Re: [Hipsec] A question on the replay protection in HIP-base 08.draft

[Pekka Nikander <pekka.nikander at nomadiclab.com>]

AFAICS, your understanding is basically correct.

However, IIRC, another consideration we had in mind was very limited  
hosts.  It may be desirable to implement HIP on hosts where it is too  
costly to store the generation counter for all/some hosts.   
Furthermore, if you don't store the generation counter, you harm only  
yourself: you will become vulnerable to certain replay attacks, but  
that doesn't harm the Internet in the large.  So, in the spirit of  
doing no harm, it seemed safest to allow hosts to pick their own  
policy depending on their resources etc.

--Pekka

On 31 Jul 2007, at 11:26, JiangXingFeng wrote:

> Hi:
> In the section 4.1.4 of draft-ietf-hip-base-08, it says that "
> Implementations MUST accept puzzles from the current generation and  
> MAY
> accept puzzles from earlier generations'.
>
> The generation counter is designed to protect replay attack in HIP.  
> The
> initiator should not accept R1 with earlier counter, but HIP uses IP
> protocol to transport message and lack of reliability, so that the R1
> message MAY duplicate or out of order. From the above reason, the  
> draft say
> that "initiator MAY accept puzzles from earlier generations". Is my
> understanding right? If it is not, what is the reason for that  
> statement?
>
> Regards!
> --
> Jiang XingFeng
>
>
>
>
> _______________________________________________
> Hipsec mailing list
> Hipsec at lists.ietf.org
> https://www1.ietf.org/mailman/listinfo/hipsec


_______________________________________________
Hipsec mailing list
Hipsec at lists.ietf.org
https://www1.ietf.org/mailman/listinfo/hipsec