Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Mon, 5 Aug 2013 15:43:11 +0530
From: Sayantan Datta <std2048@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: Sayantan's Weekly Report #7

On Mon, Aug 5, 2013 at 1:49 PM, Solar Designer <solar@...nwall.com> wrote:

> Sayantan,
>
> On Sun, Aug 04, 2013 at 06:54:09PM +0530, Sayantan Datta wrote:
> > Accomplishment:
> > - Added wordlist mode with mask mode.
> > - 5-formats(nt, raw-md5, raw-md4, raw-sha1, mscash) can use wordlist +
> mask
> > mode.
>
> Nice.  Is the rest of JtR functionality intact in your tree, though?
> Does wordlist mode still work fine on its own?  Does the rule
> preprocessor still work fine for its original purpose?  (I haven't
> checked, hence the questions.)
>

Mask mode doesn't support restore functionality yet. Other than that I
haven't tampered any other mode. So they should work as usual once I add
the support for non-mask modes to the formats. Yes the rule preprocessor
works fine for original purposes(I have used them during the contest for
cpu formats).


>
> > - Participated in contest, found out some real world problems.
> > e.g descrypt had a bug in which regenerated passwords for positive
> matches
> > were wrong. Fixed during the contest. descrypt-opencl was the most usable
> > format among the new mask formats.
> >       new fast hashes are fast only for fewer loaded hashes. For
> thousands
> > of loaded hashes they are utterly slow and equivalent to multi-core cpus
> at
> > best.
>
> How many loaded hashes were you testing on before the contest?
>

I was loading 1 to 50 hahses before contest. In most of the cases the
bitmaps would have been rejecting them. But when number of hahses becomes
large the two small bitmaps weren't most likely effective. I'll
redesign(next week) the compare functions to be more effective against
large number of loaded hashes.


>
> > Priorities:
> > - Currently descrypt format is in a complicated state, sometimes I
> couldn't
> > understand what/why I had done certain things the way I had done in the
> > format. So un-complicate things as much as possible so that it easier for
> > others and me to debug and understand the code later.
> > - Start merging mask mode with main bleeding.
>
> Yes, code simplification and cleanups are highly desirable!
>
> > Priorities for second half of GSoC:
> >  - Instead of adding new formats(which is relatively easy once you know
> how
> > to do it and probably others can do it too), I'd be focusing on
> optimizing
> > the 5 new formats(except descrypt) which are utterly slow in real world
> > scenarios. Only descrypt performed as expected during the contest.
>
> Makes sense.  Optimizations and cleaner code, please.
>
> Thanks,
>
> Alexander
>

Regards,
Sayantan

[ CONTENT OF TYPE text/html SKIPPED ]

Powered by blists - more mailing lists

Your e-mail address:

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