[RFC PATCH 0/2] Introduce PM runtime helper functions

[RFC PATCH 0/2] Introduce PM runtime helper functions

[Daniel Baluta]

Working on a new driver I noticed that there is a fair amount of duplicate
code for adding PM runtime support.

First patch refactors the PM runtime support in a new header
 <linux/iio/pm_runtime> introducing functions for setup/cleanup and
set power state.

Second patch modifies the KXCJK-1013 driver to use the newly created API. If
this is ok follow up patches will modify the rest of the drivers.

There is a small difference for hid-sensors where the setup sequence additionally
calls pm_suspend_ignore_children. This is taken care of by introducing a 3rd
parameter to iio_pm_runtime_setup.

Daniel Baluta (2):
  iio: pm_runtime: Introduce PM runtime helper functions
  iio: accel: kxcjk1013: Use the new IIO pm runtime helpers

 drivers/iio/accel/kxcjk-1013.c | 56 ++++++++++---------------------------
 include/linux/iio/pm_runtime.h | 63 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 77 insertions(+), 42 deletions(-)
 create mode 100644 include/linux/iio/pm_runtime.h

-- 
1.9.1

[RFC PATCH 1/2] iio: pm_runtime: Introduce PM runtime helper functions

[Daniel Baluta]

We need this in order to avoid reimplementing the same functions each time
we add PM runtime support in a driver.

Simple grep shows the following users:
	* accel/mma9551.c
	* accel/mmc9553.c
	* accel/kxcjk1013.c
	* accel/bmc150-accel.c
	* gyro/bmg160.c
	* imu/kmx61.c
	* common/hid-sensors.

Signed-off-by: Daniel Baluta <daniel.baluta@intel.com>
---
 include/linux/iio/pm_runtime.h | 63 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 63 insertions(+)
 create mode 100644 include/linux/iio/pm_runtime.h

diff --git a/include/linux/iio/pm_runtime.h b/include/linux/iio/pm_runtime.h
new file mode 100644
index 0000000..dc2bca7
--- /dev/null
+++ b/include/linux/iio/pm_runtime.h
@@ -0,0 +1,63 @@
+/*
+ * Industrial I/O runtime PM helper functions
+ *
+ * Copyright (c) 2015, Intel Corporation.
+ *
+ * This file is subject to the terms and conditions of version 2 of
+ * the GNU General Public License.  See the file COPYING in the main
+ * directory of this archive for more details.
+ *
+ */
+#ifndef __IIO_PM_RUNTIME
+#define __IIO_PM_RUNTIME
+
+#include <linux/pm_runtime.h>
+
+static inline int iio_pm_runtime_setup(struct device *dev, int delay,
+				       bool ignore_children)
+{
+	int ret;
+
+	ret = pm_runtime_set_active(dev);
+	if (ret)
+		return ret;
+
+	pm_suspend_ignore_children(dev, ignore_children);
+	pm_runtime_enable(dev);
+	pm_runtime_set_autosuspend_delay(dev, delay);
+	pm_runtime_use_autosuspend(dev);
+
+	return 0;
+}
+
+static inline void iio_pm_runtime_cleanup(struct device *dev)
+{
+	pm_runtime_disable(dev);
+	pm_runtime_set_suspended(dev);
+	pm_runtime_put_noidle(dev);
+}
+
+static inline int iio_pm_runtime_set_power(struct device *dev, bool on)
+{
+#ifdef CONFIG_PM
+	int ret;
+
+	if (on)
+		ret = pm_runtime_get_sync(dev);
+	else {
+		pm_runtime_mark_last_busy(dev);
+		ret = pm_runtime_put_autosuspend(dev);
+	}
+
+	if (ret < 0) {
+		dev_err(dev, "Failed: iio_set_power_state for %d\n", on);
+		if (on)
+			pm_runtime_put_noidle(dev);
+
+		return ret;
+	}
+#endif
+	return 0;
+}
+
+#endif /* __IIO_PM_RUNTIME */
-- 
1.9.1

Re: [RFC PATCH 1/2] iio: pm_runtime: Introduce PM runtime helper functions

[Peter Meerwald]

On Mon, 18 May 2015, Daniel Baluta wrote:

> We need this in order to avoid reimplementing the same functions each time
> we add PM runtime support in a driver.

comments below
 
> Simple grep shows the following users:
> 	* accel/mma9551.c
> 	* accel/mmc9553.c
> 	* accel/kxcjk1013.c
> 	* accel/bmc150-accel.c
> 	* gyro/bmg160.c
> 	* imu/kmx61.c
> 	* common/hid-sensors.
> 
> Signed-off-by: Daniel Baluta <daniel.baluta@intel.com>
> ---
>  include/linux/iio/pm_runtime.h | 63 ++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 63 insertions(+)
>  create mode 100644 include/linux/iio/pm_runtime.h
> 
> diff --git a/include/linux/iio/pm_runtime.h b/include/linux/iio/pm_runtime.h
> new file mode 100644
> index 0000000..dc2bca7
> --- /dev/null
> +++ b/include/linux/iio/pm_runtime.h
> @@ -0,0 +1,63 @@
> +/*
> + * Industrial I/O runtime PM helper functions
> + *
> + * Copyright (c) 2015, Intel Corporation.
> + *
> + * This file is subject to the terms and conditions of version 2 of
> + * the GNU General Public License.  See the file COPYING in the main
> + * directory of this archive for more details.
> + *
> + */
> +#ifndef __IIO_PM_RUNTIME
> +#define __IIO_PM_RUNTIME
> +
> +#include <linux/pm_runtime.h>
> +
> +static inline int iio_pm_runtime_setup(struct device *dev, int delay,
> +				       bool ignore_children)
> +{
> +	int ret;
> +
> +	ret = pm_runtime_set_active(dev);
> +	if (ret)

just noting: should this be (ret) or (ret < 0)?

pm_runtime_get_sync() below may return negative, 0, and positive
pm_runtime_set_active() seems to return negative or 0

documentation doesn't tell... wondering if (ret < 0) would be safer here?

> +		return ret;
> +
> +	pm_suspend_ignore_children(dev, ignore_children);
> +	pm_runtime_enable(dev);
> +	pm_runtime_set_autosuspend_delay(dev, delay);
> +	pm_runtime_use_autosuspend(dev);
> +
> +	return 0;
> +}
> +
> +static inline void iio_pm_runtime_cleanup(struct device *dev)
> +{
> +	pm_runtime_disable(dev);
> +	pm_runtime_set_suspended(dev);
> +	pm_runtime_put_noidle(dev);
> +}
> +
> +static inline int iio_pm_runtime_set_power(struct device *dev, bool on)

why a static inline function in a header file?
these functions do not seem to be performance critical and are substantial 
enough in size to avoid copying the code to every driver

> +{
> +#ifdef CONFIG_PM
> +	int ret;
> +
> +	if (on)
> +		ret = pm_runtime_get_sync(dev);
> +	else {
> +		pm_runtime_mark_last_busy(dev);
> +		ret = pm_runtime_put_autosuspend(dev);
> +	}
> +
> +	if (ret < 0) {
> +		dev_err(dev, "Failed: iio_set_power_state for %d\n", on);

the error message doesn't match the function name, the text, 'for %d', is 
not very clear

> +		if (on)
> +			pm_runtime_put_noidle(dev);
> +
> +		return ret;
> +	}
> +#endif
> +	return 0;
> +}
> +
> +#endif /* __IIO_PM_RUNTIME */
> 

-- 

Peter Meerwald
+43-664-2444418 (mobile)

Re: [RFC PATCH 1/2] iio: pm_runtime: Introduce PM runtime helper functions

[Daniel Baluta]

Adding linux-pm people since Lars suggested that the changes should be
in the PM core.

I agree, that these changes are not "awfully" IIO specific but no
other subsystem uses
this pattern for configuration. In fact iio_pm_runtime_setup can be
done in many more
other ways depending on functionality needed.

On Mon, May 18, 2015 at 8:34 PM, Peter Meerwald <pmeerw@pmeerw.net> wrote:
> On Mon, 18 May 2015, Daniel Baluta wrote:
>
>> We need this in order to avoid reimplementing the same functions each time
>> we add PM runtime support in a driver.
>
> comments below
>
>> Simple grep shows the following users:
>>       * accel/mma9551.c
>>       * accel/mmc9553.c
>>       * accel/kxcjk1013.c
>>       * accel/bmc150-accel.c
>>       * gyro/bmg160.c
>>       * imu/kmx61.c
>>       * common/hid-sensors.
>>
>> Signed-off-by: Daniel Baluta <daniel.baluta@intel.com>
>> ---
>>  include/linux/iio/pm_runtime.h | 63 ++++++++++++++++++++++++++++++++++++++++++
>>  1 file changed, 63 insertions(+)
>>  create mode 100644 include/linux/iio/pm_runtime.h
>>
>> diff --git a/include/linux/iio/pm_runtime.h b/include/linux/iio/pm_runtime.h
>> new file mode 100644
>> index 0000000..dc2bca7
>> --- /dev/null
>> +++ b/include/linux/iio/pm_runtime.h
>> @@ -0,0 +1,63 @@
>> +/*
>> + * Industrial I/O runtime PM helper functions
>> + *
>> + * Copyright (c) 2015, Intel Corporation.
>> + *
>> + * This file is subject to the terms and conditions of version 2 of
>> + * the GNU General Public License.  See the file COPYING in the main
>> + * directory of this archive for more details.
>> + *
>> + */
>> +#ifndef __IIO_PM_RUNTIME
>> +#define __IIO_PM_RUNTIME
>> +
>> +#include <linux/pm_runtime.h>
>> +
>> +static inline int iio_pm_runtime_setup(struct device *dev, int delay,
>> +                                    bool ignore_children)
>> +{
>> +     int ret;
>> +
>> +     ret = pm_runtime_set_active(dev);
>> +     if (ret)
>
> just noting: should this be (ret) or (ret < 0)?
>
> pm_runtime_get_sync() below may return negative, 0, and positive
> pm_runtime_set_active() seems to return negative or 0
>
> documentation doesn't tell... wondering if (ret < 0) would be safer here?

I think this check is correct, I also found the same check in
power/pm2301_charger.c:1116. But there is no problem to add if (ret < 0)
it looks safer.

>
>> +             return ret;
>> +
>> +     pm_suspend_ignore_children(dev, ignore_children);
>> +     pm_runtime_enable(dev);
>> +     pm_runtime_set_autosuspend_delay(dev, delay);
>> +     pm_runtime_use_autosuspend(dev);
>> +
>> +     return 0;
>> +}
>> +
>> +static inline void iio_pm_runtime_cleanup(struct device *dev)
>> +{
>> +     pm_runtime_disable(dev);
>> +     pm_runtime_set_suspended(dev);
>> +     pm_runtime_put_noidle(dev);
>> +}
>> +
>> +static inline int iio_pm_runtime_set_power(struct device *dev, bool on)
>
> why a static inline function in a header file?
> these functions do not seem to be performance critical and are substantial
> enough in size to avoid copying the code to every driver

I know about this, but I was trying to avoid another .config option
that each IIO driver should select when implementing runtime pm support.

I can change this.

>
>> +{
>> +#ifdef CONFIG_PM
>> +     int ret;
>> +
>> +     if (on)
>> +             ret = pm_runtime_get_sync(dev);
>> +     else {
>> +             pm_runtime_mark_last_busy(dev);
>> +             ret = pm_runtime_put_autosuspend(dev);
>> +     }
>> +
>> +     if (ret < 0) {
>> +             dev_err(dev, "Failed: iio_set_power_state for %d\n", on);
>
> the error message doesn't match the function name, the text, 'for %d', is
> not very clear

Correct. Thanks for spotting this!
>
>> +             if (on)
>> +                     pm_runtime_put_noidle(dev);
>> +
>> +             return ret;
>> +     }
>> +#endif
>> +     return 0;
>> +}
>> +
>> +#endif /* __IIO_PM_RUNTIME */
>>
>
> --
>
> Peter Meerwald
> +43-664-2444418 (mobile)
> --
> To unsubscribe from this list: send the line "unsubscribe linux-iio" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

[RFC PATCH 2/2] iio: accel: kxcjk1013: Use the new IIO pm runtime helpers

[Daniel Baluta]

Signed-off-by: Daniel Baluta <daniel.baluta@intel.com>
---
 drivers/iio/accel/kxcjk-1013.c | 56 +++++++++++-------------------------------
 1 file changed, 14 insertions(+), 42 deletions(-)

diff --git a/drivers/iio/accel/kxcjk-1013.c b/drivers/iio/accel/kxcjk-1013.c
index a98b5d2..308499a 100644
--- a/drivers/iio/accel/kxcjk-1013.c
+++ b/drivers/iio/accel/kxcjk-1013.c
@@ -22,12 +22,12 @@
 #include <linux/acpi.h>
 #include <linux/gpio/consumer.h>
 #include <linux/pm.h>
-#include <linux/pm_runtime.h>
 #include <linux/iio/iio.h>
 #include <linux/iio/sysfs.h>
 #include <linux/iio/buffer.h>
 #include <linux/iio/trigger.h>
 #include <linux/iio/events.h>
+#include <linux/iio/pm_runtime.h>
 #include <linux/iio/trigger_consumer.h>
 #include <linux/iio/triggered_buffer.h>
 #include <linux/iio/accel/kxcjk_1013.h>
@@ -376,29 +376,6 @@ static int kxcjk1013_get_startup_times(struct kxcjk1013_data *data)
 }
 #endif
 
-static int kxcjk1013_set_power_state(struct kxcjk1013_data *data, bool on)
-{
-#ifdef CONFIG_PM
-	int ret;
-
-	if (on)
-		ret = pm_runtime_get_sync(&data->client->dev);
-	else {
-		pm_runtime_mark_last_busy(&data->client->dev);
-		ret = pm_runtime_put_autosuspend(&data->client->dev);
-	}
-	if (ret < 0) {
-		dev_err(&data->client->dev,
-			"Failed: kxcjk1013_set_power_state for %d\n", on);
-		if (on)
-			pm_runtime_put_noidle(&data->client->dev);
-		return ret;
-	}
-#endif
-
-	return 0;
-}
-
 static int kxcjk1013_chip_update_thresholds(struct kxcjk1013_data *data)
 {
 	int ret;
@@ -700,19 +677,22 @@ static int kxcjk1013_read_raw(struct iio_dev *indio_dev,
 		if (iio_buffer_enabled(indio_dev))
 			ret = -EBUSY;
 		else {
-			ret = kxcjk1013_set_power_state(data, true);
+			ret = iio_pm_runtime_set_power(&data->client->dev,
+						       true);
 			if (ret < 0) {
 				mutex_unlock(&data->mutex);
 				return ret;
 			}
 			ret = kxcjk1013_get_acc_reg(data, chan->scan_index);
 			if (ret < 0) {
-				kxcjk1013_set_power_state(data, false);
+				iio_pm_runtime_set_power(&data->client->dev,
+							 false);
 				mutex_unlock(&data->mutex);
 				return ret;
 			}
 			*val = sign_extend32(ret >> 4, 11);
-			ret = kxcjk1013_set_power_state(data, false);
+			ret = iio_pm_runtime_set_power(&data->client->dev,
+						       false);
 		}
 		mutex_unlock(&data->mutex);
 
@@ -855,7 +835,7 @@ static int kxcjk1013_write_event_config(struct iio_dev *indio_dev,
 	 * enable operation. When runtime pm is disabled the mode
 	 * is always on so sequence doesn't matter
 	 */
-	ret = kxcjk1013_set_power_state(data, state);
+	ret = iio_pm_runtime_set_power(&data->client->dev, state);
 	if (ret < 0) {
 		mutex_unlock(&data->mutex);
 		return ret;
@@ -863,7 +843,7 @@ static int kxcjk1013_write_event_config(struct iio_dev *indio_dev,
 
 	ret =  kxcjk1013_setup_any_motion_interrupt(data, state);
 	if (ret < 0) {
-		kxcjk1013_set_power_state(data, false);
+		iio_pm_runtime_set_power(&data->client->dev, false);
 		data->ev_enable_state = 0;
 		mutex_unlock(&data->mutex);
 		return ret;
@@ -1005,7 +985,7 @@ static int kxcjk1013_data_rdy_trigger_set_state(struct iio_trigger *trig,
 		return 0;
 	}
 
-	ret = kxcjk1013_set_power_state(data, state);
+	ret = iio_pm_runtime_set_power(&data->client->dev, state);
 	if (ret < 0) {
 		mutex_unlock(&data->mutex);
 		return ret;
@@ -1015,7 +995,7 @@ static int kxcjk1013_data_rdy_trigger_set_state(struct iio_trigger *trig,
 	else
 		ret = kxcjk1013_setup_new_data_interrupt(data, state);
 	if (ret < 0) {
-		kxcjk1013_set_power_state(data, false);
+		iio_pm_runtime_set_power(&data->client->dev, false);
 		mutex_unlock(&data->mutex);
 		return ret;
 	}
@@ -1293,16 +1273,11 @@ static int kxcjk1013_probe(struct i2c_client *client,
 		dev_err(&client->dev, "unable to register iio device\n");
 		goto err_buffer_cleanup;
 	}
-
-	ret = pm_runtime_set_active(&client->dev);
+	ret = iio_pm_runtime_setup(&client->dev, KXCJK1013_SLEEP_DELAY_MS,
+				   false);
 	if (ret)
 		goto err_iio_unregister;
 
-	pm_runtime_enable(&client->dev);
-	pm_runtime_set_autosuspend_delay(&client->dev,
-					 KXCJK1013_SLEEP_DELAY_MS);
-	pm_runtime_use_autosuspend(&client->dev);
-
 	return 0;
 
 err_iio_unregister:
@@ -1326,10 +1301,7 @@ static int kxcjk1013_remove(struct i2c_client *client)
 	struct iio_dev *indio_dev = i2c_get_clientdata(client);
 	struct kxcjk1013_data *data = iio_priv(indio_dev);
 
-	pm_runtime_disable(&client->dev);
-	pm_runtime_set_suspended(&client->dev);
-	pm_runtime_put_noidle(&client->dev);
-
+	iio_pm_runtime_cleanup(&client->dev);
 	iio_device_unregister(indio_dev);
 
 	if (data->dready_trig) {
-- 
1.9.1

Re: [RFC PATCH 0/2] Introduce PM runtime helper functions

[Lars-Peter Clausen]

On 05/18/2015 06:25 PM, Daniel Baluta wrote:
> Working on a new driver I noticed that there is a fair amount of duplicate
> code for adding PM runtime support.
>
> First patch refactors the PM runtime support in a new header
>   <linux/iio/pm_runtime> introducing functions for setup/cleanup and
> set power state.

Makes sense, but none of the functions look awfully IIO specific. I'd say 
these are helper functions that belong into the PM core itself.

- Lars