On 07/01/2024 10:49, Karel Balej wrote:
> Mark,
>
> On Fri Jan 5, 2024 at 4:18 PM CET, Mark Brown wrote:
>> On Thu, Dec 28, 2023 at 10:39:13AM +0100, Karel Balej wrote:
>>
>>> @@ -68,6 +68,21 @@ static struct mfd_cell pm886_devs[] = {
>>> .num_resources = ARRAY_SIZE(pm88x_onkey_resources),
>>> .resources = pm88x_onkey_resources,
>>> },
>>> + {
>>> + .name = "88pm88x-regulator",
>>> + .id = PM88X_REGULATOR_ID_LDO2,
>>> + .of_compatible = "marvell,88pm88x-regulator",
>>> + },
>>
>> Why are we adding an of_compatible here? It's redundant, the MFD split
>> is a feature of Linux internals not of the hardware, and the existing
>> 88pm8xx MFD doesn't use them.
>
> in a feedback to my MFD series, Rob Herring pointed out that there is no
> need to have a devicetree node for a subdevice if it only contains
> "compatible" as the MFD driver can instantiate subdevices itself. I
> understood that this is what he was referring to, but now I suspect that
> it is sufficient for the mfd_cell.name to be set to the subdevice driver
> name for this - is that correct?
I think Rob was only referring to "no need to have a devicetree node".
But you added here a devicetree node, plus probably undocumented compatible.
Does it even pass the checkpatch?
Best regards,
Krzysztof
