[Cryptography] Dieharder & /dev/urandom

[Sidney Markowitz]

Michel Arboi wrote on 15/05/19 1:17 AM:
> I played with Dieharder (evolution of the famous Diehard statistics tests)
> There was a known bug wth 500 & 501 generators (/dev/random and /dev/urandom):

Here is an article which goes through a very thorough step by step analysis of 
an RNG using Dieharder to demonstrate how to interpret WEAK results by using 
them to refine the testing to get more certain results.

http://www.bitbabbler.org/test-data/dieharder.html

One important point of the article is that a result of PASSED does not mean 
that the RNG has passed a test, it means that it has with high probability not 
failed; a result of FAILED means that it has with high probability failed; and 
a result of WEAK indicates uncertainty in the results, not that the RNG is 
close to passing or close to failing a test. WEAK is an indication that you 
need to make the test stronger (e.g. with more p samples) until the 
uncertainty is resolved one way or another. The article is a demonstration of 
how to do that.