Hello.  I just tried the GENERIC kernel from the 20071208 snapshot.
The results are the same as for NetBSD-4.  Interrupts stop and the audio
and ath(4) devices stop working until the audio-using application is
killed.  So, what ever is wrong, it still exists in -current.

I'll keep poking, but any ideas would be greatly appreciated.
-thanks
-Brian
On Dec 8, 10:24pm, Brian Buhrow wrote:
} Subject: Re: Problems with ath(4) and interrupt sharing
} 	Hello.  I've been trying various fixes all afternoon in an effort to
} get the ath(4) driver to play well with the auich(4) driver using the same
} interrupt.  It seems that I have the same problem I described earlier in
} this thread even if I arrange things so that the ath(4) driver never
} acknowledges the interrupts.
} 	This leads me to wonder if sharing interrupts with the auich(4) driver
} works at all under NetBSD-4.  In looking at the audio system, it appears as
} though there's no sort of watchdog recovery mechanism in the event that a
} piece of hardware misses an interrupt.  Sharing interrupts on NetBSD-3, at
} least if the two devices are at the same interrupt priority, works fine.
} 	Does anyone hav any ideas on what I might try from here?  
} 
} -thanks
} -Brian
} 
} 	
>-- End of excerpt from Brian Buhrow