I've been failing to repro the attached bug locally,
but I did notice that we're missing an -O1 like we
usually add to Debug bot builds.
Bug: skia:9793, skia:9664
Change-Id: If30ccfd55bad3e49862d453a754ed9fcbed9bbde
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269938
Auto-Submit: Mike Klein <mtklein@google.com>
Commit-Queue: Weston Tracey <westont@google.com>
Reviewed-by: Weston Tracey <westont@google.com>
Change-Id: Ie6740ccb3302895dc9560e9a9278ab68dd64012b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269903
Auto-Submit: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Change-Id: Ic89359cd08298e73011952452ce909291d6f3fe9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269947
Auto-Submit: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
If a path is small and simple enough, we now try to atlas it. There
is only one atlas and it caps at 2048x2048. Once it runs out of room,
everything just draws direct. The atlas is rendered using the existing
GrTessellatePathOp. It provides alpha8 coverage even for msaa render
targets.
Change-Id: I715da9ce7347b6f1ef8e28b3e13ab47f6eade1c7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/268724
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
There are probably ways to make this more efficient by only optimizing
what's necessary (e.g. try JIT first, then interpreter only if it fails)
and some other performance improvements to make, but for now I want to
focus mostly on keeping things simple and correct.
The line between Builder::done() and Program::Program() is particularly
fuzzy and becoming fuzzier here, and I think that'll be something
that'll change eventually.
This makes SkVMTest debug dumps more portable, though perhaps less
useful. Might kill that feature soon now that SkVM is tested more
thoroughly in unit tests and GMs and bots and such.
Change-Id: Id9ce8daaf8570e5bea8b10f1a80b97f5b33d45dc
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269941
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Minor optimization. When non-AA, the quad corners are never moved so it
won't go through the math that requires w > 0. The GPU is perfectly
capable of clipping to w > 0, except that it produces a non-AA edge;
for AA quads this is a problem, but not so when the draw was non-AA.
Change-Id: Ibf77b678f5b3b90a5a88fb3670a31cd12ff3775f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269762
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
An earlier CL noted that the edge correction sometimes caused popping
when undering extreme perspective, so I removed the edge correction
thinking that it was unnecessary. Turns out that for actual degenerate
quads, it is necessary. However, instead of re-determining the degenerate
edge based on the local edge length, this corrects the exact same edges
that had been corrected in device space. I confirmed that this seems to
prevent the popping under extreme perspective and draws triangles correctly.
It also avoids the coordinate scale issue that comes about when the edge
tolerances had been chosen for pixel space, but many of the local coordinates
being processed had already been normalized so differed by a factor of 3ish.
Bug: skia:9889
Change-Id: Ida4c626aa982fe4fdac6695e2ad95e162e42fca2
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269758
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Bug: skia:9556
Change-Id: I1ac3744632b6e926f79e182d4cd9d59498699aee
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269896
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
This reverts commit 33c7a468c8.
Reason for revert: wow, lots of gm diffs, need to investigate
Original change's description:
> Defined N32 to always be BGRA
>
>
> https://critique.corp.google.com/#review/294013931
>
> Change-Id: I0e837b2a97b1d95918e9054c036e2a3d913009e2
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269636
> Commit-Queue: Mike Reed <reed@google.com>
> Reviewed-by: Mike Klein <mtklein@google.com>
TBR=djsollen@google.com,mtklein@google.com,bsalomon@google.com,reed@google.com
Change-Id: Ic8aeafbbed91ed4619737be168ca4cacedac68fe
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269916
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
Bug: skia:9556
Change-Id: I4042f1339844186b73b807e93b1a3701c32bf112
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269366
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
If the scripts fail to produce valid bytecode, don't overwrite the
interpreter.
Change-Id: Icd008a5188166ce086ff4df87dcb2b43d7f80820
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269487
Auto-Submit: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
e02f4996c6..a8e6a46312
git log e02f4996c6e9..a8e6a463121b --date=short --first-parent --format='%ad %ae %s'
2020-02-09 jmadill@chromium.org Suppress Vulkan RobustResourceInit tests.
2020-02-09 b.schade@samsung.com Change g_Mutex from std::mutex to std::recursive_mutex
2020-02-08 cnorthrop@google.com Capture/Replay: Add 2DArray and 3D texture support for MEC
2020-02-08 jmadill@chromium.org Make Debug marker functions return angle::Result.
2020-02-07 jmadill@chromium.org Vulkan: Clamp vertex buffer offsets.
2020-02-07 jmadill@chromium.org Vulkan: Add missing wrapper functions for XFB.
2020-02-07 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/spirv-tools/src 1f03ac10270a..fe10239f92f4 (1 commits)
2020-02-07 angle-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/glslang/src 656650f461c8..95c414ffee64 (2 commits)
Created with:
gclient setdep -r third_party/externals/angle2@a8e6a463121b
If this roll has caused a breakage, revert this CL and stop the roller
using the controls here:
https://autoroll.skia.org/r/angle-skia-autoroll
Please CC stani@google.com on the revert to ensure that a human
is aware of the problem.
To report a problem with the AutoRoller itself, please file a bug:
https://bugs.chromium.org/p/skia/issues/entry?template=Autoroller+Bug
Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md
Cq-Include-Trybots: skia/skia.primary:Build-Debian9-Clang-x86_64-Release-ANGLE;skia/skia.primary:Test-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-Golo-GPU-QuadroP400-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC8i5BEK-GPU-IntelIris655-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE
Bug: None
Tbr: stani@google.com
Change-Id: I33a6d40f4f8908792c29014029a27ecece5e340d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269717
Reviewed-by: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
On some paths with many self-intersections, performance can approach
O(N^2) in intersections, due to
"GrTessellator: always rewind to edge top when splitting",
aka b67b2354cf.
This patch reverts that change, brings back rewind_if_necessary(), and
fixes the underlying issue by continuing to rewind past any AEL
violations.
Bug: 1030306
Change-Id: I64cea1d09bf383c245a5fd2ddb6569b4041b6638
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269238
Reviewed-by: Chris Dalton <csmartdalton@google.com>
Commit-Queue: Stephen White <senorblanco@chromium.org>
Change-Id: Id84c3fb35cb61fa839691471d03a44152964bedb
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/268941
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
We can skip effect layers when none of the child nodes overlap (treat as
atomic draws, with effects applied on their paint).
The initial heuristic was simply checking for more than one child.
This version relaxes the heuristic to check each child bounds against
the union of preceding siblings.
Change-Id: I2dc6d30d945697f410100db0c46a8a8377d5b569
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269482
Commit-Queue: Florin Malita <fmalita@chromium.org>
Reviewed-by: Mike Klein <mtklein@google.com>
By using hex color instead of RGB in SkSVGDevice, SVG byte size can be
reduced. For example, Quora landing page SVG size: 1588515 bytes (RGB)
vs. 1587899 bytes (Hex). Size reduces by 616 bytes.
Bug: NONE
Test: dm -v --match SVGDevice
Change-Id: I1d557c9d8a925c2dde96899e23833dfb89264903
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269260
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
std::call_once() will trigger false positive races without
a libc++ (specifically __call_once()) built with TSAN itself.
Interestingly, there's a built-in interceptor for this
use case, only on Mac: https://reviews.llvm.org/D24188.
Fixes: skia:9884
Change-Id: I7b2c031c15b78ec7302c42b4e4ac365d596c2c33
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/265567
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Ben Wagner aka dogben <benjaminwagner@google.com>
Change-Id: Id046199edd63535ef07e1dfa65fbc7c0f8cefd00
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269371
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Replace with SkISize.
Also change some const SkISize& params to just SkISize.
Change-Id: I3c72d961662eefeda545fba17d63e877cd5ca813
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269374
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Bug: b/135133301
Follow-on to 196f319b.
- Add SkCodec::getICCProfile to match the SkAndroidCodec version.
- Update comments on getPixels() regarding how the SkColorSpace on the
SkImageInfo is treated.
- Add two new images that have ICC profiles that do not map to an
SkColorSpace. Add a test to verify that they have the un-transformed
color we expect.
- Stop uploading ColorCodecSrc images decoded to a null SkColorSpace to
Gold. Though they may be correct, they do not match other images they're
compared against. The new test above verifies that we do not do color
conversion with a null SkColorSpace.
Change-Id: I08635e4262f16500fab32ef97511d305c2c06483
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269236
Reviewed-by: Derek Sollenberger <djsollen@google.com>
Commit-Queue: Leon Scroggins <scroggo@google.com>
and improve perspective handling in rest of quad pipeline
This function produces 0, 1, or 2 quads clipped to w > 0, with proper
local coords. To make its signature a little easier to reason about,
I added a simple 'DrawQuad' struct that represents the combination of
device and local coordinates, and edge AA flags. I am open to suggestions
for the name.
GrQuad::bounds() remains perspective aware so that it is always correct.
I updated CropToRect to check for w < 0 and not attempt to crop in that
scenario. Theoretically, we could clip to w = 0 first and then go through
the CropToRect optimization path. However, with the current state of the
GrFillRectOp and GrTextureOp, that made it more annoying to have the bulk
APIs handle the w clipping as well.
So for now, the w plane clipping is entirely the responsibility of the
ops. A benefit of this approach is that GrRenderTargetContext doesn't need
to be modified, and in the case where the clipping produces 2 quads they
are automatically put in the same op w/o going through any batching code.
However, it is becoming clear to me that managing 4 effective code paths
(fill + texture X simple + bulk API) is more maintenance than it's worth.
I added skbug:9869 to work out how to simplify these op creation APIs further,
and if we succeed there, I think it will make applying the W plane clipping
more convenient as well.
For now, since this affects SkiaRenderer on Linux stable, I am pushing
for correctness.
Bug: skia:9779, chromium:224618
Change-Id: I4218a956cbe0bbc2b5c9cf133a069d54c93848e8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/268686
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Use std::min and std::max everywhere.
SkTPin still exists. We can't use std::clamp yet, and even when
we can, it has undefined behavior with NaN. SkTPin is written
to ensure that we return a value in the [lo, hi] range.
Change-Id: I506852a36e024ae405358d5078a872e2c77fa71e
Docs-Preview: https://skia.org/?cl=269357
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269357
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Mike Reed <reed@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Looks like the OS now holds on to the current GL context and doesn't
release it unless explicitly told to.
Bug: skia:9885
Change-Id: Ic7bc6860ec96d88e278682b7f2f67e1ac365d27b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269369
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Auto-Submit: Jim Van Verth <jvanverth@google.com>
Change-Id: I75d7a4bdbfb2a5d22b85bca0cae2975378b19a53
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269365
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Bug: skia:9556
Change-Id: Ibb0e0cff3301554e988d5900455766b6428484ee
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269360
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
Change-Id: I9b5929bbfea046e3ca7df45de2507536f12b3cf9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269359
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Stephen White <senorblanco@chromium.org>
This adds a specialization pass to Builder::optimize() and moves the
x86-specific _imm ops there, rewriting with the Builder API itself. I'm
only using the private Builder::push() call for the moment, but that's
enough to make me feel confident that this is a good way forward: it's
still all going through CSE that way.
We're still doing this any time we're on x86, not when targeting the
JIT, but that'll come next, see the new TODOs. It's mildly better for
the interpreter to not use the _imm ops, but this is really all still
warmup for optimizations with less mild opinions.
I'm not proud of the switch/goto impl but it's the clearest I found.
Change-Id: I30594b403832343528b95967724fd50324cd79d1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269232
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Previously, the normalization, origin flip, and insetting occurred
simultanesouly. The problem was that whether or not the insetting
was needed had a cyclic dependency with the overall filter level
(not known until the end of the loop). So it had an ugly hack that
kept track of the early appended quads and would update their domains
if needed.
This will no longer work when perspective clipping is implemented because
we won't have a 1-1 mapping from quads+viewcount pairs back to the
texture set entries. However, by splitting the insetting out from the
rest of the normalization step, we can defer it until we're in
onPrePrepare w/o losing any optimization capabilities and scales with
the perspective clipping change in the future. It should also make it
easier to upgrade an op from kNearest to kBilerp in onCombineIfPossible
because the prior op will not have applied the bilerp-insetting yet.
Bug: skia:9779
Change-Id: I71b269730de200b8452e2a6b39b777928967f621
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269151
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This reverts commit 77fdf66946.
Reason for revert: Skia-Dawn breakage should not be a tree closer.
Original change's description:
> Revert "Cleanup program building a bit"
>
> This reverts commit 4777e3adde.
>
> Reason for revert: This CL is breaking the build on Linux FYI SkiaRenderer Dawn Release
>
> Original change's description:
> > Cleanup program building a bit
> >
> > This CL:
> > now passes the GrProgramDesc as a const&
> > returns GrGLProgram as an sk_sp
> > makes the parameter ordering more consistent
> > makes GrVkPipelineState no longer ref-counted
> >
> > This is pulled out of the DDL pre-compile CL which touches this portion of the code.
> >
> > Bug: skia:9455
> > Change-Id: Id4d06f93450e276de5a2662be330ae9523026244
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/268777
> > Reviewed-by: Greg Daniel <egdaniel@google.com>
> > Commit-Queue: Robert Phillips <robertphillips@google.com>
>
> TBR=egdaniel@google.com,robertphillips@google.com
>
> # Not skipping CQ checks because original CL landed > 1 day ago.
>
> Bug: skia:9455
> Change-Id: I7019d9876b68576274e87c3b2e6bbbf9695522ba
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269261
> Reviewed-by: Austin Eng <enga@google.com>
> Reviewed-by: Kenneth Russell <kbr@google.com>
> Reviewed-by: Stephen White <senorblanco@chromium.org>
> Commit-Queue: Stephen White <senorblanco@chromium.org>
> Auto-Submit: Austin Eng <enga@google.com>
TBR=egdaniel@google.com,robertphillips@google.com,senorblanco@chromium.org,kbr@google.com,enga@google.com
Change-Id: I62f6d38a8ac351e411f4605425caec3b4783fd70
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:9455
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/269358
Reviewed-by: Stephen White <senorblanco@chromium.org>
Commit-Queue: Stephen White <senorblanco@chromium.org>
