Re: [TLS] TLS process thread

[Brian Sniffen <[e-mail]>]

Watson Ladd <[e-mail]> writes:

> On Sun, Apr 13, 2014 at 8:31 PM, Peter Gutmann
> <[e-mail]> wrote:
>> Sean Turner <[e-mail]> writes:
>>
>>>In this case, however, our charter provides a clear starting point, namely
>>>RFC 5246, and a mandate to minimize the changes to that document.
>>
>> If the charter really does require that then it sounds like it's high time to
>> update the charter...
>
> That's going to be a pretty big change from RFC 5246. Trying to keep
> the textual changes minimal probably a lost cause at this point, and
> will hurt, rather than help clarity. This is before we try to reduce
> round-trips, ...

[...]

> I don't see a way to do that and adhere to "minimize changes".

The most complicating factor I can imagine is multiple regular protocol
roles, any of which could be run in parallel: three different ways a
client could behave and three different corresponding server behaviors.
They can even run in parallel, given an anycasted server!

The most important simplification I can imagine is to make the handshake
implementation crystal clear and dead simple, up through an
authenticated and encrypted channel: amenable to formal methods, to
automated search for differences from a verified implementation, and to
high-performance implementation without much risk.  I'd rather give up
fast session resumption for half a decade than continue with a fire-drill
of ~annual protocol vulnerabilities.

-Brian
"Implementation is hard" can be my motto for the week.

-- 
Brian Sniffen
Information Security
Akamai Technologies