Re: [PATCH 1/2] dt-bindings: mfd: rohm,bd71847-pmic: Document rohm,clock-output-is-critical property

[Marek Vasut <marex@denx.de>]

On 10/20/21 12:14 PM, Vaittinen, Matti wrote:
[...]

> I wonder if this really is something specific to ROHM ICs? Do you think
> this would warrant a generic, non vendor specific property? I am Ok with
> the ROHM specific property too but it just seems to me this might not be
> unique to ROHM IC(s).
> 
> By the way, the very same clk driver where you implemented the property
> reading (patch 2/2) is used by few other ROHM PMICs. At least by
> BD71837, BD71828, BD71815, BD9576 and BD9573. So the code change here
> adds support for this property to all of those PMICs. I wonder if the
> property should be mentioned in all of the binding docs... That could be
> another argument for making this a generic property and describing it in
> clk yaml ;)
> 
> Well, just my 10 Cents - I am ok with this change as you presented it
> here if you don't think this should be generic one.

I think we need something like gpio-hog, except for clock. Some clk-hog 
maybe ? That would be useful not only here, but also for things where 
some output generates clock for random stuff which cannot be described 
in the DT for whatever reason (like e.g. the SoC is used as a substitute 
for CPLD XTAL and the CPLD isn't connected to the SoC in any other way).