- From: Paul Crowley <paul@lshift.net>
- Date: Fri, 09 Dec 2005 02:07:10 +0000
- To: xml-encryption@w3.org
I want to use EncryptedKey to transport an HMAC-SHA1 key, but the obvious way of doing it seems to be open to algorithm substitution attacks. Specifically, I'd like Alice to be able to agree an MAC key with Bob by sending Bob a document signed by Alice which includes the MAC key inside an EncryptedKey signed with Bob's public key. Bob should then be able to use this key to authenticate documents as originating from Alice. However, this construction doesn't seem sufficient by itself to do this securely, because this protocol doesn't communicate which MAC algorithm is to be used until Alice sends a message to Bob. This means that once Alice and Bob have agreed a MAC key, our attacker can choose whichever MAC algorithm is weakest among all those that Bob accepts, and use that to authenticate his bogus message. AFAIK, if it is possible for the attacker to cause the same key to be used for two different algorithms, then security cannot be guaranteed. How can I ensure that whenever I use EncryptedKey or SymmetricKeyWrap to communicate a key, the "type information" gets passed alongside it? Thanks in advance! -- [][][] Paul Crowley [][] LShift Ltd [] [] www.lshift.net
Received on Friday, 9 December 2005 09:15:13 UTC