Bug: angleproject:3333
Change-Id: Ib21d863ada0d9f23a84647265b692cd5485a064d
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1545994Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>

https://chromium.googlesource.com/external/github.com/KhronosGroup/SPIRV-Tools.git/+log/2ff54e34ed37..fcb845310450


Created with:
  gclient setdep -r ./third_party/spirv-tools/src@fcb845310450

The AutoRoll server is located here: https://autoroll.skia.org/r/spirv-tools-angle-autoroll

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md

If the roll is causing failures, please contact the current sheriff, who should
be CC'd on the roll, and stop the roller if necessary.


TBR=jonahr@google.com

Change-Id: I7aa44ede3a9f7c4e3b68dbcd5c2b3962cd1ce7e8
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1545971Reviewed-by: angle-autoroll <angle-autoroll@skia-public.iam.gserviceaccount.com>
Commit-Queue: angle-autoroll <angle-autoroll@skia-public.iam.gserviceaccount.com>

Bug: angleproject:3332
Change-Id: Ia937727062d3296af9691a95d296c89b8e9476ac
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1544779Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>

This is desired for implementing WEBGL_color_buffer_float support on
WebGL1 on ANGLE's ES2 driver, for Firefox.

Bug: angleproject:3322
Change-Id: I599f86db62402333d3ef9235956c63e667f1513d
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1541722Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>

We were casting stencil ref value instead of clamping it as the spec
requires. Fixing that which allows a number of new dEQP tests to pass.

Bug: angleproject:3244
Bug: angleproject:3245
Bug: angleproject:3247
Bug: angleproject:3248
Change-Id: I59242d59982b990e82395f5a1e77b9a54254dc2b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1541751
Commit-Queue: Tobin Ehlis <tobine@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>

https://chromium.googlesource.com/external/github.com/KhronosGroup/SPIRV-Tools.git/+log/2ac348b5c0a5..2ff54e34ed37


Created with:
  gclient setdep -r ./third_party/spirv-tools/src@2ff54e34ed37

The AutoRoll server is located here: https://autoroll.skia.org/r/spirv-tools-angle-autoroll

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md

If the roll is causing failures, please contact the current sheriff, who should
be CC'd on the roll, and stop the roller if necessary.


TBR=jonahr@google.com

Change-Id: Ic241dffc79ae1f24eb61e2c28fc184c9cebceaaa
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1541749Reviewed-by: angle-autoroll <angle-autoroll@skia-public.iam.gserviceaccount.com>
Commit-Queue: angle-autoroll <angle-autoroll@skia-public.iam.gserviceaccount.com>

The dEQP-GLES2.functional.fbo.render.texsubimage.between_render_tex2d*
test was failing because a TexSubImage call modifying the framebuffer
wasn't being applied before a subsequent DrawElements call and thus
getting an incorrect result.

Bug: angleproject:3249
Test: angle_deqp_gles2_tests --gtest_filter=dEQP.GLES2/functional_fbo_render_texsubimage_between_render_tex2d_rgb --use-angle=vulkan
Change-Id: I90f791957e6536c5c62fb731d52cd486ab5c05c1
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1542361
Commit-Queue: Courtney Goeltzenleuchter <courtneygo@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

https://chromium.googlesource.com/external/github.com/KhronosGroup/glslang/+log/5efb004d5960..e0d59bbe1857


Created with:
  gclient setdep -r ./third_party/glslang/src@e0d59bbe1857

The AutoRoll server is located here: https://autoroll.skia.org/r/glslang-angle-autoroll

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md

If the roll is causing failures, please contact the current sheriff, who should
be CC'd on the roll, and stop the roller if necessary.


TBR=jonahr@google.com

Change-Id: Ifa2bf394fe4b895046d50c434db473057bf1b90b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1541750Reviewed-by: angle-autoroll <angle-autoroll@skia-public.iam.gserviceaccount.com>
Commit-Queue: angle-autoroll <angle-autoroll@skia-public.iam.gserviceaccount.com>

On tiling GPUs, render pass loadOp and stencilLoadOp can be used to very
cheaply clear an image as it is being render to.  This change uses this
feature to clear render targets when possible.

Bug: angleproject:2361
Change-Id: Ic4bdc908873f4802760d549f4893f84a47beac0f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1500576
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

The EGL_KHR_fence_sync spec says that if a flush is needed before waiting
on the sync, it's done on the current context for the current thread.

This helps simplify the multithreading design, we don't need to worry about
flusing on a context that may no longer exist or is executing on a different
thread.  It does allow infinite waits because the context with the fence is
never flushed but the spec allows this.

BUG=angleproject:2464

Change-Id: I8bf2f93c408fee2dae95caa5bb9c76ba67687931
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1542256Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>

Bug: angleproject:3317
Change-Id: Ie60ad9670d19d0eae9926bcf14f65fc2e84215d8
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1542602Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

This is possible thanks to SecondaryCommandBuffer.  Makes life easier
when debugging by not just showing resource type in the nodes, but
actual stream of commands recorded in each.

Bug: angleproject:3136
Change-Id: I125a32ec2966a55330e60930ca088d1a3673a8ba
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1538832
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>

Bug: angleproject:3320
Change-Id: Ia24f9bc2499f36c6aab810db00bb6408920222fd
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1541719
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

If using vk::priv::SecondaryCommandBuffer.  This would allow ES3 support
where inheritedQueries is not supported.

Bug: angleproject:3136
Change-Id: I10508058301ea6da8f3415cfdcc052500a67f810
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1538829
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tobin Ehlis <tobine@google.com>

Except WGL AMD, WGL Intel, and GLX Nvidia.

BUG=922936

Change-Id: I1bdbddd1485578465bcf6ece4b4b7173ae6f9f05
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1442312Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Jie A Chen <jie.a.chen@intel.com>

If an FBO only has a depth/stencil attachment, the attachment index will
be 0, but that value was incorrectly being used as a flag for no
depth/stencil attachment.

Bug: angleproject:2361
Change-Id: I4ee803a392a29f7b261bc0cfabf9f2507dc7b178
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1541670
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

RenderTargetVk::getImageExtents() was always returning the base level
extents of the image it was viewing.  This resulted in incorrect extents
being provided to various framebuffer- and renderpass-related
declarations.  The function is renamed to `getExtents()` and returns the
appropriate mip extents.

Bug: angleproject:2361
Change-Id: I059a8d19fcb140c9095107d935aa3e5cb1852fc2
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1537327
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

A bunch of changes to rework and improve SecondaryCommandBuffers.

Inlined all of the command functions and moved them into the header.
Created new specialized commands for updating Compute/Graphics
DescriptorSets and setting a memoryBarrier.

Updated all of the pointer storage to be inferred rather than
explicitly stored in order to save space. Also removed various params
that are fixed in ANGLE to save space.

Bug: angleproject:3136
Change-Id: I994bb70d5e4db6d9e71d38ac62451aaec780a5e1
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1535704Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Tobin Ehlis <tobine@google.com>

The new validation added in http://crbug.com/809237 failed to consider
that sampler object state overrides texture object state if a sampler
object is bound. State caching makes this complicated to fix.

Fixes WebGL conformance test incompatible-texture-type-for-sampler.html
https://github.com/KhronosGroup/WebGL/pull/2823

Bug: 940080, 809237
Change-Id: I26b0fb35c5630c36248edae80f0298a0cb7e14b8
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1522364
Commit-Queue: James Darpinian <jdarpinian@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

Deleting a buffer that is bound to a VAO should act as if
the application unbound the buffer. Unbinding the buffer
should update relevant validation caches. But we were
missing the logic that updates the validation caches.

This CL adds the necessary cache updates. It does not include a
regression test. The test was causing an unrelated regression that is
going to be a bit longer. It should not block this fix.

Bug: chromium:943538
Change-Id: Ib073cd07a230ca073a5b14bc054e961158a0097d
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1536491
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

EGLThreadTest doesn't clean up well and the test framework can't handle
it well. It caused test failures related to eglMakeCurrent (ELG_BAD_ACCESS)
on Android that were masked by retries.

Calling eglGetDisplay within a test can cause it to return a cached
display from the test runner.  The test is unable to know if it should
terminate this display.

bug: angleproject:3321
Change-Id: I14a539a00acaed4ee71622e7416a6c7a75596606
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1541717
Commit-Queue: Jonah Ryan-Davis <jonahr@google.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

Update the parameter checking performed within ValidateTexImage2D() to
pass the following tests:

dEQP-GLES2.functional.negative_api.texture.teximage2d_invalid_format
dEQP-GLES2.functional.negative_api.texture.teximage2d_invalid_internalformat
dEQP-GLES2.functional.negative_api.texture.texsubimage2d_invalid_type

Bug: angleproject:3250
Bug: angleproject:3251
Test: angle_deqp_gles2_tests --use-angle=vulkan
Change-Id: I4d9be4fe0a9b377e61e3132db262750e6285464b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1534519
Commit-Queue: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>

Bug: angleproject:3317
Change-Id: I0c335c6af5d183a8ef7fa4154b6dc897cad0b8ff
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1541721Reviewed-by: Jonah Ryan-Davis <jonahr@google.com>

Since we can't test these failures on the CQ this CL likely won't hit
all the necessary suppressions. Will follow up with a second CL.

Bug: angleproject:3317
Change-Id: I241c139459d7cc7ab3c17d61a9837cdd3ab6c9a7
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1542017
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Jonah Ryan-Davis <jonahr@google.com>

These seem to be flaky and failing on all D3D11 devices.

https://ci.chromium.org/p/chromium/builders/ci/Win10%20FYI%20dEQP%20Release%20%28NVIDIA%29/25142
https://ci.chromium.org/p/chromium/builders/ci/Win10%20FYI%20dEQP%20Release%20%28NVIDIA%29/25098
https://ci.chromium.org/p/chromium/builders/ci/Win10%20FYI%20dEQP%20Release%20%28NVIDIA%29/25092

Bug: angleproject:1729
Change-Id: I4ae344ec3d950a6e647649a4ecf3ecafaca2de4f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1541666Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Can rename/clean up the library to move from chrome style to angle
style. Can also start pulling out chrome-specific definitions. More work
to be done when moving gpu_info to use Angle's SystemInfo.

Bug: angleproject:2677
Change-Id: I75fd4446aa4be85fba590cd00cd3a213a0b3752e
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1538677
Commit-Queue: Jonah Ryan-Davis <jonahr@google.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

The test texture size is reduced from 2k x 2k to 1k x 1k to avoid OOM on
lower-memory devices.

Bug: 945415
Change-Id: I66611417114212427f0a2cfc677450f5e705a7ca
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1537331
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>

Causes crashes.

Bug: angleproject:3310
Bug: 945415
Change-Id: I59c5fee0afaf74c99ac19168fbfd96de04796add
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1534946
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

Bug: angleproject:1944
Change-Id: Ie10bd447b08fd881961e7fa0849113f40f98ba1b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1538679Reviewed-by: Cody Northrop <cnorthrop@google.com>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>

The intent is actually to suppress on 64-bit,
but test expectations don't have this capability.
Since the only Win7 NVIDIA dEQP bot is x64, and that's the only x64 bot,
this works.

Bug: angleproject:1729, chromium:943312
Change-Id: Ib985bc9ff62333deaa4a943c3f3093de51159005
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1538314Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>

glmark2 is being included as part of angle_perftests, which is built as
an isolate.

Isolates do not work correctly unless there is a trailing slash for
folders in the data section.

Bug: angleproject:3311
Change-Id: I55e5a655ae15683d5cb070589970ba5d9eba4da0
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1538676
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

Enabling deqp tests previously skipped due to lack of atomic counters. Fixing
bug found in translator found by new tests.

* Switching atomicCounterDecrement from pre to post decrement
* Added 4 byte alignment check to atomic_uint offset
* Added workaround for NVIDIA D3D bug
* Added globallycoherent to atomic counters

Bug: angleproject:1729
Change-Id: If62ea003826fbe2df0834b905ff3ad7b76328399
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1480867
Commit-Queue: Enrico Galli <enrico.galli@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>

Disabled on Android because of issues with test parsing.

Bug: angleproject:2677
Change-Id: I75197e423f35bd8b84e27bb9b14d8c91779ad9c8
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1537696
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Jonah Ryan-Davis <jonahr@google.com>

1. Copy the code from src/tests/third_party/gpu_test_expectations to
src/test/test_expectations, rename .cc files to .cpp.  Put these in a new
static library and update dEQP to link against it in src/tests/BUILD.gn.
2. Merge the code in angle_config.h into the rest of the expectations parser,
this code was added so that the rest of the parser would compile in ANGLE's
tree with minimal modification.

Still need to follow up with the third step to close the issue.

Bug: angleproject:2677
Change-Id: Icf09b4eeed83a6d09b1964ad2adcfa85cabb4b63
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1536312
Commit-Queue: Jonah Ryan-Davis <jonahr@google.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

There are an number of valid tests that are excluded from the Android
test lists. Likely some older and low spec devices weren't passing
these tests and for historical reasons they were omitted from the
Android CTS. ANGLE however needs to pass these tests to be conformant
and up until now we weren't testing them on our bots.

The EGL tests still use the same Android version of the test list. It
seems the Khronos version just lists wildcards. ANGLE needs an explicit
test list.

Bug: angleproject:3242
Change-Id: Ie0aa2aa94689be0a40e7cc39e05f0042c3f03f5d
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1520986Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Some tests fail only on specific devices, so the tests should be updated
to reflect that, and allow more tests to run on more devices.

Bug: angleproject:3275
Change-Id: I8e3183c1769c0bb8ed6d2605afcaf399cb1d9ed0
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1534463Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jonah Ryan-Davis <jonahr@google.com>

We would end up with a large reserved space even though the
allocation failed. Insead set the reserved space size after
the allocation succedes.

This was showing up as angle_end2end_tests failures on
Windows 7 due to display reuse and buffer allocation.

Bug: chromium:944454
Change-Id: Idb3bd530fe7b9cc2fce9a579787684e632b1a637
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1534684Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

This reverts commit 3e8a8d5b.

Reason for revert: Didn't seem to fix the bot.

Original change's description:
> Force new displays on each Windows 7 end2end test.
> 
> This should fix the buffer allocation crash manifesting on Windows 7
> x64.
> 
> Bug: angleproject:3261
> Bug: chromium:944454
> Change-Id: I8a0713e1d1f18285f128c0cfb98b398c6a336a36
> Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1534461
> Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
> Commit-Queue: Jamie Madill <jmadill@chromium.org>

TBR=ynovikov@chromium.org,syoussefi@chromium.org,jmadill@chromium.org

# Not skipping CQ checks because original CL landed > 1 day ago.

Bug: angleproject:3261, chromium:944454
Change-Id: I835331cbe444cc9415bef77f06462be580b43f49
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1535235Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Moved vk::CommandBuffer and vk::SecondaryCommandBuffer to vk::priv:: and
aliased vk::CommandBuffer to one or the other.  This allows the rest of
the classes to continue seeing vk::CommandBuffer as they used to do.
Used a special alias for the primary command buffer that gets submitted
(vk::PrimaryCommandBuffer).

Bug: angleproject:3136
Change-Id: I61236fd182230991db7395d05e3da3be5e3f45be
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1534456Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>

RELAND of this commit. Had to fix fuzzer build errors.

Optimize performance of SecondaryCommandBuffers and enable them as the
default build option.
To disable this set angle_enable_custom_vulkan_cmd_buffers=false in
your build args.

This CL enhances the PoolAllocator to have a "fast" mode that can
be enabled at class creation. This mode uses an alignment of 1 byte and
enables a fastAllocation() call that avoids some bookkeeping overhead.
The SecondaryCommandBuffer uses this fastAllocation() function.
Furthermore the fast path of fast allocate, using the current page,
is inlined for maximum speed.
Jamie Madill also updated the SecondaryCommandBuffers to pre-allocate
blocks so that the commands occur linearly in memory. This speeds up
processing with improved cache coherency and minimizes overhead when
recording commands.
Also the core Draw functions and their state updates are all inlined
as well as the common functions to initialize commands and to copy
command pointer data.

This change also includes some new, custom commands. One is
imageBarrier that is a specialized version of pipelineBarrier that only
performs a single image layout transition.
There are customized versions of various Draw commands to minimize
copying of parameters.
There are also specialized commands to bind[Graphics|Compute]Pipeline
that have the pipeline type built in to the command.
More custom commands and command data size optimizations will be made
in follow-on commits.

Bug: angleproject:3136

Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1497418
Commit-Queue: Tobin Ehlis <tobine@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Change-Id: I621d8f8893308fca240b32390928e8ba0036cf06
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1535385Reviewed-by: Tobin Ehlis <tobine@google.com>