According to the documentation (http://msdn.microsoft.com/en-us/library/windows/desktop/dd183362(v=vs.85).aspx), the Win32 api functions BeginPaint and EndPaint should only be called in response to a WM_PAINT message. Currently, the WebView window procedure might call these functions (via WebView::paint(0, 0)) for all types of messages. Also, there is potential double drawing. If some message is dispatched before WM_PAINT, WebView::paint(0, 0) will be called. This call does not validate the area drawn, so later the WM_PAINT will arrive, drawing the same area.
Created attachment 213770 [details] Patch
Comment on attachment 213770 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=213770&action=review > Source/WebKit/win/WebView.cpp:2467 > + ::UpdateWindow(hWnd); Will this trigger the logic in webView->performLayeredWIndowUpdate()?
(In reply to comment #2) > (From update of attachment 213770 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=213770&action=review > > > Source/WebKit/win/WebView.cpp:2467 > > + ::UpdateWindow(hWnd); > > Will this trigger the logic in webView->performLayeredWIndowUpdate()? Yes, UpdateWindow will send the WM_PAINT message (synchronously) if there is an invalid area.
Comment on attachment 213770 [details] Patch r=me
<rdar://problem/15187300>
(In reply to comment #4) > (From update of attachment 213770 [details]) > r=me Thanks for reviewing!
Comment on attachment 213770 [details] Patch Clearing flags on attachment: 213770 Committed r157174: <http://trac.webkit.org/changeset/157174>
All reviewed patches have been landed. Closing bug.