Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Fri, 5 Jun 2015 15:44:31 +0300
From: Solar Designer <>
Subject: Re: poor man's fuzzer

On Fri, Jun 05, 2015 at 01:50:36PM +0300, Solar Designer wrote:
> Another curious detail is that I couldn't get it to use more than ~4 CPU
> cores on average, not even with hundreds of simultaneous processes,
> until I moved the "run" directory for john to /dev/shm (tmpfs).  I guess
> I was bumping into some lock congesion in the kernel, possibly e.g. for
> atime updates of john and john.conf.  (If so, remounting the filesystem
> with noatime should probably have helped.  I didn't try yet.)  So I am
> currently running it in /dev/shm/fuzz, which is a bit risky - in case
> the system gets rebooted, work will be lost.  I should revise the script
> to write its log and sample files into a directory different than one it
> invokes john from.

I've just checked - we already have the filesystem mounted noatime.
So the cause of lock contention (or whatever it was) must be different.
Maybe the open of john.pot for writing.


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.