[v7,4/7] pwm: ntxec: Add driver for PWM function in Netronix EC

The Netronix EC provides a PWM output which is used for the backlight
on some ebook readers. This patches adds a driver for the PWM output.

The .get_state callback is not implemented, because the PWM state can't
be read back from the hardware.

Signed-off-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
---
v7:
- no changes

v6:
- https://lore.kernel.org/lkml/20201208011000.3060239-5-j.neuschaefer@gmx.net/
- Move period / duty cycle setting code to a function
- Rename pwmchip_to_priv to ntxec_pwm_from_chip
- Set period and duty cycle only before enabling the output
- Mention that duty=0, enable=1 is assumed not to happen
- Interleave writes to the period and duty cycle registers, to minimize the
  window of time that an inconsistent state is configured

v5:
- https://lore.kernel.org/lkml/20201201011513.1627028-5-j.neuschaefer@gmx.net/
- Avoid truncation of period and duty cycle to 32 bits
- Make ntxec_pwm_ops const
- Use regmap_multi_reg_write
- Add comment about get_state to ntxec_pwm_ops
- Add comments about non-atomicity of (period, duty cycle) update

v4:
- https://lore.kernel.org/lkml/20201122222739.1455132-5-j.neuschaefer@gmx.net/
- Document hardware/driver limitations
- Only accept normal polarity
- Fix a typo ("zone" -> "zero")
- change MAX_PERIOD_NS to 0xffff * 125
- Clamp period to the maximum rather than returning an error
- Rename private struct pointer to priv
- Rearrage control flow in _probe to save a few lines and a temporary variable
- Add missing MODULE_ALIAS line
- Spell out ODM

v3:
- https://lore.kernel.org/lkml/20200924192455.2484005-5-j.neuschaefer@gmx.net/
- Relicense as GPLv2 or later
- Add email address to copyright line
- Remove OF compatible string and don't include linux/of_device.h
- Fix bogus ?: in return line
- Don't use a comma after sentinels
- Avoid ret |= ... pattern
- Move 8-bit register conversion to ntxec.h

v2:
- https://lore.kernel.org/lkml/20200905133230.1014581-6-j.neuschaefer@gmx.net/
- Various grammar and style improvements, as suggested by Uwe Kleine-König,
  Lee Jones, and Alexandre Belloni
- Switch to regmap
- Prefix registers with NTXEC_REG_
- Add help text to the Kconfig option
- Use the .apply callback instead of the old API
- Add a #define for the time base (125ns)
- Don't change device state in .probe; this avoids multiple problems
- Rework division and overflow check logic to perform divisions in 32 bits
- Avoid setting duty cycle to zero, to work around a hardware quirk
---
 drivers/pwm/Kconfig     |   8 ++
 drivers/pwm/Makefile    |   1 +
 drivers/pwm/pwm-ntxec.c | 182 ++++++++++++++++++++++++++++++++++++++++
 3 files changed, 191 insertions(+)
 create mode 100644 drivers/pwm/pwm-ntxec.c

--
2.29.2

Hello,

On Sat, Jan 09, 2021 at 07:02:17PM +0100, Jonathan Neuschäfer wrote:
> The Netronix EC provides a PWM output which is used for the backlight
> on some ebook readers. This patches adds a driver for the PWM output.
> 
> The .get_state callback is not implemented, because the PWM state can't
> be read back from the hardware.

There is only very little I would have done differently (only indention
which is too minor to make a fuss), so:

Reviewed-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
 
Thanks for your work
Uwe

On Sat, Jan 09, 2021 at 07:02:17PM +0100, Jonathan Neuschäfer wrote:
> The Netronix EC provides a PWM output which is used for the backlight
> on some ebook readers. This patches adds a driver for the PWM output.
> 
> The .get_state callback is not implemented, because the PWM state can't
> be read back from the hardware.
> 
> Signed-off-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
> ---
> v7:
> - no changes
> 
> v6:
> - https://lore.kernel.org/lkml/20201208011000.3060239-5-j.neuschaefer@gmx.net/
> - Move period / duty cycle setting code to a function
> - Rename pwmchip_to_priv to ntxec_pwm_from_chip
> - Set period and duty cycle only before enabling the output
> - Mention that duty=0, enable=1 is assumed not to happen
> - Interleave writes to the period and duty cycle registers, to minimize the
>   window of time that an inconsistent state is configured
> 
> v5:
> - https://lore.kernel.org/lkml/20201201011513.1627028-5-j.neuschaefer@gmx.net/
> - Avoid truncation of period and duty cycle to 32 bits
> - Make ntxec_pwm_ops const
> - Use regmap_multi_reg_write
> - Add comment about get_state to ntxec_pwm_ops
> - Add comments about non-atomicity of (period, duty cycle) update
> 
> v4:
> - https://lore.kernel.org/lkml/20201122222739.1455132-5-j.neuschaefer@gmx.net/
> - Document hardware/driver limitations
> - Only accept normal polarity
> - Fix a typo ("zone" -> "zero")
> - change MAX_PERIOD_NS to 0xffff * 125
> - Clamp period to the maximum rather than returning an error
> - Rename private struct pointer to priv
> - Rearrage control flow in _probe to save a few lines and a temporary variable
> - Add missing MODULE_ALIAS line
> - Spell out ODM
> 
> v3:
> - https://lore.kernel.org/lkml/20200924192455.2484005-5-j.neuschaefer@gmx.net/
> - Relicense as GPLv2 or later
> - Add email address to copyright line
> - Remove OF compatible string and don't include linux/of_device.h
> - Fix bogus ?: in return line
> - Don't use a comma after sentinels
> - Avoid ret |= ... pattern
> - Move 8-bit register conversion to ntxec.h
> 
> v2:
> - https://lore.kernel.org/lkml/20200905133230.1014581-6-j.neuschaefer@gmx.net/
> - Various grammar and style improvements, as suggested by Uwe Kleine-König,
>   Lee Jones, and Alexandre Belloni
> - Switch to regmap
> - Prefix registers with NTXEC_REG_
> - Add help text to the Kconfig option
> - Use the .apply callback instead of the old API
> - Add a #define for the time base (125ns)
> - Don't change device state in .probe; this avoids multiple problems
> - Rework division and overflow check logic to perform divisions in 32 bits
> - Avoid setting duty cycle to zero, to work around a hardware quirk
> ---
>  drivers/pwm/Kconfig     |   8 ++
>  drivers/pwm/Makefile    |   1 +
>  drivers/pwm/pwm-ntxec.c | 182 ++++++++++++++++++++++++++++++++++++++++
>  3 files changed, 191 insertions(+)
>  create mode 100644 drivers/pwm/pwm-ntxec.c

Lee, I assume you'll want to pick the whole set up into the MFD tree? If
so:

Acked-by: Thierry Reding <thierry.reding@gmail.com>

On Mon, 11 Jan 2021, Thierry Reding wrote:

> On Sat, Jan 09, 2021 at 07:02:17PM +0100, Jonathan Neuschäfer wrote:
> > The Netronix EC provides a PWM output which is used for the backlight
> > on some ebook readers. This patches adds a driver for the PWM output.
> > 
> > The .get_state callback is not implemented, because the PWM state can't
> > be read back from the hardware.
> > 
> > Signed-off-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
> > ---
> > v7:
> > - no changes
> > 
> > v6:
> > - https://lore.kernel.org/lkml/20201208011000.3060239-5-j.neuschaefer@gmx.net/
> > - Move period / duty cycle setting code to a function
> > - Rename pwmchip_to_priv to ntxec_pwm_from_chip
> > - Set period and duty cycle only before enabling the output
> > - Mention that duty=0, enable=1 is assumed not to happen
> > - Interleave writes to the period and duty cycle registers, to minimize the
> >   window of time that an inconsistent state is configured
> > 
> > v5:
> > - https://lore.kernel.org/lkml/20201201011513.1627028-5-j.neuschaefer@gmx.net/
> > - Avoid truncation of period and duty cycle to 32 bits
> > - Make ntxec_pwm_ops const
> > - Use regmap_multi_reg_write
> > - Add comment about get_state to ntxec_pwm_ops
> > - Add comments about non-atomicity of (period, duty cycle) update
> > 
> > v4:
> > - https://lore.kernel.org/lkml/20201122222739.1455132-5-j.neuschaefer@gmx.net/
> > - Document hardware/driver limitations
> > - Only accept normal polarity
> > - Fix a typo ("zone" -> "zero")
> > - change MAX_PERIOD_NS to 0xffff * 125
> > - Clamp period to the maximum rather than returning an error
> > - Rename private struct pointer to priv
> > - Rearrage control flow in _probe to save a few lines and a temporary variable
> > - Add missing MODULE_ALIAS line
> > - Spell out ODM
> > 
> > v3:
> > - https://lore.kernel.org/lkml/20200924192455.2484005-5-j.neuschaefer@gmx.net/
> > - Relicense as GPLv2 or later
> > - Add email address to copyright line
> > - Remove OF compatible string and don't include linux/of_device.h
> > - Fix bogus ?: in return line
> > - Don't use a comma after sentinels
> > - Avoid ret |= ... pattern
> > - Move 8-bit register conversion to ntxec.h
> > 
> > v2:
> > - https://lore.kernel.org/lkml/20200905133230.1014581-6-j.neuschaefer@gmx.net/
> > - Various grammar and style improvements, as suggested by Uwe Kleine-König,
> >   Lee Jones, and Alexandre Belloni
> > - Switch to regmap
> > - Prefix registers with NTXEC_REG_
> > - Add help text to the Kconfig option
> > - Use the .apply callback instead of the old API
> > - Add a #define for the time base (125ns)
> > - Don't change device state in .probe; this avoids multiple problems
> > - Rework division and overflow check logic to perform divisions in 32 bits
> > - Avoid setting duty cycle to zero, to work around a hardware quirk
> > ---
> >  drivers/pwm/Kconfig     |   8 ++
> >  drivers/pwm/Makefile    |   1 +
> >  drivers/pwm/pwm-ntxec.c | 182 ++++++++++++++++++++++++++++++++++++++++
> >  3 files changed, 191 insertions(+)
> >  create mode 100644 drivers/pwm/pwm-ntxec.c
> 
> Lee, I assume you'll want to pick the whole set up into the MFD tree? If
> so:

Yes, I'll pick this up once we have all the Acks.

The Arm parts usually go in separately.

> Acked-by: Thierry Reding <thierry.reding@gmail.com>

Thanks.

On Sat,  9 Jan 2021 19:02:17 +0100
Jonathan Neuschäfer <j.neuschaefer@gmx.net> wrote:

> The Netronix EC provides a PWM output which is used for the backlight
> on some ebook readers. This patches adds a driver for the PWM output.
> 
> The .get_state callback is not implemented, because the PWM state can't
> be read back from the hardware.
> 
> Signed-off-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
> ---
> v7:
> - no changes
> 
> v6:
> - https://lore.kernel.org/lkml/20201208011000.3060239-5-j.neuschaefer@gmx.net/
> - Move period / duty cycle setting code to a function
> - Rename pwmchip_to_priv to ntxec_pwm_from_chip
> - Set period and duty cycle only before enabling the output
> - Mention that duty=0, enable=1 is assumed not to happen
> - Interleave writes to the period and duty cycle registers, to minimize the
>   window of time that an inconsistent state is configured
> 
> v5:
> - https://lore.kernel.org/lkml/20201201011513.1627028-5-j.neuschaefer@gmx.net/
> - Avoid truncation of period and duty cycle to 32 bits
> - Make ntxec_pwm_ops const
> - Use regmap_multi_reg_write
> - Add comment about get_state to ntxec_pwm_ops
> - Add comments about non-atomicity of (period, duty cycle) update
> 
> v4:
> - https://lore.kernel.org/lkml/20201122222739.1455132-5-j.neuschaefer@gmx.net/
> - Document hardware/driver limitations
> - Only accept normal polarity
> - Fix a typo ("zone" -> "zero")
> - change MAX_PERIOD_NS to 0xffff * 125
> - Clamp period to the maximum rather than returning an error
> - Rename private struct pointer to priv
> - Rearrage control flow in _probe to save a few lines and a temporary variable
> - Add missing MODULE_ALIAS line
> - Spell out ODM
> 
> v3:
> - https://lore.kernel.org/lkml/20200924192455.2484005-5-j.neuschaefer@gmx.net/
> - Relicense as GPLv2 or later
> - Add email address to copyright line
> - Remove OF compatible string and don't include linux/of_device.h
> - Fix bogus ?: in return line
> - Don't use a comma after sentinels
> - Avoid ret |= ... pattern
> - Move 8-bit register conversion to ntxec.h
> 
> v2:
> - https://lore.kernel.org/lkml/20200905133230.1014581-6-j.neuschaefer@gmx.net/
> - Various grammar and style improvements, as suggested by Uwe Kleine-König,
>   Lee Jones, and Alexandre Belloni
> - Switch to regmap
> - Prefix registers with NTXEC_REG_
> - Add help text to the Kconfig option
> - Use the .apply callback instead of the old API
> - Add a #define for the time base (125ns)
> - Don't change device state in .probe; this avoids multiple problems
> - Rework division and overflow check logic to perform divisions in 32 bits
> - Avoid setting duty cycle to zero, to work around a hardware quirk
> ---
>  drivers/pwm/Kconfig     |   8 ++
>  drivers/pwm/Makefile    |   1 +
>  drivers/pwm/pwm-ntxec.c | 182 ++++++++++++++++++++++++++++++++++++++++
>  3 files changed, 191 insertions(+)
>  create mode 100644 drivers/pwm/pwm-ntxec.c
> 
> diff --git a/drivers/pwm/Kconfig b/drivers/pwm/Kconfig
> index 0937e1c047acb..a2830b8832b97 100644
> --- a/drivers/pwm/Kconfig
> +++ b/drivers/pwm/Kconfig
> @@ -393,6 +393,14 @@ config PWM_MXS
>  	  To compile this driver as a module, choose M here: the module
>  	  will be called pwm-mxs.
> 
> +config PWM_NTXEC
> +	tristate "Netronix embedded controller PWM support"
> +	depends on MFD_NTXEC
> +	help
> +	  Say yes here if you want to support the PWM output of the embedded
> +	  controller found in certain e-book readers designed by the original
> +	  design manufacturer Netronix.
> +
>  config PWM_OMAP_DMTIMER
>  	tristate "OMAP Dual-Mode Timer PWM support"
>  	depends on OF
> diff --git a/drivers/pwm/Makefile b/drivers/pwm/Makefile
> index 18b89d7fd092a..7d97eb595bbef 100644
> --- a/drivers/pwm/Makefile
> +++ b/drivers/pwm/Makefile
> @@ -35,6 +35,7 @@ obj-$(CONFIG_PWM_MESON)		+= pwm-meson.o
>  obj-$(CONFIG_PWM_MEDIATEK)	+= pwm-mediatek.o
>  obj-$(CONFIG_PWM_MTK_DISP)	+= pwm-mtk-disp.o
>  obj-$(CONFIG_PWM_MXS)		+= pwm-mxs.o
> +obj-$(CONFIG_PWM_NTXEC)		+= pwm-ntxec.o
>  obj-$(CONFIG_PWM_OMAP_DMTIMER)	+= pwm-omap-dmtimer.o
>  obj-$(CONFIG_PWM_PCA9685)	+= pwm-pca9685.o
>  obj-$(CONFIG_PWM_PXA)		+= pwm-pxa.o
> diff --git a/drivers/pwm/pwm-ntxec.c b/drivers/pwm/pwm-ntxec.c
> new file mode 100644
> index 0000000000000..1db30a6caa3ad
> --- /dev/null
> +++ b/drivers/pwm/pwm-ntxec.c
> @@ -0,0 +1,182 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +/*
> + * The Netronix embedded controller is a microcontroller found in some
> + * e-book readers designed by the original design manufacturer Netronix, Inc.
> + * It contains RTC, battery monitoring, system power management, and PWM
> + * functionality.
> + *
> + * This driver implements PWM output.
> + *
> + * Copyright 2020 Jonathan Neuschäfer <j.neuschaefer@gmx.net>
> + *
> + * Limitations:
> + * - The get_state callback is not implemented, because the current state of
> + *   the PWM output can't be read back from the hardware.
> + * - The hardware can only generate normal polarity output.
> + * - The period and duty cycle can't be changed together in one atomic action.
> + */
> +
> +#include <linux/mfd/ntxec.h>
> +#include <linux/module.h>
> +#include <linux/platform_device.h>
> +#include <linux/pwm.h>
> +#include <linux/regmap.h>
> +#include <linux/types.h>
> +
> +struct ntxec_pwm {
> +	struct device *dev;
> +	struct ntxec *ec;
> +	struct pwm_chip chip;
> +};
> +
> +static struct ntxec_pwm *ntxec_pwm_from_chip(struct pwm_chip *chip)
> +{
> +	return container_of(chip, struct ntxec_pwm, chip);
> +}
> +
> +#define NTXEC_REG_AUTO_OFF_HI	0xa1
> +#define NTXEC_REG_AUTO_OFF_LO	0xa2
> +#define NTXEC_REG_ENABLE	0xa3
> +#define NTXEC_REG_PERIOD_LOW	0xa4
> +#define NTXEC_REG_PERIOD_HIGH	0xa5
> +#define NTXEC_REG_DUTY_LOW	0xa6
> +#define NTXEC_REG_DUTY_HIGH	0xa7
> +
> +/*
> + * The time base used in the EC is 8MHz, or 125ns. Period and duty cycle are
> + * measured in this unit.
> + */
> +#define TIME_BASE_NS 125
> +
> +/*
> + * The maximum input value (in nanoseconds) is determined by the time base and
> + * the range of the hardware registers that hold the converted value.
> + * It fits into 32 bits, so we can do our calculations in 32 bits as well.
> + */
> +#define MAX_PERIOD_NS (TIME_BASE_NS * 0xffff)
> +
> +static int ntxec_pwm_set_raw_period_and_duty_cycle(struct pwm_chip *chip,
> +						   int period, int duty)
> +{
> +	struct ntxec_pwm *priv = ntxec_pwm_from_chip(chip);
> +
> +	/*
> +	 * Changes to the period and duty cycle take effect as soon as the
> +	 * corresponding low byte is written, so the hardware may be configured
> +	 * to an inconsistent state after the period is written and before the
> +	 * duty cycle is fully written. If, in such a case, the old duty cycle
> +	 * is longer than the new period, the EC may output 100% for a moment.
> +	 *
> +	 * To minimize the time between the changes to period and duty cycle
> +	 * taking effect, the writes are interleaved.
> +	 */
> +
> +	struct reg_sequence regs[] = {
> +		{ NTXEC_REG_PERIOD_HIGH, ntxec_reg8(period >> 8) },
> +		{ NTXEC_REG_DUTY_HIGH, ntxec_reg8(duty >> 8) },
> +		{ NTXEC_REG_PERIOD_LOW, ntxec_reg8(period) },
> +		{ NTXEC_REG_DUTY_LOW, ntxec_reg8(duty) },
> +	};
> +
> +	return regmap_multi_reg_write(priv->ec->regmap, regs, ARRAY_SIZE(regs));
> +}
> +
> +static int ntxec_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm_dev,
> +			   const struct pwm_state *state)
> +{
> +	struct ntxec_pwm *priv = ntxec_pwm_from_chip(chip);
> +	unsigned int period, duty;
> +	int res;
> +
> +	if (state->polarity != PWM_POLARITY_NORMAL)
> +		return -EINVAL;
> +
> +	period = min_t(u64, state->period, MAX_PERIOD_NS);
> +	duty   = min_t(u64, state->duty_cycle, period);
> +
> +	period /= TIME_BASE_NS;
> +	duty   /= TIME_BASE_NS;
> +
> +	/*
> +	 * Writing a duty cycle of zero puts the device into a state where
> +	 * writing a higher duty cycle doesn't result in the brightness that it
> +	 * usually results in. This can be fixed by cycling the ENABLE register.
> +	 *
> +	 * As a workaround, write ENABLE=0 when the duty cycle is zero.
> +	 * The case that something has previously set the duty cycle to zero
> +	 * but ENABLE=1, is not handled.
> +	 */
> +	if (state->enabled && duty != 0) {
> +		res = ntxec_pwm_set_raw_period_and_duty_cycle(chip, period, duty);
> +		if (res)
> +			return res;
> +
> +		res = regmap_write(priv->ec->regmap, NTXEC_REG_ENABLE, ntxec_reg8(1));
> +		if (res)
> +			return res;
> +
> +		/* Disable the auto-off timer */
> +		res = regmap_write(priv->ec->regmap, NTXEC_REG_AUTO_OFF_HI, ntxec_reg8(0xff));
> +		if (res)
> +			return res;
> +
> +		return regmap_write(priv->ec->regmap, NTXEC_REG_AUTO_OFF_LO, ntxec_reg8(0xff));
> +	} else {
> +		return regmap_write(priv->ec->regmap, NTXEC_REG_ENABLE, ntxec_reg8(0));
> +	}
> +}
> +
> +static const struct pwm_ops ntxec_pwm_ops = {
> +	.owner = THIS_MODULE,
> +	.apply = ntxec_pwm_apply,
> +	/*
> +	 * No .get_state callback, because the current state cannot be read
> +	 * back from the hardware.
> +	 */
> +};
> +
> +static int ntxec_pwm_probe(struct platform_device *pdev)
> +{
> +	struct ntxec *ec = dev_get_drvdata(pdev->dev.parent);
> +	struct ntxec_pwm *priv;
> +	struct pwm_chip *chip;
> +
> +	priv = devm_kzalloc(&pdev->dev, sizeof(*priv), GFP_KERNEL);
> +	if (!priv)
> +		return -ENOMEM;
> +
> +	priv->ec = ec;
> +	priv->dev = &pdev->dev;
> +
> +	platform_set_drvdata(pdev, priv);
> +
> +	chip = &priv->chip;
> +	chip->dev = &pdev->dev;

Hmm, I needed
chip->dev = &pdev->dev.parent to use the backlight example
in patch 2/7.
Not sure what the correct solution is. Maybe the pwm deserves its own
devicetree node.

Regards,
Andreas

On Tue, 12 Jan 2021 20:39:02 +0100
Andreas Kemnade <andreas@kemnade.info> wrote:

[...]
> > +static int ntxec_pwm_probe(struct platform_device *pdev)
> > +{
> > +	struct ntxec *ec = dev_get_drvdata(pdev->dev.parent);
> > +	struct ntxec_pwm *priv;
> > +	struct pwm_chip *chip;
> > +
> > +	priv = devm_kzalloc(&pdev->dev, sizeof(*priv), GFP_KERNEL);
> > +	if (!priv)
> > +		return -ENOMEM;
> > +
> > +	priv->ec = ec;
> > +	priv->dev = &pdev->dev;
> > +
> > +	platform_set_drvdata(pdev, priv);
> > +
> > +	chip = &priv->chip;
> > +	chip->dev = &pdev->dev;  
> 
> Hmm, I needed
> chip->dev = &pdev->dev.parent to use the backlight example
> in patch 2/7.
> Not sure what the correct solution is. Maybe the pwm deserves its own
> devicetree node.
> 
probably just assigning the node from the parent.

   pdev->dev.of_node = pdev->dev.parent->of_node;

Regards,
Andreas

On Wed, Jan 13, 2021 at 11:46:45PM +0100, Andreas Kemnade wrote:
> On Tue, 12 Jan 2021 20:39:02 +0100
> Andreas Kemnade <andreas@kemnade.info> wrote:
> 
> [...]
> > > +static int ntxec_pwm_probe(struct platform_device *pdev)
[...]
> > Hmm, I needed
> > chip->dev = &pdev->dev.parent to use the backlight example
> > in patch 2/7.
> > Not sure what the correct solution is. Maybe the pwm deserves its own
> > devicetree node.
> > 
> probably just assigning the node from the parent.
> 
>    pdev->dev.of_node = pdev->dev.parent->of_node;

Ah, good catch, thanks. This works, I'll add it to the next version.


Jonathan