Is there a reason that dump(8) looks in the mount table instead of the
fstab when it's matching devices to directories?  Is it really that
common that you'll want to dump a subdirectory which matches an unmounted
device?

I'm too used to sunisms, I guess.  I like being able to type:

	# dump 0bsdf 63 6000 54000 /usr

and have it deduce that /usr is really associated with an unmounted
device instead of trying to do a partial dump of the /usr directory
under the root filesystem.

Comments?  I thought this was what my first patch did when I submitted
it quite a few years back.  If it isn't what it did, I will submit a
corrected patch soon; otherwise, I'm interested in why it changed.

Tangentially, I'm also interested in what the logic was behind
using the -A flag to do what most umount programs do with -a.  If the
logic is not sound, I would like to propose that the meanings of -a/-A
be swapped in umount(8).

				--*greywolf;
--
BSD: No Sh;t!