Re: OpenSSH Vulns (new?) Priv seperation

From: Valdis.Kletnieks@vt.edu
Date: 06/26/02 

To: wirepair <wirepair@roguemail.net>
From: Valdis.Kletnieks@vt.edu
Date: Wed, 26 Jun 2002 13:07:31 -0400

On Tue, 25 Jun 2002 12:00:54 PDT, wirepair <wirepair@roguemail.net> said:

> "However, with privileges separation turned on, you are
> immune from at least one remote hole."
> at least one? Jesus how many are there? any information
> would be appreciated....

We know there's a known-but-not-widely-disclosed hole, so the statement
is technically correct as it stands - at least one remote hole.

The point they were trying to make (perhaps poorly) was that if you enable
privilege separation, it closes off *entire classes* of attacks - things that
will be stopped because they can't work around the separation. Even if a
second remote exploit is found/disclosed, all it gets the attacker is a
very stripped down chroot'ed running-as-nobody jail cell.

Now of course, it may be possible to mount an attack on the separation
mechanism itself - but that *still* raises the bar considerably to get a full
remote-root compromise. 

-- 
				Valdis Kletnieks
				Computer Systems Senior Engineer
				Virginia Tech