> In one sense, it's not.  In another, the way dumps are done to disk typically
> involves a relatively short code path...

A relatively short, special cased, code path.  However, it's still not
particularly short, in the case of SCSI including paths through each of:

	scsi disk driver dump code
	scsi middle layer polling-mode code
	scsi hardware driver polling-mode code

right?


> put NFS in there, and suddenly you
> involve the krpc code, udp code, ip code, ethernet drivers (which may play
> tricky mbuf games), buffer cache, etc.

There's no reason a special-case path for NFS couldn't be used, as
well.  of course, to do it, you'd need polling-mode to be implemented
in the ethernet drivers...


Not having NFS dump support is makes life a ... lot harder for people
trying to debug problems on diskless workstations...



cgd