Re: [PATCH] iio: temperature: mlx90614: use guard(mutex) for EEPROM access locking

[Andy Shevchenko <andriy.shevchenko@intel.com>]

On Sat, Apr 18, 2026 at 05:51:36PM -0300, Luiz Mugnaini wrote:

> Replace mutex_lock()/mutex_unlock() pairs with guard(mutex) from
> cleanup.h for cleaner and safer mutex handling. The lock protects
> EEPROM access across five call sites in mlx90614_read_raw(),
> mlx90614_write_raw(), and mlx90614_sleep().
> 
> In all cases, the code between mutex_unlock() and the end of scope
> is either trivial computation, a return statement, or a call to
> mlx90614_power_put() (which only schedules a deferred autosuspend
> via pm_runtime_put_autosuspend()). The slightly extended lock scope
> is negligible compared to the existing hold times, which include
> EEPROM write cycles with 40ms sleeps.

...

> -		mutex_lock(&data->lock);
> +		guard(mutex)(&data->lock);
>  		ret = i2c_smbus_read_word_data(data->client,
>  					       chip_info->op_eeprom_emissivity);
> -		mutex_unlock(&data->lock);
>  		mlx90614_power_put(data);

How do you sure that the code:
- is not slowed down due to unneeded things in the critical section
- doesn't introduce no deadlocks
?

How did you test this, please?

-- 
With Best Regards,
Andy Shevchenko