Frequent crashes of PulseAudio

[Jakub 'Livio' Rusinek]

Hi,

I want to report problem which appeared after upgrading PulseAudio to
Rawhide's version.

GNOME starts PA correctly, but after starting some playback (video or
audio) PA crashes. Same when nothing is playing and I run Evolution.

Part of log:

Apr  7 20:28:09 fedora pulseaudio[5335]: main.c: setrlimit(RLIMIT_NICE,
(31, 31)) failed: Operacja niedozwolona
Apr  7 20:28:09 fedora pulseaudio[5335]: main.c:
setrlimit(RLIMIT_RTPRIO, (9, 9)) failed: Operacja niedozwolona

I didn't set PA to run as system instance or with bigger priority...
Also cpu_limit option is commented.

I can't normally play videos or listen to music, because PA may crash.

-- 
Jakub 'Livio' Rusinek
http://liviopl.jogger.pl/

[Dimi Paun]

On Mon, 2008-04-07 at 20:33 +0200, Jakub 'Livio' Rusinek wrote:
>  I can't normally play videos or listen to music, because PA may
> crash. 
Welcome to the club:
    https://bugzilla.redhat.com/show_bug.cgi?id=438594

-- 
Dimi Paun <dimi(a)lattica.com&gt;
Lattica, Inc.

[Jakub 'Livio' Rusinek]

Dnia 2008-04-07, pon o godzinie 14:38 -0400, Dimi Paun pisze:
>  On Mon, 2008-04-07 at 20:33 +0200, Jakub 'Livio' Rusinek
> wrote:
>  > I can't normally play videos or listen to music, because PA may crash.
>  
>  Welcome to the club:
>      https://bugzilla.redhat.com/show_bug.cgi?id=438594 
Thanks, I'll CC myself.

-- 
Jakub 'Livio' Rusinek
http://liviopl.jogger.pl/

[Lennart Poettering]

On Mon, 07.04.08 14:38, Dimi Paun (dimi(a)lattica.com) wrote:

>  On Mon, 2008-04-07 at 20:33 +0200, Jakub 'Livio' Rusinek
> wrote:
>  > I can't normally play videos or listen to music, because PA may crash.
>  
>  Welcome to the club:
>      https://bugzilla.redhat.com/show_bug.cgi?id=438594 
Hmm, AFAICS this is totally unrelated.

If you get a CPU time limit abort than this can have two reasons:

1) You got an awfully slow CPU (very unlikely, if you bought your
machine in the last 10 years and is x86 or x86-64)

2) Something causes PA to enter an endless loop which causes it to eat
100% CPU.

In all the cases I managed to track down #2 is caused by a faulty ALSA
driver where poll() signalled readiness while there actually was
nothing to do. This happened with drivers which required the "multi"
libasound module, such as emu10k.

Lennart

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

[Dimi Paun]

On Mon, 2008-04-07 at 22:39 +0200, Lennart Poettering wrote:
>  
>  If you get a CPU time limit abort than this can have two reasons:
>  
>  1) You got an awfully slow CPU (very unlikely, if you bought your
>  machine in the last 10 years and is x86 or x86-64)
>  
>  2) Something causes PA to enter an endless loop which causes it to eat
>  100% CPU.
>  
>  In all the cases I managed to track down #2 is caused by a faulty ALSA
>  driver where poll() signalled readiness while there actually was
>  nothing to do. This happened with drivers which required the "multi"
>  libasound module, such as emu10k. 
Well, 1) clearly doesn't apply to me, I'm running on Core Duo 6400.

As for 2) it is true that I have noticed PA start using 100% CPU for
a few seconds before exiting. Randomly. A lot of times I find it gone
in the morning, with no music playing during the night.

As for sound card, I'm using a bog-standard Intel Corporation 82801H
(ICH8 Family) HD Audio Controller (rev 02).

-- 
Dimi Paun <dimi(a)lattica.com&gt;
Lattica, Inc.

[Jakub 'Livio' Rusinek]

Dnia 2008-04-07, pon o godzinie 16:52 -0400, Dimi Paun pisze:
>  On Mon, 2008-04-07 at 22:39 +0200, Lennart Poettering wrote:
>  > 
>  > If you get a CPU time limit abort than this can have two reasons:
>  > 
>  > 1) You got an awfully slow CPU (very unlikely, if you bought your
>  > machine in the last 10 years and is x86 or x86-64)
>  > 
>  > 2) Something causes PA to enter an endless loop which causes it to eat
>  > 100% CPU.
>  > 
>  > In all the cases I managed to track down #2 is caused by a faulty ALSA
>  > driver where poll() signalled readiness while there actually was
>  > nothing to do. This happened with drivers which required the "multi"
>  > libasound module, such as emu10k.
>  
>  Well, 1) clearly doesn't apply to me, I'm running on Core Duo 6400. 
Duron 1,6GHz here.

>  As for 2) it is true that I have noticed PA start using 100% CPU for
>  a few seconds before exiting. Randomly. A lot of times I find it gone
>  in the morning, with no music playing during the night. 
Didn't noticed...

>  As for sound card, I'm using a bog-standard Intel Corporation
> 82801H
>  (ICH8 Family) HD Audio Controller (rev 02). 
snd-via82xx: VIA 8235 integrated soundcard. AC97 standard.

-- 
Jakub 'Livio' Rusinek
http://liviopl.jogger.pl/

[Lennart Poettering]

On Mon, 07.04.08 20:33, Jakub 'Livio' Rusinek (jakub.rusinek(a)gmail.com) wrote:

>  Hi,
>  
>  I want to report problem which appeared after upgrading PulseAudio to
>  Rawhide's version.
>  
>  GNOME starts PA correctly, but after starting some playback (video or
>  audio) PA crashes. Same when nothing is playing and I run Evolution. 
Crashes? In which way?

Please run PA in a shell with "-vv":

pulseaudio -k
pulseaudio -vv

and file a bug with the output when it crashes in bugzilla. Even
better is a complete stack trace:

http://fedoraproject.org/wiki/StackTraces

>  
>  Part of log:
>  
>  ???Apr  7 20:28:09 fedora pulseaudio[5335]: main.c: setrlimit(RLIMIT_NICE,
>  (31, 31)) failed: Operacja niedozwolona
>  Apr  7 20:28:09 fedora pulseaudio[5335]: main.c:
>  setrlimit(RLIMIT_RTPRIO, (9, 9)) failed: Operacja niedozwolona 
This is not an error. Just an informational warning you can saefuly ignore.

Lennart

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