1728891 - Window chrome occasionally gets stuck

Status: RESOLVED FIXED

(Core :: Graphics, defect, P2)

Description

[Robert Mader [:rmader]]

Under yet unknown circumstances windows sometimes can get stuck. Alt-tabbing out and back makes us render another frame, but doesn't unblock things. So far this was observed in multi-window cases.

Will dig into it.

Comment 1

[Robert Mader [:rmader]]

Attachment: Bug 1728891 - Some cleanups for WaylandVsyncSource, r=stransky

To make it easier to understand. No functional changes intended.

Comment 2

[Robert Mader [:rmader]]

Attachment: Bug 1728891 - Move container initial draw callback to SetupFrameCallback(), r=stransky

This makes sure we handle a case when the wl_surface gets deleted
and recreated without restart of the VsyncSource.

As side-effect we should draw one frame earlier in some cases.

Note: due to missing STR for the original issue this was tested
by re-enabling the WaylandVsyncSource for popups. This patch
makes this case moch more usable, but I'm not yet sure if all
(racy) issues are solved.

Depends on D124571

Comment 3

[Robert Mader [:rmader]]

Ok, I think I was on a totally wrong track here. I'm writing this from such a "stuck" window right now and, assuming I'm seeing the same what others have reported, not all rendering is stuck, but only that of the browser chrome / the main process. I.e. it affects tab switching and scrolling about:support/about:config, but not website content. Scrolling or writing a text there works just fine (and vsynced).

That strongly indicates not a problem in WaylandVsyncSource (we only have one for each window) but rather in WR or some other graphics related part. Another observation is that the next update happens on tab-out, not tab-in.

Jonas, can you confirm that what you experienced was the same?

Comment 4

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/0ed4219472fd
Some cleanups for WaylandVsyncSource, r=stransky

Comment 5

[Jonas Ådahl]

Jonas, can you confirm that what you experienced was the same?

Yes, only the chrome is stuck and inoperable. I can still scroll and interact with the web page itself.

Comment 6

[Cristian Tuns]

https://hg.mozilla.org/mozilla-central/rev/0ed4219472fd

Comment 7

[Robert Mader [:rmader]]

(In reply to Jonas Ådahl from comment #5)

Jonas, can you confirm that what you experienced was the same?

Yes, only the chrome is stuck and inoperable. I can still scroll and interact with the web page itself.

Martin, needinfoing you just so you are aware of the issue - maybe you have an idea what could be causing this?

Comment 8

[Martin Stránský [:stransky] (ni? me) (PTO, back on Feb 17)]

Jonas, can you try to get profile data and check about:performance page?
https://fedoraproject.org/wiki/How_to_debug_Firefox_problems?rd=Bug_info_Firefox#Firefox_performance_issues
Especially the profile may reveal a lag in main process.
Thanks.

Comment 9

[Jonas Ådahl]

(In reply to Martin Stránský [:stransky] (ni? me) from comment #8)

Jonas, can you try to get profile data and check about:performance page?
https://fedoraproject.org/wiki/How_to_debug_Firefox_problems?rd=Bug_info_Firefox#Firefox_performance_issues
Especially the profile may reveal a lag in main process.
Thanks.

I ran into this again, and about:performance didn't report anything as consuming abnormal. The most tab eating most resources was a youtube page playing in a non-stuck window.

Didn't manage to record anything yet, as the issue suddenly stopped itself as I was trying to prepare to start recording.

Comment 10

[Robert Mader [:rmader]]

I haven't seen this for quite a while. Jonas, did you run into this lately?

Comment 11

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

The leave-open keyword is there and there is no activity for 6 months.
:bhood, maybe it's time to close this bug?
For more information, please visit auto_nag documentation.

Comment 12

[Bob Hood [:bhood]]

Robert, should we continue to hold this report open?

Comment 13

[Jonas Ådahl]

Sorry, had missed the question. I haven't seen this for quite some time too.

Comment 14

[Robert Mader [:rmader]]

Sorry, had missed the question. I haven't seen this for quite some time too.

Me neither, thanks for confirming!