| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
| |
Set fimg.isFilter from paint.isFilterBitmap() in drawBitmap like
it's done later in drawRect_withG2D.
Change-Id: I8c27933dba28ae8834f02cb1ba66ec58dd2c933c
|
|
|
|
|
|
|
| |
* replace undefined TRUE/FALSE with true/false in SkFimgApi3x.h
* fix warning in SkFimgApi3x.cpp
Change-Id: Iabaf66df526892f006573c97450fd36f461d9bb2
|
|
|
|
| |
Change-Id: Ie0509e7d537fe2a73b23c30f5128d80ee48c1d7d
|
|
|
|
|
|
| |
http://git.insignal.co.kr/insignal/origen_quad-jb_mr1.1/android/platform/external/skia/commit/?h=origen_quad-jb_mr1.1&id=7a2cdab99f7a27e2566afbf8c0a2d3ea64c8a375
Change-Id: If605106881540f2170249e6446770ec95b4667b3
|
|
|
|
|
|
|
|
|
|
| |
Modifies the intermediate buffer size for sample pixel indexes, used in the
sample proc function calls. If the buffer is not large enough, the job is split
into several calls. This would impact the performance of SIMD optimizations.
Change-Id: I86848d1c44fe338258b40f6ee6a576289cebd496
Signed-off-by: Henrik Smiding <henrik.smiding@stericsson.com>
Signed-off-by: Patrik Ryd <patrik.ryd@stericsson.com>
|
|
|
|
|
|
|
|
|
|
| |
Adds an optimized shader for S32_Opaque_D32_nofilter_DX when
running with clamped tile mode.
Improves performance of system and various applications.
Change-Id: I8dd7be24870315b2b8273bf72d9c22eb0aeb97fb
Signed-off-by: Henrik Smiding <henrik.smiding@stericsson.com>
Signed-off-by: Patrik Ryd <patrik.ryd@stericsson.com>
|
|
|
|
|
|
|
|
|
|
| |
Adds a NEON optimized shader for SI8_opaque_D32_filter_DX when
running with clamped tile mode.
Improves performance of browser.
Change-Id: I0e1fa03d4104210953b7ec81ce10eba5f77459cc
Signed-off-by: Henrik Smiding <henrik.smiding@stericsson.com>
Signed-off-by: Patrik Ryd <patrik.ryd@stericsson.com>
|
|
|
|
|
|
|
|
|
|
|
| |
This allows skia to choose whether to use the prescaled bitmap or the
unscaled bitmap depending on which would be optimal in the current
situation. This commit also introduces two functions necessary for
properly resizing Rects when using the non-pre-scaled bitmaps.
CRs-fixed: 451681
Change-Id: I36779d8a680a3ced7c28da492e879de02c7e87a5
|
|
|
|
|
|
|
| |
This will help avoid bugs like the one fixed in change-id:
Iaeb2b84b6696e3eebf29617dfdc41c55e5226776.
Change-Id: I892423e18f9927f827393d0ff0c923259e07389c
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The destination is passed as the first argument to memcpy_76() and the
source is passed as the second argument. As originally written in
commit cb5701e, the first argument was called 'src' (despite being the
destination) and the second was called 'dst' (despite being the
source). The behavior of the memcpy_76() function was correct despite
the backward variable naming.
Commit ad673c1 reversed the naming of the two arguments to clarify the
semantics of the function but failed to reverse the naming in the ARM
assembly code. This broke the function's behavior on ARM devices.
Commits a8ce1a0 and f4be52e reversed the argument names again. This
fixed ARM support, but broke non-ARM architectures (the arguments to
memcpy() in the body of the memcpy_76() function were not also
reversed).
This change swaps the argument names again and swaps the names in the
ARM assembly, restoring correct behavior on non-ARM architectures and
aligning the variable names with the function's semantics.
Change-Id: Iaeb2b84b6696e3eebf29617dfdc41c55e5226776
|
|
|
|
|
|
|
|
|
| |
Before, memcpy_76() assumed ARM instruction set. Add preprocessor
guards to use generic memcpy() if not on ARM.
Original patch had args to memcpy_76 backwards.
Change-Id: Ib5375e9de28ab6c2b348731dcdf282d6f7a529c1
|
|
|
|
| |
This reverts commit ad673c1cb9dd151fd4e74380c4ba82c1fefb580b.
|
|
|
|
|
|
|
| |
Before, memcpy_76() assumed ARM instruction set. Add preprocessor
guards to use generic memcpy() if not on ARM.
Change-Id: Id4d7335332f4086fa33641630f41810b2fe07031
|
|
|
|
|
|
|
|
| |
* Optiming constructor of Paint object
* Special copy path for small object
* Global Language list instead of constructing a new obj everytime.
Change-Id: I8d4687fffe055166a4758059872af09bfdc973d4
|
|
|
|
|
|
|
|
|
|
|
|
| |
original function has calculation are called repetitively, and
Processed individual pixels only. So Function call overhead was reduced.
And In contrast to the serial data processing in the existing code,
the Neon code processes 8 or 4 pixels in parallel. In contrast to the
serial data processing in the existing code, the Neon code processes
8 or 4 pixels in parallel
Change-Id: Ib54aeb6bacfd4ed6d257324e8146b211f90b68e9
Signed-off-by: Lee Hwangjae <hj-yo.lee@samsung.com>
|
|\
| |
| |
| |
| |
| | |
https://android.googlesource.com/platform/external/skia into mr1-staging
Change-Id: I5cf0639d6ad25f58af4d5e0e480ee063148991f8
|
| |
| |
| |
| | |
Change-Id: I2f164c30342620e1c0663e133ba671b4a3869b70
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
It looks like Id8c91ae0be6cad8a7ef77a0cd5803676290986c1 upgraded this
message from an actual error notification into a full-fledged adb log
spammer. Oops! This CL moves it back to the location it should be at,
and requires SK_DEBUG to see it even there.
Bug: 7038312
Change-Id: Iefc0b001a9874e0f177507d43231e6fdb4a1d92a
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Cherry-pick Id8c91ae0be6cad8a7ef77a0cd5803676290986c1 from master.
During font initialization, create a seperate fallback list for each
locale. At runtime, use the fallbacklist associated with the locale
set in the paint object. Fallback files are associated with locales in
fallback_fonts.xml. Multiple files can be associated with the same
locale, ordering within that langauge and apart from that language in
the fallback order is preserved.
This changelist also includes some refactoring, notably of the
functions that call getNextContext().
Change-Id: I121f0e491a522c4a8558a0066b2d8969fb8a3667
|
| |
| |
| |
| |
| | |
bug: 6987261
Change-Id: I8cb5b6da2347a1fa88a13aa3548312a7773831e5
|
| |
| |
| | |
This reverts commit 81957ccb58db76c172e195dc3e31eba85d3e4ac9
|
| |
| |
| | |
This reverts commit 421654d4bf3ca50e2247a7389851d2e8f4812564
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Fonts can be marked with elegant or compact in fallback_fonts.xml.
Webkit uses elegant fonts, Textview uses compact fonts (the default),
unmarked fonts are used by both.
Bug: 6649136
Change-Id: I6ab6173c2efc50eba8fcc488c89dfdd083fb9c06
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This version of the fake-bold-for-fallback-fonts fix addresses the
issue succinctly inside of Skia, with no need for Skia's clients to
be aware of the issue. Attempts to draw a bold parent font flag the
embolden flag in the associated SkScalerContext while leaving the
SkPaint's fake bold flag unmodified.
This depends upon change Icdd13f6b, which arranges for FreeType's
FontHost to ignore redundant embolden requests.
Bug: 6629786
Change-Id: I4132a232896f72f632caefd79add2f00c36a578a
|
| |
| |
| |
| |
| | |
Bug: 6609231
Change-Id: Ie37ada42e7e78ab78318f3ed76eb627bffaddb1a
|
| |
| |
| |
| | |
This reverts commit 6ab870723b749ffd632fecd648704d80b581fcf0.
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
* SkPaint constructor & copy constructor
Set up a global SkTextLocale array to maintain all TextLocale
explicit small copy for struture.
* SkDraw constructor
explicit small NULL set for structure
* quickReject
Go back to explicit exponent check way instead of times zero way
Change-Id: I9b2dd08b104f32bd7b6248400189504fe25cd9f9
Conflicts:
Android.mk
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
S32A_D565_Opaque
S32A_Opaque_BlitRow32
nofilter_gether
S32A_Blend_BlitRow32
* S32A_D565_Opaque:
Use neon instructions to optimize BLIT routine which operates on
32-bit source ARGB pixel and 16-bit destination RGB pixel. Here,
we process 16-pixels in an iteration. We use regular (32-bit)
arm instructions when number of pixels is less than 16 or the
architecture doesn't support SIMD/NEON instructions. When alpha
value is 255, this version of BLIT routine performs 35% better
in comparison to S32A_D565_Opaque_neon routine.
* S32A_Opaque_BlitRow32 and nofilter_gether:
Use neon instructions to optimize BlitRow32 and nofilter_gether
functions for 32-bit source and destination pixel arrays. An improvement
of 76% is observed with neon optimized BlitRow32 function and 6%
improvement is observed with nofilter_gether function.
* S32A_Blend_BlitRow32:
Optimize S32A_Blend_BlitRow32 using neon instructions for armv7
based targets and using regular arm instructions for armv6 based
targets.
This is the consolidated commit from several patches.
Change-Id: Id5aab49656247b9a69b43cfb22fe7a8608f847e5
Apply PLD to blittering functions
Extend software pipelining for S32A_Blending to process 16 points
perloop and per-read points of next loop
Change-Id: Ib7a6a6efb8249b96578e75613683099251dc2a82
Two skia optimizations:
1. Combine S16_opaque_D32_nofilter_DX and clampx_nofilter_trans
2. Change MAX_BUF between mproc and sproc to 1024 from 128
Opt 1 is giving about 8-10 FPS in gmail and facebook apk.
Opt 2 is helping all combinations for shaderspan.
Change-Id: I88dc097ffd52984a7b786e13c70384688f2e1b12
File changes for SKIA Neon optimziation for S32_Opaque_filter_DX
Change-Id: I0e8a0506aca2a8d4ee4f4b6b1aa4dcc9ce9c0329
Fix CTS failure.
Revert opt 2 in the following commit.
Author: Xin Qi <xqi@codeaurora.org>
Date: Wed Nov 16 10:07:54 2011 -0800
Two skia optimizations:
1. Combine S16_opaque_D32_nofilter_DX and clampx_nofilter_trans
2. Change MAX_BUF between mproc and sproc to 1024 from 128
Opt 1 is giving about 8-10 FPS in gmail and facebook apk.
Opt 2 is helping all combinations for shaderspan.
CRs-Fixed: 330807
Change-Id: Iece797781e97918f4c77fc767bf9e6772927d6f3
skia: Fix a bug in clampx_nofilter_trans_S16_D32_DX
When scaling a line, in the case of the last pixel needing to be
duplicated, the beginning of the line is getting clobbered with wrong
value. This is because dest buffer pointer is not moved forward. Fix
that by adding pointer update line.
CRs-Fixed: 349571
Change-Id: Id6dc41d7e6bd2b777097f62e788bd2476609a9ae
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This version of the fake-bold-for-fallback-fonts fix addresses the
issue succinctly inside of Skia, with no need for Skia's clients to
be aware of the issue. Attempts to draw a bold parent font flag the
embolden flag in the associated SkScalerContext while leaving the
SkPaint's fake bold flag unmodified.
This depends upon change Icdd13f6b, which arranges for FreeType's
FontHost to ignore redundant embolden requests.
This is a cherry-pick of change I4132a232 from master.
Bug: 6629786
Change-Id: I389d6793e5472d664f6ac7b894927dbfd811be12
|
|
|
|
|
|
|
| |
This is a cherry-pick of Skia r3364.
bug: 6522740
Change-Id: I9f9360a0fd5d8bfd936072f0f9bae0709768a1b6
|
|
|
|
|
| |
bug:6242553
Change-Id: Ied1865b667c21d0a59586ec8b36db276f892ccbc
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Mutexes in pixelrefs were done very sloppily initially. The code (a) assumes all
pixelref subclasses want a mutex to guard their lock/unlock virtuals, and (b)
most subclasses use the same mutex for *all* of their instances, even when there
is no explicit need to guard modifying one instances with another.
When we try drawing bitmaps from multiple threads, we are seeing a lot of slow-
down from these mutexes. This CL has two changes to try to speed things up.
1. Add setPreLocked(), for pixelrefs who never need the onLockPixels
virtual to be called. This speeds up those subclasses in multithreaded environs
as it avoids the mutex lock all together (e.g. SkMallocPixelRef).
2. Add setMutex() to allow a subclass to change the mutex choice. ashmem wants
this, since its unflattening constructor cannot pass down the null, it needs
to cleanup afterwards.
see https://codereview.appspot.com/6199075/
bug: 6469917
Change-Id: I81a7cfa0b2ead5a42059697eafa58de1e7a87da2
|
|/
|
|
|
|
| |
Bug #6408362
Change-Id: I69c21d9aeeb663aa6244132ce8d598be888886e2
|
|
|
|
|
|
|
|
|
| |
This is a cherry-pick of two CL's from upstream Skia:
https://codereview.appspot.com/6092056/
https://codereview.appspot.com/6131062/
bug: 6400332
Change-Id: Ia833e7c0d4fd6e21a06980653450095176f5bd42
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| | |
- will be used for better CJK shaping
- forced to use a "placement new" operator because of use of sk_bzero / memcpy
in the Skia contructor and copy operator
Change-Id: If6f62634c9088d2401250c35239a1722102fe80c
|
|/
|
|
|
|
|
|
|
| |
anticipate overflow (or if the path is completely clipped out).
http://code.google.com/p/skia/source/detail?r=3658
bug: 6208579
Change-Id: If9749231bea465c3d35cc98c1b8e2fad3c7aeff6
|
|
|
|
|
|
|
|
|
|
|
| |
If the paint has a text point size that is small (e.g. < 1)
then the results returned from FreeType are not accurate.
This code is an adapation of the Skia CL that was reviewed
at http://codereview.appspot.com/5844060/
Change-Id: I3ca1d69b2097b47b344ec9f56a04fe94da76013d
http://code.google.com/p/android/issues/detail?id=26764
|
|
|
|
|
|
|
| |
This CL also includes changes made to Android's copy of
Skia in their J release branch.
Change-Id: Ib2baecf48004951a3ad4a1574cdc38790c814cbc
|
|
|
|
|
|
|
|
| |
This CL has companion changes to account for API updates in...
(1) frameworks/base
(2) external/webkit
Change-Id: Ibb989e76e8bd24313849f9631dbef42cdef9eb7d
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Shadowed text currently does not get subjected to culling until
immediately prior to rendering each glyph. This is problematic
for any page with an axis greater than 32k as we can't covert
the glyph coordinates to fixed point. Additionally, this is a
large perf hit as we look at every shadowed glyph on the page
for every draw call regardless of the canvas' clip.
This fix enables shadowed text to be quickly rejected based on
the canvas' clip when the draw text command is executed.
Finally, a mirror image of this CL is currently under review for
inclusion in the open-source Skia project.
bug: 5571685
Change-Id: I5df94eccecbd7d77a08004b5cbcca02120e390f7
|
|
|
|
|
|
|
| |
This CL pulls in r1835 and r1870 from Skia's trunk.
bug: 5197241
Change-Id: I79b51cf8605bc16641cfe7b3e1b369f801bf0922
|
|
|
|
|
|
|
| |
Bug: 5122491
Remove font cache purge log message
Change-Id: Ib5a23433b8cb1402c96b8d510ed875be7400c80c
|
|
|
|
|
| |
bug: 5009656
Change-Id: Ic91bdb84e42b4d07f88bd6df1a2bedc2031404df
|
|
|
|
|
|
|
| |
This change has also been submitted upstream to Skia in r1954.
bug: 5060654
Change-Id: Iac9ce0d9150c59e4db6653081d7f46843ea8f2bf
|
|
|
|
| |
Change-Id: Ief005abc9eb741ccf5efec3d1f29fe2dfc23103d
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This CL includes bug fixes and closely mirrors the version of
Skia used in Chrome M13, which is likely to be our baseline for
ICS.
The CL also adds source files for the SampleApp which will allow
us to execute basic skia tests. The SampleApp requires the
utils/views directory in order to run.
Finally, we have included the PDF backend for Skia in order to
experiment with using it to generate PDF files for certain
applications.
Note: The SampleApp and PDF code are not built as part of libskia.
Change-Id: I1895ccfbd8074e25f19148cc7bd1b4af571fb307
|
|
|
|
| |
Change-Id: I34121c66fc53aaa47aed798f7cd8e33d5977e0d8
|
|
|
|
| |
Change-Id: I46f41274d07a3d7bac4728f8841c7f5e89dc9181
|