So I successfully fdisk'd two partitions, a 10-meg dos-fat and the rest of
the drive a type-169 netbsd partition.  Then, I successfully disklabel'd
the drive, with the dos partition at "i", and netbsd at "a" through "h",
including "c" for the whole netbsd parition, and "d" for the whole disk.
Everything fine so far.  All the netbsd slices are cylinder aligned.

Now, when I try to do "newfs /dev/rsd0a", I get...

/dev/rsd0a:     263640 sectors in 169 cylinders of 5 tracks, 312 sectors        
        128.7MB in 11 cyl groups (16 c/g, 12.19MB/g, 2944 i/g)                  
super-block backups (for fsck -b #) at:                                         
     32,  25312,

Then it gets stuck.  I can still telnet in to the Magnum, but I can't kill
the newfs process, even SIGKILL won't work.  And I can't ctrl-C or ctrl-\
out of it on the console.  "top" shows that newfs is using 0% cpu.  It's
just "stuck".  I've tried both with and without the drive jumpered for
termination, and both with and without a passive terminator on the back
of the chassis.  Same deal.  The scsi drive is properly probed on boot-up.
Both fdisk and disklabel worked without a hitch.  And this drive is the ONLY
scsi device on the bus.  It's connected to cable "D0" off of the internal
scsi chain.  

Is it possible that something is in disagreement in the scsipi MI code
between the 1.5R kernel and 1.5-release userland?  Or is it possible that
something is wrong with the "asc" driver?

Anyone?  Anyone?  Bueller?

Cheers,
-Mark