john-dev - starting attacks from given point, generic split of attack for distribution

Follow @Openwall on Twitter for new release announcements and other news

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

Message-ID: <20150719173806.GA2716@openwall.com>
Date: Sun, 19 Jul 2015 20:38:06 +0300
From: Aleksey Cherepanov <lyosha@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: starting attacks from given point, generic split of attack for
 distribution

I think john lacks options to start an attack from given point. Some
trickery with crafted .rec files can be done but it is considered bad
style.

I'd like to have a generic interface for all modes that provides the
following thing for fully specified attack (with hash file, wordlist
and so on):
- precise number of candidates (counting those that are to be skipped
by rules)
- ability to start attack from any point by number
- ability to specify end of attack, preferably by number of candidate
or by time (like "exit after 5 minutes") saying end point back to
caller

With such interface, it will be straight forward to split tasks for
distribution. So far, it is possible with --node= option. Will such
interface be better than --node= ?

john has ability to estimate remaining work. john has ability to save
and restore state. I wonder how hard it would be to restore state from
random point.

Also this question slightly relates to generation of candidates on
gpu/fpga, or not?

Ideas? Suggestions?

Thanks!

-- 
Regards,
Aleksey Cherepanov

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.