Re: [PATCH v3 RESEND] media: i2c: Add OV05C10 camera sensor driver

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

Hi Mehdi,

On Tue, Jun 24, 2025 at 06:34:51PM +0200, Mehdi Djait wrote:
> Hi Sakari,
> 
> On Tue, Jun 24, 2025 at 11:46:27AM +0000, Sakari Ailus wrote:
> 
> [...]
> 
> > 
> > I'm in favour of the latter but both should be workable.
> > 
> > Speaking of return values, devm_clk_get_optional() may also return
> > -EPROBE_DEFER. That needs to be handled.
> > 
> 
> Ack.
> 
> > And further on -EPROBE_DEFER, I think the helper should return
> > -EPROBE_DEFER if the "clock-frequency" property doesn't exist on non-OF
> > nodes. That signals the required software nodes required on Intel Windows
> > definitions/ipu-bridge or AMD systems aren't in place yet so really probing
> > should be deferred. This would allow removing the hacks that return
> > -EPROBE_DEFER in sensor drivers when no graph endpoint is found.
> 
> device_property_read_u32() returns the following:
> 
>  * Return: number of values if @val was %NULL,
>  *         %0 if the property was found (success),
>  *	   %-EINVAL if given arguments are not valid,
>  *	   %-ENODATA if the property does not have a value,
>  *	   %-EPROTO if the property is not an array of numbers,
>  *	   %-EOVERFLOW if the size of the property is not as expected.
>  *	   %-ENXIO if no suitable firmware interface is present.
> 
> 
> Do you mean something like this in the helper:
> 
> if (ret == -ENODATA && !of_node)
> 	return ERR_PTR(-EPROBE_DEFER);

I think -EINVAL is returned if a property doesn't exist on ACPI, same on
software nodes. On software nodes you could also have -ENODATA if the
property has no value. So test both of these?

I'd also test !of_node first.

-- 
Regards,

Sakari Ailus