> I think I must have sounded more authoritative that I intended.  Looking
> again at the code confirms what Allen says (and it looks to me like, in
> this case, there is _no_ 'a' partition assigned at all!). 

Well, I've done some hacking on the code today, but I've not tested it
to see if it works any better than the current code (it does appear to
work as well as the current code ;-).  It should now scan the first 32
partitions instead of the first 8 for its root, swap and usr partitions.
The rest of the 8 will be assigned first-come/first-serve after those.

Note that this may alter the assignments of some partitions.

I will soon put up a 'puma GENERIC #1' kernel for folks to try, if
interested.  It will also correct the problems that some people had
with the netbsd.please.test kernel generating illegal instruction traps
on some systems (thanks to those who pointed out the innocuous change
that I'd made to cause those--I still don't know what was wrong with it).

-allen

-- 
Allen Briggs - end killing - allen.briggs@bev.net ** MacBSD == NetBSD/mac68k **
   Where does all my time go?  <a href="http://www.netbsd.org/">Guess.</a>