Re: [PATCH 2/2] regulator: core: Provide per-regulator runtime PM support

[Paul Kocialkowski <contact@paulk.fr>]

[-- Attachment #1: Type: text/plain, Size: 3394 bytes --]

Le jeudi 21 janvier 2016 à 20:24 +0000, Mark Brown a écrit :
> Provide a flag auto_runtime_pm in the regulator_desc which causes the
> regulator core to take a runtime PM reference to a regulator while it
> is enabled. This helps integration with chip wide power management
> for
> auxiliary PMICs, they may be able to implement chip wide power
> savings
> if nothing on the PMIC is in use.

Thanks for working on this!

I'm having a major drawback on my development unit (fried it
accidentally) so I'm unable to test this with the LP8720 regulator on
the Optimus Black, but I'll manage to get a working development unit
soon.

I'll let you know how it goes. Feel free to merge this before I do the
work on the lp8720 regulator, though.

> Signed-off-by: Mark Brown <broonie@kernel.org>
> ---
> 
> Not tested at all yet, pushing out for testing by others who have
> devices that could benefit from this.
> 
>  drivers/regulator/core.c         | 20 ++++++++++++++++++--
>  include/linux/regulator/driver.h |  1 +
>  2 files changed, 19 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/regulator/core.c b/drivers/regulator/core.c
> index 3308c6bb83db..968ff3081e6c 100644
> --- a/drivers/regulator/core.c
> +++ b/drivers/regulator/core.c
> @@ -26,6 +26,7 @@
>  #include <linux/gpio.h>
>  #include <linux/gpio/consumer.h>
>  #include <linux/of.h>
> +#include <linux/pm_runtime.h>
>  #include <linux/regmap.h>
>  #include <linux/regulator/of_regulator.h>
>  #include <linux/regulator/consumer.h>
> @@ -2059,17 +2060,23 @@ static int _regulator_do_enable(struct
> regulator_dev *rdev)
>  		}
>  	}
>  
> +	if (rdev->desc->auto_runtime_pm) {
> +		ret = pm_runtime_get_sync(rdev->dev.parent);
> +		if (ret < 0)
> +			goto err;
> +	}
> +
>  	if (rdev->ena_pin) {
>  		if (!rdev->ena_gpio_state) {
>  			ret = regulator_ena_gpio_ctrl(rdev, true);
>  			if (ret < 0)
> -				return ret;
> +				goto err_pm;
>  			rdev->ena_gpio_state = 1;
>  		}
>  	} else if (rdev->desc->ops->enable) {
>  		ret = rdev->desc->ops->enable(rdev);
>  		if (ret < 0)
> -			return ret;
> +			goto err_pm;
>  	} else {
>  		return -EINVAL;
>  	}
> @@ -2084,6 +2091,12 @@ static int _regulator_do_enable(struct
> regulator_dev *rdev)
>  	trace_regulator_enable_complete(rdev_get_name(rdev));
>  
>  	return 0;
> +
> +err_pm:
> +	if (rdev->desc->auto_runtime_pm)
> +		pm_runtime_put_autosuspend(rdev->dev.parent);
> +err:
> +	return ret;
>  }
>  
>  /* locks held by regulator_enable() */
> @@ -2177,6 +2190,9 @@ static int _regulator_do_disable(struct
> regulator_dev *rdev)
>  			return ret;
>  	}
>  
> +	if (rdev->desc->auto_runtime_pm)
> +		pm_runtime_put_autosuspend(rdev->dev.parent);
> +
>  	/* cares about last_off_jiffy only if off_on_delay is
> required by
>  	 * device.
>  	 */
> diff --git a/include/linux/regulator/driver.h
> b/include/linux/regulator/driver.h
> index 3ac0f306f033..dccea032a143 100644
> --- a/include/linux/regulator/driver.h
> +++ b/include/linux/regulator/driver.h
> @@ -282,6 +282,7 @@ struct regulator_desc {
>  			    struct regulator_config *);
>  	int id;
>  	unsigned int continuous_voltage_range:1;
> +	unsigned int auto_runtime_pm:1;
>  	unsigned n_voltages;
>  	const struct regulator_ops *ops;
>  	int irq;

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 819 bytes --]