Alos Vandra:

> Thanks for your reply.

> I have two further questions I would like to ask:

> How can I make openssl read from that pipe?

It depends on exactly what you're asking. If you're talking about how to do
it in a program somebody else wrote, you'd have to ask them. If you're
talking about how to do it in your own program, you can replace the OpenSSL
PRNG with RAND_set_rand_method. If you're talking about how to do it in the
applications included with OpenSSL, use the 'RANDFILE' environment variable
There's built-in support for EGD, so it may be worth 'faking' EGS's
protocol.

> And why would a certified hardware quantum random generator hardware
> be less reliable than openssl's prng?

Because "reliable" is an attribute of the system as a whole, not its
individual components. When I read your question, I see, "Why would a system
I put together myself be less reliable than a system that has been publicaly
vetted and stood the test of time?" And the answer is obvious -- you might
make mistakes putting the system together and your shoulder is not being
looked over the way the OpenSSL project is.

My advice to you would be to compromise. A known property of OpenSSL's PRNG
is that it can never be hurt by adding seed material to it. So add as much
seed material to it as often as you like. This cannot possibly hurt, so your
baseline is the reliability of OpenSSL. (And, in fact, this is what I do.)

If you really want to, you can intercept OpenSSL's PRNG calls with your own
rand method. Keep track of how much entropy is being pulled out of the pool
and add back in the same amount from your own source. This way even if you
screw up the worst imaginable way (assuming you don't actually distort the
data from the PRNG going back to OpenSSL) you cannot make things any less
secure.

DS


______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
User Support Mailing List                    [email protected]
Automated List Manager                           [email protected]

Reply via email to