platform/upstream/libSkiaSharp.git
10 years agomore gms to ignore
joshualitt [Tue, 11 Nov 2014 17:28:30 +0000 (09:28 -0800)]
more gms to ignore

TBR=
BUG=skia:

Review URL: https://codereview.chromium.org/691143006

10 years agoremove separate color from coverage
joshualitt [Tue, 11 Nov 2014 16:51:30 +0000 (08:51 -0800)]
remove separate color from coverage

BUG=skia:

Review URL: https://codereview.chromium.org/699023003

10 years agoRemove bogus assert
bsalomon [Tue, 11 Nov 2014 16:29:01 +0000 (08:29 -0800)]
Remove bogus assert

content key can be set after pending io scheduled (e.g. when scaling up a npot texture to pot).

BUG=skia:2889
TBR=robertphillips@google.com

Review URL: https://codereview.chromium.org/693843004

10 years agoRebaseline problem (Galaxy S4) multipicturedraw GMs
robertphillips [Tue, 11 Nov 2014 16:17:39 +0000 (08:17 -0800)]
Rebaseline problem (Galaxy S4) multipicturedraw GMs

This CL rebaselines the buggy multipicturedraw GMs related to:

Always enable layer hoisting in Skia (https://codereview.chromium.org/698493002/)

TBR=jcgregorio@google.com

BUG=skia:3116

Review URL: https://codereview.chromium.org/715823002

10 years agoAdd support for rebaselining from trybots.
bungeman [Tue, 11 Nov 2014 16:11:55 +0000 (08:11 -0800)]
Add support for rebaselining from trybots.

This adds support to rebaseline_server to fetch actual results from
tryjobs run on rietveld issues.

Review URL: https://codereview.chromium.org/688353003

10 years agoRebaseline multipicturedraw GMs
robertphillips [Tue, 11 Nov 2014 15:55:45 +0000 (07:55 -0800)]
Rebaseline multipicturedraw GMs

This CL rebaselines the multipicturedraw GMs as a follow up to:

Always enable layer hoisting in Skia (https://codereview.chromium.org/698493002/)

TBR=jcgregorio@google.com

Review URL: https://codereview.chromium.org/714153002

10 years agoupdate dox
reed [Tue, 11 Nov 2014 15:54:10 +0000 (07:54 -0800)]
update dox

BUG=skia:
TBR=djsollen@google.com

Review URL: https://codereview.chromium.org/697363004

10 years agorebase from early clip
joshualitt [Tue, 11 Nov 2014 15:40:15 +0000 (07:40 -0800)]
rebase from early clip

TBR=
BUG=skia:

Review URL: https://codereview.chromium.org/707413003

10 years agoRemove cached OptDrawState from GrDrawState.
egdaniel [Tue, 11 Nov 2014 15:32:45 +0000 (07:32 -0800)]
Remove cached OptDrawState from GrDrawState.

Since we now snap and create the ODS one time in the inorder draw buffer,
there is no need for us to keep a cached version of it around.

BUG=skia:

Review URL: https://codereview.chromium.org/701123003

10 years agoguard call to CGColorSpaceCopyICCProfile in case we're building for iOS
reed [Tue, 11 Nov 2014 15:27:25 +0000 (07:27 -0800)]
guard call to CGColorSpaceCopyICCProfile in case we're building for iOS

BUG=skia:

Review URL: https://codereview.chromium.org/712363003

10 years agoRemove all knowledge of resource keys from the legacy cache.
bsalomon [Tue, 11 Nov 2014 15:27:16 +0000 (07:27 -0800)]
Remove all knowledge of resource keys from the legacy cache.

BUG=skia:2889

Review URL: https://codereview.chromium.org/705413002

10 years agoReenable deprecation warnings on Android framework.
scroggo [Tue, 11 Nov 2014 15:04:32 +0000 (07:04 -0800)]
Reenable deprecation warnings on Android framework.

Most recently introduced here: https://codereview.chromium.org/304333002

It appears these are not currently needed (I just disabled it and
successfully built).

Since we appear to be out of the transitional period, I propose we
remove them so we can know if we deprecate something Android uses
(and fix the call site).

Review URL: https://codereview.chromium.org/696063006

10 years agoMove from libstlport to libc++ for Android framework builds
djsollen [Tue, 11 Nov 2014 15:02:03 +0000 (07:02 -0800)]
Move from libstlport to libc++ for Android framework builds

Review URL: https://codereview.chromium.org/714113002

10 years agoAvoid warning in nanobench related to loop count with nvprmsaa4
kkinnunen [Tue, 11 Nov 2014 14:57:07 +0000 (06:57 -0800)]
Avoid warning in nanobench related to loop count with nvprmsaa4

The tests path_hairline_{small,big}_AA_conic were calling the test
function with NVPR. This caused a warning in nanobench.

The here removed hunk comes from commit referring to skia:2042 ("Enable
NVPR by default"). This is a workaround for a bug. The bug is fixed by
the commit referring to skia:2078 ("Logan bot fails NVPR assertion in
bench").

The proper fix is indeed make sure that path renderer chain ends up
trying software path renderer, if the path contains conics and is a
hairline.

The removed hunk refers also to skia:2033 ("Figure out what is happening
with conic path segments in NVPR"). The above solution is correct also in case
NVPR would support conics, as NVPR would not still support hairlines.

BUG=skia:2078

Review URL: https://codereview.chromium.org/685213005

10 years agostart again on Sk4x with portable version and unit tests
mtklein [Tue, 11 Nov 2014 14:16:42 +0000 (06:16 -0800)]
start again on Sk4x with portable version and unit tests
work on tests

CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu13.10-GCE-NoGPU-x86_64-Debug-ASAN-Trybot,Test-Ubuntu12-ShuttleA-GTX660-x86-Debug-Trybot,Test-Win7-ShuttleA-HD2000-x86_64-Debug-Trybot,Test-Win7-ShuttleA-HD2000-x86-Debug-Trybot
BUG=skia:

Review URL: https://codereview.chromium.org/704923003

10 years agoinitial checkin for experimenting
reed [Tue, 11 Nov 2014 12:56:05 +0000 (04:56 -0800)]
initial checkin for experimenting

BUG=skia:

Review URL: https://codereview.chromium.org/696063005

10 years agoChange where layer hoisting data is gathered
robertphillips [Tue, 11 Nov 2014 12:54:49 +0000 (04:54 -0800)]
Change where layer hoisting data is gathered

This CL:
1) removes the EXPERIMENTAL_optimize on SkCanvas & SkDevice
2) moves the saveLayer gathering step to endRecording
3) Replaces GPUOptimize with SkRecordComputeLayers
4) Update bench_pictures & render_pictures to provide the new flag

#2 also necessitated moving the BBH computation (and record optimization) out of SkPicture's ctor (and into endRecording)

Review URL: https://codereview.chromium.org/718443002

10 years agoUpdate SKP version
skia.buildbots [Tue, 11 Nov 2014 07:03:28 +0000 (23:03 -0800)]
Update SKP version

Automatic commit by the RecreateSKPs bot.

TBR=

Review URL: https://codereview.chromium.org/711253003

10 years agoSnap optdrawstate in inorder draw buffer and pass into gpu
joshualitt [Tue, 11 Nov 2014 00:03:14 +0000 (16:03 -0800)]
Snap optdrawstate in inorder draw buffer and pass into gpu

BUG=skia:

Review URL: https://codereview.chromium.org/709133003

10 years agotests/Time.cpp: more verbose errors
halcanary [Mon, 10 Nov 2014 22:39:24 +0000 (14:39 -0800)]
tests/Time.cpp: more verbose errors

Review URL: https://codereview.chromium.org/711143002

10 years agoSimplify canvas calls in SkDocument_PDF
halcanary [Mon, 10 Nov 2014 22:22:14 +0000 (14:22 -0800)]
Simplify canvas calls in SkDocument_PDF

As suggested by reed@ in http://crrev.com/711133002

TBR=reed@google.com

Review URL: https://codereview.chromium.org/707533003

10 years agoRemove outdated comment in SkImageDecoder.h
scroggo [Mon, 10 Nov 2014 22:18:24 +0000 (14:18 -0800)]
Remove outdated comment in SkImageDecoder.h

The comment was no longer relevant with
https://codereview.chromium.org/483593002

Review URL: https://codereview.chromium.org/715673002

10 years agoSkDocument_PDF does somthing with the trimBox parameter.
halcanary [Mon, 10 Nov 2014 22:01:57 +0000 (14:01 -0800)]
SkDocument_PDF does somthing with the trimBox parameter.

Fixes change introduced in http://crrev.com/645563002

This will allow SkDocument to be a drop-in repalcement to
SkPDFDocument/SkPDFDevice in Chromium.

Review URL: https://codereview.chromium.org/711133002

10 years agoRemove SK_FM_NEW_MATCH_FAMILY_STYLE_CHARACTER.
bungeman [Mon, 10 Nov 2014 21:29:33 +0000 (13:29 -0800)]
Remove SK_FM_NEW_MATCH_FAMILY_STYLE_CHARACTER.

This removes the old guarded code and enables the new api
introduced with "Update fontMgr to take list of bcp47 language tags."
c20386e3937d3d398ac9b35f9c7d997e972ade98 . Blink on Android is
already using the new code.

Review URL: https://codereview.chromium.org/705843004

10 years agoAdd tests (and fix!) for known bad ICO files.
scroggo [Mon, 10 Nov 2014 21:12:25 +0000 (13:12 -0800)]
Add tests (and fix!) for known bad ICO files.

We previously saw crashes decoding bad ICO files. Add tests for
known bad files.

While testing, I learned that one of them still crashes. Check for
large offset and size separately to fix the crash.

BUG=skia:2878

Review URL: https://codereview.chromium.org/712123002

10 years agoRe-enable no-longer-flaky GMs
jvanverth [Mon, 10 Nov 2014 20:12:55 +0000 (12:12 -0800)]
Re-enable no-longer-flaky GMs

TBR=jcgregorio@google.com

Review URL: https://codereview.chromium.org/684413004

10 years agoMove GrResourceKey static functions from .h to .cpp to fix shared lib builds
bsalomon [Mon, 10 Nov 2014 19:59:06 +0000 (11:59 -0800)]
Move GrResourceKey static functions from .h to .cpp to fix shared lib builds
TBR=robertphillips@google.com

Review URL: https://codereview.chromium.org/718493002

10 years agoUse GrResourceCache2 to service content key lookups
bsalomon [Mon, 10 Nov 2014 18:19:06 +0000 (10:19 -0800)]
Use GrResourceCache2 to service content key lookups

BUG=skia:2889

Review URL: https://codereview.chromium.org/707493002

10 years agoFix SkTime::GetDateTime Posix implementation, add unit test.
halcanary [Mon, 10 Nov 2014 17:20:22 +0000 (09:20 -0800)]
Fix SkTime::GetDateTime Posix implementation, add unit test.

The year was off by 1900

Review URL: https://codereview.chromium.org/714633002

10 years agoflag imageinfo as srgb
reed [Mon, 10 Nov 2014 16:57:21 +0000 (08:57 -0800)]
flag imageinfo as srgb

intended uses:
- flag a SkSurface as sRGB (only supported by Ganesh for now)
- flag images (e.g. png or jpeg) as sRGB if the codec tells us that

wins:
- faster gamma-correct text (esp. w/ distance-fields) when we can use sRGB for text
- better color fidelity when the screen really is sRGB

Review URL: https://codereview.chromium.org/676883003

10 years agoSmall refactoring of layer discovery code
robertphillips [Mon, 10 Nov 2014 16:48:29 +0000 (08:48 -0800)]
Small refactoring of layer discovery code

This CL removes CollectLayers' reliance on having the top most picture (by removing the unused fPictureID member). This then allows making CollectLayers' API closer to that of SkRecordFillBounds in order to facilitate using them interchangeably.

Review URL: https://codereview.chromium.org/714533002

10 years agoUpdate some flaky GMs to see if they stick.
jvanverth [Mon, 10 Nov 2014 16:27:31 +0000 (08:27 -0800)]
Update some flaky GMs to see if they stick.

TBR=jcgregorio@google.com

Review URL: https://codereview.chromium.org/713893002

10 years agoAddress MSAA rendering in layer hoisting
robertphillips [Mon, 10 Nov 2014 16:10:42 +0000 (08:10 -0800)]
Address MSAA rendering in layer hoisting

This became relevant whilst attempting to rebaseline the multipicturedraw GMs after turning on layer hoisting inside Skia.

Review URL: https://codereview.chromium.org/709943003

10 years agoRebase verttext once more
egdaniel [Mon, 10 Nov 2014 16:07:37 +0000 (08:07 -0800)]
Rebase verttext once more

TBR=jvanverth@google.com

BUG=skia:

Review URL: https://codereview.chromium.org/717553002

10 years agoWith https://chromium.googlesource.com/skia/+/43b8b36b20ae00e2d78421c4cda1f3f922983a2...
bungeman [Mon, 10 Nov 2014 15:48:55 +0000 (07:48 -0800)]
With https://chromium.googlesource.com/skia/+/43b8b36b20ae00e2d78421c4cda1f3f922983a20 blink_perf.layout regressed. It appears that the typeface cache is being missed. This is a partial revert of the Skia change which reverts the smallest amount of code to restore performance.

BUG=chromium:425566

Committed: https://skia.googlesource.com/skia/+/47eda52e8fb88f0725b401762a3e0bddc98e9fb3

Review URL: https://codereview.chromium.org/693213002

10 years agoUpdate SKP version
skia.buildbots [Mon, 10 Nov 2014 06:41:22 +0000 (22:41 -0800)]
Update SKP version

Automatic commit by the RecreateSKPs bot.

TBR=

Review URL: https://codereview.chromium.org/717463002

10 years agoRevert of Revert of SkDrawCommand scrubbing (patchset #1 id:1 of https://codereview...
fmalita [Sun, 9 Nov 2014 00:18:56 +0000 (16:18 -0800)]
Revert of Revert of SkDrawCommand scrubbing (patchset #1 id:1 of https://codereview.chromium.org/712683002/)

Reason for revert:
Chromium clients updated, this should be canary-proof now.

Original issue's description:
> Revert of SkDrawCommand scrubbing (patchset #2 id:20001 of https://codereview.chromium.org/706363002/)
>
> Reason for revert:
> Canary borkage.
>
> Original issue's description:
> > SkDrawCommand scrubbing
> >
> > Remove unused ctor, constify, etc.
> >
> > R=robertphillips@google.com
> >
> > Committed: https://skia.googlesource.com/skia/+/1931ec5b5dac68f1e452af0c65161bdce35b2dec
>
> TBR=robertphillips@google.com
> NOTREECHECKS=true
> NOTRY=true
>
> Committed: https://skia.googlesource.com/skia/+/9063ddd511f8e0d8c7454950cbc56e273678bf60

TBR=robertphillips@google.com
NOTREECHECKS=true
NOTRY=true

Review URL: https://codereview.chromium.org/708333002

10 years agoDefault geometry processor
joshualitt [Sat, 8 Nov 2014 17:24:25 +0000 (09:24 -0800)]
Default geometry processor

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/ff343074b2a3fdaa5f120600e28717e366bceadd

Review URL: https://codereview.chromium.org/678953002

10 years agorebasing some gms
joshualitt [Sat, 8 Nov 2014 15:09:38 +0000 (07:09 -0800)]
rebasing some gms

TBR=
BUG=skia:

Review URL: https://codereview.chromium.org/713633002

10 years agoReignore some tests that use distance fields
jvanverth [Fri, 7 Nov 2014 22:57:13 +0000 (14:57 -0800)]
Reignore some tests that use distance fields

NOTREECHECKS=true
NOTRY=true

Review URL: https://codereview.chromium.org/708173002

10 years agoRebase for win7 x86 gpu
egdaniel [Fri, 7 Nov 2014 22:22:57 +0000 (14:22 -0800)]
Rebase for win7 x86 gpu

TBR=bsalomon@google.com

BUG=skia:

Review URL: https://codereview.chromium.org/708153002

10 years agosimple cl to cleanup friends in GrGpu
joshualitt [Fri, 7 Nov 2014 22:11:45 +0000 (14:11 -0800)]
simple cl to cleanup friends in GrGpu

BUG=skia:

Review URL: https://codereview.chromium.org/709003006

10 years agoMake gpudft non-default config for gm
jvanverth [Fri, 7 Nov 2014 21:49:49 +0000 (13:49 -0800)]
Make gpudft non-default config for gm

Review URL: https://codereview.chromium.org/692153003

10 years agoFinish rebase from df and add baseline images for n10 debug
egdaniel [Fri, 7 Nov 2014 21:19:22 +0000 (13:19 -0800)]
Finish rebase from df and add baseline images for n10 debug

TBR=jvanverth@google.com

BUG=skia:

Review URL: https://codereview.chromium.org/705353003

10 years agocleaning up geometry handling in gpu
joshualitt [Fri, 7 Nov 2014 20:58:46 +0000 (12:58 -0800)]
cleaning up geometry handling in gpu

BUG=skia:

Review URL: https://codereview.chromium.org/707953004

10 years agoRebaseline s4 gms
egdaniel [Fri, 7 Nov 2014 20:56:35 +0000 (12:56 -0800)]
Rebaseline s4  gms

TBR=bsalomon@google.com

BUG=skia:

Review URL: https://codereview.chromium.org/712693002

10 years agoUpdate GrTRecorder to preserve memory allocations
cdalton [Fri, 7 Nov 2014 20:51:18 +0000 (12:51 -0800)]
Update GrTRecorder to preserve memory allocations

Rather than freeing almost all of its memory on calls to reset(), this
change updates GrTRecorder so it keeps around enough to satisfy the
storage requirements from last time, plus up to ~50% growth. This is
based on the assumption that subsequent draw calls require roughly the
same amount of memory.

BUG=skia:

Review URL: https://codereview.chromium.org/684203003

10 years agoRevert of SkDrawCommand scrubbing (patchset #2 id:20001 of https://codereview.chromiu...
fmalita [Fri, 7 Nov 2014 20:26:46 +0000 (12:26 -0800)]
Revert of SkDrawCommand scrubbing (patchset #2 id:20001 of https://codereview.chromium.org/706363002/)

Reason for revert:
Canary borkage.

Original issue's description:
> SkDrawCommand scrubbing
>
> Remove unused ctor, constify, etc.
>
> R=robertphillips@google.com
>
> Committed: https://skia.googlesource.com/skia/+/1931ec5b5dac68f1e452af0c65161bdce35b2dec

TBR=robertphillips@google.com
NOTREECHECKS=true
NOTRY=true

Review URL: https://codereview.chromium.org/712683002

10 years agoSkDrawCommand scrubbing
fmalita [Fri, 7 Nov 2014 20:09:20 +0000 (12:09 -0800)]
SkDrawCommand scrubbing

Remove unused ctor, constify, etc.

R=robertphillips@google.com

Review URL: https://codereview.chromium.org/706363002

10 years agoRevert of Default geometry processor (patchset #9 id:160001 of https://codereview...
joshualitt [Fri, 7 Nov 2014 20:08:03 +0000 (12:08 -0800)]
Revert of Default geometry processor (patchset #9 id:160001 of https://codereview.chromium.org/678953002/)

Reason for revert:
breaks nexus 5

Original issue's description:
> Default geometry processor
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/ff343074b2a3fdaa5f120600e28717e366bceadd

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

Review URL: https://codereview.chromium.org/691313003

10 years agoDefault geometry processor
joshualitt [Fri, 7 Nov 2014 19:47:10 +0000 (11:47 -0800)]
Default geometry processor

BUG=skia:

Review URL: https://codereview.chromium.org/678953002

10 years agoadd patch and clicktracking to lua
reed [Fri, 7 Nov 2014 18:23:55 +0000 (10:23 -0800)]
add patch and clicktracking to lua

BUG=skia:

Review URL: https://codereview.chromium.org/712613002

10 years agoOne more attempt at rebaselining
jvanverth [Fri, 7 Nov 2014 18:17:36 +0000 (10:17 -0800)]
One more attempt at rebaselining

TBR=egdaniel@google.com
NOTREECHECKS=true

Review URL: https://codereview.chromium.org/708903006

10 years agoMesa ES 3.0 requires sized internal formats
derekf [Fri, 7 Nov 2014 17:39:26 +0000 (09:39 -0800)]
Mesa ES 3.0 requires sized internal formats

Mesa's ES 3.0 implementation requires GL_R8 as an internal format instead
of GL_RED

Review URL: https://codereview.chromium.org/705183002

10 years agoMissed some expectations
jvanverth [Fri, 7 Nov 2014 16:50:16 +0000 (08:50 -0800)]
Missed some expectations

TBR=egdaniel@google.com

Review URL: https://codereview.chromium.org/705293002

10 years agoRefactor DrawTarget and GPU to be independent
joshualitt [Fri, 7 Nov 2014 16:48:51 +0000 (08:48 -0800)]
Refactor DrawTarget and GPU to be independent

BUG=skia:

Review URL: https://codereview.chromium.org/705593002

10 years agoRebaseline for https://codereview.chromium.org/703463002/.
jvanverth [Fri, 7 Nov 2014 16:16:49 +0000 (08:16 -0800)]
Rebaseline for https://codereview.chromium.org/703463002/.

TBR=egdaniel@google.com

Review URL: https://codereview.chromium.org/709043002

10 years agodocument the SkStrAppend methods
reed [Fri, 7 Nov 2014 15:37:33 +0000 (07:37 -0800)]
document the SkStrAppend methods

BUG=skia:

Review URL: https://codereview.chromium.org/709063002

10 years agoGet gpudft support working in dm, gm, nanobench and bench_pictures
jvanverth [Fri, 7 Nov 2014 15:12:46 +0000 (07:12 -0800)]
Get gpudft support working in dm, gm, nanobench and bench_pictures

Adds a new config to test distance field text.
Clean up some flags and #defines to read "distance field text",
not "distance field fonts" to be consistent with Chromium

NOTREECHECKS=true

Committed: https://skia.googlesource.com/skia/+/06ba179838ba4fe187cf290750aeeb4a02a2960b

Review URL: https://codereview.chromium.org/699453005

10 years agoUpdate SKP version
skia.buildbots [Fri, 7 Nov 2014 14:12:39 +0000 (06:12 -0800)]
Update SKP version

Automatic commit by the RecreateSKPs bot.

TBR=

Review URL: https://codereview.chromium.org/693583003

10 years agoCleanup: Use SkAutoGraphics in gm and tests.
tfarina [Fri, 7 Nov 2014 14:12:30 +0000 (06:12 -0800)]
Cleanup: Use SkAutoGraphics in gm and tests.

BUG=None
R=bsalomon@google.com

Review URL: https://codereview.chromium.org/704413002

10 years agoWhen running DM, write test failures to json.
scroggo [Fri, 7 Nov 2014 14:07:32 +0000 (06:07 -0800)]
When running DM, write test failures to json.

Add skiatest::Failure to keep track of data about a test failure.

Reporter::reportFailed and ::onReportFailed now take Failure as a
parameter. This allows the implementation to treat the failure as it
wishes. Provide a helper to format the failure the same as prior to
the change.

Update the macros for calling reportFailed (REPORTER_ASSERT etc) to
create a Failure object.

Convert a direct call to reportFailed to the macro ERRORF.

Write Failures to Json.
Sample output when running dm on the dummy test crrev.com/705723004:
{
   "test_results" : {
      "failures" : [
         {
            "condition" : "0 > 3",
            "file_name" : "../../tests/DummyTest.cpp",
            "line_no" : 10,
            "message" : ""
         },
         {
            "condition" : "false",
            "file_name" : "../../tests/DummyTest.cpp",
            "line_no" : 4,
            "message" : ""
         },
         {
            "condition" : "1 == 3",
            "file_name" : "../../tests/DummyTest.cpp",
            "line_no" : 5,
            "message" : "I can too count!"
         },
         {
            "condition" : "",
            "file_name" : "../../tests/DummyTest.cpp",
            "line_no" : 6,
            "message" : "seven is 7"
         },
         {
            "condition" : "1 == 3",
            "file_name" : "../../tests/DummyTest.cpp",
            "line_no" : 14,
            "message" : "I can too count!"
         }
      ]
   }
}

Report all of the failures from one test.
Previously, if one test had multiple failures, only one was reportered.
e.g:

Failures:
  test Dummy: ../../tests/DummyTest.cpp:6   seven is 7
  test Dummy2: ../../tests/DummyTest.cpp:10 0 > 3
  test Dummy3: ../../tests/DummyTest.cpp:14 I can too count!: 1 == 3
3 failures.

Now, we get all the messages:

Failures:
  test Dummy: ../../tests/DummyTest.cpp:4   false
        ../../tests/DummyTest.cpp:5 I can too count!: 1 == 3
        ../../tests/DummyTest.cpp:6 seven is 7
  test Dummy2: ../../tests/DummyTest.cpp:10 0 > 3
  test Dummy3: ../../tests/DummyTest.cpp:14 I can too count!: 1 == 3
3 failures.

(Note that we still state "3 failures" because 3 DM::Tasks failed.)

BUG=skia:3082
BUG=skia:2454

Review URL: https://codereview.chromium.org/694703005

10 years agofor X in {Dec,BigDec,Scalar}, SkWStream::write"X"AsText no longer mallocs
halcanary [Fri, 7 Nov 2014 14:07:23 +0000 (06:07 -0800)]
for X in {Dec,BigDec,Scalar}, SkWStream::write"X"AsText no longer mallocs

Review URL: https://codereview.chromium.org/706063002

10 years agoAdd LOCAL_PICKUP_FILES to dm and bench makefiles.
scroggo [Fri, 7 Nov 2014 14:02:20 +0000 (06:02 -0800)]
Add LOCAL_PICKUP_FILES to dm and bench makefiles.

Update the generator to include a line for LOCAL_PICKUP_FILES. When
generating makefiles for our tests, they will now have the following line:

LOCAL_PICKUP_FILES := \
    $(LOCAL_PATH)/../resources

This allows testing infrastructure to pick up resource files used by
our tests.

Update expectations files to test LOCAL_PICKUP_FILES.

BUG=skia:2454

Review URL: https://codereview.chromium.org/704393002

10 years agoRemove 1d glyph positions from nvpr text
cdalton [Fri, 7 Nov 2014 14:02:15 +0000 (06:02 -0800)]
Remove 1d glyph positions from nvpr text

Removes the case for x-only glyph positions from nvpr text, opting to
always send 2d glyph positions instead. The 1d glyph positions saved a
bit on memory bandwidth, but ended up a net loss because they required
more updates to the view matrix. Now we can draw an entire paragraph
without touching the GL state, whereas before we would have to update
the view matrix at every new line.

BUG=skia:

Review URL: https://codereview.chromium.org/700283002

10 years agoRevert of Get gpudft support working in dm, gm, nanobench and bench_pictures (patchse...
jvanverth [Thu, 6 Nov 2014 21:52:45 +0000 (13:52 -0800)]
Revert of Get gpudft support working in dm, gm, nanobench and bench_pictures (patchset #2 id:20001 of https://codereview.chromium.org/699453005/)

Reason for revert:
Not compiling in ANGLE build

Original issue's description:
> Get gpudft support working in dm, gm, nanobench and bench_pictures
>
> Adds a new config to test distance field text.
> Clean up some flags and #defines to read "distance field text",
> not "distance field fonts" to be consistent with Chromium
>
> NOTREECHECKS=true
>
> Committed: https://skia.googlesource.com/skia/+/06ba179838ba4fe187cf290750aeeb4a02a2960b

TBR=bsalomon@google.com,mtklein@google.com,reed@google.com
NOTREECHECKS=true
NOTRY=true

Review URL: https://codereview.chromium.org/707723005

10 years agoGet gpudft support working in dm, gm, nanobench and bench_pictures
jvanverth [Thu, 6 Nov 2014 21:38:52 +0000 (13:38 -0800)]
Get gpudft support working in dm, gm, nanobench and bench_pictures

Adds a new config to test distance field text.
Clean up some flags and #defines to read "distance field text",
not "distance field fonts" to be consistent with Chromium

NOTREECHECKS=true

Review URL: https://codereview.chromium.org/699453005

10 years agoClip in grdrawtarget
joshualitt [Thu, 6 Nov 2014 20:57:13 +0000 (12:57 -0800)]
Clip in grdrawtarget

NOTREECHECKS=true
BUG=skia:

Review URL: https://codereview.chromium.org/685883003

10 years agoRevert of Revert of Turn on NVPR 4x MSAA by default when supported in DM and nanobenc...
mtklein [Thu, 6 Nov 2014 16:06:39 +0000 (08:06 -0800)]
Revert of Revert of Turn on NVPR 4x MSAA by default when supported in DM and nanobench. (patchset #1 id:1 of https://codereview.chromium.org/693933004/)

Reason for revert:
Try again with loops overflow fix landed.

Original issue's description:
> Revert of Turn on NVPR 4x MSAA by default when supported in DM and nanobench. (patchset #2 id:20001 of https://codereview.chromium.org/704563003/)
>
> Reason for revert:
> Timing out on nvpr when drawing conics. Fix will take some thought so reverting for now.
>
> Original issue's description:
> > Turn on NVPR 4x MSAA by default when supported in DM and nanobench.
> >
> > This brings DM and nanobench's default configs in line with GM's.
> >
> > BUG=skia:
> >
> > Committed: https://skia.googlesource.com/skia/+/31f88675718966bbb7f09718b40de10c7e214739
>
> TBR=bsalomon@google.com,mtklein@google.com,mtklein@chromium.org
> NOTREECHECKS=true
> NOTRY=true
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/d5cd4ee5b81b51e43be7ed13f3c0f0f6c1b3fe14

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

Review URL: https://codereview.chromium.org/685923003

10 years agoDetect loops overflow for gpu benches.
mtklein [Thu, 6 Nov 2014 16:04:34 +0000 (08:04 -0800)]
Detect loops overflow for gpu benches.

NOTREECHECKS=true

BUG=skia:

Review URL: https://codereview.chromium.org/709473002

10 years agoSend coverage VA as floats instead of bytes
egdaniel [Thu, 6 Nov 2014 16:00:48 +0000 (08:00 -0800)]
Send coverage VA as floats instead of bytes
NOTREECHECKS=True

BUG=skia:

Review URL: https://codereview.chromium.org/700943005

10 years agoUpdate SKP version
skia.buildbots [Thu, 6 Nov 2014 07:19:12 +0000 (23:19 -0800)]
Update SKP version

Automatic commit by the RecreateSKPs bot.

TBR=

Review URL: https://codereview.chromium.org/701363002

10 years agoAdd mock context and use in ResourceCacheTest.
bsalomon [Wed, 5 Nov 2014 22:47:41 +0000 (14:47 -0800)]
Add mock context and use in ResourceCacheTest.

BUG=skia:2889

Committed: https://skia.googlesource.com/skia/+/820dd6c335411aad889c1d7e8a857642ecd87e30

Review URL: https://codereview.chromium.org/702083003

10 years agoRevert of Turn on NVPR 4x MSAA by default when supported in DM and nanobench. (patchs...
egdaniel [Wed, 5 Nov 2014 22:21:33 +0000 (14:21 -0800)]
Revert of Turn on NVPR 4x MSAA by default when supported in DM and nanobench. (patchset #2 id:20001 of https://codereview.chromium.org/704563003/)

Reason for revert:
Timing out on nvpr when drawing conics. Fix will take some thought so reverting for now.

Original issue's description:
> Turn on NVPR 4x MSAA by default when supported in DM and nanobench.
>
> This brings DM and nanobench's default configs in line with GM's.
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/31f88675718966bbb7f09718b40de10c7e214739

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

Review URL: https://codereview.chromium.org/693933004

10 years agoRevert of Add mock context and use in ResourceCacheTest. (patchset #4 id:60001 of...
bsalomon [Wed, 5 Nov 2014 20:30:32 +0000 (12:30 -0800)]
Revert of Add mock context and use in ResourceCacheTest. (patchset #4 id:60001 of https://codereview.chromium.org/702083003/)

Reason for revert:
Breaking tests

Original issue's description:
> Add mock context and use in ResourceCacheTest.
>
> BUG=skia:2889
>
> Committed: https://skia.googlesource.com/skia/+/820dd6c335411aad889c1d7e8a857642ecd87e30

TBR=robertphillips@google.com
NOTREECHECKS=true
NOTRY=true
BUG=skia:2889

Review URL: https://codereview.chromium.org/704563004

10 years agoTurn on NVPR 4x MSAA by default when supported in DM and nanobench.
mtklein [Wed, 5 Nov 2014 20:28:26 +0000 (12:28 -0800)]
Turn on NVPR 4x MSAA by default when supported in DM and nanobench.

This brings DM and nanobench's default configs in line with GM's.

BUG=skia:

Review URL: https://codereview.chromium.org/704563003

10 years agoAdd mock context and use in ResourceCacheTest.
bsalomon [Wed, 5 Nov 2014 20:09:45 +0000 (12:09 -0800)]
Add mock context and use in ResourceCacheTest.

BUG=skia:2889

Review URL: https://codereview.chromium.org/702083003

10 years agoS4 Rebaselines
bsalomon [Wed, 5 Nov 2014 19:59:25 +0000 (11:59 -0800)]
S4 Rebaselines

TBR=

Review URL: https://codereview.chromium.org/706653003

10 years agoAvoid dec = -dec overflow when appending most negative signed integers.
mtklein [Wed, 5 Nov 2014 19:35:21 +0000 (11:35 -0800)]
Avoid dec = -dec overflow when appending most negative signed integers.

BUG=skia:3096

Review URL: https://codereview.chromium.org/700953003

10 years agoAdd new baselines for remaining dashing5 GM images
robertphillips [Wed, 5 Nov 2014 17:00:22 +0000 (09:00 -0800)]
Add new baselines for remaining dashing5 GM images

This are the baselines for the problem children:
The N5 & N10 gpu images related to Issue 3097
The perspective Windows images

TBR=egdaniel@google.com

Review URL: https://codereview.chromium.org/702143002

10 years agoRemove gm from ignore expectation file
egdaniel [Wed, 5 Nov 2014 16:57:16 +0000 (08:57 -0800)]
Remove gm from ignore expectation file

No gm's actually changed from my recent change to dashing gm. This was the desired
outcome since the added calls were lines of length zero

TBR=robertphillips@google.com

BUG=skia:

Review URL: https://codereview.chromium.org/706633002

10 years agoAdd baselines for the new dashing5 GM
robertphillips [Wed, 5 Nov 2014 16:25:40 +0000 (08:25 -0800)]
Add baselines for the new dashing5 GM

These are all the correct images

TBR=egdaniel@google.com

Review URL: https://codereview.chromium.org/707513002

10 years agoAdd unit test for SkDashPathEffect::asPoints' culling
robertphillips [Wed, 5 Nov 2014 16:06:40 +0000 (08:06 -0800)]
Add unit test for SkDashPathEffect::asPoints' culling

This is a follow on to (Crop the fast path dashed lines to the cull rect - https://codereview.chromium.org/699623003/)

Review URL: https://codereview.chromium.org/701133002

10 years agoWhitelist intentionally racy TRACE_EVENT reads and writes.
mtklein [Wed, 5 Nov 2014 16:03:27 +0000 (08:03 -0800)]
Whitelist intentionally racy TRACE_EVENT reads and writes.

Chrome's tracing framework appears to be intentionally racy on its
quick-reject checks, trading some data loss for better performance
when disabled.  People will never notice the data loss, but TSAN does.
Let's assuage TSAN with some annotations.

The 'volatile' val in SK_ANNOTATE_UNPROTECTED_WRITE was making this
not compile, but that volatile doesn't really make sense there: the value we're
writing is not what we care about, it's the destination.

CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu13.10-GCE-NoGPU-x86_64-Release-TSAN-Trybot

No API changes.
TBR=reed

BUG=skia:

Review URL: https://codereview.chromium.org/702883002

10 years agoRevert of Bisect Mac performance regression. (patchset #6 id:100001 of https://codere...
egdaniel [Wed, 5 Nov 2014 15:40:07 +0000 (07:40 -0800)]
Revert of Bisect Mac performance regression. (patchset #6 id:100001 of https://codereview.chromium.org/693213002/)

Reason for revert:
Block deps roll by breaking webkit test hit-test-counts on mac.

Original issue's description:
> With https://chromium.googlesource.com/skia/+/43b8b36b20ae00e2d78421c4cda1f3f922983a20 blink_perf.layout regressed. It appears that the typeface cache is being missed. This is a partial revert of the Skia change which reverts the smallest amount of code to restore performance.
>
> BUG=chromium:425566
>
> Committed: https://skia.googlesource.com/skia/+/47eda52e8fb88f0725b401762a3e0bddc98e9fb3

TBR=reed@google.com,bungeman@google.com
NOTREECHECKS=true
NOTRY=true
BUG=chromium:425566

Review URL: https://codereview.chromium.org/683753003

10 years agoCleanup: Remove unused InstallNewHandler() function from SkGraphics.
tfarina [Wed, 5 Nov 2014 15:20:51 +0000 (07:20 -0800)]
Cleanup: Remove unused InstallNewHandler() function from SkGraphics.

BUG=None
TEST=None
R=bsalomon@google.com

Review URL: https://codereview.chromium.org/701813006

10 years agoWorkaround for PowerVR clear issue.
bsalomon [Wed, 5 Nov 2014 15:05:34 +0000 (07:05 -0800)]
Workaround for PowerVR clear issue.

BUG=skia:

Review URL: https://codereview.chromium.org/701573002

10 years agoAdd suppression to valgrind.supp for fun:FcConfigEvaluate.
bungeman [Wed, 5 Nov 2014 14:44:11 +0000 (06:44 -0800)]
Add suppression to valgrind.supp for fun:FcConfigEvaluate.

Any allocation under FcConfigValues will be 'lost'.
Update the valgrind suppression to reflect this.

Review URL: https://codereview.chromium.org/703893003

10 years agoBug fix for cull_line in SkDashPathEffect.
egdaniel [Wed, 5 Nov 2014 13:02:27 +0000 (05:02 -0800)]
Bug fix for cull_line in SkDashPathEffect.

In cull_line we must also check if both points are the same. Otherwise we
fail the assert in the else "SkASSERT(dy && !dx)".

This is currently blocking the roll as it fails a webkit test.

BUG=skia:

Review URL: https://codereview.chromium.org/703783002

10 years agoCrop the fast path dashed lines to the cull rect
robertphillips [Tue, 4 Nov 2014 21:33:50 +0000 (13:33 -0800)]
Crop the fast path dashed lines to the cull rect

Without:

maxrss  loops   min     median  mean    max     stddev  samples         config  bench
  56M   1       13.3ms  13.6ms  13.6ms  14.2ms  2%      Ooooo.....      8888    GM_dashing5_bw
  56M   13      390us   417us   416us   459us   5%      ooooO..o.o      gpu     GM_dashing5_bw
  56M   1       13.4ms  13.9ms  14.1ms  15ms    3%      Oooo..ooOo      8888    GM_dashing5_aa
  56M   13      402us   421us   416us   425us   2%      Ooo.ooOOOO      gpu     GM_dashing5_aa

With:

  40M   1       1.53ms  1.54ms  1.54ms  1.55ms  0%      oo.O...o..      8888    GM_dashing5_bw
  40M   12      407us   412us   415us   445us   3%      ...Oo.....      gpu     GM_dashing5_bw
  40M   1       1.7ms   1.7ms   1.7ms   1.72ms  0%      o.O.......      8888    GM_dashing5_aa
  43M   13      405us   409us   409us   415us   1%      ooo.Ooo..o      gpu     GM_dashing5_aa

The GM images (including the new one) are the same with and without this CL.

BUG=428296

Review URL: https://codereview.chromium.org/699623003

10 years agoDFText: fix placement of glyphs when we fall back to paths
jvanverth [Tue, 4 Nov 2014 21:32:53 +0000 (13:32 -0800)]
DFText: fix placement of glyphs when we fall back to paths

Rendering a glyph with a path wants to place it at the (sx, sy) we get
as input to the method, but we add (dx, dy) for the clipRect check.
Hence, we need to subtract that out before we render the path.

Review URL: https://codereview.chromium.org/699283003

10 years agoupdate slides
reed [Tue, 4 Nov 2014 21:24:47 +0000 (13:24 -0800)]
update slides

BUG=skia:

Review URL: https://codereview.chromium.org/686853005

10 years agoRevert of Override SkCanvas::drawImage() in SkDeferredCanvas and SkGPipe (patchset...
piotaixr [Tue, 4 Nov 2014 21:21:41 +0000 (13:21 -0800)]
Revert of Override SkCanvas::drawImage() in SkDeferredCanvas and SkGPipe (patchset #8 id:140001 of https://codereview.chromium.org/613673005/)

Reason for revert:
The patch is breaking the telemetry blink build on some windows configurations

Original issue's description:
> Override SkCanvas::drawImage() in SkDeferredCanvas and SkGPipe
>
> Depend on https://codereview.chromium.org/663233002
>
> BUG=skia:2947
>
> Committed: https://skia.googlesource.com/skia/+/687732fe046c9cfec940f2d7f661cd97651d10a8
>
> Committed: https://skia.googlesource.com/skia/+/9bb7539a59ab15749fe26fecfec05330cffae684

BUG=skia:2947, 426708

Review URL: https://codereview.chromium.org/705633002

10 years agoAdditional baselines for 4b86bacb0428.
bungeman [Tue, 4 Nov 2014 20:28:51 +0000 (15:28 -0500)]
Additional baselines for 4b86bacb0428.

The Nexus9 lacked a trybot at the time.

10 years agoAdd tracing to the SkCanvas drawFoo() methods to find long draw ops.
danakj [Tue, 4 Nov 2014 19:48:43 +0000 (11:48 -0800)]
Add tracing to the SkCanvas drawFoo() methods to find long draw ops.

BUG=skia:3088

Review URL: https://codereview.chromium.org/702473004

10 years agoPictureRecordBench's benchmarks are no longer relevant with SkRecord.
mtklein [Tue, 4 Nov 2014 19:14:48 +0000 (11:14 -0800)]
PictureRecordBench's benchmarks are no longer relevant with SkRecord.

BUG=skia:

Review URL: https://codereview.chromium.org/698163004

10 years agoadd more typeface methods to lua
reed [Tue, 4 Nov 2014 18:58:42 +0000 (10:58 -0800)]
add more typeface methods to lua

BUG=skia:
NOTRY=True
TBR=

Review URL: https://codereview.chromium.org/697053004

10 years agoAllow all font weights on Android and respect overrides.
bungeman [Tue, 4 Nov 2014 18:54:31 +0000 (10:54 -0800)]
Allow all font weights on Android and respect overrides.

This removes the code which forces 400 and 700 weights only,
and also overrides the font weight with the configured weight.

Review URL: https://codereview.chromium.org/694533006

10 years agoWith https://chromium.googlesource.com/skia/+/43b8b36b20ae00e2d78421c4cda1f3f922983a2...
bungeman [Tue, 4 Nov 2014 18:27:35 +0000 (10:27 -0800)]
With https://chromium.googlesource.com/skia/+/43b8b36b20ae00e2d78421c4cda1f3f922983a20 blink_perf.layout regressed. It appears that the typeface cache is being missed. This is a partial revert of the Skia change which reverts the smallest amount of code to restore performance.

BUG=chromium:425566

Review URL: https://codereview.chromium.org/693213002

10 years agoFix PictureTest for new GPU veto
jvanverth [Tue, 4 Nov 2014 16:50:15 +0000 (08:50 -0800)]
Fix PictureTest for new GPU veto

NOTREECHECKS=true

Review URL: https://codereview.chromium.org/702733002