Re: CVE request for dnsmasq DoS

[Jamie Strandboge <jamie () canonical com>]

On Thu, 03 Jul 2008, Jamie Strandboge wrote:

> On Tue, 01 Jul 2008, Steven M. Christey wrote:
>
> > I'm not sure I fully understand Thierry Carrez' comment about the security
> > implications of this issue.  It seems like an exploit would require a
> > malicious DHCP server, in which case isn't DHCP service already
> > compromised?  If so, then a crash of dnsmasq (null dereference?) doesn't
> > seem to be any worse than the loss of DHCP itself.
> I haven't had time to develop a PoC, but from the dnsmasq 2.26 announce
> page at [1], a client need only send a crafted renewal request to crash
> the server. Thierry's comments were only for trying to reproduce the
> problem and test the patch.
(resending as the first one didn't make it to the list)

I finally had time to develop a PoC and confirm this on my own. A client
need only send a DHCPREQUEST for an IP address not on the same network
as dnsmasq. Eg:

1. dnsmasq listening on and giving IP addresses for 192.168.122.0/24
2. client requests IP address on another network, such as 192.168.0.1
3. dnsmasq 2.25 (and presumably earlier) crashes

This can happen in normal operation with roaming users, but can also
happen with a malicious request. Attached is a script to easily test for
this (requires python scapy).

Jamie

-- 
Ubuntu Security Engineer     | http://www.ubuntu.com/
Canonical Ltd.               | http://www.canonical.com/

Attachment: dhcp_request.py
Description:

Attachment: signature.asc
Description: Digital signature