In article <01041619422001.00346@pinky.paradox.demon.co.uk> you write:
>While I'm in this area the stray interrupt code is broken, currently it 
>returns, without pulling the frame off the stack etc, I'm thinking maybe it 
>should panic, but I'm not sure about doing a panic when handling interrupts.  
>Perhaps it should be tidied up so that it can return, and we attempt to 
>handle any more interrupts?

Given the behaviour of a certain one of my podules (it flags an interrupt
for 5us or so, then thinks better of it and denies all knowledge), I'd
prefer it to log a warning and return (like the arm26 one does).

Oh, and you can certainly panic in an interrupt handler.  panic() is
designed to work from anywhere.  log() is similarly useful.

Now, maybe I should take a look at the patch...

-- 
Ben Harris                                                   <bjh21@netbsd.org>
Portmaster, NetBSD/arm26               <URL:http://www.netbsd.org/Ports/arm26/>