On Fri, Jun 11, 2021 at 2:08 PM Russell King (Oracle) <linux@xxxxxxxxxxxxxxx> wrote: > > On Fri, Jun 11, 2021 at 01:40:59PM +0200, Rafael J. Wysocki wrote: > > I'm not sure why you want the above to be two if () statements instead of one? > > > > I would change the ordering anyway, that is > > > > if (!IS_ERR(phy_node) || is_acpi_node(fwnode)) > > return phy_node; > > > > And I think that the is_acpi_node() check is there to return the error > > code right away so as to avoid returning a "not found" error later. > > > > But I'm not sure if this is really necessary. Namely, if nothing > > depends on the specific error code returned by this function, it would > > be somewhat cleaner to let the code below run if phy_node is an error > > pointer in the ACPI case, because in that case the code below will > > produce an error pointer anyway. > > However, that opens the door to someone shipping "working" ACPI with > one of these names that we've taken the decision not to support on > ACPI firmware. Surely, it's much better that we don't accept the > legacy names so we don't allow such configurations to work. Fair enough.
