Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Tue, 6 May 2014 10:59:11 -0700 (PDT)
From: deepika <anamikaspace@...oo.com>
To: "john-dev@...ts.openwall.com" <john-dev@...ts.openwall.com>
Subject: Re: Oracle Bitslice DES

Hi, I am working on oracle DES bitslicing. Since I have to use CBC encryption mode here, what should be the right strategy of doing it? In this statement in oracle_fmt_plug.c, memcpy((char *)cur_salt + salt_length, &cur_key[i][0], key_length[i]); salt is being appended with key. Here keys can have different lengths. So for n cur_salt's, should I find out the maximum length L and make other n-1 cur_salt's equal to L by appending 0's or if something else need to be done?

I doubt if doing as above will be right, as I read that DES_ncbc_encrypt() zero pads the last block of input to make the total length of input multiple of 8. Lengths of  n cur_salt's should be thus some multiple of 8 rather than equal to some L. For doing bitslicing I need all inputs to be of same length, which ofcourse won't happen here. How to tackle this issue?

On Friday, 31 January 2014 3:24 PM, Erik Winkler <ewinkler@...ls.com> wrote:
 
Deepika,

The Oracle DES format would be my first choice as it will see quite a substantial performance gain.  Dennis Yurichev has written separate code that does just that (see http://conus.info/utils/ops_SIMD/) with Oracle DES hashes with over 40M hashes/s as the noted performance.

Thanks,
Erik



On Jan 31, 2014, at 12:58 AM, deepika dutta <deepikadutta_19@...oo.com> wrote:

Hi magnum and erik,
>
>
>If you want I can do slicing of other formats too. Yes now it is going to be quite a straightforward task. You can provide me a list of formats to bitslice (in order of immediate need). I will begin working on it.  
> 
>Cheers,
>Deepika
Content of type "text/html" skipped

Powered by blists - more mailing lists

Your e-mail address:

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