[pulseaudio-discuss] Sound works, then doesn't until I log out and back in

[Lennart Poettering]

On Tue, 28.08.07 00:05, Nikita V. Youshchenko (yoush at cs.msu.su) wrote:

> 
> 
> > On Sun, 26.08.07 23:00, Nikita V. Youshchenko (yoush at cs.msu.su) wrote:
> > > > Playing file /home/dstromberg/Sound/Music/Dan Brown/The Da Vinci
> > > > Code (disc 3)/24 - Chapter 22b.flac
> > > > *** PULSEAUDIO: Unable to connect: Connection terminated
> > > > *** PULSEAUDIO: Unable to connect: Connection terminated
> > > > *** PULSEAUDIO: Unable to connect: Connection terminated
> > > > *** PULSEAUDIO: Unable to connect: Connection terminated
> > >
> > > This is exactly about what I wrote several mail ago:
> > > > - when user starts mplayer within session, 'session-side' pulseaudio
> > > > server crashes at top of pa_memblockq_drop() on
> > > >   assert(length % bq->base == 0);
> >
> > I am pretty sure these two things are unrelated. As mentioned, the
> > assert() issues seems to be a Sunray specific problem.
> 
> Maybe, maybe.
> I've just tried to reproduce it without sunray device involved, and could 
> not. Although I didn't try very hard.
> 
> However, another question rises.
> It looks like 'session-side' pulseaudio crashes because of 'terminal-side' 
> pulseaudio has issues writing to it's device.
> Is it really correct to crash (on assertions or whatever) because of remote 
> misbehaviour? This very looks like a DoS. Shouldn't some recovery measures 
> taken instead? E.g. close problematic connection?

What version of PA is this? Pleawse make sure to run PA 0.9.6. Older
versions of module-tunnel didn't handle the "buffer underflow" message
properly and hit an assert instead.

Lennart

-- 
Lennart Poettering                        Red Hat, Inc.
lennart [at] poettering [dot] net         ICQ# 11060553
http://0pointer.net/lennart/           GnuPG 0x1A015CC4