Sharing tab hangs gnome-control-center

[Dave Jones]

On the Ubuntu Desktop for Raspberry Pi current Mantic beta images (with gnome-control-center version 1.45.0-1ubuntu1), on a Raspberry Pi 4B, switching to the "Sharing" tab hangs the application (presenting the Force Quit / Wait overlay prompt).

Worse, re-launching the application simply re-opens the "Sharing" tab resulting in another hang. A workaround for now is to launch the application from the command-line, forcing initial selection of a different tab, e.g. "gnome-control-center info-overview".

Launching the application from the command line with "gnome-control-center sharing" also causes the hang, but unfortunately there's no additional console output to shed any light on the situation. I'm not sure if this is Pi specific (haven't attempted it on any other platforms yet).

[Ubuntu QA Website (ubuntuqa)]

This bug has been reported on the Ubuntu ISO testing tracker.

A list of all reports related to this bug can be found here:
https://iso.qa.ubuntu.com/qatracker/reports/bugs/2037017

[Sebastien Bacher (seb128)]

Thank you for your bug report. I haven't see the issue on amd64 but I don't know if that means it's arm64 specific.

Could you check if there is anything printed in the journal at this time? It could be an issue with rygel or another of the sharing services.

Also could you get a gdb 't a a bt' of the process while it's hanging? Ideally with libglib2.0-0-dbgsym libgtk-4-1-dbgsym gnome-control-center-dbgsym installed? It should give us a clue of where it's hanging (likely waiting for a reply from a dbus service)

[Dave Jones (waveform)]

Attaching backtrace from gdb with debug symbols

[Dave Jones (waveform)]

Additional observation: I re-tested this on the Pi 4 and 5. The hang only occurs on the Pi 4 and 400, but does *not* occur on the 5

[Sebastien Bacher (seb128)]

the backtrace suggests it's blocked trying to access the keyring to read/write a rdp password

[Dave Jones (waveform)]

Oh, one other note: no output in the journal during the hang; I'm afraid the only info is from the backtrace.

[Dave Jones (waveform)]

Argh! During ISO testing I thought I'd just quickly check this still didn't occur on a Pi 5... only to find it did, this time. I've still not encountered a Pi 4 where it *didn't* occur, but I wonder if this is a race-condition somewhere and just more prone to occurring on the slower 4.

[Dave Jones (waveform)]

Failed to reproduce this on the jammy .4 images during ISO testing. Possibility it's either fixed or only occurs on later versions. Will attempt to reproduce on mantic later, if I've got the time.

[Dave Jones (waveform)]

Failed to reproduce on noble beta images; marking invalid as that's two failures now.