Re: Adding support for configuring polarity in PWM framework.

[Thierry Reding <thierry.reding@avionic-design.de>]

[-- Attachment #1: Type: text/plain, Size: 3413 bytes --]

On Mon, Jul 16, 2012 at 12:23:46PM +0000, Philip, Avinash wrote:
> On Mon, Jul 16, 2012 at 17:09:21, Thierry Reding wrote:
> > On Mon, Jul 16, 2012 at 11:15:50AM +0000, Philip, Avinash wrote:
> > > Hi Thierry,
> > > 
> > > On one of the custom boards we are using, uses PWM to drive the backlight. However, for
> > > this device, PWM signal needs to be inversed.
> > > So, we need to a platform data to indicate this parameter.
> > > Current PWM framework doesn't provide .support for setting polarity (or inverse polarity).
> > > 
> > > Have you come across any such requirements? If so, do you have any plans to implement it?
> > 
> > I don't have any plans to implement such a feature.
> 
> Ok. Thanks for the quick response.
> > 
> > > I am planning to add support for the same but want to avoid duplication of work.
> > > 
> > > If you have no plans, then I will send a patch to support the same.
> > 
> > I wonder how you want to implement this. You'll need special hardware
> > support for it 
> 
> Yes. Our custom hardware (backlight booster) requires the pwm signal to be 
> inverted.
> 
> > you may be able to implement it in the driver itself 
> > instead of putting it into the framework. 
> 
> This is a client specific data (backlight needs pwm signal inversed) 
> and not the main device feature (not PWM IP). So we cannot send this in 
> pwm platform data. This would come as call from client driver (which in 
> our case is from pwm_bl.c)

Okay, I see.

> > Anyway I'm interested in seeing your patch.
> 
> I am planning to modify PWM framework as below.
> 1. Configure PWM polarity from client driver (using platform data provided 
> to pwm backlight driver).
> 2. PWM device needs to be disabled before calling the set-polarity API.

Okay, that sounds sensible. A couple of comments though.

> This involves
> 
> 1. PWM framework API addition.
> 	PWM frame work API support.
> 	/**
> 	 * pwm_setpolarity() - change a PWM device Polarity	
> 	 * @pwm: PWM device
> 	 * @polarity: Configure polarity of PWM
>  	*
> 	 * polarity     - false -> "on" time defined by duty ns
> 	*              - true  -> "off' time defined by duty ns.
>  	*/
> 	int pwm_setpolarity(struct pwm_device *pwm, bool inversepol);

This should match the pwm_ops name, i.e.: pwm_set_polarity().

Making the polarity argument a boolean is slightly confusing. For
instance I'd say the logical value if I want normal behaviour would be
to set it to true, which doesn't match your example. So I propose you
define the polarity parameter as an enumeration to make its meaning more
explicit:

	enum pwm_polarity {
		PWM_POLARITY_NORMAL,
		PWM_POLARITY_INVERSE,
	};

PWM_POLARITY_{HIGH,LOW} and PWM_POLARITY_{POSITIVE,NEGATIVE} would be
other good name pairs.

> 2. Add "set_polarity" operation support in pwm_ops.
> 
> 3. Modification in backlight driver (pwm_bl.c) to support polarity
> configuration.

We also need to think about how this could be represented in the device
tree. The most obvious choice seems to be a third cell for the specifier
and use a custom of_xlate callback for controllers that support polarity
inversion (and later perhaps other flags).

Also would you mind sharing the board setup code that you need this for?
I find it easier to get into the right mindset when looking at code that
actually uses this.

Thierry

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]