[Click] wifi and madwifi.stripped

[Bob Keyes]

On Thu, 12 May 2005, John Bicket wrote:

> I have been incorporating the changes from the madwifi.stripped
> driver into the actual madwifi driver.  As a result, I won't
> be maintaining the madwifi.stripped driver any longer.

Does this incorporate all the features need for RoofNet? so
madwifi.stripped is no longer required for roofnet, but a somehwat modern
version of madwifi? If so, what versions of madwifi have your changes?

> Instead of making the device driver send and receive raw 802.11
> packets on the normal device (ie. athX), I have added a second "phony"
> device to the madwifi driver that handles all of that (called athXraw).
> This allows the use of the normal 802.11 stack as well as the ability
> to run other protocols (ie roofnet) at the same time. If you want to
> bypass the normal 802.11 stack you can use monitor mode, which
> still allows packet transmission using the raw interface.
>
> The code is currently in the madwifi BSD branch, which you can checkout
> using:
> cvs -z3 -d:pserver:anonymous at cvs.sourceforge.net:/cvsroot/madwifi co -r
> BSD madwifi

Were these changes ever incorporated into the Linux version of madwifi, or
is the a canonical way to compile the BSD version of madwifi in Linux?
> You must use a sysctl to activate the rawdevice if you are running the
> driver:
> /sbin/sysctl -w dev.ath0.rawdev=1
>
> Also, you can make the rawdevice use headers in addition
> to just 802.11 frames. You can run
> /sbin/sysctl -w dev.ath0.rawdev_type=X
> where X is
> 0: raw 802.11 frames
> 1: prism headers (ie you can use PrismEncap/PrismDecap)
> 2: radiotap headers (ie you can use RadiotapEncap/RadiotapDecap)

These sysctls are available in my madwifi driver, but I am still getting
errors:

ToDevice(ath0) send: Message too long

and occasionally:

ToDevice(ath0) send: Network is down

and I want to make sure these are not due to using an incorrect version of
madwifi