> > if nfsd and moutnd talked a bit more, mountd could tell nfsd which
> > directory a host had mounted and nfsd could use vn_isunder() to find
> > out if the access was to a file in the right place, no?
> 
> Possibly. The performance would suck, since the server would have to
> do that for ~every NFS request, or spend a lot of memory caching the
> information. It would probably also break statelessness in the face of
> server reboots if done this way; mountd would be better off just
> passing the access list to the kernel and letting it check/cache
> client information.
> 
> And after all that you'd still be using address-based authentication.

Not to mention that vn_isunder() only works for directories (where you
can easily find the parent directory), not for regular files.  After all,
there might be multiple "parent" directories for regular files, or even
none at all (albeit the latter is being worked aroung by NFS anyway).

Ciao,
Wolfgang
-- 
ws@TooLs.DE     Wolfgang Solfrank, TooLs GmbH 	+49-228-985800