at91: smc configuration using DT

[Nicolas Ferre]

Le 16/12/2014 00:10, Richard Weinberger a écrit :
> Am 15.12.2014 um 23:56 schrieb Arnd Bergmann:
>> On Monday 15 December 2014 23:47:48 Richard Weinberger wrote:
>>>
>>> I'm sure I'm missing something.
>>>
>>> One of my boards is still using board files and uses sam9_smc_configure() to
>>> setup NAND timings. I'd like to convert it to DT but I fail to see how to configure the smc.
>>> What is the device tree equivalent of sam9_smc_configure()?
>>>
>>>
>>
>> I think it's actually missing, see the series "memory: add Atmel EBI
>> (External Bus Interface) driver" from Boris Brezillon.
> 
> Thanks a lot for your quick reply, Arnd!
> 
> arch/arm/mach-at91/board-stamp9g20.c was removed by:
> 
> commit fb3642ebb52cfab8eaa52fea677abec4d92182a0
> Author: Nicolas Ferre <nicolas.ferre at atmel.com>
> Date:   Wed Nov 19 10:30:41 2014 +0100
> 
>     ARM: at91: remove at91sam9260/at91sam9g20 legacy boards files
> 
>     Remove old board files that use at91sam9260 or at91sam9g20 Atmel SoCs. The
>     device tree is mature on these SoCs. It must be used now.
> 
>     Signed-off-by: Nicolas Ferre <nicolas.ferre at atmel.com>
> 
> How can this DT work for my board without proper NAND timings?
> 
> If I need Boris's patch set first I consider commit fb3642ebb a bit rude as it breaks
> things.
> 
> *confused*,
> //richard

Richard,

This action was announced several months ago on the at91.com forum and
on this mailing-list: https://lkml.org/lkml/2014/9/10/293
Please read this announce message for more background.
Moreover, 3.19 is scheduled mid-february. It was also offered to help
during this transition period... So I would not call it "rude".

Yes, it definitively breaks things. But as noted by Arnd, we are
actively working on a clean solution.
In the meantime, you have the possibility to configure the SMC with
proper NAND timings in the bootloader(s).

I hope that it helps. Bye,
-- 
Nicolas Ferre