Modified: branches/safari-601.1.32-branch/Source/WebCore/ChangeLog (184253 => 184254)
--- branches/safari-601.1.32-branch/Source/WebCore/ChangeLog 2015-05-13 04:17:20 UTC (rev 184253)
+++ branches/safari-601.1.32-branch/Source/WebCore/ChangeLog 2015-05-13 04:25:32 UTC (rev 184254)
@@ -2,369 +2,374 @@
Dana Burkart <[email protected]
[email protected]>
- Merge r184005. rdar://problem/20486538
+ Merge r184104. rdar://problem/20727702
- 2015-05-08 Alexey Proskuryakov <[email protected]>
+ 2015-05-11 Antti Koivisto <[email protected]>
- Crashes in SocketStreamHandleBase::close
- https://bugs.webkit.org/show_bug.cgi?id=144767
- rdar://problem/20486538
+ WebContent crash under com.apple.WebCore: WebCore::WebKitCSSResourceValue::isCSSValueNone const + 6
+ https://bugs.webkit.org/show_bug.cgi?id=144870
+ rdar://problem/20727702
- Reviewed by Brady Eidson.
+ Reviewed by Simon Fraser.
- This is a speculative fix, I could not reproduce the crash.
+ No repro but we are seeing null pointer crashes like this:
- * Modules/websockets/WebSocketChannel.cpp: (WebCore::WebSocketChannel::processFrame):
- Normally, processOutgoingFrameQueue() closes the handle in the end when called in
- OutgoingFrameQueueClosing state. But there is no definitive protection against
- processing two CLOSE frames, in which case we'd try to close the handle twice.
+ Thread 0 Crashed:: Dispatch queue: com.apple.main-thread
+ 0 com.apple.WebCore 0x00007fff92da5706 WebCore::WebKitCSSResourceValue::isCSSValueNone() const + 6
+ 1 com.apple.WebCore 0x00007fff93382b48 WebCore::MaskImageOperation::isCSSValueNone() const + 24
+ 2 com.apple.WebCore 0x00007fff92e0475e WebCore::FillLayer::hasNonEmptyMaskImage() const + 30
- * platform/network/cf/SocketStreamHandleCFNet.cpp:
- (WebCore::SocketStreamHandle::readStreamCallback): Passing empty data to the client
- results in the socket being closed, which makes no sense here.
+ * platform/graphics/MaskImageOperation.cpp:
+ (WebCore::MaskImageOperation::MaskImageOperation):
+ (WebCore::MaskImageOperation::isCSSValueNone):
+ This would crash like this if both m_styleImage and m_cssMaskImageValue are null.
+ There are no obvious guarantees that this doesn't happen. Two of the constructor variants allow it
+ and there is setImage which may turn m_styleImage null later too.
+
+ Fix by making null m_cssMaskImageValue always signify CSSValueNone.
+
+ (WebCore::MaskImageOperation::cssValue):
+
2015-05-12 Dana Burkart
Dana Burkart <[email protected]
[email protected]>
- Merge r184001. rdar://problem/20862460
+ Merge r184005. rdar://problem/20486538
- 2015-05-08 Eric Carlson <[email protected]>
+ 2015-05-08 Alexey Proskuryakov <[email protected]>
- [Mac] Playback target clients do not unregister on page reload
- https://bugs.webkit.org/show_bug.cgi?id=144761
+ Crashes in SocketStreamHandleBase::close
+ https://bugs.webkit.org/show_bug.cgi?id=144767
+ rdar://problem/20486538
Reviewed by Brady Eidson.
- * dom/Document.cpp:
- (WebCore::Document::prepareForDestruction): Unregister all target picker clients.
+ This is a speculative fix, I could not reproduce the crash.
- * html/HTMLMediaElement.cpp:
- (WebCore::HTMLMediaElement::registerWithDocument): Register for page cache callback.
- (WebCore::HTMLMediaElement::unregisterWithDocument): Unregister for page cache callback.
- (WebCore::HTMLMediaElement::documentWillSuspendForPageCache): New.
- (WebCore::HTMLMediaElement::documentDidResumeFromPageCache): New.
+ * Modules/websockets/WebSocketChannel.cpp: (WebCore::WebSocketChannel::processFrame):
+ Normally, processOutgoingFrameQueue() closes the handle in the end when called in
+ OutgoingFrameQueueClosing state. But there is no definitive protection against
+ processing two CLOSE frames, in which case we'd try to close the handle twice.
+ * platform/network/cf/SocketStreamHandleCFNet.cpp:
+ (WebCore::SocketStreamHandle::readStreamCallback): Passing empty data to the client
+ results in the socket being closed, which makes no sense here.
+
2015-05-12 Dana Burkart
Dana Burkart <[email protected]
[email protected]>
- Merge r183971. rdar://problem/20065572
+ Merge r184001. rdar://problem/20862460
- 2015-05-07 Dean Jackson <[email protected]>
+ 2015-05-08 Eric Carlson <[email protected]>
- [iOS] MediaControls: disappear while scrubbing
- https://bugs.webkit.org/show_bug.cgi?id=144777
- <rdar://problem/20065572>
+ [Mac] Playback target clients do not unregister on page reload
+ https://bugs.webkit.org/show_bug.cgi?id=144761
- Reviewed by Eric Carlson.
+ Reviewed by Brady Eidson.
- If we are scrubbing we shouldn't hide the controls.
+ * dom/Document.cpp:
+ (WebCore::Document::prepareForDestruction): Unregister all target picker clients.
- * Modules/mediacontrols/mediaControlsApple.js:
- (Controller.prototype.hideControls): Return early if we are scrubbing.
- * Modules/mediacontrols/mediaControlsiOS.js:
- (ControllerIOS.prototype): Add initial value for _potentiallyScrubbing and
- rename from non-underscored value throughout the file.
- (ControllerIOS.prototype.handleTimelineTouchEnd): When we finish scrubbing, reset
- the timer to hide the controls.
+ * html/HTMLMediaElement.cpp:
+ (WebCore::HTMLMediaElement::registerWithDocument): Register for page cache callback.
+ (WebCore::HTMLMediaElement::unregisterWithDocument): Unregister for page cache callback.
+ (WebCore::HTMLMediaElement::documentWillSuspendForPageCache): New.
+ (WebCore::HTMLMediaElement::documentDidResumeFromPageCache): New.
2015-05-12 Dana Burkart
Dana Burkart <[email protected]
[email protected]>
- Merge r183970. rdar://problem/20769741
+ Merge r183971. rdar://problem/20065572
- 2015-05-07 Simon Fraser <[email protected]>
+ 2015-05-07 Dean Jackson <[email protected]>
- REGRESSION (r183300): Fixed elements flash when scrolling
- https://bugs.webkit.org/show_bug.cgi?id=144778
- rdar://problem/20769741
+ [iOS] MediaControls: disappear while scrubbing
+ https://bugs.webkit.org/show_bug.cgi?id=144777
+ <rdar://problem/20065572>
- Reviewed by Dean Jackson.
+ Reviewed by Eric Carlson.
- After r183300 we can detached layer backing store when outside the coverage region.
- However, position:fixed layers are moved around by the ScrollingCoordinator behind
- GraphicsLayer's back, so we can do layer flushes with stale information about layer
- geometry.
+ If we are scrubbing we shouldn't hide the controls.
- To avoid dropping backing store for layers in this situation, prevent backing
- store detachment on layers registered with the ScrollingCoordinator as viewport-constrained
- layers. Preventing detachment on a layer also prevents detachment on all descendant
- layers.
+ * Modules/mediacontrols/mediaControlsApple.js:
+ (Controller.prototype.hideControls): Return early if we are scrubbing.
+ * Modules/mediacontrols/mediaControlsiOS.js:
+ (ControllerIOS.prototype): Add initial value for _potentiallyScrubbing and
+ rename from non-underscored value throughout the file.
+ (ControllerIOS.prototype.handleTimelineTouchEnd): When we finish scrubbing, reset
+ the timer to hide the controls.
- * platform/graphics/GraphicsLayer.h:
- (WebCore::GraphicsLayer::setAllowsBackingStoreDetachment):
- (WebCore::GraphicsLayer::allowsBackingStoreDetachment):
- * platform/graphics/ca/GraphicsLayerCA.cpp:
- (WebCore::GraphicsLayerCA::GraphicsLayerCA):
- (WebCore::GraphicsLayerCA::setVisibleAndCoverageRects): Set m_intersectsCoverageRect to true
- if backing store detachment is prevented.
- (WebCore::GraphicsLayerCA::recursiveCommitChanges): Set a bit in the CommitState to
- communicate to descendants that detachment is prevented.
- * platform/graphics/ca/GraphicsLayerCA.h:
- (WebCore::GraphicsLayerCA::CommitState::CommitState): Deleted.
- * rendering/RenderLayerBacking.cpp:
- (WebCore::RenderLayerBacking::setIsScrollCoordinatedWithViewportConstrainedRole):
- * rendering/RenderLayerBacking.h:
- (WebCore::RenderLayerBacking::setScrollingNodeIDForRole): If registering with a non-zero
- nodeID for the ViewportConstrained role, turn off backing store detachment.
-
2015-05-12 Dana Burkart
Dana Burkart <[email protected]
[email protected]>
- Merge r183965. rdar://problem/20866590
+ Merge r183970. rdar://problem/20769741
- 2015-05-07 Dean Jackson <[email protected]>
+ 2015-05-07 Simon Fraser <[email protected]>
- [iOS] While scrubbing and holding down, video continues to play
- https://bugs.webkit.org/show_bug.cgi?id=144776
- <rdar://problem/20863757>
+ REGRESSION (r183300): Fixed elements flash when scrolling
+ https://bugs.webkit.org/show_bug.cgi?id=144778
+ rdar://problem/20769741
- Reviewed by Simon Fraser.
+ Reviewed by Dean Jackson.
- When we are scrubbing a video, we should pause playback. As we
- let go of the scrubber playback can resume (but only if it was
- playing originally).
+ After r183300 we can detached layer backing store when outside the coverage region.
+ However, position:fixed layers are moved around by the ScrollingCoordinator behind
+ GraphicsLayer's back, so we can do layer flushes with stale information about layer
+ geometry.
- * Modules/mediacontrols/mediaControlsiOS.js:
- (ControllerIOS.prototype.createControls): Listen for touchstart on the scrubber.
- (ControllerIOS.prototype.handleTimelineInput): Call the prototype, but pause if necessary.
- (ControllerIOS.prototype.handleTimelineChange): Just moved this to be with the other timeline functions.
- (ControllerIOS.prototype.handleTimelineTouchStart): Add the listeners for end and cancel. Remember that we are
- potentially about to scrub.
- (ControllerIOS.prototype.handleTimelineTouchEnd): Remove the listeners.
+ To avoid dropping backing store for layers in this situation, prevent backing
+ store detachment on layers registered with the ScrollingCoordinator as viewport-constrained
+ layers. Preventing detachment on a layer also prevents detachment on all descendant
+ layers.
+ * platform/graphics/GraphicsLayer.h:
+ (WebCore::GraphicsLayer::setAllowsBackingStoreDetachment):
+ (WebCore::GraphicsLayer::allowsBackingStoreDetachment):
+ * platform/graphics/ca/GraphicsLayerCA.cpp:
+ (WebCore::GraphicsLayerCA::GraphicsLayerCA):
+ (WebCore::GraphicsLayerCA::setVisibleAndCoverageRects): Set m_intersectsCoverageRect to true
+ if backing store detachment is prevented.
+ (WebCore::GraphicsLayerCA::recursiveCommitChanges): Set a bit in the CommitState to
+ communicate to descendants that detachment is prevented.
+ * platform/graphics/ca/GraphicsLayerCA.h:
+ (WebCore::GraphicsLayerCA::CommitState::CommitState): Deleted.
+ * rendering/RenderLayerBacking.cpp:
+ (WebCore::RenderLayerBacking::setIsScrollCoordinatedWithViewportConstrainedRole):
+ * rendering/RenderLayerBacking.h:
+ (WebCore::RenderLayerBacking::setScrollingNodeIDForRole): If registering with a non-zero
+ nodeID for the ViewportConstrained role, turn off backing store detachment.
+
2015-05-12 Dana Burkart
Dana Burkart <[email protected]
[email protected]>
- Merge r183954. rdar://problem/20281886
+ Merge r183965. rdar://problem/20866590
- 2015-05-07 Beth Dakin <[email protected]>
+ 2015-05-07 Dean Jackson <[email protected]>
- New force-related DOM events should fire in WK1 views
- https://bugs.webkit.org/show_bug.cgi?id=144663
- -and corresponding-
- rdar://problem/20281886
+ [iOS] While scrubbing and holding down, video continues to play
+ https://bugs.webkit.org/show_bug.cgi?id=144776
+ <rdar://problem/20863757>
- Reviewed by Sam Weinig.
+ Reviewed by Simon Fraser.
- All of the WK1 mouse events need to take the correspondingPressureEvent.
- * page/EventHandler.h:
+ When we are scrubbing a video, we should pause playback. As we
+ let go of the scrubber playback can resume (but only if it was
+ playing originally).
- Make correspondingPressureEvent a part of CurrentEventScope. This is needed to
- have accurate pressure information for all of the mouse events in subframes.
- * page/mac/EventHandlerMac.mm:
- (WebCore::correspondingPressureEventSlot):
- (WebCore::EventHandler::correspondingPressureEvent):
- (WebCore::CurrentEventScope::CurrentEventScope):
- (WebCore::CurrentEventScope::~CurrentEventScope):
+ * Modules/mediacontrols/mediaControlsiOS.js:
+ (ControllerIOS.prototype.createControls): Listen for touchstart on the scrubber.
+ (ControllerIOS.prototype.handleTimelineInput): Call the prototype, but pause if necessary.
+ (ControllerIOS.prototype.handleTimelineChange): Just moved this to be with the other timeline functions.
+ (ControllerIOS.prototype.handleTimelineTouchStart): Add the listeners for end and cancel. Remember that we are
+ potentially about to scrub.
+ (ControllerIOS.prototype.handleTimelineTouchEnd): Remove the listeners.
- These events don’t have an associated pressure, so send nil for the
- correspondingPressureEvent.
- (WebCore::EventHandler::wheelEvent):
- (WebCore::EventHandler::keyEvent):
+ 2015-05-12 Dana Burkart
+ Dana Burkart <[email protected]
+ [email protected]>
- Pipe through correspondingPressureEvent.
- (WebCore::EventHandler::mouseDown):
- (WebCore::EventHandler::mouseDragged):
- (WebCore::EventHandler::mouseUp):
- (WebCore::EventHandler::mouseMoved):
+ Merge r183954. rdar://problem/20281886
- New function to handle pressure change events.
- (WebCore::EventHandler::pressureChange):
+ 2015-05-07 Beth Dakin <[email protected]>
- Pipe through correspondingPressureEvent.
- (WebCore::EventHandler::passMouseMovedEventToScrollbars):
- (WebCore::EventHandler::currentPlatformMouseEvent):
+ New force-related DOM events should fire in WK1 views
+ https://bugs.webkit.org/show_bug.cgi?id=144663
+ -and corresponding-
+ rdar://problem/20281886
- Take the correspondingPressureEvent in order to build a PlatformMouseEvent with
- the correct pressure information.
- * platform/mac/PlatformEventFactoryMac.h:
- * platform/mac/PlatformEventFactoryMac.mm:
- (WebCore::globalPointForEvent):
- (WebCore::pointForEvent):
- (WebCore::mouseButtonForEvent):
- (WebCore::PlatformMouseEventBuilder::PlatformMouseEventBuilder):
- (WebCore::PlatformEventFactory::createPlatformMouseEvent):
+ Reviewed by Sam Weinig.
- 2015-05-12 Dana Burkart
- Dana Burkart <[email protected]
- [email protected]>
+ All of the WK1 mouse events need to take the correspondingPressureEvent.
+ * page/EventHandler.h:
- Merge r183953. rdar://problem/19997548
+ Make correspondingPressureEvent a part of CurrentEventScope. This is needed to
+ have accurate pressure information for all of the mouse events in subframes.
+ * page/mac/EventHandlerMac.mm:
+ (WebCore::correspondingPressureEventSlot):
+ (WebCore::EventHandler::correspondingPressureEvent):
+ (WebCore::CurrentEventScope::CurrentEventScope):
+ (WebCore::CurrentEventScope::~CurrentEventScope):
- 2015-05-06 Roger Fong <[email protected]>
+ These events don’t have an associated pressure, so send nil for the
+ correspondingPressureEvent.
+ (WebCore::EventHandler::wheelEvent):
+ (WebCore::EventHandler::keyEvent):
- Media Controls: Scrubber should be independent of actual video time, causes scrubber to be jumpy.
- https://bugs.webkit.org/show_bug.cgi?id=144700.
- <rdar://problem/19997548>
+ Pipe through correspondingPressureEvent.
+ (WebCore::EventHandler::mouseDown):
+ (WebCore::EventHandler::mouseDragged):
+ (WebCore::EventHandler::mouseUp):
+ (WebCore::EventHandler::mouseMoved):
- Reviewed by Jer Noble.
+ New function to handle pressure change events.
+ (WebCore::EventHandler::pressureChange):
- Update time and timeline during the timeline input event instead of the wrapper's mousemove.
- (Controller.prototype.handleWrapperMouseMove):
- (Controller.prototype.handleTimelineMouseMove):
- (Controller.prototype.drawTimelineBackground):
+ Pipe through correspondingPressureEvent.
+ (WebCore::EventHandler::passMouseMovedEventToScrollbars):
+ (WebCore::EventHandler::currentPlatformMouseEvent):
- (Controller.prototype.updateControlsWhileScrubbing):
- Updates time and scrubber to reflect timeline user input.
+ Take the correspondingPressureEvent in order to build a PlatformMouseEvent with
+ the correct pressure information.
+ * platform/mac/PlatformEventFactoryMac.h:
+ * platform/mac/PlatformEventFactoryMac.mm:
+ (WebCore::globalPointForEvent):
+ (WebCore::pointForEvent):
+ (WebCore::mouseButtonForEvent):
+ (WebCore::PlatformMouseEventBuilder::PlatformMouseEventBuilder):
+ (WebCore::PlatformEventFactory::createPlatformMouseEvent):
2015-05-12 Dana Burkart
Dana Burkart <[email protected]
[email protected]>
- Merge r183943. rdar://problem/19913748
+ Merge r183953. rdar://problem/19997548
- 2015-05-07 Simon Fraser <[email protected]>
+ 2015-05-06 Roger Fong <[email protected]>
- Remove the WK1-only code path for independently composited iframes
- https://bugs.webkit.org/show_bug.cgi?id=144722
+ Media Controls: Scrubber should be independent of actual video time, causes scrubber to be jumpy.
+ https://bugs.webkit.org/show_bug.cgi?id=144700.
+ <rdar://problem/19997548>
- Reviewed by Dean Jackson.
+ Reviewed by Jer Noble.
- In WebKit1 on Mac, we allowed iframes to be composited independently of their
- parent document, relying on the fact that the frame's platform view can host
- a layer-backed view. However, this ran into bugs (rdar://problem/18862298),
- and triggers the assertion at the end of FrameView::updateLayoutAndStyleIfNeededRecursive(),
- because the compositing update after a layout can dirty style in notifyIFramesOfCompositingChange().
+ Update time and timeline during the timeline input event instead of the wrapper's mousemove.
+ (Controller.prototype.handleWrapperMouseMove):
+ (Controller.prototype.handleTimelineMouseMove):
+ (Controller.prototype.drawTimelineBackground):
- Removing the WK1-only code path solves these problems. It also eliminates the need
- to do compositing-specific frame overlap testing.
+ (Controller.prototype.updateControlsWhileScrubbing):
+ Updates time and scrubber to reflect timeline user input.
- * page/FrameView.cpp:
- (WebCore::FrameView::setIsOverlapped): No need to do compositing-related things here.
- Any iframe that gets composited will participate in the normal compositing overlap
- testing in its parent frame.
- (WebCore::FrameView::hasCompositedContentIncludingDescendants): Deleted.
- (WebCore::FrameView::hasCompositingAncestor): Deleted.
- * page/FrameView.h:
- * rendering/RenderLayerCompositor.cpp: Replace ownerElement() checks in this file
- with an isMainFrameCompositor() for readability. Some 0->nullptr.
- (WebCore::RenderLayerCompositor::cacheAcceleratedCompositingFlags):
- (WebCore::RenderLayerCompositor::chromeClient):
- (WebCore::RenderLayerCompositor::enclosingCompositorFlushingLayers):
- (WebCore::RenderLayerCompositor::updateCompositingLayers):
- (WebCore::RenderLayerCompositor::appendDocumentOverlayLayers):
- (WebCore::RenderLayerCompositor::updateBacking):
- (WebCore::RenderLayerCompositor::layerTreeAsText):
- (WebCore::RenderLayerCompositor::frameContentsCompositor):
- (WebCore::RenderLayerCompositor::setIsInWindow):
- (WebCore::RenderLayerCompositor::requiresCompositingForScrollableFrame):
- (WebCore::RenderLayerCompositor::requiresCompositingForFrame): frameRenderer.requiresAcceleratedCompositing()
- already bails on no content RenderView, so the shouldPropagateCompositingToEnclosingFrame() check does
- nothing and is removed.
- (WebCore::RenderLayerCompositor::isAsyncScrollableStickyLayer):
- (WebCore::RenderLayerCompositor::requiresScrollLayer):
- (WebCore::RenderLayerCompositor::documentUsesTiledBacking):
- (WebCore::RenderLayerCompositor::isMainFrameCompositor):
- (WebCore::RenderLayerCompositor::shouldCompositeOverflowControls):
- (WebCore::RenderLayerCompositor::requiresOverhangAreasLayer):
- (WebCore::RenderLayerCompositor::requiresContentShadowLayer):
- (WebCore::RenderLayerCompositor::updateLayerForTopOverhangArea):
- (WebCore::RenderLayerCompositor::updateLayerForBottomOverhangArea):
- (WebCore::RenderLayerCompositor::updateLayerForHeader):
- (WebCore::RenderLayerCompositor::updateLayerForFooter):
- (WebCore::RenderLayerCompositor::ensureRootLayer): Main frame attaches via ChromeClient,
- all other frames attach via parent frame.
- (WebCore::RenderLayerCompositor::notifyIFramesOfCompositingChange): This call to
- scheduleSetNeedsStyleRecalc(SyntheticStyleChange) was the source of dirtying style after
- layout, but is no longer needed so remove it.
- (WebCore::RenderLayerCompositor::registerAllViewportConstrainedLayers):
- (WebCore::RenderLayerCompositor::unregisterAllViewportConstrainedLayers):
- (WebCore::RenderLayerCompositor::scrollingCoordinator):
- (WebCore::RenderLayerCompositor::graphicsLayerFactory):
- (WebCore::RenderLayerCompositor::allowsIndependentlyCompositedFrames): Deleted.
- (WebCore::RenderLayerCompositor::shouldPropagateCompositingToEnclosingFrame): Deleted.
- (WebCore::RenderLayerCompositor::mainFrameBackingIsTiled): Deleted. This was mis-named;
- it really asks whether the document uses tiled backing, but does not check for main frame.
- * rendering/RenderLayerCompositor.h:
- * rendering/RenderWidget.cpp:
- (WebCore::RenderWidget::paintContents): No need to do frame overlap testing for
- compositing now.
-
2015-05-12 Dana Burkart
Dana Burkart <[email protected]
[email protected]>
- Merge r183927. rdar://problem/20854785
+ Merge r183943. rdar://problem/19913748
- 2015-05-07 Eric Carlson <[email protected]>
+ 2015-05-07 Simon Fraser <[email protected]>
- [Mac] Playback target isn't set on new element
- https://bugs.webkit.org/show_bug.cgi?id=144724
+ Remove the WK1-only code path for independently composited iframes
+ https://bugs.webkit.org/show_bug.cgi?id=144722
- Reviewed by Jer Noble.
+ Reviewed by Dean Jackson.
- * platform/graphics/avfoundation/objc/MediaPlayerPrivateAVFoundationObjC.h:
- * platform/graphics/avfoundation/objc/MediaPlayerPrivateAVFoundationObjC.mm:
- (WebCore::MediaPlayerPrivateAVFoundationObjC::createAVPlayer): Call setShouldPlayToPlaybackTarget
- if necessary.
- (WebCore::MediaPlayerPrivateAVFoundationObjC::setShouldPlayToPlaybackTarget): Remember
- the setting in case we don't have an AVPlayer yet.
+ In WebKit1 on Mac, we allowed iframes to be composited independently of their
+ parent document, relying on the fact that the frame's platform view can host
+ a layer-backed view. However, this ran into bugs (rdar://problem/18862298),
+ and triggers the assertion at the end of FrameView::updateLayoutAndStyleIfNeededRecursive(),
+ because the compositing update after a layout can dirty style in notifyIFramesOfCompositingChange().
+ Removing the WK1-only code path solves these problems. It also eliminates the need
+ to do compositing-specific frame overlap testing.
+
+ * page/FrameView.cpp:
+ (WebCore::FrameView::setIsOverlapped): No need to do compositing-related things here.
+ Any iframe that gets composited will participate in the normal compositing overlap
+ testing in its parent frame.
+ (WebCore::FrameView::hasCompositedContentIncludingDescendants): Deleted.
+ (WebCore::FrameView::hasCompositingAncestor): Deleted.
+ * page/FrameView.h:
+ * rendering/RenderLayerCompositor.cpp: Replace ownerElement() checks in this file
+ with an isMainFrameCompositor() for readability. Some 0->nullptr.
+ (WebCore::RenderLayerCompositor::cacheAcceleratedCompositingFlags):
+ (WebCore::RenderLayerCompositor::chromeClient):
+ (WebCore::RenderLayerCompositor::enclosingCompositorFlushingLayers):
+ (WebCore::RenderLayerCompositor::updateCompositingLayers):
+ (WebCore::RenderLayerCompositor::appendDocumentOverlayLayers):
+ (WebCore::RenderLayerCompositor::updateBacking):
+ (WebCore::RenderLayerCompositor::layerTreeAsText):
+ (WebCore::RenderLayerCompositor::frameContentsCompositor):
+ (WebCore::RenderLayerCompositor::setIsInWindow):
+ (WebCore::RenderLayerCompositor::requiresCompositingForScrollableFrame):
+ (WebCore::RenderLayerCompositor::requiresCompositingForFrame): frameRenderer.requiresAcceleratedCompositing()
+ already bails on no content RenderView, so the shouldPropagateCompositingToEnclosingFrame() check does
+ nothing and is removed.
+ (WebCore::RenderLayerCompositor::isAsyncScrollableStickyLayer):
+ (WebCore::RenderLayerCompositor::requiresScrollLayer):
+ (WebCore::RenderLayerCompositor::documentUsesTiledBacking):
+ (WebCore::RenderLayerCompositor::isMainFrameCompositor):
+ (WebCore::RenderLayerCompositor::shouldCompositeOverflowControls):
+ (WebCore::RenderLayerCompositor::requiresOverhangAreasLayer):
+ (WebCore::RenderLayerCompositor::requiresContentShadowLayer):
+ (WebCore::RenderLayerCompositor::updateLayerForTopOverhangArea):
+ (WebCore::RenderLayerCompositor::updateLayerForBottomOverhangArea):
+ (WebCore::RenderLayerCompositor::updateLayerForHeader):
+ (WebCore::RenderLayerCompositor::updateLayerForFooter):
+ (WebCore::RenderLayerCompositor::ensureRootLayer): Main frame attaches via ChromeClient,
+ all other frames attach via parent frame.
+ (WebCore::RenderLayerCompositor::notifyIFramesOfCompositingChange): This call to
+ scheduleSetNeedsStyleRecalc(SyntheticStyleChange) was the source of dirtying style after
+ layout, but is no longer needed so remove it.
+ (WebCore::RenderLayerCompositor::registerAllViewportConstrainedLayers):
+ (WebCore::RenderLayerCompositor::unregisterAllViewportConstrainedLayers):
+ (WebCore::RenderLayerCompositor::scrollingCoordinator):
+ (WebCore::RenderLayerCompositor::graphicsLayerFactory):
+ (WebCore::RenderLayerCompositor::allowsIndependentlyCompositedFrames): Deleted.
+ (WebCore::RenderLayerCompositor::shouldPropagateCompositingToEnclosingFrame): Deleted.
+ (WebCore::RenderLayerCompositor::mainFrameBackingIsTiled): Deleted. This was mis-named;
+ it really asks whether the document uses tiled backing, but does not check for main frame.
+ * rendering/RenderLayerCompositor.h:
+ * rendering/RenderWidget.cpp:
+ (WebCore::RenderWidget::paintContents): No need to do frame overlap testing for
+ compositing now.
+
2015-05-12 Dana Burkart
Dana Burkart <[email protected]
[email protected]>
- Merge r183909. rdar://problem/18894598
+ Merge r183927. rdar://problem/20854785
- 2015-05-06 Daniel Bates <[email protected]>
+ 2015-05-07 Eric Carlson <[email protected]>
- [iOS][WK2] Pause/resume database thread when UIProcess enters/leaves the background
- https://bugs.webkit.org/show_bug.cgi?id=144657
- <rdar://problem/18894598>
+ [Mac] Playback target isn't set on new element
+ https://bugs.webkit.org/show_bug.cgi?id=144724
- Reviewed by Andy Estes.
+ Reviewed by Jer Noble.
- Export WebCore functionality to pause and resume the database thread so that we can
- make use of this functionality from WebKit2.
+ * platform/graphics/avfoundation/objc/MediaPlayerPrivateAVFoundationObjC.h:
+ * platform/graphics/avfoundation/objc/MediaPlayerPrivateAVFoundationObjC.mm:
+ (WebCore::MediaPlayerPrivateAVFoundationObjC::createAVPlayer): Call setShouldPlayToPlaybackTarget
+ if necessary.
+ (WebCore::MediaPlayerPrivateAVFoundationObjC::setShouldPlayToPlaybackTarget): Remember
+ the setting in case we don't have an AVPlayer yet.
- * Modules/webdatabase/AbstractDatabaseServer.h:
- * Modules/webdatabase/DatabaseManager.cpp:
- (WebCore::DatabaseManager::setPauseAllDatabases): Added; turns around and calls DatabaseServer::setPauseAllDatabases().
- * Modules/webdatabase/DatabaseManager.h:
- * Modules/webdatabase/DatabaseServer.cpp:
- (WebCore::DatabaseServer::setPauseAllDatabases): Added; turns around and calls
- DatabaseTracker::tracker().setDatabasesPaused() to pause or resume the database thread.
- For now, we guard this call with PLATFORM(IOS). We'll look to remove this guard once
- we fix <https://bugs.webkit.org/show_bug.cgi?id=144660>.
- * Modules/webdatabase/DatabaseServer.h:
-
2015-05-12 Dana Burkart
- Dana Burkart <[email protected]>
+ Dana Burkart <[email protected]
+ [email protected]>
- Merge r183942. rdar://problem/20049088
+ Merge r183909. rdar://problem/18894598
- 2015-05-06 Dean Jackson <[email protected]>
+ 2015-05-06 Daniel Bates <[email protected]>
- Handle backdrop views that have to tile
- https://bugs.webkit.org/show_bug.cgi?id=142317
- <rdar://problem/20049088>
+ [iOS][WK2] Pause/resume database thread when UIProcess enters/leaves the background
+ https://bugs.webkit.org/show_bug.cgi?id=144657
+ <rdar://problem/18894598>
- Reviewed by Simon Fraser.
+ Reviewed by Andy Estes.
- Take 2 - this was rolled out because Mavericks was crashing.
+ Export WebCore functionality to pause and resume the database thread so that we can
+ make use of this functionality from WebKit2.
- Make sure backdrop layers don't tile. If they are big
- enough, we'll leave it to the platform compositor to handle.
+ * Modules/webdatabase/AbstractDatabaseServer.h:
+ * Modules/webdatabase/DatabaseManager.cpp:
+ (WebCore::DatabaseManager::setPauseAllDatabases): Added; turns around and calls DatabaseServer::setPauseAllDatabases().
+ * Modules/webdatabase/DatabaseManager.h:
+ * Modules/webdatabase/DatabaseServer.cpp:
+ (WebCore::DatabaseServer::setPauseAllDatabases): Added; turns around and calls
+ DatabaseTracker::tracker().setDatabasesPaused() to pause or resume the database thread.
+ For now, we guard this call with PLATFORM(IOS). We'll look to remove this guard once
+ we fix <https://bugs.webkit.org/show_bug.cgi?id=144660>.
+ * Modules/webdatabase/DatabaseServer.h:
- This also fixes a bug where if a layer changed from a backdrop
- type to a tiled type, it would still retain its custom appearance
- and we'd try to add children to the wrong layer.
-
- Test: compositing/media-controls-bar-appearance-big.html
-
- * platform/graphics/ca/GraphicsLayerCA.cpp:
- (WebCore::GraphicsLayerCA::commitLayerChangesBeforeSublayers): Check if
- a layer needs a backdrop before checking if it needs to tile.
-
2015-05-12 Dana Burkart
Dana Burkart <[email protected]>
- Merge r183894. rdar://problem/20049088
+ Merge r183942. rdar://problem/20049088
2015-05-06 Dean Jackson <[email protected]>
@@ -374,6 +379,8 @@
Reviewed by Simon Fraser.
+ Take 2 - this was rolled out because Mavericks was crashing.
+
Make sure backdrop layers don't tile. If they are big
enough, we'll leave it to the platform compositor to handle.
@@ -387,6 +394,32 @@
(WebCore::GraphicsLayerCA::commitLayerChangesBeforeSublayers): Check if
a layer needs a backdrop before checking if it needs to tile.
+ 2015-05-12 Dana Burkart
+ Dana Burkart <[email protected]>
+
+ Merge r183894. rdar://problem/20049088
+
+ 2015-05-06 Dean Jackson <[email protected]>
+
+ Handle backdrop views that have to tile
+ https://bugs.webkit.org/show_bug.cgi?id=142317
+ <rdar://problem/20049088>
+
+ Reviewed by Simon Fraser.
+
+ Make sure backdrop layers don't tile. If they are big
+ enough, we'll leave it to the platform compositor to handle.
+
+ This also fixes a bug where if a layer changed from a backdrop
+ type to a tiled type, it would still retain its custom appearance
+ and we'd try to add children to the wrong layer.
+
+ Test: compositing/media-controls-bar-appearance-big.html
+
+ * platform/graphics/ca/GraphicsLayerCA.cpp:
+ (WebCore::GraphicsLayerCA::commitLayerChangesBeforeSublayers): Check if
+ a layer needs a backdrop before checking if it needs to tile.
+
2015-05-06 Brent Fulgham <[email protected]>
Scroll-snap points do not handle margins and padding propertly
Modified: branches/safari-601.1.32-branch/Source/WebCore/platform/graphics/MaskImageOperation.cpp (184253 => 184254)
--- branches/safari-601.1.32-branch/Source/WebCore/platform/graphics/MaskImageOperation.cpp 2015-05-13 04:17:20 UTC (rev 184253)
+++ branches/safari-601.1.32-branch/Source/WebCore/platform/graphics/MaskImageOperation.cpp 2015-05-13 04:25:32 UTC (rev 184254)
@@ -27,6 +27,7 @@
#include "config.h"
#include "MaskImageOperation.h"
+#include "CSSValuePool.h"
#include "CachedImage.h"
#include "CachedSVGDocument.h"
#include "RenderBoxModelObject.h"
@@ -77,7 +78,6 @@
: m_isExternalDocument(false)
, m_renderLayerImageClient(nullptr)
{
- m_cssMaskImageValue = WebKitCSSResourceValue::create(CSSPrimitiveValue::createIdentifier(CSSValueNone));
}
MaskImageOperation::~MaskImageOperation()
@@ -98,8 +98,7 @@
if (image())
return false;
- ASSERT(m_cssMaskImageValue.get());
- return m_cssMaskImageValue->isCSSValueNone();
+ return !m_cssMaskImageValue || m_cssMaskImageValue->isCSSValueNone();
}
PassRefPtr<CSSValue> MaskImageOperation::cssValue()
@@ -108,7 +107,7 @@
return image()->cssValue();
if (isCSSValueNone())
- return m_cssMaskImageValue->innerValue();
+ return cssValuePool().createIdentifierValue(CSSValueNone);
ASSERT(m_cssMaskImageValue.get());
return m_cssMaskImageValue.get();
