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

[Amelie DELAUNAY]

On 5/22/19 10:41 AM, Lee Jones wrote:
> On Wed, 22 May 2019, Amelie DELAUNAY wrote:
>> On 5/22/19 7:48 AM, Lee Jones wrote:
>>> On Mon, 20 May 2019, Amelie Delaunay wrote:
>>>
>>>> When CONFIG_GPIO_OF is not defined, struct gpio_chip 'of_node' member does
>>>> not exist:
>>>> drivers/pinctrl/pinctrl-stmfx.c: In function 'stmfx_pinctrl_probe':
>>>> drivers/pinctrl/pinctrl-stmfx.c:652:17: error: 'struct gpio_chip' has no member named 'of_node'
>>>>        pctl->gpio_chip.of_node = np;
>>>>
>>>> Fixes: 1490d9f841b1 ("pinctrl: Add STMFX GPIO expander Pinctrl/GPIO driver")
>>>> Reported-by: kbuild test robot <lkp@xxxxxxxxx>
>>>> Signed-off-by: Amelie Delaunay <amelie.delaunay@xxxxxx>
>>>> ---
>>>>    drivers/pinctrl/pinctrl-stmfx.c | 2 ++
>>>>    1 file changed, 2 insertions(+)
>>>>
>>>> diff --git a/drivers/pinctrl/pinctrl-stmfx.c b/drivers/pinctrl/pinctrl-stmfx.c
>>>> index eba872c..bb64aa0 100644
>>>> --- a/drivers/pinctrl/pinctrl-stmfx.c
>>>> +++ b/drivers/pinctrl/pinctrl-stmfx.c
>>>> @@ -648,7 +648,9 @@ static int stmfx_pinctrl_probe(struct platform_device *pdev)
>>>>    	pctl->gpio_chip.base = -1;
>>>>    	pctl->gpio_chip.ngpio = pctl->pctl_desc.npins;
>>>>    	pctl->gpio_chip.can_sleep = true;
>>>> +#ifdef CONFIG_OF_GPIO
>>>>    	pctl->gpio_chip.of_node = np;
>>>> +#endif
>>>
>>> This is pretty ugly.  Will STMFX ever be used without OF support?  If
>>> not, it might be better to place this restriction on the driver as a
>>> whole.
>>>
>>> Incidentally, why is this blanked out in the structure definition?
>>> Even 'struct device' doesn't do this.
>>>
>> config PINCTRL_STMFX
>> 	tristate "STMicroelectronics STMFX GPIO expander pinctrl driver"
>> 	depends on I2C
>> 	depends on OF || COMPILE_TEST
>> 	select GENERIC_PINCONF
>> 	select GPIOLIB_IRQCHIP
>> 	select MFD_STMFX
>>
>> The issue is due to COMPILE_TEST: would "depends on OF || (OF &&
>> COMPILE_TEST)" be better ?
> 
> Linus would be in a better position to respond, but from what I can
> see, maybe:
> 
>    depends on OF || (OF_GPIO && COMPILE_TEST)
> 
> Although, I'm unsure why other COMPILE_TESTs haven't highlighted this
> issue.  Perhaps because they have all been locked down to a particular
> arch:
> 
> $ grep COMPILE_TEST -- drivers/pinctrl/Kconfig
> 	bool "Support pin multiplexing controllers" if COMPILE_TEST
> 	bool "Support pin configuration controllers" if COMPILE_TEST
> 	depends on OF && (ARCH_DAVINCI_DA850 || COMPILE_TEST)
> 	depends on OF && (ARCH_DIGICOLOR || COMPILE_TEST)
> 	depends on OF && (ARCH_LPC18XX || COMPILE_TEST)
> 	depends on ARCH_R7S72100 || COMPILE_TEST
> 	depends on ARCH_R7S9210 || COMPILE_TEST
> 	depends on ARCH_RZN1 || COMPILE_TEST
> 	depends on MIPS || COMPILE_TEST
> 
> What about adding this to your Kconfig entry:
> 
>    select OF_GPIO
> 

Yes COMPILE_TEST is combined with arch when depending on OF. But STMFX 
isn't arch dependent, it is just OF and I2C dependent.

Randy also see a build error in pinctrl-stmfx.c when CONFIG_OF is not 
set/enabled (randconfig):

../drivers/pinctrl/pinctrl-stmfx.c:409:20: error: 
âpinconf_generic_dt_node_to_map_pinâ 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


What is the prettiest way ?

>>>>    	pctl->gpio_chip.need_valid_mask = true;
>>>>    
>>>>    	ret = devm_gpiochip_add_data(pctl->dev, &pctl->gpio_chip, pctl);
>>>
>