On Fri, 27 Sep 2002, Jeremy Hall wrote:
> Hello all,
>
> I am using two RME9652 cards together as a single pcm_multi device so that
> I can get 24 channels both at 48 and 96khz without a cable reconfig. I
> have an SMP machine and have noticed that when using the pcm_multi device,
> all the channels are mapped correctly and seem to work correctly until I
> use jack and ardour. If I use a single card with jack and ardour the
> sessions can start nicely, but if I use the bonded (YICK) pcm_multi
> setting, the machine will lock up and the card will endlessly replay some
> material at a rapid rate causing the famous hammerfall hum that is higher
> in pitch the smaller the frames_per_cycle.
>
> After consulting Paul Davis, the conclusion is that jack exercizes the
> alsa drivers in unique and interesting ways, exposing subtlties. I have
> never been able to get 96khz to work at any frames_per_cycle setting,
> although I can get 48khz to work if I use 1024 or greater
> frames_per_cycle. This is unacceptable to me as I require smaller latency
> and would expect at least 512 or 256 to work given that I can get a single
> card to work reliably at 64 frames_per_cycle.
>
> The problem seems to occur when XRUNS are detected rapidly such as session
> initialization or if the CPU can't keep up with some real-time CPU
> constraints. I haven't tried this with a UP kernel because I don't know
> what would happen if I tried to boot a SMP box on a UP kernel--I recall
> texts saying this was impossible, although this was years ago in the
> beginning of SMP development.
Try specify 'nosmp' with SMP kernel at lilo prompt.
Anyway, the lock up should be avoided. Perhaps, you may try to enable
spinlock debugging in your kernel (CONFIG_DEBUG_SPINLOCK in kernel
debugging menu) and memory allocation (CONFIG_DEBUG_SLAB) for sure.
If it does not help to identify the problem, other reasons might be
possible:
- wrong PCI configuration (PCI bus will cause dead-lock)
- IRQ storm
Jaroslav
-----
Jaroslav Kysela <[EMAIL PROTECTED]>
Linux Kernel Sound Maintainer
ALSA Project http://www.alsa-project.org
SuSE Linux http://www.suse.com
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Alsa-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/alsa-devel
