john-dev - Re: [JtR patch] Fast cracker for Mozilla Firefox, Thunderbird and SeaMonkey master passwords.

Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Tue, 27 Mar 2012 13:55:56 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: [JtR patch] Fast cracker for Mozilla Firefox, Thunderbird and SeaMonkey master passwords.

On Sun, Mar 25, 2012 at 09:35:52AM +0530, Dhiru Kholia wrote:
> On Sun, Mar 25, 2012 at 9:27 AM, Dhiru Kholia <dhiru.kholia@...il.com> wrote:
> > On Sun, Mar 25, 2012 at 8:55 AM, Solar Designer <solar@...nwall.com> wrote:
> >> On Sun, Mar 25, 2012 at 08:18:09AM +0530, Dhiru Kholia wrote:
> >>> Can the test vectors structure (fmt_tests) be initialized dynamically
> >>> from init()?
> >
> > I tried doing it. It works for bench-marking (shows very low speeds
> > for some reason) but for actual cracking valid() is somehow failing.
> 
> Also this hack won't work if unused folder is moved somewhere else
> after compilation.

It is completely wrong to use anything from under the unused/ directory.
I thought this would be obvious given the name.  If you start using
those files, then you must first move them out of that directory
(to run/ if you use the files at runtime).  However, as magnum
suggested, it's better to have the test vectors compiled in instead.

> It is really worth implementing the fmt_tests for
> mozilla format or should we skip it?

I don't know about this format in particular, but in general we should
try to have self-tests and benchmarks for everything.

Alexander