LayerTreeRenderer should not depend on CoordinatedLayerTreeHostProxy so that it can be moved to WebCore.
Created attachment 185192 [details] Patch
Comment on attachment 185192 [details] Patch Attachment 185192 [details] did not pass mac-wk2-ews (mac-wk2): Output: http://queues.webkit.org/results/16201189 New failing tests: fast/workers/worker-document-leak.html
LGTM.
Created attachment 185322 [details] Patch
(In reply to comment #4) > Created an attachment (id=185322) [details] > Patch Rebased.
Comment on attachment 185322 [details] Patch LGTM, though a WK2 owner has to sign off.
Comment on attachment 185322 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=185322&action=review Okay from me given Noam already reviewed. > Source/WebKit2/UIProcess/CoordinatedGraphics/CoordinatedLayerTreeHostProxy.h:98 > + // LayerTreeRendererClient Methods Period. > Source/WebKit2/UIProcess/CoordinatedGraphics/LayerTreeRenderer.h:98 > - // The painting thread must lock the main thread to use below two methods, because two methods access members that the main thread manages. See m_coordinatedLayerTreeHostProxy. > + // The painting thread must lock the main thread to use below two methods, because two methods access members that the main thread manages. See m_client. > // Currently, QQuickWebPage::updatePaintNode() locks the main thread before calling both methods. > void purgeGLResources(); > void setActive(bool); This is backward. You should not document weird thread consistency issues, you should encapsulate them.
Created attachment 185668 [details] Patch
Comment on attachment 185668 [details] Patch Clearing flags on attachment: 185668 Committed r141376: <http://trac.webkit.org/changeset/141376>
All reviewed patches have been landed. Closing bug.