All of lore.kernel.org
 help / color / mirror / Atom feed
From: Clemens Gruber <clemens.gruber@pqgruber.com>
To: Sven Van Asbroeck <thesven73@gmail.com>
Cc: linux-pwm@vger.kernel.org,
	"Thierry Reding" <thierry.reding@gmail.com>,
	"Uwe Kleine-König" <u.kleine-koenig@pengutronix.de>,
	"Lee Jones" <lee.jones@linaro.org>,
	"Linux Kernel Mailing List" <linux-kernel@vger.kernel.org>,
	"Mika Westerberg" <mika.westerberg@linux.intel.com>,
	"David Jander" <david@protonic.nl>
Subject: Re: [PATCH v5 5/7] pwm: pca9685: Support staggered output ON times
Date: Thu, 17 Dec 2020 18:50:22 +0100	[thread overview]
Message-ID: <X9uaXpZ5FmKZgZqs@workstation.tuxnet> (raw)
In-Reply-To: <CAGngYiXCS-FwQkA7nzizXw_5417_u7VmPAXPbZeDv4dUFs2cog@mail.gmail.com>

On Wed, Dec 16, 2020 at 11:02:57PM -0500, Sven Van Asbroeck wrote:
> Hi Clemens, see below.
> 
> On Wed, Dec 16, 2020 at 7:53 AM Clemens Gruber
> <clemens.gruber@pqgruber.com> wrote:
> >
> > The PCA9685 supports staggered LED output ON times to minimize current
> > surges and reduce EMI.
> > When this new option is enabled, the ON times of each channel are
> > delayed by channel number x counter range / 16, which avoids asserting
> > all enabled outputs at the same counter value while still maintaining
> > the configured duty cycle of each output.
> >
> > Signed-off-by: Clemens Gruber <clemens.gruber@pqgruber.com>
> > ---
> >  drivers/pwm/pwm-pca9685.c | 62 +++++++++++++++++++++++++++++++--------
> >  1 file changed, 50 insertions(+), 12 deletions(-)
> >
> > diff --git a/drivers/pwm/pwm-pca9685.c b/drivers/pwm/pwm-pca9685.c
> > index 38aadaf50996..ff916980de49 100644
> > --- a/drivers/pwm/pwm-pca9685.c
> > +++ b/drivers/pwm/pwm-pca9685.c
> > @@ -79,6 +79,7 @@
> >  struct pca9685 {
> >         struct pwm_chip chip;
> >         struct regmap *regmap;
> > +       bool staggered_outputs;
> >  #if IS_ENABLED(CONFIG_GPIOLIB)
> >         struct mutex lock;
> >         struct gpio_chip gpio;
> > @@ -93,45 +94,79 @@ static inline struct pca9685 *to_pca(struct pwm_chip *chip)
> >
> >  static void pca9685_pwm_set_duty(struct pca9685 *pca, int channel, unsigned int duty)
> >  {
> > +       unsigned int on, off;
> > +
> >         if (duty == 0) {
> >                 /* Set the full OFF bit, which has the highest precedence */
> >                 regmap_write(pca->regmap, REG_OFF_H(channel), LED_FULL);
> > +               return;
> >         } else if (duty >= PCA9685_COUNTER_RANGE) {
> >                 /* Set the full ON bit and clear the full OFF bit */
> >                 regmap_write(pca->regmap, REG_ON_H(channel), LED_FULL);
> >                 regmap_write(pca->regmap, REG_OFF_H(channel), 0);
> > -       } else {
> > -               /* Set OFF time (clears the full OFF bit) */
> > -               regmap_write(pca->regmap, REG_OFF_L(channel), duty & 0xff);
> > -               regmap_write(pca->regmap, REG_OFF_H(channel), (duty >> 8) & 0xf);
> > -               /* Clear the full ON bit */
> > -               regmap_write(pca->regmap, REG_ON_H(channel), 0);
> > +               return;
> > +       }
> > +
> > +       if (pca->staggered_outputs) {
> > +               if (channel < PCA9685_MAXCHAN) {
> > +                       /*
> > +                        * To reduce EMI, the ON times of each channel are
> > +                        * spread out evenly within the counter range, while
> > +                        * still maintaining the configured duty cycle
> > +                        */
> > +                       on = channel * PCA9685_COUNTER_RANGE / PCA9685_MAXCHAN;
> > +                       off = (on + duty) % PCA9685_COUNTER_RANGE;
> > +                       regmap_write(pca->regmap, REG_ON_L(channel), on & 0xff);
> > +                       regmap_write(pca->regmap, REG_ON_H(channel), (on >> 8) & 0xf);
> > +                       regmap_write(pca->regmap, REG_OFF_L(channel), off & 0xff);
> > +                       regmap_write(pca->regmap, REG_OFF_H(channel), (off >> 8) & 0xf);
> > +                       return;
> > +               }
> > +               /* No staggering possible if "all LEDs" channel is used */
> > +               regmap_write(pca->regmap, PCA9685_ALL_LED_ON_L, 0);
> >         }
> > +       /* Set OFF time (clears the full OFF bit) */
> > +       regmap_write(pca->regmap, REG_OFF_L(channel), duty & 0xff);
> > +       regmap_write(pca->regmap, REG_OFF_H(channel), (duty >> 8) & 0xf);
> > +       /* Clear the full ON bit */
> > +       regmap_write(pca->regmap, REG_ON_H(channel), 0);
> >  }
> 
> I find the set_duty() function quite hard to follow.
> Can we simplify this by eliminating the !staggered_outputs special case?
> E.g. always calculate and write 'on' and 'off'.
> But if !staggered_outputs => on = 0 and off = duty.
> 
> Yes this adds one extra/unnecessary register write in the !staggered case,
> but simplicity/maintainability >>> efficiency here. And this "issue" will
> disappear when we switch on regmap caching.

No objections, I will eliminate the special case.

> 
> >
> >  static unsigned int pca9685_pwm_get_duty(struct pca9685 *pca, int channel)
> >  {
> > -       unsigned int off_h, val;
> > +       unsigned int off, on, val;
> >
> >         if (WARN_ON(channel >= PCA9685_MAXCHAN)) {
> >                 /* Hardware readout not supported for "all LEDs" channel */
> >                 return 0;
> >         }
> >
> > -       regmap_read(pca->regmap, LED_N_OFF_H(channel), &off_h);
> > -       if (off_h & LED_FULL) {
> > +       regmap_read(pca->regmap, LED_N_OFF_H(channel), &off);
> > +       if (off & LED_FULL) {
> >                 /* Full OFF bit is set */
> >                 return 0;
> >         }
> >
> > -       regmap_read(pca->regmap, LED_N_ON_H(channel), &val);
> > -       if (val & LED_FULL) {
> > +       regmap_read(pca->regmap, LED_N_ON_H(channel), &on);
> > +       if (on & LED_FULL) {
> >                 /* Full ON bit is set */
> >                 return PCA9685_COUNTER_RANGE;
> >         }
> >
> >         regmap_read(pca->regmap, LED_N_OFF_L(channel), &val);
> > -       return ((off_h & 0xf) << 8) | (val & 0xff);
> > +       off = ((off & 0xf) << 8) | (val & 0xff);
> > +
> > +       if (pca->staggered_outputs) {
> > +               regmap_read(pca->regmap, LED_N_ON_L(channel), &val);
> > +               on = ((on & 0xf) << 8) | (val & 0xff);
> > +
> > +               if (off >= on)
> > +                       return off - on;
> > +               else
> > +                       return off + PCA9685_COUNTER_RANGE - on;
> 
> I think the if/else is unnecessary. unsigned int is twos-complement,
> so we can just write:
> 
>         return (off - on) & 0xfff; (or PCA9685_COUNTER_RANGE - 1)
> 
> and it will "magically" work even if off < on.

Ah, yes of course. Good catch!

> 
> 
> > +       }
> > +
> > +       return off;
> >  }
> 
> As in set_duty(), consider removing the !staggered_outputs special case,
> to make this function clearer and simpler.

Yes, thanks.

> 
> >
> >  #if IS_ENABLED(CONFIG_GPIOLIB)
> > @@ -442,6 +477,9 @@ static int pca9685_pwm_probe(struct i2c_client *client,
> >
> >         regmap_write(pca->regmap, PCA9685_MODE2, reg);
> >
> > +       pca->staggered_outputs = device_property_read_bool(
> > +               &client->dev, "nxp,staggered-outputs");
> > +
> >         /* Disable all LED ALLCALL and SUBx addresses to avoid bus collisions */
> >         regmap_read(pca->regmap, PCA9685_MODE1, &reg);
> >         reg &= ~(MODE1_ALLCALL | MODE1_SUB1 | MODE1_SUB2 | MODE1_SUB3);
> > --
> > 2.29.2
> >

  reply	other threads:[~2020-12-17 17:51 UTC|newest]

Thread overview: 25+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-12-15 21:22 [PATCH v5 1/7] pwm: pca9685: Switch to atomic API Clemens Gruber
2020-12-15 21:22 ` [PATCH v5 2/7] pwm: pca9685: Support hardware readout Clemens Gruber
2020-12-15 21:22 ` [PATCH v5 3/7] pwm: pca9685: Improve runtime PM behavior Clemens Gruber
2020-12-15 21:22 ` [PATCH v5 4/7] pwm: pca9685: Reset registers to POR state in probe Clemens Gruber
2020-12-17  4:02   ` Sven Van Asbroeck
2020-12-17 17:45     ` Clemens Gruber
2021-03-01 21:46   ` Uwe Kleine-König
2021-03-04 13:16     ` Clemens Gruber
2020-12-15 21:22 ` [PATCH v5 5/7] pwm: pca9685: Support staggered output ON times Clemens Gruber
2020-12-17  4:02   ` Sven Van Asbroeck
2020-12-17 17:50     ` Clemens Gruber [this message]
2020-12-15 21:22 ` [PATCH v5 6/7] dt-bindings: pwm: pca9685: Add nxp,staggered-outputs property Clemens Gruber
2020-12-15 21:22 ` [PATCH v5 7/7] pwm: pca9685: Restrict period change for prescaler users Clemens Gruber
2020-12-17  4:03   ` Sven Van Asbroeck
2020-12-17 18:07     ` Clemens Gruber
2020-12-17 18:17       ` Sven Van Asbroeck
2020-12-17  3:58 ` [PATCH v5 1/7] pwm: pca9685: Switch to atomic API Sven Van Asbroeck
2020-12-17 16:48   ` Clemens Gruber
2020-12-17 17:10     ` Sven Van Asbroeck
2021-03-01 21:41       ` Uwe Kleine-König
2021-03-04 13:10         ` Clemens Gruber
2021-03-22  7:58       ` Thierry Reding
2021-03-27 15:54         ` Clemens Gruber
2021-03-01 21:44 ` Uwe Kleine-König
2021-03-04 13:12   ` Clemens Gruber

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=X9uaXpZ5FmKZgZqs@workstation.tuxnet \
    --to=clemens.gruber@pqgruber.com \
    --cc=david@protonic.nl \
    --cc=lee.jones@linaro.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-pwm@vger.kernel.org \
    --cc=mika.westerberg@linux.intel.com \
    --cc=thesven73@gmail.com \
    --cc=thierry.reding@gmail.com \
    --cc=u.kleine-koenig@pengutronix.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.