NetNews Usenet Archive 1992 #26/NN_1992_26.iso/spool/comp/security/misc/1643

Text File | 1992-11-05 | 2.9 KB | 73 lines

Newsgroups: comp.security.misc
From: [e-mail] (Granville Moore)
Path: sparky!uunet!pipex!demon!nemesys.demon.co.uk!gvm
ReplyTo: [e-mail]
Subject: Re: GNU su doesn't restrict root access? Why?
References: <[e-mail]>
Distribution: world
X-Mailer: cppnews $Revision: 1.19 $
Organization: Dis-organisation
Lines: 58
Date: Wed, 4 Nov 1992 14:26:36 +0000
Message-ID: <[e-mail]>
Sender: [e-mail]


In article <[e-mail]> [e-mail] (John F. Haugh II) writes:

> In article <[e-mail]> [e-mail] (Granville Moore) writes:
> >Aren't you assuming that you can check each dictionary word against
> >the list of 10 encrypted passwords in the same amount of time as you can
> >check it against 1 encrypted password?  This isn't the case with Unix,
> >since the `salt' used in the encryption process means that the encryption
> >will (in the vast majority of cases) have to be performed 10 times.
>
> Without addressing the other issue, there is no assurance that the
> salts will be different.  The odds are much better that 1 in 4096 (or
> even 10 in 4096) that two will be the same (It's about 1 in 75) assuming
> salts are completely random.  And if you have a non-distributed system
> as the earlier poster does, it's quite possible that each system has a
> different salt for each encrypted password ...
> --

So what you are saying is that you have a 1 in 75 chance (I make it
about 1 in ~91, but we won't quibble) that two out of the 10
will be the same.  So, if the cracker is lucky (just over 1% of
the time) then he/she'll only have to perform 9 encryptions
instead of 10 - he/she has two which match, but he/she still has 8 different
other ones.  He/she would probably decide to concentrate only on the two that
match, effectively getting almost a 50% speed-up, but this will only work
in about 1% of cases.

I'm not sure what you mean by a non-distributed system, here.  Assuming
you mean "distributed", then if it uses NIS, or similar, then only
the encrypted passwords are copied to the other systems, so each one
will have exactly the same salt as it had before, and there is no
change in vulnerabilty (from this point of view, anyway).  If the
systems don't use NIS, then the password lists are independent, and
your cracker has a list of 20 to work with, rather than 10.  He/she
therefore naturally has a better chance of getting 2 salts which
coincide.  It still doesn't help a lot, though - about a 1 in 20
chance of getting 2 the same, and 18 different.  The chances of 3
the same will be about 1 in 4000ish.

In summary, some speed-up can be obtained, in some "lucky" cases, but
it's nowhere near the factor of 10 in every case which you implied.

I'd still use this method rather than try to get 10 users to share
a password, change it regularly, and keep it secret.


Regards,

Granville