PowerOp Design and working patch

[david singleton <dsingleton@mvista.com>]

[-- Attachment #1: Type: text/plain, Size: 2026 bytes --]

Here is a patch that implements a version of the PowerOp concept.

The goal of PowerOp power management is to provide a framework that 
unifies
and simplifies the various power management infrastructures in Linux.  
The
three infrastructures Power Op is concerned with are:

         1) basic suspend/resume power management (CONFIG_PM)

         2) basic processor frequency management (CONFIG_CPUFREQ)

         3) SourceForge's Dynamic Power Management (CONFIG_DPM)

All three power management infrastructures are concerned with 
controlling
power states of the system, and interestingly enough they all perform 
the
same basic operational steps to control changes in power state.

PowerOp uses the existing power management sysfs infrastructure and 
extends it
to perform cpufreq and dynamic power management operations. The 
traditional
suspend to memory or disk (or swap) infrastructure has the correct 
operational
structure that supports all types of power state change.

The CPUFREQ table based frequency control makes controlling cpu 
frequency
simple and straight forward.  The user doesn't get to set the cpu to
any speed, but only to supported speeds that have been provided by
the hardware vendor and validated.

Dynamic Power Management eventually treats all types of power states as 
operating points,
wether it's a suspend operating point, a particular frequency, or a 
specific
voltage.

By combining the best of all of these power management infrastructures
PowerOp uses the operational structure of tradition CONFIG_PM power
management and converts all power states, frequency, voltage, idle or
suspend to the CPUFREQ concept of only supported and validated operating
points.

PowerOp allows setting any operating point supported by the system by 
setting
the name of the operating point into /sys/power/state.

I apologize for being so late to the table, but I find a working patch 
easier to discuss
and I needed to see if I could unify and simplify all three power 
management infrastructures.

David



[-- Attachment #2: powerop.patch --]
[-- Type: application/octet-stream, Size: 24610 bytes --]

 Documentation/power/powerop.txt                   |  168 +++++++++++++++++++
 Makefile                                          |    2 
 arch/i386/kernel/cpu/cpufreq/speedstep-centrino.c |  108 ++++++++++++
 drivers/cpufreq/cpufreq.c                         |   34 +++
 include/linux/cpufreq.h                           |    2 
 include/linux/pm.h                                |   26 ++-
 kernel/power/main.c                               |  189 +++++++++++++++++-----
 kernel/power/power.h                              |    2 
 8 files changed, 486 insertions(+), 45 deletions(-)

Index: linux-2.6.17/kernel/power/main.c
===================================================================
--- linux-2.6.17.orig/kernel/power/main.c
+++ linux-2.6.17/kernel/power/main.c
@@ -49,7 +49,7 @@ void pm_set_ops(struct pm_ops * ops)
  *	the platform can enter the requested state.
  */
 
-static int suspend_prepare(suspend_state_t state)
+static int suspend_prepare(struct power_op * state)
 {
 	int error = 0;
 	unsigned int free_pages;
@@ -82,7 +82,7 @@ static int suspend_prepare(suspend_state
 	}
 
 	if (pm_ops->prepare) {
-		if ((error = pm_ops->prepare(state)))
+		if ((error = pm_ops->prepare(state->type)))
 			goto Thaw;
 	}
 
@@ -94,7 +94,7 @@ static int suspend_prepare(suspend_state
 	return 0;
  Finish:
 	if (pm_ops->finish)
-		pm_ops->finish(state);
+		pm_ops->finish(state->type);
  Thaw:
 	thaw_processes();
  Enable_cpu:
@@ -104,7 +104,7 @@ static int suspend_prepare(suspend_state
 }
 
 
-int suspend_enter(suspend_state_t state)
+int suspend_enter(struct power_op * state)
 {
 	int error = 0;
 	unsigned long flags;
@@ -115,7 +115,7 @@ int suspend_enter(suspend_state_t state)
 		printk(KERN_ERR "Some devices failed to power down\n");
 		goto Done;
 	}
-	error = pm_ops->enter(state);
+	error = pm_ops->enter(state->type);
 	device_power_up();
  Done:
 	local_irq_restore(flags);
@@ -131,36 +131,94 @@ int suspend_enter(suspend_state_t state)
  *	console that we've allocated. This is not called for suspend-to-disk.
  */
 
-static void suspend_finish(suspend_state_t state)
+static void suspend_finish(struct power_op * state)
 {
 	device_resume();
 	resume_console();
 	thaw_processes();
 	enable_nonboot_cpus();
 	if (pm_ops && pm_ops->finish)
-		pm_ops->finish(state);
+		pm_ops->finish(state->type);
 	pm_restore_console();
 }
 
 
+struct power_op *current_state;
+struct power_op pm_states = {
+	.name = "default",
+	.type = PM_SUSPEND_ON,
+};
 
-
-static const char * const pm_states[PM_SUSPEND_MAX] = {
-	[PM_SUSPEND_STANDBY]	= "standby",
-	[PM_SUSPEND_MEM]	= "mem",
+static struct power_op standby = {
+	.name = "standby",
+	.description = "Power-On Suspend ACPI State: S1",
+	.type = PM_SUSPEND_STANDBY,
+};
+static struct power_op mem = {
+	.name = "mem   ",
+	.description = "Suspend-to-RAM ACPI State: S3",
+	.type = PM_SUSPEND_MEM,
+};
 #ifdef CONFIG_SOFTWARE_SUSPEND
-	[PM_SUSPEND_DISK]	= "disk",
-#endif
+static struct power_op disk = {
+	.name = "disk  ",
+	.description = "Suspend-to-disk ACPI State: S4",
+	.type = PM_SUSPEND_DISK,
 };
+#endif
 
-static inline int valid_state(suspend_state_t state)
+/*
+ *
+ */
+static int pm_change_state(struct power_op *state)
+{
+	int error = -EINVAL;
+	int len = strlen(state->name);
+	struct power_op *this, *next;
+	struct list_head *head = &pm_states.list;
+
+	/*
+	 * list_find new operating point.
+	 * compare to current operating point.
+	 * if different change to new operating point.
+	 */
+	list_for_each_entry_safe(this, next, head, list) {
+		if (strncmp(state->name, this->name, len) == 0) {
+			if ((strcmp(current_state->name, this->name)) == 0) {
+				return 0;
+			}
+
+			if (this->prepare_transition(current_state, this)) {
+				break;
+			}
+
+			if (this->transition(current_state, this)) {
+				break;
+			}
+
+			/*
+			 * now lets wait for the transition latency
+			 */
+			udelay(this->latency);
+
+			error = this->finish_transition(current_state, this);
+
+			if (error == 0)
+				current_state = this;
+			break;
+		}
+	}
+	return error;
+}
+
+static inline int valid_state(struct power_op * state)
 {
 	/* Suspend-to-disk does not really need low-level support.
 	 * It can work with reboot if needed. */
-	if (state == PM_SUSPEND_DISK)
+	if (state->type == PM_SUSPEND_DISK)
 		return 1;
 
-	if (pm_ops && pm_ops->valid && !pm_ops->valid(state))
+	if (pm_ops && pm_ops->valid && !pm_ops->valid(state->type))
 		return 0;
 	return 1;
 }
@@ -168,7 +226,7 @@ static inline int valid_state(suspend_st
 
 /**
  *	enter_state - Do common work of entering low-power state.
- *	@state:		pm_state structure for state we're entering.
+ *	@state:		power_op structure for state we're entering.
  *
  *	Make sure we're the only ones trying to enter a sleep state. Fail
  *	if someone has beat us to it, since we don't want anything weird to
@@ -177,7 +235,7 @@ static inline int valid_state(suspend_st
  *	we've woken up).
  */
 
-static int enter_state(suspend_state_t state)
+static int enter_state(struct power_op *state)
 {
 	int error;
 
@@ -186,16 +244,21 @@ static int enter_state(suspend_state_t s
 	if (down_trylock(&pm_sem))
 		return -EBUSY;
 
-	if (state == PM_SUSPEND_DISK) {
+	if (state->type == PM_SUSPEND_DISK) {
 		error = pm_suspend_disk();
 		goto Unlock;
 	}
 
-	pr_debug("PM: Preparing system for %s sleep\n", pm_states[state]);
+	if (state->type == PM_FREQ_CHANGE) {
+		error = pm_change_state(state);
+		goto Unlock;
+	}
+
+	pr_debug("PM: Preparing system for %s sleep\n", state->name);
 	if ((error = suspend_prepare(state)))
 		goto Unlock;
 
-	pr_debug("PM: Entering %s sleep\n", pm_states[state]);
+	pr_debug("PM: Entering %s sleep\n", state->name);
 	error = suspend_enter(state);
 
 	pr_debug("PM: Finishing wakeup.\n");
@@ -211,7 +274,15 @@ static int enter_state(suspend_state_t s
  */
 int software_suspend(void)
 {
-	return enter_state(PM_SUSPEND_DISK);
+	struct power_op *this, *next;
+	struct list_head *head = &pm_states.list;
+	int error = 0;
+
+	list_for_each_entry_safe(this, next, head, list) {
+		if (this->type == PM_SUSPEND_DISK)
+			error= enter_state(this);
+	}
+	return error;
 }
 
 
@@ -223,16 +294,48 @@ int software_suspend(void)
  *	structure, and enter (above).
  */
 
-int pm_suspend(suspend_state_t state)
+int pm_suspend(struct power_op * state)
 {
-	if (state > PM_SUSPEND_ON && state <= PM_SUSPEND_MAX)
+	if (state->type > PM_SUSPEND_ON && state->type <= PM_SUSPEND_MAX)
 		return enter_state(state);
 	return -EINVAL;
 }
 
+decl_subsys(power,NULL,NULL);
 
+/**
+ *	supported_states - control system power state.
+ *
+ *	show() returns what states are supported, which are no longer
+ * 	hard-coded to just 'standby' (Power-On Suspend), 'mem' (Suspend-to-RAM),
+ *	and *'disk' (Suspend-to-Disk), but show all the power states.
+ *
+ *	store() unwritable
+ */
 
-decl_subsys(power,NULL,NULL);
+static ssize_t supported_states_show(struct subsystem * subsys, char * buf)
+{
+	struct power_op *this, *next;
+	struct list_head *head = &pm_states.list;
+	const char *header =  "< Name >    <Frequency>  <Voltage>  <Transition Latency>  < Description >\n";
+	char * s = buf;
+
+	s += sprintf(s, "%s", header);
+	list_for_each_entry_safe(this, next, head, list) {
+		s += sprintf(s,"%s %dKHz %dmV %dus %s\n", this->name,
+		   this->frequency, this->voltage, this->latency,
+		   this->description);
+	}
+
+	return (s - buf);
+}
+
+static ssize_t supported_states_store(struct subsystem * subsys, const char *buf, size_t n)
+{
+	return -EINVAL;
+}
+
+power_attr(supported_states);
 
 
 /**
@@ -248,36 +351,28 @@ decl_subsys(power,NULL,NULL);
 
 static ssize_t state_show(struct subsystem * subsys, char * buf)
 {
-	int i;
 	char * s = buf;
 
-	for (i = 0; i < PM_SUSPEND_MAX; i++) {
-		if (pm_states[i] && valid_state(i))
-			s += sprintf(s,"%s ", pm_states[i]);
-	}
-	s += sprintf(s,"\n");
+	s += sprintf(s,"%s\n", current_state->name);
 	return (s - buf);
 }
 
 static ssize_t state_store(struct subsystem * subsys, const char * buf, size_t n)
 {
-	suspend_state_t state = PM_SUSPEND_STANDBY;
-	const char * const *s;
+	struct power_op *this, *next;
+	struct list_head *head = &pm_states.list;
 	char *p;
-	int error;
+	int error = -EINVAL;
 	int len;
 
 	p = memchr(buf, '\n', n);
 	len = p ? p - buf : n;
-
-	for (s = &pm_states[state]; state < PM_SUSPEND_MAX; s++, state++) {
-		if (*s && !strncmp(buf, *s, len))
+	list_for_each_entry_safe(this, next, head, list) {
+		if (!strncmp(buf, this->name, len)) {
+			error = enter_state(this);
 			break;
+		}
 	}
-	if (state < PM_SUSPEND_MAX && *s)
-		error = enter_state(state);
-	else
-		error = -EINVAL;
 	return error ? error : n;
 }
 
@@ -285,6 +380,7 @@ power_attr(state);
 
 static struct attribute * g[] = {
 	&state_attr.attr,
+	&supported_states_attr.attr,
 	NULL,
 };
 
@@ -295,9 +391,20 @@ static struct attribute_group attr_group
 
 static int __init pm_init(void)
 {
+
 	int error = subsystem_register(&power_subsys);
 	if (!error)
 		error = sysfs_create_group(&power_subsys.kset.kobj,&attr_group);
+
+	INIT_LIST_HEAD(&pm_states.list);
+
+#ifdef CONFIG_SOFTWARE_SUSPEND
+	list_add(&disk.list, &pm_states.list);
+#endif
+	list_add(&mem.list, &pm_states.list);
+	list_add(&standby.list, &pm_states.list);
+	current_state = &pm_states;
+
 	return error;
 }
 
Index: linux-2.6.17/include/linux/pm.h
===================================================================
--- linux-2.6.17.orig/include/linux/pm.h
+++ linux-2.6.17/include/linux/pm.h
@@ -108,7 +108,29 @@ typedef int __bitwise suspend_state_t;
 #define PM_SUSPEND_STANDBY	((__force suspend_state_t) 1)
 #define PM_SUSPEND_MEM		((__force suspend_state_t) 3)
 #define PM_SUSPEND_DISK		((__force suspend_state_t) 4)
-#define PM_SUSPEND_MAX		((__force suspend_state_t) 5)
+#define PM_FREQ_CHANGE		((__force suspend_state_t) 5)
+#define PM_VOLT_CHANGE		((__force suspend_state_t) 6)
+#define PM_SUSPEND_MAX		((__force suspend_state_t) 7)
+
+#define PM_NAME_SIZE            16
+#define PM_DESCRIPTION_SIZE     48
+
+struct power_op {
+	struct list_head list;
+	suspend_state_t type;
+	char name[PM_NAME_SIZE];
+	char description[PM_DESCRIPTION_SIZE];
+	unsigned int frequency;		/* in KHz */
+	unsigned int voltage;		/* mV */
+	unsigned int latency;		/* transition latency in us */
+	int     (*prepare_transition)(struct power_op *cur, struct power_op *new);
+	int     (*transition)(struct power_op *cur, struct power_op *new);
+	int     (*finish_transition)(struct power_op *cur, struct power_op *new);
+
+	void *md_data;			/* arch dependent data (dpm_opt) */
+};
+extern struct power_op pm_states;
+extern struct power_op *current_state;
 
 typedef int __bitwise suspend_disk_method_t;
 
@@ -128,7 +150,7 @@ struct pm_ops {
 
 extern void pm_set_ops(struct pm_ops *);
 extern struct pm_ops *pm_ops;
-extern int pm_suspend(suspend_state_t state);
+extern int pm_suspend(struct power_op *state);
 
 
 /*
Index: linux-2.6.17/kernel/power/power.h
===================================================================
--- linux-2.6.17.orig/kernel/power/power.h
+++ linux-2.6.17/kernel/power/power.h
@@ -113,4 +113,4 @@ extern int swsusp_resume(void);
 extern int swsusp_read(void);
 extern int swsusp_write(void);
 extern void swsusp_close(void);
-extern int suspend_enter(suspend_state_t state);
+extern int suspend_enter(struct power_op * state);
Index: linux-2.6.17/Makefile
===================================================================
--- linux-2.6.17.orig/Makefile
+++ linux-2.6.17/Makefile
@@ -1,7 +1,7 @@
 VERSION = 2
 PATCHLEVEL = 6
 SUBLEVEL = 18
-EXTRAVERSION = -rc1
+EXTRAVERSION = -rc1-PowerOp
 NAME=Crazed Snow-Weasel
 
 # *DOCUMENTATION*
Index: linux-2.6.17/arch/i386/kernel/cpu/cpufreq/speedstep-centrino.c
===================================================================
--- linux-2.6.17.orig/arch/i386/kernel/cpu/cpufreq/speedstep-centrino.c
+++ linux-2.6.17/arch/i386/kernel/cpu/cpufreq/speedstep-centrino.c
@@ -20,6 +20,7 @@
 #include <linux/sched.h>	/* current */
 #include <linux/delay.h>
 #include <linux/compiler.h>
+#include <linux/pm.h>
 
 #ifdef CONFIG_X86_SPEEDSTEP_CENTRINO_ACPI
 #include <linux/acpi.h>
@@ -160,6 +161,80 @@ static struct cpufreq_frequency_table ba
 	OP(1400, 1484),
 	{ .frequency = CPUFREQ_TABLE_END }
 };
+#ifdef CONFIG_PM
+/*
+ * this is the formula for OP.  We need the perfctl
+ * msr value to change to a new frequency.
+ * We'll save it in the machine dependent variable md_data.
+ */
+static void set_perfctl_msr(struct power_op *pm)
+{
+	unsigned int msr = pm->frequency/100;
+	unsigned int v = pm->voltage - 700;
+
+	msr = (unsigned int)(msr << 8);
+	msr |= (unsigned int)(v / 16);
+	pm->md_data = (void *)msr;
+}
+
+static int centrino_transition(struct power_op *cur, struct power_op *new);
+
+static struct power_op mhz600 = {
+        .name = "600MHz",
+        .description = "Low Frequency state",
+        .type = PM_FREQ_CHANGE,
+        .frequency = 600,
+        .voltage = 956,
+	.latency = 100,
+	.prepare_transition  = cpufreq_prepare_transition,
+	.transition = centrino_transition,
+	.finish_transition = cpufreq_finish_transition,
+};
+static struct power_op mhz800 = {
+        .name = "800MHz",
+        .description = "Lower Frequency state",
+        .type = PM_FREQ_CHANGE,
+        .frequency = 800,
+        .voltage = 1180,
+	.latency = 100,
+	.prepare_transition  = cpufreq_prepare_transition,
+	.transition = centrino_transition,
+	.finish_transition = cpufreq_finish_transition,
+};
+static struct power_op ghz1 = {
+        .name = "1GHz",
+        .description = "Med Frequency state",
+        .type = PM_FREQ_CHANGE,
+        .frequency = 1000,
+        .voltage = 1308,
+	.latency = 100,
+	.prepare_transition  = cpufreq_prepare_transition,
+	.transition = centrino_transition,
+	.finish_transition = cpufreq_finish_transition,
+};
+static struct power_op ghz12 = {
+        .name = "1.2GHz",
+        .description = "High Frequency state",
+        .type = PM_FREQ_CHANGE,
+        .frequency = 1200,
+        .voltage = 1436,
+	.latency = 100,
+	.prepare_transition  = cpufreq_prepare_transition,
+	.transition = centrino_transition,
+	.finish_transition = cpufreq_finish_transition,
+};
+static struct power_op ghz14 = {
+        .name = "1.4GHz",
+        .description = "Highest Frequency state",
+        .type = PM_FREQ_CHANGE,
+        .frequency = 1400,
+        .voltage = 1484,
+	.latency = 100,
+	.prepare_transition  = cpufreq_prepare_transition,
+	.transition = centrino_transition,
+	.finish_transition = cpufreq_finish_transition,
+};
+#endif
 
 /* Intel Pentium M processor 1.50GHz (Banias) */
 static struct cpufreq_frequency_table banias_1500[] =
@@ -266,6 +341,19 @@ static int centrino_cpu_init_table(struc
 	dprintk("found \"%s\": max frequency: %dkHz\n",
 	       model->model_name, model->max_freq);
 
+#ifdef CONFIG_PM
+	list_add_tail(&mhz600.list, &pm_states.list);
+	set_perfctl_msr(&mhz600);
+	list_add_tail(&mhz800.list, &pm_states.list);
+	set_perfctl_msr(&mhz800);
+	list_add_tail(&ghz1.list, &pm_states.list);
+	set_perfctl_msr(&ghz1);
+	list_add_tail(&ghz12.list, &pm_states.list);
+	set_perfctl_msr(&ghz12);
+	list_add_tail(&ghz14.list, &pm_states.list);
+	set_perfctl_msr(&ghz14);
+	current_state = &ghz14;
+#endif
 	return 0;
 }
 
@@ -620,6 +708,26 @@ static int centrino_verify (struct cpufr
 	return cpufreq_frequency_table_verify(policy, centrino_model[policy->cpu]->op_points);
 }
 
+static int centrino_transition(struct power_op *cur, struct power_op *new)
+{
+	unsigned int msr, oldmsr = 0, h = 0;
+
+	if (cur == new)
+		return 0;
+
+	msr = (unsigned int)new->md_data;
+	rdmsr(MSR_IA32_PERF_CTL, oldmsr, h);
+
+	/* all but 16 LSB are reserved, treat them with care */
+	oldmsr &= ~0xffff;
+	msr &= 0xffff;
+	oldmsr |= msr;
+
+	wrmsr(MSR_IA32_PERF_CTL, oldmsr, h);
+
+	return 0;
+}
+
 /**
  * centrino_setpolicy - set a new CPUFreq policy
  * @policy: new policy
Index: linux-2.6.17/drivers/cpufreq/cpufreq.c
===================================================================
--- linux-2.6.17.orig/drivers/cpufreq/cpufreq.c
+++ linux-2.6.17/drivers/cpufreq/cpufreq.c
@@ -226,6 +226,33 @@ static void adjust_jiffies(unsigned long
 static inline void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci) { return; }
 #endif
 
+int cpufreq_prepare_transition(struct power_op *cur, struct power_op *new)
+{
+	struct cpufreq_freqs freqs;
+
+	freqs.old = cur->frequency;
+	freqs.new = new->frequency;
+	freqs.cpu = 0;
+	freqs.flags = cpufreq_driver->flags;
+	blocking_notifier_call_chain(&cpufreq_transition_notifier_list,
+			CPUFREQ_PRECHANGE, &freqs);
+	adjust_jiffies(CPUFREQ_PRECHANGE, &freqs);
+	return 0;
+}
+
+int cpufreq_finish_transition(struct power_op *cur, struct power_op *new)
+{
+	struct cpufreq_freqs freqs;
+
+	freqs.old = cur->frequency;
+	freqs.new = new->frequency;
+	freqs.cpu = 0;
+	freqs.flags = cpufreq_driver->flags;
+	adjust_jiffies(CPUFREQ_POSTCHANGE, &freqs);
+	blocking_notifier_call_chain(&cpufreq_transition_notifier_list,
+			CPUFREQ_POSTCHANGE, &freqs);
+	return 0;
+}
 
 /**
  * cpufreq_notify_transition - call notifier chain and adjust_jiffies
@@ -884,6 +911,12 @@ static void cpufreq_out_of_sync(unsigned
 }
 
 
+#ifdef CONFIG_PM
+unsigned int cpufreq_quick_get(unsigned int cpu)
+{
+	return (current_state->frequency * 1000);
+}
+#else
 /**
  * cpufreq_quick_get - get the CPU frequency (in kHz) frpm policy->cur
  * @cpu: CPU number
@@ -905,6 +938,7 @@ unsigned int cpufreq_quick_get(unsigned 
 
 	return (ret);
 }
+#endif
 EXPORT_SYMBOL(cpufreq_quick_get);
 
 
Index: linux-2.6.17/include/linux/cpufreq.h
===================================================================
--- linux-2.6.17.orig/include/linux/cpufreq.h
+++ linux-2.6.17/include/linux/cpufreq.h
@@ -271,6 +271,8 @@ static inline unsigned int cpufreq_quick
 	return 0;
 }
 #endif
+int cpufreq_prepare_transition(struct power_op *cur, struct power_op *new);
+int cpufreq_finish_transition(struct power_op *cur, struct power_op *new);
 
 
 /*********************************************************************
Index: linux-2.6.17/Documentation/power/powerop.txt
===================================================================
--- /dev/null
+++ linux-2.6.17/Documentation/power/powerop.txt
@@ -0,0 +1,168 @@
+
+The PowerOp Power Management infrastructure.
+
+David Singleton <dsingleton@mvista.com>
+
+25 July 2006
+
+Copyright (c) 2006 MontaVista Software Inc.
+
+0. Introduction
+
+The goal of PowerOp power management is to provide a framework that unifies
+and simplifies the various power management infrastructures in Linux.  The
+three infrastructures Power Op is concerned with are:
+
+	1) basic suspend/resume power management (CONFIG_PM)
+
+	2) basic processor frequency management (CONFIG_CPUFREQ)
+
+	3) SourceForge's Dynamic Power Management (CONFIG_DPM)
+
+All three power management infrastructures are concerned with controlling
+power states of the system, and interestingly enough they all perform the
+same basic operational steps to control changes in power state.
+
+PowerOp uses the existing power management sysfs infrastructure and extends it
+to perform cpufreq and dynamic power management operations. The traditional
+suspend to memory or disk (or swap) infrastructure has the correct operational
+structure that supports all types of power state change.
+
+The CPUFREQ table based frequency control makes controlling cpu frequency
+simple and straight forward.  The user doesn't get to set the cpu to
+any speed, but only to supported speeds that have been provided by
+the hardware vendor and validated.
+
+Dynamic Power Management treats all types of power states as operating points,
+wether it's a suspend operating point, a particular frequency, or a specific
+voltage.
+
+By combining the best of all of these power management infrastructures
+PowerOp uses the operational structure of tradition CONFIG_PM power
+management and converts all power states, frequency, voltage, idle or
+suspend to the CPUFREQ concept of only supported and validated operating
+points.
+
+PowerOp then becomes a simplified power management infrastructure in that
+only operating points that are supported and validated are available
+to the user.  Control of all operating points are done by the operating
+point name.  The user cannot supply invalid, or malicious,
+parameters that would hang or crash the system.
+
+1) PowerOp interface.
+
+To simplify power management all operations take place through two sysfs
+files, /sys/power/state and /sys/power/supported_states.  The 'state' file
+shows the current operating point of the system.  The readonly
+'supported_states' file shows the operating points the system supports.
+
+Supported operating points are displayed in tuple format of:
+
+<name, frequency, voltage, transition latency, description>
+
+The supported_states file contains rows of tuples with each
+tuple describing a supported operating point of the system.
+The supported_states file looks like a merge between the old
+/sys/power/state file and a cpufreq table.
+
+The system can transition to any of the supported states by simply
+storing the operating point name in the /sys/power/state file.
+
+To allow user space notification of events, like low battery, lid of
+the notebook being closed, etc.  PowerOp notifies the user through
+the hotplug interface.
+
+
+2) PowerOP Operating Points.
+
+An operating point is represented by the power_op struct which contains:
+
+struct power_op {
+        struct list_head list;
+        suspend_state_t type;
+        char name[PM_NAME_SIZE];
+        char description[PM_DESCRIPTION_SIZE];
+        unsigned int frequency;         /* in KHz */
+        unsigned int voltage;           /* mV */
+        unsigned int latency;           /* transition latency in us */
+        int     (*prepare_transition)(struct power_op *cur, struct power_op *new);
+        int     (*transition)(struct power_op *cur, struct power_op *new);
+        int     (*finish_transition)(struct power_op *cur, struct power_op *new);
+
+        void *md_data;                  /* arch dependent data  */
+};
+
+Each operating point has its own functions for preparing to transition,
+transitioning and finishing transition.  Cpu frequency operating points
+will probably share their op vectors, idle and suspend operating points my have
+different op vectors.
+
+
+3) Traditional Operation of Power Management Code.
+
+All three power management infrastructures have the same operational model.
+All three follow the PM model of preparing to suspend,  suspending,
+and finish the state change.  It was easiest to follow the model
+enforced by the traditional power management and use the three step process of:
+
+ 	1) get ready to change state
+	2) change state
+	3) finish changes
+
+Cpufreq infrastructure makes three calls to change the frequency of the
+processor:
+
+	1) cpufreq_notify_transition(&freq, CPUFREQ_PRECHANGE);
+
+	2) acpi_processor_set_performance (data, j, next_state);
+
+	3) cpufreq_notify_transition(&freq, CPUFREQ_POSTCHANGE);
+
+DPM uses these three calls to change frequency and/or voltage:
+
+	1) dpm_driver_scale(SCALE_PRECHANGE, new);
+
+	2) clk_set_rate(prcm_set, new->md_opt.prcm_clock);
+
+	3) dpm_driver_scale(SCALE_POStCHANGE, new);
+
+PM uses these three calls to suspend:
+
+	1) suspend_prepare(state);
+
+	2) suspend_enter(state->type);
+
+	3) suspend_finish(state);
+
+
+4) PowerOP Operation.
+
+PowerOP uses the following three calls to transition to a new operating
+point.
+
+	prepare_to_transition(cur_state, new_state);
+
+	transition(cur_state, new_state);
+
+	finish_transistion(cur_state, new_state);
+
+The parameters are pointers to operating point structures, struct power_op.
+
+Power OP is a simplified version of all three of these infrastructures in
+that it only deals with operating points, and more specifically with
+supported operating points.  Power Op presents a set of supported operating
+points to the user.  This is similar to the cpufreq table concept in that
+only supported and validated frequencies are avaliable.
+
+The definition of the operating point is done in a manner similar to cpufreqs
+in that the supported operating frequency, voltage and transition latency,
+are predefined (by the hardware vendor) and validated.
+
+The user maninuplates the operting points of the system by the
+name of the operating points.  This simplifies both the code and the
+control of the system's operating points in the PowerOp daemon.
+
+All supported operating points are defined at compile time and
+the user sets the system to different operating points by
+the operating point name.
+

[-- Attachment #3: Type: text/plain, Size: 0 bytes --]