On Thu, 7 Dec 2023 15:41:25 -0300
Marcelo Schmitt <marcelo.schmitt@xxxxxxxxxx> wrote:
> The ADC needs a voltage reference to work correctly.
> Enable AD7091R internal voltage reference if no external vref is
> supplied.
>
> Signed-off-by: Marcelo Schmitt <marcelo.schmitt@xxxxxxxxxx>
> ---
> drivers/iio/adc/ad7091r-base.c | 9 ++++++---
> drivers/iio/adc/ad7091r-base.h | 1 +
> 2 files changed, 7 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/iio/adc/ad7091r-base.c b/drivers/iio/adc/ad7091r-base.c
> index f2cb638b8d77..59a7ec44955d 100644
> --- a/drivers/iio/adc/ad7091r-base.c
> +++ b/drivers/iio/adc/ad7091r-base.c
> @@ -215,10 +215,13 @@ int ad7091r_probe(struct device *dev, const char *name,
> iio_dev->channels = st->chip_info->channels;
>
> st->vref = devm_regulator_get_optional(dev, "vref");
This does not return NULL, only a valid regulator or an error code.
> - if (IS_ERR(st->vref)) {
> - if (PTR_ERR(st->vref) == -EPROBE_DEFER)
> - return -EPROBE_DEFER;
> + if (IS_ERR_OR_NULL(st->vref)) {
You still need to explicitly handle deferal here.
There might be a perfectly good regulator that just isn't ready yet and
if that happens we want to try probing this driver again later rather
than papering over it.
> + /* Enable internal vref */
> st->vref = NULL;
> + ret = regmap_update_bits(st->map, AD7091R_REG_CONF,
> + AD7091R_REG_CONF_INT_VREF, BIT(0));
> + if (ret)
> + return ret;
> } else {
> ret = regulator_enable(st->vref);
> if (ret)
> diff --git a/drivers/iio/adc/ad7091r-base.h b/drivers/iio/adc/ad7091r-base.h
> index 9546d0bf1da7..e153c2d7deb5 100644
> --- a/drivers/iio/adc/ad7091r-base.h
> +++ b/drivers/iio/adc/ad7091r-base.h
> @@ -20,6 +20,7 @@
> #define AD7091R_REG_CH_HYSTERESIS(ch) ((ch) * 3 + 6)
>
> /* AD7091R_REG_CONF */
> +#define AD7091R_REG_CONF_INT_VREF BIT(0)
> #define AD7091R_REG_CONF_ALERT_EN BIT(4)
> #define AD7091R_REG_CONF_AUTO BIT(8)
> #define AD7091R_REG_CONF_CMD BIT(10)
