Temporary rollout support for region styling (background-color) since it causes performance regression (see https://bugs.webkit.org/show_bug.cgi?id=74141). I will add the support back after: 1. Improve speed of methods that call RenderObject::style() extensively 2. Finish analyze the regression in performance of the current patch.
Created attachment 120179 [details] Patch
Comment on attachment 120179 [details] Patch Thanks!
Comment on attachment 120179 [details] Patch Rejecting attachment 120179 [details] from commit-queue. Failed to run "['/mnt/git/webkit-commit-queue/Tools/Scripts/webkit-patch', '--status-host=queues.webkit.org', '-..." exit_code: 2 Last 500 characters of output: WebCore/rendering/RenderObject.cpp patching file Source/WebCore/rendering/RenderObject.h patching file Source/WebCore/rendering/RenderObjectChildList.cpp patching file Source/WebCore/rendering/RenderRegion.cpp patching file Source/WebCore/rendering/RenderRegion.h patching file Source/WebCore/rendering/RenderView.cpp patching file Source/WebCore/rendering/RenderView.h Failed to run "[u'/mnt/git/webkit-commit-queue/Tools/Scripts/svn-apply', u'--reviewer', u'Tony Chang', u'--force']" exit_code: 1 Full output: http://queues.webkit.org/results/11039357
Created attachment 121104 [details] Patch for landing
Comment on attachment 121104 [details] Patch for landing Clearing flags on attachment: 121104 Committed r104036: <http://trac.webkit.org/changeset/104036>
All reviewed patches have been landed. Closing bug.
This didn't seem to change the perf results, so the previous changes (removing the branch in RenderObject::style() and inlining initForRegionStyling()) seem to be sufficient. I think the path forward is to first change callers of RenderObject::style() to cache the pointer, then try to reland the patch and watch for regressions.