[PATCH v2 0/1] rtc: tps6594: Add power management support

[PATCH v2 0/1] rtc: tps6594: Add power management support

[Richard Genoud]

Changes in v2:
 - use DEFINE_SIMPLE_DEV_PM_OPS instead of deprecated SIMPLE_DEV_PM_OPS
   (This fixes the defined but unsed warning when CONFIG_PM is unset)

Richard Genoud (1):
  rtc: tps6594: Add power management support

 drivers/rtc/rtc-tps6594.c   | 46 +++++++++++++++++++++++++++++++++++++
 include/linux/mfd/tps6594.h |  1 +
 2 files changed, 47 insertions(+)

[PATCH v2 1/1] rtc: tps6594: Add power management support

[Richard Genoud]

Add power management support to the driver. This allows a SoC to wake
from suspend using the nINT provided by the RTC.
It takes care of the case when the interrupt has not been caught because
the kernel has not yet woke up.
(This is the case when only edges interrupt are caught)

Signed-off-by: Richard Genoud <richard.genoud@bootlin.com>
---
 drivers/rtc/rtc-tps6594.c   | 46 +++++++++++++++++++++++++++++++++++++
 include/linux/mfd/tps6594.h |  1 +
 2 files changed, 47 insertions(+)

diff --git a/drivers/rtc/rtc-tps6594.c b/drivers/rtc/rtc-tps6594.c
index 838ae8562a35..d221dc560682 100644
--- a/drivers/rtc/rtc-tps6594.c
+++ b/drivers/rtc/rtc-tps6594.c
@@ -415,6 +415,8 @@ static int tps6594_rtc_probe(struct platform_device *pdev)
 	if (irq < 0)
 		return dev_err_probe(dev, irq, "Failed to get irq\n");
 
+	tps->irq_rtc = irq;
+
 	ret = devm_request_threaded_irq(dev, irq, NULL, tps6594_rtc_interrupt,
 					IRQF_ONESHOT, TPS6594_IRQ_NAME_ALARM,
 					dev);
@@ -434,6 +436,49 @@ static int tps6594_rtc_probe(struct platform_device *pdev)
 	return devm_rtc_register_device(rtc);
 }
 
+static int tps6594_rtc_resume(struct device *dev)
+{
+	struct tps6594 *tps = dev_get_drvdata(dev->parent);
+	struct rtc_device *rtc_dev = dev_get_drvdata(dev);
+	int ret;
+
+	ret = regmap_test_bits(tps->regmap, TPS6594_REG_INT_STARTUP,
+			       TPS6594_BIT_RTC_INT);
+	if (ret < 0) {
+		dev_err(dev, "failed to read REG_INT_STARTUP: %d\n", ret);
+		goto out;
+	}
+
+	if (ret > 0) {
+		/*
+		 * If the alarm bit is set, it means that the IRQ has been
+		 * fired. But, the kernel may not have woke up yet when it
+		 * happened. So, we have to clear it.
+		 */
+		ret = regmap_write(tps->regmap, TPS6594_REG_RTC_STATUS,
+				   TPS6594_BIT_ALARM);
+		if (ret < 0)
+			dev_err(dev, "error clearing alarm bit: %d", ret);
+
+		rtc_update_irq(rtc_dev, 1, RTC_IRQF | RTC_AF);
+	}
+out:
+	disable_irq_wake(tps->irq_rtc);
+
+	return 0;
+}
+
+static int tps6594_rtc_suspend(struct device *dev)
+{
+	struct tps6594 *tps = dev_get_drvdata(dev->parent);
+
+	enable_irq_wake(tps->irq_rtc);
+
+	return 0;
+}
+
+static DEFINE_SIMPLE_DEV_PM_OPS(tps6594_rtc_pm_ops, tps6594_rtc_suspend, tps6594_rtc_resume);
+
 static const struct platform_device_id tps6594_rtc_id_table[] = {
 	{ "tps6594-rtc", },
 	{}
@@ -444,6 +489,7 @@ static struct platform_driver tps6594_rtc_driver = {
 	.probe		= tps6594_rtc_probe,
 	.driver		= {
 		.name	= "tps6594-rtc",
+		.pm = pm_sleep_ptr(&tps6594_rtc_pm_ops),
 	},
 	.id_table = tps6594_rtc_id_table,
 };
diff --git a/include/linux/mfd/tps6594.h b/include/linux/mfd/tps6594.h
index 3f7c5e23cd4c..85933f1519c4 100644
--- a/include/linux/mfd/tps6594.h
+++ b/include/linux/mfd/tps6594.h
@@ -1011,6 +1011,7 @@ struct tps6594 {
 	bool use_crc;
 	struct regmap *regmap;
 	int irq;
+	int irq_rtc;
 	struct regmap_irq_chip_data *irq_data;
 };
 

Re: [PATCH v2 1/1] rtc: tps6594: Add power management support

[Esteban Blanc]

Hi Richard,

This looks good to me.

Reviewed-by: Esteban Blanc <eblanc@baylibre.com>

Best regards,

-- 
Esteban "Skallwar" Blanc
BayLibre

Re: [PATCH v2 1/1] rtc: tps6594: Add power management support

[Lee Jones]

On Wed, 15 May 2024, Richard Genoud wrote:

> Add power management support to the driver. This allows a SoC to wake
> from suspend using the nINT provided by the RTC.
> It takes care of the case when the interrupt has not been caught because
> the kernel has not yet woke up.
> (This is the case when only edges interrupt are caught)
> 
> Signed-off-by: Richard Genoud <richard.genoud@bootlin.com>
> ---
>  drivers/rtc/rtc-tps6594.c   | 46 +++++++++++++++++++++++++++++++++++++
>  include/linux/mfd/tps6594.h |  1 +

Acked-by: Lee Jones <lee@kernel.org>

>  2 files changed, 47 insertions(+)

-- 
Lee Jones [李琼斯]

Re: [PATCH v2 1/1] rtc: tps6594: Add power management support

[Thomas Richard]

> diff --git a/include/linux/mfd/tps6594.h b/include/linux/mfd/tps6594.h
> index 3f7c5e23cd4c..85933f1519c4 100644
> --- a/include/linux/mfd/tps6594.h
> +++ b/include/linux/mfd/tps6594.h
> @@ -1011,6 +1011,7 @@ struct tps6594 {
>  	bool use_crc;
>  	struct regmap *regmap;
>  	int irq;
> +	int irq_rtc;

irq_rtc should be stored in driver data.

>  	struct regmap_irq_chip_data *irq_data;
>  };
>  

Thomas

Re: [PATCH v2 1/1] rtc: tps6594: Add power management support

[Richard GENOUD]

Le 14/06/2024 à 15:21, Thomas Richard a écrit :
>> diff --git a/include/linux/mfd/tps6594.h b/include/linux/mfd/tps6594.h
>> index 3f7c5e23cd4c..85933f1519c4 100644
>> --- a/include/linux/mfd/tps6594.h
>> +++ b/include/linux/mfd/tps6594.h
>> @@ -1011,6 +1011,7 @@ struct tps6594 {
>>   	bool use_crc;
>>   	struct regmap *regmap;
>>   	int irq;
>> +	int irq_rtc;
> 
> irq_rtc should be stored in driver data.
I understand your point.
It may indeed be cleaner to separate the rtc-specific data into a 
private struct.

I'll send a v3 with this modification, and I'll let Alexandre decide 
which he prefers.

> 
>>   	struct regmap_irq_chip_data *irq_data;
>>   };
>>   
> 
> Thomas

Thanks !

-- 
Richard Genoud, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com

Re: [PATCH v2 1/1] rtc: tps6594: Add power management support

[Alexandre Belloni]

On 18/06/2024 11:48:33+0200, Richard GENOUD wrote:
> Le 14/06/2024 à 15:21, Thomas Richard a écrit :
> > > diff --git a/include/linux/mfd/tps6594.h b/include/linux/mfd/tps6594.h
> > > index 3f7c5e23cd4c..85933f1519c4 100644
> > > --- a/include/linux/mfd/tps6594.h
> > > +++ b/include/linux/mfd/tps6594.h
> > > @@ -1011,6 +1011,7 @@ struct tps6594 {
> > >   	bool use_crc;
> > >   	struct regmap *regmap;
> > >   	int irq;
> > > +	int irq_rtc;
> > 
> > irq_rtc should be stored in driver data.
> I understand your point.
> It may indeed be cleaner to separate the rtc-specific data into a private
> struct.
> 
> I'll send a v3 with this modification, and I'll let Alexandre decide which
> he prefers.

My plan is to take v2 as-is


-- 
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com