PM / Runtime: Idle devices asynchronously after probe|release
PATCH RESEND 1/1 PM / Runtime: Idle devices asynchronously after probe|release
From: Ulf Hansson <[email protected]>
Putting devices into idle|suspend in a synchronous manner means we are waiting for each device to become idle|suspended before the probe|release is fully done.
This patch switch to use the asynchronous runtime PM API:s instead and thus improves the parallelism since we can move on and handle the next device in queue in an earlier phase.
--- drivers/base/dd.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/base/dd.c +3 -3
@@ -380,7 +380,7 @@ int driver_probe_device(struct device_driver *drv, struct device *dev)pm_runtime_barrier (dev); ret =really_probe (dev, drv); -pm_runtime_idle (dev); +pm_request_idle (dev);return ret; }
@@ -428,7 +428,7 @@ int device_attach(struct device *dev) } }else { ret =bus_for_each_drv (dev->bus,NULL , dev, __device_attach); -pm_runtime_idle (dev); +pm_request_idle (dev); } out_unlock:device_unlock (dev);
@@ -499,7 +499,7 @@ static void __device_release_driver(struct device *dev)BUS_NOTIFY_UNBIND_DRIVER , dev); -pm_runtime_put_sync (dev); +pm_runtime_put (dev);if (dev->bus && dev->bus->remove) dev->bus->remove (dev);
Signature · 1 line
1.7.10
PATCH RESEND 1/1 PM / Runtime: Idle devices asynchronously after probe|release
Quoted 9 lines · Ulf Hansson
From: Ulf Hansson <[email protected]>
Putting devices into idle|suspend in a synchronous manner means we are waiting for each device to become idle|suspended before the probe|release is fully done.
This patch switch to use the asynchronous runtime PM API:s instead and thus improves the parallelism since we can move on and handle the next device in queue in an earlier phase.
Quoted 40 lines
Signed-off-by Ulf Hansson <[email protected]>Cc Kevin Hilman <[email protected]>Cc Alan Stern <[email protected]>Cc Greg Kroah-Hartman <[email protected]>--- drivers/base/dd.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)drivers/base/dd.c +3
-3@@ -380,7 +380,7 @@ int driver_probe_device(struct device_driver *drv, struct device *dev)pm_runtime_barrier (dev); ret =really_probe (dev, drv); -pm_runtime_idle (dev); +pm_request_idle (dev);return ret; }@@ -428,7 +428,7 @@ int device_attach(struct device *dev) } }else { ret =bus_for_each_drv (dev->bus,NULL , dev, __device_attach); -pm_runtime_idle (dev); +pm_request_idle (dev); } out_unlock:device_unlock (dev);@@ -499,7 +499,7 @@ static void __device_release_driver(struct device *dev)BUS_NOTIFY_UNBIND_DRIVER , dev); -pm_runtime_put_sync (dev); +pm_runtime_put (dev);if (dev->bus && dev->bus->remove) dev->bus->remove (dev);
Signature · 2 lines
I speak only for myself. Rafael J. Wysocki, Intel Open Source Technology Center.
PATCH RESEND 1/1 PM / Runtime: Idle devices asynchronously after probe|release
In reply to Ulf HanssonQuoted 6 lines · Ulf Hansson
From: Ulf Hansson <[email protected]>
Putting devices into idle|suspend in a synchronous manner means we are waiting for each device to become idle|suspended before the probe|release is fully done.
Quoted 8 lines
This patch switch to use the asynchronous runtime PM API:s instead and thus improves the parallelism since we can move on and handle the next device in queue in an earlier phase.
Signed-off-by Ulf Hansson <[email protected]>Cc Kevin Hilman <[email protected]>Cc Alan Stern <[email protected]>Cc Greg Kroah-Hartman <[email protected]>