[ALSA - driver 0000589]: ICH5 S/PDIF audio drops out occasionally

[bugtrack@alsa-project.org]

A NOTE has been added to this issue.
======================================================================
<https://bugtrack.alsa-project.org/alsa-bug/view.php?id=589> 
======================================================================
Reported By:                squarooticus
Assigned To:                
======================================================================
Project:                    ALSA - driver
Issue ID:                   589
Category:                   PCI - intel8x0
Reproducibility:            always
Severity:                   minor
Priority:                   normal
Status:                     new
Distribution:               Debian Unstable (sid)
Kernel Version:             2.6.7
======================================================================
Date Submitted:             10-21-2004 02:42 CEST
Last Modified:              11-16-2004 05:14 CET
======================================================================
Summary:                    ICH5 S/PDIF audio drops out occasionally
Description: 
I'm using the on-board audio of my Abit IC7-MAX3 ("Intel Corp. 82801EB/ER
(ICH5/ICH5R) AC'97 Audio Controller (rev 02)") with ALSA driver
snd_intel8x0 for my home theater setup.  In every application I've tried,
which includes mythtv, xine, and ogg123, through both native alsa and OSS
emulation, the audio output from the S/PDIF port drops out for a second
every so often---maybe once a minute, or every few minutes; it's not
periodic.

I don't know whether analog output also drops out, since I don't have
analog output hooked up or enabled.

This is with the CVS build as of October 19.  My asound.conf is:

pcm.krose-dmix {
  type dmix
  ipc_key 3516
  ipc_key_add_uid yes
  slave {
        pcm "hw:0,4"
        period_time 0
        period_size 4096
        #buffer_size 16384
        buffer_size 131072
        period_time 84000
        buffer_time 340000
        rate 48000
        channels 2
  }
  bindings.0 0
  bindings.1 1
}

pcm.dsp0 {
  type plug
  slave.pcm "krose-dmix"
}

ctl.mixer0 {
 type hw
 card 0
}

pcm.!default {
 type plug
 slave.pcm "krose-dmix"
}

I can't find any options to compile in debug logging: I'm not really sure
what the cause of the drop out is, e.g., buffer underrun, PCI contention,
etc.  I'm happy to spend time to try to debug this, but I need a starting
point.

Of course, it'd also be great to know if anyone has ever seen this kind of
behavior before.  I don't believe the SPD/IF port is turning itself off
because the receiver doesn't blip, which normally it would if the input
were to disappear temporarily.

The CPU is a 3.2G P4 with hyperthreading turned on.  I never see the CPU
usage go above 50%.

Interestingly, I see similar behavior on my dual Opteron machine with an
AMD-8111 audio controller using the same driver (snd_intel8x0).  I never
had this problem with the emu10k1 driver when I was using SB Lives, but I
have run into PCI slot limitations in my HTPC, so I have to use mobo
audio.

Suggestions?
======================================================================

----------------------------------------------------------------------
 squarooticus - 11-16-04 02:50 
----------------------------------------------------------------------
Incidentally, I compiled a low-latency kernel (Ingo Molnar's
realtime-preempt patches) to see if this would solve the problem, and it
doesn't appear to have.  It's really unclear what's responsible for this
behavior at this point, because I'd figure a low-latency kernel would at
least make the problem much less likely were buffer underrun really the
issue.  I'd really like to be able to determine for sure whether underrun
has occurred or not.

----------------------------------------------------------------------
 squarooticus - 11-16-04 05:14 
----------------------------------------------------------------------
More info: this problem occurs with another digital receiver, so the
problem is either with (a) the software, (b) the mobo hardware, or (c) the
cable.  I will try to eliminate (c) this week. :)  FWIW, the drop-out is
much shorter on this receiver (Yamaha HTR-5650) than on the original one
(Marantz SR-7000).  It's unclear to me whether this means the S/PDIF
signal is being interrupted, or if an underrun of the AC3 data is enough
to cause this effect.

Issue History
Date Modified  Username       Field                    Change              
======================================================================
10-21-04 02:42 squarooticus   New Issue                                    
10-21-04 02:42 squarooticus   Distribution              => Debian Unstable (sid)
10-21-04 02:42 squarooticus   Kernel Version            => 2.6.7           
10-21-04 23:40 squarooticus   Issue Monitored: squarooticus                    
10-27-04 16:54 squarooticus   Note Added: 0002275                          
11-10-04 12:26 tiwai          Note Added: 0002367                          
11-12-04 06:23 squarooticus   Note Added: 0002392                          
11-15-04 20:24 squarooticus   Note Added: 0002419                          
11-16-04 02:50 squarooticus   Note Added: 0002423                          
11-16-04 05:14 squarooticus   Note Added: 0002424                          
======================================================================




-------------------------------------------------------
This SF.Net email is sponsored by: InterSystems CACHE
FREE OODBMS DOWNLOAD - A multidimensional database that combines
robust object and relational technologies, making it a perfect match
for Java, C++,COM, XML, ODBC and JDBC. www.intersystems.com/match8