Re: Extensible hashing and RCU

["Michael K. Edwards" <medwards.linux@gmail.com>]

On 2/20/07, Evgeniy Polyakov <johnpol@2ka.mipt.ru> wrote:
> How I like personal insults - it is always fun to read about myself from
> people who never knew me :)

On this occasion, I did not set out to insult you.  I set out to
suggest an explanation for why cooler and grayer heads than mine are
not falling over themselves to merge your designs into the mainline
kernel.  Did I succeed?

> I just shown a problem in jenkins hash - it is not how to find a
> differnet input for the same output - it is a _law_ which allows to
> break a hash. You will add some constant, and that law will be turned
> into something different (getting into account what was written, it will
> end up with the same law).

Correct.  That's called a "weak hash", and Jenkins is known to be a
thoroughly weak hash.  That's why you never, ever use it without a
salt, and you don't let an attacker inspect the hash output either.

> Using jenkins hash is equal to the situation, when part of you hash
> chains will be 5 times longer than median square value, with XOR one
> there is no such distribution.

Show us the numbers.  Salt properly this time to reduce the artifacts
that come of applying a weak hash to a poor PRNG, and histogram your
results.  If you don't get a Poisson distribution you probably don't
know how to use gnuplot either.  :-)

> Added somthing into permutations will not endup in different
> distribution, since it is permutations which are broken, not its result
> (which can be xored with something).

I can't parse this.  Care to try again?

Cheers,
- Michael