1680743 - 46.52 - 48.24% about_preferences_basic (windows10-64-shippable-qr) regression on push e4c8bc7feaa60470c026236a7d89881a74b29d34 (Wed December 2 2020)

Status: RESOLVED FIXED

(Core :: Layout, defect)

Description

[Alexandru Ionescu (needinfo me) [:alexandrui]]

Perfherder has detected a talos performance regression from push e4c8bc7feaa60470c026236a7d89881a74b29d34. As author of one of the patches included in that push, we need your help to address this regression.

Regressions:

Ratio	Suite	Test	Platform	Options	Absolute values (old vs new)
48%	about_preferences_basic		windows10-64-shippable-qr	e10s stylo webrender-sw	109.55 -> 162.39
47%	about_preferences_basic		windows10-64-shippable-qr	e10s stylo webrender-sw	107.10 -> 156.93

Details of the alert can be found in the alert summary, including links to graphs and comparisons for each of the affected tests. Please follow our guide to handling regression bugs and let us know your plans within 3 business days, or the offending patch(es) will be backed out in accordance with our regression policy.

For more information on performance sheriffing please see our FAQ.

Comment 1

[Robert Mader [:rmader]]

Hm interesting. The changes in bug 1645528 are on a quite general level, so all timers on all OSs/platform in e10s contexts are affected. If there's something wrong about bug 1645528, I'd expect issues to show up on other platforms as well.

Do I understand things correctly that currently only Win10 with sw-wr are affected (AFAIK sw-wr is quite newish and not used anywhere by default, nor as well tested as wr on hardware)?

Is there any chance that this is reproducible on Linux, even to a lesser extend? Or in hardware WR?

Comment 2

[Robert Mader [:rmader]]

To follow up on this - if this is the only configuration affected (also only the qr build?), then it would be a strong indicator that bug 1645528 just uncovered a pre-existing bug.
Which in the case of sw-wr would AFAIK also not be critical, or at least not critical enough to justify backing out of bug 1645528, as it provides a quite important feature for the Wayland backend.

Comment 3

[Alexandru Ionescu (needinfo me) [:alexandrui]]

(In reply to Robert Mader [:rmader] from comment #1)

Do I understand things correctly that currently only Win10 with sw-wr are affected (AFAIK sw-wr is quite newish and not used anywhere by default, nor as well tested as wr on hardware)?

Is there any chance that this is reproducible on Linux, even to a lesser extend? Or in hardware WR?

The only OS I tend to say is not affected (yet) is macos. Otheriwse, the graph shows that both windows and linux webrender (with and without sw) are affected.

Comment 4

[Robert Mader [:rmader]]

Thanks, will look into the test then.

Comment 5

[BugBot [:suhaib / :marco/ :calixte]]

Set release status flags based on info from the regressing bug 1645528

Comment 6

[Robert Mader [:rmader]]

I may have a fix for this: https://treeherder.mozilla.org/jobs?repo=try&revision=6724de3b5425dd460a935c597f2e7619cbfc0641

Comment 8

[Dave Hunt [:davehunt] [he/him] ⌚BST]

:rmader could you confirm if this is something we should back out from mozilla-beta?

Comment 9

[Robert Mader [:rmader]]

Yes, please. See bug 1645528 comment 68. I'm still looking into it but unfortunately didn't solve things in time.

Comment 10

[Julien Cristau [:jcristau]]

Regressing bug backed out of beta:
https://hg.mozilla.org/releases/mozilla-beta/rev/01efe9d5b4e1006d7908576c8540139b633294df

Comment 11

[Robert Mader [:rmader]]

This is hopefully/likely fixed in bug 1681030