Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Tue, 8 Dec 2015 14:00:17 +0100
From: Marek Wrzosek <>
Subject: Re: using --fork for the first time

W dniu 08.12.2015 o 07:34, Patrick Proniewski pisze:
> Hello,
> I've managed to compile 1.8.0-jumbo (bleeding, from 2 or 3 days ago) on my FreeBSD server, so I can now use the --fork option. It works ok so far:
> $ ./john --fork=4 sup.ssha.20151205 
> Using default input encoding: UTF-8
> Loaded 15151 password hashes with 15151 different salts (Salted-SHA1 [SHA1 128/128 AVX 4x])
> Node numbers 1-4 of 4 (fork)
> But when I look at the results, I find that I've got a lot of duplicates in john.pot:
> $ wc -l john.pot
>     1254 john.pot
> $ sort -u john.pot | wc -l
>      979
> Is that to be expected? Digging into john.pot shows that most dupes are probably cracked during 'single'.
> regards,
> pat

Sometimes pot syncing is too slow for fork with dictionary attacks and
rules. Two or more processes are generating the same password candidates
using different word and rule.

Eg. If one process is at word 'foo' and will apply rule that would add
01 at the end and second process is at word 'foo0' and will apply rule
that would add 1 at the end - this will cause duplicates in pot file.

For some hashes and dictionary attacks with rules it's better to use
OpenMP than fork. It depends on hash, type of CPU and load.

Best Regards
Marek Wrzosek

Powered by blists - more mailing lists

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