Json ide script for gn to output cmake.

Can be run like

gn gen out/debug --ide=json --json-ide-script=../../gn/gn_to_cmake.py

or

gn gen out/debug --ide=json
python gn/gn_to_cmake.py out/debug/project.json

This first pass creates CMakeLists.txt with just enough information for an
ide. Each target is just a static library with sources, include
directories, and defines.

NOTRY=true
This isn't automatically tested.

Review-Url: https://codereview.chromium.org/2363873002

Change SkSpecialImage::makeSurface and makeTightSurface to take output
properties (color space), bounds, and (optional) alphaType.

We were being pretty inconsistent before. Raster was honoring all
components of the info. GPU was using the supplied color type, but
propagating the source's color space. All call sites were saying N32.

What we want to do is propagate the original device's color space, and
pick a good format from that. Rather than force all the clients to
jump through hoops constructing an SkImageInfo that meets our criteria,
just have them supply the few bits we care about, and do everything else
internally.

This also lets us always use RGBA on GPU, but N32 on raster.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2349373004

Review-Url: https://codereview.chromium.org/2349373004

Make GrRenderableConfigForColorSpace argument const

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2367813002

Review-Url: https://codereview.chromium.org/2367813002

Add swarming task for nanobench upload

BUG=skia:5719
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2364693003

Review-Url: https://codereview.chromium.org/2364693003

split tight quads and conics

Tight quads and conics may nearly fold over on themselves, confusing
coincidence against other curves. Split them at their max curvature
early on to avoid complicating later logic.

TBR=reed@google.com
BUG=skia:5131
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2357353002

Review-Url: https://codereview.chromium.org/2357353002

sRGB cleanup in image filters. Ensure we configure paints correctly.

Net effect of the two calls is (basically) the same, but given that we're
propagating from isGammaCorrect on the DC, I figured it makes sense to use
the same-named API.

More importantly, a couple places had slipped through, so those draws would
ignore sRGB-ness of inputs entirely.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2357413004

Review-Url: https://codereview.chromium.org/2357413004

It is possible to try and draw to an unsupported format. Let this go.

Squelches assert firing in gradient code on low-end Android bots.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2365703002

Review-Url: https://codereview.chromium.org/2365703002

SkPDF: do not assert when openStream return nullptr

BUG=163892
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2332813002

Review-Url: https://codereview.chromium.org/2332813002

document.close from bool to void

Nothing meaningful is returned, it is inconsistent with the bulk of the rest of skia's api. The C api is waiting on this change as well.

Required chrome CL: https://codereview.chromium.org/2355343003/

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2354403002

Review-Url: https://codereview.chromium.org/2354403002

Revert of Some Vulkan memory fixes and cleanup (patchset #2 id:20001 of https://codereview.chromium.org/2356343003/ )

Reason for revert:
fHeapIndex is not used in release, need to hide behind SK_DEBUG. Failing on Perf-Android-Clang-NVIDIA_Shield-GPU-TegraX1-arm64-Release-GN_Android_Vulkan.

Original issue's description:
> Some Vulkan memory fixes and cleanup
>
> * Switch back to not setting transfer_dst on all buffers
> * Add some missing unit tests
> * Add tracking of heap usage for debugging purposes
> * Fall back to non-device-local memory if device-local allocation fails
>
> BUG=skia:5031
> GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2356343003
>
> Committed: https://skia.googlesource.com/skia/+/c5850e9fdb62cc4ae5ed2b6af51aea92cac07455

TBR=egdaniel@google.com,brianosman@google.com
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:5031

Review-Url: https://codereview.chromium.org/2358123004

G instead of R, thanks Gold

NOTRY=true
TBR=mtklein@google.com
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2362863002

Review-Url: https://codereview.chromium.org/2362863002

don't close a contour with a line if nothing's there

R=kjlubick@google.com
BUG=skia:5789
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2360083006

Review-Url: https://codereview.chromium.org/2360083006

Some Vulkan memory fixes and cleanup

* Switch back to not setting transfer_dst on all buffers
* Add some missing unit tests
* Add tracking of heap usage for debugging purposes
* Fall back to non-device-local memory if device-local allocation fails

BUG=skia:5031
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2356343003

Review-Url: https://codereview.chromium.org/2356343003

Update Chrome section to include Skia roll/sheriff/bug info

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2351593003
NOTRY=true
DOCS_PREVIEW= https://skia.org/?cl=2351593003

Review-Url: https://codereview.chromium.org/2351593003

Make AALinearizingConvexPathRenderer able to handle stroke and fill

This is intended to catch stroke-and-fill convex paths with potentially small stroke widths (e.g., .1).

It does have the disconcerting side effect of changing bevel-joined stroke-and-filled degenerate-single-line-rects into plain rects (w/o triangular end-caps).

GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2301353004

Committed: https://skia.googlesource.com/skia/+/522bcd99fa65a8abd130880f59b500cf367d0845
Review-Url: https://codereview.chromium.org/2301353004

Change implementation of flush-count based GrGpuResource purging

Change default to approx 30seconds (given some API usage assumptions)
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2361093002

Review-Url: https://codereview.chromium.org/2361093002

skpbench fixes for linux

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2354273004

Review-Url: https://codereview.chromium.org/2354273004

Add BGRA as input format to SkColorSpaceXform

This is immediately useful for webp and I think it's a fair guess
that BGRA src formats are not uncommon.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2353363008

Review-Url: https://codereview.chromium.org/2353363008

Remove assert that current color is never used.

The COLR table format allows for use of the 'current color'.
Currently the code asserts if this is used, but will use a nearby
luminance grey to this color instead if the assert is ignored.
Remove the assert now that this path is known to be used.

BUG=skia:5788

Review-Url: https://codereview.chromium.org/2359923004

Support for color-spaces with multi-stop (texture) gradients

Texture is F16 linear, unless that's not supported. In that
case, we pack down to sRGB.

Added more test patches to the gamut GM with many stops,
to test this case. Now they render correctly.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2343253002

Review-Url: https://codereview.chromium.org/2343253002

GN: build skiaserve

I trimmed the libmicrohttpd sources and defines down to the minimum needed to build and run.  This builds and runs on Linux and Android for me.

Request.h was missing an include for SkTypes.h, which supplies the default for SK_GPU_SUPPORTED if not otherwise defined.

To build on Android, exit() -> _exit().

build.py was unused.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2367513002

NOTREECHECKS=true

Review-Url: https://codereview.chromium.org/2367513002

Reland of ix for conic fuzz (patchset #1 id:1 of https://codereview.chromium.org/2361473004/ )

Reason for revert:
Landed suppression in Chrome's LayoutTests/TestExpectations

Original issue's description:
> Revert of fix for conic fuzz (patchset #3 id:40001 of https://codereview.chromium.org/2350263003/ )
>
> Reason for revert:
> See if this fixes the layout tests.
>
> Original issue's description:
> > fix for conic fuzz
> >
> > A fuzzer generates a conic that hangs when drawn.
> > The quads that approximate the conics move up and down
> > in y, confusing the renderer.
> >
> > This fix ensures that the split conic maintains the
> > same y direction as the original conic.
> >
> > R=reed@google.com
> > BUG=647922
> > GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2350263003
> >
> > Committed: https://skia.googlesource.com/skia/+/ac78863acdef4b428aaf66985b80c76d1be0fdea
>
> TBR=reed@google.com
> # Skipping CQ checks because original CL landed less than 1 days ago.
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=647922
>
> Committed: https://skia.googlesource.com/skia/+/08b345588414b861af8a55950e7dc21a1bd85a28

TBR=reed@google.com
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=647922

Review-Url: https://codereview.chromium.org/2359253002

Revert of fix for conic fuzz (patchset #3 id:40001 of https://codereview.chromium.org/2350263003/ )

Reason for revert:
See if this fixes the layout tests.

Original issue's description:
> fix for conic fuzz
>
> A fuzzer generates a conic that hangs when drawn.
> The quads that approximate the conics move up and down
> in y, confusing the renderer.
>
> This fix ensures that the split conic maintains the
> same y direction as the original conic.
>
> R=reed@google.com
> BUG=647922
> GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2350263003
>
> Committed: https://skia.googlesource.com/skia/+/ac78863acdef4b428aaf66985b80c76d1be0fdea

TBR=reed@google.com
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=647922

Review-Url: https://codereview.chromium.org/2361473004

Add output format properties to SkImageFilter::Context

For now, this is just the color space (of the original
requesting device). This is used when constructing
intermediate rendering surfaces, so that we ensure we
land in a surface that's similar/compatible to the
final consumer of the DAG's output.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2357273002

Review-Url: https://codereview.chromium.org/2357273002

Make SkColorSpaceXform::New() take bare ptrs

There's no need to take sk_sp if we're not going to
ref the ptr.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2360863003

Review-Url: https://codereview.chromium.org/2360863003

Run commandbuffer config on CommandBuffer Perf bots.

Uh, so hey, I just noticed that while the CommandBuffer Test bot runs
with --config commandbuffer, the Perf bot just runs as a vanilla GPU bot.
Shouldn't we pass --config commandbuffer?

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2364483002

CQ_INCLUDE_TRYBOTS=master.client.skia:Perf-Mac-Clang-MacMini6.2-GPU-HD4000-x86_64-Debug-CommandBuffer-Trybot

Review-Url: https://codereview.chromium.org/2364483002

Clean up dead code.

SkOSEnvironment was used to set --enable-gl-path-rendering, and is now unused.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2365483002

Review-Url: https://codereview.chromium.org/2365483002

Is char* or uintptr_t easier to read?

Using a char* instead of uintptr_t allows us to use fewer
reinterpret_casts which may make the code easier to read.

BUG=648512
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2356703002

Review-Url: https://codereview.chromium.org/2356703002

Add recipe support for Intel HD Graphics 530.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2358183002

Review-Url: https://codereview.chromium.org/2358183002

Do not quickReject until virtual drawPatch

TBR=reed@google.com
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2358873003

Review-Url: https://codereview.chromium.org/2358873003

fix next kevin fuzz

Fix one more fuzzer crash.

R=kjlubick@google.com
BUG=skia:5775
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2357373002

Review-Url: https://codereview.chromium.org/2357373002

fix for conic fuzz

A fuzzer generates a conic that hangs when drawn.
The quads that approximate the conics move up and down
in y, confusing the renderer.

This fix ensures that the split conic maintains the
same y direction as the original conic.

R=reed@google.com
BUG=647922
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2350263003

Review-Url: https://codereview.chromium.org/2350263003

speed up debug dm

SkPathMeasure can take minutes with pathological cases.
Limit the debug check to a reasonable number.

R=reed@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2356343004

Review-Url: https://codereview.chromium.org/2356343004

Add hardware monitoring to skpbench

Adds a Hardware class with hooks for entering and exiting
"benchmarking" mode (e.g. locking clocks, etc.) as well as periodic
polling of hardware to verify the environment is stable.

Adds a partial implementation for generic Android hardware, but
ultimately we will need to write specific classes tailored to each
unique platform we need to test.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2360473002

Review-Url: https://codereview.chromium.org/2360473002

Add Jobs to tasks.json

BUG=skia:5723
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2363443002

Review-Url: https://codereview.chromium.org/2363443002

GN: take over CommandBuffer bot

As you suspected, I see nothing preventing us from building our side of things unconditionally.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2358173002
NOTREECHECKS=true

Review-Url: https://codereview.chromium.org/2358173002

Use sk_careful_memcpy when writing optional conic weights for GrShape path data key.

TBR=robertphillips@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2359933002

NO_DEPENDENCY_CHECKS=true

Review-Url: https://codereview.chromium.org/2359933002

Add optional sw generated path coverage mask caching

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2335343008

All the bots except the painfully slow windows compiler have finished so,
NOTRY=true

Review-Url: https://codereview.chromium.org/2335343008

allow clip calls w/o op param, remove unnecessary kReplace ops

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2357333002

Review-Url: https://codereview.chromium.org/2357333002

Remove unused SkImage_Base and SkImage_Gpu onNewSurface methods
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2353403002

Review-Url: https://codereview.chromium.org/2353403002

Conditionally insert gl_PointSize into shaders.

BUG=chromium:648816
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2358843002

Review-Url: https://codereview.chromium.org/2358843002

Always use transfer_dst for buffers.

This will allow Adreno to use non-coherent buffers.

BUG=skia:5034
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2361473002

Review-Url: https://codereview.chromium.org/2361473002

Build tools on NoGPU bots.

I was just lazy disabling this before.  It turns out to not be difficult.

This gets us slightly better testing out of the NoGPU bot, as we now know
not only that Skia compiles, but also that DM and nanobench can link (and
thus that Skia's got enough compiled in it to be a coherent library).

skpbench requires GPU support.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2356283002

Review-Url: https://codereview.chromium.org/2356283002

Add a transient image filter cache to SkImage::makeWithFilter & PDF

Added a bench for makeWithFilter (~23 ms -> ~6 ms median locally).

Also fixed indentation.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2360663002

Review-Url: https://codereview.chromium.org/2360663002

GN: is_skia_standalone

is_skia_standalone will be defined and set to true iff we're using our
standalone BUILDCONFIG.gn.  Any other build (Fuchsia, Chrome, Pdfium)
will have their own BUILDCONFIG.gn, which presumably does not define
is_skia_standalone.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2357963002

Review-Url: https://codereview.chromium.org/2357963002

fix skia pathops fuzzers

Add isolated tests.

R=kjlubick@google.com
BUG=skia:5775
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2358043002

Review-Url: https://codereview.chromium.org/2358043002

Make GrShape compute keys for short paths from path data instead of using the gen id.
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2357643002

Review-Url: https://codereview.chromium.org/2357643002

Stop closing filled paths in GrShape

This can force a copy of the path data and isn't really being taken advantage of. It complicates a forthcoming change to key small paths based on their data rather than gen id.
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2355833002

Committed: https://skia.googlesource.com/skia/+/ad001fdc77b2dfe32c1723cb58636728dfb776d8
Review-Url: https://codereview.chromium.org/2355833002

Add define of SK_ENABLE_VK_LAYERS to gn build

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2357953002

Review-Url: https://codereview.chromium.org/2357953002

Add setBounds to SkLiteDL for Android

The RootRenderNode in Android is changing bounds dynamically. This is a
temporary fix to accomodate that behaviour. (See also b/31304115)

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2358793002

Review-Url: https://codereview.chromium.org/2358793002

Fix memory leaks reported by clang static analyzer.

The CFURLCopyFileSystemPath & CFURLCopyFileSystemPath methods
respect the "Create Rule" [1] regarding the ownership of the
returned reference. This means that the objects need to be
deallocated explicitly by calling CFRelease.

[1]: https://developer.apple.com/library/content/documentation/CoreFoundation/Conceptual/CFMemoryMgmt/Concepts/Ownership.html#//apple_ref/doc/uid/20001148-103029

BUG=648210
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2360573002

Review-Url: https://codereview.chromium.org/2360573002

Fix some GPU image filter code to preserve precision and color space

On the pure-GPU path, we just have an SkSpecialImage (that's definitely
texture backed), and we need a renderable config for the draw context we
make. Added a helper function to pick - this is basically the high
precision analog of what we were doing before (always using 8888).

The assert that I added catches many other problems in image filter code,
but those fixes are coming in subsequent CLs.

12 GMs render correctly (or more correctly) in gpusrgb and gpuf16
configs. In most cases, they were drawing previously nothing.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2359443003

Review-Url: https://codereview.chromium.org/2359443003

Remove SK_ADDING_src_ports_SkFontConfigInterface_cpp

This flag is no longer defined anywhere, so remove it and the code it
was guarding.

TBR=reed
This is just removing dead transitional code.

Review-Url: https://codereview.chromium.org/2352163002

[SVGDom] Opacity optimization

Apply opacity as fill/stroke paint alpha instead of saveLayer, when
possible.

R=robertphillips@google.com,stephana@google.com,reed@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2353503005

Committed: https://skia.googlesource.com/skia/+/3dbb7b9f196d793fbd16243157ee67638891f5dc
Review-Url: https://codereview.chromium.org/2353503005

Blacklist serialize-8888 typefacerendering_pfaMac

BUG=skia:5778
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2356753003

TBR=halcanary@google.com
NOTREECHECKS=true
CQ_INCLUDE_TRYBOTS=master.client.skia:Test-Mac-Clang-MacMini6.2-CPU-AVX-x86_64-Debug-GN-Trybot

Review-Url: https://codereview.chromium.org/2356753003

GN: fold :vulkan into :gpu.

Obviously, :vulkan can't be used without :gpu, but more subtly,
:gpu needs to see SK_VULKAN defined if we're using Vulkan.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2352923003
NOTREECHECKS=true

Review-Url: https://codereview.chromium.org/2352923003

Revert of Stop closing filled paths in GrShape (patchset #1 id:1 of https://codereview.chromium.org/2355833002/ )

Reason for revert:
Triggers msaa16 asserts: https://build.chromium.org/p/client.skia/builders/Test-Win10-MSVC-ShuttleA-GPU-GTX660-x86_64-Debug/builds/4236/steps/test_skia%20on%20Windows/logs/stdio

c:\b\work\skia\src\gpu\grmesh.h:76: fatal error: "assert(indexCount)"

Caught exception 2147483651 EXCEPTION_BREAKPOINT, was running:
msaa16 svg  cartman.svg
Likely culprit:
msaa16 svg  cartman.svg

Original issue's description:
> Stop closing filled paths in GrShape
>
> This can force a copy of the path data and isn't really being taken advantage of. It complicates a forthcoming change to key small paths based on their data rather than gen id.
> GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2355833002
>
> Committed: https://skia.googlesource.com/skia/+/ad001fdc77b2dfe32c1723cb58636728dfb776d8

TBR=robertphillips@google.com,bsalomon@google.com
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true

Review-Url: https://codereview.chromium.org/2357713002

Revert of [SVGDom] Opacity optimization (patchset #4 id:60001 of https://codereview.chromium.org/2353503005/ )

Reason for revert:
Failing assertion

Original issue's description:
> [SVGDom] Opacity optimization
>
> Apply opacity as fill/stroke paint alpha instead of saveLayer, when
> possible.
>
> R=robertphillips@google.com,stephana@google.com,reed@google.com
> GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2353503005
>
> Committed: https://skia.googlesource.com/skia/+/3dbb7b9f196d793fbd16243157ee67638891f5dc

TBR=reed@google.com,robertphillips@google.com,stephana@google.com,fmalita@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true

Review-Url: https://codereview.chromium.org/2359603002

GN: add skia_use_mesa for OSMesa in tools.

I tested by running
  $ out/dm --src gm --config mesa
  $ out/nanobench --config mesa --match Xfermode
... but I ran out of patience waiting for nanobench to finish.  Probably works.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2357703002

Review-Url: https://codereview.chromium.org/2357703002

Attempt to fix nanobench upload

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2356813003

Review-Url: https://codereview.chromium.org/2356813003

Revert of Clear stencil buffer before using it for drawing (patchset #2 id:20001 of https://codereview.chromium.org/2342873004/ )

Reason for revert:
performance regression in Android macrobench tests

Original issue's description:
> Clear stencil buffer before using it for drawing
>
> Clear stencil buffer before using it for drawing. This is
> resolving an issue with a failing android CTS test in drawArc.
> GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2342873004
>
> Committed: https://skia.googlesource.com/skia/+/c326a383e1069497884905f9dd3e720e17d8fcb5

TBR=bsalomon@google.com,reed@google.com
# Not skipping CQ checks because original CL landed more than 1 days ago.

Review-Url: https://codereview.chromium.org/2355923002

Fix VkWrapTests

BUG=skia:
TBR=bsalomon@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2352173002

Review-Url: https://codereview.chromium.org/2352173002

GM: add test for type1 font

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2350453002

Review-Url: https://codereview.chromium.org/2350453002

Disable framebuffer fetch on the Adreno5xx devices.

This should help with some of the rendering bugs currently in GL for the
S7.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2348293004

Review-Url: https://codereview.chromium.org/2348293004

Fix possible divide by zero in vulkan granularity adjustments

This should fix the divide by zero crash on the nexus player bots

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2358613002

Review-Url: https://codereview.chromium.org/2358613002

Stop closing filled paths in GrShape

This can force a copy of the path data and isn't really being taken advantage of. It complicates a forthcoming change to key small paths based on their data rather than gen id.
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2355833002

Review-Url: https://codereview.chromium.org/2355833002

[SVGDom] Opacity optimization

Apply opacity as fill/stroke paint alpha instead of saveLayer, when
possible.

R=robertphillips@google.com,stephana@google.com,reed@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2353503005

Review-Url: https://codereview.chromium.org/2353503005

Disable failing unit tests on adreno vulkan

As mentioned in the attached bug, these failures all seem to be caused
by driver issues that have no workarounds at this time.

BUG=skia:5777
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2355883002

Review-Url: https://codereview.chromium.org/2355883002

Use STDArray to for tracked resources in vulkan command buffer

This is a minor speed improvement to help with the number of times we
are freeing and re-allocating the tracked resource arrays

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2338963004

Review-Url: https://codereview.chromium.org/2338963004

Revert of Skia recipe: Use [depot_tools]/gsutil.py instead of [depot_tools]/third_party/gsutil/gsutil (patchset #1 id:1 of https://codereview.chromium.org/2298293002/ )

Reason for revert:
Broke all nanobench uploads, eg. https://uberchromegw.corp.google.com/i/client.skia.android/builders/Perf-Android-Clang-Nexus6p-CPU-Snapdragon810-arm64-Release-GN_Android/builds/154/steps/Upload%20perf%20results/logs/stdio

Original issue's description:
> Skia recipe: Use [depot_tools]/gsutil.py instead of [depot_tools]/third_party/gsutil/gsutil
>
> The latter is being removed in https://codereview.chromium.org/2280023003/
> Is also pinned to 3.25, whereas the former is updated more often.
>
> BUG=skia:
> GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2298293002
>
> Committed: https://skia.googlesource.com/skia/+/f15590f803596f51d0950994b2aeb623d2d1fc24

TBR=hinoka@google.com
NOTREECHECKS=true
NOTRY=true
BUG=skia:

Review-Url: https://codereview.chromium.org/2355943002

GN: add skia_enable_gpu (default true)

This lets us replace the NoGPU bot.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2356853002

NOTREECHECKS=true

Review-Url: https://codereview.chromium.org/2356853002

GN: compiler_prefix -> cc_wrapper

I was a bit mixed up thinking Chrome called this feature compiler_prefix.
They actually call it cc_wrapper.  So let's switch to their name before we
get too used to this.

No bots are affected.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2355003002
DOCS_PREVIEW= https://skia.org/?cl=2355003002

NOTREECHECKS=true

Review-Url: https://codereview.chromium.org/2355003002

Snazzier dump log step on Android bots.

(The failing NexusPlayer bot is a demo.)

This should make stack traces more useful, turning this sort of thing

    09-20 11:29:39.536  2978  2978 F DEBUG   :     #00 pc 00970fd0  /data/local/tmp/dm

into something like this

    09-20 11:29:39.536  2978  2978 F DEBUG   :     #00 pc 00970fd0 adjust_bounds_to_granularity(SkIRect*, SkIRect const&, VkExtent2D const&, int, int) at /b/work/skia/out/Build-Ubuntu-Clang-x86-Debug-GN_Android_Vulkan/Debug/../../../src/gpu/vk/GrVkGpu.cpp:1803  /data/local/tmp/dm

Some bots like the S7 already have good enough stack traces, e.g.

09-20 11:35:12.567   936   936 F DEBUG   :     #00 pc 00000000000bed6c  /system/vendor/lib64/hw/vulkan.msm8996.so (_ZN13QglManagedBuf14ConfirmEntriesEv+108)
09-20 11:35:12.567   936   936 F DEBUG   :     #01 pc 00000000000b098c  /system/vendor/lib64/hw/vulkan.msm8996.so (_ZN9QglBltLib6FillHwEPK10QglBltFillPK15QglBltColorFillPK15QglBltDepthFillPjP12QglBltStatusSB_+588)
09-20 11:35:12.567   936   936 F DEBUG   :     #02 pc 00000000000b23bc  /system/vendor/lib64/hw/vulkan.msm8996.so (_ZN9QglBltLib9FillImageEPK10QglBltFillPK15QglBltColorFillPK15QglBltDepthFillPjP12QglBltStatusSB_+348)
09-20 11:35:12.567   936   936 F DEBUG   :     #03 pc 000000000009bb00  /system/vendor/lib64/hw/vulkan.msm8996.so (_ZN16A5xCommandBuffer22PerformConditionalFillEiiP10QglBltFillPK15QglBltColorFillP15QglBltDepthFillP12QglBltStatusS8_+256)
09-20 11:35:12.567   936   936 F DEBUG   :     #04 pc 000000000009c0c0  /system/vendor/lib64/hw/vulkan.msm8996.so (_ZN16A5xCommandBuffer23HwWriteSubpassInitClearEP10QglBltFillPK15QglBltColorFillP15QglBltDepthFillP12QglBltStatusS8_+224)
09-20 11:35:12.567   936   936 F DEBUG   :     #05 pc 0000000000072610  /system/vendor/lib64/hw/vulkan.msm8996.so (_ZN16QglCommandBuffer18WriteSubpassClearsEv+464)
09-20 11:35:12.567   936   936 F DEBUG   :     #06 pc 0000000000073ae0  /system/vendor/lib64/hw/vulkan.msm8996.so (_ZN16QglCommandBuffer12BeginSubpassEv+32)
09-20 11:35:12.567   936   936 F DEBUG   :     #07 pc 0000000000063120  /system/vendor/lib64/hw/vulkan.msm8996.so (vkCmdBeginRenderPass+224)
09-20 11:35:12.567   936   936 F DEBUG   :     #08 pc 0000000000635f60  /data/local/tmp/nanobench (_ZN24GrVkPrimaryCommandBuffer15beginRenderPassEPK7GrVkGpuPK14GrVkRenderPassjPK12VkClearValueRK16GrVkRenderTargetRK7SkIRectb+132)

or

09-20 11:42:24.557   937   937 F DEBUG   : backtrace:
09-20 11:42:24.557   937   937 F DEBUG   :     #00 pc 0000000000069404  /system/lib64/libc.so (tgkill+8)
09-20 11:42:24.557   937   937 F DEBUG   :     #01 pc 0000000000066b94  /system/lib64/libc.so (pthread_kill+68)
09-20 11:42:24.557   937   937 F DEBUG   :     #02 pc 0000000000023a28  /system/lib64/libc.so (raise+28)
09-20 11:42:24.557   937   937 F DEBUG   :     #03 pc 000000000001e358  /system/lib64/libc.so (abort+60)
09-20 11:42:24.557   937   937 F DEBUG   :     #04 pc 000000000076430c  /data/local/tmp/dm (_Z17sk_abort_no_printv+8)

These won't be affected.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2351243002

NOTREECHECKS=true

Review-Url: https://codereview.chromium.org/2351243002

Cleanup some sRGB TODOs, mostly around setting GrPaint flags correctly

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2358433004

Review-Url: https://codereview.chromium.org/2358433004

Support use of non-coherent memory allocations in Vulkan.

BUG=skia:5034
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2348523002

Review-Url: https://codereview.chromium.org/2348523002

Stop flattening GrCoordTransforms in parent GrFragmentProcessors.

This changes moves to a model that iterates over GrCTs in a GrFP hierarchy when inserting transformations by GrGLSLPrimitiveProcessors.
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2339203002

Committed: https://skia.googlesource.com/skia/+/d91237ee051523f439238042674ade99207fe4a6
Review-Url: https://codereview.chromium.org/2339203002

Workaround for Adreno INITIALIZATION_FAILED bug

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2352083002

Review-Url: https://codereview.chromium.org/2352083002

Fix a memory leak reported by clang static analyzer.

The CFStringCreateWithCString method respect the "Create Rule" [1]
regarding the ownership of the returned reference. This means that
the object needs to be deallocated explicitly by calling CFRelease.

[1]: https://developer.apple.com/library/content/documentation/CoreFoundation/Conceptual/CFMemoryMgmt/Concepts/Ownership.html#//apple_ref/doc/uid/20001148-103029

BUG=648210
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2354963002

Review-Url: https://codereview.chromium.org/2354963002

abstract name of clipping ops, to transtion to a more restricted set

SkRegion::Op --> SkCanvas::ClipOp (alias) --> SkClipOp

pre-CL needed in chrome : https://codereview.chromium.org/2355583002/

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2355483002

Review-Url: https://codereview.chromium.org/2355483002

skia_enable_android_framework_defines in all code, public and private, optional and required.

The bug fixed here can be demonstrated by
  $ ninja -C out pdf -v
before and after this CL.  Before you won't see the Android defines, afterwards you do.  Previously the defines applied only privately to Skia, and only to its non-optional components.

TBR=reed@google.com

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2356783002

Review-Url: https://codereview.chromium.org/2356783002

Auto-generate much of the help text for --config to reduce maintenance

The list of configs, and the details about all predefined configs are now
built directly from the gPredefinedConfigs array.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2354943002

Review-Url: https://codereview.chromium.org/2354943002

Fix Win toolchain asset, fix script

BUG=skia:5764
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2351033004

Review-Url: https://codereview.chromium.org/2351033004

Harden LinearGradient4fContext

Three distinct issues exposed by Fuzzer:

1) truncating the flags to 8 bits may trip the assert in SkToU8

Since clients can pass in any garbage, we cannot assume it fits in 8
bits.  Just static_cast<> instead.

2) vertical gradients with dx == -0.0 don't trigger interval inversion.

For dx < 0 we want reversed intervals.  But, alas, -0.0 < 0 == false
(thanks, Ob^WIEEE 754!).  Use signbit() instead.

3) half closed range checking needs to be inverted for reversed
intervals.

Normally we check for [p0, p1), but for reversed intervals (p1 < p0) we
want to check for [p1, p0), and not (p1, p0].  Adjust the logic
accordingly.

BUG=skia:5647
R=reed@google.com,kjlubick@google.com
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2349153004

Review-Url: https://codereview.chromium.org/2349153004

Blacklist incomplete pngs on CG

BUG=skia:5774
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2351083002

Review-Url: https://codereview.chromium.org/2351083002

SkASSERT firing because pointer wraps negative.

Some pointers are being cast to intptr_t so they can be used with
SkAlign8(). However, a large pointer value might become a negative
integer since intptr_t is signed.

When comparing these intptr_ts, we expect the larger pointer value to be
greater. But it might be so large that it becomes negative, causing it
to be less than.

A SkASSERT is firing for this exact reason.

BUG=648452
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2353453004

Review-Url: https://codereview.chromium.org/2353453004

Add ability to pick a different gamut in GPU f16 or sRGB configs

For now, the only options are sRGB or WideGamutRGB

Basically, the color option to the gpu config is now of
the form: 8888|srgb[_gamut]|f16[_gamut]

color=8888 still implies legacy behavior
srgb implies 8-bit gamma-correct rendering (via sRGB format)
f16 implies 16-bit gamma-correct rendering (via F16 format)

Either of the last two options can then optionally include
a gamut specifier, either _srgb or _wide.
_srgb selects the (default) sRGB gamut
_wide selects the Adobe Wide Gamut RGB gamut, which is nice
for testing, in that it's significantly wider than sRGB,
so rendering differences are obvious.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2350003003

Review-Url: https://codereview.chromium.org/2350003003

Reland of Run SVGs through nanobench on all Perf bots (patchset #1 id:1 of https://codereview.chromium.org/2348563003/ )

Reason for revert:
skbug/5770 has been fixed.
Trying out trybots to see if this can be relanded.

Original issue's description:
> Revert of Run SVGs through nanobench on all Perf bots (patchset #5 id:120001 of https://codereview.chromium.org/2343903002/ )
>
> Reason for revert:
> Causing failures:
> https://uberchromegw.corp.google.com/i/client.skia/builders/Perf-Mac-Clang-MacMini4.1-GPU-GeForce320M-x86_64-Debug-GN/builds/812
>
> Original issue's description:
> > Run SVGs through nanobench on all Perf bots.
> >
> > Tested with:
> > python infra/bots/recipes.py run --workdir=../ swarm_trigger path_config=kitchen buildername=Perf-Ubuntu-GCC-ShuttleA-GPU-GTX660-x86_64-Release-GN-Trybot mastername=client.skia slavename=skiabot-linux-swarm-010 buildnumber=1 revision=d8c2476a8b1e1e1a1771b17e8dd4db8645914f8c issue=2343903002 rietveld=https://codereview.chromium.org patchset=80001
> >
> > Example output is here:
> > https://luci-milo.appspot.com/swarming/task/3147c9789b2d0410
> > Nanobench did run on SVGs.
> >
> > BUG=skia:5757
> > GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2343903002
> >
> > Committed: https://skia.googlesource.com/skia/+/1c58dcb909642454d91ba36f71e55597511b5d52
>
> TBR=jcgregorio@google.com,fmalita@chromium.org,mtklein@google.com
> # Skipping CQ checks because original CL landed less than 1 days ago.
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=skia:5757
>
> Committed: https://skia.googlesource.com/skia/+/be7e3c0db3caa7a86ae233a6bd26281817b83882

TBR=jcgregorio@google.com,fmalita@chromium.org,mtklein@google.com
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=skia:5757
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2350233002

Review-Url: https://codereview.chromium.org/2350233002

Create GrVkCopyPipeline class

This creates a special subclass of GrVkPipeline that we can use for doing
copies on draws. Currently this is not hooked into the system.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2353563003

Review-Url: https://codereview.chromium.org/2353563003

remove dead code around SK_ENABLE_CLIP_QUICKREJECT

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2349383002

TBR=

Review-Url: https://codereview.chromium.org/2349383002

Add FrameworkDefs bot to the CQ.

Word is that this bot only exists so it can be on the CQ...
but we forgot to actually put it on the CQ until last week.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2352703002

Review-Url: https://codereview.chromium.org/2352703002

Skia recipe: Use [depot_tools]/gsutil.py instead of [depot_tools]/third_party/gsutil/gsutil

The latter is being removed in https://codereview.chromium.org/2280023003/
Is also pinned to 3.25, whereas the former is updated more often.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2298293002

Review-Url: https://codereview.chromium.org/2298293002

Split SkFontConfigInterface globals and factory.

Chromium needs to be able to set up their build such that the globals
continue existing but the SkFontMgr::Factory can be defined separately.

GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2346333002
CQ_INCLUDE_TRYBOTS=master.client.skia.compile:Build-Ubuntu-GCC-x86_64-Release-CMake-Trybot,Build-Mac-Clang-x86_64-Release-CMake-Trybot

Review-Url: https://codereview.chromium.org/2346333002

Add adb support to skpbench

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2350003002

Review-Url: https://codereview.chromium.org/2350003002

Remove unnecessary assert in GrBatchAtlas.

BUG=skia:5770
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2354653002

Review-Url: https://codereview.chromium.org/2354653002

skpbench

skpbench is a benchmarking suite for skps that aims to generate 100%
repeatable results. The initial commit consists of three parts:

skpbench

A minimalist program whose sole purpose is to open an skp file,
benchmark it on a single config, and exit. No tiling, looping, or
other fanciness is used; it just draws the skp whole into a size-
matched render target and syncs the GPU after each draw.

Limiting the entire process to a single config/skp pair helps to keep
the results repeatable.

skpbench.py

A wrapper to execute the skpbench binary with various configs and skps.
It also monitors the output in order to filter out and re-run results
with an unacceptable stddev.

In the future this script will lock down and monitor clocks and
temperatures.

parseskpbench.py

A utility for parsing skpbench output into a spreadsheet.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2341823002

Review-Url: https://codereview.chromium.org/2341823002

SkPDF: SkBitSet gets reset() to make drop() better.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2354583002

Review-Url: https://codereview.chromium.org/2354583002

Move vulkan spir-v compile call to GrVkUtil

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2355493002

Review-Url: https://codereview.chromium.org/2355493002

Refactor vulkan buffer mapping and unmapping

A lot of this is so we don't have duplicated code in both the map/unmap and
updateData functions of GrVkBuffer. Also there were slightly differences in
how we handled various things in the two cases that this now unifies.

Also I added a barrier after the vkUpdateBuffer call which I believe was
missing.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2344323002

Review-Url: https://codereview.chromium.org/2344323002

Revert of Stop flattening GrCoordTransforms in parent GrFragmentProcessors. (patchset #3 id:40001 of https://codereview.chromium.org/2339203002/ )

Reason for revert:
Crashing blink
 https://codereview.chromium.org/2351743002/

Original issue's description:
> Stop flattening GrCoordTransforms in parent GrFragmentProcessors.
>
> This changes moves to a model that iterates over GrCTs in a GrFP hierarchy when inserting transformations by GrGLSLPrimitiveProcessors.
> GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2339203002
>
> Committed: https://skia.googlesource.com/skia/+/d91237ee051523f439238042674ade99207fe4a6

TBR=egdaniel@google.com,robertphillips@google.com
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true

Review-Url: https://codereview.chromium.org/2351753002

To prep for removing CMake bots, take them off the CQ and presubmit.

GN is now far more capable than our CMake builds.
Time to officially end that experiment.

Reminder how this lands:
  1) this CL
  2) remove the bots from the masters
  3) remove the CMake recipe files
  4) remove the CMake files

I'm willing to wait on 4) until we can do GN -> CMake generation, but I don't want the bots to hold us back.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2350073002

Review-Url: https://codereview.chromium.org/2350073002