Apologies for the digest response, I get the digest version of the list.

Date: Thu, 09 Oct 2003 09:05:48 +0200
From: Artur Hecker <[EMAIL PROTECTED]>

> no, PEAP is a different protocol. you could use TTLS with whatever EAP
> method tunneled in it.

How does this help with ease of use with XP?  Do people commonly tunnel
MD5 over TTLS?  Or something else -- password auth?  What software
supports this?

Is anyone working on PEAP support native in FreeRADIUS?


> you still have DEFAULT values in your users file, right? if you
> explicitly reject the user, he will NOT be authenticated.

I blanked out my users file before I started, but putting a DEFAULT
Auth-Type Reject in the end of the file solved it.  I'm going to have
to look at the config a little more...

> however, it's true that the User-Name content, the certified name AND
> the EAP-Identity information is not checked for consistency by the
> server. (EAP-Identity should be equal User-Name - that's the function
> of the AP, that is something you have a trust with; however, these
> both  compared to the certified name in the certificate could NOT
> match and the certificate would still be accepted. the question here
> is: do they have to match as strings or which is the good metrics?
> perhaps a configurable comparison handler?)

I don't see an EAP-Identity value in my server debugging.  What does XP
send for that value?  The name of the cert, or the machine
identification?


> i didn't try WPA yet, but do you have the XP WPA-patches? i suppose
> you have *sigh* perhaps also the newest firmware for 1200.

I wasn't aware a patch was needed, but I've just downloaded it.  The
1200 is up-to-date; it shipped with VxWorks and I updated it with the
latest update image from Cisco.

> the per-session keys (PMKs sent to the APs and the derived TKIP keys)
> will be different since they are derived from the TLS master which is
> based upon random numbers chosen by the peers during the
> authentication process, so with high probability different for every
> session.

That was my concern.  I don't mind everyone using the same credentials
to access the wireless network, but I didn't want the shared encryption
environment we currently have with WEP.


> however, virtually it would all be one person for you, ie all users
> connecting is the one and the same - normal, since you have ONE
> certified identity. unless you want to use the "bug" in the server,
> described above (User-Name/EAP-Id don't have to match CN) by
> activating the XP option 'use a different user name on connection' and
> typing in  the desired name. however, be assured that then every user
> could type  ANYTHING he wants and probably he would. so, i wouldn't
> call it secure, unless you have full trust in your co-workers :-) but
> it will be still difficult to break your links from outside, almost as
> difficult as when you used different certificates - thanks to TLS.

True.  We're currently using MAC authentication to track users back to
devices, and control access.  We could still do that with EAP; the
certificate would be the replacement for the shared WEP key, but the
per-user encryption would be better.

I still think PEAP is a better route, without having to put any
certificate on the user machine, but I guess that's not an option right
now.

Dave

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

Reply via email to