Document: draft-ietf-keyprov-dskpp-10.txt Reviewer: Brian Carpenter Review Date: 2010-04-22 IETF LC End Date: 2010-04-26 IESG Telechat date: Summary: Almost ready, one issue about algorithms. -------- Note: This is a long draft and I am not a cryptography or XML expert. I do not claim to have reviewed the draft in detail. Major issue: ------------ "3.4.3. The DSKPP Message Hash Algorithm When sending its last message in a protocol run, the DSKPP server generates a MAC that is used by the client for key confirmation. ... d. The resultant string is hashed using SHA-256 in accordance with [FIPS180-SHA]." I am concerned that this is rigidly bound to SHA-256. What happens when SHA-256 goes the way of MD-5? Shouldn't this be a pluggable algorithm? In fact there's another similar issue a few lines earlier: "For the purposes of this document, the secret key k MUST be at least 16 octets long." What would happen if that minimum needs to be raised? Looking at Section 9 (Conformance Requirements), which BTW seems to be extremely useful to implementors, I get the feeling that the point I'm raising is a bit more general. What is the method for adding and deprecating crypto algorithms in DSKPP? If the only method is by bumping up the DSKPP version number, is that going to be a blocking problem in years to come? Maybe Section 1.2 (Versions) should say something about the intent and methodology for versioning the protocol and its range of algorithms. (Full disclosure: I'm exercised about this as an author of draft-iab-extension-recs, in particular because of a pertinent recent comment that it "should specifically note that any protocol using cryptography needs to be designed to be algorithm-independent.") Nit: ---- == Using lowercase 'not' together with uppercase 'MUST', 'SHALL', 'SHOULD', or 'RECOMMENDED' is not an accepted usage according to RFC 2119. Please use uppercase 'NOT' together with RFC 2119 keywords (if that is what you mean). Found 'SHOULD not' in this paragraph: If keys generated in DSKPP will be associated with a particular...