From: Ian Dall <ian@beware.dropbear.id.au>

> Jon Buller <jonb@paclink.com> writes:
> 
> >      Jan  3 07:19:17 bullbox /netbsd: ncr1 at mainbus0 addr 0xffd00000, irq 4
> >      panic: interrupt already allocated
> 
> So why does it think there is more than one ncr device? I wouldn't
> have called your work-around a bug-antibug pair, but then I wouldn't
> call it a fix either! Still, since we will never have more than one ncr device,
> it seems a good case for being pragmatic.

Yes, it's not a fix, since IMHO the probe/attach code is broken
somewhere, and just telling config there is only one ncr to attach
just covers that up.  OTOH, there is only one ncr to attach, so
why have config search for more...  I suspect "The Right Thing To
Do(tm)" is to fix the probe code, then tell config there is only
one, so don't bother looking for more.

I haven't committed the change yet, because if we want to attack
the probe/attach code, this will only give us an excuse to put it
off. 8-)

Jon