Intel 810 Random Number Generator

[nathan.zook en amd.com]

I have recieved a communication from a prefer-to-remain-annonymous source
that the press report I recall reading indicating a table-indexed function
is erroneous.  This source asserted that the data is run through a von
Neumann-styled whitener.  (The von Neumann whitener is the famous 00-> nil
01-> 0 10-> 1 11->nil technique).  I say von Neumann-styled because it
generates on average 1 bit for every six input instead of 1 for every four
input.  The problem with such a whitener is that it only guaranteed to work
if successive samples are time-domain independent.  On a square-wave
generator such as a computer motherboard, this is not the case.  As has been
mentioned, this is HIGHLY dependent upon board layout, and is likely to
change with board revisions.

Let me be clear: given that the RNG sets a register indicating data "ready",
there is a very good probablility that the thing does contain entropy.  The
only question is how much to tell /dev/random.  Without the design
specifications, it is hard to guess.  (With them, it would still be a tough
job, beyond most of us on this list.)

Nathan




-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo en vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/