On Wed, 2006-05-17 at 11:50 -0400, Stephen Smalley wrote:
> On Tue, 2006-05-16 at 15:30 -0500, George C. Wilson wrote:
> > Yes, you're right.  David Howells' keyring hooks went in.  The LSMs need to 
> > be
> > updated to implement those hooks.  I was under the impression that work was
> > done.  But it wasn't.  The ultimate answer is that somebody needs to do that
> > for SELinux, which I'd like to think should be a fairly small task.  But I
> > thought we had a way to restrict keyring operations to admins w/DAC for the
> > evaluated configuration, which is why it is marked 90%.  If not, we have 
> > hooks
> > to implement.  There would be no policy for them in the evaluated
> > configuration. 
> 
> Implementing a basic set of checks in the current set of hooks would be
> straightforward.  Whether or not such checks would a) prevent
> information flow in violation of policy, and/or b) enable the expected
> usage model for the keys to continue to work in the presence of
> processes with multiple security contexts running in the same user
> session is not at all clear.  That's the harder question.

Trimmed cc list as the redhat-lspp list bounced it due to too many
recipients (odd).

Quick question:  What is using the kernel keyring support?  OpenAFS?  I
don't see anything in Linus' tree using it in-kernel at present.  So why
is it there?  Looks like ecryptfs is using it over in -mm.  If nothing
in the evaluated configuration is using it, can it just be disabled in
the evaluated kernel?

-- 
Stephen Smalley
National Security Agency

--
redhat-lspp mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/redhat-lspp

Reply via email to