This CL removes the uses of SkNEW that have resprouted since commit
385fe4d, and removes the macros entirely now that Android and Chromium
have been cleaned up to no longer depend on them.
A bunch of files implicitly depend on #include <new> from SkPostConfig.h
still though, so keep that for now. To be fixed in a followup CL.
[mtklein mucking around]
Only public API removed.
TBR=reed@google.com
Review URL: https://codereview.chromium.org/1360653004
Reason for revert:
Who wants to land forever?
Original issue's description:
> update memset16/32 inlining heuristics
>
> I spent some time looking at perf.skia.org and it looks like we can do better.
>
> It is weird, weird, weird that on x86, we see three completely different behaviors:
> - x86 Android: inlining better for small N, custom better for large N;
> - Windows: inlining better for large N, custom better for small N;
> - other x86: inlining generally better
>
> BUG=skia:4316,chromium:516426
>
> (Temporary, plan to revert.)
> TBR=reed@google.com
>
> Committed: https://skia.googlesource.com/skia/+/b68fa409fc00ce2f38e2a0fd6f9dc2379b372481TBR=reed@google.com,jcgregorio@google.com,mtklein@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:4316,chromium:516426
Review URL: https://codereview.chromium.org/1358793002
I spent some time looking at perf.skia.org and it looks like we can do better.
It is weird, weird, weird that on x86, we see three completely different behaviors:
- x86 Android: inlining better for small N, custom better for large N;
- Windows: inlining better for large N, custom better for small N;
- other x86: inlining generally better
BUG=skia:4316,chromium:516426
(Temporary, plan to revert.)
TBR=reed@google.com
Review URL: https://codereview.chromium.org/1357193002
Remove some bogus tests on the cache, as they are not thread-reliable. Running w/ discardable these are racy.
BUG=532981
Review URL: https://codereview.chromium.org/1351453004
The new FP is used to implement SkXM::Mode color filters and SkXM::Mode image filters. Also, these now support all advanced SkXM::Mode xfermodes.
Review URL: https://codereview.chromium.org/1334293003
Adds an entry point to GrContext to allow enumeration and tracing of GPU resources
via the newly added SkTraceMemoryDump.
Plan is for Chrome to call this on each of its GrContexts.
Dumps both the total size of GPU resources, as well as the total purgeable size.
BUG=526261
Review URL: https://codereview.chromium.org/1313743002
This makes the blurcircles bench go from ~33us to ~8us on Windows desktop.
It will require layout test suppressions
Review URL: https://codereview.chromium.org/1311583005
Blink is migrating away from SkBitmaps, so we need an SkImage-based
SkImageFilter source. This is pretty much a 1-1 equivalent of
SkBitmapSource.
To avoid duplication, relocate the SkImage deserialization logic
from SkPictureData to SkReadBuffer.
R=reed@google.com,robertphillips@google.com,senorblanco@chromium.org
Review URL: https://codereview.chromium.org/1343703005
The new flow of calls for attaching a Stencil looks like:
Client
rt->attachStencilAttachment()
gpu->getStencilAttachment()
glgpu->createStencilAttachment()
glrt->completeStencilAttachment() //actually attaches
BUG=skia:
Review URL: https://codereview.chromium.org/1333383002
Possible follow-up changes to consider
1. Roll SkImage_Raster and _Gpu into _Generator, where the generator (or cacherator) is backed by a pre-existing texture or raster.
2. Evolve SkImageUsageType into a verb requiring stretching, and have the caller (common code) digest the caps() and usage, so that subclasses are just told what to do (stretch or not)
3. Common code/utility to convert an unstretched texture into a stretch one (and cache it) if the generator can only make an unstretched one.
BUG=skia:
Review URL: https://codereview.chromium.org/1282363002
This was a pre-SkOpts attempt that we can bring under its wing now.
This should be a perf no-op, deo volente.
BUG=skia:4117
Review URL: https://codereview.chromium.org/1314863006
SkOncePtr<T[]> is identical to SkOncePtr<T> except we'll default to delete[]
for cleanup.
CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Debug-ASAN-Trybot
BUG=skia:
No public API changes.
TBR=reed@google.com
Review URL: https://codereview.chromium.org/1311893010
Reason for revert:
Breaks Chrome roll.
obj/skia/ext/skia_chrome.skia_memory_dump_provider.o
does not have -I include/private on its include path, but transitively includes SkMessageBus.h.
Original issue's description:
> Port uses of SkLazyPtr to SkOncePtr.
>
> This gives SkOncePtr a non-trivial destructor that uses std::default_delete
> by default. This is overrideable, as seen in SkColorTable.
>
> SK_DECLARE_STATIC_ONCE_PTR still just leaves its pointers hanging at EOP.
>
> BUG=skia:
>
> No public API changes.
> TBR=reed@google.com
>
> Committed: https://skia.googlesource.com/skia/+/a1254acdb344174e761f5061c820559dab64a74cTBR=herb@google.com,mtklein@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:
Review URL: https://codereview.chromium.org/1334523002
This gives SkOncePtr a non-trivial destructor that uses std::default_delete
by default. This is overrideable, as seen in SkColorTable.
SK_DECLARE_STATIC_ONCE_PTR still just leaves its pointers hanging at EOP.
BUG=skia:
No public API changes.
TBR=reed@google.com
Review URL: https://codereview.chromium.org/1322933005
Image encoding may fail during serialization, resulting in zero-length
encoded data in the SKP.
Instead of invalidating the stream (and preventing deserialization of
the whole picture) we can instantiate placeholder images.
BUG=skia:4285
R=reed@google.com,robertphillips@google.com
Review URL: https://codereview.chromium.org/1308273011
It is easy to think in some cases that SkString::resize(int) is not
destructive, since optimizations mean that most of the time the data
is still there after a resize. However, in the general case, the
original string's data is lost and the new SkString contains garbage.
Review URL: https://codereview.chromium.org/1304833004
The million SKPs generated require >5T of storage. A good deal
of that are copies of system fonts.
Chrome built with
#DEFINE SK_WHITELIST_SERIALIZED_TYPEFACES
will omit the font data if the font matches a precomputed
checksum.
The captured SKP prepends sk_ to the names of fonts that
have their data omitted. The SKP consumer can either add
renamed fonts from the recording machine, or add
gDeserializeTypefaceDelegate = WhitelistDeserializeTypeface;
which strips the sk_ prefix when deserializing typefaces.
whitelist_typefaces --check
Computes the checksums of fallback
fonts and returns 0 if the checksums match the checked-in
file SkWhitelistChecksum.cpp.
whitelist_typefaces --generate
Writes an updated version of SkWhitelistChecksum.cpp.
(Added Mike since this modifies a public header)
R=bungeman@google.com,rmistry@google.com,reed@google.com
Review URL: https://codereview.chromium.org/1317913005
Redesigns SkScanlineDecoder.h to indicate the ordering
in which the scanlines are provided
Refactors SkSwizzler::Fill() to include the zeroInit check
and to actually be correct.
BUG=skia:3257
BUG=skia:4198
Review URL: https://codereview.chromium.org/1287423002
This change regularizes Skia's type traits so that when <type_traits>
can finally be used the transition is easier. Various traits are
renamed to match <type_traits> and placed in the skstd namespace.
Current users of these traits are updated.
Review URL: https://codereview.chromium.org/1317593004
This macro was responsible for producing code like:
static SkAlignedStorage<sizeof(Foo)> g_gFoo_Storage;
static Foo* gFoo = new(g_gFoo_Storage.get()) Foo;
static SkAutoTDestroy<Foo> gFoo_ad(gFoo);
which would allocate static storage for an object of type Foo
(g_gFoo_Storage), lazily instantiate the object in that memory (via
gFoo's initializer), and then ensure that at global destruction time
the object is destroyed (via gFoo_Ad's destructor).
However, the exact same effect is achieved by just writing:
static Foo gFoo;
Review URL: https://codereview.chromium.org/1314763009
Dump the memory statistics of resource cache and glyph cache using the
SkTraceMemoryDump interface.
BUG=chromium:503168
Review URL: https://codereview.chromium.org/1313793004
Re-land; layout tests are suppressed and gm differences are understood.
A merge conflict prevented a automatic reland.
If the endcap is not butt, draw the endcaps even when the line
has zero length.
If the dash length is zero, generate a zero length line segment.
Treat a move followed by a close as a move followed by a zero-length
line.
TBR=reed@google.com
BUG=422974
Review URL: https://codereview.chromium.org/1314213002
Unfortunately, immintrin.h (which is also included by SkTypes)
includes xmmintrin.h which includes mm_malloc.h which includes
stdlib.h for malloc even though, from the implementation, it is
difficult to see why.
Fortunately, arm_neon.h does not seem to be involved in such
shenanigans, so building for Android will keep things sane.
TBR=reed@google.com
Doesn't change Skia API, just moves an include.
Review URL: https://codereview.chromium.org/1313203003
The motivation for this was to remove SK_OFFSETOF from SkTypes, but
this CL is mostly about cleaning up our use of offsetof generally.
SK_OFFSETOF is removed to SkTypes and added to the two places it is
actually used (for the non standard behavior of finding the offset of
fields in types which are not standard layout).
Older versions of gcc required POD for offsetof to be used without
warning. Newer versions require the more relaxed standard layout.
Now that we no longer build on older versions of gcc, remove the
old warning suppressions.
PODMatrix is renamed to AggregateMatrix. SkMatrix is already POD
(trivial and standard layout). The PODMatrix name implies that the
POD-ness is needed for the offsetof, but it is actually the aggregate
attribute which is needed for compile time constant initialization.
This makes it more obvious that this can be revisited after we can
rely on constexpr constructors.
This also adds skstd::declval since this allows removal of existing
awkward code which casts a constant to a pointer to find the size of
a field.
TBR=reed@google.com
No API change, only removes unused macro.
Review URL: https://codereview.chromium.org/1309523003
Looks like the -Wmicrosoft warnings in the bug are coming up specificically
because many the effects in SkLightingImageFilter are defined inside an
anonymous namespace (spanning, I think, lines 33-702 today).
BUG=skia:4091
No public API changes.
TBR=reed@google.com
Review URL: https://codereview.chromium.org/1311783004
CL (1 of 3) adds empty lists in our .gypi,
and builds the files in those empty lists with the appropriate flags.
CL (2 of 3) will have Chrome's GYP and GN files read these lists,
and build them with the appropriate flags.
CL (3 of 3) will add runtime detection and stub files to the lists
with empty Init_sse42(), Init_avx(), Init_avx2() methods.
After that, we should be able to use SSE 4.2, AVX, and AVX2 if desired.
Some motivation:
- SSE 4.2 adds some sweet string-oriented methods that
can help us write fast high quality 32-bit hashes.
- AVX is SSE doubled, e.g. 8 floats or two SkPMFloat at a time.
- AVX2 is SSE2 doubled, e.g. 8 pixels at a time.
BUG=skia:4117
Review URL: https://codereview.chromium.org/1290423007
Reason for revert:
More GMs changed than I expected. Will probably affect layout tests as well; reverting until I can verify that the changes are correct.
Original issue's description:
> experiment with zero-length round capped line segments
>
> If the endcap is not butt, draw the endcaps even when the line
> has zero length.
>
> If the dash length is zero, generate a zero length line segment.
>
> Treat a move followed by a close as a move followed by a zero-length
> line.
>
> R=reed@google.com,schenney@google.com
> BUG=422974
>
> Committed: https://skia.googlesource.com/skia/+/dd3c165828fffb369d0f4b13b48381169a0249a9TBR=reed@google.com,schenney@google.com,schenney@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=422974
Review URL: https://codereview.chromium.org/1304163008
If the endcap is not butt, draw the endcaps even when the line
has zero length.
If the dash length is zero, generate a zero length line segment.
Treat a move followed by a close as a move followed by a zero-length
line.
R=reed@google.com,schenney@google.com
BUG=422974
Review URL: https://codereview.chromium.org/1309753002
Reason for revert:
This causes a syntax error.
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_compile_dbg_ng/builds/87819/steps/compile%20%28with%20patch%29/logs/stdio
Original issue's description:
> Implement canComputeFastBounds() for image filters.
>
> Image filters have never implemented this check, which means that
> filters which affect transparent black falsely claim they can compute
> their bounds.
>
> Implemented an affectsTransparentBlack() virtual for image
> filters, and a similar helper function for color filters.
>
> This will affect the following GMs: imagefiltersscaled
> (lighting, perlin noise now filter to clip),
> colorfilterimagefilter (new test case), imagefiltersclipped
> (perlin noise now filters to clip).
>
> Note: I de-inlined SkPaint::canComputeFastBounds() to avoid adding
> a dependency from SkPaint.h to SkImageFilter.h.h. Skia benches show
> no impact from this change, but will watch the perf bots carefully.
>
> BUG=4212
>
> Committed: https://skia.googlesource.com/skia/+/915881fe743f9a789037695f543bc6ea189cd0cbTBR=reed@google.com,senorblanco@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=4212
Review URL: https://codereview.chromium.org/1300403003
Image filters have never implemented this check, which means that
filters which affect transparent black falsely claim they can compute
their bounds.
Implemented an affectsTransparentBlack() virtual for image
filters, and a similar helper function for color filters.
This will affect the following GMs: imagefiltersscaled
(lighting, perlin noise now filter to clip),
colorfilterimagefilter (new test case), imagefiltersclipped
(perlin noise now filters to clip).
Note: I de-inlined SkPaint::canComputeFastBounds() to avoid adding
a dependency from SkPaint.h to SkImageFilter.h.h. Skia benches show
no impact from this change, but will watch the perf bots carefully.
BUG=4212
Review URL: https://codereview.chromium.org/1296943002
SkTemplates.h contains a number of Skia specific utilities which are
not designed for external use. In addition to reducing the external
support burden, this will allow Skia to freely refactor this file.
Review URL: https://codereview.chromium.org/1272293004
This also:
makes the SkLightingShader handle normal maps where the rects aren't aligned between the diffuse and normal maps.
adds a light aggregating class (Lights) to SkLightingShader (along with a Builder nested class).
Split out of https://codereview.chromium.org/1261433009/ (Add SkCanvas::drawLitAtlas call)
Review URL: https://codereview.chromium.org/1291783003
By adding a cast to SK_ARRAY_COUNT, the compiler treats the
compile-time value as if it were a const int, and like regular
numbers, permits it to be compared to signed and unsigned
numbers freely.
R=reed@google.com
Review URL: https://codereview.chromium.org/1299943002
We don't want to test small images on Gold because they are
not interested to look at. Instead, I wrote a unit test to
verify that scaling small images does not cause crashes.
BUG=skia:
Review URL: https://codereview.chromium.org/1287863004
Old flow to serialize a picture:
1) serialize picture ops
2) serialize all sub pictures recursively
3) flatten the rest of this picture into a buffer, deduping flattenable factories and typefaces as we go
4) serialize the factories and typefaces
5) serialize the bytes from 3)
This allows the data in step 5) to refer to the deduplicated factories and typefaces from step 4). But, each sub picture in step 2) is completely siloed, so they can't dedup with the parent picture or each other.
New flow:
1) serialize picture ops
2) flatten the rest of this picture into a buffer, deduping flattenable factories and typefaces as we go
3) dummy-serialize sub pictures into /dev/null, with the effect of adding any new typefaces to our dedup set
4) serialize the factories and typefaces
5) serialize the bytes from 2)
6) serialize all sub pictures recursively, with perfect deduplication because of step 3).
Now all typefaces in the top-level picture and all sub pictures recursively should end up deduplicated in the top-level typeface set.
Decoding changes are similar: we just thread through the top-level typefaces to the sub pictures. What's convenient / surprising is that this new code correctly reads old pictures if we just have each picture prefer its local typeface set over the top-level one: old pictures always just use their own typefaces, and new pictures always use the top-level ones.
BUG=skia:4092
Review URL: https://codereview.chromium.org/1233953004
Added comment about how computeInvariantOutput() is non-recursive in GrFragmentProcessor
Made isEqual() recursive in GrFragmentProcessor
BUG=skia:4182
Review URL: https://codereview.chromium.org/1287343005
Fixed wrong indent
Changed auto child advance back to backwards linear search for getting subset of coords and samplers array of a child
Used offset from parent instead of backwards linear search to find a child proc's coords and transforms in Auto...Advance
append mangleString to variable name in nameVariable()
BUILDS! Added AutoFragmentChildProcAdvance class; fixed a few errors from previous commits
BUG=skia:4182
Review URL: https://codereview.chromium.org/1286293002
Currently, SkPaintFilterCanvas does not provide any help in cloning
target canvas state. While that could be handled in subclasses, it is
easy to miss (see linked bug).
This CL adds a new constructor variant which ensures that the initial
matrix and clip bounds are inherited from the target canvas.
BUG=516790
R=reed@google.com,robertphillips@google.com
Review URL: https://codereview.chromium.org/1294013002
To do this, create SkImageCacherator, which wraps a generator and provides an
interface to get a cached answer for either the raster or texture output of
the generator.
BUG=skia:
Review URL: https://codereview.chromium.org/1291803002
Prior to this patch clients who were solely uploading to textures (e.g., SW Mask Mgr) would cause extra flushes b.c., even though kPreferNoIO was being specified, resources with pending IO would still be returned even though there was plenty of space in the resource cache.
Review URL: https://codereview.chromium.org/1286203002
This allows removal the difficult to use (and so currently unused)
placement new and related macros to allow any constructor of T to
be used to initilize the storage of SkTLazy. This also properly
aligns the SkTLazy storage.
Review URL: https://codereview.chromium.org/1283183003
Changed childProcessor(i) to return const referencd
Fixed rootProc/parentProc offset issues; renamed a few things.
added nonempty check to gatherTransforms to avoid segfault
removed recursive append_gr_coord_transforms() from GrGLProgramBuilder
BUILDS! Changed num*includeProc() calls to num() calls
added gatherCoordTransforms(). added coordTransforms() for root proc only
Modified GrFragmentProcessor to append child proc transforms and textures to root proc's arrays.
BUG=skia:4182
Review URL: https://codereview.chromium.org/1275853005
Previously, many of our codec implementations followed the same
pattern (often in a function named handleRewind):
switch (this->rewindIfNeeded()) {
case CouldNotRewind:
return CouldNotRewind;
case NoRewindNecessary:
// keep going
break;
case Rewound:
<re-read header etc>
break;
}
In this CL, remove the enum, and put the piece that happens in the
Rewound case into a virtual function, onRewind. rewindIfNeeded now
contains the common pieces from various functions named handleRewind.
In SkBmpCodec, add a function that returns whether the BMP is in ICO,
so it can have a common implementation for onRewind.
BUG=skia:3257
Review URL: https://codereview.chromium.org/1288483002
This requires adding the stroke info to the cache key, and doing the
stroking and dashing before rendering as triangles.
BUG=skia:3755
Review URL: https://codereview.chromium.org/1275553002
Paths are cached as tessellated triangle meshes in vertex buffers on the GPU. Stroked paths are not (yet) cached.
Paths containing no curved segments (linear paths) are reused at all scales. Paths containing curved segments are reused within a scale tolerance threshold.
In order to invalidate the cache when an SkPath is changed or deleted,
this required implementing genID change notification in SkPath. This is
modelled almost exactly on SkPixelRef::GenIDChangeListener.
However, It does not currently implement the check for unique genIDs,
so notifiers will fire when the first instance of an SkPathRef
using a given genID is destroyed.
Another caveat is that you cannot successfully add a change notifier
to an empty path, since it uses the "canonical" empty path which is
never modified or destroyed. For this reason, we prevent adding
listeners to it.
BUG=skia:4121,skia:4122, 497403
DOCS_PREVIEW= https://skia.org/?cl=1114353004
Committed: https://skia.googlesource.com/skia/+/468dfa72eb6694145487be17876804dfca3b7adb
Review URL: https://codereview.chromium.org/1114353004
Prior to this CL, if a client wanted to decode scanlines, they had to
create an SkCodec in order to get an SkScanlineDecoder. This introduces
complications if input data is not easily shared between the two
objects.
Instead, add methods to SkScanlineDecoder for creating a new one from
input data, and remove the creation functions from SkCodec.
Update DM and tests.
Review URL: https://codereview.chromium.org/1267583002
Does not try to cache calls to readPixels at the moment:
- not triggered by drawing
- not clear if we want to perform any pixel transformations (that readPixels allows) on the GPU or CPU
Can consider that another time.
BUG=513695
Review URL: https://codereview.chromium.org/1262923003
Paths are cached as tessellated triangle meshes in vertex buffers on the GPU. Stroked paths are not (yet) cached.
Paths containing no curved segments (linear paths) are reused at all scales. Paths containing curved segments are reused within a scale tolerance threshold.
In order to invalidate the cache when an SkPath is changed or deleted,
this required implementing genID change notification in SkPath. This is
modelled almost exactly on SkPixelRef::GenIDChangeListener.
However, It does not currently implement the check for unique genIDs,
so notifiers will fire when the first instance of an SkPathRef
using a given genID is destroyed.
Another caveat is that you cannot successfully add a change notifier
to an empty path, since it uses the "canonical" empty path which is
never modified or destroyed. For this reason, we prevent adding
listeners to it.
BUG=skia:4121,skia:4122, 497403
DOCS_PREVIEW= https://skia.org/?cl=1114353004
Review URL: https://codereview.chromium.org/1114353004
Renames Sk4pxXfermode.h to SkXfermode_opts.h,
and refactors it a tiny bit internally.
This moves xfermode optimization from being "compile-time everywhere but NEON"
to simply "runtime everywhere". I don't anticipate any effect on perf or
correctness.
BUG=skia:4117
Review URL: https://codereview.chromium.org/1264543006
With this new arrangement, the benefits of inlining sk_memset16/32 have changed.
On x86, they're not significantly different, except for small N<=10 where the inlined code is significantly slower.
On ARMv7 with NEON, our custom code is still significantly faster for N>10 (up to 2x faster). For small N<=10 inlining is still significantly faster.
On ARMv7 without NEON, our custom code is still ridiculously faster (up to 10x) than inlining for N>10, though for small N<=10 inlining is still a little faster.
We were not using the NEON memset16 and memset32 procs on ARMv8. At first blush, that seems to be an oversight, but if so it's an extremely lucky one. The ARMv8 code generation for our memset16/32 procs is total garbage, leaving those methods ~8x slower than just inlining the memset, using the compiler's autovectorization.
So, no need to inline any more on x86, and still inline for N<=10 on ARMv7. Always inline for ARMv8.
BUG=skia:4117
Review URL: https://codereview.chromium.org/1270573002
This enables the NEON sk_float_rsqrt() code for configurations that have NEON at run-time but not compile-time.
These devices will see about a 2x (1.26 -> 2.33) slowdown in sk_float_rsqrt(), but it should be more precise than our portable fallback.
(When inlined, the portable fallback and the NEON code are almost identical in speed. The only difference is precision. Going through a function pointer is causing all this slowdown. This is a good example of a place where Skia really benefits from compile-time NEON.)
BUG=skia:4117,skia:4114
No public API changes.
TBR=reed@google.com
Review URL: https://codereview.chromium.org/1264893002
