Hello ,

On Fri, Feb 11, 2022 at 11:06:33AM +0800, Song Chen wrote:
> 在 2022/2/10 18:03, Uwe Kleine-König 写道:
> > On Thu, Feb 10, 2022 at 05:05:02PM +0800, Song Chen wrote:
> > > Introduce apply in pwm_ops to replace legacy operations,
> > > like enable, disable, config and set_polarity.
> > > 
> > > Signed-off-by: Song Chen <chensong_2000@189.cn>
> > > ---
> > >   drivers/staging/greybus/pwm.c | 46 +++++++++++++++--------------------
> > >   1 file changed, 19 insertions(+), 27 deletions(-)
> > > 
> > > diff --git a/drivers/staging/greybus/pwm.c b/drivers/staging/greybus/pwm.c
> > > index 891a6a672378..e1889cf979b2 100644
> > > --- a/drivers/staging/greybus/pwm.c
> > > +++ b/drivers/staging/greybus/pwm.c
> > > @@ -204,43 +204,35 @@ static void gb_pwm_free(struct pwm_chip *chip, struct pwm_device *pwm)
> > >   	gb_pwm_deactivate_operation(pwmc, pwm->hwpwm);
> > >   }
> > > -static int gb_pwm_config(struct pwm_chip *chip, struct pwm_device *pwm,
> > > -			 int duty_ns, int period_ns)
> > > -{
> > > -	struct gb_pwm_chip *pwmc = pwm_chip_to_gb_pwm_chip(chip);
> > > -
> > > -	return gb_pwm_config_operation(pwmc, pwm->hwpwm, duty_ns, period_ns);
> > > -};
> > > -
> > > -static int gb_pwm_set_polarity(struct pwm_chip *chip, struct pwm_device *pwm,
> > > -			       enum pwm_polarity polarity)
> > > +static int gb_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm,
> > > +			const struct pwm_state *state)
> > >   {
> > > +	int ret;
> > >   	struct gb_pwm_chip *pwmc = pwm_chip_to_gb_pwm_chip(chip);
> > > -	return gb_pwm_set_polarity_operation(pwmc, pwm->hwpwm, polarity);
> > > -};
> > > -
> > > -static int gb_pwm_enable(struct pwm_chip *chip, struct pwm_device *pwm)
> > > -{
> > > -	struct gb_pwm_chip *pwmc = pwm_chip_to_gb_pwm_chip(chip);
> > > +	/* set period and duty cycle*/
> > > +	ret = gb_pwm_config_operation(pwmc, pwm->hwpwm, state->duty_cycle, state->period);
> > 
> > gb_pwm_config_operation's 3rd parameter is an u32, so you're loosing
> > bits here as state->duty_cycle is a u64. Ditto for period.
> 
> originally, pwm_apply_state --> pwm_apply_legacy --> gb_pwm_config -->
> gb_pwm_config_operation is also loosing bits, does it mean greybus can live
> with that?

This is true, I tried to address that, but Thierry had concerns.
(https://lore.kernel.org/all/20210312212119.1342666-1-u.kleine-koenig@pengutronix.de/
was the patch I suggested.)

> Or redefine gb_pwm_config_request, switch duty and period to __le64?

Don't use __le64, this is only for representing (little endian) register
values. u64 would be the right one.

> > Also it would be nice if you go from
> > 
> > 	.duty_cycle = A, .period = B, .enabled = 1
> > 
> > to
> > 
> > 	.duty_cycle = C, .period = D, .enabled = 0
> > 
> > that C/D wasn't visible on the output pin. So please disable earlier
> > (but keep enable at the end).
> 
> sorry, i don't quite understand this part,

To reexplain: If your hardware is configured for

	.duty_cycle = A, .period = B, .enabled = 1

and pwm_apply is called with

	.duty_cycle = C, .period = D, .enabled = 0

you configured the registers for .duty_cycle and .period first and only
then disable the PWM. This usually results in glitches because the
hardware shortly runs with

	.duty_cycle = C, .period = D, .enabled = 1

. So the idea is, to disable before configuring duty and period if the
eventual goal is a disabled state.

> but is below code looking good to
> you?
> 
> static int gb_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm,
> 			const struct pwm_state *state)
> {
> 	int err;
> 	bool enabled = pwm->state.enabled;
> 	struct gb_pwm_chip *pwmc = pwm_chip_to_gb_pwm_chip(chip);
> 
> 	/* set polarity */
> 	if (state->polarity != pwm->state.polarity) {
> 		if (enabled) {
> 			gb_pwm_disable_operation(pwmc, pwm->hwpwm);
> 			enabled = false;
> 		}
> 		err = gb_pwm_set_polarity_operation(pwmc, pwm->hwpwm, state->polarity);
> 		if (err)
> 			return err;
> 	}
> 
> 	if (!state->enabled) {
> 		if (enabled)
> 			gb_pwm_disable_operation(pwmc, pwm->hwpwm);
> 		return 0;
> 	}
> 
> 	/* set period and duty cycle*/
> 	err = gb_pwm_config_operation(pwmc, pwm->hwpwm, state->duty_cycle, state->period);
> 	if (err)
> 		return err;
> 
> 	/* enable/disable */
> 	if (!enabled)
> 		return gb_pwm_enable_operation(pwmc, pwm->hwpwm);
> 
> 	return 0;
> }

This looks good.

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | https://www.pengutronix.de/ |