On Thu, May 05, 2022 at 04:59:35PM +0200, Arnd Bergmann wrote:
> On Thu, May 5, 2022 at 4:39 PM Mark Brown <broonie@kernel.org> wrote:
> > On Thu, May 05, 2022 at 04:33:06PM +0200, Linus Walleij wrote:
> > > On Thu, May 5, 2022 at 8:04 AM Arnd Bergmann <arnd@kernel.org> wrote:

> > > > static struct snd_soc_jack_pin hs_jack_pin[] = {
> > > >         {
> > > >                 .pin    = "Headphone Jack",
> > > >                 .mask   = SND_JACK_HEADPHONE,
> > > >         },
> > > >         {
> > > >                 .pin    = "Speaker",
> > > >                 /* disable speaker when hp jack is inserted */
> > > >                 .mask   = SND_JACK_HEADPHONE,
> > > >                 .invert = 1,
> > > >         },

> > > Hm some ASoC thingie. No idea what that is, but I suppose another
> > > place where a subsystem for legacy reasons try to do the gpiolib
> > > inversion on it's own accord. That one isn't flagged as active low in the
> > > descriptor so it's fine I guess.

> > It's saying that when the headphone is inserted the headphone output
> > should be enabled and the speaker output should be disabled, and vice
> > versa.

> Ok, that sounds like I should remove the flag here if I declare the
> GPIO line as GPIO_ACTIVE_LOW instead of GPIO_ACTIVE_HIGH, right?

If you change the sense of the GPIO you'll need to flip the invert to
the headphone instead of the speaker - whichever way round the GPIO
sense is each of the pins should be taking the opposite sense from the
GPIO state to the other.