No work needs to be done in Metal for cross-context textures. Metal
automatically places a barrier between different command buffer
encoders, and Skia and the external client will implicitly be using
different encoders.
Bug: skia:8243
Change-Id: I0c855e2d78aa8973f389108483184947f019342c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/256699
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
In order to be able to create GrProgramDescs pre-flush we would like as little reliance on the GrRenderTarget as possible (also GrPixelConfig is going away).
Bug: skia:9455
Change-Id: Ie35c87758bb1ed97c99c6e9e130eaeb3355f4139
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/256457
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Stephen White <senorblanco@chromium.org>
Commit-Queue: Robert Phillips <robertphillips@google.com>
This reverts commit 3e7af41224.
Change-Id: Id4f66b3956f4bdbe690db20fc478b7365ee89717
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/256676
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
Auto-Submit: Mike Reed <reed@google.com>
To facilitate this, the GrQuadBuffer::Iter's local GrQuads that are
modified on each next() are now allowed to be operated on for the AA
inset/outsetting. Previously this required additional GrQuads on the
stack to hold the results, and additional guards for accessing localQuad()
when the entry didn't have actual coords.
With this change, a 2D op should have its device and src GrQuads' Ws
set to 1 once, and then they are completely ignored for all iteration
and tessellation, without any more redundant initialization. In all
likelihood we won't see the needle move on powerful platforms, but may
help lower end devices.
Change-Id: I457205786766403a760918e779d36ba056d69cde
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/256097
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
We need the wrapped GrVkSeocndaryCommandBuffer to live, holding refs to
GrVkResources until the GPU is done with the work. However, we don't know
ourselves when the command buffer gets submitted or when the GPU is done
since the client handles on that in the wrapped case. Our spec for wrapped
secondary command buffers requires to the client to keep the
GrSecondaryCommandBufferContext alive till the GPU is done the work so we
can use that to manage our needed lifetime as well. By putting the
GrVkSecondaryCommandBuffer on the GrVkRenderTarget, which is owned by the
SkGpuDevice in the above GrSCBContext, or lifetime should match what we need.
Change-Id: I7cdd92089bed390548ebf33f9d5c7e714bacc9c5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/256361
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
This reverts commit b1edfde279.
Reason for revert: senorblanco@ is adding a WriteOnly flag which will
be more appropriate for this functionality.
Original change's description:
> Rename GLRTFBOIDIs0 to WrapsSwapchainSurface and use for Metal.
>
> This provides a way to indicate that a Metal RenderTarget, and in
> particular a RenderTargetProxy, is framebufferOnly, so it can be
> restricted to certain operations.
>
> Bug: skia:9573
> TBR: bsalomon@google.com
> Change-Id: I733fae2fce402c375534889346255afe28a57944
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255783
> Commit-Queue: Jim Van Verth <jvanverth@google.com>
> Reviewed-by: Robert Phillips <robertphillips@google.com>
TBR=egdaniel@google.com,jvanverth@google.com,bsalomon@google.com,robertphillips@google.com
# Not skipping CQ checks because original CL landed > 1 day ago.
Bug: skia:9573
Change-Id: Ide775fbbe81114c10fe81e7faa517c20784e8498
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/256496
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Now that the UniformHandlers can accept SurfaceProxies directly, there's
no need to have the restricted type on TextureSampler anymore. This will
make them one step closer to compatibility with GrSurfaceProxyView.
While making these changes, several effects performed their own source
coord normalization by peeking the texture. I updated these to just
use the proxy's backing store dimensions.
Bug: skia:9556
Change-Id: I97183e8453b7a3db86cd90d614c8c959b3707abb
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255978
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
* Remove Run.
* Have SubRun handle both mask and path data.
* Collect up runs of glyph with the same format to
simplify building SubRuns.
The nexus of this change is GrTextBlob::makeSubRun. All the
invariants for a SubRun are set in this code, and those
invariants stem from the SubRunType. Much of the code has been
simplified by not having to maintain complicated invariant
state between adding glyphs.
The change that reduces the most number of problems with the
old code is having SubRun own its SkStrikeSpec. Having bulk
glyph insertion facilitated this.
Change-Id: I2439392a9ee3485e7b07a1f3fdaf1351128c4d47
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255309
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Jim Van Verth <jvanverth@google.com>
This is a reland of 997b37fdb9
Layout tests have been suppressed.
Original change's description:
> Revert "Revert "Make FP optimizations helpers use SkAlphaType not GrColorType""
>
> This reverts commit 078e8faa26.
>
> Change-Id: I67b2970584db5a1afbf9928b77c5444947ef71a3
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255897
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Commit-Queue: Brian Salomon <bsalomon@google.com>
Change-Id: I6db4cff8b71ef1f29ac47f7729b9514d009f730a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/256225
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
The bulkrect_1000_random_uniqueimages_batch on a Nexus 6P/Adreno 430
w/o this CL
curr/maxrss loops min median mean max stddev samples config
304/304 MB 1 151ms 159ms 158ms 163ms 3% ▆█▇▄▆▆▁▂█▅ gles
w/ this CL
curr/maxrss loops min median mean max stddev samples config
286/286 MB 1 18.1ms 18.1ms 18.1ms 18.1ms 0% ▂▄▅▃▅▅▃▄▁█ gles
Change-Id: I0f6d690b953444ec7a3176cb27c8a253caa55f5d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255986
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Replaces VisitSurfaceProxyFunc with GrOp::VisitProxyFunc since they are
now the same.
Records sampled textures as GrSurfaceProxies, but asserts they are in fact
texturable. Updates all backends to process the pipelines with surface
proxies (which always have had to do a surface->asTexture() virtual call
that is unchanged, this just avoids the unnecessary proxy virtual).
Also updates the GrTextureOp to not call asTextureProxy() anymore, to
take advantage of the relaxed pipeline types.
A number of instances of asTextureProxy() remain. It is still the primary
method of determining if a proxy is texturable, so code paths that branch
on that still use it. Some resource management code still requires mipmap
functionality specific to the GrTextureProxy API, so they still use it.
To keep this CL sane, the GrFP::TextureSampler and all generated FPs have
not been updated to use GrSurfaceProxy, although that will be in a follow
up since it should be technically possible.
Bug: skia:9556
Change-Id: I1e917aff9e70b08915e98ccc17c6b8f3be29c4f6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255830
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Change-Id: I2df94791af133fcfb3fa491163394369f324bbc3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/256096
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Add a common way to make rect op for testing that uses a GP with a local
matrix.
Change-Id: I958d1230bd5067b2e4b60fcd374e2f7718681e43
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255782
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Chrome is seeing unexplained segfaults when trying to access the n-1th
textureSampler while creating textOp GeoProcs -- it's possible that's
because n is 0. Also add checks to make sure n is not > kMaxTextures.
Bug: chromium:1027314
Change-Id: I65a5472980bab37d580b3f237a132a8b23794abe
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255985
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Change-Id: Ib29f41f6e71b176fec1ead26259ad1945a41e634
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/254677
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Change-Id: I3c8bf48dda061aa9d318b9f81f22608fdd68fcbd
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255786
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
(also renames CreateTextureSetOps to AddTextureSetOps, to match naming of
GrFillRectOp::AddFillRectOps).
Now that GrTextureOp can add more than one op to the GrRTC, it can take
over ownership of its fallback code for the texture set. It already had
taken over the code for non src-over blends when drawing a single texture.
Besides consolidating where the logic of converting TextureSetEntries into
op data lives, this makes the fallback case more consistent in terms of
performance. Previously, it would go through GrRTC::drawTexturedQuad,
which attempts to merge the clip with the draw for correctness reasons.
A batch never attempted these optimizations, so now even when one op per
quad is required, there won't be the overhead of comparing clips.
Change-Id: I30883e3bd45ed4386f81584e3d68229c46b17e47
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255781
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
This provides a way to indicate that a Metal RenderTarget, and in
particular a RenderTargetProxy, is framebufferOnly, so it can be
restricted to certain operations.
Bug: skia:9573
TBR: bsalomon@google.com
Change-Id: I733fae2fce402c375534889346255afe28a57944
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255783
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
The *previous* SkWuffsCodec.cpp commit, 39da10b5 "Optimize SkWuffsCodec
pixbuf zero-initialization", removed the need to *explicitly*
zero-initialize the fPixelBuffer on the first frame decode. Subsequent
frames, including the first frame of the animation again if the
animation is looped, still need to call sk_bzero.
Prior to *this* commit, each row of the pixel buffer was zeroed
separately. When it is necessary to memset to zero, it can be faster to
issue one big call instead of many small calls.
For example, the single-frame droids.gif file mentioned in the previous
commit (39da10b5) now skips the explicit sk_bzero calls, due to the
optimization in exactly that previous commit. But if it was a looped
animation, decoding subsequent (repeated) frames would need to memset
the re-used buffer to zero. The time taken for those sk_bzero calls
would be affected by this commit: from 2.900ms to 2.454ms (out of a
total decode time of 33.342ms as per 39da10b5's commit message).
Bug: skia:8235
Change-Id: I72ec3dc4230f06f358b5e8fee1b2799607d86615
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255358
Reviewed-by: Leon Scroggins <scroggo@google.com>
Commit-Queue: Leon Scroggins <scroggo@google.com>
This reverts commit 1792b19485.
Reason for revert: need to update legacy_convexity, still used by google3
Original change's description:
> Revert "Revert "Use flat version of path-direction enum""
>
> This reverts commit 0dacc6b7d3.
>
> Change-Id: Ie103e9f36b07e4ee256a3688a4decf3a6dd74314
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255832
> Auto-Submit: Mike Reed <reed@google.com>
> Reviewed-by: Mike Reed <reed@google.com>
> Commit-Queue: Mike Reed <reed@google.com>
TBR=reed@google.com
Change-Id: I0ecea0eb8a237298c6b908cc4bfd1cacdfc5b900
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255976
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
This reverts commit 0dacc6b7d3.
Change-Id: Ie103e9f36b07e4ee256a3688a4decf3a6dd74314
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255832
Auto-Submit: Mike Reed <reed@google.com>
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
This reverts commit e0fbe94351.
Reason for revert: need to add guard flag to flutter
Original change's description:
> Use flat version of path-direction enum
>
> Bug: skia:9663
> Change-Id: I00077d9f2b14b3e983e6a46ef6f560cabdb1678d
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/242557
> Commit-Queue: Mike Reed <reed@google.com>
> Reviewed-by: Florin Malita <fmalita@chromium.org>
TBR=fmalita@chromium.org,reed@google.com
Change-Id: If47173d9b203b2d3a175af290a15d986accb4703
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:9663
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255831
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Mike Reed <reed@google.com>
This adds a bunch of tests for ops that can all be evaluated directly in
skvm::Builder. You can see the sort of effect this has by looking at
the diffs for SkVMTest.expected... lots of `v3 = sub_f32 v2 v2`
transformed to `v3 = splat 0 (0)` and that sort of thing.
My favorite part is handling many assert_true() calls at compile time!
While the old inter-Op code parallels aren't as clear now, these new
early-out tests kind of work like comments explaining each op. I find
that nice. I found it hard to parse so many uses of the word "splat" so
I did go back to isImm() from isSplat(), and added allImm() to test for
and read several immediates all at once.
Some of this is less C++17 than I'd like. :/
Change-Id: Ie8187d5d184195e3c0c92d613508fb708c28302f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255814
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
OK this one is cool.
GrGLCaps::onSurfaceSupportsWritePixels() starts off like
bool GrGLCaps::onSurfaceSupportsWritePixels(const GrSurface* surface) const {
if (fDisallowTexSubImageForUnormConfigTexturesEverBoundToFBO) {
...
Clang compiles this into
53e5: cmpl $0x0,0x94(%rdi)
53ec: js 5435 <GrGLCaps::onSurfaceSupportsWritePixels(GrSurface const*) const+0x5d>
I think what's going on is that
fDisallowTexSubImageForUnormConfigTexturesEverBoundToFBO is the 32nd of
a bunch of neighboring bitfields, and that makes it the top bit of the 4
bytes at a 0x94 byte offset from GrGLCaps (rdi == this). The cmpl/js
combo is reading that whole 4-byte chunk of bitfields and asking "is
this negative", a fun and I'm sure efficient way to test that bit.
Now I bet Valgrind thinks all those bits need to be initialized to
answer that question, and I also bet GCC just slapped a big 4-byte 0
down onto all these bitfields, incidentally initializing the fields we
leave uninitialized. In any case I think it's healthy to initialize
these fields. Shame there aren't any bitfield initializers until C++20.
Think it's too early to upgrade?
Change-Id: I767515190cf1ed08cf1d6981e550a65ccaffdec5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255419
Reviewed-by: Ben Wagner aka dogben <benjaminwagner@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This is a reland of 190157bff2
Original change's description:
> Fixed some GrSkSLFP limitations.
>
> This makes it possible to pass more types as uniforms / inputs and fixes an
> issue with non-main functions calling other non-main functions.
>
> Change-Id: I6b5623b3c967d8219b992a455cc68bb0a29706b8
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255300
> Reviewed-by: Brian Osman <brianosman@google.com>
> Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Change-Id: Icc06cd54abb7acc83415915e16ca64c3eb6b943b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255779
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Change-Id: Ibdbd9b56e8848c4efd842fa658c6d76be4523043
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255580
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
As the first real op to be updated this required the plumbing of the DstProxyView and the de-constification of the GrAppliedClip.
Bug: skia:9455
Change-Id: Ieaa16adfa1d114021b2ad60d0ef8ecbe31d9cc82
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255136
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
The index API are no longer needed now that the Geometries
hold pointers instead of indexes.
Change-Id: I940520816ca58959f2c7c415a8f356ea5bf3d293
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255816
Commit-Queue: Herb Derby <herb@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
Still reasoning through that bitfield one, but this is unrelated, and
exactly like the other version. Clang may decide to branch on `0 ==
strcmp(kChromium, suffix)` before it branches on the result of `3 ==
sscanf(..., suffix)` that only has initialized suffix when it is true.
Change-Id: If76e7b9087b59ddae4249d9f2828c881fa649c8c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255418
Reviewed-by: Mike Klein <mtklein@google.com>
Reviewed-by: Ben Wagner aka dogben <benjaminwagner@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Auto-Submit: Mike Klein <mtklein@google.com>
- BUILD.gn `gif` optional source_set: use new skia_libgifcodec_path
code, not third_party/gif code.
- SkCodec.cpp: use third_party version of SkGifCodec.h.
- rm src/codec/SkGifCodec.* third_party/gif/*
Original CL: https://review.skia.org/254582
Cq-Include-Trybots: luci.skia.skia.primary:Build-Debian9-Clang-x86-devrel-Android_SKQP
Bug: skia:9654
Change-Id: Ia0379af70a5ec5656a7d123b8b6816b9fa24eaee
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255780
Reviewed-by: Hal Canary <halcanary@google.com>
Commit-Queue: Hal Canary <halcanary@google.com>
I started adding more _imm Ops like mul_f32_imm and noticed that there
was some low hanging peephole fruit here that we can harvest first.
Each of these peepholes I've added is sound in the way that I've
documented, and the div_f32 and mul_f32 peepholes are actively being
hit. I think it's inevitable that the add_f32 and sub_f32 ones will be
hit eventually too as we expand how skvm is used.
The easiest place to see this in action is
when SkColorFilter_Matrix::program() is passed an opaque color:
skvm::F32 invA = p->div(p->splat(1.0f), *a)
now leaves invA as splat(1.0f) directly, skipping the premul divide.
We still do the next part checking to see if invA is infinity; more
on that at the end of the CL description.
In this way I think we ought to never need to thread through a
shader_is_opaque parameter to color filters, instead letting the program
builder simply look and see if it is, among any number of other similar
optimizations.
Interestingly, though we may know sometimes the input to
SkColorFilter_Matrix::program() is opaque, as written today we'll never
know if the output is. Since we don't specialize the code based on the
0/1 values of the matrix, we lose the knowledge that alpha is 1.0f as it
goes through the matrix. There are two good ways to fix this:
1) do specialize on 0/1 values of the matrix so that the program
sees alpha is not changed when alpha is not changed;
2) wrap today's virtual program() with a base-class non-virtual
that queries getFlags() & kAlphaUnchanged_Flag to save and restore
the input alpha, essentially marking any code that changes alpha
as dead code.
2) is kind of the brute force, trusting version of 1), but does have
the advantage that the generated code need not change. Still, in this
case of a matrix, I think we'll want to look at 0/1 values anyway...
they'll come up more often than just for the alpha channel.
You can see a mul_f32 peephole happen when the blitter goes to store any
known-opaque color to memory. We go through some logic like
alpha_as_byte = round(mul(alpha, splat(255.0f)))
and if alpha is known to be splat(1.0f), we'll now skip that mul(),
alpha_as_byte = round(splat(255.0f))
I think this is all the strictly viable float peepholes where one
argument is a constant. Obviously there are lots of peepholes we can
write for int32, int16, and bitwise instructions, and then there's a
whole untapped world of peepholes to explore when _all_ arguments are
constant. These all-constant peepholes let our program notice that,
e.g., 1.0f < infinity and we can skip that part of the unpremul too, or
that round(splat(255.0f)) == 0xff and we can skip that work.
These all-constant peepholes may not be super important, as anywhere
they can trigger, the instruction must be hoistable: since all arguments
are constant, none depend on loop variables. But still, nicer to run
once ever at compile time than once per invocation at runtime. And it's
less code to analzye, less code to JIT, fewer instructions to interpret,
etc.
Change-Id: Ia2dc5af2cfff71a12693a2903f579a57c9302d12
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255616
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Something is breaking client code in unexpected ways. Let's try this
piecemeal.
- skia_libgifcodec_path gn var. Default is
third_party/externals/libgifcodec. Anfroid framework uses
third_party/libgifcodec.
- skia_use_libgifcodec gn var; defaults to !skia_use_wuffs. Allow
disabling both codecs.
- GN template("optional"): add public_include_dirs, similar to
public_defines.
- SK_USE_LIBGIFCODEC, ifdef for skia_use_libgifcodec.
- SkCodec.cpp: use SK_USE_LIBGIFCODEC
Original CL: https://review.skia.org/254582
Cq-Include-Trybots: luci.skia.skia.primary:Build-Debian9-Clang-x86-devrel-Android_SKQP
Bug: skia:9654
Change-Id: I70561f5f945656065e4b318d5fff505961fb0b6b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255578
Reviewed-by: Hal Canary <halcanary@google.com>
Commit-Queue: Hal Canary <halcanary@google.com>
This reverts commit 190157bff2.
Reason for revert: regression in arithmode
Original change's description:
> Fixed some GrSkSLFP limitations.
>
> This makes it possible to pass more types as uniforms / inputs and fixes an
> issue with non-main functions calling other non-main functions.
>
> Change-Id: I6b5623b3c967d8219b992a455cc68bb0a29706b8
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255300
> Reviewed-by: Brian Osman <brianosman@google.com>
> Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
TBR=bsalomon@google.com,brianosman@google.com,ethannicholas@google.com
# Not skipping CQ checks because original CL landed > 1 day ago.
Change-Id: I85d1a10fab072b82f35c9caf2c27b07d380a7bf4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255776
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Bug: skia:9603
Change-Id: I8571150bb7efe2f127c00d1a06e77ec8fa905d8f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255526
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
The ability to have device paths was removed a while ago.
Remove the code for device paths.
Change-Id: I09418eaa464b040c7bb13eac2e167997363e49d1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255530
Reviewed-by: Ben Wagner <bungeman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
We support ops with up three arguments today, all three of which can be
registers, or one can be an immediate. Extend this to allow up to two
immediates, fitting the uniform?? ops much better than my old 16-bit
hack.
(The next thing on my mind is adding Op::extract_imm where both the
shift and now too the mask are known to be immediates.)
This refactor lets me straighten out the kinks of some ops, mainly the
uniform?? ops and extract. Things should generally now use x,y,z in
that order: most of the old immediates are now immy, and we use immz
only where we've already used the y register or immy immediate.
Small renaming isImm() -> isSplat(). I think this helps mutually
distinguish splat instructions, constants held by skvm instructions
(immy,immz), and constants as used by the JIT or assembler (still
generally `imm`).
Change-Id: I540a22a01459a39a499801362023ced7f2d6f49c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255538
Auto-Submit: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
I broke this adding VTune support.
Change-Id: I53399a8717c86cce9864a0fe96decec828c75737
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255617
Reviewed-by: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This CL is 100% plumbing. We ultimately want all GrPrimitiveProcessor-derived objects to not be refCounted. This will make several helper objects POD and, by putting them into an arena, will make managing their lifetime easier (e.g., for DDL prePreparing).
Note: the CCPR GrGeometryProcessor-derived classes only ever appear on the stack so aren't forced into arenas.
Bug: skia:9455
Change-Id: Ib9be503d2fbf8c2578642df93fc301156629829d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/255304
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
