Re: [PATCH v2 2/2] mfd: lm3533: convert to use OF

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

On Wed, Feb 19, 2025 at 07:39:29PM +0200, Svyatoslav Ryhel wrote:
> ср, 19 лют. 2025 р. о 17:56 Andy Shevchenko
> <andriy.shevchenko@linux.intel.com> пише:
> > On Wed, Feb 19, 2025 at 05:13:15PM +0200, Svyatoslav Ryhel wrote:
> > > ср, 19 лют. 2025 р. о 17:07 Andy Shevchenko
> > > <andriy.shevchenko@linux.intel.com> пише:
> > > > On Wed, Feb 19, 2025 at 04:36:38PM +0200, Svyatoslav Ryhel wrote:
> > > > > ср, 19 лют. 2025 р. о 16:27 Andy Shevchenko
> > > > > <andriy.shevchenko@linux.intel.com> пише:

...

> > > > > MFD part is removed since MFD cells binding is unconditional, while
> > > > > the device supports any amount of children grater then one. For
> > > > > example, my  device uses only backlight at bank A, while all other
> > > > > subdevices are not present and used. This patch switches to dynamic
> > > > > bind of children.
> > > >
> > > > MFD does the same. Please, take your time and get familiar with how MFD works.
> > >
> > > It does not. I have tried. If mfd cell binding is missing, driver will
> > > complain and fail.
> >
> > I really don't know what exactly is going on here, you can check it later, but
> > I'm 100% sure that MFD works for only driver that are present. What you are
> > describing is how component framework is designed.
> >
> > To proove my words you can check drivers/mfd/intel_soc_pmic_*.c and find listed
> > cells that never had a driver in the Linux kernel. They are just placeholders.
> 
> This example is not valid since those drivers do not use device tree.

You didn't get the point. I'm telling that it should not matter if there is a
device driver present or absent. The rest will be enumerated as usual.

I even checked the code that handles of_compatible member from struct mfd_cell and
at a brief glance I do not see that it can do otherwise.

Care to elaborate what and where is the error happened exactly?

-- 
With Best Regards,
Andy Shevchenko