At Tue, 5 Aug 2008 07:56:12 -0700, Dan Wing wrote: > > ... > > Sure, but again, that requires examining every single piece > > of the message > > which you wish to exampt from the signature and determine whether > > there is some important attack that can be mounted by modifying > > that section. As I noted above, those questions are not necessarily > > immediately apparent. > > Implicit in that argument is that 4474 got it right.
No, I don't think that's true. 4474 did the conservative thing: it signed everything. > We know it already isn't done correctly with RFC4474 for unidirectional media > (draft-kaplan-sip-baiting-attack). To get bi-directional media, an attacker > would need to share a NAT or a TURN server with the identity they want to > spoof (e.g., a bank, a pizza restaurant, a political organization), and the > attacker would need to obtain the same UDP port from the NAT or TURN server > within RFC4474's replay window (which is recommended to be 10 minutes). I haven't spent a long time examining draft-kaplan-sip-baiting, but as I recall, it's not the fault of 4474 failing to sign something that it should have but rather that it's inherent in the message-oriented nature of SIP. With that said, ISTM that this cuts against your argument that we should be signing less of the message, since the failure of RFC 4474 (to the extent there is one) in this case is that it doesn't protect *enough* information. -Ekr _______________________________________________ Sip mailing list https://www.ietf.org/mailman/listinfo/sip This list is for NEW development of the core SIP Protocol Use [EMAIL PROTECTED] for questions on current sip Use [EMAIL PROTECTED] for new developments on the application of sip
