Re: [TLS] Implementation survey: Client Certificate URL extension

[pgut001 at cs.auckland.ac.nz (Peter Gutmann)]

Martin Rex <Martin.Rex at sap.com> writes:

>Although this Papers describes a serious design flaw in the rfc3280
>suggestion to put URLs of intermediate CAs into X.509v3 cert extensions and
>have peers use them in order to be able to build a certification path, the
>very same problem will apply to every concept that a communication peer can
>be coerced to access one or more arbitrary URLs prior to authentication, and
>the Client Certificate URL extension appears to suffer the same
>vulnerabilities and security problems.

This has been known for a long time, and is even worse in protocols like OCSP
where you can use this facility to scan and penetrate corporate networks, use
machines as attack amplifiers, and so on.  The PKIX response in the past has
been "not our problem".

I don't know if the TLS WG wants to make it their problem or not.  It's always
seemed a very obvious flaw, my code deliberately doesn't support any of these
"turn a host into an attacker-controlled proxy" extensions.

Peter.
_______________________________________________
TLS mailing list
TLS at ietf.org
https://www.ietf.org/mailman/listinfo/tls