On Mon, Oct 1, 2012 at 5:44 PM, Tony Lindgren <tony@xxxxxxxxxxx> wrote: >> OK that is typical pinctrl driver implementation work. >> I hope Tony can advice on this? > > I think we're best off to just stick to alternative named modes > passed from device tree. For example, for GPIO wake-ups you can > have named modes such as "default", "enabled" and "idle" where > "idle" muxes things for GPIO wake-ups for the duration of idle. > > It seems that should also work for leds-gpio. And you can > define more named modes as needed. This is what we're doing for ux500 and should be a good model. > You really don't want the client driver or the GPIO driver doing > things like pull-up/down automatically as that is board specific and > can also depend on things like externall pull resistors. Nope. We've had instances of people getting bad leakage because of pulling down a line where there is already a pull-down resistor on the board :-( >> OK so the threshold is that we need to get it right for the first >> one and then the others will look good too. > > It seems we want to keep leds-gpio, gpio framework and pinctrl > framework generic. It also seems you should be able to do > what you're describing using the pinctrl named modes. I think so too. Yours, Linus Walleij -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html