Re: [TLS] WGLC for draft-ietf-tls-grease

[Hubert Kario <[e-mail]>]

On Wednesday, 27 February 2019 13:52:57 CET Stephen Farrell wrote:
> On 27/02/2019 12:30, Hubert Kario wrote:
> > I'm not sure which part for the key_share extension you mean as being
> > empty, but the key_exchange in the KeyShareEntry can't be empty, per RFC
> > 8446,> 
> > Section 4.2.8:
> >       struct {
> >       
> >           NamedGroup group;
> >           opaque key_exchange<1..2^16-1>;
> >       
> >       } KeyShareEntry;
> 
> Ah sorry, you're right - there is one byte in the
> key_exchange with a value 0x00 in a greasy KeyShareEntry
> I just looked at.
> 
> And a different browser doing tls1.2 sent 2 greasy
> extensions, one with no data and another with a single
> byte (0x00 again).

that's ok, the extensions in general can have empty payloads
 
> I guess my question remains though, as to whether more
> bytes ought be sent in these fields sometimes and whether
> or not they also ought be random values.
> 
> FWIW, I guess it'd make sense to send a range of random
> length random values some of the time, but mostly I'm
> wondering what other folks are doing/expecting.

for extensions doing an empty one and one with just few bytes should be 
sufficient

For key_share, to simulate actual use I'd say the values should be at least 
few dozen bytes long. That does inflate unnecessarily the ClientHello, so it's 
a balancing act...

-- 
Regards,
Hubert Kario
Senior Quality Engineer, QE BaseOS Security team
Web: www.cz.redhat.com
Red Hat Czech s.r.o., Purkyňova 115, 612 00  Brno, Czech Republic