249cb20017..5f857839b9
git log 249cb200173f..5f857839b9ec --date=short --first-parent --format='%ad %ae %s'
2019-12-05 ianelliott@google.com Improve current multisampled renderbuffer/texture support
2019-12-04 syoussefi@chromium.org Remove build_apk_secondary_abi from gn files
2019-12-04 jmadill@chromium.org gn: Minor improvements.
2019-12-04 syoussefi@chromium.org Vulkan: Optimize unused uniform cleanup in glslang wrapper
2019-12-04 syoussefi@chromium.org Re-enable atomic counter tests on !windows
2019-12-04 jmadill@chromium.org Vulkan: Fold read access into write flags on buffer writes
2019-12-04 shrekshao@google.com Turn on fboRenderMipmap for D3D11
2019-12-04 courtneygo@google.com Split out core.compute_shader expectations
2019-12-04 courtneygo@google.com Add Compute Shared Memory Size Validation
2019-12-04 thomasanderson@chromium.org Expose GL_ARB_sync for GL
2019-12-04 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/vulkan-validation-layers/src 4fde9b750992..e72b61c7c20d (1 commits)
2019-12-04 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/SwiftShader b64fbfec4dcd..bbd0694f9ab2 (3 commits)
2019-12-04 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/spirv-tools/src 3ed4586044f2..e82a428605f6 (3 commits)
2019-12-04 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/glslang/src a4af7676feb0..e58c080d4f02 (1 commits)
Created with:
gclient setdep -r third_party/externals/angle2@5f857839b9ec
If this roll has caused a breakage, revert this CL and stop the roller
using the controls here:
https://autoroll.skia.org/r/angle-skia-autoroll
Please CC bsalomon@google.com on the revert to ensure that a human
is aware of the problem.
To report a problem with the AutoRoller itself, please file a bug:
https://bugs.chromium.org/p/skia/issues/entry?template=Autoroller+Bug
Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md
Cq-Include-Trybots: skia/skia.primary:Build-Debian9-Clang-x86_64-Release-ANGLE;skia/skia.primary:Test-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-Golo-GPU-QuadroP400-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC8i5BEK-GPU-IntelIris655-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE
Bug: None
Tbr: bsalomon@google.com
Change-Id: Ie309f15e7769ab1b972b07b2779825291ed7ba35
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/258098
Reviewed-by: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
This reverts commit d83912bc81.
Reason for revert: Chrome+Android complains about mutable constant GrQuadUtils::TessellationHelper::kFunctionTables; maybe also layout test failures?
Original change's description:
> Specialize inset/outset/reset based on known quad type
>
> Change-Id: I5ca35be216cf6e7d863f2d7638905a61a0b9ef00
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257392
> Commit-Queue: Michael Ludwig <michaelludwig@google.com>
> Reviewed-by: Robert Phillips <robertphillips@google.com>
TBR=robertphillips@google.com,michaelludwig@google.com
Change-Id: Ibc3ec464c42ba8f3a16ac9c773700868400f3ed9
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/258116
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
This is an automated CL created by the recipe roller. This CL rolls recipe
changes from upstream projects (e.g. depot_tools) into downstream projects
(e.g. tools/build).
More info is at https://goo.gl/zkKdpD. Use https://goo.gl/noib3a to file a bug.
depot_tools:
https://crrev.com/32a801b5dadfd40ef7999339559413408555ebdc [bot_update] Also bootstrap in the case that there are 0 pack files (iannucci@chromium.org)
TBR=borenet@google.com
Recipe-Tryjob-Bypass-Reason: Autoroller
Bugdroid-Send-Email: False
Change-Id: I6e17ba1d5505e2e422d1b047588ca421b49720f1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/258004
Reviewed-by: Recipe Roller <recipe-mega-autoroller@chops-service-accounts.iam.gserviceaccount.com>
Commit-Queue: Recipe Roller <recipe-mega-autoroller@chops-service-accounts.iam.gserviceaccount.com>
This is an automated CL created by the recipe roller. This CL rolls recipe
changes from upstream projects (e.g. depot_tools) into downstream projects
(e.g. tools/build).
More info is at https://goo.gl/zkKdpD. Use https://goo.gl/noib3a to file a bug.
depot_tools:
https://crrev.com/a0b8c5dcf1890916119ed180ab936948ec62cba1 Don't fetch tags in presubmit (danakj@chromium.org)
TBR=borenet@google.com
Recipe-Tryjob-Bypass-Reason: Autoroller
Bugdroid-Send-Email: False
Change-Id: Ie164f2a66a404ce5ee9256aed9dc621c832c43dd
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/258040
Reviewed-by: Recipe Roller <recipe-mega-autoroller@chops-service-accounts.iam.gserviceaccount.com>
Commit-Queue: Recipe Roller <recipe-mega-autoroller@chops-service-accounts.iam.gserviceaccount.com>
Switches ViewCountPair to store a sk_sp<GrSurfaceProxy> instead of a
GrSurfaceProxyView. It applies the origin from the view at initialization
time and then no longer needs it. All ViewCountPairs have the same swizzle
so that is stored once at the op level instead of per entry.
Additionally, re-arranges the metadata of GrTextureOp to ensure that it
packs into an 8byte aligned field. After first switching to
GrSurfaceProxyView, the op went to 144 bytes, and each entry was 32 bytes.
After Greg's GrSwizzle shrinkage, the op went to 136 bytes and each entry
went to 24 bytes. This CL further reduces the top to 120 bytes and each
entry is 16 bytes.
Change-Id: I912864db8846affc32620f60cc7d1d17cf0654fd
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257919
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Change-Id: I5ca35be216cf6e7d863f2d7638905a61a0b9ef00
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257392
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
I've landed fixes for most of these,
leaving -Wclass-memaccess as a TODO.
Bug: skia:9674
Change-Id: Ifb951bc66e022b48ff4b66e4555d3fe3c7ef5aaf
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257501
Reviewed-by: Ben Wagner aka dogben <benjaminwagner@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
The old code assumes a conservative size for vertices of 20 bytes,
but this only happens in perspective or strange DFT modes. This can
be reduced to 16 bytes in the non-perspective case a savings of 20%
Change-Id: Ife7e7cc4bd00798f173c5390c3100605908e6276
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257923
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
This feels kind of bad, but at least it's just a unit test?
Change-Id: I18740f846680218a00add40eb2398876c4fd19be
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/258023
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This is a big plumbing CL to pass the viewMatrix to the right places.
Change-Id: I6af67c6ba44c9c830ea52e8b50baafb88b40afc6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257894
Auto-Submit: Herb Derby <herb@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
* Turns out that dealloc is unnecessary. Some compilers
give an error when it is there.
* Use __bridge cast, since some compilers require it.
Cq-Include-Trybots: luci.skia.skia.primary:Build-Mac-Clang-arm64-Debug-iOS_Metal
Cq-Include-Trybots: luci.skia.skia.primary:Build-Mac-Clang-arm64-Release-iOS_Metal
Change-Id: I42d9ef16a5d6425c98a8a2b564bf4a06f26886f5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257677
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Auto-Submit: Hal Canary <halcanary@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
In order to be able to expose AAHairlineOp's programInfos we'll need to be able allocate the GPs in a different arena.
Bug: skia:9455
Change-Id: I786318dc4bf18dbc245cb26bbc995649de9104d4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257996
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Plan is to next move the clipper into utils / SkCanvas itself
Change-Id: Ia79c5bc2750a3a2719f901d255890a8c43ef1b42
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257916
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
This is an automated CL created by the recipe roller. This CL rolls recipe
changes from upstream projects (e.g. depot_tools) into downstream projects
(e.g. tools/build).
More info is at https://goo.gl/zkKdpD. Use https://goo.gl/noib3a to file a bug.
depot_tools:
https://crrev.com/0e60ecd30efa9a469b4bcc93006c253861a26164 [git-cl] Add graceful error handling to "git cl archive". (kmarshall@chromium.org)
TBR=borenet@google.com
Recipe-Tryjob-Bypass-Reason: Autoroller
Bugdroid-Send-Email: False
Change-Id: Ia83885e1f99fc4ebaae22cf5f914f0d4206a8772
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257908
Reviewed-by: Recipe Roller <recipe-mega-autoroller@chops-service-accounts.iam.gserviceaccount.com>
Commit-Queue: Recipe Roller <recipe-mega-autoroller@chops-service-accounts.iam.gserviceaccount.com>
Having it means you cannot write:
SkSize size;
float x, y;
size = {x , y};
clang allows it but GCC does not, claiming it is ambiguous between the
implicitly generated
SkSize& SkSize::operator=(const SkSize&)
and
SkSize& SkSize::operator=(const SkISize&)
clang gives the same error if the former is explicitly declared default.
Change-Id: I3b64436ef6aa669b3d87e7f37057c5dcb4add987
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257880
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
On 32-bit Release builds, GCC warns about this line in this test,
../../tests/SkRasterPipelineTest.cpp:225:40:
error: ‘int __builtin_memcmp_eq(const void*, const void*, unsigned int)’
reading 8 bytes from a region of size 0 [-Werror=stringop-overflow=]
!memcmp(&data[j][0], &buffer[j][0], sizeof(buffer[j])));
~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
I don't really see anything wrong, and I'm not really sure why it takes
offense here testing load_f16 and not in the very similar next chunk of
code testing load_af16. Anyway, it's simple enough to write a loop.
Change-Id: Ie4f598328dd7109660b9ae97a5a27b386e833fd5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257869
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Auto-Submit: Mike Klein <mtklein@google.com>
Bug: skia:9603
Change-Id: Ia8e020e15361e4a535719958a7e4035888dde14f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257681
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
GCC 8 is still tripping on GrWindowRectangles for mips64el:
../SRC/skia/tests/WindowRectanglesTest.cpp:55:42:
error: 'int __builtin_memcmp_eq(const void*, const void*, long unsigned int)'
reading 112 bytes from a region of size 16 [-Werror=stringop-overflow=]
REPORTER_ASSERT(reporter, !memcmp(B.data(), windowData,
(GrWindowRectangles::kMaxWindows - 1) * sizeof(SkIRect)));
../SRC/skia/tests/WindowRectanglesTest.cpp:69:42:
error: 'int __builtin_memcmp_eq(const void*, const void*, long unsigned int)'
reading 128 bytes from a region of size 16 [-Werror=stringop-overflow=]
REPORTER_ASSERT(reporter, !memcmp(B.data(), A.data(),
GrWindowRectangles::kMaxWindows * sizeof(SkIRect)));
Both A and B are filled up to the kMaxWindows limit (8) at this point,
but GCC still clearly thinks .data() is returning the 16-byte
fLocalWindows SkIRect[1] array.
I was curious to see if removing the (currently pointless) array helps GCC
get the gist. It doesn't, but I still kind of want to remove it.
Change-Id: I4ee9478b8c69061cb351d203d55749f68dcb939e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257873
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
converting from YUV to RGB on GPU.
Change-Id: I872e1e85f3efccce7bf93d2f3d5f2a841b8a5b0f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257680
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
In DDL-mode, this pulls the tessellation of the GrFillRectOp's vertices forward to record time.
Bug: skia:9601
Change-Id: I1a505845cc20f795d6f171b4c19d779475fdd240
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257685
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
In Release mode GCC 8 finds some issue I don't exactly understand when
using SkSTArray (or even SkTArray) here with MEM_COPY=true, for both
arrays, issuing one -Warray-bounds warning and one -Wstringop-overflow
warning for each array, ostensibly about the calls to resize_back().
(Oddly if you remove the initial reservations in the constructors, this
drops to just one warning each.)
My initial suspicion was some sort of aliasing issue, but I think the
SkPoint / 2x GrGLfloat pun might be legal, and even if it weren't, it
would not explain why the verbs array warns in the same manner.
Anyway I'd care more about this if the pattern of allocating an array
of a fixed runtime size weren't served better by SkAutoSTArray. No
warnings from GCC now.
Change-Id: I6959acc0500df4539e43e5d280c1e872b8c2fca3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257862
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Also fix some compile breakages from recent Skia changes.
No changes to Skia API.
Change-Id: Ifd0b1d89fb4ea3cf1f6a7170f2f74614276364ae
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257877
Commit-Queue: Stephen White <senorblanco@chromium.org>
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Mike Reed <reed@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Change-Id: I85477806b151773bc715f31bcdb05554107cc6b4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257700
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Previously, this was only checked for desktop GL, not for GLES. With [1], ANGLE
now exposes GL_ARB_sync for GLES, so we should extend the check. Also always
check for GL_APPLE_sync for consistency. It's technically out-of-spec for
GL_ARB_sync to exist outside of desktop GL, and for GL_APPLE_sync to exist
outside of GLES, but if the platform is listing the extensions, we should expect
them to work anyway.
Bug: chromium: 1028799
Change-Id: I411d2f106df0a2d7eb9b288053478ee79e060b81
R: reed@google.com
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257693
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Change-Id: Ib4baeec7dc7247b81c7e4a1ef7f85ef378e4be19
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257621
Commit-Queue: Eric Boren <borenet@google.com>
Reviewed-by: Heather Miller <hcm@google.com>
Reviewed-by: Ben Wagner aka dogben <benjaminwagner@google.com>
fb40d231c3..249cb20017
git log fb40d231c3e2..249cb200173f --date=short --first-parent --format='%ad %ae %s'
2019-12-03 syoussefi@chromium.org Translator: Rename TIntermInvariantDeclaration to ..GlobalQualifier..
2019-12-03 syoussefi@chromium.org glslang.l: refactor extension checks
2019-12-03 syoussefi@chromium.org Automatically call flex/bison if necessary
2019-12-03 syoussefi@chromium.org Vulkan: Fix WaW same-layout image transitions
2019-12-03 jmadill@chromium.org Revert "Add new test runner harness."
2019-12-03 geofflang@chromium.org D3D9: Log the exact error code generated by D3D9 device creation.
2019-12-03 geofflang@chromium.org Skip flaky BlendMinMaxTest.RGBA16F/ES2_Metal.
2019-12-03 geofflang@chromium.org Link to the autoroller configuration directory in the wrangling doc.
2019-12-03 thakis@chromium.org Remove an unused .gn file.
2019-12-03 flibitijibibo@gmail.com Fix UWP build
2019-12-03 geofflang@chromium.org GL: Make GL_EXT_texture_format_BGRA8888 enableable.
2019-12-03 jmadill@chromium.org Use SwiftShader-based ICD generation.
2019-12-03 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/glslang/src 83af46951202..a4af7676feb0 (1 commits)
2019-12-03 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/vulkan-tools/src 979b531e9fff..0a0625a3dca6 (1 commits)
2019-12-03 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/spirv-tools/src 47f3eb42641b..3ed4586044f2 (1 commits)
Created with:
gclient setdep -r third_party/externals/angle2@249cb200173f
If this roll has caused a breakage, revert this CL and stop the roller
using the controls here:
https://autoroll.skia.org/r/angle-skia-autoroll
Please CC bsalomon@google.com on the revert to ensure that a human
is aware of the problem.
To report a problem with the AutoRoller itself, please file a bug:
https://bugs.chromium.org/p/skia/issues/entry?template=Autoroller+Bug
Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md
Cq-Include-Trybots: skia/skia.primary:Build-Debian9-Clang-x86_64-Release-ANGLE;skia/skia.primary:Test-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-Golo-GPU-QuadroP400-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC8i5BEK-GPU-IntelIris655-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE
Bug: None
Tbr: bsalomon@google.com
Change-Id: I627cd0c33d0b6d8b00f46eda1bfaa399ae92203d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257798
Reviewed-by: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
* Convert from fInitialX, fInitialY to an SkPoint fInitialOrigin
* Start putting params in a consistent order
* inline and remove single use functions
Change-Id: I53eecf6098b475ef05f4f5c4133510ee1104073f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257756
Auto-Submit: Herb Derby <herb@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
* Make a single private section
* Reorganzie private section
* Start making const things const
Change-Id: I88081d02ee6034234d35d15306bffb47ce63ac8c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257688
Commit-Queue: Herb Derby <herb@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
Auto-Submit: Herb Derby <herb@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
Instead of always grabbing the first frame at load time, only do so for
single-frame images.
For multi-frame images, defer frame resolution until actually needed
(at seek time).
There's a slight complication in dealing with image scaling: since we
no longer know the intrinsic size of the frames upfront, we need to
set up a transform effect preemptively for multi-frame images, and
update on the fly when the frames are resolved. This is not
necessarily bad, because theoretically frames could have different
sizes.
Bug: skia:9686
Change-Id: Ib831d0e2ecad061ba52bdd8721e7598ea38c1e6e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257622
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Florin Malita <fmalita@chromium.org>
For clients not using JSON, the factories were inconsistent.
Change-Id: Ifd920fa1e18f5edffa12de238af8488406951e5b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257683
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This does several pure refactoring changes:
1. Adds reset() functions to each helper struct type instead of having
TessellationHelper impl. the struct's reset logic.
2. Moves computeDegenerateQuad() to EdgeEquations.
3. Separate adjustVertices() into adjust and adjustDegenerate varieties, and
have it take a signed distance. This makes it no longer branch on degeneracy
or on whether or not it's insetting or outsetting.
4. The top-level inset()/outset() functions apply the appropriate sign to the
edge distances, and use the appropriate adjust or adjustDegenerate function
based on the inset/outset degenerate flag in OutsetRequest. Essentially all
the old branching in adjustVertices() is baked into the choice of inset() or
outset().
5. Rearranged the structs and function definitions so no forward declarations
were needed, and functions in CPP matched H order.
The major functional change with this is that adjustDegenerate() now has
an optimized case for rectilinear quads, where it can continue to use
moveAlong() but with smaller inset distances.
Change-Id: Ifc7f4537a4e89e82c74e831ab1cd00ffc4daaa4f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/256777
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
This may fix a fuzzer bug that trips an
assert for empty rectangles. I could not reproduce
the fuzzer problem.
Bug: chromium:1029831
Change-Id: I88befb7c27d9ddc357e0a9e494e68125f8252f43
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257676
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
It's error-prone, and now unused.
Change-Id: Ibae848bdb121695521e66d6e8544a82678e22f7c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257660
Reviewed-by: Mike Klein <mtklein@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Change-Id: Ib9013fa9bedbbbf8a7787ef2a75d707dd6360cf8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257635
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: Ravi Mistry <rmistry@google.com>
Calls to Required() are subtly error prone without taking the alpha
types into account. Just so happens that we're doing half of that here
ourselves, and letting Required() handle the rest... best to just let
CSXformSteps handle all the logic.
This _does_ slightly_ change the values of unpremul and premul, in that
it's now possible for them both to be true when we need a color space
transform (when the source and dst are both premul). We don't think
this actually changes any of the logic that uses these bools though...
the canvas2D fast paths are unchanged because if needsColorConversion is
false, we'll still never set both at once. And the `convert` bool only
ever looks at the logical or of the two, so it won't change.
Change-Id: I051793cd84693e1a5dd367d036ab7f2e2308bbf8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257659
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
While we're asking if there are any color space xform steps to do,
let's look at all 5 of them, including whether we need to unpremul or
premul. The sprite blitter paths this logic guards can't handle
any change in color space _or_ alpha type, so this is a perfect fit.
Without this fix, we'd happily draw premul into unpremul using memcpy().
(Unpremul -> premul was already guarded by the check with my new note.)
Change-Id: Ic8d9e69f6c7bd2902ff77393de1da9cd940ea826
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/257657
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
