Enable VaapiDrmPicture on more cases |
|||
Issue descriptionCurrently the existing VaapiTFPPicture(GLX) works fine on linux desktop (non-ozone) but it does not work when using EGL. The solution is to enable VaapiDrmPicture on non-ozone and use it when the user has passed the cmd line argument --use-gl=egl. In order to enable this we need to use the extension EGL_MESA_image_dma_buf_export. The sequence will be: 1- vaGetDisplayDRM 2- Export output gl texture as dmabuf (through EGLImage) 3- Import dmabuf into output VA surfaces. Existing sequence for Ozone is: 1- vaGetDisplayDRM 2- Create Dumb buffer (ioctl) and export it as dmabuf (gbm_bo_get_fd). 3- Import dmabuf into output gl texture (through EGLImage). 4- Import dmabuf into output VA surfaces. As you can see only step 2) from the first sequence above is missing in current Chromium. Everything else is already in place. In order to prove that it can work I went ahead and I implemented it. See step 2) here https://chromium-review.googlesource.com/c/chromium/src/+/766787/8/media/gpu/vaapi/vaapi_drm_picture.cc#125 To sum up, vaapi is already working on non-ozone in current Chromium but it requires GLX (VaapiTFPPicture(GLX)), i.e. --use-gl=desktop. I am suggesting to extend the vaapi support to --use-gl=egl.
,
Nov 22 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/a9dd08addbd53885176f45a205ea992ee815961c commit a9dd08addbd53885176f45a205ea992ee815961c Author: Julien Isorce <julien.isorce@chromium.org> Date: Wed Nov 22 20:22:40 2017 Allow DRM and TFP pictures support to be compiled in together No functional change. This is useful to select DRM(egl) or TFP(glx) at runtime, i.e. with --use-gl=egl or --use-gl=desktop. This is a preliminary step for the subsequent CL https://chromium-review.googlesource.com/c/chromium/src/+/766787 Bug: 785201 Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Change-Id: I6963187b88c3c69da8584e2cc7d14d709ac2fc17 Reviewed-on: https://chromium-review.googlesource.com/785690 Reviewed-by: Dan Sanders <sandersd@chromium.org> Commit-Queue: Julien Isorce <julien.isorce@chromium.org> Cr-Commit-Position: refs/heads/master@{#518726} [modify] https://crrev.com/a9dd08addbd53885176f45a205ea992ee815961c/media/filters/BUILD.gn [modify] https://crrev.com/a9dd08addbd53885176f45a205ea992ee815961c/media/gpu/BUILD.gn [modify] https://crrev.com/a9dd08addbd53885176f45a205ea992ee815961c/media/gpu/vaapi/vaapi_drm_picture.cc [modify] https://crrev.com/a9dd08addbd53885176f45a205ea992ee815961c/media/gpu/vaapi/vaapi_picture.cc [modify] https://crrev.com/a9dd08addbd53885176f45a205ea992ee815961c/media/gpu/vaapi/vaapi_tfp_picture.cc [modify] https://crrev.com/a9dd08addbd53885176f45a205ea992ee815961c/media/gpu/vaapi_video_decode_accelerator.cc [modify] https://crrev.com/a9dd08addbd53885176f45a205ea992ee815961c/media/gpu/vaapi_wrapper.cc [modify] https://crrev.com/a9dd08addbd53885176f45a205ea992ee815961c/media/gpu/vaapi_wrapper.h
,
Nov 24 2017
Do we have any bot combination of Linux + --use-gl=egl ?
,
Nov 28 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/65e985d72d99fad9c5aae48ef7759c03a441f770 commit 65e985d72d99fad9c5aae48ef7759c03a441f770 Author: Julien Isorce <julien.isorce@chromium.org> Date: Tue Nov 28 10:15:21 2017 Remove unnecessary ozone guard around libva-drm.so This is non-ozone specific and this will help maintenance. Also this is useful to select DRM(egl) or TFP(glx) at runtime, i.e. with --use-gl=egl or --use-gl=desktop. Note that a switch(gl::GetGLImplementation()) is not practicable here because of the way generate_stubs.py works. This is another preliminary step for the subsequent CL https://chromium-review.googlesource.com/c/chromium/src/+/766787 Bug: 785201 Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Change-Id: Icc4f2c6a0ff448ce1507798bfbc708d2fe16ff29 Reviewed-on: https://chromium-review.googlesource.com/787432 Reviewed-by: Dan Sanders <sandersd@chromium.org> Commit-Queue: Julien Isorce <julien.isorce@chromium.org> Cr-Commit-Position: refs/heads/master@{#519636} [modify] https://crrev.com/65e985d72d99fad9c5aae48ef7759c03a441f770/media/gpu/vaapi_wrapper.cc
,
Nov 28 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/8580c2220857886dcfa8fea7374bcc559a291416 commit 8580c2220857886dcfa8fea7374bcc559a291416 Author: Julien Isorce <julien.isorce@chromium.org> Date: Tue Nov 28 12:11:32 2017 Can select DRM(egl) or TFP(glx) at runtime for vaapi video decoding Fixes overlap with https://chromium-review.googlesource.com/767375 Added unit test this time. This is a preliminary step for the subsequent CL https://chromium-review.googlesource.com/c/chromium/src/+/766787 Bug: 785201 , 788123 Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Change-Id: I70620697109fdef7d036be4c26fcecb763ddb4ba Tests: media_unittests --gtest_filter=*Vaapi*SupportedPlatforms* Reviewed-on: https://chromium-review.googlesource.com/788051 Reviewed-by: Dan Sanders <sandersd@chromium.org> Reviewed-by: Miguel Casas <mcasas@chromium.org> Commit-Queue: Julien Isorce <julien.isorce@chromium.org> Cr-Commit-Position: refs/heads/master@{#519657} [modify] https://crrev.com/8580c2220857886dcfa8fea7374bcc559a291416/media/gpu/vaapi/vaapi_picture.cc [modify] https://crrev.com/8580c2220857886dcfa8fea7374bcc559a291416/media/gpu/vaapi/vaapi_picture_factory.cc [modify] https://crrev.com/8580c2220857886dcfa8fea7374bcc559a291416/media/gpu/vaapi/vaapi_picture_factory.h [modify] https://crrev.com/8580c2220857886dcfa8fea7374bcc559a291416/media/gpu/vaapi_video_decode_accelerator_unittest.cc
,
Nov 29 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/73d609f366ba6a86324048bbad81527f76d237b5 commit 73d609f366ba6a86324048bbad81527f76d237b5 Author: Julien Isorce <julien.isorce@chromium.org> Date: Wed Nov 29 09:34:41 2017 Add RGBX to gfx va converters When decoding into the gl texture RGB, vaapi needs to output RGBX. Also move format constants to VaapiPictureFactory. This is another preliminary step for the subsequent CL https://chromium-review.googlesource.com/c/chromium/src/+/766787 Bug: 785201 Tests: media_unittests --gtest_filter=*Vaapi*PictureBufferFormat* Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Change-Id: Ibd1b0f249f104fa40608b951a14cb76da4a5974d Reviewed-on: https://chromium-review.googlesource.com/787290 Commit-Queue: Julien Isorce <julien.isorce@chromium.org> Reviewed-by: Dan Sanders <sandersd@chromium.org> Cr-Commit-Position: refs/heads/master@{#520051} [modify] https://crrev.com/73d609f366ba6a86324048bbad81527f76d237b5/media/gpu/format_utils.cc [modify] https://crrev.com/73d609f366ba6a86324048bbad81527f76d237b5/media/gpu/vaapi/vaapi_drm_picture.cc [modify] https://crrev.com/73d609f366ba6a86324048bbad81527f76d237b5/media/gpu/vaapi/vaapi_picture_factory.cc [modify] https://crrev.com/73d609f366ba6a86324048bbad81527f76d237b5/media/gpu/vaapi/vaapi_picture_factory.h [modify] https://crrev.com/73d609f366ba6a86324048bbad81527f76d237b5/media/gpu/vaapi_video_decode_accelerator.cc [modify] https://crrev.com/73d609f366ba6a86324048bbad81527f76d237b5/media/gpu/vaapi_video_decode_accelerator_unittest.cc [modify] https://crrev.com/73d609f366ba6a86324048bbad81527f76d237b5/media/gpu/vaapi_wrapper.cc
,
Nov 30 2017
I'd like to follow up on the comments of https://crrev.com/c/766787, concretely [1] and [2]. Chromium is a large project and has plenty of downstream sub projects and ongoing refactorings and sub projects but, in general, we don't host code that is not shipped _and_ has no plans to be shipped in any near future. With that, let's follow the facts: 1. IIUC, hardware accelerated video decoding (and encoding, for the case) is disabled by default in Linux (except CrOs), has been so for a long time, and will not be enabled in the foreseeable future, see: https://crbug.com/137247 . 2. IIUC, VaapiTfpPicture is only used in Linux non-CrOs platforms, for Vaapi acceleration purposes, correct? If Chrome never uses Hw decode acceleration in Linux, it follows from 1 and 2 that we never use any vaapi_* files in Linux in "production", i.e. it's not shipped. (Although we compile those files in some bots because, well, they the files are there now). Something else is that, perhaps, some downstream Cr embedders might be using this functionality? Or some users might enable it via flags? This is a circumstance that: a) Chromium doesn't support actively b) Chromium tries not to break unexpectedly, but see a) Take into account that the majority of users don't use Chrome command line flags beyond whatever is hardcoded by us in chrome://version/. An example of how things are done to support a downstream project without having to carry the code ourselves is Geolocation [3, 4]. Summing up, either we (plan to) support the Vaapi Linux code path 100% or we don't support it at all. (When I say "Linux" I mean non-CrOs Linux). [1] https://chromium-review.googlesource.com/c/chromium/src/+/766787#message-67edc5ee8cf34cdc4f9d559317daafec3a3bdec0 [2] https://chromium-review.googlesource.com/c/chromium/src/+/766787#message-ba65d17348c54009ac6426571b830c7d1c495192 [3] https://cs.chromium.org/chromium/src/content/public/browser/content_browser_client.h?type=cs&q=OverrideSystemLocationProvider&l=499 [4] https://codereview.chromium.org/2126893003/diff/120001/content/public/browser/geolocation_delegate.h
,
Dec 1 2017
Good points. There has to be some intermediate steps between "not supporting Vaapi on Linux at all" and "supporting the Vaapi Linux code path 100%" though. Ultimately, support for Vaapi on Linux *is* desirable. 4k (and more) decoding on the CPU is still painful, and libva provides a good interface to improve that. IIUC the main reason why it is not enabled on Linux is because we still need external patches and bug fixes on Chromium OS to make it work, and the libva ABI used to unexpectedly change between stable releases. These two points make libva on Linux more of a liability than anything. On Chromium OS, we have full control over the sysroot and compiled packages, so we can mitigate these issues. And the fact is that libva works pretty well there and is heavily tested with Intel GPUs. So provided that libva improves on the two fronts I mentioned above, support on Linux desktop should not be too painful. It just happens that the libva 2.0 series *should* remain completely ABI-stable. We still need to integrate a few external patches. However, I can see a near future where, while not supporting this use case and requiring explicit activation, libva could be enabled on Linux desktop builds. If the experiment goes well, we could even switch to full support. There is a huge intersection with the Chrome OS use case, so I am willing to help sandersd@ on this if he is ok with that plan. Looking at the latest comments on https://chromium-review.googlesource.com/c/chromium/src/+/766787, it seems like our stance is becoming "we don't support that, but we won't hinder and are willing to see how it goes". Is that fair enough?
,
Dec 5 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/51357dc19efbf30328ca05655fbf69886f6e9113 commit 51357dc19efbf30328ca05655fbf69886f6e9113 Author: Julien Isorce <julien.isorce@chromium.org> Date: Tue Dec 05 23:39:45 2017 Allow RGBX for VaapiTFPPicture Fixes regression on non-ozone platforms and introduced by 73d609f366ba6a86324048bbad81527f76d237b5 https://chromium-review.googlesource.com/787290 Note that the format is only used for sanity check. All the logic is done automatically in the vaapi driver's implementation of vaPutSurface. Bug: 785201 Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Change-Id: Idc0bdf184874bf2c238e28da6f42f4e3572e9743 Reviewed-on: https://chromium-review.googlesource.com/807928 Reviewed-by: Dan Sanders <sandersd@chromium.org> Commit-Queue: Julien Isorce <julien.isorce@chromium.org> Cr-Commit-Position: refs/heads/master@{#521897} [modify] https://crrev.com/51357dc19efbf30328ca05655fbf69886f6e9113/media/gpu/vaapi/vaapi_tfp_picture.cc
,
Dec 6 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/4c259455066d41ac777613d0d44506c04d13dd57 commit 4c259455066d41ac777613d0d44506c04d13dd57 Author: Julien Isorce <julien.isorce@chromium.org> Date: Wed Dec 06 09:36:02 2017 Can export a gl texture as dmabuf First create an EGLImage from a given gl texture. Then export the EGLImage as dmabuf. In addition to the new unit test included in this CL, this will be useful for the subsequent CL, https://chromium-review.googlesource.com/c/chromium/src/+/766787 which will enable existing VaapiDrmPicture when passing --use-gl=egl. While existing VaapiTFPPicture(X11) will be left for --use-gl=desktop. Added unit test ui/gl/gl_image_native_pixmap_unittest.cc The test is enabled on non-ozone but should work on ozone too. If any of the pre-conditions fail then the test is skipped. Specifically it requires the extension EGL_MESA_image_dma_buf_export. So it works on Intel, AMDGPU and nouveau drivers but skipped on NVIDIA. Bug: 785201 Test: gl_unittests --gtest_filter=*NativePixmap* Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Change-Id: Id9b4e56aaaaf5fd9994f4b144ce74a234d848253 Reviewed-on: https://chromium-review.googlesource.com/766687 Commit-Queue: Julien Isorce <julien.isorce@chromium.org> Reviewed-by: Kenneth Russell <kbr@chromium.org> Cr-Commit-Position: refs/heads/master@{#522052} [modify] https://crrev.com/4c259455066d41ac777613d0d44506c04d13dd57/gpu/ipc/service/gpu_memory_buffer_factory_android_hardware_buffer.cc [modify] https://crrev.com/4c259455066d41ac777613d0d44506c04d13dd57/ui/gl/BUILD.gn [modify] https://crrev.com/4c259455066d41ac777613d0d44506c04d13dd57/ui/gl/generate_bindings.py [modify] https://crrev.com/4c259455066d41ac777613d0d44506c04d13dd57/ui/gl/gl_bindings_api_autogen_egl.h [modify] https://crrev.com/4c259455066d41ac777613d0d44506c04d13dd57/ui/gl/gl_bindings_autogen_egl.cc [modify] https://crrev.com/4c259455066d41ac777613d0d44506c04d13dd57/ui/gl/gl_bindings_autogen_egl.h [modify] https://crrev.com/4c259455066d41ac777613d0d44506c04d13dd57/ui/gl/gl_image_ahardwarebuffer_unittest.cc [modify] https://crrev.com/4c259455066d41ac777613d0d44506c04d13dd57/ui/gl/gl_image_egl.cc [modify] https://crrev.com/4c259455066d41ac777613d0d44506c04d13dd57/ui/gl/gl_image_egl.h [modify] https://crrev.com/4c259455066d41ac777613d0d44506c04d13dd57/ui/gl/gl_image_native_pixmap.cc [modify] https://crrev.com/4c259455066d41ac777613d0d44506c04d13dd57/ui/gl/gl_image_native_pixmap.h [add] https://crrev.com/4c259455066d41ac777613d0d44506c04d13dd57/ui/gl/gl_image_native_pixmap_unittest.cc
,
Dec 11 2017
julien.isorce@ I have a question, you said "... GLX (VaapiTFPPicture(GLX)), i.e. --use-gl=desktop. I am suggesting to extend the vaapi support to --use-gl=egl." In which circumstances would a user want to use one or the other? IOW, would it be better (more performant and/or more maintainable) to switch to EGL altogether? Or is this a runtime config that would only be exercised if we run with the command line options to force-using-hardware-accelerated-video-decoding (don't remember the exact name) and --use-gl=egl ?
,
Dec 11 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/bc2855ec135981984ae905cf9fe2c0de9fcfde76 commit bc2855ec135981984ae905cf9fe2c0de9fcfde76 Author: Julien Isorce <julien.isorce@chromium.org> Date: Mon Dec 11 19:29:00 2017 Use the generic VaapiPictureFactory::GetGLTextureTarget Instead of hardcoding GL_TEXTURE_EXTERNAL_OES or GL_TEXTURE_2D This will be useful for the subsequent CL, https://chromium-review.googlesource.com/c/chromium/src/+/766787 Bug: 785201 Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Change-Id: I2678181b4b9389c39999d07cb9e56c102fc5f299 Tests: media_unittests --gtest_filter=*VaapiVideoDecodeAcceleratorTest* Reviewed-on: https://chromium-review.googlesource.com/819131 Reviewed-by: Dan Sanders <sandersd@chromium.org> Commit-Queue: Julien Isorce <julien.isorce@chromium.org> Cr-Commit-Position: refs/heads/master@{#523168} [modify] https://crrev.com/bc2855ec135981984ae905cf9fe2c0de9fcfde76/media/gpu/vaapi/vaapi_drm_picture.cc [modify] https://crrev.com/bc2855ec135981984ae905cf9fe2c0de9fcfde76/media/gpu/vaapi/vaapi_drm_picture.h [modify] https://crrev.com/bc2855ec135981984ae905cf9fe2c0de9fcfde76/media/gpu/vaapi/vaapi_picture.cc [modify] https://crrev.com/bc2855ec135981984ae905cf9fe2c0de9fcfde76/media/gpu/vaapi/vaapi_picture.h [modify] https://crrev.com/bc2855ec135981984ae905cf9fe2c0de9fcfde76/media/gpu/vaapi/vaapi_picture_factory.cc [modify] https://crrev.com/bc2855ec135981984ae905cf9fe2c0de9fcfde76/media/gpu/vaapi/vaapi_picture_factory.h [modify] https://crrev.com/bc2855ec135981984ae905cf9fe2c0de9fcfde76/media/gpu/vaapi/vaapi_tfp_picture.cc [modify] https://crrev.com/bc2855ec135981984ae905cf9fe2c0de9fcfde76/media/gpu/vaapi/vaapi_tfp_picture.h [modify] https://crrev.com/bc2855ec135981984ae905cf9fe2c0de9fcfde76/media/gpu/vaapi/vaapi_video_decode_accelerator.cc [modify] https://crrev.com/bc2855ec135981984ae905cf9fe2c0de9fcfde76/media/gpu/vaapi/vaapi_video_decode_accelerator_unittest.cc
,
Dec 12 2017
#c11, mcasas@: Chromium defaults to glx (i.e. --use-gl=desktop) on Linux Desktop X11. So yes currently this is a runtime config that would only be exercised if we run with the command line --disable-gpu-sandbox --ignore-gpu-blacklist --use-gl=egl I think the long term plan is to use --use-gl=angle by default on all platforms (MANGLE project). Note that --use-gl=angle sets gl::GLImplementation = kGLImplementationEGLGLES2 like --use-gl=egl. If I am not mistaken the above effort is completed for Windows (Though not enabled for all drivers crbug.com/693090), almost for Linux (works but not default yet crbug.com/522133) and on going for Mac (crbug.com/757974) Note that this current issue will benefit Chromium for Linux Desktop on Wayland as it defaults to egl and ofcourse no glx on Wayland (Wayland support in Chromium is an on going work but nearly there, there is a few cr bugs for it). I will just need to refactor the changes in vaapi_drm_picture.cc with Ozone a little bit more but all other CLs will be used as it is. At some points I believe vaapi_tfp_picture.cc (glx) will be removed so it should be easier to maintain as you mentioned. Also vaapi_drm_picture.cc (egl) is faster in theory but the path X Pixmap to gl texture is well optimized nowadays (gl_image_glx.cc) so the overhead is negligible. Once this bug is completed I will add support for VaapiDrmPicture with --use-gl=angle too. It should be straightforward because angle is alsmot a 'passthrough' to egl. I believe I will just need to add some plumbing in Angle project to map all required extensions that are not already mapped.
,
Dec 13 2017
#13, julien.isorce@:OK, that makes sense, IIUC in time we'll use MANGLE and that'll boil down to EGL and that will use your code path, so that's cool EGL-wise. That's one half of the landscape. We still have the situation that Vaapi*Pictures are only used by the VaAPI video encode/decode accelerators, correct? Hence users would need to explicitly pass the command line flags "--disable-gpu-sandbox --ignore-gpu-blacklist" in order to use the Vaapi*Pictures. I guess my question/concern is that the latter code depends on another chunk of code, the video decode acceleration support, that is not supposed to be used nor maintained, if I'm reading crbug.com/137247 and crbug.com/590209 correctly, or has something changed, or what's the plan here?
,
Dec 19 2017
#14: About your question, Daniele asked something similar here https://chromium-review.googlesource.com/c/chromium/src/+/766787#message-67edc5ee8cf34cdc4f9d559317daafec3a3bdec0 Then Stéphane and Dan replied the following: https://chromium-review.googlesource.com/c/chromium/src/+/766787#message-0050c8d5c74bef8f76d32e75be4d1e84eb3f609f https://chromium-review.googlesource.com/c/chromium/src/+/766787#message-f7102eafa34671b9cb61a4e5ce742e4a2728e878 So the CL https://chromium-review.googlesource.com/c/chromium/src/+/766787 will allow to do more tests on Linux while there is still no 'enablement plan' to users for now.
,
Dec 19 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/1e2bbbdf6ae30470ff59f9fb556be642e2784178 commit 1e2bbbdf6ae30470ff59f9fb556be642e2784178 Author: Julien Isorce <julien.isorce@chromium.org> Date: Tue Dec 19 23:52:34 2017 Fix random crashes in gl_unittests when passing --single-process-tests EGLApiTest calls ClearBindings() which set eglTerminateFn to null but left g_display not null. So the next call to gl::init::ShutdownGL will crash because it calls eglTerminate from GLSurfaceEGL::ShutdownOneOff. [ RUN ] EGLApiTest.DisabledExtensionBitTest [ OK ] EGLApiTest.DisabledExtensionBitTest [ RUN ] GLContextGLXTest.DoNotDestroyOnFailedMakeCurrent Received signal 11 SEGV_MAPERR 000000000000 gl::GLSurfaceEGL::ShutdownOneOff() gl::init::ShutdownGLPlatform() gl::init::ShutdownGL() gl::GLSurfaceTestSupport::InitializeOneOffImplementation() gl::GLImageTestSupport::InitializeGL() gl::GLContextGLXTest_DoNotDestroyOnFailedMakeCurrent_Test::TestBody() testing::Test::Run() Depending on the execution order the crash does not happen making the tests flaky. The solution is to call init::ShutdownGL which internally calls ClearBindings but it also makes sure to reset g_display first. Bug: 785201 Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Change-Id: I388be6271c03b10f5f23355539cb05104db945b1 Tests: gl_unittests --single-process-tests Reviewed-on: https://chromium-review.googlesource.com/834028 Reviewed-by: Kenneth Russell <kbr@chromium.org> Commit-Queue: Julien Isorce <julien.isorce@chromium.org> Cr-Commit-Position: refs/heads/master@{#525188} [modify] https://crrev.com/1e2bbbdf6ae30470ff59f9fb556be642e2784178/ui/gl/egl_api_unittest.cc [modify] https://crrev.com/1e2bbbdf6ae30470ff59f9fb556be642e2784178/ui/gl/gl_egl_api_implementation.cc [modify] https://crrev.com/1e2bbbdf6ae30470ff59f9fb556be642e2784178/ui/gl/gl_surface_egl.cc
,
Dec 20 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/7089b2da3e207e29d0124e3ac6907cc2f25138e1 commit 7089b2da3e207e29d0124e3ac6907cc2f25138e1 Author: Julien Isorce <julien.isorce@chromium.org> Date: Wed Dec 20 11:19:59 2017 Add specific initialisers to GLImageEGL's childs GLImageAHardwareBuffer: bool Initialize(AHardwareBuffer* buffer, bool preserved) GLImageNativePixmap: bool Initialize(uint32_t texture_id) Not keeping the same signature as GLImageEGL::Initialize to indicate how it is used. And if needed, more initialisers can be added. This prevents the user from doing mistakes when instantiating a subclass of GLImageEGL but calling the parent's initialiser. For consolidation this CL also moves GLImageEGL::Initialize from public to protected. Note that GLImageEGL cannot be instantiated directly because it does not implement a few pure virtual functions it got from its parent, gl::GLImage. Bug: 785201 Test: gl_unittests --gtest_filter=*NativePixmap* Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Change-Id: Ida5834bbf964893680f449f1f5da7371e46075bc Reviewed-on: https://chromium-review.googlesource.com/817563 Commit-Queue: Julien Isorce <julien.isorce@chromium.org> Reviewed-by: David Reveman <reveman@chromium.org> Cr-Commit-Position: refs/heads/master@{#525307} [modify] https://crrev.com/7089b2da3e207e29d0124e3ac6907cc2f25138e1/gpu/ipc/service/gpu_memory_buffer_factory_android_hardware_buffer.cc [modify] https://crrev.com/7089b2da3e207e29d0124e3ac6907cc2f25138e1/ui/gl/gl_image_ahardwarebuffer.cc [modify] https://crrev.com/7089b2da3e207e29d0124e3ac6907cc2f25138e1/ui/gl/gl_image_ahardwarebuffer.h [modify] https://crrev.com/7089b2da3e207e29d0124e3ac6907cc2f25138e1/ui/gl/gl_image_ahardwarebuffer_unittest.cc [modify] https://crrev.com/7089b2da3e207e29d0124e3ac6907cc2f25138e1/ui/gl/gl_image_egl.h [modify] https://crrev.com/7089b2da3e207e29d0124e3ac6907cc2f25138e1/ui/gl/gl_image_native_pixmap.cc [modify] https://crrev.com/7089b2da3e207e29d0124e3ac6907cc2f25138e1/ui/gl/gl_image_native_pixmap.h [modify] https://crrev.com/7089b2da3e207e29d0124e3ac6907cc2f25138e1/ui/gl/gl_image_native_pixmap_unittest.cc
,
Dec 21 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/bc7b0afd8a1be8cfee2164dead6f4bc3b1e1d3f4 commit bc7b0afd8a1be8cfee2164dead6f4bc3b1e1d3f4 Author: Julien Isorce <julien.isorce@chromium.org> Date: Thu Dec 21 10:15:36 2017 Enable VaapiDrmPicture when passing --use-gl=egl The existing VaapiTFPPicture(X11) only works with --use-gl=desktop. CQ-DEPEND=CL:817563 Bug: 785201 Tests: gfx_unittests --gtest_filter=*NativePixmapDmabuf* Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Change-Id: I304f729d3a9db04169fb3070c0382c561a659d5b Reviewed-on: https://chromium-review.googlesource.com/766787 Reviewed-by: Daniele Castagna <dcastagna@chromium.org> Reviewed-by: Dan Sanders <sandersd@chromium.org> Commit-Queue: Julien Isorce <julien.isorce@chromium.org> Cr-Commit-Position: refs/heads/master@{#525662} [modify] https://crrev.com/bc7b0afd8a1be8cfee2164dead6f4bc3b1e1d3f4/media/gpu/vaapi/vaapi_drm_picture.cc [modify] https://crrev.com/bc7b0afd8a1be8cfee2164dead6f4bc3b1e1d3f4/ui/gfx/BUILD.gn [add] https://crrev.com/bc7b0afd8a1be8cfee2164dead6f4bc3b1e1d3f4/ui/gfx/linux/native_pixmap_dmabuf.cc [add] https://crrev.com/bc7b0afd8a1be8cfee2164dead6f4bc3b1e1d3f4/ui/gfx/linux/native_pixmap_dmabuf.h [add] https://crrev.com/bc7b0afd8a1be8cfee2164dead6f4bc3b1e1d3f4/ui/gfx/linux/native_pixmap_dmabuf_unittest.cc
,
Jan 4 2018
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/a190f3eac76edf03471022b15e69085599df2189 commit a190f3eac76edf03471022b15e69085599df2189 Author: Julien Isorce <julien.isorce@chromium.org> Date: Thu Jan 04 13:56:18 2018 Can select GLImplementation in GLImageTestSupport::InitializeGL This allows to re-use generic GLImageTest templates in tests that requires EGL,GLES2. CQ-DEPEND=CL:817563 gl_unittests --gtest_filter=*GLImage* ozone_gl_unittests Bug: 785201 Tests: Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Change-Id: I3ce122b69bef136fdf8c544f0b432d438c528f07 Reviewed-on: https://chromium-review.googlesource.com/832226 Commit-Queue: Julien Isorce <julien.isorce@chromium.org> Reviewed-by: Michael Spang <spang@chromium.org> Reviewed-by: Robert Kroeger <rjkroege@chromium.org> Reviewed-by: Ian Vollick <vollick@chromium.org> Reviewed-by: Antoine Labour <piman@chromium.org> Cr-Commit-Position: refs/heads/master@{#526977} [modify] https://crrev.com/a190f3eac76edf03471022b15e69085599df2189/chrome/browser/vr/test/gl_test_environment.cc [modify] https://crrev.com/a190f3eac76edf03471022b15e69085599df2189/ui/gl/BUILD.gn [modify] https://crrev.com/a190f3eac76edf03471022b15e69085599df2189/ui/gl/gl_context_glx_unittest.cc [modify] https://crrev.com/a190f3eac76edf03471022b15e69085599df2189/ui/gl/gl_image_ahardwarebuffer_unittest.cc [modify] https://crrev.com/a190f3eac76edf03471022b15e69085599df2189/ui/gl/gl_image_dxgi_unittest.cc [modify] https://crrev.com/a190f3eac76edf03471022b15e69085599df2189/ui/gl/gl_image_io_surface_unittest.cc [modify] https://crrev.com/a190f3eac76edf03471022b15e69085599df2189/ui/gl/gl_image_native_pixmap_unittest.cc [modify] https://crrev.com/a190f3eac76edf03471022b15e69085599df2189/ui/gl/gl_image_ref_counted_memory_unittest.cc [modify] https://crrev.com/a190f3eac76edf03471022b15e69085599df2189/ui/gl/gl_image_shared_memory_unittest.cc [modify] https://crrev.com/a190f3eac76edf03471022b15e69085599df2189/ui/gl/test/gl_image_test_support.cc [modify] https://crrev.com/a190f3eac76edf03471022b15e69085599df2189/ui/gl/test/gl_image_test_support.h [add] https://crrev.com/a190f3eac76edf03471022b15e69085599df2189/ui/gl/test/gl_image_test_template.cc [modify] https://crrev.com/a190f3eac76edf03471022b15e69085599df2189/ui/gl/test/gl_image_test_template.h [modify] https://crrev.com/a190f3eac76edf03471022b15e69085599df2189/ui/ozone/gl/gl_image_ozone_native_pixmap_unittest.cc
,
Jan 15 2018
,
Mar 9 2018
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/e330bcb63ec23aec1ec260f5880f8768c6d36679 commit e330bcb63ec23aec1ec260f5880f8768c6d36679 Author: Miguel Casas <mcasas@chromium.org> Date: Fri Mar 09 14:55:06 2018 vaapi: split VaapiDrmPicture into common and ozone/egl parts This CL splits VaapiDrmPicture into a common part named VaapiPictureNativePixmap, and two subclasses, namely: - VaapiPictureNativePixmapOzone, containing the current #if defined(USE_OZONE) sandwiched parts, and - VaapiPixtureNativePixmapEgl with the other parts. So the class tree should look like: +-----------------+ | VaapiPicture | +-----------------+ ^ ^ | +----+ +--------------+ +------------------+ | VaapiPicture | | VaapiPictureTFP | | NativePixmap | | (existing on ToT)| +--------------+ +------------------+ ^ +---------------+ | | +-------+ +-------+ | VPNP | | VPNP | | Ozone | | Egl | +-------+ +-------+ The point here is to reduce the amount of if-defs to one single point, the VaapiPictureFactory. BUILD.gn is changed to reflect that VaapiDrmPicture is egl-specific (see bug, |use_egl| definition in [1]), while VaapiDrmPictureOzone is |use_ozone| [2] specific. Note that |use_egl| can coexist with |use_x11| [3]. This CL just moves code around, no new code introduced in either vaapi_picture*.cc. [1] https://cs.chromium.org/chromium/src/ui/gl/features.gni?q=ui/gl/features.gni&sq=package:chromium&dr&l=16 [2] https://cs.chromium.org/chromium/src/build/config/ui.gni?q=build/config/ui.gni&sq=package:chromium&dr&l=26 [3] https://cs.chromium.org/chromium/src/build/config/ui.gni?type=cs&sq=package:chromium&l=47 Bug: 785201 Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Change-Id: I67f6f4d994322ef7f23a8ed255405963a4be0152 Reviewed-on: https://chromium-review.googlesource.com/951928 Reviewed-by: Julien Isorce <julien.isorce@chromium.org> Commit-Queue: Miguel Casas <mcasas@chromium.org> Cr-Commit-Position: refs/heads/master@{#542107} [modify] https://crrev.com/e330bcb63ec23aec1ec260f5880f8768c6d36679/media/gpu/vaapi/BUILD.gn [delete] https://crrev.com/6402e8cbceaf038bc8342816050848de0aa8c9f3/media/gpu/vaapi/vaapi_drm_picture.cc [delete] https://crrev.com/6402e8cbceaf038bc8342816050848de0aa8c9f3/media/gpu/vaapi/vaapi_drm_picture.h [modify] https://crrev.com/e330bcb63ec23aec1ec260f5880f8768c6d36679/media/gpu/vaapi/vaapi_picture_factory.cc [add] https://crrev.com/e330bcb63ec23aec1ec260f5880f8768c6d36679/media/gpu/vaapi/vaapi_picture_native_pixmap.cc [add] https://crrev.com/e330bcb63ec23aec1ec260f5880f8768c6d36679/media/gpu/vaapi/vaapi_picture_native_pixmap.h [add] https://crrev.com/e330bcb63ec23aec1ec260f5880f8768c6d36679/media/gpu/vaapi/vaapi_picture_native_pixmap_egl.cc [add] https://crrev.com/e330bcb63ec23aec1ec260f5880f8768c6d36679/media/gpu/vaapi/vaapi_picture_native_pixmap_egl.h [add] https://crrev.com/e330bcb63ec23aec1ec260f5880f8768c6d36679/media/gpu/vaapi/vaapi_picture_native_pixmap_ozone.cc [add] https://crrev.com/e330bcb63ec23aec1ec260f5880f8768c6d36679/media/gpu/vaapi/vaapi_picture_native_pixmap_ozone.h [rename] https://crrev.com/e330bcb63ec23aec1ec260f5880f8768c6d36679/media/gpu/vaapi/vaapi_picture_tfp.cc [rename] https://crrev.com/e330bcb63ec23aec1ec260f5880f8768c6d36679/media/gpu/vaapi/vaapi_picture_tfp.h
,
Mar 9 2018
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/89caee51f7f91f6525996dd7d2b27f3b9a46ac7f commit 89caee51f7f91f6525996dd7d2b27f3b9a46ac7f Author: Miguel Casas <mcasas@chromium.org> Date: Fri Mar 09 18:13:08 2018 Revert "vaapi: split VaapiDrmPicture into common and ozone/egl parts" This reverts commit e330bcb63ec23aec1ec260f5880f8768c6d36679. Reason for revert: typo forced using EGL VaapiPictures, bad! Original change's description: > vaapi: split VaapiDrmPicture into common and ozone/egl parts > > This CL splits VaapiDrmPicture into a common part named > VaapiPictureNativePixmap, and two subclasses, namely: > - VaapiPictureNativePixmapOzone, containing the current > #if defined(USE_OZONE) sandwiched parts, and > - VaapiPixtureNativePixmapEgl with the other parts. > > So the class tree should look like: > > +-----------------+ > | VaapiPicture | > +-----------------+ > ^ ^ > | +----+ > +--------------+ +------------------+ > | VaapiPicture | | VaapiPictureTFP | > | NativePixmap | | (existing on ToT)| > +--------------+ +------------------+ > ^ > +---------------+ > | | > +-------+ +-------+ > | VPNP | | VPNP | > | Ozone | | Egl | > +-------+ +-------+ > > The point here is to reduce the amount of if-defs to one single > point, the VaapiPictureFactory. > > BUILD.gn is changed to reflect that VaapiDrmPicture is > egl-specific (see bug, |use_egl| definition in [1]), while > VaapiDrmPictureOzone is |use_ozone| [2] specific. Note > that |use_egl| can coexist with |use_x11| [3]. > > This CL just moves code around, no new code introduced in > either vaapi_picture*.cc. > > [1] https://cs.chromium.org/chromium/src/ui/gl/features.gni?q=ui/gl/features.gni&sq=package:chromium&dr&l=16 > [2] https://cs.chromium.org/chromium/src/build/config/ui.gni?q=build/config/ui.gni&sq=package:chromium&dr&l=26 > [3] https://cs.chromium.org/chromium/src/build/config/ui.gni?type=cs&sq=package:chromium&l=47 > > Bug: 785201 > Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel > Change-Id: I67f6f4d994322ef7f23a8ed255405963a4be0152 > Reviewed-on: https://chromium-review.googlesource.com/951928 > Reviewed-by: Julien Isorce <julien.isorce@chromium.org> > Commit-Queue: Miguel Casas <mcasas@chromium.org> > Cr-Commit-Position: refs/heads/master@{#542107} TBR=posciak@chromium.org,kcwu@chromium.org,mcasas@chromium.org,chcunningham@chromium.org,acourbot@chromium.org,julien.isorce@chromium.org Change-Id: If5103a891bcf6f758feea44cc514503457f00fbc No-Presubmit: true No-Tree-Checks: true No-Try: true Bug: 785201 Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Reviewed-on: https://chromium-review.googlesource.com/957322 Reviewed-by: Miguel Casas <mcasas@chromium.org> Commit-Queue: Miguel Casas <mcasas@chromium.org> Cr-Commit-Position: refs/heads/master@{#542148} [modify] https://crrev.com/89caee51f7f91f6525996dd7d2b27f3b9a46ac7f/media/gpu/vaapi/BUILD.gn [add] https://crrev.com/89caee51f7f91f6525996dd7d2b27f3b9a46ac7f/media/gpu/vaapi/vaapi_drm_picture.cc [add] https://crrev.com/89caee51f7f91f6525996dd7d2b27f3b9a46ac7f/media/gpu/vaapi/vaapi_drm_picture.h [modify] https://crrev.com/89caee51f7f91f6525996dd7d2b27f3b9a46ac7f/media/gpu/vaapi/vaapi_picture_factory.cc [delete] https://crrev.com/8d0e11cc9ffb098fd0f4405b95807cc67965fd64/media/gpu/vaapi/vaapi_picture_native_pixmap.cc [delete] https://crrev.com/8d0e11cc9ffb098fd0f4405b95807cc67965fd64/media/gpu/vaapi/vaapi_picture_native_pixmap.h [delete] https://crrev.com/8d0e11cc9ffb098fd0f4405b95807cc67965fd64/media/gpu/vaapi/vaapi_picture_native_pixmap_egl.cc [delete] https://crrev.com/8d0e11cc9ffb098fd0f4405b95807cc67965fd64/media/gpu/vaapi/vaapi_picture_native_pixmap_egl.h [delete] https://crrev.com/8d0e11cc9ffb098fd0f4405b95807cc67965fd64/media/gpu/vaapi/vaapi_picture_native_pixmap_ozone.cc [delete] https://crrev.com/8d0e11cc9ffb098fd0f4405b95807cc67965fd64/media/gpu/vaapi/vaapi_picture_native_pixmap_ozone.h [rename] https://crrev.com/89caee51f7f91f6525996dd7d2b27f3b9a46ac7f/media/gpu/vaapi/vaapi_tfp_picture.cc [rename] https://crrev.com/89caee51f7f91f6525996dd7d2b27f3b9a46ac7f/media/gpu/vaapi/vaapi_tfp_picture.h
,
Mar 13 2018
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/c9a424a38055c56a79f8aff102768c180331fae7 commit c9a424a38055c56a79f8aff102768c180331fae7 Author: Miguel Casas <mcasas@chromium.org> Date: Tue Mar 13 14:44:34 2018 RELAND: vaapi: split VaapiDrmPicture into common and ozone/egl parts I reverted the original CL because of a silly typo. This CL fixes it (see diff between PS1 and PS2): https://chromium-review.googlesource.com/c/chromium/src/+/957144/1..2/media/gpu/vaapi/vaapi_picture_factory.cc Original CL description ------------------------------------------------ This CL splits VaapiDrmPicture into a common part named VaapiPictureNativePixmap, and two subclasses, namely: - VaapiPictureNativePixmapOzone, containing the current #if defined(USE_OZONE) sandwiched parts, and - VaapiPixtureNativePixmapEgl with the other parts. So the class tree should look like: +-----------------+ | VaapiPicture | +-----------------+ ^ ^ | +----+ +--------------+ +------------------+ | VaapiPicture | | VaapiPictureTFP | | NativePixmap | | (existing on ToT)| +--------------+ +------------------+ ^ +---------------+ | | +-------+ +-------+ | VPNP | | VPNP | | Ozone | | Egl | +-------+ +-------+ The point here is to reduce the amount of if-defs to one single point, the VaapiPictureFactory. BUILD.gn is changed to reflect that VaapiDrmPicture is egl-specific (see bug, |use_egl| definition in [1]), while VaapiDrmPictureOzone is |use_ozone| [2] specific. Note that |use_egl| can coexist with |use_x11| [3]. This CL just moves code around, no new code introduced in either vaapi_picture*.cc. [1] https://cs.chromium.org/chromium/src/ui/gl/features.gni?q=ui/gl/features.gni&sq=package:chromium&dr&l=16 [2] https://cs.chromium.org/chromium/src/build/config/ui.gni?q=build/config/ui.gni&sq=package:chromium&dr&l=26 [3] https://cs.chromium.org/chromium/src/build/config/ui.gni?type=cs&sq=package:chromium&l=47 Bug: 785201 Change-Id: I9c1dbc2e7f6cf1fb851d0d6dffa068fad9652edd Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Reviewed-on: https://chromium-review.googlesource.com/951928 Reviewed-by: Julien Isorce <julien.isorce@chromium.org> Commit-Queue: Miguel Casas <mcasas@chromium.org> Reviewed-on: https://chromium-review.googlesource.com/957144 Cr-Commit-Position: refs/heads/master@{#542794} [modify] https://crrev.com/c9a424a38055c56a79f8aff102768c180331fae7/media/gpu/vaapi/BUILD.gn [delete] https://crrev.com/e02429434b868d9bcc0e6431e6037ff6ec2898bd/media/gpu/vaapi/vaapi_drm_picture.cc [delete] https://crrev.com/e02429434b868d9bcc0e6431e6037ff6ec2898bd/media/gpu/vaapi/vaapi_drm_picture.h [modify] https://crrev.com/c9a424a38055c56a79f8aff102768c180331fae7/media/gpu/vaapi/vaapi_picture_factory.cc [add] https://crrev.com/c9a424a38055c56a79f8aff102768c180331fae7/media/gpu/vaapi/vaapi_picture_native_pixmap.cc [add] https://crrev.com/c9a424a38055c56a79f8aff102768c180331fae7/media/gpu/vaapi/vaapi_picture_native_pixmap.h [add] https://crrev.com/c9a424a38055c56a79f8aff102768c180331fae7/media/gpu/vaapi/vaapi_picture_native_pixmap_egl.cc [add] https://crrev.com/c9a424a38055c56a79f8aff102768c180331fae7/media/gpu/vaapi/vaapi_picture_native_pixmap_egl.h [add] https://crrev.com/c9a424a38055c56a79f8aff102768c180331fae7/media/gpu/vaapi/vaapi_picture_native_pixmap_ozone.cc [add] https://crrev.com/c9a424a38055c56a79f8aff102768c180331fae7/media/gpu/vaapi/vaapi_picture_native_pixmap_ozone.h [rename] https://crrev.com/c9a424a38055c56a79f8aff102768c180331fae7/media/gpu/vaapi/vaapi_picture_tfp.cc [rename] https://crrev.com/c9a424a38055c56a79f8aff102768c180331fae7/media/gpu/vaapi/vaapi_picture_tfp.h |
|||
►
Sign in to add a comment |
|||
Comment 1 by julien.isorce@chromium.org
, Nov 15 2017