We can now pass in a context to `isAllowedInES2,` and it will return
true if the context supports the type. That is, ES2 types always return
true; non-ES2 types return true if the context is not strict-ES2.
Also, renamed `allowedInES2` to `isAllowedInES2` to better match similar
methods in Type.
Change-Id: Ia108bbde1354f6a14df05e8fb05742f04eefc375
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455661
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
DSLWriter is about to be renamed to something non-DSL-specific, and
(especially since we're in the middle of trying to break up some of our
big classes anyway) it didn't feel right to leave a pile of extremely
DSL-specific utility methods in it. And since it turns out that none of
these methods really do much of anything anymore, it seemed best to
just kill them all.
Change-Id: I5e257f87108a7a6db7fc1c01bab4e6c1544d60b0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455158
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
`Type::isPrivate` now works properly even on structs and arrays, so we
don't need two separate methods anymore.
Change-Id: Ic3e16e1315ebb0c8cec575f109af7e472a11ac8c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455660
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>
In the stack trace in the bug, #16 findOrCreateStrike() grabs the
lock, and then #8 Dump() makes calls that grab the lock. This is
only a problem with debug code.
Bug: chromium:1255085
Change-Id: Idcaad8bf2e8c2823bec899ee2b96a74ae666c164
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455501
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
We no longer need to explicitly pass this variable in since it lives in
the Context.
Change-Id: Ib8e753907578518e593cc3cd36f6f17871d95eec
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455659
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>
When requesting a font match for fonts like Sylfaen or Gabriola, which
absolutely do not have a bold or italic version, Windows silently
produces DirectWrite simulations which synthetic bolding or synthetic
italic or both.
From Chromium's perspective, we clearly do not want Windows to produce
silent simulations for us and this makes it impossible for us to
determine what the font's actual weight or slant is. As a result, we
are blocked from implementing the font-synthesis CSS property
correctly. For the latter, we need to be able to programmatically
enable setEmbolden() or setSkew() on SkFont and have Skia apply
artifical synthesis, instead of Windows instantiating such a font
silently.
Addresses TODO to avoid simulations in
SkFontStyleSet_DirectWrite::matchStyle.
Bug: skia:12455
Cq-Include-Trybots: luci.skia.skia.primary:Test-Win10-Clang-NUC5i7RYH-CPU-AVX2-x86_64-Debug-All-NativeFonts,Test-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Release-All-NativeFonts,Test-Win7-Clang-Golo-CPU-AVX512-x86_64-Debug-All-NativeFonts,Test-Win7-Clang-Golo-CPU-AVX512-x86_64-Debug-All-NativeFonts_GDI
Change-Id: If965d2501dff45ee7881544c6473b3fb36a8d90d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/450676
Commit-Queue: Ben Wagner <bungeman@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
This is a reland of 36f53ec7e1
Original change's description:
> Disallow constructors of ES3 types in ES2 code.
>
> The fuzzer found that we constructed TypeReferences without first
> checking for disallowed tyoes. (In fact, TypeReference creation had no
> error checking at all; it didn't even have Convert/Make functions.)
>
> Added proper Convert/Make to TypeReference, and used those calls to
> report errors or cause assertions if trying to make a TypeReference to a
> type that the program did not support.
>
> (While tracking down this bug, I added strict-ES2 type assertions to our
> constructor IR nodes as well. This helped pinpoint the error and seem
> reasonable to leave in, just in case.)
>
> Change-Id: I896b68ae9d3d9e1f30d7eba9fa594617ab851c74
> Bug: oss-fuzz:39540
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455498
> 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>
Bug: oss-fuzz:39540
Change-Id: Id8e323c22b18726214613b6061c08873048b7c69
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455617
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
We shouldn't return expressions that violate basic invariants (such as a
Constructor with an invalid Type) because future code should be able to
rely on those invariants to hold. Returning a Constructor with an
invalid Type broke some assertions I had added; these assertions checked
that the Constructor's type was valid before operating on it.
Change-Id: I861927ad042f30d4a1e20896149ce404f4160ffb
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455657
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
This makes the flag accessible from anywhere with a Context. Previously,
it was known by both the DSLWriter and IRGenerator, but otherwise
inaccessible unless manually passed to each function that needed it.
Change-Id: I029715c9049cb0810cb2960864632a82064921f9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455656
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This reverts commit 36f53ec7e1.
Reason for revert: breaks DS3Types test
Original change's description:
> Disallow constructors of ES3 types in ES2 code.
>
> The fuzzer found that we constructed TypeReferences without first
> checking for disallowed tyoes. (In fact, TypeReference creation had no
> error checking at all; it didn't even have Convert/Make functions.)
>
> Added proper Convert/Make to TypeReference, and used those calls to
> report errors or cause assertions if trying to make a TypeReference to a
> type that the program did not support.
>
> (While tracking down this bug, I added strict-ES2 type assertions to our
> constructor IR nodes as well. This helped pinpoint the error and seem
> reasonable to leave in, just in case.)
>
> Change-Id: I896b68ae9d3d9e1f30d7eba9fa594617ab851c74
> Bug: oss-fuzz:39540
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455498
> 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>
Bug: oss-fuzz:39540
Change-Id: I1dc3ccca477fcb9fe3f39cfe8af1fd54dcb18d6b
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455616
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
The fuzzer found that we constructed TypeReferences without first
checking for disallowed tyoes. (In fact, TypeReference creation had no
error checking at all; it didn't even have Convert/Make functions.)
Added proper Convert/Make to TypeReference, and used those calls to
report errors or cause assertions if trying to make a TypeReference to a
type that the program did not support.
(While tracking down this bug, I added strict-ES2 type assertions to our
constructor IR nodes as well. This helped pinpoint the error and seem
reasonable to leave in, just in case.)
Change-Id: I896b68ae9d3d9e1f30d7eba9fa594617ab851c74
Bug: oss-fuzz:39540
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455498
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>
This was inherited from SkSLParser, but is never called with anything
other than zero.
Change-Id: I423f943fa5cf9eeb611d1fa7a1e8a8884c8dd912
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/454744
Reviewed-by: John Stiles <johnstiles@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Change-Id: I5a57e2db46734ca08825e6aef7a6363bcaada45a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/453759
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
`optimize_comparison` asserted that its inputs were numbers. However,
it's also valid to compare boolean inputs. Fortunately, other than the
over-zealous assertion, the actual logic worked fine.
Change-Id: I8a9db000274b4993a4c303efa223a1ed72461a87
Bug: oss-fuzz:39513
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455296
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This should fix a failure in the ES2 conformance suite's "const_in_int".
Change-Id: I8b5487749291ef57712b8fe6c3949dc7c3e76883
Bug: skia:12499
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455157
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Previously, `Type::applyPrecisionQualifiers` would return a new type
(e.g. `mediump + float` returned `half`) but left the precision
qualifier flags as-is. This was implemented that way because the
modifiers were already baked into a pool, so mutating them was
difficult.
The rewritten DSLParser does not share this limitation--every place
where applyPrecisionQualifiers is used, the Modifiers are easily
mutable. As a result, `applyPrecisionQualifiers` can now clear the
precision-qualifier bits on the Modifier, meaning that `half` and a
`mediump float` will generate the exact same Type/Modifier combination.
This change fixes a bug where precision qualifiers were not allowed on
function parameters. (See `check_parameters` in FunctionDeclaration.cpp
to pinpoint the cause of the error. A less-invasive fix could have just
marked those modifier bits as allowed in `check_parameters`, but this
fix addresses the root of the issue and is honestly how I wanted
`applyPrecisionQualifiers` to work all along.)
Change-Id: I331813efa54138f469a0d5bff2d274cd3ce64b70
Bug: skia:12489
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455156
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This reverts commit 5f15c695f9.
Reason for revert: landed http://ag/15959743 to fix Android roll
Original change's description:
> Revert "Mark GLSL reserved names as reserved in SkSL grammar."
>
> This reverts commit 57f3fc4cde.
>
> Reason for revert: breaking Android roll
>
> Original change's description:
> > Mark GLSL reserved names as reserved in SkSL grammar.
> >
> > We now reject every reserved name in the ES2 docs as an unexpected
> > token, except for the rule that all names beginning with `gl_` are
> > reserved. (Unfortunately, sksl_frag bends the rules by directly
> > declaring a builtin variable named `gl_SecondaryFragColorEXT`.)
> >
> > Change-Id: I5dcb40b754720ca97fe3d80e2f9072beaa39fcdb
> > Bug: skia:11115
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/454737
> > Auto-Submit: John Stiles <johnstiles@google.com>
> > Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
> > Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
>
> Bug: skia:11115
> Change-Id: Ica56f48dc76ef1e52780acaf59b8ad9143637637
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/454860
> Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
> Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Bug: skia:11115
Change-Id: I012b8d4e03be7f9c888c26d912552412529b4fb6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/455159
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
This reverts commit 57f3fc4cde.
Reason for revert: breaking Android roll
Original change's description:
> Mark GLSL reserved names as reserved in SkSL grammar.
>
> We now reject every reserved name in the ES2 docs as an unexpected
> token, except for the rule that all names beginning with `gl_` are
> reserved. (Unfortunately, sksl_frag bends the rules by directly
> declaring a builtin variable named `gl_SecondaryFragColorEXT`.)
>
> Change-Id: I5dcb40b754720ca97fe3d80e2f9072beaa39fcdb
> Bug: skia:11115
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/454737
> Auto-Submit: John Stiles <johnstiles@google.com>
> Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
> Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Bug: skia:11115
Change-Id: Ica56f48dc76ef1e52780acaf59b8ad9143637637
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/454860
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
In complex programs with multiple functions, the Inliner can cause code
to be reordered in ways that cause a function call to be raised above
its declaration.
The Pipeline stage code generator will now emit a prototype for every
function defined in the program, before emitting any function bodies at
all.
With this change, ES2 conformance test `copy_global_inout_on_call` now
passes.
Change-Id: I85485710a34b778adef3cbc4a7ebe110a21a2a03
Bug: skia:12488
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/454742
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>
Cleanup:
Remove two unused from_8888
Change-Id: Ib87986a122169154e493d489fd1d24718c28f5c7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/453638
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
No functional changes, just continuing the process of moving code out
of IRGenerator in hopes of its eventual disappearance.
Change-Id: I3507b85c5493c61812869c9bae53de8af89ee129
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/453941
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
Previously we did not have a Pipeline callback function for prototyping
a function, so prototypes would be discarded during translation. This
failure mode can be seen in http://review.skia.org/454741, where
FunctionPrototype.sksl is made more complex (thwarting the inliner).
This causes us to emit invalid GLSL, and dm asserts/fails in the SkSL
tests: http://screen/4PkEEWn4m4tF5e7
This CL makes the same changes to FunctionPrototype, but does not crash.
Change-Id: Ia342c7811a454f62f52677440d247e628a1bdc4f
Bug: skia:12488
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/454740
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
We now reject every reserved name in the ES2 docs as an unexpected
token, except for the rule that all names beginning with `gl_` are
reserved. (Unfortunately, sksl_frag bends the rules by directly
declaring a builtin variable named `gl_SecondaryFragColorEXT`.)
Change-Id: I5dcb40b754720ca97fe3d80e2f9072beaa39fcdb
Bug: skia:11115
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/454737
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
`input` is a reserved word in GLSL. http://screen/85m4iRwvJRadKbV
Change-Id: I94ac0901964daa6c5fa01fc40ea3dbd733f821b7
Bug: skia:11115
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/454739
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Fuchsia was interpreting this assert differently than all other
compilers.
int ia = ?,
ib = ?;
SkASSERT(-ib <= ia && ia <= UINT16_MAX - ib);
It interpreted UINT16_MAX - ib as an unsigned int forcing ia to be
interpreted as an unsigned int causing a negative number to be a very
large unsigned number.
Just use the value 65535 instead of UINT16_MAX.
Change-Id: I97b244d21f5f1624427d62d67c754def585ce03c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/454256
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Moving stuff out of IRGenerator in preparation for its eventual
deletion.
Change-Id: Ib4290158670fecbca02bc71452b9bf4691094fe3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/453943
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
This is a reland of ef96fa9e83
Addition: add the flag SK_SUPPORT_LEGACY_BILERP_HIGHP to
enable fixing chrome unittests and blink tests
Original change's description:
> lowp bilerp_clamp_8888
>
> Use scaling multipy, _mm_mulhrs_epi16(Intel) and vqrdmulhq_s16(ARM),
> to implement bilerp_8888_clamp.
>
> This CL results in 756usec to 590usec improvement for
> samplingoptions_filter_1_mipmap_0 on Intel. For ARM, this improvement
> is 1180usec -> 897usec.
>
> This CL introduces scaled_mult which takes fixed-point numbers on the
> interval [-1, 1) and returns a result which is rescaled to the
> interval [-1, 1).
>
> It also introduces the notion of constrained_add(I16, U16) where the
> result is guaranteed to be U16. This avoids moving to a 32-bit integer
> for during the computation.
>
> Change-Id: I410e494364039df63e5976f433f7e68355e9cfbf
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/443896
> Reviewed-by: Brian Osman <brianosman@google.com>
> Commit-Queue: Herb Derby <herb@google.com>
Change-Id: I43e572226e11a75a6e7de5b124f2b1ae3cc37c87
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/453556
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
This is a reland of I060988c63edbfeb628d5a6c34cda3e159066114a
Original change's description:
> Don't keep a non-msaa FBO with EXT_multisampled_render_to_texture
>
> This seems to cause bugs on various platforms. Instead, do the same
> thing we do for dmsaa: reconfigure a single FBO as msaa or not,
> based on need.
>
> Bug: chromium:1222095
> Change-Id: I4ca4aff32772ee2a9607c92c586ae9494f955229
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/442565
> Commit-Queue: Jim Van Verth <jvanverth@google.com>
> Reviewed-by: Brian Salomon <bsalomon@google.com>
Change-Id: I0bb9e9178170da391be0a295758879994bf4be0d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/452724
Reviewed-by: Chris Dalton <csmartdalton@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This will move highp AVX users from AVX to SSE41. This is about
7.5%[1] of users for software.
The underlying problem is that AVX works well with floating point,
it has little support for integer operations. So for lowp, AVX works
as SSE41 and in highp it works as HSW and SKX. The rcp_precise
has the distinction as being the only function that works as highp and
lowp. So, when compiled with -mavx it doesn't compile because in
highp it has 8 elements and in lowp it has 4 elements. So, move
it to be with SSE41 for highp.
[1] https://uma.googleplex.com/p/chrome/histograms?sid=ec095beddef4b2771a6a3f12a78f32c1
Bug: skia:12453
Change-Id: I2929312ef36cc0c0a68515ac74e5d953c0b7fd09
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/453716
Reviewed-by: Brian Osman <brianosman@google.com>
Now that DSLParser is used for everything we can get rid of this
unused code.
Change-Id: I7b586a7d7a51ac9df7d2c52baa1f525233f489b9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/443076
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This reverts commit 65967ab2c9.
Reason for revert: used in Android - http://screen/3fk6K3RUWZG8x5f
Original change's description:
> Remove SkRuntimeShader::makeImage.
>
> We don't have any known users, and it no longer exposes anything that
> a user can't just do directly.
>
> Change-Id: Id653a6be3f265a2847b1670f3e6c054cf2d094a2
> Bug: skia:12482
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/453142
> Auto-Submit: John Stiles <johnstiles@google.com>
> Reviewed-by: Brian Osman <brianosman@google.com>
> Commit-Queue: John Stiles <johnstiles@google.com>
Bug: skia:12482
Change-Id: I6cc2e79796f53c42e9f53b7d9927c853e7e2d71c
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/453317
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Commit-Queue: Kevin Lubick <kjlubick@google.com>
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Owners-Override: Kevin Lubick <kjlubick@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
This reverts commit ef96fa9e83.
Reason for revert: Breaks Chrome's pixel and layout tests.
Original change's description:
> lowp bilerp_clamp_8888
>
> Use scaling multipy, _mm_mulhrs_epi16(Intel) and vqrdmulhq_s16(ARM),
> to implement bilerp_8888_clamp.
>
> This CL results in 756usec to 590usec improvement for
> samplingoptions_filter_1_mipmap_0 on Intel. For ARM, this improvement
> is 1180usec -> 897usec.
>
> This CL introduces scaled_mult which takes fixed-point numbers on the
> interval [-1, 1) and returns a result which is rescaled to the
> interval [-1, 1).
>
> It also introduces the notion of constrained_add(I16, U16) where the
> result is guaranteed to be U16. This avoids moving to a 32-bit integer
> for during the computation.
>
> Change-Id: I410e494364039df63e5976f433f7e68355e9cfbf
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/443896
> Reviewed-by: Brian Osman <brianosman@google.com>
> Commit-Queue: Herb Derby <herb@google.com>
Change-Id: I715be7bdbb706725770560e49a4d9f0777d2c22b
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/453256
Auto-Submit: Herb Derby <herb@google.com>
Commit-Queue: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
We don't have any known users, and it no longer exposes anything that
a user can't just do directly.
Change-Id: Id653a6be3f265a2847b1670f3e6c054cf2d094a2
Bug: skia:12482
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/453142
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
This reverts commit cc91452f0a.
Change-Id: I7eff0ddeebef4ce298893f9b3ba410b09647e9a4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/453138
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This reverts commit a909dd6b8d.
Turns out those reportPendingErrors() calls I removed were in fact
necessary, just not on any of the CQ bots.
Change-Id: I8be0898ac0b41dbb703a35f705cac06ca716c0b7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/453077
Reviewed-by: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
The Skia build of HarfBuzz does not use ICU, so there is no need to
initialize it. If a user uses a HarfBuzz build which does use ICU, they
will need enuse that ICU is initialized in their own way.
Change-Id: I89bd518dbf3c58aaaee2463a26d6402738dfbf97
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/453140
Reviewed-by: Julia Lavrova <jlavrova@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
Use scaling multipy, _mm_mulhrs_epi16(Intel) and vqrdmulhq_s16(ARM),
to implement bilerp_8888_clamp.
This CL results in 756usec to 590usec improvement for
samplingoptions_filter_1_mipmap_0 on Intel. For ARM, this improvement
is 1180usec -> 897usec.
This CL introduces scaled_mult which takes fixed-point numbers on the
interval [-1, 1) and returns a result which is rescaled to the
interval [-1, 1).
It also introduces the notion of constrained_add(I16, U16) where the
result is guaranteed to be U16. This avoids moving to a 32-bit integer
for during the computation.
Change-Id: I410e494364039df63e5976f433f7e68355e9cfbf
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/443896
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Prereq changes have landed in chrome.
Bug: skia:11809
Change-Id: Id727cd4146104ef93c682f495b4ed37e793a5aa1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/453076
Commit-Queue: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Auto-Submit: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
This removes SkSLParser from the last code path where it was still in
use, now instead using DSLParser for module loading.
Change-Id: I3858729ee2949e2900f0bcfc3f622aa4b5ef3d5d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/452719
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
This is a reland of 949dfb5ffd
Original change's description:
> Use a single FBO for EXT_multisampled_render_to_texture targets
>
> Uses a single FBO and changes the attachments based on whether we need
> multisample or not. This is an effort to work around presumed driver
> bugs from having the same texture bound to different FBOs with different
> sample counts.
>
> Bug: chromium:1222095
> Change-Id: I1347a149524be95181a4d3fb0a0da7bfe92a2ec9
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/442737
> Commit-Queue: Jim Van Verth <jvanverth@google.com>
> Reviewed-by: Brian Salomon <bsalomon@google.com>
Change-Id: I678ae8b5f7a3e8cca354526a47c9edb822607a38
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/452317
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Right now the crop image filter is hidden in src/ and is only used in
a new GM to test out its functionality. In later CLs the plan will be to
migrate individual filter effects away from the built-in crop rect to
composing themselves with this image filter.
Eventually, the crop filter will be made public as part of the
SkImageFilters factories so there can be independent control on input
and output cropping. Its features will also be expanded to support more
tile modes than just kDecal.
Once all other effects rely on this crop filter, the legacy system can
be deleted and the effects can be more easily updated to the rest of the
new image filter API. This crop filter is actually the first image
filter to be implemented only in terms of the new API (and skif coord
space types). This highlighted a few pain points that still exist in
terms of the API, where I have added comments to capture how they could
be improved. Unfortunately, they aren't addressable until the rest of
the implementations have been migrated.
Bug: skia:9296
Change-Id: I12e96b679c246a9ec4e1e61f640414a11f1b6bb5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/451597
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
If a VarDeclaration line contained multiple variables, and the first
variable had an illegal initializer-expression, the Declare() would
return a Nop. AddVarDeclaration did not expect to see a Nop and would
assert once we tried to process the second var-declaration. Now, we
allow adding var declarations to a Nop.
Bulked up some tests to cover local and global variables (since those
are parsed in separate functions) and to check both the first
initializer as well as follow-on initializers (since those are parsed in
separate parts of the var-decl handler).
Change-Id: I66341191698175b490a659715cb8edaafe2f75ae
Bug: oss-fuzz:39032
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/452696
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Previously, we were abundantly cautious about producing premultiplied
colors with RGB > A, when the color-type was normalized fixed-point.
I'm not seeing any evidence that code cares about that. If it does, I
suggest we fix the issue at the API border, or via dedicated pass. In
the common case, this makes Skia less opinionated, which (in general)
makes it more versatile.
Change-Id: Iae524ff61f6c81073c34d0f2dced973c229cdfb7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/452558
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
