random(4) plugin infrastructure for mulitple RNG in a modular fashion

[Dag-Erling Smørgrav]

Mark R V Murray <mark at grondar.org> writes:
> OK - in the context of what is currently there, it makes less sense than
> that; loading RDRAND/Ivy and Nehemiah simultaneously is silly because
> they are different architectures, and only one can ever work on a particular
> box; so what happens is some script selects the wrong one? I suppose
> the probe is there to prevent this.

I don't know offhand whether we support them, but there are discrete
HWRNGs which might be present alongside an on-die HWRNG; and in all
cases, Yarrow and / or Fortuna may be present in the kernel alongside a
supported HWRNG.

> We still have the anachronism where the older hardware RNGs are turned
> into /dev/random devices and the newer ones supply their entropy to
> the software (Yarrow) for further processing.

Provided the HWRNG is of sufficient quality, the user should be allowed
to use it directly (through /dev/random) without Yarrow / Fortuna.  At
the same time, we do not want to lose the ability to feed their output
to Yarrow / Fortuna.  Plugging all {P,HW}RNGs into a common framework
makes it *easier*, not *harder*, to support both options.

DES
-- 
Dag-Erling Smørgrav - des at des.no