Talos has detected a Firefox performance regression from push a4df3aa93c77d99b7a8e96f3e90ecc29c5b36575. As author of one of the patches included in that push, we need your help to address this regression. Regressions: 47% tcanvasmark summary linux64 pgo e10s 11447.29 -> 6088.71 43% tcanvasmark summary linux64 opt e10s 9294.29 -> 5260.42 33% tcanvasmark summary windows7-32 pgo e10s 9809 -> 6526.04 32% tcanvasmark summary windows8-64 pgo e10s 9010.83 -> 6147.71 32% tcanvasmark summary windows7-32 opt e10s 8608.88 -> 5875.5 31% tcanvasmark summary windows8-64 opt e10s 8862.04 -> 6138.17 5% tp5o Main_RSS osx-10-10 opt e10s 321749743.28 -> 339191380.26 2% tp5o Main_RSS osx-10-10 opt 386237849.55 -> 394083169.37 You can find links to graphs and comparison views for each of the above tests at: https://treeherder.mozilla.org/perf.html#/alerts?id=5099 On the page above you can see an alert for each affected platform as well as a link to a graph showing the history of scores for this test. There is also a link to a treeherder page showing the Talos jobs in a pushlog format. To learn more about the regressing test(s), please see: https://wiki.mozilla.org/Buildbot/Talos/Tests For information on reproducing and debugging the regression, either on try or locally, see: https://wiki.mozilla.org/Buildbot/Talos/Running *** Please let us know your plans within 3 business days, or the offending patch(es) will be backed out! *** Our wiki page outlines the common responses and expectations: https://wiki.mozilla.org/Buildbot/Talos/RegressionBugsHandling

:smaug, can you take a look at this bug? The canvasmark regression is concerning, I didn't see any mention in the root cause bug to indicate this was expected.

Looks very similar to Bug 1316575 which happened last time prioritized vsync was enabled.

:smaug, is there work to do here? I am not clear based on comment 2 if this is expected and we need to accept it, or if there is pending work?

I think I need to look at the canvasmark test. (I have no idea what that test is)

Canvasmark is documented here: https://wiki.mozilla.org/Buildbot/Talos/Tests#CanvasMark I added :snorp to this bug as he requested the test a few years ago.

tryserver debugging, since I apparently can't run canvasmark locally. This might help, https://treeherder.mozilla.org/perf.html#/comparechooser?newProject=try&newRevision=1e9469375fc0750349114e439fdd3ac2a6df29e6

I'm pretty sure this should fix the issue... but let's see what tryserver says. remote: Follow the progress of your build on Treeherder: remote: https://treeherder.mozilla.org/#/jobs?repo=try&revision=1933d9ea0b702d879f910edfbdc43928d998bf3a remote: remote: It looks like this try push has talos jobs. Compare performance against a baseline revision: remote: https://treeherder.mozilla.org/perf.html#/comparechooser?newProject=try&newRevision=1933d9ea0b702d879f910edfbdc43928d998bf3a

Comment on attachment 8837592 [details] [diff] [review] v2 I think I want to make still one tweak

remote: Follow the progress of your build on Treeherder: remote: https://treeherder.mozilla.org/#/jobs?repo=try&revision=2b3beb5e72e8503cf4909bbe0517856daa15e3c3 remote: remote: It looks like this try push has talos jobs. Compare performance against a baseline revision: remote: https://treeherder.mozilla.org/perf.html#/comparechooser?newProject=try&newRevision=2b3beb5e72e8503cf4909bbe0517856daa15e3c3

And the name "mLastParentTick" because it is the TimeStamp coming from parent. I wish we had some sane units saying whether we're dealing with timestamps from parent or child.

Thanks for the comments on IRC. I agree this makes the code easier to understand. And I removed the GC/CC handling. remote: Follow the progress of your build on Treeherder: remote: https://treeherder.mozilla.org/#/jobs?repo=try&revision=994be72cd6f8023f1995eee6ca769152695b162a remote: remote: It looks like this try push has talos jobs. Compare performance against a baseline revision: remote: https://treeherder.mozilla.org/perf.html#/comparechooser?newProject=try&newRevision=994be72cd6f8023f1995eee6ca769152695b162a

Comment on attachment 8837706 [details] [diff] [review] v4 Review of attachment 8837706 [details] [diff] [review]: ----------------------------------------------------------------- Yep, this is easier to read.

Pushed by opettay@mozilla.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/05a99b44e424 if processing high priority vsyncs takes too much time, fallback to normal priority, r=afarre

thank you for the fix, I have verified this on the perfherder graphs!

Based on https://treeherder.mozilla.org/perf.html#/graphs?series=%5Bmozilla-inbound,ce1b313532802d346b16d4cedd6ebe9c20d40b38,1,1%5D the canvasmark case looks ok, and Main_RSS was discussed already in the other bug.