On Sat, Sep 18, 2021 at 2:00 AM Heiko Stübner [off-list ref] wrote:

The issue is that the pinconf part for PIN_CONFIG_OUTPUT is actually
using the gpio controller to realize this setting. So when this ends up
in a pinctrl-hog, stuff explodes while probing the first pinctrl part.

The Nomadik driver has something similar, I came up with a solution
ages ago which isn't elegant either, so it's not like I'm any better :/

commit ab4a936247561cd998913bab5f15e3d3eaed1f9e
"pinctrl: nomadik: assure GPIO chips are populated"

Thinking about this, the component framework may be one option?
And then adding a pinctr-register / init+enable variant where the
pinctrl hogs can be aquired separately, not as part of pinctrl_enable?

Check out my commit, but the component framework is what we
should ideally use (IMO) when drivers depend on each other
so I think you are right.

Yours,
Linus Walleij

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel