Re: [PATCH v2] leds: bd2802: Convert to use GPIO descriptors

[Dan Murphy <dmurphy@xxxxxx>]

Linus

On 12/8/19 6:42 PM, Linus Walleij wrote:
> The Rohm BD2802 have no in-kernel users so we can drop the
> GPIO number from the platform data and require users to
> provide the GPIO line using machine descriptors.
>
> As the descriptors come with inherent polarity inversion
> semantics, we invert the calls to set the GPIO line such
> that 0 means "unasserted" and 1 means "asserted".
>
> Put a note in the driver that machine descriptor tables
> will need to specify that the line is active low.
>
> Cc: Kim Kyuwon <chammoru@xxxxxxxxx>
> Signed-off-by: Linus Walleij <linus.walleij@xxxxxxxxxx>
> ---
> ChangeLog v1->v2:
> - Ask simply for a GPIO descriptor named "reset", drop the
>    idiomatic naming from the code.
> ---
>   drivers/leds/leds-bd2802.c  | 27 ++++++++++++++++++---------
>   include/linux/leds-bd2802.h |  1 -
>   2 files changed, 18 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/leds/leds-bd2802.c b/drivers/leds/leds-bd2802.c
> index e7ec6bff2b5f..bd61a823d0ca 100644
> --- a/drivers/leds/leds-bd2802.c
> +++ b/drivers/leds/leds-bd2802.c
> @@ -10,7 +10,7 @@
>   
>   #include <linux/module.h>
>   #include <linux/i2c.h>
> -#include <linux/gpio.h>
> +#include <linux/gpio/consumer.h>
>   #include <linux/delay.h>
>   #include <linux/leds.h>
>   #include <linux/leds-bd2802.h>
> @@ -67,6 +67,7 @@ struct led_state {
>   struct bd2802_led {
>   	struct bd2802_led_platform_data	*pdata;
>   	struct i2c_client		*client;
> +	struct gpio_desc		*reset;
>   	struct rw_semaphore		rwsem;
>   
>   	struct led_state		led[2];
> @@ -200,7 +201,7 @@ static void bd2802_update_state(struct bd2802_led *led, enum led_ids id,
>   		return;
>   
>   	if (bd2802_is_all_off(led) && !led->adf_on) {
> -		gpio_set_value(led->pdata->reset_gpio, 0);
> +		gpiod_set_value(led->reset, 1);
>   		return;
>   	}
>   
> @@ -226,7 +227,7 @@ static void bd2802_configure(struct bd2802_led *led)
>   
>   static void bd2802_reset_cancel(struct bd2802_led *led)
>   {
> -	gpio_set_value(led->pdata->reset_gpio, 1);
> +	gpiod_set_value(led->reset, 0);
>   	udelay(100);
>   	bd2802_configure(led);
>   }
> @@ -420,7 +421,7 @@ static void bd2802_disable_adv_conf(struct bd2802_led *led)
>   						bd2802_addr_attributes[i]);
>   
>   	if (bd2802_is_all_off(led))
> -		gpio_set_value(led->pdata->reset_gpio, 0);
> +		gpiod_set_value(led->reset, 1);
>   
>   	led->adf_on = 0;
>   }
> @@ -670,8 +671,16 @@ static int bd2802_probe(struct i2c_client *client,
>   	pdata = led->pdata = dev_get_platdata(&client->dev);
>   	i2c_set_clientdata(client, led);
>   
> -	/* Configure RESET GPIO (L: RESET, H: RESET cancel) */
> -	gpio_request_one(pdata->reset_gpio, GPIOF_OUT_INIT_HIGH, "RGB_RESETB");
> +	/*
> +	 * Configure RESET GPIO (L: RESET, H: RESET cancel)
> +	 *
> +	 * We request the reset GPIO as OUT_LOW which means de-asserted,
> +	 * board files specifying this GPIO line in a machine descriptor
> +	 * table should take care to specify GPIO_ACTIVE_LOW for this line.
> +	 */
> +	led->reset = devm_gpiod_get(&client->dev, "reset", GPIOD_OUT_LOW);
> +	if (IS_ERR(led->reset))
> +		return PTR_ERR(led->reset);

What about EPROBE_DEFER?

Is the reset line optional or required?  I see that this device does not 
have DT docs and the call infers it is required.

But can this be optional?

Dan