Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 18 Aug 2013 20:48:50 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: Parallella: bcrypt

Katja,

On Fri, Aug 16, 2013 at 07:19:36PM +0200, Katja Malvoni wrote:
> On Thu, Aug 15, 2013 at 1:39 AM, Solar Designer <solar@...nwall.com> wrote:
> > I suggest that you use a smaller wordlist and a smaller list of hashes -
> > e.g., 64 of each (with every word actually cracking a password) - and do
> > many more invocations.  This will likely trigger the problem more
> > frequently (because more of the bcrypt computation results will actually
> > need to match a loaded hash).  Maybe you'll trigger it on the "embecosm"
> > system too.
> 
> I tried this on both system. Results are attached. And it didn't trigger
> the problem... Only problem is occasional stall (lines with 0 cracked
> passwords).
> What should I do next?

Regarding this specific problem, here's another test to run:

On a fast x86 system with native support for bcrypt hashes, I am now
generating this file:

perl -e 'for ($i = 0; $i < 1000000; $i++) { print crypt($i, "\$2a\$04\$012345678901234567890u"), "\n"; }' > pw-bcrypt-1M

It's not done yet (will be in a few minutes), but I've already ran some
tests using a 100k lines version of it:

solar@...aro-ubuntu-desktop:~/k2/JohnTheRipper/run$ rm -f john.log john.pot; seq 0 99999 | ./parallella_john.sh -stdin -form=bcrypt-parallella ~/pw-bcrypt-100k > /dev/null
Press 'q' or Ctrl-C to abort, almost any other key for status
100000g 0:00:00:47 2103g/s 2103p/s 2103c/s 106417KC/s 99968..99999
Use the "--show" option to display all of the cracked passwords reliably
Session completed

I did not trigger any problem yet.  While you may put this test in a
loop and maybe eventually trigger the problem, I'm not sure if it'd be
any better than the tests you ran so far... even though we're testing
every bcrypt computation here.  What I was hoping for is that this would
trigger the problem much quicker, but it does not.

I'll upload the 1M file shortly and you may play with that, but I am not
sure what to do about this next.

Are you making any progress (and what kind of progress) on bcrypt in
Zynq FPGA?

Alexander

Powered by blists - more mailing lists

Your e-mail address:

Powered by Openwall GNU/*/Linux - Powered by OpenVZ