Re: Feedback on draft-ietf-nfsv4-01 - stateid - newgrp?

New Message Reply About this list Date view Thread view Subject view Author view Attachment view

From: Neil Brown (neilb@cse.unsw.edu.au)
Date: 10/14/99-10:45:55 PM Z


From: Neil Brown <neilb@cse.unsw.edu.au>
Date: Fri, 15 Oct 1999 13:45:55 +1000 (EST)
Message-ID: <14342.41843.191214.917026@notabene.cse.unsw.EDU.AU>
Subject: Re: Feedback on draft-ietf-nfsv4-01 - stateid - newgrp?

On Thursday October 14, mre@eng.sun.com wrote:
> >     (ASIDE:
> >      actually, this is a bit like chgrp in Unix, which requires you to
> >      own the file and be a member of the group, so you need two
> >      credentials.   AUTH_SYS allows you to send multiple credentials,
> >      pretending that it is one big credential.  How does KerberosV
> >      cope with the possibility of a "newgrp" command? Does it?)
> 
> The Microsoft version of Kerberos V5 allows this by including user id and
> group id information in the ticket from the KDC. (However the method by which
> this information is encoded is, to my knowledge, non-standard, and not
> documented). The KDC thus becomes both an authenticator and authorizer.
> Generic versions of the Kerberos V5 KDC tend to stay away from operating
> system specific concepts like uids and gids. So, if using say Solaris7, with
> the Solaris Kerberos V5 (SEAM 1.0), newgrp won't work for you, because the NFS
> server just gets your principal name, and maps you to uid and gid, and gid
> list that the NFS server determines from the passwd and group databases.
> 
> When you raised your question, I was surprised to find that Solaris supports
> newgrp. I'd assumed for the last 16 years that use newgrp had been
> replaced by the much more sensible multiple group stuff that
> 4.2 BSD introduced. Does Linux support newgrp?
> 
> NFS over AUTH_DES works the same way as NFS over KerberosV5. NFS/AUTH_DES
> is over 10 years old, and I've never come across a customer complaint
> about the newgrp issue. 
> 
> RPCSEC_GSS is flexible enough to support newgrp semantics. It's just
> a small matter of finding a GSS-API mechanism to plug into it. I think
> we should stick with generic stuff, to increase the opportunities for
> interoperability in heterogeneous environments.
> 
> 	-mre

Linux does support newgrp, though it probably isn't used much.  I was
just trying to think of a current API that didn't match the services
provided.

I guess that if the security service used doesn't support something
like newgrp (or user-selected capabilities in general), then there is
nothing NFSv4 can do about it, and if the security service does
support it, then NFSv4 would just treat a different security context
as a different nfs_lockowner.

NeilBrown


New Message Reply About this list Date view Thread view Subject view Author view Attachment view

This archive was generated by hypermail 2.1.2 : 03/04/05-01:47:44 AM Z CST