2012-03-08 22:25:40

by Martin Schleier

[permalink] [raw]
Subject: Re: Google Summer of Code 2012 - Application for the Linux Foundation submitted

Till Kamppeter <till.kamppeter@...> writes:

>
> On 03/08/2012 10:45 PM, Luis R. Rodriguez wrote:
> >
> > So one idea I had was working on open firmware for ath9k_htc. We don't
> > have the resources to open that up but can likely provide under NDA
> > source code to let it get cleaned up for public release. If this is
> > reasonable I'll throw it on.
> >
>
> OK, please do it.
>
> Till

Wait wait wait... qualcomm/Atheros does not have the resources, so how on earth
should a student (or students) actually do that?

Given the TODO over at
<http://wireless.kernel.org/en/users/Drivers/ath9k_htc#TODO> [Scroll down till
you get "fixed in the firmware. ]. This would be at least one years of full-time
work for a small team of trained code-monkey and therefore this "project" is
highly unrealistic.

Maybe we should stick to something less complicated.

Like extending PID to use MCS rates, or implementing a TPC control algorithm
[this is stuff, every drivers could use].




2012-03-08 22:33:51

by Luis R. Rodriguez

[permalink] [raw]
Subject: Re: Google Summer of Code 2012 - Application for the Linux Foundation submitted

On Thu, Mar 8, 2012 at 2:25 PM, Martin Schleier <[email protected]> wrote:
> Till Kamppeter <till.kamppeter@...> writes:
>
>>
>> On 03/08/2012 10:45 PM, Luis R. Rodriguez wrote:
>> >
>> > So one idea I had was working on open firmware for ath9k_htc. We don't
>> > have the resources to open that up but can likely provide under NDA
>> > source code to let it get cleaned up for public release. If this is
>> > reasonable I'll throw it on.
>> >
>>
>> OK, please do it.
>>
>>     Till
>
> Wait wait wait... qualcomm/Atheros does not have the resources, so how on earth
> should a student (or students) actually do that?
>
> Given the TODO over at
> <http://wireless.kernel.org/en/users/Drivers/ath9k_htc#TODO> [Scroll down till
> you get "fixed in the firmware. ]. This would be at least one years of full-time
> work for a small team of trained code-monkey and therefore this "project" is
> highly unrealistic.

It didn't take one year to open up the ar9170.fw [0] which lead to
carl9170.fw [1] and we even had two drivers from this work. I think
you underestimate our potential ;)

[0] http://wireless.kernel.org/en/users/Drivers/ar9170.fw
[1] http://wireless.kernel.org/en/users/Drivers/carl9170.fw

> Maybe we should stick to something less complicated.

Lets not undermine potential.

> Like extending PID to use MCS rates, or implementing a TPC control algorithm
> [this is stuff, every drivers could use].

PID? Hah, PID is dead. And TPC -- well... that requires a long answer.
Just finding documentation on TPC and what vendors do for TPC is as
much hard work as writing open firmware. But hey, its a wiki, feel
free to add what you think might be good, you can, and please be
verbose ;)

http://wireless.kernel.org/en/developers/GSoC/2012

Luis