Re: PIII kernel support- hardware randomizer

Steve Underwood (steveu@infowebtelecom.com)
Sun, 17 Oct 1999 03:15:42 +0000


Alan Cox wrote:

> > Brian Hall <brianw.hall@compaq.com> said:
> > > I am currently using an overclocked Celeron 300 (now 450). The only real
> > > interest I have in the PIII and above is the hardware random number
> > > generator (using thermal noise). I have seen discussion of PIII kernel
> > > patches, but not about the random number generator. Does support for
> > > this exist or is it planned, and is there a repository for PIII patches?
> >
> > I'd be _very_ suspicious of something like that. To get a decent
> > distribution is probably quite hard and not done routinely. Besides, there
> > is _no_ way to check if it is misbehaving or feeding you a deterministic
> > sequence. Fine for games, perhaps.
>
> Its true the NSA might have the keys to the "RNG" or it might have magic
> switches. However if the rest of our RNG is secure then I can't see it
> being anything but useful (once info becomes available) to feed their
> hardware RNG into the entropy pool of /dev/random as another source

>From what I have seen (the usual rather vague information) the 820's random
number generator is ktb (thermal noise) based. In that case there are no keys.
You still shouldn't trust it, though. It means the chip is a mixed signal
(analogue + digital) one. The digital noise will probably get into the analogue
sampling, corrupt the results, and add structure to the randomness. The results
would vary from board to board, depending on both board layout and sample to
sample variations. The seems rather unlikely to be a Las Vegas grade random
generator.

Steve

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