On Thu, 19 Jun 2025 at 22:36, Alan DeKok <alan.dekok=
[email protected]> wrote:

>   Just reviewing it yet again, both RFC 7170 and 7170bis have the
> following text for the Crypto-Binding TLV:
>
> Nonce
>
> > The Nonce field is 32 octets.  It contains a 256-bit nonce that is
> > temporally unique, used for Compound-MAC key derivation at each
> > end.  The nonce in a request MUST have its least significant bit
> > set to zero (0), and the nonce in a response MUST have the same
> > value as the request nonce except the least significant bit MUST
> > be set to one (1).
>
>
>   Except that the Nonce is *not* used for the Compound-MAC key derivation
> at each end.
>
>   Do implementations set / check the Nonce field as discussed above?
> Would it make sense to just ignore this field?


See section '5.3. Computing the Compound MAC' in the original TEAP RFC and
step 1 therein.
https://datatracker.ietf.org/doc/html/rfc7170#section-5.3

   1  The entire Crypto-Binding TLV attribute with both the EMSK and MSK
      Compound MAC fields zeroed out.

This could be clarified to say something like this:

   1  The entire Crypto-Binding TLV attribute with both the EMSK and MSK
      Compound MAC fields zeroed out. Nonce is used untouched.

If flip a bit in received or sent nonce right when it's packed into a
message, I see, for example, in eapol_test this before failure:

    EAP-TEAP: MSK Compound MAC did not match

Or do you mean something else about nonce?

-- 
Heikki Vatiainen
[email protected]
_______________________________________________
Emu mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to