Re: [linux-pm] [GIT PULL] One more power management fix for 2.6.37

Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]
From: Rafael J. Wysocki
Date: Wednesday, November 10, 2010 - 1:51 pm

On Wednesday, November 10, 2010, Alan Stern wrote:

No, I think you're right.  I rethought that particular thing in the meantime
and came to the conclusion that it'd be better to keep it where it was.


I guess it would allow us to do that, but that's going to be a major change.

What about applying the patch below now and moving to the multiple list heads
approach in the next cycle?

Rafael

---
From: Rafael J. Wysocki <rjw@sisk.pl>
Subject: PM: Allow devices to be removed during late suspend and early resume

Holding dpm_list_mtx across late suspend and early resume of devices
is problematic for the PCMCIA subsystem and doesn't allow device
objects to be removed by late suspend and early resume driver
callbacks.  This appears to be overly restrictive, as drivers are
generally allowed to remove device objects in other phases of suspend
and resume.  Therefore rework dpm_{suspend|resume}_noirq() so that
they don't have to hold dpm_list_mtx all the time.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
---
 drivers/base/power/main.c |   34 ++++++++++++++++++++++++++++++----
 1 file changed, 30 insertions(+), 4 deletions(-)

Index: linux-2.6/drivers/base/power/main.c
===================================================================
--- linux-2.6.orig/drivers/base/power/main.c
+++ linux-2.6/drivers/base/power/main.c
@@ -475,20 +475,33 @@ End:
  */
 void dpm_resume_noirq(pm_message_t state)
 {
-	struct device *dev;
+	struct list_head list;
 	ktime_t starttime = ktime_get();
 
+	INIT_LIST_HEAD(&list);
 	mutex_lock(&dpm_list_mtx);
 	transition_started = false;
-	list_for_each_entry(dev, &dpm_list, power.entry)
+	while (!list_empty(&dpm_list)) {
+		struct device *dev = to_device(dpm_list.next);
+
+		get_device(dev);
 		if (dev->power.status > DPM_OFF) {
 			int error;
 
 			dev->power.status = DPM_OFF;
+			mutex_unlock(&dpm_list_mtx);
+
 			error = device_resume_noirq(dev, state);
+
+			mutex_lock(&dpm_list_mtx);
 			if (error)
 				pm_dev_err(dev, state, " early", error);
 		}
+		if (!list_empty(&dev->power.entry))
+			list_move_tail(&dev->power.entry, &list);
+		put_device(dev);
+	}
+	list_splice(&list, &dpm_list);
 	mutex_unlock(&dpm_list_mtx);
 	dpm_show_time(starttime, state, "early");
 	resume_device_irqs();
@@ -789,20 +802,33 @@ End:
  */
 int dpm_suspend_noirq(pm_message_t state)
 {
-	struct device *dev;
+	struct list_head list;
 	ktime_t starttime = ktime_get();
 	int error = 0;
 
+	INIT_LIST_HEAD(&list);
 	suspend_device_irqs();
 	mutex_lock(&dpm_list_mtx);
-	list_for_each_entry_reverse(dev, &dpm_list, power.entry) {
+	while (!list_empty(&dpm_list)) {
+		struct device *dev = to_device(dpm_list.prev);
+
+		get_device(dev);
+		mutex_unlock(&dpm_list_mtx);
+
 		error = device_suspend_noirq(dev, state);
+
+		mutex_lock(&dpm_list_mtx);
 		if (error) {
 			pm_dev_err(dev, state, " late", error);
+			put_device(dev);
 			break;
 		}
 		dev->power.status = DPM_OFF_IRQ;
+		if (!list_empty(&dev->power.entry))
+			list_move(&dev->power.entry, &list);
+		put_device(dev);
 	}
+	list_splice_tail(&list, &dpm_list);
 	mutex_unlock(&dpm_list_mtx);
 	if (error)
 		dpm_resume_noirq(resume_event(state));
--
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]

Messages in current thread:
Re: [linux-pm] [GIT PULL] One more power management fix fo ..., Rafael J. Wysocki, (Wed Nov 10, 1:51 pm)