Re: Allwinner wip

2013-07-20 Thread Martin Pieuchot
On 20/07/13(Sat) 14:10, Artturi Alm wrote:
 Hi,
 
 work-in-progress-quality 'port' to add support for Allwinner A10, tested
 on cubieboard only.

Nice work.  For what I've seen this port is really similar to the beagle
one.  Do you think it's possible to merge your work into this port or is
there any technical issue?

M.



Re: Allwinner wip

2013-07-20 Thread Patrick Wildt

Am 20.07.2013 um 14:20 schrieb Martin Pieuchot mpieuc...@nolizard.org:

 On 20/07/13(Sat) 14:10, Artturi Alm wrote:
 Hi,
 
 work-in-progress-quality 'port' to add support for Allwinner A10, tested
 on cubieboard only.
 
 Nice work.  For what I've seen this port is really similar to the beagle
 one.  Do you think it's possible to merge your work into this port or is
 there any technical issue?
 
 M.
 

I guess the biggest issue would be the load address, which currently
is statically set in the kernel config.  Sure, we could add a new config
file for that SoC, but I'd rather not put multiple SoCs into the
beagle/omap SoC port.

I'm working on a generic arm(v7) arch, but that'll still take some time.



Re: Allwinner wip

2013-07-20 Thread Artturi Alm

On 07/20/13 15:20, Martin Pieuchot wrote:

On 20/07/13(Sat) 14:10, Artturi Alm wrote:

Hi,

work-in-progress-quality 'port' to add support for Allwinner A10, tested
on cubieboard only.


Nice work.  For what I've seen this port is really similar to the beagle
one.  Do you think it's possible to merge your work into this port or is
there any technical issue?

M.



That was my initial suggestion to Patrick, but I agree with him
that having separate configs is not the way to go.

This might also be beneficial if support for Allwinner A20 is
added, going MP in allwinner-arch alone without possibly having so
many Cortex-A9(pandaboard es) vs Cortex-A7(cubieboard2 w/A20 chip)
issues to solve, this is purely speculation thought, i have not
looked up on what kind of differences there is, nor if they would
be at '/arch/arm-level' anyway.

Inspired by Patrick's question about timer frequency, i found what was
missing from awtimer.c, and with USELOWSPEEDOSC + related changes,
times are finally reported correctly :)

I was also given a possible direction to go with uart problems, so the
horrible comdiff can be dropped i think, this is nr1 on my todo.

i think i need to rework all the diffs outside arch/allwinner(including
those that i did not send as they are not useful for anything as is)
too, as
i realized i had not rebuilt /usr/cross/allwinner as often as i updated
/usr/src and it seems not to build the cross-env for me anymore, will
sort this out with clean install or something after uart is working and
the code is backed up somewhere before even trying native compilation.

if anyone is really going to give it a try soonish, ask for current
copy off list, as anything i could send now is most likely old by
tomorrow morning.


-Artturi