Re: uid 0 (was Re: CVS commit: src/sys/kern)

[Antti Kantee <pooka%cs.hut.fi@localhost>]

On Tue Mar 17 2009 at 17:31:19 +0100, Quentin Garnier wrote:
> > > >Is it correct to use uid == 0 as a test outside of the secmodel?
> > > 
> > > Probably not, but I just made the code look exactly like the fork()
> > > case.
> > 
> > Doesn't kauth(KAUTH_GENERIC_ISSUSER) do the right thing?  Yes, I agree,
> > it wastes much non-ansified non-whitespace in the source tree, but things
> > like gaols would work better if we don't propagate code which we know
> > to be incorrect.
> 
> KAUTH_GENERIC_ISSUSER was a temporary kludge for the kauth move.  It's
> slightly better than !uid, but not by much.

Sure, completely agree, but it's what we're using now.  I don't really
see it going away except in fantasy, though, since we need an effective
concept of superuser for unix.