Re: [PATCH v2 09/10] gpio: bd73800: Support ROHM BD73800 PMIC GPIOs

[Matti Vaittinen]

On 07/08/2026 23:29, Linus Walleij wrote:
> Hi Matti,
>
> thanks for your patch!
>
> On Tue, Aug 4, 2026 at 12:23 PM Matti Vaittinen
> <matti.vaittinen@xxxxxxxxx> wrote:
>
>
> > +static int bd73800gpio_get(struct gpio_chip *chip, unsigned int offset)
> > +{
> > +       struct bd73800_gpio *data = gpiochip_get_data(chip);
> > +       struct bd73800_gpio_pin_cfg *pin = &data->pin[offset];
> > +       int ret, val;
> > +
> > +       /* Only pins configured as GPI via OTP can have their status read */
> > +       if (pin->state != BD73800_PIN_GPI) {
> > +               dev_dbg(data->dev, "pin %d (%x) not input. State %d\n", offset,
> > +                       pin->mask, pin->state);
> > +               return -EINVAL;
> > +       }
> > +
> > +       ret = regmap_read(data->regmap, BD73800_REG_INT_5_SRC, &val);
> > +       if (ret)
> > +               return ret;
> > +
> > +       return val & pin->mask;
> > +}
> > +
> > +static int bd73800gpo_set(struct gpio_chip *chip, unsigned int offset,
> > +                         int value)
> > +{
> > +       struct bd73800_gpio *data = gpiochip_get_data(chip);
> > +       struct bd73800_gpio_pin_cfg *pin = &data->pin[offset];
> > +
> > +       if (pin->state != BD73800_PIN_GPO) {
> > +               dev_dbg(data->dev, "pin %d (%d) not output. State %d\n",
> > +                       offset, pin->mask, pin->state);
> > +
> > +               return -EINVAL;
> > +       }
> > +
> > +       if (value)
> > +               return regmap_set_bits(data->regmap, BD73800_REG_GPO_OUT,
> > +                                      pin->mask);
> > +
> > +       return regmap_clear_bits(data->regmap, BD73800_REG_GPO_OUT, pin->mask);
> > +}
>
> These can probably be handled by the latest version of gpio-regmap
> helpers (the version in linux-next).
>
> We now support both input-only and output-only in gpio-regmap.
>
> Latest <linux/gpio/regmap.h>:
>
>   * @fixed_direction_mask:
>   *                      (Optional) Bitmap representing the GPIO lines that
>   *                      make use of the @fixed_direction_output list to
>   *                      enforce direction of the GPIO. If this is NULL
>   *                      and @fixed_direction_output is defined, ALL GPIOs
>   *                      are assumed to be fixed direction (out or in).
>   * @fixed_direction_output:
>   *                      (Optional) Bitmap representing the fixed direction of
>   *                      the GPIO lines. Useful when there are GPIO lines with a
>   *                      fixed direction mixed together in the same register.
>
> It seems you can set up the fixed_direction_output mask from OTP
> and just use the library, right?

The BD73800 uses different registers for 'out' and 'in'. So there aren't 
really "GPIO lines with a fixed direction mixed together in the same 
register".

I, however, see some good stuff being in the linux-next! Finally a bit 
of flexibility in a form of callbacks a driver can register is getting 
in. So I am somewhat hopeful the usability of gpio-regmap will finally 
increase.

I think the BD73800 can indeed use the gpio-regmap, by keeping the pin 
direction information in private data and adding a custom xlate, which 
errors out if get is requested for output pin, or set is requested for 
input. I am not sure how much of an improvement it is compared to the 
patch v2, but I'll give it a shot :) I believe the new -rc1 isn't that 
far away, and hopefully the gpio-regmap -stuff from the next will be in 
it together with the regulator stuff bindings depended.

Thanks again for the pointer!

Yours,
	-- Matti

--
Matti Vaittinen
Linux kernel developer at ROHM Semiconductors
Oulu Finland

~~ When things go utterly wrong vim users can always type :help! ~~