[coyotos-dev] Some comments on Coyotos Microkernel Specification

[Jonathan S. Shapiro]

On Mon, 2005-12-05 at 14:53 -0700, Christopher Nelson wrote:
> > Christopher Nelson wrote:
> > 
> > >I fail to see how this disproves what I said.
> > >
> > Odds of 4294967296 to 1 are not "astronimical", they are 
> > squarely in the realm of possible for someone with some time 
> > on their hands and a spare computer.
> 
> I guess I don't understand your point.  What I'm saying is that
> SHA-1[64] is "sufficiently unique" for interface id generation.  What
> are you saying?
> 
> I also don't see where the 42949...96 number enters.  A 64-bit number is
> much larger than 4gigs of address. 

That isn't relevant. Let's back up and do this right.

First, as Dominique said, SHA-1 generates a 160 bit hash, so forget the
64 bit thing entirely.

Given a 160 bit cryptographic hash, you get about 2^80 reliably
non-colliding values. Yes, this is quite a lot.

But in any case the entire discussion is moot, because the *input* to
the hash is not taken from a non-colliding space, and the hash output is
a deterministic function of its input.

shap