[noise] Reworking PSK usage

[Jason A. Donenfeld]

Hi Trevor,

On Tue, May 2, 2017 at 10:50 PM, Trevor Perrin <trevp at trevp.net> wrote:
> On Tue, May 2, 2017 at 7:56 PM, Jason A. Donenfeld <Jason at zx2c4.com> wrote:
>> Actually, my suggestion and what I placed in that pull request is:
>>
>> MixKeyAndHash(psk):
>>   ck, temp, k = HKDF(ck, psk)
>>   MixHash(temp)
>>
>> The reason is that there are cases in which k is overwritten again without
>> being used.
>
> That makes sense - h is always used after every token, but sometimes k isn't.

Great, okay. So this is what we shall use then.

Since this is a pretty late-stage change for WireGuard, I'm anxious to
transition to this ASAP, since implementations are propagating. Have
any rough ideas when you think rev32 will be ready for pushing? Before
I cut the latest WireGuard snapshot (which usually happen once a week
or every other week), I'd like to at least verify compatibility with
another person's noise implementation, which means waiting for
somebody to update one, which means rev32.

Jason