vchans seem to affect /dev/dsp* availability

[Cliff L. Biffle]

On Sunday 27 April 2003 05:30 pm, Cliff L. Biffle wrote:
> When I have vchans enabled, every few attempts at opening /dev/dsp* result
> in 'Operation not supported by device.'  Also, while they're enabled, some
> programs (most reproduceably artsd) peg my CPU after outputting sound data
> in certain situations.

I've had a few people report that they get this issue even with vchans 
enabled, and I can confirm this.

After a kernel rebuild with some debugging switches, I'm now getting a lock 
order reversal message that seems to coincide with the dsp breakage.

Here's the messages, if anyone can make sense of 'em. :-)

-Cliff L. Biffle

---

lock order reversal
 1st 0xc1dea840 pcm0:play:0 (pcm channel) @ 
/usr/src/sys/dev/sound/pcm/sound.c:191
 2nd 0xc1dea9c0 pcm0 (sound cdev) @ /usr/src/sys/dev/sound/pcm/sound.c:163
Stack backtrace:

lock order reversal
 1st 0xc1cbb9c0 pcm0:play:0 (pcm channel) @ 
/usr/src/sys/dev/sound/pcm/sound.c:191
 2nd 0xc21aab40 pcm0 (sound cdev) @ /usr/src/sys/dev/sound/pcm/sound.c:163
Stack backtrace: