[17540] in bugtraq

home help back first fref pref prev next nref lref last post

Re: FreeBSD Security Advisory: FreeBSD-SA-00:62.top [REISSUED]

daemon@ATHENA.MIT.EDU (vort-fu)
Tue Nov 7 13:14:35 2000

Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Message-Id:  <Pine.BSO.4.21.0011071255400.30141-100000@new.wiretapped.net>
Date:         Tue, 7 Nov 2000 13:12:56 +1100
Reply-To: vort-fu <vort@WIRETAPPED.NET>
From: vort-fu <vort@WIRETAPPED.NET>
To: BUGTRAQ@SECURITYFOCUS.COM

> II.  Problem Description
>
> A "format string vulnerability" was discovered in the top(1) utility
> which allows unprivileged local users to cause the top process to
> execute arbitrary code.  The top utility runs with increased
> privileges as a member of the kmem group, which allows it to read from
> kernel memory (but not write to it).  A process with the ability to
> read from kernel memory can monitor privileged data such as network
> traffic, disk buffers and terminal activity, and may be able to
> leverage this to obtain further privileges on the local system or on
> other systems, including root privileges.


From my initial findings, this was not exploitable (not easily if at all),
as the command input is restricted in length, 50 bytes from memory. No
supplied addresses can be accessed nor any that could be used to
easily modify the ret of a call.

>NOTE: The original version of this advisory contained an incomplete
>patch which does not fully eliminate the security vulnerability. The
>additional vulnerability was pointed out by Przemyslaw Frasunek
><venglin@freebsd.lublin.pl>.

It seems that somebody from the freebsd team didnt copy and paste well
enough from my openbsd patch, but the missing segment in the original
fbsd patch was only added for 'completeness' on my behalf.

The error lied in the kill command alone, the renice command was
unaffected, thus the updated patch is not necessarily needed.

Personally I would put this down to a programming error, and not an
(exploitable) vulnerability. Though I admit that I am not an overly
superkraduberleet(tm) exploit coder, and dont pretend to be, but if anyone
can exploit this bug, please forward to the list.

ps. This was sent to the openbsd team, and patched, a month or so ago. How
can the freebsd team justify the lateness in applying their patch
(especially considering that they felt it was exploitable)?

vort-fu
vort@wiretapped.net

home help back first fref pref prev next nref lref last post