This allows the iterator type/boilerplate to be reused for any other
data collection that sits above GrBlockAllocator, as long as its a fixed
"type" with indices into a block.
Also adds reverse iteration (which is useful for stack-like use cases).
Change-Id: Id9a205e8fb396a8558e360439240fd20c92c9700
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/302665
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
This is just minor little stuff I've been meaning to do,
with essentially no impact anywhere.
- Add an easy-to-flip switch to disable the JIT.
- Stop checking so carefully whether we hasJIT()
in test_jit_and_interpreter(). This was helpful
for making progress but now just gets in the way.
Change-Id: I08065ba1f42700f9d7d63f8303af357ec5fe11ae
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/302944
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
With this change if a backend Gr*Gpu wants to using staging buffers
they just add a generic GrStagingBufferManager member object. This
object can be used to get slices of upload buffers. Then they just need
to implement the virtual for taking ownership of buffers during submit.
We rely on our GrResourceCache to handle caching and reuse of these
buffers.
This change allows us to remove all other virtuals on GrGpu around
managing staging buffers.
Change-Id: I5db9a3c52133978ea89d6c0de440f434fbf91a51
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/300226
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Stephen White <senorblanco@google.com>
This reverts commit a56da9ee92.
Reason for revert: UBSAN complains in Vulkan OOPRDDL mode
Original change's description:
> Add a direct context arg to makeColorTypeAndColorSpace
>
> This is part of a larger effort to de-power SkImage and force users to
> specify the GPU context for all new images.
>
> Staging flag landed in Chrome CL 2296632.
>
> Bug: skia:10466
> Change-Id: I6b7bbec10369f7d8ee884dd1bcc234d332c30a6c
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/302290
> Commit-Queue: Adlai Holler <adlai@google.com>
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Reviewed-by: Robert Phillips <robertphillips@google.com>
TBR=bsalomon@google.com,robertphillips@google.com,adlai@google.com
Change-Id: Ide36bed6966d3d92ad6b8d05f897d22d287b40b1
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:10466
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/302824
Reviewed-by: Adlai Holler <adlai@google.com>
Commit-Queue: Adlai Holler <adlai@google.com>
This is part of a larger effort to de-power SkImage and force users to
specify the GPU context for all new images.
Staging flag landed in Chrome CL 2296632.
Bug: skia:10466
Change-Id: I6b7bbec10369f7d8ee884dd1bcc234d332c30a6c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/302290
Commit-Queue: Adlai Holler <adlai@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
C++ algorithms have largely standardized on a [begin, end) half-open
range, as seen in standard library containers. SkTQSort now adheres to
this model, and takes vec.begin() and vec.end() as its inputs.
To avoid confusion between inclusive and half-open ranges inside the
implementation, internal helper functions now take "left" and "count"
arguments instead of "left"/"right" or "begin"/"end". This avoids any
ambiguity.
(Although performance was not the main goal, this CL appears to
slightly improve our sorting benchmark on my machine.)
Change-Id: I5e96b6730be96cf23d001ee0915c69764b2c024a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/302579
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
We now rely on GrTextureEffect and GrBicbicEffect
to determine if shader-based tiling is required.
GrTextureProducer is still responsible for noticing that
the proxy is approximate because GrTextureEffect doesn't
consider that in its basic Make() factory (as opposed to
MakeSubset()).
Change-Id: I8e1aeb9edbcfa73ea0bf80b5256ee1ca21fe9c81
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/301985
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
This simplifies things like ConstantOutputForConstantInput and
invokeChild. It also removes the need for child indices: generated
FPs now directly refer to their children by slot number.
Change-Id: I69bbb042d5d72d21b999256f969c467702d0774d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/302436
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
Change-Id: If51be35205b40c4a22979a4b49b031126af1dde7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/302500
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
In order to stage the transition from GrContext to GrDirectContext, both
of them will have to have the factories for a while.
This CL also removes all internal uses of the old (GrContext) factories.
Change-Id: Ibe1edd0818ea23a0d54257c55f35f12526047ef3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/302263
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Adlai Holler <adlai@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
This reverts commit 70474c1cb0.
Reason for revert: bot build failure
Original change's description:
> Remove custom SkSort algorithms.
>
> SortBench shows that SkTQSort and SkTHeapSort are inferior to std::sort.
> The difference is small on randomized inputs, but quite significant for
> semi-ordered inputs (forward/backward/repeated). There doesn't seem to
> to be any compelling advantage to SkTQSort.
>
> Nanobench results: https://screenshot.googleplex.com/9JOLV1d6Z0u
>
> (These performance numbers are from an optimized build my local machine;
> it's possible that we might see different results on the test bots.)
>
> Change-Id: Iaf19563041547eae7de2953be249129108f093b1
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/302295
> Commit-Queue: John Stiles <johnstiles@google.com>
> Reviewed-by: Mike Klein <mtklein@google.com>
TBR=mtklein@google.com,brianosman@google.com,johnstiles@google.com
Change-Id: I1126dd4cda95716dac225ad32d5b0e5cf3f09421
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/302447
Reviewed-by: John Stiles <johnstiles@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
SortBench shows that SkTQSort and SkTHeapSort are inferior to std::sort.
The difference is small on randomized inputs, but quite significant for
semi-ordered inputs (forward/backward/repeated). There doesn't seem to
to be any compelling advantage to SkTQSort.
Nanobench results: https://screenshot.googleplex.com/9JOLV1d6Z0u
(These performance numbers are from an optimized build my local machine;
it's possible that we might see different results on the test bots.)
Change-Id: Iaf19563041547eae7de2953be249129108f093b1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/302295
Commit-Queue: John Stiles <johnstiles@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
ConstColorProcessor contained three separate InputModes with their own
unique behaviors, but every (non-test) call site simply hardcoded one of
the InputModes.
This change also allows the actual const-color processor to remove the
inputFP entirely; it is never sampled.
The GM slide has been split into three separate slides as well.
Change-Id: I2b77f4eab4d655f06e3704fb6fde8d4f8c70a075
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/301987
Commit-Queue: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
This is part of a larger effort to force users to provide a context
when manipulating GPU images which may be shared, instead of having
images themselves retain powerful context pointers.
Chrome flag landed in Chrome CL 2292800
Bug: skia:10466
Change-Id: Ic530a2c5eb1f4399db899d243ea944760fdf2055
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/300707
Commit-Queue: Adlai Holler <adlai@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
While testing on Linux with a configuration like
skia_enable_fontmgr_custom_directory=false
skia_enable_fontmgr_custom_embedded=false
skia_enable_fontmgr_custom_empty=true
skia_enable_gpu=true
skia_use_fontconfig=false
skia_use_freetype=true
skia_use_system_freetype2=false
the default typeface will be an empty typeface with no glyphs. This of
course leads to many test failures, which is fine.
However, this also leads to crashes when testing GPU Ops since the Op
factories may return nullptr to indicate no-op but the callers of those
factories currently do not expect nullptr or handle it as a no-op.
Change the callers of Op factories to treat nullptr as no-op.
Change-Id: I9eb1dfca4a8a9066a9cfb4c902d1f52d07763667
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/301586
Reviewed-by: Herb Derby <herb@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
This makes batch texture ops consistent with singleton texture ops, and
images drawn through a texture producer.
Bug: chromium:1102578
Change-Id: I490b20940ef6f1899396b786369271ce7130e8a9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/301540
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Change-Id: I9fa9f8785f48e884cf296a638347003d1687e7c0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/301536
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Brian Salomon <bsalomon@google.com>
Auto-Submit: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Change-Id: Ice75d0caa7b4beb2d982be094d62b54e71b45045
Bug: chromium:1101491
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/301456
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
The SkPDFDocument::fTagTree::fRoot is set at document creation time.
This root will initially always be discardable, since no annotations
have been made. As a result it does not make sense to assert on it being
non-discardable, since it should be possible to create a PDF which just
happens to have no annotations added to it.
Change-Id: I2fe336c872805b6937f7c8ea275c0cb4d3438682
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/301383
Commit-Queue: Ben Wagner <bungeman@google.com>
Reviewed-by: Dominic Mazzoni <dmazzoni@chromium.org>
We are replacing GrContext with the GrDirectContext/GrRecordingContext
pair. This starts making that change visible to clients (and weaning
Skia off of GrContext).
Change-Id: I00cc9bf208499984de855a1646229bd7557fc925
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/300706
Reviewed-by: Adlai Holler <adlai@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Bug: skia:10416
Change-Id: I0ca7535a0e6507e6b2a9f4682788826972c5f3b8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/300296
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This reverts commit c8b721b086.
Reason for revert: Looks like it's causing build failures around
MakeRenderTargetContext on the roll.
Original change's description:
> Make SkGpuDevice hold a GrRecordingContext
>
> This makes the code reflect what is actually going on. During DDL
> recording the SkGpuDevice only holds a recording context.
>
> This can't land until the following Chrome-side CL lands:
>
> https://chromium-review.googlesource.com/c/chromium/src/+/2277964 (Add GrContext.h include to skia renderer for upcoming Skia roll)
>
> Change-Id: I69cfa744226c315c25f68fc509b7b59ec38bbf31
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/299867
> Commit-Queue: Robert Phillips <robertphillips@google.com>
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Reviewed-by: Adlai Holler <adlai@google.com>
TBR=bsalomon@google.com,robertphillips@google.com,adlai@google.com
Change-Id: I6a362daf7c40e36ed9f068c5b2d477c16a3f778e
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/300853
Reviewed-by: Chris Dalton <csmartdalton@google.com>
Commit-Queue: Chris Dalton <csmartdalton@google.com>
This makes the code reflect what is actually going on. During DDL
recording the SkGpuDevice only holds a recording context.
This can't land until the following Chrome-side CL lands:
https://chromium-review.googlesource.com/c/chromium/src/+/2277964 (Add GrContext.h include to skia renderer for upcoming Skia roll)
Change-Id: I69cfa744226c315c25f68fc509b7b59ec38bbf31
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/299867
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Adlai Holler <adlai@google.com>
defines one method that is never used.
Change-Id: If8522c5f1ac7447b0d5584e76cdbd2f7d127036b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/300259
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Auto-Submit: Brian Salomon <bsalomon@google.com>
All coord transforms were identity, so this enshrines that
knowledge, then transitively removes a large amount of code.
Bug: skia:10416
Change-Id: Iae4af9ca21590bced1ce9fce3ab807f6cceaebd4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/300234
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This CL makes it explicit that the unit tests always get a direct context.
It is mainly a mechanical CL.
Change-Id: I49e0628851d9c81eb47386ef978edf905c6469d8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/299866
Reviewed-by: Adlai Holler <adlai@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Previously, we had no unit test coverage for the Metal `struct Globals`
object. This CL adds simple tests for numeric globals and samplers.
Change-Id: I259c3cf416d5a1a03b1f815cc4c03891d60cbd08
Bug: skia:10382
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/300616
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
GM was updated in:
https://skia-review.googlesource.com/c/skia/+/300172 (Make GM::onGpuSetup take a GrDirectContext)
This CL updates: skpbench, nanobench, and some testing infrastructure.
Only minor changes were made to the unit tests as they will be updated
en masse in a follow up cl.
Change-Id: Ieffc98865d4c9fc73e292d3c807ed4ae2081745a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/300220
Reviewed-by: Adlai Holler <adlai@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
External clients will need access to these classes once GrContext
goes away.
This is a purely mechanical CL.
Bug: skia:10441
Change-Id: I7ffeb29d88bcc0f012412fba911e8362d046e24a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/300206
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
It now tracks all sample calls of a child (matrix, explicit coords,
pass through). There is now just one registerChild() call, and the
sampling pattern of that child is fully determined by the SampleUsage
parameter.
Change-Id: Iaadcd325fca64a59f24192aadd06923c66362181
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/299875
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
External clients will eventually have to call this to get access
to a direct context from an SkCanvas or SkSurface (which will
only have 'recordingContext' accessors).
Bug: skia:10441
Change-Id: I10e34081277b685fa59d03e1fce1887f3524e0fa
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/300178
Reviewed-by: Adlai Holler <adlai@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
Just a little follow up, adding a mem->xmm vmovups
instruction to make it possible. Nothing tricky.
Change-Id: I319e11839e44ccda46e664c82fb858a18499f9be
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/299883
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Herb Derby <herb@google.com>
This is a reland of 9716414e93
Original change's description:
> Simplify GrClip API
>
> Removes quickContains(SkRect), quickContains(SkRRect), and isRRect().
> Replaces these three functions with preApply() that conservatively
> determines the clip effect up to a single rrect intersection. The major
> motivation for this is the new GrClipStack implementation. preApply()
> and apply() will be able to reuse much more code compared to separating
> the preApply functionality across the older three functions that were
> removed. Additionally, preApply is able to convey more information for
> less work, since it can usually determine being skipped or unclipped while
> determining if the clip is a single rrect.
>
> As part of using this API, the attemptQuadOptimiziation and the equivalent
> rrect optimization are overhauled. Hopefully legibility is improved, and
> the rrect case is now applied outside of the android framework (but with
> tighter AA requirements).
>
> Bug: skia:10205
> Change-Id: I33249dd75a28a611495f87b211cb7ec74ebb7ba4
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/298506
> Reviewed-by: Brian Salomon <bsalomon@google.com>
> Reviewed-by: Chris Dalton <csmartdalton@google.com>
> Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Bug: skia:10205, 10456
Change-Id: I500eeda36ea50e95eb8cb658b36aa2373d5166c0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/298823
Reviewed-by: Chris Dalton <csmartdalton@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
This also adds back default flush() calls which simply do a flush
without any submit.
Change-Id: Ia8c92bbdecd515d871abfa6364592f502e98656b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/298818
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
Previously, we had separate "ComposeOne" and "ComposeTwo" fragment
processors. These performed extremely similar roles, but varied in
the number of child FPs used, and treatment of the input color's alpha
channel.
This CL combines these two FPs into a single unified "Compose" fragment
processor. To avoid breaking many existing GMs, the semantics of the
prior FPs have been preserved as closely as possible. (Specifically, the
FP treads very carefully to ensure that the alpha channel handling
matches the old code, as dozens of GMs fail otherwise.)
Change-Id: I7ab453f3313e70ae25e5e70f86373a64ff02072f
Bug: skia:10217
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/299703
Commit-Queue: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
isConstant really means "has a known value at compile time", but for
variables declared with "const", the two meanings had been conflated.
There were several ways for this to produce surprising error messages.
The included unit test crashed before removing the override, and now
passes.
Change-Id: I49b926e51c421db93240cbbf42231de0444358d6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/299860
Reviewed-by: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
This removes the code generation support for @coordTransform and sk_TransformedCoords2D
when processing .fp files. Instead, the main() function can optionally add a float2
parameter. This is marked as the SK_MAIN_COORDS builtin, just like the main function
for a runtime pipeline stage.
Bug: skia:10416
Change-Id: I0c192d890bb798a1167bc445003f6ddffe6118f0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/299687
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
This reverts commit ca5b36c474.
Reason for revert: This may be blocking the Chrome roll
Original change's description:
> Add storage on the surface for its last render task
>
> Let's land this and see if it gets us back to baseline on the
> lastRenderTask regression from the bug. If it doesn't, we'll revert it
> since the extra complexity won't have been worth it.
>
> Bug: skia:10372
> Change-Id: I9d5ae93435b833d575afdc7f219dc8e7c453c92b
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/297836
> Reviewed-by: Robert Phillips <robertphillips@google.com>
> Commit-Queue: Adlai Holler <adlai@google.com>
TBR=robertphillips@google.com,adlai@google.com
Change-Id: Id418d042d1123d946cd99b7b1ba438211cb628ec
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:10372
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/299763
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
I will follow this up with a Chrome-side CL to fix the call sites of the following factories:
MakeFromCompressedTexture
MakeFromTexture
MakeFromYUVATexturesCopyWithExternalBackend
MakeFromYUVTexturesCopyWithExternalBackend
MakeFromNV12TexturesCopyWithExternalBackend
Here is the Chrome-side CL: https://chromium-review.googlesource.com/c/chromium/src/+/2264598/ (Skia's SkImage::Make* factories now guarantee cleanup)
Here is the Chrome-side CL that adds the guard flag:
https://chromium-review.googlesource.com/c/chromium/src/+/2273067/ (Add flag in order to roll a Skia CL into Chrome)
TBR=bsalomon@google.com
Bug: 1097484
Change-Id: Ic2fcdc116f0f866b33d752b6d5abc784c7f65be6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/299663
Reviewed-by: Robert Phillips <robertphillips@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
