On Thu, 13 Apr 2006 20:04:24 -0700 Ben Collver wrote:

[snip]
> I reproduced the problem in NetBSD 2.0, and the backtrace looks pretty
> much the same as the 3.0 backtrace.

Oh, thank you very much for testing this!

>  Since this release pre-dates Tamura
> Kent's changes, I am guessing he is not responsible for the problem.

Yes, it is. We have also built the kernels just before and after
kent-audio1 merging and problem was obscured on both too. So mentioned
changes definitively not related here.

> The error message I get is:
> 
> isa0: dma start on free channel 5
> panic: _isa_dmastart
> Stopped in pid 526.1 (audioplay) at   netbsd:cpu_Debugger+0x4:   leave
> db>

So we should investigate it in usual way :-) Could you support my very
limited kernel knowledge by recompiling kernel with following options
and posting resulting kernel messages, panic output, and backtrace here:

options DIAGNOSTIC
options DEBUG
options ISADMA_DEBUG
options DEBUG_ISAPNP
options AUDIO_DEBUG
options GUSPLAYDEBUG

Thanks in advance.

--
Mishka.

P.S. I've asked to add this PR to the 4.0 showstoppers list.

P.P.S. Any help from competent people would be appreciated :-)