Full Disclosure mailing list archives

Re: Sidewinder G2 Firewall


From: Goetz Von Berlichingen <goetzvonberlichingen () comcast net>
Date: Sat, 15 Nov 2003 15:54:00 -0700

Daniel Sichel wrote:
We are in the process of implementing new Sidewinder G2 firewalls.
So far I have not been able to find any record of successful hacks on
these things, so I am pretty happy. The downside is the suckers run sendmail. It is in a jail but still...
Its sendmail. Anybody out there who has substituted Qmail on one of
these? If not, any advise on what stupid things I can avoid while
configuring these. I say these because we are in a high availability
scenario.

Haven't played with these in a while. When we did, they were not our primary targets because there was lower-hanging fruit. No implementation I have seen included the servers on the firewall - they all filtered to a DMZ with servers - so I don't know how it performs as a server. Are you purchasing the appliance version or the software to run on your hardware? Either way, the firewall runs on SecureOS, which is basically a Mandatory Access Control (MAC) version of BSD. The research which eventually led to SecureOS was done by Earl Boebert, et al, back in the early 80s for various Three Letter Agencies (TLAs). SecureOS uses the Honeywell version of Domain Type Enforcement (a standard mechanism of secure OSes). Honeywell added DTE to MULTICS as part of the World-Wide Military Comand and Control System (WWMCCS, pronounced Wimiks). The Honeywell variety (as opposed to the Trusted Information Systems variety) of DTE later became the basis for SecureComputing's SecureOS. My team has successfully attacked DTE (but not in the form of a MAC OS like SecureOS). These systems are only as secure as their role authentication mechanism. The bottom line all comes back to the first principle of cyberwarfe I proposed at the First Annual IEEE SMC Information Assurance Workshop. In all systems, some human or cyber entity has the ability and privilege to perform the action the attacker wants to perform. The attacker needs to assume the identity of that entity. In your case, if the sendmail program is vulnerable, the attacker will be able to do anything that sendmail is able to do. Yes, that limits the attackers' options, but lots of attacks are still available to them. This type of system is more secure than an OS without MAC. Since this is the state-of-the-art in secure operating systems, you are certainly practicing due diligence. Personally, I'd recommend limiting systems to single functions and not running the MTA on the firewall. If you must combine functions, you should be able to run anything compatible with BSD. However, you may have to reconfigure domain access policy to accomodate non-standard software, which at the least is a pain in the ass and at worst could violate warranties and such.

Goetz


_______________________________________________
Full-Disclosure - We believe in it.
Charter: http://lists.netsys.com/full-disclosure-charter.html


Current thread: