[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
```Message-ID: <675377cb183e2454d4c13cab3c73944d@smtp.hushmail.com>
Date: Thu, 14 Jun 2012 21:54:42 +0200
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: notes on sharding the incremental search space

On 2012-06-14 14:49, Tavis Ormandy wrote:
> Hey, just FYI, I recently implemented parallelisation for john on an
> unusual cluster architecture, and had to solve one of the problems
> listed as open on the wiki. I don't know if my solution is general
> enough to be helpful to anyone else, but I think my notes might at least
> be useful to someone else trying to solve a similar problem in future!

Welcome, this looks like Good Stuff.

> // Given a { length, fixed, count } triplet, calculate the maximum amount of
> // work expected in inc_key_loop().

I'm hoping this might also lead to correct progress reports and ETA (in
the cases it's not zillions of years) as a spin-off.

>      // A table of coefficients required to compensate for fixed characters in
>      // JtR character files. These were found using Mathematica integer sequence
>      // analsysis.
>      static const int64_t kOrderCoefficients[][8] = {
>          { 0,   0,   0,   0,   0,   0,   0,   0 },
>          { 0,  -1,   0,   0,   0,   0,   0,   0 },
>          { 0,  -2,   1,   0,   0,   0,   0,   0 },
>          { 0,  -3,   3,  -1,   0,   0,   0,   0 },
>          { 0,  -4,   6,  -4,   1,   0,   0,   0 },
>          { 0,  -5,  10, -10,   5,  -1,   0,   0 },
>          { 0,  -6,  15, -20,  15,  -6,   1,   0 },
>          { 0,  -7,  21, -35,  35, -21,   7,  -1 },
>      };

This might be a silly question: Will this table be correct for any
charset file, or just all.chr distributed with Jumbo-5?

> If exposing this via a supported interface (e.g. the config language, or
> a command line option) has a chance of making the official distribution,
> let me know and I would be happy to write patch.

I'm looking forward to Solar's answer to all this. Unless he sees
problems we don't, it sure would be welcome.

> p.s. I also have a sha-1 implementation that's a little faster than the
> jumbo version, would this be the right list to send that to? Is there a
> jumbo cvs repo I can checkout to patch against?

Like Dhiru said, the repo is
https://github.com/magnumripper/magnum-jumbo and the branches are
currently magnum-jumbo (patches upon 1.7.9-Jumbo-5) and bleeding-jumbo
(the same, but rebased on Solar's CVS, currently 1.7.9.4, plus some
"less stable" code). Just send [git format-]patches to this list, or
send me a pull request. Actually in this case a patch here might be
better, so Solar can review it with ease before I merge it, and you/we
can discuss it on this list.

Oh, and there's this wiki page: http://openwall.info/wiki/john/patches -
you might want to put large-ish patches there and reference to it for
discussions.

magnum

```

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