Re: [PATCH 0/3] iio: light: tcs3472: devm conversion, wait time, locking cleanup

[Aldo Conte <aldocontelk@gmail.com>]

On 5/6/26 12:21, Andy Shevchenko wrote:
> On Wed, May 06, 2026 at 11:43:08AM +0200, Aldo Conte wrote:
>> This series of changes modernizes the tcs3472 driver by completing the
>> devm conversion and updating the locking style to use an automatic
>> guard (mutex). Furthermore, it implement control of the WAIT state,
>> resolving a TODO.
>>
>> Patch1: Converts the driver to use device-managed resource allocation.
>> This removes the need for an explicit remove() callback, since cleanup
>> is now handled automatically. It also adds a new function called
>> tcs3472_powerdown_action() that powers down the chip when the driver is
>> unloaded, removed or when probe fails after the chip has been enabled.
>>
>> Patch2: adds support for the wait time resolving the old TODO comment.
>> The user can control the WTIME indirectly by writing to the
>> sampling_frequency attribute. Changing the sampling
>> frequency attribute results in a change to the wtime while preserving
>> the current integration time.
>> Similarly, if the user has previously set a sampling frequency and
>> then changes the integration time, the driver re-computes WTIME so the
>> previous frequency is preserved. The patch also handles the deactivation
>> of wtime when the requested frequency is very high, as well as the
>> extension of wtime to its maximum values (by enabling the use of WLONG)
>> when the requested frequency is very low.
>>
>> Patch3: converts the remaining mutex_lock()/mutex_unlock() pairs to
>> guard(mutex).
>>
>> All patches have been tested on a Raspberry Pi 3B with a TCS3472
>> connected to I2C-1 at address 0x29. Sampling frequency and integration
>> time changes have been exercised checking the value of WTIME and
>> consequently WEN and WLONG. Raw RGBC reads, calibscale, integration_time
>> and threshold events continue to work as before.
> 
> Thanks for contribution, unfortunately this series needs much more work
> and more preparatory patches as well. I have reviewed the current series.
> Please, follow and address (or comment why it can't be done as suggested).
> 

Thank you for your review.
These mistakes are undoubtedly due to my still limited experience. I 
will start working on them right away.

Best regards,
Aldo Conte