[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Possible bugs and misfeatures



On Wed, 2002-06-19 at 10:50, John Leach wrote:
> Hi Gianni,
> 
> I installed Firestorm from the latest binary rpm
> firestorm-0.4.4-1.i386.rpm  and came up with a few possible bugs and/or
> misfeatures (or features, you never know).
> 
> I'm running firestorm in a chroot in /usr/lib/firestorm, loading snort
> rules from /rules  (i.e: /usr/lib/firestorm/rules) and logging to the
> directory /logs.
> 
> I obtained the latest experimental snort rules list, combined it all
> into one big file and loaded it in.  Firestorm had a few troubles:

FYI: You can include more than one signatures file from firestorm.conf
;)

> The rule:
> alert tcp any any -> [232.0.0.0/8,233.0.0.0/8,239.0.0.0/8] any (msg:"BAD
> TRAFFIC syn to multicast address"; flags:S+; classtype:bad-unknown;
> sid:1431; rev:4;)
> caused:
> snort: /rules/bigass.rules:663 : Rule failed to commit: BAD TRAFFIC syn
> to multicast address

Ah, this is due to the fact that ip lists aren't supported. The latest
code from CVS should fix that.

> The rule:
> alert tcp $EXTERNAL_NET any -> $HOME_NET 111 (msg:"RPC portmap request
> tooltalk"; flow:to_server,established; rpc:100083,*,*;
> reference:cve,CAN-2001-0717; reference:cve,CVE-1999-0003;
> reference:cve,CVE-1999-0687; reference:cve,CAN-1999-1075;
> reference:url,www.cert.org/advisories/CA-2001-05.html;
> classtype:rpc-portmap-decode; sid:1298;  rev:7;)
> 
> caused:
> snort: /rules/bigger.rules:1422 : Rule failed to commit: RPC portmap
> request tooltalk
> 
> In fact, none of the RPC rules I tried worked.  Neither did any of the
> shellcode rules.

The RPC rules will have failed because the RPC matcher isn't supported
yet at all, just don't include rpc.rules.

The shellcode rules, that is a bug, I belive the problem is that you
don't have the $SHELLOCDE_PORTS variable set. The error should be more
descriptive. Again, fixed in CVS.

> I also noticed that firestorm.log gets overwritten each time firestorm
> is run.  I don't think this is intentional.

It is intentional, it is now documented in the manpage. The
firestorm.log doesn't have timestamps so multiple runs logging to the
same file would be a bit confusing anyway. The only purpose for
firestorm.log is for debugging etc... If your firestorm run goes
successfully then you shouldn't really need the log for anything...

> Also, autocreation of a "firestorm" user with a specific uid+guid and a
> small redhat init script would be nice :)

Yeah the RPM setup could be a little more automated, sure.

-- 
// Gianni Tedesco (gianni at ecsc dot co dot uk)
8646BE7D: 6D9F 2287 870E A2C9 8F60 3A3C 91B5 7669 8646 BE7D

Attachment: signature.asc
Description: This is a digitally signed message part