Great Circle Associates Firewalls
(February 1998)
 

Indexed By Date: [Previous] [Next] Indexed By Thread: [Previous] [Next]

Subject: RE: FW-1 and FIN scanning (was: nmap tool)
From: Robert Ståhlbrand <robert . stahlbrand @ nmac . ericsson . se>
Date: Wed, 4 Feb 1998 12:21:38 +0100
To: "'Steve Birnbaum'" <sbirn @ security . org . il>, "'Marek Kubita'" <marek @ corpus . cz>
Cc: "'firewalls @ greatcircle . com'" <firewalls @ greatcircle . com>


> -----Original Message-----
> From:	Steve Birnbaum [SMTP:sbirn @
 security .
 org .
 il]
> Sent:	den 4 februari 1998 11:03
> To:	Robert Ståhlbrand
> Cc:	'Marek Kubita'; 'firewalls @
 greatcircle .
 com'
> Subject:	Re: FW-1 and FIN scanning (was: nmap tool) 
> 
> 
> 
> robert .
 stahlbrand @
 nmac .
 ericsson .
 se said:
> > And when I think I don't understand why the connection would close.
> I
> > can unplug my wire for an hour and when I attach it again my
> > telnet-session (for example) is not closed (unless you hit a key
> while
> 
> Unlike a packet filter where you generally allow all established
> traffic,
> Firewall-1 keeps a state table of open connections and dynamically
> allows 
> only the required established traffic.  My understanding is that 
> re-installing the policy clears the state table.  Therefore, you'd
> have to 
> send out a new SYN in order for it to add a new entry to the state
> table 
> if it didn't have some way to dynamically re-add running sessions.
> 
> [Robert Ståhlbrand]  
> That is one of the strength with Checkpoint 8-) (not allowing
> established packets as a general).
> What you say is that Checkpoint caches all the current connections and
> when you clear the state table reestablish the connections. Is this
> done with a new SYN-packet? Not very likely is it? How would you do
> that? Shall the firewall send out the SYN-packet to the
> destination-address with the initiators ip-address. That is spoofing
> and how should the initiator know that the firewall has sent out this
> packet and accept the SYN,ACK he recieves back? This won't work and
> you cannot force the initiator to resend a SYN-packet without loosing
> the current connection (or force him at all).
> If think this is done with a cache with all current connections. When
> you clear the table (installing a policy) he just puts this cache
> somewhere and after it has been installed lifting the cache back in
> the system. Why should you put in more effort?
> 
> If I'm wrong here please describe the progress!
> 
> I still don't see the connection between dropping packets and
> reestablishing connections and why only with dropping packets rules
> you can pass a not established connection???
> 
> 
> However, I agree that this behaviour is not well known and I would
> personally rather see sessions lost than allow the ability to scan the
> network.  The least that could be done is to make this an option,
> disabled
> by default, with a nice red friendly warning that pops up when you try
> to
> enable it.
> 
> I am, however concerned about your statement that data is passing
> through.
> Hopefully this behaviour is unique to pre-version 3.
> [Robert Ståhlbrand]  
> 
> We will recieve an answer on that pretty soon, I assume from Marek
> Kubita?
> 
>   Steve
> 
> -- 
> sbirn @
 security .
 org .
 il Phone: +972-2-6795860    (PGP key available)
> Fight Internet Spam!  http://www.vix.com/spam/  Disclaimer: My
> opinions only.
> 
>  << File: ATT00173.ATT >> 
> [Robert Ståhlbrand]  
> 
> /Robert Ståhlbrand, Ericsson Telecom AB


Follow-Ups:
Indexed By Date Previous: RE: FW-1 and FIN scanning (was: nmap tool)
From: Robert Ståhlbrand <robert . stahlbrand @ nmac . ericsson . se>
Next: Re: Firewalls-Digest V7 #53
From: Stefan Schmidt <schmidts @ informatik . tu-muenchen . de>
Indexed By Thread Previous: Re: FW-1 and FIN scanning (was: nmap tool)
From: Marek Kubita <marek @ corpus . cz>
Next: Re: FW-1 and FIN scanning (was: nmap tool)
From: Steve Birnbaum <sbirn @ security . org . il>

Google
 
Search Internet Search www.greatcircle.com