change in iwi-firmware operation

[Kevin Oberman]

> Date: Mon, 15 May 2006 07:44:34 -0600
> From: "Wilde, Donald" <dwilde at sandia.gov>
> Sender: owner-freebsd-mobile at freebsd.org
> 
> 
> 
> -----Original Message-----
> From: Frank Altpeter [mailto:frank at altpeter.de] 
> Sent: Monday, May 15, 2006 7:25 AM 
> To: Wilde, Donald Cc: freebsd-mobile at freebsd.org
> Subject: Re: change in iwi-firmware operation
> 
> Hi there,
> 
> Wilde, Donald wrote on 2006-05-15 at 15:12:13 CEST: 
> Hi, All - 
>  
>
> I have a Centrino (2200G) equipped Dell Inspiron 6000, and had it
> working reasonably well with the iwi firmware.  
>  
> I just updated to the version in ports as of a few days ago, and  
> discovered that the firmware won't stay firm.
> 
> Sure. The firmware is not permanent. It must be loaded on boot every
> time ...  I'm running ipw which provides an ipw.sh startup script in
> $PREFIX/etc/rc.d which has to enabled with ipw_enable="YES" - i
> assume the same fits for iwi driver...
> =========================
> =========================
> =========================
> 
> Thanks for the pointer, Frank,
> 
> Hmmm. Kinda defeats the meaning of 'firmware', and I'm going to start
> worrying about write-cycles of the FLASH. Dunno what's in a Centrino
> system, (anybody?) although it's likely Intel NOR FLASH, which should
> be good for 100K-cycles. Even so, WHY would one need to load FLASH
> every boot?

It is not hard on the flash as the iwi (and many other newer cards)
don't have flash at all. They use RAM, which is much less expensive. It
saves money, but means that the "firmware" must be loaded after every
power cycle.

It's a real pain and, because of the wording of the license, the
firmware may not be included in the FreeBSD distribution. That's why the
port to get the firmware. 

Note that this applies to Intel and Broadcom based cards, but others do
allow the inclusion of the binary in distributions, so you don't need to
install a port for the Atheros or most Asian parts.

With margins on these things is razor thin, saving a few cents on RAM
vs. flash can be the difference between a successful product and a
failure.
-- 
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman at es.net			Phone: +1 510 486-8634