[20175] in bugtraq
Re: flaw in RH ``mkpasswd'' command
daemon@ATHENA.MIT.EDU (Shez)
Thu Apr 12 15:29:56 2001
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-ID: <01041213184501.00669@kara>
Date: Thu, 12 Apr 2001 13:18:45 +0100
Reply-To: Shez <shez@MOLIONS.COM>
From: Shez <shez@MOLIONS.COM>
X-To: Lee Howard <faxguy@deanox.com>
To: BUGTRAQ@SECURITYFOCUS.COM
In-Reply-To: <3.0.6.32.20010411175527.00aa7c70@server.deanox.com>
Hi,
PeterW@usa.net pointed out to me that if the man page for mkpasswd is
correct then this shouldn't be the problem, and that it may be a seeding
problem. The code appears to seed of pid and uses its own generator:
set _ran [pid]
proc rand {m} {
global _ran
set period 259200
set _ran [expr ($_ran*7141 + 54773) % $period]
expr int($m*($_ran/double($period)))
}
Anyhow, writing a perl/any language replacement would be a good move for
anyone who's been using this command.
Cheers
Shez
On Thursday 12 April 2001 00:55, you wrote:
> Although I wonder if they're not outweighing the risk of dropping some of
> those rules over the risk of having fewer possible passwords. (Not to say
> that I agree with that, though.)
>
> Lee.
