Re: [PATCH v3 02/11] mfd: lm3533: Remove driver specific regmap wrappers

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Incorrect argument order for `regmap_update_bits` silently prevents disabling interrupts.
--

commit 02ae49c5dba8ff55717ce569312b1ed0ac85b17f
Author: Svyatoslav Ryhel <clamor95@gmail.com>

mfd: lm3533: Remove driver specific regmap wrappers

This commit replaces custom wrapper functions for regmap access with standard
regmap helpers across the lm3533 MFD driver and its sub-device drivers. The
commit message notes that regmap_update_bits requires the mask and value
arguments to be swapped compared to the old wrappers.

> diff --git a/drivers/iio/light/lm3533-als.c b/drivers/iio/light/lm3533-als.c
> index 99f0b903018cf..fb61904f110f1 100644
> --- a/drivers/iio/light/lm3533-als.c
> +++ b/drivers/iio/light/lm3533-als.c

[ ... ]

> @@ -280,7 +281,8 @@ static int lm3533_als_set_int_mode(struct iio_dev *indio_dev, int enable)
>  	else
>  		val = 0;
>  
> -	ret = lm3533_update(als->lm3533, LM3533_REG_ALS_ZONE_INFO, val, mask);
> +	ret = regmap_update_bits(als->lm3533->regmap, LM3533_REG_ALS_ZONE_INFO,
> +				 val, mask);

[Severity: High]
Are the mask and val arguments swapped in this regmap_update_bits() call?

The commit message mentions that regmap_update_bits() has the mask and value
arguments swapped compared to the old wrapper, but in this instance they are
passed as val, then mask.

If enable is false, val is 0. Calling regmap_update_bits() with 0 as the
bitmask will modify zero bits, which fails to clear the interrupt enable bit.
Could this cause an unhandled interrupt storm when the interrupt is disabled
via sysfs, or when the driver is removed and free_irq() is called while the
level-triggered IRQ line remains asserted?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260601151831.76350-1-clamor95@gmail.com?part=2