3

Currently I am using haveged on my server as source of entropy.

My Server is used as KVM hypervisor, to run virtual machines.

I did not use haveged at the beginning, and I noticed the VMs were draining the entropy pool from the server. Sometimes, when VMs were started SSH waited for enough entropy (to generate session keys, I guess).

Now with haveged, I don't have this problem anymore.

But I would like to try to use a HW random number generator. I am not saying haveged is bad, but true HW random number generator can only make the entropy better. I have seen some HW RNG which work on basis of Geiger counter, some which collect noise from microphone, and so on.

Which are most reasonable to use ? Could somebody perhaps recommend some specific one ?

Ideally, I would like it to be connected over serial port. Second best would be over USB.

  • Doesn't the processor on your host have a random generator built in (RDRAND)? – Gilles 'SO- stop being evil' Nov 12 '19 at 17:32
  • @Gilles 'SO- stop being evil' - I think the processor has RNG, but linux does not use it. – Martin Vegter Nov 13 '19 at 1:43
-1

Haveged has already been contested and does not guarantee TRNG.

You should get a hardware certified device like QRNG . It is a certified device and has passed many compliance tests including NIST SP800-22.

You can also find an USB stick device which claims it passes tests like Dieharder, ENT, Rngtest, FIPS-140-2.

These 2 should definitely be better than Haveged.

Other things you mentioned like Geiger or noise are also determinable/predictable.

| improve this answer | |
  • 2
    "Geiger or noise are also determinable/predictable." I would like to know how Geiger is predictable.For noise, it depends on what kind of system you're imagining. – Future Security Nov 12 '19 at 16:00
  • Sources used like an Americium 241 pellet have very determinable output. Unless you work in a lab with hundreds of different sources, you will be able to determine a Geiger's response. – Overmind Nov 13 '19 at 8:04
  • Citation needed on your claims about haveged (not in VMs) and Geiger counters. – A. Hersean Nov 13 '19 at 13:21
  • You may be able to determine the general rate of decay, but it is fundamentally impossible to predict a (sensitive) Geiger's response with sufficient precision to make it useless as an entropy source. – forest Nov 17 '19 at 23:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.