hi

right, the EAP/Identity and User-Name must be the same, that's the job
of the client, and we could thus verify only one, completely ignoring
the other. however, the rlm_eap_tls currently authentifies the
_certified_ name, which does not have to match either of the both...

the bug i mentioned refers to the missing comparison of one of the both
(from here on i will use the term "User-Name") to the certified name (CN
in the certificate). as i already explained twice on this list, the
problem is that the both do not HAVE to be strictly string-equal: e.g.
in the case of proxying the User-Name is likely to have a suffix which
the CN of the certificate is very unlikely to have in practice. thus, as
i proposed before, there should be a definable equivalence (e.g. in the
tls-module options) or even better a regular expression (or an external
handler) which specifies exactly when the both can be considered equal.


ciao
artur


Alan DeKok wrote:
> 
> Artur Hecker <[EMAIL PROTECTED]> wrote:
> > that's right, you don't. eap module will authentify independently. it
> > can be seen as a bug, since the authentication is not very consistent.
> > everything else in the server - e.g. the accounting - is based on the
> > user-name...
> 
>   Further, the RFC's say that if an EAP client has a user name, it
> MUST include that in the EAP-Identity, and also in the User-Name of a
> RADIUS packet.
> 
>   The latest CVS snapshot is a little more forgiving, in that it
> allows *SOME* EAP authentication types without a User-Name.
> 
>   Alan DeKok.
> 
> -
> List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html

-- 
Artur Hecker
artur[at]hecker.info

- 
List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html

Reply via email to