5.2-RELEASE TODO

[Robert Watson]

On Fri, 16 May 2003, Ilmar S. Habibulin wrote:

> On Thu, 15 May 2003, Robert Watson wrote:
> 
> >   Desired Features for 5.2-RELEASE
> >
> >    +------------------------------------------------------------------------+
> >    |    Issue    | Status |  Responsible  |           Description           |
> >    |-------------+--------+---------------+---------------------------------|
> >    |             |        |               | Currently, MAC protections are  |
> >    |             |        |               | enforced only on locally        |
> >    |             |        |               | originated file system          |
> >    |             |        |               | operations (VOPs), and not on   |
> >    |             |        |               | RPCs generated via the NFS      |
> >    | MAC support |        |               | server. Improvements in NFS     |
> >    | for NFS     | --     | Robert Watson | server credential handling are  |
> >    | Server      |        |               | required to correct this        |
> >    |             |        |               | problem, as well as the         |
> >    |             |        |               | introduction of new entry       |
> >    |             |        |               | points to properly label NFS    |
> >    |             |        |               | credentials and perform         |
> >    |             |        |               | enforcement properly.           |
> >    |-------------+--------+---------------+---------------------------------|
> 
> Do you plan to transfer labels over NFS? If so, why not to make some
> generic extended attributes transfer mechanism over NFS and use it for
> ACL too? 

Right now, no mediation is performed on requests received via NFS RPC's,
so the first bit of the work is to provide for that mediation based on
current credentials.  The second bit is to find useful sources of label
data for those credentials -- that might include host mapping tables,
additional NFS RPCs, packet labels, etc.  The mediation element falls out
easily once you have some of the NFS credential fixes we have in the MAC
tree, and shouldn't be hard.  Then the question of where the labels come
from and how to integrate the policies are raised, and I think some
experimentation is called for.

> And what about packet labeling - it's still desired feature for
> trustedbsd/6.0-current? 

Yes.  It could be we get it in for 5.2, though.  Both improvements in
IPSEC integration, and CIPSO are highly desirable features. 

Robert N M Watson             FreeBSD Core Team, TrustedBSD Projects
robert at fledge.watson.org      Network Associates Laboratories