Re: [PATCH 1/3] dt-bindings: gpio: sunxi：create a DT header for Allwinner pin controller

[Samuel Holland <samuel@sholland.org>]

Maxime,

On 7/17/20 11:07 AM, Maxime Ripard wrote:
> Hi!
> 
> On Wed, Jul 15, 2020 at 07:54:12PM +0800, Frank Lee wrote:
>> From: Yangtao Li <frank@allwinnertech.com>
>>
>> The sunxi gpio binding defines a few custom cells for its gpio specifier.
>> Provide bank name for those.
>>
>> Signed-off-by: Yangtao Li <frank@allwinnertech.com>
> 
> Thanks for working on this, I wanted to do it at some point but it kept
> getting pushed further into my todo list.
> 
>> ---
>>  include/dt-bindings/gpio/sunxi-gpio.h | 29 +++++++++++++++++++++++++++
>>  1 file changed, 29 insertions(+)
>>  create mode 100644 include/dt-bindings/gpio/sunxi-gpio.h
>>
>> diff --git a/include/dt-bindings/gpio/sunxi-gpio.h b/include/dt-bindings/gpio/sunxi-gpio.h
>> new file mode 100644
>> index 000000000000..c692b4360da6
>> --- /dev/null
>> +++ b/include/dt-bindings/gpio/sunxi-gpio.h
> 
> So generally we've been using the compatible name as the file name. You
> should follow that convention too, and since it was added with the A10,
> using the A10 compatible.
> 
>> @@ -0,0 +1,29 @@
>> +/* SPDX-License-Identifier: GPL-2.0 */
>> +/*
>> + * GPIO definitions for Allwinner SoCs
>> + *
>> + * Copyright (C) 2020 Yangtao Li <frank@allwinnertech.com>
>> + */
>> +
>> +#ifndef _DT_BINDINGS_SUNXI_GPIO_H
>> +#define _DT_BINDINGS_SUNXI_GPIO_H
>> +
>> +#include <dt-bindings/gpio/gpio.h>
>> +
>> +/* pio */
>> +#define PA	0
>> +#define PB	1
>> +#define PC	2
>> +#define PD	3
>> +#define PE	4
>> +#define PF	5
>> +#define PG	6
>> +#define PH	7
>> +#define PI	8
>> +
>> +/* r-pio */
>> +#define PL	0
>> +#define PM	1
>> +#define PN	2
>> +
>> +#endif /* _DT_BINDINGS_SUNXI_GPIO_H */
> 
> Maybe we can go one step further and use a macro to have something like
> PIN(A, 12) ?

Since we have separate cells for the bank and pin, I don't think it would be
appropriate to have a single macro generating both. And I'm not sure what the
benefit of the macro would be, if all it does is forward its arguments. Are you
concerned that P[A-M] could conflict with something else in the device tree?

Cheers,
Samuel