I would note that this is one of those cases that a required use of the typ field makes a lot of sense.
Set typ to one of JWN, JWM or JWS Only allow those algorithms which match the typ to be validated by the library Set up the convention that specs that want to impose more requirements would use "Foo+JWS" as their type indicator so that the library still knows which of the types to be checking against algorithms Jim From: [email protected] [mailto:[email protected]] On Behalf Of Jim Schaad Sent: Tuesday, September 03, 2013 3:15 PM To: 'Richard Barnes'; 'Mike Jones' Cc: [email protected] Subject: [jose] Mxed Signature Algorithm TYpes A side question from this discussion that I would like to see addressed is should we state any recommendations about mixing different types of signatures algorithms in the same message. I would think that it makes sense to say that type types "none", "mac" and "asymmetric-Signature" should never be mixed in the same message because these different types of algorithms have significantly different security properties. From: [email protected] [mailto:[email protected]] On Behalf Of Richard Barnes Sent: Tuesday, September 03, 2013 1:33 PM To: Mike Jones Cc: [email protected] Subject: Re: [jose] Text about applications and "alg":"none" This text is still far too weak, and does not reflect what I remember EKR saying (in particular, there is no MUST). It does not address the attack where an application may, in general, be willing to accept both signed and unsigned content, but each in specific contexts. Proposed text: """ JWS implementations MUST provide an interface for applications to specify a list of "alg" values that are acceptable for the validation of a given JWS object. JWS implementations MUST NOT indicate that a JWS object is valid if the "alg" value for the object is "none", unless the application has specifically indicated that the value "none" is acceptable for the particular JWS object being validated. Applications using "none" MUST indicate support on a per-object basis, in order to avoid downgrade attacks that arise if more broadly-applicable preferences are specified. """ I continue to believe that this is far too subtle, and that applications are very likely to get it wrong. It is far simpler and safer to require that a JWS implementation MUST reject an object with "alg":"none", and have another content type for unsigned content. Also, if "none" is going to remain, then it needs to be OPTIONAL. Given all the above limitations, I don't see how you could justify it being mandatory. --Richard On Tue, Sep 3, 2013 at 2:02 PM, Mike Jones <[email protected]> wrote: I took an action item during the last call to write text along the lines suggested by ekr about applications and "alg":"none". I propose that the following text be included: It is RECOMMENDED that libraries provide applications a means of specifying the list of acceptable algorithms used in a JWS object in a way that causes inputs using algorithms outside the specified set to be rejected. In particular, it is intended for applications to use this mechanism to exclude accepting inputs using "alg":"none" in security contexts where non-integrity protected inputs are not acceptable. Feedback/proposed wording refinements welcomed. -- Mike _______________________________________________ jose mailing list [email protected] https://www.ietf.org/mailman/listinfo/jose
_______________________________________________ jose mailing list [email protected] https://www.ietf.org/mailman/listinfo/jose
