[alsa-devel] [snd_via82xx] EAGAIN on snd_pcm_prepare after EPIPE on mmap_commit
Hello ! I am encountering some issues with pulseaudio : after some time and for no evident reason PA stops routing the sound. That's what I get with $pulseaudio -vvv : http://pastebin.mandriva.com/13648 (got from a 2.6.29.6_mdv kernel) running $pacmd list-sinks when stuck gives : http://pastebin.mandriva.com/13649
Our pulseaudio integrator Colin Guthrie concluded : "OK, so what is happening is that some alsa level problem is occurring (due to mmap). Pulse tries to recover from this but cannot. The alsa sink (and the whole card) is subsequently unloaded due to this fatal error." And now.. a quick paste from irc : "[00:44:20] <coling> Jul 22 14:04:29 <mezcalero> coling: driver issue i am pretty sure [00:44:20] <coling> Jul 22 14:04:33 <mezcalero> i saw that before once [00:44:20] <coling> Jul 22 14:04:55 <mezcalero> coling: i.e. EAGAIN on snd_pcm_prepare after EPIPE on mmap_commit [00:44:23] <coling> Jul 22 14:05:08 <mezcalero> coling: needs to be brought to attention of alsa folks i guess [00:44:30] <coling> That's what I was told."
I'm now on 2.6.30 and I can reproduce that bug, then I'm bringing that to you ! :)
Thanks !
At Thu, 23 Jul 2009 17:21:44 +0200, rom1dep wrote:
Hello ! I am encountering some issues with pulseaudio : after some time and for no evident reason PA stops routing the sound. That's what I get with $pulseaudio -vvv : http://pastebin.mandriva.com/13648 (got from a 2.6.29.6_mdv kernel) running $pacmd list-sinks when stuck gives : http://pastebin.mandriva.com/13649
Our pulseaudio integrator Colin Guthrie concluded : "OK, so what is happening is that some alsa level problem is occurring (due to mmap). Pulse tries to recover from this but cannot. The alsa sink (and the whole card) is subsequently unloaded due to this fatal error." And now.. a quick paste from irc : "[00:44:20] <coling> Jul 22 14:04:29 <mezcalero> coling: driver issue i am pretty sure [00:44:20] <coling> Jul 22 14:04:33 <mezcalero> i saw that before once [00:44:20] <coling> Jul 22 14:04:55 <mezcalero> coling: i.e. EAGAIN on snd_pcm_prepare after EPIPE on mmap_commit
This is odd. EAGAIN is returned only at open, read/write and drain ioctls.
Could you track with strace?
Takashi
2009/7/23 Takashi Iwai tiwai@suse.de:
At Thu, 23 Jul 2009 17:21:44 +0200, rom1dep wrote:
Hello ! I am encountering some issues with pulseaudio : after some time and for no evident reason PA stops routing the sound. That's what I get with $pulseaudio -vvv : http://pastebin.mandriva.com/13648 (got from a 2.6.29.6_mdv kernel) running $pacmd list-sinks when stuck gives : http://pastebin.mandriva.com/13649
Our pulseaudio integrator Colin Guthrie concluded : "OK, so what is happening is that some alsa level problem is occurring (due to mmap). Pulse tries to recover from this but cannot. The alsa sink (and the whole card) is subsequently unloaded due to this fatal error." And now.. a quick paste from irc : "[00:44:20] <coling> Jul 22 14:04:29 <mezcalero> coling: driver issue i am pretty sure [00:44:20] <coling> Jul 22 14:04:33 <mezcalero> i saw that before once [00:44:20] <coling> Jul 22 14:04:55 <mezcalero> coling: i.e. EAGAIN on snd_pcm_prepare after EPIPE on mmap_commit
This is odd. EAGAIN is returned only at open, read/write and drain ioctls.
Could you track with strace?
Takashi
It wasn't easy : that big output made my config swap and finally pulseaudio leave for "E: cpulimit.c: Received request to terminate due to CPU overload." but : http://pastebin.mandriva.com/13665
'Twas brillig, and rom1dep at 23/07/09 18:23 did gyre and gimble:
2009/7/23 Takashi Iwai tiwai@suse.de:
At Thu, 23 Jul 2009 17:21:44 +0200, rom1dep wrote:
Hello ! I am encountering some issues with pulseaudio : after some time and for no evident reason PA stops routing the sound. That's what I get with $pulseaudio -vvv : http://pastebin.mandriva.com/13648 (got from a 2.6.29.6_mdv kernel) running $pacmd list-sinks when stuck gives : http://pastebin.mandriva.com/13649
Our pulseaudio integrator Colin Guthrie concluded : "OK, so what is happening is that some alsa level problem is occurring (due to mmap). Pulse tries to recover from this but cannot. The alsa sink (and the whole card) is subsequently unloaded due to this fatal error." And now.. a quick paste from irc : "[00:44:20] <coling> Jul 22 14:04:29 <mezcalero> coling: driver issue i am pretty sure [00:44:20] <coling> Jul 22 14:04:33 <mezcalero> i saw that before once [00:44:20] <coling> Jul 22 14:04:55 <mezcalero> coling: i.e. EAGAIN on snd_pcm_prepare after EPIPE on mmap_commit
This is odd. EAGAIN is returned only at open, read/write and drain ioctls.
Could you track with strace?
Takashi
It wasn't easy : that big output made my config swap and finally pulseaudio leave for "E: cpulimit.c: Received request to terminate due to CPU overload." but : http://pastebin.mandriva.com/13665
You probably want to repost using a different pastebin... our stupid one cuts the posts short at some pointlessly low value... I've asked about this to get it made bigger but I don't know who looks after it. I'd recommend pastebin.com.
Col
2009/7/24 Colin Guthrie gmane@colin.guthr.ie:
'Twas brillig, and rom1dep at 23/07/09 18:23 did gyre and gimble:
2009/7/23 Takashi Iwai tiwai@suse.de:
At Thu, 23 Jul 2009 17:21:44 +0200, rom1dep wrote:
Hello ! I am encountering some issues with pulseaudio : after some time and for no evident reason PA stops routing the sound. That's what I get with $pulseaudio -vvv : http://pastebin.mandriva.com/13648 (got from a 2.6.29.6_mdv kernel) running $pacmd list-sinks when stuck gives : http://pastebin.mandriva.com/13649
Our pulseaudio integrator Colin Guthrie concluded : "OK, so what is happening is that some alsa level problem is occurring (due to mmap). Pulse tries to recover from this but cannot. The alsa sink (and the whole card) is subsequently unloaded due to this fatal error." And now.. a quick paste from irc : "[00:44:20] <coling> Jul 22 14:04:29 <mezcalero> coling: driver issue i am pretty sure [00:44:20] <coling> Jul 22 14:04:33 <mezcalero> i saw that before once [00:44:20] <coling> Jul 22 14:04:55 <mezcalero> coling: i.e. EAGAIN on snd_pcm_prepare after EPIPE on mmap_commit
This is odd. EAGAIN is returned only at open, read/write and drain ioctls.
Could you track with strace?
Takashi
It wasn't easy : that big output made my config swap and finally pulseaudio leave for "E: cpulimit.c: Received request to terminate due to CPU overload." but : http://pastebin.mandriva.com/13665
You probably want to repost using a different pastebin... our stupid one cuts the posts short at some pointlessly low value... I've asked about this to get it made bigger but I don't know who looks after it. I'd recommend pastebin.com.
Col
Or not :) the total strace takes 43Mo, I kept a few before and a few after pulseaudio's "oops" If it's not enough I can paste more on another pastebin but that's a terribly big output...
--
Colin Guthrie gmane(at)colin.guthr.ie http://colin.guthr.ie/
Day Job: Tribalogic Limited [http://www.tribalogic.net/] Open Source: Mandriva Linux Contributor [http://www.mandriva.com/] PulseAudio Hacker [http://www.pulseaudio.org/] Trac Hacker [http://trac.edgewall.org/]
Alsa-devel mailing list Alsa-devel@alsa-project.org http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
participants (3)
-
Colin Guthrie -
rom1dep -
Takashi Iwai