On Thu, Dec 08, 2005 at 03:01:14PM -0800, Mike Cheponis wrote:
> Hi, I volunteered to contact the folks at MIT to offer them NetBSD for use 
> on the $100 World Computer, since I'm an MIT alum.
> 
> Here's the response.

Mike,

NetBSD is in a good position to supply important parts of the $100
Laptop's software.  I lead a project called CUWiN (www.cuwireless.net)
that makes open-source, NetBSD-based, IPv4/IPv6 wireless "mesh" routers,
favoring the Atheros chipset.  A couple of the routers we have around
town (http://www.meshsandbox.com/cuwin.html) use an AMD Geode processor.
It would be easy to port CUWiN's wireless routing to the $100 Laptop as
it is presently specified.

(Incidentally, the $100 Laptop project's goals are resonant with
CUWiN's.  Our principle funder is George Soros' Open Society Institute,
whose interest in wireless networking is for building communications
infrastructure in the developing world.  Our system is in-use in Apirede,
Ghana, and we anticipate a deployment in South Africa.)

I haven't missed JFFS2, but we use a (mostly) read-only FFS on a
CompactFlash card, which has wear-leveling built in.  I am more concerned
about kern/30525.

Dave

> ---------- Forwarded message ----------
> Date: Thu, 08 Dec 2005 17:37:50 -0500
> From: Jim Gettys <jg@freedesktop.org>
> To: mac@wireless.com
> Subject: OS Software for the $100 World Computer  (NetBSD)
> 
> Mike,
> 
> To begin with, a hearty welcome to NetBSD!
> 
> The machine is an open platform, for people to do as they will with, and
> NetBSD is certainly more than welcome to play, as is everyone else. To
> the extent possible, we'll make specifications available as soon as they
> solidify.  In some cases, some components may not yet be in the market,
> and you'll probably have to get NDA's to get the detailed specs (for
> example, we may choose to use a new version of the AMD Geode, for which
> specifications are not yet public).
> 
> To first order, you can get a good idea of the baseline (lower bound) of
> the hardware at http://www.gettysfamily.org/wordpress/. More exact
> details will be forthcoming as the machine design solidifies over the
> next few months (modulo the issue mentioned in the paragraph above).
> 
> The #1 challenge for NetBSD is that, as far as I know, none of the BSD's
> have an equivalent to the Linux jffs2 (journalling flash file system,
> V2) file system found in Linux: the machine is flash based, and wear
> leveling essential for general purpose use.  I recommend the BSD systems
> pool effort here if they can; this lack leaves *BSD maybe 2 years behind
> Linux on one really key technology, somewhat to my surprise (I only
> discovered this last week).  Googling for "flash file system" on the BSD
> portal was more than a bit disappointing.
> 
> #2: On the networking side, we expect to want/need/require mesh
> networking; this derives on Linux from the good work FreeBSD has done,
> as I understand it. The reason is we expect to drop bunches of systems
> into places like jungles in the developing world, and do not expect that
> there may be the expertise or any infrastructure: the systems should be
> able to talk to each other and be useful "out of the box".
> 
> #3 And, of course, for a battery (and crank) powered machine, power
> management is way up there too.
> 
> It is entirely possible we'll want to deploy on IPv6, given the scale of
> the project and the fact it is going into the developing world.
> 
> Over the next month or two, I hope we get better web and project
> infrastructure set up, but this should get you going in the right
> directions without more fussing around, and without having to wait for
> us to get that part of our act together.  This instant, we're focused
> almost entirely on solidifying the design for the first machine, and
> making sure the display will really work.
> 
> Feel free to forward this to the appropriate NetBSD mailing lists.
> 
> 			Best regards,
> 				Jim Gettys
> 				One Laptop Per Child
> 
> 

-- 
David Young             OJC Technologies
dyoung@ojctech.com      Urbana, IL * (217) 278-3933