This is more general than checking for __linux, __FreeBSD__, etc. (In
principle we could remove some of the existing checks such as
__FreeBSD__, but I have not tried that so far.)
In particular, it allows Skia to build with the NaCl or PNaCl
toolchains, which is something we would like for Mojo.
BUG=https://github.com/domokit/mojo/issues/431
TEST=none
Review URL: https://codereview.chromium.org/1523733003
Running `Release/dm --gpu 0`, the number of times we call
SkBitmap::operator=(const SkBitmap&)
(which refs the pixelref) is reduced from ~214929 to ~214626.
Review URL: https://codereview.chromium.org/1514503004
Reason for revert:
speculative revert to see if it unblocks the DEPS roll
https://codereview.chromium.org/1529443002
Original issue's description:
> Add config options to run different GPU APIs to dm and nanobench
>
> Add extended config specification form that can be used to run different
> gpu backend with different APIs.
>
> The configs can be specified with the form:
> gpu(api=string,dit=bool,nvpr=bool,samples=int)
>
> This replaces and removes the --gpuAPI flag.
>
> All existing configs should still work.
>
> Adds following documentation:
>
> out/Debug/dm --help config
>
> Flags:
> --config: type: string default: 565 8888 gpu nonrendering
> Options: 565 8888 debug gpu gpudebug gpudft gpunull msaa16 msaa4
> nonrendering null nullgpu nvprmsaa16 nvprmsaa4 pdf pdf_poppler skp svg
> xps or use extended form 'backend(option=value,...)'.
>
> Extended form: 'backend(option=value,...)'
>
> Possible backends and options:
>
> gpu(api=string,dit=bool,nvpr=bool,samples=int) GPU backend
> api type: string default: native.
> Select graphics API to use with gpu backend.
> Options:
> native Use platform default OpenGL or OpenGL ES backend.
> gl Use OpenGL.
> gles Use OpenGL ES.
> debug Use debug OpenGL.
> null Use null OpenGL.
> dit type: bool default: false.
> Use device independent text.
> nvpr type: bool default: false.
> Use NV_path_rendering OpenGL and OpenGL ES extension.
> samples type: int default: 0.
> Use multisampling with N samples.
>
> Predefined configs:
>
> gpu = gpu()
> msaa4 = gpu(samples=4)
> msaa16 = gpu(samples=16)
> nvprmsaa4 = gpu(nvpr=true,samples=4)
> nvprmsaa16 = gpu(nvpr=true,samples=16)
> gpudft = gpu(dit=true)
> gpudebug = gpu(api=debug)
> gpunull = gpu(api=null)
> debug = gpu(api=debug)
> nullgpu = gpu(api=null)
>
> BUG=skia:2992
>
> Committed: https://skia.googlesource.com/skia/+/e13ca329fca4c28cf4e078561f591ab27b743d23TBR=bsalomon@google.com,scroggo@google.com,joshualitt@google.com,kkinnunen@nvidia.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:2992
Review URL: https://codereview.chromium.org/1528473002
Add extended config specification form that can be used to run different
gpu backend with different APIs.
The configs can be specified with the form:
gpu(api=string,dit=bool,nvpr=bool,samples=int)
This replaces and removes the --gpuAPI flag.
All existing configs should still work.
Adds following documentation:
out/Debug/dm --help config
Flags:
--config: type: string default: 565 8888 gpu nonrendering
Options: 565 8888 debug gpu gpudebug gpudft gpunull msaa16 msaa4
nonrendering null nullgpu nvprmsaa16 nvprmsaa4 pdf pdf_poppler skp svg
xps or use extended form 'backend(option=value,...)'.
Extended form: 'backend(option=value,...)'
Possible backends and options:
gpu(api=string,dit=bool,nvpr=bool,samples=int) GPU backend
api type: string default: native.
Select graphics API to use with gpu backend.
Options:
native Use platform default OpenGL or OpenGL ES backend.
gl Use OpenGL.
gles Use OpenGL ES.
debug Use debug OpenGL.
null Use null OpenGL.
dit type: bool default: false.
Use device independent text.
nvpr type: bool default: false.
Use NV_path_rendering OpenGL and OpenGL ES extension.
samples type: int default: 0.
Use multisampling with N samples.
Predefined configs:
gpu = gpu()
msaa4 = gpu(samples=4)
msaa16 = gpu(samples=16)
nvprmsaa4 = gpu(nvpr=true,samples=4)
nvprmsaa16 = gpu(nvpr=true,samples=16)
gpudft = gpu(dit=true)
gpudebug = gpu(api=debug)
gpunull = gpu(api=null)
debug = gpu(api=debug)
nullgpu = gpu(api=null)
BUG=skia:2992
Review URL: https://codereview.chromium.org/1490113005
Add a test that we get the same color back after calling SkBitmap::eraseColor, modulo rounding.
Also update some incorrect docs.
BUG=skia:4297
Review URL: https://codereview.chromium.org/1521673002
If the client requests a color type or alpha type that is
not supported, we should decode to the default/appropriate
color and alpha types to match existing behavior in Android.
BUG=skia:
Review URL: https://codereview.chromium.org/1513023002
This CL may not really be necessary since it seems that the Android Ones we have fall back to mediump when highp isn't supported.
FYI, for bits-of-precision/precision we have:
Low Med High Precision-Used
A1 (Cherry One - Mali 400) 10 10 0 Med - conic looks bad
N5 (Adreno 330) 10 10 23 High - conic looks bad!!!
N6 (Adreno 420) 10 10 23 High - conic looks good
N7v2 (Adreno 320) 10 10 23 High - conic looks bad!!!
N9 (K1) 23 23 23 High - looks good
N10 (Mail-604) 10 10 23 High - looks good
I suspect derivative computation is broken on the Adreno 300s.
BUG=555779,skia:4668
TBR=bsalomon@google.com
Review URL: https://codereview.chromium.org/1513573009
In the current implementation, a blur filter is always created even in the
case when sigma.fX == 0 && sigma.fY == 0. This CL makes the blur filter
return input in this case.
BUG=568393
Review URL: https://codereview.chromium.org/1518643002
Make SkAutoTMalloc's interface look more like SkAutoMalloc:
- add free(), which does what you expect
- make reset() return a pointer fPtr
No public API changes (SkAutoTMalloc is in include/private)
BUG=skia:2148
Review URL: https://codereview.chromium.org/1516833003
Extend the ends of hairline and haircurve segments when the paint is set to square or round, and the line or curve is at the start or end of a contour.
R=reed@google.com
BUG=skia:4599
Review URL: https://codereview.chromium.org/1491843006
Use an inline function that does a normal shift. When built for the sanitizer, add casts so that the shift is unsigned.
Also make a few fixes to do unsigned shifts or avoid the shift altogether; and add an argument spec to some macros.
R=reed@google.com,mtklein@google.com
BUG=skia:4633
Review URL: https://codereview.chromium.org/1503423003
Needed to enlarge radii to surface bugs without exceeding max texture size on low end devices.
BUG=chromium:477684
Review URL: https://codereview.chromium.org/1508003008
Make each filter responsible for expanding its destination
bounds. Previously, we were using a union of all
intermediate bounds sizes via join() calls in many image
filters' computeFastBounds(), due to the fact that those
filters could only produce bitmaps the same size as their
inputs. Now, we compute optimal bounds for each filter as
follows:
1) Pass the (unmodified) clip bounds to the root node
of the DAG in the first recursive call to onFilterImage()
as the Context's fClipBounds.
2) Reverse-map the clip: when recursing up the DAG in
filterInput[GPU](), apply filter-specific expansion to the
clip by calling calling onFilterNodeBounds(... kReverse).
This allows upstream nodes to have a clip that respects the
current node's requirements. This is done via helper
function mapContext().
3) Forward-map the source bitmap: just prior to applying
the crop rect in applyCropRect(), we determine the filter's
preferred bounds by mapping the source bitmap bounds
forwards via onFilterNodeBounds(..., kForward).
NOTE: GMs affected by this change:
fast_slow_blurimagefilter: fast and slow paths now produce the same result
spritebitmap: drawSprite() and drawBitmap() paths now produce the same result
filterfastbounds: fast bounds are optimized; all drop-shadow results now appear
apply-filter: snug and not-snug cases give same results
dropshadowimagefilter: drawSprite() results now show shadows
draw-with-filter: no artifacts on erode edges; blur edges no longer clipped
displacement, imagefiltersbase, imagefiltersclipped, imagefilterscropexpand, imagefiltersscaled, matriximagefilter,
resizeimagefilter, localmatriximagefilter, testimagefilters: fixed incorrect clipping
imagefilterstransformed, morphology: no artifacts on erode edges
BUG=skia:1062,skia:3194,skia:3939,skia:4337,skia:4526
Review URL: https://codereview.chromium.org/1308703007
- Update SkCodec's dox to point out that some of the data must be
read twice in order to decode
- Add an accessor that reports how much is needed for reading twice
- Make SkCodec default to use peek()
If an input stream supports peek()ing, peek() instead of reading.
This way the stream need not implement rewind()
- Make SkCodec use read() + rewind() as a backup
So that streams without peek() implemented can still function
properly (assuming they can rewind).
- read everything we may need to determine the format once
In SkCodec::NewFromStream, peek()/read() 14 bytes, which is enough
to read all of the types we support. Pass the buffer to each subtype,
which will have enough info to determine whether it is the right
type. This simplifies the code and results in less reading and
rewinding.
- NOTE: SkWbmpCodec needs the following number of bytes for the header
+ 1 (type)
+ 1 (reserved)
+ 3 (width - bytes needed to support up to 0xFFFF)
+ 3 (height - bytes needed to support up to 0xFFFF)
= 8
- in SkWebpCodec, support using read + rewind as a backup if peek does
not work.
A change in Android will add peek() to JavaInputStreamAdapter.
BUG=skia:3257
Review URL: https://codereview.chromium.org/1472123002
Will no longer be needed once ag/817367 lands - Android will be
inheriting directly from SkPngChunkReader, so no need for the
intermediate.
BUG=skia:4574
Review URL: https://codereview.chromium.org/1470913004
It's undefined behavior to pass null as src or dst to memcpy, even if len is 0.
This currently triggers -fsanitize=attribute-nonnull warnings, but also can
lead to very unexpected code generation with GCC.
sk_careful_memcpy() checks len first before calling memcpy(),
which prevents that weird undefined situation.
This allows me to mark all sanitizers as no-recover, i.e. make-the-bots-red fatal.
CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Debug-ASAN-Trybot
BUG=skia:4641
NOTREECHECKS=true
Review URL: https://codereview.chromium.org/1510683002
Add SkImageEncoder::EncodeData(const SkPixmap&, ...) function.
Add SkImageEncoder::CreatePixelSerializer() to return a
PixelSerializer that calls into SkImageEncoder::EncodeData.
SkImage::encode() make use of SkImageEncoder::CreatePixelSerializer.
Review URL: https://codereview.chromium.org/1507123002
Replaces drawPathsFromRange with a more general drawPathBatch method.
While this still isn't perfect, it's a step in the right direction that
removes the need for path range draws to fit in a public API.
BUG=skia:
Review URL: https://codereview.chromium.org/1506823004
If the stream can peek less than requested, peek that amount. Return
the number of bytes peeked.
This simplifies crrev.com/1472123002. For a stream that is smaller than
14 bytes, it can successfully peek, meaning the client will not need to
fall back to read() + rewind(), which may fail if the stream can peek
but not rewind.
This CL revives code from patch set 3 of crrev.com/1044953002, where I
initially introduced peek() (including tests).
Add a test for SkFrontBufferedStream that verifies that peeking does
not make rewind() fail (i.e. by reading past the internal buffer).
BUG=skia:3257
Review URL: https://codereview.chromium.org/1490923005
Because the convolution kernel is (currently) applied in device space,
there's no way to know which object-space pixels will be touched. So
return false from canComputeFastBounds().
The results from the matrixconvolution GM were actually wrong, since
they were showing edge differences on the clip boundaries, where they
should really only show on crop boundaries. I added a crop to the GM
to keep the results the same (which are useful to test the different
convolution tile modes).
While I was at it, SkImageFilter::affectsTransparentBlack() was
inapplicable on most things except color filters, and its use on
leaf nodes was confusing. So I removed it, and made
SkImageFilter::canComputeFastBounds() virtual instead.
BUG=skia:4630
Review URL: https://codereview.chromium.org/1500923004
visual bench run on Mac Pro
curr/maxrss loops min median mean max stddev samples config bench
100/100 MB 16 412µs 413µs 413µs 414µs 0% ▄▁▇▄▄▄▄█▄▃▅ gpu warmupbench
101/102 MB 32 547µs 548µs 611µs 1.24ms 34% █▁▁▁▁▁▁▁▁▁▁ gpu image-filter-sprite-draw-image
102/103 MB 32 547µs 548µs 721µs 1.23ms 41% █▁▇▁▁█▁▁▁▁▁ gpu image-filter-sprite-draw-bitmap
103/103 MB 64 546µs 546µs 546µs 547µs 0% ▆▄▂▁▇█▅▇▅▇▃ gpu image-filter-sprite-draw-sprite
Should have no effect on Chrome while SK_SUPPORT_LEGACY_LAYER_BITMAP_IMAGEFILTERS is defined (which it is in chrome)
BUG=skia:1073
Review URL: https://codereview.chromium.org/1491293002
Reason for revert:
Introduced memory leak; pixel changes in Chrome.
Original issue's description:
> Matrix convolution bounds fix; affectsTransparentBlack fixes.
>
> Because the convolution kernel is (currently) applied in device space,
> there's no way to know which object-space pixels will be touched. So
> return false from canComputeFastBounds().
>
> The results from the matrixconvolution GM were actually wrong, since
> they were showing edge differences on the clip boundaries, where they
> should really only show on crop boundaries. I added a crop to the GM
> to keep the results the same (which are useful to test the different
> convolution tile modes).
>
> While I was at it, SkImageFilter::affectsTransparentBlack() was
> inapplicable on most things except color filters, and its use on
> leaf nodes was confusing. So I removed it, and made
> SkImageFilter::canComputeFastBounds() virtual instead.
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/8705ec80518ef551994b82ca5ccaeb0241d6adecTBR=reed@google.com,reed@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:
Review URL: https://codereview.chromium.org/1497083005
Because the convolution kernel is (currently) applied in device space,
there's no way to know which object-space pixels will be touched. So
return false from canComputeFastBounds().
The results from the matrixconvolution GM were actually wrong, since
they were showing edge differences on the clip boundaries, where they
should really only show on crop boundaries. I added a crop to the GM
to keep the results the same (which are useful to test the different
convolution tile modes).
While I was at it, SkImageFilter::affectsTransparentBlack() was
inapplicable on most things except color filters, and its use on
leaf nodes was confusing. So I removed it, and made
SkImageFilter::canComputeFastBounds() virtual instead.
BUG=skia:
Review URL: https://codereview.chromium.org/1500923004
