This is sorta what the crit header is for.

You would set some value like "use mld-dsa when present", and then even an
es256 would fail to verify unless the verifier decided to ignore this
critical information.

You could also register a hybrid signature scheme that explicitly allowed
verifying only 1 component, but required both to be present.

crit header seems like a better solution for PQ migration though.

OS

On Mon, Dec 30, 2024, 12:23 PM Michael Richardson <[email protected]>
wrote:

>
> lgl island-resort.com <[email protected]> wrote:
>     > There is RFC 9338 for counter signatures. Your use case is not listed
>     > as one that RFC 9338 addresses. Maybe RFC 9338 can be used if you
> sign
>     > with the strongest signature first, then counter sign with the next
>     > weakest and so on.
>
> So the attacker can remove the weaker ones, unwrapping things essentially.
> It implies a partial ordering on algorithm strength, and the assumption
> that
> the "stronger" (assumed by me quantum-safe) ones are not defeated by
> something else.
>
> It seems to me that an inner attribute is needed to signal what algorithms
> the signer expects to be used.
>
>
> --
> Michael Richardson <[email protected]>   . o O ( IPv6 IøT consulting )
>            Sandelman Software Works Inc, Ottawa and Worldwide
>
>
>
>
>
_______________________________________________
COSE mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to