[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <135187.1298654740@localhost>
Date: Fri, 25 Feb 2011 12:25:40 -0500
From: Valdis.Kletnieks@...edu
To: Vasiliy Kulikov <segoon@...nwall.com>
Cc: "David S. Miller" <davem@...emloft.net>, netdev@...r.kernel.org,
linux-kernel@...r.kernel.org,
Alexey Kuznetsov <kuznet@....inr.ac.ru>,
"Pekka Savola (ipv6)" <pekkas@...core.fi>,
James Morris <jmorris@...ei.org>,
Hideaki YOSHIFUJI <yoshfuji@...ux-ipv6.org>,
Patrick McHardy <kaber@...sh.net>,
Eric Dumazet <eric.dumazet@...il.com>,
Tom Herbert <therbert@...gle.com>,
Changli Gao <xiaosuo@...il.com>, Jesse Gross <jesse@...ira.com>
Subject: Re: [PATCH] don't allow CAP_NET_ADMIN to load non-netdev kernel modules
On Fri, 25 Feb 2011 18:14:14 +0300, Vasiliy Kulikov said:
> Since a8f80e8ff94ecba629542d9b4b5f5a8ee3eb565c any process with
> CAP_NET_ADMIN may load any module from /lib/modules/. This doesn't mean
> that CAP_NET_ADMIN is a superset of CAP_SYS_MODULE as modules are limited
> to /lib/modules/**. However, CAP_NET_ADMIN capability shouldn't allow
> anybody load any module not related to networking.
>
> This patch restricts an ability of autoloading modules to netdev modules
> with explicit aliases. Currently there are only three users of the
> feature: ipip, ip_gre and sit.
And you stop an attacker from simply recompiling the module with a suitable
MODULE_ALIAS line added, how, exactly? This patch may make sense down the
road, but not while it's still trivial for a malicious root user to drop stuff
into /lib/modules.
And if you're going the route "but SELinux/SMACK/Tomoyo will prevent a malicious
root user from doing that", then the obvious reply is "this should be part of those
subsystems rather than something done one-off like this (especially as it has a chance
of breaking legitimate setups that use the current scheme).
Content of type "application/pgp-signature" skipped
Powered by blists - more mailing lists