This is in service of: https://skia-review.googlesource.com/c/11125/ (Add parallel proxyID to StencilOps & RenderTargetOpList) where I want a better choke point for texture creation to improve discard handling.
This is a reland of: https://skia-review.googlesource.com/c/11200/ (Rm readPixels from GrSurface & move read/writeSurfacePixels to GrContextPriv)
Change-Id: Icd0a90d2beb483dc24ed87c3bace9c817019e148
Reviewed-on: https://skia-review.googlesource.com/11326
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This moves the signing and packaging of
ios apps from the host that drives the device
to the cloud machine where the app is built.
This is in preparation of moving to Raspberry Pi
hosts.
BUG=skia:5331
Change-Id: I4918272a6c734f5344351f0726b7721fde042234
Reviewed-on: https://skia-review.googlesource.com/9906
Commit-Queue: Stephan Altmueller <stephana@google.com>
Reviewed-by: Eric Boren <borenet@google.com>
Reviewed-by: Mike Klein <mtklein@chromium.org>
Gives Chrome the flexibility to xform and cache
SkImages before they reach the
SkColorSpaceXformCanvas.
Bug: skia:
Change-Id: I1f188f385b953b5a958c15578ea66deffb4dc6c5
Reviewed-on: https://skia-review.googlesource.com/11290
Commit-Queue: Matt Sarett <msarett@google.com>
Reviewed-by: Mike Klein <mtklein@chromium.org>
Reviewed-by: Mike Reed <reed@google.com>
Previously, these were decoded "AsIs". That feature went away (for
simplicity). Unfortunately, that means that images with gamma 2.2 were
being asked to decode to sRGB, incurring a costly (and destructive)
conversion.
Bug: skia:
Change-Id: I6c8cdfefc052c82a4de37c697c2f659dcdd38d4d
Reviewed-on: https://skia-review.googlesource.com/11352
Reviewed-by: Matt Sarett <msarett@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
No one seems to be using this.
Bug: skia:
Change-Id: Iab0644dda2c62bd58159e0a1537138aa2e1ebcbc
Reviewed-on: https://skia-review.googlesource.com/11203
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Matt Sarett <msarett@google.com>
I saved the easiest for last. No generated code diff for store_f32.
This just moves the platform-specific code over to SkJumper_vectors.h
Also clarify types in the existing load4()/store4() functions.
SkJumper_stages.cpp looks good to start growing again!
Change-Id: I6a8599d090b4e17663703b0c0325dbe550a6cdd8
Reviewed-on: https://skia-review.googlesource.com/11348
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
Currently, GrConfigConversionEffect is able to round-trip on many
mobile GPUs because it uses highp for all intermediate variables
(including the texture fetch result). Separating the texture sample
into a different processor breaks that.
This is a blunt instrument, not to be used lightly.
This reverts commit dffe9827b1.
Bug: skia:
Change-Id: I940af3256c47e6672a008d516db9e55669672ca3
Reviewed-on: https://skia-review.googlesource.com/11345
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This should fix the "git cl upload" failures on the recipe roller.
Change-Id: I1c433a5a6dc7f52034bef8b2178b3041b8f4661e
Reviewed-on: https://skia-review.googlesource.com/11020
Reviewed-by: Ravi Mistry <rmistry@google.com>
Commit-Queue: Eric Boren <borenet@google.com>
This reverts commit fb0bd98a43.
Reason for revert: ANGLE failures
Original change's description:
> Rm readPixels from GrSurface & move read/writeSurfacePixels to GrContextPriv
>
> This is in service of: https://skia-review.googlesource.com/c/11125/ (Add parallel proxyID to StencilOps & RenderTargetOpList) where I want a better choke point for texture creation to improve discard handling.
>
> Change-Id: If57a7de47edc0853dae7bc61337d9acdc03d63b0
> Reviewed-on: https://skia-review.googlesource.com/11200
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Commit-Queue: Robert Phillips <robertphillips@google.com>
>
TBR=bsalomon@google.com,robertphillips@google.com,reviews@skia.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
Change-Id: I7241070dc1f9df47181061e07adab141f9857974
Reviewed-on: https://skia-review.googlesource.com/11324
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This is in service of: https://skia-review.googlesource.com/c/11125/ (Add parallel proxyID to StencilOps & RenderTargetOpList) where I want a better choke point for texture creation to improve discard handling.
Change-Id: If57a7de47edc0853dae7bc61337d9acdc03d63b0
Reviewed-on: https://skia-review.googlesource.com/11200
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This device is rather old and was continuously failing anyway.
Bug: skia:
NOTRY=true
Change-Id: Ifd0bed7ded76c805993973f835c1d11f35802d00
Reviewed-on: https://skia-review.googlesource.com/11341
Reviewed-by: Eric Boren <borenet@google.com>
Commit-Queue: Kevin Lubick <kjlubick@google.com>
This build includes the GPU-specific code (EGL, GLES)
The armhf_sysroot should be generic enough to support
other ARM targets, like the Raspberry Pi.
I split out the EGL dependencies into their own CIPD package
(chromebook_c100p_lib) to facilitate the modularity of the approach.
When we add another ARM chromebook, maybe one that has different libraries,
we should be able to re-use armhf_sysroot and then a newly created CIPD asset for
that specific GPU (if needed).
Maybe this also can be used to build for the chromecasts (thus the TODO)
Bug: skia:
NOTRY=true
Change-Id: Icc131025932dc8d41da5be39f3c5cad0aa95d848
Reviewed-on: https://skia-review.googlesource.com/11064
Commit-Queue: Kevin Lubick <kjlubick@google.com>
Reviewed-by: Eric Boren <borenet@google.com>
The previous fix to blit framebuffer didn't take cases where
the canvas had a clip applied into account. Fix and update
the unit test to add this case.
Bug: 658277
Change-Id: If3a9d2c8ddf955164cf529c9d6036618f957e426
Reviewed-on: https://skia-review.googlesource.com/11300
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
This reverts commit a08104abf2.
Reason for revert: MSAN's failing immediately, and ASAN is catching a new leak:
==20306==ERROR: LeakSanitizer: detected memory leaks
Direct leak of 4112 byte(s) in 257 object(s) allocated from:
#0 0x138828b (/b/s/w/irNps76D/out/Release/dm+0x138828b)
#1 0x2c07b76 (/b/s/w/irNps76D/out/Release/dm+0x2c07b76)
Original change's description:
> Update clang linux to release_40
>
> Bug: skia:
> Change-Id: Ibe2406592024a4eb20b278ad189d3abddec129d9
> Reviewed-on: https://skia-review.googlesource.com/11285
> Reviewed-by: Mike Klein <mtklein@chromium.org>
> Commit-Queue: Mike Klein <mtklein@chromium.org>
>
TBR=mtklein@chromium.org,mtklein@google.com,kjlubick@google.com,reviews@skia.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
Change-Id: I178c129cf19ed38f0a91ea3eea583643daadc2c7
Reviewed-on: https://skia-review.googlesource.com/11288
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
Bug: skia:
Change-Id: Ibe2406592024a4eb20b278ad189d3abddec129d9
Reviewed-on: https://skia-review.googlesource.com/11285
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
This makes the legacy code use GrProcessorSet::Analysis in the same manner as the non-GrLegacyMeshDrawOps which enables changes to how analysis works.
Change-Id: I8171e285ac8930beb3ac33cd3c4ee88f217b9e40
Reviewed-on: https://skia-review.googlesource.com/11205
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
I set -mdspr2 unconditionally on 32-bit MIPS builds only to not have to
deal with setting it conditionally on the source files specifically
requiring those instructions.
Since then, I've deleted all the source files specifically requiring
-mdspr2 (and -mdsp). No one understood them and they were only
accelerating really uninteresting cases, like 565 blits.
So, we haven't needed -mdspr2 anywhere for a while, and removing it
will unblock running these builds on the CI20 dev boards, which
support neither -mdsp nor -mdsp2.
BUG=skia:1437
CQ_INCLUDE_TRYBOTS=skia.primary:Build-Ubuntu-Clang-mipsel-Debug-Android,Build-Ubuntu-Clang-mipsel-Release-Android
Change-Id: Idda56fd8bb37ba064b78b4cddd8878b37dfe9502
Reviewed-on: https://skia-review.googlesource.com/11284
Commit-Queue: Mike Klein <mtklein@chromium.org>
Reviewed-by: Derek Sollenberger <djsollen@google.com>
Reviewed-by: Mike Klein <mtklein@chromium.org>
Pretty much the same deal as the last CL going the other direction:
split store_f16 into to_half() and store4(). Platforms that had fused
strategies here get a little less optimal, but the code's easier to
follow, maintain, and reuse.
Also adds widen_cast() to encapsulate the fairly common pattern of
expanding one of our logical vector types (e.g. 8-byte U16) up to the
width of the physical vector type (e.g. 16-byte __m128i). This
operation is deeply understood by Clang, and often is a no-op.
I could make bit_cast() do this, but it seems clearer to have two names.
Change-Id: I7ba5bb4746acfcaa6d486379f67e07baee3820b2
Reviewed-on: https://skia-review.googlesource.com/11204
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
Crurently, when preparing a texture for blitFramebuffer, we ignore the
kRectsMustMatchForMSAASrc_BlitFramebufferFlag, and may attempt to
copy from one src rect to a different dst rect.
This change updates initDescForDstCopy and setupDstTexture to allocate
larger textures if necessary and accomodate this flags requirements.
Bug: 658277
Change-Id: If4489ac3192dcf6f9996494c63821279721d0a12
Reviewed-on: https://skia-review.googlesource.com/11141
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
This saves ~1:30 of overhead when the isolate cache is warm.
Bug: skia:5813
Change-Id: I5053d66e865ca6500b82567f83be8042568e22cf
Reviewed-on: https://skia-review.googlesource.com/11063
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Commit-Queue: Eric Boren <borenet@google.com>
GrXferProcessor can no longer use this functionality so it is moved to a new intermediate class inherited by GrFragmentProcessor and GrPrimitiveProcessor.
Change-Id: I4f30c89bdceb2d77b602bf0646107e0780881c26
Reviewed-on: https://skia-review.googlesource.com/11202
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This fixes a bug that was exposed when I added color space
support for 565 decodes.
Before this CL, we would sometimes fill incomplete images with
R and B swapped.
This fixes that issue. Part of the fix is the decision to do 565
xforms when building the color table (then just swizzle to 565),
rather than do them per pixel after swizzling.
Bug: skia:
Change-Id: I09e1ec75aba09a4e288015ea746465d0c3f7d59f
Reviewed-on: https://skia-review.googlesource.com/11137
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Matt Sarett <msarett@google.com>
load_f16 gets slightly worse codegen for ARMv7, SSE2, SSE4.1, and AVX
from splitting it apart compared to the previous fused versions. But
the stage code becomes much simpler.
I'm happy to make those trades until someone complains.
load4() will be useful on its own to implement a couple other stages.
Everything draws the same. I intend to follow up with more of the
same sort of refactoring, but this was tricky enough a change I want
to do them in small steps.
Change-Id: Ib4aa86a58d000f2d7916937cd4f22dc2bd135a49
Reviewed-on: https://skia-review.googlesource.com/11186
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
Coord transforms are always computed and interpolated at high precision.
Bug: skia:
Change-Id: I5f7eadc2080df8ad5cbb080835c0dba09c59e63e
Reviewed-on: https://skia-review.googlesource.com/11180
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This will allow the XP to be created before the dst texture.
Change-Id: I3e5bdfa8e5d47e58a3560792ce5cf3899d30a024
Reviewed-on: https://skia-review.googlesource.com/11011
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
SkJumper_stages.cpp is starting to get unweildy.
This spins some logical parts out into their own headers.
I will follow up by moving more of the very specific
f16/f32 load/store logic into SkJumper_vectors.h too.
Change-Id: I2a3a055e9d1b65f56983d05649270772a4c69f31
Reviewed-on: https://skia-review.googlesource.com/11133
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
Fixes:
- create temp api for android to pass nullptr
- don't release and access sk_sp<SkData> at the same time in parameters
This reverts commit b14131c185.
Bug: skia:
Change-Id: Ic0e4f62520ba9f35455499ed30d306ad19d998a8
Reviewed-on: https://skia-review.googlesource.com/11129
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Matt Sarett <msarett@google.com>
This fixes SkColorSpaceXformCanvas gms that expect
non-linear premuls from the codec.
BUG=skia:
Change-Id: I5dc236d0cd760c23605a26e9c33ddb18955f9231
Reviewed-on: https://skia-review.googlesource.com/10164
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Matt Sarett <msarett@google.com>
Testing this to see what the performance/quality tradeoff is. This is
known to fix some gradient bugs on Mali, but we want to observe perf and
see if it slows thigs down.
Bug: skia:5654
Change-Id: I988410478e89576b8e81786f49e627915e224d42
Reviewed-on: https://skia-review.googlesource.com/11128
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
BUG=skia:6456
Change-Id: Ib94bba9db669562a9b2b64fff56ebe40a2bc0096
Reviewed-on: https://skia-review.googlesource.com/11122
Commit-Queue: Matt Sarett <msarett@google.com>
Reviewed-by: Mike Klein <mtklein@chromium.org>
BUG=skia:
Change-Id: I882b6563c735796f3a4dcd19f6c79e7efd8306ae
Reviewed-on: https://skia-review.googlesource.com/10505
Commit-Queue: Yuqian Li <liyuqian@google.com>
Reviewed-by: Mike Klein <mtklein@chromium.org>
The monolithic GrPipeline is moved to a subclass GrLegacyDrawMeshOp.
The pipeline used to record a GrMesh draw in a GrMeshDrawOp must now be passed rather than implicitly using the op's pipeline.
Change-Id: I50d77e4dcc8d91a523fa7566ce43a9a291174706
Reviewed-on: https://skia-review.googlesource.com/11002
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
This reverts commit ec53c636b7.
Reason for revert: Mali GMs look bad.
Original change's description:
> Revert "Revert "Allow FPs to elevate default precision for the entire fragment program""
>
> This reverts commit 903c3f7040.
>
> Reason for revert: Vulkan issue fixed in compiler.
>
> Original change's description:
> > Revert "Allow FPs to elevate default precision for the entire fragment program"
> >
> > This reverts commit 92d7ccafdf.
> >
> > Reason for revert: Vulkan errors.
> >
> > Original change's description:
> > > Allow FPs to elevate default precision for the entire fragment program
> > >
> > > Currently, GrConfigConversionEffect is able to round-trip on many mobile
> > > GPUs because it uses highp for all intermediate variables (including the
> > > texture fetch result). Separating the texture sample into a different
> > > processor breaks that.
> > >
> > > This is a blunt instrument, not to be used lightly.
> > >
> > > Bug: skia:
> > > Change-Id: I2ab365e3da79628069e2eb727c43c2bf45bfd789
> > > Reviewed-on: https://skia-review.googlesource.com/10162
> > > Reviewed-by: Brian Salomon <bsalomon@google.com>
> > > Commit-Queue: Brian Osman <brianosman@google.com>
> > >
> >
> > TBR=bsalomon@google.com,robertphillips@google.com,brianosman@google.com,ethannicholas@google.com,reviews@skia.org
> > NOPRESUBMIT=true
> > NOTREECHECKS=true
> > NOTRY=true
> >
> > Change-Id: Iee5bb409f86a9cabecc76bd1273a5b3cef6af179
> > Reviewed-on: https://skia-review.googlesource.com/10967
> > Reviewed-by: Brian Osman <brianosman@google.com>
> > Commit-Queue: Brian Osman <brianosman@google.com>
> >
>
> TBR=bsalomon@google.com,robertphillips@google.com,reviews@skia.org,brianosman@google.com,ethannicholas@google.com
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
>
> Change-Id: I733a0ecc40b58d8727f0259b5498c8e6610cedce
> Reviewed-on: https://skia-review.googlesource.com/11010
> Reviewed-by: Brian Osman <brianosman@google.com>
> Commit-Queue: Brian Osman <brianosman@google.com>
>
TBR=bsalomon@google.com,robertphillips@google.com,reviews@skia.org,brianosman@google.com,ethannicholas@google.com
# Not skipping CQ checks because original CL landed > 1 day ago.
Change-Id: Ic3274a0a8b776e811354c3441391ffdc80678292
Reviewed-on: https://skia-review.googlesource.com/11061
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This was confusing and unnecessary. Added an explicit sBGR format to
handle one case. Now the CF always fully describes the color type and
transfer function (or it's legacy).
Bug: skia:
Change-Id: I8f792e8aa18cd2391aa541a9368e345d92bba828
Reviewed-on: https://skia-review.googlesource.com/11013
Reviewed-by: Matt Sarett <msarett@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
We don't _really_ need the Android NDK. We just need <arm_neon.h>
(which comes from Clang, not the NDK) and a smattering of <stdint.h>
([u]intN_t), <string.h> (memcpy) and <stddef.h> (size_t).
The idea here is solely to make it easier to run build_stages.py.
If this becomes a pain to maintain, I'm happy to go back to the NDK.
Change-Id: Ic6bb287646b6160ac42ac6e4d5290a66a7e92425
Reviewed-on: https://skia-review.googlesource.com/10980
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
