Re: [core] draft-ietf-core-oscore-groupcomm: Question signature length

[Marco Tiloca <[e-mail]>]

Hello Jim,

Thanks for your comment!

We have considered it and plan the following updates:

1) In Section 2, the Common Security Context can include a further
optional parameter "Counter Signature Curve", with value from the COSE
Elliptic Curves registry. This parameter indicates the curve to use for
the algorithm specified in the "Counter Signature Algorithm" parameter.

2) In Section 3, the "algorithms" array of the external_aad would be
further extended as:

algorithms : [alg_aead : int / tstr , alg_countersign : int / tstr ,
?crv_countersign : int / tstr]

where "crv_countersign" contains the Counter Signature Curve from the
Common Security Context.


Best,
/Marco

On 1/10/19 4:44 AM, Jim Schaad wrote:
> I want to verify that you understand the following is true.
>
> It is not possible based on just the content of the message to determine
> where the encrypted body ends and the signature begins.  
>
> To see why this is the case, if the signature algorithm in the context is
> EdDSA then it could be either signed with the x25519 curve or the x448
> curve.  The same thing is true with ECDSA signatures as the value ES256
> implies that the hash used is SHA-256 but it does not require that the curve
> used to sign is P-256.  It can be any of the curves that would be valid for
> this hash function.
>
> It might be nicer if there was a signature length encoded someplace so that
> the two values can easily be separated.
>
> 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