Re: Updated scalable urandom patchkit

["George Spelvin" <linux@horizon.com>]

> Can we just use my multi pool patch for now? It works and is actually scalable
> and does not require any new "cryptographic research" or other risks.

To clarify, it's actually quite easy to come up with something that works,
cryptographically.  All the discussion is:
1) Erring on the side of public discussion for a security-sensitive area, and
2) Trying to find "The Right Thing" among the many workable solutions.

Earlier, Ted seemed to be interested in more significant changes like
replacing sha_transform and "abuse mitigation mode", so I've been
thinking expansively.  The immediate plans aren't necessarily quite
as complex.

Mostly, it's an embarrassment of ways to do it, and trying to find
a reason to choose one over the other.  I wrote one very quickly, said
"I can do better", thought about coding that up, thought of an even
better solution, and decided to post the performance-affecting part
while I played donkey-between-two-piles-of-hay.

> or other risks.

Ted gets to decide, but my objection is that the already-sparse entropy
supply is considerably diluted.  That *is* a risk.