Re: [PATCH 0/3] [v11] pinctrl: qcom: add support for sparse GPIOs

[Stephen Boyd <sboyd@codeaurora.org>]

On 12/28, Linus Walleij wrote:
> On Wed, Dec 27, 2017 at 3:01 AM, Stephen Boyd <sboyd@codeaurora.org> wrote:
> 
> > The different approaches come down to expressing
> > which pins are available through the gpio valid mask, or through
> > the npins field of the msm pinctrl driver. Also, my approach
> > covers more than just GPIOs, it covers irqs and adjusts the
> > pinctrl pin request function so that pinctrl can't request
> > unavailable pins.
> 
> I agree, this is better.

Thanks for the feedback. I'll update and resend my patch to the
list.

> 
> Would even patch 1 be needed after this? Maybe I should
> revert that too. Leaving that code in has the upside of showing
> the actual initial directions of GPIO lines even if they have
> not been requested, in e.g. debugfs.
> 

Patch 1 is still needed. Without that patch, we'll be poking each
GPIO to figure out the direction at boot without checking any
valid mask or calling the request APIs. I don't see the part in
debugfs where we show the direction of a GPIO if it hasn't been
requested. Don't we skip over the unrequested GPIOs because of
this code in gpiolib_dbg_show()?

	if (!test_bit(FLAG_REQUESTED, &gdesc->flags)) {
		...
		continue;
	}

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project