On Tue, Mar 17, 2020 at 10:30:56PM +0100, Uwe Kleine-König wrote:
> Hello Thierry,
> 
> On Tue, Mar 17, 2020 at 06:43:44PM +0100, Thierry Reding wrote:
> > On Tue, Mar 17, 2020 at 02:32:26PM +0200, Oleksandr Suvorov wrote:
> > > Add the description of PWM_POLARITY_NORMAL flag.
> > > 
> > > Signed-off-by: Oleksandr Suvorov <oleksandr.suvorov@toradex.com>
> > > ---
> > > 
> > >  Documentation/devicetree/bindings/pwm/pwm.txt | 1 +
> > >  1 file changed, 1 insertion(+)
> > > 
> > > diff --git a/Documentation/devicetree/bindings/pwm/pwm.txt b/Documentation/devicetree/bindings/pwm/pwm.txt
> > > index 084886bd721e..440c6b9a6a4e 100644
> > > --- a/Documentation/devicetree/bindings/pwm/pwm.txt
> > > +++ b/Documentation/devicetree/bindings/pwm/pwm.txt
> > > @@ -46,6 +46,7 @@ period in nanoseconds.
> > >  Optionally, the pwm-specifier can encode a number of flags (defined in
> > >  <dt-bindings/pwm/pwm.h>) in a third cell:
> > >  - PWM_POLARITY_INVERTED: invert the PWM signal polarity
> > > +- PWM_POLARITY_NORMAL: don't invert the PWM signal polarity
> > 
> > This doesn't make sense. PWM_POLARITY_NORMAL is not part of the DT ABI.
> 
> "is not part of the DT ABI" is hardly a good reason. If it's sensible to
> be used, it is sensible to define it.

That's exactly it. It's not sensible at all to use it. If you define it
here it means people are allowed to do stuff like this:

	pwms = <&pwm 1234 PWM_POLARITY_INVERTED | PWM_POLARITY_NORMAL>;

which doesn't make sense. What's more, it impossible for the code to
even notice that you're being silly because | PWM_POLARITY_NORMAL is
just | 0 and that's just a nop.

> (And as far as I understand it, the term PWM_POLARITY_INVERTED isn't
> part of the DT ABI either. Only the value 1 has a meaning (for some PWM
> controlers).)
> 
> > The third cell of the specifier is a bitmask of flags.
> > 
> > PWM_POLARITY_NORMAL is an enumeration value that evaluates to 0, so it
> > makes absolutely no sense as a flag.
> 
> Using 0 or PWM_POLARITY_NORMAL doesn't have an effect on the compiled
> device tree, that's true. But having the term PWM_POLARITY_NORMAL (in
> contrast to a plain 0) in a dts file is useful in my eyes for human
> readers.
> 
> > PWM signals are considered to be "normal" by default, so no flag is
> > necessary to specify that.
> 
> GPIOs are considered to be active high by default, still there is
> GPIO_ACTIVE_HIGH (which also evaluates to 0). Also there is
> IRQ_TYPE_NONE.

I'm aware of those. That doesn't mean that everything needs to have
symbolic names.

Thierry