Re: FreeBSD rnd bug

[Patrick Welche <prlw1%cam.ac.uk@localhost>]

On Fri, Feb 20, 2015 at 09:41:48AM -0600, J. Lewis Muir wrote:
> On 2/20/15 7:55 AM, Taylor R Campbell wrote:
> > Yes, for cprng_strong.  However, statistical tests on the output of
> > a cryptographic PRNG will not detect failure to seed it.  They will
> > detect only catastrophic bugs in the PRNG itself.  (They will also
> > sometimes spuriously fire, as is the nature of statistical tests on
> > uniform random data.)
> 
> See Dilbert's tour of accounting:
> 
>   http://dilbert.com/strip/2001-10-25
> 
> :-)

Thanks for sorting that out - for other reasons, I read rnd(4) yesterday,
and wondered:

         o   An utterance from an accounting troll who always says `nine' has
             zero bits of entropy.

"What troll?!"

P