[ALSA - driver 0001047]: module hangs at seemingly random times

[bugtrack@alsa-project.org]

A NOTE has been added to this issue.
======================================================================
<https://bugtrack.alsa-project.org/alsa-bug/view.php?id=1047> 
======================================================================
Reported By:                alien999999999
Assigned To:                mjander
======================================================================
Project:                    ALSA - driver
Issue ID:                   1047
Category:                   PCI - au88x0
Reproducibility:            sometimes
Severity:                   block
Priority:                   normal
Status:                     assigned
Distribution:               Mandrake
Kernel Version:             2.6.7
======================================================================
Date Submitted:             04-12-2005 20:43 CEST
Last Modified:              08-24-2005 11:13 CEST
======================================================================
Summary:                    module hangs at seemingly random times
Description: 
sometimes i start playing a song, and it starts playing a few second or so
and hangs, then i kill the application and modprobe -r all sound modules
and modprobe them again to make it work again.

BUT: sometimes not only that happens, but also when i try to kill the apps
it will not kill. when that happens, all kill, killall, top, ps aux
commands hang at the command line and cannot be killed by CTRL-C or
otherwise, i have been able to see that when i stopped my display
managener I did an lsmod and it gave something like this:

snd-pcm-oss            59752  11
snd-mixer-oss          20480  1 snd-pcm-oss
snd-au8810             43760  220
snd-ac97-codec         83408  1 snd-au8810
snd-pcm               108172  112 snd-pcm-oss,snd-au8810,snd-ac97-codec
snd-page-alloc         10384  1 snd-pcm
gameport                3840  1 snd-au8810
snd-mpu401-uart        11904  1 snd-au8810


as you can see the snd-au8810 module seem to have an impossible number of
"dependencies" (i think has to do with the number of unclosed sound-apps
trying to be played; this could be since gaim is programmed to execute an
'aplay %s')

i've had this major crash below only 3 times; and the logs didn't detect
anything specific at the time. the only thing the logs mentioned at that
time was an ntpd sync going on; so the only thing i can think of is that
at a certain moment when a sync is going on, some kind of lock is holding
cause this to happen... the only thing that i can do to fix this is
reset...

it is interesting to note that i also have an snd-emu10k1 as second card,
which never gave problems like this, and i am always able to "modprobe -r
snd-emu10k1" ...
======================================================================

----------------------------------------------------------------------
 alien999999999 - 04-12-05 20:51 
----------------------------------------------------------------------
i have attached the changed files (since i don't have a CVS version of alsa
to diff against)

these files are located in the alsa-driver-1.0.8/alsa-kernel/pci/au88x0/
path

----------------------------------------------------------------------
 Raymond - 08-24-05 11:13 
----------------------------------------------------------------------
When playing mono streams on hw:0,2 (16 subdevices) on my au8830 (32-bits
platform), the 17th mono stream (aplay) seem to be blocked
(snd_pcm_open).

Is this the normal behaviour of snd_pcm_open when maximum number of
subdevice have been opened ?

When playing mono streams on hw:0,0 (32 subdevices) on au8830, the 17th
mono stream (aplay) will aborted immediately since snd_pcm_hw_params()
return no more SRC available error.

Please note that the capture stream also using DMA, SRC and MIXER.

I guess some weird things happen when there is only 1 SRC left and the
ALSA application (stereo stream) request 2 SRC.

au88x0 driver should check the available number of SRC >= the number of
channels before it allocate the SRC and MIXER. THe error recovery may not
released the SRC and MIXER properly.


You should add printk to log the resource DMA, SRC, MIXER usuage at the
start and the end of a successful snd_pcm_vortex_hw_params().

Issue History
Date Modified  Username       Field                    Change              
======================================================================
04-12-05 20:43 alien999999999 New Issue                                    
04-12-05 20:43 alien999999999 Distribution              => Mandrake        
04-12-05 20:43 alien999999999 Kernel Version            => 2.6.7           
04-12-05 20:48 alien999999999 Note Added: 0004461                          
04-12-05 20:49 alien999999999 File Added: au88x0.c                         
04-12-05 20:49 alien999999999 File Added: au88x0.h                         
04-12-05 20:50 alien999999999 File Added: au88x0_core.c                    
04-12-05 20:50 alien999999999 File Added: au88x0_mixer.c                    
04-12-05 20:51 alien999999999 Note Added: 0004462                          
08-24-05 11:13 Raymond        Note Added: 0005928                          
======================================================================




-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf