Re: [PATCH] pinctrl: stmfx: Fix compile issue when CONFIG_OF_GPIO is not defined

[Amelie DELAUNAY <amelie.delaunay@xxxxxx>]

On 5/23/19 9:18 PM, Linus Walleij wrote:
> On Thu, May 23, 2019 at 4:11 PM Amelie DELAUNAY <amelie.delaunay@xxxxxx> wrote:
>> On 5/22/19 11:48 PM, Linus Walleij wrote:
>>> On Wed, May 22, 2019 at 11:21 AM Amelie DELAUNAY <amelie.delaunay@xxxxxx> wrote:
>>>
>>>> ../drivers/pinctrl/pinctrl-stmfx.c:409:20: error:
>>>> ‘pinconf_generic_dt_node_to_map_pin’ undeclared here (not in a function)
>>>>      .dt_node_to_map = pinconf_generic_dt_node_to_map_pin,
>>>>
>>>> OF_GPIO depends on OF.
>>>>
>>>> So either
>>>>        depends on OF || (OF && COMPILE_TEST)
>>>> or
>>>>        depends on OF || (OF_GPIO && COMPILE_TEST)
>>>>
>>>> and
>>>>
>>>>        select OF_GPIO
>>>
>>> I would use just:
>>>
>>> depends on OF_GPIO
>>>
>>> Because OF_GPIO already depends on OF, and
>>> compile tests will not work without OF_GPIO which
>>> require OF so...
>>>
>>> Besides it is what most other GPIO drivers do.
>>>
>>> So just keep that one line and drop the rest.
>>>
>>> Yours,
>>> Linus Walleij
>>>
>>
>> Ok so I can get rid of COMPILE_TEST ?
>>          depends on I2C
>>          depends on OF_GPIO
>>          select GENERIC_PINCONF
>>          select GPIOLIB_IRQCHIP
>>          select MFD_STMFX
> 
> Yep just like that.
> 
>> Because I've no arch to balance COMPILE_TEST. Or maybe something like
>>          depends on OF_GPIO && (OF || COMPILE_TEST)
>> even if OF_GPIO && OF is redundant ?
> 
> COMPILE_TEST is just to make something available for testing
> on other architectures, such as testing ARM-specific drivers
> on x86.
> 
> With just OF_GPIO as dependency, it will be compile tested anyways
> because x86 allyesconfig will enable OF and OF_GPIO, and also
> all the STMFX drivers.
> 
> Yours,
> Linus Walleij
> 

Thanks Linus for this clarification!

Regards,
Amelie