- From: John Franks <john@math.nwu.edu>
- Date: Fri, 19 Dec 1997 10:08:15 -0600 (CST)
- To: Scott Lawrence <lawrence@agranat.com>
- cc: jg@w3.org, paulle@microsoft.com, ietf-http-wg@w3.org
Note the temporary mailing list: ietf-http-wg@w3.org On Fri, 19 Dec 1997, Scott Lawrence wrote: > > > John Franks: > > > It is the client who must be concerned about reused nonces to avoid > > a replay attack. To avoid a replay attack the client would have to > > keep a data base of all previous nonces and make sure they are not > > reused. > > No - it only needs to keep the nonce it used for the outstanding > request; if that does not produce the correct digest then it is not > valid even if it would have been valid for some earlier request. > Maybe I am not understanding you. It seems to me that if a client, for example, regularly places an order and receives and acknowledgement there is a possible replay attack. If an intermediary attacker intercepts the order he can impersonate the server and offer a nonce from a previous transaction. The client then submits the order which is grabbed by the attacker and then the attacker replays a previous acknowledgement with the dates changed. A similar attack might involve sending a fake 304 when in fact the resource has changed. This is what hash dates is supposed to prevent. Am I missing something. John Franks john@math.nwu.edu
Received on Friday, 19 December 1997 11:12:08 UTC