Matthias Drochner wrote:
 
> > I merged pcred and ucred and poolified the result.
> 
> It would be really nice if the special values FSCRED and NOCRED
> would go - perhaps replaced by a (more or less) static struct ucred
> which has a special flag in it or so for SECURELEVEL support.
> I recently found that the nfs code blindly dereferences the -1/-2
> (see my recent posting to tech-kern), and I'm sure there are more
> places where this can happen.

OK.  I have your original analysis.  At the very least some
documentation would be useful.

Would it be possible to eliminate these uses, or are they here
to stay, and we should just work with it?
 
	-- Gregory McGarry <g.mcgarry@ieee.org>