Re: [PATCH v7 3/3] platform/chrome: cros_ec_uart: Add DT enumeration support

[Tzung-Bi Shih <tzungbi@kernel.org>]

On Tue, Nov 29, 2022 at 11:13:04AM -0700, Mark Hasemeyer wrote:
> > > @@ -392,6 +393,12 @@ static int __maybe_unused cros_ec_uart_resume(struct device *dev)
> > >  static SIMPLE_DEV_PM_OPS(cros_ec_uart_pm_ops, cros_ec_uart_suspend,
> > >                        cros_ec_uart_resume);
> > >
> > > +static const struct of_device_id cros_ec_uart_of_match[] = {
> > > +     { .compatible = "google,cros-ec-uart" },
> > > +     {}
> > > +};
> > > +MODULE_DEVICE_TABLE(of, cros_ec_uart_of_match);
> >
> > It would need a guard for checking CONFIG_OF.  Similar to what
> > `cros_ec_uart_acpi_id` does.
> >
> > > @@ -405,6 +412,7 @@ static struct serdev_device_driver cros_ec_uart_driver = {
> > >       .driver = {
> > >               .name   = "cros-ec-uart",
> > >               .acpi_match_table = ACPI_PTR(cros_ec_uart_acpi_id),
> > > +             .of_match_table = cros_ec_uart_of_match,
> >
> > It would need be wrapped by `of_match_ptr()`.  Similar to what
> > `ACPI_PTR()` does.
> 
> I'm not sure we want a guard in this case because we compile the kernel without
> CONFIG_OF enabled for (most?) x86 platforms. Yet we still need the device
> tree code to enumerate using the PRP0001 _HID method.

I'm not familiar to ACPI.  However, I thought it should add the compatible
string in the _DSD instead of using of_match for the case.  See example
in [1].

[1]: https://www.kernel.org/doc/html/latest/firmware-guide/acpi/enumeration.html#device-tree-namespace-link-device-id