[PATCH] Fix remaining cases of direct logging of untrusted strings by avc_audit

[Stephen Smalley]

On Tue, 2005-05-24 at 22:16 +0100, David Woodhouse wrote:
> On Tue, 2005-05-24 at 16:11 -0500, Klaus Weidner wrote:
> > > I wouldn't claim to be 100% convinced that it's always nul-terminated,
> > > but we were relying on that already so OK.
> > 
> > Good thing we're not trying to get certified at EAL7, this doesn't quite
> > meet "formally verified design" requirements ;-)
> 
> Heh, yeah. I do remember going through the code at one point for some
> reason any trying to convince myself it was always NUL-terminated. I
> _think_ it is, but I don't 100% recall my conclusion, or even why I was
> checking. Once upon a time it certainly wasn't, but now I think it is.
> 
> I wouldn't have introduced such an assumption without going through and
> checking for myself, but since we were already making that assumption
> I'm prepared to trust it for now.

Other option would be to change audit_log_untrustedstring() to take a
length parameter or provide a variant interface that takes one, and use
that to explicitly pass the length.

-- 
Stephen Smalley
National Security Agency