platform/upstream/libSkiaSharp.git
8 years agoPossible fix Moto E compilation failure
robertphillips [Tue, 21 Jul 2015 19:06:52 +0000 (12:06 -0700)]
Possible fix Moto E compilation failure

It appears that the Adreno compiler is even more twitchy about gl_FragCoord handling than expected.

BUG=skia:4078

Review URL: https://codereview.chromium.org/1246773003

8 years agoDe-templatize Sk4pxXfermode code a bit.
mtklein [Tue, 21 Jul 2015 19:03:37 +0000 (12:03 -0700)]
De-templatize Sk4pxXfermode code a bit.

This deduplicates a few pieces of code:
  - we end up with one copy of each xfer32() driver loop instead of one per xfermode;
  - we end up with two* copies of each xfermode implementation instead of ten**.

* For a given Mode: Mode() itself and xfer_aa<Mode>().
** From unrolling: twice at a stride of 8, once at 4, once at 2, and once at 1, then all again for when we have AA.

This decreases the size of SkXfermode.o from 1.5M to 620K on x86-64 and from 1.3M to 680K on ARMv7+NEON.

If we wanted to, we could eliminate the xfer_aa<Mode>() copy by tagging each Mode() function as __attribute__((noinline)) or its equivalent.  This would result in another ~100K space savings.

Performance is affected in proportion to the original xfermode speed:
fast modes like Plus take the largest proportional hit, and slow modes
like HardLight or SoftLight see essentially no hit at all.

This adds SK_VECTORCALL to help keep this code fast on ARMv7 and Windows.  I've looked at the ARMv7 generated code... it looks good, even pretty.

For compatibility with SK_VECTORCALL, we now pass the vector-sized arguments by value instead of by reference.  Some refactoring now allows us to declare each mode as just a static function instead of a struct, which simplifies things.

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

BUG=skia:

Review URL: https://codereview.chromium.org/1242743004

8 years agoReenable yasm for Android x86 and x86-64 on Linux host
msarett [Tue, 21 Jul 2015 19:01:48 +0000 (12:01 -0700)]
Reenable yasm for Android x86 and x86-64 on Linux host

It turns out that gyp (kind of) has support for cross
compiling with a different host and target.  We simply
need to specify CC_host and CC_target instead of CC.

Making this change allows us to compile yasm on a Linux
host for Android.

We run into problems on Mac because
the linker on a Mac host requires different command line
arguments than the linker on the Android target.  In
looking through the code for gyp itself and speaking to
Ben, it doesn't appear to me that gyp supports passing
different arguments to host and target linkers.

I would imagine that we would have similar problems on
Windows.

Below is a link to a CL that would fix this issue in gyp.
It looks like it has been dropped for a long time.
Thanks to Ben for this link!
https://chromiumcodereview.appspot.com/10795044/

Also I'm adding a link to the build instructions for Chrome
(thanks again Ben).  It looks like they only support
building for Android from Linux.
https://code.google.com/p/chromium/wiki/AndroidBuildInstructions

My next steps are:
1) Getting in touch with Torne or someone else with gyp to
see if people are aware of this issue or interested in
fixing it.
2) Deciding if skia should care about this issue.
3) Deciding if skia should work around this issue.

It'd be really great to hear your thoughts on (2) and (3).
My first thought is that we shouldn't care because, as
long as we always compile the production copy of skia for
Android on Linux, we will get the fast code.  Is this
a valid conclusion?  Is there a way to write Android apps
on Mac that accidentally use the slower code?

If we do care, there are workarounds:
For Mac, we can check in a yasm binary - it's a little
smaller than the one I am deleting in this CL :-/
For Windows, we *might* be able to use the yasm.exe binary
already in externals (we get this from DEPS because this is
how chromium uses yasm on Windows).
Are there other platforms that we care about?

Let me know what you think!

BUG=skia:4028
DOCS_PREVIEW= https://skia.org/?cl=1239333002

Review URL: https://codereview.chromium.org/1239333002

8 years agoAdd sanity check to GrAtlasTextBlob
joshualitt [Tue, 21 Jul 2015 18:39:34 +0000 (11:39 -0700)]
Add sanity check to GrAtlasTextBlob

BUG=skia:

Review URL: https://codereview.chromium.org/1250693002

8 years agoRemove SK_LEGACY_IMAGE_FILTER_CROP_RECT_EDGES code.
senorblanco [Tue, 21 Jul 2015 18:11:17 +0000 (11:11 -0700)]
Remove SK_LEGACY_IMAGE_FILTER_CROP_RECT_EDGES code.

This is no longer used in Chrome.

BUG=skia:4089

Review URL: https://codereview.chromium.org/1245193002

8 years agomake varied_text* gm portable
caryclark [Tue, 21 Jul 2015 16:15:53 +0000 (09:15 -0700)]
make varied_text* gm portable

TBR=reed@google.com

Review URL: https://codereview.chromium.org/1244063002

8 years agoRun noise shader separately in imagefiltersclipped
jvanverth [Tue, 21 Jul 2015 15:38:06 +0000 (08:38 -0700)]
Run noise shader separately in imagefiltersclipped

BUG=skia:4079

Review URL: https://codereview.chromium.org/1245983002

8 years agorename BitmapTextBlob and move it to its own file
joshualitt [Tue, 21 Jul 2015 15:05:03 +0000 (08:05 -0700)]
rename BitmapTextBlob and move it to its own file

BUG=skia:

Review URL: https://codereview.chromium.org/1249663002

8 years agomake xfermode* portable
caryclark [Tue, 21 Jul 2015 14:42:45 +0000 (07:42 -0700)]
make xfermode* portable

TBR=reed@google.com

Review URL: https://codereview.chromium.org/1243103002

8 years agoSkip imagefiltersclipped GM on Galaxy S4
borenet [Tue, 21 Jul 2015 12:57:22 +0000 (05:57 -0700)]
Skip imagefiltersclipped GM on Galaxy S4

BUG=skia:4079

Review URL: https://codereview.chromium.org/1250473005

8 years agoRevert of 565 support for SIMD xfermodes (patchset #3 id:40001 of https://codereview...
mtklein [Tue, 21 Jul 2015 12:02:40 +0000 (05:02 -0700)]
Revert of 565 support for SIMD xfermodes (patchset #3 id:40001 of https://codereview.chromium.org/1245673002/)

Reason for revert:
942930d (included in this roll) introduced a 140 kB sizes regression in
libskia.so. Please investigate and reland if this regression is necessary.

Original issue's description:
> 565 support for SIMD xfermodes
>
> This uses the most basic approach possible:
>   - to load an Sk4px from 565, convert to SkPMColors on the stack serially then load those SkPMColors.
>   - to store an Sk4px to 565, store to SkPMColors on the stack then convert to 565 serially.
>
> Clearly, we can optimize these loads and stores.  That's a TODO.
>
> The code using SkPMFloat is the same idea but a little more long-term viable, as we're only operating on one pixel at a time anyway.  We could probably write 565 <-> SkPMFloat methods, but I'd rather not until it's really compelling.
>
> The speedups are varied but similar across SSE and NEON: a few uninteresting, many 50% faster, some 2x faster, and SoftLight ~4x faster.
>
> This will cause minor GM diffs, but I don't think any layout test changes.
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/942930dcaa51f66d82cdaf46ae62efebd16c8cd0

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

Review URL: https://codereview.chromium.org/1242973004

8 years agoAdd Rec709 YUV color space support to GrYUVtoRGBEffect.
rileya [Mon, 20 Jul 2015 22:00:03 +0000 (15:00 -0700)]
Add Rec709 YUV color space support to GrYUVtoRGBEffect.

This change is motivated by a recent switch in how chromium handles
<video> color spaces, making rec709 more commonly used. This will
allow video -> canvas copies to take the fast GPU path when we're using
709, just as we do with 601 and jpeg.

Chromium-side change: https://codereview.chromium.org/1236313002

Review URL: https://codereview.chromium.org/1241723005

8 years agoGenerate platform specific fontmgr gm.
caryclark [Mon, 20 Jul 2015 20:38:56 +0000 (13:38 -0700)]
Generate platform specific fontmgr gm.

The gm output on different font platforms is so different
that comparing images in Gold has little value. Separate the
images by appending platform information to the gm name to
group somewhat similar images together.

Note that this does not attempt to make sure that all images
generated by Gold are nearly pixel identical; it only reduces
the number of nonsensical comparisons.

R=bungeman@google.com

Review URL: https://codereview.chromium.org/1245643002

8 years agoremoving png_read_end from SkPngScanlineDecoder
emmaleer [Mon, 20 Jul 2015 20:38:44 +0000 (13:38 -0700)]
removing png_read_end from SkPngScanlineDecoder

BUG=skia:4041

Review URL: https://codereview.chromium.org/1242423002

9 years agomake stroketext textblob* texteffects tileimagefilter gm portable
caryclark [Mon, 20 Jul 2015 17:48:01 +0000 (10:48 -0700)]
make stroketext textblob* texteffects tileimagefilter gm portable

TBR=reed@google.com

Review URL: https://codereview.chromium.org/1244833002

9 years ago565 support for SIMD xfermodes
mtklein [Mon, 20 Jul 2015 17:35:30 +0000 (10:35 -0700)]
565 support for SIMD xfermodes

This uses the most basic approach possible:
  - to load an Sk4px from 565, convert to SkPMColors on the stack serially then load those SkPMColors.
  - to store an Sk4px to 565, store to SkPMColors on the stack then convert to 565 serially.

Clearly, we can optimize these loads and stores.  That's a TODO.

The code using SkPMFloat is the same idea but a little more long-term viable, as we're only operating on one pixel at a time anyway.  We could probably write 565 <-> SkPMFloat methods, but I'd rather not until it's really compelling.

The speedups are varied but similar across SSE and NEON: a few uninteresting, many 50% faster, some 2x faster, and SoftLight ~4x faster.

This will cause minor GM diffs, but I don't think any layout test changes.

BUG=skia:

Review URL: https://codereview.chromium.org/1245673002

9 years agoFix textureDomain/bleed prevention in msaa
robertphillips [Mon, 20 Jul 2015 17:22:29 +0000 (10:22 -0700)]
Fix textureDomain/bleed prevention in msaa

This updates Ganesh's bleed avoidance check to handle the case where the
sample location may be outside of the rect geometry but used because it
partially covers the pixel.

BUG=skia:4066

Review URL: https://codereview.chromium.org/1237623012

9 years agoSplit llvm_coverage_run into two scripts
borenet [Mon, 20 Jul 2015 16:43:36 +0000 (09:43 -0700)]
Split llvm_coverage_run into two scripts

... One for running to generate the coverage data, another to parse the data into various formats.

NOTRY=true
BUG=skia:2430

Review URL: https://codereview.chromium.org/1239963002

9 years agoClean up dead xfermode opts code.
mtklein [Mon, 20 Jul 2015 14:14:19 +0000 (07:14 -0700)]
Clean up dead xfermode opts code.

Now that SK_SUPPORT_LEGACY_XFERMODES is unused, tons of code becomes dead.

Nothing is needed in opts/ anymore for x86.
We still do runtime NEON detection, which just duplicates Sk4pxXfermode.

TBR=reed@google.com

BUG=skia:

Review URL: https://codereview.chromium.org/1230023011

9 years agoUpdate SKP version
skia.buildbots [Sun, 19 Jul 2015 07:19:11 +0000 (00:19 -0700)]
Update SKP version

Automatic commit by the RecreateSKPs bot.

TBR=
NO_MERGE_BUILDS

Review URL: https://codereview.chromium.org/1241073002

9 years agoReland "SkPDF: Annotations are clipped by canvas clip stack."
wangxianzhu [Sat, 18 Jul 2015 00:23:15 +0000 (17:23 -0700)]
Reland "SkPDF: Annotations are clipped by canvas clip stack."

Original patch was created by halcanary@google.com, and was reverted
because it triggered crbug.com/503541.

This patch fixes a bug in the original patch about clip path
transformation.

> Also, remove some SkPDFDevice functions.
> Will fix this GM: http://crrev.com/1159273003
> BUG=skia:3872
> Review URL: https://codereview.chromium.org/1148263005

BUG=skia:3872
BUG=503514

Review URL: https://codereview.chromium.org/1238503007

9 years agomake drawbitmaprect lighting lumafilter matrixconvolution srcmode gm portable
caryclark [Fri, 17 Jul 2015 20:20:48 +0000 (13:20 -0700)]
make drawbitmaprect lighting lumafilter matrixconvolution srcmode gm portable

TBR=reed@google.com

Review URL: https://codereview.chromium.org/1240843005

9 years agoadding unit test for textblob cache / GrBatchAtlas
joshualitt [Fri, 17 Jul 2015 19:47:39 +0000 (12:47 -0700)]
adding unit test for textblob cache / GrBatchAtlas

BUG=skia:

Review URL: https://codereview.chromium.org/1233253003

9 years agomake arithmode bigtext colortype* degeneratesegments gm portable
caryclark [Fri, 17 Jul 2015 16:39:16 +0000 (09:39 -0700)]
make arithmode bigtext colortype* degeneratesegments gm portable

TBR=reed@google.com

Review URL: https://codereview.chromium.org/1235343004

9 years agoSome cleanups of VisualBench
joshualitt [Fri, 17 Jul 2015 16:09:23 +0000 (09:09 -0700)]
Some cleanups of VisualBench

BUG=skia:

Review URL: https://codereview.chromium.org/1240633005

9 years agollvm_coverage_build: Use CC and CXX variables by default
borenet [Fri, 17 Jul 2015 14:31:07 +0000 (07:31 -0700)]
llvm_coverage_build: Use CC and CXX variables by default

NOTRY=true
BUG=skia:2430

Review URL: https://codereview.chromium.org/1242573003

9 years agoReland of Add fast normalize for SkLightingImageFilter. (patchset #1 id:1 of https...
jvanverth [Fri, 17 Jul 2015 14:22:30 +0000 (07:22 -0700)]
Reland of Add fast normalize for SkLightingImageFilter. (patchset #1 id:1 of https://codereview.chromium.org/1244523002/)

Reason for revert:
Florin will rebaseline the images.

Original issue's description:
> Revert of Add fast normalize for SkLightingImageFilter. (patchset #2 id:20001 of https://codereview.chromium.org/1240023002/)
>
> Reason for revert:
> Speculative revert -- DEPS roll block on linux_blink_rel
>
> https://storage.googleapis.com/chromium-layout-test-archives/linux_blink_rel/71483/layout-test-results/results.html
>
> Original issue's description:
> > Add fast normalize for SkLightingImageFilter.
> >
> > The normalize routine in SkPoint3 is very robust. However, for simple
> > lighting cases we prefer speed over robustness. This fixes a perf
> > regression in smoothness.tough_filters_cases.
> >
> > BUG=chromium:510562
> >
> > Committed: https://skia.googlesource.com/skia/+/dfa0ecf169db87f7afddd93bc1c500de481a62c7
>
> TBR=reed@google.com,senorblanco@google.com,senorblanco@chromium.org,jvanverth@google.com
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=chromium:510562
>
> Committed: https://skia.googlesource.com/skia/+/ac66a8122b27c388cc74b3913d9a9be351a44e54

TBR=reed@google.com,senorblanco@google.com,senorblanco@chromium.org,reed@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:510562

Review URL: https://codereview.chromium.org/1241583007

9 years agoRevert[2] of guard to remove DrawBitmapRectFlags
reed [Fri, 17 Jul 2015 14:09:43 +0000 (07:09 -0700)]
Revert[2] of guard to remove DrawBitmapRectFlags

This reverts commit 91110195a2eee170c11885da9d16f94b00a39f87.

BUG=skia:
TBR=

Review URL: https://codereview.chromium.org/1240753003

9 years agoAdd an entry in directory for BugChomper.
tfarina [Fri, 17 Jul 2015 14:07:25 +0000 (07:07 -0700)]
Add an entry in directory for BugChomper.

BUG=None
R=borenet@google.com
NOTRY=true
DOCS_PREVIEW= https://skia.org/?cl=1239803008

Review URL: https://codereview.chromium.org/1239803008

9 years agoNo longer using non-antialiased fast path for lines when antialiasing is on.
ethannicholas [Fri, 17 Jul 2015 13:44:02 +0000 (06:44 -0700)]
No longer using non-antialiased fast path for lines when antialiasing is on.

BUG=skia:3776

Review URL: https://codereview.chromium.org/1223323003

9 years agoRemoving GrTemplates.h
joshualitt [Fri, 17 Jul 2015 13:40:29 +0000 (06:40 -0700)]
Removing GrTemplates.h

BUG=skia:

Review URL: https://codereview.chromium.org/1233923004

9 years agoCleanup of GrFontDescKey
joshualitt [Fri, 17 Jul 2015 13:19:19 +0000 (06:19 -0700)]
Cleanup of GrFontDescKey

BUG=skia:

Review URL: https://codereview.chromium.org/1238473004

9 years agoRevert of guard to remove DrawBitmapRectFlags (patchset #1 id:1 of https://codereview...
scroggo [Fri, 17 Jul 2015 12:57:34 +0000 (05:57 -0700)]
Revert of guard to remove DrawBitmapRectFlags (patchset #1 id:1 of https://codereview.chromium.org/1235393003/)

Reason for revert:
breaking android framework build

Original issue's description:
> guard to remove DrawBitmapRectFlags
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/6fb0b6779e40ce05c20cf279f0ecff31fa3cd60d

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

Review URL: https://codereview.chromium.org/1230823007

9 years agoRevert of Add fast normalize for SkLightingImageFilter. (patchset #2 id:20001 of...
reed [Thu, 16 Jul 2015 23:02:39 +0000 (16:02 -0700)]
Revert of Add fast normalize for SkLightingImageFilter. (patchset #2 id:20001 of https://codereview.chromium.org/1240023002/)

Reason for revert:
Speculative revert -- DEPS roll block on linux_blink_rel

https://storage.googleapis.com/chromium-layout-test-archives/linux_blink_rel/71483/layout-test-results/results.html

Original issue's description:
> Add fast normalize for SkLightingImageFilter.
>
> The normalize routine in SkPoint3 is very robust. However, for simple
> lighting cases we prefer speed over robustness. This fixes a perf
> regression in smoothness.tough_filters_cases.
>
> BUG=chromium:510562
>
> Committed: https://skia.googlesource.com/skia/+/dfa0ecf169db87f7afddd93bc1c500de481a62c7

TBR=reed@google.com,senorblanco@google.com,senorblanco@chromium.org,jvanverth@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:510562

Review URL: https://codereview.chromium.org/1244523002

9 years agoNow that the SkImageFilter::CropRect crop edges refer to
senorblanco [Thu, 16 Jul 2015 22:19:11 +0000 (15:19 -0700)]
Now that the SkImageFilter::CropRect crop edges refer to
width & height, name them appropriately.

BUG=240827

Review URL: https://codereview.chromium.org/1234873005

9 years agoFix visual bench memory leak
joshualitt [Thu, 16 Jul 2015 21:23:22 +0000 (14:23 -0700)]
Fix visual bench memory leak

BUG=skia:

Review URL: https://codereview.chromium.org/1239103002

9 years agomake glyph_pos imageblur* largeglyphblur portable
caryclark [Thu, 16 Jul 2015 21:16:04 +0000 (14:16 -0700)]
make glyph_pos imageblur* largeglyphblur portable

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

Review URL: https://codereview.chromium.org/1243493003

9 years agoSmall fix for visualbench
joshualitt [Thu, 16 Jul 2015 20:40:51 +0000 (13:40 -0700)]
Small fix for visualbench

BUG=skia:

Review URL: https://codereview.chromium.org/1237563005

9 years agorefactor code to apply the croprect
reed [Thu, 16 Jul 2015 20:17:13 +0000 (13:17 -0700)]
refactor code to apply the croprect

BUG=skia:

Review URL: https://codereview.chromium.org/1218993002

9 years agoAdd fast normalize for SkLightingImageFilter.
jvanverth [Thu, 16 Jul 2015 19:56:46 +0000 (12:56 -0700)]
Add fast normalize for SkLightingImageFilter.

The normalize routine in SkPoint3 is very robust. However, for simple
lighting cases we prefer speed over robustness. This fixes a perf
regression in smoothness.tough_filters_cases.

BUG=chromium:510562

Review URL: https://codereview.chromium.org/1240023002

9 years agoStop running image tests on msaa.
scroggo [Thu, 16 Jul 2015 19:36:10 +0000 (12:36 -0700)]
Stop running image tests on msaa.

Blacklist all image tests on msaa. We do not run them anyway (since
they will not do anything interestingly different from drawing to the
raster backend) - we early exit from Src::draw(), but we still need to
create a render target that matches the size of the image (when not
blacklisted).

Remove the more specific blacklist of a particular image, which is
covered by this one.

BUG=skia:4045

Review URL: https://codereview.chromium.org/1234313006

9 years agomake imagefilters* imagemagnifier imageresizetiled portable
caryclark [Thu, 16 Jul 2015 19:35:58 +0000 (12:35 -0700)]
make imagefilters* imagemagnifier imageresizetiled portable

TBR=reed@google.com

Review URL: https://codereview.chromium.org/1238083004

9 years agoRevert of Remove rt flags that are not necessary (patchset #1 id:1 of https://coderev...
bsalomon [Thu, 16 Jul 2015 17:00:28 +0000 (10:00 -0700)]
Revert of Remove rt flags that are not necessary (patchset #1 id:1 of https://codereview.chromium.org/1226123012/)

Reason for revert:
breaks tests, e.g.
https://build.chromium.org/p/client.skia.android/builders/Test-Android-GCC-Nexus9-GPU-TegraK1-Arm64-Release/builds/752/steps/dm/logs/stdio

Original issue's description:
> Remove rt flags that are not necessary
>
> Committed: https://skia.googlesource.com/skia/+/b6b0a6ef179f4143c94ba7e0e1be6e7ce4b0a23a

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

Review URL: https://codereview.chromium.org/1234503004

9 years agoRemove rt flags that are not necessary
bsalomon [Thu, 16 Jul 2015 16:34:57 +0000 (09:34 -0700)]
Remove rt flags that are not necessary

Review URL: https://codereview.chromium.org/1226123012

9 years agoAll batches do their own pipeline compare
joshualitt [Thu, 16 Jul 2015 16:13:50 +0000 (09:13 -0700)]
All batches do their own pipeline compare

BUG=skia:

Review URL: https://codereview.chromium.org/1242033002

9 years agoMake readpixels work on GrTextures
bsalomon [Thu, 16 Jul 2015 15:23:13 +0000 (08:23 -0700)]
Make readpixels work on GrTextures

Review URL: https://codereview.chromium.org/1234313002

9 years agoGive GrBatch a pointer to GrPipeline
joshualitt [Thu, 16 Jul 2015 15:12:45 +0000 (08:12 -0700)]
Give GrBatch a pointer to GrPipeline

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/1170a12839218f7a23c93487bf95fd83aae0201f

Review URL: https://codereview.chromium.org/1237283007

9 years agoVarious GL testing fixes
jvanverth [Thu, 16 Jul 2015 14:46:07 +0000 (07:46 -0700)]
Various GL testing fixes

Check for GL context when printing NVPR error string.
Fix some indenting.

Review URL: https://codereview.chromium.org/1235283004

9 years agohave canvas send discard instead of retain if the draw would overwrite everything
reed [Thu, 16 Jul 2015 14:40:45 +0000 (07:40 -0700)]
have canvas send discard instead of retain if the draw would overwrite everything

Motivation:
- perf win for clients that overwrite the surface after a snapshot.
- may allow us to eliminate SkDeferredCanvas, as this was its primary advantage.

BUG=skia:

Review URL: https://codereview.chromium.org/1236023004

9 years agoRevert of Give GrBatch a pointer to GrPipeline (patchset #1 id:1 of https://coderevie...
joshualitt [Thu, 16 Jul 2015 14:16:53 +0000 (07:16 -0700)]
Revert of Give GrBatch a pointer to GrPipeline (patchset #1 id:1 of https://codereview.chromium.org/1237283007/)

Reason for revert:
breaking nanobench

Original issue's description:
> Give GrBatch a pointer to GrPipeline
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/1170a12839218f7a23c93487bf95fd83aae0201f

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

Review URL: https://codereview.chromium.org/1239073002

9 years agoGive GrBatch a pointer to GrPipeline
joshualitt [Thu, 16 Jul 2015 14:06:50 +0000 (07:06 -0700)]
Give GrBatch a pointer to GrPipeline

BUG=skia:

Review URL: https://codereview.chromium.org/1237283007

9 years agoincrease gm offsetimagefilter stroke width
caryclark [Thu, 16 Jul 2015 14:06:46 +0000 (07:06 -0700)]
increase gm offsetimagefilter stroke width

With a hairline width, offsetimagefilter fails to draw the
rectangle on some platforms. Make the width 2 so that it
shows up everywhere.

TBR=reed@google.com

Review URL: https://codereview.chromium.org/1236133003

9 years agoCoverage script: fix missing data
borenet [Thu, 16 Jul 2015 14:01:44 +0000 (07:01 -0700)]
Coverage script: fix missing data

BUG=skia:2430

Review URL: https://codereview.chromium.org/1242023002

9 years agoReplace buggy_blend_modes GM with an exhaustive test.
mtklein [Thu, 16 Jul 2015 14:01:39 +0000 (07:01 -0700)]
Replace buggy_blend_modes GM with an exhaustive test.

The new test is disabled by default, as it's quite slow.
We can run it if we suspect problems by passing -x to DM.

This test would have been failing before the bug fix, and now is passing.

Assuming the Priv on the end means it's not considered public API...
TBR=reed@google.com

BUG=skia:4052

Review URL: https://codereview.chromium.org/1228333003

9 years agoguard to remove DrawBitmapRectFlags
reed [Thu, 16 Jul 2015 13:10:02 +0000 (06:10 -0700)]
guard to remove DrawBitmapRectFlags

BUG=skia:

Review URL: https://codereview.chromium.org/1235393003

9 years agomake *imagefilter morphology iamge bitmapsource gm portable
caryclark [Wed, 15 Jul 2015 21:23:50 +0000 (14:23 -0700)]
make *imagefilter morphology iamge bitmapsource gm portable

TBR=reed@google.com

Review URL: https://codereview.chromium.org/1236403002

9 years agomake strokefill gm portable
caryclark [Wed, 15 Jul 2015 21:03:12 +0000 (14:03 -0700)]
make strokefill gm portable

Convert the Mac-specific font text statically into
paths so that they draw the same on all platforms.

Add drawing portable text using the fake bold paint
setting so that the code path is tested everywhere
and but draws the same thing.

R=reed@google.com

Review URL: https://codereview.chromium.org/1234463006

9 years agorename utility to see if a paint will overwrite its pixels
reed [Wed, 15 Jul 2015 20:42:32 +0000 (13:42 -0700)]
rename utility to see if a paint will overwrite its pixels

BUG=skia:

Review URL: https://codereview.chromium.org/1228853005

9 years agoAdded GLVec4ScalarBench that doesn't use VAOs
wangyix [Wed, 15 Jul 2015 19:26:07 +0000 (12:26 -0700)]
Added GLVec4ScalarBench that doesn't use VAOs

BUG=skia:

Review URL: https://codereview.chromium.org/1235153008

9 years agomake gradtext quadpaths gm portable
caryclark [Wed, 15 Jul 2015 18:57:00 +0000 (11:57 -0700)]
make gradtext quadpaths gm portable

TBR=reed@google.com

Review URL: https://codereview.chromium.org/1234233003

9 years agoFix leak in Android FontMgr test.
bungeman [Wed, 15 Jul 2015 18:29:25 +0000 (14:29 -0400)]
Fix leak in Android FontMgr test.

FontMgrAndroidParserTest deos not free the font families it creates.
This issue was discovered by asan when enabling this test on Linux
platforms.

9 years agoAllow Android font manager on *nix.
bungeman [Wed, 15 Jul 2015 17:58:47 +0000 (10:58 -0700)]
Allow Android font manager on *nix.

There is nothing left in the SkFontMgr_android.cpp which is specific
to Android, so allow it to be built and used on *nix platforms. This
allows for easier development and testing.

The only reason to limit to *nix platforms are the dependencies on
Expat and FreeType. This should be buildable and runnable on other
platforms when these dependencies are also available.

Review URL: https://codereview.chromium.org/1228833003

9 years agoFix for android command line arguments to VisualBench
joshualitt [Wed, 15 Jul 2015 17:17:20 +0000 (10:17 -0700)]
Fix for android command line arguments to VisualBench

BUG=skia:

Review URL: https://codereview.chromium.org/1241923002

9 years agoOnly run test_wrapped_resources() when GL is active
jvanverth [Wed, 15 Jul 2015 17:16:56 +0000 (10:16 -0700)]
Only run test_wrapped_resources() when GL is active

Review URL: https://codereview.chromium.org/1234233002

9 years agoDon't upload texture smaller than the min texture size when stretching a bmp up to...
bsalomon [Wed, 15 Jul 2015 17:12:16 +0000 (10:12 -0700)]
Don't upload texture smaller than the min texture size when stretching a bmp up to a POT

BUG=skia:3932

Review URL: https://codereview.chromium.org/1238913002

9 years agomake colorwheelnative displacement dropshadowimagefilter getposttextpath gm portable
caryclark [Wed, 15 Jul 2015 16:29:32 +0000 (09:29 -0700)]
make colorwheelnative displacement dropshadowimagefilter getposttextpath gm portable

TBR=reed@google.com

Review URL: https://codereview.chromium.org/1237233004

9 years agomake path-reverse gm portable
caryclark [Wed, 15 Jul 2015 15:58:12 +0000 (08:58 -0700)]
make path-reverse gm portable

Put font glyph data in test as path
instead of getting data from the Mac
font. THe font may change, and is
not available on most platforms.

Also remove unreachable code.

R=reed@google.com

Review URL: https://codereview.chromium.org/1230703005

9 years agoMatch works correctly on visualbench
joshualitt [Wed, 15 Jul 2015 15:38:02 +0000 (08:38 -0700)]
Match works correctly on visualbench

BUG=skia:

Review URL: https://codereview.chromium.org/1234123004

9 years agoexperimental: remove old PDF benchmarking tools
halcanary [Wed, 15 Jul 2015 15:06:29 +0000 (08:06 -0700)]
experimental: remove old PDF benchmarking tools

TBR=

Review URL: https://codereview.chromium.org/1233093004

9 years agollvm-coverage script: respect SKIA_OUT
borenet [Wed, 15 Jul 2015 14:43:06 +0000 (07:43 -0700)]
llvm-coverage script: respect SKIA_OUT

NOTRY=true
BUG=skia:2430

Review URL: https://codereview.chromium.org/1239803002

9 years agoFix undefined behavior.
herb [Wed, 15 Jul 2015 14:40:37 +0000 (07:40 -0700)]
Fix undefined behavior.

BUG=skia:4050

Review URL: https://codereview.chromium.org/1232003008

9 years agoOptimize RGB16 blitV functions with NEON for ARM platform.
yang.zhang [Wed, 15 Jul 2015 14:07:30 +0000 (07:07 -0700)]
Optimize RGB16 blitV functions with NEON for ARM platform.

Here are some performance resultsi on Nexus 9:
SkRGB16BlitterBlitV_neon:
+--------+-----------+
|height  |  C/NEON   |
+--------+-----------+
|1       | 0.765230  |
+--------+-----------+
|8       | 1.273330  |
+--------+-----------+
|18      | 1.441462  |
+--------+-----------+
|32      | 1.627798  |
+--------+-----------+
|76      | 1.683131  |
+--------+-----------+
|85      | 1.679456  |
+--------+-----------+
|120     | 1.721311  |
+--------+-----------+
|128     | 1.725482  |
+--------+-----------+
|512     | 1.784117  |
+--------+-----------+

BUG=skia:

Review URL: https://codereview.chromium.org/1213723002

9 years agoRevert of Added a GLBench for testing performance of vec4 vs scalar for coverage...
scroggo [Wed, 15 Jul 2015 13:36:42 +0000 (06:36 -0700)]
Revert of Added a GLBench for testing performance of vec4 vs scalar for coverage in generated shaders. (patchset #6 id:90001 of https://codereview.chromium.org/1225383002/)

Reason for revert:
Still failing on Windows. e.g. https://uberchromegw.corp.google.com/i/client.skia/builders/Perf-Win8-MSVC-ShuttleB-GPU-HD4600-x86_64-Release-ANGLE/builds/231/steps/nanobench/logs/stdio

skbug.com/4053

Original issue's description:
> Added a glBench for testing performance of vec4 vs scalar coverage in generated shaders.
>
> Added bench for timing vec4 vs scalar type for coverage in shaders
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/6104ced165f17eb2f765ace354d5895c0bc890c5
>
> Committed: https://skia.googlesource.com/skia/+/c734e69e8cf94bacaf68d3d8ee3310d1ad1fe8b8

TBR=joshualitt@google.com,tomhudson@google.com,wangyix@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1239503003

9 years agoFix VisualBench to hold onto a surface
joshualitt [Wed, 15 Jul 2015 13:18:57 +0000 (06:18 -0700)]
Fix VisualBench to hold onto a surface

BUG=skia:

Review URL: https://codereview.chromium.org/1236723003

9 years agomake linepaths gm portable
caryclark [Wed, 15 Jul 2015 13:18:47 +0000 (06:18 -0700)]
make linepaths gm portable

TBR=reed@google.com

Review URL: https://codereview.chromium.org/1225303010

9 years agoremove funkster from dcshader gm
caryclark [Wed, 15 Jul 2015 12:43:38 +0000 (05:43 -0700)]
remove funkster from dcshader gm

The multicolored font that funster enables
only draws and measures on some platforms.

Removing it from the dcshader gm allows
the layout to be uniform on all devices,
making it easier to compare the output
in gold.

TBR=bsalomon@google.com

Review URL: https://codereview.chromium.org/1231853004

9 years agoDocument new CQ keyword
rmistry [Wed, 15 Jul 2015 11:40:54 +0000 (04:40 -0700)]
Document new CQ keyword

BUG=skia:
NOTRY=true
DOCS_PREVIEW= https://skia.org/?cl=1237693005
TBR=jcgregorio

Review URL: https://codereview.chromium.org/1237693005

9 years agoFixed valgrind issues with libjpeg-turbo
msarett [Tue, 14 Jul 2015 21:39:59 +0000 (14:39 -0700)]
Fixed valgrind issues with libjpeg-turbo

The fixes are in the updated libjpeg-turbo repository
pulled in by DEPS.  The fixes are detailed in the linked
skia bug.  To summarize briefly, we now use calloc()
instead of malloc().

BUG=skia:4030

Review URL: https://codereview.chromium.org/1237213004

9 years agoUpdated PDFViewer drawImageRect() overrides
fmalita [Tue, 14 Jul 2015 21:15:08 +0000 (14:15 -0700)]
Updated PDFViewer drawImageRect() overrides

R=reed@google.com

Review URL: https://codereview.chromium.org/1240573008

9 years agoLet Android use old version of onDrawBitmapRect.
scroggo [Tue, 14 Jul 2015 20:59:10 +0000 (13:59 -0700)]
Let Android use old version of onDrawBitmapRect.

Define SK_SUPPORT_LEGACY_ONDRAWIMAGERECT when building for the
Android framework, since SkiaCanvasProxy overrides the old version
of the method.

Fixes master-skia build.

NOTRY=True

Review URL: https://codereview.chromium.org/1237903002

9 years agoFix buggy blend modes.
mtklein [Tue, 14 Jul 2015 20:38:28 +0000 (13:38 -0700)]
Fix buggy blend modes.

The problem turns out to be not over- or underflow like I thought, but that I used different math for x*y/255 for alphas and colors, sometimes resulting in colors where alpha was one less than the maximum color component, which is not a valid SkPMColor.

To be safe, I've switched over all four of these similar complex modes to use exact math everywhere.  They now match the byte-by-byte code in SkXfermode.cpp exactly.

This will slow down Darken and Lighten by about 2x.  I plan to follow up with a CL to see if I can eek out some speed there, and another CL to add asserts that Sk4px code always writes valid SkPMColors.

BUG=skia:4052

Review URL: https://codereview.chromium.org/1241683003

9 years agoPreserve SKP cullrects on deserialization
fmalita [Tue, 14 Jul 2015 20:12:25 +0000 (13:12 -0700)]
Preserve SKP cullrects on deserialization

Let's not ignore the offset.

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

Review URL: https://codereview.chromium.org/1235953004

9 years agoVery tiny paths are subsumed by the Convexicator
caryclark [Tue, 14 Jul 2015 18:19:26 +0000 (11:19 -0700)]
Very tiny paths are subsumed by the Convexicator
and are treated as convex when they are not.

Allow the SkPath::Iter to leave degenerate path
segments unmolested by passing an additional exact
bool to next().

Treat any non-zero length as significant in addPt().

R=reed@google.com,robertphillips@google.com
BUG=493450

Review URL: https://codereview.chromium.org/1228383002

9 years agoRemove GrRedBlackTree
joshualitt [Tue, 14 Jul 2015 18:07:53 +0000 (11:07 -0700)]
Remove GrRedBlackTree

BUG=skia:

Review URL: https://codereview.chromium.org/1226203013

9 years agoRename backing texture testing routines
jvanverth [Tue, 14 Jul 2015 18:02:52 +0000 (11:02 -0700)]
Rename backing texture testing routines

Review URL: https://codereview.chromium.org/1230193006

9 years agoAnother trivial cleanup
joshualitt [Tue, 14 Jul 2015 18:02:45 +0000 (11:02 -0700)]
Another trivial cleanup

TBR=bsalomon@google.com
BUG=skia:

Committed: https://skia.googlesource.com/skia/+/52e7657cd850f95e66eb23c6d138ee45149a1039

Review URL: https://codereview.chromium.org/1229303003

9 years agoDisable optimization as HWUI does not yet support color shaders.
djsollen [Tue, 14 Jul 2015 17:59:28 +0000 (10:59 -0700)]
Disable optimization as HWUI does not yet support color shaders.

Review URL: https://codereview.chromium.org/1235913005

9 years agoImplement support for CHROMIUM_path_rendering pseudo extension
kkinnunen [Tue, 14 Jul 2015 17:59:23 +0000 (10:59 -0700)]
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

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

9 years ago3-15% speedup to HardLight / Overlay xfermodes.
mtklein [Tue, 14 Jul 2015 17:54:19 +0000 (10:54 -0700)]
3-15% speedup to HardLight / Overlay xfermodes.

While investigating my bug (skia:4052) I saw this TODO and figured
it'd make me feel better about an otherwise unsuccessful investigation.

This speeds up HardLight and Overlay (same code) by about 15% with SSE, mostly
by rewriting the logic from 1 cheap comparison and 2 expensive div255() calls
to 2 cheap comparisons and 1 expensive div255().

NEON speeds up by a more modest ~3%.

BUG=skia:

Review URL: https://codereview.chromium.org/1230663005

9 years agoadd src-rect-constraint to drawImageRect
reed [Tue, 14 Jul 2015 17:54:12 +0000 (10:54 -0700)]
add src-rect-constraint to drawImageRect

Follow-on work
- unify around SrcRectConstraint (i.e. drawBitmapRect)
- remove silly drawBitmapRectToRect alias
- clean-up (possibly remove) alias problems around drawBitmapRect + IRect parameter

BUG=skia:

Review URL: https://codereview.chromium.org/1228083004

9 years agoAdd a GYP flag to define SKNX_NO_SIMD when set.
mtklein [Tue, 14 Jul 2015 12:52:52 +0000 (05:52 -0700)]
Add a GYP flag to define SKNX_NO_SIMD when set.

Will use this with a new -SKNX_NO_SIMD bot.

BUG=skia:

Review URL: https://codereview.chromium.org/1227163016

9 years agoFix typo in non-SIMD Sk4f::thenElse().
mtklein [Tue, 14 Jul 2015 12:23:50 +0000 (05:23 -0700)]
Fix typo in non-SIMD Sk4f::thenElse().

This was causing the 3 xfermodes that use floats and conditionals
to draw wrong when SKNX_NO_SIMD was defined.

BUG=skia:4051

Review URL: https://codereview.chromium.org/1229013003

9 years agoRevert of Another trivial cleanup (patchset #6 id:100001 of https://codereview.chromi...
joshualitt [Mon, 13 Jul 2015 20:29:13 +0000 (13:29 -0700)]
Revert of Another trivial cleanup (patchset #6 id:100001 of https://codereview.chromium.org/1229303003/)

Reason for revert:
breaking things

Original issue's description:
> Another trivial cleanup
>
> TBR=bsalomon@google.com
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/52e7657cd850f95e66eb23c6d138ee45149a1039

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

Review URL: https://codereview.chromium.org/1233853004

9 years agoAdd new SkPoint3 class
robertphillips [Mon, 13 Jul 2015 20:16:44 +0000 (13:16 -0700)]
Add new SkPoint3 class

The existing Light filter and the upcoming Lighting Shader both need a Point3 class

Review URL: https://codereview.chromium.org/1229693009

9 years agoAnother trivial cleanup
joshualitt [Mon, 13 Jul 2015 20:11:40 +0000 (13:11 -0700)]
Another trivial cleanup

TBR=bsalomon@google.com
BUG=skia:

Review URL: https://codereview.chromium.org/1229303003

9 years agoAdded a glBench for testing performance of vec4 vs scalar coverage in generated shaders.
wangyix [Mon, 13 Jul 2015 20:09:17 +0000 (13:09 -0700)]
Added a glBench for testing performance of vec4 vs scalar coverage in generated shaders.

Added bench for timing vec4 vs scalar type for coverage in shaders

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/6104ced165f17eb2f765ace354d5895c0bc890c5

Review URL: https://codereview.chromium.org/1225383002

9 years agoGet rid of GrMurmur3Hash
joshualitt [Mon, 13 Jul 2015 19:49:13 +0000 (12:49 -0700)]
Get rid of GrMurmur3Hash

BUG=skia:

Review URL: https://codereview.chromium.org/1233933002

9 years agouse portable typeface
caryclark [Mon, 13 Jul 2015 19:08:45 +0000 (12:08 -0700)]
use portable typeface

TBR=reed@google.com

Review URL: https://codereview.chromium.org/1235873004

9 years agoAdd a GM that reproduces layout test failures with my new xfermode code.
mtklein [Mon, 13 Jul 2015 19:06:33 +0000 (12:06 -0700)]
Add a GM that reproduces layout test failures with my new xfermode code.

Inspired by https://storage.googleapis.com/chromium-layout-test-archives/linux_blink_rel/69169/layout-test-results/results.html

I think the root cause is overflow.

Also, adds tests for Sk16b::operator<().  It wasn't wrong, but it was suspect
(used in all three of these xfermode implementations) and so it's best to have
tests.

BUG=skia:

Review URL: https://codereview.chromium.org/1228393006

9 years agoRevert of Added a GLBench for testing performance of vec4 vs scalar for coverage...
scroggo [Mon, 13 Jul 2015 19:03:39 +0000 (12:03 -0700)]
Revert of Added a GLBench for testing performance of vec4 vs scalar for coverage in generated shaders. (patchset #4 id:40002 of https://codereview.chromium.org/1225383002/)

Reason for revert:
This is breaking a buildbot. See https://chromegw.corp.google.com/i/client.skia/builders/Test-Win8-MSVC-ShuttleB-GPU-HD4600-x86_64-Debug-ANGLE/builds/172/steps/nanobench/logs/stdio

We get the following error: ERROR::SHADER::COMPLIATION_FAILED: ERROR: 0:8: '0.f' : Floating-point suffix unsupported prior to GLSL ES 3.00
ERROR: 0:8: '0.f' : syntax error

Original issue's description:
> Added a glBench for testing performance of vec4 vs scalar coverage in generated shaders.
>
> Added bench for timing vec4 vs scalar type for coverage in shaders
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/6104ced165f17eb2f765ace354d5895c0bc890c5

TBR=joshualitt@google.com,tomhudson@google.com,wangyix@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review URL: https://codereview.chromium.org/1235533004