Re: [PATCH v3 3/5] drivers: regulator: Add ADI MAX77541/MAX77540 Regulator Support

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

On Tue, Jan 31, 2023 at 01:59:45PM +0000, Sahin, Okan wrote:
> On Tue, 31 Jan 2022 4:30 PM
> Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote:
> >On Tue, Jan 31, 2023 at 01:23:33PM +0000, Sahin, Okan wrote:
> >> On Tue, 31 Jan 2022 3:27 PM
> >> Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote:
> >> >On Tue, Jan 31, 2023 at 09:27:48AM +0000, Sahin, Okan wrote:

...

> >> >> Sorry for second question. I do not want to bother you, but I
> >> >> realized that I need to be sure about driver_data before sending
> >> >> new patch. You said that you need to use pointers directly for
> >> >> driver_data then I fixed that part in mfd, but I do not need or
> >> >> use driver_data in regulator since chip_id comes from mfd device so
> >> >> I think using like below should be enough for my implementation.
> >> >>
> >> >> static const struct platform_device_id max77541_regulator_platform_id[] =
> >{
> >> >> 	{ "max77540-regulator", },
> >> >> 	{ "max77541-regulator", },
> >> >> 	{  /* sentinel */  }
> >> >> };
> >> >> MODULE_DEVICE_TABLE(platform, max77541_regulator_platform_id);
> >> >>
> >> >> static const struct of_device_id max77541_regulator_of_id[] = {
> >> >> 	{ .compatible = "adi,max77540-regulator", },
> >> >> 	{ .compatible = "adi,max77541-regulator", },
> >> >> 	{ /* sentinel */  }
> >> >> };
> >> >> MODULE_DEVICE_TABLE(of, max77541_regulator_of_id);
> >> >>
> >> >> What do you think?
> >> >
> >> >If you have got all necessary data from the upper layer, why do you
> >> >need to have an ID table here? I'm not sure I understand how this OF
> >> >ID table works in this case.
> >
> >> I added it since there is regulator node in device tree. With the help
> >> of devm_regulator_register(..), driver takes parameters of regulator
> >> node. I also used id to select and to initialize regulator descriptors
> >> which are chip specific. So far there is no comment about OF ID table
> >> so I kept it. I thought I need to add both of id table and platform id
> >> table as name matching is required to initialize platform device from mfd.
> >
> >For platform device is one mechanism how to enumerate device, and bind it to
> >the driver. The OF ID table needs to be present in case you are using it for direct
> >DT enumeration (there is also something related to MFD child nodes, but you
> >need to check and explain how your device is enumerated by this driver).
> >
> >I.o.w. please clarify how the OF ID table is being used.
> 
> I do not use "of id table" directly in max77541-regulator.c so do I need to exclude it?

Exactly my point. How does this OF ID table affect the device enumeration?

> However, devm_regulator_register(..) method initialize each regulator with
> the nodes under "regulators node". If of_match in desc and name of node
> matches, then regulator will be initialized with parameters in the node under
> the regulators node in the device tree. Since I am using device tree to
> initialize regulators, I added of id table. I hope I explained the situation
> clearly.

This is confusing. If your regulator is enumerated via DT, why do you need MFD?

-- 
With Best Regards,
Andy Shevchenko