Re: [PATCH v2 2/2] media: i2c: add support for ov4689

[Sakari Ailus <sakari.ailus@linux.intel.com>]

Hi Tommaso,

On Fri, Sep 16, 2022 at 03:34:01PM +0200, Tommaso Merciai wrote:
> > >> +	ret = clk_set_rate(ov4689->xvclk, OV4689_XVCLK_FREQ);
> > >> +	if (ret < 0) {
> > >> +		dev_err(dev, "Failed to set xvclk rate (24MHz)\n");
> > >> +		return ret;
> > >> +	}
> > >> +	if (clk_get_rate(ov4689->xvclk) != OV4689_XVCLK_FREQ)
> > >> +		dev_warn(dev, "xvclk mismatched, modes are based on 24MHz\n");
> > >
> > >
> > > What do you think about?
> > > Thanks.
> > 
> > Unfortunately, I have no experience with ACPI-based devices. :(
> > 
> > Do you mean that in the case of an ACPI device and devm_clk_get_optional
> > returning NULL we should assume that the clock is already enabled and
> > will stay enabled during sensor operation? How should we distinguish it
> > from the case of an OF-based system and clock just missing from device
> > tree?
> 
> Not exaclty :)
> 
> I copy comment from [1]
> 
> if you use ov5693->xvclk to identify the ACPI vs OF use case shouldn't
> you use the get_optionl() version ? Otherwise in the ACPI case you will have
> -ENOENT if there's not 'xvclk' property and bail out.
> 
> Unless my understanding is wrong on ACPI we have "clock-frequency" and
> on OF "xvclk" with an "assigned-clock-rates",

Generally yes. It's also possible to have a clock in ACPI based system
although those clocks do not come from ACPI. See e.g.
drivers/platform/x86/intel/int3472/clk_and_regulator.c .

-- 
Sakari Ailus