Make SkImage::getTexture() const. At the moment the function
does not mutate the state.
One use-case is that this makes it possible in the future to add draw
function to SkBaseDevice functions. The device draw functions take
const ref objects, but SkGpuDevice likely would benefit of using the
getTexture().
BUG=skia:3388
R=reed@google.com
Review URL: https://codereview.chromium.org/925853002
The new enum describes the nature of the failure. This is in
preparation for writing a replacement for SkImageDecoder, which will
use this interface.
Update the comments for getPixels() to specify what it means to pass
an SkImageInfo with a different size.
Make SkImageGenerator Noncopyable.
Leave onGetYUV8Planes alone, since we have separate discussions
regarding modifying that API.
Make callers of SkImageDecoder consistently handle kPartialSuccess.
Previously, some callers considered it a failure, and others considered
it a success.
BUG=skia:3257
Review URL: https://codereview.chromium.org/919693002
SkTHashTable is very similar to SkTDynamicHash, except it's generalized to support non-pointer value types.
It doesn't support remove(), just to keep things simple (it's not hard to add).
Instead of an iterator, it has foreach(), again, to keep things simple.
SkTHashMap<K,V> and SkTHashSet<T> build a friendlier experience on top of SkTHashTable.
BUG=skia:
Review URL: https://codereview.chromium.org/925613002
This allows for the removal of SkPath::mSourcePath on Android
as they now have a better indicator of whether or not the path
can be used again via the Java API.
Review URL: https://codereview.chromium.org/913413004
The language was being set to garbage, now set to part of the file name.
Add a test to ensure we continue to parse fallback directories correctly.
BUG=chromium:422180
Review URL: https://codereview.chromium.org/912053003
No algorithmic changes. The new APIs let us avoid a few ugly trips through void*,
and I've made the consume/acquire/release decision explicitly conditioned on TSAN.
This should fix the attached bug, which is TSAN seeing us implementing the
sk_consume_load() with a relaxed load, where we used to pass __ATOMIC_CONSUME
to TSAN. This restores us to the status quo of a couple weeks ago, where we
use relaxed loads (to avoid an extra dmb on ARM) for all setups except TSAN,
who gets the logically correct memory order, consume.
No public API changes.
TBR=reed@google.com
BUG=chromium:455606
Review URL: https://codereview.chromium.org/908943002
Implement SVG clips based on clip stack flattening -
which is now exposed in SkClipStack::asPath() and shared
with SkCanvas's simplify-clip code.
R=reed@google.com,mtklein@google.com
Review URL: https://codereview.chromium.org/876923003
This adds sk_memory_barrier(), implemented using sk_atomic_fetch_add() on an uninitialized variable. If that becomes a problem we can drop this to the porting layer, using std::atomic_thread_fence() / __atomic_thread_fence() / __sync_synchronize().
The big win is that ref() doesn't generate a memory barrier any more on ARM.
This is an instance of SkSafeRef() in SkPaint(const SkPaint&) after this CL:
4d0: 684a ldr r2, [r1, #4]
4d2: 6018 str r0, [r3, #0]
4d4: b13a cbz r2, 4e6 <_ZN7SkPaintC1ERKS_+0x2e>
4d6: 1d10 adds r0, r2, #4
4d8: e850 4f00 ldrex r4, [r0]
4dc: 3401 adds r4, #1
4de: e840 4500 strex r5, r4, [r0]
4e2: 2d00 cmp r5, #0
4e4: d1f8 bne.n 4d8 <_ZN7SkPaintC1ERKS_+0x20>
Here's the before, pretty much the same with two memory barriers surrounding the ref():
4d8: 684a ldr r2, [r1, #4]
4da: 6018 str r0, [r3, #0]
4dc: b15a cbz r2, 4f6 <_ZN7SkPaintC1ERKS_+0x3e>
4de: 1d10 adds r0, r2, #4
4e0: f3bf 8f5f dmb sy
4e4: e850 4f00 ldrex r4, [r0]
4e8: 3401 adds r4, #1
4ea: e840 4500 strex r5, r4, [r0]
4ee: 2d00 cmp r5, #0
4f0: d1f8 bne.n 4e4 <_ZN7SkPaintC1ERKS_+0x2c>
4f2: f3bf 8f5f dmb sy
The miscellaneous files in here are just fixups to explicitly include SkMutex.h,
instead of leeching it off SkRefCnt.h.
No public API changes.
TBR=reed@google.com
Build trybots seem hosed.
NOTRY=true
BUG=skia:
Review URL: https://codereview.chromium.org/896803002
This merges and refactors SkAtomics.h and SkBarriers.h into SkAtomics.h and
some ports/ implementations. The major new feature is that we can express
memory orders explicitly rather than only through comments.
The porting layer is reduced to four template functions:
- sk_atomic_load
- sk_atomic_store
- sk_atomic_fetch_add
- sk_atomic_compare_exchange
From those four we can reconstruct all our previous sk_atomic_foo.
There are three ports:
- SkAtomics_std: uses C++11 <atomic>, used with MSVC
- SkAtomics_atomic: uses newer GCC/Clang intrinsics, used on not-MSVC where possible
- SkAtomics_sync: uses older GCC/Clang intrinsics, used where SkAtomics_atomic not supported
No public API changes.
TBR=reed@google.com
BUG=skia:
Review URL: https://codereview.chromium.org/896553002
Decades ago Intel decided that the bsr (Bit Scan Reverse) instruction
should have undefined results if its argument is zero. This probably
makes the instruction harder to implement and it definitely makes it
more difficult to use.
In SkCLZ_portable it requires a check for a zero argument, but despite
that check /analyze still warns that _BitScanReverse might fail
(because it doesn't know what can cause failures). Because this warning
occurs in a frequently included header file it ends up being very noisy,
accounting for ~30% of all warnings (before deduplication).
Suppressing this useless warning will make the raw results easier to
look through.
Review URL: https://codereview.chromium.org/872673007
This method is not called anywhere on Android or Chrome, and it
has a FIXME that it may not be correct.
A client can still getLocalMatrix().isIdentity() if they need this
information. (It has the same FIXME, and perhaps we should revisit
it. In the meantime, this convenience method is not needed.)
Review URL: https://codereview.chromium.org/882443007
SkWriter32::snapshotAsData() is no longer performance critical.
It's only used when we're serializing to disk.
BUG=skia:2289
Review URL: https://codereview.chromium.org/875403005
This is necessary for multisampling, so that each multisampled render
target resolves before Chrome's compositor attempts to draw the
texture.
BUG=skia:
Review URL: https://codereview.chromium.org/878653004
SkTypeface already requires typeface streams to support SkStreamAsset
in practice, and in practice all users are already supplying them.
Review URL: https://codereview.chromium.org/869763002
This fixes two problems:
1) #include SK_SOME_DEFINE doesn't work well for all our clients.
2) Things in include/ are #including things in src/, which we don't like.
TBR=reed@google.com
BUG=skia:
Review URL: https://codereview.chromium.org/862983002
SkStream is a stateful object, so it does not make sense for it to have
multiple owners. Make SkStream inherit directly from SkNoncopyable.
Update methods which previously called SkStream::ref() (e.g.
SkImageDecoder::buildTileIndex() and SkFrontBufferedStream::Create(),
which required the existing owners to call SkStream::unref()) to take
ownership of their SkStream parameters and delete when done (including
on failure).
Switch all SkAutoTUnref<SkStream>s to SkAutoTDelete<SkStream>s. In some
cases this means heap allocating streams that were previously stack
allocated.
Respect ownership rules of SkTypeface::CreateFromStream() and
SkImageDecoder::buildTileIndex().
Update the comments for exceptional methods which do not affect the
ownership of their SkStream parameters (e.g.
SkPicture::CreateFromStream() and SkTypeface::Deserialize()) to be
explicit about ownership.
Remove test_stream_life, which tested that buildTileIndex() behaved
correctly when SkStream was a ref counted object. The test does not
make sense now that it is not.
In SkPDFStream, remove the SkMemoryStream member. Instead of using it,
create a new SkMemoryStream to pass to fDataStream (which is now an
SkAutoTDelete).
Make other pdf rasterizers behave like SkPDFDocumentToBitmap.
SkPDFDocumentToBitmap delete the SkStream, so do the same in the
following pdf rasterizers:
SkPopplerRasterizePDF
SkNativeRasterizePDF
SkNoRasterizePDF
Requires a change to Android, which currently treats SkStreams as ref
counted objects.
Review URL: https://codereview.chromium.org/849103004
Restructure SkGpuDevice creation:
*SkSurfaceProps are optional.
*Use SkSurfaceProps to communicate DF text rather than a flag.
*Tell SkGpuDevice::Create whether RT comes from cache or not.
Review URL: https://codereview.chromium.org/848903004
This code requires fewer macros to use it (just one), has less code in macro
definitions, and has simpler synchronization code (just atomic ints, no SkOnce,
no SkMutex, etc.)
A minor downside, we lose indentation and reverse-ordering in the final report:
Leaked SkRefCntBase: 7
Leaked SkFontMgr: 1
Leaked SkWeakRefCnt: 1
Leaked SkTypeface: 1
Leaked SkFlattenable: 3
Leaked SkXfermode: 3
Leaked SkPathRef: 1
Leaked SkPixelRef: 1
Leaked SkMallocPixelRef: 1
becomes
Leaked SkXfermode: 3
Leaked SkMallocPixelRef: 1
Leaked SkPixelRef: 1
Leaked SkPathRef: 1
Leaked SkFlattenable: 3
Leaked SkTypeface: 1
Leaked SkWeakRefCnt: 1
Leaked SkFontMgr: 1
Leaked SkRefCntBase: 7
This is motivated by wanting to land https://codereview.chromium.org/806473006/,
which makes sure all static use of SkOnce are in global scope. The current
implementation of SkInstCnt uses them in function scope, which isn't safe.
BUG=skia:
No public API changes.
TBR=reed@google.com
Review URL: https://codereview.chromium.org/841263004
This fixes every case where virtual and SK_OVERRIDE were on the same line,
which should be the bulk of cases. We'll have to manually clean up the rest
over time unless I level up in regexes.
for f in (find . -type f); perl -p -i -e 's/virtual (.*)SK_OVERRIDE/\1SK_OVERRIDE/g' $f; end
BUG=skia:
Review URL: https://codereview.chromium.org/806653007
An integer overflow is causing a memory allocation to succeed while it should fail for being too large.
BUG=445810
Review URL: https://codereview.chromium.org/831583004
There are only a handful of SkFlattenables that are not flattenable. That
there are any seems highly illogical. To make this look less like a normal
thing, this removes both macros that marked SkFlattenables as non-flattenable
(in slightly different ways).
The handful of SkFlattenables in our codebase that can't be flattened now
assert violently that they can't be flattened. They're internal or
part of animator... places where we'll never actually flatten them.
TestLooper and DummyRasterizer were so trivial that I just made them flattenable.
BUG=skia:
Review URL: https://codereview.chromium.org/841753002
Defining SK_DYNAMIC_ANNOTATIONS_ENABLED as 1 whenever DYNAMIC_ANNOTATIONS_ENABLED was 1
seems to be working fine for Chrome. Should be we can just use DYNAMIC_ANNOTATIONS_ENABLED.
BUG=skia:
Review URL: https://codereview.chromium.org/810513002
We are trying to replace Skia's NaN checker with our own in Mozilla,
so it would be nice to have to patch a single place by making sure
these NaN checks used SkScalarIsNaN().
BUG=skia:
Review URL: https://codereview.chromium.org/809443002
We can't do this unconditionally or pipe will become stupidly slow.
DM's serialize mode fails subtly on Mac when we force embedding, so I've
#ifdef'd that away. Other platforms look fine.
BUG=skia:
Review URL: https://codereview.chromium.org/796523002
Gives more flexibility to the caller to decide whether to use the
encoded data returned by refEncodedData().
Provides an implementation that supports the old version of
SkPicture::serialize().
TODO: Update Chrome, so we can remove SK_LEGACY_ENCODE_BITMAP entirely
BUG=skia:3190
Review URL: https://codereview.chromium.org/784643002
This allows us to control SK_DYNAMIC_ANNOTATIONS_ENABLED by Sk*Config.h files.
This is a no-op today, because we control it from the compiler command line.
BUG=430815
Review URL: https://codereview.chromium.org/787003003
Reason for revert:
Compilation is failing on some bots
Original issue's description:
> Replace EncodeBitmap with an interface.
>
> Gives more flexibility to the caller to decide whether to use the
> encoded data returned by refEncodedData().
>
> Provides an implementation that supports the old version of
> SkPicture::serialize().
>
> TODO: Update Chrome, so we can remove SK_LEGACY_ENCODE_BITMAP entirely
>
> BUG=skia:3190
>
> Committed: https://skia.googlesource.com/skia/+/0c4aba6edb9900c597359dfa49d3ce4a41bc5dd1TBR=reed@google.com,scroggo@google.com
NOTREECHECKS=true
NOTRY=true
BUG=skia:3190
Review URL: https://codereview.chromium.org/787833002
Gives more flexibility to the caller to decide whether to use the
encoded data returned by refEncodedData().
Provides an implementation that supports the old version of
SkPicture::serialize().
TODO: Update Chrome, so we can remove SK_LEGACY_ENCODE_BITMAP entirely
BUG=skia:3190
Review URL: https://codereview.chromium.org/784643002
Reason for revert:
break many gm's
Original issue's description:
> Make all blending up to GrOptDrawState be handled by the xp/xp factory.
>
> In this cl the blending information is extracted for the xp and stored in the ODS
> which is then used as it currently is. In the follow up cl, an XP backend will be added
> and at that point all blending work will take place inside XP's.
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/7c66342a399b529634bed0fabfaa562db2c0dbd4TBR=bsalomon@google.com
NOTREECHECKS=true
NOTRY=true
BUG=skia:
Review URL: https://codereview.chromium.org/766653008
In this cl the blending information is extracted for the xp and stored in the ODS
which is then used as it currently is. In the follow up cl, an XP backend will be added
and at that point all blending work will take place inside XP's.
BUG=skia:
Review URL: https://codereview.chromium.org/759713002
Mostly this means using SkLazyPtr for the 16-bit cache.
We can remove the copy constructor now and just ref it instead.
BUG=skia:
Review URL: https://codereview.chromium.org/769323002
The only thing the unlock methods were doing was assert their balance.
This removes the unlock methods and renames the lock methods "read".
BUG=skia:
Review URL: https://codereview.chromium.org/719213008
No subclass overrides either method.
This is just warmup. Perf is neutral. The real meat of the time spent
is inside canComputeFastBounds / computeFastBounds, not getting to them.
BUG=skia:
Review URL: https://codereview.chromium.org/772573003
This switches to a new way of doing this, enforcing the caching with the type
recorded rather than having to do it in SkRecorder. Should be more foolproof.
Updated SkPath and SkBitmap's equivalents too. ImmutableBitmap was close,
but using inheritance now makes the rest of the code less weird.
BUG=437511
I'm not sure whether or not this will _fix_ the SkMatrix aspect of that bug.
There may be other SkMatrices that we're racing on. It does cover the obvious
ones, though, and removing the SkTRacy<> wrapper will allow TSAN to show
us any other races.
It turned out to be easier to turn missing optional matrices into I early rather
than late. I figure this should be harmless. Recording and playback perf both
look neutral.
Review URL: https://codereview.chromium.org/773433003
This was needed for pictures before v33, and we're now requiring v35+.
Will follow up with the same for skia/ext/pixel_ref_utils_unittest.cc
BUG=skia:
Review URL: https://codereview.chromium.org/769953002
this just accelerates what would happen on the next beginRecording() call or the destructor. chrome has unittests that break (shaders on the stack) if we hold on to the internals past endRecording().
BUG=skia:
TBR=
Review URL: https://codereview.chromium.org/758813002
16 bits are not enough to hold the integral part of fx in
Clamp_S32_opaque_D32_nofilter_DX_shaderproc.
Weirdly, no GM diffs on my desktop.
BUG=skia:3096
Review URL: https://codereview.chromium.org/733163003
Tested with -fsanitize=signed-integer-overflow.
This new assert used to trigger in MipMap unit test.
Don't appear to be any GM diffs.
BUG=skia:
Review URL: https://codereview.chromium.org/729373004
Restores type safety with all the same features.
(Also note, less code: 29 insertions, 50 deletions.)
BUG=skia:
Review URL: https://codereview.chromium.org/746553002
SkNVRefCnt is a variant of SkRefCnt that's Not Virtual, so weighs 4 bytes
instead of 8 or 16. There's only benefit to doing this if the deriving class
does not otherwise need a vtable, e.g. SkPicture.
I've stripped out some cruft from SkPicture, rearranged fields to pack tightly,
and added compile asserts for the sizes of SkPicture, SkRecord, and
SkVarAlloc.
BUG=skia:3144
Review URL: https://codereview.chromium.org/741793002
It makes the code harder to read, and makes Debug and Release SkTDArrays
different sizes. Looks like fData is left over from when debuggers weren't
very good at inspecting data structures.
No API changes.
TBR=reed@google.com
BUG=skia:
Review URL: https://codereview.chromium.org/739263002
Idea:
1. in its mutable recording state, keep a table of drawables on the side, and store an index in the record list.
2. In "immediate-mode" draw, just call the clients drawable directly (need access to our private list to turn the stored index into a proc)
3. when we "snap", we replace the list of drawables with a list of (sub) pictures, and then during playback of the snapped picture, we invoke a private drawable which just calls "drawPicture" on the index'd subpicture.
Review URL: https://codereview.chromium.org/727363003
- move field declarations together and pack them a little tighter
- get rid of fData
- remove dead code in debugger, including unused SkPicturePlayback subclass
There are now no more long-lived SkPictureData! (Really, there never were,
but now we don't pretend to support them.)
BUG=skia:
No API changes.
TBR=reed@google.com
Review URL: https://codereview.chromium.org/725143002
This CL cleans up the existing violations and enables the
build time check to ensure that we don't regress.
The motiviation behind this change is to allow clients who include
our headers to be able to build with this warning enabled.
Review URL: https://codereview.chromium.org/726923002
We're currently overwriting the paint LCD text flag based on the the run
font data => this cancels any LCD filtering we might have performed
higher up the stack.
BUG=423362
R=reed@google.com
Review URL: https://codereview.chromium.org/718913003
This CL updates various files in the includes directory to ensure that (1) they do
not depend on headers in /src and (2) that they minimize their dependence on external
headers.
To ensure that we don't regress this behavior a new build target has been added to
build a single cpp file that contains all* public includes and is compiled with
only those directories in the include path.
* The exception is those includes that depend on OS specific headers
BUG=skia:2941
NOTRY=true
Review URL: https://codereview.chromium.org/721903002
Under the hood, add SkPixelGeometry to the CreateInfo for new devices, allowing them to see their geometry (SkDeviceProperties) up front, rather than having it changed later.
The only exception is for devices that are used on the root-layer, where we don't see the device until after the fact (at least as long as we allow clients to attach a device to a canvas externally).
We also filter the geometry when we're creating a layer, so we can disable LCD text automatically if the layer is not marked as opaque.
NOTRY=True
-- gammatext flake?
Review URL: https://codereview.chromium.org/719253002
the new virtual takes a struct which we can amend in the future w/o having to
update our subclasses in chrome.
BUG=skia:
NOTRY=True
Review URL: https://codereview.chromium.org/723743002
This CL:
1) removes the EXPERIMENTAL_optimize on SkCanvas & SkDevice
2) moves the saveLayer gathering step to endRecording
3) Replaces GPUOptimize with SkRecordComputeLayers
4) Update bench_pictures & render_pictures to provide the new flag
#2 also necessitated moving the BBH computation (and record optimization) out of SkPicture's ctor (and into endRecording)
Review URL: https://codereview.chromium.org/718443002
intended uses:
- flag a SkSurface as sRGB (only supported by Ganesh for now)
- flag images (e.g. png or jpeg) as sRGB if the codec tells us that
wins:
- faster gamma-correct text (esp. w/ distance-fields) when we can use sRGB for text
- better color fidelity when the screen really is sRGB
Review URL: https://codereview.chromium.org/676883003
This CL removes CollectLayers' reliance on having the top most picture (by removing the unused fPictureID member). This then allows making CollectLayers' API closer to that of SkRecordFillBounds in order to facilitate using them interchangeably.
Review URL: https://codereview.chromium.org/714533002
Chrome's tracing framework appears to be intentionally racy on its
quick-reject checks, trading some data loss for better performance
when disabled. People will never notice the data loss, but TSAN does.
Let's assuage TSAN with some annotations.
The 'volatile' val in SK_ANNOTATE_UNPROTECTED_WRITE was making this
not compile, but that volatile doesn't really make sense there: the value we're
writing is not what we care about, it's the destination.
CQ_EXTRA_TRYBOTS=client.skia:Test-Ubuntu13.10-GCE-NoGPU-x86_64-Release-TSAN-Trybot
No API changes.
TBR=reed
BUG=skia:
Review URL: https://codereview.chromium.org/702883002
This optimization can reduce comparison and assignments. For
geo_rect_sort benchmark, performance improved to 1.63us from 3.28us.
BUG=skia:
Review URL: https://codereview.chromium.org/695443005
This will be a bit hairy to review.
The FillBounds and CollectLayers code has diverged significantly resulting in the rendering path seeing different bounds than the hoisting path. This CL merges the FillBounds changes into CollectLayers. A follow on CL will, hopefully, find a way to layer CollectLayers on top of FillBounds.
The only code in CollectLayers that is different from FillBounds is bracketed by "LAYER HOISTING" comments.
NOTREECHECKS=true
Review URL: https://codereview.chromium.org/685263004
Motivation: SkDocument_PDF can be refactored to stream all
PDF objects as SkCanvas calls are made, to save memory!
BUG=skia:3030
BUG=skia:2683
Review URL: https://codereview.chromium.org/691783002
When unique() returns true, it must also issue an acquire barrier.
Note that this change may adversly impact SkPath performance,
but editing SkPaths is already a performance issue.
BUG=chromium:258499
No API changes.
TBR=reed@google.com
Review URL: https://codereview.chromium.org/687293002
Any path that is generated frame-to-frame should not be rendered by using the
DistanceFieldPathRenderer, because generating the initial distance field,
uploading it and rendering it takes longer than the SoftwarePathRenderer.
BUG=skia:2935
Review URL: https://codereview.chromium.org/677463002
This is basically how blink uses the filter. Currently, I can't use it for "ShadowOnly" mode with the filter at all, but instead of copying the code and risking to have the codepaths diverge, I'm simply going to add the option here.
BUG=skia:
Review URL: https://codereview.chromium.org/646213004
Add a new enum to differentiate between a complete decode and a
partial decode (with the third value being failure). Return this
value from SkImageDecoder::onDecode (in all subclasses, plus
SkImageDecoder_empty) and ::decode.
For convenience, if the enum is treated as a boolean, success and
partial success are both considered true.
Note that the static helper functions (DecodeFile etc) still return
true and false (for one thing, this allows us to continue to use
SkImageDecoder::DecodeMemory as an SkPicture::InstallPixelRefProc in
SkPicture::CreateFromStream).
Also correctly report failure in SkASTCImageDecoder::onDecode when
SkTextureCompressor::DecompressBufferFromFormat fails.
BUG=skia:3037
BUG:b/17419670
Review URL: https://codereview.chromium.org/647023006
Currently, the PDF backend does not support image filters (since PDF
does not have that functionality), so it simply removes them. This is
causing Chrome print preview to render incorrectly (see bug). The fix
here is to fall back to a raster device for image filters, as we used
to do in Blink. The resulting bitmap will be drawn to the destination
device as a normal main-memory-backed bitmap.
Note: this change invalidates the PDF results of all GMs containing
image filters (since they'll actually be rendered).
BUG=422144
Review URL: https://codereview.chromium.org/644323006
Reason for revert:
A large number of GMs on Ubuntu12 are failing. The text layout on GPU is visibly different than that for 8888.
Original issue's description:
> Change drawText() to generate positions and send to drawPosText()
>
> The idea here is to have a central place that does layout for drawText(), and
> then always feed text through drawPosText(). This both makes all of the
> GrTextContexts consistent in drawText() output, and does a better job of
> stressing drawPosText().
>
> Because of the effect of matrices on hinting and approximation error, the
> generated text is not 100% identical to that produced by the raster pipeline.
>
> BUG=skia:2778
>
> Committed: https://skia.googlesource.com/skia/+/7851a56895c9c076f73a835a7dd51d3c6180c16fTBR=cdalton.nvidia@gmail.com,bungeman@google.com,reed@google.com
NOTREECHECKS=true
NOTRY=true
BUG=skia:2778
Review URL: https://codereview.chromium.org/659993003
The idea here is to have a central place that does layout for drawText(), and
then always feed text through drawPosText(). This both makes all of the
GrTextContexts consistent in drawText() output, and does a better job of
stressing drawPosText().
Because of the effect of matrices on hinting and approximation error, the
generated text is not 100% identical to that produced by the raster pipeline.
BUG=skia:2778
Review URL: https://codereview.chromium.org/653133004
A careful reading of the preprocessor specification indicates that
any use of the 'defined' operator outside the form of 'defined X' or
'defined ( X )' directly in the constant expression of a '#if' or
'#elif' may cause undefined behavior.
In particular, msvc is very unpredictable. The 'defined X' and
'defined ( X )' forms behave differently when created from marco
expansion, with 'defined ( X )' generally evaluating to '0L'. The
'defined X' form generally behaves more the way one would expect,
but still has a number of quirks which should simply be considered
undefined behavior.
BUG=chromium:419245
Review URL: https://codereview.chromium.org/657183002
Function- or method- local scope isn't threadsafe; the pointer is generally
zero-initialized on first use in function scope (i.e. lazily... we have to go
deeper), but for globals we can be pretty sure the linker will do that for us.
BUG=skia:
No public API changes.
TBR=reed@google.com
Review URL: https://codereview.chromium.org/651723003
Before getting too far into changing how SkTileGrid stores its tiles, I figured I'd
better see how much I can tweak out the existing format. Cleverly, that way
any improvements I make by changing the format will look that much less
impressive.
This CL looks like it will be a 5-15% win in time spent recording, with no effect
on playback.
This CL also shrinks the tiles to fit exactly when we're done inserting,
using newly added SkTDArray::shrinkToFit(). It's quite cheap to run (maybe
taking back 1-2% from those 5-15% wins), and means we'll lug around about 15%
fewer bytes in the tile grids. Note though this strategy temporarily uses up to
30% more memory while building the tile grid. For our largest SKPs, that's
maybe 75-100K extra.
BUG=skia:
Committed: https://skia.googlesource.com/skia/+/52455cbc02d7f480d988ae7cdacc11ad69078c2c
CQ_EXTRA_TRYBOTS=tryserver.skia:Canary-Chrome-Ubuntu13.10-Ninja-x86_64-ToT-Trybot
Review URL: https://codereview.chromium.org/639823005
Reason for revert:
failed assertion "fXTiles * fYTiles != 0"
Original issue's description:
> Use BBH reserve hook to preallocate space for tiles.
>
> Before getting too far into changing how SkTileGrid stores its tiles, I figured I'd
> better see how much I can tweak out the existing format. Cleverly, that way
> any improvements I make by changing the format will look that much less
> impressive.
>
> This CL looks like it will be a 5-15% win in time spent recording, with no effect
> on playback.
>
> This CL also shrinks the tiles to fit exactly when we're done inserting,
> using newly added SkTDArray::shrinkToFit(). It's quite cheap to run (maybe
> taking back 1-2% from those 5-15% wins), and means we'll lug around about 15%
> fewer bytes in the tile grids. Note though this strategy temporarily uses up to
> 30% more memory while building the tile grid. For our largest SKPs, that's
> maybe 75-100K extra.
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/52455cbc02d7f480d988ae7cdacc11ad69078c2cTBR=reed@google.com,robertphillips@google.com,mtklein@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=skia:
Review URL: https://codereview.chromium.org/642933002
Before getting too far into changing how SkTileGrid stores its tiles, I figured I'd
better see how much I can tweak out the existing format. Cleverly, that way
any improvements I make by changing the format will look that much less
impressive.
This CL looks like it will be a 5-15% win in time spent recording, with no effect
on playback.
This CL also shrinks the tiles to fit exactly when we're done inserting,
using newly added SkTDArray::shrinkToFit(). It's quite cheap to run (maybe
taking back 1-2% from those 5-15% wins), and means we'll lug around about 15%
fewer bytes in the tile grids. Note though this strategy temporarily uses up to
30% more memory while building the tile grid. For our largest SKPs, that's
maybe 75-100K extra.
BUG=skia:
Review URL: https://codereview.chromium.org/639823005
This removes:
1) ability to record old pictures with SkPictureRecorder;
2) a couple tests specific to the old backend.
The functionality of DEPRECATED_beginRecording() now lives in
(private) SkPicture::Backport(), which is the only place we
need it now.
BUG=skia:
TBR=reed@google.com
Review URL: https://codereview.chromium.org/618303002
Having hoisted layers from different pictures invalidates the assumptions of the old GrReplacements object. This is fixed by switching to a SkTDynamicHash-based back-end.
Sub-picture-layers also require that the replacement drawing occur for the sub-pictures too. The ReplaceDraw object is added to make this happen and limit the replacement lookup to saveLayer draw commands.
This is split out of (Fix sub-picture layer rendering bugs - https://codereview.chromium.org/597293002/).
BUG=skia:2315
Review URL: https://codereview.chromium.org/607763008
Feature-wise, this removes:
1) BBH support;
2) peephole optimizations;
3) record-time text op specializations;
4) the guarantee that SkPaints are flattened.
This deletes the optimizations GM, which only exists to test the peepholes of
the old backend. SkRecord optimizations are unit tested, and if that ever fails we
can think about adding another GM like this, but they're different enough we'd
want to start from scratch anyway.
We need to keep the code that plays back the specialized text ops around for
a while for compatibility with existing .SKPs that have those ops recorded.
BUG=skia:
CQ_EXTRA_TRYBOTS=tryserver.skia:Canary-Chrome-Ubuntu13.10-Ninja-x86_64-ToT-Trybot
R=robertphillips@google.com, reed@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/617953002
Implement proper x/y drawTextBlob() handling by plumbing a
drawPosText() offset parameter (to act as an additional glyph pos
translation) throughout the device layer.
The new offset superceeds the existing constY, with a minor semantic
tweak: whereas previous implementations were ignoring constY in 2D
positioning mode (scalarsPerGlyph == 2), now the offset is always
observed, in all positioning modes. We can do this because existing
drawPosText() clients always pass constY == 0 for full positioning mode.
R=reed@google.com, jvanverth@google.com, robertphillips@google.com
Review URL: https://codereview.chromium.org/605533002
fDirtyBits is only used by SkPaint::FlatteningTraits, which in turn was
only used as a smaller, faster format to flatten paints in-memory to dedup
them in the old picture backend.
SkRecord obsoleted all this. Neither flatten()/unflatten() (disk format)
nor FlatteningTraits is used anywhere performance or size matters.
Here I revert the deduping code back to using the disk format for flattened paints.
We stil do have to flatten and unflatten paints while coverting from SkRecord
backend to the old backend, so we can't just delete this all yet, but any
faithful round trip flatten()/unflatten() pair will be fine, however slow.
NOTRY=true
BUG=skia:
R=reed@google.com, mtklein@google.com
Author: mtklein@chromium.org
Review URL: https://codereview.chromium.org/604813003
Refactored text blob backend for improved performance: instead of using
separate buffers for runs/positions/glyphs, everything is now packed in
a consolidated slab (including the SkTextBlob object itself!).
Benefits:
* number of allocations per blob construction reduced from ~4 to 1
(also minimizes internal fragmentation)
* run record size reduced by 8 bytes
This takes the blob construction overhead down to negligible levels
(for the current Blink uncached textblob implementation).
Unfortunately, the code is much more finicky (run merging in
particular) -- hence the assert spree.
Multi-run blobs are vulnerable to realloc storms but this is not a
problem at the moment because Blink is using one-run blobs 99% of the
time. Will be addressed in the future.
R=mtklein@google.com, reed@google.com, robertphillips@google.com
Committed: https://skia.googlesource.com/skia/+/13645ea0ea87038ebd71be3bd6d53b313069a9e4
Author: fmalita@chromium.org
Review URL: https://codereview.chromium.org/581173003
Reason for revert:
Broke the new blobshader gm.
Original issue's description:
> Souped-up SkTextBlob.
>
> Refactored text blob backend for improved performance: instead of using
> separate buffers for runs/positions/glyphs, everything is now packed in
> a consolidated slab (including the SkTextBlob object itself!).
>
> Benefits:
>
> * number of allocations per blob construction reduced from ~4 to 1
> (also minimizes internal fragmentation)
> * run record size reduced by 8 bytes
>
> This takes the blob construction overhead down to negligible levels
> (for the current Blink uncached textblob implementation).
>
> Unfortunately, the code is much more finicky (run merging in
> particular) -- hence the assert spree.
>
> Multi-run blobs are vulnerable to realloc storms but this is not a
> problem at the moment because Blink is using one-run blobs 99% of the
> time. Will be addressed in the future.
>
>
> R=reed@google.com,mtklein@google.com,robertphillips@google.com
>
> Committed: https://skia.googlesource.com/skia/+/13645ea0ea87038ebd71be3bd6d53b313069a9e4R=mtklein@google.com, reed@google.com, robertphillips@google.comTBR=mtklein@google.com, reed@google.com, robertphillips@google.com
NOTREECHECKS=true
NOTRY=true
Author: fmalita@chromium.org
Review URL: https://codereview.chromium.org/588853002
Refactored text blob backend for improved performance: instead of using
separate buffers for runs/positions/glyphs, everything is now packed in
a consolidated slab (including the SkTextBlob object itself!).
Benefits:
* number of allocations per blob construction reduced from ~4 to 1
(also minimizes internal fragmentation)
* run record size reduced by 8 bytes
This takes the blob construction overhead down to negligible levels
(for the current Blink uncached textblob implementation).
Unfortunately, the code is much more finicky (run merging in
particular) -- hence the assert spree.
Multi-run blobs are vulnerable to realloc storms but this is not a
problem at the moment because Blink is using one-run blobs 99% of the
time. Will be addressed in the future.
R=mtklein@google.com, reed@google.com, robertphillips@google.com
Author: fmalita@chromium.org
Review URL: https://codereview.chromium.org/581173003
typeface. Loads the paths using the driver's glyph loading routines.
Refactors GrPathRange to accept a PathGenerator class that it uses to
lazily initialize its paths. The client code is no longer expected to
initialize the paths in a GrPathRange; instead it must provide a
PathGenerator* instance to createPathRange().
Adds a new createGlyphs() method to GrPathRendering that creates a
range of glyph paths, indexed by glyph id. GrPathRendering implements
createGlyphs() with a PathGenerator that loads glyph paths using the
skia frameworks. GrGLPathRendering uses glMemoryGlyphIndexArrayNV()
instead, when possible, to load the glyph paths.
Removes all GlyphPathRange logic from GrStencilAndCoverTextContext.
It instead uses createGlyphs().
BUG=skia:2939
R=bsalomon@google.com, jvanverth@google.com
Author: cdalton@nvidia.com
Review URL: https://codereview.chromium.org/563283004