Fix for: <http://bugzilla.opendarwin.org/show_bug.cgi?id=8425>
and <http://bugzilla.opendarwin.org/show_bug.cgi?id=6947>
Test: svg/custom/non-opaque-filters.svg
* kcanvas/device/quartz/KCanvasFilterQuartz.mm:
(WebCore::KCanvasFilterQuartz::prepareFilter):
We create an autorelease pool so we can control the deallocation of the
CIContext that we're creating. The CIContext retains the CGContext that
you pass it internally so when the CIContext is released, the
CGContext is released as well.
This is all fine and dandy unless you wrap the creation of the CIFilter
with a pair of CGBegin/EndTransparencyLayer calls which swap the context
out from underneath you. So if you start with context A,
CGBeginTransparencyLayer swaps it out and gives you B. You create a CIFilter
with it and add a reference to B. CGEndTransparencyLayer swaps out B and
gives you back A. Autorelease pool comes and cleans up the Filter, and calls
release on A, but A never got the refcount in the first place. B did. BOOM!
So we create a pool, then do a retain, then release the pool so that we
don't have to worry about the pool releasing it at a later time.
See <rdar://problem/
4647735> for reduction of CGEndTransparencyLayer case
git-svn-id: https://svn.webkit.org/repository/webkit/trunk@15603
268f45cc-cd09-0410-ab3c-
d52691b4dbfc
+2006-07-24 Dave MacLachlan <dmaclach@mac.com>
+
+ Reviewed by Darin and Alexey.
+
+ Test for: <http://bugzilla.opendarwin.org/show_bug.cgi?id=8425>
+ and <http://bugzilla.opendarwin.org/show_bug.cgi?id=6947>
+
+ * svg/custom/non-opaque-filters-expected.checksum: Added.
+ * svg/custom/non-opaque-filters-expected.png: Added.
+ * svg/custom/non-opaque-filters-expected.txt: Added.
+ * svg/custom/non-opaque-filters.svg: Added.
+
2006-07-24 Mitz Pettel <opendarwin.org@mitzpettel.com>
Reviewed by Hyatt.
--- /dev/null
+02d502476ee4f6dc4f5cfc3de57131d0
\ No newline at end of file
--- /dev/null
+KCanvasResource {id="gblur" [type=FILTER] [bounding box=at (-0.10,-0.10) size 1.20x1.20] [effect bounding box mode=0] [effects=[[type=GAUSSIAN-BLUR] [subregion="at (0,0) size 800x600"] [std dev. x=0.00 y=0.00]]]}
+layer at (0,0) size 800x600
+ RenderView at (0,0) size 800x600
+ KCanvasContainer {svg} at (0,0) size 87x40
+ KCanvasContainer {g} at (0,0) size 87x40 [opacity=0.50]
+ KCanvasItem {rect} at (0,0) size 87x40 [fill={[type=SOLID] [color=#008000]}] [filter=#gblur] [data="M0.00,0.00L87.00,0.00L87.00,40.00L0.00,40.00"]
+ RenderSVGText {text} at (0,0) size 785x18
+ RenderText {#text} at (0,0) size 139x18
+ text run at (0,0) width 139: "This should not crash."
--- /dev/null
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+<svg xmlns="http://www.w3.org/2000/svg">
+ <defs>
+ <filter id="gblur">
+ <feGaussianBlur stdDeviation="0" />
+ </filter>
+ </defs>
+ <g opacity="0.5">
+ <rect x="0" y="0" width="87" height="40" fill="green" stroke="none" filter="url(#gblur)"/>
+ </g>
+ <text fill="green" y="60">This should not crash.</text>
+</svg>
+2006-07-24 Dave MacLachlan <dmaclach@mac.com>
+
+ Reviewed by Darin and Alexey.
+
+ Fix for: <http://bugzilla.opendarwin.org/show_bug.cgi?id=8425>
+ and <http://bugzilla.opendarwin.org/show_bug.cgi?id=6947>
+
+ Test: svg/custom/non-opaque-filters.svg
+
+ * kcanvas/device/quartz/KCanvasFilterQuartz.mm:
+ (WebCore::KCanvasFilterQuartz::prepareFilter):
+ We create an autorelease pool so we can control the deallocation of the
+ CIContext that we're creating. The CIContext retains the CGContext that
+ you pass it internally so when the CIContext is released, the
+ CGContext is released as well.
+ This is all fine and dandy unless you wrap the creation of the CIFilter
+ with a pair of CGBegin/EndTransparencyLayer calls which swap the context
+ out from underneath you. So if you start with context A,
+ CGBeginTransparencyLayer swaps it out and gives you B. You create a CIFilter
+ with it and add a reference to B. CGEndTransparencyLayer swaps out B and
+ gives you back A. Autorelease pool comes and cleans up the Filter, and calls
+ release on A, but A never got the refcount in the first place. B did. BOOM!
+ So we create a pool, then do a retain, then release the pool so that we
+ don't have to worry about the pool releasing it at a later time.
+ See <rdar://problem/4647735> for reduction of CGEndTransparencyLayer case
+
2006-07-24 Mitz Pettel <opendarwin.org@mitzpettel.com>
Reviewed by Hyatt.
/*
* Copyright (C) 2005, 2006 Apple Computer, Inc. All rights reserved.
+ * Copyright (C) 2006 Dave MacLachlan (dmaclach@mac.com)
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
if (useSoftware)
contextOptions = [NSDictionary dictionaryWithObjectsAndKeys:[NSNumber numberWithBool:YES], kCIContextUseSoftwareRenderer, nil];
+ // Use of CGBegin/EndTransparencyLayer around this call causes over release
+ // of cgContext due to it being created on an autorelease pool, and released
+ // after CGEndTransparencyLayer. Create local pool to fix.
+ // <http://bugzilla.opendarwin.org/show_bug.cgi?id=8425>
+ // <http://bugzilla.opendarwin.org/show_bug.cgi?id=6947>
+ // <rdar://problem/4647735>
+ NSAutoreleasePool *filterContextPool = [[NSAutoreleasePool alloc] init];
m_filterCIContext = HardRetain([CIContext contextWithCGContext:cgContext options:contextOptions]);
+ [filterContextPool drain];
+
m_filterCGLayer = [m_filterCIContext createCGLayerWithSize:CGRect(bbox).size info:NULL];
KRenderingDeviceContext *filterContext = new KRenderingDeviceContextQuartz(CGLayerGetContext(m_filterCGLayer));
git://git.webkit.org / WebKit-https.git / commitdiff
