followups

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1693823005

Review URL: https://codereview.chromium.org/1693823005

Remove check for platform_tools/android/third_party/externals.

We don't sync anything into it anymore, so a fresh gclient sync won't create it.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1694813004

Review URL: https://codereview.chromium.org/1694813004

Add support for caching font files in the Android SkFontMgr.

SkFontMgr_Android lazily access font files from disk, which is not
possible when it is used in the renderer sandbox on Linux. Add a flag to
SkFontMgr_Android_CustomFonts for caching readonly FILE streams when
creating the font mgr.

Since the font mgr is created before the sandbox is initialized, it can
access these files on initialization, and use these cached streams safely
for its lifetime.

Review URL: https://codereview.chromium.org/1673373003

Make SkJpegCodec compatible with libjpeg

BUG=skia:4470
BUG=skia:4520
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1687863003

Review URL: https://codereview.chromium.org/1687863003

Change ScalarTo256 to more efficient implementation.

The previous implementation was likely more efficient when SkScalar was SkFixed.

BUG=skia:4632
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1693683002

Review URL: https://codereview.chromium.org/1693683002

Put all XSAN bots on the same DEPS-sourced build of Clang.

This adds LLVM, Clang, and compiler-rt (*san tools) to DEPS,
then uses them from xsan_build, building them if needed.

This is similar to how the CMake bots bootstrap CMake if needed.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1693733003

Review URL: https://codereview.chromium.org/1693733003

Use noperspective interpolation for 2D draws

Adds a mechanism to notify GrGLSLVaryingHandler that a shader will not
emit geometry in perspective. This gives it a chance to use the
noperspective keyword if it is supported.

Updates the existing processors to notify the varying handler when
there is no perspective.

Begins using the noperspective keyword in GrGLGpu internal shaders.

Web scenes with observable benefit (Pixel C, gpu config):

  tabl_nofolo.skp        4.62 -> 3.33 ms   28%
  desk_tigersvg.skp      26.5 -> 24 ms      9%
  desk_pokemonwiki.skp   16.1 -> 14.9 ms    7%
  tabl_deviantart.skp    3.97 -> 3.7 ms     7%
  desk_gws.skp           3.85 -> 3.65 ms    5%

Also adds new methods for creating varyings with flat interpolation.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1673093002

Review URL: https://codereview.chromium.org/1673093002

Make copySurface work in more situations.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1693923002

Review URL: https://codereview.chromium.org/1693923002

added clip visualization to skiaserve
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1690023004

Review URL: https://codereview.chromium.org/1690023004

Add a dependency from the debugger on tools.gyp:url_data_manager.

BUG=skia:4947
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1695753003

Review URL: https://codereview.chromium.org/1695753003

Put back the 0.5 bias in the shader. Only adjust the radii for the geometry, not the radii constants.

After working through this with Jim, this is more mathematically correct, and also accounts for the inner radius.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1692683003

Review URL: https://codereview.chromium.org/1692683003

skiaserve now retains the graphics context rather than creating a new one each request
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1693893002

Review URL: https://codereview.chromium.org/1693893002

Update Android framework makefile to build static and shared libs

Also use static lib for testing tools and expose includes needed for
other framework testing tools to statically link in Skia.
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1696483002

Review URL: https://codereview.chromium.org/1696483002

Documantation: nanobench

DOCS_PREVIEW= https://skia.org/dev/testing/tests?cl=1696723002
              https://skia.org/dev/testing/tests?cl=1696723002#bench
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1696723002
NOTRY=true

Review URL: https://codereview.chromium.org/1696723002

Remove the 0.5 bias from the ellipse edge function. This makes coverage drop to zero before we get clipped by the rasterizer (i.e. ensures the generated ellipse is inscribed in the geometry).

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1683843006

Review URL: https://codereview.chromium.org/1683843006

SkDocument: remove use of SkTArray (part 3/3).

TBR=reed
Removal only. Verbal lgtm.

Review URL: https://codereview.chromium.org/1692703003

fixed SkDebugCanvas' handling of drawPicture
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1685133008

Review URL: https://codereview.chromium.org/1685133008

Enable RAW codec for Google3.

This CL depends on internal cl/113089689.
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1692513003

Review URL: https://codereview.chromium.org/1692513003

Delete SkDecodingImageGenerator

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1692053002

Review URL: https://codereview.chromium.org/1692053002

Switch to vertex colors rather than uniforms for color in Ellipse/Circle/RRect/DRRect

[Notes:]
- Performance delta on desktop looked like noise. I'm curious what the results will be on low-end mobile GPUs. I'm still trying to figure out how I determine that with perf, etc...
- There appeared to be a single one-pixel image diff while running dm, if I'm understanding the results of skdiff correctly, so that's probably good news?
- Should I include anyone else on the review?

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1675833002

Review URL: https://codereview.chromium.org/1675833002

Add #define directives to GrGLSLShaderBuilder

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1660813002

Review URL: https://codereview.chromium.org/1660813002

Implement onSkipScanlines() for bmp and wbmp

TBR=reed@google.com
BUG=skia:4270
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1691083002

Review URL: https://codereview.chromium.org/1691083002

fixed off-by-one error in skiaserve /cmd
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1692683002

Review URL: https://codereview.chromium.org/1692683002

Add GrShaderFlags enum

Replaces GrGLSLUniformHandler::ShaderVisibility with GrShaderFlags.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1684063006

Review URL: https://codereview.chromium.org/1684063006

new version of SkHalfToFloat_01

This is a little faster than the previous version, and much better explained.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1688233002

Review URL: https://codereview.chromium.org/1688233002

start to chip away at friending of GrAtlasTextBatch/GrAtlasTextBlob

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1684253002

Review URL: https://codereview.chromium.org/1684253002

Revert of Make copySurface work for texture dsts, return a bool, & add unit test. (patchset #6 id:100001 of https://codereview.chromium.org/1684313002/ )

Reason for revert:
Copy surface tests are not happy for Windows, Android and probably others:
https://uberchromegw.corp.google.com/i/client.skia.android/builders/Test-Android-GCC-Nexus7-GPU-Tegra3-Arm7-Release/builds/4161
https://uberchromegw.corp.google.com/i/client.skia/builders/Test-Win8-MSVC-ShuttleA-GPU-GTX960-x86_64-Release/builds/3694

Original issue's description:
> Make copySurface work for texture dsts, return a bool, & add unit test.
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1684313002
>
> Committed: https://skia.googlesource.com/skia/+/7ea5e28065e5eb797e95f5d81c1a65cf3209d741

TBR=robertphillips@google.com,bsalomon@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/1690053002

Remove GrContext* from GrTextContext

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1686143003

Review URL: https://codereview.chromium.org/1686143003

Make copySurface work for texture dsts, return a bool, & add unit test.
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1684313002

Review URL: https://codereview.chromium.org/1684313002

enable ThermalManager on Android

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1684653002

Review URL: https://codereview.chromium.org/1684653002

Support more color types for ICOs

BUG=skia:4620
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1694513002

Review URL: https://codereview.chromium.org/1694513002

Avoid initializing memory twice on incomplete RLE bmps

BUG=skia:4582
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1689983002

Review URL: https://codereview.chromium.org/1689983002

integrated SkDebugCanvas JSON support with UrlDataManager
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1686143004

Review URL: https://codereview.chromium.org/1686143004

Test SkAndroidCodec for more types

Now that SkAndroidCodec supports all types, stop selectively running
its tests for the types supported by BRD.

For the tests that actually require BRD support (i.e. partial
scanlines), change the method name from supports_scaled_codec to
supports_partial_scanlines to be more accurate today.

Use the name SkAndroidCodec instead of SkScaledCodec.
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1691823002

Review URL: https://codereview.chromium.org/1691823002

Further reduce the number of primitives in the donut tests. Avoids crashes on iOS.

BUG=skia:4937
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1691813002

Review URL: https://codereview.chromium.org/1691813002

Clarify bmp comment about when we support the alpha mask

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1689183002

Review URL: https://codereview.chromium.org/1689183002

Fix scanline decoding of rare RLE bmps

This also exposed a bug in rewinding RLE bmps, which I have also
fixed in this CL.

This should fix testcase7.bmp on Gold.

The image that I am adding to resources is in the
public domain.

BUG=skia:4730
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1689953002

Review URL: https://codereview.chromium.org/1689953002

SkNx refactoring

  - add back Sk4i typedef
  - define SSE casts in terms of Sk4i
      * uint8  <-> float becomes uint8  <-> int <-> float
      * uint16 <-> float becomes uint16 <-> int <-> float

    This has the nice side effect of specializing uint8 <-> int
    and uint16 <-> int, which are useful in their own right.

There are many cast specializations now, some of which call each other.
I have tried to arrange them in some sort of sensible order, subject to
the constraint that those called must precede those who call.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1690633003
CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Review URL: https://codereview.chromium.org/1690633003

Remove SK_SUPPORT_LEGACY_BITMAP_FILTER guard

TBR=reed@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1695443002

Review URL: https://codereview.chromium.org/1695443002

SkDocument: remove use of SkTArray (part 1/3).

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1689683002

Review URL: https://codereview.chromium.org/1689683002

wire up new json code in skiaserve

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1691773002

Review URL: https://codereview.chromium.org/1691773002

fix for gpu valgrind bot

TBR=egdaniel@google.com
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1689823004

Review URL: https://codereview.chromium.org/1689823004

Pass a GrContext pointer in GrDrawContext constructor

TBR=bsalomon@google.com
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1691503002

Review URL: https://codereview.chromium.org/1691503002

Add AndroidCodecBench to time scaled decodes

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1685693003

Review URL: https://codereview.chromium.org/1685693003

SkHalfToFloat_01 / SkFloatToHalf_01

These are basically inlined, 4-at-a-time versions of our existing functions,
but cut down to avoid any work that's only necessary outside [0,1].

Both f16 and f32 denorms should work fine modulo the usual ARMv7 NEON denorm==zero caveat.

In exchange for a little speed, f32->f16 does not round properly.
Instead it truncates, so it's never off by more than 1 bit.

Support for finite values >1 or <0 is straightforward to add back.
>1 might already work as-is.

Getting close to _u16 performance:
    micros    bench
    261.13   xferu64_bw_1_opaque_u16
   1833.51   xferu64_bw_1_alpha_u16
   2762.32 ? xferu64_aa_1_opaque_u16
   3334.29   xferu64_aa_1_alpha_u16
    249.78   xferu64_bw_1_opaque_f16
   3383.18   xferu64_bw_1_alpha_f16
   4214.72   xferu64_aa_1_opaque_f16
   4701.19   xferu64_aa_1_alpha_f16

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1685133005

Committed: https://skia.googlesource.com/skia/+/9ea11a4235b3e3521cc8bf914a27c2d0dc062db9

CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Review URL: https://codereview.chromium.org/1685133005

Revert of SkHalfToFloat_01 / SkFloatToHalf_01 (patchset #11 id:200001 of https://codereview.chromium.org/1685133005/ )

Reason for revert:
Gotta fix Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD

Original issue's description:
> SkHalfToFloat_01 / SkFloatToHalf_01
>
> These are basically inlined, 4-at-a-time versions of our existing functions,
> but cut down to avoid any work that's only necessary outside [0,1].
>
> Both f16 and f32 denorms should work fine modulo the usual ARMv7 NEON denorm==zero caveat.
>
> In exchange for a little speed, f32->f16 does not round properly.
> Instead it truncates, so it's never off by more than 1 bit.
>
> Support for finite values >1 or <0 is straightforward to add back.
> >1 might already work as-is.
>
> Getting close to _u16 performance:
>     micros    bench
>     261.13   xferu64_bw_1_opaque_u16
>    1833.51   xferu64_bw_1_alpha_u16
>    2762.32 ? xferu64_aa_1_opaque_u16
>    3334.29   xferu64_aa_1_alpha_u16
>     249.78   xferu64_bw_1_opaque_f16
>    3383.18   xferu64_bw_1_alpha_f16
>    4214.72   xferu64_aa_1_opaque_f16
>    4701.19   xferu64_aa_1_alpha_f16
>
>
> BUG=skia:
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1685133005
>
> Committed: https://skia.googlesource.com/skia/+/9ea11a4235b3e3521cc8bf914a27c2d0dc062db9

TBR=jvanverth@google.com,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/1693443003

SkHalfToFloat_01 / SkFloatToHalf_01

These are basically inlined, 4-at-a-time versions of our existing functions,
but cut down to avoid any work that's only necessary outside [0,1].

Both f16 and f32 denorms should work fine modulo the usual ARMv7 NEON denorm==zero caveat.

In exchange for a little speed, f32->f16 does not round properly.
Instead it truncates, so it's never off by more than 1 bit.

Support for finite values >1 or <0 is straightforward to add back.
>1 might already work as-is.

Getting close to _u16 performance:
    micros    bench
    261.13   xferu64_bw_1_opaque_u16
   1833.51   xferu64_bw_1_alpha_u16
   2762.32 ? xferu64_aa_1_opaque_u16
   3334.29   xferu64_aa_1_alpha_u16
    249.78   xferu64_bw_1_opaque_f16
   3383.18   xferu64_bw_1_alpha_f16
   4214.72   xferu64_aa_1_opaque_f16
   4701.19   xferu64_aa_1_alpha_f16

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1685133005

Review URL: https://codereview.chromium.org/1685133005

Move surface props off of GrTextContext

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1685653004

Review URL: https://codereview.chromium.org/1685653004

skia: Add support for CHROMIUM_image backed textures.

I created a new abstract base class TextureStorageAllocator that consumers of
Skia can subclass and pass back to Skia. When a surface is created with a
pointer to a TextureStorageAllocator, any textures it creates, or that are
derived from the original surface, will allocate and deallocate storage using
the methods on TextureStorageAllocator.

BUG=https://code.google.com/p/chromium/issues/detail?id=579664
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1623653002

Committed: https://skia.googlesource.com/skia/+/92098e691f10a010e7421125ba4d44c02506bb55

Committed: https://skia.googlesource.com/skia/+/7fec91ce6660190f8d7c5eb6f3061e4550cc672b

Committed: https://skia.googlesource.com/skia/+/b8d6e088590160f1198110c2371b802c1d541a36

Review URL: https://codereview.chromium.org/1623653002

Add bench and gm for shapes

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1677253002

patch from issue 1686673002 at patchset 20001 (http://crrev.com/1686673002#ps20001)

Review URL: https://codereview.chromium.org/1677253002

skiaserve: Fix /cmd/N and /cmd/N[0|1], the response was never queued.

BUG=skia:
R=joshualitt@google.com, ethannicholas@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1686553006

Review URL: https://codereview.chromium.org/1686553006

added JSON support for various lesser-used features
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1690613002

Review URL: https://codereview.chromium.org/1690613002

More efficient Jpeg skipScanlines for older versions of libjpeg-turbo

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1683283002

Review URL: https://codereview.chromium.org/1683283002

Fix bug processing bmp bit masks

BUG=skia:4866
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1688003002

Review URL: https://codereview.chromium.org/1688003002

added SkDebugCanvas::toJSON
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1684233002

Review URL: https://codereview.chromium.org/1684233002

Documentation for accessing the different repositories of SKPs

BUG=skia:2306
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1683233002
NOTRY=true
DOCS_PREVIEW= https://skia.org/?cl=1683233002

Review URL: https://codereview.chromium.org/1683233002

Start whittling down GrTextContext

TBR=
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1684733005

Review URL: https://codereview.chromium.org/1684733005

added some missing static modifiers
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1690683002

Review URL: https://codereview.chromium.org/1690683002

Remove GrTextBlobCache/GrAtlasTextBlob friendliness

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1686113002

Review URL: https://codereview.chromium.org/1686113002

disable flaky gm for now

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1689653002

TBR=

Review URL: https://codereview.chromium.org/1689653002

Revert of Make SkPicture/SkImageGenerator default to SkCodec (patchset #7 id:120001 of https://codereview.chromium.org/1671193002/ )

Reason for revert:
Breaks Ubuntu and Mac CMAKE

Original issue's description:
> Make SkPicture/SkImageGenerator default to SkCodec
>
> Remove reference to SkImageDecoder from SkPicture. Make the default
> InstallPixelRefProc passed to CreateFromStream use
> SkImageGenerator::NewFromEncoded instead.
>
> Make SkImageGenerator::NewFromEncoded create an SkCodecImageGenerator.
> Remove the old version that used SkImageDecoder.
>
> Remove all versions of lazy_decode_bitmap/LazyDecodeBitmap. The default
> now behaves lazily.
>
> Update all clients to use the default.
>
> Move SkImageDecoderGenerator into KtxTest.cpp, and use it directly.
>
> BUG=skia:4691
> BUG=skia:4290
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1671193002
>
> Committed: https://skia.googlesource.com/skia/+/026388a01864c74208ad57d1ba4f711602d101c6

TBR=msarett@google.com,reed@google.com,scroggo@google.com
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:4691

Review URL: https://codereview.chromium.org/1685963004

Make SkPicture/SkImageGenerator default to SkCodec

Remove reference to SkImageDecoder from SkPicture. Make the default
InstallPixelRefProc passed to CreateFromStream use
SkImageGenerator::NewFromEncoded instead.

Make SkImageGenerator::NewFromEncoded create an SkCodecImageGenerator.
Remove the old version that used SkImageDecoder.

Remove all versions of lazy_decode_bitmap/LazyDecodeBitmap. The default
now behaves lazily.

Update all clients to use the default.

Move SkImageDecoderGenerator into KtxTest.cpp, and use it directly.

BUG=skia:4691
BUG=skia:4290
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1671193002

Review URL: https://codereview.chromium.org/1671193002

Turn on conservative raster clip for GPU canvases
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1682223002

Review URL: https://codereview.chromium.org/1682223002

Finalize gpu shaders in ProgramBuilding.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1689703002

Review URL: https://codereview.chromium.org/1689703002

Fix for mips build

TBR=scroggo@google.com
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1688823002

Review URL: https://codereview.chromium.org/1688823002

move compose-shader into base-class as factory

next steps:
1. this lands
2. update android to call the new factory
3. move SkComposeShader.h into private

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1686803003

Review URL: https://codereview.chromium.org/1686803003

cleanup textblob cache sanity check and prevent it from bitrotting

TBR=bsalomon@google.com
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1684513004

Review URL: https://codereview.chromium.org/1684513004

Clean up comments in SkSwizzler

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1690583002

Review URL: https://codereview.chromium.org/1690583002

Sk4f: floor() via int32_t roundtrip.

About 25% faster on both x86 and ARMv7.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1682953002
CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Review URL: https://codereview.chromium.org/1682953002

Enable ImageGenerator tests on direct raster and direct gpu backends

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1682973004

Review URL: https://codereview.chromium.org/1682973004

Moved Canvas->JSON and JSON->Canvas functionality into SkDebugCanvas.
SkJSONCanvas and SkJSONRenderer will be deleted once the debug server
has been updated.
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1681643002

Committed: https://skia.googlesource.com/skia/+/32fff1ba0c759d97ab78320b7ae46dd7696a4e6e

Committed: https://skia.googlesource.com/skia/+/46cb6d6b82b0ab71033aef91419e8beae25bb8a4

Review URL: https://codereview.chromium.org/1681643002

Revert of Moved Canvas->JSON and JSON->Canvas functionality into SkDebugCanvas. (patchset #3 id:40001 of https://codereview.chromium.org/1681643002/ )

Reason for revert:
breaks various android builds, e.g.

https://android-build.storage.googleapis.com/builds/git_master-skia-linux-bullhead-userdebug/2596406/acb02999d48ace628601452700d4b4f8dcd5ed71b426e6ecdb1c7f5bc5452f7b/logs/build_error.log?Signature=BkLEdMOdi0LelP0v939QECdTZzznSD2dLFIkPu52E7%2BPr5Oj1ktq5nNlxGZn%2F4ZhCKXAw2%2FNHxy27nNJ8lffcgXn7PhQpa82bP%2B8F%2Bbyo9j%2BT%2B3etmOdQFEejJSRsEv2RrVWtKsKYRX%2FuZinrF9yD37jUpSbKNZvmvIbKa4V4IQ%3D&GoogleAccessId=701025073339-mqn0q2nvir9iurm6q5d00tdv7blbgvjr%40developer.gserviceaccount.com&Expires=1455108483

external/skia/bench/../tools/debugger/SkDrawCommand.cpp:602:9: error: use of undeclared identifier 'free'
        free(data);

Original issue's description:
> Moved Canvas->JSON and JSON->Canvas functionality into SkDebugCanvas.
> SkJSONCanvas and SkJSONRenderer will be deleted once the debug server
> has been updated.
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1681643002
>
> Committed: https://skia.googlesource.com/skia/+/32fff1ba0c759d97ab78320b7ae46dd7696a4e6e
>
> Committed: https://skia.googlesource.com/skia/+/46cb6d6b82b0ab71033aef91419e8beae25bb8a4

TBR=jcgregorio@google.com,joshualitt@chromium.org,ethannicholas@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/1683093002

Revert of skia: Add support for CHROMIUM_image backed textures. (patchset #19 id:380001 of https://codereview.chromium.org/1623653002/ )

Reason for revert:
Breaks ASAN bot:

Direct leak of 56 byte(s) in 1 object(s) allocated from:
...
test_CustomTexture

https://uberchromegw.corp.google.com/i/client.skia/builders/Test-Ubuntu-GCC-Golo-GPU-GT610-x86_64-Debug-ASAN/builds/2676/steps/dm/logs/stdio

Original issue's description:
> skia: Add support for CHROMIUM_image backed textures.
>
> I created a new abstract base class TextureStorageAllocator that consumers of
> Skia can subclass and pass back to Skia. When a surface is created with a
> pointer to a TextureStorageAllocator, any textures it creates, or that are
> derived from the original surface, will allocate and deallocate storage using
> the methods on TextureStorageAllocator.
>
> BUG=https://code.google.com/p/chromium/issues/detail?id=579664
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1623653002
>
> Committed: https://skia.googlesource.com/skia/+/92098e691f10a010e7421125ba4d44c02506bb55
>
> Committed: https://skia.googlesource.com/skia/+/7fec91ce6660190f8d7c5eb6f3061e4550cc672b
>
> Committed: https://skia.googlesource.com/skia/+/b8d6e088590160f1198110c2371b802c1d541a36

TBR=bsalomon@chromium.org,cblume@chromium.org,bsalomon@google.com,robertphillips@google.com,egdaniel@google.com,reed@google.com,erikchen@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=https://code.google.com/p/chromium/issues/detail?id=579664

Review URL: https://codereview.chromium.org/1684993002

turn off debug svg file

TBR=fmalita@chromium.org
BUG=skia:4549
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1674303004

Review URL: https://codereview.chromium.org/1674303004

Sk4f: add floor()

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1685773002
CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Committed: https://skia.googlesource.com/skia/+/86c6c4935171a1d2d6a9ffbff37ec6dac1326614

CQ_EXTRA_TRYBOTS=client.skia.android:Test-Android-GCC-Nexus7-GPU-Tegra3-Arm7-Release-Trybot,Test-Android-GCC-Nexus9-GPU-TegraK1-Arm64-Release-Trybot

Review URL: https://codereview.chromium.org/1685773002

Revert of Sk4f: add floor() (patchset #3 id:40001 of https://codereview.chromium.org/1685773002/ )

Reason for revert:
build break must be this, right?

Original issue's description:
> Sk4f: add floor()
>
> BUG=skia:
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1685773002
> CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot
>
> Committed: https://skia.googlesource.com/skia/+/86c6c4935171a1d2d6a9ffbff37ec6dac1326614

TBR=herb@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/1679343004

Moved Canvas->JSON and JSON->Canvas functionality into SkDebugCanvas.
SkJSONCanvas and SkJSONRenderer will be deleted once the debug server
has been updated.
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1681643002

Committed: https://skia.googlesource.com/skia/+/32fff1ba0c759d97ab78320b7ae46dd7696a4e6e

Review URL: https://codereview.chromium.org/1681643002

make test consistent

Don't pass rand() to a function as a pair of
parameters, since the order of evaluation
is unknown

TBR=robertphillips@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1682203002

Review URL: https://codereview.chromium.org/1682203002

Sk4f: add floor()

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1685773002
CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Review URL: https://codereview.chromium.org/1685773002

skia: Add support for CHROMIUM_image backed textures.

I created a new abstract base class TextureStorageAllocator that consumers of
Skia can subclass and pass back to Skia. When a surface is created with a
pointer to a TextureStorageAllocator, any textures it creates, or that are
derived from the original surface, will allocate and deallocate storage using
the methods on TextureStorageAllocator.

BUG=https://code.google.com/p/chromium/issues/detail?id=579664
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1623653002

Committed: https://skia.googlesource.com/skia/+/92098e691f10a010e7421125ba4d44c02506bb55

Committed: https://skia.googlesource.com/skia/+/7fec91ce6660190f8d7c5eb6f3061e4550cc672b

Review URL: https://codereview.chromium.org/1623653002

add drawCircle to c api

BUG=skia:4927
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1684763002

TBR=

Review URL: https://codereview.chromium.org/1684763002

add helper to create fancy underlines

Add a couple of utility functions to SkPaint that return the bounds
of glyphs between a pair of lines.

The common use case envisioned generates the edges of descenders
between the top and bottom bounds of an underline to allow computing
a stroke that skips those descenders.

The implementation stores a linked list in each glyph containing
the bounds of the lines parallel to the advance and the outermost
intersections within those bounds.

When the glyph cache is constructed, the glyph path is intersected
with the bounds and the extreme min and max values within the bounds
is added to an intercept.

Share the text to path iter to construct the data.

Make a half-hearted attempt to support vertical text; while the
vertical implementation is complete; surrounding code (e.g. paint
align) has short-comings with vertical.

R=fmalita@chromium.org, reed@google.com

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1654883003

Review URL: https://codereview.chromium.org/1654883003

moved debugger support files from src/utils/debugger to tools/debugger
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1678893003

Review URL: https://codereview.chromium.org/1678893003

Disable indirect backends for image decode tests

BUG=skia:4924
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1684713002

Review URL: https://codereview.chromium.org/1684713002

update DM deps in g3 build

Added https://skia.googlesource.com/skia/+/f1d415188ffb4c34e2886c2cfceb363a148333f1

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1683803002

Review URL: https://codereview.chromium.org/1683803002

implement more xfermodeproc4f and add GM

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1677293004

Review URL: https://codereview.chromium.org/1684623002

Remove SkNEW_APPEND_TO_TARRAY.

The use of SkNEW_APPEND_TO_TARRAY is now better served by
SkTArray::emplace_back(...) which should now be used instead. The
existing users of SkNEW_APPEND_TO_TARRAY are converted and the code
relating to SkNEW_APPEND_TO_TARRAY is removed.

TBR=reed
This only removes code. The file should also be made private.

Review URL: https://codereview.chromium.org/1682083002

restore sk4i SSE specialization

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1679343003
CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Review URL: https://codereview.chromium.org/1679343003

Slim down Sk4fRoundtrip benches.

Looks like the conversions are complicated enough to not compile away.
No need to mess with random number generation, etc.
We still need to blackhole it into a volatile in the end.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1682793002

Review URL: https://codereview.chromium.org/1682793002

sknx refactoring

  - trim unused specializations (Sk4i, Sk2d) and apis (SkNx_dup)
  - expand apis a little
    * v[0] == v.kth<0>()
    * SkNx_shuffle can now convert to different-sized vectors, e.g. Sk2f <-> Sk4f
  - remove anonymous namespace

I believe it's safe to remove the anonymous namespace right now.
We're worried about violating the One Definition Rule; the anonymous namespace protected us from that.

In Release builds, this is mostly moot, as everything tends to inline completely.
In Debug builds, violating the ODR is at worst an inconvenience, time spent trying to figure out why the bot is broken.

Now that we're building with SSE2/NEON everywhere, very few bots have even a chance about getting confused by two definitions of the same type or function.  Where we do compile variants depending on, e.g., SSSE3, we do so in static inline functions.  These are not subject to the ODR.

I plan to follow up with a tedious .kth<...>() -> [...] auto-replace.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1683543002
CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Review URL: https://codereview.chromium.org/1683543002

Add unit test to feed valid SVG sequences to make sure that
path strings can be parsed without returning an error.

Draw the output through Skia and SVG to make sure they are
parsed correctly.

R=fmalita@chromium.org
BUG=skia:4549
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1675053002

Review URL: https://codereview.chromium.org/1675053002

Shot in the dark to fix 581472

BUG=581472
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1679743005

Review URL: https://codereview.chromium.org/1679743005

Add some backend flexibility for shader declarations in ProgramBuilding.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1682703003

Review URL: https://codereview.chromium.org/1682703003

start experimenting with 64bit frame-buffers

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1680523002

Review URL: https://codereview.chromium.org/1680523002

SkTArray to move when moving.

This updates SkTArray to move elements when possible, instead of always
copying them.

TBR=reed
Agreed moving is good. This should also become private.

Committed: https://skia.googlesource.com/skia/+/3c69348e725131150e4ab962dec1b3ff1148a6bd

Review URL: https://codereview.chromium.org/1672063002

Use skstd::remove_pointer_t in GrGLFunction
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1684513003

Review URL: https://codereview.chromium.org/1684513003

Relocate anisotropic mipmap logic to SkMipMap::extractLevel()

Pass a full x/y scale and defer the anisotropic heuristic to SkMipMap.

R=reed@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1686563002

Review URL: https://codereview.chromium.org/1686563002

skeleton for float <-> half optimized procs

Nothing fancy yet, just calls the serial code in a loop.

I will try to folow this up with at least some of:
   - SSE2 version of serial code
   - NEON version of serial code
   - NEON version using vcvt.f32.f16/vcvt.f16.f32
   - F16C (between AVX and AVX2) version using vcvtph2ps/vcvtps2ph
The last two are fastest but need runtime detection.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1686543003

Review URL: https://codereview.chromium.org/1686543003

Use min scale when selecting mip level

R=reed@google.com,robertphillips@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1681953004

Review URL: https://codereview.chromium.org/1681953004

Create a thermal manager class and wire it in to nanobench behind a flag

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1671573002

Review URL: https://codereview.chromium.org/1671573002