On Tue, Feb 21, 2017 at 10:17:49AM +0200, Sakari Ailus wrote: > Hi Rafael, > > On Wed, Feb 08, 2017 at 01:19:57PM +0100, Rafael J. Wysocki wrote: > ... > > > > I think I agree with Rob's prior comments about making an ops struct for DT > > > > vs ACPI. Out of the 16 patches, 2/16, 3/16, 5/16 (multiple times), and this > > > > patch all end up using the same construct. Maybe it needs to be a separate > > > > refactoring effort, but if it's happening this often just in this patch set, > > > > it seems like it's getting time to clean things up. > > > > > > As long as there are two cases only (ACPI vs DT), an ops struct wouldn't > > > really make things simpler and it would make the code more difficult to > > > follow. > > > > > > But we do have a third case (static or built-in properties) and it doesn't > > > seem to be covered at all. > > > > That said the ops struct could be introduced on top of this series just fine. > > It even might be cleaner to do it this way, so I'm not asking for a redesign > > here. > > > > I'd like the built-in properties to be covered too, however. > > That sounds good to me. After looking at the implementation, the fwnode interface is rather complete for property sets --- graphs are inherently cross-device concepts and property sets don't really lend themselves to that. I think it'd be good to have at least a use case for that before implementing it based on a thought that it might be needed for something. -- Regards, Sakari Ailus e-mail: sakari.ailus@xxxxxx XMPP: sailus@xxxxxxxxxxxxxx -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
