1820293 - [sway wayland] Tooltip crash: Buffer size (1173x128) is not divisible by scale (2)

Status: RESOLVED DUPLICATE of bug 1803016

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

Description

[Simon McFarlane]

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:109.0) Gecko/20100101 Firefox/112.0

Steps to reproduce:

Running sway master on Arch Linux. Firefox Nightly is the only window open.

Hover the mouse over something (a link, a tab).

Actual results:

The entire browser crashed with the Wayland error: Buffer size (1173x128) is not divisible by scale (2). This only happens occasionally (sometimes the tooltip does appear).

I've filed a bug against a similar error before, however the buffer size was different each time. In this case, the buffer size is the same every time: 1173x128. If you search the crash reports, you will find many users seeing this error, including the exact buffer size, on Nightly since 2023-02-24 on a variety of GPUs and Linux distributions: https://crash-stats.mozilla.org/search/?moz_crash_reason=~Buffer%20size%20%281173x128%29%20is%20not%20divisible%20by%20scale&date=%3E%3D2022-09-03T23%3A39%3A00.000Z&date=%3C2023-03-03T23%3A39%3A00.000Z&_facets=signature&_sort=-date&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform#crash-reports

Expected results:

The browser should not crash and the tooltip should be displayed.

Comment 1

[Adam Harvey]

I'm seeing this as well since upgrading from Arch's firefox-developer-edition 111.0b8-3 package to 112.0b1-1. Doesn't happen every time on a tooltip being displayed, but has happened a couple of times today. Feels like a regression in 112.

Most recent crash report: https://crash-stats.mozilla.org/report/index/12b6ed4d-d356-48ee-b11b-f10d00230316

Comment 2

[philipp+mozilla]

https://bugzilla.mozilla.org/show_bug.cgi?id=1795851 is back in 112. It was fixed and worked well in 111, but since upgrading to 112, it crashes Firefox on sway again.

Comment 3

[Emilio Cobos Álvarez (:emilio)]

If someone that can repro can run mozregression --good 111 --bad 112 it'd be greatly appreciated

Comment 4

[Adam Harvey]

I got b2 last night and it seems that I can't reproduce this any more, but I'll run mozregression if it rears its head again.

Comment 5

[Adam Harvey]

(In reply to Adam Harvey from comment #4)

I got b2 last night and it seems that I can't reproduce this any more, but I'll run mozregression if it rears its head again.

And then, of course, it happened again five minutes later.

Since this is fairly non-deterministic, I can't be totally sure that this is actually correct, but this is the last few lines of mozregression log output:

2023-03-16T10:49:55.910000: INFO : Narrowed integration regression window from [b4f28139, 1ff252c1] (3 builds) to [b4f28139, a85ff6a7] (2 builds) (~1 steps left)
2023-03-16T10:49:55.919000: DEBUG : Starting merge handling...
2023-03-16T10:49:55.919000: DEBUG : Using url: https://hg.mozilla.org/integration/autoland/json-pushes?changeset=a85ff6a754efad86dcdf5ad2cd9cfbe61c9b0bd7&full=1
2023-03-16T10:49:55.919000: DEBUG : redo: attempt 1/3
2023-03-16T10:49:55.919000: DEBUG : redo: retry: calling _default_get with args: ('https://hg.mozilla.org/integration/autoland/json-pushes?changeset=a85ff6a754efad86dcdf5ad2cd9cfbe61c9b0bd7&full=1',), kwargs: {}, attempt #1
2023-03-16T10:49:55.920000: DEBUG : urllib3.connectionpool: Resetting dropped connection: hg.mozilla.org
2023-03-16T10:49:56.931000: DEBUG : urllib3.connectionpool: https://hg.mozilla.org:443 "GET /integration/autoland/json-pushes?changeset=a85ff6a754efad86dcdf5ad2cd9cfbe61c9b0bd7&full=1 HTTP/1.1" 200 None
2023-03-16T10:49:56.957000: DEBUG : Found commit message:
Bug 1813407 [Linux] Remove unused nsWindow::GetMozContainerSize() r=emilio

Depends on D169907

Differential Revision: https://phabricator.services.mozilla.com/D169908

2023-03-16T10:49:56.957000: DEBUG : Did not find a branch, checking all integration branches
2023-03-16T10:49:56.958000: INFO : The bisection is done.
2023-03-16T10:49:56.959000: INFO : Stopped

Comment 6

[Darkspirit]

Thanks, that's plausible! bug 1813407 landed in 20230224092408 which is the first bad build id from comment 0.

Comment 7

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

:stransky, since you are the author of the regressor, bug 1813407, could you take a look?

For more information, please visit auto_nag documentation.

Comment 8

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

The bug is linked to a topcrash signature, which matches the following criterion:

• Top 5 desktop browser crashes on Linux on beta

:stransky, could you consider increasing the severity of this top-crash bug?

For more information, please visit auto_nag documentation.