leds-gpio: no GPIO labels (in debugfs)

leds-gpio: no GPIO labels (in debugfs)

[Alexander Stein]

Hi,

I jut noticed that the used GPIOs for leds-gpio do not have an actual label 
which is visible in debugfs (/sys/kernel/debug/gpio):
> gpiochip4: GPIOs 368-383, parent: spi/spi0.2, mcp23s17, can sleep:
>  gpio-369 P0.1 (?           ) out hi
>  gpio-370 P0.2 (?           ) out lo
>  gpio-371 P0.3 (?           ) out lo
>  gpio-372 P0.4 (?           ) out lo
>  gpio-373 P0.5 (?           ) out lo
>  gpio-374 P0.6 (?           ) out lo

The reason is that the call to devm_get_gpiod_from_child() in 
gpio_leds_create() eventually calls "gpiod_request(desc, NULL);" in 
fwnode_get_named_gpiod. So no label attached to GPIO, hence "?" output.

As we have already a gpiod in create_gpio_led(), devm_gpio_request_one() with 
the proper label is not called.
I don't know much of that code, so what would be needed to have a proper GPIO 
label? And how?

Best regards,
Alexander Stein
-- 
Dipl.-Inf. Alexander Stein
SYS TEC electronic GmbH
alexander.stein@systec-electronic.com

Legal and Commercial Address:
Am Windrad 2
08468 Heinsdorfergrund
Germany

Office: +49 (0) 3765 38600-0
Fax:    +49 (0) 3765 38600-4100
 
Managing Directors:
	Director Technology/CEO: Dipl.-Phys. Siegmar Schmidt;
	Director Commercial Affairs/COO: Dipl. Ing. (FH) Armin von Collrepp
Commercial Registry:
	Amtsgericht Chemnitz, HRB 28082; USt.-Id Nr. DE150534010

Re: leds-gpio: no GPIO labels (in debugfs)

[Jacek Anaszewski]

Hi Alexander,

On 11/30/2016 05:24 PM, Alexander Stein wrote:
> Hi,
>
> I jut noticed that the used GPIOs for leds-gpio do not have an actual label
> which is visible in debugfs (/sys/kernel/debug/gpio):
>> gpiochip4: GPIOs 368-383, parent: spi/spi0.2, mcp23s17, can sleep:
>>  gpio-369 P0.1 (?           ) out hi
>>  gpio-370 P0.2 (?           ) out lo
>>  gpio-371 P0.3 (?           ) out lo
>>  gpio-372 P0.4 (?           ) out lo
>>  gpio-373 P0.5 (?           ) out lo
>>  gpio-374 P0.6 (?           ) out lo
>
> The reason is that the call to devm_get_gpiod_from_child() in
> gpio_leds_create() eventually calls "gpiod_request(desc, NULL);" in
> fwnode_get_named_gpiod. So no label attached to GPIO, hence "?" output.
>
> As we have already a gpiod in create_gpio_led(), devm_gpio_request_one() with
> the proper label is not called.
> I don't know much of that code, so what would be needed to have a proper GPIO
> label? And how?

 From the LED subsystem perspective everything looks OK both
in the case when label is defined and without it (node name is
used for the LED class device name then).

Since this seems to be an issue within GPIOLIB internals,
I'm adding Linus Walleij.

-- 
Best regards,
Jacek Anaszewski

Re: leds-gpio: no GPIO labels (in debugfs)

[Linus Walleij]

On Thu, Dec 1, 2016 at 9:45 AM, Jacek Anaszewski
<j.anaszewski@samsung.com> wrote:
> On 11/30/2016 05:24 PM, Alexander Stein wrote:

>> I jut noticed that the used GPIOs for leds-gpio do not have an actual
>> label
>> which is visible in debugfs (/sys/kernel/debug/gpio):
>>>
>>> gpiochip4: GPIOs 368-383, parent: spi/spi0.2, mcp23s17, can sleep:
>>>  gpio-369 P0.1 (?           ) out hi
>>>  gpio-370 P0.2 (?           ) out lo
>>>  gpio-371 P0.3 (?           ) out lo
>>>  gpio-372 P0.4 (?           ) out lo
>>>  gpio-373 P0.5 (?           ) out lo
>>>  gpio-374 P0.6 (?           ) out lo
>>
>> The reason is that the call to devm_get_gpiod_from_child() in
>> gpio_leds_create() eventually calls "gpiod_request(desc, NULL);" in
>> fwnode_get_named_gpiod. So no label attached to GPIO, hence "?" output.
>>
>> As we have already a gpiod in create_gpio_led(), devm_gpio_request_one()
>> with
>> the proper label is not called.
>> I don't know much of that code, so what would be needed to have a proper
>> GPIO
>> label? And how?
>
> From the LED subsystem perspective everything looks OK both
> in the case when label is defined and without it (node name is
> used for the LED class device name then).
>
> Since this seems to be an issue within GPIOLIB internals,
> I'm adding Linus Walleij.

The consumer label is entirely optional, it's just a very nice thing to
get right for debugging and feeling you have everything in check on
the system.

So it's not a bug, it is an optional feature.

If you want to fix it better looking I guess dig into the code :)

Yours,
Linus Walleij