[Arm-netbook] Port to 2.6.39.4 WIP

[Gordan Bobic]

Alejandro Mery wrote:
> Hi Alec,
> 
> El 12/01/12 10:47, Alec Ari escribió:
>> Hi! I'm more than half way done porting the Allwinner A10 kernel code
>> to a newer kernel release; 2.6.39.4. I am making changes to the code
>> only when necessary and trying to keep things as close to upstream as
>> possible, making future updates easier and less of a hassle. I was
>> wondering if it would be OK to push the changes into the custom head
>> I have created; lichee/for-next, that way at least developers can
>> look it over and make more changes where necessary. I'm also removing
>> an absolutely insane amount of whitespace at the same time during the
>> port, along with a lot of code clean-up.
> 
> may I ask why 2.6.39? 3.0 is old but at least it's LTS...

Well, I'm actually quite interested in backporting it to 2.6.32, since 
that is what the RHEL6 kernel is based on. Having as many SoCs supported 
in the same kernel would be really nice, for my project at least. :)
I have Kirkwood support on the RHEL6 kernel working (SheevaPlug was 
already there in vanilla 2.6.32, but I had to patch in GuruPlug and 
DreamPlug myself). I know for a fact there is a 2.6.32 kernel for Tegra 
for AC100 (which means it'll probably work on TrimSlice, too, with small 
or no changes) so I'm hoping to backport that into the RHEL6 kernel, as 
well. If the A10 support could be backported, too, that would cover the 
majority of the SoCs I'm hoping to get my RHEL6 ARM port running on at 
the moment (hardware availability allowing [hint, nudge]).

> Tom is trying to convince his manager to make their internal 2.6.36 git 
> repo (with history!) public so I'm holding my own clean up and porting 
> to master until there is an answer on that. It might make the process 
> much easier if we can see and understand how was it developed, and will 
> also allow us to preserve authorship correctly.

That'd be quite handy for forward/back-porting.

Gordan