Re: [babel] Babel-MAC: Blake2s is 128-bits by default

Random thought: the draft mentions that the MAC algorithm
is configured alongside the key for it. Perhaps we should
add an explicit sentence such as "For MAC algorithms that
support multiple digest sizes (such as BLAKE2s), the digest
size needs to be communicated alongside the MAC algorithm.".

I agree that the "SHOULD implement BLAKE2s" should
mention a key length, but I have no preference between:
- SHOULD implement BLAKE2s with 16-octet hashes
- SHOULD implement BLAKE2s with 32-octet hashes
- SHOULD implement BLAKE2s with both 16 and 32-octet hashes

David

On Fri, Nov 6, 2020 at 3:09 PM Antonin Décimo <antonin.decimo@gmail.com>
wrote:

> > > As Valery Smyslov noted, BLAKE2s is able to produce hashes of any size
> > > between 1 and 32 octets (8 and 256 bits).  However, both
> implementations
> > > of Babel-MAC only ever produce 16-octet BLAKE2s hashes.
> >
> > Okay, something seemed off about this when I read it just now, so I just
> > double checked: And in fact both implementations use 32-byte Blake2s
> > digest sizes...
>
> That also seemed off to me. To be fair regarding babeld, the "old"
> hmac branch used 16-bytes blake2s, and my rewrite uses 32-bytes. We’re
> using the reference blake2s implementation, and I changed 16 to the
> provided BLAKE2S_OUTBYTES = 32 constant.
>
> > I still support making the change and specifying the size, but to
> > correspond to current implementation practice the change would need to
> > be "with 256-bit (32 octet) hashes".
>
> Me too. I an exchange earlier this year, Donald and Juliusz explained
> to me that the size of the output hash is considered part of the
> hashing algorithm, so yes, I support making that explicit for blake2s.
>
>
> -- Antonin
>
> _______________________________________________
> babel mailing list
> babel@ietf.org
> https://www.ietf.org/mailman/listinfo/babel
>