DEPS: sfntly 468cad5 → b18b09b

SfntlyWrapper::SubsetFont now takes TTC index, needed for SkPDF.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2186103002

Review-Url: https://codereview.chromium.org/2186103002

Rename GrContext's newDrawContext & drawContext to makeDrawContext

These both return sk_sp.

GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2186073002

Review-Url: https://codereview.chromium.org/2186073002

Make the yasm copy target be type 'none' rather than type 'executable'
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2182873005

Review-Url: https://codereview.chromium.org/2182873005

require semi at the end of SkASSERT and friends

R=bungeman@google.com
TBR=reed@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2185063002

Review-Url: https://codereview.chromium.org/2185063002

GN: fixes for Mac

 - Make fiddle build on Mac (skipping GL).
 - Now that we're building in SkCodec, we depend on libpng and libjpeg-turbo unconditionally, not just on Linux.
 - Re-arrange third_party a bit so that our targets are Fuchsia/Chrome compatible.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2184133002

NOTREECHECKS=true
This doesn't affect Chrome/Blink, so landing through the closed tree.

Review-Url: https://codereview.chromium.org/2184133002

SkPDF: SkPDFStream takes a unique_ptr

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2188623004

Review-Url: https://codereview.chromium.org/2188623004

Rename textual includes as *.inc.

This makes it easier to distinguish source files that can stand on their own vs files meant to be #included into other source files.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2184143002

Review-Url: https://codereview.chromium.org/2184143002

DEPS: ICU c291cde → ec9c113

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2189643003

Review-Url: https://codereview.chromium.org/2189643003

DEPS:  harfbuzz 1.2.7 → 1.3.0

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2179963005

Review-Url: https://codereview.chromium.org/2179963005

DEPS:  sfntly 130f832 → 468cad5

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2189643002

Review-Url: https://codereview.chromium.org/2189643002

Include "revision" tag for all Swarming tasks

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2186993002

Review-Url: https://codereview.chromium.org/2186993002

Fix ParseConfigs_ViaParsing test when skia_angle=1

The handling of angle-gl changes with SK_ANGLE. We don't have any bots
that test this particular combination, but I see it all the time while
running DM for other things.

Previously considered changing things so that the config parsing results
are consistent, regardless of GYP_DEFINES, but this is much simpler (and
more consistent with the other code we already have for testing config
parsing).

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2183323004

Review-Url: https://codereview.chromium.org/2183323004

Fix leak of srgb/adobesrgb colorspace objects

GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2185043002

Review-Url: https://codereview.chromium.org/2185043002

Fix Perf data upload

BUG=skia:5563
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2187603005

Review-Url: https://codereview.chromium.org/2187603005

SW backend for viewer on Windows
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2182273002

Review-Url: https://codereview.chromium.org/2182273002

fuzz wednesday

Add debug 'skip assert' access to SkCoincidentSpans.
That permits suppressing asserts to make it easier
to detect when this fuzz data generates unparsable
data.

TBR=reed@google.com
BUG=631564, 631992
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2186973002

Review-Url: https://codereview.chromium.org/2186973002

Clean up some unused atomic routines.

AtomicTest was the only use of sk_atomic_add().
AtomicInc64 bench was the only use of sk_atomic_inc(int64_t*).

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2183473005
CQ_INCLUDE_TRYBOTS=master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-TSAN-Trybot,Test-Ubuntu-GCC-Golo-GPU-GT610-x86_64-Release-TSAN-Trybot

Review-Url: https://codereview.chromium.org/2183473005

Add gyp define to enable vulkan debug layers

On Debug vulkan bots, running with the debug layers on seems to be adding
more than an hour to the total running time. Since we suppress any output
on the bots anyways the debug layers are serving no purpose. Thus I am
adding a gyp define to disable the layers on the bot.

With this change, by default when running vulkan in Debug, the debug_layers
will be enabled. The bots should disable the layers. Android framework
should also have them disabled by default.

TBR=djsollen@google.com

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2185953003

Review-Url: https://codereview.chromium.org/2185953003

Tweaks for the experimental SwarmBucket bot

BUG=skia:5581
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2181073005

Review-Url: https://codereview.chromium.org/2181073005

Extend __mulodi4() hack to 32-bit x86.

This gets the Android-Clang-x86 perf bot from failing on __mulodi4() to failing on the next problem...

See https://luci-milo.appspot.com/swarming/task/3042e5f68636f810/steps/nanobench/0/stdout

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2190493002

Review-Url: https://codereview.chromium.org/2190493002

Simplify color space xform storage and usage (float[] vs. SkMatrix44)

Public API is really just internal.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2185533005
TBR=bsalomon@google.com

Review-Url: https://codereview.chromium.org/2185533005

Fix Housekeeper-PerCommit

BUG=skia:5563
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2183023004

Review-Url: https://codereview.chromium.org/2183023004

Only run recipe simulation tests if infra/ changes.

This should lessen the need for mortals to have coverage.py,
and saves 1.8s (of 3.6s) on my desktop for this non-infra change.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2187713002

Review-Url: https://codereview.chromium.org/2187713002

add unittests for roundtripping premul

BUG=skia:5579
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2179123005

Review-Url: https://codereview.chromium.org/2179123005

Initial SVG model

A minimal subset needed to render tiger.svg: <svg>, <g>, <path>, 'd', 'fill'/'stroke' (color-only), 'transform'.

R=reed@google.com,robertphillips@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2164193002

Review-Url: https://codereview.chromium.org/2164193002

Implement coverage AA for skewed rects with local coords

Adds a path fallback for rects with local coords that can't be drawn
with an analytic shader. This is accomplished by modifying the view
matrix and then drawing the local rect/quad.

BUG=skia:5500, 7508
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2183223002

Review-Url: https://codereview.chromium.org/2183223002

Avoid warnings on GCC 4.8.4

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2183243002

[mtklein adds...]
TBR=reed@google.com
No public API changes.

Review-Url: https://codereview.chromium.org/2183243002

Start on fiddle.

Mostly stolen from Joe.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2188493002

Review-Url: https://codereview.chromium.org/2188493002

Make yasm-android executable

TBR=mtklein@google.com
NOTRY=true
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2188583002

Review-Url: https://codereview.chromium.org/2188583002

Rename yasm binary and make it executable

TBR=mtklein@google.com
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2187663002

Review-Url: https://codereview.chromium.org/2187663002

Update iOS to new SkSurface interface

BUG=skia:5558
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2179363003

Review-Url: https://codereview.chromium.org/2179363003

Start measuring f16 and srgb speed.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2181073003

Review-Url: https://codereview.chromium.org/2181073003

Add sw support to viewer on Linux.

This doesn't yet do anything with srgb/colorspace.

Replaces the --vulkan flag with --backend which takes 'gl', 'vk', or 'sw'.

GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2165813002

Review-Url: https://codereview.chromium.org/2165813002

Move SkJSONCPP.h to tools.  It tool-only.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2184483003
CQ_INCLUDE_TRYBOTS=master.client.skia.compile:Build-Ubuntu-GCC-x86_64-Release-CMake-Trybot,Build-Mac-Clang-x86_64-Release-CMake-Trybot

TBR=reed@google.com

Review-Url: https://codereview.chromium.org/2184483003

don't specialize drawBitmapRect for minirecorder; no longer called by blink

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2183193002

Review-Url: https://codereview.chromium.org/2183193002

Use Windowing system-specific WindowContext factories.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2169543002

Review-Url: https://codereview.chromium.org/2169543002

Move Skia recipes from build repo

BUG=skia:5563
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2175373002

Review-Url: https://codereview.chromium.org/2175373002

Reduce usage of MakeRenderTargetDirect

GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2187573002

Review-Url: https://codereview.chromium.org/2187573002

Remove SkSurfaceProps gamma-correctness flag entirely.

This is no longer required - gamma-correctness is now just based on the
presence or absence of a color space.

Public API change is just removal of (unused) flag.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2188463002
TBR=bsalomon@google.com

Review-Url: https://codereview.chromium.org/2188463002

Fix yasm for Android build

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2180333002

Review-Url: https://codereview.chromium.org/2180333002

Fix docs for checking out Skia

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2179023003
NOTRY=true
DOCS_PREVIEW= https://skia.org/?cl=2179023003

Review-Url: https://codereview.chromium.org/2179023003

Remove all usage of SkSurfaceProps::isGammaCorrect()

DrawContext's isGammaCorrect now just based on presence of color space.
Next change will remove the function and flag entirely, but I wanted to
land this separately. This alters a few GMs in srgb/f16 mode, generally
those that are creating off-screen surfaces in ways that were somewhat
lossy before. No unexplained changes.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2186633002

Review-Url: https://codereview.chromium.org/2186633002

Fix alignment problems in NEON Sk4b.

As written at head, the compiler can assume these loads and stores are 4
byte aligned [1].  We want Sk4b to load from any 1-byte aligned address,
to prevent crashes like [2].

[1] https://llvm.org/bugs/show_bug.cgi?id=24421
[2] https://luci-milo.appspot.com/swarming/task/304079e125b1b910/steps/nanobench/0/stdout

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2183133002
CQ_INCLUDE_TRYBOTS=master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Review-Url: https://codereview.chromium.org/2183133002

fix fuzz bugs

Add ability for intersection template to detect that
the test contains bounded numbers so that extra
asserts can trigger.

Add some exit points for out of range numbers
in those templates.

TBR=reed@google.com
BUG=631383,631374,631360
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2185703002

Review-Url: https://codereview.chromium.org/2185703002

Don't unnecessarily resize windows in sk_app on X
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2182163002

Review-Url: https://codereview.chromium.org/2182163002

Update bleed-downscale GM to use power of 2 texture

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2179303004

Review-Url: https://codereview.chromium.org/2179303004

Always supply a color space (sRGB for now) with F16

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2177193004

Review-Url: https://codereview.chromium.org/2177193004

Add Sk4h_load4 for loading F16.

Should feel very similar to Sk4h_store4:
NEON uses its native instruction, SSE unpacks manually.

Since we'll have our F16s in 4 Sk4h by the time we're done here,
this also extracts an Sk4h->Sk4f routine from the old uint64_t->Sk4f one.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2184753002
CQ_INCLUDE_TRYBOTS=master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Review-Url: https://codereview.chromium.org/2184753002

Move prepareForExternalIO from GrRenderTarget to GrDrawContext

This is part of the push to remove GrRenderTarget from SkGpuDevice

GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2182543003

Review-Url: https://codereview.chromium.org/2182543003

Reuse sampler descriptor set allocations in Vulkan

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2172873003

Review-Url: https://codereview.chromium.org/2172873003

Add special Windows instructions for getting SKPs

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2184723002
NOTRY=true
DOCS_PREVIEW= https://skia.org/?cl=2184723002

Review-Url: https://codereview.chromium.org/2184723002

respect kFast_SrcRectConstraint by allowing shader access to entire bitmap

BUG=355305
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2176193002

Review-Url: https://codereview.chromium.org/2176193002

Use sk_srgb_to_linear_trunc in SkColorXform_opts

This gives us a little more control over instruction order, allowing
us to pipeline the muls and get better performance.  Technically,
clang should be able to do this for us anyway...

Performance on HP z620 (201295.jpg):
toSRGB: 371us -> 356us

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2175413002
CQ_INCLUDE_TRYBOTS=master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Review-Url: https://codereview.chromium.org/2175413002

Delete SkDefaultXform, handle edge cases in SkColorSpaceXform_Base

"Edge" cases include:
(1) Matrices with translation
(2) colorLUTs

Performance on HP z620:

201295.jpg
to2Dot2: 386us -> 414us
toSRGB:  346us -> 371us
toHalf:  282us -> 302us

strange0-translate.jpg
toSRGB:  1060us -> 244us

strange1-colorLUT.jpg
toSRGB:  2.74ms -> 2.00ms

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2177173003
CQ_INCLUDE_TRYBOTS=master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Review-Url: https://codereview.chromium.org/2177173003

Fix F16 raster text masks. We don't want fake gamma in that case.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2184443002

Review-Url: https://codereview.chromium.org/2184443002

Update typeface serialization style.

GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2030683002

Committed: https://skia.googlesource.com/skia/+/7ae1c72e593823e6e300e4199558555765bcec17
Review-Url: https://codereview.chromium.org/2030683002

DM: print which task caused the crash.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2177843003

Review-Url: https://codereview.chromium.org/2177843003

Add test for typeface style round trip.

This also fixes the CG and GDI ports so they pass the test.

GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2171163002

Review-Url: https://codereview.chromium.org/2171163002

Have SkRasterPipelineBlitter take over for F16 when it can.

3 minor diffs to a couple GMs from fixed transfermodes (arithmetic and plus).

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2181033002

Review-Url: https://codereview.chromium.org/2181033002

lock special-raster while the caller's bitmap is in-scope

previous CL that removed the specialimage subclass that wrapped the original image: https://codereview.chromium.org/2164073004

BUG=skia:5574
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2183483003

Review-Url: https://codereview.chromium.org/2183483003

SkPDF: re-work SkPDFUtils::FloatToDecimal

  * do a lot less floating-point math by converting to
    an integer as early as possible [faster].
  * round rather than truncate.
  * use 8 significant digits rather than 9 when possible.
  * remove trailing zeros in fractions.

before:
      0.12 !    PDFScalar   nonrendering
after:
      0.07 !    PDFScalar   nonrendering

Accuracy guaranteed by existing unit test.

Example diffs:

    -/Shading <</Function <</C0 [.321568638 .333333343 .321568638]
    +/Shading <</Function <</C0 [.32156864 .33333334 .32156864]

    -/C1 [.258823543 .270588248 .258823543]
    +/C1 [.25882354 .27058825 .25882354]

    -1 0 0 -1 20 120.394500 Tm
    +1 0 0 -1 20 120.394501 Tm

    -1 0 0 -1 20 184.789001 Tm
    +1 0 0 -1 20 184.789 Tm

    -291.503997 0 l
    +291.504 0 l

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2146103004

Review-Url: https://codereview.chromium.org/2146103004

Google3 <functional> has a global "using std::plus"...

... so we can't use the name plus().  -_-

TBR=tomhudson@google.com

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2183453003

Review-Url: https://codereview.chromium.org/2183453003

Just pass the draw context to paint conversion

Concentrate on using draw context functionality to answer gamma and color
related queries in more places.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2182603003

Review-Url: https://codereview.chromium.org/2182603003

Fix debugger build by specifying nullptr for color space.

GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2181733002

Review-Url: https://codereview.chromium.org/2181733002

stages for most xfermodes

also assumes the clamp from https://codereview.chromium.org/2178793002/ is present

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2177103002

Review-Url: https://codereview.chromium.org/2177103002

SkSL performance improvements (plus a couple of minor warning fixes)
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2131223002

Committed: https://skia.googlesource.com/skia/+/9fd67a1f53809f5eff1210dd107241b450c48acc
Review-Url: https://codereview.chromium.org/2131223002

add SK_SUPPORT_LEGACY_BITMAP_GETTEXTURE for out-of-date google3 callers

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2177993002

Review-Url: https://codereview.chromium.org/2177993002

remove/deprecate SkBitmap::getTexture, as it now always returns false

oh happy day

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2175873002

Review-Url: https://codereview.chromium.org/2175873002

add gm to test bleed with downscale

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2182503002

TBR=

Review-Url: https://codereview.chromium.org/2182503002

remove fClipStack and attach/deattach-from-canvas

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2180793002

Review-Url: https://codereview.chromium.org/2180793002

typedef uint16_t SkGlyphID

This goes well with our existing `typedef int32_t SkUnichar`.
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2095263003

Review-Url: https://codereview.chromium.org/2095263003

Remove long-running sk_linear_to_srgb test for ASAN in public.bzl.

This unit test takes >~5m and causes the ASAN test to time out.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2175303002

Review-Url: https://codereview.chromium.org/2175303002

Add a clamp stage to SkRasterPipelineBlitter.

This clamps to [0,1] premul just before every store to memory.

By making the clamp a stage itself, this design makes it easy to move the clamp
around, to replace it with a debug-only assert-we're-clamped stage for certain
formats, clamp in more places, programatically not clamp, etc. etc.

Before this change, clamping was a little haphazard: store_srgb clamped
R, G and B to [0,1], but not A, and didn't clamp the colors to A.  565
didn't clamp at all.

6 GMs draw subtly differently in sRGB, I think because we've started clamping
colors to alpha to enforce premultiplication better.  No changes for 565.

My hope is that now no other stage need ever concern itself with clamping.

So we don't double-clamp, I've added a _noclamp version of sk_linear_to_srgb()
that simply asserts a clamp isn't necessary.  This happens to expose the Sk4f
_needs_trunc version that might be useful for power users (*cough* Matt *cough*).

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2178793002

Review-Url: https://codereview.chromium.org/2178793002

fix fuzz

Extreme inputs trigger asserts intended for in range data.
Return an error without asserting unless the test signals
otherwise.

TBR=reed@google.com
BUG=630736
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2178923002

Review-Url: https://codereview.chromium.org/2178923002

Arithmetic xfermode stage.

565 is identical, and as usual the old sRGB path was broken.  :)
It was unimplemented, so the default missing Sk4f implementation was mimicking Src.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2175133002

Review-Url: https://codereview.chromium.org/2175133002

webpages_playback: Make --upload_to_partner_bucket work without --upload

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2182463002

Review-Url: https://codereview.chromium.org/2182463002

Add destination color space to AsFPArgs. Eliminates last XFORMTODO.

This is going to be needed in many more places as I finish connecting the
dots. Even better - I'd like to switch to a world where SkColorSpace !=
nullptr is the only signal we use for gamma-correct rendering, so I can
eliminate SkSourceGammaTreatment and SkSurfaceProps::isGammaCorrect.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2180503002

Review-Url: https://codereview.chromium.org/2180503002

lockPixelsAreWritable is deprecated

BUG=skia:4328
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2177063002

TBR=

Review-Url: https://codereview.chromium.org/2177063002

Fix misdetection of rectangles in SkPath::IsSimpleClosedRect.

BUG=chromium:630369
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2175923002

Review-Url: https://codereview.chromium.org/2175923002

Update SKP version

Automatic commit by the RecreateSKPs bot.

TBR=
NO_MERGE_BUILDS
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2177053002

Review-Url: https://codereview.chromium.org/2177053002

Remove asserts on scissor size in Vulkan

These are asserts are firing from a recent change to our scissor code.
Since these asserts were added, the Vulkan spec has been updated to no
longer require the scissor is insides the bounds of the image, just that
x + width does not overflow.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2171283004

Review-Url: https://codereview.chromium.org/2171283004

Have SkRasterPipelineBlitter take over for 565 when it can.

This is an experiment / demo to have our 565 backend fold into
SkRasterPipelineBlitter as it grows more powerful.  I plan to follow up with
the same for the other 8888 format.

Blur mask filters look significantly different (better) after this change.
We keep the full 13-14-13 bits of precision for mask blits, where the old code
uses 11-11-10 bit intermediates.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2172343002

Review-Url: https://codereview.chromium.org/2172343002

Convert XYZ values from PNGs to D50

Previously, we were storing whatever values we got out of the file
and pretending they were D50.  Brian has demonstrated that this
doesn't really work.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2180483003

Review-Url: https://codereview.chromium.org/2180483003

In the current code, tiling and bilerp sampling are strongly tied together. They can be separated by taking advantage of observation that translating a sample point into filter points in the bilerp stage the filter points will be at most 0.5 outside the tile. This allows simplified repositioning for the various tiling modes; clamp and mirror use min and max while repeat has max -> 0 and 0-> max. This allows bilerp to simply treat the filter points that fall off the tile. This allows tiling and bilerp sampling to be totally separate.

This CL has several parts that are intertwined:
* move pin/wrap functionality into BilerpSampler.
* remove the nearest neighbor and bilerp tilers
* create a simplified general tiler
* remove the pipeline virtual calls bilerpEdge and bilerpSpan because everything works of sample points now.
* redo all the bilerp sampling to use the new local to methods to wrap/pin.
* introduce a new medium rate sample that handles spans with 1 < |dx| < 2.

This change improves the performance as displayed below:
Most of top 25 desktop improves or are the same. A few are worse, but close to the noise floor. In addition, this change has about 3% smaller code.

old time     new time   new/old
13274693  8414645  0.633886  top25desk_google_com_search_q_c.skp_1
4946466   3258018  0.658656  top25desk_wordpress.skp_1
6977187   5737584  0.822335  top25desk_youtube_com.skp_1
3770021   3296831  0.874486  top25desk_google_com__hl_en_q_b.skp_1
8890813   8600143  0.967307  top25desk_answers_yahoo_com.skp_1
3178974   3094300  0.973364  top25desk_facebook.skp_1
8871835   8711260  0.981901  top25desk_twitter.skp_1
838509    829290   0.989005  top25desk_blogger.skp_1
2821870   2801111  0.992644  top25desk_plus_google_com_11003.skp_1
511978    509530   0.995219  top25desk_techcrunch_com.skp_1
2408588   2397435  0.995369  top25desk_ebay_com.skp_1
4446919   4448004  1.00024   top25desk_espn.skp_1
2863241   2875696  1.00435   top25desk_google_com_calendar_.skp_1
7170086   7208447  1.00535   top25desk_booking_com.skp_1
7356109   7417776  1.00838   top25desk_pinterest.skp_1
5265591   5340392  1.01421   top25desk_weather_com.skp_1
5675244   5774144  1.01743   top25desk_sports_yahoo_com_.skp_1
1048531   1067663  1.01825   top25desk_games_yahoo_com.skp_1
2075501   2115131  1.01909   top25desk_amazon_com.skp_1
4262170   4370441  1.0254    top25desk_news_yahoo_com.skp_1
3789319   3897996  1.02868   top25desk_docs___1_open_documen.skp_1
919336    949979   1.03333   top25desk_wikipedia__1_tab_.skp_1
4274454   4489369  1.05028   top25desk_mail_google_com_mail_.skp_1
4149326   4376556  1.05476   top25desk_linkedin.skp_1

BUG=skia:5566
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Committed: https://skia.googlesource.com/skia/+/8602ede5fdfa721dcad4dcb11db028c1c24265f1
Review-Url: https://codereview.chromium.org/2134893002

Always init SkPathRef variables.

BUG=623195
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2173853003

Review-Url: https://codereview.chromium.org/2173853003

DM: don't redirect stderr to verbose.log

verbose.log is currently tricky to get to, so it's hurting more than it
helps to direct stderr there.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2178583002

Review-Url: https://codereview.chromium.org/2178583002

Correct advances for 'monospace' fonts in PDF.

FT_IS_FIXED_WIDTH, kCTFontMonoSpaceTrait, and TMPF_FIXED_PITCH
are style bits, they do not imply that all advances are the same.

BUG=skia:5537
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2162023002

Review-Url: https://codereview.chromium.org/2162023002

Add SkRasterPipeline blitter.

This is now pixel-exact with the existing sRGB SW impl as far as I've tested.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2146413002
CQ_INCLUDE_TRYBOTS=master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot,Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Debug-ASAN-Trybot

Committed: https://skia.googlesource.com/skia/+/3011e337693a9786f62d8de9ac4b239ad6dbdaca
Review-Url: https://codereview.chromium.org/2146413002

Bundle SkShader::asFragmentProcessor arguments in a struct

The signature of this thing keeps changing (and is about to change again).
This just makes maintenance much easier.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2175563003

Review-Url: https://codereview.chromium.org/2175563003

limit number of searched roots

Extreme numbers can generate more than
three found cubic roots when the roots
are found through a binary search.

Fail in this case.

TBR=reed@google.com
BUG=630649
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2176733002

Review-Url: https://codereview.chromium.org/2176733002

Recycle small uniform buffers.

BUG=skia:5031
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2159333002

Review-Url: https://codereview.chromium.org/2159333002

Connect dst color space in a few places (removing TODOs)

There is one more XFORMTODO left, but it requires more plumbing.
With this change, certain drawing paths do gamut conversion with
no special hacking required.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2174843002

Review-Url: https://codereview.chromium.org/2174843002

Revert of Add SkRasterPipeline blitter. (patchset #18 id:340001 of https://codereview.chromium.org/2146413002/ )

Reason for revert:
Leaking the blitter

https://build.chromium.org/p/client.skia/builders/Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Debug-ASAN/builds/7908/steps/test_skia%20on%20Ubuntu/logs/stdio

Original issue's description:
> Add SkRasterPipeline blitter.
>
> This is now pixel-exact with the existing sRGB SW impl as far as I've tested.
>
> BUG=skia:
> GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2146413002
> CQ_INCLUDE_TRYBOTS=master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot
>
> Committed: https://skia.googlesource.com/skia/+/3011e337693a9786f62d8de9ac4b239ad6dbdaca

TBR=reed@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/2178523002

Creating framework for drawShadowedPicture
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2146073003

Committed: https://skia.googlesource.com/skia/+/0ae097d116f4332be02a135ffc99c162473dee6a
Committed: https://skia.googlesource.com/skia/+/95302da19d8b0a3bcd9d9be0e79f486760787f09
Review-Url: https://codereview.chromium.org/2146073003

Allow GrReducedClip to take non-integer query bounds

Fixes places where AA bloat was being conflated with geometric
boundaries and updates GrReducedClip to work with non-integer query
bounds. This allows for better clip reduction with AA shared edges.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2160093002

Review-Url: https://codereview.chromium.org/2160093002

Add SkRasterPipeline blitter.

This is now pixel-exact with the existing sRGB SW impl as far as I've tested.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2146413002
CQ_INCLUDE_TRYBOTS=master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Review-Url: https://codereview.chromium.org/2146413002

Fix SkClipStack bug with inverse-filled difference elements

Previously, SkClipStack would call "setEmpty" on itself when an
inverse-filled difference element made the stack empty. This was
a problem because setEmpty would forget the element had an inverse
fill, yet leave the op as "difference". This change modifies it to
manually update the clip bounds and set the gen-ID to kEmptyGenID,
rather than calling setEmpty.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2175493002

Review-Url: https://codereview.chromium.org/2175493002

Don't ask GLX for EGL procs.

Seems to fix my woes.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2168293003

Review-Url: https://codereview.chromium.org/2168293003

Change SKPs to new asset management format

BUG=skia:5427
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2169653003
DOCS_PREVIEW= https://skia.org/?cl=2169653003

Review-Url: https://codereview.chromium.org/2169653003

Fix SW sRGB dst + LCD coverage bug.

We're using the linear procs for sRGB destintations
and the sRGB procs for linear destinations.  Fix that.

C.f. State32::getLCDProc(), which flags |= kDstIsSRGB_LCDFlag.
kDistIsSRGB is (1<<2) == 4, so the sRGB procs must be 4-7, not 0-3.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2177493002

Review-Url: https://codereview.chromium.org/2177493002