Hello
On Mon, Jun 11, 2018 at 3:38 PM Marcel Holtmann <mar...@holtmann.org> wrote: > > Hi Andy, > > >> Rave-sp behaves differently based on the device variant. > > > > > >> sp->variant = of_device_get_match_data(dev); > >> + if (!sp->variant) { > >> + id = serdev_match_id(rave_sp_serdev_id, serdev); > > > > I think you may leave the ID table where it is in the code and use link. > > > >> + if (id) > >> + sp->variant = (const struct rave_sp_variant *) > >> + id->driver_data; > >> + } > > > > Perhaps a helper like it's done for ACPI / OF cases? > > > > [device_get_match_data() -> ] > > of_fwnode_device_get_match_data() > > acpi_fwnode_device_get_match_data() > > something like that and frankly this is trying to hard. This driver is > currently really DT specific and can be fixed up later. It is causing a lot > of noise in this patch series. I would really urge to focus on the core > changes get prominent drivers support. I think hci_bcm.c is a good example > since a) we need to fix up Edison and b) new ACPI based tablets might be able > to allow for easy testing. I can try to implement something like andy proposes, but if it turns out too complicated we can remove this driver from the series. Thanks! > > Regards > > Marcel > -- Ricardo Ribalda