[929] in Kerberos
Re: login authentication
daemon@ATHENA.MIT.EDU (Greg Wohletz)
Mon May 7 21:50:03 1990
To: "Jon A. Rochlis" <jon@MIT.EDU>
Cc: kerberos@MIT.EDU
In-Reply-To: Your message of Fri, 04 May 90 13:26:41 -0400.
Date: Mon, 07 May 90 12:32:29 -0700
From: Greg Wohletz <greg%duke.cs.unlv.edu@RELAY.CS.NET>
> >In login.c is the following comment:
> > ---
> >shouldn't login be able to do mutual authentication and avoid this problem
?
> >Or am I missing something?
> Stan Zanarotti (srz@athena.mit.edu) sent me a patch to login.c that
> does just this - tries to get a rcmd ticket for the machine you are
> about to login to and has the machine decrypt it - so if he will permit
> I can pass it on to you; or maybe he has sent it to you himself
> already...
>
>That is a fine solution, if you have a place to store the secret (the
>rcmd srvtab) . It doesn't work on "public" workstations ...
how about a login ticket. The login process would have to authenticate
itself to the login server which presuable would run on a secure machine.
Tis way you only need a srvtab entry on the machine that runs the login
server (which presumably is secure). Actually it could just get a rcmd for
a machine in a locked room somewhere and get rid of the srvtab problem,
right?
>the
>Athena theory is that if you are phsyically in front a machine it's
>yours anyway, so you might as well let anybody log in.
unless the machine has hardware passwords that keep people from doing
anything other than booting from the default place...
--Greg
