Re: R. Thomas's NFS question

[il1 () dsroc6 dsdoe ornl gov (Dave Dillow)]

From: PETER.T.WHITING () sprint sprint com
Subject:  R. Thomas's NFS question
Sender: owner-bugtraq () fc net
Content-Length: 907
    R. Thomas> hostA --> exports /usr/share to -access=hostB 
    R. Thomas> hostB --> a linux box.  re-exports /usr/share to everyone
    R. Thamas> hostC --> not implicitly trusted by hostA, mounts /usr/share

    R. Thomas> aside from any security concerns, this would certainly
    R. Thomas> thrash your nfsd's.  does anyone have any experience
    R. Thomas> with this?  i have only recently discovered this, and
    R. Thomas> have not had time to peruse it in depth.

Not a problem.  Host C gets to look at ***HostB's*** /usr/share - the
one that has HostA's /usr/share mounted over it, not HostA's
/usr/share.  NFS gives you a single hop.  In the above example HostA
could then mount (if perms were granted) HostC's /usr/share and
everything would work.

pete

Peter T. Whiting
[snip]
End of excerpt....

After reading the man page for nfsd on a Linux box, I have to agree with
Mr. Thomas. The man page even makes reference to using this feature to 
function as a NFS multiplier, whatever that may be.... Anyway, the security
concern raised by Mr. Thomas is valid. Mr. Whiting is correct in that most
of the nfsd's I know about do not behave this way, and I belive Linux's can,
by simply not specifing '-r' on the command line. I have not tried this, and
I may be wrong, so it is worth a check yourself.

Dave Dillow
il1 () ornl gov
dillow () cs utk edu