[dix] Re: [Ietf-http-auth] Notes on Web authentication enhancements
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[dix] Re: [Ietf-http-auth] Notes on Web authentication enhancements




On 26-Jun-06, at 12:20 PM, Eric Rescorla wrote:

Dick Hardt <dick at sxip.com> writes:

On 23-Jun-06, at 9:17 PM, Eric Rescorla wrote:

In DIX, the RP includes a nonce in the request, which must then also
be in the nonce which would prevent replay attacks assuming the RP is
managing nonce state would it not?

Only if each authentication token is only single-use. Otherwise, an attacker can replay it during the validity period. Even then, cut-and-paste attacks are still possible if you block the initial request.

My understanding of the definition of a nonce is that it is single- use.
Would you humour me with an explanation of a cut-and-paste attack per
above?

So, let's take the diagram from my recent mail message:


User RP IdP ---- -- --- Sell me beer -> <- Prove you're 21

IdP, help me!                                   ->
          <---------- Auth exchange ------------->
                                         <- Credential

Credential ->
                       <- OK


The IdP gives the User some credential that he gives the RP. That's fine for the first request, but what happens when the user wants to make a second request? You clearly don't want to go back to the IdP every time. The classic solution is for the server to give you some cookie: those cookies can obviously be cut-and-pasted from one message to another. Even if you make them single-time (evolve them every time) there's a window between the cookie delivery (in the HTTP response) and the next HTTP request.

Another option is to bypass the cookie thing and just make
the Credential reusable, but this has the same problem...

Thanks Eric. Very useful to understand.
This seems to be a more general issue of securely managing state over HTTP would it not?


-- Dick

_______________________________________________
dix mailing list
dix at ietf.org
https://www1.ietf.org/mailman/listinfo/dix




Note: Messages sent to this list are the opinions of the senders and do not imply endorsement by the IETF.