[Olsr-users] Olsr-users Digest, Vol 6, Issue 1

[John Clark]

Bernd Petrovitsch schrieb:
> On Die, 2007-11-27 at 11:02 +0100, Aaron Kaplan wrote:
>   
>> hm, well at that time there apparently was only that one SoC which  
>>     
>
> I'm pretty ignorant (read: it's low on my priority list) with OLPC but
> which month/year was "that time"?
>
>   
>> could continue to mesh layer 2 frames while everything else was  
>> suspended (CPU, RAM, flash hard disc, ...).
>>     
>
> Uuuuh, I hope there is no hidden agenda somewhere which brought this
> feature request into the requirements specification to force everyone to
> use the single only possible solution.
> Well, time will reveal ......
> SCNR ...

Atheros and Broadcom both have ARM/MIPS processor based wireless chips. 
I don't know what
the current status of the Madwifi/FreeBSD Sam Leffler based HAL to 
support the chips that have
the processor on them. Who knows... maybe the Marvell chip is the 
Atheros chip underneath...

I've been 'tracking' if that is the right word, several types of chips 
for nearly a decade that have
used some form of microprocessor to control. Along time ago, there was 
the Alteon Gigabit Ethernet
board that used two MIPS processors. The 'code' was made available to 
the universe, and even
after Alteon divested themselves of their PCI board products, people 
where able to 'support' the
firmware without having to go to Alteon... To my knowledge, Alteon never 
manufactured actual
silicon, so there was always some 'hidden' foundry that did...

That particular product was probably the only product that had such 
openness on the firmware.

Since that time, most chip manufacturers have made deals with 'big 
computer manufacturers' and
of course, computer maufacturers who support Windows... and because 
those markets can put the
management in very comfortable retirement plans... there is no need to 
worry about the less than
1% of the market that would want access to hardware/firmware details.

In the case of the Wireless devices however, there is the added item of 
FCC (or other country regulatory
agencies), and the FCC's requirement of disallowing 'fiddling' with the 
radios and circumventing any
limits on power/spectrum usage of the device. Of course since the power 
can be limited in hardware,
it doesn't take a rocket scientist to understand that for the most part, 
the chip manufacturers are
probably hiding behind that sort of regulation to avoid further 
discussion on opening up the 'box'...

In terms of 'open' Atheros has been one of the most 'open' in the sense 
they have supported Sam Leffler
and his minons in creating a closed HAL with hooks for *BSD and Linux, 
for chips that are based
on the PCI interface.

Unfortunately that process doesn't seem thave gone over to the SoC type 
Atheros devices...

Oh, and as a final note... Broadcom has been particularlly unhelpful in 
the past to the open source community.

Also, and this is a conjecture, Marvell seems to be one of those types 
of manufacturers which buy up
'intellectual property' from others. The Marvell Gig-E chips are 
definately decendents from some other
company's IP. Now it could be that Marvell was the silicon foundary or a 
middle man in the IP holder to
Foundary path... but the company does not seem to be a primary inovator 
in technology...

John Clark.