client_secret as the HMAC key...?

[Brian Campbell]

Messages 4.3. Signing[1] says tor "Symmetric Signatures ... the client and server MUST establish a shared secret out of band."

But isn't the shared secret intended to be the client_secret and shouldn't that be stated explicitly?

Step 6 in 5.2 ID Token Validation [2] says "If the alg parameter of the JWT header is a MAC based algorithm such as HS256, HS384, or HS512, the client_secret for the client_id contained in the aud (audience) Claim is used as the key to validate the signature.", which is consistent with client_secret as the MAC key. client_secret_jwt in 2.2.1. Client Authentication [3] does also.

[1] http://openid.net/specs/openid-connect-messages-1_0-15.html#sigs [2] http://openid.net/specs/openid-connect-messages-1_0-15.html#id.token.validation [3] http://openid.net/specs/openid-connect-messages-1_0-15.html#client_authentication

[Vladimir Dzhuvinov]

Yes, the messages paragraphs should link to the client_secret in registration, for OPs who choose to implement the reg endpoint.

[Brian Campbell]

even for those that don't implement the reg endpoint, the client sec is (I think) well understood and it should be called out that it is the shared secret that is used, or use to bootstrap, the symmetric crypto

[Michael Jones]

We will make this clarification in 4.3.

[Michael Jones]

Fixed #761 - client_secret as the HMAC key. Fixed #762 - client_secret to key.

→ <<cset 45d34fab8fac>>

[Michael Jones]

Fixed #761 - client_secret as the HMAC key. Fixed #762 - client_secret to key.

→ <<cset 9d0ddfc1c934>>