Re: Security issues with local filesystem caching

[Stephen Smalley <sds@tycho.nsa.gov>]

On Wed, 2006-11-01 at 10:58 -0500, Karl MacMillan wrote:
> I suggested change instead of transition because, like most uses of
> change, this was a manual relabel rather than automatic. Transition
> probably makes more sense, though.

Yes, if we need it at all.

> This is all predicated on the notion that there is a need to have the
> normal SELinux checks performed. Since this serves only as a sanity
> check and doesn't add any real security the best option seems bypass,
> but I guess that isn't an option.

"Bypass" in the sense of directly calling the inode ops rather than the
vfs helpers is undesirable.  "Bypass" in the sense of temporarily
setting a task flag indicating that permission checking should be
disabled for an internal access attempt seems ok.

> fssid seems like the wrong name, though it does match the DAC concept.
> This is really more general impersonation of another domain by the
> kernel and might have other uses.

NFS will want a fssid in order to have file access checks applied
against the client process' SID if/when the client process' context
becomes available.  But it isn't really necessary here as far as I can
see; the cachefiles module is not trying to act on behalf of a task, but
instead is performing an internal access to local cache that should
always succeed, and the usual permission checking for userspace is
handled by the fs before cachefiles is called.

-- 
Stephen Smalley
National Security Agency

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/