[PLUG] Wifi card

[Russell Senior]

>>>>> "Alan" == Alan Olsen <alan.olsen at gmail.com> writes:

Alan> There has been a big pissing match over binary kernel blobs
Alan> v.s. binary firmware blobs.  Some people do not see the
Alan> difference.  Some claim that there is a difference.

There is a difference.  Binary firmware blobs run on the wireless
device and can possibly not work.  Binary kernel blobs run on your CPU
and can panic your box.

Alan> It is my understanding that the FCC is resisting allowing open
Alan> sourced drivers because people would be able to modify the
Alan> output power of the radios.  (Not that closed source is a
Alan> barrier given the right disassembler.)

The FCC rule requires hardware vendors to try to prevent end-users to
exceed mandated limits, which has been interpretted as prohibiting
their (the vendors) release of open source drivers.  Some devices,
software defined radios, are particularly impacted by this, Atheros
radios being an example.  My understanding is that the FCC rule is not
binding on people other than radio hardware vendors creating
open-source drivers.  So, open-source driver hackers are having to
figure a lot of stuff out on their own, which naturally slows things
down.

Alan> The biggest problem right now is that there is not one clear
Alan> interface for wireless drivers in the kernel.  There is that way
Alan> the kernel developers do it, there is the Madwifi people and the
Alan> hostap interface.  These need to be combined into one before
Alan> things will get any better.

There *is* one clear interface direction these days, and that's
mac80211.  Everything else is legacy.  Madwifi folks have said that
their development effort is being directed mostly towards the ath5k
(mac80211 stack) driver these days.  The problem is that mac80211
isn't quite ready for prime time, nor are many of the drivers that use
it.


-- 
Russell Senior, Secretary
russell at personaltelco.net