Commit Graph

8622 Commits

Author SHA1 Message Date
John Stiles
21a59d650d Fix double-negation of constant-value construcors.
Previously, the code which calculated Constructor constant values
assumed that a constant-value PrefixExpression would always have an
operand of Constructor. It turns out that another valid case is multiple
PrefixExpressions nested within each other (representing repeated
negation). Updated the code to work regardless of the type of the prefix
operand.

Change-Id: Ic9bf54725ae59330ac817bc4ec7a64def384ab54
Bug: oss-fuzz:27663
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/337177
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
2020-11-23 17:02:21 +00:00
Mike Klein
ee40ec6dd6 remove Op::pack
pack(x,y,bits) as an alias for x|(y<<bits) only existed originally to
implement it with the SLI arm64 instruction, but I've since realized
that was misguided.

I had thought the assumption on pack ("(x & (y << bits)) == 0"), i.e.
"no overlap between x and the shifted y", was enough to make using SLI
legal, but it's actually not strong enough a requirement.

The SLI docs say "...inserts the result into the corresponding vector
element in the destination SIMD&FP register such that the new zero bits
created by the shift are not inserted but retain their existing value."
The key thing not mentioned there happens with zero bits _not_ created
by the shift, the ones already present at the top of y.  They're of
course inserted, overwriting any previous values.

This means SLI (and so pack()) become strictly order dependent in a way
I had never intended.  This will work as you'd think,

    skvm::I32 px = splat(0);
    px = pack(px, r,  0);
    px = pack(px, a, 24);

but this version swapping the two calls to pack() will overwrite alpha,

    skvm::I32 px = splat(0);
    px = pack(px, a, 24);
    px = pack(px, r,  0);

I find that error-prone, so I've removed Op::pack and replaced it
with a simple expansion to x|(y<<bits).  That of course works in either
order.

This new test can't JIT at head, but if we implement the other missing
instructions (soon, dependent CL) it would start failing when JIT'd.
The interpreter and x86 were both fine, since they're both doing what's
now the only approach to pack(), the simple x|(y<<bits).

I've left assembler support for SLI in case we want to try it again.

Change-Id: Iaf879309d3e1d0a458a688f3a62556e55ab05e23
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/337197
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
2020-11-21 12:38:19 +00:00
John Stiles
dda1d31623 Enable SPIR-V disassembly output from skslc.
We now have SPIR-V golden outputs for `blend` and `shared` tests.
This exposes a handful of SPIR-V limitations for us to address.

Change-Id: Ie5278889b8a61432403d06231b17765885bee0ac
Bug: skia:10694
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/337182
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>
2020-11-20 22:54:26 +00:00
Robert Phillips
2c21a11923 Remove uses of SkCanvas::flush
afaict the perf surprises associated with:

https://skia-review.googlesource.com/c/skia/+/334417 (Remove SkBaseDevice::flush)

were bc Ganesh resolves MSAA buffers for SkCanvas::flush but doesn't do so for GrDirectContext::flush.

Where possible this CL switches SkCanvas::flush to SkSurface::flush (which will also resolve MSAA buffers) so that when https://skia-review.googlesource.com/c/skia/+/334417 relands there should not be any performance surprises.

Change-Id: I705ad6219f0f625a88cf3f9e8b2418a3182d298c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335866
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Adlai Holler <adlai@google.com>
2020-11-20 21:42:38 +00:00
John Stiles
0f46450775 Reland "Remove inliner from IR generation stage."
This reverts commit 4c412bce4c.

Reason for revert: investigating Pinpoint failure cases, if any

Original change's description:
> Revert "Reland "Remove inliner from IR generation stage.""
>
> This reverts commit e497a08065.
>
> Reason for revert: Pinpoint disagrees
>
> Original change's description:
> > Reland "Remove inliner from IR generation stage."
> >
> > This reverts commit 941fc7174f.
> >
> > Reason for revert: performance now seems to be roughly equal or better
> > (~1%) over several trials.
> > Nanobench: http://screen/A8e8sojaXBgbMgF
> >
> > Original change's description:
> > > Revert "Remove inliner from IR generation stage."
> > >
> > > This reverts commit 21d7778cb5.
> > >
> > > Reason for revert: Pinpoint absolutely hates this change
> > >
> > > Original change's description:
> > > > Remove inliner from IR generation stage.
> > > >
> > > > There is no need to inline code during IR generation, as the optimizer
> > > > can now handle this.
> > > >
> > > > Change-Id: If272bfb98e945a75ec91fb4aa026e5631ac51b5b
> > > > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/315971
> > > > Commit-Queue: John Stiles <johnstiles@google.com>
> > > > Commit-Queue: Brian Osman <brianosman@google.com>
> > > > Reviewed-by: Brian Osman <brianosman@google.com>
> > > > Auto-Submit: John Stiles <johnstiles@google.com>
> > >
> > > TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com
> > >
> > > Change-Id: I62c235415bcdc92a088e2a7f9c3d7dbf7e1bf669
> > > No-Presubmit: true
> > > No-Tree-Checks: true
> > > No-Try: true
> > > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/317976
> > > Reviewed-by: John Stiles <johnstiles@google.com>
> > > Commit-Queue: John Stiles <johnstiles@google.com>
> >
> > TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com
> >
> > Change-Id: I6189806c678283188f4b67ee61e5886f88c2d6fc
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/324891
> > Reviewed-by: John Stiles <johnstiles@google.com>
> > Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
> > Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
> > Auto-Submit: John Stiles <johnstiles@google.com>
>
> TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com
>
> Change-Id: I79149467565f22f53b8c28868dd53b80f3421137
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/325626
> Reviewed-by: John Stiles <johnstiles@google.com>
> Commit-Queue: John Stiles <johnstiles@google.com>
> Auto-Submit: John Stiles <johnstiles@google.com>

TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com

Change-Id: I2727bd4a2b43e8d12b36b1979ce6fe4a2d935380
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335936
Reviewed-by: John Stiles <johnstiles@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
2020-11-20 18:44:07 +00:00
Robert Phillips
5c80964048 Update GrProgramInfo's constructor
Change-Id: I3a8ca6913ad64ac4fc29e63f7e26db128774c407
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/337118
Reviewed-by: Adlai Holler <adlai@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
2020-11-20 18:22:17 +00:00
John Stiles
c30fbcaf77 Allow swizzle optimizations to apply to any 'trivial' ctor fields.
This allows swizzle removal to apply in more cases; in particular, we
can now optimize away extra swizzles caused by zero/one swizzle-
components quite effectively.

The "trivial expression" code was lifted from the inliner. Some subtle
changes in trivial-expression determination affect the inliner's results
in boring, non-meaningful ways. In particular, multi-argument
constructors containing all-constant values are now considered trivial,
whereas previously only single-argument constructors made the trivial-
ness cut. This allows the inliner to propagate some values that it
wouldn't have before.

Change-Id: I9a009b6803d9ac9595d65538252ba81c2b7166a7
Bug: skia:10954
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336156
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
2020-11-20 16:35:16 +00:00
Greg Daniel
42dbca51f4 Pass in the color load op to GrProgramInfo.
In follow on CLs we need to know what the load op is when we try to use
discardable msaa attachments. For vulkan the load op affects how we
copy the resolve attachment into the msaa attachment, which changes the
render pass we use (adds extra subpass). We need to be able to make a
compatible render pass to compile programs.

Bug: skia:10979
Change-Id: I40c23a18b251af6a2ad3b78a1f6382bdba0b90c4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336598
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
2020-11-20 16:29:56 +00:00
Brian Osman
0006ad01ce Stop cloning builtin functions
Previously, any builtin functions would be optimized as a side-effect of
optimizing programs that used them. Now that shared elements aren't
being optimized in that way, we explicitly optimize any shared modules
when they are first created. We don't remove dead elements, but we
we do substitute settings, simplify, and inline.

Bug: skia:10905
Change-Id: I701b5e9f52fb880ef3e6f4c67694d08602f47e95
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336440
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
2020-11-20 15:02:54 +00:00
Adlai Holler
e2296f7a70 Pass the writeView as a const-ref instead of a pointer …
The important part is in GrOpFlushState.h were previously
we were taking a mutable pointer to the view, which should
at least be a const pointer and was making us do funky things
in some of the calling code. But I decided to go all the way
and do a const ref instead which is The Way It Should Be (tm).

Change-Id: I399d102e8b5e0a5059168cc450ae66f12ad47e13
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336451
Reviewed-by: Greg Daniel <egdaniel@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Adlai Holler <adlai@google.com>
2020-11-20 14:59:34 +00:00
John Stiles
8c58899371 Fix fuzzer crash when casting between int and float.
The fix submitted at http://review.skia.org/335868 did not support
casts. The fuzzer discovered this shortcoming right away.

Change-Id: I2f5166528cee41367348564d4e664476fd5704ff
Bug: oss-fuzz:27650
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336656
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
2020-11-20 14:07:03 +00:00
John Stiles
ed289e777c Simplify _blend_set_color_saturation, removing an instruction.
This tightens up our intrinsics slightly; after inlining, it eliminates
one scratch variable. (We no longer need to copy `sda` into `hueColor`
as hueColor is now unchanged.)

Change-Id: Iece5ba2fe11cde54481704a1787114a2c2a66d9b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336599
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>
2020-11-19 22:57:10 +00:00
Michael Ludwig
ad5ec1ae85 Fix color stop positions when computing degenerate gradient color
For non-degenerate gradients, the positions are pinned to [0, 1] and
forced to be monotonically increasing in the constructor of
SkGradientShaderBase. This isn't ever called for degenerate gradients,
so the average_gradient_color helper function needs to make the same
fixes or it may calculate an invalid color (e.g. negative alpha because
the original positions are not sorted).

Bug: chromium:1149216
Change-Id: I95c6e66ebb57722370ef2f5dbba3d8d66727e48b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336437
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
2020-11-19 20:13:20 +00:00
John Stiles
d9076cb637 Merge foo.x, foo.y, foo.z into foo.xyz when optimizing swizzles.
When values from the same argument are used consecutively by the outer
swizzle, they can be merged in the inner swizzle. Merging isn't always
possible, of course, but it will be used where it can be:

    `half4(1, colRGB).yzwx` --> `half4(colRGB.xyz, 1)`
    `half4(1, colRGB).yxzw` --> `half4(colRGB.x, 1, colRGB.yz)`

Change-Id: Id164b046bc15022ded331c06d722f1ae3605a3bd
Bug: skia:10954
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335872
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
2020-11-19 18:20:24 +00:00
Mike Reed
8dacba3547 Migrate away for heuristics for bicubic filtering
Bug: skia: 7650
Change-Id: Id20aed3a7a2676ab053a32a95f225e60186c0851
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335593
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
2020-11-19 18:12:04 +00:00
Chris Dalton
9458c8d44a Add an epsilon to GrPathUtils::findCubicConvex180Chops
Cubic tangents become unstable if we chop too close to 0 or 1. This
adds an epsilon to simply not chop them if it's too close.

Bug: skia:10419
Change-Id: I99e98c5d433e2cb59c767ee564e015d7ec4f7765
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336280
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
2020-11-19 17:43:21 +00:00
John Stiles
0777ac4778 Optimize swizzled multiple-argument constructors.
This will reorder constructors with swizzles applied, such as
    `half4(1, 2, 3, 4).xxyz` --> `half4(1, 1, 2, 3)`
    `half4(1, colRGB).yzwx` --> `half4(colRGB.x, colRGB.y, colRGB.z, 1)`

Note that, depending on the swizzle components, some elements of the
constructor may be duplicated and others may be eliminated. The
optimizer makes sure to leave the swizzle alone if it would duplicate
anything non-trivial, or if it would eliminate anything with a side
effect.

Change-Id: I470fda217ae8cf5828406b89a5696ca6aebf608d
Bug: skia:10954
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335860
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
2020-11-19 17:10:11 +00:00
John Stiles
d1d872905b Add fix for fuzzer-discovered crash with negated constructors.
This was found at https://oss-fuzz.com/testcase-detail/5155684475469824
but the associated oss-fuzz issue ID appears to be misdirected (it's
showing oss-fuzz:24498, an unrelated issue).

PrefixExpressions can return true for `isCompileTimeConstant` but did
not implement `compareConstant`; the fuzzer discovered this. Because
compile-time constants can only be compared if they are of the same
kind, this means that `compareConstant` is actually comparing a pair of
expressions that are both negated. These negations will just cancel
out, so `compareConstant` on a pair of PrefixExpressions can just call
`compareConstant` on the inner operand of each expression.

Change-Id: I7793e25314e6c8a74278b73299d310794baf71f4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335870
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>
2020-11-18 21:53:45 +00:00
Brian Osman
1f71f433ff Always enable SkSL's ByteCodeGenerator, disable interpreter in Google3
The ByteCodeGenerator is needed for SkSL-via-skvm, but almost no one
needs the ByteCode interpreter.

Bug: b/172773885
Change-Id: Ia7b6768dbc00c6c78b971ba50f0b702536bbd5b1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336016
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
2020-11-18 21:30:45 +00:00
John Stiles
d0f712f3fe Add fix for fuzzer-discovered crash at oss-fuzz:27614.
The fuzzer managed to create a test case which temporarily evaluates to
expression `half2(half(0.2)) + 2` as it is optimized. This requires a
bunch of temporary nonsense math as the IR Generator is attempting to
simplify as it goes; various attempts to remove terms from the fuzzer
test-case would cause it to stop reproducing the error.

Constructor::getVecComponent assumed that any constructor with a single
scalar argument would always implement `getConstantFloat` and
`getConstantInt`; however, constructors themselves did not actually
implement these methods. This meant that nesting a scalar constructor
inside a non-scalar constructor would abort when it tried to deduce the
value inside the inner constructor.

This has been fixed by implementing `getConstantFloat` and
`getConstantInt` for Constructors. These methods will assert if the
constructor has more than one argument or is a non-scalar type. This
should allow any number of nested constructors, e.g.
`half4(half(half(half(1))))` should recursively evaluate properly,
should we somehow generate this as an intermediate expression.

Change-Id: Iaee4284cba03974443cd7b5dccfd7909c1a5f3a6
Bug: oss-fuzz:27614
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335868
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>
2020-11-18 21:01:15 +00:00
John Stiles
108bbe2522 Optimize away swizzles on single-argument constructors.
The optimizer can now turn the expression `half4(1).xyz` into
`half3(1)`, or `half4(1).w` into `1`. This is actually a somewhat common
case when inlining chains of fragment processors, as inputs are often
overridden to `half4(1)` or `half4(0)`. This optimization also applies
to more complex cases, e.g.:

     `half2(anyFunc(sqrt(2))).yxyx` --> `half4(anyFunc(sqrt(2)))`

Since the interior of the constructor is always evaluated once in either
case, it does not actually matter what the constructor contains.

Change-Id: I8d5f358502eaa8e35d4968e74fbd6b0ce2ab6365
Bug: skia:10954
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335818
Commit-Queue: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
2020-11-18 17:02:45 +00:00
Brian Osman
8dbcebac34 Remove (unused) SkSL::NodeArrayWrapper
Change-Id: Ibe3c3d27a112df8838bc86d6c2482277fdae62af
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335821
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
2020-11-18 16:56:15 +00:00
Brian Osman
0540efe78a Remove nullptr default Type from IntLiteral
This was slightly dangerous, and only used for tests. Fix the tests to
just put a Context on the stack.

Change-Id: Ifc3d600c3498e1dedeee8350f3284163f3195bec
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335819
Auto-Submit: Brian Osman <brianosman@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
2020-11-18 16:14:35 +00:00
Robert Phillips
a186c35574 Update the TopoSort test ...
... to ensure that, if the topological sort fails, all the nodes still appear in the result.

Change-Id: Ic4acaab2662c872662246b7c1f5471e17c5ba98e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335639
Reviewed-by: Adlai Holler <adlai@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
2020-11-18 12:39:15 +00:00
John Stiles
e1bbd5c128 Disallow unsized array dimensions on size fields past the frontmost.
This was slightly complicated by the fact that this syntax indicates an
array with a known size:

    float[] x = float[](1, 2, 3, 4);

Of course, the size is 4; it's just never explicitly stated in the
code. (The SkSL parser never actually deduces the size, but it doesn't
apparently have a need to; we don't do much in the way of optimization
for arrays.) However, this prevents us from simply failing whenever we
parse "[]" in non-builtin code; we need to keep scanning and see if the
variable is initialized. We already check this in the
ArrayConstructors.sksl test file.

Change-Id: I5b86958e81bd9bf5edf28a617cecf95c1875583e
Bug: skia:10957
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335240
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
2020-11-17 16:44:13 +00:00
John Stiles
08070f6f65 Report the correct line number when vardecls have an error.
Previously, the Type's fOffset was set to -1 during parsing, so any
errors related to the Type would be reported on line 1.

Change-Id: I9834f733bc763c5946b3ff81d8aef4807cdc13d1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335584
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>
2020-11-17 16:32:23 +00:00
John Stiles
1d75778cbf Disallow opaque types in structs and interface blocks.
This is a followup to http://review.skia.org/335196. This detects opaque
types (samplers and textures) at parsing or IR generation time and
reports an error regardless of backend. This check occurs before Metal
or SPIR-V would have a chance to detect the error, so it changes their
output to a slightly more focused error message. The Metal/SPIR-V fix in
the prior CL is still a nice broad catch-all for preventing spurious
ABORTs, though.

Change-Id: I4cce92a8767d72b5d3d7277a8afde8ce5ce86db2
Bug: skia:10956
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335217
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
2020-11-17 15:25:43 +00:00
Robert Phillips
fbbc3bbecf Move SkTTopoSort to src/gpu and rename it to GrTTopoSort
Change-Id: I76deff6bbb2d796cead0e9415c19daf5edceaa73
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335296
Reviewed-by: Adlai Holler <adlai@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
2020-11-17 14:17:03 +00:00
John Stiles
0023c0c827 Detect unsupported types for MemoryLayout and report errors.
Previously, MemoryLayout would ABORT if it encountered any types that
we can't layout in memory (e.g. opaque types like samplers). Instead of
an abort, this case is now detected cleanly and an error is reported
identifying the offending type.

This should unwedge the fuzzer, which appears to be very
enthusiatically generating interface blocks with nonsense types inside.

(Note that code generators which don't actually try to compute a memory
layout--that is, GLSL--will still accept these types. This should still
be caught and reported as an error, since it's still illegal in GLSL,
but that's for a future CL.)

Change-Id: I88a9649bcd8c75dadc8cca679f3c5e94570742bc
Bug: skia:10956, oss-fuzz:27525
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335196
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>
2020-11-16 19:14:48 +00:00
Robert Phillips
b7c1310834 Revert "Remove SkBaseDevice::flush"
This reverts commit 971b19d0c6.

Reason for revert: See if this CL was causing the Perf-Win10-Clang-ShuttleA-GPU-GTX660-x86_64-Release-All issue

Original change's description:
> Remove SkBaseDevice::flush
>
> Another small step in removing SkCanvas::flush
>
> Change-Id: I6f17edcd1996e1009dad7cc96a97be3b0c4664f4
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334417
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Reviewed-by: Adlai Holler <adlai@google.com>
> Commit-Queue: Robert Phillips <robertphillips@google.com>

TBR=egdaniel@google.com,bsalomon@google.com,robertphillips@google.com,adlai@google.com

Change-Id: I0b9116f8ce4eed3a0d49ccf1cc55d8d89675617e
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335216
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
2020-11-16 18:46:19 +00:00
Leon Scroggins
1340dbde91 Reland "SkAndroidCodec: Support decoding all frames"
This is a reland of fc4fdc5b25

Original change's description:
> SkAndroidCodec: Support decoding all frames
>
> Bug: b/160984428
> Bug: b/163595585
>
> Add support to SkAndroidCodec for decoding all frames with an
> fSampleSize, so that an entire animation can be decoded to a smaller
> size.
>
> dm/:
> - Test scaled + animated decodes
>
> SkAndroidCodec:
> - Make AndroidOptions inherit from SkCodec::Options. This allows
>   SkAndroidCodec to use fFrameIndex. (It also combines the two versions
>   of fSubset, which is now const for both.)
> - Respect fFrameIndex, and call SkCodec::handleFrameIndex to decode
>   the required frame.
> - Disallow decoding with kRespect + fFrameIndex > 0 if there is a
>   non-default orientation. As currently written (except without
>   disabling this combination), SkPixmapPriv::Orient would draw the new
>   portion of the frame on top of uninitialized pixels, instead of the
>   prior frame. This could be fixed by
>   - If SkAndroidCodec needs to decode the required frame, it could do so
>     without applying the orientation, then decode fFrameIndex, and then
>     apply the orientation.
>   - If the client provided the required frame, SkAndroidCodec would need
>     to un-apply the orientation to get the proper starting state, then
>     decode and apply.
>   I think it is simpler to force the client to handle the orientation
>   externally.
>
> SkCodec:
> - Allow SkAndroidCodec to call its private method handleFrameIndex. This
>   method handles decoding a required frame, if necessary. When called by
>   SkAndroidCodec, it now uses the SkAndroidCodec to check for/decode the
>   required frame, so that it will scale properly.
> - Call rewindIfNeeded inside handleFrameIndex. handleFrameIndex calls a
>   virtual method which may set some state (e.g. in SkJpegCodec). Without
>   this change, that state would be reset by rewindIfNeeded.
> - Simplify handling a kRestoreBGColor frame. Whether provided or not,
>   take the same path to calling zero_rect.
> - Updates to zero_rect:
>   - Intersect after scaling, which will also check for empty.
>   - Round out instead of in - this ensures we don't under-erase
>   - Use kFill_ScaleToFit, which better matches the intent.
>
> Change-Id: Ibe1951980a0dca8f5b7b1f20192432d395681683
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/333225
> Commit-Queue: Leon Scroggins <scroggo@google.com>
> Reviewed-by: Derek Sollenberger <djsollen@google.com>

Bug: b/160984428
Bug: b/163595585
Change-Id: I7c1e79e0f92c75b4840eef65c8fc2b8497189e81
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334842
Auto-Submit: Leon Scroggins <scroggo@google.com>
Commit-Queue: Derek Sollenberger <djsollen@google.com>
Reviewed-by: Derek Sollenberger <djsollen@google.com>
2020-11-16 16:55:48 +00:00
John Stiles
34de5cb57b Convert remaining Metal tests to golden outputs.
Metal-specific tests are pretty thin on the ground here, and some of
the remaining tests no longer added value as they were already covered
pretty well by existing tests in Shared. The majority of remaining tests
were specific to Metal's lack of flexible matrix casting (and SkSL's
ability to paper over this with helper functions).

Change-Id: I7b3c445268b95320e7f46ec88d793c315d43ee8a
Bug: skia:10694
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334956
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
2020-11-16 16:32:56 +00:00
Robert Phillips
971b19d0c6 Remove SkBaseDevice::flush
Another small step in removing SkCanvas::flush

Change-Id: I6f17edcd1996e1009dad7cc96a97be3b0c4664f4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334417
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Adlai Holler <adlai@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
2020-11-16 15:26:06 +00:00
John Stiles
031a76756e Stop the inliner after it has inlined 2500 statements in a program.
This prevents OOMing when given a pathological input, but is large
enough that almost all inputs should continue to compile as-is.

Change-Id: If5c46711b886ee08495bfd09af537e9dc7ea5649
Bug: skia:10945, oss-fuzz:27442
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334838
Commit-Queue: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
2020-11-13 23:02:11 +00:00
John Stiles
053739dfa8 Add unit test for O(n^3) behavior in the inliner.
In practice, the inline threshold does a good job of limiting the
blast radius here.

Change-Id: I495184116e733262ea9d84fec30885ea047ca116
Bug: skia:10945
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334597
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
2020-11-13 22:13:21 +00:00
John Stiles
b4b627e62a Disallow usage of private types ($vec, etc) in non-builtin code.
This fixes a fuzzer crash in Metal.

Private types aren't meant to be used directly; we can't generate a
valid MemoryLayout for them. We will now detect them during IR
generation and report an error. (Note that unreferenced structs
currently don't have any IR representation at all, so structs have to be
used somewhere in the code to trigger the error.)

Bug: oss-fuzz:27288
Change-Id: I432f0a69fbb54cd33ff5b90a9f3d4757a9370117
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334830
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
2020-11-13 21:55:50 +00:00
Leon Scroggins III
267a8589d3 Revert "SkAndroidCodec: Support decoding all frames"
This reverts commit fc4fdc5b25.

Reason for revert: Google3 and ASAN failures

Change-Id: I890cd76109c0375391637f879550837d01e650f9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334840
Reviewed-by: Leon Scroggins <scroggo@google.com>
Commit-Queue: Leon Scroggins <scroggo@google.com>
2020-11-13 16:20:39 +00:00
Leon Scroggins
fc4fdc5b25 SkAndroidCodec: Support decoding all frames
Bug: b/160984428
Bug: b/163595585

Add support to SkAndroidCodec for decoding all frames with an
fSampleSize, so that an entire animation can be decoded to a smaller
size.

dm/:
- Test scaled + animated decodes

SkAndroidCodec:
- Make AndroidOptions inherit from SkCodec::Options. This allows
  SkAndroidCodec to use fFrameIndex. (It also combines the two versions
  of fSubset, which is now const for both.)
- Respect fFrameIndex, and call SkCodec::handleFrameIndex to decode
  the required frame.
- Disallow decoding with kRespect + fFrameIndex > 0 if there is a
  non-default orientation. As currently written (except without
  disabling this combination), SkPixmapPriv::Orient would draw the new
  portion of the frame on top of uninitialized pixels, instead of the
  prior frame. This could be fixed by
  - If SkAndroidCodec needs to decode the required frame, it could do so
    without applying the orientation, then decode fFrameIndex, and then
    apply the orientation.
  - If the client provided the required frame, SkAndroidCodec would need
    to un-apply the orientation to get the proper starting state, then
    decode and apply.
  I think it is simpler to force the client to handle the orientation
  externally.

SkCodec:
- Allow SkAndroidCodec to call its private method handleFrameIndex. This
  method handles decoding a required frame, if necessary. When called by
  SkAndroidCodec, it now uses the SkAndroidCodec to check for/decode the
  required frame, so that it will scale properly.
- Call rewindIfNeeded inside handleFrameIndex. handleFrameIndex calls a
  virtual method which may set some state (e.g. in SkJpegCodec). Without
  this change, that state would be reset by rewindIfNeeded.
- Simplify handling a kRestoreBGColor frame. Whether provided or not,
  take the same path to calling zero_rect.
- Updates to zero_rect:
  - Intersect after scaling, which will also check for empty.
  - Round out instead of in - this ensures we don't under-erase
  - Use kFill_ScaleToFit, which better matches the intent.

Change-Id: Ibe1951980a0dca8f5b7b1f20192432d395681683
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/333225
Commit-Queue: Leon Scroggins <scroggo@google.com>
Reviewed-by: Derek Sollenberger <djsollen@google.com>
2020-11-12 21:31:41 +00:00
Leon Scroggins
afdbd9d75d SkAnimatedImage: Reject invalid crops
A crop is invalid if it is empty/unsorted or does not intersect with
the image. Android (the only known client of the cropping API) will not
pass such a rectangle to the API, but this prevents other clients from
passing an invalid rectangle.

Change-Id: I09e007ecd378c358a1c604aff518035090f1e0a4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/333224
Commit-Queue: Leon Scroggins <scroggo@google.com>
Reviewed-by: Derek Sollenberger <djsollen@google.com>
2020-11-12 20:24:21 +00:00
Derek Sollenberger
0f3afa7e55 Revert "make quickReject treat empty rects and paths the same"
This reverts commit d986835a6c.

Reason for revert: breaking a CTS tests and stopping Android autoroller

Original change's description:
> make quickReject treat empty rects and paths the same
>
> Change-Id: Ie334a2f0aa354ea3b52f6143e2659fd74fdd6185
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334100
> Reviewed-by: Mike Klein <mtklein@google.com>
> Commit-Queue: Mike Reed <reed@google.com>

TBR=mtklein@google.com,reed@google.com

Change-Id: I7c296aa93089170c4bcd0968b2782809cab7efb0
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334421
Reviewed-by: Derek Sollenberger <djsollen@google.com>
Commit-Queue: Derek Sollenberger <djsollen@google.com>
2020-11-12 16:10:39 +00:00
Robert Phillips
80bfda87c8 Remove GrSurfaceContext::flush calls ...
... replacing them w/ calls to GrDirectContextPriv::flushSurface.

Since recording contexts can also possess surface- and renderTarget-
Contexts it is misleading for them to have a flush method.

Change-Id: I10f4fad12d4d5efdd999ba212fda9ce5cdd83130
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334068
Reviewed-by: Adlai Holler <adlai@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
2020-11-12 15:33:07 +00:00
Leon Scroggins
bc098ef6d4 Handle EXIF orientation in SkAnimCodecPlayer
Bug: skia:10914

SkAnimCodecPlayer:
- Properly handle orientation, whether the image is still or not
- Mark const methods as const
- Fix seek() so that if you seek to the duration of frame 0, it will
  show frame 1
- Fix the SkImageInfo so if the first frame is opaque, but following
  frames are not, those frames can still be decoded

resources:
- Rename "webp-animated.webp" to "stoplight.webp", which better
  describes the animation
  - Update test files accordingly
- Add "stoplight_h.webp", which is the same animation with an EXIF
  that converts it to a horizontal stoplight

AnimCodecPlayer test:
- Test the new image files
- Verify SkAnimCodecPlayer::dimensions behaves as expected
- Remove extra debugging line
- Provide better error messages

AnimCodecPlayerExifGM:
- Add a new GM that shows all frames of the new animation with an EXIF
  orientation
- Add a new GM that shows all frames of an animation with an opaque
  first frame followed by frames with alpha

Change-Id: I43cf91c16d52aa1901eef8e13e1e644eea6058b3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332753
Reviewed-by: Derek Sollenberger <djsollen@google.com>
Commit-Queue: Leon Scroggins <scroggo@google.com>
2020-11-12 14:43:47 +00:00
Brian Salomon
e4387382c2 Split SkYUVAInfo::PlanarConfig into PlaneConfig and Subsampling enums
Sometimes it's helpful to think about subsampling separately from
how the channels are spread across planes.

Bug: skia:10632
Change-Id: Ib03f71195f9706ef6def418b1f2125c29e0cf738
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334102
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
2020-11-11 22:42:36 +00:00
Mike Reed
d986835a6c make quickReject treat empty rects and paths the same
Change-Id: Ie334a2f0aa354ea3b52f6143e2659fd74fdd6185
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334100
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Reed <reed@google.com>
2020-11-11 18:40:45 +00:00
Brian Osman
b4ce94431f Runtime effects: Fix error when mutating main's coords
These often became direct references to a varying, which can't be
mutated in GLSL. Fix this by copying them to a local variable in the
FP, and pointing all references at that.

Bug: skia:10918
Change-Id: I705e3c966b1d44fc4dfc3d4b40eb8e46110febdd
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334043
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
2020-11-11 16:56:25 +00:00
John Stiles
76013704ad Add unit tests for overflowing int and uint literal limits.
At present, we do not report any error; the values wrap silently.

Change-Id: I8c435cfdd81f6c2e5fd87e9c39c708138bf4ec82
Bug: skia:10932
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/333676
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
2020-11-11 16:11:15 +00:00
Adlai Holler
9902cff102 Use SkSpan to clean up GrOnFlushCallbackObject API
Also adjust the OpsTask terminology to the broader RenderTask term.

Change-Id: I8549e74a3e2f6b2caf765103f31243b776823c16
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332724
Commit-Queue: Adlai Holler <adlai@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
2020-11-11 14:34:15 +00:00
Brian Osman
24c18526a5 Make runtime effect error tests simpler / more explicit
I've wanted to write a few of these lately, and (just like the shader
tests later in the file), the rigid structure was making things
difficult. It also forced some of these to be overly clever (like
commenting out main). Switch to just providing all of the SkSL for each
test case.

Change-Id: I8aff34224fa6efbdc4d67628744a9cf499eb53b5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/333796
Auto-Submit: Brian Osman <brianosman@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
2020-11-11 01:14:44 +00:00
Mike Reed
282f8f4685 Recompute the last-move-to inside addPath()
Inspired by https://skia-review.googlesource.com/c/skia/+/330696

Bug: 1134474
Change-Id: I819cefa33f7b4245b3e527fade2260dde30e8a9f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/331776
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
2020-11-10 21:25:25 +00:00
John Stiles
7f7b48537c Fix flipped array dimensions in SkSL.
Change-Id: I6e44dd5c347b43b3a5cb135724083adbaf65cf27
Bug: skia:10924
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/333536
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
2020-11-10 18:08:29 +00:00