[PATCH] PM: runtime: return at once for active device in rpm_resume

[PATCH] PM: runtime: return at once for active device in rpm_resume

[tom.leiming]

From: Ming Lei <tom.leiming@gmail.com>

This patch moves ahead check for active rpm state of device,
so that rpm_resume can return immediately for device with
active rpm state.

Considered rpm_resume is run in atomic context and is often in
hot path(such as in usbnet, called before transmitting every
network frame), this patch should make sense.

Signed-off-by: Ming Lei <tom.leiming@gmail.com>
---
 drivers/base/power/runtime.c |   10 +++++-----
 1 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/base/power/runtime.c b/drivers/base/power/runtime.c
index 0d4587b..dcc3008 100644
--- a/drivers/base/power/runtime.c
+++ b/drivers/base/power/runtime.c
@@ -455,6 +455,11 @@ static int rpm_resume(struct device *dev, int rpmflags)
 	dev_dbg(dev, "%s flags 0x%x\n", __func__, rpmflags);
 
  repeat:
+	if (dev->power.runtime_status == RPM_ACTIVE) {
+		retval = 1;
+		goto out;
+	}
+
 	if (dev->power.runtime_error)
 		retval = -EINVAL;
 	else if (dev->power.disable_depth > 0)
@@ -472,11 +477,6 @@ static int rpm_resume(struct device *dev, int rpmflags)
 	if (!dev->power.timer_autosuspends)
 		pm_runtime_deactivate_timer(dev);
 
-	if (dev->power.runtime_status == RPM_ACTIVE) {
-		retval = 1;
-		goto out;
-	}
-
 	if (dev->power.runtime_status == RPM_RESUMING
 	    || dev->power.runtime_status == RPM_SUSPENDING) {
 		DEFINE_WAIT(wait);
-- 
1.7.4.1

Re: [PATCH] PM: runtime: return at once for active device in rpm_resume

[Rafael J. Wysocki]

On Tuesday, June 14, 2011, tom.leiming@gmail.com wrote:
> From: Ming Lei <tom.leiming@gmail.com>
> 
> This patch moves ahead check for active rpm state of device,
> so that rpm_resume can return immediately for device with
> active rpm state.
> 
> Considered rpm_resume is run in atomic context and is often in
> hot path(such as in usbnet, called before transmitting every
> network frame), this patch should make sense.

It ignores the fact that rpm_resume() cancels pending runtime PM requests
in addition to resuming suspended devices.

Thanks,
Rafael


> Signed-off-by: Ming Lei <tom.leiming@gmail.com>
> ---
>  drivers/base/power/runtime.c |   10 +++++-----
>  1 files changed, 5 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/base/power/runtime.c b/drivers/base/power/runtime.c
> index 0d4587b..dcc3008 100644
> --- a/drivers/base/power/runtime.c
> +++ b/drivers/base/power/runtime.c
> @@ -455,6 +455,11 @@ static int rpm_resume(struct device *dev, int rpmflags)
>  	dev_dbg(dev, "%s flags 0x%x\n", __func__, rpmflags);
>  
>   repeat:
> +	if (dev->power.runtime_status == RPM_ACTIVE) {
> +		retval = 1;
> +		goto out;
> +	}
> +
>  	if (dev->power.runtime_error)
>  		retval = -EINVAL;
>  	else if (dev->power.disable_depth > 0)
> @@ -472,11 +477,6 @@ static int rpm_resume(struct device *dev, int rpmflags)
>  	if (!dev->power.timer_autosuspends)
>  		pm_runtime_deactivate_timer(dev);
>  
> -	if (dev->power.runtime_status == RPM_ACTIVE) {
> -		retval = 1;
> -		goto out;
> -	}
> -
>  	if (dev->power.runtime_status == RPM_RESUMING
>  	    || dev->power.runtime_status == RPM_SUSPENDING) {
>  		DEFINE_WAIT(wait);
>