added capability to collect minidump and callstack if buildbot fails with heap
coruption in windows, and a NPE bug was fixed in SkPDFDocument, when document was destroyed without ever beeing used and a field was NULL + a few minor conflicts have been resolved)
git-svn-id: http://skia.googlecode.com/svn/trunk@8487 2bbb7eff-a529-9590-31e7-b0007b416f81
The problem fixed - http://code.google.com/p/skia/issues/detail?id=940 - is that getResources will recursively obtain all child resource recursively without checking for duplicates.
If we have lots of duplicates, then we try to build a very large vector (exponential with the number of nodes usually) and sooner or later we end up using too much memory and crash.
A possible solution could have been to make sure resources do not have duplicates, but that requirement is impractical, and it this leaves the solution fragile, if there is any issue in the tree, we crash.
When we emit the pdf, the large number of duplicates is not an issue, because SkPDFCatalog::addObject will deal with duplicates.
I have run the gm with --config pdf, and the images are 100% same bits, while the pdfs have the same size but some very small changes, the order of some objects.
Review URL: https://codereview.appspot.com/6744050
git-svn-id: http://skia.googlecode.com/svn/trunk@7883 2bbb7eff-a529-9590-31e7-b0007b416f81
When moving the content of a device into a PDF object like SkPDFFormXObject
or SkPDFShader does, we only need the top level resources in the new object's
resource list, not the recursive set of objects. Otherwise, when you
put a form on a form on form, etc, references to the objects multiply.
This fixed http://crbug.com/117321
Review URL: https://codereview.appspot.com/5796048
git-svn-id: http://skia.googlecode.com/svn/trunk@3360 2bbb7eff-a529-9590-31e7-b0007b416f81
This patch removes static initializers related to static and
global mutexes from the final library's machine code when
building on a pthread-capable system.
We use PTHREAD_MUTEX_INITIALIZER to perform POD-style
initialization. You need a line like the following to declare
a global mutex with it:
SkBaseMutex gMutex = { PTHREAD_MUTEX_INITIALIZER };
We introduce the SK_DECLARE_STATIC_MUTEX and SK_DECLARE_GLOBAL_MUTEX
macros to be able to declare static/global mutexes in the source tree
uniformly.
SkMutex is now defined as a sub-class of SkBaseMutex, with standard
construction/destruction semantics. This is useful if the mutex
object is a member of another C++ class, or allocated dynamically.
We also modify a few places to refer to SkBaseMutex instead of a
SkMutex, where it makes sense. Generally speaking, client code
should hold and use pointers to SkBaseMutex whenever they can
now.
We defined a new built-time macro named SK_USE_POSIX_THREADS
to indicate that we're using a pthread-based SkThread.h
interface. The macro will also be used in future patches
to implement other helper thread synchronization classes.
Finally, we inline the acquire() and release() functions in the
case of Posix to improve performance a bit.
Running: 'bench -repeat 10 -match mutex' on an Android device or
a 2.4GHz Xeon Linux desktop shows the following improvements:
Before After
Galaxy Nexus 1.64 1.45
Nexus S 1.47 1.16
Xoom 1.86 1.66
Xeon 0.36 0.31
This removes 5 static mutex initializers from the library
Review URL: https://codereview.appspot.com/5501066
git-svn-id: http://skia.googlecode.com/svn/trunk@3091 2bbb7eff-a529-9590-31e7-b0007b416f81
which require no arguments and so may be called without requiring the matrix
and clip to be completely evaluated.
De-inlines virtual functions in SkDevice interface.
git-svn-id: http://skia.googlecode.com/svn/trunk@2250 2bbb7eff-a529-9590-31e7-b0007b416f81
I have manually examined all of these diffs and restored a few files that
seem to require manual adjustment.
The following files still need to be modified manually, in a separate CL:
android_sample/SampleApp/AndroidManifest.xml
android_sample/SampleApp/res/layout/layout.xml
android_sample/SampleApp/res/menu/sample.xml
android_sample/SampleApp/res/values/strings.xml
android_sample/SampleApp/src/com/skia/sampleapp/SampleApp.java
android_sample/SampleApp/src/com/skia/sampleapp/SampleView.java
experimental/CiCarbonSampleMain.c
experimental/CocoaDebugger/main.m
experimental/FileReaderApp/main.m
experimental/SimpleCocoaApp/main.m
experimental/iOSSampleApp/Shared/SkAlertPrompt.h
experimental/iOSSampleApp/Shared/SkAlertPrompt.m
experimental/iOSSampleApp/SkiOSSampleApp-Base.xcconfig
experimental/iOSSampleApp/SkiOSSampleApp-Debug.xcconfig
experimental/iOSSampleApp/SkiOSSampleApp-Release.xcconfig
gpu/src/android/GrGLDefaultInterface_android.cpp
gyp/common.gypi
gyp_skia
include/ports/SkHarfBuzzFont.h
include/views/SkOSWindow_wxwidgets.h
make.bat
make.py
src/opts/memset.arm.S
src/opts/memset16_neon.S
src/opts/memset32_neon.S
src/opts/opts_check_arm.cpp
src/ports/SkDebug_brew.cpp
src/ports/SkMemory_brew.cpp
src/ports/SkOSFile_brew.cpp
src/ports/SkXMLParser_empty.cpp
src/utils/ios/SkImageDecoder_iOS.mm
src/utils/ios/SkOSFile_iOS.mm
src/utils/ios/SkStream_NSData.mm
tests/FillPathTest.cpp
Review URL: http://codereview.appspot.com/4816058
git-svn-id: http://skia.googlecode.com/svn/trunk@1982 2bbb7eff-a529-9590-31e7-b0007b416f81
There are a lot of small pieces to make this change work:
- SkPDFDocument (and SkPDFCatalog) take flags to disable compression (and font embedding - not implemented yet, can disable font subsetting for now).
- SkPDFStream now defers compression until the size/emit step.
- Classes that *had* a stream (because they didn't know the stream size at construction time) now *are* streams to make the substitution work correctly.
- The SkPDFShader implementation got pulled apart into two classes, one that is a SkPDFDict, and one that is a SkPDFStream (making the common ancestor SkPDFObject).
- Added helper methods in SkPDFObject for children that have simple resource lists.
- Added an iterator to SkPDFDict so that a substitute SkPDFStream can get a copy of the stream dictionary.
- Change SkPDFDocument to have a pointer to an SkPDFCatalog to remove a new circular header reference.
Review URL: http://codereview.appspot.com/4700045
git-svn-id: http://skia.googlecode.com/svn/trunk@1911 2bbb7eff-a529-9590-31e7-b0007b416f81
This stack object helper class calls finishContentEntry when it goes out of scope, maintains the current content entry, and manages the dst form xobject when it is needed. This can be made cleaner by moving the guts of SkPDFDevice into a core object, which can expose setUp/finishContentEntry as public, but that is left as a todo.
Review URL: http://codereview.appspot.com/4515126
git-svn-id: http://skia.googlecode.com/svn/trunk@1409 2bbb7eff-a529-9590-31e7-b0007b416f81
For Clear, Src, Dst/Src-In/Out, we have to consider the current clip, and
potentially draw Dst, clipped to the inverse of the current clip before
doing the operation of interest.
For clear or src, if we haven't drawn anything, or the clip is empty, there's
nothing to be done.
For Src/Dst-In/Out, if either is empty, the result is empty.
Review URL: http://codereview.appspot.com/4538082
git-svn-id: http://skia.googlecode.com/svn/trunk@1407 2bbb7eff-a529-9590-31e7-b0007b416f81
This change uses the soft mask (aka soft clip) functionality of PDF to implement the xfermodes. It has to put existing content (dst) into a form xobject as well as putting the new (src) content into a different form xobject. It then draws one of them with the other as the soft mask.
To accomplish this, we add a call to finishContentEntry after each call to setUpContentEntry - this is kind of a hack, but I don't see a better way to extract src.
Unfortunately, soft mask is specified in the Graphic State PDF object (and not in the form xobject), so when handling one of these modes, we add a one time GS object to set the soft mask and invoke a simple GS to reset the soft mask when done.
Review URL: http://codereview.appspot.com/4496041
git-svn-id: http://skia.googlecode.com/svn/trunk@1320 2bbb7eff-a529-9590-31e7-b0007b416f81
The current approach of setting the existing clip just before drawing a layer into a device doesn't work. SkDevice::clear() uses existing clip before that and if we need to put the content in a transparency group (i.e. for SrcIn xfermode), we need a valid existing clip. Instead, change the factory to use a special constructor when creating a layer device.
Review URL: http://codereview.appspot.com/4495041
git-svn-id: http://skia.googlecode.com/svn/trunk@1270 2bbb7eff-a529-9590-31e7-b0007b416f81
Instead of writing all drawing and state updates into the final content stream immediately, this change creates a new ContentEntry each time the transform, clip, or paint changes. Drawing is done into a stream in the ContentEntry. When the consumer asks for the content, we combine all the ContentEntries with appropriate updates to the state (clip, transform, paint) in between. This allows us to modify the clip even after a drawing has completed. It also lets us remove ContentEntries with no drawing. Further optimization can be done to better use the stack features of PDF, for now we follow the previous model of having a single clip followed by a single transform on the graphic state stack.
Push rectangle logic into SkPDFUtil::AppendRectangle.
Change private functions to adhere to coding standards.
Review URL: http://codereview.appspot.com/4459041
git-svn-id: http://skia.googlecode.com/svn/trunk@1269 2bbb7eff-a529-9590-31e7-b0007b416f81
The width and height we pass to SkDevice must be postive.
Shader can no longer use negative coordinates (without transform).
Shader unflip matrix should use same values as passed to SkPDFDevice (height).
Most Shader dictionary entries should be scalars and not ints.
Review URL: http://codereview.appspot.com/4454047
git-svn-id: http://skia.googlecode.com/svn/trunk@1219 2bbb7eff-a529-9590-31e7-b0007b416f81
The argument still has a default value that does what most users will want, but provides more flexibility.
Chrome will use this change to support an initial translation of the origin to simulate a margin and to scale the entire content (needed on Windows).
When landing to Chrome, this will need http://codereview.chromium.org/6820038
Review URL: http://codereview.appspot.com/4373052
git-svn-id: http://skia.googlecode.com/svn/trunk@1111 2bbb7eff-a529-9590-31e7-b0007b416f81
- Shaders, or as they are referred to in PDF, patterns, are drawn in the coordinate system of the initial page, so when we canonicalize them, we have to consider the current transform and where they are constructed.
- Image shaders are tiled by default, this makes repeat and mirror modes easy, but means we have to draw a pattern as large as the current clip to support clamp mode.
- Gradient shaders are implemented with type 4 functions, which are basically small snippets of post script code. I've tried to make the code generation modular and heavily commented to make it easy to understand or expand.
Review URL: http://codereview.appspot.com/4239061
git-svn-id: http://skia.googlecode.com/svn/trunk@905 2bbb7eff-a529-9590-31e7-b0007b416f81
