[Soekris] Wireless drivers

Stuart Henderson stu at spacehopper.org
Fri Oct 29 09:48:40 UTC 2004


While not strictly Soekris-related, it's strongly associated and I
think it may be interesting to any list members who haven't
noticed... 

Short summary: email campaign has succeeded in getting free
distribution rights for Symbol/Zydas/Atmel firmware, Intersil and
Intel in discussion (Intel needing more encouragement from users), TI
still to open a dialogue.


Many recent wireless chipsets have no onboard flash memory to hold
firmware, so obviously this needs to be downloaded each time they're
initialised. This is handled by the driver (e.g. Windows drivers come
with the firmware in a file). 

Redistribution rights for this firmware haven't been available, so in
the cases where a free driver exists, the firmware must usually be
extracted by the user from a Windows machine.

In past days it would be something that's already programmed into the
wireless card, but nowadays as a mass-market product the cost of
flash chips makes quite a difference over a production run, and I
think it's probably easier on unskilled users to install a driver
update than it is for them to flash a new firmware. (Not that there's
technically much difference, but perhaps the concept is easier to
grasp).

In the last week or so OpenBSD have been very active about persuading
wireless chipset vendors to free up firmware for distribution with
free OS.

(This isn't about open-sourcing the firmware: while that might be
nice, it's not necessary to have the drivers working, and since it
doesn't run on the host it's less of a problem to not have code. It's
just about the right to put an already-distributed binary firmware
into distributions other than the Windows drivers - drivers for some
SCSI cards and other hardware work this way).

Things started rolling after a pair of Centrino drivers were
committed to the tree, obviously without the necessary firmware as
redistribution wasn't allowed. Shortly after, Theo posted a list of
Intel email addresses to try and open a dialogue, which seems like
could be happening now.

Requests were then made to contact other vendors, again with some
email addresses supplied.

Current status seems to be that Intel may still need a little
encouragement, and there has been no response from TI about the
ACX100 firmware, so those companies still need some attention. 

If you want to follow this but don't want to sift through mailing
lists, it's being tracked on undeadly.org. Some recent posts from
Theo de Raadt on the openbsd-misc list:

> I am happy to announce that my discussion with Atmel has gone very
> well.  A firmware license has been decided on that is acceptable for
> all free operating systems.
> 
> Handled:
> Symbol
> Zydas
> Atmel
> 
> Terms being discussed and worked on
> Conexant/Intersil Prism GT (FullMAC)
> Intel (yes they called me, but keep up the mails please)
> 
> No response yet from vendor:
> TI
> 
> Will be handled in another creative way:
> Atheros
> 
> Thanks for your help guys.  Activism works.


>> Anyone who mailed to TI before, I think we have a better email
>> address now.
>> 
>> Re-examining how their company appears to be organized, I believe
>> that the person who correspondence should be directed at is
>> rrobertson at ti.com
> 
> That should be
> 
> rroberson at ti.com
> Randy Roberson
> GM of WNBU (2004-06-03)
> 
>> I believe he is the GM of the WNBU (wireless group)
>> 
>> Good luck.


And there's a list of relevant Intel addresses on the article at:
<http://undeadly.org/cgi?action=article&sid=20041027193425&mode=expanded>

This is a different situation to Atheros where the binary code runs
on the host CPU as part of the kernel - I started writing an
explanation of this but then I found <http://tinyurl.com/46rw5> which
describes it fairly well.




More information about the Soekris-tech mailing list