1958043 - [wayland][kwin] aVsyncTimestamp <= tickStart assert in nsRefreshDriver

Status: RESOLVED FIXED

(Core :: Widget: Gtk, defect, P3)

Description

[Emilio Cobos Álvarez [:emilio]]

I see this assert fail intermittently in my debug builds, only on Wayland. Might be related to the weird wayland vsync stuff Olli was seeing?

Comment 1

[Martin Stránský [:stransky] (ni? me)]

Yes, that's possible. Can you try to capture a log for it? MOZ_LOG="WidgetVSync:5 WidgetWayland:5"
Thanks.

Comment 2

[Emilio Cobos Álvarez [:emilio]]

Attachment: Log

I just caught this, here's the log I could rescue (very verbose)

Comment 3

[Martin Stránský [:stransky] (ni? me)]

Thanks. The log looks correct, ale time stamps from Wayland backend are in row. I'd expect a problem when we have "WaylandVsyncSource::EmulatedWindowCallback" in the log but all callbacks are from compositor.

Comment 4

[Martin Stránský [:stransky] (ni? me)]

I don't think it's related - the assert triggers when system time is changed. It's not related to our event/vsync code but rather system config. I've seen that on X11 too.

Comment 5

[Martin Stránský [:stransky] (ni? me)]

The assertion may be adjusted for Wayland - we can't expect any relation between Wayland event timestamp and system monotonic time.

Comment 6

[Emilio Cobos Álvarez [:emilio]]

Hmm, I only see this on KWin, and not when system time changes necessarily... But ok, maybe we can just suppress it on wayland. It's also not a new problem

Comment 7

[Emilio Cobos Álvarez [:emilio]]

Attachment: Bug 1958043 - Weaken vsync timestamp assert on wayland. r=smaug!

Comment 8

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/244a7e35efec
Weaken vsync timestamp assert on wayland. r=smaug

Comment 9

[Sandor Molnar[:smolnar]]

https://hg.mozilla.org/mozilla-central/rev/244a7e35efec