This was the old way of handling color spaces.
This is now handled by chooseCacheFormat() / buildCacheInfo().
Bug: skia:
Change-Id: Ifab1ab25d547a00296af237578dd8befec9c86c6
Reviewed-on: https://skia-review.googlesource.com/13004
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Matt Sarett <msarett@google.com>
Snap input values to N / 255 before doing PM/UPM conversions, and add
a new round-to-nearest conversion pair. With the pre-snapping, that
pair is chosen by (almost?) every GPU we have. This now lets us round
trip perfectly on Mali GPUs (which were falling back to SW before).
Bug: skia:
Change-Id: I0cebf1382e1c829aedfef4ec1614d7fa980057b5
Reviewed-on: https://skia-review.googlesource.com/11520
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This CL reverts legacy destinations back to lerping bytes.
Lerping linear is only the correct behavior for gradients
drawing into colorspace-aware destinations.
As written we're lerping between de-sRGB'd colors but never
transforming the outputs back. That leaves us in a weird
halfway-right-is-worse-than-wrong spot for legacy.
Change-Id: I79b85552b6913649afd2414205cf57108a8b93c6
Reviewed-on: https://skia-review.googlesource.com/13064
Commit-Queue: Mike Klein <mtklein@chromium.org>
Reviewed-by: Florin Malita <fmalita@chromium.org>
Bug: skia:6423
NOTRY=true
Change-Id: I2193c49151359a174ffa071650e1dbd366c91b28
Reviewed-on: https://skia-review.googlesource.com/11783
Commit-Queue: Kevin Lubick <kjlubick@google.com>
Reviewed-by: Mike Klein <mtklein@chromium.org>
Bug: skia:
Change-Id: I3ae75d4edde1b3c8cc44413db74e5ab9789a9020
Reviewed-on: https://skia-review.googlesource.com/13065
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Eric Boren <borenet@google.com>
N=8 on that bot.
CQ_INCLUDE_TRYBOTS=skia.primary:Build-Ubuntu-Clang-x86_64-Release-Fast
Change-Id: If54ae800b50d9dffb9f983b23ff6f522657943b1
Reviewed-on: https://skia-review.googlesource.com/13061
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
Also remove some unused functions from GrBlend.h and related unit test.
Bug: skia:
Change-Id: Id8ad0057a02f65a9e19dc75e4b88709a762f4139
Reviewed-on: https://skia-review.googlesource.com/12623
Reviewed-by: Chris Dalton <csmartdalton@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Constructor changes the color type to N32 now, so this
code is never needed.
Bug: skia:
Change-Id: If464440fed799122fed810d870f5d6d9a932bf12
Reviewed-on: https://skia-review.googlesource.com/13000
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Matt Sarett <msarett@google.com>
Bug: skia:6473
Change-Id: I722059844bb0bc9d58347b4f6002ee62043b9e20
Reviewed-on: https://skia-review.googlesource.com/13006
Reviewed-by: Ravi Mistry <rmistry@google.com>
Commit-Queue: Eric Boren <borenet@google.com>
It doesn't create an actual dependency, but it does cause the
Release-Shared build to be added to the tasks DAG, where it would be
orphaned if not for the other housekeeper.
Bug: skia:
Change-Id: Ia2974728ec353c7f209db45474e572d83cacbda5
Reviewed-on: https://skia-review.googlesource.com/13003
Reviewed-by: Ravi Mistry <rmistry@google.com>
Commit-Queue: Eric Boren <borenet@google.com>
This will allow different subclasses to use different models for lifetime management.
GrXferProcessor moves to simple ref counting since they don't own GrGpuResources.
This also constifies GrXferProcessor factories.
Change-Id: I6bea0ea8de718874063224232f9da50887868b16
Reviewed-on: https://skia-review.googlesource.com/11792
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Plus two sample implementations. Will fill out the
rest if the model looks ok.
Bug: skia:
Change-Id: Ie24ecd0cbc36263220cb1e506f7812c582554e50
Reviewed-on: https://skia-review.googlesource.com/12100
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Matt Sarett <msarett@google.com>
When aspectRatio is not finite, insert() can fall into an infinite loop.
This happens if you pass SkRect::MakeLargest() to the factory as bounds.
BUG=skia:5974
Change-Id: Ibcc9e5c5943c718608d4c1448305f7b8f11413bc
Reviewed-on: https://skia-review.googlesource.com/11784
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
At one point I moved the contents of the recipes into modules so that
they could be shared between repos. It turns out that we don't need
that, and it adds complexity.
Bug: skia:6473
Change-Id: I75a920b6a8474dcdd8b37ee9edd52aac801d1ab0
Reviewed-on: https://skia-review.googlesource.com/12622
Reviewed-by: Ravi Mistry <rmistry@google.com>
Commit-Queue: Eric Boren <borenet@google.com>
Change-Id: I118fcd49990597d4dfea92efd3f9d99e52fdbfab
Reviewed-on: https://skia-review.googlesource.com/11481
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Bug: skia:6473
Change-Id: Id78f0b5cb50ae2243b9a5aed8e552f7ed8202228
Reviewed-on: https://skia-review.googlesource.com/12060
Commit-Queue: Eric Boren <borenet@google.com>
Reviewed-by: Ravi Mistry <rmistry@google.com>
By fixing this some ugly casts are avoided in setting up Chrome's GrGLInterface
Change-Id: Ie540fc41b3cbb27f976d784cad90ae10c2cef9ed
Reviewed-on: https://skia-review.googlesource.com/12621
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
DirectWrite can now produce system font typefaces which are variations,
so mark these as multiple masters so printing knows what to do.
BUG=chromium:697916
Change-Id: Idf09ebba3c7002a09ff2e4a2dbae13dbce4e79d4
Reviewed-on: https://skia-review.googlesource.com/12101
Commit-Queue: Ben Wagner <bungeman@google.com>
Reviewed-by: Hal Canary <halcanary@google.com>
This gives compilers a strong hint that SkColor constants to be computed
at compile time.
Change-Id: I09c8a7fd23e4488f8b101f53d67a155f64d5d423
Reviewed-on: https://skia-review.googlesource.com/12056
Commit-Queue: Lei Zhang <thestig@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
Reviewed-by: Mike Klein <mtklein@chromium.org>
It should be allowed for the client to upload original data to a texture via
writePixels and then we just regenerate the mipmaps. I think it also resonable
to limit this to either writting to all levels or just the top level.
Bug: skia:
Change-Id: I66943cca54c2a7187a781788653948fb69c17c68
Reviewed-on: https://skia-review.googlesource.com/11798
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
We were just combing through possible changes that might have affected
the attached bug, this popped out as pretty obviously unsafe.
This doesn't explain the Chrome bug...
SkAutoTMalloc and SKAutoSTMalloc are separate types. :(
The new test fails and crashes before, passes after.
Change-Id: I033f488a7f644b7a70e612c8535fedfac35c76db
Reviewed-on: https://skia-review.googlesource.com/11797
Commit-Queue: Mike Klein <mtklein@chromium.org>
Reviewed-by: Herb Derby <herb@google.com>
Factors out a function F from_byte(U8) too.
Change-Id: Ib739ccbd509ddf25d2bfb7751ba6eaf51b16c12f
Reviewed-on: https://skia-review.googlesource.com/11791
Reviewed-by: Matt Sarett <msarett@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
Motivation: We may want to make SkMultiPictureDocument.h public in the
future.
Change-Id: Ie97b88d51a179c2283155d65bcadee32178115ca
Reviewed-on: https://skia-review.googlesource.com/11402
Commit-Queue: Hal Canary <halcanary@google.com>
Reviewed-by: Herb Derby <herb@google.com>
If xOffset or yOffset was not zero, these computations would not
be evaluated anyway, so no need to add zero for these comparisons.
(Split off from CL 9810)
Change-Id: I8481eb257d96766696945cef4eafaff4281fb7d9
Reviewed-on: https://skia-review.googlesource.com/11782
Reviewed-by: Matt Sarett <msarett@google.com>
Commit-Queue: Leon Scroggins <scroggo@google.com>
Chrome's got threadsafe statics, Android too, Flutter too, and heck, we even have them on all our Windows builds. No one responded to my mail about them. Let's stop turning them off on not-Windows!
Change-Id: Iadf00336aa73df3f694ac019c9d54a3f2257a3bd
Reviewed-on: https://skia-review.googlesource.com/11683
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
Bug: skia:5813
Change-Id: I734c716d953091e84d30ad36c55a6c74727f376c
Reviewed-on: https://skia-review.googlesource.com/11781
Reviewed-by: Eric Boren <borenet@google.com>
Commit-Queue: Eric Boren <borenet@google.com>
Minor cleanup of left over Tab3 tests.
Bug: skia:6442
Change-Id: I03bb4a034101a2773619abd2be7a300a70f6ac3d
Reviewed-on: https://skia-review.googlesource.com/11414
Reviewed-by: Eric Boren <borenet@google.com>
BUG=chromium:538133
BUG=chromium:359065
Change-Id: I8208efc63579e0f1ebb77680beae2cd08cdba547
Reviewed-on: https://skia-review.googlesource.com/11681
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Hal Canary <halcanary@google.com>
I think this happens to fix the particular issue in the attached bug.
memcpy() is kind of the swiss army knife as far as strict aliasing is
concerned... you're always allowed to use it.
The generated code for writeToMemory() is unchanged, and
readFromMemory() gets a bit better looking.
BUG=skia:5105
Change-Id: Ib5bf96600f1138650c004ced2d696e9a4ba83ca7
Reviewed-on: https://skia-review.googlesource.com/11682
Reviewed-by: Cary Clark <caryclark@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
I don't think we've tried this since switching to GN,
where we decided to just go with jsimd_none.c on all x86 bots.
CQ_INCLUDE_TRYBOTS=skia.primary:Test-Ubuntu-Clang-GCE-CPU-AVX2-x86_64-Debug-MSAN,Perf-Ubuntu-Clang-GCE-CPU-AVX2-x86_64-Debug-MSAN
BUG=skia:4550
Change-Id: Iff4be3b640fb77d1094565a5db4d32778e7f2b4f
Reviewed-on: https://skia-review.googlesource.com/11680
Reviewed-by: Matt Sarett <msarett@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
It won't save as much time for faster machines, but it should still
shave a few seconds, and it'll be nice to be consistent.
Bug: skia:
Change-Id: I0b2aa010579b527c66325cb5b691bfb549b31316
Reviewed-on: https://skia-review.googlesource.com/11418
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Commit-Queue: Eric Boren <borenet@google.com>
Here we use 64-bit gather instructions for HSW,
which I think we haven't done before.
Change-Id: I7b22b3cc0b7a151952518bb9afb90624ebdb4a22
Reviewed-on: https://skia-review.googlesource.com/11602
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
Change-Id: Iefa8044bac0555c5fff370217a6270b4f3c64300
Reviewed-on: https://skia-review.googlesource.com/11582
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
This is all the gathers except index 8 and f16, which aren't conceptually hard
but I want to land separately.
Change-Id: I525f2496e55451041bd6ea07985858fda7b56a40
Reviewed-on: https://skia-review.googlesource.com/11524
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@chromium.org>
boxBlurInterp showed up hot when running Speedometer. I noticed that
BoxBlur calls boxBlurInterp repeatedly, the call does not get inlined,
and the function could be simplified a bit if 'transpose' is a
compile-time constant.
In particular, dst_x_stride or dst_y_stride now become a constant 1,
which simplifies the dptr updates.
Bug: skia:none
Change-Id: I8782e5b133a2195328e13878aca8a749af150b86
Reviewed-on: https://skia-review.googlesource.com/11362
Reviewed-by: Stephen White <senorblanco@chromium.org>
Reviewed-by: Mike Klein <mtklein@chromium.org>
Commit-Queue: Mike Klein <mtklein@chromium.org>
