[Cryptography] [RNG] on RNGs, VM state, rollback, etc.

[Kent Borg]

On 10/20/2013 05:28 AM, ianG wrote:
> Good example.  I'm going to get off the fence and say that the RNG 
> should never block.

The RNG should be configurable to block.

In the case of Linux's urandom Ted suggested blocking on bits-in and 
time, which ever comes first.  The question of defaults becomes key.

I suggest that the kernel's default values for these two parameters 
should be small enough that nearly no existing user is harmed by the 
change, yet many could benefit from not running on empty immediately 
after boot.

One question I have is who are typical first users of urandom after 
boot?  (That is, who will notice if the delay in seconds or bits is too 
large, what are they doing, when are they doing it?)

At the larger system level, these parameters could be set explicitly 
according to what that system is doing, how it is designed, etc.

-kb