Wifi again

Martin Lefkowitz lefko at sbcglobal.net
Sun Feb 18 17:12:28 CET 2007

They've gotten GSM to the point to where it uses a standard AT command
set (I think it's standard).  This means the driver is very simple.  The
current Wifi chips have a different approach.  They leverage the
resources on the host (i.e. scatter gather) and have the upper layer MAC
functions in the driver.  There is no standard interface there. I have
not looked at what Intel did (I would be surprised if it were any
different), but what both Atheros and Broadcom have done is have a
binary that handles the interface for the chip that needs to be included
in the opensource project.  I know with Atheros it taints the kernel.  I
know it used to be a pain in the ass to build on not as popular distros,
don't know what it's state is today though.

BTW it's interesting that the HTC uses the ACX100.  I actually designed,
and developed the first version of, that interface for TI.  That one is
different than the others in that the ACX100 (at least when I was
involved in it around 2000-02) was that it was so cheap for TI to put an
ARM core into a chipset, as well as the fact that at the time the 802.11
standard was not as mature as it is today (and it's still not mature
today).  802.11e was no wheres near complete.  That the ARM was used to
hide the registers/low level real time functions and  hardware functions
from the host --  that was my idea anyway.  I have not looked at the
opensource project but there should only be a software to software
interface necessary during it's operational phase.


> Date: Sat, 17 Feb 2007 19:29:25 -0800
> From: "Wolfgang S. Rupprecht"
>> > This has been discussed many, many times in the last month.  According
>> > to Sean, no wifi chips that both are low-power and have open specs.
> It sounds like the GPS and GSM chips aren't open spec either.  That
> didn't stop them from being included.
> I wonder if openmoko could benefit from using whatever the OLPC (One
> Laptop Per Child) folks are using.  They are also going for low-power,
> but perhaps their definition of low-power is quite a bit higher than a
> cell phone's definition of low power.
> -wolfgang
> -- Wolfgang S. Rupprecht http://www.wsrcc.com/wolfgang/

More information about the community mailing list