I have a chip like this too (TP 560X).  Mine is a Cirrus CS4237.
I'd like to suggest that we always probe the sb last and that
we create a mechanism that allows the SB to be ignored if
some flag is set.
(PS
I believe if I play real hard with the isapnp I can make the
hardware make the sb addresses disappear.  I'd like a simpler
solution -- and I need the wss probe first.)
(PPS
One reason you want to use wss bs sb, is that wss mode gives you
a 20DB gain on the mike input that sb does not.)


Lennart Augustsson <augustss@cs.chalmers.se> writes:

> Is this NetBSD 1.3.2?
> 
> > sb 0 at isa0 port 0x200 - 0x237 irq 7 drq 1: dsp v3.02
> > audio0 at sb0
> > wss0 at isa 0 port 0x530-0x537 irq 10 drq 0 drq2 1isa0: drq 1 is not free
> > panic: isa_dmamap_create
> > Stopped at         _Debugger+0x4: leave
> You have one of those sound cards that responds as both an SB and a WSS.
> If you compile a kernel with one or the other driver (but not both)
> it should work.
> 
>         -- Lennart