith Packard Date: Wed Jan 3 22:15:53 2007 -0800 Merge branch 'modesetting' into crestline commit 5a793b0dcf2d5de408b55073858fcfba6d99f994 Merge: f188525 394124c Author: Keith Packard Date: Wed Jan 3 21:54:05 2007 -0800 Merge branch 'modesetting-origin' into modesetting commit f188525030a8fac59e41520449b1aec9b123e4ea Author: Keith Packard Date: Wed Jan 3 21:53:07 2007 -0800 Add interlace defines for pipeconf regs commit 9e48808fccf5b153c2fb4027f9a09944574fdace Author: Wang Zhenyu Date: Thu Jan 4 11:25:31 2007 +0800 Fix EXA mem binding We should check if EXA is really enabled. commit ebd636dbd62cb69ed7defbd86f297c6bc7dac5dd Merge: 75f0033 3d9ea6f Author: Wang Zhenyu Date: Thu Jan 4 10:13:55 2007 +0800 Merge branch 'modesetting' into crestline Conflicts: src/i830.h src/i830_randr.c commit 3d9ea6f3c8a274f930d286fc15d3665dfaa8327a Merge: 75f4df2 394124c Author: Wang Zhenyu Date: Thu Jan 4 10:12:50 2007 +0800 Merge branch 'modesetting' of git+ssh://zhen@git.freedesktop.org/git/xorg/driver/xf86-video-intel into modesetting Conflicts: src/i830_driver.c commit 394124ceaadb46d976ad5c3bdeb1b77d351c57f6 Author: Eric Anholt Date: Tue Jan 2 16:57:33 2007 -0800 Add per-CRTC gamma control support for RandR. This makes the CRTCs now always run in gamma-enabled mode, rather than having flaky logic for switching modes. Also, it should clear up issues with the LUTs being uninitialized when outputs are first brought up. commit 69f250af60220a875f4a04c6d682bffa352281e4 Merge: 232e209 d960dea Author: Keith Packard Date: Wed Jan 3 00:39:15 2007 -0800 Merge branch 'modesetting-origin' into modesetting commit 232e2094321dbcdd6a67ef230eb50494a1c7d6df Author: Keith Packard Date: Wed Jan 3 00:38:34 2007 -0800 Allow initial position to be set in config file. Increase 965 max size. Add relative and absolute position configuration code, using per-output monitor sections. Options include: PreferredMode selects a preferred mode for this output by name Position absolute position, x and y in a single string. Below relative positions; argument names other monitor. RightOf Above LeftOf Enable force the monitor to be disabled by setting Disable enable to no or disable to yes. MinClock Set valid clock ranges MaxClock Monitor sections can also include sync ranges, physical size and mode lines as documented in xorg.conf(5). Monitors are associated with outputs through options in the Device section: Option "monitor-VGA" "My VGA Monitor" Output named 'VGA' will use monitor section "My VGA Monitor". commit 58e797b2caa6effa5455fc1f13dc4c58d0658744 Author: Keith Packard Date: Wed Jan 3 00:04:58 2007 -0800 Sync dspbase/dspsurf registers by re-reading them. This seems to eliminate base/surf value confusion during EnterVT. commit d960deab39eef91fb82b9f23118323aeb4c9c63e Author: Jesse Barnes Date: Tue Jan 2 20:23:37 2007 -0800 Various sparse cleanups: NULL vs. 0 remove redundant pScrn in I830DRISwapContext make I830DRISwapContext static commit 5057769d3a7c1b3a94f49bbff47b9697f368d975 Author: Dave Airlie Date: Tue Jan 2 19:25:08 2007 -0800 Use correct variable to initialize DVO I2C bus. intel_output->pI2CBus isn't initialized until the output is all happy and running, so using it before that is rather unlikely to do much good. commit 2bb9bb951e664d704bdf7a808edbe8b2f50c4366 Author: Keith Packard Date: Tue Jan 2 14:04:10 2007 -0800 Typo disabled new DRI MM on all chips. Detection for new DRI MM wasn't converted to #define correctly because of a typo. commit 5080307230d534b84090bfefc13ac437d53f915d Author: Keith Packard Date: Tue Jan 2 00:16:41 2007 -0800 Fix driver to build on RandR 1.2 systems again. Changes to move initial config code to RandR 1.1/1.2 common path broke code required only for RandR 1.2. commit 237847007afb5a429a50b6aa0b1acc239326571b Author: Keith Packard Date: Tue Jan 2 00:09:35 2007 -0800 Make driver build and run on Xorg 7.1. Delay DSP*BASE setting until last. Autodetect libdrm version, disable new memory manager on older libraries. Move new M_T_ defines from i830.h to i830_xf86Crtc.h. Add many system headers to define functions. Use i830PipeSetBase at end of mode setting code to set DSP*BASE and flush changes. Don't duplicate PipeSetBase call from screen init function. Make initial RandR configuration code usable on older versions of extension so the server doesn't start in a panning mode. Use xfree instead of free in i830_tv.c. commit 25d5a892319b02dc6eb81390dea29cd88a1e7da4 Author: Keith Packard Date: Sun Dec 31 15:39:20 2006 -0800 Elide identical modes from reported list. Where two modes would produce precisely the same crtc settings and have the same name, remove the latter mode from the mode list. commit feeefc92e450e9de58da51147325300ffabd2059 Author: Keith Packard Date: Sun Dec 31 14:22:53 2006 -0800 Replace author-specific license with generic license. This file retained the old MIT license with the embedded author name. Replace that with the version that uses generic 'copyright holders' instead. commit 5e456251f0d6c8bef3241267f6c1b3bca4670c0d Author: Keith Packard Date: Sun Dec 31 14:21:12 2006 -0800 Replace author-specific license with generic license. This file retained the old MIT license with the embedded author name. Replace that with the version that uses generic 'copyright holders' instead. commit 75f0033ec41a34bbe695cfd4a9bc9aa60e919c99 Author: Wang Zhenyu Date: Sun Dec 31 15:16:59 2006 +0800 Issue current virtual size to mode valid. Revert change in rotation patch. commit 2330b341c925e60bc2c991b4ca8450badbc459a4 Author: Keith Packard Date: Sat Dec 30 22:47:58 2006 -0800 Prune default modes to EDID clock limit. Pick out the EDID max clock value and use that to elide default modes which are too fast. commit c70e7bc701cba3092acfdd66ed4d63487ac8ff9f Merge: 8b6c456 4ed79a2 Author: Keith Packard Date: Sat Dec 30 21:58:36 2006 -0800 Merge branch 'modesetting-origin' into modesetting commit 8b6c456e16c81f719bb68178e3e4d26f4108a9c2 Author: Keith Packard Date: Sat Dec 30 21:57:42 2006 -0800 Detect in ProbeModes. Let get_modes set edid. Prune interlace/dblscan. Move output connection status detection from RandR code up to ProbeModes so it is done before mode sets are built. Otherwise, the mode building code will elide all modes the first time through as it ignores outputs that are disconnected. Most get_modes functions fetch EDID data; make sure that any EDID changes are used in the ProbeModes filtering of default modes. Otherwise, stale EDID data will be used. Allow outputs to advertise support for interlaced and double scan modes; prune such modes from the default mode list when outputs do not support them. commit 4ed79a2ba6250354ffc24fa1f7a21ca914ad157e Author: Eric Anholt Date: Fri Dec 29 12:19:41 2006 -0800 Relax tiling requirements on G965. For the 965, we can tile with the pitch at any integer multiple of a tile size (128 or 512B), up to 128KB. commit b8692e646227e56c9ae4f72b9aaa75457b4c0f5f Author: Keith Packard Date: Thu Dec 28 22:23:16 2006 -0800 Lack of configured monitor implies no configured modes. Missing check for missing monitor configuration would result in segfault. commit 973da654219ea43916b0b44acfa09a415bed3d7a Author: Keith Packard Date: Thu Dec 28 22:22:22 2006 -0800 Sync ranges from EDID/default should not limit configured modelines. Limit the effect of sync ranges so that sync ranges found via EDID will not eliminate modes explicitly added by the user. Limit default sync range to eliminating only default modes, not configured or EDID modes. commit f7b1d4c1f7d17a811e17c6a17861ff70be9fbdd7 Author: Keith Packard Date: Thu Dec 28 22:18:57 2006 -0800 Belinea 10 15 55 monitor quirk - override preferred mode with largest @60Hz Belinea 10 15 55 model monitor reports a preferred mode of 640x350, when in fact it wants a 1024x768 mode @ 60Hz. Add an edid quirk that selects the largest size mode, preferring those closer to 60hz among equal sized modes. commit fbd3f588f0d412c20cc3f86dfad73700734a9a76 Author: Wang Zhenyu Date: Fri Dec 29 10:47:52 2006 +0800 rotation support for randr-1.2 This patch trys to resolve rotation regression with randr-1.2. Tested with Crestline VGA output. See bug 83. commit a1796bfb51d90fd545973909404e1a77ad27c9cc Merge: 84915ac fabfccd Author: Wang Zhenyu Date: Fri Dec 29 10:10:26 2006 +0800 Merge branch 'crestline' of git+ssh://zhen@otc-graphics.jf.intel.com/git/xorg/driver/xf86-video-intel into crestline commit bedab1654e2dfcf7800bd0101e6991800a544019 Author: Eric Anholt Date: Thu Dec 28 16:34:40 2006 -0800 Remove 8MB/32MB default VideoRam limits. With modern monitors and increased XV and EXA memory requirements, these small limits were resulting in DRI and other initialization failures because we wouldn't allow them enough memory. Instead, allow each piece of the system (DRI, EXA, XAA, etc) to request as much memory as it wants, and choose the actual videoRam to be used for laying out the memory afterwards. With this change, in the absence of a VideoRam option, 32MB will be allocated for textures. commit 4cc5dcf6d7423239f0181f246638366edcb8be87 Author: Keith Packard Date: Thu Dec 28 16:43:39 2006 -0800 Oops, dont use symlinks for stolen xf86 header files commit 86da0d05e4e338e7f4c841577002eb1ba9f6b2ac Author: Keith Packard Date: Thu Dec 28 16:40:54 2006 -0800 Add stolen xf86 header files to driver source commit 2b0a997e5b4b51267fa4f6725f8a965093392434 Merge: bb238a8 9cfbf1c Author: Keith Packard Date: Thu Dec 28 16:34:38 2006 -0800 Merge branch 'modesetting-origin' into modesetting commit bb238a8fc234a8e5e86cd2f42c58c9816a15732c Author: Keith Packard Date: Thu Dec 28 16:27:22 2006 -0800 Configuration support: per-output mode lines, preferred mode. Add the modelines specified in the per-output monitor and all of the default modes to the list to each output. Prune the resulting list to specified sync limits and virtual sizes. Sort the resulting mode list on preferred/size/refresh. commit 9cfbf1ceda8f153438df1ba4b1712c55cc872017 Author: Eric Anholt Date: Thu Dec 28 12:59:23 2006 -0800 In EXA mode, force backing pixmaps into memory when doing XV to them. commit e889bde13dcc2438181a03319e204ae7b9235a78 Author: Eric Anholt Date: Thu Dec 28 12:49:24 2006 -0800 Draw textured video to the backing pixmap in the composited case. Currently, when the backing pixmap is not in framebuffer, we just BadAlloc rather than drawing garbage to the front buffer. This can be fixed with EXA. commit c7083a6f30fdf1859f60beba4c352bd790af4773 Author: Eric Anholt Date: Thu Dec 28 11:53:33 2006 -0800 Add WIP CH7017 LVDS driver. This is disconnected because it's untested. commit 778db49616b233f5a2f4f007ceb62a1a878a6e25 Author: Eric Anholt Date: Thu Dec 28 11:36:15 2006 -0800 Add WIP IVCH (i82807aa) driver. It is not currently hooked up as it hasn't been tested. The hardware that this was written for decided to stop working. commit 5806f670eb3a5171ffb0b75701640d5311b34c80 Author: Eric Anholt Date: Thu Dec 21 16:22:14 2006 -0800 Reformat the DVO drivers table to be a bit more normal style. commit b928cef9cd5b3f88dc0880f406300d246a59e0fe Author: Eric Anholt Date: Thu Dec 21 16:21:28 2006 -0800 Move dvo driver detection into i830_dvo_init(), and use GPIOB for LVDS drivers. The documentation states that GPIOB is (generally) used for devices on DVOA on the motherboard, which appears to be the case on the laptop we have with LVDS on the motherboard. This patch is probably not entirely accurate, as there was apparently an LVDS DVO card sold that could be put in desktop machines, which would likely be on GPIOE like other ADD cards. Given that we couldn't find one of these cards for purchase, I'm not worrying about it. commit fabfccd3d8a55a83138d361f66c53cb7e6d371a8 Author: Zou Nan hai Date: Thu Dec 28 14:48:54 2006 +0800 Fix commit ddefce4f02eb3272ad37e890350f73f03eff73c9 Author: Zou Nan hai Date: Thu Dec 28 14:45:53 2006 +0800 support NTSC 480i M-J 1280x1024 1024x768 800x600 640x480 PAL 576i 1280x1024 1024x768 800x600 640x480 I still have problem with non-interlace mode and Hi Res mode. also I don't know how to pickup those mode in xorg.conf commit c87462ded20904dbc6c6cbdb9547523b75fe5471 Author: Eric Anholt Date: Wed Dec 27 19:14:14 2006 -0800 Remove the nasty #define away of ErrorF in i830_video.c. Every new person touching this file hits this trap, so it's now removed. commit 6dc3387219b2b31e0c7dbaa88b568fbd54d72878 Author: Eric Anholt Date: Wed Dec 27 18:58:40 2006 -0800 Fix operator precedence issue that stuck 965 t-v state into the front buffer. commit 210f30bd75529d0064bc46af15464a2a47f9d4c6 Author: Eric Anholt Date: Wed Dec 27 18:57:50 2006 -0800 debugging for location of various 965 textured-video state. commit 82738e60769a4798d7425ffc67882d1f74043b76 Author: Eric Anholt Date: Wed Dec 27 18:35:35 2006 -0800 Fix discrepancy between XAA/EXA linear allocations that broke EXA XV. The XAA allocator returned an offset from the beginning of the frontbuffer allocation (where the linear allocations come from), while EXA returned one from the start of framebuffer. Now, both of them use the start of framebuffer as the reference. commit 8165e043670f6a7fee5e7e8f33644adda5757591 Author: Eric Anholt Date: Wed Dec 27 18:17:37 2006 -0800 Bump the size of the extra space for 965 so it actually fits. commit f830c55e5f4efc97451240f8f18b49f46e8bce0f Author: Eric Anholt Date: Wed Dec 27 18:14:59 2006 -0800 Make the assert() in i965_video.c actually work and expose an error. commit a75b416c91085df1db4027a25eafcd04fef1e57c Author: Eric Anholt Date: Wed Dec 27 17:58:57 2006 -0800 Replace XAA mark/waitsyncs with the XAA/EXA wrappers. commit 150010d2f20f4a57f8507eba8762e36966b0f426 Author: Eric Anholt Date: Wed Dec 27 17:57:34 2006 -0800 reformat i965_video.c to 4-space indents, 80 columns, no trailing whitespace. commit 05066632b068ae91d7be77e3a199ea0d337d975b Author: Eric Anholt Date: Wed Dec 27 17:41:45 2006 -0800 Move 965 textured-video out to a separate file. commit 46df75ccd4647ea033583130253a2a2218b34a20 Author: Eric Anholt Date: Wed Dec 27 13:45:06 2006 -0800 Bug #7524: Major improvements to EXA/XAA static memory allocation. With this, we no longer allocate XV through the XF86 linear allocator in the EXA case. We also no longer allocate extra space for the XF86 2D allocator that we don't use in EXA mode, or space for the EXA allocator in XAA mode. The EXA offscreen allocator now gets enough space for several screenfuls of pixmaps plus one 1920x1088 movie. A large duplicated section of code for allocating each framebuffer in the old dual-screen mode was also factored out. commit 84915ac8afeb4bbc03df8f94ab3ba351788d6501 Author: Keith Packard Date: Thu Dec 21 23:42:53 2006 -0800 If DDC detailed modes are missing physical size, pull from features. Monitors without detailed modes, or those which do not bother to set a physical size in their detailed modes may still have physical size in the global data. commit 1e9105395a7900521206b28bae4bb4f85669a0cd Merge: df9ecf8 c237e93 Author: Keith Packard Date: Thu Dec 21 23:34:30 2006 -0800 Merge branch 'modesetting-origin' into modesetting commit df9ecf8ab93f0aeffde11d438513d41220ba75cb Author: Keith Packard Date: Thu Dec 21 23:32:35 2006 -0800 Change belinea edid quirk to cover the 10 20 30W model as well. This larger model reported different (but still incorrect) sync polarities, so instead of flipping them, just set them to the right value. commit c237e930615f1b048895efe988d28503f5d67c37 Author: Keith Packard Date: Thu Dec 21 23:30:37 2006 -0800 Use Vesa DDC timeouts for all I2C busses. The default I2C timeouts are very short, and while most modern monitors have no trouble responding at that rate, some older ones cannot manage. commit d8c5dba4d797fc50d7b2b5855f34e2d2e2ad3e4f Author: Keith Packard Date: Thu Dec 21 21:20:43 2006 -0800 When cleaning duplicate modes, make sure ->Last is reset correctly. When removing the very last mode for a monitor, move the ->Last pointer to the previous list element. commit fab9a6b6210daea423b609208ef57fa26571f5d3 Author: Keith Packard Date: Thu Dec 21 21:19:32 2006 -0800 Add EDID quirk support for broken EDID data. For EDID with known errors, add a quirk mechanism to automatically compensate. The first quirk is for a Belinea 1440x900 monitor which incorrectly specifies sync polarities in the detailed mode. commit d9b27667e6cc6c7e171b0f513d40be7658cf4574 Author: Keith Packard Date: Thu Dec 21 21:17:03 2006 -0800 PLL computations missed one possible 'm2' value. m2 was ranging from min <= m2 < max instead of <= max resulting in inaccurate PLL frequencies for some modes. commit c5205595f2abacc9f736e1f53666302ec551cbc4 Merge: b8f3ec7 4c0c1aa Author: Keith Packard Date: Thu Dec 21 02:51:33 2006 -0800 Merge branch 'modesetting' into crestline commit 4c0c1aa882cfec77b2183baec93cbc4cfaf4abe0 Author: Keith Packard Date: Thu Dec 21 02:33:39 2006 -0800 Computed corred color conversion values. Extract correct color conversion values for all video formats from documentation. Use those, with appropriate conversions, for the color conversion register values. commit 98fd44d681220aa31200e4262f1a7ec952a09530 Author: Keith Packard Date: Thu Dec 21 01:24:24 2006 -0800 TV subcarrier was computed from wrong clock value. The constants provided in the documentation for the subcarrier DDA values assumed the clock was programmed to precisely 108MHz, but the PLL can't hit that value exactly (or our PLL computation can't, in any case). The result was an incorrect subcarrier frequency which resulted in synthetic subcarrier phase shift and a lovely rainbow effect on the screen. Unfortunately, the documentation didn't exactly describe the function performed by the subcarrier clock hardware, so a bit of detective work was needed. New constants were computed using the code in tv.5c and those, along with lots of other values from the documentation were inserted into the necessary tables. The result appears to generate stable NTSC video on the svideo connector. commit b8f3ec77104882fc755ddc88f2583cd5d15387a4 Author: Zou Nan hai Date: Thu Dec 21 10:46:17 2006 +0800 This fix should be better, I830SetMode will call i830_sdvo_dump, if the sdvo is not initialized, X will crash. So any application which SetMode will crash X if there is no SDVO output. commit b83cbec34a8d1ff39bd303fcd27fc03dcdb1d6c3 Author: Zou Nan hai Date: Wed Dec 20 17:32:05 2006 +0800 fix null pointer reference, I hit this when running doom3-demo commit 4ba72fc408e7fab7c384aff5e73f0cfb8fd86cf3 Author: Keith Packard Date: Tue Dec 19 22:39:57 2006 -0800 Reinitialize DGA mode list whenever we update the global list. DGA has a copy of the current mode list (yes, this is broken). Regenerate it whenever the ddx mode list changes. commit 07d3f141712e2abf704cb68a59cd5d5c3390e2ca Merge: 0cd1a11 c28075e Author: Keith Packard Date: Tue Dec 19 22:01:02 2006 -0800 Merge branch 'modesetting' into crestline commit c28075e1d70c98b39fc9829a6a01da171a9b9426 Author: Keith Packard Date: Tue Dec 19 21:53:10 2006 -0800 TV output I830OutputPrivate had wrong type (crashed in SetMode). TV output private type field was set to I830_OUTPUT_SDVO instead of I830_OUTPUT_TVOUT. When DGA mode setting occurred, the sdvo output debug code would be invoked and attempt to dereference sdvo-specific bits of the private structure leading to a segfault. commit 2ef4c5e8f6444aad192304e5a2f7a0c77bfb917d Author: Keith Packard Date: Tue Dec 19 16:45:39 2006 -0800 Re-initialize physical screen size only if -dpi was passed. Physical screen size from the config file or DDC will already be set correctly in the screen structure, unless it was computed from the virtual size using the -dpi command line option. Recompute physical size as we reset the screen size if the -dpi option was used. commit b00951a5c87ca8c596ee9cc73d96d598eec60751 Merge: 05cd921 c0a0ddc Author: Keith Packard Date: Tue Dec 19 16:22:04 2006 -0800 Merge branch 'modesetting-origin' into modesetting commit 05cd921140fae2746c3421715d7f18248a66735a Author: Keith Packard Date: Tue Dec 19 16:21:46 2006 -0800 Align textured video dither matrix to window commit c0a0ddc0ce69b60ac3a58d45cb32ccd0e0bfb0c9 Author: Eric Anholt Date: Tue Dec 19 16:06:13 2006 -0800 Bug #9382: Save the CRTC's desired mode in the old RandR 1.0 mode set path. commit 3c86fdda1d09fd22d4c15a8e412611db4cea7dcf Author: Keith Packard Date: Tue Dec 19 15:53:07 2006 -0800 Turn on dithering for 915 textured video commit 293ad158ece941eeb9894f5af4d663464b9954e5 Author: Eric Anholt Date: Tue Dec 19 14:35:51 2006 -0800 Remove 8k framebuffer stride restriction on 965. Also, clean up the message if the limitation is hit. commit ff1223142aedcf888665de6e4d7d5d03a55b2688 Author: Eric Anholt Date: Tue Dec 19 12:49:55 2006 -0800 Only erase curMode in i830DisableUnusedFunctions when the CRTC is disabled. This bug resulted in broken xinerama, among other issues. commit 3b5703c7fa08e63f2a31be7496ded9c70015946d Author: Daniel Stone Date: Tue Dec 19 09:49:49 2006 -0800 Respect the srcdir. commit d89e211b12ff715d7475421af69157b102accaca Author: Eric Anholt Date: Fri Dec 15 14:39:14 2006 -0800 Disable configure-disabled outputs before mode setting. We're not supposed to turn off pipes/planes/dplls with outputs attached, which we could have done before this. commit 3fc9feec7b29e0b908e3b90e0f6634117d66ba6d Author: Eric Anholt Date: Fri Dec 15 14:38:23 2006 -0800 Simplify i830DisableUnusedFunctions() by using the CRTC dpms routines. commit 7b8056e3e347b15b37be0116315d7d3a23aef99f Author: Keith Packard Date: Tue Dec 19 01:32:54 2006 -0800 Writing 1 to I2C line means to tristate the bus so others can manipulate it. We were forcing bus lines to 1 which was breaking DDC for some monitors. Instead, make the PutBits function just tristate when writing 1 bits. commit 0cd1a11d4493ef4a2d05ccf8aadec743504dba5f Author: Zou Nan hai Date: Tue Dec 19 14:03:19 2006 +0800 fix for bug #108 commit eedef7adc53dd8337d27c02551c5778fb43bae05 Author: Eric Anholt Date: Mon Dec 18 15:57:44 2006 -0800 Add a detect() function for DVO chips, and implement it on sil164. commit 8983845f91cacf8110c70121e0f5f293fe443e6d Author: Eric Anholt Date: Mon Dec 18 15:57:08 2006 -0800 Fix crash in xf86SetScrnInfoModes when pScrn->modes ends up empty. commit fa4642048b183134544fc5ee47558446d27f6194 Author: Eric Anholt Date: Mon Dec 18 15:32:35 2006 -0800 Clean up i2c_vid interface, including de-StudlyCapsing and removing dead code. The old Init() function is removed and the previous Detect() function is now init(). This leaves us room in the namespace for a detect() like other outputs have ("is the monitor connected?"). Also, Power() became dpms(), taking a DPMSMode*. In general, the mode setting path now matches the intel internal path, except for the lack of mode_fixup(). commit dcb069a1da6b3300b0772843dd9b67efdf319637 Author: Dave Airlie Date: Tue Dec 19 10:29:56 2006 +1100 ch7xxx: add lookup table for name Don't load if we don't match name commit 0305298405bd7018e5c520bdd5ea46a75be09f37 Author: Eric Anholt Date: Mon Dec 18 14:49:01 2006 -0800 Limit modes to 165Mhz on CH7009, according to spec. commit 23114fbccad849249b4b7200e3d085a9ac0e2e99 Author: Eric Anholt Date: Mon Dec 18 14:47:44 2006 -0800 Align the driver to the CH7009[AB] spec. With this, the driver works at 640x480 and 1280x1024 on my panel. commit 633d718f6fb019844304228a81c84a8605da004f Author: Eric Anholt Date: Mon Dec 18 13:59:17 2006 -0800 Clean up a bunch of log noise from ch7xxx startup. commit 3167bc6b24bd74c50ca4c4d361b14aa24d77cfa3 Author: Eric Anholt Date: Mon Dec 18 13:45:55 2006 -0800 Fix DVO mode valid function to not throw out all modes. commit 4ca92da5b93b07f0bff51843f8fb5d04a2a29443 Author: Eric Anholt Date: Mon Dec 18 13:45:34 2006 -0800 Add save/restore to ch7xxx to avoid segfaults on server start. commit 31849edb3b503ade659981dfa36babf5ad8757c5 Author: Eric Anholt Date: Mon Dec 18 13:32:21 2006 -0800 Move the ch7xxx code to the style I've been settling on for output code. commit 1975fa5b010100196af201e40f43b30a149b7750 Author: Eamon Walsh Date: Mon Dec 18 14:08:55 2006 -0500 Revert "Convert callers of LookupWindow() to dixLookupWindow()." This reverts commit 08575331ac75783c9910cfb6e78db701a29983ac. commit 2d0ca2202cb0d780bb4387bbe04e5caa6512e3b1 Author: Eric Anholt Date: Mon Dec 18 10:37:33 2006 -0800 Bug #8786: Treat pictures with no color data as non-component alpha. This is an un(der?)-documented part of the render protocol: If the mask is alpha only, then the component alpha flag is ignored on it. commit 65cd18b97bc533b24031d6eb882fd3074779ceba Merge: bf43f8c 6823ca8 Author: Keith Packard Date: Sun Dec 17 17:32:24 2006 -0800 Merge branch 'modesetting' into crestline commit 6823ca87f3b1ef3b28ed167254dcfce2a80467df Author: Keith Packard Date: Sat Dec 16 21:12:47 2006 -0800 Follow mode setting order in RestoreHWState. Add delays after output and CRTC disable. Restore panel fit register before PLLs are restarted. Move all VGA restore code last. Shuffle various register writes around and add delays to match PipeSetMode code. commit 86558cc622b516b568cc26efdf9b64d4b660f50f Author: Keith Packard Date: Sat Dec 16 21:09:31 2006 -0800 Prefer earliest CRTC when mapping to outputs. For some reason, the code was preferring the last possible output when mapping outputs to crtcs. Use the earlier CRTC instead to make the i830 driver consistent with BIOS usage. commit 8e6ab99b3195325f9fe5432725fe328591c0c7e2 Author: Keith Packard Date: Sat Dec 16 21:06:36 2006 -0800 Not restoring active outputs. Wait for input sync before enabling outputs. Oops--looks like a typo to me; the code was callint set_target_output instead of set_active_outputs. BIOS loops waiting for the SDVO input to sync before enabling outputs, this makes sense to me. commit bffd611b0a1cb05868e0f93e6ff9357a3116eaa6 Author: Keith Packard Date: Sat Dec 16 21:03:15 2006 -0800 Follow BIOS order in writing DPLL/DPLL_MD registers. 965 BIOS writes DPLL and then DPLL_MD. 945 BIOS writes DPLL twice. commit 9b1a1b170befae2e705c23ce295837d0d13b60c0 Author: Keith Packard Date: Sat Dec 16 16:53:49 2006 -0800 reorder restore writes for 965 VGA commit 08575331ac75783c9910cfb6e78db701a29983ac Author: Eamon Walsh Date: Fri Dec 15 17:44:44 2006 -0500 Convert callers of LookupWindow() to dixLookupWindow(). commit 60411bc4d0b3c53850c73b7246d5f7ed5c2d4084 Author: Keith Packard Date: Thu Dec 14 18:36:44 2006 -0800 Follow BIOS PLL write protocol (disable,enable,enable). This gets VGA output and text restore working, but SDVO is still not working. commit 082519f34b468d0c5aa08db74ff2b51cf411d743 Author: Keith Packard Date: Thu Dec 14 16:51:48 2006 -0800 Turn pll/pipe/plane on in crtc_set_mode Instead of delaying pll/pipe/plane enables to the dpms function, turn them on right away in the crtc_set_mode function. To avoid rewriting these registers in the subsequent dpms function, check each register written there to see if the enable bit is already on and don't rewrite. commit 55e7a32096a58f7bb2380b04df6b3dd9dc8e5b6a Author: Keith Packard Date: Thu Dec 14 16:50:26 2006 -0800 Respect Virtual in xorg.conf Instead of growing virtual to fit our desired sizes, if it is set in the config file, use the value as the virtual size and make the default configuration fit within that size (if possible). commit d4142abc68888178a8f6c54fdcece91e898bc9f5 Author: Keith Packard Date: Thu Dec 14 13:49:08 2006 -0800 Set frame buffer size ranges in xf86_config. Initialization and default configuration code needs to know the range of legitimate sizes for the frame buffer. commit 2b5a5542ad7d1c558d1684a0a69b66bab6e9d001 Author: Keith Packard Date: Thu Dec 14 13:43:50 2006 -0800 Change TV modes from Preferred to Driver. This avoids having the artificial TV mode size drive overall screen size. commit 3fe802453a85183a69c36a098639895f49b17df1 Author: Keith Packard Date: Wed Dec 13 13:15:14 2006 -0800 Move xf86CrtcConfig to ScrnInfo private. Pull xf86CrtcConfig out of the driver private structure and allocate a ScrnInfo private index for it. Also, make the arrays of outputs and crtcs dynamic instead of fixed. commit 0f6addc8a6aeb9bd041d0f8e8e5850e76764ba51 Merge: 001e272 3fa5b39 Author: Keith Packard Date: Wed Dec 13 12:10:31 2006 -0800 Merge branch 'modesetting-origin' into modesetting commit 001e272437a0247ffbc4b9ff8a3f2b437cf4c533 Author: Keith Packard Date: Wed Dec 13 12:08:58 2006 -0800 Auto-detect working TV output by checking TV regs functionality. We can't figure out which chips are supposed to have TV out, so instead we prod the TV_DAC register to see if it will hold the value written to it, if not, we assume the chip doesn't have TV out. commit 3fa5b3998353518c40e2fb0c28b425ee22c8d625 Author: Eric Anholt Date: Wed Dec 13 11:54:36 2006 -0800 Add magic double-write of the dpll register to fix mac mini cold boot. commit e27372e85a007d8a7e31678dbdb62755c22918bf Author: Eric Anholt Date: Wed Dec 13 00:26:27 2006 -0800 Add some caution with PCI write posting and DPLL delays to i830_crtc_dpms. While it doesn't specifically help/hurt my test case, we've seen enough mysterious behavior that caution is probably warranted. commit df0a5a25aa72563a33731f8063602dd7faad8459 Author: Keith Packard Date: Wed Dec 13 11:38:18 2006 -0800 Rename outputs to TMDS-1/TMDS-2, LVDS, TV and VGA. These names are reasonably short and describe the connector rather than the target media, but we don't appear to have any way to determine what is on the other end of the wire. More importantly, they're all unique now. commit e3604fc63243ab0f31673a923d20e23131b607f6 Author: Alan Hourihane Date: Wed Dec 13 15:42:53 2006 +0000 Delete ChangeLog commit 4cdcaac26b3f425ec8fcc2fd3eaa9bfb7a137422 Author: Keith Packard Date: Tue Dec 12 23:26:00 2006 -0800 Write dpll_md register while updating dpll as that appears to be required. SDVO multiplier on 965 is in the dpll_md register; for some reason, that needs to be written along with the dpll value or the multiplier doesn't get set correctly. commit 9dd1520421dcb8c7f36a684f3d2ffb6b42a6671a Merge: d57a258 63bf484 Author: Keith Packard Date: Tue Dec 12 22:52:59 2006 -0800 Merge branch 'modesetting-origin' into modesetting commit d57a25815398ae83eae8bdcb3a1b607760aa30b6 Author: Keith Packard Date: Tue Dec 12 22:48:21 2006 -0800 Elide duplicate modes in pScrn->monitor. xf86DDCMonitorSet dumps all of the DDC-discovered modes into the monitor mode list without checking to see if they are already present. This provides an ever-changing list of modes for outputs which have no DDC and which simply duplicate the monitor mode list. commit 63bf48481d4e5e285fdfe7037ea4150283163a54 Author: Eric Anholt Date: Tue Dec 12 22:42:41 2006 -0800 Correct typo in mac mini hack, and return modes when the hack is unnecessary. commit 41444183b59ed84c09749ca89afbef036d42ec5f Author: Eric Anholt Date: Tue Dec 12 18:08:57 2006 -0800 Replace custom, partially broken DPMS implementation with a generic one. commit 7ed1b05922c07ff45a5794a992fd3d59ab55aa73 Author: Eric Anholt Date: Tue Dec 12 16:03:52 2006 -0800 Fix mac mini SDVO output: write the SDVO[BC] enabled register state twice. commit 9776f6c68b3cdd5585e58e677c1b1318d9aedaf4 Author: Eric Anholt Date: Tue Dec 12 16:02:47 2006 -0800 Flush the plane changes in i830_crtc_dpms() Otherwise, the changes may not have taken effect. commit 54823ac39ce9666099d69196694643f04123cb4d Author: Eric Anholt Date: Tue Dec 12 14:38:08 2006 -0800 Move the SDVO sync detection to after we've turned the port on. commit ec45d7274333cccd20ad080facc8e995ad7b26b0 Author: Eric Anholt Date: Tue Dec 12 13:54:49 2006 -0800 Extend the error state reporting to cover ESR and decode PGTBL_ERR for 945. commit 838af10b85ee8c085c6d2c4fbe18fd90b2be9d3f Author: Eric Anholt Date: Tue Dec 12 13:45:41 2006 -0800 Move PrintErrorState and CheckInheritedErrors code to i830_debug.c. commit 668ae6ce3c3c4b93a58749fd44ef32451adb6f9c Author: Eric Anholt Date: Tue Dec 12 11:39:40 2006 -0800 Set the SDVO port to on before sending the SDVO output on command. commit 91a538beadc17155be395ea2c8109a0958a969bd Author: Eric Anholt Date: Sat Dec 9 15:58:47 2006 -0800 More debugging output for SDVO. commit ce54538b905eb329f45c1b9b15e95ddce4a7927f Author: Eric Anholt Date: Fri Dec 8 08:26:30 2006 -0800 Add a hack for DDC on my Mac Mini. commit c8c1fb64ccecfb88f9923ad65c9898aa44692260 Author: Eric Anholt Date: Mon Dec 11 16:02:29 2006 -0800 Move the overlay mode switch status updates to when we enable/disable crtcs. This should help if we mode switch with randr during video playback, and also simplifies the code by moving the protection all to one place. commit 3f3791ba2ece8e8a7aa3a5676f5f6361862a292c Author: Eric Anholt Date: Mon Dec 11 15:20:15 2006 -0800 Remove remnants of old savedCurrentMode hack. Man, that was gross. commit c21baa5a3044eb66f2518b018dae07eadb989aa1 Author: Eric Anholt Date: Mon Dec 11 15:15:01 2006 -0800 Remove dead remnants of old clone mode support. commit b3865f34669319f449f2da8f87a4eef6ec8e7804 Author: Joshua J. Berry Date: Mon Dec 11 14:46:20 2006 -0800 Include i830.h to get M_T_* compat defines for older X Servers. Not duplicating the defines ensures that we won't let the compat bits slip into the X Server when we move this code over there. commit 24d9521c7ae72e146025f10ea85b07ebd7041716 Author: Eric Anholt Date: Mon Dec 11 09:34:50 2006 -0800 Add error messages for previous silent DRI initialization failures. commit bf43f8ce736d91659c2757fb39b13e63bc5891a4 Merge: 684ea26 9452321 Author: Wang Zhenyu Date: Fri Dec 8 14:17:43 2006 +0800 Merge branch 'modesetting' into crestline commit 94523215127a3458a28d0f1217224a27c8870884 Merge: 75f4df2 fde52de Author: Wang Zhenyu Date: Fri Dec 8 14:15:06 2006 +0800 Merge branch 'modesetting' of git+ssh://zhen@git.freedesktop.org/git/xorg/driver/xf86-video-intel into modesetting commit 684ea26b740d07ec8f6b0d223c1171e70c073bfa Merge: 64269de 6b977f0 Author: Wang Zhenyu Date: Fri Dec 8 14:03:10 2006 +0800 Merge branch 'exa-i965' of git+ssh://zhen@git.freedesktop.org/git/xorg/driver/xf86-video-intel into crestline commit 99f6f84076b25af69dc226e6d4704c74d207e657 Author: Alan Hourihane Date: Thu Dec 7 09:47:34 2006 +0000 Bump to 1.7.3 commit fde52de870c84821ab457e17634c334a10cf71ab Merge: 7dbe6c3 85de8a0 Author: Keith Packard Date: Wed Dec 6 23:40:13 2006 -0800 Merge branch 'modesetting-origin' into modesetting commit 7dbe6c378d1bc0c06ed6be4c1f9ca38ac31a5571 Author: Keith Packard Date: Wed Dec 6 23:39:14 2006 -0800 Move vga restore to end of RestoreHWState. I was having consistent system lockups when the vga restore was first. Moving it to the end has reduced them to an infrequent occurrence (but, alas, it has happened once since). This does not make me happy. commit 28c7002b1cf81e286a083384575d53021d3ce906 Author: Keith Packard Date: Wed Dec 6 23:35:15 2006 -0800 Use DOUBLE_WIDE mode on all hardware. DOUBLE_WIDE mode is needed when the pixel clock is > 90% of the core clock rate. The code guesses what the core clock rate is based on the bus (AGP -> 133MHz, PCI-E -> 200MHz). commit 85de8a05804eaebcd61430793ef5406f196f9a95 Author: Eric Anholt Date: Wed Dec 6 20:26:53 2006 -0800 Include render.h to get SubPixelUnknown define. commit 64269de3c34db047e78b788dc5681abdc009f8aa Merge: 8ddbc26 182d2f5 Author: Keith Packard Date: Wed Dec 6 18:09:28 2006 -0800 Merge branch 'modesetting' into crestline commit 182d2f503a7a1be90d93a4a15059585a564fd3ce Merge: 677c454 6777d80 Author: Keith Packard Date: Wed Dec 6 17:18:58 2006 -0800 Merge branch 'modesetting-origin' into modesetting commit 6777d8044d5cc063698e05afb7a93a58a7a3b64a Author: Eric Anholt Date: Wed Dec 6 17:10:06 2006 -0800 Harmless warning fix. commit c6ce8a3c3355e8a56e86856caa56baf1cd3c0231 Author: Eric Anholt Date: Wed Dec 6 17:07:48 2006 -0800 Reduce dumpregs to X startup, after a mode set, and LeaveVT. While here, remove some other register dumping that is better done by i830DumpRegs(). commit b23dec37b28a76433ad5d537ab508294e843cabe Author: Eric Anholt Date: Wed Dec 6 16:58:24 2006 -0800 Save/restore registers around calls in PreInit with register side-effects. commit d3c21c09e3904b3d32aca7563044194b061ac2fa Author: Eric Anholt Date: Wed Dec 6 16:50:52 2006 -0800 Remove stale i830_tv_pre_set_mode. Replaced by generic dpms hook call. Also replaced by not spamming unrelated ADPA register. At least, it had sure better not be related. commit 9c3e733aaa2068fcb0164577237ed70d177e9b5a Author: Eric Anholt Date: Wed Dec 6 16:36:42 2006 -0800 Save/restore more TV registers. commit d5ec9d2160f47f21a6015c1cc05b57274cbb0471 Author: Keith Packard Date: Wed Dec 6 14:44:40 2006 -0800 Cannot modify DPLL register in output mode_set function. DPLL register contains clock setup, so it cannot be written from the output mode_set function or 'bad things' happen. commit 5d345c13a949caef384a63769d7185885de6d26d Author: Keith Packard Date: Wed Dec 6 13:19:58 2006 -0800 Clean up SDVO multiplier debug output commit 6ee63364f5fabbc5578bcc9ded38c778595f5a6d Merge: 56f7119 b75ecdb Author: Keith Packard Date: Wed Dec 6 12:53:22 2006 -0800 Merge branch 'modesetting' into nonrandr-setup Also, fix buffer overflow in i830_debug.c commit b75ecdb48309a15eb7c52b279c7f8523a95bcd48 Merge: 1cb6311 b3bb10e Author: Eric Anholt Date: Wed Dec 6 11:59:33 2006 -0800 Merge branch 'generic-mode-set' into modesetting commit b3bb10e33e44c78f132d239e30931f97065a9fd6 Author: Eric Anholt Date: Wed Dec 6 11:53:40 2006 -0800 Fix LVDS: Don't change bits in LVDS other than port enable/pipe select. This behavior matches what we did before when LVDS worked. commit 47d07b1073c162ec00a8b173b8b6389dc4adbe88 Author: Eric Anholt Date: Wed Dec 6 10:56:38 2006 -0800 More LVDS fixed mode fixing: use the EDID DTD's blank length. commit 727bf1cbf72362edcbdd68001a3558fd2b2b4eca Author: Eric Anholt Date: Wed Dec 6 10:12:49 2006 -0800 Add missing newlines to new debug output. commit 236c53be5d94798d55219651a6885fee32ce175b Author: Eric Anholt Date: Wed Dec 6 10:10:27 2006 -0800 Fix LVDS fixed mode code after generic-mode-set. commit 1cb6311c1182fe98d2b8d237cef42509c3178f9a Author: Wang Zhenyu Date: Wed Dec 6 16:31:00 2006 +0800 fix typo in checking xf86CrtcScreenInit return commit 56f71194157ef929b62fe34a89c840bd59e56301 Author: Keith Packard Date: Wed Dec 6 00:13:57 2006 -0800 Use xf86 structures for default config instead of RandR. To avoid requiring RandR 1.2 in the X server, use the xf86 Crtc and Output structures as the basis for the default configuration computation (and, eventually, the config-file based configuration as well). commit 6b977f0eca388a295ca7348f5960bd8e0192cf1d Author: Wang Zhenyu Date: Wed Dec 6 13:24:44 2006 +0800 fix Makefile.am commit ffd8aacbe7c72b696ff7257609e3c1d45c057609 Author: Keith Packard Date: Tue Dec 5 19:48:39 2006 -0800 new server APIs are not in server-1.2 (version 7.1.99.2) commit a73649611414cb92697575efa90addfc3cf91cf9 Author: Keith Packard Date: Tue Dec 5 19:47:04 2006 -0800 Capture VGA registers sooner so checks are more informative commit 62fba80d3ea9c185ec8ae1e06c099c02bf844d8b Author: Keith Packard Date: Tue Dec 5 19:43:41 2006 -0800 fix up misleading comment in i830_crt.c about border register commit 041f0029c78620ca1b50e7e5d11f5d022e189641 Author: Wang Zhenyu Date: Wed Dec 6 10:43:29 2006 +0800 Formats fixes We should use card_fmt for src/mask picture, and use dest color buffer format helper. Also fix wrong name for G965 texture formats, and pict_x1r5g5b5 isn't supported by sampler engine. commit 1d94ec7de3387b70815679977cfc5d88200efa99 Author: Eric Anholt Date: Tue Dec 5 18:22:59 2006 -0800 Fix ordering of PIPE[AB]SRC debug output to be (x, y). commit eee23fdd0dadd38a2b80525ec886120150f91077 Author: Eric Anholt Date: Tue Dec 5 18:19:49 2006 -0800 Fix copy'n'paste-o of the ordering of h/v fields in DSP[AB]SIZE. commit 58c247c6de56508b00eadf77b4c145c94134270d Merge: 0b4c3e7 2e8c927 Author: Eric Anholt Date: Tue Dec 5 15:28:21 2006 -0800 Merge branch 'modesetting-origin' into modesetting Conflicts: src/i830_display.c commit 0b4c3e7bff790e1e99ace5036a41e96046335b04 Author: Eric Anholt Date: Tue Dec 5 14:39:47 2006 -0800 Add a bunch of per-register debug code to i830DumpRegs(). commit 7844e576e71c004ed495bbc31f7bbe890d18b8d9 Author: Eric Anholt Date: Tue Dec 5 14:39:09 2006 -0800 Move reg dump from i830PipeSetMode to after we set up the modes with RandR. commit e777d38ce98d7220621b049b09df1deca5a5df42 Author: Eric Anholt Date: Tue Dec 5 10:01:31 2006 -0800 WIP code to move mode set sequencing to XFree86 handlers. It compiles. It definitely doesn't run. commit 2e8c927f9308069a82f25b65bb0c62bc5a156832 Author: Keith Packard Date: Mon Dec 4 14:02:30 2006 -0800 Re-create RandR Crtc/output structures on server regen. RandR structures must be re-created when the server reinitializes, but the driver PreInit function is not re-invoked. Recreate them manually in this case during ScreenInit. commit 583619cdb2b6a469299dd5bf658bf5b51ee999fe Author: Wang Zhenyu Date: Mon Dec 4 15:48:04 2006 +0800 set correct default border color commit f28d5fc30ccf60d72d5832bb6542522dc7f1c463 Author: Wang Zhenyu Date: Mon Dec 4 15:47:31 2006 +0800 fix typo in ps kernel fix corrupt in some subspans commit 677c454469a6223460778033758e384e989cea47 Merge: 16f9a56 8fcf9a8 Author: Keith Packard Date: Sat Dec 2 23:28:38 2006 -0800 Merge branch 'modesetting-origin' into modesetting commit 8fcf9a81179ee8577ddab5e904c58fbfd14cf59c Author: Keith Packard Date: Sat Dec 2 22:58:31 2006 -0800 DOUBLE_WIDE mode for high pixel clock 8xx. Rewrite PLL search. High pixel clock modes on pipe A of an 8xx chip require DOUBLE_WIDE mode. It's supposed to be modes > 180MHz or so, but the board I have requires DOUBLE_WIDE mode for clocks > 108MHz or so. The limit is related to the core clock speed of the chip, which can be found indirectly through PCI config space. None of the possible values explain why this board needs this mode for these relatively low clock rates though. Also, create tables of data for the PLL computation and use them instead of code. I think it's cleaner looking. It is also untested on 9xx. It'll work. Really. commit 81dde11d419c8f9198ab3502d9813d66d0bc6d6d Author: Eric Anholt Date: Fri Dec 1 10:37:19 2006 -0800 Fix copy'n'paste-o from restructure-outputs that resulted in crashes. Our driver private is stored in the bus that we set up, not the DDC device that xf86 code does. commit a504e31f0fefdec802ae403573f52fac3d577b65 Author: Eric Anholt Date: Fri Dec 1 12:49:43 2006 -0800 Don't try to write the read-only PP_STATUS register. Also, don't bother reading the PP_CONTROL register to try to get LVDS status -- that's what PP_STATUS is for. commit f21230d3e1cc5470e8f88404b5d9f1702bec44bb Author: Eric Anholt Date: Fri Dec 1 12:48:57 2006 -0800 Add the [ax]4r4g4b4 source picture formats. commit 8ddbc26b276016b922487f7b1e8997998b655a7d Merge: 93975a5 16f9a56 Author: Keith Packard Date: Fri Dec 1 10:57:16 2006 -0800 Merge branch 'modesetting' into crestline Use new TV output condition (all 9xx except 915g) commit 16f9a56070a01c7ada8e566d9b73fcf7900136e0 Author: Keith Packard Date: Fri Dec 1 10:50:27 2006 -0800 TV output exists on all 9xx chips except for 915G commit 93975a526086432cdeb8d7ead721a862d632c1e7 Merge: 08b6569 e603cd0 Author: Keith Packard Date: Fri Dec 1 10:28:58 2006 -0800 Merge branch 'modesetting' into crestline commit e603cd0c73344ef137d3276b5cfcbcf4df340778 Author: Keith Packard Date: Fri Dec 1 10:08:36 2006 -0800 Leave detected TV status alone when no free CRTC is available. Yes, this means not detecting TV hotplug when two outputs are already running. An alternative would be to turn off one of the other outputs temporarily, but that would cause flashing. Something to consider. commit ec30356d950199903978265076b8a4e3960cf84f Author: Keith Packard Date: Fri Dec 1 10:06:29 2006 -0800 Do output detection before any crtc allocation. Some output detection requires a crtc for load detection, perform all of the output detection before allocating any crtcs so that there will be a free crtc for any load detection. Avoids losing TV detection when two monitors are connected. commit 7642da82781826cb3b45078750cc54d8f592893a Author: Keith Packard Date: Thu Nov 30 23:21:39 2006 -0800 When CRTC is disable through RandR, the mode is NULL. Don't dereference it. Correctly interpret mode == NULL as CRTC disable, make sure the CRTC is disabled in this case. commit de50ca9ed9e78c17be1badec62e35ed5e4d62b05 Author: Eric Anholt Date: Thu Nov 30 17:56:47 2006 -0800 Tristate the clock/data pins during GPIO when released while getting values. While the register is laid out suggesting that you can read a low value while driving the output high, and the I2C spec seems to indicate that you should be able to as well, and on some hardware this works successfully, on the i865 and perhaps some other chips it doesn't. So, if we're not holding the clock or data pin low during GetBits, tristate the pin so that we can successfully read. This fixes i865 analog (VGA) DDC so it successfully sees slave acks. Also, improve the I2C bit-banging debugging. commit 8c44556408601db2be799b5ac5c4da1d92886d31 Author: Eric Anholt Date: Thu Nov 30 15:56:52 2006 -0800 Preserve some GPIO bits that the docs tell us to. commit 28401b25cc5c1afb272e506cef196ba3a1258bd3 Author: Eric Anholt Date: Thu Nov 30 15:45:13 2006 -0800 Add debug code for the bit-banging I2C implementation. commit ff64bc7397cafbec94e388e1625b3a1999f1aca7 Merge: b94b7c4 16e01b1 Author: Keith Packard Date: Thu Nov 30 14:09:31 2006 -0800 Merge branch 'restructure-outputs' into modesetting. Outputs and Crtcs now have a driver-independent representation which should permit generic code to control RandR 1.2 and startup configuration. commit 16e01b117bb7ae90b150dd4f25a887dd895cf473 Author: Keith Packard Date: Thu Nov 30 11:55:29 2006 -0800 Oops. Don't try to use xf86RandR names, use xf86RandR12 instead. Attempts to override functions in core server fail, so use new names instead. commit bcf206ff17c6eca8212db72dd86fcd8fdc0162da Merge: 27ca1c1 2c8c310 Author: Keith Packard Date: Thu Nov 30 11:39:56 2006 -0800 Merge branch 'restructure-outputs-origin' into restructure-outputs commit 27ca1c17aab30670dd1d07e0809fd58cdecfc3b6 Author: Keith Packard Date: Thu Nov 30 11:39:37 2006 -0800 Rename I830 randr functions to xf86 as they are becoming generic. Change function names, create separate i830_randr.h header file to contain definitions. commit b94b7c4bcfdb7ba59ed818f72309b5060a2ab7ee Merge: b6fc8df 25e6e49 Author: Eric Anholt Date: Thu Nov 30 09:15:30 2006 -0800 Merge branch 'exa' of ../xf86-video-intel into modesetting Conflicts: man/i810.man src/Makefile.am src/i830.h src/i830_driver.c src/i830_rotate.c src/i830_video.c commit b6fc8df9a52f5fe1b4d26ae06bc4d48235b44a67 Author: Eric Anholt Date: Wed Nov 29 15:06:32 2006 -0800 Properly detect the GTT size on the G965. In the past, the GTT has always been sized just large enough to map the whole graphics aperture. However, apparently on the G965 that isn't the case, and it is actually 512KB on hardware with a 256MB aperture. This resulted in X not bothering to allocate memory for 256KB that it thought was already mapped into stolen memory, and thus garbage rendering (particularly visible in large video modes that displayed this unallocated memory). The kernel happens to get the right answer by hardwiring a 512KB GTT size already, but that may not be true on future hardware. Instead, we use a convenient field in PGETBL_CTL that's specifically for the GTT size rather than the aperture size, which gets us the answer we want. commit 359dc81c07901665da0f86c573c096fa1661cdd2 Author: Eric Anholt Date: Wed Nov 29 15:01:39 2006 -0800 Revert "Don't allocate stuff in the first 256K of video memory (GATT?)" This reverts commit 997e8c9bb4235cab1fff4738387df9afcbea0a03. The GTT is definitely located at the end of stolen memory. This commit apparently worked around mis-estimation of the GTT size. commit db391e8e4c4d87bfe3ccad0de14dd5b47b69b8fe Author: Wang Zhenyu Date: Wed Nov 29 17:16:46 2006 +0800 shut up warning commit 290f15cd4cda97727ebcaadacbbbf7650278934b Author: Wang Zhenyu Date: Wed Nov 29 17:14:55 2006 +0800 fix alpha blending state commit 3d4edd325f3859c749ee42df102bb4239eac5287 Author: Wang Zhenyu Date: Wed Nov 29 17:05:32 2006 +0800 Add in sf/wm program for mask picture without CA commit a704120b15efae47344a90d972e7f3da64a202a6 Author: Wang Zhenyu Date: Wed Nov 29 16:52:44 2006 +0800 misc cleanup for G965 vs/sf/wm states commit e3c70c68e39183226e498271c44e98ef1b96a681 Author: Wang Zhenyu Date: Wed Nov 29 16:40:15 2006 +0800 WM kernel needs scratch space commit aa515c54f0cfd9025fc38dc4b7938ff17a8a13fb Author: Wang Zhenyu Date: Wed Nov 29 16:37:06 2006 +0800 Setup default border color for our samplers commit b6eba96584bcd2c024f6443d9f3728eb65b234fb Author: Wang Zhenyu Date: Wed Nov 29 16:30:53 2006 +0800 fix vertex buffer size commit 453842c9ff733af45fa665d9db6a35164f45c60a Author: Wang Zhenyu Date: Wed Nov 29 16:24:24 2006 +0800 clean up issue cmd to ring buffer Make it easy to track different part of ring state, and use rectlist primitive instead. commit 18ad7d5cf04081d89a9f978ccc7794116f7c498b Author: Wang Zhenyu Date: Wed Nov 29 16:04:16 2006 +0800 remove wrong scale_units commit 3d5bd0c14eea7951540f7a12eee257428f78e2d1 Author: Wang Zhenyu Date: Wed Nov 29 16:02:16 2006 +0800 Rename exa sf/wm program files Also fix some minors in wm program. commit a06beb5f80f097fac3b718e742742bb32f1c1194 Author: Wang Zhenyu Date: Wed Nov 29 15:47:19 2006 +0800 EXA state mem for G965 Not split offscreen mem for exa, but alloc a dedicated one for G965 states. commit 9e4e7d4fa25a64a2494e7531967599142e60e716 Author: Eric Anholt Date: Tue Nov 28 16:36:09 2006 -0800 Conditionally compile old-server compatibility code in. This also replaces calls to compat code with the real names of the functions, and slips #defines to an i830-namespaced version in when doing compat. The current server version (7.1.99.2) is still left as requiring compat code, since the version hasn't been bumped yet. This also fixes some failures to call the compat code, and some failures to actually compile the compat code. Oops. commit 8d9ba2405d1f1e58a0cfb570e83e9206af8525b5 Author: Eric Anholt Date: Tue Nov 28 15:44:06 2006 -0800 sparse cleanups (use NULL instead of 0 for pointers) commit dbb8135e10e943a4d4e65f95853eeccc764ccf24 Author: Eric Anholt Date: Tue Nov 28 13:21:31 2006 -0800 Remove printing of the XFree86 screen mode list, and some other detritus. commit e8fb40ab1063b18372be2b5996257eb69d3f15c4 Author: Eric Anholt Date: Tue Nov 28 11:17:21 2006 -0800 Replace custom DDC mode list code with a copy of the X Server's. The X Server now has most of the mode helper functions we want, and we can start removing compilation of our copies when the server is new enough. commit 2c8c310bd8a105dbeda5c989dca31cc766eb6bb8 Author: Keith Packard Date: Mon Nov 27 22:18:59 2006 -0800 Driver-independent code tracks num_crtc itself, don't do it in driver. Now that driver-independent code manages array of crtcs, let it track the count and just have the driver compute how many there are in a separate variable. commit c421724d8151d1d4578468abdbe8fe1120e373ca Author: Keith Packard Date: Mon Nov 27 22:15:40 2006 -0800 Move #include "randrstr.h" from i830.h to i830_xf86Crtc.h As part of the separation of driver-dependent from driver-independent code, the interactions with RandR are moving entirely out of the driver, which means the driver shouldn't even know about RandR, while the driver-independent code does. commit e8d1db326388c9a112f2b8e2eff5a81210b8d6e4 Author: Keith Packard Date: Mon Nov 27 12:01:47 2006 -0800 Move crtc/output config to sub-structure. Place crtc/output in separate structure at head of driver private structure. Use this from the config code to make it driver-independent. Still lots of effectively driver independent code that continues to use driver dependent stuff, but that will change. commit 72692ba2e0254460c4f5a8cd476f5748383390ae Author: Eric Anholt Date: Mon Nov 27 11:23:55 2006 -0800 Postpone SDVO DDC bus creation until we've detected the SDVO device. This reduces log noise for those of us with no SDVO devices. commit 5f38bc3e2a685da5c03efa8103a353f79c842aa7 Author: Eric Anholt Date: Mon Nov 27 11:06:50 2006 -0800 Reduce the severity of many informational log messages. commit 85f404bc67d5ab54175c19e75aad406abfcf2133 Author: Eric Anholt Date: Mon Nov 20 09:32:26 2006 -0800 Re-indent ch7xxx driver. commit a0518f5a440630365b1d935b7c2d0725f326ad51 Author: Keith Packard Date: Sun Nov 26 19:44:38 2006 -0800 Remove mode setting from load detect CRTC allocation. To share load-detect CRTC allocation with TV driver, move it to the output driver commit a47c549df036990e29f05bc3df80e1a2ab9f3b3c Author: Keith Packard Date: Sun Nov 26 18:26:26 2006 -0800 Clean up reworked data structure code so the server actually starts. Use i830GeLoadDetectPipe again (instead of missing xf86AllocCrtc). Actually create new Crtc structures. Fix a few other NULL pointer dereferences. commit 2529863a1ade782819d76be2d0dc16e89028c1e3 Author: Keith Packard Date: Sun Nov 26 16:44:17 2006 -0800 Start output/crtc restructuring work. It compiles. Outputs and Crtcs are now split between 'generic' and 'driver specific' pieces in the hope that more code will be able to migrate to the xf86-generic layer. Right now, the code remains tangled together, significant work remains to tease the pieces apart. First the code needs to be made to actually work as-is though. commit 9aca4e207440119f4280b78199a221f85d50c511 Author: Keith Packard Date: Wed Nov 22 15:52:32 2006 -0800 Remove custom configuration code (which was #ifdef'd out). Multi-screen configuration is moving to generic code. commit 850e3652f4d5bb57f3c4ccb9cf29e2c75fecba9d Author: Keith Packard Date: Wed Nov 22 15:36:43 2006 -0800 Note which pipe has failed to be configured for VBlank interrupt commit 659e3db925de91954421f7f33c072576dfefcc4a Author: Keith Packard Date: Wed Nov 22 15:35:59 2006 -0800 Set up maximum screen size for clone instead of panorama. This avoids heading past the 8Kbyte stride limit of DRI so that 3D will be available by default. commit 03c12f866529230df7a1503de61954cd5ade242e Author: Keith Packard Date: Wed Nov 22 15:34:03 2006 -0800 Limit DRI use to cases when pitch < 8Kbytes, not 4096 pixels. 3D rendering pipeline is limited to pitch of 8Kbytes and not 4096 pixels, which rather makes a difference at 32bpp. commit 6781575f734f05547d7d5ceef4116fc157bba44d Author: Alan Hourihane Date: Wed Nov 22 16:49:57 2006 +0000 Be a little more verbose when rejecting modes. commit e065324661ad08b3b359136f48090232f6138959 Author: Alan Hourihane Date: Wed Nov 22 16:49:36 2006 +0000 update port attributes immediately when overlay is on. commit a0a9d1ea2ac9a0c665e5a1de5b8b8358dee3b9b4 Author: Keith Packard Date: Mon Nov 20 17:25:00 2006 -0800 Set configured values for screen virtual size and initial frame. Computation for virtual size and initial frame origin is quite broken in xf86 common code. commit d6a0f917e601ea36643c6ad857756e19d24ecd73 Author: Keith Packard Date: Mon Nov 20 15:17:32 2006 -0800 Enable second SDVO channel. Rework SDVO support so that it can deal with two channels correctly, also save/restore all connected output timings. commit 08b6569dc663ddf38cb36a6875de6d4ab55acac9 Author: Eric Anholt Date: Sun Nov 19 16:15:45 2006 -0800 Enable TV output on Crestline (untested). commit 05202cabbd23f15330b811ae6b8d708ad042bc40 Merge: ee502dd b945a65 Author: Eric Anholt Date: Sun Nov 19 16:14:18 2006 -0800 Merge branch 'modesetting-origin' into crestline commit b945a650e952f98c2d101b71bd3ec0f390478da5 Author: Keith Packard Date: Sun Nov 19 00:54:30 2006 -0800 Fix TV color key. Subcarrier defines were incorrect in header file leaving one of the DDA phases disabled. commit 28224af3d90a1a08d54a865dfaf20184330fe8a4 Author: Keith Packard Date: Sun Nov 19 00:40:46 2006 -0800 Preliminary 945 TV output. Color key is broken. Fixed mode. TV output is generating video with this patch, but the color burst signal is incorrect somehow. commit ee502dd92a3dfccdc2efcfb76fc652694bc89e3b Merge: 15ff17c 2fe6107 Author: Eric Anholt Date: Fri Nov 17 14:57:41 2006 -0800 Merge branch 'crestline-origin' into crestline commit 15ff17c756e42f392306820e3f7ffbdcc56b9892 Merge: 9cc2f33 816fc1a Author: Eric Anholt Date: Fri Nov 17 11:48:53 2006 -0800 Merge branch 'modesetting' into crestline This works for analog, but SDVO output appears to not work yet. Conflicts: src/i830_driver.c commit 816fc1a76a5ac738e41b172ba8f43137c1521328 Merge: 9948d83 7a7bb33 Author: Eric Anholt Date: Fri Nov 17 00:05:53 2006 -0800 Merge branch 'modesetting-origin' into modesetting commit 7a7bb331e10498e5b8ccec58130bb23334d36562 Author: Keith Packard Date: Thu Nov 16 21:19:20 2006 -0800 Don't dereference null DisplayModePtr on disabled output. During initial configuration, outputs which are disabled have null modes. commit 2fe6107ac68e86ed183d8602436633348340678d Author: Wang Zhenyu Date: Fri Nov 17 13:11:36 2006 +0800 Rotation support for 965GM commit c4508c1cadf323e9ef1d0e69dd77d5e841a6a978 Author: Keith Packard Date: Thu Nov 16 21:09:23 2006 -0800 RandR-based initial output configuration. Using pre-init computed RandR information, make reasonable default choices for the output configuration at startup time. Either some preferred size or a size which yields 96dpi is chosen, from which other monitors are set to a similar size. The largest size sets the screen size. This needs to be extended to respect config file settings, but those have not been defined yet. commit 9948d8377d70e898260c12210151b952b3fb8bf1 Merge: 45a27f8 b649f95 Author: Eric Anholt Date: Thu Nov 16 19:47:24 2006 -0800 Merge branch 'master' into modesetting Conflicts: src/i830.h src/i830_cursor.c src/i830_dri.c src/i830_driver.c src/i830_video.c commit 9cc2f3313d77487dea372e6ab32d6d9c06617ae3 Author: Eric Anholt Date: Thu Nov 16 15:12:43 2006 -0800 Replace broken PCI resource size detection with pciGetBaseSize() call. commit 45a27f80e1c783627f570c309e7a853dcc9af0c1 Author: Keith Packard Date: Thu Nov 16 13:38:35 2006 -0800 Remove output options. Let outputs be connected to other crtcs. Output options will be replaced by properties. Permits outputs to be connected to arbitrary CRTCs (within hardware limits). No cloning yet. commit 9aea79d1e954fe4fb5c101edcb7c2d0f706a5c4c Author: Keith Packard Date: Thu Nov 16 11:40:26 2006 -0800 Inverted boolean sense when selecting pipe for CRT detection. Pipes are available when they are not in use. commit 4889b9f33336c92f07aac86d75f50316db6ef81a Author: Eric Anholt Date: Wed Nov 15 15:26:31 2006 -0800 Enable the LVDS if we find it and assign it to a pipe (oops). commit 854ff826c0031a44e874239c0b0a3533f4d9e14b Author: Eric Anholt Date: Tue Nov 14 09:14:24 2006 -0800 Disable setup of the second SDVO device until we fix it. commit d51555fba4e57c059fd184c1e54822d7e5b62a2f Author: Eric Anholt Date: Fri Nov 10 14:40:40 2006 -0800 Fix clock range for single-channel LVDS. commit dd1dcfab0ab0f2d0c25077fa663209e2762f26e8 Author: Eric Anholt Date: Fri Nov 10 11:08:02 2006 -0800 Fill in some of the high bits of mode timings for SDVO. commit 3955f044cfe1d592bcb36c43c539fb4a75840b8d Author: Eric Anholt Date: Thu Nov 9 20:28:46 2006 -0800 Expose the DDC-probed EDID data as the EDID_DATA output property. commit 0f5886689d7ef7dbbef6425d5c855ac6b67d3350 Author: Keith Packard Date: Wed Nov 8 23:19:59 2006 -0800 Create RandR 1.2 objects in I830PreInit. Creating the objects early will allow the driver to use randr structures to select a reasonable configuration. That part has not been done yet. commit 679c7bd82639a09cdce133becb8a08629ce3a4e9 Author: Keith Packard Date: Wed Nov 8 21:39:28 2006 -0800 ignore edited man page commit 81b7b489afa2cab4d8614c64f4906be627f1d07e Author: Keith Packard Date: Wed Nov 8 21:38:00 2006 -0800 Adapt to RandR updates that split object creation from screen association. RandR DIX code is preparing for xf86 drivers that want to allocate RandR objects at PreInit time. This patch adapts to that change without taking advantage of it. commit 81bace0c316c3ed80201a34eca533254d12cd193 Merge: 713c5b0 beb8916 Author: Keith Packard Date: Wed Nov 8 20:23:20 2006 -0800 Merge branch 'modesetting-keithp' into modesetting Conflicts in PipeSetMode were resolved to use the keithp changes that pushed more modesetting stuff into the per-pipe function. Switched availablePipes to num_pipes. Used modesetting default output configuration. commit 713c5b0899428edfea7cea0780244488115dbe1d Author: Eric Anholt Date: Wed Nov 8 19:55:31 2006 -0800 Change the output and pipe "is it on/off" field name to "enabled". commit ff77e9d84f2037b99a8e4ac55da0b0ac92ed6b9b Author: Eric Anholt Date: Wed Nov 8 19:53:31 2006 -0800 Clean up i830_crt_detect_load() a bit more. ADPA might not have been set right in some cases (DPMS-off monitor, for example), and a wait for vsync that the bios does was missing. commit 9b267014b55f49d9362c1d432d6ba19ddd0ad95d Author: Eric Anholt Date: Wed Nov 8 19:39:18 2006 -0800 Go back to only setting up outputs that have a display connected. commit 35ab689bbde5f74752598cd743d735640486b639 Author: Eric Anholt Date: Wed Nov 8 19:35:49 2006 -0800 Fix i830DisableUnusedFunctions after pipe structure change. Using "pipe" instead of the index "i" meant pipe(3) got referenced instead of a nice small integer. Oops. commit b649f95ea6fd3555d073fdbf8f2f035dfe1afd33 Author: Alan Hourihane Date: Wed Nov 8 19:01:25 2006 +0000 Disable some debug message commit 75f4df278e9db360967d77cdba4756cbde622d56 Author: root Date: Wed Nov 8 13:56:32 2006 +0800 855 fix commit f1ff01e31eb8e9dc05190bf1a8b318d4f587f64a Author: Keith Packard Date: Sat Nov 4 00:46:18 2006 -0800 Eliminate operatingDevices member and PIPE_* values. operatingDevices and MonType1/MonType2 duplicate information already stored in the device structures. Eliminate them and replace uses with direct references to the appropriate other data. (cherry picked from 3ab7f9693217d8fe993bdc94c376b219b0082961 commit) commit 0b2d36d4f038c4e8fa08632b6f1368627f010392 Author: Keith Packard Date: Fri Nov 3 23:29:12 2006 -0800 Use pI830->availablePipes instead of MAX_DISPLAY_PIPES everywhere (cherry picked from e4bcec796e80e9fd66ab0c36394f5946915531f1 commit) commit d0ef9e99acb9e999e1b6d3eb76edc6355555043b Author: Eric Anholt Date: Mon Nov 6 18:30:46 2006 -0800 Restore PFIT_CONTROL before turning the LVDS back on in the restore method. commit 94a3731c2b4f2ea2e696a8c87dccc0d214d41e8e Author: Eric Anholt Date: Mon Nov 6 18:26:48 2006 -0800 Move PFIT_CONTROL disable for G965 up before post_set_mode. Also, remove setting of some other random registers that appears to have been spammed in at the same time, and don't try to disable on the I830, before this register existed. commit beb89163d73376e70870e6e2a6b19863f3a058b1 Author: Keith Packard Date: Sun Nov 5 19:06:45 2006 -0800 DSPSURF must be page aligned. Place intra-screen offset in DSPBASE. DSPASURF/DSPBSURF can only take page aligned values, ignoring the lower order bits. So, place the offset for the output within the frame buffer in the DSPABASE/DSPBBASE registers instead. commit 997e8c9bb4235cab1fff4738387df9afcbea0a03 Author: Keith Packard Date: Sun Nov 5 18:56:33 2006 -0800 Don't allocate stuff in the first 256K of video memory (GATT?) Letting the ring buffer or other objects be allocated within the lowest portion of memory appears to trash some memory mapping data; I'm assuming this is the GATT table on the 965. Just marking this out of bounds for allocation fixes this problem. commit 5a355c72614ed77f2000e5ede45f3ff5990c79d9 Author: Keith Packard Date: Sun Nov 5 18:51:28 2006 -0800 Fix CRT output on 965 chipset. A few more register settings are needed to get CRT output working on the 965 chipset, in particular the the SDVO/UDI clock multiplier register needed to get set to the default value (3). No, I really don't know what this does, but it does get the CRT running at a wide range of sizes. commit 49a6bea7d969dbfd1dd542c0c3e02abc330d6850 Merge: 64447c7 fbb376b Author: Wang Zhenyu Date: Mon Nov 6 10:25:23 2006 +0800 Merge branch 'master' into crestline commit 68c3185046b27ab936ca6c92b924b443b3cd6fce Author: Keith Packard Date: Sun Nov 5 13:30:32 2006 -0800 Avoid crashing when disabling sdvo output. XXX commit 15ef08046bcc3e746453301379f7c5d1bf929ee1 Author: Keith Packard Date: Sun Nov 5 13:29:56 2006 -0800 Move remaining pipe mode setting logic to i830PipeSetMode commit 7fcb555735a58e19ccc10875b211402983170a87 Author: Keith Packard Date: Sat Nov 4 00:52:21 2006 -0800 Rename availablePipes to num_pipes commit 3ab7f9693217d8fe993bdc94c376b219b0082961 Author: Keith Packard Date: Sat Nov 4 00:46:18 2006 -0800 Eliminate operatingDevices member and PIPE_* values. operatingDevices and MonType1/MonType2 duplicate information already stored in the device structures. Eliminate them and replace uses with direct references to the appropriate other data. commit e4bcec796e80e9fd66ab0c36394f5946915531f1 Author: Keith Packard Date: Fri Nov 3 23:29:12 2006 -0800 Use pI830->availablePipes instead of MAX_DISPLAY_PIPES everywhere commit b7262a9a9110dac66e1a92c39dcb3ab59d95d081 Author: Keith Packard Date: Fri Nov 3 23:24:07 2006 -0800 Finish removing persistant vbe data commit 4625073244d4f521a07e12adcf0609e85658acbe Author: Keith Packard Date: Fri Nov 3 23:23:38 2006 -0800 Oops, martian memset of randr modes pointer commit 719ad68515be9b996a6314de5448843de1146b88 Author: Keith Packard Date: Fri Nov 3 19:41:41 2006 -0800 Use VBE only temporarily to fetch BIOS rom image (cherry picked from 6a9386651785afc70a29e355255e8295b321f28e commit) commit 27df2ff7908ea7ea2943a5f3445e12dbc24d97c9 Author: Eric Anholt Date: Fri Nov 3 15:55:10 2006 -0800 Report pipe status (and status mismatches) in i830DescribeOutputConfiguration() commit ecbe73b940b2d642115de4b73c2f757eb46ff956 Merge: 561af00 9681602 Author: Eric Anholt Date: Fri Nov 3 15:59:59 2006 -0800 Merge branch 'modesetting-origin' into modesetting Conflicts: src/i830_display.c commit 561af007974b8cdad1eea907fb73ed9d430c21ac Author: Eric Anholt Date: Fri Nov 3 15:26:14 2006 -0800 Add support for load-based CRT detection. commit e416b426d83de031441ada7a77b6bd66cec8b5c9 Author: Eric Anholt Date: Fri Nov 3 15:25:41 2006 -0800 Print out modelines as info, not error (which had been used for debugging). commit 282a9e073ea985cbf0d0f3f296d593af1426bad5 Author: Eric Anholt Date: Fri Nov 3 13:46:09 2006 -0800 Don't memset the modes pointer on init, which was dereferencing NULL. commit 9681602177124e84a817a1e1d428f1779f2a45c9 Author: Keith Packard Date: Fri Nov 3 12:55:25 2006 -0800 Create I830PipeRec to hold pipe-specific data. Remove unused I830 members. I830 contained six parallel arrays for pipe-specific data; these have been moved to a I830PipeRec structure instead. I830 also contained several unused members: unsigned int bios_version; Bool newPipeSwitch; Bool fakeSwitch; int fixedPipe; These have been removed, along with the code that set them. commit 0510671a6c5233468ac20f0ec8096e084df03ce6 Author: Eric Anholt Date: Fri Nov 3 10:58:23 2006 -0800 Fix a pasteo in I965 register restore. commit 2c9ab6e0594769274f2dbcdf7c00fe297fc385d5 Author: Keith Packard Date: Thu Nov 2 13:44:55 2006 -0800 set the v_sync_off_high to zero. XXX should check docs commit 56f6d4f1bb67f447500af3f4f7fa557c3e887baa Author: Keith Packard Date: Thu Nov 2 13:42:17 2006 -0800 Disable the panel fitter when not using it. Cleans up SDVO DVI output. The panel fitter appears to exist on the 965 hardware (at least) and causes troubles with DVI output over SDVO when enabled. This patch checks to see if the panel fitter is pointing at the pipe being configured and disables it unconditionally in that case. The LVDS driver will configure it correctly if necessary afterwards. commit f22d9bcc25aea19ba38d35282367b591fd1b7ca0 Author: Keith Packard Date: Thu Nov 2 13:34:45 2006 -0800 Add another couple of new registers commit a9eac38bcdb49df2ce1122b49bd8b1eb19e8cae5 Author: Eric Anholt Date: Thu Nov 2 13:24:54 2006 -0800 Remove duplicated register defs that were just added. commit 7887c76062b7c79e14fb8e4f13486aa592dcbce8 Author: Eric Anholt Date: Thu Nov 2 12:27:21 2006 -0800 Add airlied's I2C code, ifdeffed out. I've gone back to compare our behavior to it several times, so I'll just keep the code in tree for now. commit 87b15cfbf762468d4b8728b3e7a39c76654017de Author: Eric Anholt Date: Thu Nov 2 11:30:21 2006 -0800 Remove dead specifiedMonitor field. commit 2636d68663a02f6d9eaf36971706b67036ebf56c Author: Keith Packard Date: Thu Nov 2 11:57:11 2006 -0800 Dump more registers for debug purposes commit 786ec54c4c1540f4aced63ef21d567c3b9f3282e Author: Keith Packard Date: Thu Nov 2 11:56:50 2006 -0800 Add a few more registers from the 965 spec commit 85e32ad2dadcce1134fcadb14ece8ff30f3925f2 Author: Keith Packard Date: Thu Nov 2 11:56:12 2006 -0800 ch7xxxSaveRegs receives real type instead of void * commit ffbd6ca09bc2300bf967d7c248a559d85b8706e0 Author: Eric Anholt Date: Wed Nov 1 13:05:44 2006 -0800 Remove dead VESARec struct. commit 97c3a1b2421031e41f0b2b1630fde1dc4262d264 Author: Eric Anholt Date: Wed Nov 1 13:04:08 2006 -0800 Remove the refresh rate appended to some mode names. This gets the SDVO and CRT outputs I have to have at least 1 common mode according to RandR. commit 7971c401554c218c84a8c45335c9b31bbccfece7 Author: Eric Anholt Date: Wed Nov 1 12:42:56 2006 -0800 Attempt to pull monitor physical size information out of DDC EDID data. commit fb94c1210966f7875e5f034f10ea31c06c502c3a Author: Eric Anholt Date: Wed Nov 1 12:23:50 2006 -0800 Move mode lists from per-pipe to per-output. This should let RandR do the right thing in exposing the modes to userland. As a side effect of getting this working, the SDVO pixel clock range code was fixed and the mode valid tests for various outputs got extended. Also, LVDS grew a get_modes for the fixed panel mode. Note that we now no longer do automatic enabling of outputs at xrandr -s 0, hotkey, or VT switch. That will be left to generic RandR code later. Also, generic modes and user-defined modes are once again not validated into the lists, so this is a regression there. commit f30d7f912f36b110c3af7dc795e35456593781ab Author: Eric Anholt Date: Wed Nov 1 11:50:51 2006 -0800 Update for the move of RandR phyiscal size information. commit 64447c7a059775e7ea8649f4714df7565e932c60 Author: Zou Nanhai Date: Wed Nov 1 14:36:20 2006 +0800 walk around to VBIOS bug in Crestline B0 commit 7195dfabd56239f08cdd8175a2ef3a66ef9600de Author: Eric Anholt Date: Tue Oct 31 17:10:08 2006 -0800 Give each output a get_modes function and expose those modes through RandR. The get_modes should return the probed modes only. The driver should then append to the list (for example, compatible modes listed in other outputs, or standard VESA modes) to create the list to expose through RandR. That isn't done yet. commit cc3728be2481637dda321d3bc2e4e89a220699cd Author: Eric Anholt Date: Tue Oct 31 15:00:36 2006 -0800 Add compat definitions for M_T_PREFERRED and M_T_DRIVER for older X Servers. commit a71f283650e8cb7b760e5a53c4db79202c4cc5c4 Author: Eric Anholt Date: Tue Oct 31 14:46:23 2006 -0800 Connect output detection up to RandR. commit 68cef9f4e028755bbf3e1862da2ef47d46ddaa6a Author: Eric Anholt Date: Tue Oct 31 14:32:00 2006 -0800 Move output connection detection to a per-output method. This will be used by RandR, and should let us clean up some of the initial display configuration, hopefully. Also, analog hotplug-based detection is now enabled on G965. commit 9fd719fce27f916ab5120f6e1234affa14eaed9d Author: Eric Anholt Date: Tue Oct 31 14:29:44 2006 -0800 Move SDVOB_PRESERVE_MASK next to SDVOC_PRESERVE_MASK. commit 4f5d4d8870fc2784192f95a561163cf4fc8737ac Author: Eric Anholt Date: Tue Oct 31 13:37:23 2006 -0800 i830SetLVDSPanelPower is now a static function in i830_lvds.c, so remove it. commit 49bbdf16c02107c08169f8d2b6e9c6dbd7d8cd95 Author: Eric Anholt Date: Tue Oct 31 10:44:45 2006 -0800 Fix many inconsistencies in the SDVO code compared to the spec. Also, fix some struct padding so that the right bits are sent out. commit e7d546cac06767ec58325396a3bb5780b2257c53 Author: Eric Anholt Date: Mon Oct 30 11:24:43 2006 -0800 Remove some dead code from BIOS modesetting. commit 2ca57040b0cd24ad3dbe693789091e28be4e69f8 Author: Eric Anholt Date: Mon Oct 30 11:19:19 2006 -0800 Remove GetBIOSVersion(). This info hardly useful now that we don't use the BIOS for mode setting. commit 837b2f632062bc29268f109895a577bd90cabd6d Author: Eric Anholt Date: Mon Oct 30 11:17:55 2006 -0800 Warning fix. commit bca9e6ccbd14eb8f2f103e8e64b28a623113d494 Author: Eric Anholt Date: Mon Oct 30 11:17:27 2006 -0800 Remove SetPipeAccess and now-unnecessary VBE reinit. commit 819a47b27cd4728feb269a08be32403304993ffa Author: Eric Anholt Date: Mon Oct 30 09:50:33 2006 -0800 Use the new fields for SDVO pixel multiply on the G965. This should fix display at resolutions/refresh rates in a different multiplier class than the console display (generally, high resolution modes). commit 71545db4614cfc4650acc4325912474e777c3b36 Author: Eric Anholt Date: Mon Oct 30 09:46:10 2006 -0800 Return and use valid status bits for i830_sdvo_get_trained_inputs(). commit c357eca10ca1c535d305e1f3028471a912ae4102 Author: Eric Anholt Date: Mon Oct 30 09:44:55 2006 -0800 Clean up whitespace in i830_randr.c. commit fbb376bd1a4daad4c86e349df98438989ce173f1 Author: Wang Zhenyu Date: Mon Oct 30 14:15:12 2006 +0800 Bug 8594: Fix Xv hang with G965 commit c0ee50c4ee5ff2c594fdf60c9cb8b952e25644de Author: Thomas Hellstrom Date: Fri Oct 27 19:13:33 2006 +0200 Add support for the new DRM memory manager. Some code are duplicated with the new libdrm. Once this code has been released with xserver, it can be removed. See the man page for new options and backwards 3D driver compatibility. commit 25e6e497824a23eb231fc6fd6e483f601d612cee Author: Eric Anholt Date: Thu Oct 26 15:47:49 2006 -0700 Major cleanup of 3D invariant state, fixing hangs with rotation and render. Now, the generic invariant state is always set while the X Server is active, and happens automatically when the X Server grabs the DRI lock. More 3D state is moved to the generic code. Then, the 3D consumers (video, rotation, render) set last_3d to their enum entry, and can update their own invariant state when another consumer was active. commit 0bdcce2e4541c6e441c44aad49254ad3093cedb2 Author: Eric Anholt Date: Thu Oct 26 14:30:46 2006 -0700 Note alignment requirement for i915 3D (texturing). commit c3666a968b1cfac61b7867874d5fbe2acd894720 Author: Eric Anholt Date: Thu Oct 26 14:30:25 2006 -0700 Don't write unused values beyond the end of scale_units array. commit 3ab9f5a4a8cd62c8a8c2a09d3d105adbe815a83d Author: Eric Anholt Date: Tue Oct 24 00:26:09 2006 -0700 Move vbeInfo out of the driver struct to the one place it's used. commit df14838eb5d0a056f663d9f12bd8b5c25cf97330 Author: Eric Anholt Date: Tue Oct 24 00:18:33 2006 -0700 Remove dead memsize reporting. The calculation no longer made sense, as we don't use the BIOS for mode selection. commit 3a6104ab89b159241845314ccf88fa62da14cf7d Author: Eric Anholt Date: Mon Oct 23 19:15:30 2006 -0700 Remove disabled I830DetectMonitorChange(). This used to be called when switching back in to X. It might make some sense to detect monitors at this time (it happens to occur at resume time, when monitors are likely to have changed), but it should probably live in either userland policy with RandR 1.2 or RandR 1.2 XFree86-DDX generic code. commit 88c12f577f80fa132ec45cdf456d8060f2ece4fd Author: Eric Anholt Date: Mon Oct 23 18:57:57 2006 -0700 Move LVDS initialization and blacklisting damage to the LVDS support file. commit 2631014e9d5b2e64908ea413729eb5fd819b17fc Author: Eric Anholt Date: Wed Oct 25 12:21:44 2006 -0700 Clean up the SDVO code. The main change is to send SDVO commands using data passed into the send command function, and receive responses into memory passed into the read response function, rather than stuff things in/out through dev_priv->sdvo_regs. This lets us use structures to represent some arguments, which results in a nice cleanup (and 100% fewer arguments named magicN as a side effect). Also, the mode set path is changed to not do any preferred input timing work. We weren't doing anything legitimate with the results, since we didn't modify the CRTC timing appropriately, so now we just stuff the CRTC timing into both and hope for the best. This should probably be revisited later. commit ddb986e54f5320359abac06f512f2d3f446872db Author: Eric Anholt Date: Tue Oct 24 17:47:58 2006 -0700 Nuke trailing whitespace in SDVO code not already cleaned up. commit cd2f0d948a6cae61b0d77d2c7184ed8d695be8c3 Author: Eric Anholt Date: Tue Oct 24 17:47:47 2006 -0700 Clean up SDVO code, moving the private structure internal and cleaning up caps. All the SDVO code should now be in lower case rather than StudlyCaps. This also adjusts the I2C setup to create a bus per SDVO output we set up. The previous setup with shared buses was failing in some circumstances, which is probably due to the lack of refcounting in xf86i2c.c. commit cd9c6e29146e1debaba4b0b9ad0d241f07bdbc14 Merge: c5cca4c eec5580 Author: Keith Packard Date: Tue Oct 24 16:53:46 2006 -0700 Merge branch 'modesetting-origin' into modesetting commit eec5580cefffc293bf547372ab63b2fedaef4a83 Merge: 53c28b3 a91c0cb Author: Eric Anholt Date: Mon Oct 23 14:52:55 2006 -0700 Merge branch 'overhaul' into modesetting commit a91c0cbab5de51885bfce7c7dce76f82c1b19553 Author: Eric Anholt Date: Mon Oct 23 14:30:38 2006 -0700 Add work-in-progress integrated TV-out support. This is the TV connector on board for the 915GM and 945GM. It is currently not hooked up to output initialization as it's entirely untested. However, I think this is a reasonable starting point for getting TV-out actually working. commit 7d67324fa3a5b5cf0227550316c366752fe4abfb Author: Alan Hourihane Date: Mon Oct 23 17:29:17 2006 +0100 Add some MergedFB checks for accelerator limitations commit 20d263805a365be548279758bc39d027d6e56d63 Merge: 8548fa5 5ac927d Author: Alan Hourihane Date: Mon Oct 23 14:24:41 2006 +0100 Merge branch 'master' of git+ssh://git.freedesktop.org/git/xorg/driver/xf86-video-intel commit 8548fa53746c9886eccfb67d022136471d8f3316 Author: Alan Hourihane Date: Mon Oct 23 14:23:29 2006 +0100 Fix manpage to report 8160 instead of 6144 for LinearAlloc. commit 5ac927d6f05ba6f9608d7e758cbac66e8b3a9db1 Author: Michel Dänzer Date: Tue Oct 17 13:27:40 2006 +0200 Don't crash if SetCursorPosition is called with pScrn->currentMode == NULL. This allows the driver to work with servers that don't have the fix for this that was only recently pushed to git. commit 8149681f2eac0af3b70a9457c5204e17da56142b Author: Eric Anholt Date: Mon Oct 16 16:51:04 2006 -0700 Add a per-output mode-valid method. This is currently disconnected, but will be used in more overhaul work. This should be where any output limitations, such as clocks, resolution, scaling limits, or other options, are validated. Other limitations, such as chipset resolution limits, CRTC clock limits, etc. should be elsewhere. commit c5cca4c20ae6b519e3b021a9d90809c1b3d1facb Author: Keith Packard Date: Fri Oct 13 15:31:43 2006 -0700 Advertise textured video adapter first commit adae75ef2f8a655a699e982ca8f26b934dbe950a Author: Alan Hourihane Date: Fri Oct 13 17:10:27 2006 +0100 Make VIDEO_DEBUG into a configure option commit 1d4cfaa391ac41099de11b8f3b21e5eb56543909 Author: Alan Hourihane Date: Fri Oct 13 13:55:10 2006 +0100 Bump to 1.7.2 commit cd9e51cd3ab0b36d2f97dfc6d8850dafe169c221 Author: Alan Hourihane Date: Fri Oct 13 13:54:31 2006 +0100 Ugh. Turn off debugging. commit 334aa062c333f2c59bc40e1b9239d3ea71c88a8f Author: Alan Hourihane Date: Fri Oct 13 12:59:30 2006 +0100 Bump to 1.7.1 commit 8045a7a0af1b276cd7f46ce99d034fd5d503e67c Author: Alan Hourihane Date: Fri Oct 13 12:53:53 2006 +0100 Fix bug #8553 & #8542 - suspend/resume regression from 1.5 driver commit 6596333b077a692c669f8cb4ae32e145f8da6731 Author: Alan Hourihane Date: Fri Oct 13 12:48:16 2006 +0100 Fix some warnings commit 1838671476875e9f5b3dde235eacf9fb43afb66c Author: Eric Anholt Date: Tue Oct 10 12:41:01 2006 -0700 Only disable the output when marked disabled. Also, remove a couple of dead variables. commit 4198f1216eb13b30d1e92d4395e98861f4324c38 Author: Wang Zhenyu Date: Tue Oct 10 15:50:10 2006 +0800 Mark current ps kernel is experimential with little test. commit d485c1f2de11db92b515690fbe38b6aeb6207605 Author: Wang Zhenyu Date: Tue Oct 10 14:11:35 2006 +0800 Use sf_prog.h instead commit 2591c1fcf15608d96031be4760cf08534461ca34 Author: Eric Anholt Date: Mon Oct 9 14:20:49 2006 -0700 Fix LVDS restore path, and move pipe assignment for outputs to the right place. commit 09e3d10b0ff69d180467fa9099d12da08e4f681b Author: Eric Anholt Date: Mon Oct 9 13:09:18 2006 -0700 Add a function for describing the output connection configuration. commit 317cc119c575650c1aa8bf992a0f42bdfffcd7ba Author: Eric Anholt Date: Mon Oct 9 11:49:18 2006 -0700 Move per-output mode setting code to per-output methods. This is not a very clean interface, as a number of outputs require tweaks to the DPLL registers. When possible, the DPLLs are just adjusted in the per-output post_set_mode, which happens just after the DPLL is enabled. However, this seems better than the previous method of having all outputs programmed in the same function. commit 53c28b3980d2682de830e8f86553ccad71527ac1 Author: Adam Jackson Date: Sat Oct 7 11:18:26 2006 -0400 Expand the check for AOpen Mini-PC. Just match on subsystem vendor, don't bother inspecting the subsystem device ID, since apparently they're all busted. commit bf3820f1f505649ac0730add23d97d6de3f6d22c Author: Keith Packard Date: Fri Oct 6 21:57:26 2006 -0700 Compute LVDS resolution from server DPI and native panel size. I was unable to find the native LVDS panel physical size in the BDB information. I would prefer to report accurate information through RandR if possible though. commit c7a23908a0ef45b14b438f656a5f15582bfcbc1c Author: Alan Hourihane Date: Fri Oct 6 13:35:39 2006 +0100 Fix server regeneration problem mapping the aperture commit d649fb0d964a9b40b8e04314b5fc14bbbfd41bc3 Author: Keith Packard Date: Thu Oct 5 22:46:07 2006 -0700 Don't require MonitorLayout when two monitors are plugged in. With randr12 working, we can just leave the second monitor off for now. commit 9bb7736ab36f172db58703c4664bb1b0cd7f80c3 Author: Eric Anholt Date: Thu Oct 5 16:02:02 2006 -0700 Remove checks for output privates in various bits of code. Now, the output is only set up if it fully initializes, so it will never exist if the private it requires doesn't. commit ada8f62da263d1e93e22df4e0b1149bf1dbe24d4 Author: Eric Anholt Date: Thu Oct 5 15:55:07 2006 -0700 Give each output type an init method in its file, making other methods static. commit 0a5504e59f90abecedd81a8435b88a60fe098f29 Author: Keith Packard Date: Thu Oct 5 09:11:29 2006 -0700 Fix DPI at startup in RandR 1.2 code. Use requested monitor resolution to compute the appropriate screen size when resizing the screen during RandR initialization. commit 4ac81d58b7e0fbffbb4981deffe6a576be821a9b Author: Keith Packard Date: Wed Oct 4 23:22:52 2006 -0700 Remove mode origins, add preferred mode count. Just tracking changes in the 1.2 protocol spec. commit 103b4edce7859ddf58f3e1fadeb427a5e85c7acd Author: Eric Anholt Date: Wed Oct 4 15:00:19 2006 -0700 Move the save, restore, and DPMS per-output settings to per-output files. commit 16988b27258acfe3d56fc2415aa9ade1ae8d03a9 Merge: cc4148e 5a06000 Author: Keith Packard Date: Wed Oct 4 18:44:43 2006 -0700 Merge branch 'modesetting-guitar' into modesetting commit 5a060002487e16c53dc96e32af72cd1bfcf6a227 Author: Keith Packard Date: Wed Oct 4 18:43:07 2006 -0700 Add mode origins for randr commit 07cdc60be8ebe5d0c21149f3cd2aeb8fed05022a Author: Alan Hourihane Date: Wed Oct 4 11:05:30 2006 +0100 Fix bug #5795, VT switching fails because of a bad test on some platforms. commit 8da10d15f84ccc992b828546af5ba0d848e7263e Author: Keith Packard Date: Tue Oct 3 22:00:14 2006 -0700 Enable XV_PORT attribute even when not in Clone mode. As Clone mode is now something that can change after server startup, always enable the XV_PORT attribute as we cannot change the list of reported attributes. (cherry picked from 1bc1cedbcdf6ea4d2a3e8b07b553ac50b3c126f4 commit) commit cc4148e25dc226cb1365e090db21f01b77cb3cbe Merge: 1bc1ced 3e6f81f Author: Keith Packard Date: Tue Oct 3 22:00:41 2006 -0700 Merge branch 'modesetting-origin' into modesetting commit 1bc1cedbcdf6ea4d2a3e8b07b553ac50b3c126f4 Author: Keith Packard Date: Tue Oct 3 22:00:14 2006 -0700 Enable XV_PORT attribute even when not in Clone mode. As Clone mode is now something that can change after server startup, always enable the XV_PORT attribute as we cannot change the list of reported attributes. commit 33629ed304b64e45d5640397bd1807c5a98907d1 Author: Keith Packard Date: Tue Oct 3 09:39:22 2006 -0700 Always register list of modes for outputs, even when disabled. commit 3e6f81f70f65a5ba6b5c3a4e0eeaf67776a5f54d Author: Eric Anholt Date: Mon Oct 2 09:53:57 2006 -0700 Move cursor base address register setting to a separate function. commit f337eea9e73cbf5c71d07df50d2e5b661a1f34cf Author: Alan Hourihane Date: Mon Oct 2 11:16:42 2006 +0100 silence warning commit 2b9bdd8a4021231e4b5f4c5831a46c9ff1e38081 Author: Alan Hourihane Date: Mon Oct 2 11:12:23 2006 +0100 bump to 1.7.0 commit eca082ca0fec12973a4dbb0106b841136b5b6cdb Author: Alan Hourihane Date: Sat Sep 30 13:44:43 2006 +0100 Missing gamma bit commit cece9a455b54ab184a3277c164ee6dfa65b70f48 Author: Alan Hourihane Date: Sat Sep 30 13:27:27 2006 +0100 Fix issues with Xv locking up the engine. bug #7915. commit 2013b839de3733fd12becb2cc3c1daadde329eb4 Author: Michel Dänzer Date: Thu Sep 28 13:49:44 2006 +0200 Always call I830UpdateXineramaScreenInfo() unconditionally, and document why. commit 92c5020bbd31e02c7224798a8f094d237afa19a0 Author: Wang Zhenyu Date: Thu Sep 28 13:55:52 2006 +0800 Fix picture's transform checking commit 518802843284973f1f86132afd805c0e662e1ba6 Author: Wang Zhenyu Date: Thu Sep 28 11:15:33 2006 +0800 Fallback in mask picture for now Do it later after finish wm kernel program. commit 25ff5baad95c205c7c690da72fc0f252b3ed9289 Author: Wang Zhenyu Date: Thu Sep 28 11:09:52 2006 +0800 Fix compile, add wm header file. commit c3a49b5123d094280b5ff358ceef961958319ce4 Author: Wang Zhenyu Date: Thu Sep 28 10:36:00 2006 +0800 Add simplest wm kernel program for no mask picture composite This is a try to use new gen4asm language, and will finish composite program for mask picture with or without CA case later. commit 6ea16bf6b06c8b3aed4d2c98679ab28304d1b56c Merge: fdb6de6 b970166 Author: Eric Anholt Date: Wed Sep 27 16:38:01 2006 -0700 Merge branch 'master' into modesetting This reverts most of the mergedfb code. This will instead be done in device- independent RandR code. Conflicts: src/Makefile.am src/i810_driver.c src/i810_reg.h src/i830.h src/i830_cursor.c src/i830_driver.c src/i830_modes.c src/i830_video.c commit fdb6de663579d3b9f31bf9e8a93430b8505ca73f Author: Eric Anholt Date: Wed Sep 27 13:59:41 2006 -0700 Re-disable broken load-based CRT detection. commit 1407a42c7378706644fd8be554b43b0e7b581011 Merge: 25890ec 4bd3b89 Author: Eric Anholt Date: Wed Sep 27 13:33:23 2006 -0700 Merge branch 'randr-1.2' into modesetting commit 25890ecda9fd00fad9bc53dea83fc58e0013fcdf Merge: 965609f 117ff04 Author: Eric Anholt Date: Wed Sep 27 13:33:17 2006 -0700 Merge branch 'acpi-hotkey' into modesetting Conflicts: src/i830.h src/i830_driver.c commit 117ff04b504578a24dff70659e2db1b81aaa1177 Author: Eric Anholt Date: Thu Sep 21 17:03:34 2006 -0700 Attempt to make the ACPI hotkey support a little more modesetting-compatible. Previously, we watched for the BIOS to have changed the layout, and repaired the resulting configuration. Now, we request that the BIOS make no changes, but leave a note in a register for when the key has been pressed. When we notice this, we reprobe monitors and turn on/off the things we find. This is a temporary solution until we can get the hotkey hooked up as an input key to external applications to control the change using RandR 1.2. It is also untested as neither of my laptops do anything with the hotkey. However, this code does result in many fewer BIOS calls. commit fe691953f08b4d299295f91450412404ba42810b Author: Michel Dänzer Date: Thu Aug 31 18:35:10 2006 +0200 Store viewport limits of both pipes in SAREA. This allows DRI clients to determine which pipe they should synchronize buffer swaps of each window to. commit 06c5f7bd076e9ba4b925e061a40b837714bc8267 Author: Michel Dänzer Date: Wed Aug 30 19:45:33 2006 +0200 If the DRM can handle it, enable vertical blank interrupts for both pipes. commit 89cbc6e215a5f313ccc17370424c35630cf75892 Author: Michel Dänzer Date: Thu Aug 24 10:30:27 2006 +0200 Only enable the IRQ after DRIFinishScreenInit. This makes sure we've been assigned a context ID, so the interrupt context won't mess things up if it grabs the HW lock. commit 185df8a18102eea79b8c8e11304dfee148ac2002 Author: Wang Zhenyu Date: Wed Sep 27 16:48:43 2006 +0800 Add mask sampler state commit 1422d4f04ac2d57899eee91f6522f8670f7263da Author: Wang Zhenyu Date: Wed Sep 27 13:54:14 2006 +0800 change some src sampler states sampler for mask should also be set up, and fix default border texel. commit 6614b4bda2e079fd1e56e0825894481b80e2df58 Merge: d6ca9fe b3ddcf3 Author: Alan Hourihane Date: Mon Sep 25 12:04:01 2006 +0100 Merge branch 'master' of git+ssh://git.freedesktop.org/git/xorg/driver/xf86-video-intel commit d6ca9feb78a9211758510751bbf47013caabff49 Author: Alan Hourihane Date: Mon Sep 25 12:03:27 2006 +0100 Fix bug #7943 where video flickers when changing attributes. This code was a leftover from testing lockups and shouldn't be needed anymore. commit 52a4f2a02b8f38108bfa866bf74b1596b8125512 Author: Wang Zhenyu Date: Mon Sep 25 14:35:51 2006 +0800 Add file for i965 exa composite This does not include ps program, which will be added in g4a form. commit f272f0d811f9ee059e8f8617a516e6c8bff917a1 Merge: ff202f8 b3ddcf3 Author: Wang Zhenyu Date: Mon Sep 25 14:15:21 2006 +0800 Merge branch 'master' into exa commit 4bd3b89c73b6c5aa9b0eb553ad5d553ee0e8a489 Author: Keith Packard Date: Sat Sep 23 16:41:38 2006 +0100 Oops, duplicated CRT-redetect code. commit c34490bbda6604a21809d15c798607806fa6c725 Author: Keith Packard Date: Sat Sep 23 12:00:43 2006 +0100 Construct default monitor description for hotplug non-DDC monitor. When detecting a monitor that doesn't support DDC, construct a default monitor with "sensible" values instead of using whatever the builtin LCD screen uses. Clearly we need a way to set the monitor parameters when we cannot detect them. commit 965609f6fa63e28e5a28128f5bc44f8c4d7b9f68 Author: Eric Anholt Date: Fri Sep 22 09:51:45 2006 -0700 Restructure i830_bios.c so we don't leak a copy of the BIOS per generation. commit c52242c22779a51aa12b18a7a589080ce44c8484 Author: Eric Anholt Date: Fri Sep 22 09:41:07 2006 -0700 Remove some dead code related to clock ranges. commit c2446be9b444b16c95f78dab17bf130f9f491ee2 Author: Eric Anholt Date: Fri Sep 22 09:38:11 2006 -0700 Remove the GetDevicePresence BIOS call which just printed BIOS information. Because we aren't using the BIOS to set modes any more, what the BIOS thinks is present is probably even less important than before. commit b6ba268d0d5f22c6a18ce45416452fce83438620 Author: Eric Anholt Date: Fri Sep 22 09:31:37 2006 -0700 Remove the no-longer-connected VBERestore option. commit 20956a5d6f1eb518717a680e58938f31461ca5e4 Author: Eric Anholt Date: Fri Sep 22 09:27:30 2006 -0700 Remove unused display{Attached,Present} fields. commit 2cd28be71472d67956f47c7d49283ebabefa089a Author: Eric Anholt Date: Fri Sep 22 08:55:55 2006 -0700 Remove empty SAVERESTORE_HWSTATE code. commit ff202f8e2ae4117b464c94047001023d5d0531cc Author: Wang Zhenyu Date: Fri Sep 22 15:02:19 2006 +0800 Fix a typo to mark sync for XAA. This should fix the scrolling screen corrupt in XAA method. commit d87d1f5bb0475c6f651fcb7e2cab2a7d46edcc69 Author: Keith Packard Date: Fri Sep 22 02:20:35 2006 +0100 Remove BIOS from non-BIOS related names. Lots of names included BIOS for no apparent reason; as we try to eliminate BIOS calls from the driver, these only serve to confuse us. (cherry picked from 8e5d280d94ad3d3ba3c75871c17abec9da62ed34 commit) commit b3ddcf3348365873aed4a2b8b2750b330fb4cf26 Author: A Costa Date: Thu Sep 21 23:18:43 2006 +1000 Typo fix in man page (Debian bug #364559). commit 4820caf46e050761d9b347b8a440381e1b1f4727 Author: Keith Packard Date: Thu Sep 21 01:47:27 2006 -0700 Make planeEnabled track pipes controlled by randr. Also add code to deal with disabling pipes. commit c11c445bdeac34253b48192a5d406b55ff8b2be7 Author: Keith Packard Date: Thu Sep 21 01:23:10 2006 -0700 Consistently use Cursor A on Pipe 0 and Cursor B on Pipe 1. Mixing random cursors and pipes didn't work very well. I'm left wondering whether the palette stuff will work on pre-9xx series hardware though; it is special cased everwhere else. commit 4a8b9515e914369e1faf1dca6b11204b233c73ee Merge: 18a510b bdca869 Author: Keith Packard Date: Wed Sep 20 22:46:55 2006 -0700 Merge branch 'modesetting' into randr-1.2 commit bdca8697cc6ac5a98e0548b2ce2e0032e9bcfa52 Merge: d9db5ec daade50 Author: Keith Packard Date: Wed Sep 20 22:46:42 2006 -0700 Merge branch 'modesetting-origin' into modesetting commit 18a510bf9af2f288c0e94359f672775d6e9c9ab2 Merge: f6500e9 d9db5ec Author: Keith Packard Date: Wed Sep 20 22:45:48 2006 -0700 Merge branch 'modesetting' into randr-1.2 commit f6500e94fec0d6db8c1f1350bee1d137bf06a09e Author: Keith Packard Date: Wed Sep 20 22:38:55 2006 -0700 Update driver for RandR 1.2 X server API. This is not entirely what I'd like to see, but it's at least functional. Limitations: Can't disable/enable crtcs Can't move outputs on/off crtcs But, it does handle monitor hot-plug, detecting changes in VGA and SDVO status on-the fly. Which makes for good demo material. commit daade50ca271d1cdf236bbe84afade85d4111ac9 Author: Linus Torvals Date: Wed Sep 20 12:07:09 2006 -0700 Add standard C headers to fix build on some systems after xf86_ansic.h removal. This appears to have been hidden for others by header pollution in X Server headers. commit 1681fe6bb041385e5aefb11baa9afd1f614abc5b Author: Wang Zhenyu Date: Wed Sep 20 09:36:02 2006 +0800 Revert "fix default tex wrap mode with clamp" oops, we've set border color to all zero, which should give us transparent in RepeatNone case. commit d9db5ec6866555ec13ba3ddabb1516eb45637afa Merge: cbaf3cf e3f4caf Author: Keith Packard Date: Tue Sep 19 16:03:04 2006 -0700 Merge branch 'modesetting-origin' into modesetting commit 5adb05f0b82ee3a168142d803ab3434f75cf15a3 Author: Wang Zhenyu Date: Tue Sep 19 17:14:05 2006 +0800 fix default tex wrap mode with clamp we don't set any 'default' border color, default tex wrap mode should be clamp to edge. commit e3ab89b0327ef6ff790ab53bba29c721aef032cd Author: Eric Anholt Date: Tue Sep 19 10:27:36 2006 +0800 shader program fix for component alpha set If CA is set and blend op needs src alpha, the src value is not needed and should be (src.A * mask.X). This is found in handling exa magic two pass composite. commit 37429c3ecb97df8faf0f3f56128d7a2b22eb8a5e Author: Wang Zhenyu Date: Tue Sep 19 10:08:31 2006 +0800 misc cleanup commit e3f4caf40708478ef327b029d0a75944c51ea905 Author: Adam Jackson Date: Fri Sep 15 13:51:18 2006 -0400 Add model-specific tweaks for some funky 945GM boards. For the Aopen Mini-PC, ignore the claimed attached 800x600 LVDS panel. Likewise for the Apple Mac Mini, but done slightly differently since it shares PCI IDs with the Macbook Pro. commit bd758030a6a59af32f9d73655cb691e018e94a7d Author: Eric Anholt Date: Thu Sep 14 11:12:03 2006 -0700 Add a compile flag to enable syncing after each operation in EXA. This replaces other debug sync options sprinkled around the EXA code. It doesn't change the mis-rendering of text on the 915. commit f46c70e877a3432ba23696e1a16d5906183876af Author: Eric Anholt Date: Thu Sep 14 10:05:51 2006 -0700 Add a note on maxX/maxY for the 2D rendering. commit ce00db054bf0b7f62ff0be312d7e3470141922dc Author: Eric Anholt Date: Thu Sep 14 10:00:06 2006 -0700 Replace a couple of mis-uses of the screen's bpp when the pixmap's was needed. commit 71c2dd63361be9616c77db80a93445307d63dd41 Author: Eric Anholt Date: Thu Sep 14 09:54:19 2006 -0700 Replace duplicated sync code in exa with a call to I830Sync. The I830Sync version has additional code for the 965. commit 564c830a91afda76706a8c4c2f3287d0b8d514e9 Merge: bf49084 b970166 Author: Alan Hourihane Date: Wed Sep 13 16:34:10 2006 +0100 Merge branch 'master' of git+ssh://git.freedesktop.org/git/xorg/driver/xf86-video-intel commit bf49084ad700f064b732bf39a163209eee1d6845 Author: root Date: Wed Sep 13 16:32:39 2006 +0100 bump max Xvideo height to 1088 to cope with some players. commit 59ce4180b95270432540624a1fdf7ae8dd18a7f3 Author: Wang Zhenyu Date: Mon Sep 11 22:57:23 2006 +0800 reimplement Solid function Replace COLOR_BLT_CMD with XY_COLOR_BLT_CMD, by which device would care for direction and provide a (x,y) coord interface. This fixes pixmap artifacts even in no composite case. commit 2b9c87bbf8ee5f7f56631114eb98303cd80e4a48 Author: Eric Anholt Date: Fri Sep 8 11:14:05 2006 -0700 Remove xf86_ansic.h usage. commit b970166eab95ac024ff481b0f5fd9aaf3644aabf Author: Eric Anholt Date: Fri Sep 8 11:09:19 2006 -0700 Fix a compile failure from the last commit. commit 2b7602847aa7ae9b4e88fbea070710695d2cb8a4 Merge: e4cdc42 b45f4b5 Author: Eric Anholt Date: Fri Sep 8 11:02:41 2006 -0700 Merge branch 'exa-origin' into exa Conflicts: src/i915_exa_render.c commit e4cdc42a98dd1be74e5c2bf5d9566f7449523b14 Author: Eric Anholt Date: Thu Sep 7 10:32:02 2006 -0700 Remove another debug printf. commit 099be52a3bb52516fb5e8d56c154a4d6b6707e09 Author: Eric Anholt Date: Thu Sep 7 10:28:09 2006 -0700 Re-remove xf86_ansic.h, and move copyright to the top of the file. commit 4ad577f83f7190df4d64e66fbf7c515ee6f411ce Author: Eric Anholt Date: Thu Sep 7 10:24:23 2006 -0700 Remove the UTS/DFS implementations that just match the fallback versions. commit 8b9d7a5f8560ee55ca5cb5376517a4246cd79fb9 Merge: 6500a3a 1e6e288 Author: Alan Hourihane Date: Thu Sep 7 17:45:57 2006 +0100 Merge branch 'master' of git+ssh://git.freedesktop.org/git/xorg/driver/xf86-video-intel commit 6500a3aa134e11a75b7ddaf7fff253003b223241 Author: Alan Hourihane Date: Thu Sep 7 17:45:28 2006 +0100 When in MergedFB mode, fix offscreen memory allocation. commit c378328ece701756ddca966eb108e2b1f5c3ed6c Author: Eric Anholt Date: Wed Sep 6 11:15:17 2006 -0700 Remove a debug printf. commit 1e6e288b8826789f3b2520d12426ff7852a67ccd Author: Eric Anholt Date: Fri Sep 1 16:47:17 2006 -0700 Add a check for intel-gen4asm, and rules to compile {wm,sf}_prog.h using it. This adds assembly source for the two programs used in the textured video implementation on the 965, which should make them easier to modify in the future. The compiled versions are also included, so that intel-gen4asm isn't a build requirement for people that aren't modifying these programs. There are minor differences in the compiled versions of these programs compared to their previous versions which were compiled with a different tool. I believe the changes should be harmless, and video continues to work on my system. commit cbaf3cf74bd420533d299c4113761ec536097e33 Author: Keith Packard Date: Thu Aug 31 18:25:21 2006 -0700 verbose debug message for panel sync data commit 1feb733eb8b09a8b07b7a6987add5149c53b0157 Author: Alan Hourihane Date: Tue Aug 29 14:54:15 2006 +0100 Fix bug with probing info of DFP2 and LFP2 connected devices (Aaron Ridout) commit 26e6c074c17b2d4e9b9f165468ad45a3b0ecc0ef Author: Alan Hourihane Date: Mon Aug 28 11:55:38 2006 +0100 Fix SetVBlank to operate all the time commit b45f4b54b189a496255ddae12af38fc4c4408a51 Author: Wang Zhenyu Date: Fri Aug 25 16:23:18 2006 +0800 Making render log a little quiet in normal commit 6f0d352b83fc9f39dd86edbda9af83243b50c764 Author: Ross Burton Date: Thu Aug 24 18:16:42 2006 -0700 Bug #7957: Fix distcheck. commit cc79b3ec19a7e858972228a270ef774041eb8946 Author: Wang Zhenyu Date: Thu Aug 24 10:28:59 2006 +0800 Current 915 render operations for 915G/945G only commit 078af29fa65653c5322b006404f7ad29aec24cce Merge: 7458a6a ea63e82 Author: Wang Zhenyu Date: Thu Aug 24 10:23:22 2006 +0800 Merge branch 'master' into exa Conflicts: man/i810.man src/Makefile.am src/i830_accel.c src/i830_dga.c src/i830_driver.c commit ea63e82ef417a9918e7d7105910a8ddeba2994f6 Author: Alan Hourihane Date: Tue Aug 22 16:40:08 2006 +0100 fix typo commit b7e57deebbda527e878326cf3e6358c0a48d7817 Author: Wang Zhenyu Date: Tue Aug 22 19:38:13 2006 +0800 pci id update for Crestline The "Crestline" banner should be replaced with official name later. commit dc4128ea0781124984dee0cb41e7b6ee87e362a1 Author: Alan Hourihane Date: Mon Aug 21 12:13:42 2006 +0100 remove GL* usage commit dad0e7fc09fa7794f20278d9d99abd96a21b691e Merge: 361c49a 43daaec Author: Alan Hourihane Date: Mon Aug 21 12:08:01 2006 +0100 Merge branch 'master' of git+ssh://git.freedesktop.org/git/xorg/driver/xf86-video-intel commit 361c49ad3cfe58b9ab98dfb6e95de505c0d63730 Author: Alan Hourihane Date: Mon Aug 21 12:07:29 2006 +0100 Fix bug #7930, i810 doesn't compile without DRI (Samuel Thibault) commit 43daaec63929c1f0e54a5125375d8147629da4b9 Author: Eric Anholt Date: Thu Aug 17 15:57:31 2006 -0700 Bug #7905: Mark DRI state as dirty in 965 textured video, fixing hangs with 3d. commit cdcc6f23e8cad797e6cfd57e2ff454ec8509b5ba Author: Alan Hourihane Date: Wed Aug 16 17:16:28 2006 +0100 Disable device check when mergedfb in operation. commit a94c96abf90c691faecbebf0c8d08f8010bc67ef Author: Alan Hourihane Date: Wed Aug 16 17:15:26 2006 +0100 Fix bug #5150. Disable LVDS usage on i915G, i945G and i965G. commit 2a4e486d2990af433e915cfa26dc57cae02e01b7 Author: Wang Zhenyu Date: Fri Aug 11 11:07:08 2006 +0800 Fix undefined alloc type for agp memory This shuts up the nonfatal warning that type 3 for alloc agp memory is undefined for intel-agp. commit 4cfed93df6b23903704348cafe050a6546e84479 Author: Eric Anholt Date: Thu Aug 10 15:41:32 2006 -0700 Bump to 1.6.5 for release. commit 38e7e48418cd48a46e48f5bc8a6547721db8f76d Author: Eric Anholt Date: Thu Aug 10 15:38:14 2006 -0700 Fix README typos from in the generated file. commit 975e60261088dee124b329d28c64e508bce1f90b Author: Eric Anholt Date: Thu Aug 10 13:49:43 2006 -0700 Bug #7829: Fix reported driver version. The driver now reports itself as the PACKAGE_VERSION from autoconf. The DRI DDX -> client interface version is dissociated from this so that we can do appropriate major/minor versioning of the interface that's not tied to the package version. Bumped the i830 ddx dri version patchlevel to note the fix in the previous commit. commit 820e9a22fdb759cbdaac4a488322825bc0908b0d Author: Eric Anholt Date: Thu Aug 10 13:37:18 2006 -0700 Bug #7835: Restore unused fields in I830DRIRec to avoid i915 DRI breakage. As it was, 1.6.4 broke compatibility with the released DRI driver. commit caa3b35cd772fc75d65a7ff791f00addbb39a764 Author: Eric Anholt Date: Wed Aug 9 16:11:30 2006 -0700 Correct typos in README source. commit 2a7426cf138e518a5eafb40f478359160a7ec98b Author: Eric Anholt Date: Wed Aug 9 15:47:09 2006 -0700 Clean up warnings. commit 24e59a0daa20b7c3e5028c9ca7972052801d02a1 Merge: bb60807 32f1199 Author: Eric Anholt Date: Wed Aug 9 14:19:06 2006 -0700 Merge branch 'textured-video', bringing in fixed-up i915 textured video. Conflicts: src/i830_video.c commit bb6080735efc40e103e92b65d0c2f1f729156632 Author: Eric Anholt Date: Wed Aug 9 12:21:16 2006 -0700 Bump to 1.6.4 for release. commit 3ca14275d13b1261b69b0e3fda90a112cb567472 Author: Keith Packard Date: Wed Aug 9 10:35:57 2006 -0700 Add missing headers to i810_drv_la_SOURCES commit 309374f78df35207b1398e14bba986fb891f3643 Author: Keith Packard Date: Wed Aug 9 10:30:41 2006 -0700 Use double quotes to avoid sgml syntax error commit dca9f856ca21e63abeb87e4ef2c40944c26d4429 Author: Keith Packard Date: Wed Aug 9 10:29:59 2006 -0700 Reformat README commit b19ea222727ed47b69d28a03242c09d3a6ab2673 Author: Eric Anholt Date: Wed Aug 9 10:01:29 2006 -0700 Update the README.sgml file for the upcoming release. commit f8b47f607297e0591b63e5f25296af07ee74f433 Author: Keith Packard Date: Tue Aug 8 16:19:51 2006 -0700 Update manual page to include i945 and later details. (cherry picked from 7833d0733bc146cf1ae7f588516c49797886b396 commit) commit d150b53d102c511f3c9245ef1f6fd36c12b01ca3 Merge: bb65a4d bc5f565 Author: Eric Anholt Date: Tue Aug 8 15:51:58 2006 -0700 Merge branch 'i965', adding i965G support. Conflicts: src/i830_cursor.c src/i830_driver.c commit bc5f56568021d8c63313e2b6bf30710e7516c04a Author: Eric Anholt Date: Tue Aug 8 15:48:04 2006 -0700 Intel bug #49: Fix video output at 32bpp by using B8G8R8A8 instead of B8G8R8X8. While here, don't overallocate video memory for the i965G state. commit c3b3d479788fcea7e543f29acf83c85b8b148fbe Author: Alan Hourihane Date: Tue Aug 8 15:28:14 2006 -0700 Intel bug #35: Fix accelerator syncing with DGA. Fixes glitches seen with Mark Vojkovich's "texture" demo. commit bc12208f6e145ec29c3ebe38ae04dc2ebca1b4cc Author: Wang Zhenyu Date: Mon Aug 7 13:27:00 2006 +0800 Disable error register dumping in dri TransitionTo2d. This's for debug which might confuse QA. commit bb65a4dc55b416437d61ccbee8b6056fe8e08357 Author: Alan Hourihane Date: Tue Aug 8 14:51:18 2006 +0100 bump to 1.6.3 commit 37644293da751bfc7268c9fc74ecda1044607dee Author: Alan Hourihane Date: Tue Aug 8 14:42:32 2006 +0100 check for xineramaproto commit 3661d4df86db012682ab4b393287aa02e9fc8453 Author: Alan Hourihane Date: Tue Aug 8 12:25:03 2006 +0100 Re-enable monitor detect after a few fixups. commit 995b142f510d1daab3914d336f3c0d017a043b41 Author: Alan Hourihane Date: Tue Aug 8 12:01:39 2006 +0100 Disable detecting of new monitors on VT switch. It's doesn't always work. The modesetting branch will deal with hotplug displays correctly. commit 8e5844bbf9c8880c9eb6eff29b6db2db06b01933 Author: Alan Hourihane Date: Tue Aug 8 11:43:10 2006 +0100 bump to 1.6.2 to indicate mergedfb support commit 421b415e23c1ddc78837cd222167d6ed71a3ef88 Author: Alan Hourihane Date: Tue Aug 8 11:28:42 2006 +0100 Fix a build problem. commit e4f63eaf5efb654121db148f2f8e32cc3b23b6b5 Merge: 633a683 e26f3e3 Author: Alan Hourihane Date: Tue Aug 8 11:07:56 2006 +0100 Merge branch 'master' of git+ssh://xorg.freedesktop.org/git/xorg/driver/xf86-video-intel commit 633a683a4adcb9a44a54519fd7ff66aab2d12f97 Author: Alan Hourihane Date: Tue Aug 8 10:23:29 2006 +0100 Ensure palette is updated in mergedfb & clone modes commit 71e3e2d4e3b2a2c538fe6f1cea41f442fdb8d756 Author: Alan Hourihane Date: Fri Aug 4 20:39:50 2006 +0100 Fix a problem creating the I2C bus for the SDVOC interface, due to a name match with SDVOB. Bus names must be unique. commit 0fd4831fdcf4c8f43d80c66e43eff8942f89b324 Author: Eric Anholt Date: Fri Aug 4 00:21:53 2006 -0700 Disable dynamic front buffer mapping on i965. Moving front buffers should only be necessary for rotation. Currently, the server isn't ready for it, and the method attempted to work around it caused crashes with DRI. Since i965 doesn't support rotation yet, this should be harmless for now. commit 8d0a5138503586cbf980eb9464f2db91b72509c7 Author: Eric Anholt Date: Thu Aug 3 19:16:59 2006 -0700 Add parenthesis so that IS_I965G doesn't make the test pass for 8-bit. commit aa69018c01d2fa963fb940718dbd653d6ca2c9eb Author: Eric Anholt Date: Thu Aug 3 19:12:15 2006 -0700 Turn off video debugging now that it appears to work fine. commit d15f87110807111ab7d71f1254ebe3dfef1fee80 Merge: a79aa0d d56ffa5 Author: Eric Anholt Date: Thu Aug 3 19:08:34 2006 -0700 Merge branch 'broadwater-video-rehash' into i965 The previous merge wasn't done on a synced-up tree, and missed necessary changes. commit a79aa0d7b27b0b9b032472776d7dda410fd66a98 Merge: 760021e f9e94c1 Author: Eric Anholt Date: Thu Aug 3 17:58:56 2006 -0700 Merge branch 'broadwater-video-rehash' into broadwater Conflicts: src/Makefile.am src/common.h src/i810_driver.c src/i810_reg.h src/i830.h src/i830_accel.c src/i830_cursor.c src/i830_dri.c src/i830_dri.h src/i830_driver.c src/i830_memory.c src/i830_rotate.c src/i830_video.c commit 760021e3983f7783900075b8c9603bd4fbe7e0a2 Author: Alan Hourihane Date: Thu Aug 3 17:08:39 2006 -0700 Add current Tungsten Graphics code drop for i965 support. commit d56ffa5f35e3cf4262d66469052b2122fdb24027 Author: Eric Anholt Date: Thu Aug 3 16:03:50 2006 -0700 Bump PS_MAX_THREADS to 32 now that the program doesn't fail. commit 4525379d95ff292d7322e1a7a516c0bedd1f7543 Author: Eric Anholt Date: Thu Aug 3 16:03:15 2006 -0700 Make the sampler's payload be the WM payload rather than uninitialized data. The sampler's payload happens to be in the same format as the WM payload, though most of the fields are ignored. This appears to fix the program in the presence of multiple PS threads. commit ad2c70b4121121f1fb53190ea49edf2323c804a9 Author: Eric Anholt Date: Thu Aug 3 12:47:19 2006 -0700 Remove some stale XXX-prefixed comments. commit f9e94c17c55e4c75802d8574c908744e286e7843 Author: Eric Anholt Date: Wed Aug 2 21:18:19 2006 -0700 Set the WM scratch space that we had already allocated. It appears to be required, even if the kernel doesn't use any scratch space. commit aefa6fdfc5300546caeb64ace14a7854d3dc7dae Author: Eric Anholt Date: Wed Aug 2 21:14:14 2006 -0700 Clean up GRF allocation (which was wrong at 16-register boundaries). Also use PS_MAX_THREADS rather than hard-coding 1 thread, and remove the dead SF_KERNEL_NUM_URB macro. commit 7a64e14624514ef31f6fa9f15e8804c45f930212 Author: Eric Anholt Date: Wed Aug 2 20:48:13 2006 -0700 Crank down the SF allocation and comment on why this is a fine lower limit. commit bc6a2bb7576a7c1e7971f6d1e0b893b2ada1aaa3 Author: Eric Anholt Date: Wed Aug 2 20:34:57 2006 -0700 Remove the clip URB allocation. Previously, the VS was misconfigured and exceeding its allocation, which the (unused) clip was providing padding for. commit defe2795429484ffe4c1438bafb86bb5e5469ba9 Author: Eric Anholt Date: Wed Aug 2 20:32:41 2006 -0700 Correct the VS setup, and allocate a correct, minimal number of URB entries. The VS number of URB entries and URB entry size are always used, even when the VS is disabled. Similarly, the cache enable bit is always used. commit b57ccb682cb3dea3e26c6f1b0c709e63dfde0d31 Author: Eric Anholt Date: Wed Aug 2 19:46:15 2006 -0700 Replace the SF max threads setting with a define for easier tweaking. Tweak it to 1 for now. commit 82037a12758c41a304f2e0bbd033d3345cccbe1a Author: Eric Anholt Date: Wed Aug 2 19:33:28 2006 -0700 Remove CS URB allocation since we don't use any constants. commit 1d45668d7a42bfa5d7f5bfb68d8bae38bda0936b Author: Eric Anholt Date: Wed Aug 2 19:18:20 2006 -0700 We only need 3 vertices to fit in the URB, since we only dispatch 3. commit a076d35bed6f13cf943a0f8948176aa0c999e2da Author: Eric Anholt Date: Wed Aug 2 19:16:03 2006 -0700 No GS URB allocation is necessary when the function is disabled. commit befa655168fb8dcb6806592eb44f7ac49f191822 Author: Eric Anholt Date: Wed Aug 2 19:11:38 2006 -0700 Reduce URB_VS_ENTRY_SIZE to 1 as our vertices are under 8 floats. commit 33acbdca0a0f82725e5bf7887b325726403a6ffd Author: Eric Anholt Date: Wed Aug 2 19:09:19 2006 -0700 Remove the VS kernel and binding table. The VS URB entries have to remain as they're used to store the VF output which isn't modified by a VS program. commit aafa48cb85cd03c735fb968a4275c19e1a68cd02 Author: Eric Anholt Date: Wed Aug 2 18:26:26 2006 -0700 Fix wm prog to correct the ordering of the Cr and Cb channels. commit bc2c842d93de04d48c7de60482814db346bd0b78 Author: Eric Anholt Date: Wed Aug 2 18:10:01 2006 -0700 Allocate space for the 965's state at the end of the video buffer. Fixes corruption in the first few lines of the video. Based on 1b506798d98d911be733543da2c40cb451a28912 commit 524460ea1f02bb6e8e2239d7763334666012cec4 Author: Eric Anholt Date: Wed Aug 2 17:47:55 2006 -0700 Updated WM kernel to load video and do colorspace conversion. commit ba896c779c697e1d7458028798ec49013bd9da9f Author: Eric Anholt Date: Wed Aug 2 17:47:37 2006 -0700 Updated grf/urb state for WM. commit e5c572f841b626b8b6f21a6966a33956d3b0b35b Author: Adam Jackson Date: Wed Aug 2 20:47:12 2006 -0400 Fix a braino in mode list pruning. Interpreting the size of the display in centimeters as the size in pixels, and then clipping the modes list based on that, rarely does what you want. commit 21b62df7c34217be5dd95985c35e33be11c25846 Author: Eric Anholt Date: Wed Aug 2 17:36:49 2006 -0700 Move the WM kernel to a separate file. commit 5d3424492f9586a4c5a28962a9757f48f2c12e83 Author: Eric Anholt Date: Wed Aug 2 17:34:12 2006 -0700 Replace SF kernel with the one from broadwater-video HEAD. commit 7458a6adb5ea62f56bb3d4ab19ad7d1aa6ad2d19 Author: Eric Anholt Date: Mon Jul 31 10:09:17 2006 -0700 Move MAPSURF_XBIT settings into the texture formats structure. commit 7ea74843bbdbf54b6804727be9b41d26832c8f76 Author: Eric Anholt Date: Mon Jul 31 10:06:34 2006 -0700 Whitespace cleanup. commit e71108f1e05b7a8d8edd174eb64edd6cccacbcdc Author: Alan Hourihane Date: Fri Jul 28 10:32:12 2006 +0100 Fix DGA with MergedFB Turn off rotation support when MergedFB enabled commit a91a4f95c664f6905fef61dab251707bf2548bb8 Author: Alan Hourihane Date: Thu Jul 27 16:11:48 2006 +0100 Fix pipe reversal for Xv commit ac3ad32f667b306e771617d784648f7111743f1a Author: Alan Hourihane Date: Thu Jul 27 15:28:42 2006 +0100 Calculate allowable refresh rates on the private mode data for each independent screen in mergedfb. Lots of other fixes too. commit 30952e58ed83e2e18d1007f662d2cc9a773c876c Author: Eric Anholt Date: Wed Jul 26 13:42:12 2006 -0700 Limit drawing to the destination pixmap's boundaries, not the screen's. This shouldn't matter, as miComputeCompositeRegion shouldn't giving us anything that would draw outside the bounds, anyway. commit cffd2cd36d0437b38ac8164d66ea71be50b19330 Author: Eric Anholt Date: Wed Jul 26 13:22:51 2006 -0700 Fix hangs when compositing with a mask. The modify bits in the mask field for sampler/map state appear to actually be enable bits. So, prepare the state values in I915TextureSetup, then write the sampler/map state out all at once in PrepareComposite. commit 96754b822df7ac110a70b2d08dda2ebb299772be Author: Eric Anholt Date: Wed Jul 26 13:19:12 2006 -0700 Fix a couple of typos from code review. commit e786e2f9f3a4df31702736db6f68a44c9ebba546 Author: Alan Hourihane Date: Wed Jul 26 16:45:38 2006 +0100 When detecting new monitors in mergedfb renew the modepool. commit 8e6e990db34d63174670512f494fa9adb44786f5 Author: Alan Hourihane Date: Wed Jul 26 10:48:47 2006 +0100 Update Xvideo to deal with MergedFB modes. commit e26f3e30b30a57ab4aad0267d689a9a5d7a5e877 Author: Alan Hourihane Date: Wed Jul 26 09:17:52 2006 +0100 Fix a build problem commit b919db75d2f6dc1019f981534b0d5d87c6029727 Author: Alan Hourihane Date: Wed Jul 26 09:07:19 2006 +0100 Bump to 1.6.1 commit 9c93d1498fa2363c52ef7fbe97d781560f67acf5 Author: Eric Anholt Date: Tue Jul 25 21:17:05 2006 -0700 Oops, partial revert of last commit. Some were obviously not NOOPS. commit 38d1a5e0dbe059f5c01bd5120a108a386ff10718 Author: Eric Anholt Date: Tue Jul 25 21:09:12 2006 -0700 Replace some OUT_RING(0)s with OUT_RING(MI_NOOP) to remind me why they're there. commit fd19b12793f09b6714468556ace875ef36ed9e1c Author: Alan Hourihane Date: Tue Jul 25 11:14:11 2006 +0100 Add mergedfb support to the intel driver with additional pseudo-Xinerama support. commit aec88cf4c66833f89afa5f8fa4f8bcf30b2b6b4b Author: Wang Zhenyu Date: Mon Jul 24 15:46:19 2006 +0800 remove crap blit in UploadToScreen, we should find a more efficent way and implement blit correctly. commit 32f1199937e92b9100aba52cbbb97157014e3182 Author: Wang Zhenyu Date: Mon Jul 24 15:42:15 2006 +0800 remove an extra '-' commit 02c5c4cd1dd44058d80ec23cccfd45a13bc3bb2c Author: Wang Zhenyu Date: Fri Jul 21 12:54:06 2006 +0800 add copyright info commit 4f0ad337098eb0bacb1c11bc61d780ac18041e1d Author: Wang Zhenyu Date: Fri Jul 21 11:59:04 2006 +0800 fixup segfault in xaa when xaa not initialized in i830WaitSync. commit 914327f40d07a3f7c069752ce005bf8e21352b03 Author: Eric Anholt Date: Thu Jul 20 17:58:07 2006 -0400 Disable drawing to PICT_a8, and turn on support for non-extended repeats. We don't know what COLOR_BUF_8BIT does, so it's disabled for now while we're stabilizing the code. The non-extended repeats appear to be working according to the rendercheck tests on a8 and a8r8g8b8 sources. Masks continue to hang the card, as before this change. commit 45cb032e2b7ac1e4c765fc29a2e4dd0cef19066d Author: Eric Anholt Date: Thu Jul 20 17:34:50 2006 -0400 Use MT_8BIT_A8 so PICT_a8 expands to (0,0,0,a) not (a,a,a,a) commit fd6d825ae0780f3daf436f4224bf5f177cb8630e Author: Eric Anholt Date: Thu Jul 20 17:25:31 2006 -0400 Move ss6 setting with other immediate state, and ensure that stencil is off. commit d6d6a44c66446f34b57394e5ea9a8e32917f7569 Author: Eric Anholt Date: Thu Jul 20 11:23:05 2006 -0400 Try to fix up the basic RepeatNormal and RepeatNone support. With RepeatNone, access outside of the drawable should return 0 alpha values, which is done by using the border color. For RepeatNormal, we use the wrap mode (which requires that we use normalized texture coordinates). commit 2b2173c618c080b1678990d65fe49b52bd8001b4 Author: Eric Anholt Date: Thu Jul 20 11:15:18 2006 -0400 The pixmap's drawable x/y coordinates are always 0, so don't bother using them. commit 9f1cec83e0aef36c7d3482e62e8f01595f1fd076 Author: Eric Anholt Date: Thu Jul 20 11:05:49 2006 -0400 The vertex data are all floats, so write the dest coords as floats. commit 4737955a62c39177e7a7ce7749a2f20e111afc68 Author: Wang Zhenyu Date: Fri Jul 21 01:02:38 2006 +0800 move i830 default coord set into I830EmitInvarientState commit a7e30bb051bc03063fe699473610a57054a64973 Author: Wang Zhenyu Date: Fri Jul 21 00:56:58 2006 +0800 Take from i915, blend ctl code cleanup. commit ac34a37cb60d763cbd99b6e2f6fdcb639592b99b Author: Wang Zhenyu Date: Fri Jul 21 00:52:59 2006 +0800 Take instruction fixes into i830, with indent changes. commit 3e86bf4273241daa23645f5a8104913e4205a74b Author: Wang Zhenyu Date: Thu Jul 20 13:43:27 2006 +0800 remove i915 default ctx setup. commit 5c945bf9eb9f5bedd34d98dcc7abc3f1ea3cb962 Author: Wang Zhenyu Date: Thu Jul 20 13:40:39 2006 +0800 Revert "Fix several size error with STATE3D_LOAD_STATE_IMMEDIATE cmd." This reverts commit 433cb6ba82698676f6f72e09834aba4d64611d54. commit d5243d1b35ab1db608c81889819e4d4d87048154 Author: Eric Anholt Date: Wed Jul 19 19:27:53 2006 -0400 The height and width of texture maps are one more than the programmed values. commit 13e372c2095fde94ae41bea959fba0ad95a59c08 Author: Eric Anholt Date: Wed Jul 19 16:42:57 2006 -0400 Fix the texture map pitch setting as documented, and as used by the 3d driver. commit da5efb09d55291ce5bcabff7db8d1490bb5e838b Author: Eric Anholt Date: Wed Jul 19 16:36:52 2006 -0400 Fix the alignment padding fix (didn't bump BEGIN_LP_RING count). commit 36aa43bf73ee1268f0a250788eb637a4123f08eb Author: Eric Anholt Date: Wed Jul 19 16:22:46 2006 -0400 Force texture alpha channels to 1 if not present. commit adce6bd0fc9533156da09da7d029ba0d76606b4f Author: Eric Anholt Date: Wed Jul 19 16:21:49 2006 -0400 Add a trailing MI_NOOP to frag shader output for alignment if necessary. commit 4e409ea44180f7ef2780d70c9d2f54629bbd0fc3 Author: Eric Anholt Date: Wed Jul 19 15:57:09 2006 -0400 The map bits per pixel must come from the drawable, not the visible screen. commit 449e4be503c8b938d5bd46ee810244f8bcf54ceb Author: Eric Anholt Date: Wed Jul 19 13:34:06 2006 -0400 Correct the DWORD count of several 3D instructions. commit f1b62d890ca22e12d61f7ef67bc4e35d68cde019 Author: Eric Anholt Date: Wed Jul 19 13:02:24 2006 -0400 Replace hand register setting with new i915 fragmet program API. commit 76a316d0cec92a04c8735926a76e76c21f960d7e Merge: 433cb6b 8480516 Author: Eric Anholt Date: Wed Jul 19 12:31:36 2006 -0400 Merge branch 'master' into exa Conflicts: src/Makefile.am commit baf65ce98abcdd21dff2531a43bb9c5044732c28 Author: Eric Anholt Date: Tue Jul 18 19:42:37 2006 -0400 Re-convert i915 video to new fragment shader API. Although in the history of this branch it had happened before, this time it's for real. commit bb81e8d6c777a5e16b8193c07667fbee8e21203e Merge: 2a1b3cf 8480516 Author: Eric Anholt Date: Tue Jul 18 19:23:21 2006 -0400 Merge branch 'master' into textured-video This moves the i915 textured video implementation into i915_video.c to avoid conflicts in register definitions with i830_reg.h when we use i915_reg.h. This also means that i810_reg.h's i915 3D regs definitions are removed and replaced with i915_reg.h usage. Conflicts: src/i830_rotate.c commit 84805167ab8a422966355b9753bfcb4dad802413 Author: Eric Anholt Date: Tue Jul 18 18:27:10 2006 -0400 Convert i915 rotate code to the new fragment program API. commit 5176d62ba58c100c87f75a4f333d00129d780c99 Author: Eric Anholt Date: Tue Jul 18 16:18:18 2006 -0400 Add an API for programming i915 fragment programs. commit 148ef9bdd9e0ef3e7ac86b56a8662b53a3ea9168 Author: Eric Anholt Date: Mon Jul 17 22:32:25 2006 -0700 Convert magic numbers to symbolic names in i915 rotate code. This doesn't cover the fragment shader yet, which we need to make a sensible set of macros for (at least the basic bits). Reviewed by: md5 commit 433cb6ba82698676f6f72e09834aba4d64611d54 Author: Wang Zhenyu Date: Tue Jul 18 17:00:50 2006 +0800 Fix several size error with STATE3D_LOAD_STATE_IMMEDIATE cmd. Issue texture coord set with proper scale value. commit 855a9f13cb7fd4f6c489ff280bf389475bbd9bbe Author: Wang Zhenyu Date: Tue Jul 18 10:02:47 2006 +0800 fallback in 'repeat' case for now commit a6d438ebe3cf141a0331e0cd55eb9b5e137a5e37 Author: Keith Packard Date: Mon Jul 17 14:11:19 2006 -0400 Disable spread spectrum clock usage. The spread spectrum clock generator that improves EMI characteristics for laptop screens lives in an external chip that is programmed over an i2c bus. Without correct programming, attempts to use this mode for the LVDS can result in a DC signal being sent to the panel. Until we find programming information for this external chip, we should leave this mode disabled. commit 16d6263e6518a4a05562e2842ff2d0fdb4710304 Author: Alan Hourihane Date: Sun Jul 16 20:39:52 2006 +0100 whoops, reverse part of that. commit 2f50f6d1b1b3fa4fbec98bd8fa5818df890070e7 Author: Alan Hourihane Date: Sun Jul 16 20:17:38 2006 +0100 move ContextMem out of XF86DRI commit c7083a267209c93b2a91ef00dea2ca840400d160 Author: Eric Anholt Date: Fri Jul 14 13:23:40 2006 -0700 Bug #7404: Only save/restore VGA fonts and not other VGA regs. This fixes a hang on the i945 during restore. It appears that saving/restoring the VGA registers is not important, as we're correctly saving/restoring the registers we touch within the driver anyway. commit 49b827605628d3e1a6d4d41447cf46c5f38cc0c2 Author: Eric Anholt Date: Fri Jul 14 14:23:04 2006 -0700 Add a register restore implementation so we don't crash on LeaveVT. We shouldn't ever need to save/restore the hi/lo frequency regs, as they're read-only. commit 6a92a779646ec03a03a3b1f45170b2e705ce8934 Author: Eric Anholt Date: Fri Jul 14 14:20:41 2006 -0700 Don't try to probe on more pipes than we really have. commit 04d1584737fd0d14e99608a97281fd7b1549ae0e Author: Wang Zhenyu Date: Fri Jul 14 16:14:18 2006 +0800 Current exa render implement for i830 and i915, test on 865GM and 915G. There is issue in picture 'repeat' support. And also stop recursive behavior in I830WaitLpRing to allow server to abort instead of system hang. commit de470aaf5c47f4d2b0f477ac678039ef43af773d Author: Adam Jackson Date: Thu Jul 13 19:10:11 2006 -0400 In I830xf86SortModes, catch cases where two modes are equal in only one dimension, by comparing the areas of the modes. Otherwise, 800x600 would sort before 1024x600 if it was added later. commit 22843830ebdd14247aa76d19f89494a56e2ac887 Author: Adam Jackson Date: Thu Jul 13 17:20:17 2006 -0400 Fix a thinko; would only inject the FP native mode if a mode list was already found, which is never the case when there's no xorg.conf. commit 05bcbadd130524694e11e372d54cb419cea566cc Author: Eric Anholt Date: Tue Jul 11 14:05:38 2006 -0700 Avoid NULL dereference if cursor position changes during a mode change. commit b65f18b05a5fba506b71293b495cab95197037ac Author: Eric Anholt Date: Tue Jul 11 13:29:57 2006 -0700 Bug #7443: Respect the user's Modes configuration, and make it more useful. Now, mode names generated by DDC get names of the form "WIDTHxHEIGHTxREFRESH". The matching for user Modes lines takes the user Modes as the prefix that needs to match, rather than an exact string match or "WIDTHxHEIGHT" match. So one can, for example, specify "1024x768" to get any old 1024x768, or 1024x768x60 to get one of the modes named 1024x768x60. commit 5a2e04bd1b700a8a6e26136b8831ef5e4d11b565 Author: Eric Anholt Date: Tue Jul 11 10:21:51 2006 -0700 Fix crash with DDC when there are no user modes to add. commit f9499a68da0ce459fed0b29b998678fd81898a51 Author: Luká\u0161 Hejtmánek Date: Tue Jul 11 10:13:18 2006 -0700 Make gamma settings apply to the cursor as well, and fix clone-mode gamma. commit 7068468ac1951bfca0071bb9b1a99df4f37368a0 Author: Luká\u0161 Hejtmánek Date: Tue Jul 11 09:51:26 2006 -0700 Add support for adjusting saturation value of overlay video. commit b1c2ea653502dd8547079e7014b698f241433dff Author: Alan Hourihane Date: Tue Jul 11 08:13:30 2006 +0100 whoops, revert some unnecessary changes commit 8a44a7acfcadbba2410dca750afc9d32bc83706e Merge: 584b544 e7723a4 Author: Alan Hourihane Date: Tue Jul 11 07:41:27 2006 +0100 Merge branch 'master' of git+ssh://git.freedesktop.org/git/xorg/driver/xf86-video-intel commit 584b544987be5cf23dce29ddaf3130e59cfe6fa8 Author: Alan Hourihane Date: Tue Jul 11 07:40:40 2006 +0100 Add an additional check before rotating commit b912bf5673e38e03b0b25c2f5d05fe7e26994ba1 Author: Eric Anholt Date: Mon Jul 10 18:35:20 2006 -0700 Clean up warnings in sil164 module. commit d75490701cdbf2ab6eab82eaa078790a5fe0aea0 Author: Eric Anholt Date: Mon Jul 10 18:21:04 2006 -0700 Hook up SiI164 mode setting (just a matter of turning the chip on). Also adds register dumping in case this turns out to not be enough, and fixes a couple of prototypes. commit 426d26ea446d646fa8f561ea0e03c8e4a2c0c315 Author: Eric Anholt Date: Mon Jul 10 18:19:51 2006 -0700 Fix prototype for SaveRegs. commit 8d7987d00242020d29a2574ac0c8b6e55cc22112 Author: Eric Anholt Date: Mon Jul 10 17:34:57 2006 -0700 Move to 4-space indents in sil164. commit f5a01a2ef02125611d5fb74c20d53d52e544701a Author: Eric Anholt Date: Mon Jul 10 17:17:51 2006 -0700 Make DVO code light up my sil164-based DVI output, when already set up by BIOS. commit 23a0ee73bce12f9e0b881af420413aeec4c0517f Author: Eric Anholt Date: Mon Jul 10 17:16:18 2006 -0700 Fix modelist with a configured monitor to not begin with all unvalidated modes. commit 48ba9273ddfb36d3525e19238b94b18c56667c4d Author: Eric Anholt Date: Mon Jul 10 15:01:51 2006 -0700 Fix validation when the first mode is thrown out, and print hsync in modelines. commit 3924ffb7d7b71cb9c6ab9eeb12d0f1b2c26244c0 Merge: 5a8f6a4 f76f94a Author: Eric Anholt Date: Mon Jul 10 12:12:38 2006 -0700 Merge branch 'dvo-merge' into modesetting Conflicts: src/i830_driver.c commit f76f94a743505da16e121992eb789c1f74eb7673 Author: Eric Anholt Date: Mon Jul 10 12:04:09 2006 -0700 Supply proper NULL-terminated symbol lists to avoid crashing. commit 5a8f6a486d79f50d2d659e615283289d59f9caa4 Author: Eric Anholt Date: Sat Jul 8 15:26:19 2006 -0700 Improve output bus setup to include LVDS setup for pre-i915. commit df333cc9a848bc2299a52a7613fe4ffdff8038a2 Author: Eric Anholt Date: Fri Jul 7 13:41:33 2006 -0700 Initial add of DVO support code. Probes my sil164. This is a mostly-untested merge of airlied's work. The I2C modules are intended to be moved into the core server or a separate driver module when they're functional and we're happy with the API. commit e1064f52b0ff69ea7937897b8c951cc3e32cd752 Author: Eric Anholt Date: Wed Jul 5 16:00:03 2006 -0700 Don't try to probe modes on an SDVO device with NULL sdvo_drv. commit ffa6ecc18bc54151061d9956f1d12575fc057da3 Author: Eric Anholt Date: Wed Jul 5 14:41:08 2006 -0700 More fixes to "choose closest mode for the pipe" code to select correct refresh. commit dfd7fef457c048c9f0d826e37d91453d9e1485b9 Author: Eric Anholt Date: Mon Jul 3 12:01:57 2006 -0700 Bug #7375: Don't double-free the current XF86 mode after a randr reprobe. commit 5d07ebdf4f23e16fb8f60eafeadc947701e7877c Author: Eric Anholt Date: Mon Jul 3 11:59:21 2006 -0700 Only override display size with the XFree86 mode's for actual panel scaling. commit 48f27ac62128251640a9b1ca54f63376676b47eb Author: Eric Anholt Date: Wed Jun 28 15:07:01 2006 +0200 Replace i830InjectModes with i830DuplicateModes usage. The remaining functionality of DuplicateModes was OBE, and the name was bad. commit ce5bd108c55d2378db072617c380514a39672603 Author: Eric Anholt Date: Wed Jun 28 14:21:49 2006 +0200 Validate and insert user and VESA standard modes for DDC or configured fallback. This isn't really tested because I lack a good CRT to test against currently. commit 367f69f8e7710e53dcd286f1b62506a3276e80f9 Author: Eric Anholt Date: Wed Jun 28 13:10:02 2006 +0200 Replace xf86ValidateModes usage with a set of custom validators and pruning. This moves us to maintaining MonPtrs per pipe instead of using the EDID structure "xf86MonPtr", which is closer to what we want to be looking at when doing validation. The new validation isn't enough yet -- particularly, we aren't importing and validating the custom modelines to the pipes when applicable, but this will be easier than (for example) trying to make flat panel modes pass xf86ValidateModes through various gross hacks. Hotplug turn-on/off also happens at SwitchMode time now, instead of at randr probe time. commit 9fbd3d8f4befb75ed6f6bd9a9ffe0175626e8785 Author: Eric Anholt Date: Tue Jun 27 23:31:40 2006 +0200 If the panel power registers are all zeroes on Mobile parts, disable LVDS. This is the case on the Mac mini, which is an i945GM but has no LVDS attached. Powering on with the power timing registers zeroed would probably be a bad idea, even if there was a panel attached. commit 56f7aedd0ad1f5645a90a5509b1263ec6b7b7ee1 Author: Eric Anholt Date: Tue Jun 27 00:37:56 2006 +0200 Fix randr current mode reporting and mode loss/mis-configuration on re-randr. commit b85f268051a785f90e4c55bd1cac80d673388f16 Author: Eric Anholt Date: Mon Jun 26 16:26:58 2006 +0200 Fix remaining warning. commit 6a3f89f4b6035534c7b93a4c05fd704305349785 Merge: 3cdc537 e7723a4 Author: Eric Anholt Date: Mon Jun 26 16:25:58 2006 +0200 Merge branch 'master' into exa Conflicts: src/i830.h commit 3cdc53797bf6fdbfbcb53e07b269a5071f5ec97d Author: Eric Anholt Date: Mon Jun 26 16:23:55 2006 +0200 Fix build with new EXA headers. commit e7723a4e5725147d3bd9ba22c5a3314b0556e440 Merge: 5111b88 dae9cb7 Author: Eric Anholt Date: Mon Jun 26 16:04:33 2006 +0200 Merge branch 'origin' commit 7104b915da412a26661bab7b5a940e935e8cbc8a Author: Eric Anholt Date: Mon Jun 26 15:11:13 2006 +0200 Clean up warnings. commit 21dc3edfc41987bb8cf0f8d265fba9775cbe8cdb Merge: f113e90 5111b88 Author: Eric Anholt Date: Mon Jun 26 15:06:42 2006 +0200 Merge branch 'master' into modesetting Conflicts: src/i830.h commit 5111b883480a5a9cc82200f2684cba67b515aa73 Author: Eric Anholt Date: Mon Jun 26 14:53:10 2006 +0200 Turn on extra warning flags for GCC, and clean up the resulting fallout. commit f113e9002cf53510e30984f816d44b06f1e71216 Author: Eric Anholt Date: Mon Jun 26 12:54:30 2006 +0200 Fix the plane/pipe disabling and turn off missing outputs when no longer DDCed. commit 0e5cda3796ba0164496f0814eb57d1dfa7ab9257 Author: Eric Anholt Date: Mon Jun 26 10:30:46 2006 +0200 Fix FP scaling by using the desired mode to get at the real [HV]Display. commit 51d14f803aae6d0a738520c3cad110289ee444db Author: Eric Anholt Date: Mon Jun 26 10:21:23 2006 +0200 Add a function to turn off unused outputs, DPLLs, planes, and pipes. commit d2c18d8d79596513149273e4a0d322f04bf27e80 Author: Eric Anholt Date: Mon Jun 26 10:15:11 2006 +0200 Fix DDC probing after last (untested) commit. commit 6d1d105d68d9c25890c3fd2ebad1367529c2a991 Author: Eric Anholt Date: Mon Jun 26 10:14:27 2006 +0200 Unset pipe current mode on EnterVT, so we reprogram the mode for sure. commit 52e8231a19f28bd4744f983aee2197a18c20aa3a Author: Eric Anholt Date: Mon Jun 26 07:46:28 2006 +0200 Major cleanup of mode reprobing: - Don't mess with pScrn->monitor->Modes, and instead make our own availModes. - Don't re-program the pipe with the same values (no flicker at xrandr) - Move a bunch of stuff that should be exposed through the public API (probably) to i830_xf86Modes.c - Use a table with established modes plus GTF to come up with modes from EDID, instead of trying to walk and find one in pScrn->monitor->Modes. I think this is correct. - Reset clone state if we've detected new pipes, which should turn on the cursor. commit f5e5f8aeddb3e0d6d073471aeff6176fb54576e2 Author: Eric Anholt Date: Fri Jun 23 23:29:55 2006 -0700 WIP to allow re-probing and validation of modes for new heads at "xrandr" time. Now, DDC modes always end up being preferred to custom modelines, even if smaller. This should probably be fixed by inserting custom modelines into the probed mode list if they're valid according to the probed parameters of the monitor. Too much code is lifted from static functions in xf86Mode.c, and those should be made unstatic if possible. Using xf86ValidateModes is also rather hacky, and I want to break the function down, but this is a first step. commit 89791914d2a78f19f4f60ca370d387e5b1ccfb46 Author: Eric Anholt Date: Fri Jun 23 18:21:17 2006 -0700 Split probed modes out per pipe, and union them into the available modes. This is the first stage of getting runtime monitor attachment. The old i830 GTF code is returned to use to provide suitable modelines for xf86ValidateModes in the LVDS case, even though the LVDS doesn't care about the modeline and just always programs its fixed values. commit dae9cb7712d5d8f88697ca83808c59af08364c0e Author: Alan Coopersmith Date: Thu Jun 22 15:07:16 2006 -0700 Provide definitions of __FUNCTION__ for non-gcc compilers commit bb4810521633b6c3db2fc7d01ddc71325583d265 Author: Eric Anholt Date: Thu Jun 22 09:38:27 2006 -0700 Move FP mode validation next to other mode validation code. commit 66d9a1be302ad34573de98de21cbdf6419592092 Author: Eric Anholt Date: Wed Jun 21 17:11:54 2006 -0700 Detect SDVO display presence at startup and default to displaying to it, too. commit 72e25a7488c2eabcc92e9e0769a89dee687f52fd Author: Eric Anholt Date: Wed Jun 21 16:04:18 2006 -0700 Fix SDVO output at low pixel clocks. I had interpreted the docs as saying that the multiplier setting would further divide the clock and stuff dummy bytes in. Instead, we have to set the DPLL at the higher clock rate, and the pixel multiplier just controls the stuffing of dummy bytes. Also, we have to set the multiplier both in the graphics chip and on the SDVO device on the other side. commit 726443309d72134341cff1f6db978aa1d6e3ce52 Author: Eric Anholt Date: Wed Jun 21 15:38:19 2006 -0700 Add decoding of SDVO command names for debug output. commit 5a1b68993f3a3a2e8dcd428a7118e29c36703cd6 Author: Alan Hourihane Date: Wed Jun 21 08:41:16 2006 +0100 Fix build without DRI commit 16b310823bacab6be4947da234b3a081b0a3cd62 Author: Matthieu Herrb Date: Wed Jun 21 00:12:27 2006 +0200 Fix build without DRI commit 896ffe78fe96469cdd3ade77c8e68e1503967223 Merge: 89c2c4b 52243d4 Author: Eric Anholt Date: Tue Jun 20 15:10:35 2006 -0700 Merge branch 'modesetting-origin' into modesetting commit 89c2c4bc40b8c032915ccb3ed4f3c143c3d8db12 Author: Eric Anholt Date: Tue Jun 20 15:10:19 2006 -0700 Add #if 0-ed code I've been using for CRT detection debugging. commit be08661e3126907c50c54485042fcde00b0da2b4 Author: Eric Anholt Date: Tue Jun 20 14:48:03 2006 -0700 Only default to enabling CRT or LVDS output if they're actually detected. Still, if we haven't detected any outputs automatically (including CRT through DDC), default to CRT anyway. commit b454c9601f005c69c11556a558150403378d34d9 Author: Eric Anholt Date: Tue Jun 20 14:32:40 2006 -0700 Add support for CRT detection using DDC. This method is slower (~5ms), but works on older chipsets. Also, load-based detection is disabled, as it can be fooled by other outputs on the pipe being active, such as LVDS. commit 0b76646666e9d330e77c6f81af8b91e34623be92 Author: Eric Anholt Date: Tue Jun 20 13:57:26 2006 -0700 Add CRT detection function by testing for load, and clean up hotplug version. commit e4584a4f44a70d746396ed48b8e40033504d68b2 Author: Eric Anholt Date: Tue Jun 20 10:39:28 2006 -0700 Remove dead DisplayInfo option. commit ab60e34dcfc52ab5f22a82145d5b4db51b4c62c5 Author: Eric Anholt Date: Tue Jun 20 10:07:47 2006 -0700 Add debugging info for pipe/display plane size. commit 8a6edba33213911cc2210b5e903428b81d45862f Author: Keith Packard Date: Mon Jun 19 13:47:28 2006 -0700 Set vblank interrupt configuration to match pipe configuration New i915 drm ioctl (in version 1.5) allows the X server to select which pipe drives vblank interrupts. Use this to drive from the 'preferred' pipe. Yes, per-window vblanks would be nice in a shared fb environment. Maybe someday. (cherry picked from 2fb375b665f4802819b89f2277fd6154006c11ee commit) commit 52243d407cad93283956660de4771097ac0b4b2d Merge: 2fb375b 34f6a82 Author: Keith Packard Date: Mon Jun 19 13:28:09 2006 -0700 Merge branch 'modesetting-origin' into modesetting commit 2fb375b665f4802819b89f2277fd6154006c11ee Author: Keith Packard Date: Mon Jun 19 13:24:57 2006 -0700 Set vblank interrupt configuration to match pipe configuration New i915 drm ioctl (in version 1.5) allows the X server to select which pipe drives vblank interrupts. Use this to drive from the 'preferred' pipe. Yes, per-window vblanks would be nice in a shared fb environment. Maybe someday. commit b5acc6b3a3a3a109014d6b971f4722d0f0a4c29a Author: Keith Packard Date: Mon Jun 19 13:22:17 2006 -0700 Add backlight control to DPMS logic. Turn backlight on and off in response to DPMS state changes. commit d7e5d996aa123e52857362fa8a1e69f709bfe218 Merge: 89a5a00 a73ab7f Author: Wang Zhenyu Date: Tue Jun 20 01:22:53 2006 +0800 Merge branch 'master' into exa commit a73ab7f0e6e3b0462e05c0031ffd602ed3e2bcd4 Author: Alan Hourihane Date: Mon Jun 19 11:35:42 2006 +0100 additions for rotation fixes commit 1fe3dd38eb613475d62140850e64767defed7d34 Author: Alan Hourihane Date: Mon Jun 19 11:27:28 2006 +0100 Set some invarient state, cures some problems with rotation at startup. This mimicks the 3D drivers setup. commit 89a5a0024a7e077a0726ffe1427e573df3bcbef5 Author: Eric Anholt Date: Sun Jun 18 00:25:34 2006 -0700 Fix crash with EXA during randr, when XAA structures were being accessed. EXA shouldn't have any issues with there also being pixmaps in framebuffer while rotation is active. commit f2ab3aa3199e1c4b0edec72b5acea6d725cc514a Author: Eric Anholt Date: Sun Jun 18 00:32:01 2006 -0700 Fix DRI in EXA mode: Don't attempt to use XAA symbols (use the driver's internal ones for EXA support when available). Also, add an abstraction of XAA/EXA MarkSync and WaitSync functionality so we don't need to sprinkle the ifdefs all over, and correctly use them. commit 3592b432b48d51d2273c1e1064f85e656fbba130 Author: Alan Hourihane Date: Tue Jun 13 21:42:53 2006 +0100 fix 8bpp & 16bpp rotation modes for i8xx series chips commit a50610b7719bfe800c3496c17d0ba77739167b35 Author: Alan Hourihane Date: Mon Jun 12 13:53:20 2006 +0100 Use 800x600 mode to double check commit f02268b2091c9a785d26e82bcb35a8b713463072 Author: Alan Hourihane Date: Mon Jun 12 12:16:58 2006 +0100 Don't rely on register check to find out if we're resuming - it's not reliable. But then, neither is the BIOS, but it's the best we can hope for until Eric's work is complete. Try setting another mode to cater for some broken BIOS' too. commit 672c3d18dbb405095e465126053ff887d891409e Author: Alan Hourihane Date: Mon Jun 12 10:02:06 2006 +0100 Only mark rotation flags after initial screen setup. Fixes bug #7053 commit 34f6a8204f1edec015283fc6b5f196e47897e3de Author: Keith Packard Date: Sun Jun 4 00:15:06 2006 -0700 Get sDVO output working on mac mini. Add lots of register debugging to track delta from BIOS settings. Fix various mode settings to mirror BIOS sDVO values. Disable analog/lvds output on pipe with sDVO. Borrow Dave Airlie's I830xf86ValidateDDCModes code. Fix various sDVO I2C messages to mirror Dave's code. commit 6812b5382077e5d3f421aceeeb2f337e9b3f570e Author: Dave Airlie Date: Fri Jun 2 12:22:14 2006 +1000 intel: fix VT switch DRI locking The DRI locking is incorrect at VT switch, due to reference counting inside the driver. Just call the DRI directly. commit 3f158fd610a3363a23daa7205bcd9f213686cf1c Author: Keith his master's voice Packard Date: Mon May 29 18:05:57 2006 -0700 Nice texture coordinate gradient, broken slightly in y commit bb0ad04d46eba2fed57a888ff960d2436ec7d70d Author: Keith his master's voice Packard Date: Sun May 28 22:59:58 2006 -0700 Ok, finally something sensible up on the screen. Replace PS kernel with constant data source (pink). Dodge g0/g1 so URB data doesn't land on top of thread data. Flip source/dest coordinates (dunno why they're fetched this way). commit ddf3e5b2737399dca6d401f91db51a51f93b6373 Author: Keith his master's voice Packard Date: Sun May 28 21:03:39 2006 -0700 Using tiny rectangle, still locks up in pixel shader program somehow commit 2e16c79dc2f24b0a04111aa6236a44870c6c64df Author: Keith his master's voice Packard Date: Sun May 28 20:31:23 2006 -0700 Lots more debug code. Appears to execute pixel shader thread now though. hurray! commit 79a514412bda7e38e018c105a603970c4a9d758a Author: Keith his master's voice Packard Date: Sun May 28 16:09:59 2006 -0700 dump out piles of debug. Create VS thread just to see how it works commit 9c111d89fe19f1773af2eefb000e1c2389b4b6e1 Author: Keith his master's voice Packard Date: Sat May 27 19:52:13 2006 -0700 Push all of the obvious Mesa state setting into the video code commit 01101196b16010ac3dadab647bfe7000a53fa94d Author: Keith his master's voice Packard Date: Sat May 27 01:05:09 2006 -0700 flesh out cc state. set cull mode to none. enable sf kernel commit 462a860af89ed855fe2b718342fcaf9c169af3fb Author: Keith his master's voice Packard Date: Sat May 27 00:17:25 2006 -0700 Rename BRW instructions, check video instruction generation. Doesnt lock up, but doesnt display anything either commit 9ec7cf22e3f03c13524bb2d15711699dfcc02984 Author: Keith his master's voice Packard Date: Fri May 26 21:30:55 2006 -0700 Use broadwater video code on broadwater hardware. Pad ring to even length. compute state base as address rather than offset commit f5fe700b9a943c956bcfcc3a0d2de13c23b978bc Author: Keith his master's voice Packard Date: Fri May 26 13:47:39 2006 -0700 Prepare real SF kernel and fake WM kernel commit 1549accb6f52498fef3dcbd87bb72d89fcd5bccd Author: Keith his master's voice Packard Date: Thu May 25 16:10:31 2006 -0700 Scale video source vertices. Allocate space for kernels commit c1c46f882f9a11c383c8d1d1ce393be8fda55ed0 Merge: 1e2da24 f2967a2 Author: Eric Anholt Date: Mon May 22 10:42:24 2006 -0700 Merge branch 'master' into modesetting Conflicts: man/.gitignore commit 1e2da2450d7213c0aa4d4bd1fba5723dcda13ddf Author: Lukáš Hejtmanek Date: Mon May 22 09:48:09 2006 -0700 Replace VBE call to do DPMS with native code, and fix screensaver in clone mode. commit 2a1b3cfccb7de53f7ce8f9e4816e4278afb1fcab Author: Eric Anholt Date: Mon May 22 10:32:13 2006 -0700 Use RECTLIST instead of TRIFAN for video so we get horizontal shearing instead of diagonal. Also remove the unnecessary vertex elements that were being emitted. commit bc42dbe07cbd61b357d8eed02608e026c4cf4485 Author: Wang Zhenyu Date: Mon May 22 11:22:03 2006 +0800 fix a typo commit bce209cd3f60cb5d51aadc5fc8ec1a4151435ec3 Author: Eric Anholt Date: Fri May 19 17:13:37 2006 -0700 Put in code for idling accelerator on subsequent cliprects. commit 3640117bd9f2073ff54dc474f0cdefff49742584 Author: Eric Anholt Date: Fri May 19 17:10:04 2006 -0700 Set up the state buffer in framebuffer. commit e70d99d51edc39ea53536c2f4d62ac622ba3e0d9 Author: Wang Zhenyu Date: Fri May 19 15:47:19 2006 +0800 revert wrong setting for memorySize, which should be at the range of pI830->Offscreen. commit de06cd70a9edb8b56d05d3f505137f7c7f083c2f Author: Eric Anholt Date: Thu May 18 18:27:11 2006 -0700 Checkpoint for filling out more 3D state. commit ad7ec6a24b436d5492d38e4fa56845b229cf5fb8 Author: Eric Anholt Date: Thu May 18 15:26:28 2006 -0700 Checkpoint of BW textured video work, filling out vertex submission stuff and some more other state. commit 291770efc691a02650e3c580ca40c2f9fce3896c Author: Eric Anholt Date: Thu May 18 10:43:07 2006 -0700 Start laying out some of the bits that need to be done for BW textured video. Headers taken from TG code drop. commit b3d79ad9ecc70e9f841092fc75b7e23576b685dd Author: Wang Zhenyu Date: Thu May 18 16:31:40 2006 +0800 oop I830FALLBACK define error commit 1bc2a8f7a20b2ea9251e93157c0f2342fdbb951c Author: Dave Airlie Date: Thu May 18 16:58:22 2006 +1000 fixup I830SDVOGetActiveInputs to use return values not args commit 33ad959323f70f76c494c66ec35a04d7d839f612 Author: Wang Zhenyu Date: Thu May 18 11:32:59 2006 +0800 Add DEBUG_I830FALLBACK for easier tracking commit 52af679c94a25c0c6e5df5f3647f62fdd256c826 Author: Wang Zhenyu Date: Thu May 18 11:31:14 2006 +0800 Check alignment require in PrepareSolid, noticed by Eric, as pixmap offset align set is violated in Solid. commit fdb52a8c6da0f3d3f4cfbf1583b2dc4d91f7e933 Author: Wang Zhenyu Date: Thu May 18 11:24:46 2006 +0800 Solid planemask in PrepareSolid/Copy might be 0xffffffff or bits up to the depth. This should save some fallback case, noticed by Eric. commit bc51d6525a12c748d0a293b7e560f6dcea33eecb Author: Eric Anholt Date: Wed May 17 13:42:51 2006 -0700 Turn off overlay video on BW until we have stable PCI IDs so we can know whether the hardware supports overlay. commit f97895efd5532cca145b6f224f9615739b1e8f26 Author: Dave Airlie Date: Wed May 17 14:46:37 2006 +1000 fixup chipid override This makes the ChipID override work so that we actually override the pci id that gets used everywhere in the driver. commit 011a1c99d49c1c69b5fa81ade6a2ed73e84276ae Author: Wang Zhenyu Date: Wed May 17 11:16:00 2006 +0800 Fix memorySize to enable offscreen mem manage. commit b0ac5303f33f75dc607cf0c705c23db1da836983 Merge: 4c72725 c2cd10e Author: Eric Anholt Date: Tue May 16 15:21:17 2006 -0700 Merge branch 'textured-video' into broadwater-video Conflicts: src/i830_video.c commit c2cd10e1fba0e75c0ed3db5d17211bddf7ab1e33 Author: Eric Anholt Date: Tue May 16 10:08:58 2006 -0700 Flag the 3D state as dirty when we draw textured video, which should help rotation (I have other issues with rotation anyway). commit 63a72e46fa20a4a4ba74efed386f6c3c167be5b5 Author: Eric Anholt Date: Tue May 16 08:54:43 2006 -0700 Turn debugging back off. commit 01c043de0393170e98515169f8239fef4d3e2053 Author: Eric Anholt Date: Tue May 16 08:53:40 2006 -0700 Use linear min/mag blending. commit db3683907d15959e79adfb8f0cd94e861fae5c36 Author: Eric Anholt Date: Tue May 16 08:40:53 2006 -0700 For textured video, disable double buffering and sync before uploading new video data. Allows more videos to play simultaneously. commit 29a8e88ed01c9e15a2ceba5eb62b19773e14c1f8 Author: Eric Anholt Date: Mon May 15 17:19:33 2006 -0700 Relax the alignment requirements for textured video. commit c9be11459bc2198b435c97c5a3432425246c4d2d Author: Eric Anholt Date: Mon May 15 17:04:27 2006 -0700 Enable overlay and/or textured video at runtime according to hardware capabilities. Sets up 16 textured-video ports. Left in one hack (disconnected but advertised BRIGHTNESS and CONTRAST atoms) which may actually not be necessary. commit f268979a0c779641c84e8d5b763acbda131474cf Author: Eric Anholt Date: Mon May 15 10:05:19 2006 -0700 Correct drawing issues with planar formats when top or left != 0, and Y didn't get its offset. commit b09fd42d7088ead6c23e040ac4b71114f62de82b Author: Eric Anholt Date: Mon May 15 09:29:43 2006 -0700 Fix the planar formats to display correctly in textured mode. Still has issues with clipping, and some sampling differences between ximagesink and xvimagesink. commit dd48790f4600a880fc4907c6e3b1cd51e9c0f0b7 Author: Eric Anholt Date: Fri May 12 13:27:33 2006 -0700 Divide width by 2 in planar-to-packed conversion loop, since each pass through the loop writes two source pixels. commit eec5e996ec9361099bf81d8d3b66933d5981c5a8 Author: Eric Anholt Date: Thu May 11 20:26:26 2006 -0700 Merge textured-video-wip to textured-video-planar-full. commit 4154a2f74811b91c0ef5bef32a919d6f8baf1a70 Author: Eric Anholt Date: Wed May 10 12:09:00 2006 -0700 Experimental work to use a full pixel shader for planar to YUV conversion, which also doesn't quite work. commit 3e0a9c9082942eb6f52612235d84b8408e1e03e9 Author: Eric Anholt Date: Tue May 9 16:57:19 2006 -0700 Do a separate BEGIN/ADVANCE_LP_RING set in the planar vs packed blocks, so I can adjust the planar code more easily. commit 3af4a967e73b367bb531f2760b4803db1388bcf9 Author: Eric Anholt Date: Tue May 9 16:50:48 2006 -0700 Add a couple of macros to simplify writing of video pixel shaders. commit 06e62ec521ed3f7ed232ace8e188891bedb53097 Author: Eric Anholt Date: Tue May 9 12:04:58 2006 -0700 Commit a WIP implementation of the planar video shader that does the planar-to-packed conversion for us. Unfortunately the documentation is unclear, and I haven't managed to get any implementation of it working correctly. commit b1090a42b200710628dd8b0c7ced15db7bbe71a1 Author: Eric Anholt Date: Mon May 8 14:22:00 2006 -0700 More magic number reduction in rotation code. commit 3a2d8af214a79591322ce6e5546f856a1ee41736 Author: Eric Anholt Date: Mon May 8 09:42:40 2006 -0700 Add initial textured XV support for i915, which can do YUY2 and UYVY, but fails on I420 and YV12 currently, doesn't support the composite extension, and should break XV support on non-i915. commit 66875c1559bc20b531ab72e5d6b921d9f50b29f3 Author: Eric Anholt Date: Fri May 5 10:15:23 2006 -0700 Convert magic numbers in i915 rotation 3D state to symbolic names. commit 4c727254da354cfd6f35148a334d046d67a50e99 Author: Eric Anholt Date: Fri May 12 16:26:58 2006 -0700 Remove the local, renamed copy of lnx_agp.c. The diff between lnx_agp.c and it appeared to be removal of bugfixes. commit a115c4b872a385530dcf94b7e7f2fa9b3b7e3155 Author: Eric Anholt Date: Fri May 12 16:11:29 2006 -0700 Remove the code that changes our behavior based on whether a magic file exists with magic contents in /tmp (created by some install script). commit 0cb251fadca1cbb3d4c5b97982cd0d8c2fc3e840 Author: Eric Anholt Date: Fri May 12 16:10:02 2006 -0700 Do a couple of reverts to get the DRI code building. At this point, the whole driver builds. commit 78b95386b630039864b31954ebcd02ec8829b0c8 Author: Eric Anholt Date: Fri May 12 16:09:40 2006 -0700 Remove intel_randr.c and stick with the previous code. Broadwater shouldn't be changing our randr, and the new version didn't compile. commit 2e58aa401dfbab438752038a9034df571c8f8bde Author: Eric Anholt Date: Fri May 12 15:54:37 2006 -0700 Make the intel_acpi.c code non-modular and make it compile. I think we'll end up nuking this code anyway, as keithp (and I, as well) disagree about how ACPI should be handled, but the goal is to compile at the moment. commit 2cd6c8fa2321ca217ef89db1027dbe9e716ad7aa Author: Eric Anholt Date: Fri May 12 15:37:44 2006 -0700 Revert internal shadow module changes back like master, along with RandR initialization. RandR initialization retains the no-rotation setting for BW. commit 88558ebeed12d6cefd73bba0ddac3c043861ac89 Author: Eric Anholt Date: Fri May 12 15:00:17 2006 -0700 Start fixing up the build and remove a regression from master (I think) in rotation. commit 9e387ef92be9b38c68bda8a6a28b0d9eb98d53a4 Author: Eric Anholt Date: Fri May 12 13:41:38 2006 -0700 Re-add authorship note in i830_driver.c accidentally left out of last commit. commit af2432322ba1d561057c34ab185561a8e799e8cd Author: Eric Anholt Date: Fri May 12 13:32:38 2006 -0700 First pass of integrating the Tungsten Graphics driver for Broadwater. This patch is based off of diffing from the branchpoint to the supplied code, but with many chunks containing reversions of commits removed. Won't work yet. commit fbba4312e7cb326fdcf6a71194c4fc4a7a9b5488 Author: Dave Airlie Date: Fri May 12 18:44:27 2006 +1000 add sdvo capability reading support This reads the SDVO cap bits and uses them to figure out the input/output to save/restore commit f2967a2f5f47b636b2445fa69dbc3ec79e065c90 Author: Keith Packard Date: Tue May 9 13:51:25 2006 -0700 Video overlay gamma bounds checking must be done bytewise. Also, pend bound computations to register writes to allow updates to individual values that are 'out of spec' so the client can update multiple values. (cherry picked from 190f9ad0606e96e684e0b028d576d822dc9aa3cf commit) commit 190f9ad0606e96e684e0b028d576d822dc9aa3cf Author: Keith Packard Date: Tue May 9 13:51:25 2006 -0700 Video overlay gamma bounds checking must be done bytewise. Also, pend bound computations to register writes to allow updates to individual values that are 'out of spec' so the client can update multiple values. commit fb10966e9f43f44488097b0daf2fbe10c6a804dc Author: Eric Anholt Date: Thu May 4 18:52:08 2006 -0700 Use xf86int10Addr() when calculating an address to read the video BIOS from, out of int10's copy. Fixes a crash on FreeBSD. commit a66f2c01f7d557d860883346671fb864807dbdca Author: Eric Anholt Date: Mon May 1 12:35:21 2006 -0700 Stop doing the BIOS memory size tweaking now that we don't ask the BIOS about what modes are available. commit 99b0e53244aa49f4e7461a1c96aba69453b3516f Author: Matthieu Herrb Date: Mon May 1 10:47:09 2006 +0200 Fix non-dri build. commit 86f0119f47d8be975016ffdea850f0227e32eebc Author: Matthieu Herrb Date: Mon May 1 10:41:10 2006 +0200 Fix non-dri build. commit fcec956807038a2e90287850df8b4410b1f8b3c2 Author: Wang Zhenyu Date: Sat Apr 29 10:55:25 2006 +0800 Save initialize EXADriverPtr, and free it in failure. commit 0be9bc604420e4433697cc31e8ac58c9f36185e1 Author: Wang Zhenyu Date: Sat Apr 29 10:54:02 2006 +0800 I830CheckDevicesTimer() sync should be in I830_USE_XAA case. commit a00a3d198aee7c173c2c36858f4447e13c86eb7f Author: Wang Zhenyu Date: Sat Apr 29 10:52:45 2006 +0800 fix I830BIOSAdjustFrame() sync in I830_USE_EXA case, use exaWaitSync() helper. commit c94cc67bcc507e8256ae4b51b84dfe3f4aa12ab5 Author: Eric Anholt Date: Fri Apr 28 15:33:57 2006 -0700 Update to EXA 2.0 API. commit a4a4ced50cc235e79013ce8dd78e35c6eb4c98ca Author: Eric Anholt Date: Fri Apr 28 14:59:52 2006 -0700 Get it to compile. commit 941a49f91eac02e256706d8821a2cb069ff7ffef Author: Eric Anholt Date: Fri Apr 28 14:57:03 2006 -0700 Commit what applied from Jesse Barnes's i830-exa-latest.patch. commit a555e28e5afc81969ef7b28482e654cc26b3a446 Author: Eric Anholt Date: Mon Apr 24 16:55:44 2006 -0700 Correct some SDVO-related register definitions. commit cc70e6b789a1901bbe4e3501b6b654542d3cdc20 Author: Eric Anholt Date: Mon Apr 24 16:49:48 2006 -0700 Save/restore the output's SDVO reg. Note that we might be programming the other SDVO reg instead of the one assigned to the output. When trying to fix that, I ended up getting no output at all. commit b498d2b1d1170123595ada65353428578b59a361 Author: Eric Anholt Date: Mon Apr 24 15:42:46 2006 -0700 Start trying to save/restore SDVO state on VT switches. commit 56c1f8b0de9b6acc50f51561caf14d3e6bac09ec Author: Eric Anholt Date: Mon Apr 24 14:10:20 2006 -0700 Clean up some argument passing, and remove extra SetTarget{In,Out}Puts that had no effect. Note that we are currently trying to program both outputs of any SDVO device the same way. commit 9ba5319b36e7286e33cf0dcdd804dfc0458a81f5 Author: Eric Anholt Date: Mon Apr 24 13:55:05 2006 -0700 Only write out as many arguments as the commands need, and fix up the numbers in some cases. Pretty-print the return status. commit d32514aee4b00b035652830e8b5e6c0b43cf159c Author: Eric Anholt Date: Mon Apr 24 12:21:45 2006 -0700 Start trying to implement DDC over SDVO. It's slightly tricky because the control bus will reset from DDC mode to internal-registers mode after every Stop afer a Start on the DDC bus. The xf86 DDC code causes multiple Start/Stops in one probe. So, we create a wrapper bus that does the control bus switch at every Start. It's not working yet on my hardware, but I'm pretty sure this is the right way to go. commit effab21c3d108fac7a4e28ae4dabb0b5f74a5380 Author: Eric Anholt Date: Mon Apr 24 11:42:24 2006 -0700 Set displayWidth to a sufficient value for the modes we come up with for LVDS. Reported by: Lukáš Hejtmánek commit bcb441225d1365435bc3373901180de944298e86 Author: Eric Anholt Date: Mon Apr 24 10:54:45 2006 -0700 Simplify the i2c code by using the GetBits/PutBits interface rather than reimplementing it. commit 47bd9059431eadfd8824e496eb91bb50efa0e282 Merge: 729c373 32a0ad5 Author: Eric Anholt Date: Mon Apr 24 10:36:24 2006 -0700 Merge branch 'lukas-resume' commit 1dbb19059f5dd5cd2ad386c4f73498095987b943 Author: Alan Hourihane Date: Thu Apr 20 10:43:19 2006 +0100 fix bug 6365 commit 729c373121ce2bbb0d813cc923f1254e8b37a025 Author: Dave Airlie Date: Thu Apr 20 13:34:55 2006 +1000 move sdvo output setting we have to set the sdvo register a lot earlier in order for them to sync properly otherwise my monitor doesn't sync unfortunately, also disable the sdvo while tweaking the PLLs. This also comments out a setting that seems to break my system here for Eric to look at later. commit 0ba7b13fb4410c6a48b2fb098d2033e040eca6d2 Author: Dave Airlie Date: Thu Apr 20 12:29:06 2006 +1000 fix type 0 instead of O commit 2991d81a3b643161babab6b8f44c057aaaf351c3 Author: Dave Airlie Date: Thu Apr 20 12:27:47 2006 +1000 correct height parameter in sdvo packet commit b5f099e03a9f58b6b99933fb06526bce4db72bd3 Author: Dave Airlie Date: Thu Apr 20 09:50:36 2006 +1000 cleanup sDVO for device on C only This destroys the i2c device properly if the device isn't detected, and allows sDVO to work on GM chipsets, and doesn't initialise the i2c bus twice for sDVO. commit 2909802de63756972b38651a496b4ff1b36ac8a2 Author: Eric Anholt Date: Wed Apr 19 16:29:06 2006 -0700 Clean up SDVO initialization, include config.h, and make it check the right slave address on the first device as well. This gets me to the point of bringing up some modes on my device. commit 32a0ad570d9c010e7d26d980830f719782d9f2f3 Author: Lukáš Hejtmanek Date: Wed Apr 19 19:43:45 2006 -0300 Add more registers to save/restore. Save/restore palette as well commit 132dc0599cf44389c4cc03919f1da8d3a0762b44 Author: Eric Anholt Date: Wed Apr 19 15:04:17 2006 -0700 Whine if SDVO I2C device init fails, rather than be silent. commit 88bb4b578857588f34ac84b7a20577139eccab6d Author: Eric Anholt Date: Wed Apr 19 14:23:45 2006 -0700 Add more SDVO code. It's taken from airlied's driver, but with magic numbers replaced by symbolic names in many places. I tried to restrain myself from functional changes in airlied's code in this pass. commit d8f7dfac769d7b03f069306b1296bb2e1e08b009 Author: Eric Anholt Date: Wed Apr 19 10:45:13 2006 -0700 Start bringing in some SDVO code, mostly from airlied. commit a371a04a57620b7128e3c4395bc7c2ac55effe19 Author: Eric Anholt Date: Mon Apr 17 14:10:50 2006 -0700 Use the IS_I9XX macro intead of >= i915G, since by PCI ID number, 855GM > 915. commit 33413a3cf34b06e3207fe1cdb733d586d55a4337 Author: Eric Anholt Date: Mon Apr 17 13:42:57 2006 -0700 Remove some dead code and one particularly useless debug printf. commit a85be820033af8bf30d1fec4a899349af35a6ef6 Author: Ian Romanick Date: Mon Apr 17 12:52:30 2006 -0700 Add missing new-line character in log message. commit 365b4a53ee965002a5452e6f6016b528e51cee69 Author: Eric Anholt Date: Mon Apr 17 01:12:28 2006 -0500 Disable the BIOS dump-to-file since it's a trivial local DOS, and I can just ask people to turn it on and recompile when I need to. commit ee7be006b63b6b1ce7f786b045fb8f26d337433c Author: Eric Anholt Date: Sun Apr 16 13:13:42 2006 -0500 Bug #6589: Use alternate offsets to successfully get at the panel data for some broken video BIOSes. commit 786ee3df1726f08953167d05f7fa1930452703bb Author: Eric Anholt Date: Sun Apr 16 13:05:35 2006 -0500 Add a standalone program for parsing VBT. commit 62652127cd12f5a0fc9364285b81d2661372148a Author: Eric Anholt Date: Sat Apr 15 12:16:46 2006 -0700 Check for LVDS BIOS tables exactly once at startup. While there, dump the BIOS data to /tmp/xf86-video-intel-VBIOS, for offline debugging. commit 830fa81792a613fe2127a2b89a3eaa326f56114c Author: Alan Hourihane Date: Wed Apr 12 21:55:31 2006 +0100 fix typo commit 3c9bc19db63a317db9e2d67a5bc15b1de1a1060a Author: Alan Hourihane Date: Wed Apr 12 21:54:50 2006 +0100 fix typo commit d6edffee7d987ef551e1a94d9fac21beb72a9598 Author: Eric Anholt Date: Wed Apr 12 12:16:51 2006 -0700 Initial support for pre-i915 PLL programming. Untested. commit 59f88955f57cf0f98458b57418dae25cf53ca180 Author: Eric Anholt Date: Wed Apr 12 11:11:14 2006 -0700 Automatically enable clone mode if we detect two active outputs. commit a797a6626a8117cd16456a206ed96300e4a0ceeb Author: Wang Zhenyu Date: Tue Apr 11 16:44:42 2006 -0700 Fix a broken test that resulted in page flipping always being enabled, not just when the user enabled it. commit 66d216da3bfe7c143d681b610130e6daf6dfa52f Author: Eric Anholt Date: Tue Apr 11 16:37:10 2006 -0700 Replace .cvsignore with .gitignore. commit 185b5251419724fa9377421d67981daa674908c8 Author: Eric Anholt Date: Mon Apr 10 20:25:13 2006 -0700 Add missing MIT copyright licenses, which should have been on all versions. commit c0a2dc608d95f92b0f5a151c623745f09df3afc5 Author: Keith Packard Date: Mon Apr 10 20:09:37 2006 -0700 Save/restore PFIT_CONTROL. Eliminate compiler warnings commit 498abf474b90b04bc4085397a209cc493d98d15c Merge: c26f073 3de82ff Author: Keith Packard Date: Mon Apr 10 19:41:50 2006 -0700 Merge branch 'modesetting-origin' into modesetting commit c26f0737e2c0f000f67f21ef6007b351160036a4 Merge: 11ad8a5 692a4a4 Author: Keith Packard Date: Mon Apr 10 19:41:41 2006 -0700 Merge in master to modesetting commit 3de82ff3938c6559c90079be0c28dc507d62f79e Author: Eric Anholt Date: Mon Apr 10 15:10:15 2006 -0700 Improve LVDS modes when outputting a CRT+LVDS combo at larger than the LVDS's panel size. This is a hack until we get better clone mode, but it correctly displays a subset of the root on the LVDS by using a correct pixel clock and pipe/display size. commit 692a4a4bafd4cab197b761cda22368581b78f996 Author: Adam Jackson Date: Mon Apr 10 11:27:25 2006 -0700 Bump to 1.6.0. commit 43c93057c044f33c20b1dd639eaeabefc0901589 Author: Aaron Plattner Date: Mon Apr 10 11:26:33 2006 -0700 Add a DrawablePtr argument to the XV functions to pave the way for redirected video. commit 39bdfbb4c23c9ba2a44888f22330a892c4becedc Author: Alan Hourihane Date: Mon Apr 10 11:26:06 2006 -0700 x86_64 warning fixes commit 11ad8a590d65849be00e1be4e9dd52c1159a4f24 Author: Eric Anholt Date: Mon Apr 10 09:39:24 2006 -0700 Remove the int10 POST on EnterVT. It has been reported to cause crashes now, and I believe it is the responsibility of the kernel to bring the device back to a mostly-sane state on resume anyway. commit 4217ce18cec257cad435adf9ddc9258a3c8164ec Author: Eric Anholt Date: Mon Apr 10 09:37:02 2006 -0700 Warnings cleanup. commit 334512e0604c208ffec914374a76d85720c1dcf9 Author: Eric Anholt Date: Mon Apr 10 09:32:45 2006 -0700 Don't try to use LVDS by default on chipsets that won't have an LVDS attached. The BIOS tables may still exist, so we can't rely on their presence to indicate LVDS attachment. commit 69083a2fc3ca4a3d06c1985c8a630d5628c1110c Author: Eric Anholt Date: Mon Apr 3 16:29:14 2006 -0700 Remove BIOS save/restore code, fixing VT switching. Removes some other dead code, too. commit 2e5d85fb83def483ab1fd96877aae4a89a962b7f Author: Eric Anholt Date: Mon Apr 3 16:27:24 2006 -0700 Turn off the VGA plane when we're setting our native modes. commit ceb08d28f4a4e9f42c9417938b8541bf1b4e8245 Author: Eric Anholt Date: Mon Apr 3 15:32:41 2006 -0700 Remove the hacky-looking Set640x480. I'll be removing things it depends on (setpipe), and I suspect we'll end up with different hacks for resume, anyway. commit 1f0ba458d02f7d4777c1669aae02138f3a6628c1 Author: Eric Anholt Date: Mon Apr 3 15:28:47 2006 -0700 Remove more BIOS stuff, particularly Get/SetDisplayDevices. Now, if there's no hardcoding in the config file, we default to LFP if we detect it from BIOS, and LFP or CRT if we can get EDID out of them. commit b20b466aaed54708ae9e4676623c8c394a6f00d5 Author: Eric Anholt Date: Mon Apr 3 14:28:55 2006 -0700 Remove the now-unused displaySize and pipeDisplaySize. commit d960c3ca1512a58a53b5c24702cb5c97124817ee Author: Eric Anholt Date: Mon Apr 3 14:12:29 2006 -0700 Remove more BIOS modesetting stuff. commit 4e3a4827007d624aa3da1a9f5a299837bd601a33 Author: Keith Packard Date: Fri Mar 31 14:05:46 2006 -0800 autodetect LVDS dither. Fix 16bpp depth selection commit 8d27f8246ed5a73f7a78043e128b47be784186c0 Author: Eric Anholt Date: Tue Mar 28 13:49:41 2006 -0800 Clean up (and I believe fix a couple of bugs in) the divisor selection code. commit e45581cd073b23a62719f20783d617bd35316fec Author: Eric Anholt Date: Tue Mar 28 13:40:32 2006 -0800 Fill in clock information from VBT table, so that a good clock value gets requested for choosing divisors. commit 33977d23830b5f9bb7d9e2e9c141f91cb127b7de Author: Eric Anholt Date: Mon Mar 27 18:50:53 2006 -0800 Port code from radeon driver for panel mode validation, which will hopefully get the right mode chosen on the VAIO. Untested. commit febdfa967dbe1df487db71ba5f677ef37450bf7a Author: Eric Anholt Date: Mon Mar 27 15:19:52 2006 -0800 Remove some VBE DDC code that I believe is OBE. commit f86892c9163473a683ab591adeb9fb1c0be2ef99 Author: Eric Anholt Date: Fri Mar 24 16:00:30 2006 -0800 Make auto panel fitting work by pulling LVDS timing EDID info out of the static BIOS table, and always using that. commit f6a3243f40074aee471ec1cda2cc8e5f05116284 Author: Eric Anholt Date: Fri Mar 24 13:49:10 2006 -0800 Add more VBT reading, so we find the panel size. commit 64756e215016730b5cc2e174f08d47c0288c0ba4 Author: Eric Anholt Date: Fri Mar 24 11:55:01 2006 -0800 Start interpreting VBT information with the goal of pulling out LVDS timings for panel fitting. commit c2d554be671f5368bd98867b8ecf9428e4560a81 Author: Eric Anholt Date: Thu Mar 23 14:33:04 2006 -0800 Save/restore a couple of important VGA plane registers, and have vgahw save/restore everything it knows about. Also moves the save to just after the other reg saves, above the VBE save stuff which may have side effects. commit 8411c126ae66239f8b3a2261e338a723c36aa44f Author: Eric Anholt Date: Thu Mar 23 11:31:24 2006 -0800 Update to newer airlied DDC code, and do some cleanups as well. Now tries DDC on LVDS, though my current LVDS gives no results. commit 6414ad89b9d368a032adf2358a65404f5443ef35 Author: Eric Anholt Date: Thu Mar 23 09:48:18 2006 -0800 Add enough of airlied's DDC/I2C work to pull EDID info from my CRT. commit 5215e19af58f5c2746c8d281e33ecee86e55f0a5 Author: Eric Anholt Date: Mon Mar 13 10:29:51 2006 -0800 Add some register information for VGACNTRL, and do the reg-debug thing on it. commit b77bdc2c15640a3b15233f2190179d66bc8b2a4b Author: Eric Anholt Date: Wed Mar 8 15:08:06 2006 -0800 Move the PFIT enabling before the enabling of the pipe, as the specs say we should. This doesn't work yet, as we apparently need to adjust the pipe to output at the full resolution of the panel, not the displayed image size, because PFIT controls scaling between the display planes and the pipe. commit de4a9e4a7891daa1488d17bf4c22283759f97373 Author: Eric Anholt Date: Wed Mar 8 12:54:14 2006 -0800 Add a couple more bits to panel auto-fitting, though it isn't working yet. commit 02341aa68cf2171e037196019f2553f0d1b1bd0d Author: Eric Anholt Date: Mon Mar 6 13:23:29 2006 -0800 Make the reg-snapshot code work, add a couple more registers, and use it on VT switching to hopefully help point out where we aren't restoring enough. Currently doesn't reveal anything. commit 2b6fc7ddb020a5511f6f599a2f5c37db27eddbcc Author: Eric Anholt Date: Fri Mar 3 15:42:15 2006 -0800 Add missed save/restore of DSP*SIZE registers. commit 12ce799818722473dde2f82739d50ba4ec7f6ecd Author: Eric Anholt Date: Fri Mar 3 15:38:31 2006 -0800 Add some untested debugging code to for helping figure out VT switch issues. commit 585cc5f256b8e91460414a26409d0e484a86718c Author: Eric Anholt Date: Fri Mar 3 10:37:26 2006 -0800 Add more integrated LVDS support code. commit 33c62e8e654fa2228bc3fd31ec0fe010703f363a Author: Eric Anholt Date: Thu Mar 2 17:53:38 2006 -0800 Change the order of reg writes for restore, possibly increasing chances of success. Also save and restore ADPA. commit 7f4dc09dcafc7f0423b7f3e970a0b19fbbf515dd Author: Eric Anholt Date: Thu Mar 2 15:55:18 2006 -0800 Remove some dead VBE code. commit f3ec8653ab4c9b26d249bcf4393326de37ac8321 Author: Eric Anholt Date: Thu Mar 2 14:31:39 2006 -0800 Include config.h, fixing many issues with reading from pI830. commit 84fc3971d76f78546a0bb7dd57dba52dd893a921 Author: Eric Anholt Date: Thu Mar 2 13:35:40 2006 -0800 Turn off panel power before poking pipe B, and turn it back on if we're doing LVDS. Also, refuse to do LVDS on pipe A, as the docs say you can't. commit aca0f942f377d997c5c0b5be790ee7f255d6e046 Author: Eric Anholt Date: Thu Mar 2 13:31:51 2006 -0800 Fix prototype for i830DetectCRT. commit df6347139bc9de22af981f776d711ac9d44fad1c Author: Eric Anholt Date: Thu Mar 2 12:52:19 2006 -0800 Use DISPPLANE_32BPP_NO_ALPHA for depth 24, else the empty alpha channel results in a black display plane. commit 8e819ced8397f0330fd44efa60fef89bd10ff70e Author: Eric Anholt Date: Thu Mar 2 11:55:00 2006 -0800 It appears from the specs that you have to wait for vblank after disabling the plane, not the pipe. commit ee2410b965321e3d9d64fc500fdcfa6d6dc3a039 Author: Eric Anholt Date: Thu Mar 2 11:43:41 2006 -0800 Move display base setting to a separate function and use it from both mode setup and AdjustFrame. Rename AdjustFrame to reflect the fact that it doesn't touch the BIOS. commit 980fb78e759b752a6cae3c87d188c50ab131ba80 Author: Eric Anholt Date: Thu Mar 2 10:49:31 2006 -0800 Correct capitalization of i830SetMode. commit 2b842c7ee3f9c33cabf16561f11bf1c3a2502d4a Author: Eric Anholt Date: Thu Mar 2 10:08:20 2006 -0800 Start implementing support for setting modes for multiple output devices, particularly LVDS. Untested. commit a09d9cf27af7016bbd9e562c7b1c338154fd3d40 Author: Eric Anholt Date: Thu Mar 2 09:05:13 2006 -0800 Replace I830VESASetMode with simpler i830SetMode, which is all native and doesn't have BIOS-setup workarounds. Multihead behavior may be wrong, and it is all untested. commit 7edb25789bd13f26134e8f5bf493897c8df322e0 Author: Eric Anholt Date: Wed Mar 1 14:02:51 2006 -0800 Start making i830SetMode aware of multiple pipes. commit b5915ac77a3887cd3aa4ce0bb77b8a36e1aa5a1b Author: Eric Anholt Date: Wed Mar 1 13:12:25 2006 -0800 Clean up a couple of warnings. commit c20e15fbe4daeb7288f5c56cf5467eed13686080 Author: Eric Anholt Date: Wed Mar 1 13:08:38 2006 -0800 Disable I830VESASetVBEMode on restore, since I've removed that code. Note that this means that VT switching is broken. We're fine with that for now, but at least now we get a bad display rather than X crashing about an undefined symbol. commit 7d37c5036956609d0c7ae874d0dc1c616f76f849 Author: Eric Anholt Date: Wed Mar 1 13:06:04 2006 -0800 Re-disable chatty debug code. commit 2538cbeb5014e40280a3ae782a755512b1dc85fa Author: Eric Anholt Date: Wed Mar 1 13:01:52 2006 -0800 Enable -Wall on GCC, and do a cleanup of existing warnings. commit 7746da3b346968ab8d2534fc158d026da67cc7b5 Author: Eric Anholt Date: Wed Mar 1 12:49:06 2006 -0800 Move LVDS panel power state setting to a separate function, and reduce magic numbers. commit 142bc4f91a5d776e7ab44cc5fb2328e3f7267557 Author: Eric Anholt Date: Tue Feb 28 18:50:57 2006 -0800 Write the ADPA (CRTC) register on mode setup. Obtained from airlied. commit 6877f532c5f41a445d41eb6a9982bf6bcf691bf2 Author: Eric Anholt Date: Tue Feb 28 15:56:06 2006 -0800 Add untested CRT detection code. commit a085813714818a23aa8d326804f642cd699e0f17 Author: Eric Anholt Date: Tue Feb 28 15:16:59 2006 -0800 Add adjustments of PLL divisor limits for "Almador". commit 2d4415ff1729554537b20be2b6c878444485e406 Author: Eric Anholt Date: Mon Feb 27 16:57:43 2006 -0800 Remove the hacked local xf86SetModeCRTC now that xf86SetCrtcForModes is taking effect on all of our modes. commit 611264ab9b82bca6648a9b27e5ba4b7457c46aa6 Author: Eric Anholt Date: Mon Feb 27 15:44:11 2006 -0800 Remove more VBE stuff and start using xf86ValidateModes. Guessed some params to xf86ValidateModes. commit 9838f639911e8c165ab615fdd9a00e02fa4b3f8a Author: Eric Anholt Date: Mon Feb 27 14:12:50 2006 -0800 Always ErrorF output debugging info when doing BIOS calls, including file/line. commit 767944e3782f9941e9fc72a6705cc3115a6e24ac Author: Eric Anholt Date: Mon Feb 27 13:08:46 2006 -0800 Replace a few magic numbers with symbolic names. Reviewed by md5. commit 4955cd267e7f8ed70e90b2a3de6f93de2ef859c8 Author: Eric Anholt Date: Mon Feb 27 09:37:47 2006 -0800 Add untested save/restore code, and starting on not using VBEValidateModes commit 73496b765c9783a8a271b4774a44fa263dd47684 Author: Eric Anholt Date: Mon Feb 27 09:26:51 2006 -0800 Change an if statement to a more obvious but equivalent case statement. commit 9fe316da7390918dd88940087567314b6f253981 Author: Eric Anholt Date: Thu Feb 23 13:49:31 2006 -0800 Add the manpage's possible names when generated. commit 1555229f29fa7479d6a7a51f451d04a5ef3460bd Author: Eric Anholt Date: Thu Feb 23 13:48:26 2006 -0800 Move .cvsignore to .gitignore commit 139d33ac807fb0fc35c37c3689a6e80238199442 Author: Eric Anholt Date: Thu Feb 23 13:46:30 2006 -0800 Initial add of native CRT modesetting code. commit 35e4756645891bfc85f25f1ab60c287eba9ef749 Author: Dave Airlie Date: Wed Apr 5 07:12:00 2006 +0000 Fix some argument inversions in xf86DrvMsg function calls commit 6e5f9b026372eea295c390b18fc41f7018e0e797 Author: Kristian Høgsberg Date: Tue Apr 4 18:22:32 2006 +0000 Add more missing #include's, in particular assert.h. commit fdfce49c54dabf0f6b0d20aa7e5eeb62b5b57eda Author: Kristian Høgsberg Date: Wed Mar 22 23:24:03 2006 +0000 Drop libc wrapper; don't include xf86_ansic.h and add includes now missing. commit 58817c9909ba659d0070afab0ff7f4e26f722403 Author: Alan Hourihane Date: Fri Mar 10 13:07:00 2006 +0000 Enable gamma for hw cursor when ARGB cursor in use (Lukas Hejtmanek) commit ab4c2f3bf4b26fdd597ea7b7a726ae77778617b2 Author: Alan Hourihane Date: Fri Mar 3 16:27:37 2006 +0000 Check the version of shadow being used and turn off rotation if it isn't the right one. Bump the Xvideo resolution support from 1920x1080 to 1920x1088. commit be6216cd8c6a47dd053240f7a5d1445b61f79038 Author: Alan Hourihane Date: Mon Feb 20 03:25:50 2006 +0000 Fix a rotation problem when DRI is disabled commit 073c4a6a18e98e2dde114756b5d787b13bb2b7a5 Author: Alan Hourihane Date: Mon Feb 20 03:22:26 2006 +0000 Fix a rotation problem when DRI is disabled commit 7cca2ae73e0114a9df2152ed5ed5b66fc44f65dd Author: Alan Hourihane Date: Sun Feb 19 00:04:49 2006 +0000 fix another build problem commit 2e01277da0be55687ca901efadcaf5c919826824 Author: Alan Hourihane Date: Fri Feb 17 16:16:10 2006 +0000 fix build failure commit 86fa25369353471dc9a1041df74e66e9029863a7 Author: Alan Hourihane Date: Wed Feb 15 18:31:22 2006 +0000 a little more of a workaround until we unplug the frontbuffer from libdri commit 32f97e02ac3611fd5e3870826dedf701f3b69a51 Author: Alan Hourihane Date: Fri Feb 10 21:59:37 2006 +0000 temporary workaround until DRI frontbuffer mapping code is removed from libdri.a commit 50e5c2696389cb4d5907a822aa79d5e0e03cee8c Author: Alan Hourihane Date: Thu Jan 26 09:55:15 2006 +0000 update Changelog commit 91d2ba5743041077754de3bd3565df930b036c7d Author: Alan Hourihane Date: Thu Jan 26 09:05:50 2006 +0000 add changelog commit 076971b95ed3b89e1a333adbffceab1c2a5b7743 Author: Alan Hourihane Date: Wed Jan 25 22:22:42 2006 +0000 When going dual head only take a portion of memory for the second head instead of doubling the requirements. commit 77d1c579f3031789031dd203175d3c9555226f79 Author: Alan Hourihane Date: Wed Jan 25 22:17:00 2006 +0000 Fix a glitch in a dual head fix from #3105 commit d8d749080923e88376db1aa18ae2ad722d5113d9 Author: Alan Hourihane Date: Tue Jan 24 15:20:34 2006 +0000 Bump version to 1.5.0.0 Add Intel 945GM support Add RandR rotation support (full 3D acceleration, HWcursor & Xvideo rotated too) Remove shadow framebuffer rotation code Add a new LinearAlloc option to allow more offscreen memory to be allocated for XVideo applications. This allows HDTV movies to be played via Xvideo. commit 919d36b6717e9a7d930c5120309778da7b0fe16c Author: Kevin E Martin Date: Wed Dec 21 02:30:00 2005 +0000 Update package version for X11R7 release. commit 92962f32fccac6df751268190928d8becb558b64 Author: Adam Jackson Date: Mon Dec 19 16:25:53 2005 +0000 Stub COPYING files commit d2ef609211306b6bd3861c0e5f6d3a9af3e145fb Author: Alan Coopersmith Date: Mon Dec 19 09:07:50 2005 +0000 Typo fixes, mailing list & url updates, and other changes to prepare for X11R6.9 & 7.0 releases. commit 1c51ff5c652690d66d1f7cba803c0bb26208b509 Author: Kevin E Martin Date: Thu Dec 15 00:24:17 2005 +0000 Update package version number for final X11R7 release candidate. commit c5007a4b46ffb66f5d98abe8bcae6df8c6eb1b31 Author: Kevin E Martin Date: Tue Dec 6 22:48:35 2005 +0000 Change *man_SOURCES ==> *man_PRE to fix autotools warnings. commit 15df4215347854604fd3f6795ac5f971444cf39d Author: Kevin E Martin Date: Sat Dec 3 05:49:33 2005 +0000 Update package version number for X11R7 RC3 release. commit c50cd5a7f5af7d45550f25f173b512eb89bf41c5 Author: Kevin E Martin Date: Fri Dec 2 02:16:08 2005 +0000 Remove extraneous AC_MSG_RESULT. commit da2c00c0e85a284e3fbd237f3a77212f9647da37 Author: Adam Jackson Date: Wed Nov 30 18:47:54 2005 +0000 Bump libdrm dep to 2.0. commit ca36f1da4e254c779b38df30fd655533e548a294 Author: Adam Jackson Date: Tue Nov 29 23:29:59 2005 +0000 Only build dlloader modules by default. commit 0b337390d6919e9a3d31ca654c8bc44b30756862 Author: Alan Coopersmith Date: Mon Nov 28 22:04:09 2005 +0000 Change *mandir targets to use new *_MAN_DIR variables set by xorg-macros.m4 update to fix bug #5167 (Linux prefers *.1x man pages in man1 subdir) commit 159bea027893ce5996a452a6f493372792e95122 Author: Eric Anholt Date: Mon Nov 21 10:49:14 2005 +0000 Add .cvsignores for drivers. commit e62939e30e664592085ce394680fd8e7f3d31339 Author: Kevin E Martin Date: Sat Nov 19 07:15:38 2005 +0000 Update pkgconfig files to separate library build-time dependencies from application build-time dependencies, and update package deps to work with separate build roots. commit 08f03e4237ce9a54f0dffa37307af402b0bd2e55 Author: Kevin E Martin Date: Wed Nov 9 21:31:20 2005 +0000 Bump version numbers for X11R6.9/X11R7 RC2. commit a600e89a1547908f2018de96b5ecb814e2427b34 Author: Kevin E Martin Date: Wed Nov 9 21:15:15 2005 +0000 Update package version number for X11R7 RC2 release. commit f73c23b4c588c2f5dbc3c36a2f43b8bbb1e0f36a Author: Kevin E Martin Date: Tue Nov 1 15:08:53 2005 +0000 Update pkgcheck depedencies to work with separate build roots. commit 25a63bfefda420f85b3cba89bdae4b35ef80c84f Author: Adam Jackson Date: Fri Oct 21 18:40:19 2005 +0000 Bug #4829: Remove some useless DRI symbol checks. commit 53c122fa162709f618be26d5940e3488c0cbbafa Author: Kevin E Martin Date: Wed Oct 19 02:48:03 2005 +0000 Update package version number for RC1 release. commit 4b97a544cac9fb507b4a2a7c8017607ea2cac1bb Author: Alan Coopersmith Date: Tue Oct 18 00:01:53 2005 +0000 Use @DRIVER_MAN_SUFFIX@ instead of $(DRIVER_MAN_SUFFIX) in macro substitutions to work better with BSD make commit c7b7ece6330561b3745f0e1136b57575eadec5ad Author: Adam Jackson Date: Mon Oct 17 22:57:33 2005 +0000 More 1.7 braindamage: define EXTRA_DIST in terms of @DRIVER_NAME@ instead of indirectly commit 0586dbe1c5d0432e8d8da2f3576ac2ccbe139919 Author: Alan Coopersmith Date: Mon Oct 17 00:09:03 2005 +0000 Use sed & cpp to substitute variables in driver man pages commit 1317f2d1f35241969dcf376abd106ffe3f918432 Author: Søren Sandmann Pedersen Date: Tue Oct 4 20:01:03 2005 +0000 Check in generated README files commit bd48e0531a16bdf27fc8549826cc1123ba12d5bd Author: Søren Sandmann Pedersen Date: Mon Oct 3 21:22:07 2005 +0000 - For all drivers that have a .sgml file, add code in their build system to build the README file at make dist time - in util/macros/xorg-macros.m4, add a new XORG_CHECK_LINUXDOC macro that will check if the required tools and files exist, and if so set a conditional. - util/modular/symlink.sh - Link all the .sgml to xf86-video-/README.sgml - Add all the README. to the list of excluded files - xc/programs/Xserver/hw/xfree86/doc/sgml/SiS.sgml: Various changes to make it spew less warnings when the text file is built. commit 6fda4e886c30669e346f7f30a38b93a77004a7f6 Author: Alan Hourihane Date: Mon Oct 3 09:39:17 2005 +0000 Bug #4353: Fix lockup when suspend and resume when playing video. Shutdown the overlay on LeaveVT. commit 154724cc4be1e951649c29e4abdbde964a8c112e Author: Alan Hourihane Date: Tue Sep 27 08:58:23 2005 +0000 Put back the print code to display the DDC information, and remove some bogus DDC messages. commit 2174848266940e1896970c12717e106a4742dcbc Author: Adam Jackson Date: Sat Sep 24 18:45:45 2005 +0000 Disable the {Open,Close}FullScreen DRI protocol. Remove empty FullScreen stubs from drivers, comment the non-empty ones. commit b6443515e9d485b1ab8ee266c5b21750d2a5a551 Author: Alan Coopersmith Date: Thu Sep 22 01:58:17 2005 +0000 Don't build xvmc module if DRI support is not present commit 76033c2c885524c2116ccf0bdd30c89fa2636150 Author: Eric Anholt Date: Thu Sep 15 06:04:59 2005 +0000 Bug #4458: Check for the xvmc proto package, since it's required to build these. (Joshua Baergen) commit 50f0bf51a67c89bf445a682f8fec34c52ff909a8 Author: Alan Hourihane Date: Thu Sep 8 14:51:40 2005 +0000 fix another possible crash commit 521e357c1846960b48e83fb97f77025f6f91e638 Author: Alan Hourihane Date: Thu Sep 8 09:31:04 2005 +0000 Fix a crash situation commit e8c46c897564f038744b9274949482568d654ecc Author: Alan Hourihane Date: Tue Sep 6 10:31:33 2005 +0000 Fix DirectColor visual colormap issues at 16bpp in the i830 driver. commit bf6fef41935ebf624cc1ae6faa084dfa0a7eee72 Author: Alan Hourihane Date: Tue Sep 6 10:23:57 2005 +0000 Sync the accelerator before calling AdjustFrame to avoid lockups. commit 5fc5f02d76e077b09bf3ab1c7deeb1263033d57a Author: Dave Airlie Date: Mon Aug 29 12:06:13 2005 +0000 Add i810 XvMC support, it passes distcheck for me. commit 86e6935bd50b129519b24ee3d694a3f75b8d4f1c Author: Dave Airlie Date: Mon Aug 29 12:02:29 2005 +0000 Fix include paths for i810 XvMC add config.h to driDrawable.c commit 6f1406753da102ccce1898b3f160d6eaaccc9b7c Author: Kristian Høgsberg Date: Fri Aug 26 15:32:40 2005 +0000 Added i830_io.c, which just includes i810_io.c with BUILD_FOR_I830 set. commit 219beaca91b88dabfe05b8e756d558269f7a5148 Author: Daniel Stone Date: Sun Aug 21 06:26:33 2005 +0000 Fix typo in debug ErrorF. commit fdc6997f2edf4dfdcfd877152028bff2a0306999 Author: Daniel Stone Date: Thu Aug 18 09:03:47 2005 +0000 Update autogen.sh to one that does objdir != srcdir. commit 461d43be78f49e6cb965b7a206c77f6fe1622601 Author: Alan Hourihane Date: Mon Aug 15 07:30:05 2005 +0000 Egbert's 64bit fixes for mixed 32/64bit clients commit 95bd8dbee69b4b35e4c37e19d15e0449c38726b6 Author: Alan Hourihane Date: Fri Aug 12 21:23:59 2005 +0000 add missing i830_shadow.c commit 7476d0413f161ddb3f3cc4a77c3f2a2912d17e60 Author: Alan Hourihane Date: Mon Aug 8 09:17:06 2005 +0000 Enable gamma correction bit even for 8bit modes. The pipeconf register handles the real enablement. commit a89f89a2423aecb199d5f4af2e9430e180aa1131 Author: Ian Romanick Date: Fri Aug 5 23:04:33 2005 +0000 I have resolved this by making all the "Direct rendering disabled" messages X_WARNING and all the "Direct rendering enabled" messages X_INFO. Ideally, we need to factor some of this stuff out to common code. There is a *LOT* of duplication among the *_dri.c files under programs/Xserver/hw/xfree86/drivers. I'm trying to clean up as much low hanging fruit as I can and got tired of waiting for ajax. :P commit fd6d7d7f7ea03961c5ecf7db029f12ece93d414a Author: Alan Hourihane Date: Tue Aug 2 16:22:42 2005 +0000 Adjust newPipe code for BIOS's starting at 1200 commit da7b85176fc061995e1b04fa22dc736b6c9353e6 Author: Adam Jackson Date: Sun Jul 31 17:19:29 2005 +0000 Post-merge fixups: - Disable building the gamma driver - Fix numerous Imakefiles to get header paths and subdirs right - Symlink client-side GLX lib source from Mesa - Add new shader subdirs to the Mesa core build - Tweak the DDX drivers to not ask for "xf86dri.h"; they don't need it, since they don't generate protocol, and it was breaking the build. commit 1561ec3ca4f0d774f2664e9354d537061a0102d1 Author: Kevin E Martin Date: Fri Jul 29 21:22:43 2005 +0000 Various changes preparing packages for RC0: - Verify and update package version numbers as needed - Implement versioning scheme - Change bug address to point to bugzilla bug entry form - Disable loadable i18n in libX11 by default (use --enable-loadable-i18n to reenable it) - Fix makedepend to use pkgconfig and pass distcheck - Update build script to build macros first - Update modular Xorg version commit d165b7dc4ec3793e921363cdeb1ad41c0a3653d6 Author: Søren Sandmann Pedersen Date: Wed Jul 27 21:23:29 2005 +0000 Add xf86driproto to dependencies for drivers with DRI support. Patch from Donnie Berkholz commit 0cd371c6e57f899c34c4d2e2a8531ef7ce3967a4 Author: Alan Hourihane Date: Tue Jul 26 21:55:42 2005 +0000 Bump the DDX version to 1.4.0 to indicate the removal of drmAddress. Thus allowing 32/64bit clients to cohabit. commit adef005c2f96395d396834f2173125f0c06d40cb Author: Alan Hourihane Date: Tue Jul 26 08:07:37 2005 +0000 Bug #3850: Fix non-DRI builds. commit 09e871b1b4a67cfe818613bc012d72965b449f1c Author: Søren Sandmann Pedersen Date: Tue Jul 26 00:41:42 2005 +0000 Build system for i810 commit ac920dae7c22ed3956e4f31a0d3baf1909762b76 Author: Adam Jackson Date: Mon Jul 11 02:29:51 2005 +0000 Prep for modular builds by adding guarded #include "config.h" everywhere. commit f90f661652e5fe0986eb804f8dff3eeffe0af820 Author: Alan Hourihane Date: Sat Jul 9 21:19:11 2005 +0000 Copy mode strings upon mode copy commit 8fced5b648b0666ca88221e37bc1131ab5c2072d Author: Alan Hourihane Date: Tue Jul 5 09:25:03 2005 +0000 Fix Xvideo when operating in one line mode and video is pushed off the top of the screen. Also some fixes for Xvideo skew in certain resolutions. commit 68bc259fa1082f7780e524fc703e926a27fc7f9a Author: Alan Hourihane Date: Mon Jul 4 20:45:39 2005 +0000 NULL check commit 83c3e3ee49a0aa3ceaf52e2408cc253fc5dc8af4 Author: Alan Hourihane Date: Tue Jun 28 16:13:02 2005 +0000 Add support to check the in-built VESA mode list. commit 0d64e9e1f5494cff633a1499545d55a0edd0d92e Author: Adam Jackson Date: Sat Jun 25 21:16:55 2005 +0000 Bug #3626: _X_EXPORT tags for video and input drivers. commit 67174d67e740b9082ed0a65a79f08b5df3b4fcfb Author: Alan Hourihane Date: Mon Jun 13 12:15:25 2005 +0000 commit missing piece from resume patch commit ddab0558d584fb906cc0dfb383cc0e7a122d5336 Author: Alan Hourihane Date: Thu Jun 9 09:26:03 2005 +0000 Bug #3496 Remove a bogus check for DRM 1.2 for resume functionality as it's been present since 1.1 commit 7f49513afe9d4490bdfdac182d50383c089875e0 Author: Alan Hourihane Date: Wed Jun 8 17:50:32 2005 +0000 Fix the mode validation such that using function 5f05 to set the refresh rate works again. commit f8de9b4b9a5ce2756652c6b9ec1e84aedbc3519a Author: Alan Hourihane Date: Tue Jun 7 18:54:10 2005 +0000 Fix text restoration when display switching has been activated and we could be potentially on a different display than the initial one. commit 1f538007e9da7d4ad30ec4d4d281fb07ef300ddb Author: Alan Hourihane Date: Mon Jun 6 09:47:29 2005 +0000 missed a hunk commit d8a32bf4274ad91cc2b9ca425f5b0a017c2dd293 Author: Alan Hourihane Date: Mon Jun 6 09:40:23 2005 +0000 Bug #3436 Fix some offset, pitch and overlay scaler size problems with the video overlay. Solves various spontaneous lockups. commit 005d121d280ff4502f4d6e3d3d585ff0175367bb Author: Alan Hourihane Date: Tue May 31 20:29:01 2005 +0000 and another. sheesh. commit 27f86d693c32e550d9ebf4d6ff05c2d827754c8b Author: Alan Hourihane Date: Tue May 31 20:27:24 2005 +0000 missed hunk commit 27feb0f3474c1e21557225e10b13558e002c0ebf Author: Alan Hourihane Date: Tue May 31 17:58:26 2005 +0000 Add Intel(R) 945G support commit 2d10e8dc8184b0514e8ac9f835a06b85d609bb1b Author: Alan Hourihane Date: Wed May 25 09:10:44 2005 +0000 Change the sequence on how refresh rates are set in the i830 driver. This works much better with custom modelines. commit 89ec4ac201c09329d5d732323bdce6250b8592fe Author: Alan Hourihane Date: Fri May 20 15:24:02 2005 +0000 Bugzilla #3292: https://bugs.freedesktop.org/show_bug.cgi?id=3292 Add shadowFB support to the Intel driver (Dima Dorfman) commit df0964ea8f92c6b278dd87b95380dbc3ada625e5 Author: Alan Hourihane Date: Fri May 20 13:24:02 2005 +0000 Bugzilla #2339: https://bugs.freedesktop.org/show_bug.cgi?id=2339 Bugzilla #1064: https://bugs.freedesktop.org/show_bug.cgi?id=1064 Bugzilla #3055: https://bugs.freedesktop.org/show_bug.cgi?id=3055 Much better display switching support for mobile chipsets using the Fn+F? combination keys. Add DirectColor support. Add gamma correction support. commit b99ef9cd59fce0e02e8829210c7cb8aadb6a3499 Author: Alan Hourihane Date: Fri May 20 11:56:45 2005 +0000 Bugzilla #2585: https://bugs.freedesktop.org/show_bug.cgi?id=2585 Support the Intel E7221 server chipset. commit 3b87fac327b5549678f91a9c961b9fe5c0bf8a37 Author: Adam Jackson Date: Wed May 18 17:47:35 2005 +0000 Bug #3163: Create new DRIINFO_*_VERSION macros to indicate the version number of the DRIInfoRec, disambiguating it from the XF86DRI protocol version number. Modify DRIQueryVersion to return the libdri version number, which all DDXes were requesting implicitly. Fix the DDXes to check for the DRIINFO version they were compiled against. commit 72100ac7cff7485db3e869f1331caddb0bc21a24 Author: Daniel Stone Date: Wed Apr 20 12:25:28 2005 +0000 Fix includes right throughout the Xserver tree: * change "foo.h" to for core headers, e.g. X.h, Xpoll.h; * change "foo.h", "extensions/foo.h" and "X11/foo.h" to for extension headers, e.g. Xv.h; * change "foo.[ch]" to for Xtrans files. commit c813bf655908cd72049f346919abbdff8365b944 Author: Alan Hourihane Date: Thu Apr 14 22:55:01 2005 +0000 Fix initial DDC parameters when validating modes. commit e637630908f40f42367c94344d44918b6cdb0eb2 Author: Alan Hourihane Date: Fri Apr 1 14:17:24 2005 +0000 Check the VESA BIOS scanline requirements and adjust the displayWidth accordingly. Fixes an assumption that the screen width was always used as the displayWidth. commit 5090c8ba6778091aa4ce24b16b058d534ba78dc7 Author: Alan Hourihane Date: Mon Mar 21 21:20:39 2005 +0000 a one line fixup to the UseDDC code commit 8bc9b0d7da5de9a3294687f450afb50452898c7d Author: Alan Hourihane Date: Mon Mar 21 18:57:39 2005 +0000 Fix some bad alignment problems when allocating memory from the end of AGP space. Bugzilla #2764 (https://bugs.freedesktop.org/show_bug.cgi?id=2764) commit 50a105e88cfdfc22fef8cfd2397fc5e9fae17663 Author: Alan Hourihane Date: Tue Mar 15 13:10:23 2005 +0000 LFP's may exist on Pipe A in older i830's. Check that case so that Panel ID can be obtained and Xvideo is correctly setup for panel fitting. commit eb8b5df739e7be4509423c2bf757a7df35ce02f6 Author: Alan Hourihane Date: Tue Mar 15 11:32:10 2005 +0000 Enforce DDC monitor ranges usage if we have them and reject bad modes. Use NoDDC option to override DDC timings. We currently only use DS_RANGES, but could use other DDC information, as does the common layer, to deduce the h/v ranges. commit 49dcbacf7fa7f29d18c18eec7e2f99e859bf4c7e Author: Alan Hourihane Date: Thu Mar 3 18:01:51 2005 +0000 Limit the maximum refresh rate to 85Hz and remove the bogus 100Hz and 120Hz settings as they don't match close enough and are really unsupported. commit 7eaf88d00d7539fde68422ae6566f23993aa2633 Author: Alan Hourihane Date: Tue Mar 1 19:53:28 2005 +0000 Fix an off by one error when scaling which caused occasional overlay color lines at the far right and bottom of the window. commit 28d65bceaec3ef636f0f5bdd03b38fd76f87c8c6 Author: Alan Hourihane Date: Thu Feb 10 13:16:08 2005 +0000 Fix unaligned YV12 videos that caused skew when playing. commit 6352e65b9cd635e9f266e0e9ed64c5beae9e5c44 Author: Alan Hourihane Date: Thu Feb 10 10:41:20 2005 +0000 Fix pixmap cache corruption when Xv was in use due to a memory allocator calculation problem. commit f691f959213d3eb5c365edaf5a5a53fa96300491 Author: Alan Hourihane Date: Tue Feb 8 20:13:33 2005 +0000 Setup the Xv atoms correctly. commit 31a4191a97f538b2c6d4b8036796f06443f783c1 Author: Egbert Eich Date: Mon Feb 7 11:56:31 2005 +0000 Check the pixel clock choosen for a specific refresh rate against the maximally allowed pixel clock when choosing mode lines for VBESetVBEMode() and VESA BIOS version >= 3.0 (Bugzilla #2486). commit a103496e08f699b609be875827ecf1f3afa45e68 Author: Alan Hourihane Date: Tue Feb 1 19:47:32 2005 +0000 Check return value of i830InitDma() and fail to initialize the DRI if that fails. Also, check the DRM module name that it matches i915 and not the older i830 driver which some people were still trying to use and was causing Dma Initialization failures and lockups. commit 7ff7cec9220815eb5f74cfd95860aaa5f1f263bd Author: Egbert Eich Date: Mon Jan 24 15:02:17 2005 +0000 Unmap XCursor AGP memory on VT switch to prevent system from lockup (Matthias Hopf). commit d4c87adf6792ae9af641b9824c547d18cdd1e521 Author: Alan Hourihane Date: Mon Jan 10 11:29:02 2005 +0000 Revert some changes to TweakMemorySize commit 076c5116ae5eda1822da70184cbcc8942b470541 Author: Alan Hourihane Date: Thu Jan 6 14:43:14 2005 +0000 Add: - Intel i915GM support to 2D DDX and 3D Mesa drivers. - PanelID identification - DRI suspend/resume support - Detection of monitor changes on VT switches - Support custom video modes if available in the Video BIOS - 3D enhancements: * GL_ARB_texture_cube_map * GL_EXT_blend_equation_separate * GL_ATI_blend_equation_separate * GL_ARB_point_parameters * GL_NV_blend_square * GL_EXT_cull_vertex * GL_ARB_depth_texture * GL_SGIX_depth_texture * GL_ARB_shadow * GL_EXT_shadow_funcs * GL_3DFX_texture_compression_FXT1 By Tungsten Graphics, Keith Whitwell & Alan Hourihane. commit edb9b6fde0948bb1121a8456500ef9f55f734376 Author: Adam Jackson Date: Mon Jan 3 16:03:56 2005 +0000 Bug #1488: Fix crash on i830 during startup. (Eric Anholt) commit 59cecb714b78f7f384c24bf159bd39c81e9c7ee1 Author: Markus Kuhn Date: Sat Dec 4 00:43:09 2004 +0000 Encoding of numerous files changed to UTF-8 commit c24b0e04f3a1b064412e0bea44c4af546ec9fb77 Author: Egbert Eich Date: Tue Nov 2 08:54:53 2004 +0000 Fixed KGA handling for i810. KGA handling for chips derived from C&T chips is slightly different. The changes make the code consistent with the C&T (chips) and i740 drivers. commit 7ce4898ee319d149cf86e938e7257ea7490d327f Author: Egbert Eich Date: Wed Oct 27 09:52:15 2004 +0000 Removing vtSema shilding of accel sync - should have been '&&' instead of '||' anyway. commit 88714c3b2274fcfe50837631b1d38a7cd96f692a Author: Egbert Eich Date: Tue Oct 12 19:13:43 2004 +0000 Change order of SetDisplayDevice(), HWRestore(), UnbindGART() and RestoreBIOSMemSize() to be exactly opposite to the Save procedure in EnterVT() (Matthias Hopf, Alan Hourihane). Fixed server crash on reset when a structure allocated in PreInit() was freed on CloseScreen(). Fixed ring buffer lock ups that happened because the structure that contained ringbuffer data was not zeroed after allocation. Fixed numerous warnings due to signed unsigned comparisons. commit 687298daed40aaade44e61342a9abe984088541b Author: Egbert Eich Date: Mon Oct 11 09:58:04 2004 +0000 Checking if server isn't switched away before calling sync. Sanity check for possible bugs in aother areas of the code. Fixing default amount of of allocated video memory from AGP for i810: Use 16MB if less than 192MB are installed else use 24MB (Matthias Hopf). commit d2866271b8d42a0d44b65e30667adab96ac8fd54 Author: Adam Jackson Date: Wed Sep 22 04:38:03 2004 +0000 Bug #1252, #1253, #1255, #1256: Various typo fixes from Dave Jones. commit 88fd270d23ed1a1761cfdb98367c89ca89735b12 Author: Adam Jackson Date: Fri Sep 17 03:04:51 2004 +0000 Bug #1192: Remove cfb support from drivers where its use is an option. Delete xf24_32bpp, as s3virge was the last user. Fix up some comments to refer to fb rather than cfb. commit 9e8e19c6fde15fe5b9bb7532316dedeeca54b8d9 Author: Kevin E Martin Date: Mon Aug 30 03:19:08 2004 +0000 Fix ARGB cursor problem on i810 (Bug #1195, Alan Hourihane). commit 15c61546ea1a64d6c7dd60764164143ea7174123 Author: Kevin E Martin Date: Thu Aug 26 05:10:56 2004 +0000 Fix i810 compilation without Xinerama (Bug #1180, Simon Schubert). commit 828c0fb8050dcbed8026365cbcb17d4d4330c268 Author: Kevin E Martin Date: Wed Aug 25 00:30:41 2004 +0000 Fix i810 driver startup on Sony VAIO laptop (Bug #1084, Dave Airlie). commit c4626af83d760c2a6aae90ad478430eac1c4f4ea Author: Egbert Eich Date: Fri Jul 30 21:36:40 2004 +0000 Do I810EmitFlush() during screen init to make sure 2D engine is in a sane state (Peter Kunzmann). commit e0e01b7d6188a83c7539256ac1710ace04ce132a Author: Egbert Eich Date: Fri Jul 30 20:38:27 2004 +0000 add i845 to the list of chips that allow memory size tweaking. commit 199620cdc924e6aa98683e305843ea91c7e1771d Author: Adam Jackson Date: Fri Jul 30 20:30:52 2004 +0000 Bug #400 (partial): Driver fixes for the dlloader. When using dlloader, all framebuffer formats except cfb and the overlay modes should work, and r128 and radeon need to be loaded from the ati driver (both issues to be fixed soon). Tested on i740, s3virge, mach64, tdfx, vesa, and vga drivers. elfloader users shouldn't be affected. commit 54e644ffbb63290b7e2ec51b041f5b2df26e164b Author: Egbert Eich Date: Fri Jul 30 19:04:14 2004 +0000 Removed bugs on TweakMemorySize() which prevented it from working at all. Initialized last element of BIOS version number string to 0 to avoid random problems. commit 3a3b00f18ac5edd7dfef06a96342c95f3cb97f79 Author: Kevin E Martin Date: Fri Jul 30 04:44:13 2004 +0000 Fix "DoLoadableServer NO" build commit 986bc24892e7ce3784fc723fda54b1f89dd75b66 Author: Egbert Eich Date: Tue Jul 6 14:37:47 2004 +0000 Increase default video RAM size to 16MB when DRI is enabled and more than 128MB are available (Egbert Eich). Fixed lockups during mode switch. Problem was introduced when attempting to copy the behavior during LeaveVT()/EnterVT() but but forgetting to call I810DRILeave() before I810DRIEnter(). The entire DRILeave()/Enter() scenario has been commented out as it didn't seem to be necessary (Egbert Eich). Fix TweakMemorySize() (tested with i855/i865) (Egbert Eich). commit b7310758e60956a7de14dcf968bb713dfc56738f Author: Alan Hourihane Date: Tue Jun 22 15:12:42 2004 +0000 add CmdBuffer commit 5ed7e6f8424236dd57d83f6ab1282693260969cb Author: Alan Hourihane Date: Tue Jun 22 14:28:46 2004 +0000 i915 support Dual Head support for i830, i855, i915. ARGB cursor support (including i810) for all chipsets. commit a46e95722d3c2b9dfb9eacf9a390a7c47d427773 Author: Eric Anholt Date: Wed Jun 16 19:16:54 2004 +0000 Mop-up from the DRI 20040613 import, taking care of some drm.h header location and type changes, and update the ChangeLog. commit 35a5cf103e9e41d178f980f685307f8c8063a006 Author: Eric Anholt Date: Wed Jun 16 09:44:00 2004 +0000 Merge DRI-trunk-20040613 changes in programs/Xserver/hw/xfree86/drivers, with the following notes: - Savage and Mach64 (= ati/ati*.[ch] changes) DRI not merged due to insecurity. - VIA driver converted to new drmContext and drmHandle names. - Radeon driver merge conflicted in many places, and MergedFB at least could probably use some checking at this point. commit bec47f9369b6f02ce4537ab1ac547ba933609edc Author: Eric Anholt Date: Wed Jun 16 09:26:18 2004 +0000 DRI trunk-20040613 import commit c5b1972b50a6dbe57c67d3cc88ae010386b908a6 Author: Eric Anholt Date: Wed Jun 16 09:23:18 2004 +0000 DRI XFree86-4_3_99_12-merge import commit 41eba03d0d62fe03a7b437c9c4914016c3125141 Author: Egbert Eich Date: Fri Apr 23 19:35:55 2004 +0000 Merging XORG-CURRENT into trunk commit e27494f2a9262b6310c2f1e0a40c406f578fc693 Author: Egbert Eich Date: Sun Mar 14 08:33:22 2004 +0000 Importing vendor version xf86-4_4_99_1 on Sun Mar 14 00:26:39 PST 2004 commit 5ece535dc7f77185cc8c327a57ae3c02c8b9f918 Author: Egbert Eich Date: Wed Mar 3 12:12:19 2004 +0000 Importing vendor version xf86-4_4_0 on Wed Mar 3 04:09:24 PST 2004 commit ba904078dd22e013f4112479e7c2669c33e708a8 Author: Egbert Eich Date: Thu Feb 26 13:35:53 2004 +0000 readding XFree86's cvs IDs commit 13ea363db30dc0010cd09df29df90074db77f646 Author: Egbert Eich Date: Thu Feb 26 09:23:19 2004 +0000 Importing vendor version xf86-4_3_99_903 on Wed Feb 26 01:21:00 PST 2004 commit 64b129e800d915f0d82a67ce0f8b1850e67012fc Author: Egbert Eich Date: Thu Jan 29 08:08:35 2004 +0000 Importing vendor version xf86-012804-2330 on Thu Jan 29 00:06:33 PST 2004 commit 64d7bbf0ac060a57761aa2ecf21ddbe8169674c0 Author: Kaleb Keithley Date: Fri Dec 19 20:55:10 2003 +0000 XFree86 4.3.99.902 (RC 2) commit 418eaaee109acd3ea178b147dd29f90318d159f9 Author: Kaleb Keithley Date: Thu Dec 4 22:03:17 2003 +0000 XFree86 4.3.99.901 (RC 1) commit 72063c52b82faf80a0183f66d2fd5789def76d8f Author: Kaleb Keithley Date: Tue Nov 25 19:28:37 2003 +0000 XFree86 4.3.99.16 Bring the tree up to date for the Cygwin folks commit e04a1a8f862eb63b546fb74e73ee602fefb82e00 Author: Kaleb Keithley Date: Fri Nov 14 16:48:55 2003 +0000 XFree86 4.3.0.1 commit 3ad0d9a73bc0a4f3edb858d5f3a7c36827b7a429 Author: Kaleb Keithley Date: Fri Nov 14 16:48:55 2003 +0000 Initial revision