Merge bk://kernel.bkbits.net//home/mochel/linux-2.5-power

into home.osdl.org:/home/torvalds/v2.5/linux

drivers/acpi/Kconfig

@@ -68,7 +68,7 @@ config ACPI_BOOT
 config ACPI_SLEEP
 	bool "Sleep States (EXPERIMENTAL)"
 	depends on X86 && ACPI
-	depends on EXPERIMENTAL
+	depends on EXPERIMENTAL && PM
 	default y
 	---help---
 	  This option adds support for ACPI suspend states. 

drivers/acpi/sleep/main.c

@@ -22,6 +22,7 @@ ACPI_MODULE_NAME		("sleep")
 u8 sleep_states[ACPI_S_STATE_COUNT];
 
 extern void do_suspend_lowlevel_s4bios(int);
+extern void do_suspend_lowlevel(int);
 
 /**
  * acpi_system_restore_state - OS-specific restoration of state
@@ -71,10 +72,6 @@ acpi_system_restore_state (
  * First, we call to the device driver layer to save device state.
  * Once we have that, we save whatevery processor and kernel state we
  * need to memory.
- * If we're entering S4, we then write the memory image to disk.
- *
- * Only then is it safe for us to power down devices, since we may need
- * the disks and upstream buses to write to.
  */
 acpi_status
 acpi_system_save_state(
@@ -185,12 +182,11 @@ acpi_system_suspend(
 		status = acpi_enter_sleep_state(state);
 		break;
 
-#ifdef CONFIG_SOFTWARE_SUSPEND
 	case ACPI_STATE_S2:
 	case ACPI_STATE_S3:
 		do_suspend_lowlevel(0);
 		break;
-#endif
+
 	case ACPI_STATE_S4:
 		do_suspend_lowlevel_s4bios(0);
 		break;

drivers/base/Makefile

 # Makefile for the Linux device tree
 
-obj-y			:= core.o sys.o interface.o power.o bus.o \
+obj-y			:= core.o sys.o interface.o bus.o \
 			   driver.o class.o platform.o \
 			   cpu.o firmware.o init.o map.o
+obj-y			+= power/
 obj-$(CONFIG_FW_LOADER)	+= firmware_class.o
 obj-$(CONFIG_NUMA)	+= node.o  memblk.o

drivers/base/base.h

@@ -13,3 +13,5 @@ struct class_device_attribute *to_class_dev_attr(struct attribute *_attr)
 {
 	return container_of(_attr,struct class_device_attribute,attr);
 }
+
+

drivers/base/bus.c

@@ -16,6 +16,7 @@
 #include <linux/init.h>
 #include <linux/string.h>
 #include "base.h"
+#include "power/power.h"
 
 #define to_dev(node) container_of(node,struct device,bus_list)
 #define to_drv(node) container_of(node,struct device_driver,kobj.entry)
@@ -364,6 +365,7 @@ void device_release_driver(struct device * dev)
 	if (drv) {
 		sysfs_remove_link(&drv->kobj,dev->kobj.name);
 		list_del_init(&dev->driver_list);
+		device_detach_shutdown(dev);
 		if (drv->remove)
 			drv->remove(dev);
 		dev->driver = NULL;

drivers/base/core.c

@@ -20,6 +20,7 @@
 #include <asm/semaphore.h>
 
 #include "base.h"
+#include "power/power.h"
 
 int (*platform_notify)(struct device * dev) = NULL;
 int (*platform_notify_remove)(struct device * dev) = NULL;
@@ -230,6 +231,8 @@ int device_add(struct device *dev)
 
 	bus_add_device(dev);
 
+	device_pm_add(dev);
+
 	/* notify platform of device entry */
 	if (platform_notify)
 		platform_notify(dev);
@@ -304,6 +307,8 @@ void device_del(struct device * dev)
 {
 	struct device * parent = dev->parent;
 
+	device_pm_remove(dev);
+
 	down_write(&devices_subsys.rwsem);
 	if (parent)
 		list_del_init(&dev->node);

drivers/base/interface.c

@@ -21,77 +21,39 @@ static ssize_t device_read_name(struct device * dev, char * buf)
 
 static DEVICE_ATTR(name,S_IRUGO,device_read_name,NULL);
 
-static ssize_t
-device_read_power(struct device * dev, char * page)
+/**
+ *	detach_state - control the default power state for the device.
+ *	
+ *	This is the state the device enters when it's driver module is 
+ *	unloaded. The value is an unsigned integer, in the range of 0-4.
+ *	'0' indicates 'On', so no action will be taken when the driver is
+ *	unloaded. This is the default behavior.
+ *	'4' indicates 'Off', meaning the driver core will call the driver's
+ *	shutdown method to quiesce the device.
+ *	1-3 indicate a low-power state for the device to enter via the 
+ *	driver's suspend method. 
+ */
+
+static ssize_t detach_show(struct device * dev, char * buf)
 {
-	return sprintf(page,"%d\n",dev->power_state);
+	return sprintf(buf,"%u\n",dev->detach_state);
 }
 
-static ssize_t
-device_write_power(struct device * dev, const char * buf, size_t count)
+static ssize_t detach_store(struct device * dev, const char * buf, size_t n)
 {
-	char	str_command[20];
-	char	str_level[20];
-	int	num_args;
-	u32	state;
-	u32	int_level;
-	int	error = 0;
-
-	if (!dev->driver)
-		goto done;
-
-	num_args = sscanf(buf,"%10s %10s %u",str_command,str_level,&state);
-
-	error = -EINVAL;
-
-	if (!num_args)
-		goto done;
-
-	if (!strnicmp(str_command,"suspend",7)) {
-		if (num_args != 3)
-			goto done;
-		if (!strnicmp(str_level,"notify",6))
-			int_level = SUSPEND_NOTIFY;
-		else if (!strnicmp(str_level,"save",4))
-			int_level = SUSPEND_SAVE_STATE;
-		else if (!strnicmp(str_level,"disable",7))
-			int_level = SUSPEND_DISABLE;
-		else if (!strnicmp(str_level,"powerdown",8))
-			int_level = SUSPEND_POWER_DOWN;
-		else
-			goto done;
-
-		if (dev->driver->suspend)
-			error = dev->driver->suspend(dev,state,int_level);
-		else
-			error = 0;
-	} else if (!strnicmp(str_command,"resume",6)) {
-		if (num_args != 2)
-			goto done;
-
-		if (!strnicmp(str_level,"poweron",7))
-			int_level = RESUME_POWER_ON;
-		else if (!strnicmp(str_level,"restore",7))
-			int_level = RESUME_RESTORE_STATE;
-		else if (!strnicmp(str_level,"enable",6))
-			int_level = RESUME_ENABLE;
-		else
-			goto done;
-
-		if (dev->driver->resume)
-			error = dev->driver->resume(dev,int_level);
-		else
-			error = 0;
-	}
- done:
-	return error < 0 ? error : count;
+	u32 state;
+	state = simple_strtoul(buf,NULL,10);
+	if (state > 4)
+		return -EINVAL;
+	dev->detach_state = state;
+	return n;
 }
 
-static DEVICE_ATTR(power,S_IWUSR | S_IRUGO,
-		   device_read_power,device_write_power);
+static DEVICE_ATTR(detach_state,0644,detach_show,detach_store);
+
 
 struct attribute * dev_default_attrs[] = {
 	&dev_attr_name.attr,
-	&dev_attr_power.attr,
+	&dev_attr_detach_state.attr,
 	NULL,
 };

drivers/base/power/Makefile

+obj-y			:= shutdown.o
+obj-$(CONFIG_PM)	+= main.o suspend.o resume.o runtime.o sysfs.o

drivers/base/power/main.c

+/*
+ * drivers/base/power/main.c - Where the driver meets power management.
+ *
+ * Copyright (c) 2003 Patrick Mochel
+ * Copyright (c) 2003 Open Source Development Lab
+ *
+ * This file is released under the GPLv2
+ *
+ *
+ * The driver model core calls device_pm_add() when a device is registered.
+ * This will intialize the embedded device_pm_info object in the device
+ * and add it to the list of power-controlled devices. sysfs entries for
+ * controlling device power management will also be added.
+ *
+ * A different set of lists than the global subsystem list are used to 
+ * keep track of power info because we use different lists to hold 
+ * devices based on what stage of the power management process they 
+ * are in. The power domain dependencies may also differ from the 
+ * ancestral dependencies that the subsystem list maintains.
+ */
+
+#define DEBUG
+
+#include <linux/device.h>
+#include "power.h"
+
+LIST_HEAD(dpm_active);
+LIST_HEAD(dpm_suspended);
+LIST_HEAD(dpm_off);
+LIST_HEAD(dpm_off_irq);
+
+DECLARE_MUTEX(dpm_sem);
+
+/*
+ * PM Reference Counting.
+ */
+
+static inline void device_pm_hold(struct device * dev)
+{
+	atomic_inc(&dev->power.pm_users);
+}
+
+static inline void device_pm_release(struct device * dev)
+{
+	atomic_inc(&dev->power.pm_users);
+}
+
+
+/**
+ *	device_pm_set_parent - Specify power dependency.
+ *	@dev:		Device who needs power.
+ *	@parent:	Device that supplies power.
+ *
+ *	This function is used to manually describe a power-dependency
+ *	relationship. It may be used to specify a transversal relationship
+ *	(where the power supplier is not the physical (or electrical)
+ *	ancestor of a specific device.
+ *	The effect of this is that the supplier will not be powered down
+ *	before the power dependent.
+ */
+
+void device_pm_set_parent(struct device * dev, struct device * parent)
+{
+	struct device * old_parent = dev->power.pm_parent;
+	if (old_parent)
+		device_pm_release(old_parent);
+	dev->power.pm_parent = parent;
+	if (parent)
+		device_pm_hold(parent);
+}
+EXPORT_SYMBOL(device_pm_set_parent);
+
+int device_pm_add(struct device * dev)
+{
+	int error;
+
+	pr_debug("PM: Adding info for %s:%s\n",
+		 dev->bus ? dev->bus->name : "No Bus", dev->kobj.name);
+	down(&dpm_sem);
+	list_add_tail(&dev->power.entry,&dpm_active);
+	device_pm_set_parent(dev,dev->parent);
+	if ((error = dpm_sysfs_add(dev)))
+		list_del(&dev->power.entry);
+	up(&dpm_sem);
+	return error;
+}
+
+void device_pm_remove(struct device * dev)
+{
+	pr_debug("PM: Removing info for %s:%s\n",
+		 dev->bus ? dev->bus->name : "No Bus", dev->kobj.name);
+	down(&dpm_sem);
+	dpm_sysfs_remove(dev);
+	list_del(&dev->power.entry);
+	up(&dpm_sem);
+}
+
+

drivers/base/power/power.h

+
+
+enum {
+	DEVICE_PM_ON,
+	DEVICE_PM1,
+	DEVICE_PM2,
+	DEVICE_PM3,
+	DEVICE_PM_OFF,
+};
+
+/*
+ * shutdown.c
+ */
+
+extern int device_detach_shutdown(struct device *);
+extern void device_shutdown(void);
+
+
+#ifdef CONFIG_PM
+
+/*
+ * main.c
+ */
+
+/*
+ * Used to synchronize global power management operations.
+ */
+extern struct semaphore dpm_sem;
+
+/* 
+ * The PM lists.
+ */
+extern struct list_head dpm_active;
+extern struct list_head dpm_suspended;
+extern struct list_head dpm_off;
+extern struct list_head dpm_off_irq;
+
+
+static inline struct dev_pm_info * to_pm_info(struct list_head * entry)
+{
+	return container_of(entry,struct dev_pm_info,entry);
+}
+
+static inline struct device * to_device(struct list_head * entry)
+{
+	return container_of(to_pm_info(entry),struct device,power);
+}
+
+extern int device_pm_add(struct device *);
+extern void device_pm_remove(struct device *);
+
+/*
+ * sysfs.c
+ */
+
+extern int dpm_sysfs_add(struct device *);
+extern void dpm_sysfs_remove(struct device *);
+
+/*
+ * resume.c 
+ */
+extern int dpm_resume(void);
+extern void dpm_power_up(void);
+extern void dpm_power_up_irq(void);
+extern void power_up_device(struct device *);
+extern int resume_device(struct device *);
+
+/*
+ * suspend.c
+ */
+extern int suspend_device(struct device *, u32);
+extern int power_down_device(struct device *, u32);
+
+
+/*
+ * runtime.c
+ */
+
+extern int dpm_runtime_suspend(struct device *, u32);
+extern void dpm_runtime_resume(struct device *);
+
+#else /* CONFIG_PM */
+
+
+static inline int device_pm_add(struct device * dev)
+{
+	return 0;
+}
+static inline void device_pm_remove(struct device * dev)
+{
+
+}
+
+static inline int dpm_runtime_suspend(struct device * dev, u32 state)
+{
+	return 0;
+}
+
+static inline void dpm_runtime_resume(struct device * dev)
+{
+
+}
+
+#endif

drivers/base/power/resume.c

+/*
+ * resume.c - Functions for waking devices up.
+ *
+ * Copyright (c) 2003 Patrick Mochel
+ * Copyright (c) 2003 Open Source Development Labs
+ *
+ * This file is released under the GPLv2
+ *
+ */
+
+#include <linux/device.h>
+#include "power.h"
+
+extern int sysdev_resume(void);
+extern int sysdev_restore(void);
+
+
+/**
+ *	resume_device - Restore state for one device.
+ *	@dev:	Device.
+ *
+ */
+
+int resume_device(struct device * dev)
+{
+	struct device_driver * drv = dev->driver;
+
+	if (drv && drv->resume)
+		return drv->resume(dev,RESUME_RESTORE_STATE);
+	return 0;
+}
+
+/**
+ *	dpm_resume - Restore all device state.
+ *
+ *	Walk the dpm_suspended list and restore each device. As they are 
+ *	resumed, move the devices to the dpm_active list.
+ */
+
+int dpm_resume(void)
+{
+	while(!list_empty(&dpm_suspended)) {
+		struct list_head * entry = dpm_suspended.next;
+		struct device * dev = to_device(entry);
+		list_del_init(entry);
+		resume_device(dev);
+		list_add_tail(entry,&dpm_active);
+	}
+	return 0;
+}
+
+
+/**
+ *	device_pm_resume - Restore state of each device in system.
+ *
+ *	Restore system device state, then common device state. Finally,
+ *	release dpm_sem, as we're done with device PM.
+ */
+
+void device_pm_resume(void)
+{
+	sysdev_restore();
+	dpm_resume();
+	up(&dpm_sem);
+}
+
+
+/**
+ *	power_up_device - Power one device on.
+ *	@dev:	Device.
+ */
+
+void power_up_device(struct device * dev)
+{
+	struct device_driver * drv = dev->driver;
+	if (drv && drv->resume)
+		drv->resume(dev,RESUME_POWER_ON);
+}
+
+
+/**
+ *	device_power_up_irq - Power on some devices. 
+ *
+ *	Walk the dpm_off_irq list and power each device up. This 
+ *	is used for devices that required they be powered down with
+ *	interrupts disabled. As devices are powered on, they are moved to
+ *	the dpm_suspended list.
+ *
+ *	Interrupts must be disabled when calling this. 
+ */
+
+void dpm_power_up_irq(void)
+{
+	while(!list_empty(&dpm_off_irq)) {
+		struct list_head * entry = dpm_off_irq.next;
+		list_del_init(entry);
+		power_up_device(to_device(entry));
+		list_add_tail(entry,&dpm_suspended);
+	}
+}
+
+
+/**
+ *	dpm_power_up - Power on most devices.
+ *
+ *	Walk the dpm_off list and power each device up. This is used
+ *	to power on devices that were able to power down with interrupts
+ *	enabled. 
+ */
+
+void dpm_power_up(void)
+{
+	while (!list_empty(&dpm_off)) {
+		struct list_head * entry = dpm_off.next;
+		list_del_init(entry);
+		power_up_device(to_device(entry));
+		list_add_tail(entry,&dpm_suspended);
+	}
+}
+
+
+/**
+ *	device_pm_power_up - Turn on all devices.
+ *
+ *	First, power on system devices, which must happen with interrupts 
+ *	disbled. Then, power on devices that also require interrupts disabled.
+ *	Turn interrupts back on, and finally power up the rest of the normal
+ *	devices.
+ */
+
+void device_pm_power_up(void)
+{
+	sysdev_resume();
+	dpm_power_up_irq();
+	local_irq_enable();
+	dpm_power_up();
+}
+
+/**
+ * device_resume - resume all the devices in the system
+ * @level:	stage of resume process we're at 
+ * 
+ *	This function is deprecated, and should be replaced with appropriate
+ *	calls to device_pm_power_up() and device_pm_resume() above.
+ */
+
+void device_resume(u32 level)
+{
+
+	printk("%s is deprecated. Called from:\n",__FUNCTION__);
+	dump_stack();
+}
+

drivers/base/power/runtime.c

+/*
+ * drivers/base/power/runtime.c - Handling dynamic device power management.
+ *
+ * Copyright (c) 2003 Patrick Mochel
+ * Copyright (c) 2003 Open Source Development Lab
+ *
+ */
+
+#include <linux/device.h>
+#include "power.h"
+
+
+static void runtime_resume(struct device * dev)
+{
+	if (!dev->power.power_state)
+		return;
+
+	power_up_device(dev);
+	resume_device(dev);
+}
+
+
+/**
+ *	dpm_runtime_resume - Power one device back on.
+ *	@dev:	Device.
+ *
+ *	Bring one device back to the on state by first powering it 
+ *	on, then restoring state. We only operate on devices that aren't
+ *	already on. 
+ *	FIXME: We need to handle devices that are in an unknown state.
+ */
+
+void dpm_runtime_resume(struct device * dev)
+{
+	down(&dpm_sem);
+	runtime_resume(dev);
+	up(&dpm_sem);
+}
+
+
+/**
+ *	dpm_runtime_suspend - Put one device in low-power state.
+ *	@dev:	Device.
+ *	@state:	State to enter.
+ */
+
+int dpm_runtime_suspend(struct device * dev, u32 state)
+{
+	int error = 0;
+
+	down(&dpm_sem);
+	if (dev->power.power_state == state)
+		goto Done;
+
+	if (dev->power.power_state)
+		dpm_runtime_resume(dev);
+
+	error = suspend_device(dev,state);
+	if (!error) {
+		error = power_down_device(dev,state);
+		if (error)
+			goto ErrResume;
+		dev->power.power_state = state;
+	}
+ Done:
+	up(&dpm_sem);
+	return error;
+ ErrResume:
+	resume_device(dev);
+	goto Done;
+}
+
+
+/**
+ *	dpm_set_power_state - Update power_state field.
+ *	@dev:	Device.
+ *	@state:	Power state device is in.
+ *
+ *	This is an update mechanism for drivers to notify the core
+ *	what power state a device is in. Device probing code may not
+ *	always be able to tell, but we need accurate information to 
+ *	work reliably.
+ */
+void dpm_set_power_state(struct device * dev, u32 state)
+{
+	down(&dpm_sem);
+	dev->power.power_state = state;
+	up(&dpm_sem);
+}

drivers/base/power.c → drivers/base/power/shutdown.c

 /*
- * power.c - power management functions for the device tree.
+ * shutdown.c - power management functions for the device tree.
  * 
  * Copyright (c) 2002-3 Patrick Mochel
  *		 2002-3 Open Source Development Lab
  * 
  * This file is released under the GPLv2
- * 
- *  Kai Germaschewski contributed to the list walking routines.
  *
  */
 
 #undef DEBUG
 
 #include <linux/device.h>
-#include <linux/module.h>
 #include <asm/semaphore.h>
-#include "base.h"
+
+#include "power.h"
 
 #define to_dev(node) container_of(node,struct device,kobj.entry)
 
 extern struct subsystem devices_subsys;
 
+
+int device_detach_shutdown(struct device * dev)
+{
+	if (!dev->detach_state)
+		return 0;
+
+	if (dev->detach_state == DEVICE_PM_OFF) {
+		if (dev->driver && dev->driver->shutdown)
+			dev->driver->shutdown(dev);
+		return 0;
+	}
+	return dpm_runtime_suspend(dev,dev->detach_state);
+}
+
+
 /**
  * We handle system devices differently - we suspend and shut them 
  * down first and resume them first. That way, we do anything stupid like
@@ -31,88 +44,6 @@ extern struct subsystem devices_subsys;
  */
 
 extern int sysdev_shutdown(void);
-extern int sysdev_save(u32 state);
-extern int sysdev_suspend(u32 state);
-extern int sysdev_resume(void);
-extern int sysdev_restore(void);
-
-/**
- * device_suspend - suspend/remove all devices on the device ree
- * @state:	state we're entering
- * @level:	what stage of the suspend process we're at
- *    (emb: it seems that these two arguments are described backwards of what
- *          they actually mean .. is this correct?)
- *
- * The entries in the global device list are inserted such that they're in a
- * depth-first ordering.  So, simply interate over the list, and call the 
- * driver's suspend or remove callback for each device.
- */
-int device_suspend(u32 state, u32 level)
-{
-	struct device * dev;
-	int error = 0;
-
-	down_write(&devices_subsys.rwsem);
-	list_for_each_entry_reverse(dev,&devices_subsys.kset.list,kobj.entry) {
-		if (dev->driver && dev->driver->suspend) {
-			pr_debug("suspending device %s\n",dev->name);
-			error = dev->driver->suspend(dev,state,level);
-			if (error)
-				printk(KERN_ERR "%s: suspend returned %d\n",
-				       dev->name,error);
-		}
-	}
-	up_write(&devices_subsys.rwsem);
-
-	/*
-	 * Make sure system devices are suspended.
-	 */
-	switch(level) {
-	case SUSPEND_SAVE_STATE:
-		sysdev_save(state);
-		break;
-	case SUSPEND_POWER_DOWN:
-		sysdev_suspend(state);
-		break;
-	default:
-		break;
-	}
-
-	return error;
-}
-
-/**
- * device_resume - resume all the devices in the system
- * @level:	stage of resume process we're at 
- * 
- * Similar to device_suspend above, though we want to do a breadth-first
- * walk of the tree to make sure we wake up parents before children.
- * So, we iterate over the list backward. 
- */
-void device_resume(u32 level)
-{
-	struct device * dev;
-
-	switch (level) {
-	case RESUME_POWER_ON:
-		sysdev_resume();
-		break;
-	case RESUME_RESTORE_STATE:
-		sysdev_restore();
-		break;
-	default:
-		break;
-	}
-
-	down_write(&devices_subsys.rwsem);
-	list_for_each_entry(dev,&devices_subsys.kset.list,kobj.entry) {
-		if (dev->driver && dev->driver->resume) {
-			pr_debug("resuming device %s\n",dev->name);
-			dev->driver->resume(dev,level);
-		}
-	}
-	up_write(&devices_subsys.rwsem);
-}
 
 /**
  * device_shutdown - call ->remove() on each device to shutdown. 
@@ -135,6 +66,3 @@ void device_shutdown(void)
 	sysdev_shutdown();
 }
 
-EXPORT_SYMBOL(device_suspend);
-EXPORT_SYMBOL(device_resume);
-EXPORT_SYMBOL(device_shutdown);

drivers/base/power/suspend.c

+/*
+ * suspend.c - Functions for putting devices to sleep. 
+ *
+ * Copyright (c) 2003 Patrick Mochel
+ * Copyright (c) 2003 Open Source Development Labs
+ *
+ * This file is released under the GPLv2
+ *
+ */
+
+#include <linux/device.h>
+#include "power.h"
+ 
+extern int sysdev_save(u32 state);
+extern int sysdev_suspend(u32 state);
+
+/*
+ * The entries in the dpm_active list are in a depth first order, simply
+ * because children are guaranteed to be discovered after parents, and 
+ * are inserted at the back of the list on discovery. 
+ * 
+ * All list on the suspend path are done in reverse order, so we operate
+ * on the leaves of the device tree (or forests, depending on how you want
+ * to look at it ;) first. As nodes are removed from the back of the list, 
+ * they are inserted into the front of their destintation lists. 
+ *
+ * Things are the reverse on the resume path - iterations are done in
+ * forward order, and nodes are inserted at the back of their destination
+ * lists. This way, the ancestors will be accessed before their descendents.
+ */
+
+
+/**
+ *	suspend_device - Save state of one device.
+ *	@dev:	Device.
+ *	@state:	Power state device is entering.
+ */
+
+int suspend_device(struct device * dev, u32 state)
+{
+	struct device_driver * drv = dev->driver;
+	int error = 0;
+
+	if (drv && drv->suspend) 
+		error = drv->suspend(dev,state,SUSPEND_SAVE_STATE);
+
+	if (!error) {
+		list_del(&dev->power.entry);
+		list_add(&dev->power.entry,&dpm_suspended);
+	}
+	return error;
+}
+
+
+/**
+ *	device_pm_suspend - Save state and stop all devices in system.
+ *	@state:		Power state to put each device in. 
+ *
+ *	Walk the dpm_active list, call ->suspend() for each device, and move
+ *	it to dpm_suspended. If we hit a failure with any of the devices, call
+ *	dpm_resume() above to bring the suspended devices back to life. 
+ *
+ *	Have system devices save state last. 
+ *
+ *	Note this function leaves dpm_sem held to
+ *	a) block other devices from registering.
+ *	b) prevent other PM operations from happening after we've begun.
+ *	c) make sure we're exclusive when we disable interrupts.
+ *
+ *	device_pm_resume() will release dpm_sem after restoring state to 
+ *	all devices (as will this on error). You must call it once you've 
+ *	called device_pm_suspend().
+ */
+
+int device_pm_suspend(u32 state)
+{
+	int error = 0;
+
+	down(&dpm_sem);
+	while(!list_empty(&dpm_active)) {
+		struct list_head * entry = dpm_active.prev;
+		struct device * dev = to_device(entry);
+		if ((error = suspend_device(dev,state)))
+			goto Error;
+	}
+
+	if ((error = sysdev_save(state)))
+		goto Error;
+ Done:
+	return error;
+ Error:
+	dpm_resume();
+	up(&dpm_sem);
+	goto Done;
+}
+
+
+/**
+ *	power_down_device - Put one device in low power state.
+ *	@dev:	Device.
+ *	@state:	Power state to enter.
+ */
+
+int power_down_device(struct device * dev, u32 state)
+{
+	struct device_driver * drv = dev->driver;
+	int error = 0;
+
+	if (drv && drv->suspend)
+		error = drv->suspend(dev,state,SUSPEND_POWER_DOWN);
+	if (!error) {
+		list_del(&dev->power.entry);
+		list_add(&dev->power.entry,&dpm_off);
+	}
+	return error;
+}
+
+
+/**
+ *	dpm_power_down - Put all devices in low power state.
+ *	@state:	Power state to enter.
+ *
+ *	Walk the dpm_suspended list (with interrupts enabled) and try
+ *	to power down each each. If any fail with -EAGAIN, they require
+ *	the call to be done with interrupts disabled. So, we move them to
+ *	the dpm_off_irq list.
+ *
+ *	If the call succeeds, we move each device to the dpm_off list.
+ */
+
+static int dpm_power_down(u32 state)
+{
+	while(!list_empty(&dpm_suspended)) {
+		struct list_head * entry = dpm_suspended.prev;
+		int error;
+		error = power_down_device(to_device(entry),state);
+		if (error) {
+			if (error == -EAGAIN) {
+				list_del(entry);
+				list_add(entry,&dpm_off_irq);
+				continue;
+			}
+			return error;
+		}
+	}
+	return 0;
+}
+
+
+/**
+ *	dpm_power_down_irq - Power down devices without interrupts.
+ *	@state:	State to enter.
+ *
+ *	Walk the dpm_off_irq list (built by dpm_power_down) and power
+ *	down each device that requires the call to be made with interrupts
+ *	disabled. 
+ */
+
+static int dpm_power_down_irq(u32 state)
+{
+	struct device * dev;
+	int error = 0;
+
+	list_for_each_entry_reverse(dev,&dpm_off_irq,power.entry) {
+		if ((error = power_down_device(dev,state)))
+			break;
+	}
+	return error;
+}
+
+
+/**
+ *	device_pm_power_down - Put all devices in low power state.
+ *	@state:		Power state to enter.
+ *
+ *	Walk the dpm_suspended list, calling ->power_down() for each device.
+ *	Check the return value for each. If it returns 0, then we move the
+ *	the device to the dpm_off list. If it returns -EAGAIN, we move it to 
+ *	the dpm_off_irq list. If we get a different error, try and back out. 
+ *
+ *	dpm_irq_off is for devices that require interrupts to be disabled to
+ *	either to power down the device or power it back on. 
+ *
+ *	When we're done, we disable interrrupts (!!) and walk the dpm_off_irq
+ *	list to shut down the devices that need interrupts disabled. 
+ *
+ *	This function leaves interrupts disabled on exit, since powering down
+ *	devices should be the very last thing before the system is put into a 
+ *	low-power state. 
+ *
+ *	device_pm_power_on() should be called to re-enable interrupts and power
+ *	the devices back on. 
+ */
+
+int device_pm_power_down(u32 state)
+{
+	int error = 0;
+
+	if ((error = dpm_power_down(state)))
+		goto ErrorIRQOn;
+	local_irq_disable();
+	if ((error = dpm_power_down_irq(state)))
+		goto ErrorIRQOff;
+
+	sysdev_suspend(state);
+ Done:
+	return error;
+
+ ErrorIRQOff:
+	dpm_power_up_irq();
+	local_irq_enable();
+ ErrorIRQOn:
+	dpm_power_up();
+	goto Done;
+}
+
+
+/**
+ * device_suspend - suspend all devices on the device ree
+ * @state:	state we're entering
+ * @level:	Stage of suspend sequence we're in.
+ *
+ *
+ *	This function is deprecated. Calls should be replaced with
+ *	appropriate calls to device_pm_suspend() and device_pm_power_down().
+ */
+
+int device_suspend(u32 state, u32 level)
+{
+
+	printk("%s Called from:\n",__FUNCTION__);
+	dump_stack();
+	return -EFAULT;
+}
+

drivers/base/power/sysfs.c

+/*
+ * drivers/base/power/sysfs.c - sysfs entries for device PM
+ */
+
+#include <linux/device.h>
+#include "power.h"
+
+
+/**
+ *	state - Control current power state of device
+ *
+ *	show() returns the current power state of the device. '0' indicates
+ *	the device is on. Other values (1-3) indicate the device is in a low
+ *	power state. 
+ *
+ *	store() sets the current power state, which is an integer value 
+ *	between 0-3. If the device is on ('0'), and the value written is 
+ *	greater than 0, then the device is placed directly into the low-power
+ *	state (via its driver's ->suspend() method).
+ *	If the device is currently in a low-power state, and the value is 0,
+ *	the device is powered back on (via the ->resume() method).
+ *	If the device is in a low-power state, and a different low-power state
+ *	is requested, the device is first resumed, then suspended into the new
+ *	low-power state.
+ */
+
+static ssize_t state_show(struct device * dev, char * buf)
+{
+	return sprintf(buf,"%u\n",dev->power.power_state);
+}
+
+static ssize_t state_store(struct device * dev, const char * buf, size_t n)
+{
+	u32 state;
+	char * rest;
+	int error = 0;
+
+	state = simple_strtoul(buf,&rest,10);
+	if (rest)
+		return -EINVAL;
+	if (state)
+		error = dpm_runtime_suspend(dev,state);
+	else
+		dpm_runtime_resume(dev);
+	return error ? error : n;
+}
+
+static DEVICE_ATTR(state,0644,state_show,state_store);
+
+
+static struct attribute * power_attrs[] = {
+	&dev_attr_state.attr,
+	NULL,
+};
+static struct attribute_group pm_attr_group = {
+	.name	= "power",
+	.attrs	= power_attrs,
+};
+
+int dpm_sysfs_add(struct device * dev)
+{
+	return sysfs_create_group(&dev->kobj,&pm_attr_group);
+}
+
+void dpm_sysfs_remove(struct device * dev)
+{
+	sysfs_remove_group(&dev->kobj,&pm_attr_group);
+}

include/linux/device.h

@@ -19,6 +19,7 @@
 #include <linux/types.h>
 #include <linux/ioport.h>
 #include <linux/module.h>
+#include <linux/pm.h>
 #include <asm/semaphore.h>
 #include <asm/atomic.h>
 
@@ -41,13 +42,6 @@ enum {
 	RESUME_ENABLE,
 };
 
-enum device_state {
-	DEVICE_UNINITIALIZED	= 0,
-	DEVICE_INITIALIZED	= 1,
-	DEVICE_REGISTERED	= 2,
-	DEVICE_GONE		= 3,
-};
-
 struct device;
 struct device_driver;
 struct class;
@@ -64,8 +58,8 @@ struct bus_type {
 	struct device * (*add)	(struct device * parent, char * bus_id);
 	int		(*hotplug) (struct device *dev, char **envp, 
 				    int num_envp, char *buffer, int buffer_size);
-};
 
+};
 
 extern int bus_register(struct bus_type * bus);
 extern void bus_unregister(struct bus_type * bus);
@@ -269,13 +263,16 @@ struct device {
 	void		*driver_data;	/* data private to the driver */
 	void		*platform_data;	/* Platform specific data (e.g. ACPI,
 					   BIOS data relevant to device) */
-
+	struct dev_pm_info	power;
 	u32		power_state;	/* Current operating state. In
 					   ACPI-speak, this is D0-D3, D0
 					   being fully functional, and D3
 					   being off. */
 
 	unsigned char *saved_state;	/* saved device state */
+	u32		detach_state;	/* State to enter when device is
+					   detached from its driver. */
+
 	u64		*dma_mask;	/* dma mask (if dma'able device) */
 
 	void	(*release)(struct device * dev);

include/linux/pm.h

@@ -25,6 +25,7 @@
 
 #include <linux/config.h>
 #include <linux/list.h>
+#include <asm/atomic.h>
 
 /*
  * Power management requests
@@ -118,29 +119,29 @@ extern int pm_active;
 /*
  * Register a device with power management
  */
-struct pm_dev *pm_register(pm_dev_t type,
-			   unsigned long id,
-			   pm_callback callback);
+struct pm_dev __deprecated *pm_register(pm_dev_t type,
+					unsigned long id,
+					pm_callback callback);
 
 /*
  * Unregister a device with power management
  */
-void pm_unregister(struct pm_dev *dev);
+void __deprecated pm_unregister(struct pm_dev *dev);
 
 /*
  * Unregister all devices with matching callback
  */
-void pm_unregister_all(pm_callback callback);
+void __deprecated pm_unregister_all(pm_callback callback);
 
 /*
  * Send a request to a single device
  */
-int pm_send(struct pm_dev *dev, pm_request_t rqst, void *data);
+int __deprecated pm_send(struct pm_dev *dev, pm_request_t rqst, void *data);
 
 /*
  * Send a request to all devices
  */
-int pm_send_all(pm_request_t rqst, void *data);
+int __deprecated pm_send_all(pm_request_t rqst, void *data);
 
 /*
  * Find a device
@@ -188,6 +189,26 @@ static inline void pm_dev_idle(struct pm_dev *dev) {}
 extern void (*pm_idle)(void);
 extern void (*pm_power_off)(void);
 
+struct device;
+
+struct dev_pm_info {
+#ifdef	CONFIG_PM
+	u32			power_state;
+	u8			* saved_state;
+	atomic_t		pm_users;
+	struct device		* pm_parent;
+	struct list_head	entry;
+#endif
+};
+
+extern void device_pm_set_parent(struct device * dev, struct device * parent);
+
+extern int device_pm_suspend(u32 state);
+extern int device_pm_power_down(u32 state);
+extern void device_pm_power_up(void);
+extern void device_pm_resume(void);
+
+
 #endif /* __KERNEL__ */
 
 #endif /* _LINUX_PM_H */

include/linux/reboot.h

@@ -21,7 +21,7 @@
  * CAD_OFF     Ctrl-Alt-Del sequence sends SIGINT to init task.
  * POWER_OFF   Stop OS and remove all power from system, if possible.
  * RESTART2    Restart system using given command string.
- * SW_SUSPEND  Suspend system using Software Suspend if compiled in
+ * SW_SUSPEND  Suspend system using software suspend if compiled in.
  */
 
 #define	LINUX_REBOOT_CMD_RESTART	0x01234567

include/linux/suspend.h

@@ -11,9 +11,6 @@
 
 extern unsigned char software_suspend_enabled;
 
-#define NORESUME	 1
-#define RESUME_SPECIFIED 2
-
 #ifdef CONFIG_SOFTWARE_SUSPEND
 /* page backup entry */
 typedef struct pbe {
@@ -50,8 +47,7 @@ extern int shrink_mem(void);
 extern void drain_local_pages(void);
 
 /* kernel/suspend.c */
-extern void software_suspend(void);
-extern void software_resume(void);
+extern int software_suspend(void);
 
 extern int register_suspend_notifier(struct notifier_block *);
 extern int unregister_suspend_notifier(struct notifier_block *);
@@ -72,10 +68,10 @@ extern void do_suspend_lowlevel(int resume);
 extern void do_suspend_lowlevel_s4bios(int resume);
 
 #else	/* CONFIG_SOFTWARE_SUSPEND */
-static inline void software_suspend(void)
+static inline int software_suspend(void)
 {
+	return -EPERM;
 }
-#define software_resume()		do { } while(0)
 #define register_suspend_notifier(a)	do { } while(0)
 #define unregister_suspend_notifier(a)	do { } while(0)
 #endif	/* CONFIG_SOFTWARE_SUSPEND */

init/do_mounts.c

@@ -385,10 +385,6 @@ void __init prepare_namespace(void)
 
 	is_floppy = MAJOR(ROOT_DEV) == FLOPPY_MAJOR;
 
-	/* This has to be before mounting root, because even readonly mount of reiserfs would replay
-	   log corrupting stuff */
-	software_resume();
-
 	if (initrd_load())
 		goto out;
 

kernel/Makefile

@@ -14,7 +14,7 @@ obj-$(CONFIG_SMP) += cpu.o
 obj-$(CONFIG_UID16) += uid16.o
 obj-$(CONFIG_MODULES) += ksyms.o module.o
 obj-$(CONFIG_KALLSYMS) += kallsyms.o
-obj-$(CONFIG_PM) += pm.o power/
+obj-$(CONFIG_PM) += power/
 obj-$(CONFIG_CPU_FREQ) += cpufreq.o
 obj-$(CONFIG_BSD_PROCESS_ACCT) += acct.o
 obj-$(CONFIG_COMPAT) += compat.o

kernel/power/Makefile

-obj-y				:= process.o console.o
+obj-y				:= main.o process.o console.o pm.o
 obj-$(CONFIG_SOFTWARE_SUSPEND)	+= swsusp.o
+
+obj-$(CONFIG_MAGIC_SYSRQ)	+= poweroff.o

kernel/power/console.c

+/*
+ * drivers/power/process.c - Functions for saving/restoring console.
+ *
+ * Originally from swsusp.
+ */
+
 #include <linux/vt_kern.h>
 #include <linux/kbd_kern.h>
 #include "power.h"
@@ -14,13 +20,13 @@ int pm_prepare_console(void)
 #ifdef SUSPEND_CONSOLE
 	orig_fgconsole = fg_console;
 
-	if(vc_allocate(SUSPEND_CONSOLE))
+	if (vc_allocate(SUSPEND_CONSOLE))
 	  /* we can't have a free VC for now. Too bad,
 	   * we don't want to mess the screen for now. */
 		return 1;
 
-	set_console (SUSPEND_CONSOLE);
-	if(vt_waitactive(SUSPEND_CONSOLE)) {
+	set_console(SUSPEND_CONSOLE);
+	if (vt_waitactive(SUSPEND_CONSOLE)) {
 		pr_debug("Suspend: Can't switch VCs.");
 		return 1;
 	}
@@ -34,7 +40,7 @@ void pm_restore_console(void)
 {
 	console_loglevel = orig_loglevel;
 #ifdef SUSPEND_CONSOLE
-	set_console (orig_fgconsole);
+	set_console(orig_fgconsole);
 #endif
 	return;
 }

kernel/power/main.c

+/*
+ * kernel/power/main.c - PM subsystem core functionality.
+ *
+ * Copyright (c) 2003 Patrick Mochel
+ * Copyright (c) 2003 Open Source Development Lab
+ * 
+ * This file is release under the GPLv2
+ *
+ */
+
+#include <linux/kobject.h>
+#include <linux/string.h>
+#include <linux/errno.h>
+#include <linux/init.h>
+#include <linux/pm.h>
+
+
+
+static int standby(void)
+{
+	return 0;
+}
+
+static int suspend(void)
+{
+	return 0;
+}
+
+static int hibernate(void)
+{
+	return 0;
+}
+
+#define decl_state(_name) \
+	{ .name = __stringify(_name), .fn = _name }
+
+struct pm_state {
+	char * name;
+	int (*fn)(void);
+} pm_states[] = {
+	decl_state(standby),
+	decl_state(suspend),
+	decl_state(hibernate),
+	{ NULL },
+};
+
+
+static int enter_state(struct pm_state * state)
+{
+	return state->fn();
+}
+
+
+
+decl_subsys(power,NULL,NULL);
+
+
+#define power_attr(_name) \
+static struct subsys_attribute _name##_attr = {	\
+	.attr	= {				\
+		.name = __stringify(_name),	\
+		.mode = 0644,			\
+	},					\
+	.show	= _name##_show,			\
+	.store	= _name##_store,		\
+}
+
+/**
+ *	state - control system power state.
+ *
+ *	show() returns what states are supported, which is hard-coded to
+ *	'standby' (Power-On Suspend), 'suspend' (Suspend-to-RAM), and
+ *	'hibernate' (Suspend-to-Disk).
+ *
+ *	store() accepts one of those strings, translates it into the 
+ *	proper enumerated value, and initiates a suspend transition.
+ */
+
+static ssize_t state_show(struct subsystem * subsys, char * buf)
+{
+	struct pm_state * state;
+	char * s = buf;
+
+	for (state = &pm_states[0]; state->name; state++)
+		s += sprintf(s,"%s ",state->name);
+	s += sprintf(s,"\n");
+	return (s - buf);
+}
+
+static ssize_t state_store(struct subsystem * s, const char * buf, size_t n)
+{
+	struct pm_state * state;
+	int error;
+	char * end = strchr(buf,'\n');
+	
+	if (end)
+		*end = '\0';
+
+	for (state = &pm_states[0]; state; state++) {
+		if (!strcmp(buf,state->name))
+			break;
+	}
+	if (!state)
+		return -EINVAL;
+	error = enter_state(state);
+	return error ? error : n;
+}
+
+power_attr(state);
+
+static struct attribute * g[] = {
+	&state_attr.attr,
+	NULL,
+};
+
+static struct attribute_group attr_group = {
+	.attrs = g,
+};
+
+
+static int pm_init(void)
+{
+	int error = subsystem_register(&power_subsys);
+	if (!error)
+		error = sysfs_create_group(&power_subsys.kset.kobj,&attr_group);
+	return error;
+}
+
+core_initcall(pm_init);

kernel/pm.c → kernel/power/pm.c

@@ -24,7 +24,6 @@
 #include <linux/slab.h>
 #include <linux/pm.h>
 #include <linux/interrupt.h>
-#include <linux/sysrq.h>
 
 int pm_active;
 
@@ -295,39 +294,3 @@ EXPORT_SYMBOL(pm_find);
 EXPORT_SYMBOL(pm_active);
 
 
-#ifdef CONFIG_MAGIC_SYSRQ
-
-/**
- * handle_poweroff	-	sysrq callback for power down
- * @key: key pressed (unused)
- * @pt_regs: register state (unused)
- * @kbd: keyboard state (unused)
- * @tty: tty involved (unused)
- *
- * When the user hits Sys-Rq o to power down the machine this is the
- * callback we use.
- */
-
-static void handle_poweroff (int key, struct pt_regs *pt_regs,
-			     struct tty_struct *tty)
-{
-	if (pm_power_off)
-		pm_power_off();
-}
-
-static struct sysrq_key_op	sysrq_poweroff_op = {
-	.handler        = handle_poweroff,
-	.help_msg       = "powerOff",
-	.action_msg     = "Power Off\n"
-};
-
-#endif  /* CONFIG_MAGIC_SYSRQ */
-
-
-static int pm_init(void)
-{
-	register_sysrq_key('o', &sysrq_poweroff_op);
-	return 0;
-}
-
-subsys_initcall(pm_init);

kernel/power/poweroff.c

+/*
+ * poweroff.c - sysrq handler to gracefully power down machine.
+ *
+ * This file is released under the GPL v2
+ */
+
+#include <linux/kernel.h>
+#include <linux/sysrq.h>
+#include <linux/init.h>
+#include <linux/pm.h>
+
+
+/**
+ * handle_poweroff	-	sysrq callback for power down
+ * @key: key pressed (unused)
+ * @pt_regs: register state (unused)
+ * @kbd: keyboard state (unused)
+ * @tty: tty involved (unused)
+ *
+ * When the user hits Sys-Rq o to power down the machine this is the
+ * callback we use.
+ */
+
+static void handle_poweroff (int key, struct pt_regs *pt_regs,
+			     struct tty_struct *tty)
+{
+	if (pm_power_off)
+		pm_power_off();
+}
+
+static struct sysrq_key_op	sysrq_poweroff_op = {
+	.handler        = handle_poweroff,
+	.help_msg       = "powerOff",
+	.action_msg     = "Power Off\n"
+};
+
+
+static int pm_sysrq_init(void)
+{
+	register_sysrq_key('o', &sysrq_poweroff_op);
+	return 0;
+}
+
+subsys_initcall(pm_sysrq_init);

kernel/power/process.c

@@ -8,6 +8,7 @@
 
 #undef DEBUG
 
+#include <linux/smp_lock.h>
 #include <linux/interrupt.h>
 #include <linux/suspend.h>
 #include <linux/module.h>

kernel/power/swsusp.c

@@ -67,7 +67,7 @@
 
 extern long sys_sync(void);
 
-unsigned char software_suspend_enabled = 0;
+unsigned char software_suspend_enabled = 1;
 
 #define __ADDRESS(x)  ((unsigned long) phys_to_virt(x))
 #define ADDRESS(x) __ADDRESS((x) << PAGE_SHIFT)
@@ -85,8 +85,7 @@ spinlock_t suspend_pagedir_lock __nosavedata = SPIN_LOCK_UNLOCKED;
 static int pagedir_order_check;
 static int nr_copy_pages_check;
 
-static int resume_status;
-static char resume_file[256] = "";			/* For resume= kernel option */
+static char resume_file[256];			/* For resume= kernel option */
 static dev_t resume_device;
 /* Local variables that should not be affected by save */
 unsigned int nr_copy_pages __nosavedata = 0;
@@ -352,15 +351,10 @@ static int count_and_copy_data_pages(struct pbe *pagedir_p)
 	int pfn;
 	struct page *page;
 	
-#ifdef CONFIG_DISCONTIGMEM
-	panic("Discontingmem not supported");
-#else
 	BUG_ON (max_pfn != num_physpages);
-#endif
+
 	for (pfn = 0; pfn < max_pfn; pfn++) {
 		page = pfn_to_page(pfn);
-		if (PageHighMem(page))
-			panic("Swsusp not supported on highmem boxes. Send 1GB of RAM to <pavel@ucw.cz> and try again ;-).");
 
 		if (!PageReserved(page)) {
 			if (PageNosave(page))
@@ -479,19 +473,23 @@ static void drivers_unsuspend(void)
 /* Called from process context */
 static int drivers_suspend(void)
 {
-	device_suspend(4, SUSPEND_NOTIFY);
-	device_suspend(4, SUSPEND_SAVE_STATE);
-	device_suspend(4, SUSPEND_DISABLE);
-	if(!pm_suspend_state) {
+	if (device_suspend(4, SUSPEND_NOTIFY))
+		return -EIO;
+	if (device_suspend(4, SUSPEND_SAVE_STATE)) {
+		device_resume(RESUME_RESTORE_STATE);
+		return -EIO;
+	}
+	if (!pm_suspend_state) {
 		if(pm_send_all(PM_SUSPEND,(void *)3)) {
 			printk(KERN_WARNING "Problem while sending suspend event\n");
-			return(1);
+			return -EIO;
 		}
 		pm_suspend_state=1;
 	} else
 		printk(KERN_WARNING "PM suspend state already raised\n");
+	device_suspend(4, SUSPEND_DISABLE);
 	  
-	return(0);
+	return 0;
 }
 
 #define RESUME_PHASE1 1 /* Called from interrupts disabled */
@@ -504,7 +502,7 @@ static void drivers_resume(int flags)
 		device_resume(RESUME_ENABLE);
 	}
   	if (flags & RESUME_PHASE2) {
-		if(pm_suspend_state) {
+		if (pm_suspend_state) {
 			if(pm_send_all(PM_RESUME,(void *)0))
 				printk(KERN_WARNING "Problem while sending resume event\n");
 			pm_suspend_state=0;
@@ -696,7 +694,7 @@ void do_magic_suspend_2(void)
 	mark_swapfiles(((swp_entry_t) {0}), MARK_SWAP_RESUME);
 }
 
-static void do_software_suspend(void)
+static int do_software_suspend(void)
 {
 	arch_prepare_suspend();
 	if (pm_prepare_console())
@@ -715,7 +713,7 @@ static void do_software_suspend(void)
 		blk_run_queues();
 
 		/* Save state of all device drivers, and stop them. */		   
-		if(drivers_suspend()==0)
+		if (drivers_suspend()==0)
 			/* If stopping device drivers worked, we proceed basically into
 			 * suspend_save_image.
 			 *
@@ -731,20 +729,35 @@ static void do_software_suspend(void)
 	software_suspend_enabled = 1;
 	MDELAY(1000);
 	pm_restore_console();
+	return 0;
 }
 
-/*
- * This is main interface to the outside world. It needs to be
- * called from process context.
+
+/**
+ *	software_suspend - initiate suspend-to-swap transition.
+ *
+ *	This is main interface to the outside world. It needs to be
+ *	called from process context.
  */
-void software_suspend(void)
+
+int software_suspend(void)
 {
 	if(!software_suspend_enabled)
-		return;
+		return -EINVAL;
+
+	if (num_online_cpus() > 1) {
+		printk(KERN_WARNING "swsusp does not support SMP.\n");	
+		return -EPERM;
+	}
+
+#if defined (CONFIG_HIGHMEM) || defined (COFNIG_DISCONTIGMEM)
+	printk("swsusp is not supported with high- or discontig-mem.\n");
+	return -EPERM;
+#endif
 
 	software_suspend_enabled = 0;
 	might_sleep();
-	do_software_suspend();
+	return do_software_suspend();
 }
 
 /* More restore stuff */
@@ -890,31 +903,9 @@ static int bdev_read_page(struct block_device *bdev, long pos, void *buf)
 	return 0;
 } 
 
-static int bdev_write_page(struct block_device *bdev, long pos, void *buf)
-{
-#if 0
-	struct buffer_head *bh;
-	BUG_ON (pos%PAGE_SIZE);
-	bh = __bread(bdev, pos/PAGE_SIZE, PAGE_SIZE);
-	if (!bh || (!bh->b_data)) {
-		return -1;
-	}
-	memcpy(bh->b_data, buf, PAGE_SIZE);	/* FIXME: may need kmap() */
-	BUG_ON(!buffer_uptodate(bh));
-	generic_make_request(WRITE, bh);
-	if (!buffer_uptodate(bh))
-		printk(KERN_CRIT "%sWarning %s: Fixing swap signatures unsuccessful...\n", name_resume, resume_file);
-	wait_on_buffer(bh);
-	brelse(bh);
-	return 0;
-#endif
-	printk(KERN_CRIT "%sWarning %s: Fixing swap signatures unimplemented...\n", name_resume, resume_file);
-	return 0;
-}
-
 extern dev_t __init name_to_dev_t(const char *line);
 
-static int __read_suspend_image(struct block_device *bdev, union diskpage *cur, int noresume)
+static int __read_suspend_image(struct block_device *bdev, union diskpage *cur)
 {
 	swp_entry_t next;
 	int i, nr_pgdir_pages;
@@ -939,18 +930,9 @@ static int __read_suspend_image(struct block_device *bdev, union diskpage *cur,
 	else if (!memcmp("S2",cur->swh.magic.magic,2))
 		memcpy(cur->swh.magic.magic,"SWAPSPACE2",10);
 	else {
-		if (noresume)
-			return -EINVAL;
-		panic("%sUnable to find suspended-data signature (%.10s - misspelled?\n", 
+		printk("swsusp: %s: Unable to find suspended-data signature (%.10s - misspelled?\n", 
 			name_resume, cur->swh.magic.magic);
-	}
-	if (noresume) {
-		/* We don't do a sanity check here: we want to restore the swap
-		   whatever version of kernel made the suspend image;
-		   We need to write swap, but swap is *not* enabled so
-		   we must write the device directly */
-		printk("%s: Fixing swap signatures %s...\n", name_resume, resume_file);
-		bdev_write_page(bdev, 0, cur);
+		return -EFAULT;
 	}
 
 	printk( "%sSignature found, resuming\n", name_resume );
@@ -1000,7 +982,7 @@ static int __read_suspend_image(struct block_device *bdev, union diskpage *cur,
 	return 0;
 }
 
-static int read_suspend_image(const char * specialfile, int noresume)
+static int read_suspend_image(const char * specialfile)
 {
 	union diskpage *cur;
 	unsigned long scratch_page = 0;
@@ -1019,7 +1001,7 @@ static int read_suspend_image(const char * specialfile, int noresume)
 			error = PTR_ERR(bdev);
 		} else {
 			set_blocksize(bdev, PAGE_SIZE);
-			error = __read_suspend_image(bdev, cur, noresume);
+			error = __read_suspend_image(bdev, cur);
 			blkdev_put(bdev, BDEV_RAW);
 		}
 	} else error = -ENOMEM;
@@ -1048,64 +1030,47 @@ static int read_suspend_image(const char * specialfile, int noresume)
 	return error;
 }
 
-/*
- * Called from init kernel_thread.
- * We check if we have an image and if so we try to resume
+/**
+ *	software_resume - Check and load saved image from swap.
+ *
+ *	Defined as a late_initcall, so it gets called after all devices
+ *	have been probed and initialized, but before we've mounted anything.
  */
 
-void software_resume(void)
+static int software_resume(void)
 {
-	if (num_online_cpus() > 1) {
-		printk(KERN_WARNING "Software Suspend has malfunctioning SMP support. Disabled :(\n");	
-		return;
-	}
-	/* We enable the possibility of machine suspend */
-	software_suspend_enabled = 1;
-	if (!resume_status)
-		return;
-
-	printk( "%s", name_resume );
-	if (resume_status == NORESUME) {
-		if(resume_file[0])
-			read_suspend_image(resume_file, 1);
-		printk( "disabled\n" );
-		return;
-	}
-	MDELAY(1000);
+	if (!strlen(resume_file))
+		return 0;
 
 	if (pm_prepare_console())
 		printk("swsusp: Can't allocate a console... proceeding\n");
 
-	if (!resume_file[0] && resume_status == RESUME_SPECIFIED) {
-		printk( "suspension device unspecified\n" );
-		return;
-	}
+	printk("swsusp: %s\n", name_resume );
+
+	MDELAY(1000);
 
-	printk( "resuming from %s\n", resume_file);
-	if (read_suspend_image(resume_file, 0))
+	printk("swsusp: resuming from %s\n", resume_file);
+	if (read_suspend_image(resume_file))
 		goto read_failure;
 	do_magic(1);
-	panic("This never returns");
+	printk("swsusp: Resume failed. Continuing.\n");
 
 read_failure:
 	pm_restore_console();
-	return;
+	return -EFAULT;
 }
 
+late_initcall(software_resume);
+
 static int __init resume_setup(char *str)
 {
-	if (resume_status == NORESUME)
-		return 1;
-
 	strncpy( resume_file, str, 255 );
-	resume_status = RESUME_SPECIFIED;
-
 	return 1;
 }
 
 static int __init noresume_setup(char *str)
 {
-	resume_status = NORESUME;
+	resume_file[0] = '\0';
 	return 1;
 }