On 2010-09-10 05:51, Mikael Abrahamsson wrote:
> On Thu, 9 Sep 2010, Fernando Gont wrote:
> 
>> Mikael Abrahamsson wrote:
>>
>>> Last I checked, the standards said that if precedence/dscp changed, the
>>> host should reset the session (correct me if I'm wrong, I don't really
>>> have time to check it right now).
>>
>> You're right. And it doesn't make sense. See Section 11.1.4 of
>> http://www.gont.com.ar/papers/tn-03-09-security-assessment-TCP.pdf
> 
> Yes, it doesn't make sense, and luckily nobody (as far as I've been able
> to discern) actually follows this part. So even though it might still be
> in the standard, it's not a real-world problem.

I suspect it was intended to prevent "misuse" of high IP Precedence during
TCP SYN/ACK exchanges, or something like that. But it really does make
no sense.

> So whatever happens, please make the flow label something that can be
> changed without any adverse reaction from the end system.

I agree, and that's behind my suggestion earlier in this thread that
"downstream nodes MUST NOT rely on the value being unchanged".

   Brian

--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to