Re: [Softwires] Internal routing based on the V octet can be easy

[Rémi Després <[e-mail]>]

Le 10 oct. 2011 à 10:58, Ole Troan a écrit :

>>> The 4rd function examines ALL packets that reach the CE.
>>> It processes all those that have V in their destination addresses, and only those.
>>> 
>> This is a big change to the current forwarding behavior of CPE devices, doable, but quite different from, for example, some encapsulation/decapsulation implementations.
> 
> +1.
> doable in theory, but I share your concerns.

Well, it wasn't quite right to say, as I did, "the 4rd function examines ALL packets that reach the CE".
More exactly, with existing kernel architectures, the 4rd function should only see packets that have the V octet in their IPv6 destinations.

With Linux, it is feasible AFAIK as follows:
- Before submitting a received packet to the routing function, it is submitted to the Netfilter pre-routing function. 
- Netfilter adds to the packet a mark, specific of 4rd, if it has the V  octet (mask + value).
- The routing function, duly configured, then sends packets that have this mark to the 4rd-function interface. 
With BSD, the PF packet filter has a role similar to that of Netfilter.
(Thanks to Francis Dupont for providing this information, hopefully not damaged by my transcription.)  

Useful capabilities of 4rd solutions SHOUDN'T therefore be abandoned for the unjustified fear that internal routing based on the V octet might be difficult.

Regards,
RD