Re: [TLS] Additional Elliptic Curves (Curve25519 etc) for TLS ECDH key agreement

[Dr Stephen Henson <[e-mail]>]

On 12/01/2014 19:21, Salz, Rich wrote:
> I don't want this question to get lost in the general discussion so I am explicitly addressing this to the AD's
> 
>>> I think we should keep this consistency and have the point format for 
>>> the new curves defined in one standard, whatever that might be.
> 
>> Since this format may be used by more than just TLS, and was originally defined outside the IETF, what would
>> be the most appropriate place to discuss extending it?
> 
> Some folks have been working on a concrete proposal, which is that we "take" 0x41 as a new marker byte in ECPoint, and use that to mean "key format defined by the curve"  Ask IANA to maintain a curve point-format registry.  For example, Curve25519 would have 0x41 followed by the 32 bytes of x-value for the key.
> 
> So, where should such a document be started?
> 
> Many of us are feeling pressure to implement ECDHA to get PFS for our customers, and see the efficiency of Curve25519 as very important.  That is, the  clock is ticking...
> 

Based on the current text, perhaps draft-ladd-safecurves (which might get
renamed) being discussed in CFRG? It already has some brief notes on key formats
though not the same as above.

I'm thinking we'd have a section indicating a (recommended, suggested or just
possible) key format and then the TLS, a certificate format document and any
other interested protocols could reference it.

Steve.
-- 
Dr Stephen N. Henson.
Core developer of the   OpenSSL project: http://www.openssl.org/
Freelance consultant see: http://www.drh-consultancy.co.uk/
Email: [e-mail], PGP key: via homepage.