This ensures an effect setup code path that calls GrMatrix::Make()
doesn't produce an identity and non-identity shader variant.
Change-Id: I4d17395ed1b5bfd2bf9ceac68e85756c18eb2d2f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/396018
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Change-Id: Iee1ea7ee2d545138a8243c373f7a163f7120548d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/396337
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Preparation for optimizing the case of PaintGlyph + PaintSolid /
PaintGradient so that the remainder of the colrv1_draw_paint() function
can be turned into configuring an SkPaint.
Perform canvas CTM concatenation at the end of the function instead of
in each individual case as before.
No functional changes.
Bug: 11851
Change-Id: I04a77134d406f1625efebd5f623cfdbcca19b7cb
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/396036
Commit-Queue: Dominik Röttsches <drott@chromium.org>
Commit-Queue: Ben Wagner <bungeman@google.com>
Auto-Submit: Dominik Röttsches <drott@chromium.org>
Reviewed-by: Ben Wagner <bungeman@google.com>
A subset of these signatures will be available to public SkSL, and
structured in a way that really pushes the coords as the primary
argument (and color as an optional one). In any case, I find this
ordering more natural.
Change-Id: I7b3bc962c5b305b9eeed1ae11ae1dc2ce7269364
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/396021
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Bug: skia:11844
Change-Id: I13165fb17b06371854a4911c5d91bf2415d4df2a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/396017
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
The GPU will provide a linear mix of two texels as part of sampling.
We do have to change sampling from nearest to linear, but cutting the number
of texture lookups in half should more than make up for that - texture
fetches tend to be slow operations and the hardware has been optimized for
linear texture fetches.
Bug: skia:11796
Change-Id: Id6fc3e3cdd2491c841b31fe97ca718d6916f27be
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/390796
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This is a reland of 45889d1269
Original change's description:
> Remove code to push pixmaps to backend textures from GrGpu classes
>
> Replace GrGpu::updateBackendTexture with narrower method that clears
> a backend texture.
>
> Creation of data for a solid color compressed texture is lifted up to
> GrDirectContext and goes through updateCompressedBackendTexture.
>
> Bug: skia:11786
> Change-Id: I1d617623df5e65686f30e57c361a64f78d77f7bd
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/392836
> Reviewed-by: Jim Van Verth <jvanverth@google.com>
> Reviewed-by: Greg Daniel <egdaniel@google.com>
> Commit-Queue: Brian Salomon <bsalomon@google.com>
Bug: skia:11786
Change-Id: Ibf85794a33e68acb17f0fb704f6815cd6460b3f2
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/396098
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Also switches GrFillRRectOp to call usesMSAASurface.
Bug: skia:11396
Change-Id: I66d0ee2c6dc76819efc0f3283f6e1a0af11f23f7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395136
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
This reverts commit 45889d1269.
Reason for revert: compressed texture assertion on D3D bot
Original change's description:
> Remove code to push pixmaps to backend textures from GrGpu classes
>
> Replace GrGpu::updateBackendTexture with narrower method that clears
> a backend texture.
>
> Creation of data for a solid color compressed texture is lifted up to
> GrDirectContext and goes through updateCompressedBackendTexture.
>
> Bug: skia:11786
> Change-Id: I1d617623df5e65686f30e57c361a64f78d77f7bd
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/392836
> Reviewed-by: Jim Van Verth <jvanverth@google.com>
> Reviewed-by: Greg Daniel <egdaniel@google.com>
> Commit-Queue: Brian Salomon <bsalomon@google.com>
TBR=egdaniel@google.com,jvanverth@google.com,bsalomon@google.com
Change-Id: Ie58f52245c44c77f09742b0cb590cc97b97e6e37
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:11786
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/396097
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
We are up to having seven distinct types of codegen, and will soon have
an 8th (DSL C++).
Change-Id: I6758328390c234ba1d5c30c118199dbc820af52a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395817
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Change-Id: I8ca4fbba551624c61a4ebe7a4716750fe0b48196
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395818
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Change-Id: I047c4397fba87ce5db6f422162f942ac0ca0b0b9
Bug: skia:10446
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395816
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Add a Surface abstraction to manage the canvas backing store (equivalent
to SkSurface).
This allows relieving Canvas of buffer management duties - we can now
focus solely on rendering APIs at this level.
At the moment, only a Bitmap-backed surface is implemented -- but other
native surface types will be added.
Also relocate native code to 'src'.
Change-Id: I46738472536cf24524428dbd36969f2b99d251e3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395536
Reviewed-by: Jorge Betancourt <jmbetancourt@google.com>
Commit-Queue: Florin Malita <fmalita@google.com>
Replace GrGpu::updateBackendTexture with narrower method that clears
a backend texture.
Creation of data for a solid color compressed texture is lifted up to
GrDirectContext and goes through updateCompressedBackendTexture.
Bug: skia:11786
Change-Id: I1d617623df5e65686f30e57c361a64f78d77f7bd
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/392836
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Noticed during code read-through; unaware of any actual issues caused by
this.
Change-Id: I455e7fbc3a1d83f6319aaed1518d382ec231c8be
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395836
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
f92eb43ce9..15f4925ac3
2021-04-13 nicolascapens@google.com Restore EGL_EXT_buffer_age for SwiftShader Vulkan
2021-04-13 syoussefi@chromium.org Vulkan: More perf warning for vertex format conversion
2021-04-12 doughorn@google.com Reland "Fix multithreaded crash on draw commands on D3D11 backend."
2021-04-12 sunnyps@chromium.org Ignore internal format attribute for YUV D3D11 EGLImages
2021-04-12 syoussefi@chromium.org Remove EvqBaseVertex and EvqBaseInstance qualifiers
2021-04-12 stevensd@chromium.org Ignore subclass when finding GPU PCI devices
2021-04-12 jmadill@chromium.org infra: Remove named vs_toolchain cache.
2021-04-12 jmadill@chromium.org DEPS: Include a pinned version of depot_tools.
2021-04-12 angle-autoroll@skia-public.iam.gserviceaccount.com Roll SwiftShader from 4c0f888d7ef8 to 77edad299907 (1 revision)
2021-04-12 angle-autoroll@skia-public.iam.gserviceaccount.com Roll Chromium from 920c6c5ffe65 to 497fa2f8d899 (507 revisions)
If this roll has caused a breakage, revert this CL and stop the roller
using the controls here:
https://autoroll.skia.org/r/angle-skia-autoroll
Please CC brianosman@google.com on the revert to ensure that a human
is aware of the problem.
To report a problem with the AutoRoller itself, please file a bug:
https://bugs.chromium.org/p/skia/issues/entry?template=Autoroller+Bug
Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+doc/master/autoroll/README.md
Cq-Include-Trybots: skia/skia.primary:Build-Debian10-Clang-x86_64-Release-ANGLE;skia/skia.primary:Test-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-Golo-GPU-QuadroP400-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC8i5BEK-GPU-IntelIris655-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE
Tbr: brianosman@google.com
Change-Id: Idb39e313dd004dee1fd4b0d14d7d4b0c75bda30e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395787
Reviewed-by: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Not a big deal, just a tiny simplification.
Change-Id: I34e1404d79e22449f19a22ccf2cf6ffd98acf879
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395727
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Bug: skia:11844
Change-Id: I959966d9c2366c249bbdb3b78d47d0e29e9e43f1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395719
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Also add unit test of all GLSL type aliases.
Bug: skia:10679
Change-Id: I93e21621c11adfe3f114d0c55fb8043518e62696
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395718
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Allows GrProcessors to change behavior based on cap.
Bug: skia:11844
Change-Id: I4378c47c50a9bf33fa7461c6b9c522413e932bcd
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395717
Commit-Queue: Brian Salomon <bsalomon@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Auto-Submit: Brian Salomon <bsalomon@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
In http://review.skia.org/393397, I replaced a Constructor::Convert call
with a call directly to ConstructorCompoundCast::Make.
This worked fine if the input expression was actually a compound, but
if it was not, the code would assert/crash. The fuzzer detected this
error right away. (Enums are not considered to be a scalar, a vector or
a matrix in SkSL.)
Change-Id: Ie0df4c5771ff4f4d8f5251d4703e9c3516b6baad
Bug: oss-fuzz:33113
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395720
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Most of the COLRv1 paint operations only need the clip and matrix to be
kept on SkCanvas, only PaintComposite needs a full saveLayer().
Adjust calls accordingly.
Bug: skia:11851
Change-Id: I85b3c27495c78306457e17fea8062c957e2f3b05
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395617
Auto-Submit: Dominik Röttsches <drott@chromium.org>
Commit-Queue: Ben Wagner <bungeman@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
This was not used anywhere. (u_skRTHeight, on the other hand, is used to
support Y-flipped surfaces.)
Change-Id: I0cb803e8f82834715a47366a00c8ec913ee3c6e5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/394898
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
The persistent cache stores the struct `SkSL::Program::Inputs` on disk
as a binary blob. If the struct changes size or shape in any way, this
causes a compatibility break in the cache files and we need to increment
kCurrentVersion to invalidate them. We will now automatically detect any
change to Inputs which causes its size to differ and static_assert.
Change-Id: Ie8b9958029c209f2c763f5bea78fcaa8bc3444d7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/394897
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This code is not performance-sensitive, and there doesn't seem to be any
benefit to making it header-only.
Change-Id: Ic83be5cbb96f08c18a144954be7453aaabaa5a72
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/394900
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
ConstantFolder::Simplify already makes this check internally (see
SkSLConstantFolder:387). We don't need to do it here as well.
The `errors/DivideByZero.sksl` test reports the same errors as before.
Change-Id: If82d24b3c8613574b069cce0116aad4e789964c4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395416
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Two of the tests are now (correctly) detected as invalid code, but were
previously checked into /shared/ because they compiled when the fuzzer
first reported them (that is, after the crash was fixed). These have
been moved into the /errors/ test folder.
One of the tests was only generating an error because its main function
was named `a` instead of `main`, so I renamed it to `main`.
Change-Id: I1a2346fb16e304b0c66ff377a3f9bf7e7ee89ba9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/394899
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Bug: skia:11850
Change-Id: I3eed3d07481b6dd27e3c51d25d64b69ee8a16280
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395537
Reviewed-by: Ravi Mistry <rmistry@google.com>
Commit-Queue: Eric Boren <borenet@google.com>
Pinpoint discovered some regressions in Block::Make.
Block::Make attempts to identify Blocks that contain zero or one actual
Statements; in that case it can avoid allocating a Block and just return
the Statement. The original logic which did this check worked by
removing the empty Statements from the array, then shrinking the array.
For a large array of nested Statements, this could end up taking a
nontrivial amount of time as it would scan every statement for
emptiness.
I've rewritten this logic to be more efficient; it now stops as soon as
two non-empty Statements are found, and makes no attempt to eliminate
empty statements. It's more lines of code, but much less work.
If this doesn't resolve Pinpoint's woes we can remove this logic
altogether, as its only value is to avoid an unnecessary allocation and
generate slightly-smaller but overall equivalent IR.
Change-Id: I2d18a0e09df266629d63b6796f40501d8610f756
Bug: chromium:1187973, chromium:1187991, chromium:1188002, chromium:1188025
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395417
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
4340fcfaa8..f92eb43ce9
2021-04-11 m.maiya@samsung.com Vulkan: All builtins of gl_out needs to be considered as active
2021-04-10 syoussefi@chromium.org Vulkan: Use RGBA16_FLOAT for buffer format fallback
2021-04-10 ianelliott@google.com Vulkan: damage rectangles must never be pre-rotated
2021-04-10 ianelliott@google.com Vulkan: Warn for unsupported vertex attrib format
2021-04-10 syoussefi@chromium.org Remove unused translator option bits
2021-04-10 ynovikov@chromium.org Really fix iOS build
2021-04-09 ianelliott@google.com Add TextureType::Buffer to RecordBindTextureTypeError()
2021-04-09 jmadill@chromium.org Android: Call getExternalStorageDirectory natively.
2021-04-09 ynovikov@chromium.org Skip TracePerfTest.Run/native_real_gangster_crime on Linux AMD
2021-04-09 syoussefi@chromium.org Add tests that use, then update, then use buffers
2021-04-09 cclao@google.com Vulkan: Add featue flag to force highp to mediump in fragment shader.
2021-04-09 timvp@google.com Fix angle_expose_non_conformant_extensions_and_versions
2021-04-09 m.maiya@samsung.com Consider AOSP ANGLE path as well during SHA generation
2021-04-09 syoussefi@chromium.org Translator: Validate multi-declarations are separated
2021-04-09 syoussefi@chromium.org Remove unused translator option bit
2021-04-09 angle-autoroll@skia-public.iam.gserviceaccount.com Roll Chromium from 6cad7c765c8b to 920c6c5ffe65 (341 revisions)
If this roll has caused a breakage, revert this CL and stop the roller
using the controls here:
https://autoroll.skia.org/r/angle-skia-autoroll
Please CC brianosman@google.com on the revert to ensure that a human
is aware of the problem.
To report a problem with the AutoRoller itself, please file a bug:
https://bugs.chromium.org/p/skia/issues/entry?template=Autoroller+Bug
Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+doc/master/autoroll/README.md
Cq-Include-Trybots: skia/skia.primary:Build-Debian10-Clang-x86_64-Release-ANGLE;skia/skia.primary:Test-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-Golo-GPU-QuadroP400-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC8i5BEK-GPU-IntelIris655-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE
Tbr: brianosman@google.com
Test: Test: dEQP-GLES31.functional.tessellation.shader_input_output.gl_position_tcs_to_tes
Change-Id: I70b44bfb7334cbb9a4e06c80c171ff65a4443637
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/394970
Reviewed-by: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
In this CL we don't actually bail and use the record-time-ordered
DAG in this case, but we at least will see which perf bots with
reordering turned on are going over budget.
Bug: skia:10877
Change-Id: I38666166f15a7c168eadf262f7e975822c4308db
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/394158
Commit-Queue: Adlai Holler <adlai@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
This allows the caller to decide whether a plan is feasible before
committing to it. In this CL the drawing manager doesn't actually
call it, but we test it. We'll call it in a follow-up CL
Bug: skia:10877
Change-Id: Ie3a6c14a0196f595c522a0c961aba7b10c980711
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/394157
Commit-Queue: Adlai Holler <adlai@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
A follow-up CL allows the user to check the budget headroom
in between these two steps.
Bug: skia:10877
Change-Id: Ie0e6f23029065158b2eadb0dc26caf32f54433e1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/394999
Commit-Queue: Adlai Holler <adlai@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
These builtins have been obsolete for a long time; they were only
implemented on the GLSL backend. These values are provided in
u_skRTHeight and u_skRTWidth instead. (See GrGLSLProgramBuilder's
addRTHeightUniform and addRTWidthUniform.)
Change-Id: I8cceca348cbf9071939618f913693c316d35dbc6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395001
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Encode the RSXForm text blob data into the SkGlyphRuns. Change
SkBaseDevice's drawGlyphRunList() to break up glyph runs with
TSXForm information into single glyph runs. Add
a virtual onDrawGlyphRunList() = 0 so drawGlyphRunList can use
it to render SkGlyphRunLists per device.
Change-Id: I7250460f7b4be2f820e4bb8ca4223654a9d9d145
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/395098
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Currently doesn't do anything other than add new configs that set
a GrContextOption that becomes available in GrCaps. Runs new configs
on Perf/Test bots (MTL/iPhone11 and GL/Ubuntu).
Bug: skia:11844
Change-Id: I58586cae0980e52701abd1633dbc79b381f6015b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/394996
Reviewed-by: Eric Boren <borenet@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
