Re: [tcpm] ICMP attacks draft

At 20:52 18/07/2006, Joe Touch wrote:

> > * The condition for all the so-called hard errors to be treated as soft
> > errors is that the conenction must be in any of the synchronized states.
> > * The reason for not honoring the SHOULD is that in hostile
> > environments, its an attack vector that is easier to perform than
> > others. (I will note the ambiguity in the case of port unreachables)
> > * The good thing of the change is that it mitigates the above threat,
> > and also improves TCP's robustness in the case of multipath scenarios.
>
>I agree with these. It would also be useful to add - to the last one -
>that this *changes* how TCP would behave in multipath scenarios and/or
>path changes, which *may* make TCP less likely to _correctly_ rapidly
>disconnect from a session when a routing change occurs that disconnects
>endpoints.
>
>(i.e., there are cases that this might not be the best choice; trading
>robustness for reactiveness to failure is typical, but not always the
>only choice).

I understand what you mean. However, the message codes to which we 
are changing from hard to soft errors (protocol unreachable and port 
unreachable) are not the ones that would reflect a routing change.

In the case of ICMP frag needed (if TCP does not implement PMTUD), it 
might be an indication of a routing change. Although implementing TCP 
without PMTUD, and at the same time setting the DF bit in the packets 
seems to me, at the very least, very weird.

Or am I missing something?

Kindest regards,

--
Fernando Gont
e-mail: fernando@gont.com.ar || fgont@acm.org
PGP Fingerprint: 7809 84F5 322E 45C7 F1C9 3945 96EE A9EF D076 FFF1

_______________________________________________
tcpm mailing list
tcpm@ietf.org
https://www1.ietf.org/mailman/listinfo/tcpm