On Tue, Jan 13, 2015 at 10:35:52AM +0100, Robert Jarzmik wrote: > Paul Zimmerman <Paul.Zimmerman@xxxxxxxxxxxx> writes: > > The patch below fixes it. And it seems like the right thing to me, > > since GPIOs should be optional for a generic phy, I would think. But > > my device tree foo is very weak, so I'm not sure. > > > > CCing Robert, who touched the generic phy code last. Robert, what do > > you think? > I think your patch in [1] is correct, because > "Documentation/devicetree/bindings/usb/usb-nop-xceiv.txt" states that > reset-gpios is optional, and because Felipe told me gpios for phy_generic are > optional. > > Now the original code was written by Felipe, so better ask him first. The > original code was : > af9f51c55 phy-generic.c nop->gpio_reset = of_get_named_gpio_flags(node, "reset-gpios", > af9f51c55 phy-generic.c 0, &flags); > af9f51c55 phy-generic.c if (nop->gpio_reset == -EPROBE_DEFER) > af9f51c55 phy-generic.c return -EPROBE_DEFER; yeah, at the time that was written, there was no _optional() varianto of get_gpio, if there is one now, we should use it, indeed. I can rather easily test here on BBB to make sure there are no regressions. -- balbi
Attachment:
signature.asc
Description: Digital signature
