iio: fix bug with triggers not resuming after sleep
Subsystems (2)
v3 cover iio: fix bug with triggers not resuming after sleep
When a device enters an idle state (for example ASUS RC71L with s2idle) and an iio driver has a trigger attached such as iio-trig-hrtimer, after resuming the device the trigger is not triggering data acquisition.
This patch series solves the problem reliably and is well tested after many cycles and many reboots.
Changelog:
- V2: patch 2:
+ Simplify code
+ Remove unneeded code protections around SET_RUNTIME_PM_OPS()
+ pm_ptr() to let the compiler drop bmi323_core_pm_ops if !CONFIG_PM
- V3: patch 1:
+ Add pf->irq=0 that was accidentally left out
Previous patches obsoleted:
https://lore.kernel.org/all/[email protected]/
https://lore.kernel.org/all/[email protected]/
Denis Benato (2):
iio: trigger: allow devices to suspend/resume theirs associated
trigger
iio: bmi323: suspend and resume triggering on relevant pm operations
drivers/iio/imu/bmi323/bmi323.h | 1 + drivers/iio/imu/bmi323/bmi323_core.c | 23 +++++++++++++++++++++++ drivers/iio/imu/bmi323/bmi323_i2c.c | 1 + drivers/iio/imu/bmi323/bmi323_spi.c | 1 + drivers/iio/industrialio-trigger.c | 27 +++++++++++++++++++++++++++ include/linux/iio/iio.h | 17 +++++++++++++++++ 6 files changed, 70 insertions(+)
Signature · 1 line
2.46.0
PATCH v3 1/2 iio: trigger: allow devices to suspend/resume theirs associated trigger
When a machine enters a sleep state while a trigger is associated to an iio device that trigger is not resumed after exiting the sleep state:
provide iio device drivers a way to suspend and resume
the associated trigger to solve the aforementioned bug.
Each iio driver supporting external triggers is expected to call iio_device_suspend_triggering before suspending,
and iio_device_resume_triggering upon resuming.
--- drivers/iio/industrialio-trigger.c | 27 +++++++++++++++++++++++++++ include/linux/iio/iio.h | 17 +++++++++++++++++ 2 files changed, 44 insertions(+)
drivers/iio/industrialio-trigger.c +27 -0
@@ -347,6 +347,7 @@ int iio_trigger_detach_poll_func(struct iio_trigger *trig,iio_trigger_put_irq (trig, pf->irq);free_irq (pf->irq, pf);module_put (iio_dev_opaque->driver_module); + pf->irq =0 ;return ret; }
@@ -770,3 +771,29 @@ void iio_device_unregister_trigger_consumer(struct iio_dev *indio_dev)if (indio_dev->trig)iio_trigger_put (indio_dev->trig); } + +int iio_device_suspend_triggering (struct iio_dev *indio_dev) +{ +struct iio_dev_opaque *iio_dev_opaque =to_iio_dev_opaque (indio_dev); + +guard (mutex)(&iio_dev_opaque->mlock); + +if ((indio_dev->pollfunc) && (indio_dev->pollfunc->irq >0 )) +disable_irq (indio_dev->pollfunc->irq); + +return 0 ; +} +EXPORT_SYMBOL (iio_device_suspend_triggering); + +int iio_device_resume_triggering (struct iio_dev *indio_dev) +{ +struct iio_dev_opaque *iio_dev_opaque =to_iio_dev_opaque (indio_dev); + +guard (mutex)(&iio_dev_opaque->mlock); + +if ((indio_dev->pollfunc) && (indio_dev->pollfunc->irq >0 )) +enable_irq (indio_dev->pollfunc->irq); + +return 0 ; +} +EXPORT_SYMBOL (iio_device_resume_triggering);
include/linux/iio/iio.h +17 -0
@@ -810,6 +810,23 @@ static inline struct dentry *iio_get_debugfs_dentry(struct iio_dev *indio_dev) }#endif +/** +* iio_device_suspend_triggering() - suspend trigger attached to an iio_dev +* @indio_dev: iio_dev associated with the device that will have triggers suspended +* +* Return 0 if successful, negative otherwise +**/ +int iio_device_suspend_triggering (struct iio_dev *indio_dev); + +/** +* iio_device_resume_triggering() - resume trigger attached to an iio_dev +* that was previously suspended with iio_device_suspend_triggering() +* @indio_dev: iio_dev associated with the device that will have triggers resumed +* +* Return 0 if successful, negative otherwise +**/ +int iio_device_resume_triggering (struct iio_dev *indio_dev); +#ifdef CONFIG_ACPI bool iio_read_acpi_mount_matrix(struct device *dev, struct iio_mount_matrix *orientation,
Signature · 1 line
2.46.0
PATCH v3 2/2 iio: bmi323: suspend and resume triggering on relevant pm operations
In reply to Denis BenatoPrevent triggers from stop working after the device has entered sleep:
use iio_device_suspend_triggering and iio_device_resume_triggering helpers.
--- drivers/iio/imu/bmi323/bmi323.h | 1 + drivers/iio/imu/bmi323/bmi323_core.c | 23 +++++++++++++++++++++++ drivers/iio/imu/bmi323/bmi323_i2c.c | 1 + drivers/iio/imu/bmi323/bmi323_spi.c | 1 + 4 files changed, 26 insertions(+)
drivers/iio/imu/bmi323/bmi323.h +1 -0
@@ -205,5 +205,6 @@struct device ;int bmi323_core_probe (struct device *dev);extern const struct regmap_config bmi323_regmap_config; +extern const struct dev_pm_ops bmi323_core_pm_ops;#endif
drivers/iio/imu/bmi323/bmi323_core.c +23 -0
@@ -2121,6 +2121,29 @@ int bmi323_core_probe(struct device *dev) }EXPORT_SYMBOL_NS_GPL (bmi323_core_probe,IIO_BMI323 ); +#if defined(CONFIG_PM ) +static int bmi323_core_runtime_suspend (struct device *dev) +{ +struct iio_dev *indio_dev =dev_get_drvdata (dev); + +return iio_device_suspend_triggering (indio_dev); +} + +static int bmi323_core_runtime_resume (struct device *dev) +{ +struct iio_dev *indio_dev =dev_get_drvdata (dev); + +return iio_device_resume_triggering (indio_dev); +} + +#endif + +const struct dev_pm_ops bmi323_core_pm_ops = { +SET_RUNTIME_PM_OPS (bmi323_core_runtime_suspend, + bmi323_core_runtime_resume,NULL ) +}; +EXPORT_SYMBOL_NS_GPL (bmi323_core_pm_ops,IIO_BMI323 ); +MODULE_DESCRIPTION ("Bosch BMI323 IMU driver" );MODULE_AUTHOR ("Jagath Jog J <[email protected]>" );MODULE_LICENSE ("GPL" );
drivers/iio/imu/bmi323/bmi323_i2c.c +1 -0
@@ -128,6 +128,7 @@ MODULE_DEVICE_TABLE(of, bmi323_of_i2c_match);static struct i2c_driver bmi323_i2c_driver = { .driver = { .name ="bmi323" , + .pm =pm_ptr (&bmi323_core_pm_ops), .of_match_table = bmi323_of_i2c_match, .acpi_match_table = bmi323_acpi_match, },
drivers/iio/imu/bmi323/bmi323_spi.c +1 -0
@@ -79,6 +79,7 @@ MODULE_DEVICE_TABLE(of, bmi323_of_spi_match);static struct spi_driver bmi323_spi_driver = { .driver = { .name ="bmi323" , + .pm =pm_ptr (&bmi323_core_pm_ops), .of_match_table = bmi323_of_spi_match, }, .probe = bmi323_spi_probe,
Signature · 1 line
2.46.0
Jonathan Cameron
maintainerQuoted 9 lines · On Wed, 7 Aug 2024 20:56:17 +0200 Denis Benato <[email protected]>…
When a device enters an idle state (for example ASUS RC71L with s2idle) and an iio driver has a trigger attached such as iio-trig-hrtimer, after resuming the device the trigger is not triggering data acquisition.
This patch series solves the problem reliably and is well tested after many cycles and many reboots.
Quoted 8 lines
Changelog:
- V2: patch 2:
+ Simplify code
+ Remove unneeded code protections aroundSET_RUNTIME_PM_OPS()
+pm_ptr()to let the compiler dropbmi323_core_pm_opsif !CONFIG_PM- V3: patch 1:
+ Add pf->irq=0 that was accidentally left out
Dropped v2 and picked up v3.
Thanks,
Jonathan
Quoted 18 lines
Previous patches obsoleted:
https://lore.kernel.org/all/[email protected]/
https://lore.kernel.org/all/[email protected]/Denis Benato (2):
iio: trigger: allow devices to suspend/resume theirs associated
trigger
iio: bmi323: suspend and resume triggering on relevant pm operationsdrivers/iio/imu/bmi323/bmi323.h | 1 + drivers/iio/imu/bmi323/bmi323_core.c | 23 +++++++++++++++++++++++ drivers/iio/imu/bmi323/bmi323_i2c.c | 1 + drivers/iio/imu/bmi323/bmi323_spi.c | 1 + drivers/iio/industrialio-trigger.c | 27 +++++++++++++++++++++++++++ include/linux/iio/iio.h | 17 +++++++++++++++++ 6 files changed, 70 insertions(+)