[PATCH v2 1/9] ata: at91: use syscon to configure the smc

[Nicolas Ferre]

Le 08/04/2015 14:00, Bartlomiej Zolnierkiewicz a écrit :
> On Wednesday, April 08, 2015 01:13:35 PM Alexandre Belloni wrote:
>> Hi,
>>
>> On 08/04/2015 at 13:04:19 +0200, Bartlomiej Zolnierkiewicz wrote :
>>> On Monday, March 23, 2015 08:29:07 PM Alexandre Belloni wrote:
>>>> Use syscon/regmap to configure the smc. This allows to avoid using
>>>> at91sam9_smc.h and to compile the driver in a multiplatform configuration.
>>>>
>>>> The driver will still not probe until the proper DT bindings are added. That
>>>> binding will include an atmel,smc property that is a phandle to the SMC the CF
>>>> controller is connected to.
>>>
>>> If the driver is currently working fine in !ARCH_MULTIPLATFORM
>>> configuration then this patch will make it non-functional until
>>> atmel,smc property is added to DT.  To prevent this and preserve
>>> bisectability the patchset should first add atmel,smc property
>>> and then convert pata_at91 driver to use it.
>>>
>>
>> Starting with 4.1, it will not be possible to use the driver anyway as
>> all the platforms using it have switched to multiplatform. This patch
>> makes it compilable again.
> 
> Hmm.  It seems that it was your commit which did the switch without
> converting all at91 specific code to be multiplatform ready first:
> 
>     From: Alexandre Belloni <alexandre.belloni at free-electrons.com>
>     Date: Fri, 13 Mar 2015 22:57:18 +0100
>     Subject: ARM: at91: switch to multiplatform
> 
>     Switch AT91 to multiplatform as all SoCs are properly handled.
> 
>     Signed-off-by: Alexandre Belloni <alexandre.belloni at free-electrons.com>
>     Signed-off-by: Nicolas Ferre <nicolas.ferre at atmel.com>
> 
> ?
> 
>> Anyway, it seems that there is little interest in that driver and nobody
>> I contacted has access to a board which can be used to test this.
> 
> If there are no users then probably the driver can be removed but this
> something that platform Maintainers should decide on.

This is why there are loose constrains on this driver and that we
decided to move on.
So I think that modifying it and introducing the DT property afterwards
can be done.

Bye,
-- 
Nicolas Ferre