Repository navigation
Tags
Tags
[kotlin] Abort on unhandled callback exceptions When an uncaught exception occurs on a callback thread newly attached by JNI, forward it to the thread's custom or default UncaughtExceptionHandler when one is configured, and otherwise log the Java stack trace and abort natively instead of returning into Dawn's native worker loop. Unlike a terminating Java thread where Thread.dispatchUncaughtException() ignores exceptions thrown by the handler, also abort if looking up or invoking the handler throws a secondary exception so execution does not return into Dawn's native worker loop. Add subprocess abort tests in AsyncHelperTest (~7.0 s total across the 4 subprocess probes, ~1.7-1.8 s per probe on Pixel 9). Also verified manually in an Android app process with the stock RuntimeInit$KillApplicationHandler installed that an uncaught callback exception on DawnWorker logs "FATAL EXCEPTION: DawnWorker" in logcat, reports the crash via handleApplicationCrash (ApplicationExitInfo REASON_CRASH), and writes no native tombstone. Bug: 581261070 Test: ./gradlew :webgpu:connectedDebugAndroidTest Change-Id: I20786ca55ccb9c782ca09672d09edf4380703a86 Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/353675 Commit-Queue: Tarun Saini <sainitarun@google.com> Reviewed-by: Nikita Gupta <nikigupta@google.com>
Reland "D3D11: Create shader objects asynchronously." This is a reland of commit ba3c279 What's changed: - Disable async shader creations if a debug layer is detected. - A debug layer will use an D3D11Multithread mutex that would cause deadlock if we create shaders asynchronously and wait for them in Queue::Submit. This is because Queue::Submit will hold that mutex until the command buffer is submitted. Original change's description: > D3D11: Create shader objects asynchronously. > > This change makes the creation of D3D11 shader objects > (ID3D11VertexShader, etc.) asynchronous. Currently, these objects must > be ready when a pipeline is used for command buffer encoding. However, > they are only strictly needed when the command buffer is submitted. > > By deferring the creation until submission, this change avoids blocking > the encoding thread if the shader takes a long time to be created. This > allows for more ample time for shader creation to happen in parallel > with command recording. > > Bug: 439845637, chromium:40266176, chromium:459978952 > Change-Id: Ib5c678bdcf5218d941214db46aa6b5b236bace46 > Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/270015 > Reviewed-by: Corentin Wallez <cwallez@chromium.org> > Reviewed-by: Geoff Lang <geofflang@chromium.org> > Commit-Queue: Corentin Wallez <cwallez@chromium.org> > Auto-Submit: Quyen Le <lehoangquyen@chromium.org> Bug: 439845637, chromium:40266176 Change-Id: I840373a37b7393b3e9060ea02764517d166b199f Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/273554 Commit-Queue: Quyen Le <lehoangquyen@google.com> Reviewed-by: Antonio Maiorano <amaiorano@google.com>
[dawn][metal] Fix SharedTextureMemoryTest on Mac Debug This test was failing but apparently no bot runs error injection tests (normally enabled by default on Debug) on Mac. dawn_end2end_tests --gtest_filter='Metal/SharedTextureMemoryTests.EndAccessAfterFailedSubmitThenDeviceDestroyed/*' Bug: none, 553115993 Change-Id: I1953916fd424d1a4d6a74bb3814df59d5fe02f8b Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/357896 Auto-Submit: Kai Ninomiya <kainino@chromium.org> Reviewed-by: Loko Kung <lokokung@google.com> Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Make DeviceBase::ReduceMemoryUsageImpl return ResultOrError<bool> Return pending-work status through ResultOrError<bool> and handle backend errors in DeviceBase. Update D3D11 and Vulkan implementations, mocks, and tests for pending-work status and backend error handling. Fixed: 479477220 Change-Id: I9c4e1f1c59008e5e1125d28b95bf113ca9d1d879 Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/355235 Commit-Queue: Aaron Teo <hocheung@chromium.org> Reviewed-by: Brandon Jones <bajones@chromium.org> Reviewed-by: Quyen Le <lehoangquyen@google.com>
Roll DirectX Shader Compiler from 823a3dadf981 to c8611f6ff76e (1 rev… …ision) https://chromium.googlesource.com/external/github.com/microsoft/DirectXShaderCompiler.git/+log/823a3dadf981..c8611f6ff76e 2026-11-11 jackell@microsoft.com [HLSL] Allow a small ULP tolerance in LinAlg float matrix multiply tests (#9096) If this roll has caused a breakage, revert this CL and set the roller to dry run mode using the controls here: https://autoroll.skia.org/r/directx-shader-compiler-dawn Please CC rharrison@google.com,webgpu-developers@google.com on the revert to ensure that a human is aware of the problem. To file a bug in DirectX Shader Compiler: https://github.com/microsoft/DirectXShaderCompiler/issues/new/choose To file a bug in Dawn: https://bugs.chromium.org/p/dawn/issues/entry To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Bug: None Change-Id: I79236addf8423ec6e19293ef0a364d445219bc04 Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/358755 Bot-Commit: dawn-autoroll@skia-public.iam.gserviceaccount.com <dawn-autoroll@skia-public.iam.gserviceaccount.com> Commit-Queue: dawn-autoroll@skia-public.iam.gserviceaccount.com <dawn-autoroll@skia-public.iam.gserviceaccount.com>
Fix ShaderF16Tests skipping on D3D12 when use_dxc is default
ShaderF16Tests::GetRequiredFeatures() only requested ShaderF16 on D3D12
if "use_dxc" appeared in forceEnabledWorkarounds. Since use_dxc became
enabled by default on D3D12 and test instantiations switched to
D3D12Backend() and D3D12Backend({}, {"use_dxc"}), all D3D12
ShaderF16Tests silently skipped.
Rely directly on SupportsFeatures({wgpu::FeatureName::ShaderF16}), since
ValidateFeatureSupportedWithTogglesImpl already rejects ShaderF16 when
use_dxc is disabled on the adapter.
Bug: 454348977, 577702448
Change-Id: I1dc0c9434fc46c5ec7f6c8a59190572b5ba141e2
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/358277
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Commit-Queue: Natalie Chouinard <chouinard@google.com>
Roll Mesa from cc2dcf9abdc6 to 7526bf51645c (1 revision) https://chromium.googlesource.com/external/gitlab.freedesktop.org/mesa/mesa.git/+log/cc2dcf9abdc6..7526bf51645c 2026-11-10 jeongik@google.com gfxstream: support variable descriptor counts and sparse sets If this roll has caused a breakage, revert this CL and set the roller to dry run mode using the controls here: https://autoroll.skia.org/r/mesa-dawn Please CC rharrison@google.com,webgpu-dev-team@google.com on the revert to ensure that a human is aware of the problem. To file a bug in Mesa: https://crbug.com/dawn To file a bug in Dawn: https://crbug.com/dawn To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Bug: None Change-Id: I191a98bbbbd2d611b5805bc3faad644bf41e2282 Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/357817 Commit-Queue: dawn-autoroll@skia-public.iam.gserviceaccount.com <dawn-autoroll@skia-public.iam.gserviceaccount.com> Bot-Commit: dawn-autoroll@skia-public.iam.gserviceaccount.com <dawn-autoroll@skia-public.iam.gserviceaccount.com>
Roll Mesa from b8368b4fb346 to 62dc26608fba (5 revisions) https://chromium.googlesource.com/external/gitlab.freedesktop.org/mesa/mesa.git/+log/b8368b4fb346..62dc26608fba 2026-11-07 hamber@bk.ru microsoft/compiler: Respect special-value preservation in DXIL binops 2026-11-07 jnoorman@igalia.com ir3/validate: validate SSA dominance 2026-11-07 jnoorman@igalia.com ir3: rename is_sat_compatible to sat_is_fsat 2026-11-07 jnoorman@igalia.com ir3: only fold fsat to (sat) on cat2/3 float instructions 2026-11-07 jnoorman@igalia.com ir3: move is_cat[23]_float helper up in their file If this roll has caused a breakage, revert this CL and set the roller to dry run mode using the controls here: https://autoroll.skia.org/r/mesa-dawn Please CC alanbaker@google.com,webgpu-dev-team@google.com on the revert to ensure that a human is aware of the problem. To file a bug in Mesa: https://crbug.com/dawn To file a bug in Dawn: https://crbug.com/dawn To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Bug: None Change-Id: I65ce58eaeba0435feece31eabc45fd670059c417 Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/357235 Bot-Commit: dawn-autoroll@skia-public.iam.gserviceaccount.com <dawn-autoroll@skia-public.iam.gserviceaccount.com> Commit-Queue: dawn-autoroll@skia-public.iam.gserviceaccount.com <dawn-autoroll@skia-public.iam.gserviceaccount.com>
Roll Mesa from fef00314b672 to 845c8162f2fa (6 revisions) https://chromium.googlesource.com/external/gitlab.freedesktop.org/mesa/mesa.git/+log/fef00314b672..845c8162f2fa 2026-11-06 sushma.venkatesh.reddy@intel.com anv: Enable typed clear barriers for Grid Legends 2026-11-06 Matthew.Arsenault@amd.com radv/llvm: keep the ring offsets in s[0:1] for GFX11+ compute shaders 2026-11-06 michael.blumenkrantz@gmail.com zink: reset attachment storeOp when begin_rendering has no tc info 2026-11-06 kandru@qti.qualcomm.com freedreno/ir3: Lower FP32 to FP16 for subnormals 2026-11-06 erhan.bilgili@gmail.com winsys/amdgpu: Fix UserQ signal point array allocation. 2026-11-06 oliverhenryc@gmail.com amd/common: mask count for SDMA_OPCODE_WRITE in ib parser If this roll has caused a breakage, revert this CL and set the roller to dry run mode using the controls here: https://autoroll.skia.org/r/mesa-dawn Please CC alanbaker@google.com,webgpu-dev-team@google.com on the revert to ensure that a human is aware of the problem. To file a bug in Mesa: https://crbug.com/dawn To file a bug in Dawn: https://crbug.com/dawn To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Bug: None Change-Id: I7cb7fc23734d57a8256351e63df89f93f8d95160 Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/356695 Commit-Queue: dawn-autoroll@skia-public.iam.gserviceaccount.com <dawn-autoroll@skia-public.iam.gserviceaccount.com> Bot-Commit: dawn-autoroll@skia-public.iam.gserviceaccount.com <dawn-autoroll@skia-public.iam.gserviceaccount.com>
Roll Mesa from 9ac9e279ca22 to 0848c37191ed (3 revisions) https://chromium.googlesource.com/external/gitlab.freedesktop.org/mesa/mesa.git/+log/9ac9e279ca22..0848c37191ed 2026-11-05 kherbst@redhat.com rusticl: implement cl_khr_external_memory 2026-11-05 kherbst@redhat.com rusticl/mem: use MultiValProperties for Buffer and Image 2026-11-05 michael.blumenkrantz@gmail.com glsl: treat volatile as coherent when translating to nir If this roll has caused a breakage, revert this CL and set the roller to dry run mode using the controls here: https://autoroll.skia.org/r/mesa-dawn Please CC alanbaker@google.com,webgpu-dev-team@google.com on the revert to ensure that a human is aware of the problem. To file a bug in Mesa: https://crbug.com/dawn To file a bug in Dawn: https://crbug.com/dawn To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Bug: None Change-Id: I0adb433102ec00eba22a78a4329798c2875548b2 Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/356175 Bot-Commit: dawn-autoroll@skia-public.iam.gserviceaccount.com <dawn-autoroll@skia-public.iam.gserviceaccount.com> Commit-Queue: dawn-autoroll@skia-public.iam.gserviceaccount.com <dawn-autoroll@skia-public.iam.gserviceaccount.com>
PreviousNext