Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!helios!bcm!rice!uw-beaver!ubc-cs!unixg.ubc.ca!gauguin.ucs.ubc.ca!hilchey
From: hilchey@ucs.ubc.ca (Paul Hilchey)
Newsgroups: comp.sys.sgi
Subject: Re: ELM problem (for the curious).
Message-ID: <1991Mar1.212820.11720@unixg.ubc.ca>
Date: 1 Mar 91 21:28:20 GMT
References: <9103011515.AA17958@centr00.CENTRCN.UMontreal.CA>
Sender: news@unixg.ubc.ca (Usenet News Maintenance)
Reply-To: hilchey@ucs.ubc.ca (Paul Hilchey)
Organization: University of British Columbia
Lines: 26

chouinar@centrcn.umontreal.ca (Luc Chouinard) writes:

> Setgid progams . . . can change the current egid to the current gid
> but they cannot perform the inverse (i.e. egid=gid of the program).

This works fine if you use setgid instead of setegid.  Eg:

	egid = getegid();
	printf("%d\n",getegid());
	setgid(getgid());
	printf("%d\n",getegid());
	setgid(egid);
	printf("%d\n",getegid());

produces the expected result with Irix 3.3.1.  It doesn't work with
SunOS 4.1.1, though.

A related question: Elm changes the effective uid/gid back to the real
ones before calling the access function, though on all the systems I've
checked, access is done w.r.t. the real uid/gid.  Are there any systems
on which that isn't true?

____
Paul Hilchey
University Computing Services
The University of British Columbia