Fix a bunch of places where mNativeBitmap was being
poked at directly, switch them either to the NDK API
or to GraphicsJNI where it made sense
Change-Id: I6b3df3712d6497cba828c2d3012e725cb4ebb64d
When I moved this code here from external/skia, I accidentally changed
the lowercase 'm' to uppercase 'M'. Change back, to keep consistent
with SkXfermode and Xfermode.cpp.
Change-Id: I9694d29a22aaed7fcc8708f5962e3d2032297d08
We are removing it from Skia, so we need it here to support Android.
Add some small cleanups (remove comment that doesn't apply, convert
SK_OVERRIDE to override, remove 'virtual' keyword from methods with
'override' on them).
BUG:skbug.com/3329
Change-Id: I1f883082d6fb9d49b9c9ba9e1f50bd713eabf915
Changes proposed by Ben Hawkes of Google Project Zero.
NinePatchPeeker.cpp:
Instead of asserting, return false for bad data.
ResourceTypes.h:
Store ninepatch values as unsigned.
BUG:19151999
Change-Id: Ibe35e7569f632c6bb8a34a7701e26bb6ed547ec2
SkFontHost is now almost removed from Skia, and nothing is left in it
which interests Android. Remove the last few references to it to aid
in final removal.
Change-Id: I7d59cf7e08bdd5b9c3923c9ffaf03b1cc2f9f7b4
bug:18921087
This fixes issues where bitmaps are cached as textures, and not
reuploaded when reused by BitmapRegionDecoder, showing stale content.
Change-Id: I7626fd6acd36d8e38723689734855ce9154eb8c8
After "Use at least SkStreamAsset for SkTypefaces." there is now dead
code in AssetStreamAdapter. Remove the code and update the users.
Change-Id: I17a9d82dd6ca4d87a951c1570dc91de7f2fc54f2
The existing AssetStreamAdapter is only SkStreamRewindable, and does not
support 'duplicate'. This is needed in order for SkTypeface to be used
properly. Also, SkTypeface is moving to require SkStreamAsset for
creating typefaces for performance and code reasons.
In the previous code, Asset::getBuffer is called on creation of the
typeface, so do so explicitly and manage the memory directly. This
also prevents additional copies being made of the asset data.
BUG: 18867034
Change-Id: I458a8ec024efefb761138178b87b88b48cb4a773
This enables us to...
1) simplify the lifecycle/ownership between Java and HWUI
2) remove DisplayListRenderer::drawBitmapData and associated logic
3) track pixel lifecycle using standard SkPixelRef refcounting
4) Remove uncessary calls to ref/unref the bitmap's pixels and colorTable
Change-Id: I3c95078da20995444f6388a029414280fd654318
Adaptations to changes in Skia upstream since Oct 2014.
(https://code.google.com/p/skia/issues/detail?id=2377)
cbb922d use new roundOut signature
77082de Call SkColorTable::readColors().
5456ab1 Handle the result of SkMatrix::asAffine.
25cfc78 Remove dependence on SK_LEGACY_PICTURE_SIZE_API.
BUG:18468293
Change-Id: I5a4d274e854298843891410b1ffd5e24f038b88a
Incrementally unify the upper layers for Skia and HWUI.
Remove redundant code from GLES20Canvas.java; instead
use inherited mNativeCanvasWrapper and superclass method
definitions.
Moves some unrelated SkPaint utility functions from Renderer
to new utils/PaintUtils.
bug: 15672762
Change-Id: I4ddd4214b8e9eeb95289d054ef423f2542bb5fa5
There are two callers of SkiaCanvas::create_canvas(SkCanvas*). In both
cases, we were calling ref() first. It is necessary to call ref() in
both cases, since we have an SkCanvas returned by
SkPictureRecorder::beginRecording, which does not increment the ref
count to account for the caller. (i.e. the SkPictureRecorder has the
only ref until you call ref()).
Rather than leave the ref up to the caller, since the SkiaCanvas does
the unref(), it should also do the ref(). Update it to do so and
document its behavior.
Also, make SkiaCanvas' constructors explicit.
Change-Id: I894d0a71a87587cf8c2b26deb2384dc49ae090ef
