Support for constant-expression function calls in SkSL now exists, and
support for abs() was added at http://review.skia.org/405676.
Change-Id: I3144af993db93a3d640971734d4cb03e0cfb8589
Bug: skia:10835
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/408642
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This avoids redundant code, and has a small side benefit of
deduplicating constant vectors which appear more than once in the code,
since `writeConstructorSplat` already supports this.
Change-Id: I2972ee922ac92adeb40bc765da3b490a59b957b3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/408360
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
This is a clone of the Metal test. (This can be moved into shared/ and
enabled as a real test once the codegen is fixed.)
At present, this test generates broken code; everything is writing an
SpvOpMatrixTimesScalar opcode regardless of the actual operation being
performed.
Change-Id: If06b4196e7d9be36e41c5c60c006b2a713cc25d8
Bug: skia:11985
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/408297
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
We were emitting this at global scope (not in Globals). That would lead
to errors about the variable needing to be in the constant address
space. (You can see the result in ConstArray.metal - the old code was
invalid). Also, we were already making references use _globals, so the
code was double-wrong (or half-right, depending on your perspective).
After the core change, writeVarDeclaration was only used for local
scope, and writeModifiers never used the 'globalContext' parameter.
The removal of finishLine() changed every test output, unfortunately.
Change-Id: Icc1356ba2cc3c339b2f5759b3d18523fd39395bc
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/408356
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
There were two issues leading to the corruption seen in the linked
chromium issue.
1. The draw's bounds were calculated based on the quad being clipped
to w>= epsilon, which is what happens when the AA inset/outset is done.
But for non-aa quads, the fillrect and texture ops did no clipping,
assuming that the GPU would be sufficient. However, this can produce
non-aa draws that exceed the calculated bounds, misleading the clip
stack into incorrectly removing the scissor, etc.
2. Precision issues within CropToRect meant some perspective quads'
barycentric coordinates would become degenerate and compute to (0,0,1),
making it appear as if the render target/scissor were contained within
it. This meant we'd turn it into a rectangular clear.
These changes appear to address the corruption on Linux and Windows, but
there are still rendering artifacts from poor aa inset/outset
calculations. These artifacts are at least limited to the clip properly.
A better rendering method that does not rely on line intersections
will address these artifacts, but this CL is a reasonable temporary
mitigation.
Bug: chromium:1204347
Change-Id: I3c67d4efe70313ae7c98abc0a57b5b047c83890d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/407821
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
This CL adds a polyfill for componentwise matrix/matrix division to
Metal, as well as matrix/=matrix. Matrix/scalar and scalar/matrix
division work by splatting the scalar out to a matrix (handled in the
prior CL, http://review.skia.org/407616) and then performing
componentwise matrix/matrix division.
Working demonstration (copy-pasted from the Metal output file):
http://screen/BrqyPcbPrB7Dy4m
Change-Id: I6a8b97783be3485f7ffee551b669d14bc58e7568
Bug: skia:11125
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/407796
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
GrCullTest quickly determines whether a set of points will be visible
or not. It will be used to guard against exponential recursion when
chopping paths at an extreme zoom level.
Bug: skia:10419
Change-Id: I85183826d820bbca0349a1e794b0696f776b134b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/407297
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Herb Derby <herb@google.com>
This reverts commit 90508f02dc.
Reason for revert: avoiding driver bugs this time
Original change's description:
> Revert "The Matrices test now verifies its results."
>
> This reverts commit 86121f6c0e.
>
> Reason for revert: tree sad
>
> Original change's description:
> > The Matrices test now verifies its results.
> >
> > Previously, this test did a bunch of matrix math but never actually
> > checked its results for correctness.
> >
> > Change-Id: I353be58049286266c2d561b0939b3874d2684403
> > Bug: skia:11985
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/407360
> > 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>
>
> TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com
>
> Change-Id: I1335f01c14ee955426e02efaa3c30421cd41aa34
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: skia:11985
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/407617
> Reviewed-by: John Stiles <johnstiles@google.com>
> Commit-Queue: John Stiles <johnstiles@google.com>
Bug: skia:11985
Change-Id: I214375d74977f324973da72c440d7ff5ff179016
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/408157
Reviewed-by: John Stiles <johnstiles@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
The Metal code generator will now detect matrix-op-scalar expressions
and splat the scalar across a matrix. This allows a scalar to be added
to, or subtracted from, a matrix. (It does not fix division because
Metal also does not natively support componentwise division on
matrices.)
Change-Id: I7d5b0c5bd35393475c524e34cad789bf4f72a103
Bug: skia:11125
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/407616
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
- most of the small diffs are because I moved GrWangsFormula.h out
of the tessellate/ directory and into the geometry/ directory since
it's more general than HW tessellation.
The previous implementation was based on the heuristic that the distance
from the true curve to the line segment would be divided by 4 every time
the curve was recursively subdivided. This was a reasonable
approximation if the curve had balanced curvature on both sides of the
split. However, in the case of the new GM's curve, the left half was
already very linear and the right half had much higher curves.
This lead to the approximation reporting fewer points than required.
Theoretically, those few points that weren't utilized by the left half
of the curve could have been made available to the right half, but the
implementation of that would be tricky.
Instead, it now uses Wang's formula to compute the number of points.
Since recursive subdivision leads to linearly spaced samples assuming it
can't stop early, this point count represents a valid upper bound on
what's needed. It also then ensures both left and right halves of a
curve have the point counts they might need w/o updating the
generation implementations. However, since the recursive point
generation exits once each section has reached the error tolerance, in
scenarios where the prior approximation was reasonable, we'll end up
using fewer points than reported by Wang's. Hopefully that means there
is negligible performance regression since we won't be increasing
vertex counts by that much (except where needed for correctness).
Bug: skia:11886
Change-Id: Iba39dbe4de82011775524583efd461b10c9259fe
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/405197
Reviewed-by: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
This reverts commit 86121f6c0e.
Reason for revert: tree sad
Original change's description:
> The Matrices test now verifies its results.
>
> Previously, this test did a bunch of matrix math but never actually
> checked its results for correctness.
>
> Change-Id: I353be58049286266c2d561b0939b3874d2684403
> Bug: skia:11985
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/407360
> 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>
TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com
Change-Id: I1335f01c14ee955426e02efaa3c30421cd41aa34
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:11985
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/407617
Reviewed-by: John Stiles <johnstiles@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Previously, this test did a bunch of matrix math but never actually
checked its results for correctness.
Change-Id: I353be58049286266c2d561b0939b3874d2684403
Bug: skia:11985
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/407360
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>
Change-Id: I674f038600afd6d49316c1ece515941ee5579068
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/406939
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This reverts commit 71430593f0.
Reason for revert: Clang-tidy bot unhappy. (Not sure how this slipped by).
Original change's description:
> Added DSL flags
>
> This adds some basic flags to the DSL initialization which allows us to
> toggle things like optimization and validation.
>
> Change-Id: I35b10526af1678c88901eaacd5a2c4a9f5cd21a7
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/406896
> Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
> Reviewed-by: John Stiles <johnstiles@google.com>
TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com
Change-Id: I7512b8bcc6e94707949904eec42c396faa7fdaf8
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/407176
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This adds some basic flags to the DSL initialization which allows us to
toggle things like optimization and validation.
Change-Id: I35b10526af1678c88901eaacd5a2c4a9f5cd21a7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/406896
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
This effectively reverts a80ce1a36d,
and goes back to using SkMakeSpan (which works in C++14).
Change-Id: Iaa63c86b5acaadbdd60588b0a5c703820e810770
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/406938
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
- expand SkParsePath to support relative path encoding
- introduce a new SkSVGCanvas flag for clients to opt into relative
encoding
Bug: skia:11981
Change-Id: I2d6c63254df23311f1a86dee9481bde9531c3b61
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/406876
Reviewed-by: Tyler Denniston <tdenniston@google.com>
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Florin Malita <fmalita@google.com>
Makes computeFastBounds not part of the public API, it's only accessible
to subclasses of SkPathEffect, GrStyle, and SkPaint. Subclasses can
invoke it other path effects using SkPathEffectPriv::ComputeFastBounds.
Changes the internal function to
bool computeFastBounds(SkRect* bounds) const;
Subclasses of SkPathEffect must implement this, and can choose to return
false when fast bounds aren't computable.
Provides implementations of computeFastBounds() for path effects
bundled with Skia.
Bug: skia:11974
Change-Id: I545ccf99b4e669d3af9df13acfac28573306fab8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/406140
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Mike Reed <reed@google.com>
Aside from sqrt() and normalize(), we now optimize all the intrinsics
which take a single argument as input, and return that argument with
each of its components permuted as output.
This CL also introduces a minor restriction--we no longer optimize
intrinsics which evaluate to inf or nan, such as `inversesqrt(-1)`.
These will be left in the source as-is.
Change-Id: I4919b3c18a2df81accd6daf2f650b9f587ff43fc
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/406577
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
This is similar to the intrinsic optimization for any() and all(). Tests
for all three intrinsics have been bulked up a bit as well.
Change-Id: I262b9448e543b4709d1e7c8585f74a206c4b5abd
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/406576
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Bug: skia:11968
In the simple constructor without an SkImageInfo, cropRect, etc, read
the orientation and swap the width and height if necessary. Although
SkAnimatedImage respects the orientation at decode time, initializing
with backwards width/height results in treating them as a scale.
Change-Id: I0500c3e9a99701c0ec2bba8994c356587a3c876c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/406456
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Commit-Queue: Leon Scroggins <scroggo@google.com>
This layout qualifier is not actually used anywhere.
Change-Id: I817c9affdd00e492c70f251eb52680644b7ff3f7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/406141
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
UInt support was added to DSL in http://review.skia.org/403601. We use
the UInt type in the DitherEffect fragment processor.
Change-Id: I2770eb0196177ee403b461134c9895d2e0b2e6db
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/406139
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
In particular, this optimizes abs() and sign() when all inputs are known
at compile time. This resolves a TODO on a test case in
`IllegalIndexing.rts`.
Change-Id: Ica310522a85b42dc7ae255bd25004a6629d04176
Bug: skia:10835
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/405676
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
A cast like `float(five)` or `int4(colorGreen)` will now detect const
variables and replace the expression with its compile-time constant
equivalent value. This can unblock further constant folding
opportunities.
(This CL is very similar in spirit to http://review.skia.org/404676)
Change-Id: If78a2091770777b0caaaec696fe15a0f55d88c24
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/405683
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
A cast like `float(five)` or `int4(colorGreen)` should detect const
variables and replace the expression with its compile-time constant
equivalent value. At present, this replacement is missed, which inhibits
further optimization opportunities on the expression.
(This CL is very similar in spirit to http://review.skia.org/404676)
Change-Id: I04b5c435a30d2afcdbdb3d020adc15e9c651cc31
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/405682
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
The SkEnumerate was capturing elements by value instead of
by reference. This caused odd things to happen when the value
was used on the left side of an assignment.
Added a test to demonstrate this.
Change-Id: I14bbeee8041bcbaa4e5ca845d6510096558673b1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341462
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
* Enable MipmappedTest
* Add didWriteToSurface calls to fix up MipmapInvalidation test
* Add flush & submit to fix DDLSkSurfaceFlush test
Change-Id: I9c3cad96b40491a54cab4edb8f56639d6cc16665
Bug: skia:10446
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/405496
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Reformat uses to de-yodaify and try to improve readabilty of
complicated conditionals.
Change-Id: Ifccb84836cbe4f5a01813795ceb287089984f8ae
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/405685
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This CL only handles a subset of our intrinsics. In particular, it
avoids changing the behavior of `sqrt` as many of our tests use sqrt as
an optimization barrier.
The transcendental test inputs are intentionally kept very simple to
avoid putting numbers in the test outputs which could round differently
on various platforms and cause Housekeeper to complain.
Change-Id: I539f918294332310dcd6fe12fab163c0b6216f65
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/405398
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>
Previously, the code neglected to resolve constant variables into
values. This meant that expressions like `lessThan(zero, one)` could not
be compile-time evaluated even when `zero` and `one` have known values.
Change-Id: I2f5ce303e3dcc682be14e4d2485e24dd7c59212e
Bug: skia:10835
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/405536
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Chromium has been updated to use makeAnalysisCanvas directly and there are
no more references to SkTextBlobDiffCanvas as a type in its code base.
Since the GlyphTrackingDevice extends SkNoPixelsDevice, any SkCanvas that
uses it is effectively a "no-draw" canvas. However, by returning a base
SkCanvas the text tracking now automatically happens in the context of
the base's AutoLayerForImageFilter handling it applies on every draw. This
means that drawing a text blob with an image filter that modifies the
transform state will now be analyzed in that context automatically
(simplifying code in chrome after this lands).
Another behavioral change is that all non-text draws will still go through
the base SkCanvas' virtuals and invoke the device function. Since it's an
SkNoPixelsDevice, it'll still be a no-op, it just happens a little later.
This won't really impact performance because oop-r already inspects their
operations and only plays back text and transform related ones to the
analysis canvas, so we shouldn't really see non-text draws being invoked
anyways.
Bug: chromium:1187246
Change-Id: I83f86571300751f385b3065dfe889f218fa1edc6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/405196
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
This includes several new array variants of existing APIs, DSLBlock now
supporting SymbolTables, and a couple of other minor changes needed by
the upcoming DSLParser.
Change-Id: I71feb268feb27cf7ff453cc59046091779bffe06
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/404779
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
If every argument passed to lessThan/greaterThan/lessThanEqual/
greaterThanEqual/equal/notEqual() is a compile-time constant, we now
detect this and optimize away the function call entirely.
Change-Id: I3415d21be6ef51b38b682a792bd118fad51957f5
Bug: skia:10835
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/404776
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
If every argument passed to any() or all() is a compile-time constant,
we now detect this and optimize away the function call entirely.
Future CLs will perform a similar optimization on other intrinsic calls
which can be detected and eliminated at compile time, like lessThan().
Change-Id: Ie55aff538b1ccaf2b3bcf9a69573a85f081b7ade
Bug: skia:10835
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/404417
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This is a reland of adadb95a9f
... adds a temporary workaround for some Android framework code.
Original change's description:
> Better first-class shader & color filter support in runtime effects
>
> This does a few things, because they're all intertwined:
>
> 1) SkRuntimeEffect's API now includes details about children (which Skia
> stage was declared, not just the name). The factories verify that the
> declared types in the SkSL match up with the C++ types being passed.
> Today, we still only support adding children of the same type, so the
> checks are simple. Once we allow mixing types, we'll be testing the
> declared type against the actual C++ type supplied for each slot.
> 2) Adds sample variants that supply the input color to the child. This
> is now the only way to invoke a colorFilter child. Internally, we
> support passing a color when invoking a child shader, but I'm not
> exposing that. It's not clearly part of the semantics of the Skia
> pipeline, and is almost never useful. It also exposes users to
> several inconsistencies (skbug.com/11942).
> 3) Because of #2, it's possible that we can't compute a reusable program
> to filter individual colors. In that case, we don't set the constant
> output for constant input optimization, and filterColor4f falls back
> to the slower base-class implementation.
>
> Bug: skia:11813 skia:11942
> Change-Id: I06c41e1b35056e486f3163a72acf6b9535d7fed4
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/401917
> Commit-Queue: Brian Osman <brianosman@google.com>
> Reviewed-by: Mike Klein <mtklein@google.com>
Bug: skia:11813 skia:11942
Change-Id: I2c31b147ed86fa8c4dddefb7066bc1d07fe0d285
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/404637
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
The SkM44::RectToRect function matches the semantics of
SkMatrix::RectToRect(kFill_ScaleToFit). No other ScaleToFit variants are
ported over to SkM44.
skottie uses some instances of kCenter_ScaleToFit so that functionality
may need to be added in the future (in SkM44 or in skottie). There are
no current usages of the kStart and kEnd_ScaleToFit semantics.
The SkM44::mapRect() function is implemented to correspond to the
SkMatrix::mapRect() that returns the mapped rect (instead of modifying a
pointer) and always has ApplyPerspectiveClip::kYes. This was chosen to
keep its behavior simple and because perspective clipping is almost
always the right thing to do. In the new implementation there is no
longer a performance cliff to worry about (see below). For the timebeing
mapRect is hidden behind SkMatrixPriv::MapRect().
Performance:
I added benchmarks for mapRect() on SkM44 and SkMatrix that use the same
matrices to get a fair comparison on their different specializations.
SkMatrix has a very efficient mapRect when it's scale+translate or
simpler, then another impl. for affine matrices, and then falls back to
SkPath clipping when there's perspective. On the other hand, SkM44 only
has 2 modes: affine and perspective.
On my desktop, with a Ryzen 9 3900X, here are the times for 100,000 calls
to mapRect for different types of matrices:
SkMatrix SkM44
scale+translate 0.35 ms 0.42 ms
rotate 1.70 ms 0.42 ms
perspective 63.90 ms 0.66 ms
clipped-perspective 138.0 ms 0.96 ms
To summarize, the SkM44::mapRect is almost as fast as the s+t specialization
in SkMatrix, but for all non-perspective matrices. For perspective matrices
it's only 2x slower than that specialization when no vertices are clipped,
and still almost 2x faster than the affine specialization when vertices are
clipped (and 100x faster than falling back to SkPath).
Given that, there's the open question of whether or not keeping an affine
specialization is worth it for SkM44's code size.
Bug: skia:11720
Change-Id: I6771956729ed64f3b287a9de503513375c9f42a0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/402957
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
Auto-Submit: Michael Ludwig <michaelludwig@google.com>
A constructor like `float3(five)` is not a compile-time constant, so we
miss optimization opportunities like folding. Constant variables inside
splat constructors are now replaced when optimization is on, so this
would optimize down to `float3(5.0)` and be eligible for folding.
Change-Id: I4bf6f52a48ef733e6b24791d02687081194ef488
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/404676
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
A constructor like `float2(one, two)` is not a compile-time constant, so
we miss optimization opportunities like folding. Constant variables
inside compound constructors are now replaced when optimization is on,
so this would optimize down to `float2(1.0, 2.0)` and be eligible for
folding.
Change-Id: I80dd421f61d4eed21278805e2dc26d198a678e52
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/404657
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
The Serialization_Typeface and FontMgr_AliasNames tests sometimes fail
on macOS 10.13 and 10.15.7 bots. This is likely fixed on macOS 11,
however in the interest of investigation there should be some attempt at
discovering if the retrieved fonts are actually different and which ones
were retrieved. Improve the reports with additional information about
the typefaces in question.
Change-Id: I0e4354eacf91be1ae98838569e4da9f964dc2ac8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/404338
Commit-Queue: Ben Wagner <bungeman@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
This breaks up the giant IRGenerator::convertFunction method into more-
manageable chunks, moves the functionality into FunctionDeclaration,
and funnels the DSL through it so it receives the same error checking.
Change-Id: Icf2ac650ab3d5276d8c0134062a4e7e220f9bf32
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/402778
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
This will be used by the upcoming DSLParser, which needs to be able to
put DSLExpression and DSLVar into containers such as std::vector and
std::optional.
Change-Id: I8d367cfd0b3a852a368c69a5b3be6c0eaa41d74a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/404156
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
This reverts commit adadb95a9f.
Reason for revert: breaking android
Original change's description:
> Better first-class shader & color filter support in runtime effects
>
> This does a few things, because they're all intertwined:
>
> 1) SkRuntimeEffect's API now includes details about children (which Skia
> stage was declared, not just the name). The factories verify that the
> declared types in the SkSL match up with the C++ types being passed.
> Today, we still only support adding children of the same type, so the
> checks are simple. Once we allow mixing types, we'll be testing the
> declared type against the actual C++ type supplied for each slot.
> 2) Adds sample variants that supply the input color to the child. This
> is now the only way to invoke a colorFilter child. Internally, we
> support passing a color when invoking a child shader, but I'm not
> exposing that. It's not clearly part of the semantics of the Skia
> pipeline, and is almost never useful. It also exposes users to
> several inconsistencies (skbug.com/11942).
> 3) Because of #2, it's possible that we can't compute a reusable program
> to filter individual colors. In that case, we don't set the constant
> output for constant input optimization, and filterColor4f falls back
> to the slower base-class implementation.
>
> Bug: skia:11813 skia:11942
> Change-Id: I06c41e1b35056e486f3163a72acf6b9535d7fed4
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/401917
> Commit-Queue: Brian Osman <brianosman@google.com>
> Reviewed-by: Mike Klein <mtklein@google.com>
TBR=mtklein@google.com,bsalomon@google.com,brianosman@google.com
Change-Id: I94ba57e73305b2302f86fd0c1d76f667d4e45b92
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:11813 skia:11942
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/404117
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Change-Id: I721825d1a38e9f6846b94f84d14cb8c85b7a7519
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/403601
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
This does a few things, because they're all intertwined:
1) SkRuntimeEffect's API now includes details about children (which Skia
stage was declared, not just the name). The factories verify that the
declared types in the SkSL match up with the C++ types being passed.
Today, we still only support adding children of the same type, so the
checks are simple. Once we allow mixing types, we'll be testing the
declared type against the actual C++ type supplied for each slot.
2) Adds sample variants that supply the input color to the child. This
is now the only way to invoke a colorFilter child. Internally, we
support passing a color when invoking a child shader, but I'm not
exposing that. It's not clearly part of the semantics of the Skia
pipeline, and is almost never useful. It also exposes users to
several inconsistencies (skbug.com/11942).
3) Because of #2, it's possible that we can't compute a reusable program
to filter individual colors. In that case, we don't set the constant
output for constant input optimization, and filterColor4f falls back
to the slower base-class implementation.
Bug: skia:11813 skia:11942
Change-Id: I06c41e1b35056e486f3163a72acf6b9535d7fed4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/401917
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
This string had several characters which resulted in zero
glyphs. Having zero glyphs bypassed the buffer sizing code,
and continued to process the glyph run list assuming it had
7 runs because the buffer sizing code also clears the run list.
This caused the code to process runs from a previous SkTextBlob,
which was already deleted.
If the there are no glyphs, call the buffer sizing code to
set up all the invariants. Exit if there are no runs produced.
Bug: oss-fuzz:33915
Change-Id: I9f3f38a58112c44ddd65265c68d982b3b0dcd79c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/403439
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Change-Id: I06631e7f0db518f4de19a39bf1ed368afbd5d409
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/403076
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>
