On Fri, 2004-03-05 at 13:57, Jason Holt wrote:
> [...]
> >     var final_to_encode = passwd_enc + form.one_time_token.value;
> >     form.encoded_pw.value = calcMD5(final_to_encode);
> [...]
> 
> > This way the password cannot be "sniffed" and the page is not vulnerable
> > to a replay attack.
> 
> Firstly, you'd lose to any kind of active attack, which isn't too much harder
> than passively listening.
> 
> But secondly, no, replay is still possible.  form.one_time_token.value
> corresponds to the output of calcMD5, so I just replay both values.  Keeping
> track of which value you specified on the server side can help, but is also
> trouble-prone.

It never occurred to me that a server might _not_ keep track of which
token was sent. The whole point is that the token can only be used one
time. In the case of blackboard they have actually disabled adding the
token to the password before hashing it. So blackboard could suffer a
replay attack, particularly if the token isn't handled properly.

So Jason, supposing you had to implement this kind of a system (and
let's say you can't just say "That's stupid!" and walk away.) How would
you do it? In particular, how would you handle the token properly?

Yahoo! does this exact thing for their login.

Mark, you could learn a ton about authentication (and all it's possible
weaknesses) by studying Kerberos <http://web.mit.edu/kerberos/www/>.
There's a particularly easy overview here
<http://web.mit.edu/kerberos/www/dialogue.html>.

But in your case that won't help, 'cause you still have to send a
plain-text password to the LDAP server anyway (BYU's LDAP server for
those who just tuned in). In fact I really can't think of a good way to
do what you want to do without SSL.


____________________
BYU Unix Users Group 
http://uug.byu.edu/ 
___________________________________________________________________
List Info: http://uug.byu.edu/cgi-bin/mailman/listinfo/uug-list

Reply via email to