Summary: | Make RenderLayer not arena-allocated | ||||||
---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Simon Fraser (smfr) <simon.fraser> | ||||
Component: | Layout and Rendering | Assignee: | Nobody <webkit-unassigned> | ||||
Status: | RESOLVED FIXED | ||||||
Severity: | Normal | CC: | commit-queue, eric, esprehn+autocc, glenn, hyatt, jchaffraix, kling, kondapallykalyan, ojan, simon.fraser | ||||
Priority: | P2 | ||||||
Version: | 528+ (Nightly build) | ||||||
Hardware: | Unspecified | ||||||
OS: | Unspecified | ||||||
Bug Depends on: | 122651 | ||||||
Bug Blocks: | |||||||
Attachments: |
|
Description
Simon Fraser (smfr)
2012-05-25 11:42:06 PDT
Agreed! How is RenderLayer different than the other Render* classes WRT usefulness of being arena-allocated? There don't tend to be that many RenderLayers on most pages (tens - hundreds, not thousands). Fair enough. Created attachment 213975 [details]
Patch
Comment on attachment 213975 [details] Patch Clearing flags on attachment: 213975 Committed r157297: <http://trac.webkit.org/changeset/157297> All reviewed patches have been landed. Closing bug. Re-opened since this is blocked by bug 122651 This caused crashes on tests (debug and release WK2). Very strange that EWS didn't catch it, was the landed patch identical to the one posted here? Rolled out in <https://trac.webkit.org/r157306>. Results: http://build.webkit.org/results/Apple%20MountainLion%20Debug%20WK2%20(Tests)/r157297%20(13299)/results.html Committed r157333: <http://trac.webkit.org/changeset/157333> |