70564 2011-10-20 16:54:53 -0700 [Chromium] Make root layer always opaque 2011-12-06 12:17:22 -0800 1 1 1 Unclassified WebKit New Bugs 528+ (Nightly build) Unspecified Unspecified RESOLVED FIXED P2 Normal --- 0 danakj danakj backer cc-bugs jamesr piman webkit.review.bot wjmaclean oldest_to_newest 488001 0 danakj 2011-10-20 16:54:53 -0700 [Chromium] Make root layer always opaque 488006 1 111875 danakj 2011-10-20 17:06:49 -0700 Created attachment 111875 Patch 488009 2 danakj 2011-10-20 17:10:16 -0700 Attempt to make the root layer marked as opaque, for #70085. 488024 3 111875 jamesr 2011-10-20 17:26:07 -0700 Comment on attachment 111875 Patch I'd rather we set the bit instead of overriding what 'opaque()' returns. Can you look at where we set the "NonComposited" bit and call setOpaque() there instead (http://google.com/codesearch#OAMlx_jo-ck/src/third_party/WebKit/Source/WebCore/platform/graphics/chromium/NonCompositedContentHost.cpp&q=setIsNonCompo&exact_package=chromium&l=45)? 488350 4 danakj 2011-10-21 06:38:34 -0700 Here's why I thought it would be better to override opaque() instead: If we set the bit for a NonComposited surface, we will have have to put checks in the code in every other place that unsets the bit to check if it's NonComposited. And that makes more special case code than we have now. Am I wrong about this? If not, we could check isNonComposited() rather than !m_parent in opaque(). I had thought you meant to remove the NonCompositedLayer entirely when you talked about removing special cases, but I guess that wasn't your intent. 488564 5 jamesr 2011-10-21 12:38:17 -0700 (In reply to comment #4) > Here's why I thought it would be better to override opaque() instead: If we set the bit for a NonComposited surface, we will have have to put checks in the code in every other place that unsets the bit to check if it's NonComposited. And that makes more special case code than we have now. Am I wrong about this? > > If not, we could check isNonComposited() rather than !m_parent in opaque(). > > I had thought you meant to remove the NonCompositedLayer entirely when you talked about removing special cases, but I guess that wasn't your intent. Sorry, I wasn't expressive enough. The current state of the world is that we have a NonCompositedContentHost object that hosts a GraphicsLayer and so, indirectly, a LayerChromium. We want this LayerChromium to have some special behaviors (blending disabled, no border texels). The way that is done today is by setting a special flag on the LayerChromium which is then propagated through the system and used in various places. I think this is wrong and we should instead have NonCompositedContentHost set the more relevant bits for the behaviors it wants. For example, to disable blending it should just set the opaque bit and then the compositor would know to disable blending. Similarly we should be able to detect when we need border texels without relying on this bit (although that's a separate issue). So my feedback is this: change NonCompositedContentHost to set the opaque bit correctly on its LayerChromium and then use that to control blending instead of the isNonCompositedContent() bit. You shouldn't be messing around with how isOpaque() works, that should definitely still just be a bool on LayerChromium that works in the normal way. 512092 6 danakj 2011-11-30 13:46:17 -0800 One question: should a root layer that is rotated 45 degrees have blending enabled to anti-alias its edges? 512145 7 117277 danakj 2011-11-30 14:41:46 -0800 Created attachment 117277 Patch I didn't change the blending decision in this patch, because the blending decision in LayerRenderChromium needs to be made based on rotation transform, and I'm not clear if this should affect non-composited layers or how. Bug #72965 adds some functions to decide if a layer is being drawn opaque, and they could be used if the decision should be the same for noncompositing opaque layers as for other opaque layers. There are no tests for this right now, since the opaque flag is overridden by isNonComposited still, so layout tests won't see any change. I could make a unit test but I'm not sure it's possible to properly instantiate a NonCompositedContentHost in a unit test since it touches non-chromium WebCore stuff, is it? 512215 8 jamesr 2011-11-30 16:18:59 -0800 (In reply to comment #6) > One question: should a root layer that is rotated 45 degrees have blending enabled to anti-alias its edges? We don't support rotating the root layer or edge AA'ing it. 512216 9 jamesr 2011-11-30 16:20:48 -0800 Sorry for being thick but why can't we just call GraphicsLayer::setContentsOpaque() from NonCompositedContentHost? 512221 10 jamesr 2011-11-30 16:30:25 -0800 Example: http://google.com/codesearch#OAMlx_jo-ck/src/third_party/WebKit/Source/WebKit2/WebProcess/WebPage/ca/LayerTreeHostCA.cpp&q=setContentsOpaque&ct=rc&exact_package=chromium&cd=14&sq=&l=68 512751 11 danakj 2011-12-01 07:22:29 -0800 (In reply to comment #9) > Sorry for being thick but why can't we just call GraphicsLayer::setContentsOpaque() from NonCompositedContentHost? Hm. I thought we can't because the RenderLayerBacking will be setting contentsOpaque true/false based on the contents. But perhaps there is no backing since it is non-composited. I will read more code and verify this, thanks. 512788 12 danakj 2011-12-01 08:22:40 -0800 (In reply to comment #11) > (In reply to comment #9) > > Sorry for being thick but why can't we just call GraphicsLayer::setContentsOpaque() from NonCompositedContentHost? > > Hm. I thought we can't because the RenderLayerBacking will be setting contentsOpaque true/false based on the contents. But perhaps there is no backing since it is non-composited. I will read more code and verify this, thanks. So the non-composited content host does have a backing. RenderLayerCompositor::updateBacking() creates the NonCompositingContentHost, via RenderLayerCompositor::enableCompositingMode(). Then it creates a backing for it via RenderLayer::ensureBacking(). So, if we can know the NCCH will have contents that cause it to be set opaque in RenderLayerBacking eventually (eg. solid background color, background-clip:border-box), I can just add a call in NCCH to setContentsOpaque(true) until the RenderLayerBacking stuff lands, and then remove it. I am not sure how to verify this one though. 512868 13 danakj 2011-12-01 10:05:48 -0800 (In reply to comment #12) > So, if we can know the NCCH will have contents that cause it to be set opaque in RenderLayerBacking eventually (eg. solid background color, background-clip:border-box), I can just add a call in NCCH to setContentsOpaque(true) until the RenderLayerBacking stuff lands, and then remove it. The NCCH has a background color of 0, which is transparent. I am guessing this is the default color given to things which cannot be styled. This makes the RenderLayerBacking mark the GraphicsLayer as non-opaque, over-riding the flag set in the NCCH constructor. It would be possible to change the color on the NCCH, though I think this is easier. Maybe you know something more about the situation also, so what do you think of this all? 512987 14 jamesr 2011-12-01 12:21:29 -0800 The NCCH's GraphicsLayer does not have any associated RenderLayerBacking. I'm a bit surprised that RenderLayerCompositor can manipulate it at all. Can you show a callstack of RLB overriding the opaque flag? I wonder if we're confusing some layers somewhere. The GraphicsLayer tree should look something like this: (A) GraphicsLayer - NonCompositedContentHost (B) GraphicsLayer - root of composited tree RLC/RLB should only be messing with GraphicsLayer (B) and below. The GraphicsLayerClient for (A) is the NonCompositedContentHost 513011 15 danakj 2011-12-01 12:51:30 -0800 Wall of text incoming. Here's backtrace for creating the NCCH: #0 WebCore::NonCompositedContentHost::create (contentPaint=...) at ../../third_party/WebKit/Source/WebCore/platform/graphics/chromium/NonCompositedContentHost.h:51 #1 0x00007f907f77be4a in WebKit::WebViewImpl::setIsAcceleratedCompositingActive (this=0x7f906f7ddc80, active=true) at ../../third_party/WebKit/Source/WebKit/chromium/src/WebViewImpl.cpp:2857 #2 0x00007f907f77b439 in WebKit::WebViewImpl::setRootGraphicsLayer (this=0x7f906f7ddc80, layer=0x7f906f8f8200) at ../../third_party/WebKit/Source/WebKit/chromium/src/WebViewImpl.cpp:2717 #3 0x00007f907f6b2e5b in WebKit::ChromeClientImpl::attachRootGraphicsLayer (this=0x7f906f7ddcd0, frame=0x7f90737b2800, graphicsLayer=0x7f906f8f8200) at ../../third_party/WebKit/Source/WebKit/chromium/src/ChromeClientImpl.cpp:813 #4 0x00007f90803ac8c6 in WebCore::RenderLayerCompositor::attachRootLayer (this=0x7f906f7bfe70, attachment=WebCore::RenderLayerCompositor::RootLayerAttachedViaChromeClient) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerCompositor.cpp:1957 #5 0x00007f90803ac348 in WebCore::RenderLayerCompositor::ensureRootLayer (this=0x7f906f7bfe70) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerCompositor.cpp:1897 #6 0x00007f90803a5a6f in WebCore::RenderLayerCompositor::enableCompositingMode (this=0x7f906f7bfe70, enable=true) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerCompositor.cpp:127 #7 0x00007f90803a64ac in WebCore::RenderLayerCompositor::updateBacking (this=0x7f906f7bfe70, layer=0x7f906f7f2178, shouldRepaint=WebCore::RenderLayerCompositor::CompositingChangeRepaintNow) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerCompositor.cpp:354 #8 0x00007f90803a677b in WebCore::RenderLayerCompositor::updateLayerCompositingState (this=0x7f906f7bfe70, layer=0x7f906f7f2178, shouldRepaint=WebCore::RenderLayerCompositor::CompositingChangeRepaintNow) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerCompositor.cpp:424 #9 0x00007f908039665e in WebCore::RenderLayer::styleChanged (this=0x7f906f7f2178, oldStyle=0x0) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayer.cpp:4296 The RenderLayer is 0x7f906f7f2178. Here's the backtrace for creating the RLB on the same layer: #0 WebCore::RenderLayer::ensureBacking (this=0x7f906f7f2178) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayer.cpp:3867 #1 0x00007f90803a651e in WebCore::RenderLayerCompositor::updateBacking (this=0x7f906f7bfe70, layer=0x7f906f7f2178, shouldRepaint=WebCore::RenderLayerCompositor::CompositingChangeRepaintNow) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerCompositor.cpp:366 #2 0x00007f90803a677b in WebCore::RenderLayerCompositor::updateLayerCompositingState (this=0x7f906f7bfe70, layer=0x7f906f7f2178, shouldRepaint=WebCore::RenderLayerCompositor::CompositingChangeRepaintNow) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerCompositor.cpp:424 #3 0x00007f908039665e in WebCore::RenderLayer::styleChanged (this=0x7f906f7f2178, oldStyle=0x0) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayer.cpp:4296 Here's the backtrace setting opaque in the backing owned by that same layer: #0 WebCore::GraphicsLayerChromium::setContentsOpaque (this=0x7f906f8f8e00, opaque=false) at ../../third_party/WebKit/Source/WebCore/platform/graphics/chromium/GraphicsLayerChromium.cpp:262 #1 0x00007f90803a1d62 in WebCore::RenderLayerBacking::updateContentsOpaque (this=0x7f906aff0850, opaqueRegion=...) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerBacking.cpp:1009 #2 0x00007f908039e6e0 in WebCore::RenderLayerBacking::updateCompositedBounds (this=0x7f906aff0850) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerBacking.cpp:236 #3 0x00007f90803a8d7a in WebCore::RenderLayerCompositor::rebuildCompositingLayerTree (this=0x7f906f7bfe70, layer=0x7f906f7f2178, compositingState=..., childLayersOfEnclosingLayer=...) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerCompositor.cpp:948 #4 0x00007f90803a9112 in WebCore::RenderLayerCompositor::rebuildCompositingLayerTree (this=0x7f906f7bfe70, layer=0x7f906f7f22d8, compositingState=..., childLayersOfEnclosingLayer=...) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerCompositor.cpp:1009 #5 0x00007f90803a9112 in WebCore::RenderLayerCompositor::rebuildCompositingLayerTree (this=0x7f906f7bfe70, layer=0x7f906f7f29b8, compositingState=..., childLayersOfEnclosingLayer=...) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerCompositor.cpp:1009 #6 0x00007f90803a62eb in WebCore::RenderLayerCompositor::updateCompositingLayers (this=0x7f906f7bfe70, updateType=WebCore::CompositingUpdateAfterLayoutOrStyleChange, updateRoot=0x7f906f7f29b8) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerCompositor.cpp:318 Here you can see the owning layer matches: (gdb) frame 1 #1 0x00007f90803a1d62 in WebCore::RenderLayerBacking::updateContentsOpaque (this=0x7f906aff0850, opaqueRegion=...) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerBacking.cpp:1009 1009 m_graphicsLayer->setContentsOpaque(layerRegion.isEmpty()); (gdb) print m_owningLayer $7 = (WebCore::RenderLayer *) 0x7f906f7f2178 It does the updateCompositedBounds here because the layer has a backing. I am not sure what the other layers are in the rebuildCompositingLayerTree callstack, or if that is important information. 513219 16 jamesr 2011-12-01 15:58:14 -0800 007f908039665e in WebCore::RenderLayer::styleChanged (this=0x7f906f7f2178, oldStyle=0x0) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayer.cpp:4296 > > Here's the backtrace setting opaque in the backing owned by that same layer: > #0 WebCore::GraphicsLayerChromium::setContentsOpaque (this=0x7f906f8f8e00, opaque=false) at ../../third_party/WebKit/Source/WebCore/platform/graphics/chromium/GraphicsLayerChromium.cpp:262 > #1 0x00007f90803a1d62 in WebCore::RenderLayerBacking::updateContentsOpaque (this=0x7f906aff0850, opaqueRegion=...) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerBacking.cpp:1009 > #2 0x00007f908039e6e0 in WebCore::RenderLayerBacking::updateCompositedBounds (this=0x7f906aff0850) at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerBacking.cpp:236 > #3 0x00007f90803a8d7a in WebCore::RenderLayerCompositor::rebuildCompositingLayerTree (this=0x7f906f7bfe70, layer=0x7f906f7f2178, compositingState=..., childLayersOfEnclosingLayer=...) > at ../../third_party/WebKit/Source/WebCore/rendering/RenderLayerCompositor.cpp:948 This part of the callstack (in particular RenderLayerBacking::updateContentsOpaque) does not exist in my checkout. Do you have some other patches applied locally? I think this code needs to be aware of the frameview. Are you sure this is pushing to the NCCH::m_graphicsLayer? Remember that the GraphicsLayer and RenderLayerBacking trees are *not* isomorphic. If that's the case, this leads me to believe that we probably shouldn't do anything at all in NonCompositedContentHost. We should just rely on the normal contents opaque setting logic to detect opaque FrameViews correctly and set the contentsOpaque bit. 513879 17 danakj 2011-12-02 10:02:58 -0800 If we move contentsOpaque (or whatever other incarnation) to CCLayerDelegate, as brought up in bug #72964, I think we can dodge this whole bullet, at least for this CL. We can just set the opaque flag directly on the LayerChromium then and there is no conflict to worry about. 515913 18 118051 danakj 2011-12-06 07:52:48 -0800 Created attachment 118051 Patch Just sets the opaque flag, and uses it to avoid blending. The backing can avoid conflicting with this by not setting contentsOpaque when drawsContent is false. The NCCH flag is still used to decide to use the opaque shader program or not. I think we should use the methods in bug #72965 to make that decision instead. I don't want to start duplicating the logic from LRC for checking "draw this opaque" over here. 516063 19 118051 jamesr 2011-12-06 11:23:01 -0800 Comment on attachment 118051 Patch R=me 516139 20 118051 webkit.review.bot 2011-12-06 12:17:17 -0800 Comment on attachment 118051 Patch Clearing flags on attachment: 118051 Committed r102164: <http://trac.webkit.org/changeset/102164> 516140 21 webkit.review.bot 2011-12-06 12:17:22 -0800 All reviewed patches have been landed. Closing bug. 111875 2011-10-20 17:06:49 -0700 2011-11-30 14:41:28 -0800 Patch bug-70564-20111020200648.patch text/plain 2216 danakj U3VidmVyc2lvbiBSZXZpc2lvbjogOTgwMjcKZGlmZiAtLWdpdCBhL1NvdXJjZS9XZWJDb3JlL0No YW5nZUxvZyBiL1NvdXJjZS9XZWJDb3JlL0NoYW5nZUxvZwppbmRleCBjYThjZGMzNDMyMDljMjg1 ZDNjZmUzMjQyZGQ5ZWVmOGIwMDRmZTJjLi4wZmYwZmUwNTJiZTZmZTAwYWU2ZmYyMjM4YTIyZmVk NzEwYjZjZDg0IDEwMDc1NQotLS0gYS9Tb3VyY2UvV2ViQ29yZS9DaGFuZ2VMb2cKKysrIGIvU291 cmNlL1dlYkNvcmUvQ2hhbmdlTG9nCkBAIC0xLDMgKzEsMTQgQEAKKzIwMTEtMTAtMjAgIERhbmEg SmFuc2VucyAgPGRhbmFrakBjaHJvbWl1bS5vcmc+CisKKyAgICAgICAgW0Nocm9taXVtXSBNYWtl IHJvb3QgbGF5ZXIgYWx3YXlzIG9wYXF1ZQorICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9y Zy9zaG93X2J1Zy5jZ2k/aWQ9NzA1NjQKKworICAgICAgICBSZXZpZXdlZCBieSBKYW1lcyBSb2Jp bnNvbi4KKworICAgICAgICAqIHBsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL0xheWVyQ2hyb21p dW0uY3BwOgorICAgICAgICAoV2ViQ29yZTo6TGF5ZXJDaHJvbWl1bTo6b3BhcXVlKTogUmV0dXJu IHRydWUgaWYgdGhlIGxheWVyIGlzIGEgcm9vdCBsYXllciAoaWUuIGhhcyBubyBwYXJlbnQpLgor ICAgICAgICAqIHBsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL0xheWVyQ2hyb21pdW0uaDoKKwog MjAxMS0xMC0yMCAgQWxleGV5IFByb3NrdXJ5YWtvdiAgPGFwQGFwcGxlLmNvbT4KIAogICAgICAg ICBSRUdSRVNTSU9OIChyOTY4MjMpOiBDb250ZXh0dWFsIG1lbnUgY2xvc2VzIGltbWVkaWF0ZWx5 IHdoZW4gY29udHJvbC1jbGlja2luZyBpbiBGbGFzaCBwbHVnLWluCmRpZmYgLS1naXQgYS9Tb3Vy Y2UvV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGljcy9jaHJvbWl1bS9MYXllckNocm9taXVtLmNwcCBi L1NvdXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL0xheWVyQ2hyb21pdW0u Y3BwCmluZGV4IGJlOGU3OTA2ZmNkMzIzNjIzZjRkNWQ5MTk3MDFiNzBlY2RlMGE5YmUuLjJlMzE3 ZTllYzA1NTVlODMxZDYzZjYyNTQyMTZiNjg4MmM4OTM0N2EgMTAwNjQ0Ci0tLSBhL1NvdXJjZS9X ZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL0xheWVyQ2hyb21pdW0uY3BwCisrKyBi L1NvdXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL0xheWVyQ2hyb21pdW0u Y3BwCkBAIC0yNzQsNiArMjc0LDExIEBAIHZvaWQgTGF5ZXJDaHJvbWl1bTo6cmVzZXROZWVkc0Rp c3BsYXkoKQogICAgIG1fZGlydHlSZWN0ID0gRmxvYXRSZWN0KCk7CiB9CiAKK2Jvb2wgTGF5ZXJD aHJvbWl1bTo6b3BhcXVlKCkgY29uc3QKK3sKKyAgICByZXR1cm4gbV9vcGFxdWUgfHwgIW1fcGFy ZW50OworfQorCiB2b2lkIExheWVyQ2hyb21pdW06OnB1c2hQcm9wZXJ0aWVzVG8oQ0NMYXllcklt cGwqIGxheWVyKQogewogICAgIGxheWVyLT5zZXRBbmNob3JQb2ludChtX2FuY2hvclBvaW50KTsK ZGlmZiAtLWdpdCBhL1NvdXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL0xh eWVyQ2hyb21pdW0uaCBiL1NvdXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVt L0xheWVyQ2hyb21pdW0uaAppbmRleCBjZWRmMmVjNTIwMTllNWI1ZTc4ZmQyNjhlYTE1YzliYzk2 NTMxOGU5Li5hMjI2MWYzNDA4NDkwNjdlYjM3ZDcxZWQyZDc3NjE4NmI3NjM4YmNmIDEwMDY0NAot LS0gYS9Tb3VyY2UvV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGljcy9jaHJvbWl1bS9MYXllckNocm9t aXVtLmgKKysrIGIvU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vZ3JhcGhpY3MvY2hyb21pdW0vTGF5 ZXJDaHJvbWl1bS5oCkBAIC0xMTMsNyArMTEzLDcgQEAgcHVibGljOgogICAgIGZsb2F0IG9wYWNp dHkoKSBjb25zdCB7IHJldHVybiBtX29wYWNpdHk7IH0KIAogICAgIHZvaWQgc2V0T3BhcXVlKGJv b2wgb3BhcXVlKSB7IG1fb3BhcXVlID0gb3BhcXVlOyBzZXROZWVkc0NvbW1pdCgpOyB9Ci0gICAg Ym9vbCBvcGFxdWUoKSBjb25zdCB7IHJldHVybiBtX29wYXF1ZTsgfQorICAgIGJvb2wgb3BhcXVl KCkgY29uc3Q7CiAKICAgICB2b2lkIHNldFBvc2l0aW9uKGNvbnN0IEZsb2F0UG9pbnQmIHBvc2l0 aW9uKSB7IG1fcG9zaXRpb24gPSBwb3NpdGlvbjsgIHNldE5lZWRzQ29tbWl0KCk7IH0KICAgICBG bG9hdFBvaW50IHBvc2l0aW9uKCkgY29uc3QgeyByZXR1cm4gbV9wb3NpdGlvbjsgfQo= 117277 2011-11-30 14:41:46 -0800 2011-12-06 07:52:42 -0800 Patch bug-70564-20111130174144.patch text/plain 4368 danakj U3VidmVyc2lvbiBSZXZpc2lvbjogMTAxNTM3CmRpZmYgLS1naXQgYS9Tb3VyY2UvV2ViQ29yZS9D aGFuZ2VMb2cgYi9Tb3VyY2UvV2ViQ29yZS9DaGFuZ2VMb2cKaW5kZXggMmYwZTczMGQ4MTFiMGRj YmJjYjY2OTg0ZWE5ZGI4ZTYzNzg3OGM2MC4uNmMzMmU4MWVjZmViN2Y2NDc0MmNhYWFjZTdkZTFi ZDk4N2U1MGM0ZSAxMDA2NDQKLS0tIGEvU291cmNlL1dlYkNvcmUvQ2hhbmdlTG9nCisrKyBiL1Nv dXJjZS9XZWJDb3JlL0NoYW5nZUxvZwpAQCAtMSwzICsxLDE4IEBACisyMDExLTExLTMwICBEYW5h IEphbnNlbnMgIDxkYW5ha2pAY2hyb21pdW0ub3JnPgorCisgICAgICAgIFtDaHJvbWl1bV0gTWFr ZSByb290IGxheWVyIGFsd2F5cyBvcGFxdWUKKyAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtpdC5v cmcvc2hvd19idWcuY2dpP2lkPTcwNTY0CisKKyAgICAgICAgUmV2aWV3ZWQgYnkgTk9CT0RZIChP T1BTISkuCisKKyAgICAgICAgKiBwbGF0Zm9ybS9ncmFwaGljcy9jaHJvbWl1bS9HcmFwaGljc0xh eWVyQ2hyb21pdW0uY3BwOgorICAgICAgICAoV2ViQ29yZTo6R3JhcGhpY3NMYXllckNocm9taXVt OjpHcmFwaGljc0xheWVyQ2hyb21pdW0pOgorICAgICAgICAoV2ViQ29yZTo6R3JhcGhpY3NMYXll ckNocm9taXVtOjp1cGRhdGVDb250ZW50c09wYXF1ZSk6CisgICAgICAgIChXZWJDb3JlOjpHcmFw aGljc0xheWVyQ2hyb21pdW06OnNldEZvcmNlQ29udGVudHNPcGFxdWUpOgorICAgICAgICAqIHBs YXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL0dyYXBoaWNzTGF5ZXJDaHJvbWl1bS5oOgorICAgICAg ICAqIHBsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL05vbkNvbXBvc2l0ZWRDb250ZW50SG9zdC5j cHA6CisgICAgICAgIChXZWJDb3JlOjpOb25Db21wb3NpdGVkQ29udGVudEhvc3Q6Ok5vbkNvbXBv c2l0ZWRDb250ZW50SG9zdCk6CisKIDIwMTEtMTEtMzAgIFRpbSBIb3J0b24gIDx0aW1vdGh5X2hv cnRvbkBhcHBsZS5jb20+CiAKICAgICAgICAgZHggY2F1c2VzIG5vbi1CTVAgY2hhcmFjdGVycyB0 byBmYWlsIHRvIHJlbmRlcgpkaWZmIC0tZ2l0IGEvU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vZ3Jh cGhpY3MvY2hyb21pdW0vR3JhcGhpY3NMYXllckNocm9taXVtLmNwcCBiL1NvdXJjZS9XZWJDb3Jl L3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL0dyYXBoaWNzTGF5ZXJDaHJvbWl1bS5jcHAKaW5k ZXggNGJlYWE1MjU1Y2E2Mjg4MzlmYTQ3YmFkNjVlNjE0YjZmYzU1MTc1OC4uZDQ0YmNiYTNjOGUz NWNhZjhjYzhmMmM1ZGIzMGE5MWVjMWQyOTc2MSAxMDA2NDQKLS0tIGEvU291cmNlL1dlYkNvcmUv cGxhdGZvcm0vZ3JhcGhpY3MvY2hyb21pdW0vR3JhcGhpY3NMYXllckNocm9taXVtLmNwcAorKysg Yi9Tb3VyY2UvV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGljcy9jaHJvbWl1bS9HcmFwaGljc0xheWVy Q2hyb21pdW0uY3BwCkBAIC03Myw2ICs3Myw3IEBAIEdyYXBoaWNzTGF5ZXJDaHJvbWl1bTo6R3Jh cGhpY3NMYXllckNocm9taXVtKEdyYXBoaWNzTGF5ZXJDbGllbnQqIGNsaWVudCkKICAgICAsIG1f Y29udGVudHNMYXllclB1cnBvc2UoTm9Db250ZW50c0xheWVyKQogICAgICwgbV9jb250ZW50c0xh eWVySGFzQmFja2dyb3VuZENvbG9yKGZhbHNlKQogICAgICwgbV9pblNldENoaWxkcmVuKGZhbHNl KQorICAgICwgbV9mb3JjZUNvbnRlbnRzT3BhcXVlKGZhbHNlKQogewogICAgIG1fbGF5ZXIgPSBD b250ZW50TGF5ZXJDaHJvbWl1bTo6Y3JlYXRlKHRoaXMpOwogCkBAIC01NDMsNyArNTQ0LDcgQEAg dm9pZCBHcmFwaGljc0xheWVyQ2hyb21pdW06OnVwZGF0ZU1hc2tzVG9Cb3VuZHMoKQogCiB2b2lk IEdyYXBoaWNzTGF5ZXJDaHJvbWl1bTo6dXBkYXRlQ29udGVudHNPcGFxdWUoKQogewotICAgIG1f bGF5ZXItPnNldE9wYXF1ZShtX2NvbnRlbnRzT3BhcXVlKTsKKyAgICBtX2xheWVyLT5zZXRPcGFx dWUobV9jb250ZW50c09wYXF1ZSB8fCBtX2ZvcmNlQ29udGVudHNPcGFxdWUpOwogfQogCiB2b2lk IEdyYXBoaWNzTGF5ZXJDaHJvbWl1bTo6dXBkYXRlQmFja2ZhY2VWaXNpYmlsaXR5KCkKQEAgLTY4 NSw2ICs2ODYsMTUgQEAgZmxvYXQgR3JhcGhpY3NMYXllckNocm9taXVtOjpjb250ZW50c1NjYWxl KCkgY29uc3QKICAgICByZXR1cm4gMTsKIH0KIAordm9pZCBHcmFwaGljc0xheWVyQ2hyb21pdW06 OnNldEZvcmNlQ29udGVudHNPcGFxdWUoYm9vbCBmb3JjZUNvbnRlbnRzT3BhcXVlKQoreworICAg IGlmIChmb3JjZUNvbnRlbnRzT3BhcXVlID09IG1fZm9yY2VDb250ZW50c09wYXF1ZSkKKyAgICAg ICAgcmV0dXJuOworCisgICAgbV9mb3JjZUNvbnRlbnRzT3BhcXVlID0gZm9yY2VDb250ZW50c09w YXF1ZTsKKyAgICB1cGRhdGVDb250ZW50c09wYXF1ZSgpOworfQorCiAvLyBUaGlzIGZ1bmN0aW9u IHNpbXBseSBtaW1pY3MgdGhlIG9wZXJhdGlvbiBvZiBHcmFwaGljc0xheWVyQ0EKIHZvaWQgR3Jh cGhpY3NMYXllckNocm9taXVtOjp1cGRhdGVPcGFjaXR5T25MYXllcigpCiB7CmRpZmYgLS1naXQg YS9Tb3VyY2UvV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGljcy9jaHJvbWl1bS9HcmFwaGljc0xheWVy Q2hyb21pdW0uaCBiL1NvdXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL0dy YXBoaWNzTGF5ZXJDaHJvbWl1bS5oCmluZGV4IGIwNWQwN2JmNjM3YmY1Y2ZlMGU0YmVjNGUzOTNk YmE3YjBjMDNkZjIuLjkyYWViMTc3ZDJjODg4YjU1ZjgyZjllZjMxNTU1ZmExOWMyODJlZTEgMTAw NjQ0Ci0tLSBhL1NvdXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL0dyYXBo aWNzTGF5ZXJDaHJvbWl1bS5oCisrKyBiL1NvdXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNz L2Nocm9taXVtL0dyYXBoaWNzTGF5ZXJDaHJvbWl1bS5oCkBAIC0xMDEsNiArMTAxLDggQEAgcHVi bGljOgogICAgIHZpcnR1YWwgdm9pZCBwYWludENvbnRlbnRzKEdyYXBoaWNzQ29udGV4dCYsIGNv bnN0IEludFJlY3QmIGNsaXApOwogICAgIHZpcnR1YWwgdm9pZCBub3RpZnlTeW5jUmVxdWlyZWQo KTsKIAorICAgIHZvaWQgc2V0Rm9yY2VDb250ZW50c09wYXF1ZShib29sKTsKKwogcHJpdmF0ZToK ICAgICB2b2lkIHVwZGF0ZU9wYWNpdHlPbkxheWVyKCk7CiAKQEAgLTE0Nyw2ICsxNDksNyBAQCBw cml2YXRlOgogICAgIENvbnRlbnRzTGF5ZXJQdXJwb3NlIG1fY29udGVudHNMYXllclB1cnBvc2U7 CiAgICAgYm9vbCBtX2NvbnRlbnRzTGF5ZXJIYXNCYWNrZ3JvdW5kQ29sb3IgOiAxOwogICAgIGJv b2wgbV9pblNldENoaWxkcmVuOworICAgIGJvb2wgbV9mb3JjZUNvbnRlbnRzT3BhcXVlOwogfTsK IAogfSAvLyBuYW1lc3BhY2UgV2ViQ29yZQpkaWZmIC0tZ2l0IGEvU291cmNlL1dlYkNvcmUvcGxh dGZvcm0vZ3JhcGhpY3MvY2hyb21pdW0vTm9uQ29tcG9zaXRlZENvbnRlbnRIb3N0LmNwcCBiL1Nv dXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL05vbkNvbXBvc2l0ZWRDb250 ZW50SG9zdC5jcHAKaW5kZXggYmM2YTJhNDk3ODI4ZWIyOTFlOTg5N2ViMWJiMTIyMjA3MGE0YTJh My4uZjJkNjhiNTNjN2NmYzI2NmQ2MTBiNDI1MWY4OGYwMDQzY2Q1NGI3MCAxMDA2NDQKLS0tIGEv U291cmNlL1dlYkNvcmUvcGxhdGZvcm0vZ3JhcGhpY3MvY2hyb21pdW0vTm9uQ29tcG9zaXRlZENv bnRlbnRIb3N0LmNwcAorKysgYi9Tb3VyY2UvV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGljcy9jaHJv bWl1bS9Ob25Db21wb3NpdGVkQ29udGVudEhvc3QuY3BwCkBAIC0yOSw2ICsyOSw3IEBACiAKICNp bmNsdWRlICJGbG9hdFJlY3QuaCIKICNpbmNsdWRlICJHcmFwaGljc0xheWVyLmgiCisjaW5jbHVk ZSAiR3JhcGhpY3NMYXllckNocm9taXVtLmgiCiAjaW5jbHVkZSAiTGF5ZXJDaHJvbWl1bS5oIgog I2luY2x1ZGUgIkxheWVyUGFpbnRlckNocm9taXVtLmgiCiAKQEAgLTQzLDYgKzQ0LDkgQEAgTm9u Q29tcG9zaXRlZENvbnRlbnRIb3N0OjpOb25Db21wb3NpdGVkQ29udGVudEhvc3QoUGFzc093blB0 cjxMYXllclBhaW50ZXJDaHJvbWkKICNlbmRpZgogICAgIG1fZ3JhcGhpY3NMYXllci0+c2V0RHJh d3NDb250ZW50KHRydWUpOwogICAgIG1fZ3JhcGhpY3NMYXllci0+cGxhdGZvcm1MYXllcigpLT5z ZXRJc05vbkNvbXBvc2l0ZWRDb250ZW50KHRydWUpOworCisgICAgR3JhcGhpY3NMYXllckNocm9t aXVtKiBncmFwaGljc0xheWVyQ2hyb21pdW0gPSBzdGF0aWNfY2FzdDxHcmFwaGljc0xheWVyQ2hy b21pdW0qPihtX2dyYXBoaWNzTGF5ZXIuZ2V0KCkpOworICAgIGdyYXBoaWNzTGF5ZXJDaHJvbWl1 bS0+c2V0Rm9yY2VDb250ZW50c09wYXF1ZSh0cnVlKTsKIH0KIAogTm9uQ29tcG9zaXRlZENvbnRl bnRIb3N0Ojp+Tm9uQ29tcG9zaXRlZENvbnRlbnRIb3N0KCkK 118051 2011-12-06 07:52:48 -0800 2011-12-06 12:17:17 -0800 Patch bug-70564-20111206105247.patch text/plain 2725 danakj U3VidmVyc2lvbiBSZXZpc2lvbjogMTAyMDE5CmRpZmYgLS1naXQgYS9Tb3VyY2UvV2ViQ29yZS9D aGFuZ2VMb2cgYi9Tb3VyY2UvV2ViQ29yZS9DaGFuZ2VMb2cKaW5kZXggOGNhN2M2NjlmMDE5YzAy NWQ5ZTZmNmI3NzkwMTkzYWZlZGFlZTRkZi4uMjMwOGUzNDM1YzEzMzU1OWFiNGEyZThkZGMyNjNj MGM0MjllYzhhMiAxMDA2NDQKLS0tIGEvU291cmNlL1dlYkNvcmUvQ2hhbmdlTG9nCisrKyBiL1Nv dXJjZS9XZWJDb3JlL0NoYW5nZUxvZwpAQCAtMSwzICsxLDE1IEBACisyMDExLTEyLTA2ICBEYW5h IEphbnNlbnMgIDxkYW5ha2pAY2hyb21pdW0ub3JnPgorCisgICAgICAgIFtDaHJvbWl1bV0gTWFr ZSByb290IGxheWVyIGFsd2F5cyBvcGFxdWUKKyAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtpdC5v cmcvc2hvd19idWcuY2dpP2lkPTcwNTY0CisKKyAgICAgICAgUmV2aWV3ZWQgYnkgTk9CT0RZIChP T1BTISkuCisKKyAgICAgICAgKiBwbGF0Zm9ybS9ncmFwaGljcy9jaHJvbWl1bS9Ob25Db21wb3Np dGVkQ29udGVudEhvc3QuY3BwOgorICAgICAgICAoV2ViQ29yZTo6Tm9uQ29tcG9zaXRlZENvbnRl bnRIb3N0OjpOb25Db21wb3NpdGVkQ29udGVudEhvc3QpOgorICAgICAgICAqIHBsYXRmb3JtL2dy YXBoaWNzL2Nocm9taXVtL2NjL0NDVGlsZWRMYXllckltcGwuY3BwOgorICAgICAgICAoV2ViQ29y ZTo6Q0NUaWxlZExheWVySW1wbDo6ZHJhdyk6CisKIDIwMTEtMTItMDUgIFBldGVyIEJldmVybG9v ICA8cGV0ZXJAY2hyb21pdW0ub3JnPgogCiAgICAgICAgIFtDaHJvbWl1bV0gQWRkIEFuZHJvaWQg a2V5Y29kZXMgYW5kIGJ1aWxkIExpbnV4IGNsaXBib2FyZC9maWxlc3lzdGVtIGZpbGVzLgpkaWZm IC0tZ2l0IGEvU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vZ3JhcGhpY3MvY2hyb21pdW0vTm9uQ29t cG9zaXRlZENvbnRlbnRIb3N0LmNwcCBiL1NvdXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNz L2Nocm9taXVtL05vbkNvbXBvc2l0ZWRDb250ZW50SG9zdC5jcHAKaW5kZXggYmM2YTJhNDk3ODI4 ZWIyOTFlOTg5N2ViMWJiMTIyMjA3MGE0YTJhMy4uMTlhMDY1MjA1ZTRlMDEzYzAzZTMzNzMxMWI1 NTdhNTlmZDgwYzBjYyAxMDA2NDQKLS0tIGEvU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vZ3JhcGhp Y3MvY2hyb21pdW0vTm9uQ29tcG9zaXRlZENvbnRlbnRIb3N0LmNwcAorKysgYi9Tb3VyY2UvV2Vi Q29yZS9wbGF0Zm9ybS9ncmFwaGljcy9jaHJvbWl1bS9Ob25Db21wb3NpdGVkQ29udGVudEhvc3Qu Y3BwCkBAIC00Myw2ICs0Myw3IEBAIE5vbkNvbXBvc2l0ZWRDb250ZW50SG9zdDo6Tm9uQ29tcG9z aXRlZENvbnRlbnRIb3N0KFBhc3NPd25QdHI8TGF5ZXJQYWludGVyQ2hyb21pCiAjZW5kaWYKICAg ICBtX2dyYXBoaWNzTGF5ZXItPnNldERyYXdzQ29udGVudCh0cnVlKTsKICAgICBtX2dyYXBoaWNz TGF5ZXItPnBsYXRmb3JtTGF5ZXIoKS0+c2V0SXNOb25Db21wb3NpdGVkQ29udGVudCh0cnVlKTsK KyAgICBtX2dyYXBoaWNzTGF5ZXItPnBsYXRmb3JtTGF5ZXIoKS0+c2V0T3BhcXVlKHRydWUpOwog fQogCiBOb25Db21wb3NpdGVkQ29udGVudEhvc3Q6On5Ob25Db21wb3NpdGVkQ29udGVudEhvc3Qo KQpkaWZmIC0tZ2l0IGEvU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vZ3JhcGhpY3MvY2hyb21pdW0v Y2MvQ0NUaWxlZExheWVySW1wbC5jcHAgYi9Tb3VyY2UvV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGlj cy9jaHJvbWl1bS9jYy9DQ1RpbGVkTGF5ZXJJbXBsLmNwcAppbmRleCA1MTFkODVmN2FlNDE4YTA3 M2VjZThlOTJmZTIwZjRhNTgxZGM4ZjgyLi40OWQxZGYzZmQ4YmUzYzg5MGMwNmYwNWFlNzVlZTk1 MTdlNzk3YzEzIDEwMDY0NAotLS0gYS9Tb3VyY2UvV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGljcy9j aHJvbWl1bS9jYy9DQ1RpbGVkTGF5ZXJJbXBsLmNwcAorKysgYi9Tb3VyY2UvV2ViQ29yZS9wbGF0 Zm9ybS9ncmFwaGljcy9jaHJvbWl1bS9jYy9DQ1RpbGVkTGF5ZXJJbXBsLmNwcApAQCAtMTM4LDEw ICsxMzgsNiBAQCB2b2lkIENDVGlsZWRMYXllckltcGw6OmRyYXcoTGF5ZXJSZW5kZXJlckNocm9t aXVtKiBsYXllclJlbmRlcmVyKQogICAgICAgICBsYXllclF1YWQuaW5mbGF0ZUFudGlBbGlhc2lu Z0Rpc3RhbmNlKCk7CiAgICAgfQogCi0gICAgR3JhcGhpY3NDb250ZXh0M0QqIGNvbnRleHQgPSBs YXllclJlbmRlcmVyLT5jb250ZXh0KCk7Ci0gICAgaWYgKGlzTm9uQ29tcG9zaXRlZENvbnRlbnQo KSkKLSAgICAgICAgR0xDKGNvbnRleHQsIGNvbnRleHQtPmRpc2FibGUoR3JhcGhpY3NDb250ZXh0 M0Q6OkJMRU5EKSk7Ci0KICAgICBzd2l0Y2ggKG1fc2FtcGxlZFRleGVsRm9ybWF0KSB7CiAgICAg Y2FzZSBMYXllclRleHR1cmVVcGRhdGVyOjpTYW1wbGVkVGV4ZWxGb3JtYXRSR0JBOgogICAgICAg ICBpZiAodXNlQUEpIHsKQEAgLTE3NCw5ICsxNzAsNiBAQCB2b2lkIENDVGlsZWRMYXllckltcGw6 OmRyYXcoTGF5ZXJSZW5kZXJlckNocm9taXVtKiBsYXllclJlbmRlcmVyKQogICAgIGRlZmF1bHQ6 CiAgICAgICAgIEFTU0VSVF9OT1RfUkVBQ0hFRCgpOwogICAgIH0KLQotICAgIGlmIChpc05vbkNv bXBvc2l0ZWRDb250ZW50KCkpCi0gICAgICAgIEdMQyhjb250ZXh0LCBjb250ZXh0LT5lbmFibGUo R3JhcGhpY3NDb250ZXh0M0Q6OkJMRU5EKSk7CiB9CiAKIHZvaWQgQ0NUaWxlZExheWVySW1wbDo6 c2V0VGlsaW5nRGF0YShjb25zdCBDQ0xheWVyVGlsaW5nRGF0YSYgdGlsZXIpCg==