> It doesn't give the process root's password.  It says "authenticate root",
> or it doesn't say "authenticate root".  Now, you could write a password
> cracker which sits on top of this, but it's not significantly more
> effective than, say, typing 'su' constantly.  :)

How do you authenticate root without getting some information whereby
you can authenticate root?   If xlock can accept a root password, why
can't a user write a trojan horse that accepts the root password and
looks just like xlock?   It doesn't matter that you've protected the
crypted root password if the attacker manages to get it in plaintext!

Part of security is making sure your programs don't have software bugs
that make them vulnerable to attack.  Another equally important part
of security is making sure that your procedures do not introduce
vulnerabilities.

Allowing the root password to interrupt an xlock session is an example
of a procedure that introduces a vulnerability.  Another example would
be allowing xinit to prompt for a password when a user has connected
using an insecure link.  There are many examples of problems like
this in existing software.  The fact of their existance does not
justify introducing features to aid in their perpetuation.

			       _MelloN_