On Tue, Apr 02, 2002 at 05:53:28PM -0600, McNutt, Justin M. wrote:
> The problem I have with all of this is the fact that the actual passwords can be 
>deduced using the "cleartext equivalent" that MS stores.  This is a huge weakness in 
>NT/2K-based authentication that I was hoping to get around using FreeRADIUS.
> 

The actual password cannot be deduced from the cleartext equivalent in any
way other than dictionary attack / guessing (same as for unix crypt or md5).
The only problem with this is that the M$ hash has no salt, so an attacker
can precompile a dictionary.

However, a salt is relatively useless as the md4 hash (what M$ stores)
is cleartext equivalent, so an attacker does not need to deduce the actual
password -- at least not for purposes of using M$ style authenticated
resources.  So MS-CHAP is not worse than regular CHAP (essentially EAP-MD5)
in this regard.

There is definitely a problem if someone has the md4's and a user re-uses
their password on other systems, eg to authenticate to both unix and
windows machines.

You could use PAP auth or perhaps EAP-TLS.  PAP auth has the disadvantage
of the password going over the wire in the clear to the NAS, so you have
to weigh your concerns here.  I don't know of any radius servers that do
EAP-TLS.  (But then, I know almost nothing about EAP period.)

There are better ways to do this, but there are no implementations of them.

/fc


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

Reply via email to