platform/upstream/libSkiaSharp.git
10 years agoadd SkImage::newSurface
reed [Fri, 21 Nov 2014 16:46:37 +0000 (08:46 -0800)]
add SkImage::newSurface

BUG=skia:

Review URL: https://codereview.chromium.org/741763002

10 years agosimplify uniqueID code in picture
reed [Fri, 21 Nov 2014 16:07:41 +0000 (08:07 -0800)]
simplify uniqueID code in picture

patch from issue 750003002 at patchset 1 (http://crrev.com/750003002#ps1)

BUG=skia:

Review URL: https://codereview.chromium.org/752573002

10 years agoIgnore a flaky msaa4 GM and some perma-broken S4 GMs
robertphillips [Fri, 21 Nov 2014 15:24:01 +0000 (07:24 -0800)]
Ignore a flaky msaa4 GM and some perma-broken S4 GMs

The Ubuntu12 msaa4 image seems to flake randomly
The S4 images are covered by skia:3116 (Galaxy S4 exhibits some saveLayer bugs (both normal and with hoisting))

TBR=humper@google.com

Review URL: https://codereview.chromium.org/748773003

10 years agoFix memory leak in nanobench
robertphillips [Fri, 21 Nov 2014 14:53:00 +0000 (06:53 -0800)]
Fix memory leak in nanobench

This is cleanup for (Add MultiPictureDraw to nanobench - https://codereview.chromium.org/731973005/)

Review URL: https://codereview.chromium.org/730343003

10 years agoSmall changes to get things building and running on Linux/ARM.
mtklein [Fri, 21 Nov 2014 14:48:43 +0000 (06:48 -0800)]
Small changes to get things building and running on Linux/ARM.

ninja -C out/Debug works, and I tested out/Debug/dm --nogpu (no X yet).

The only problems encountered were:
  - Like CrOS, I also needed to use the default float ABI (presumably hard).
  - The strings in GrBinHashKeyTest weren't necessarily 4-byte aligned.

BUG=skia:

Review URL: https://codereview.chromium.org/746043002

10 years agoAdd MultiPictureDraw to nanobench
robertphillips [Fri, 21 Nov 2014 14:19:36 +0000 (06:19 -0800)]
Add MultiPictureDraw to nanobench

I would like some guard against performance regressions on our side before turning layer hoisting on in Chromium.

TBR=bsalomon@google.com

Committed: https://skia.googlesource.com/skia/+/0ddad31012dabfc1267effc8071d37f7d606efbe

Review URL: https://codereview.chromium.org/731973005

10 years agoRevert of Add MultiPictureDraw to nanobench (patchset #7 id:120001 of https://coderev...
robertphillips [Fri, 21 Nov 2014 13:50:21 +0000 (05:50 -0800)]
Revert of Add MultiPictureDraw to nanobench (patchset #7 id:120001 of https://codereview.chromium.org/731973005/)

Reason for revert:
Needs more work

Original issue's description:
> Add MultiPictureDraw to nanobench
>
> I would like some guard against performance regressions on our side before turning layer hoisting on in Chromium.
>
> TBR=bsalomon@google.com
>
> Committed: https://skia.googlesource.com/skia/+/0ddad31012dabfc1267effc8071d37f7d606efbe

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

Review URL: https://codereview.chromium.org/750583002

10 years agoAdd MultiPictureDraw to nanobench
robertphillips [Fri, 21 Nov 2014 13:35:54 +0000 (05:35 -0800)]
Add MultiPictureDraw to nanobench

I would like some guard against performance regressions on our side before turning layer hoisting on in Chromium.

TBR=bsalomon@google.com

Review URL: https://codereview.chromium.org/731973005

10 years agoRevert of SkCanvas::clear() should call down to devices even when the clip is empty...
fmalita [Fri, 21 Nov 2014 04:44:04 +0000 (20:44 -0800)]
Revert of SkCanvas::clear() should call down to devices even when the clip is empty. (patchset #1 id:1 of https://codereview.chromium.org/735233003/)

Reason for revert:
Chrome unit_tests failures on Mac & Win, blocking the roll: http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_rel_ng/builds/5613

Original issue's description:
> SkCanvas::clear() should call down to devices even when the clip is empty.
>
> SkDrawIter had a lovely, completely unused, option for this.  Oversight?
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/378e70d8e561b72a904286ea231946f5c8345b9a

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

Review URL: https://codereview.chromium.org/748733002

10 years agoCleanup: Delete webtry.gyp
tfarina [Fri, 21 Nov 2014 02:20:06 +0000 (18:20 -0800)]
Cleanup: Delete webtry.gyp

Looks like this was left over when we moved it into buildbot.

Also remove a few other, now unnecessary, references to webtry.

BUG=None
R=humper@google.com

Review URL: https://codereview.chromium.org/726833004

10 years agorebasing a bunch of 0% android mismatches
joshualitt [Fri, 21 Nov 2014 00:51:50 +0000 (16:51 -0800)]
rebasing a bunch of 0% android mismatches

TBR=
BUG=skia:

Review URL: https://codereview.chromium.org/744383002

10 years agoAdding GeometryData object
joshualitt [Thu, 20 Nov 2014 22:50:39 +0000 (14:50 -0800)]
Adding GeometryData object

BUG=skia:

Review URL: https://codereview.chromium.org/745663002

10 years agoRebaseline textblobshader expectation for nvpr
cdalton [Thu, 20 Nov 2014 22:30:08 +0000 (14:30 -0800)]
Rebaseline textblobshader expectation for nvpr

Rebaseline after https://codereview.chromium.org/712223002. The
difference is a result of using different bounding boxes to cover
gradient fills, which caused a few pixels do differ by a value of one.

BUG=skia:

Review URL: https://codereview.chromium.org/746603002

10 years agoAdd SkNVRefCnt::getRefCnt()
Florin Malita [Thu, 20 Nov 2014 21:56:22 +0000 (16:56 -0500)]
Add SkNVRefCnt::getRefCnt()

Used in Blink unit tests.

TBR=mtklein@google.com,humper@google.com

Review URL: https://codereview.chromium.org/746693002

10 years agoDisable LCD text explicitly in SkPictureImageFilter::onFilterImage()
fmalita [Thu, 20 Nov 2014 18:44:58 +0000 (10:44 -0800)]
Disable LCD text explicitly in SkPictureImageFilter::onFilterImage()

As an intermediate fix for
https://code.google.com/p/skia/issues/detail?id=3142, we can use a
non-public SkCanvas constructor and force-disable LCD text.

BUG=skia:3142
R=reed@google.com,senorblanco@google.com

Review URL: https://codereview.chromium.org/725243004

10 years agoAdd SkNVRefCnt::deref().
mtklein [Thu, 20 Nov 2014 18:07:54 +0000 (10:07 -0800)]
Add SkNVRefCnt::deref().

TBR=reed@google.com

BUG=skia:

Review URL: https://codereview.chromium.org/744853002

10 years agoDon't use NULL GrOptDrawState to indicate that draw should be skipped.
bsalomon [Thu, 20 Nov 2014 17:56:11 +0000 (09:56 -0800)]
Don't use NULL GrOptDrawState to indicate that draw should be skipped.

Review URL: https://codereview.chromium.org/742853002

10 years agoAdd SkNVRefCnt, prune down SkPicture's size
mtklein [Thu, 20 Nov 2014 17:18:31 +0000 (09:18 -0800)]
Add SkNVRefCnt, prune down SkPicture's size

SkNVRefCnt is a variant of SkRefCnt that's Not Virtual, so weighs 4 bytes
instead of 8 or 16.  There's only benefit to doing this if the deriving class
does not otherwise need a vtable, e.g. SkPicture.

I've stripped out some cruft from SkPicture, rearranged fields to pack tightly,
and added compile asserts for the sizes of SkPicture, SkRecord, and
SkVarAlloc.

BUG=skia:3144

Review URL: https://codereview.chromium.org/741793002

10 years agoSkRecord: Pack SkRegion::Op and AA bit together.
mtklein [Thu, 20 Nov 2014 17:14:28 +0000 (09:14 -0800)]
SkRecord: Pack SkRegion::Op and AA bit together.

Saves 4 bytes per clip op.

BUG=skia:

Review URL: https://codereview.chromium.org/741093002

10 years agoSkCanvas::clear() should call down to devices even when the clip is empty.
mtklein [Thu, 20 Nov 2014 16:58:02 +0000 (08:58 -0800)]
SkCanvas::clear() should call down to devices even when the clip is empty.

SkDrawIter had a lovely, completely unused, option for this.  Oversight?

BUG=skia:

Review URL: https://codereview.chromium.org/735233003

10 years agoStop overloading internal_dispose in SkTextBlob and SkData
bsalomon [Thu, 20 Nov 2014 16:02:46 +0000 (08:02 -0800)]
Stop overloading internal_dispose in SkTextBlob and SkData

Review URL: https://codereview.chromium.org/737093002

10 years agoMake sure pictures draw Clears even when the clip is empty.
mtklein [Thu, 20 Nov 2014 16:01:09 +0000 (08:01 -0800)]
Make sure pictures draw Clears even when the clip is empty.

We fix this by rewriting empty queries to very tiny queries, which will certainly
hit ops that span the entire picture (like Clear) and hopefully not much more.

(This doesn't quite work in the full cull rect world if [0,0,ε,ε] doesn't
overlap the picture.  Let's cross that bridge when we get there.)

BUG=432991

Review URL: https://codereview.chromium.org/732723004

10 years agodstCopy on optdrawstate
joshualitt [Thu, 20 Nov 2014 15:28:52 +0000 (07:28 -0800)]
dstCopy on optdrawstate

BUG=skia:

Review URL: https://codereview.chromium.org/735363003

10 years agoPreventing division by 0 in non-separable blend mode shaders.
rosca [Thu, 20 Nov 2014 15:24:32 +0000 (07:24 -0800)]
Preventing division by 0 in non-separable blend mode shaders.

In the software path, the same issue has been fixed some time ago:
https://codereview.chromium.org/114173002

BUG=skia:

Review URL: https://codereview.chromium.org/666043003

10 years agoImplement conics for NVPR
kkinnunen [Thu, 20 Nov 2014 06:49:03 +0000 (22:49 -0800)]
Implement conics for NVPR

Implement conics for NVPR. Verified with currently disabled conicpaths
gm, produces mostly same shapes with nvprmsaa4 and msaa4.

BUG=skia:2033

Review URL: https://codereview.chromium.org/694503003

10 years agoeven more stragglers -- hopefully this makes the gtx machine go green?
Greg Humphreys [Wed, 19 Nov 2014 22:28:53 +0000 (17:28 -0500)]
even more stragglers -- hopefully this makes the gtx machine go green?

BUG=skia:

Review URL: https://codereview.chromium.org/716243005

10 years agomore rebaseline stragglers
Greg Humphreys [Wed, 19 Nov 2014 22:02:31 +0000 (17:02 -0500)]
more rebaseline stragglers

BUG=skia:

Review URL: https://codereview.chromium.org/726373004

10 years agoStore DM failures in an array, not a single string.
mtklein [Wed, 19 Nov 2014 21:36:19 +0000 (13:36 -0800)]
Store DM failures in an array, not a single string.

This should fix the problem where a repeatedly failing test causes this assertion:
  ../../src/core/SkString.cpp:572: failed assertion "length >= 0 && length < SkToInt(kBufferSize)"

Example output:

0 GMs x 7 configs, 2 tests, 0 pictures
0 tasks left, 100 failed
Failures:
  test RTree: ../../tests/RTreeTest.cpp:77 0 != rtree.getCount()
  test RTree: ../../tests/RTreeTest.cpp:77 0 != rtree.getCount()
  test RTree: ../../tests/RTreeTest.cpp:77 0 != rtree.getCount()
  test RTree: ../../tests/RTreeTest.cpp:77 0 != rtree.getCount()
  < 95 lines elided >
  test RTree: ../../tests/RTreeTest.cpp:77 0 != rtree.getCount()
100 failures.

BUG=skia:

Review URL: https://codereview.chromium.org/741833002

10 years agoa few straggling rebaselines
Greg Humphreys [Wed, 19 Nov 2014 21:35:20 +0000 (16:35 -0500)]
a few straggling rebaselines

BUG=skia:

Review URL: https://codereview.chromium.org/742893002

10 years agoSmall cleanups after GrDrawState/GrOptDrawState base class elimination.
bsalomon [Wed, 19 Nov 2014 20:36:22 +0000 (12:36 -0800)]
Small cleanups after GrDrawState/GrOptDrawState base class elimination.

Review URL: https://codereview.chromium.org/739133004

10 years agoImplement SkPicture::bytesUsed() for SkRecord backend
tomhudson [Wed, 19 Nov 2014 18:41:14 +0000 (10:41 -0800)]
Implement SkPicture::bytesUsed() for SkRecord backend

BUG=chromium:230419
R=mtklein@google.com,reed@google.com

Review URL: https://codereview.chromium.org/490253003

10 years agoNo need to short circuit the query rect here anymore.
mtklein [Wed, 19 Nov 2014 18:37:20 +0000 (10:37 -0800)]
No need to short circuit the query rect here anymore.

Whatever assertion was triggering in SkRTree.cpp seems to be gone now.

BUG=432991

Review URL: https://codereview.chromium.org/725373003

10 years agoRemove debug-only fData from SKTDArray.
mtklein [Wed, 19 Nov 2014 17:41:27 +0000 (09:41 -0800)]
Remove debug-only fData from SKTDArray.

It makes the code harder to read, and makes Debug and Release SkTDArrays
different sizes.  Looks like fData is left over from when debuggers weren't
very good at inspecting data structures.

No API changes.
TBR=reed@google.com

BUG=skia:

Review URL: https://codereview.chromium.org/739263002

10 years agoscissor state on optstate
joshualitt [Wed, 19 Nov 2014 17:38:51 +0000 (09:38 -0800)]
scissor state on optstate

waiting

BUG=skia:

Review URL: https://codereview.chromium.org/742763002

10 years agoAdd an image filter row to GM:lcdtextprops
fmalita [Wed, 19 Nov 2014 17:23:22 +0000 (09:23 -0800)]
Add an image filter row to GM:lcdtextprops

We should be disabling LCD text when rendering image filters, but
currently don't. Add a GM to capture this behavior.

R=reed@google.com

Review URL: https://codereview.chromium.org/741433003

10 years agoRemove /arch:SSE2 from Win 64 Release builds
bsalomon [Wed, 19 Nov 2014 16:59:58 +0000 (08:59 -0800)]
Remove /arch:SSE2 from Win 64 Release builds

Review URL: https://codereview.chromium.org/740733002

10 years agoadd runtime config option to retain stream for custom typefaces
humper [Wed, 19 Nov 2014 16:32:19 +0000 (08:32 -0800)]
add runtime config option to retain stream for custom typefaces

BUG=skia:

Review URL: https://codereview.chromium.org/729973004

10 years agomove GrGeometryProcessor to src
joshualitt [Wed, 19 Nov 2014 16:24:09 +0000 (08:24 -0800)]
move GrGeometryProcessor to src

TBR=
BUG=skia:

Review URL: https://codereview.chromium.org/739943003

10 years agoCreate GrOptDrawState before recording draw in GrInOrderDrawBuffer
bsalomon [Wed, 19 Nov 2014 16:23:49 +0000 (08:23 -0800)]
Create GrOptDrawState before recording draw in GrInOrderDrawBuffer

Review URL: https://codereview.chromium.org/739673002

10 years agoAllow pictures to have a full bounds
reed [Wed, 19 Nov 2014 16:04:34 +0000 (08:04 -0800)]
Allow pictures to have a full bounds

This reverts commit 7c4cdd2c5b7b6c71fa0e7612594f7885f2dbf264.

BUG=skia:

Review URL: https://codereview.chromium.org/738083002

10 years agoRevert of allow pictures to have a full bounds (patchset #3 id:40001 of https://coder...
reed [Wed, 19 Nov 2014 15:17:16 +0000 (07:17 -0800)]
Revert of allow pictures to have a full bounds (patchset #3 id:40001 of https://codereview.chromium.org/736583004/)

Reason for revert:
needed to update legacy width() helpers

Original issue's description:
> allow pictures to have a full bounds
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/7e76bff26e7c74902841ca4f607eb0b24a833a4a

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

Review URL: https://codereview.chromium.org/722043005

10 years agoallow pictures to have a full bounds
reed [Wed, 19 Nov 2014 14:59:41 +0000 (06:59 -0800)]
allow pictures to have a full bounds

BUG=skia:

Review URL: https://codereview.chromium.org/736583004

10 years agoAdd processor name to comment when printing out shaders.
egdaniel [Wed, 19 Nov 2014 14:22:39 +0000 (06:22 -0800)]
Add processor name to comment when printing out shaders.

BUG=skia:

Review URL: https://codereview.chromium.org/735153002

10 years agoadd roundOut that returns its result
reed [Wed, 19 Nov 2014 13:03:18 +0000 (05:03 -0800)]
add roundOut that returns its result

BUG=skia:

Review URL: https://codereview.chromium.org/742663002

10 years agoUpdate SKP version
skia.buildbots [Wed, 19 Nov 2014 07:06:53 +0000 (23:06 -0800)]
Update SKP version

Automatic commit by the RecreateSKPs bot.

TBR=

Review URL: https://codereview.chromium.org/738003002

10 years agochange roundOut() to take a dst, so we can use that signature to return IRect
reed [Wed, 19 Nov 2014 02:06:45 +0000 (18:06 -0800)]
change roundOut() to take a dst, so we can use that signature to return IRect

BUG=skia:
TBR=robertphilips

Review URL: https://codereview.chromium.org/731373003

10 years agosetting patch_primitive_nvprmsaa4 on K1 to ignore
joshualitt [Wed, 19 Nov 2014 01:38:35 +0000 (17:38 -0800)]
setting patch_primitive_nvprmsaa4 on K1 to ignore

TBR=
BUG=skia:

Review URL: https://codereview.chromium.org/743563003

10 years agodrawinfo carries bufferinfo
joshualitt [Tue, 18 Nov 2014 22:24:27 +0000 (14:24 -0800)]
drawinfo carries bufferinfo

BUG=skia:

Review URL: https://codereview.chromium.org/737723003

10 years agoignore dashcubics test pending skia bug 2049
Greg Humphreys [Tue, 18 Nov 2014 20:42:19 +0000 (15:42 -0500)]
ignore dashcubics test pending skia bug 2049

BUG=skia:2049

Review URL: https://codereview.chromium.org/741533002

10 years agoa few straggling rebaselines
Greg Humphreys [Tue, 18 Nov 2014 20:23:05 +0000 (15:23 -0500)]
a few straggling rebaselines

BUG=skia:

Review URL: https://codereview.chromium.org/738943003

10 years agomore rebaselines for the perspective test server
Greg Humphreys [Tue, 18 Nov 2014 20:19:45 +0000 (15:19 -0500)]
more rebaselines for the perspective test server

BUG=skia:

Review URL: https://codereview.chromium.org/727063004

10 years agorebaseline new GTX images from drive update on linux
Greg Humphreys [Tue, 18 Nov 2014 20:15:27 +0000 (15:15 -0500)]
rebaseline new GTX images from drive update on linux

BUG=skia:
TBR=borenet
NOTRY=true

Review URL: https://codereview.chromium.org/735833002

10 years agoAdd computation of saveLayer information to RecordingBench
robertphillips [Tue, 18 Nov 2014 19:53:02 +0000 (11:53 -0800)]
Add computation of saveLayer information to RecordingBench

In (Add flag to beginRecording to request saveLayer information - https://codereview.chromium.org/721883002/) I claimed the extra recording cost would be negligible. This CL attempts to put some numbers behind that.

Review URL: https://codereview.chromium.org/741523002

10 years agohttp://skbug.com/3130 step 4 (include/utils/SkWGL.h gone for good)
halcanary [Tue, 18 Nov 2014 19:30:20 +0000 (11:30 -0800)]
skbug.com/3130 step 4 (include/utils/SkWGL.h gone for good)

BUG=skia:3130

Review URL: https://codereview.chromium.org/729363003

10 years agowip for drawables
reed [Tue, 18 Nov 2014 19:08:05 +0000 (11:08 -0800)]
wip for drawables

Idea:

1. in its mutable recording state, keep a table of drawables on the side, and store an index in the record list.

2. In "immediate-mode" draw, just call the clients drawable directly (need access to our private list to turn the stored index into a proc)

3. when we "snap", we replace the list of drawables with a list of (sub) pictures, and then during playback of the snapped picture, we invoke a private drawable which just calls "drawPicture" on the index'd subpicture.

Review URL: https://codereview.chromium.org/727363003

10 years agorebaseline after dont-normalize-matrix cl
reed [Tue, 18 Nov 2014 17:57:49 +0000 (09:57 -0800)]
rebaseline after dont-normalize-matrix cl

BUG=skia:
TBR=
NOTRY=True

Review URL: https://codereview.chromium.org/738743003

10 years agofix for use of uninit variable
joshualitt [Tue, 18 Nov 2014 17:40:40 +0000 (09:40 -0800)]
fix for use of uninit variable

TBR=
BUG=skia:

Review URL: https://codereview.chromium.org/740463002

10 years agoSkVarAlloc::approxBytesAllocated()
mtklein [Tue, 18 Nov 2014 17:32:36 +0000 (09:32 -0800)]
SkVarAlloc::approxBytesAllocated()

This is what I was getting at on the other CL.

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/f27f1bcce50c8f95aea8469684a70b70c9baee09

CQ_EXTRA_TRYBOTS=client.skia.android:Test-Android-Nexus5-Adreno330-Arm7-Release-Trybot

Review URL: https://codereview.chromium.org/730193003

10 years agoPrune SkRTree
mtklein [Tue, 18 Nov 2014 17:27:49 +0000 (09:27 -0800)]
Prune SkRTree

  - Propagate a bunch of constant parameters through.
  - Delete code that's not used when bulk loading.
  - Allocate all Nodes together.
  - Stay in SkRect.

Doing a single malloc for the nodes can't not have improved memory usage.

Looks like this might improve record performance ~5%, probably mostly from
staying in SkRects.  This finally dethrones building the BBH as the hot spot.
(Now it's mapping user bounds back to device bounds and adjusting for paints.)

Recording time changes from my MBP:
    desk_rectangletransition.skp 11.5us -> 11.7us 1x
             desk_forecastio.skp  115us ->  114us 0.98x
                desk_booking.skp  550us ->  541us 0.98x
            tabl_mercurynews.skp  176us ->  173us 0.98x
                   tabl_hsfi.skp  294us ->  287us 0.98x
              desk_wordpress.skp  351us ->  343us 0.98x
           tabl_worldjournal.skp  439us ->  426us 0.97x
                  tabl_gmail.skp 20.3us -> 19.7us 0.97x
         desk_youtubetvvideo.skp 10.8us -> 10.4us 0.97x
             desk_googleplus.skp  1.1ms -> 1.07ms 0.97x
               tabl_slashdot.skp  106us ->  103us 0.97x
         desk_jsfiddlebigcar.skp 26.7us -> 25.7us 0.96x
               tabl_techmeme.skp 95.4us -> 91.7us 0.96x
             tabl_deviantart.skp  133us ->  127us 0.96x
              desk_pinterest.skp 40.6us -> 38.9us 0.96x
                 desk_carsvg.skp  195us ->  187us 0.96x
               tabl_engadget.skp  376us ->  359us 0.96x
                tabl_sahadan.skp 60.5us -> 57.5us 0.95x
      tabl_culturalsolutions.skp  255us ->  242us 0.95x
                  tabl_gspro.skp 58.3us -> 55.5us 0.95x
               desk_linkedin.skp  146us ->  138us 0.94x
                   desk_ebay.skp  192us ->  181us 0.94x
                    tabl_cnn.skp  467us ->  440us 0.94x
     desk_jsfiddlehumperclip.skp 29.9us -> 28.1us 0.94x
               desk_tigersvg.skp 43.2us -> 40.5us 0.94x
           desk_yahooanswers.skp  131us ->  123us 0.94x
desk_googlespreadsheetdashed.skp 1.18ms -> 1.11ms 0.94x
                desk_blogger.skp  193us ->  181us 0.94x
                tabl_mozilla.skp 1.82ms ->  1.7ms 0.94x
                    tabl_mlb.skp  145us ->  136us 0.93x
              mobi_wikipedia.skp  577us ->  539us 0.93x
               tabl_frantzen.skp 54.1us -> 50.4us 0.93x
                  desk_baidu.skp 87.9us -> 81.9us 0.93x
             desk_techcrunch.skp  224us ->  209us 0.93x
                 desk_sfgate.skp  206us ->  192us 0.93x
                  tabl_ukwsj.skp  269us ->  250us 0.93x
               desk_facebook.skp  316us ->  293us 0.93x
            desk_gmailthread.skp  205us ->  190us 0.93x
         tabl_googlecalendar.skp  158us ->  147us 0.93x
                   tabl_digg.skp  382us ->  354us 0.93x
                 desk_amazon.skp  106us -> 98.5us 0.93x
          tabl_androidpolice.skp  693us ->  642us 0.93x
                tabl_nytimes.skp  206us ->  191us 0.92x
                    desk_gws.skp  124us ->  114us 0.92x
                desk_youtube.skp  255us ->  235us 0.92x
           tabl_cuteoverload.skp  583us ->  537us 0.92x
            desk_oldinboxapp.skp   18us -> 16.6us 0.92x
             desk_mobilenews.skp  297us ->  273us 0.92x
                 tabl_pravda.skp  168us ->  154us 0.92x
              tabl_vnexpress.skp  236us ->  217us 0.92x
          desk_css3gradients.skp  202us ->  185us 0.92x
            tabl_gamedeksiam.skp  508us ->  464us 0.91x
                desk_wowwiki.skp 1.02ms ->  929us 0.91x
                   desk_espn.skp  209us ->  191us 0.91x
             desk_chalkboard.skp  315us ->  284us 0.9x
                 desk_mapsvg.skp  607us ->  543us 0.89x
            desk_pokemonwiki.skp 5.18ms -> 4.62ms 0.89x
               desk_samoasvg.skp  335us ->  298us 0.89x
        desk_youtubetvbrowse.skp 10.1us -> 8.59us 0.85x
BUG=skia:3085, skia:2834

Review URL: https://codereview.chromium.org/734723002

10 years agoRevert of SkVarAlloc::approxBytesAllocated (patchset #5 id:80001 of https://coderevie...
mtklein [Tue, 18 Nov 2014 16:15:23 +0000 (08:15 -0800)]
Revert of SkVarAlloc::approxBytesAllocated (patchset #5 id:80001 of https://codereview.chromium.org/730193003/)

Reason for revert:
Android needs dlmalloc_usable_size().

Original issue's description:
> SkVarAlloc::approxBytesAllocated()
>
> This is what I was getting at on the other CL.
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/f27f1bcce50c8f95aea8469684a70b70c9baee09
>
> CQ_EXTRA_TRYBOTS=Test-Android-Nexus5-Adreno330-Arm7-Release-Trybot

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

Review URL: https://codereview.chromium.org/741443002

10 years agominor cleanup in GrGpu.h
bsalomon [Tue, 18 Nov 2014 15:36:19 +0000 (07:36 -0800)]
minor cleanup in GrGpu.h

Review URL: https://codereview.chromium.org/731073003

10 years agorebaseline perspective gm after matrix dont-renormalize-cl
reed [Tue, 18 Nov 2014 15:32:26 +0000 (07:32 -0800)]
rebaseline perspective gm after matrix dont-renormalize-cl

BUG=skia:
TBR=
NOTRY=True

Review URL: https://codereview.chromium.org/741433002

10 years agoSkVarAlloc::approxBytesAllocated()
mtklein [Tue, 18 Nov 2014 15:24:14 +0000 (07:24 -0800)]
SkVarAlloc::approxBytesAllocated()

This is what I was getting at on the other CL.

BUG=skia:

Review URL: https://codereview.chromium.org/730193003

10 years agodon't normalize after perspective concat
reed [Tue, 18 Nov 2014 14:04:37 +0000 (06:04 -0800)]
don't normalize after perspective concat

BUG=skia:

Review URL: https://codereview.chromium.org/734513004

10 years agoRemove unused NVPR related GL tokens
kkinnunen [Tue, 18 Nov 2014 13:13:48 +0000 (05:13 -0800)]
Remove unused NVPR related GL tokens

Remove unused NVPR related GL tokens.

Also replace GR_GL_PATH_INITIAL_END_CAP and GR_GL_PATH_TERMINAL_END_CAP
with a single call setting GR_GL_PATH_END_CAPS. Skia does not and
probably will not have different initial and terminal caps. This came up
in the review of command buffer implementation of the extension.

If more NVPR features will be used, the respective tokens can be added
back per implemented feature.

Review URL: https://codereview.chromium.org/723453002

10 years agoDo not calculate many sierpinski fractals for each nanobench run unless needed
kkinnunen [Tue, 18 Nov 2014 12:50:50 +0000 (04:50 -0800)]
Do not calculate many sierpinski fractals for each nanobench run unless needed

Removes work done by the constructors of picture_nesting benches,
and moves the work to the Benchmark::onPreDraw override.

This avoids PictureNesting::sierpinsky showing up in profile traces
when profiling other benches.

Review URL: https://codereview.chromium.org/725523002

10 years agoUpdate SKP version
skia.buildbots [Tue, 18 Nov 2014 07:03:57 +0000 (23:03 -0800)]
Update SKP version

Automatic commit by the RecreateSKPs bot.

TBR=

Review URL: https://codereview.chromium.org/733323002

10 years agomore baselines for blur2rects
reed [Tue, 18 Nov 2014 02:24:33 +0000 (18:24 -0800)]
more baselines for blur2rects

BUG=skia:
TBR=
NOTRY=True

Review URL: https://codereview.chromium.org/715223003

10 years agoone more rebase on k1
joshualitt [Tue, 18 Nov 2014 01:59:49 +0000 (17:59 -0800)]
one more rebase on k1

TBR=
BUG=skia:

Review URL: https://codereview.chromium.org/734243002

10 years agorebase nvpr after ds on stack
joshualitt [Tue, 18 Nov 2014 00:48:12 +0000 (16:48 -0800)]
rebase nvpr after ds on stack

TBR=
BUG=skia:

Review URL: https://codereview.chromium.org/726103004

10 years agoDrawstate on stack
joshualitt [Mon, 17 Nov 2014 22:22:48 +0000 (14:22 -0800)]
Drawstate on stack

BUG=skia:

Review URL: https://codereview.chromium.org/732693002

10 years agobaselines for blur2rects
reed [Mon, 17 Nov 2014 21:53:14 +0000 (13:53 -0800)]
baselines for blur2rects

BUG=skia:
TBR=
NOTRY=True

Review URL: https://codereview.chromium.org/733233002

10 years agoReduce the amount of validation in GrResourceCache2
bsalomon [Mon, 17 Nov 2014 20:15:56 +0000 (12:15 -0800)]
Reduce the amount of validation in GrResourceCache2

BUG=chromium:433711

Review URL: https://codereview.chromium.org/737533002

10 years agoFix yet another convexicator issue
robertphillips [Mon, 17 Nov 2014 20:11:42 +0000 (12:11 -0800)]
Fix yet another convexicator issue

This CL fixes the case where a bad initial vector (i.e., nearly zero) managed to short circuit all of the convexicator's logic. The initial bad vector would become the last vector and then never get displaced.

The history of this is:

https://codereview.chromium.org/298973004/
Switched the convexicator to not advance the last vector when the cross product wasn't significant

https://codereview.chromium.org/573763002/
Fixed a bug (crbug.com/412640) wherein a zero area path was being incorrectly categorized as convex b.c. opposite but equal vectors were not signaling concavity.

BUG=433683

Review URL: https://codereview.chromium.org/727283003

10 years agoReally enable deprecation warnings for Android framework.
scroggo [Mon, 17 Nov 2014 18:29:18 +0000 (10:29 -0800)]
Really enable deprecation warnings for Android framework.

Left out of a previous CL to reenable.

(We had the define added in two places for some reason...)

Review URL: https://codereview.chromium.org/728503002

10 years agoRemove bug_chomper
borenet [Mon, 17 Nov 2014 17:59:53 +0000 (09:59 -0800)]
Remove bug_chomper

It has been moved into buildbot repo.

BUG=skia:

Review URL: https://codereview.chromium.org/731123003

10 years agoAllow GPU resources to not be counted against the cache budget.
bsalomon [Mon, 17 Nov 2014 17:33:27 +0000 (09:33 -0800)]
Allow GPU resources to not be counted against the cache budget.

BUG=skia:2889

Review URL: https://codereview.chromium.org/721353002

10 years agoRebase GM on win7 after res cache change.
bsalomon [Mon, 17 Nov 2014 16:54:04 +0000 (08:54 -0800)]
Rebase GM on win7 after res cache change.

TBR=

Review URL: https://codereview.chromium.org/733133003

10 years agoMake BBH parameter to SkRecordComputeLayers and SkRecordFillBounds optional
robertphillips [Mon, 17 Nov 2014 16:16:15 +0000 (08:16 -0800)]
Make BBH parameter to SkRecordComputeLayers and SkRecordFillBounds optional

Chromium creates a picture to contain their picture pile in order to use MultiPictureDraw. They currently do not create a bounding box for that picture but I still need layer information for it. This change allows Chromium to continue without a BBH but still have layer information.

In the future, the brute force BBH might be suitable for their use case.

Please see gpu_raster_worker_pool.cc in (Add flag to beginRecording to request saveLayer information - https://codereview.chromium.org/721883002/) for where this is happening in Chromium.

Review URL: https://codereview.chromium.org/733963004

10 years agoUse DrawStates ProcOptInfo when creating the Optimized Draw State.
egdaniel [Mon, 17 Nov 2014 15:45:53 +0000 (07:45 -0800)]
Use DrawStates ProcOptInfo when creating the Optimized Draw State.

BUG=skia:

Review URL: https://codereview.chromium.org/726693002

10 years agotweak gm to show fractional offsets
reed [Mon, 17 Nov 2014 15:41:35 +0000 (07:41 -0800)]
tweak gm to show fractional offsets

BUG=skia:
TBR=

Review URL: https://codereview.chromium.org/728403002

10 years agoCorrect accounting for wrapped resources
bsalomon [Mon, 17 Nov 2014 15:34:06 +0000 (07:34 -0800)]
Correct accounting for wrapped resources

BUG=skia:2889

Review URL: https://codereview.chromium.org/720033004

10 years agoMore SkPicture cleanup
mtklein [Mon, 17 Nov 2014 14:45:18 +0000 (06:45 -0800)]
More SkPicture cleanup
 - move field declarations together and pack them a little tighter
 - get rid of fData
 - remove dead code in debugger, including unused SkPicturePlayback subclass

There are now no more long-lived SkPictureData!  (Really, there never were,
but now we don't pretend to support them.)

BUG=skia:

No API changes.
TBR=reed@google.com

Review URL: https://codereview.chromium.org/725143002

10 years agoRevert "move SkPDFD*.h from include to src"
Greg Humphreys [Mon, 17 Nov 2014 14:42:09 +0000 (09:42 -0500)]
Revert "move SkPDFD*.h from include to src"

This reverts commit 693024300f79ff3fb497ddee1043b609ddf521a4.

was breaking canary build

BUG=skia:

Review URL: https://codereview.chromium.org/731173002

10 years agomove SkPDFD*.h from include to src
halcanary [Mon, 17 Nov 2014 14:26:21 +0000 (06:26 -0800)]
move SkPDFD*.h from include to src

BUG=278148

Review URL: https://codereview.chromium.org/727343002

10 years agoseparate c headers
reed [Mon, 17 Nov 2014 14:15:42 +0000 (06:15 -0800)]
separate c headers

BUG=skia:

Review URL: https://codereview.chromium.org/728323002

10 years agoAdded missing validation
sugoi [Mon, 17 Nov 2014 13:45:55 +0000 (05:45 -0800)]
Added missing validation

BUG=433359

Review URL: https://codereview.chromium.org/729583002

10 years agoUpdate SKP version
skia.buildbots [Mon, 17 Nov 2014 06:31:01 +0000 (22:31 -0800)]
Update SKP version

Automatic commit by the RecreateSKPs bot.

TBR=

Review URL: https://codereview.chromium.org/731003004

10 years agoMove SkDocument_PDF.cpp into pdf.gypi - now standard PDF API
halcanary [Sun, 16 Nov 2014 16:26:54 +0000 (08:26 -0800)]
Move SkDocument_PDF.cpp into pdf.gypi - now standard PDF API

TBR=reed@google.com

Review URL: https://codereview.chromium.org/714223002

10 years agoUpdate SKP version
skia.buildbots [Sun, 16 Nov 2014 06:52:04 +0000 (22:52 -0800)]
Update SKP version

Automatic commit by the RecreateSKPs bot.

TBR=

Review URL: https://codereview.chromium.org/734493002

10 years agohttp://skbug.com/3130 step 1 (to fix chromium roll)
halcanary [Sat, 15 Nov 2014 14:11:18 +0000 (06:11 -0800)]
skbug.com/3130 step 1 (to fix chromium roll)

BUG=skia:3130
TBR=djsollen@google.com, bsalomon@google.com

Review URL: https://codereview.chromium.org/729173002

10 years agoUpdate SKP version
skia.buildbots [Sat, 15 Nov 2014 06:56:33 +0000 (22:56 -0800)]
Update SKP version

Automatic commit by the RecreateSKPs bot.

TBR=

Review URL: https://codereview.chromium.org/730833002

10 years agoFix Chromium roll broken by http://crrev.com/728823002
halcanary [Sat, 15 Nov 2014 01:22:05 +0000 (17:22 -0800)]
Fix Chromium roll broken by crrev.com/728823002

Review URL: https://codereview.chromium.org/727783004

10 years agoPush creation of default GP to the caller
joshualitt [Sat, 15 Nov 2014 00:00:38 +0000 (16:00 -0800)]
Push creation of default GP to the caller

TBR=
BUG=skia:

Review URL: https://codereview.chromium.org/715903002

10 years agoFix Android framework build.
scroggo [Fri, 14 Nov 2014 22:49:34 +0000 (14:49 -0800)]
Fix Android framework build.

When we define SK_BUILD_FOR_ANDROID, we don't define it to anything,
so

    #if SK_BUILD_FOR_ANDROID

is not a meaningful check. Instead, use #ifdef.

TBR=caryclark@google.com, djsollen@google.com

Review URL: https://codereview.chromium.org/728013002

10 years agodisable gammatext gm for a while, looking for other flakes
reed [Fri, 14 Nov 2014 22:39:29 +0000 (14:39 -0800)]
disable gammatext gm for a while, looking for other flakes

BUG=skia:

Review URL: https://codereview.chromium.org/725303002

10 years agoGuard high water cache stats vars by GR_CACHE_STATS
bsalomon [Fri, 14 Nov 2014 21:59:57 +0000 (13:59 -0800)]
Guard high water cache stats vars by GR_CACHE_STATS

TBR=halcanary@google.com
NOTRY=true

Review URL: https://codereview.chromium.org/729753003

10 years agoMake GrResourceCache2 responsible for calling release, abandon, and ~.
bsalomon [Fri, 14 Nov 2014 21:33:09 +0000 (13:33 -0800)]
Make GrResourceCache2 responsible for calling release, abandon, and ~.

BUG=skia:2889

TBR=robertphillips@google.com

NOTRY=true

Review URL: https://codereview.chromium.org/729683002

10 years agoSkDocument SK_API
halcanary [Fri, 14 Nov 2014 21:24:18 +0000 (13:24 -0800)]
SkDocument SK_API

Review URL: https://codereview.chromium.org/726073002

10 years agoReplace GrResourceCache with GrResourceCache2.
bsalomon [Fri, 14 Nov 2014 20:10:14 +0000 (12:10 -0800)]
Replace GrResourceCache with GrResourceCache2.

BUG=skia:2889

Committed: https://skia.googlesource.com/skia/+/66a450f21a3da174b7eed89a1d5fc8591e8b6ee6

Committed: https://skia.googlesource.com/skia/+/407aa584d183c1bf314f5defd1cf0202e8a96c89

Review URL: https://codereview.chromium.org/716143004