Change skiaserve color palette for drawing batch bounds

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1756603002

Review URL: https://codereview.chromium.org/1756603002

Fix MSVC 2015 compiler complaint

This is/was slowly, but surely, driving me insane:

DM.cpp(150): warning C4838: conversion from 'DWORD' to 'int' requires a narrowing conversion

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1749363002

Review URL: https://codereview.chromium.org/1749363002

Disabling calls to TexParameteri when the values do not exist on ES2.

BUG=590804
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1750833003

Review URL: https://codereview.chromium.org/1750833003

Adding anisotropic mipmap levels to SkMipMap.

Adding 1x2, 1x3, 2x1, 3x1 filters to SkMipMap and enabling SkMipMap to generate anisotropic mip levels.

BUG=590804
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1750303002

Review URL: https://codereview.chromium.org/1750303002

Handle spans in sampling.

Older version copied over.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1743123004

Review URL: https://codereview.chromium.org/1743123004

Revert of Progress on gamma-correctness in the GPU backend. Fixed conversion of color and profile type to pix… (patchset #1 id:1 of https://codereview.chromium.org/1750383002/ )

Reason for revert:
GM breakage. Changes to SkGr.cpp appear to be altering behavior on a variety of tests. Debugging...

Original issue's description:
> Progress on gamma-correctness in the GPU backend. Fixed conversion of color and profile type to pixel config, which makes many things "just work".
>
> Added (color=8888|f16|srgb) option to gpu configurations, along with gpuf16, gpusrgb, and anglesrgb predefined configs. Runs the gpu backend in gamma-correct mode (with either FP16 linear or sRGB 8888 frambuffers).
>
> BUG=skia:
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1750383002
>
> Committed: https://skia.googlesource.com/skia/+/a6f58194733c1c50e4fe5f98585e42344f29b6f0

TBR=mtklein@google.com,bsalomon@google.com
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1755553003

Check bounds before casting float to integer in SamplePatch.cpp and SkBlurMaskFilter::CreateEmboss.

BUG=skia:4632
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1693013002

Review URL: https://codereview.chromium.org/1693013002

Fix FrontBufferedStreamTest

TBR=scroggo@google.com

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1747423002

Review URL: https://codereview.chromium.org/1747423002

Implement various vulkan stencil clears and fix various stenciling bugs.

BUG=skia:4973
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1755753002

Review URL: https://codereview.chromium.org/1755753002

Progress on gamma-correctness in the GPU backend. Fixed conversion of color and profile type to pixel config, which makes many things "just work".

Added (color=8888|f16|srgb) option to gpu configurations, along with gpuf16, gpusrgb, and anglesrgb predefined configs. Runs the gpu backend in gamma-correct mode (with either FP16 linear or sRGB 8888 frambuffers).

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1750383002

Review URL: https://codereview.chromium.org/1750383002

Make GrDrawContext::internalDrawPath cons up its own GrPipelineBuilder

Hopefully, this better defines (and reduces) the lifetime and reuse of GrPipelineBuilder objects in GrDrawContext.

TBR=bsalomon@google.com

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1730903007

Committed: https://skia.googlesource.com/skia/+/00fddebe56fabea67dcc08762805c1294eebf5bf

Review URL: https://codereview.chromium.org/1730903007

turn on f16 and sRGB

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1748113002

Review URL: https://codereview.chromium.org/1748113002

Fix bug in SkGifCodec / Switch SkImageDec tests to use Codec

SkImageDecoder is still used throughout tests, tools, gms etc.
Deleting it from tests is an easy first step.

Bonus is that we add tests of SkCodec.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1733863003

Review URL: https://codereview.chromium.org/1733863003

Add cap and builder feature for multisample interpolation

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1722363002

Review URL: https://codereview.chromium.org/1722363002

dm: gamma_correct: true/false -> "yes"/"no"

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1755693002

NOTREECHECKS=true

Review URL: https://codereview.chromium.org/1755693002

add gm for distance field bug

Added gm to illustrate this bug. If you trace through
the red path, it uses GrAADistanceFieldPathRenderer
when it draws without antialiasing.

R=jvanverth@google.com
BUG=skia:4864
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1755683002

Review URL: https://codereview.chromium.org/1755683002

Unbust the old debugger

AFAICT the recent STL changes have introduced some QT conflicts.

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1753813002

Review URL: https://codereview.chromium.org/1753813002

remove untested blit optimization

R=mtklein@google.com
BUG=skia:5016
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1746423003

Review URL: https://codereview.chromium.org/1746423003

Move some GrAuditTrail fuctions to cpp file

TBR=ethannicholas@google.com
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1753753002

Review URL: https://codereview.chromium.org/1753753002

isolate sk_sp from larger cl

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1752683002

Review URL: https://codereview.chromium.org/1752683002

SampleApp now displays gamma-correct content in "S32" and "F16" modes. The default framebuffer is always (?) sRGB, but querying it to determine that is difficult or impossible. In addition, when running in "L32" mode, we stil want to exhibit the old behavior, where we pretended that it was not sRGB. Thus, set the pixel config of the render target based on the window's color and profile types. That triggers our code to turn on linear->sRGB conversion-on-write in flushRenderTarget.

NOte: Angle does not display correctly, due to limitations in the set of ES2 extensions it exposes. However, it can still be used to generate gamma-correct images (via DM, etc...).

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1752473002

Review URL: https://codereview.chromium.org/1752473002

SkDebugCanvas ignore batch bounds if they are offscreen

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1754633002

Review URL: https://codereview.chromium.org/1754633002

Add swizzle for rgb8888.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1746153002
CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Review URL: https://codereview.chromium.org/1746153002

print stack trace on crash, tweak formatting, _Exit hard

Mac example:

1 srcs * 1 sinks + 0 tests == 1 tasks
0ns elapsed, 1 active, 0 queued, 11MB RAM, 11MB peak
f16 gm  dstreadshuffle
../../src/core/SkXfermode.cpp:935: fatal error: ""r.fVec[i] >= min && r.fVec[i] <= max""

Caught signal 6 [Abort trap: 6], was running:
f16 gm  dstreadshuffle

Stack trace:
    0   dm                                  0x000000010f23fc9d _ZZL19setup_crash_handlervENK3$_5clEi + 397
    1   dm                                  0x000000010f23fb06 _ZZL19setup_crash_handlervEN3$_58__invokeEi + 22
    2   libsystem_platform.dylib            0x00007fff9d44e52a _sigtramp + 26
    3   ???                                 0x00007fff9900b902 0x0 + 140735760349442
    4   libsystem_c.dylib                   0x00007fff96a836e7 abort + 129
    5   dm                                  0x000000010fcb67c9 _Z17sk_abort_no_printv + 9
    6   dm                                  0x000000010f90fc1f _ZL11assert_unitRK6SkPM4f + 175
    7   dm                                  0x000000010f90e957 _Z7proc_4fIXadL_ZL10lighten_4fRK4SkNxILi4EfES3_EEE6SkPM4fRKS4_S6_ + 103
    8   dm                                  0x000000010f918946 _Z10xfer_u64_nIL7DstType1EEvPK10SkXfermodePyPK6SkPM4fiPKh + 182
    9   dm                                  0x000000010f7a5f3e _ZN22SkState_Shader_BlitterI7State64E9blitAntiHEiiPKhPKs + 398
    10  dm                                  0x000000010f3c730c _ZN9SkBlitter10blitAntiH2Eiijj + 108
    11  dm                                  0x000000010f8d76f4 _ZN25Vertish_SkAntiHairBlitter7drawCapEiiii + 164
    12  dm                                  0x000000010f8d4d92 _ZL16do_anti_hairlineiiiiPK7SkIRectP9SkBlitter + 3058
    13  dm                                  0x000000010f8d40f4 _ZN6SkScan15AntiHairLineRgnEPK7SkPointiPK8SkRegionP9SkBlitter + 900
    14  dm                                  0x000000010f8d8fff _ZL8hairquadPK7SkPointPK8SkRegionP9SkBlitteriPFvS1_iS4_S6_E + 719
    15  dm                                  0x000000010f8da838 _Z9hair_pathILN7SkPaint3CapE0EEvRK6SkPathRK12SkRasterClipP9SkBlitterPFvPK7SkPointiPK8SkRegionS9_E + 1048
    16  dm                                  0x000000010f8d86dc _ZN6SkScan12AntiHairPathERK6SkPathRK12SkRasterClipP9SkBlitter + 44
    17  dm                                  0x000000010f7da8ff _ZNK6SkDraw8drawPathERK6SkPathRK7SkPaintPK8SkMatrixbbP9SkBlitter + 2239
    18  dm                                  0x000000010f774993 _ZNK6SkDraw8drawPathERK6SkPathRK7SkPaintPK8SkMatrixb + 83
    19  dm                                  0x000000010f774031 _ZN14SkBitmapDevice8drawPathERK6SkDrawRK6SkPathRK7SkPaintPK8SkMatrixb + 97
    20  dm                                  0x000000010f773f6b _ZN14SkBitmapDevice8drawOvalERK6SkDrawRK6SkRectRK7SkPaint + 139
    21  dm                                  0x000000010f7b5072 _ZN8SkCanvas10onDrawOvalERK6SkRectRK7SkPaint + 626
    22  dm                                  0x000000010f7b38a5 _ZN8SkCanvas8drawOvalERK6SkRectRK7SkPaint + 69
    23  dm                                  0x000000010f7b9ee6 _ZN8SkCanvas10drawCircleEfffRK7SkPaint + 294
    24  dm                                  0x000000010f614039 _ZN6skiagm14DstReadShuffle9drawShapeEP8SkCanvasP7SkPaintNS0_9ShapeTypeE + 249
    25  dm                                  0x000000010f613b89 _ZN6skiagm14DstReadShuffle6onDrawEP8SkCanvas + 425
    26  dm                                  0x000000010f268b17 _ZN6skiagm2GM11drawContentEP8SkCanvas + 71
    27  dm                                  0x000000010f268a6c _ZN6skiagm2GM4drawEP8SkCanvas + 60
    28  dm                                  0x000000010f25186b _ZNK2DM5GMSrc4drawEP8SkCanvas + 123
    29  dm                                  0x000000010f256de7 _ZNK2DM10RasterSink4drawERKNS_3SrcEP8SkBitmapP9SkWStreamP8SkString + 295
    30  dm                                  0x000000010f246a96 _ZN4Task3RunERKS_ + 358
    31  dm                                  0x000000010f2447e5 _ZZ7dm_mainvENK3$_3clEv + 21
    32  dm                                  0x000000010f2447bd _ZNSt3__128__invoke_void_return_wrapperIvE6__callIJRZ7dm_mainvE3$_3EEEvDpOT_ + 45
    33  dm                                  0x000000010f24475c _ZNSt3__110__function6__funcIZ7dm_mainvE3$_3NS_9allocatorIS2_EEFvvEEclEv + 44
    34  dm                                  0x000000010f406aba _ZNKSt3__18functionIFvvEEclEv + 26
    35  dm                                  0x000000010f8fc168 _ZN12_GLOBAL__N_110ThreadPool4WaitEP8SkAtomicIiL15sk_memory_order5EE + 328
    36  dm                                  0x000000010f8fc00c _ZN11SkTaskGroup4waitEv + 28
    37  dm                                  0x000000010f23e9c4 _Z7dm_mainv + 1940
    38  dm                                  0x000000010f23fa77 main + 39
    39  dm                                  0x000000010f23e1a4 start + 52
    40  ???                                 0x0000000000000007 0x0 + 7

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1753503002

Review URL: https://codereview.chromium.org/1753503002

Revert of Progress on gamma-correctness in the GPU backend. Fixed conversion of color and profile type to pix… (patchset #3 id:40001 of https://codereview.chromium.org/1746253002/ )

Reason for revert:
Fixing the build.

Original issue's description:
> Progress on gamma-correctness in the GPU backend. Fixed conversion of color and profile type to pixel config, which makes many things "just work".
>
> Added (color=8888|f16|srgb) option to gpu configurations, along with gpuf16, gpusrgb, and anglesrgb predefined configs. Runs the gpu backend in gamma-correct mode (with either FP16 linear or sRGB 8888 frambuffers).
>
>
> BUG=skia:
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1746253002
>
> Committed: https://skia.googlesource.com/skia/+/eef980270d3385fee340eb1633962fe3ba8b7132

TBR=mtklein@google.com,egdaniel@google.com,bsalomon@google.com
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1748823002

Move drawDRRect back to GrDrawContext

This still leaves GrDrawContext fiddling around with specialized fragment processors but it does allow for different handling of the DRRects (e.g., for instanced drawing).

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1750533003

Review URL: https://codereview.chromium.org/1750533003

Progress on gamma-correctness in the GPU backend. Fixed conversion of color and profile type to pixel config, which makes many things "just work".

Added (color=8888|f16|srgb) option to gpu configurations, along with gpuf16, gpusrgb, and anglesrgb predefined configs. Runs the gpu backend in gamma-correct mode (with either FP16 linear or sRGB 8888 frambuffers).

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1746253002

Review URL: https://codereview.chromium.org/1746253002

Simplify the poly union.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1752433002

Review URL: https://codereview.chromium.org/1752433002

Fix overlooked bilerp fallback.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1746093002

Review URL: https://codereview.chromium.org/1746093002

fix minor bug drawing batch bounds by default

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1746953002

Review URL: https://codereview.chromium.org/1746953002

Remove internal calls to SkImage::getTexture

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1744243002

Review URL: https://codereview.chromium.org/1744243002

Fixup some of the batch info in skiaserve

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1750653002

Review URL: https://codereview.chromium.org/1750653002

Disabling mipmap generation until anisotropic mipmap levels are generated.

BUG=590804
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1744413002

Review URL: https://codereview.chromium.org/1744413002

Improve horizontal baseline detection.

The goal is to hint the baseline when hinting is possible, which is
to say when the glyphs are be aligned with the pixel grid. The
current code has three shortcomings.

1. correctly snaps when the horizontal baseline is on the x-axis but
not when on the y-axis. Instead it is snapping the horizontal
baseline when there is y-skew.

2. tests against the full device matrix instead of the relaxed
matrix used by the scaler context.

3. has range issues when relaxing the matrix.

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1740163002

Review URL: https://codereview.chromium.org/1740163002

Render batch bounds as stroke rects

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1745063002

Review URL: https://codereview.chromium.org/1745063002

This CL adds glTexStorage support.

For us to take advantage of immutable texture storage, we would need to
know in advance that the texture will not be changing allocated size.

In some cases we cannot know this in advance: we will sometimes later
need mipmaps to be allocated and generated. However, in the cases where
we know in advance that we are allocating mipmaps we can take advantage
of immutable storage.

BUG=476416
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1570173004

Review URL: https://codereview.chromium.org/1570173004

Swarming Bots: fix resource path

TBR=rmistry
NOTRY=true
BUG=skia:4763
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1749873002

Review URL: https://codereview.chromium.org/1749873002

Swarming bots: Fix tester out dir

TBR=rmistry
NOTRY=true
BUG=skia:4763
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1751463002

Review URL: https://codereview.chromium.org/1751463002

Modernize SkSpinlock.

 - Use std::atomic directly.
 - No more need for SkPODSpinlock or SK_DECLARE_STATIC_SPINLOCK.

Now simple code like this works as you'd hope:
    static SkSpinlock gLock;

That is, it starts unlocked and there's no static initializer.

std::atomic_flag would make this terser and standard-guaranteed,
but ATOMIC_FLAG_INIT caused not-yet-implemented errors on MSVC 2013.
The generated code for this approach is identical.

It appears the implicit constructor is constexpr when all the member
initializers are.  I'm hoping this way of producing constexpr constructors
without typing "constexpr" gives us a way to eliminate more SkFoo / SkBaseFoo
distinctions and SK_DECLARE_STATIC_FOO.  This was certainly the easiest.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1734383002

Review URL: https://codereview.chromium.org/1734383002

Fix Swarming compile out dir

TBR=rmistry
NOTRY=true
BUG=skia:4763
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1743353002

Review URL: https://codereview.chromium.org/1743353002

Revert of DM: disable ignored and very niche configs on bots. (patchset #1 id:1 of https://codereview.chromium.org/1745073002/ )

Reason for revert:
remote-8888 just caught a bug.  I am disproved.

Original issue's description:
> DM: disable ignored and very niche configs on bots.
>
> These are all probably just wasting time:
>   - upright-matrix-{8888,gpu} and remote[_cache] are ignored in Gold
>   - sp and 2ndpic test very niche areas of SkPicture
>     that are not under active change
>   - twice-8888 was never useful... the twice- via was meant for local
>     testing of GPU batching / caching.  Drawing into 8888 was moot.
>
> These configs are all useful to keep around for local testing, but probably are
> not worth the extra runtime or especially burden to have to think about them.
>
> If we should happen to stumble into bugs these configs would have caught,
> let's add them back with a note about those bugs to justify their need to run.
> I'm not convinced any exist today.
>
> This cuts ~30s off the CQ.
>
> BUG=skia:
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1745073002
>
> Committed: https://skia.googlesource.com/skia/+/7b6d3a30cf18377935c800603fc00fe0966d15de

TBR=halcanary@google.com,mtklein@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1749583003

Use enclosing int rect for picture shader tiles

To avoid rounding to zero for very small tiles.

BUG=chromium:590447
R=reed@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1749763002

Review URL: https://codereview.chromium.org/1749763002

Fix uninitialized variable warning.

CQ_INCLUDE_TRYBOTS=client.skia.compile:Build-Ubuntu-GCC-Mips64-Release-Android-Trybot

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1743273002

Review URL: https://codereview.chromium.org/1743273002

remove xmlparser includes from samples

Remove all #include SkXMLParser.h from SampleApp samples.

R=tomhudson@google.com
BUG=skia:309
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1737383002

Review URL: https://codereview.chromium.org/1737383002

hack up get_images_from_skps

I couldn't get the version at head to give me any images,
so I decided to rewrite it.  Does something like this work
for you?

BUG=skia:5010,skia:5005

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1737593003

Review URL: https://codereview.chromium.org/1737593003

DM: disable ignored and very niche configs on bots.

These are all probably just wasting time:
  - upright-matrix-{8888,gpu} and remote[_cache] are ignored in Gold
  - sp and 2ndpic test very niche areas of SkPicture
    that are not under active change
  - twice-8888 was never useful... the twice- via was meant for local
    testing of GPU batching / caching.  Drawing into 8888 was moot.

These configs are all useful to keep around for local testing, but probably are
not worth the extra runtime or especially burden to have to think about them.

If we should happen to stumble into bugs these configs would have caught,
let's add them back with a note about those bugs to justify their need to run.
I'm not convinced any exist today.

This cuts ~30s off the CQ.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1745073002

Review URL: https://codereview.chromium.org/1745073002

Swarming Tests: create tmp dir if it doesn't exist

TBR=rmistry
NOTRY=true
BUG=skia:4763
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1749623003

Review URL: https://codereview.chromium.org/1749623003

Swarming tests: Fix uninteresting hash file path

TBR=rmistry
NOTRY=true
BUG=skia:4763
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1746063002

Review URL: https://codereview.chromium.org/1746063002

tile spans

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1719333002

Review URL: https://codereview.chromium.org/1719333002

Add abilitly to query audit trail for batches by draw op

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1745513002

Committed: https://skia.googlesource.com/skia/+/9b48a6e3f862076018cc7d63b180b6340f4873cd

Review URL: https://codereview.chromium.org/1745513002

Add gamma_correct option field to dm.json

E.g.

{
   "max_rss_MB" : 23,
   "results" : [
      {
         "key" : {
            "config" : "pdf",
            "name" : "gamma",
            "source_type" : "gm"
         },
         "md5" : "c5dfb531f4d76c77c3305b6a04733262",
         "options" : {
            "ext" : "pdf",
            "gamma_correct" : false
         }
      },
      {
         "key" : {
            "config" : "8888",
            "name" : "gamma",
            "source_type" : "gm"
         },
         "md5" : "6177860ed24106446d3a9087527e67bf",
         "options" : {
            "ext" : "png",
            "gamma_correct" : false
         }
      },
      {
         "key" : {
            "config" : "f16",
            "name" : "gamma",
            "source_type" : "gm"
         },
         "md5" : "213f80145953ecd4c71e0612447b2ad9",
         "options" : {
            "ext" : "png",
            "gamma_correct" : true
         }
      }
   ]
}

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1741973002

Review URL: https://codereview.chromium.org/1741973002

Enable RAW codec for Windows

* Fix the exception catching
* Set preprocessor differently for MSVC

BUG=skia:4889(b/26958348)
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1738913002

Committed: https://skia.googlesource.com/skia/+/474e4c3dd28b67f590851321f15d9983ef7fd031

Review URL: https://codereview.chromium.org/1738913002

Remove legacy Budgeted enum aliases from SkSurface

TBR=reed@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1743233002

Review URL: https://codereview.chromium.org/1743233002

DM crash handler tweaks

    - try to guard against reentrant signals.
    - re-raise the signal with the default handler when done printing.
    - support Windows

BUG=589654
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1744553002

Review URL: https://codereview.chromium.org/1744553002

Revert of Add abilitly to query audit trail for batches by draw op (patchset #4 id:60001 of https://codereview.chromium.org/1745513002/ )

Reason for revert:
kInvalidID conflicts with an older xCode #define

Original issue's description:
> Add abilitly to query audit trail for batches by draw op
>
> BUG=skia:
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1745513002
>
> Committed: https://skia.googlesource.com/skia/+/9b48a6e3f862076018cc7d63b180b6340f4873cd

TBR=ethannicholas@google.com,jcgregorio@google.com,joshualitt@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1747893002

Add test_skia.py, isolates for test_skia, images, skps

This enables running DM through Swarming.

NOTRY=true
BUG=skia:4763
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1743113003

Review URL: https://codereview.chromium.org/1743113003

A bit more privacy for SkiaServe's Request

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1741043003

Review URL: https://codereview.chromium.org/1741043003

Add abilitly to query audit trail for batches by draw op

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1745513002

Review URL: https://codereview.chromium.org/1745513002

move drawextrapatheffect header into animator

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1747753002

TBR=caryclark@google.com

Review URL: https://codereview.chromium.org/1747753002

Update SKP version

Automatic commit by the RecreateSKPs bot.

TBR=
NO_MERGE_BUILDS
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1742243002

Review URL: https://codereview.chromium.org/1742243002

remove unused view helpers

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1744043002

TBR=

Review URL: https://codereview.chromium.org/1744043002

update CMakeLists for views/aminated removal

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1749483002
CQ_EXTRA_TRYBOTS=client.skia.compile:Build-Ubuntu-GCC-x86_64-Release-CMake-Trybot,Build-Mac-Clang-x86_64-Release-CMake-Trybot

NOTRY=True
TBR=

Review URL: https://codereview.chromium.org/1749483002

remove unused view files

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1745813002

TBR=

Review URL: https://codereview.chromium.org/1745813002

remove unneeded view subclasses

no doubt, more deletes will follow...
goodbye old friends

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1740383002

TBR=

Review URL: https://codereview.chromium.org/1740383002

Add catapult to RecreateSKPs bot's PYTHONPATH

TBR=borenet
BUG=skia:5024
NOTRY=true
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1747623002

Review URL: https://codereview.chromium.org/1747623002

Revert of Enable RAW codec for Windows (patchset #8 id:140001 of https://codereview.chromium.org/1738913002/ )

Reason for revert:
I don't think there's anything wrong with this per-se, but the 32-bit Windows bots are running out of memory while running these tests now.

(You'll see something like c:\0\build\slave\workdir\build\skia\include\core\skbitmap.h:247: fatal error: ""sk_throw"" in the log.)

We run these tests in parallel, and sometimes these 32-bit processes try to use more than the 2-3G RAM they can allocate.  Seems like this is a particularly memory-intense process?

If we reland this, we might want to blacklist these tests on the 32-bit Windows bots.  The 64-bit bots should have access to tons and tons of RAM and let us keep testing for Windows.

Original issue's description:
> Enable RAW codec for Windows
>
> * Fix the exception catching
> * Set preprocessor differently for MSVC
>
> BUG=skia:4889(b/26958348)
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1738913002
>
> Committed: https://skia.googlesource.com/skia/+/474e4c3dd28b67f590851321f15d9983ef7fd031

TBR=scroggo@google.com,msarett@google.com,yujieqin@google.com
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:4889(b/26958348)

Review URL: https://codereview.chromium.org/1747443003

Revert of Add non-inline SkRTreeFactory constructor to fix clang-cl (patchset #2 id:20001 of https://codereview.chromium.org/1676833002/ )

Reason for revert:
This workaround should no longer be needed, per comment#16 from
thakis@ on the original CL.

Original issue's description:
> Add non-inline SkRTreeFactory constructor to fix clang-cl
>
> When Chromium is switched to build with VS 2015 that switches clang-cl
> to build in -fmsc-version=1900 mode. This causes some changes in
> importing of constructors and vftables which leads to link errors in
> blink_platform.dll. Adding a non-inline constructor makes 1900 mode
> behave consistently with 1800 mode and avoids the link error.
>
> The root cause needs investigation but this will unblock the VS 2015
> switch.
>
> BUG=440500,498544
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1676833002
>
> Committed: https://skia.googlesource.com/skia/+/964eec67760196585954203ba625e440607f7e92

TBR=reed@google.com,bsalomon@google.com,mtklein@chromium.org,thakis@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=440500,498544

Review URL: https://codereview.chromium.org/1739363002

DM: dm.json can be missing some final results

We called JsonWriter::DumpJson() every once in a while,
but this did not necssarily wait for offloaded PNG encoding to finish.

This means we might have been dropping some DM results for a while now.
Fix by one more JsonWriter::DumpJson() call after joining back to the
main thread.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1744723002

Review URL: https://codereview.chromium.org/1744723002

DM: support --config {f16,srgb}

Will need to follow up with enabling f16 and srgb configs.
They're still assert-y.

The GMs 'gamma' and 'gradients' look reassuringly good on f16 and srgb backends.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1711143002

Review URL: https://codereview.chromium.org/1711143002

Switch float arrays to vecs for conical gradients
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1745613002

Review URL: https://codereview.chromium.org/1745613002

Creating functions for uploading a mipmapped texture.

BUG=476416
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1249543003

Review URL: https://codereview.chromium.org/1249543003

Make GrVkStencilAttachment derive form GrVkImage

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1735283004

Review URL: https://codereview.chromium.org/1735283004

Fix memory crash in vulakn writeUniformBuffers

BUG=skia:5021
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1737593005

Review URL: https://codereview.chromium.org/1737593005

Replace fWillReadFragmentPosition with a bitfield

Replaces fWillReadFragmentPosition on GrProcessor with a
"RequiredFeatures" bitfield. This will allow us to add additional
built-in features. Completely removes information about reading the
fragment position from GrPipeline and GrProcOptInfo.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1734163002

Review URL: https://codereview.chromium.org/1734163002

New GM that exercises gamma-correctness in various parts of the pipeline.

Attempts to render 50% grey through many different techniques. Only the first one (black/white dither) is guaranteed to be correct. This serves as ground-truth for all the others (independent of whatever your display may or may not be doing when viewing the output of the GM).

Current tests:
- Texture bilerp
- Texture scale (ie mipmaps)
- Simple paint color
- Gradient
- A handful of Xfer modes
- 50% grey bitmaps, both linear and sRGB
- Bitmaps wrapped in shaders, which hits a different raster code path

TODO:
- Test coverage (rather than alpha)?
- Text, Blur, etc...?

Re-landing this without the blitter selection change, which may have caused a regression. Also updated the text color so it's legible against both black (SampleApp) and white (gold).

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1740213003

Review URL: https://codereview.chromium.org/1740213003

Include glGetMultisamplefv API

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1720963002

Committed: https://skia.googlesource.com/skia/+/69c1b24bcaa7cffd59753a75a7f3d155696fcac4

Review URL: https://codereview.chromium.org/1720963002

When a surface is backed by an external render target force a copy on image snap

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1694943002

Review URL: https://codereview.chromium.org/1694943002

Add back Vk files with LF endings
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1745433002

Review URL: https://codereview.chromium.org/1745433002

Revert of New GM that exercises gamma-correctness in various parts of the pipeline. (patchset #4 id:60001 of https://codereview.chromium.org/1734983003/ )

Reason for revert:
Testing revert to fix MacMini?

Original issue's description:
> New GM that exercises gamma-correctness in various parts of the pipeline.
>
> Attempts to render 50% grey through many different techniques. Only the first one (black/white dither) is guaranteed to be correct. This serves as ground-truth for all the others (independent of whatever your display may or may not be doing when viewing the output of the GM).
>
> Current tests:
> - Texture bilerp
> - Texture scale (ie mipmaps)
> - Simple paint color
> - Gradient
> - A handful of Xfer modes
> - 50% grey bitmaps, both linear and sRGB
>
> Also includes a bug-fix for sRGB -> Linear bitmap case
>
> BUG=skia:
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1734983003
>
> Committed: https://skia.googlesource.com/skia/+/9f73a71eec0eb54783f2dcad009aae6aafbe38d5

TBR=bsalomon@google.com,reed@google.com
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1744533002

Remove Vk files with spurious CRLFs
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1726403007

Review URL: https://codereview.chromium.org/1726403007

Make GrRRectBlurEffect::onIsEqual include the roundrect's rect

BUG=skia:5014
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1743553003

Review URL: https://codereview.chromium.org/1743553003

Get SkiaServe Request started off with a little privacy

This CL moves the surface stuff into the request object so it
can maintain its internal state consistently

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1741823002

Review URL: https://codereview.chromium.org/1741823002

Fix oval rendering in vulkan

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1739063002

Review URL: https://codereview.chromium.org/1739063002

SkPDF/Test: detailed error messages in PDFDeflateWStreamTest

BUG=skia:5009
TBR=

Review URL: https://codereview.chromium.org/1727153005

Wire up /batches in skiaserve

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1737323002

Review URL: https://codereview.chromium.org/1737323002

Fix uniform buffer layout in vulkan

TBR=jvanverth@google.com

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1734403002

Review URL: https://codereview.chromium.org/1734403002

Add batchlist managment to GrAuditTrail

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1724243004

Review URL: https://codereview.chromium.org/1724243004

Add VkProgramDataManager back into to repo

Confirmed that this is concidered a plain ASCII file with no CR

TBR=jvanverth@google.com

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1743593002

Review URL: https://codereview.chromium.org/1743593002

New GM that exercises gamma-correctness in various parts of the pipeline.

Attempts to render 50% grey through many different techniques. Only the first one (black/white dither) is guaranteed to be correct. This serves as ground-truth for all the others (independent of whatever your display may or may not be doing when viewing the output of the GM).

Current tests:
- Texture bilerp
- Texture scale (ie mipmaps)
- Simple paint color
- Gradient
- A handful of Xfer modes
- 50% grey bitmaps, both linear and sRGB

Also includes a bug-fix for sRGB -> Linear bitmap case

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1734983003

Review URL: https://codereview.chromium.org/1734983003

Try to fix broken file by just deleting it.

Will try re uploading fixed version after.

TBR=jvanverth@google.com

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1737763003

Review URL: https://codereview.chromium.org/1737763003

Enable RAW codec for Windows

* Fix the exception catching
* Set preprocessor differently for MSVC

BUG=skia:4889(b/26958348)
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1738913002

Review URL: https://codereview.chromium.org/1738913002

add gm to illustrate bug

TBR=reed@google.com
BUG=skia:517
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1734343004

Review URL: https://codereview.chromium.org/1734343004

fix undefined signed shifts

The expression (1 << 31) is technically undefined.

Fixed the undefined shift referenced by this bug and
a few friends.

R=reed@google.com
BUG=skia:2285
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1737363002

Review URL: https://codereview.chromium.org/1737363002

Revert of Skia Filter Quality and Scaling Metrics (patchset #3 id:60001 of https://codereview.chromium.org/1686203002/ )

Reason for revert:
reverting this to address Ilya's comments.

Original issue's description:
> Skia Filter Quality and Scaling Metrics
>
> Adds histogram metrics to log the filter quality and scale factor of
> each image draw. To make the data easier to consume, this is broken
> down into a number of individual histograms:
> - Filter quality across all draw calls
> - Scale amount across all draw calls
> - Scale amount per filter quality (4 histograms total)
>
> BUG=skia:
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1686203002
>
> Committed: https://skia.googlesource.com/skia/+/f57b3a6e4a002caf01378832cbd756c6c163a783

TBR=bsalomon@google.com,isherman@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1735423002

SkPDF: Let client re-encode JPGs if they want to

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1734223003

Review URL: https://codereview.chromium.org/1734223003

Fix createTexture calls in vk tests

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1737343002

Review URL: https://codereview.chromium.org/1737343002

Revert of Make GrDrawContext::internalDrawPath cons up its own GrPipelineBuilder (patchset #3 id:40001 of https://codereview.chromium.org/1730903007/ )

Reason for revert:
Got a GL-related crash in nanobench:

https://uberchromegw.corp.google.com/i/client.skia.android/builders/Perf-Android-GCC-Nexus10-GPU-MaliT604-Arm7-Release/builds/3815/steps/nanobench/logs/stdio

Given we're about to branch and this is the only GL-related change in the blamelist, I'm going to assume this is responsible and revert.

Original issue's description:
> Make GrDrawContext::internalDrawPath cons up its own GrPipelineBuilder
>
> Hopefully, this better defines (and reduces) the lifetime and reuse of GrPipelineBuilder objects in GrDrawContext.
>
> TBR=bsalomon@google.com
>
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1730903007
>
> Committed: https://skia.googlesource.com/skia/+/00fddebe56fabea67dcc08762805c1294eebf5bf

TBR=joshualitt@chromium.org,bsalomon@google.com,robertphillips@google.com
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true

Review URL: https://codereview.chromium.org/1737373002

Fix copy-paste typo in comment.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1734613005

Review URL: https://codereview.chromium.org/1734613005

4f gradient negative-dx interval fixup for kMirror_TileMode

4f intervals are meant to be monotonically increasing in X.  This means
we invert the point order when dx < 0:

  [0...1) -> (1...0]

For kMirror_TileMode, we also append a duplicate/flipped interval
sequence and expand the range to 2:

  [0...1) -> [0...1)[1...2)

When dx < 0, the current logic yields

  (1...0] -> (1...0](2...1]

which violates the interval monotonicity.

To achieve the proper order, we need to swap the two halves:

  (1...0] -> (1...0](2...1] -> (2...1](1...0]

R=reed@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1735773003

Review URL: https://codereview.chromium.org/1735773003

Make GrDrawContext::internalDrawPath cons up its own GrPipelineBuilder

Hopefully, this better defines (and reduces) the lifetime and reuse of GrPipelineBuilder objects in GrDrawContext.

TBR=bsalomon@google.com

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1730903007

Review URL: https://codereview.chromium.org/1730903007