[293] in linux-security and linux-alert archive
Re: YAWTCQ
daemon@ATHENA.MIT.EDU (=?ISO-8859-1?Q?Thomas_K=F6nig?=)
Fri Jul 21 08:17:24 1995
To: A.Main@dcs.warwick.ac.uk (Zefram)
Date: Thu, 20 Jul 1995 18:05:13 +0200 (MET DST)
Cc: aleph1@dfw.net, linux-security@tarsier.cv.nrao.edu
In-Reply-To: <20133.199507191752@stone.dcs.warwick.ac.uk> from "Zefram" at Jul 19, 95 06:52:11 pm
From: Thomas.Koenig@ciw.uni-karlsruhe.de (=?ISO-8859-1?Q?Thomas_K=F6nig?=)
> Curiously, at jobs *are* owned by the user
> (otherwise crond wouldn't know who to execute them as),
This also serves as a sort of authenticication, on a system with
restricted chown(), as Linux is, only the user can have created
that file.
The problems which occur when a program written with that assumption
moves into a universe in which this doesn't hold are easy to imagine.
> and it is possible to
> edit them, and this does not pose any serious security
> threat that I am aware of.
This does not hold true for Linux.
It is no longer possible to edit at jobs there in newer versions;
as turned out recently, this was a very wise descision, because there
did indeed lurk a potential fatal security hole there.
> It's even safe on systems where anyone can
> chown their files to anyone, as the at job must have the setuid bit set
> in order to be executed.
You're definitely not speaking of Linux at there :-)
Let's just hope that whoever implemented that particular system
also made the scripts non - executable, in that case.
--
Thomas König, Thomas.Koenig@ciw.uni-karlsruhe.de, ig25@dkauni2.bitnet.
The joy of engineering is to find a straight line on a double
logarithmic diagram.
