Re: [core] Missing must in the Group OSCORE document

[Marco Tiloca <[e-mail]>]

Hi Jim,

On 2020-08-01 21:25, Jim Schaad wrote:
> Christian,
>
>  I have been thinking about the problem case of having a duplicate IV reuse
> in the case where I suggested that we use separate IV spaces for the group
> and pairwise keying materials.  I agree that this is a problem, however the
> problem is greater than what you outlined.  This is going to be a situation
> that will arise anytime that the request comes in under one security context
> and the response goes out under a different security context.  In this
> situation you will always have the problem that a reflected IV value from
> context 1 will lead to a potential IV reuse in context 2.
>
> Missing requirement in the document:
>
> A server MUST use a PIV value from it's own sender context when ever it
> would normally use a reflected IV, but the security context for the request
> and response are not the same.

==>MT
I guess here "different security context" means "different protection
mode" - hence different PIV space - regardless a possible group rekeying
happening. Correct?

This seems to confirm the issue (1) raised in slide 9 of [1]. So this
MUST-requirement would be needed if the two different PIV spaces are
introduced.

Just to clarify, do you think that the requirement is needed also in the
current document with only one PIV space?

Thanks,
/Marco

[1]
https://www.ietf.org/proceedings/108/slides/slides-108-core-sessa-group-oscore-00
<==

>
> Jim
>
>

-- 
Marco Tiloca
Ph.D., Senior Researcher

RISE Research Institutes of Sweden
Division ICT
Isafjordsgatan 22 / Kistagången 16
SE-164 40 Kista (Sweden)

Phone: [phone]
https://www.ri.se