Bug: skia:6718
Change-Id: I4d9c31bc0ca42a80e652de29462531587efef031
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/248257
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
This defers the decision about whether pointSize is emitted until down in GrProgramDesc::Build. Although a bit wonky in GL's case, this anticipates the world where each cluster of meshes always has the same GrPrimitiveType.
Ultimately, we would like the GrPrimitiveType to be part of the GrProgramInfo and known far earlier.
Change-Id: I65566b2d77eab6a69f7af5aa60a069299b58db65
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246018
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
The dispatch_semaphore_t is a higher-level, more-efficient semaphore
primitive if the cross-process capabilities of semaphore_t are not
needed.
Bug: chromium:1012386
Change-Id: I7c39fa6b8c015bdf7b4b2275da8165165253c2d1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/248259
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Bug: skia:6718
Change-Id: I0f9c7d89e732c59bdbe76ab45d42934029106efb
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/247684
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
This is still pretty far away from this goal but, the goal here is to be able to create a program w/o requiring instantiated proxies.
Bug: skia:9455
Change-Id: I9526a3a737438c5331fef4a621be4786ef43c080
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/247397
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
This improves the encapsulation of the GrProgramInfo class.
Bug: skia:9455
Change-Id: Ic241ba4a2c2edea3e774522065371e5ed8cc2da3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/247438
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Perspective glyphs can vary in screen size so it's unclear which SDF
level is best, and even if we choose one for an entire subrun it's
possible that a given glyph will have artifacts if it's too big.
Instead we fall back to paths unless specifically requested to
support perspective SDF.
Bug: skia:9515
Change-Id: I75ac1d60f2eaa0e63de201b5db36fe0485fff463
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/247398
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Herb Derby <herb@google.com>
This is a deferred TODO/code review request. In general, this improves the encapsulation of program information w/in GrProgramInfo. Additionally, it is desirable to stop using the Desc/Key to access this information.
Bug: skia:9455
Change-Id: I854ff3225a49657d3a523e88e4bfa549128ee486
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/247336
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This reverts commit 76826fc91c.
Reason for revert: Missing some characters
Original change's description:
> Draw perspective text as paths.
>
> Perspective glyphs can vary in screen size so it's unclear which SDF
> level is best, and even if we choose one for an entire subrun it's
> possible that a given glyph will have artifacts if it's too big.
> Instead we fall back to paths.
>
> Bug: skia:9515
> Change-Id: I88f03b25651df0222459f5dbd03eee9465b97487
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/247437
> Commit-Queue: Herb Derby <herb@google.com>
> Reviewed-by: Jim Van Verth <jvanverth@google.com>
TBR=jvanverth@google.com,herb@google.com
Change-Id: I5e39566c35b49323913eb72ff89ecbc91faac8ce
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:9515
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/247462
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Perspective glyphs can vary in screen size so it's unclear which SDF
level is best, and even if we choose one for an entire subrun it's
possible that a given glyph will have artifacts if it's too big.
Instead we fall back to paths.
Bug: skia:9515
Change-Id: I88f03b25651df0222459f5dbd03eee9465b97487
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/247437
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
SkGlyphPositionRoundingSpec encapsulates the information needed for rounding
positions correctly. This allows test to avoid using an SkStrike in future
unit tests.
This is part of a major rewrite:
https://skia-review.googlesource.com/c/skia/+/238196
Change-Id: I18dd6fe394df0fcbda492cefe2484c9bf9bb40d1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/247338
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Although not necessary I would like to centralize/encapsulate the error checking so there is just one place to go to update/extend this code. It does have the nice property that, since we validate in the ctor and this object is immutable, we only have to do it once (and can delete some code spread throughout the code base).
Bug: skia:9455
Change-Id: Iaa26c7a896a9bd053358a6f8424d7f5944b0117e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/247339
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This api will replace ComputeTextureSize.
Change-Id: I9befd8706f44250805315bdb13a559cf81c7142d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/247337
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
There exist 'tricky' fonts which use bytecode hints to move all parts of
a glyph for legibility, not just to slightly tweak the outlines. Until
it is possible to detect these 'tricky' fonts, retain all font's
bytecode hints.
Bug: chromium:1010396
Change-Id: I4eaca967da1a853c8061a4997ce8b872a53fd2f1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/247298
Reviewed-by: Hal Canary <halcanary@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
We want to be able to pull more work forward when creating DDLs. The prePrepare entry points will allow us to perform this CPU-side preprocessing.
Change-Id: I2c0c7978dbf7d7c335425ea81ab2323cb9dcdbc3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/247048
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
This is the first step in moving the marshaling of program information earlier in renderTask processing (i.e., to onPrePrepare).
Change-Id: I91e3baed9a128e845bd32f9dbbacd9b21d852a3d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/244118
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Bug: skia:6718
Change-Id: Id611e0f5c317908cb63f1afdafff28ebf6cf8158
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/247049
Commit-Queue: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
When the CTM contained a "mirroring" operation - for example
scale(-1, 1) or scale(1, -1) - the radius of the morphology filter would
end up negative, essentially disabling the filter.
When mapping the radii, use the magnitude of a vector built from each
component of the radii instead of considering the radii as a plain
vector.
Bug: chromium:595244
Change-Id: Ia249f8962648f0a4e69b6d0becf87ee1ba6be9a6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246996
Reviewed-by: Florin Malita <fmalita@chromium.org>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Florin Malita <fmalita@chromium.org>
Bug: skia:6718
Change-Id: I1d540042d8b69393d1676d0f0cedca7f37ac88d7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246852
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Although shouldAlwaysUseDedicatedImageMemory() returns true for
protected images, this does not take effect because of the error
in if-else logic.
Bug: skia:9016
Change-Id: Iffdfdf5d045b38d846d438969747ded617d96739
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246916
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Add first, a ctor that will make an SkZip with more constness,
a safe default ctor, and allow SkSpan to be used as r-values.
Change-Id: Iaa89eee3a8994c3a0bd1def1a41e0f066dc872b9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246822
Commit-Queue: Herb Derby <herb@google.com>
Auto-Submit: Herb Derby <herb@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
With these changes we can build across all versions of Metal, and for
the most part run on older OSes (further testing is necessary).
Bug: skia:9408
Change-Id: I00ee03b9a3a0e49483af4a610544b20d482a0cdd
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/245957
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
kLinesAdjacency is no longer used while kPath has historically been left out.
Change-Id: I84be64a2e8ec79ff299a2bced73ded0be04df38e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246307
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Since we want to collect shader information early, this attempts to make clear that numSamples, origin and outputSwizzle can all be obtained from the target renderTargetProxy.
Change-Id: I42e0fd79e2163f17673ccdd344a31fbaadac5f53
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246298
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This is an alternative approach to https://skia-review.googlesource.com/c/skia/+/234896
It inverts the templates and coordinate spaces, so that each coordinate space is
defined by a struct that accepts the SkIRect/etc. type as an argument. For parameter
space and device space, the only exposed functionality is holding on to the coordinates
and moving into and out of the layer space (using the Mapping type, which now encapsulates
the definition of parameter, layer, and device space for a given CTM).
The LayerSpace struct has specialized definitions for each of the 6 key geometric types,
exposes accessors/mutators and operations like join/intersect that work on appropriate
layer-space geometry.
FilterResult and Context have been updated to use these new types, although they no
longer take a Usage enum. If this strategy works, FilterResult will be updated to have
more utilities to operate and read from its image in layer-space so there won't be a need
to have the image space (and the usage enum to differentiate between different image
spaces).
To further reinforce the differences between getInputBounds() and getOutputBounds() (which
previously used Usage to do this), they now accept/return values in device and parameter
space. This is no different than what SkCanvas would have done before calling the old
functions, but now that work is owned by SkImageFilter (seems like a good thing), and
achieves the same reinforcement.
Change-Id: I8bebb9fadf6c8f2bd51fa863b2d6f2e4a33dd08b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/244515
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
bug: skia:9503
Change-Id: I5b4e2434c6085d048d6d0fc42b2b928439b1066b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246296
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
This is a reland of 1e946aa24b
Original change's description:
> Add less granular version of GrFence for older Metal versions.
>
> Rather than use MTLSharedEvent, we create a semaphore and
> signal it within the current command buffer's completion handler.
> The pointer to the semaphore is cast as GrFence and returned.
> The waitFence method recasts it and waits on the semaphore.
>
> Bug: skia:8243
> Change-Id: I2c854c65a62786fe8031794008eb6844974ebefd
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/245583
> Reviewed-by: Greg Daniel <egdaniel@google.com>
> Commit-Queue: Jim Van Verth <jvanverth@google.com>
Bug: skia:8243
Change-Id: I63872c26b8f5c29cba6e618f418b291cf5ffac54
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246016
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Change-Id: I0e236bf14acf1a8b7df3433be05a192633096df0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246098
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This is a reland of 6fc04f88a8
Original change's description:
> Reland "SkSurface asynchronous read APIs allow client to extend pixel lifetime"
>
> This is a reland of ce240cc6fd
>
> Original change's description:
> > SkSurface asynchronous read APIs allow client to extend pixel lifetime
> >
> > Previously the pixel data passed to the client was only valid during
> > the client's callback. This meant if the client wanted to defer
> > processing of the data a copy was necessary.
> >
> > Now we pass an object to the callback and the pixel lifetime is tied
> > to the lifetime of that object.
> >
> > The object may be holding a GPU transfer buffer mapped. We don't assume
> > that the object will be released on the direct GrContext thread. So
> > when the object is destroyed it posts a message to a new type,
> > GrClientMappedBufferManager, hanging off the direct context. The direct
> > context will periodically check for messages and unmap and then unref
> > buffers so that they can be reused. Currently this is done in
> > GrContext::performDeferredCleanup() and GrDrawingManager::flush().
> >
> > The old API is kept around for backwards compatibility but it is
> > reimplemented as a bridge on top of the new mechanism.
> >
> > Also a utility function to SkImageInfo is added to directly make a new
> > info with a specified dimensions rather than passing the width and
> > height separately to makeWH().
> >
> > Bug: chromium:973403
> > Bug: skia:8962
> >
> > Change-Id: Id5cf04235376170142a48e90d3ecd13fd021a2a6
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/245457
> > Reviewed-by: Brian Osman <brianosman@google.com>
> > Commit-Queue: Brian Salomon <bsalomon@google.com>
>
> Bug: chromium:973403, skia:8962
> Change-Id: I5cecd36276c8b6dc942cf549c7095db2df88530c
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/245678
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Commit-Queue: Brian Salomon <bsalomon@google.com>
Bug: chromium:973403, skia:8962
Change-Id: Ie584c1c3ef8021c976f71b708e53871c693cc450
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246057
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Change-Id: I9b27c073122e615cf2db289de5607d6636658803
Bug: skia:6718
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246077
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
- The raster pipeline JIT was an experiment that never really panned
out (except for also introducing the interpreter, which survives).
The removal should be pretty thorough, including undoing some changes
to SkRasterPipeline itself. For reference, I basically undid most of
https://skia-review.googlesource.com/c/skia/+/112204/
- With runtime shaders, always convert to Program using kPipelineStage
This makes the first-pass compile consistent with runtime color filter,
and with GrSkSLFP.
- Remove sksl_mixer.inc, which hasn't been used in a while.
- In sksl_pipeline.inc, all of the math symbols are already declared
in sksl_gpu.inc, which forms the base symbol table when this include
is parsed.
sk_x and sk_y appear to never be used? Those builtin IDs are the IDs
of the x and y params to main, but that logic still works without
these declarations.
sk_OutColor only makes sense in FP files (it's still declared in
sksl_fp.inc).
Change-Id: Ie8dd68d2c4687745d46f96804a76695bce8c1ba2
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246017
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This reverts commit 35e0a58278.
Reason for revert: Breaking g3 roll
Unfortunately, the g3 tests call FcFini at the end, but FcFini asserts
that everything is really no longer in use (it doesn't just unref the
current FcConfig). This means that all local FcConfig references must
be destroyed before calling FcFini. However, one of these will be
rooted with a global, leading to destructor ordering issues.
Will need to see how to handle this.
Original change's description:
> Hold onto FcConfig in SkFontConfigInterface_direct.
>
> It is possible that some other thread will change the default FcConfig
> out from under one of this classes own calls, which is a terrible race.
> Prevent this by taking a reference to the default FcConfig at creation
> time and keeping it alive.
>
> Bug: chromium:1004254
> Change-Id: I23338c72e0b93859beae465cb6d48e87b3928ed3
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/245720
> Reviewed-by: Etienne Bergeron <etienneb@chromium.org>
> Reviewed-by: Herb Derby <herb@google.com>
> Commit-Queue: Ben Wagner <bungeman@google.com>
TBR=bungeman@google.com,herb@google.com,etienneb@chromium.org
Change-Id: If0f90c9060ff5c7a486f6d7ffe2b7b52a7ae31a4
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: chromium:1004254
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246078
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
Change-Id: Ie4653fce34e4d6f38613f63bc68ecc1b7bae361b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/246019
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
It is possible that some other thread will change the default FcConfig
out from under one of this classes own calls, which is a terrible race.
Prevent this by taking a reference to the default FcConfig at creation
time and keeping it alive.
Bug: chromium:1004254
Change-Id: I23338c72e0b93859beae465cb6d48e87b3928ed3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/245720
Reviewed-by: Etienne Bergeron <etienneb@chromium.org>
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
The idea here is to separate information we can have at opList time:
numSamples, origin, primProc, primProcProxies and pipeline
and what we can only have at flush time:
sample locations
The opList-available parameters will then be moved into a descriptor (GrProgramInfo).
Change-Id: Iae044a98a0cfc8685569c923d154a6f9dabfeb75
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/245621
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This reverts commit 6fc04f88a8.
Reason for revert: Chrome roll failure suspect because of:
* https://chromium-review.googlesource.com/c/chromium/src/+/1837131 (22 commits)
* https://chromium-review.googlesource.com/c/chromium/src/+/1837214 (24 commits)
Original change's description:
> Reland "SkSurface asynchronous read APIs allow client to extend pixel lifetime"
>
> This is a reland of ce240cc6fd
>
> Original change's description:
> > SkSurface asynchronous read APIs allow client to extend pixel lifetime
> >
> > Previously the pixel data passed to the client was only valid during
> > the client's callback. This meant if the client wanted to defer
> > processing of the data a copy was necessary.
> >
> > Now we pass an object to the callback and the pixel lifetime is tied
> > to the lifetime of that object.
> >
> > The object may be holding a GPU transfer buffer mapped. We don't assume
> > that the object will be released on the direct GrContext thread. So
> > when the object is destroyed it posts a message to a new type,
> > GrClientMappedBufferManager, hanging off the direct context. The direct
> > context will periodically check for messages and unmap and then unref
> > buffers so that they can be reused. Currently this is done in
> > GrContext::performDeferredCleanup() and GrDrawingManager::flush().
> >
> > The old API is kept around for backwards compatibility but it is
> > reimplemented as a bridge on top of the new mechanism.
> >
> > Also a utility function to SkImageInfo is added to directly make a new
> > info with a specified dimensions rather than passing the width and
> > height separately to makeWH().
> >
> > Bug: chromium:973403
> > Bug: skia:8962
> >
> > Change-Id: Id5cf04235376170142a48e90d3ecd13fd021a2a6
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/245457
> > Reviewed-by: Brian Osman <brianosman@google.com>
> > Commit-Queue: Brian Salomon <bsalomon@google.com>
>
> Bug: chromium:973403, skia:8962
> Change-Id: I5cecd36276c8b6dc942cf549c7095db2df88530c
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/245678
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Commit-Queue: Brian Salomon <bsalomon@google.com>
TBR=bsalomon@google.com,brianosman@google.com
Change-Id: I9e01d1b82fb399b94292441d91da51176bb161d9
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: chromium:973403, skia:8962
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/245956
Reviewed-by: Ravi Mistry <rmistry@google.com>
Commit-Queue: Ravi Mistry <rmistry@google.com>
This is a reland of ce240cc6fd
Original change's description:
> SkSurface asynchronous read APIs allow client to extend pixel lifetime
>
> Previously the pixel data passed to the client was only valid during
> the client's callback. This meant if the client wanted to defer
> processing of the data a copy was necessary.
>
> Now we pass an object to the callback and the pixel lifetime is tied
> to the lifetime of that object.
>
> The object may be holding a GPU transfer buffer mapped. We don't assume
> that the object will be released on the direct GrContext thread. So
> when the object is destroyed it posts a message to a new type,
> GrClientMappedBufferManager, hanging off the direct context. The direct
> context will periodically check for messages and unmap and then unref
> buffers so that they can be reused. Currently this is done in
> GrContext::performDeferredCleanup() and GrDrawingManager::flush().
>
> The old API is kept around for backwards compatibility but it is
> reimplemented as a bridge on top of the new mechanism.
>
> Also a utility function to SkImageInfo is added to directly make a new
> info with a specified dimensions rather than passing the width and
> height separately to makeWH().
>
> Bug: chromium:973403
> Bug: skia:8962
>
> Change-Id: Id5cf04235376170142a48e90d3ecd13fd021a2a6
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/245457
> Reviewed-by: Brian Osman <brianosman@google.com>
> Commit-Queue: Brian Salomon <bsalomon@google.com>
Bug: chromium:973403, skia:8962
Change-Id: I5cecd36276c8b6dc942cf549c7095db2df88530c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/245678
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
