On Wed, Aug 24, 2005 at 04:47:28PM -0700, Jonathan Stone wrote:
> If I recall correctly, the historic (MHLEN + MLEN + 1) dates back to
> when default MLEN was 128, and expected memory was much smaller.

I guessed so.

> But, I'd guess adding either "intermediate" mbufs or bnon-power-of-two
> mbuf-cluster support is more than you'd want to bite off right now?

Implementing the changes to the mbuf framework shouldn't be difficult.
But changing all the locations in the code to use the new features
will be more difficult.

About non-power-of-two mbuf-clusters:
A fellow worker and I have currently ported mbufs and NetBSD's IPSec to
an embedded operating system. And because the target platform is very
restricted on main memory we changed the mbuf cluster size from 2K
to 1600 bytes (enough for a full ethernet MTU and alignment). That
worked out of the box without any changes to the mbuf code.

	Kind regards

-- 
Matthias Scheler                                  http://scheler.de/~matthias/