On Fri, 1 Sep 2000, Mark Murray wrote:
> > > PC's are pretty low-entropy devices; users who need lots of random
> > > bits (as opposed to a steady supply of random numbers) are arguably
> > > going to need to go to extraordinary lengths to get them; their
> > > own statistical analysis is almost certainly going to be required.
> >
> > I claim this to be untrue: my tests show an ordinary sound card (with no
> > recording source, at maximum input gain) will provide far more
> > (high-quality) entropy than Yarrow can make use of under even the most
> > punishing loads.
>
> Kris; could you come up with some kind of proof-of-concept for this?
> I don't want to steal a sound card, but being able to use one at the
> user's request is good; I'd need kernel-mode code to do some harvesting.
I posted code and results a while back to freebsd-arch which sampled the
sound card from userland and analysed the shannon entropy of the noise
(for the record, all but one card I tried gave about 6 bits of entropy or
more per 16 bit sample with no recording device plugged in, and maximum
input gain, and could be sampled 44000 times per second or more. The other
card driver was probably broken). Thats what I'm basing this on, but I
haven't tried to make a kernel harvester.
Kris
--
In God we Trust -- all others must submit an X.509 certificate.
-- Charles Forsythe <[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
