If memory serves me right, Dinesh Nair wrote:
> On 11/22/05 14:08 Bruce A. Mah said the following:
> > I was wondering if you're still doing the m0n0bsd-style build? I've
> i believe the instructions for a build of m0n0bsd are still available on
> m0n0.ch. additionally, you may want to take a look at
> /usr/src/release/picobsd on a freebsd system to see if tweaking picobsd
> could assist.
I looked at picobsd once...I remember thinking that it wasn't quite what
I was after. Glancing quickly through the manpage it seems like it's
similar to m0n0bsd in that it has a compressed MFS root filesystem, and
therefore doesn't require the boot media after the kernel is loaded.
nanobsd (/usr/src/tools/tools/nanobsd) is designed to do a "normal"
UFS/UFS2 filesystem that lives on a flash drive or similar. This
requires a larger flash device (each copy of my root partition, as
configured, could fit in a 64MB device, maybe less if I put some effort
into crunching things down). However it is extremely easy to customize
because I don't have to deal with compressing and recompressing
filesystem images to make a change.
So for me, nanobsd seems to fit a little better with what I have in
mind. Mine's been up and running for a couple months now, and aside
from a memory leak in if_bridge with PF (recently fixed in RELENG_6) all
seems to work well.
> > easier to implement and use with the new if_bridge(4) driver, which was
> > ported from OpenBSD. For starters, this makes it possible to solve the
> > "can't do NAT and bridging on the same box" problem. It also integrates
> exactly my thought process. before embarking on this however, i'd like to
> be sure that usng FreeBSD 6.0 is what the m0n0 users want. since
> performance seems to be the main beef blocking a move to 6.0, the 1.3aX
> releases are prototypes to see if we can address the performance issue
> before converting the rest of the m0n0wall architecture to what freebsd 6.x
> has to offer. a lot of this would be necessary anyways as older features
> which m0n0wall currently uses are deprecated.
Right. I'm interested (as are, I'm sure, a lot of FreeBSD committers)
to see what the performance measurements turn up. For my setup,
downloads through my net4801 (filtering bridge with a few dozen PF
rules) can saturate my 6 Mbps DSL downlink. This is of course not a
very high bar! Anyways, once a few numbers have come in with the
prototype(s), I think it'd be useful to make a post to net@. If there
turn out to be some bottlenecks, that's the crowd to work with.