On December 8, 2001 07:02 pm, Jeremy Fitzhardinge wrote:
> On Fri, 2001-12-07 at 07:51, Daniel Phillips wrote:
> > I did try R5 in htree, and at least a dozen other hashes. R5 was the
> > worst of the bunch, in terms of uniformity of distribution, and caused a
> > measurable slowdown in Htree performance. (Not an order of magnitude,
> > mind you, something closer to 15%.)
>
> Did you try the ReiserFS teahash? I wrote it specifically to address
> the issue you mentioned in the paper of an attacker deliberately
> generating collisions; the intention was that each directory (or maybe
> filesystem) have its own distinct hashing key.
Yes, I tried every hash in ReiserFS. Please have a look at Larry McVoy's
'linear congruential' hash in the bitkeeper code. It's decent. In fact, the
only good hashes I've found after trolling all over the internet are that
one, and the one I wrote, both based on combining a sequence of characters
with a well-known pseudo-random number generation technique.
-- Daniel - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Dec 15 2001 - 21:00:14 EST