Other Directory Sites: SeekWonder | Directory Owners Forum
The following archive was created by hippie-mail 7.98617-22 on Thu May 27 1999 - 23:44:21
mgraffam@idsi.net
Fri, 9 Apr 1999 11:39:28 -0400 (EDT)
On Fri, 9 Apr 1999 steven.soroka@mts.mb.ca wrote:
> > Is there any analysis of /dev/random used by Linux available?
>
> I doubt it would pass a statistical test in diehard.. if anyone really wants to know, msg me
> and I'll give it a shot and post the results..
I've done it. It passes with flying colors; as has been noted.. anything
hashed with SHA kicks Diehard's ass.. /dev/zero will too if processed with
SHA correctly.. how about:
1. Let M be a block of zero's
2. Hash M -> H
3. Output H
4. M = H + Zero's (+ denotes concatenation)
5. Goto 2
This will give excellent results when fed to any suite of tests.
The entropy pool of /dev/random is hashed with SHA prior to output..
/dev/urandom uses a feedback loop with SHA, IIRC, probably much like the
above.. so it will pass too.
Michael J. Graffam (mgraffam@idsi.net)
"Our merchants and master-manufacturers complain much of the bad effects
of high wages in raising the price, and thereby lessening the sale of
their goods both at home and abroad. They say nothing concerning the bad
effects of high profits. They are silent with regard to the pernicious
effects of their own gains." - Adam Smith, Wealth of Nations, Book I Chap. IX.
The following archive was created by hippie-mail 7.98617-22 on Thu May 27 1999 - 23:44:21