[PATCH] Leave runtime suspended devices off at system resume

[PATCH] Leave runtime suspended devices off at system resume

[Todd Poynor]

Currently all devices are resumed at system resume time, including any
that were individually powered off ("at runtime") prior to the system
suspend.  In certain cases it can be nice to force back on individually
suspended devices, such as the display, but hopefully this policy can be
left up to userspace power managers; the kernel should probably honor
the settings previously made by userspace/drivers.  This seems
preferable to requiring a power-conscious system to re-suspend devices
after a system resume; furthermore, for certain platforms (such as
XScale PXA27X) there can be disastrous consequences of powering up
devices when the system is in a state incompatible with operation of the
device.

Suggested patch does this:

(1) At system resume, checks power_state to see if the device was
suspended prior to system suspend, and skips powering on the device if
so.

(2) Does not re-suspend an already-suspended device at system suspend
(using a different method than is currently employed, which reorders the
list, see #3).

(3) Preserves the active/off device list order despite the above changes
to suspend/resume behavior, to avoid dependency problems that tend to
occur when the list is reordered.

--- linux-2.6.6-orig/drivers/base/power/suspend.c	2004-05-10 11:22:58.000000000 -0700
+++ linux-2.6.6-prevstate/drivers/base/power/suspend.c	2004-05-25 19:00:20.803379624 -0700
@@ -39,7 +39,7 @@
 {
 	int error = 0;
 
-	if (dev->bus && dev->bus->suspend)
+	if (dev->bus && dev->bus->suspend && ! dev->power.power_state)
 		error = dev->bus->suspend(dev,state);
 
 	if (!error) {
--- linux-2.6.6-orig/drivers/base/power/resume.c	2004-05-10 11:22:58.000000000 -0700
+++ linux-2.6.6-prevstate/drivers/base/power/resume.c	2004-05-25 18:07:30.978266288 -0700
@@ -35,7 +35,10 @@
 		struct list_head * entry = dpm_off.next;
 		struct device * dev = to_device(entry);
 		list_del_init(entry);
-		resume_device(dev);
+
+		if (! dev->power.power_state)
+			resume_device(dev);
+
 		list_add_tail(entry,&dpm_active);
 	}
 }



-- 
Todd

Re: [PATCH] Leave runtime suspended devices off at system resume

[Greg KH]

On Wed, May 26, 2004 at 02:43:19PM -0700, Todd Poynor wrote:
> Currently all devices are resumed at system resume time, including any
> that were individually powered off ("at runtime") prior to the system
> suspend.  In certain cases it can be nice to force back on individually
> suspended devices, such as the display, but hopefully this policy can be
> left up to userspace power managers; the kernel should probably honor
> the settings previously made by userspace/drivers.  This seems
> preferable to requiring a power-conscious system to re-suspend devices
> after a system resume; furthermore, for certain platforms (such as
> XScale PXA27X) there can be disastrous consequences of powering up
> devices when the system is in a state incompatible with operation of the
> device.
> 
> Suggested patch does this:
> 
> (1) At system resume, checks power_state to see if the device was
> suspended prior to system suspend, and skips powering on the device if
> so.
> 
> (2) Does not re-suspend an already-suspended device at system suspend
> (using a different method than is currently employed, which reorders the
> list, see #3).
> 
> (3) Preserves the active/off device list order despite the above changes
> to suspend/resume behavior, to avoid dependency problems that tend to
> occur when the list is reordered.

Nice, that looks good.

Applied, thanks.

greg k-h

Re: [PATCH] Leave runtime suspended devices off at system resume

[Todd Poynor]

On Fri, May 28, 2004 at 10:03:15AM -0700, Greg KH wrote:
> Nice, that looks good.
> 
> Applied, thanks.

Sorry, Felipe Solana found that USB drivers tweak the power.power_state
field during suspend, so we can't rely on that field to tell what to
resume.  A new patch against 2.6.6 creates a separate field for the
original value, so driver mods to the field won't break resume.  I'll
also send a patch against the original patch to fix trees already
updated in a moment.  Thanks -- Todd

--- linux-2.6.6-orig/drivers/base/power/suspend.c	2004-05-10 11:22:58.000000000 -0700
+++ linux-2.6.6-prevstate/drivers/base/power/suspend.c	2004-05-27 13:58:01.931014888 -0700
@@ -39,7 +39,9 @@
 {
 	int error = 0;
 
-	if (dev->bus && dev->bus->suspend)
+	dev->power.prev_state = dev->power.power_state;
+
+	if (dev->bus && dev->bus->suspend && ! dev->power.power_state)
 		error = dev->bus->suspend(dev,state);
 
 	if (!error) {
--- linux-2.6.6-orig/drivers/base/power/resume.c	2004-05-10 11:22:58.000000000 -0700
+++ linux-2.6.6-prevstate/drivers/base/power/resume.c	2004-05-27 14:35:03.373304328 -0700
@@ -35,7 +35,10 @@
 		struct list_head * entry = dpm_off.next;
 		struct device * dev = to_device(entry);
 		list_del_init(entry);
-		resume_device(dev);
+
+		if (! dev->power.prev_state)
+			resume_device(dev);
+
 		list_add_tail(entry,&dpm_active);
 	}
 }
--- linux-2.6.6-orig/include/linux/pm.h	2004-05-10 11:23:56.000000000 -0700
+++ linux-2.6.6-prevstate/include/linux/pm.h	2004-05-27 14:35:37.143170528 -0700
@@ -231,6 +231,7 @@
 struct dev_pm_info {
 #ifdef	CONFIG_PM
 	u32			power_state;
+	u32			prev_state;
 	u8			* saved_state;
 	atomic_t		pm_users;
 	struct device		* pm_parent;