[v8,2/3] pwm: sifive: change the PWM controlled LED algorithm

The `frac` variable represents the pulse inactive time, and the result
of this algorithm is the pulse active time. Therefore, we must reverse the result.

The reference is SiFive FU740-C000 Manual[0]

Link: https://sifive.cdn.prismic.io/sifive/1a82e600-1f93-4f41-b2d8-86ed8b16acba_fu740-c000-manual-v1p6.pdf [0]

Co-developed-by: Zong Li <zong.li@sifive.com>
Signed-off-by: Zong Li <zong.li@sifive.com>
Co-developed-by: Vincent Chen <vincent.chen@sifive.com>
Signed-off-by: Vincent Chen <vincent.chen@sifive.com>
Signed-off-by: Nylon Chen <nylon.chen@sifive.com>
---
 drivers/pwm/pwm-sifive.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

Hello,

Regarding the Subject: The patch has nothing to do with an LED, has it?

On Fri, Jan 26, 2024 at 03:40:44PM +0800, Nylon Chen wrote:
> The `frac` variable represents the pulse inactive time, and the result
> of this algorithm is the pulse active time. Therefore, we must reverse the result.

Please break lines at 75 columns in the commit log.

> The reference is SiFive FU740-C000 Manual[0]
> 
> Link: https://sifive.cdn.prismic.io/sifive/1a82e600-1f93-4f41-b2d8-86ed8b16acba_fu740-c000-manual-v1p6.pdf [0]

I looked at Figure 29 in this document (version v1p6, pdf page 148). Not
sure I understand that correctly, but I expect that the output of the
">=?" node below pwmcmp0 to become 1 if pwms has reached pwmcmp0, is
that right? In that case this output is zero when pwmcount is zero and
then pwmcmp0ip is zero, too. So a period starts with the inactive part
and so it's inversed polarity.

What made you think that the current driver implementation is wrong?

> Co-developed-by: Zong Li <zong.li@sifive.com>
> Signed-off-by: Zong Li <zong.li@sifive.com>
> Co-developed-by: Vincent Chen <vincent.chen@sifive.com>
> Signed-off-by: Vincent Chen <vincent.chen@sifive.com>
> Signed-off-by: Nylon Chen <nylon.chen@sifive.com>
> ---
>  drivers/pwm/pwm-sifive.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/pwm/pwm-sifive.c b/drivers/pwm/pwm-sifive.c
> index eabddb7c7820..b07c8598bb21 100644
> --- a/drivers/pwm/pwm-sifive.c
> +++ b/drivers/pwm/pwm-sifive.c
> @@ -113,6 +113,7 @@ static int pwm_sifive_get_state(struct pwm_chip *chip, struct pwm_device *pwm,
>  	u32 duty, val;
>  
>  	duty = readl(ddata->regs + PWM_SIFIVE_PWMCMP(pwm->hwpwm));
> +	duty = (1U << PWM_SIFIVE_CMPWIDTH) - 1 - duty;

I find it irritating that both values are assigned to duty. I'd spend
another variable and make this:

	inactive = readl(ddata->regs + PWM_SIFIVE_PWMCMP(pwm->hwpwm));
	duty = (1U << PWM_SIFIVE_CMPWIDTH) - 1 - inactive;


>  
>  	state->enabled = duty > 0;
>  
> @@ -123,11 +124,10 @@ static int pwm_sifive_get_state(struct pwm_chip *chip, struct pwm_device *pwm,
>  	state->period = ddata->real_period;
>  	state->duty_cycle =
>  		(u64)duty * ddata->real_period >> PWM_SIFIVE_CMPWIDTH;
> -	state->polarity = PWM_POLARITY_INVERSED;
> +	state->polarity = PWM_POLARITY_NORMAL;
>  
>  	return 0;
>  }
> -

Please keep this empty line between functions.

>  static int pwm_sifive_apply(struct pwm_chip *chip, struct pwm_device *pwm,
>  			    const struct pwm_state *state)
>  {

Best regards
Uwe

Uwe Kleine-König <u.kleine-koenig@pengutronix.de> 於 2024年2月6日 週二 上午2:07寫道：
>
> Hello,
Hi Uwe, thanks for your feedback.
>
> Regarding the Subject: The patch has nothing to do with an LED, has it?
I will correct this.
>
> On Fri, Jan 26, 2024 at 03:40:44PM +0800, Nylon Chen wrote:
> > The `frac` variable represents the pulse inactive time, and the result
> > of this algorithm is the pulse active time. Therefore, we must reverse the result.
>
> Please break lines at 75 columns in the commit log.
got it.
>
> > The reference is SiFive FU740-C000 Manual[0]
> >
> > Link: https://sifive.cdn.prismic.io/sifive/1a82e600-1f93-4f41-b2d8-86ed8b16acba_fu740-c000-manual-v1p6.pdf [0]
>
> I looked at Figure 29 in this document (version v1p6, pdf page 148). Not
> sure I understand that correctly, but I expect that the output of the
> ">=?" node below pwmcmp0 to become 1 if pwms has reached pwmcmp0, is
> that right? In that case this output is zero when pwmcount is zero and
> then pwmcmp0ip is zero, too. So a period starts with the inactive part
> and so it's inversed polarity.
>
> What made you think that the current driver implementation is wrong?

This is the process of my speculation.

This is a HiFive Unmatched/Unleashed LED-PWM layout

            VDD
               |
               |
           _____
           \        /   LED
            \     /
              ---
               |
               |
               |
         ______
        |              |
        -             |
        ^    -->    |------ PWM
        |___|___|
               |
               |
              __
               -
            GND

- the waveform
e.g. duty=30s, period=100s, actvie-high = 30%, active-low = 70%

V
^
|
| ----------|
|             |
|             |
|______ |__________ > t

When VCC is high, the LED will be illuminated, which is an active-high
logic. This is why I want to remove "active-low".

For HW, we just focus on pwmcount/pwmcmp[0-3]
- pwmcount default is zero, that counter 0->1->0xffff
- Follow the origin algorithm the frac=0x0(on) / 0xffff(off) and when
the smaller the value of frac, the brighter the light.
   -- E.g. pwmcmp = 0x2, pwmcount 0x0->0x1->...->0xffff
       --- 0->0x2=low & 0x3->0xffff=high => 98%
   -- E.g. pwmcmp = 0xffff, pwmcount 0x0->0x1->...->0xffff
       --- 0->0xffff=low => 0%
- For SW, we reference the algorithm. (D=PW/T*100% D=duty_cycle,
T=period, PW=pulse width (pulse active time))
  -- when we consider HW behavior
  --- Direct writing SW frac into HW's pwmcmp is active low, so when
we want to get an active-high behavior that use a invert function.

If my understanding or deduction process is incorrect, please let me
know. Thank you.
>
> > Co-developed-by: Zong Li <zong.li@sifive.com>
> > Signed-off-by: Zong Li <zong.li@sifive.com>
> > Co-developed-by: Vincent Chen <vincent.chen@sifive.com>
> > Signed-off-by: Vincent Chen <vincent.chen@sifive.com>
> > Signed-off-by: Nylon Chen <nylon.chen@sifive.com>
> > ---
> >  drivers/pwm/pwm-sifive.c | 7 ++++---
> >  1 file changed, 4 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/pwm/pwm-sifive.c b/drivers/pwm/pwm-sifive.c
> > index eabddb7c7820..b07c8598bb21 100644
> > --- a/drivers/pwm/pwm-sifive.c
> > +++ b/drivers/pwm/pwm-sifive.c
> > @@ -113,6 +113,7 @@ static int pwm_sifive_get_state(struct pwm_chip *chip, struct pwm_device *pwm,
> >       u32 duty, val;
> >
> >       duty = readl(ddata->regs + PWM_SIFIVE_PWMCMP(pwm->hwpwm));
> > +     duty = (1U << PWM_SIFIVE_CMPWIDTH) - 1 - duty;
>
> I find it irritating that both values are assigned to duty. I'd spend
> another variable and make this:
>
>         inactive = readl(ddata->regs + PWM_SIFIVE_PWMCMP(pwm->hwpwm));
>         duty = (1U << PWM_SIFIVE_CMPWIDTH) - 1 - inactive;
got it.
>
>
> >
> >       state->enabled = duty > 0;
> >
> > @@ -123,11 +124,10 @@ static int pwm_sifive_get_state(struct pwm_chip *chip, struct pwm_device *pwm,
> >       state->period = ddata->real_period;
> >       state->duty_cycle =
> >               (u64)duty * ddata->real_period >> PWM_SIFIVE_CMPWIDTH;
> > -     state->polarity = PWM_POLARITY_INVERSED;
> > +     state->polarity = PWM_POLARITY_NORMAL;
> >
> >       return 0;
> >  }
> > -
>
> Please keep this empty line between functions.
got it.
>
> >  static int pwm_sifive_apply(struct pwm_chip *chip, struct pwm_device *pwm,
> >                           const struct pwm_state *state)
> >  {
>
> Best regards
> Uwe
>
> --
> Pengutronix e.K.                           | Uwe Kleine-König            |
> Industrial Linux Solutions                 | https://www.pengutronix.de/ |