> > The snps,dw-apb-gpio driver iterates over its port subnodes (like gpio2b)
> > to register gpio_chip interfaces, but it doesn't appear to register these
> > subnodes as independent platform devices.
> >
> > Because the driver core only applies pinctrl-0 properties automatically
> > during probe for fully initialized device instances, this configuration
> > might be silently ignored by the kernel.
>
> True. Will rework.

After some more research: not true

gpiolib handles this, check this comment:

5351 /*
5352  * The DT node of some GPIO chips have a "compatible" property, but
5353  * never have a struct device added and probed by a driver to register
5354  * the GPIO chip with gpiolib. In such cases, fw_devlink=on will cause
5355  * the consumers of the GPIO chip to get probe deferred forever because
5356  * they will be waiting for a device associated with the GPIO chip
5357  * firmware node to get added and bound to a driver.
5358  *
5359  * To allow these consumers to probe, we associate the struct
5360  * gpio_device of the GPIO chip with the firmware node and then simply
5361  * bind it to this stub driver.
5362  */

Further proof, 'pinmux-pins' in debugfs shows that these pins are tied
to the gpiochip with this patch. It doesn't without this patch.

I think it is okay as-is. But I need to make another update anyhow, so I
will send a v2 nonetheless.