Make sp- via support SkDrawables; add a GM to test it.

BUG=skia:

Review URL: https://codereview.chromium.org/1212363005

Add libjpeg-turbo library (depends on yasm)
Mangle external function names to avoid conflict with libjpeg
Take advantage of direct color conversion (RGBA, BGRA, 565)
Prepare to use jpeg_skip_scanlines (when it is upstreamed)

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/b60c3f8291529303299262dba19b1a896060bd2d

Committed: https://skia.googlesource.com/skia/+/f8bf9181d7b0463c8e371755cfbb9ece90b34fc5

Committed: https://skia.googlesource.com/skia/+/e9e3ee33f30c14c31afd5fc3fe4dda7f15783c75

Committed: https://skia.googlesource.com/skia/+/40141b57f061fbfcc2fa38da942d9efe25aca4d0

Review URL: https://codereview.chromium.org/1180983002

add SkImage::NewFromBitmap

BUG=skia:

Review URL: https://codereview.chromium.org/1217053003

Revert of Revert of Fixing libpng transform use  (patchset #1 id:1 of https://codereview.chromium.org/1213743004/)

Reason for revert:
This appears to have been reverted in order to fix the DEPS roll in https://codereview.chromium.org/1214943004/

However, it only affects SkCodec, which is not used by Chromium. Relanding

Original issue's description:
> Revert of Fixing libpng transform use  (patchset #5 id:80001 of https://codereview.chromium.org/1214203005/)
>
> Reason for revert:
> DEPS roll failing
>
> Original issue's description:
> > This change:
> > - supports kGray correctly
> > - avoid extra call to png_get_IHDR by storing the bit depth
> > - call transforms as needed
> > - checks for tRNS alpha value in RGB and GRAY color types
> >
> >
> > BUG=skia:
> >
> > Committed: https://skia.googlesource.com/skia/+/9693037fd41b7ce545b44beaa3489dcfd915018c
>
> TBR=scroggo@google.com,msarett@google.com,emmaleer@google.com
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/6c90e09575c1a77aee060aa475fdb3d25a17d6a0

TBR=msarett@google.com,emmaleer@google.com,jvanverth@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1215853005

Fix up SkMatrix::isFinite call

Review URL: https://codereview.chromium.org/1228473004

When three or more edges are coincident, the logic needs
to compute the overlapping ranges and combine the winding
into a single destination.

This computes coincidence more rigorously, fixing the
edge cases exposed by this bug.

Also, add the ability to debug and dump pathop structures
from the coincident context.

TBR=reed@google.com
BUG=skia:3651

Review URL: https://codereview.chromium.org/1182493015

Add nanobench-style JSON output to llvm_coverage_run

Runs the build and run scripts for coverage, then dumps the results into
a file in nanobench-compatible format.

BUG=skia:2430

Review URL: https://codereview.chromium.org/1227523004

Use kLow_SkFilterQuality in DrawBitmapAABench

BUG=skia:4001
R=reed@google.com

Review URL: https://codereview.chromium.org/1222713007

Revert of Revert of Revert of Fix SkTileImageFilter clipping/cropRect interaction issue (patchset #1 id:1 of https://codereview.chromium.org/1207353004/)

Reason for revert:
Still want to disable due to pref regression

Original issue's description:
> Revert of Revert of Fix SkTileImageFilter clipping/cropRect interaction issue (patchset #1 id:1 of https://codereview.chromium.org/1219193002/)
>
> Reason for revert:
> Blocking the roll
>
> Original issue's description:
> > Revert of Fix SkTileImageFilter clipping/cropRect interaction issue (patchset #2 id:30001 of https://codereview.chromium.org/1210053003/)
> >
> > Reason for revert:
> > Perf regression: https://code.google.com/p/chromium/issues/detail?id=505564
> >
> > Original issue's description:
> > > Fix SkTileImageFilter clipping/cropRect interaction issue
> > >
> > > BUG=499499
> > >
> > > Committed: https://skia.googlesource.com/skia/+/157bcd0840b578060dbc3365daafffc6837da391
> >
> > TBR=reed@google.com,senorblanco@google.com,senorblanco@chromium.org,robertphillips@google.com
> > NOPRESUBMIT=true
> > NOTREECHECKS=true
> > NOTRY=true
> > BUG=499499
> >
> > Committed: https://skia.googlesource.com/skia/+/ebaf6a69bf604c85185e23aca3fb93308e747ff5
>
> TBR=reed@google.com,senorblanco@google.com,senorblanco@chromium.org,robertphillips@google.com,reed@chromium.org,bsalomon@google.com
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=499499
>
> Committed: https://skia.googlesource.com/skia/+/835510085062f055c04d8ea46d82831cfbe51793

TBR=reed@google.com,senorblanco@google.com,senorblanco@chromium.org,reed@chromium.org,bsalomon@google.com,jvanverth@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=499499

Review URL: https://codereview.chromium.org/1217303004

Add scripts for running LLVM coverage

BUG=skia:2430

Review URL: https://codereview.chromium.org/1213063009

Add a bench for measuring drawBitmap anti-aliasing overhead

R=reed@google.com,mtklein@google.com

Review URL: https://codereview.chromium.org/1207403007

Update SKP version

Automatic commit by the RecreateSKPs bot.

TBR=
NO_MERGE_BUILDS

Review URL: https://codereview.chromium.org/1222483003

add colortable param to newrastercopy

BUG=skia:
TBR=
NOTREECHECKS=True

Review URL: https://codereview.chromium.org/1222683004

explicitly bump legacy genID on gpu surface

Previous attempt to "test" getTextureHandle was giving false-positive because we had an active image-snapshot, which was incidentally also triggering a new genID.

I think this CL will also enable this change: https://codereview.chromium.org/1222243002/

BUG=skia:
TBR=bsalomon@google.com
NOTREECHECKS=True

Review URL: https://codereview.chromium.org/1208993008

Revert of Revert of Fix SkTileImageFilter clipping/cropRect interaction issue (patchset #1 id:1 of https://codereview.chromium.org/1219193002/)

Reason for revert:
Blocking the roll

Original issue's description:
> Revert of Fix SkTileImageFilter clipping/cropRect interaction issue (patchset #2 id:30001 of https://codereview.chromium.org/1210053003/)
>
> Reason for revert:
> Perf regression: https://code.google.com/p/chromium/issues/detail?id=505564
>
> Original issue's description:
> > Fix SkTileImageFilter clipping/cropRect interaction issue
> >
> > BUG=499499
> >
> > Committed: https://skia.googlesource.com/skia/+/157bcd0840b578060dbc3365daafffc6837da391
>
> TBR=reed@google.com,senorblanco@google.com,senorblanco@chromium.org,robertphillips@google.com
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=499499
>
> Committed: https://skia.googlesource.com/skia/+/ebaf6a69bf604c85185e23aca3fb93308e747ff5

TBR=reed@google.com,senorblanco@google.com,senorblanco@chromium.org,robertphillips@google.com,reed@chromium.org,bsalomon@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=499499

Review URL: https://codereview.chromium.org/1207353004

Revert of Implement support for CHROMIUM_path_rendering pseudo extension (patchset #4 id:60001 of https://codereview.chromium.org/1192663002/)

Reason for revert:
DEPS roll failing

Original issue's description:
> Implement support for CHROMIUM_path_rendering pseudo extension
>
> Implement support for path rendering in Chromium through
> CHROMIUM_path_rendering pseudo extension.
>
> The extension defines a new pseudo-gl function,
> BindFragmentInputLocation. This behaves similarly to the
> BindUniformLocation pseudo-gl function. The idea is to assign fragment
> input location to a fragment input before linking the program.
>
> BUG=chromium:344330
>
> Committed: https://skia.googlesource.com/skia/+/eeef46d181f9f8db388ecea81df699fc1b3c9280

TBR=bsalomon@google.com,joshualitt@google.com,kkinnunen@nvidia.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:344330

Review URL: https://codereview.chromium.org/1223673002

Revert of Switch SkJpegCode to libjpeg-turbo (patchset #29 id:750001 of https://codereview.chromium.org/1180983002/)

Reason for revert:
DEPS roll failing

Original issue's description:
> Add libjpeg-turbo library (depends on yasm)
> Mangle external function names to avoid conflict with libjpeg
> Take advantage of direct color conversion (RGBA, BGRA, 565)
> Prepare to use jpeg_skip_scanlines (when it is upstreamed)
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/b60c3f8291529303299262dba19b1a896060bd2d
>
> Committed: https://skia.googlesource.com/skia/+/f8bf9181d7b0463c8e371755cfbb9ece90b34fc5
>
> Committed: https://skia.googlesource.com/skia/+/e9e3ee33f30c14c31afd5fc3fe4dda7f15783c75
>
> Committed: https://skia.googlesource.com/skia/+/40141b57f061fbfcc2fa38da942d9efe25aca4d0

TBR=scroggo@google.com,djsollen@google.com,emmaleer@google.com,msarett@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1226543003

Revert of Fixing libpng transform use  (patchset #5 id:80001 of https://codereview.chromium.org/1214203005/)

Reason for revert:
DEPS roll failing

Original issue's description:
> This change:
> - supports kGray correctly
> - avoid extra call to png_get_IHDR by storing the bit depth
> - call transforms as needed
> - checks for tRNS alpha value in RGB and GRAY color types
>
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/9693037fd41b7ce545b44beaa3489dcfd915018c

TBR=scroggo@google.com,msarett@google.com,emmaleer@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1213743004

Revert of Fix libjpeg-turbo deps file (patchset #2 id:2 of https://codereview.chromium.org/1216303007/)

Reason for revert:
DEPS roll failing

Original issue's description:
> Fix libjpeg-turbo deps file
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/4380f06a124c28a47dc99956d48c2d4c5bcb78e0

TBR=msarett@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1226453003

Revert of Remove SkThread.h, include SkMutex.h or SkAtomics.h as appropriate. (patchset #4 id:60001 of https://codereview.chromium.org/1215393002/)

Reason for revert:
Breaking the roll.

E.g. on android_chromium_gn_compile_dbg:

FAILED: /b/build/goma/gomacc ../../third_party/android_tools/ndk/toolchains/arm-linux-androideabi-4.9/prebuilt/linux-x86_64/bin/arm-linux-androideabi-g++ -MMD -MF obj/skia/ext/libskia.SkMemory_new_handler.o.d -DV8_DEPRECATION_WARNINGS -DCLD_VERSION=1 -DENABLE_NOTIFICATIONS -DENABLE_BROWSER_CDMS -DENABLE_PRINTING=1 -DENABLE_BASIC_PRINTING=1 -DDONT_EMBED_BUILD_METADATA -DUSE_OPENSSL=1 -DUSE_OPENSSL_CERTS=1 -DNO_TCMALLOC -DDISABLE_NACL -DENABLE_CONFIGURATION_POLICY -DENABLE_SUPERVISED_USERS=1 -DENABLE_AUTOFILL_DIALOG=1 -DUSE_PROPRIETARY_CODECS -DV8_USE_EXTERNAL_STARTUP_DATA -DVIDEO_HOLE=1 -DMOBILE_SAFE_BROWSING -DSAFE_BROWSING_SERVICE -DCHROMIUM_BUILD -D_FILE_OFFSET_BITS=64 -DANDROID -DHAVE_SYS_UIO_H -DCOMPONENT_BUILD -D__GNU_SOURCE=1 -D_DEBUG -DDYNAMIC_ANNOTATIONS_ENABLED=1 -DWTF_USE_DYNAMIC_ANNOTATIONS=1 -DSKIA_IMPLEMENTATION=1 -DSK_ARM_HAS_OPTIONAL_NEON -DSK_GAMMA_APPLY_TO_A8 -DSK_GAMMA_EXPONENT=1.4 -DSK_GAMMA_CONTRAST=0.0 -DSK_DEFAULT_FONT_CACHE_LIMIT=1048576 -DSK_IGNORE_LINEONLY_AA_CONVEX_PATH_OPTS -DSKIA_DLL -DGR_GL_IGNORE_ES3_MSAA=0 -DSK_SUPPORT_GPU=1 -DSK_BUILD_FOR_ANDROID -DUSE_CHROMIUM_SKIA -DXML_STATIC -I../.. -Igen -I../../third_party/skia/include/private -I../../third_party/skia/src/core -I../../third_party/skia/src/image -I../../third_party/skia/src/opts -I../../third_party/skia/src/pdf -I../../third_party/skia/src/ports -I../../third_party/skia/src/sfnt -I../../third_party/skia/src/utils -I../../third_party/skia/src/lazy -I../../skia/config -I../../skia/ext -I../../third_party/skia/include/c -I../../third_party/skia/include/config -I../../third_party/skia/include/core -I../../third_party/skia/include/effects -I../../third_party/skia/include/images -I../../third_party/skia/include/lazy -I../../third_party/skia/include/pathops -I../../third_party/skia/include/pdf -I../../third_party/skia/include/pipe -I../../third_party/skia/include/ports -I../../third_party/skia/include/utils -I../../third_party/skia/include/gpu -I../../third_party/skia/src/gpu -I../../third_party/zlib -I../../third_party/expat/files/lib -I../../third_party/freetype-android/include -I../../third_party/freetype-android/src/include -I../../third_party/android_tools/ndk/sources/android/cpufeatures -fno-strict-aliasing -march=armv7-a -mfloat-abi=softfp -mtune=generic-armv7-a -mthumb -mthumb-interwork -fno-tree-sra -fno-caller-saves -funwind-tables -fPIC -pipe -ffunction-sections -funwind-tables -fno-short-enums -finline-limit=64 -mfpu=vfpv3-d16 -isystem../../third_party/android_tools/ndk/sources/cxx-stl/llvm-libc++/libcxx/include -isystem../../third_party/android_tools/ndk/sources/cxx-stl/llvm-libc++abi/libcxxabi/include -isystem../../third_party/android_tools/ndk/sources/android/support/include -fvisibility=hidden --sysroot=/b/build/slave/android_chromium_gn/build/src/third_party/android_tools/ndk/platforms/android-16/arch-arm -Os -fdata-sections -ffunction-sections -fomit-frame-pointer -g1 -Wno-format -Wendif-labels -Werror -Wno-missing-field-initializers -Wno-unused-parameter -Wno-psabi -Wno-extra -Wno-ignored-qualifiers -Wno-type-limits -Wno-unused-local-typedefs -fno-threadsafe-statics -fvisibility-inlines-hidden -std=gnu++11 -fno-rtti -fno-exceptions -Wno-deprecated -Wno-narrowing -Wno-literal-suffix -Wno-error=c++0x-compat -Wno-non-virtual-dtor -Wno-sign-promo -c ../../skia/ext/SkMemory_new_handler.cpp -o obj/skia/ext/libskia.SkMemory_new_handler.o
../../skia/ext/SkMemory_new_handler.cpp:12:52: fatal error: third_party/skia/include/core/SkThread.h: No such file or directory
 #include "third_party/skia/include/core/SkThread.h"

Original issue's description:
> Remove SkThread.h, include SkMutex.h or SkAtomics.h as appropriate.
>
> SkThread.h doesn't do anything anymore execept include those two,
> and thankfully, it doesn't seem to be mentioned outside Skia.
>
> No public API changes.
> TBR=reed@google.com
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/c50acf2321d7a934c80d754e9cbe936dfb8eb4cc

TBR=mtklein@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1214603003

Implement support for CHROMIUM_path_rendering pseudo extension

Implement support for path rendering in Chromium through
CHROMIUM_path_rendering pseudo extension.

The extension defines a new pseudo-gl function,
BindFragmentInputLocation. This behaves similarly to the
BindUniformLocation pseudo-gl function. The idea is to assign fragment
input location to a fragment input before linking the program.

BUG=chromium:344330

Review URL: https://codereview.chromium.org/1192663002

Fix libjpeg-turbo deps file

BUG=skia:

Review URL: https://codereview.chromium.org/1216303007.

whitespace

This change:
- supports kGray correctly
- avoid extra call to png_get_IHDR by storing the bit depth
- call transforms as needed
- checks for tRNS alpha value in RGB and GRAY color types

BUG=skia:

Review URL: https://codereview.chromium.org/1214203005

Add libjpeg-turbo library (depends on yasm)
Mangle external function names to avoid conflict with libjpeg
Take advantage of direct color conversion (RGBA, BGRA, 565)
Prepare to use jpeg_skip_scanlines (when it is upstreamed)

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/b60c3f8291529303299262dba19b1a896060bd2d

Committed: https://skia.googlesource.com/skia/+/f8bf9181d7b0463c8e371755cfbb9ece90b34fc5

Committed: https://skia.googlesource.com/skia/+/e9e3ee33f30c14c31afd5fc3fe4dda7f15783c75

Review URL: https://codereview.chromium.org/1180983002

Remove SkThread.h, include SkMutex.h or SkAtomics.h as appropriate.

SkThread.h doesn't do anything anymore execept include those two,
and thankfully, it doesn't seem to be mentioned outside Skia.

No public API changes.
TBR=reed@google.com

BUG=skia:

Review URL: https://codereview.chromium.org/1215393002

Add support for drawBitmapRect() to SkMiniRecorder.

BUG=chromium:503705

Review URL: https://codereview.chromium.org/1220733006

Revert "Move headers used by headers in include/ to include/private."

This reverts commit 928e16565f3e69cd6661e9ad300ac17e3f33c0c8.

BUG=skia:

Review URL: https://codereview.chromium.org/1213093004.

Revert of Switch SkJpegCode to libjpeg-turbo (patchset #28 id:710001 of https://codereview.chromium.org/1180983002/)

Reason for revert:
Broke iOS build.

Original issue's description:
> Add libjpeg-turbo library (depends on yasm)
> Mangle external function names to avoid conflict with libjpeg
> Take advantage of direct color conversion (RGBA, BGRA, 565)
> Prepare to use jpeg_skip_scanlines (when it is upstreamed)
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/b60c3f8291529303299262dba19b1a896060bd2d
>
> Committed: https://skia.googlesource.com/skia/+/f8bf9181d7b0463c8e371755cfbb9ece90b34fc5
>
> Committed: https://skia.googlesource.com/skia/+/e9e3ee33f30c14c31afd5fc3fe4dda7f15783c75

TBR=scroggo@google.com,djsollen@google.com,emmaleer@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1213093003

Don't cap num_cores at 32 on 32-bit Windows.

See here for a similar change in Ninja.  Wanna bet they just got Z840s too?
https://github.com/martine/ninja/commit/8c8834acffdc0da0d94119725929acc712c9dfad

BUG=skia:

Review URL: https://codereview.chromium.org/1213063011

cleanup GLBenches

BUG=skia:

Review URL: https://codereview.chromium.org/1214203004

Add libjpeg-turbo library (depends on yasm)
Mangle external function names to avoid conflict with libjpeg
Take advantage of direct color conversion (RGBA, BGRA, 565)
Prepare to use jpeg_skip_scanlines (when it is upstreamed)

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/b60c3f8291529303299262dba19b1a896060bd2d

Committed: https://skia.googlesource.com/skia/+/f8bf9181d7b0463c8e371755cfbb9ece90b34fc5

Review URL: https://codereview.chromium.org/1180983002

Move headers used by headers in include/ to include/private.

Some of this is transitive, like SkRecords.h used by SkMiniRecorder.h
used by (public) SkPictureRecorder.h.

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/a89f55198bdc58f0b6f6196907ab25a6afc1a661

Review URL: https://codereview.chromium.org/1217293004

Revert of Fix SkTileImageFilter clipping/cropRect interaction issue (patchset #2 id:30001 of https://codereview.chromium.org/1210053003/)

Reason for revert:
Perf regression: https://code.google.com/p/chromium/issues/detail?id=505564

Original issue's description:
> Fix SkTileImageFilter clipping/cropRect interaction issue
>
> BUG=499499
>
> Committed: https://skia.googlesource.com/skia/+/157bcd0840b578060dbc3365daafffc6837da391

TBR=reed@google.com,senorblanco@google.com,senorblanco@chromium.org,robertphillips@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=499499

Review URL: https://codereview.chromium.org/1219193002

Revert "Move headers used by headers in include/ to include/private."

This reverts commit a89f55198bdc58f0b6f6196907ab25a6afc1a661.

Reason: breaking the roll.

TBR=mtklein@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true

Review URL: https://codereview.chromium.org/1216033008

Move headers used by headers in include/ to include/private.

Some of this is transitive, like SkRecords.h used by SkMiniRecorder.h
used by (public) SkPictureRecorder.h.

BUG=skia:

Review URL: https://codereview.chromium.org/1217293004

fix for angle

BUG=skia:

Review URL: https://codereview.chromium.org/1216963011

Thread through a flag to force SkPicture::playback() when recording subpictures.

This makes nanobench picture recording benchmarks somewhat useful again,
as opposed to all taking about 5us to run no matter the content.

ATTN Sheriff: this will probably trigger perf.skia.org alerts.

BUG=skia:

Review URL: https://codereview.chromium.org/1219873002

In the case of subset decodes, we will often not decode to
the bottom of the image.  In our code before this patch,
this would result in cleanup code in onFinish() never being
called.

We can allow subclasses to take ownership of the
SkScanlineDecoder in order to make sure that it is
finished/deleted before deleting the decode manager.

BUG=skia:

Review URL: https://codereview.chromium.org/1212593003

Benchmark to test cost for additional vertex attributes

BUG=skia:

Review URL: https://codereview.chromium.org/1213233011

Fix dm failures after Cleanup legacy NVPR-related definitions

Non-NV_path_rendering drivers should always return NULL context
for NVPR context type request. Fix this.

This was needed due to the previous patch trying to implement below:
NV_path_rendering drivers should return non-NULL context
for NVPR context type request, regardless whether path
rendering is actually supported or not. This is due to SAN
bots having older drivers and trying to run nvprmsaa config.

Review URL: https://codereview.chromium.org/1220513005

Cleanup legacy NVPR-related definitions

Fixed-function NVPR codepaths were removed a while ago. Only NVPR API
version 1.3 (PathFragmentInputGen) was left working. Remove
backwards-compatibility code that was left behind.

Remove some NVPR API function typedefs that were left from initial
commits.

Remove PathCoords function pointer from GrGLInterface, it has
never been called and causes problems in the future, since it will
not be implemented in the Chromium pseudo extension.

Avoid failing interface creation even if nvprmsaaXX config is
requested but the driver is not recent enough. The SAN bots have
old driver, but try to run nvprmsaa16 configs. Instead, print
out a warning.

Committed: https://skia.googlesource.com/skia/+/fb8d6884e0e01d0c2f8596adf5af1efb0d08de7e

Committed: https://skia.googlesource.com/skia/+/e35b5d99d8dfcc6b2be844df28cba47436380809

Review URL: https://codereview.chromium.org/1177243004

Revert of Skip SkSharedMutexMultiThreaded test on TSAN bot while investigating. (patchset #1 id:1 of https://codereview.chromium.org/1213033003/)

Reason for revert:
Add tsan annotations to repair tests.

Original issue's description:
> Skip SkSharedMutexMultiThreaded test on TSAN bot while investigating.
>
> The ~ means "don't run this".  This keeps all the other bots running the code, skipping it only on the whiny TSAN bot.
>
> BUG=skia:3997
>
> Committed: https://skia.googlesource.com/skia/+/37cc0b2e31e4a078b12790253d7032d36271440c

TBR=mtklein@google.com,mtklein@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:3997

Review URL: https://codereview.chromium.org/1218153004

Add annotations to SkSharedMutex.

BUG=skia:3997

Review URL: https://codereview.chromium.org/1213663004

Fix infinite matrix during an invert.

This patch ensures that when inverting a SkMatrix44, we handle small
floats properly. When inverted these can cause infinite values, but
still evaluate to true in an if condition.

BUG=chromium:498516

Review URL: https://codereview.chromium.org/1209763002

Fix CodecSubset benches seg faults for kIndex8

All of the CodecSubset benches fail when the color type is
kIndex8.  We need to pass a color table to allocPixels()
when we want to decode to kIndex8 or it will throw a failure.

BUG=skia:

Review URL: https://codereview.chromium.org/1213983003

Add SkSurface factory for wrapping an FBO in SkSurface

Review URL: https://codereview.chromium.org/1221853003

Rename flushForExternalRead->flushForExternalIO and always call in SkSurface::getTextureHandle

Review URL: https://codereview.chromium.org/1216243003

Skip SkSharedMutexMultiThreaded test on TSAN bot while investigating.

The ~ means "don't run this".  This keeps all the other bots running the code, skipping it only on the whiny TSAN bot.

BUG=skia:3997

Review URL: https://codereview.chromium.org/1213033003

Add wrapper script for computing n-grams from SKPs on Cluster Telemetry

Add --n argument

BUG=skia:3840

Review URL: https://codereview.chromium.org/1216073007

Pass arguments to SkRecords structs by const&.

This has the effect of using delay_copy() on every argument,
obviating the need for delay_copy() and, crucially, having
to remember to call delay_copy().

All these constructors are fully inlined, so we'll never pay
a penalty for passing small things by reference... the compiler
can see that and just pass by value.

BUG=skia:

Review URL: https://codereview.chromium.org/1215523004

Revert of Cleanup legacy NVPR-related definitions (patchset #4 id:60001 of https://codereview.chromium.org/1177243004/)

Reason for revert:
Breaks the Ubuntu *SAN bots.

Original issue's description:
> Cleanup legacy NVPR-related definitions
>
> Fixed-function NVPR codepaths were removed a while ago. Only NVPR API
> version 1.3 (PathFragmentInputGen) was left working. Remove
> backwards-compatibility code that was left behind.
>
> Remove some NVPR API function typedefs that were left from initial
> commits.
>
> Remove PathCoords function pointer from GrGLInterface, it has
> never been called and causes problems in the future, since it will
> not be implemented in the Chromium pseudo extension.
>
> Committed: https://skia.googlesource.com/skia/+/fb8d6884e0e01d0c2f8596adf5af1efb0d08de7e
>
> Committed: https://skia.googlesource.com/skia/+/e35b5d99d8dfcc6b2be844df28cba47436380809

TBR=joshualitt@google.com,cdalton@nvidia.com,bsalomon@google.com,kkinnunen@nvidia.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true

Review URL: https://codereview.chromium.org/1219663005

Remove default for --images in common flags

Review URL: https://codereview.chromium.org/1214363002

Use AddStory instead of deprecated AddUserStory

BUG=skia:3967
TBR=nednguyen
NOTRY=true

Review URL: https://codereview.chromium.org/1212843008

Fix blur bug on Andreno 430 with bool cast to float

BUG=skia:3944

Review URL: https://codereview.chromium.org/1214483007

Revert of Use page_set_module.PageSet instead of story.StorySet (patchset #1 id:1 of https://codereview.chromium.org/1221553003/)

Reason for revert:
This should no longer be required due to Ned's fix in https://codereview.chromium.org/1217823005/

Original issue's description:
> Use page_set_module.PageSet instead of story.StorySet
>
> BUG=skia:3967
> NOTRY=true
> TBR=nednguyen
>
> Committed: https://skia.googlesource.com/skia/+/7b971f0152299ae9a924252a9bfd220318497bdd

TBR=nednguyen@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:3967

Review URL: https://codereview.chromium.org/1214903008

Add benchmarkstream to visualbench

BUG=skia:

Review URL: https://codereview.chromium.org/1215033002

Update blur image filter bench to have crop variation

BUG=skia:3944

Review URL: https://codereview.chromium.org/1212973004

Change initializer list order to match field declaration order.

BUG=chromium:505304

Review URL: https://codereview.chromium.org/1214933004

Move visualbench to its own folder

BUG=skia:

Review URL: https://codereview.chromium.org/1216973002

Use page_set_module.PageSet instead of story.StorySet

BUG=skia:3967
NOTRY=true
TBR=nednguyen

Review URL: https://codereview.chromium.org/1221553003

Cleanup legacy NVPR-related definitions

Fixed-function NVPR codepaths were removed a while ago. Only NVPR API
version 1.3 (PathFragmentInputGen) was left working. Remove
backwards-compatibility code that was left behind.

Remove some NVPR API function typedefs that were left from initial
commits.

Remove PathCoords function pointer from GrGLInterface, it has
never been called and causes problems in the future, since it will
not be implemented in the Chromium pseudo extension.

Committed: https://skia.googlesource.com/skia/+/fb8d6884e0e01d0c2f8596adf5af1efb0d08de7e

Review URL: https://codereview.chromium.org/1177243004

Migrate PageSets to StorySets in buildbot page sets

BUG=skia:3967
NOTRY=true
TBR=

Review URL: https://codereview.chromium.org/1203433002

Refactor separable varying location info to be stored in GrGLProgram subclass

Refactor separable varying location info to be stored in GrGLProgram
subclass GrGLProgram instead of storing it in GrGLPathProcessor.

Separable varyings are exactly analoguous to uniforms: they are inputs
to the shader program. Shader compile-time information about uniforms is gathered to
GrGLProgramBuilder. This information is the converted to link-time
information, uniform locations, when constructing the program. Separable
varyings need to have same lifetime model.

This is needed in the future to support path rendering in Chromium. The
Chromium pseudo-extension will expose program fragment input binding
function similar to uniform binding function. Thus the separable varying
locations need to be decided and bound before link, e.g. before
GrGLProgram is created. This will be achieved in further patches by
overloading GrGLProgramBuilder::bindProgramResourceLocations() in
GrGLNvprProgramBuilder.

BUG=chromium:344330

Review URL: https://codereview.chromium.org/1186113007

Hide NVPR text behind distance field flag

TODO: Rename the distance field flag to be more general.

BUG=skia:

Review URL: https://codereview.chromium.org/1216973004

Remove old iOS porting files.

These files do not appear to be used, are unfinished, and should be
redundant with the mac porting files.

Review URL: https://codereview.chromium.org/1197963003

augment surface tests to check legacy genID

BUG=skia:

Review URL: https://codereview.chromium.org/1219793002

Make nanobench zoom animation time based

BUG=skia:

Review URL: https://codereview.chromium.org/1211253003

Implement shared locks in the style of pthread's rwlock.

BUG=skia:

Review URL: https://codereview.chromium.org/1215503005

Add script for triggering Cluster Telemetry jobs using local scripts

BUG=skia:3979

Review URL: https://codereview.chromium.org/1214843002

SoftLight with SkPMFloat

SSE speeds up about 4.5x over existing integer SSE,
NEON speeds up about 3x over serial integer code.

We expect 1-2 bit component diffs in the usual GMs.

Still guarded by SK_SUPPORT_LEGACY_XFERMODES,
which I'll now try to lift in Chrome.

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/3e47d49b46b3ab62071218ef3dd44642c9713e04

CQ_EXTRA_TRYBOTS=client.skia:Test-ChromeOS-GCC-Daisy-CPU-NEON-Arm7-Debug-Trybot

Review URL: https://codereview.chromium.org/1221493002

Edges matter, part 2.

Affected modes: lighten, hard-light, overlay (== hard-light).
This fixes a couple places where I used < when I should have used <=,
or swapped the logic as I've done here.

Caught by layout tests; our tests should be unchanged.
https://storage.googleapis.com/chromium-layout-test-archives/linux_blink_rel/68935/layout-test-results/css3/blending/background-blend-mode-crossfade-image-gradient-diffs.html

BUG=skia:

Review URL: https://codereview.chromium.org/1217013003

don't fix winding for empty paths

A path can be non-empty but become empty when it is simplified.
For instance, a path with the same rectangle, twice, with opposite
windings.

No contours are created for empty paths, so don't try to
fix their winding direction.

Additionally, check for a NULL coincidence since the
OpBuilder assumes that no concidence edges can be present
after the paths are simplified. This code should not get
called, but it's worth the future-proofing to check.

R=fmalita@chromium.org
BUG=502792

Review URL: https://codereview.chromium.org/1218863005

Revert of Switch SkJpegCode to libjpeg-turbo (patchset #25 id:680001 of https://codereview.chromium.org/1180983002/)

Reason for revert:
yasm is leaking memory :(.

https://uberchromegw.corp.google.com/i/client.skia/builders/Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Debug-ASAN/builds/926/steps/build%20dm/logs/stdio

Original issue's description:
> Add libjpeg-turbo library (depends on yasm)
> Mangle external function names to avoid conflict with libjpeg
> Take advantage of direct color conversion (RGBA, BGRA, 565)
> Prepare to use jpeg_skip_scanlines (when it is upstreamed)
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/b60c3f8291529303299262dba19b1a896060bd2d
>
> Committed: https://skia.googlesource.com/skia/+/f8bf9181d7b0463c8e371755cfbb9ece90b34fc5

TBR=scroggo@google.com,djsollen@google.com,emmaleer@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1214023003

Add libjpeg-turbo library (depends on yasm)
Mangle external function names to avoid conflict with libjpeg
Take advantage of direct color conversion (RGBA, BGRA, 565)
Prepare to use jpeg_skip_scanlines (when it is upstreamed)

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/b60c3f8291529303299262dba19b1a896060bd2d

Review URL: https://codereview.chromium.org/1180983002

Adding binary for yasm on x86 android devices

Revert of SoftLight with SkPMFloat (patchset #6 id:100001 of https://codereview.chromium.org/1221493002/)

Reason for revert:
xfermodes and xfermodes2 show major diffs on Nexus 5 and Daisy (both ARMv7 w/NEON).  Nexus 9 and SSE all look fine...

Original issue's description:
> SoftLight with SkPMFloat
>
> SSE speeds up about 4.5x over existing integer SSE,
> NEON speeds up about 3x over serial integer code.
>
> We expect 1-2 bit component diffs in the usual GMs.
>
> Still guarded by SK_SUPPORT_LEGACY_XFERMODES,
> which I'll now try to lift in Chrome.
>
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/3e47d49b46b3ab62071218ef3dd44642c9713e04

TBR=reed@google.com,mtklein@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1221683002

remove premature discard optimization

BUG=skia:

Review URL: https://codereview.chromium.org/1212613005

SoftLight with SkPMFloat

SSE speeds up about 4.5x over existing integer SSE,
NEON speeds up about 3x over serial integer code.

We expect 1-2 bit component diffs in the usual GMs.

Still guarded by SK_SUPPORT_LEGACY_XFERMODES,
which I'll now try to lift in Chrome.

BUG=skia:

Review URL: https://codereview.chromium.org/1221493002

add getTextureHandle to SkSurface

BUG=485243

Review URL: https://codereview.chromium.org/1210303003

Fix for valgrind use of uninitialized variable

TBR=bsalomon@google.com
BUG=skia:

Review URL: https://codereview.chromium.org/1217843002

Update SKP version

Automatic commit by the RecreateSKPs bot.

TBR=
NO_MERGE_BUILDS

Review URL: https://codereview.chromium.org/1218683003

Benchmark to test native perf of GL Instanced Ararys

BUG=skia:

Review URL: https://codereview.chromium.org/1216663002

Don't suppress nanobench output table in verbose mode

Changes verbose mode to print both the table and the individual sample
values. No need to hold back information in verbose mode.

BUG=skia:

Review URL: https://codereview.chromium.org/1208763003

add drawArraysInstanced to caps

BUG=skia:

Review URL: https://codereview.chromium.org/1202383006

Make SkPaint::glyphsToUnichars use explicit SkSurfaceProps

BUG=skia:3984,skia:3934

Review URL: https://codereview.chromium.org/1211413003

remove SkInstCnt

It's been outclassed by Valgrind and leak sanitizer,
and it seems to be causing problems for external folks building Skia.

I'm not sure why our own builds seem unaffected.

Latest thread:
https://groups.google.com/forum/#!topic/skia-discuss/oj9FsQwwSF0

BUG=skia:

Review URL: https://codereview.chromium.org/1217573002

Kill mixedxfermodes

It's too random to be helpful.

BUG=skia:

Review URL: https://codereview.chromium.org/1215703005

Turn off buffer mapping in Chromium

BUG=chromium:496224

Review URL: https://codereview.chromium.org/1211363006

Add extra braces for uintNNxMx4_t initializers.

These structs are always implemented as
    struct uintNNxMx4_t {
        uintNNxM val[4];
    };
So, the first set of braces is for the struct, the second for val.

BUG=skia:

Review URL: https://codereview.chromium.org/1221453002

Android scripts now notify gyp that your building with clang

Review URL: https://codereview.chromium.org/1217443004

Color dodge and burn with SkPMFloat.

Both 25-35% faster with SSE.
With NEON, Burn measures as a ~10% regression, Dodge a huge 2.9x improvement.

The Burn regression is somewhat artificial: we're drawing random colored rects onto an opaque white dst, so we're heavily biased toward the (d==da) fast path in the serial code.  In the vector code there's no short-circuiting and we always pay a fixed cost for ColorBurn regardless of src or dst content.

Dodge's fast paths, in contrast, only trigger when (s==sa) or (d==0), neither of which happens any more than randomly in our benchmark.  I don't think (d==0) should happen at all.  Similarly, the (s==0) Burn fast path is really only going to happen as often as SkRandom allows.

In practice, the existing Burn benchmark is hitting its fast path 100% of the time.  So I actually feel really great that this only dings the benchmark by 10%.

Chrome's still guarded by SK_SUPPORT_LEGACY_XFERMODES, which I'll lift after finishing the last xfermode, SoftLight.

BUG=skia:

Review URL: https://codereview.chromium.org/1214443002

Revert of Revert of remove SK_SUPPORT_LEGACY_PATHOP_ENUMS (patchset #1 id:1 of https://codereview.chromium.org/1203983003/)

Reason for revert:
fixes have landed in android

Original issue's description:
> Revert of remove SK_SUPPORT_LEGACY_PATHOP_ENUMS (patchset #1 id:1 of https://codereview.chromium.org/1200323005/)
>
> Reason for revert:
> Breaking the Android build.
>
> Original issue's description:
> > remove SK_SUPPORT_LEGACY_PATHOP_ENUMS
> >
> > BUG=skia:
> > TBR=
> >
> > Committed: https://skia.googlesource.com/skia/+/8bcc7a00febd737f7e82513bd5e4a697526626de
>
> TBR=reed@google.com
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/ba9a4aa8550c982c480d03cb6ba2cabaf6209b41

TBR=scroggo@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1218523002

Fix nanobench to reset gl/gr context after every config run.

BUG=skia:

Review URL: https://codereview.chromium.org/1214703002

Fix SkTileImageFilter clipping/cropRect interaction issue

BUG=499499

Review URL: https://codereview.chromium.org/1210053003

add/fix copyrights

BUG=skia:
TBR=

Review URL: https://codereview.chromium.org/1212393002

Revert of Revert of added stroking support to GrAALinearizingConvexPathRenderer (patchset #1 id:1 of https://codereview.chromium.org/1212833002/)

Reason for revert:
Manual roll is in: https://codereview.chromium.org/1212373002

Let's re-land this.

Original issue's description:
> Revert of added stroking support to GrAALinearizingConvexPathRenderer (patchset #6 id:100001 of https://codereview.chromium.org/1180903006/)
>
> Reason for revert:
> speculative revert to try to fix DEPS roll (layouttests)
>
> Original issue's description:
> > added stroking support to GrAALinearizingConvexPathRenderer
> >
> > BUG=skia:
> >
> > Committed: https://skia.googlesource.com/skia/+/2436f191e6602953b32a51cf50f2d7a4e2af90fd
>
> TBR=bsalomon@google.com,robertphillips@google.com,ethannicholas@google.com
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/9730f4a663534009d216c2f6d834bd534dd44a3d

TBR=bsalomon@google.com,robertphillips@google.com,ethannicholas@google.com,reed@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1209003004

Revert of added stroking support to GrAALinearizingConvexPathRenderer (patchset #6 id:100001 of https://codereview.chromium.org/1180903006/)

Reason for revert:
speculative revert to try to fix DEPS roll (layouttests)

Original issue's description:
> added stroking support to GrAALinearizingConvexPathRenderer
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/2436f191e6602953b32a51cf50f2d7a4e2af90fd

TBR=bsalomon@google.com,robertphillips@google.com,ethannicholas@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1212833002

Revert of Revert of Return GrGLContext from GrTestTarget (patchset #1 id:1 of https://codereview.chromium.org/1211123004/)

Reason for revert:
un-referting, did not help DEPS roll

Original issue's description:
> Revert of Return GrGLContext from GrTestTarget (patchset #3 id:40001 of https://codereview.chromium.org/1207393002/)
>
> Reason for revert:
> speculative revert to unblock DEPS roll (layout-tests)
>
> Original issue's description:
> > Return GrGLContext from GrTestTarget
> >
> > BUG=skia:
> >
> > Committed: https://skia.googlesource.com/skia/+/b6d93ea428ca5d13641c35faca7d0380980a3a17
>
> TBR=bsalomon@google.com,joshualitt@google.com,joshualitt@chromium.org
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/1852ec2b46a2a0efbb3550ff4d8b640f56a810e9

TBR=bsalomon@google.com,joshualitt@google.com,joshualitt@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1215593002

Add samplingTime mode to nanobench

Adds a nanobench mode that takes samples for a fixed amount of time,
rather than taking a fixed amount of samples.

BUG=skia:

Review URL: https://codereview.chromium.org/1204153002