Bug 485022

Summary: Wireplumber crashes when the window/screen my portal is capturing is removed and I change volume level
Product: [Plasma] xdg-desktop-portal-kde Reporter: Brodie Robertson <brodierobertson54321>
Component: generalAssignee: Plasma Bugs List <plasma-bugs-null>
Status: RESOLVED UPSTREAM    
Severity: crash CC: aleixpol, jgrulich, nate
Priority: NOR    
Version First Reported In: 6.0.3   
Target Milestone: ---   
Platform: Other   
OS: Linux   
Latest Commit: Version Fixed/Implemented In:
Sentry Crash Report:
Attachments: Attached pipewire log

Description Brodie Robertson 2024-04-04 11:05:32 UTC
Created attachment 168141 [details]
Attached pipewire log

SUMMARY
I'm not entirely what in the chain is at fault here

If I am capturing a window/screen inside of OBS using the KDE portal on the Plasma Wayland session, then I remove that window (by closing it) or that screen (by switching the output of my monitor) but the I don't update what the portal is capturing. If I change my volume level wireplumber instantly segfaults. This only occurs when I am making use of the portal that's why I thought maybe it's a KDE problem, also KDE is mentioned in the pipewire log


STEPS TO REPRODUCE
1. Open OBS
2. Begin capturing a window with the portal
3. Close that window
4. Change your volume level
5. Wireplumber crashes

OBSERVED RESULT
All audio input and output breaks and takes a moment until wireplumber comes back, however pipewire video capture continues going


EXPECTED RESULT
Things not crashing


SOFTWARE/OS VERSIONS

Linux: Arch Linux
KDE Plasma Version: 6.03
KDE Frameworks Version: 6.0.0
Qt Version: 6.6.3
Comment 1 Brodie Robertson 2024-04-04 11:08:10 UTC
I've also made a report on Wireplumber's repo https://gitlab.freedesktop.org/pipewire/wireplumber/-/issues/632
Comment 2 Nate Graham 2024-04-04 22:04:16 UTC
From that link, Wireplumber devs seem to think it's most likely their fault, which would make sense since it's Wireplumber that's crashing, and that's not KDE code. I'd say it's pretty unlikely we'd be making it crash. So let's close this as an upstream issue until there's some indication upstream that we're at fault somehow.