Re: [PATCH v4 09/10] iio: adc: ti-ads1262: support REFOUT and VBIAS regulators

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Calling regulator_get_voltage() on a regulator's own supply from within its .get_voltage() callback causes a recursive locking self-deadlock.
--

commit ea8c096c3d3561a926dd5bc84b789c153ef1db68
Author: Kurt Borja <kuurtb@gmail.com>

iio: adc: ti-ads1262: support REFOUT and VBIAS regulators

This commit registers the "refout" and "vbias" regulators in the ti-ads1262
driver to enable their use as common mode supplies.

> diff --git a/drivers/iio/adc/ti-ads1262.c b/drivers/iio/adc/ti-ads1262.c
> index 3f6d34a08b296..4784142f35c9d 100644
> --- a/drivers/iio/adc/ti-ads1262.c
> +++ b/drivers/iio/adc/ti-ads1262.c

[ ... ]

> @@ -1132,6 +1133,120 @@ static irqreturn_t ads1262_irq_handler(int irq, void *dev_id)
>  	return IRQ_HANDLED;
>  }
>  

[ ... ]

> +static int ads1262_vbias_get_voltage(struct regulator_dev *rdev)
> +{
> +	struct ads1262 *st = rdev_get_drvdata(rdev);
> +	int avdd_uV, avss_uV;
> +
> +	avdd_uV = regulator_get_voltage(st->avdd_supply);

[Severity: High]
Does this cause a recursive locking self-deadlock?

When a consumer or sysfs triggers a voltage query on vbias, the regulator
core calls regulator_lock_dependent(), which safely locks vbias and
recursively locks all of its supplies, including avdd.

The core then invokes vbias's .get_voltage callback, mapping to
ads1262_vbias_get_voltage(). Inside this callback, regulator_get_voltage()
acts as an external API: it initializes a new lock context and attempts
to lock avdd_supply again.

Because the current thread already holds the mutex for avdd, attempting
to lock it again with a new lock context results in a self-deadlock. A
thread attempting to read the vbias voltage will deadlock indefinitely,
causing a hang on that code path.

Could this approach use the unlocked internal API
regulator_get_voltage_rdev(rdev->supply->rdev) to get the supply's
voltage from within a regulator op instead?

> +	if (avdd_uV < 0)
> +		return avdd_uV;
> +
> +	avss_uV = st->avss_supply ? regulator_get_voltage(st->avss_supply) : 0;
> +	if (avss_uV < 0)
> +		return avss_uV;
> +
> +	return DIV_ROUND_CLOSEST(avdd_uV - avss_uV, 2);
> +}

[ ... ]

> +static const struct regulator_desc ads1262_vbias_regulator_desc = {
> +	.name = "vbias",
> +	.of_match = "vbias",
> +	.regulators_node = "regulators",
> +	.supply_name = "avdd",

[Severity: High]
This line establishes avdd as the upstream supply for vbias, which
instructs the regulator core that avdd is a strict dependency of vbias
and must be locked before dispatching operations to vbias ops. This is
what triggers the self-deadlock described above when
regulator_get_voltage() is called.

> +	.ops = &ads1262_vbias_regulator_ops,
> +	.type = REGULATOR_VOLTAGE,
> +	.owner = THIS_MODULE,
> +};

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260828-ads126x-v4-0-1dc27e9c0260@gmail.com?part=9