[OpenID-specs-EAP] EAP call notes 20-Jul-17
Mike Jones
Michael.Jones at microsoft.com
Thu Jul 20 15:23:26 UTC 2017
EAP call notes 20-Jul-17
Mike Jones
Brian Campbell
Nat Sakimura
John Bradley
Agenda
Detecting Downplay Attacks
Crypto Agility for the "tbh" Hash Function
Clustered OAuth Web Server Clients
OAuth Token Binding Security Considerations
Next Call
Detecting Downplay Attacks
Brian: The Security Considerations language currently seems possibly overly specific
Paragraph 3 in Section 4 describes indeterminate cases as likely attacks
John: If you're man-in-the-middled, the attacker will change the browser string to one that doesn't support Token Binding
Brian: The on/off metadata that we currently have may be useful but is not sufficient
John: An indeterminate state can occur if the client negotiates token binding and sends a referred token binding
but receives an unbound ID token
This could happen due to algorithm mismatches - although servers are required to support all the algorithms
We should say in the spec that this should be rejected
Mike: There may be deployments in which specific knowledge of capabilities of the participants is used
John: If the OP doesn't receive a Token Binding, it may need to inform the user to choose an appropriate browser or that something bad is happening
Crypto Agility for the "tbh" Hash Function
We should add similar language as that used for "x5t" and "x5t#S256"
Brian will take a crack at this (he's already doing this for the OAuth MTLS draft)
Clustered OAuth Web Server Clients
Brian talked about Token Binding for clustered Web Server clients
He's worried that Refresh Token Token Binding may be problematic for clustered deployments
The client would have to share Token Binding keys across instances in clustered deployments
Brian thinks there may be need to be a way to allow Web Server Clients to bind access tokens but not refresh tokens
The best way to indicate this is probably client metadata
A dynamic parameter to turn it off seems like it would be a bad idea
OAuth Token Binding Security Considerations
There are likely similar wording issues in the OAuth Token Binding Security Considerations language
Brian will have a look at this
Next Call
Our next call is Thursday, August 3rd, 2017 at 8am Pacific
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openid.net/pipermail/openid-specs-eap/attachments/20170720/60a96f71/attachment.html>
More information about the Openid-specs-eap
mailing list