Created attachment 44136 [details]
gstack of hang
<caillon> mezcalero, am getting this hang in mutter several times a day.... http://caillon.fedorapeople.org/mutter.gstack
<mezcalero> caillon: urks, looks like a libcanberra issue
<mezcalero> please file a bug against lc, not pa
<mezcalero> looks bad
<mezcalero> probably something wrong with your pa server...
<mezcalero> and the client is not racting to this as it should
Created attachment 44802 [details] [review]
workaround patch I've been using
FWIW, I've been using this patch for 3 weeks now in order to restore sanity to my desktop
I'm getting this too, and have been for a few weeks; every so often the shell hangs and I just have to wait until the stack gets too deep and the shell dies. Stack looks exactly like caillon's.
Created attachment 44803 [details]
screencap of the stacktrace
I'm wondering if this could be ANOTHER case of stuff scoped to the X server (as pulseaudio tries to do) breaking because of sharing between gdm and initial login.
Created attachment 44887 [details] [review]
I think this is a better patch; testing it now.
Hi, I think I am experiencing this issue too.
I've setup a pulseaudio server in my LAN to get my desktop pc connected to hifi equipment. So I am using archlinux x64 with libcanberra 0.28-2 on my client.
The pulseaudio server is only online ondemand, so most of the time I am using local soundcard of my desktop. My gtk using applications are drawing the GUI very lazy while the PA server is offline. It is fixed if I boot the PA Server.
This is really annoying because browsing in firefox isn't funny when PA Server is offline.
I've used strace to get following information while firefox (or any other app) is responding lazy: http://nopaste.info/481c0626fa_nl.html
As you see it seems to been related to libcanberra.
Could somebody tell me an ETA for a fix?
i tried the patch to add "p->reconnect = false" to pulse.c and this fixed the issue for me too :-)
Fixed in git.