ORBIT-USER: ORBIT Node kernel version

Max Ott max at semandex.net
Wed Oct 11 11:40:09 EDT 2006


Luis,

The reason Rodney is trying to build his own kernel is our size
constraints. We have nodes with only flash memory so we want to make
the image as small as possible. The default orbit image is way too big
for us. We also want to use things like unionfs so we can wipe out all
experiment changes through a simple reboot without having to image the
node (our control channel is over WiMax and not a nice 1Gbps link).

In other words are there some instructions on how to build a kernel
which works for a specific libmac and madwifi distro.

Thanks,

-max

On 10/11/06, Luis R. Rodriguez <mcgrof at gmail.com> wrote:
> Rodney,
>
> On 10/10/06, Rodney Berriman <Rodney.Berriman at nicta.com.au> wrote:
> >
> >  and now that I realise madwifi is not in the linux kernel,
>
> It's because of the proprietary HAL. Don't fret though -- there are
> attempts to fix this ;)
>
> > then I guess it's just the last question:
> >
> > Is there a patch available, or a set of instructions to patch the madwifi driver
> > for libmac?
>
> Better yet use our image, created Auguest 08, 2006:
>
> wireless-dev-madwifi-libmac.ndz
>
> This is based on madwifi-0.9.1. Madwifi is a perfect example of
> bastardization of ioctl() through Linux Wireless Extensions though. So
> to summarize, in order to support Kishore's libmac through the old
> ioctl() mechanisms we were forced to remove WMM extensions from
> madwifi. The proper solution really is to ditch madwifi altogether but
> that won't happen for a bit longer. If you need WMM we can prep up a
> configfs solution for you rather soon.
>
> Again, our list of images are available at:
>
> http://orbit-lab.org/wiki/Documentation/SupportedImages
>
>  Luis
>


-- 
Dr. Max Ott
Research Program Leader - Network and Pervasive Computing, NICTA Australia
Founder & CTO, Semandex Networks
Research Professor, WINLAB, Rutgers University



More information about the orbit-user mailing list