On Mon, Nov 11, 2013 at 8:33 PM, Stephen Warren [off-list ref] wrote:

On 11/11/2013 12:17 PM, Gerlando Falauto wrote:

quoted

Thanks for your answer. So it should be the parent driver to proactively
configure the GPIO as input, active high etc..., when an IRQ for its
GPIOs is requested, right?

That was the conclusion of this thread, or a similar thread, yes.

Yepps second that.

And we added an API to that you can flag a GPIO as being
in use by an IRQ using gpio_lock_as_irq() from your
irq_chip callbacks.

Yours,
Linus Walleij