If we start with an OpConstantComposite, then we do an
OpCompositeExtract from it, we can look up the result directly from op
cache and avoid doing any work. This helps our matrix code a lot.
Change-Id: Idfbdc0c69676b9c1e91cdc57bf0d6382b9b5d8d4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529339
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Many chromium adjacent projects have started using this repo to fetch
all of the Vulkan and SPIRV repos in a single bundle. Using this ensures
that we get a group of repos that should work together (atomically), and
simplifies autoroller configuration (we just need one roller for
vulkan-deps that does transitive rolls of the children, rather than five
separate rollers).
Change-Id: I015edb5811be60eb2275cdfe195138317d341a03
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529730
Reviewed-by: Eric Boren <borenet@google.com>
This library is now used by Dawn, so will be required to get the roller
working again. Landing this separately (and unused), first.
Change-Id: I22a820e4996299ccdc9181ac96fb617eb7933c5b
Bug: dawn:1365
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529725
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: Brian Osman <brianosman@google.com>
Reviewed-by: Eric Boren <borenet@google.com>
Commit-Queue: Eric Boren <borenet@google.com>
PaintRotate needs to rotate counter-clockwise, but was rotating
clockwise due to passing the font's value unchanged to
SkMatrix::RotateDeg. Negate the value to achieve counter-clockwise
rotation.
For PaintSkew, we compute the tangent from the values in the font and
set up a skew matrix with that. Since we're using the trigonometric
functions to set up the matrix, the angle works in counter-clockwise
direction.
So this previously worked correctly for skew in the x
direction, but not in the y-direction due Skia using a y-down coordinate
system. Negate the y-angle from PaintSkew before calculating the
tangent and setting up the matrix.
Bug: skia:13190
Cq-Include-Trybots: luci.skia.skia.primary:Test-Android-Clang-GalaxyS20-GPU-MaliG77-arm64-Release-All-Android_NativeFonts
Change-Id: I98329f9746bcc6e875c4f87982782470e0f0a17d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529076
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Dominik Röttsches <drott@google.com>
Add dimensions to the glyph digest to enable SkRemoteGlyphCache to build
SubRuns directly.
It would be nice to use SkGlyphRect to encode this information but,
the MSVC-2019 compiler has problems producing correct code on 32-bit
machines.
Change-Id: I519ffecf3c727f608a8e6972742aee04b65a283a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529613
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
https://en.cppreference.com/w/cpp/container/span/subspan
The real std::span::subspan in C++20 allows the "count" argument to
default to `std::dynamic_extent` (a fancy way of spelling ~0U). I didn't
think it would be worth adding `skstd::dynamic_extent`, but I did have
a use for an unbounded subspan, so I added a single-argument version to
SkSpan.
Change-Id: I297cc452cf2db727a3f9869ff8f46f3527e19370
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529603
Reviewed-by: Arman Uguray <armansito@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
The field SkGlyph::fForceBW is currently used by the DirectWrite
SkScalerContext implementation to let generateImage know that
generateMetrics needed to fall back to aliased rendering in order to
load the glyph. This is part of a more general issue where
generateMetrics must generate the metrics based on some glyph
representation and then generateImage must use that same representation
to draw.
Currently this consistency comes from ensuring that generateMetrics and
generateImage make the same decisions, with the only real information
passed from generateMetrics to generateImage being the mask format.
However, there have always been multiple glyph represenations for a
given mask format (like outline masks vs. embedded bitmap masks) and now
there are four (soon to be five) different color formats. While the mask
format and bounds are enough to allocate the space for a glyph
rendering, this is not enough information for generateMetrics and
generateImage to agree on which glyph rendering to choose. The
additional required information has come from implicitly coding
generateImage to make the same decisions generateMetrics would have.
However, these decisions may have been based on information which is
expensive to compute, like probing the font for which representations a
glyph may have.
To transmit this extra information, formalize that generateMetrics may
state which decisions it made (which glyph representation it used) so
that generateImage can use this information explicitly instead of
implicitly. Do this by expanding fForceBW and renaming it
fScalerContextBits and documenting what it is used for.
Change-Id: I91ba747233447f4d7bea81b0dcde51e49fa705fe
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529604
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
The invariant for drawing is that this strike must be pinned in
the cache. If it is not, then dump a stacktrace.
Bug: chromium:1315730
Change-Id: I73a79d8d3c45262753ef64b679e4af23cd1d6eea
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529557
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Based on some code archaeology, we believe that textures (not samplers)
should always be used for subpass inputs.
Change-Id: Ida461e20f3b4f236918e72076e9c008ac280b8ef
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529609
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Previously, we didn't usually generate OpConstantComposite ops for
matrices. Now, a matrix assembled from constants should come out as a
constant.
Change-Id: I458718901686dffb84e4079a81017d61195420d3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529338
Reviewed-by: Arman Uguray <armansito@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
The output changes here are almost entirely a wash, because we already
had support for caching scalars and vectors. Almost all changes are just
inconsequential reorderings of IDs, and the removal of RelaxedPrecision
decorators on constants (which were not meaningful).
Change-Id: I45340c4a240cb504b7c4a934b3db178d2f39ec99
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/528709
Reviewed-by: Arman Uguray <armansito@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
This reverts commit acb8770918.
Reason for revert: possibly breaking g3 roll
Original change's description:
> skif:LayerSpace<SkRect>::roundOut/In have epsilon tolerance.
>
> Adds a little tolerance so that e.g. left=30.999994 with roundOut
> will still round to 31 not 30. Helps avoid cases where imprecision
> leads to including an entire unwanted row/column of an input image
> to an image filter.
>
> This Chrome change must land first:
> https://chromium-review.googlesource.com/c/chromium/src/+/3577185/
>
> Bug: chromium:1313579
> Change-Id: I143c8f99b90413a6b610f2b3a5e54e648777ca68
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/528652
> Reviewed-by: Michael Ludwig <michaelludwig@google.com>
> Commit-Queue: Brian Salomon <bsalomon@google.com>
Bug: chromium:1313579
Change-Id: Ia5589858afb042fae142357983f5ccf50ecc0020
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529628
Auto-Submit: Greg Daniel <egdaniel@google.com>
Commit-Queue: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
We have a new `writeInstruction` call which can cache instructions as
they are written. This allows us to reuse existing instructions instead
of recomputing the same things twice.
This feature is largely untapped in this CL. As a demonstration that it
works, we no longer check the `fNumberConstants` map when writing
numbers, and instead rely on the cache to deduplicate OpConstant
instructions. Followup CLs will leverage the feature more heavily.
Change-Id: I6c25d0c2db7888ace05c24f8f9961d25c91b93b5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/528817
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Arman Uguray <armansito@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Evolution over time led to a bit of gnarly logic around sampler
creation in SPIR-V. I've untangled it a bit.
Change-Id: I1987dd9b06c66f55051d72a6de4e950813697dbb
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529499
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Historically this code generator has used -1 as a sentinel "invalid"
SpvID. It's clearer to give this concept a name. The "NA" name is
borrowed from SkVM, and existing code using -1 has been updated.
Change-Id: I1733df4fb0bd440f849390046a7f26973e518976
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529519
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Arman Uguray <armansito@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Add dimensions to the glyph digest to enable SkRemoteGlyphCache to build
SubRuns directly.
Bug: skia:13192
Change-Id: I8e68fe4ef144344347d4bc23865821d8517709fc
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529517
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Bug: skia:12698
Change-Id: Ibd09f64d2f8690719d0cb6ed1f34abf6cdcf4497
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/527916
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Bug fix for https://skia-review.googlesource.com/c/skia/+/529126
This only affects Graphite, so there was no need to revert and fix.
Bug: skia:12787
Change-Id: Iab20b26e90083f1659d0c6502bd8e2d85ecb6fec
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529518
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Auto-Submit: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This will allow `is_globally_reachable_op` to be used in other places.
Change-Id: I99ea05b0a1da685a24d96700b395b4d13df0993e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529122
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Change-Id: Ia799cdff5288efe5d5d53e8d8f77cf32f3343371
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529131
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
In profiling, calculating the inverse of the 4x4 every draw call or on
every clip operation was measurable, but it really only needs to be
done if the transform is stale before the next time it's used.
Code is a little cleaner now too, since we're staying skgpu:: types
more than before.
Bug: skia:12787
Change-Id: Ib6313f211c8ebd6a484878bf9975e5ad6a4e750f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529126
Reviewed-by: Robert Phillips <robertphillips@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Rearrange SkGlyphRect so that leftTop is negative. This makes
calculating width and height = rightBottom + leftTop. Then we
can use this as the dimensions of the glyph in the digest.
Rename: topLeft to leftTop.
Bug: skia:13192
Change-Id: If06ebf6e595f380f8fbb9a3381e87945e312eceb
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529336
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Move the packed glyph id out of the digest to make room for the
glyph dimensions. Use SkTHashMap instead of SkTHashTable. Add a
Hash function struct to SkPackedGlyphID.
Bug: skia:13192
Change-Id: I1b59b47cb292402bf8b8e892b3b37edad9ecb9ea
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529132
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Change-Id: I126bee10ebeb34c024c84e5cd6bc5db950a82029
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529354
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Commit-Queue: Erik Rose <erikrose@google.com>
This removes the required dependency on our JSON code. In the Bazel
rules, this dependency is pushed down into sksl instead of required
by the cc_binary rules.
It adds a stub version of SkVMDebugTrace.cpp and removes
SkVMDebugTracePlayer unless the appropriate GN or Bazel flag
is set (skia_enable_sksl_tracing and enable_sksl_tracing,
respectively). There was an existing #define that CanvasKit
used (CK_INCLUDE_SKSL_TRACE) and this was changed to
SKSL_ENABLE_TRACING.
Users of //:skia_core no longer need to specify a JSON dep,
if sksl needs it (e.g. for tracing), then it will specify
the dependency.
This is a reland of https://skia-review.googlesource.com/c/skia/+/528837
Bug: skia:12541
Change-Id: I79612c69fdbefd3db9822a2b66df7552f7c13865
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529278
Reviewed-by: John Stiles <johnstiles@google.com>
Adds a little tolerance so that e.g. left=30.999994 with roundOut
will still round to 31 not 30. Helps avoid cases where imprecision
leads to including an entire unwanted row/column of an input image
to an image filter.
This Chrome change must land first:
https://chromium-review.googlesource.com/c/chromium/src/+/3577185/
Bug: chromium:1313579
Change-Id: I143c8f99b90413a6b610f2b3a5e54e648777ca68
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/528652
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Make the glyph buffers use the same Direct position calculations
for Bitmap and GPU processing. This will create fewer changes when
switching to SkGlyphDigest based API.
Bug: skia:13192
Change-Id: I7e3e44dcfc1a4bad014d0ebe2bef3c1b28c712f4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529277
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Bug: skia:12701
Change-Id: Ib02db7160599d3c15a01597c746ce5131f2998e0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/526298
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Stop attempting to draw the glyph on any error. Clean up some
indentation and naming. Ensure that out parameters are always set
correctly.
Change-Id: I0e5c6c1da698393684f76413c4dbb156803f508a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/528880
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
This is a reland of commit ae5e846047
Original change's description:
> [graphite] Move Graphite into Skia base directories.
>
> Change-Id: Ie0fb74f3766a8b33387c145bd1151344c25808cb
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/528708
> Reviewed-by: Kevin Lubick <kjlubick@google.com>
> Reviewed-by: Greg Daniel <egdaniel@google.com>
> Commit-Queue: Jim Van Verth <jvanverth@google.com>
Change-Id: Ia575fd49206ad0b665a6a9153317e738bb321446
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529059
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
* Wire up the WGSLCodeGenerator to SkSLCompiler.
* Wire up build rules to generate WGSL from unit tests.
* Include HelloWorld.sksl as the first complete program.
Bug: skia:13092
Change-Id: I283cf5971b6856126b9fc23340afacff5cc54697
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/526760
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
Commit-Queue: Arman Uguray <armansito@google.com>
Bug: skia:12701
Change-Id: Iaac8b6fd0f2d9d1f8d5771f8204047a9127184c1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/528041
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This reverts commit 6bc4bdf645.
Reason for revert: Android roll
Original change's description:
> [sksl] Make sksl tracing optional
>
> This removes the required dependency on our JSON code. In the Bazel
> rules, this dependency is pushed down into sksl instead of required
> by the cc_binary rules.
>
> It adds a stub version of SkVMDebugTrace.cpp and removes
> SkVMDebugTracePlayer unless the appropriate GN or Bazel flag
> is set (skia_enable_sksl_tracing and enable_sksl_tracing,
> respectively). There was an existing #define that CanvasKit
> used (CK_INCLUDE_SKSL_TRACE) and this was changed to
> SKSL_ENABLE_TRACING.
>
> Users of //:skia_core no longer need to specify a JSON dep,
> if sksl needs it (e.g. for tracing), then it will specify
> the dependency.
>
> Change-Id: I2fcd29cde118fc391c269ba2d8f8a40a6f164c99
> Bug: skia:12541
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/528837
> Reviewed-by: John Stiles <johnstiles@google.com>
> Commit-Queue: Kevin Lubick <kjlubick@google.com>
Bug: skia:12541
Change-Id: Icf75495f19e409d96925ca4dca9e839eca4057ec
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/529129
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
