Commit Graph

20979 Commits

Author SHA1 Message Date
joshualitt
3854f11ce3 Move urlhandlers out of skiaserve.cpp
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1737643003

Review URL: https://codereview.chromium.org/1737643003
2016-02-25 11:28:18 -08:00
ericrk
f57b3a6e4a Skia Filter Quality and Scaling Metrics
Adds histogram metrics to log the filter quality and scale factor of
each image draw. To make the data easier to consume, this is broken
down into a number of individual histograms:
- Filter quality across all draw calls
- Scale amount across all draw calls
- Scale amount per filter quality (4 histograms total)

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1686203002

Review URL: https://codereview.chromium.org/1686203002
2016-02-25 11:21:57 -08:00
bungeman
47a1e96b95 Deduplicate axis value resolving code.
The code was originally written for Android, but is now implemented
more generally by the FreeType scanner to be shared between all
FreeType backed typefaces. This removes the now duplicate code in the
Android font manager and uses the shared code instead.

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1726213004

Review URL: https://codereview.chromium.org/1726213004
2016-02-25 11:20:01 -08:00
joshualitt
26cc3f5ad0 Create Response namespace for skiaserve response functions
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1739763002

Review URL: https://codereview.chromium.org/1739763002
2016-02-25 11:09:36 -08:00
robertphillips
b56f92783a Hide GrPipelineBuilder from GrOvalRenderer
Limiting the reach of GrPipelineBuilder is always good.

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1726223008

Review URL: https://codereview.chromium.org/1726223008
2016-02-25 11:03:52 -08:00
scroggo
e4499849df Add a DM flag to limit the types of Codec decodes
When the flag "simpleCodec" is set to true, we will only decode
to the canvas colortype and to a scale of 1. No SkAndroidCodec or
BRDCodec tests are run. This makes it easier to find the simple
case when running locally.

Clean up some SkImageDecoder references along the way.
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1735743003

Review URL: https://codereview.chromium.org/1735743003
2016-02-25 11:03:47 -08:00
robertphillips
3552ba18ee Make skpinfo able to inspect SK_PICT_TYPEFACE_TAG blocks
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1732263003

BUG=skia:4691

Review URL: https://codereview.chromium.org/1732263003
2016-02-25 10:58:49 -08:00
joshualitt
4dcbe43124 Move some methods to SkiaServe's Request, this is still very leaky
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1733813004

Review URL: https://codereview.chromium.org/1733813004
2016-02-25 10:50:28 -08:00
benjaminwagner
f6bfccd562 Cleanups related to SkFixed.
This CL relands https://codereview.chromium.org/1683743005/ with two fixes:
  - Removing the test for vertical text. Vertical text doesn't work on Windows and breakText isn't supported for non-trivial text.
  - Omit PaintBreakTextTest in Google3 because we don't have the correct font setup yet.

Remove SK_FixedNaN. It does not seem to be supported or used anywhere in Skia, Chromium, Android, or Google3, (except accidentally by TwoPtRadial::kDontDrawT). I think supporting it would erase any benefit of SkFixed over float.

Remove SkBitmapFilter::lookup. It does not appear to be used anywhere in Skia, Chromium, Android, or Google3.

Fix a bug in SkPaint::breakText that limited it to ~5kB of text. Now it can handle more than 1GB.

BUG=skia:4632
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1683743005

Committed: https://skia.googlesource.com/skia/+/7ea5cb18389db11a94175de95c9db2b44972630c

patch from issue 1683743005 at patchset 120001 (http://crrev.com/1683743005#ps120001)

Review URL: https://codereview.chromium.org/1739453002
2016-02-25 10:28:11 -08:00
robertphillips
d77061067a Mv DRRect drawing code from GrDrawContext into SkGpuDevice
This is a TODO left over from https://codereview.chromium.org/1731413002/ (Hide GrDrawTarget from GrOvalRenderer). It is desireable since the GrDrawContext doesn't, in general, manipulate GrFragmentProcessors.

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1733993003

Review URL: https://codereview.chromium.org/1733993003
2016-02-25 09:28:08 -08:00
joshualitt
24dd687ac5 Move Request to its own file in preparation for numerous cleanups
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1738483003

Review URL: https://codereview.chromium.org/1738483003
2016-02-25 08:37:54 -08:00
bsalomon
5ec26ae9bf Move Budgeted enum out of SkSurface, use in GrTextureProvider
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1728093005
DOCS_PREVIEW= https://skia.org/?cl=1728093005

Committed: https://skia.googlesource.com/skia/+/57599fe6c0336feaeeeb9b1996e77b70219b483c
CQ_EXTRA_TRYBOTS=client.skia.compile:Build-Ubuntu-GCC-x86_64-Release-CMake-Trybot,Build-Mac-Clang-x86_64-Release-CMake-Trybot

Review URL: https://codereview.chromium.org/1728093005
2016-02-25 08:33:02 -08:00
jvanverth
f7e865a33a Add stub for new method to GrVkGpu
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1736983003

Review URL: https://codereview.chromium.org/1736983003
2016-02-25 07:59:45 -08:00
bungeman
fa5e5c7521 Use kSubpixelRounding in one missed location.
The kSubpixelRounding macro was created to clarify a block of code,
but one simplification was missed.

Review URL: https://codereview.chromium.org/1733843002
2016-02-25 07:07:32 -08:00
jcgregorio
50fe38e72a use the favicon on debugger.skia.org
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1734913002

Review URL: https://codereview.chromium.org/1734913002
2016-02-25 07:02:13 -08:00
mtklein
bb5f309bce 10.9 is no more
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1732353002

Review URL: https://codereview.chromium.org/1732353002
2016-02-25 06:52:18 -08:00
bsalomon
bd500f09ce Revert of Move Budgeted enum out of SkSurface, use in GrTextureProvider (patchset #6 id:100001 of https://codereview.chromium.org/1728093005/ )
Reason for revert:
Need workaround for chrome to build

Original issue's description:
> Move Budgeted enum out of SkSurface, use in GrTextureProvider
>
> BUG=skia:
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1728093005
> DOCS_PREVIEW= https://skia.org/?cl=1728093005
>
> Committed: https://skia.googlesource.com/skia/+/57599fe6c0336feaeeeb9b1996e77b70219b483c

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

Review URL: https://codereview.chromium.org/1734043002
2016-02-25 06:52:12 -08:00
Greg Daniel
2d07b76405 Update shaderc lib and use single combined lib
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1737653002
R=jvanverth@google.com

Review URL: https://codereview.chromium.org/1737653002
2016-02-25 09:47:33 -05:00
bsalomon
57599fe6c0 Move Budgeted enum out of SkSurface, use in GrTextureProvider
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1728093005
DOCS_PREVIEW= https://skia.org/?cl=1728093005

Review URL: https://codereview.chromium.org/1728093005
2016-02-25 06:33:27 -08:00
yujieqin
ddb347b63c Add -lrt flag for ChromeOS to enable the usage of clock_gettime. Enable RawCodec for ChromeOS
BUG=b/26813410
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1733223002

Review URL: https://codereview.chromium.org/1733223002
2016-02-25 05:42:30 -08:00
kkinnunen
45c2c8166b Implement support for using GL ES 3.0 with command buffer
Adds a new 'api': --config gpu(api=commandbuffer3) for dm/nanobench.

BUG=skia:4943
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1684413003

Review URL: https://codereview.chromium.org/1684413003
2016-02-25 02:03:43 -08:00
mtklein
3eed7ddb25 trim DM fat
- cuts out a few fringe features like logs and notes tallies
 - moves vetos and blacklisting earlier so they're never really seen
 - many string strength reductions, either to const SkString& or const char*

CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Debug-MSAN-Trybot

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1727363004

Review URL: https://codereview.chromium.org/1727363004
2016-02-24 19:07:07 -08:00
Brian Salomon
dcceacf574 Fix ~,! typo when clearing kRenderTarget flag
TBR=brucedawson@chromium.org,robertphillips@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1727413004

Review URL: https://codereview.chromium.org/1727413004 .
2016-02-24 21:55:47 -05:00
mtklein
da884c49e2 DM: assert we ran as many tasks as we thought we would
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1731783003

Review URL: https://codereview.chromium.org/1731783003
2016-02-24 18:00:23 -08:00
halcanary
7d57124a2a Unit Tests: eliminate stray SkDebugf()s.
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1733113002

Review URL: https://codereview.chromium.org/1733113002
2016-02-24 17:59:16 -08:00
halcanary
8e9f5e39d7 SkPDF: fix scalar serialization
BUG=skia:4868
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1720863003

Review URL: https://codereview.chromium.org/1720863003
2016-02-24 15:46:46 -08:00
brianosman
fbe355208d Add skia.conf (configuration file that supports printShaders, etc...) to .gitignore
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1734563003

Review URL: https://codereview.chromium.org/1734563003
2016-02-24 15:38:48 -08:00
fmalita
36693b5ac8 Add 4f linear gradient benches
Plus some more regular tiling mode variations and random cleanups.

R=reed@google.com,mtklein@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1728283003

Review URL: https://codereview.chromium.org/1728283003
2016-02-24 15:32:57 -08:00
halcanary
1e440510d8 SkPDF/Bench: add bench for SkPDFSharedStream (deflate)
Also, simplify gyp.

BUG=skia:5009
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1730833003

Review URL: https://codereview.chromium.org/1730833003
2016-02-24 15:17:20 -08:00
ericrk
f7b8b8affe Add wrapBackendTextureAsRenderTarget API
Skia's GrTextureProvider currently exposes two APIs for wrapping backend
objects:
 * wrapBackendTexture - wraps a texture into a GrTexture. Depending on
   flags, this GrTexture can be converted to a GrRenderTarget. Skia
   manages the render target objects it may create to provide a render
   target for the texture. This allows Skia to create stencil buffers
   if needed and manager MSAA resolves.
 * wrapBackendRenderTarget - wraps a FBO into a GrRenderTarget. This
   object cannot be converted to a GrTexture. Skia does not manage
   the render target objects for such a GrRenderTarget, and as such
   cannot attach stencil buffers or perform MSAA resolves on the
   created GrRenderTarget.

Given these two options, wrapBackendTexture provides more versatility
and allows Skia more room for optimization. Chrome currently uses
wrapBackendTexture for this reason.

While these two functions cover most cases, they do not provide a way
for Skia to wrap a texture into a render target (and gain the MSAA and
stencil buffer management), without also creating a GrTexture. This is
problematic in cases where a texture can be bound to a render target,
but cannot be textured from, as is the case in Chrome's limited support
for GL_TEXTURE_RECTANGLE.

To address this, a new function is created:
 * wrapBackendTextureAsRenderTarget - wraps a texture into a
   GrRenderTarget. As with wrapBackendTexture, the created render
   target objects are fully managed by Skia. Unlike wrapBackendTexture
   no GrTexture is created, and the created object will never be
   textured from.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1709163003

Review URL: https://codereview.chromium.org/1709163003
2016-02-24 14:49:52 -08:00
robertphillips
0cc2f85a19 Hide GrDrawTarget from GrOvalRenderer
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1731413002

Review URL: https://codereview.chromium.org/1731413002
2016-02-24 13:36:56 -08:00
mtklein
afae30aeef Rethink DM output
Instead of printing out each task as it finishes, or as it starts and as it finishes, periodically print out what we're up to.

Example output:
https://build.chromium.org/p/client.skia/builders/Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-Shared-Trybot/builds/6436/steps/dm/logs/stdio

BUG=skia:4978
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1730943003

Review URL: https://codereview.chromium.org/1730943003
2016-02-24 12:28:32 -08:00
brianosman
ca34846584 Added a flag that forces SampleApp to continuously redraw, even in non-animated tests. Helpful when profiling or debugging.
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1731853003

Review URL: https://codereview.chromium.org/1731853003
2016-02-24 11:47:47 -08:00
benjaminwagner
43437c2f19 Round to nearest in ScalarTo256.
This is a followup to https://codereview.chromium.org/1693683002 per discussion on https://codereview.chromium.org/1691403002.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1723483004

Review URL: https://codereview.chromium.org/1723483004
2016-02-24 10:57:47 -08:00
halcanary
1cc345f699 Documentation: sync-and-gyp design document
NOTRY=true
DOCS_PREVIEW= https://skia.org/dev/design/sync-and-gyp?cl=1736483002

Review URL: https://codereview.chromium.org/1736483002
2016-02-24 10:50:35 -08:00
benjaminwagner
f156ea3c3e Change SkUnitScalarClampToByte to more accurate 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=1691403002

Review URL: https://codereview.chromium.org/1691403002
2016-02-24 10:07:52 -08:00
reed
3719886415 remove debugging printf
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1725173003

NOTRY=True
TBR=

Review URL: https://codereview.chromium.org/1725173003
2016-02-24 09:32:40 -08:00
halcanary
1abea462c8 SkPDF/Bench: add bench for SkPDFCreateBitmapObject
BUG=skia:5009

Review URL: https://codereview.chromium.org/1729943003
2016-02-24 09:25:58 -08:00
joshualitt
b8d864902b fixup GLPrograms textblob todo
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1730253002

Review URL: https://codereview.chromium.org/1730253002
2016-02-24 09:23:03 -08:00
caryclark
d78c088b61 fix path ops fuzz bug
If one path is empty and the other has extreme values, the
intermediate coincident paths cannot be resolved, but triggers
an assert that a data structure unexpectedly has zero-length.

Tunnel this failure back up to the top and return that the
entire path op fails.

A future optimization could detect the empty path and avoid
this, allowing the op to succeed -- not sure that it's worth
the additional logic though.

TBR=reed@google.com
BUG=535151
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1730293002

Review URL: https://codereview.chromium.org/1730293002
2016-02-24 09:03:07 -08:00
mtklein
d9381acb0e Turn off DM --pre_log on the bots. Too spammy.
BUG=skia:4978
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1734703002

Review URL: https://codereview.chromium.org/1734703002
2016-02-24 08:58:27 -08:00
jvanverth
970fd1fbea Use ARM fixes for SDF encoding.
This pulls out the encoding fixes by Joel Liang from
https://codereview.chromium.org/1643143002/
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1726763002

Review URL: https://codereview.chromium.org/1726763002
2016-02-24 08:33:10 -08:00
benjaminwagner
c2d35d8513 Revert of Simple cleanups related to SkFixed. (patchset #4 id:120001 of https://codereview.chromium.org/1683743005/ )
Reason for revert:
New test is failing on Windows.

Original issue's description:
> Cleanups related to SkFixed.
>
> Remove SK_FixedNaN. It does not seem to be supported or used anywhere in Skia, Chromium, Android, or Google3, (except accidentally by TwoPtRadial::kDontDrawT). I think supporting it would erase any benefit of SkFixed over float.
>
> Remove SkBitmapFilter::lookup. It does not appear to be used anywhere in Skia, Chromium, Android, or Google3.
>
> Fix a bug in SkPaint::breakText that limited it to ~5kB of text. Now it can handle more than 1GB.
>
> BUG=skia:4632
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1683743005
>
> Committed: https://skia.googlesource.com/skia/+/7ea5cb18389db11a94175de95c9db2b44972630c

TBR=mtklein@google.com,reed@google.com
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:4632

Review URL: https://codereview.chromium.org/1724283003
2016-02-24 08:29:11 -08:00
ethannicholas
f15b07b75c Revert of Include glGetMultisamplefv API (patchset #1 id:1 of https://codereview.chromium.org/1720963002/ )
Reason for revert:
This is in the blamelist for Visualbench failures:

https://build.chromium.org/p/client.skia/builders/Perf-Ubuntu-GCC-ShuttleA-GPU-GTX660-x86_64-Release-VisualBench/builds/1035

and seems the most likely candidate for the failure, so I am tentatively reverting.

Original issue's description:
> Include glGetMultisamplefv API
>
> BUG=skia:
> GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1720963002
>
> Committed: https://skia.googlesource.com/skia/+/69c1b24bcaa7cffd59753a75a7f3d155696fcac4

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

Review URL: https://codereview.chromium.org/1731333002
2016-02-24 08:19:04 -08:00
reed
8516dc1016 extend xfermodes gm to include 3 custom modes
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1736523002

Review URL: https://codereview.chromium.org/1736523002
2016-02-24 08:18:42 -08:00
benjaminwagner
9d24023c66 Ensure the Gaussian tail of circle blurs goes to zero.
Compare blurcircle GM masked so that values <255 are white and =255 are black:
https://x20web.corp.google.com/~benjaminwagner/review/1723383002/blurcircles-lt255-before.png
https://x20web.corp.google.com/~benjaminwagner/review/1723383002/blurcircles-lt255-after.png

This is a spinoff of https://codereview.chromium.org/1691403002, which was previously causing more "rectangles" in this GM.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1723383002

Review URL: https://codereview.chromium.org/1723383002
2016-02-24 07:51:33 -08:00
benjaminwagner
7ea5cb1838 Cleanups related to SkFixed.
Remove SK_FixedNaN. It does not seem to be supported or used anywhere in Skia, Chromium, Android, or Google3, (except accidentally by TwoPtRadial::kDontDrawT). I think supporting it would erase any benefit of SkFixed over float.

Remove SkBitmapFilter::lookup. It does not appear to be used anywhere in Skia, Chromium, Android, or Google3.

Fix a bug in SkPaint::breakText that limited it to ~5kB of text. Now it can handle more than 1GB.

BUG=skia:4632
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1683743005

Review URL: https://codereview.chromium.org/1683743005
2016-02-24 06:51:52 -08:00
reed
8f7b0b2d80 simplify/unify xferproc api
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1721223002

Review URL: https://codereview.chromium.org/1721223002
2016-02-24 06:21:36 -08:00
joshualitt
cbbc9df143 Create new GM to target translations problems in GrAtlasTextContext
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1721833005

Review URL: https://codereview.chromium.org/1721833005
2016-02-24 05:49:55 -08:00
fmalita
ca058f57a3 Retire SkShader::kSupports4f_Flag
No longer used except in one test.

Introduce kPrefers4f_Flag instead.

R=reed@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1726983002

Review URL: https://codereview.chromium.org/1726983002
2016-02-23 19:02:20 -08:00