On Tue, Feb 7, 2023 at 4:15 AM Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx> wrote: > > On Mon, Feb 06, 2023 at 03:18:26PM -0800, Dmitry Torokhov wrote: > > On Mon, Feb 06, 2023 at 07:50:16AM -0600, Danny Kaehn wrote: > > ... > > > > +#if IS_ENABLED(CONFIG_OF_GPIO) > > > + dev->gc.of_node = of_get_child_by_name(hdev->dev.of_node, "gpio"); > > > > > > I believe Andy is actively trying to get rid of of_node from GPIO chips. > > And in general, we should be using fwnode and generic device properties > > as much as possible. > > > > > +#endif > > Correct. And looking into the code of this patch I don't see any obstacles > to use fwnode APIs. You can Cc a v5 (which is supposed to be fwnode API based) > to me. > Sounds great, will do. I looked into doing this with the fwnode initially, but thought since the capability to describe usb devices in ACPI doesn't seem to be there, that I should be explicit that this only works for devicetree--but makes sense that it's better to be generic at the driver level if possible (especially if of_node is being removed from gpio chips), so will do! Thanks, Danny Kaehn
