After new-multi was enabled the RenderFlowThread code path started affecting some of the performance tests (some have multi-column elements). Investigate this regression and find a fix for it.
The changeset to blame: http://trac.webkit.org/changeset/168286
Created attachment 231135 [details] Patch
Created attachment 231136 [details] Patch that applies
Comment on attachment 231136 [details] Patch that applies r=me. Instead of testing the single region case directly, it may be a good idea to have a helper function.
I don't suspect I'll use this condition in more places. I'm kinda skeptical having a function hasOneRegion actually brings value. If this is an optimization that can be applied in other places as well, I'll create that helper function.
Comment on attachment 231136 [details] Patch that applies Clearing flags on attachment: 231136 Committed r168534: <http://trac.webkit.org/changeset/168534>
All reviewed patches have been landed. Closing bug.