Re: [kitten] SPAKE preauth: generation of SPAKE2 secret input

[Nico Williams <[e-mail]>]

On Wed, May 13, 2015 at 09:15:19PM -0700, Watson Ladd wrote:
> On May 13, 2015 9:05 AM, "Nico Williams" <[e-mail]> wrote:
> > > w is not distributed uniformly: far from it. The question of how far
> >
> > That depends on how its salted.  For our purposes, in practice it's not.
> 
> Not true: w comes from a much smaller list of values. If it was
> uniformly distributed over a wide range, we could use it as a key
> directly.

For Kerberos (the context in which we're discussing SPAKE2) w would be
derived from user principal long-term keys, which are basically a
PBKDF2() application to the password and salt.  The salt can in
principle be uniformly distributed and randomly chosen, but it is
public, so w has no more entropy than the password, but w can be as
uniformly distributed as the salt.

In practice the salt is not chosen randomly but derived from the user's
principal name.  So w does not (will not) be uniformly distributed.

In any case, there's a big difference between "uniformly distributed"
and "difficult to guess".  We want a PAKE precisely because user
principal long-term keys are derived from simple passwords and used to
encrypt messages which are then subject to off-line dictionary attacks.

Nico
--