>Yes.  You can also mount Ultrix partitions readonly under NetBSD, and
>you can also use the Ultrix swap partition as additional swap space
>under NetBSD.  I've been running my 5000 this way since I set it up
>early in 1997.

Errr...  It's like, a two- or three- line change to mount the
partition read/*write*.

This was found to be dangerous (causes filesystem corruption and
possibly data loss) on AMIX which also uses a 4.2BSD vintage FFS.
It's regularly used on SunOs, which uses a slightly newer variant of
Kirk McKusick's FFS.  Michael Hitch might just possibly have more
accurate info on AMIX experiences than I do.


Off-hand, I can't see how it could be dangerous if you pre-tune the
FFS system to have less than 8 rotational positions. (Any more than
that and the rotiational-optimization table runs into the space that
was redefined for Other Things in 4.3BSD.)

I can send you the patch, if I still have it, but Caveat Emptor, Back
Up Your Filesystems, etc., etc.  

Last I looked, there is still a bug in our fsck that can cause damage
under this scenario.  I suspected at the time that *that* was the real
problem, not the FFS code per se.  I ran into it myself while
bootstrapping the NetBSD kernel and mounting Ultrix partitions
read-write for lack of space.  The mounts worked fine, the fsck would
smash *all* the symlinks and garbled the allocation bitmap for the
sectors formerly holding the symlink pathname.

I'm still pretty sure the fix I proposed for fsck two or three years
ago is correct, and may even fix the rumoured AMIX lossage, but for
some reason nobody ever explained to me, it got vetoed.

Unless the fsck fix is already committed and I missed it, that is... I
havent kept up-to-date with it.  There's an easy way to see: try
fsck'ing an Ultrix FFS filesystem that has symlinks.

But be __sure__ to back it up first :).