Hi Yaroslav, thank you for your review. Please, see inline.
> Document: draft-ietf-ipsecme-ikev2-downgrade-prevention > Title: Downgrade Prevention for the Internet Key Exchange Protocol Version 2 > (IKEv2) Reviewer: Yaroslav Rosomakho Review result: Has Issues > > I have reviewed this document as part of the security directorate's ongoing > effort to review all IETF documents being processed by the IESG. These > comments > were written primarily for the benefit of the security area directors. > Document > editors and WG chairs should treat these comments just like any other last > call > comments. > > This specification describes an extension to IKEv2 that prevents particular > downgrade attacks. > > Minor issues: > > I found Section 7.2 incomplete. It requires the new mechanism to be used > during > IKE session resumption if it was used during the previous IKE SA > establishment, > but it does not specify the transcript to be signed or MAC'ed for the > IKE_SESSION_RESUME exchange. RFC 5723 (IKE Session Resumption) does not define any new transcript for authentication, it is the same as specified in Section 2.15 of RFC 7296, but RealMessage1 and RealMessage1 in this case will be the IKE_SESSION_RESUME messages and not the IKE_SA_INIT messages. For this reason, there is no need to specify a separate transcript for resumption. the draft modifies the way data to be authenticated is constructed (as per Section 2.15) and this automatically applies to IKE session resumption too. > I believe it would be beneficial to clarify the security properties when > symmetric shared secrets are used for authentication. The security argument > relies on at least one non-compromised authentication key being used. This is > clear for asymmetric authentication, where compromise of one peer's private > key > does not imply compromise of the other peer's private key. However, when a > single shared symmetric secret is used for authentication, compromise of that > secret compromises both authentication directions. In IKEv2 each peer can use its own PSK for authentication (to construct the AUTH payload). It's true, that in this case both peers have to know both PSKs, but _formally_ they can be different. Thus, I can imagine a situation when one of these PSKs is known to an attacker, while the other is not. In this case we have the situation similar to the asymmetric authentication, - the attacker can create the AUTH payload on behalf of one peer, but not on behalf of the other (provided peers check that the used PSK is associated with the peer identity). Sure, if these PSKs are the same and the attacker knows this fact and this PSK is compromised, then our defense does not work. Do you think this should be clarified somehow in the draft? > Nits: > > Some articles missing: It happens (I'm sure these parts were written by me) :-) > Section 7.1: If peers support extension defined in this document -> If peers > support the extension defined in this document > > Section 7.2: provides a client with session ticket -> provides a client with a > session ticket > > Section 8: one of extensions -> one of the extensions > > Section 9: defines new Notify Message Type -> defines a new Notify Message > Type > > Minor grammar: > > Section 7.2: each of parameter -> each parameter > > Section 7.2: it is RECOMMENDED that the host do not use -> it is RECOMMENDED > that the host does not use > > Section 8: Note, that -> Note that Thanks for these findings, a PR is created: https://github.com/smyslov/ikev2-downgrade-prevention/pull/45 Regards, Valery. _______________________________________________ IPsec mailing list -- [email protected] To unsubscribe send an email to [email protected]
