Summary: | Increase priority on SharedTimer source | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Andre Moreira Magalhaes <andrunko> | ||||||
Component: | WebKitGTK | Assignee: | Nobody <webkit-unassigned> | ||||||
Status: | REOPENED --- | ||||||||
Severity: | Normal | CC: | bugs-noreply, clopez, commit-queue, mrobinson | ||||||
Priority: | P2 | ||||||||
Version: | 528+ (Nightly build) | ||||||||
Hardware: | Unspecified | ||||||||
OS: | Unspecified | ||||||||
Attachments: |
|
Description
Andre Moreira Magalhaes
2014-06-20 10:34:16 PDT
Created attachment 233433 [details]
Patch
Created attachment 233434 [details]
Patch with proper changelog
Oops, forgot to update changelog, new patch in.
Comment on attachment 233434 [details]
Patch with proper changelog
This would be a problem for webkit1, but is a proper fix for the webkit2 model IMO, since we do not compete with GTK+ mainloop sources anymore.
Comment on attachment 233434 [details] Patch with proper changelog Clearing flags on attachment: 233434 Committed r170529: <http://trac.webkit.org/changeset/170529> All reviewed patches have been landed. Closing bug. This caused a performance regression on the test Animation/balls:FrameRate. See: https://bugs.webkit.org/show_bug.cgi?id=134972 The priority of the timers is very sensitive and can negatively affect animations, so I recommend we roll this out until we understand the isseu. (In reply to comment #7) > The priority of the timers is very sensitive and can negatively affect animations, so I recommend we roll this out until we understand the isseu. I agree here. I actually opened this bug more to start a discussion on the proper priorities to use for the various timers/sources as we had some issues with JS being starved and thus breaking some webapps that relied on setTimeout calls. This issue only happened in a very specific usecase where we had a browser running in a VM with limited resources. I vote for reverting this but would like to see your opinion on what is the best approach to solve the issue with JS being starved when running on limited resources. I reverted r170529 on r171343 <https://trac.webkit.org/r171343> because it caused a ~10% performance regression on the perf test Animation/balls: https://webkit.org/b/134972 Reopening this bug. |