[RFC PATCH] cpuidle: allow per cpu latencies

[RFC PATCH] cpuidle: allow per cpu latencies

[Peter De Schrijver]

On some systems (eg Tegra30) latencies for a given C state are not equal for
all CPUs. Therefore we introduce a new per CPU structure which contains
those parameters.

--

This patch doesn't update all cpuidle device registrations. I will do that
in a next version after agreement on the exact data structure to be
introduced.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
---
 arch/arm/mach-tegra/cpuidle.c      |   15 +++++++++++----
 drivers/cpuidle/cpuidle.c          |   27 ++++++++++++++++++++++++---
 drivers/cpuidle/driver.c           |   25 -------------------------
 drivers/cpuidle/governors/ladder.c |   21 ++++++++++++---------
 drivers/cpuidle/governors/menu.c   |    7 ++++---
 drivers/cpuidle/sysfs.c            |   34 +++++++++++++++++++++++-----------
 include/linux/cpuidle.h            |   17 ++++++++++-------
 7 files changed, 84 insertions(+), 62 deletions(-)

diff --git a/arch/arm/mach-tegra/cpuidle.c b/arch/arm/mach-tegra/cpuidle.c
index d83a8c0..b8fdf02 100644
--- a/arch/arm/mach-tegra/cpuidle.c
+++ b/arch/arm/mach-tegra/cpuidle.c
@@ -41,14 +41,20 @@ struct cpuidle_driver tegra_idle_driver = {
 	.states = {
 		[0] = {
 			.enter			= tegra_idle_enter_lp3,
-			.exit_latency		= 10,
-			.target_residency	= 10,
-			.power_usage		= 600,
-			.flags			= CPUIDLE_FLAG_TIME_VALID,
 			.name			= "LP3",
 			.desc			= "CPU flow-controlled",
 		},
 	},
+	.power_specified = 1,
+};
+
+static struct cpuidle_state_parameters __initdata tegra_idle_parameters[] = {
+	[0] = {
+		.exit_latency		= 10,
+		.target_residency	= 10,
+		.power_usage		= 600,
+		.flags			= CPUIDLE_FLAG_TIME_VALID,
+	},
 };
 
 static DEFINE_PER_CPU(struct cpuidle_device, tegra_idle_device);
@@ -95,6 +101,7 @@ static int __init tegra_cpuidle_init(void)
 		dev->cpu = cpu;
 
 		dev->state_count = drv->state_count;
+		dev->state_parameters[0] =  tegra_idle_parameters[0];
 		ret = cpuidle_register_device(dev);
 		if (ret) {
 			pr_err("CPU%u: CPUidle device registration failed\n",
diff --git a/drivers/cpuidle/cpuidle.c b/drivers/cpuidle/cpuidle.c
index 87411ce..b992d00 100644
--- a/drivers/cpuidle/cpuidle.c
+++ b/drivers/cpuidle/cpuidle.c
@@ -87,8 +87,9 @@ int cpuidle_play_dead(void)
 	for (i = CPUIDLE_DRIVER_STATE_START; i < drv->state_count; i++) {
 		struct cpuidle_state *s = &drv->states[i];
 
-		if (s->power_usage < power_usage && s->enter_dead) {
-			power_usage = s->power_usage;
+		if (dev->state_parameters[i].power_usage < power_usage
+				&& s->enter_dead) {
+			power_usage = dev->state_parameters[i].power_usage;
 			dead_state = i;
 		}
 	}
@@ -371,7 +372,7 @@ EXPORT_SYMBOL_GPL(cpuidle_disable_device);
  */
 static int __cpuidle_register_device(struct cpuidle_device *dev)
 {
-	int ret;
+	int ret, i;
 	struct device *cpu_dev = get_cpu_device((unsigned long)dev->cpu);
 	struct cpuidle_driver *cpuidle_driver = cpuidle_get_driver();
 
@@ -389,6 +390,26 @@ static int __cpuidle_register_device(struct cpuidle_device *dev)
 		return ret;
 	}
 
+	/*
+	 * cpuidle driver should set the drv->power_specified bit
+	 * before registering if the driver provides
+	 * power_usage numbers.
+	 *
+	 * If power_specified is not set,
+	 * we fill in power_usage with decreasing values as the
+	 * cpuidle code has an implicit assumption that state Cn
+	 * uses less power than C(n-1).
+	 *
+	 * With CONFIG_ARCH_HAS_CPU_RELAX, C0 is already assigned
+	 * an power value of -1.  So we use -2, -3, etc, for other
+	 * c-states.
+	 */
+	if (!cpuidle_driver->power_specified) {
+		for (i = CPUIDLE_DRIVER_STATE_START;
+				i < cpuidle_driver->state_count; i++)
+			dev->state_parameters[i].power_usage = -1 - i;
+	}
+
 	dev->registered = 1;
 	return 0;
 }
diff --git a/drivers/cpuidle/driver.c b/drivers/cpuidle/driver.c
index 40cd3f3..d81c6db 100644
--- a/drivers/cpuidle/driver.c
+++ b/drivers/cpuidle/driver.c
@@ -17,30 +17,6 @@
 static struct cpuidle_driver *cpuidle_curr_driver;
 DEFINE_SPINLOCK(cpuidle_driver_lock);
 
-static void __cpuidle_register_driver(struct cpuidle_driver *drv)
-{
-	int i;
-	/*
-	 * cpuidle driver should set the drv->power_specified bit
-	 * before registering if the driver provides
-	 * power_usage numbers.
-	 *
-	 * If power_specified is not set,
-	 * we fill in power_usage with decreasing values as the
-	 * cpuidle code has an implicit assumption that state Cn
-	 * uses less power than C(n-1).
-	 *
-	 * With CONFIG_ARCH_HAS_CPU_RELAX, C0 is already assigned
-	 * an power value of -1.  So we use -2, -3, etc, for other
-	 * c-states.
-	 */
-	if (!drv->power_specified) {
-		for (i = CPUIDLE_DRIVER_STATE_START; i < drv->state_count; i++)
-			drv->states[i].power_usage = -1 - i;
-	}
-}
-
-
 /**
  * cpuidle_register_driver - registers a driver
  * @drv: the driver
@@ -58,7 +34,6 @@ int cpuidle_register_driver(struct cpuidle_driver *drv)
 		spin_unlock(&cpuidle_driver_lock);
 		return -EBUSY;
 	}
-	__cpuidle_register_driver(drv);
 	cpuidle_curr_driver = drv;
 	spin_unlock(&cpuidle_driver_lock);
 
diff --git a/drivers/cpuidle/governors/ladder.c b/drivers/cpuidle/governors/ladder.c
index b6a09ea..2e7ea8c 100644
--- a/drivers/cpuidle/governors/ladder.c
+++ b/drivers/cpuidle/governors/ladder.c
@@ -79,9 +79,9 @@ static int ladder_select_state(struct cpuidle_driver *drv,
 
 	last_state = &ldev->states[last_idx];
 
-	if (drv->states[last_idx].flags & CPUIDLE_FLAG_TIME_VALID) {
+	if (dev->state_parameters[last_idx].flags & CPUIDLE_FLAG_TIME_VALID) {
 		last_residency = cpuidle_get_last_residency(dev) - \
-					 drv->states[last_idx].exit_latency;
+				 dev->state_parameters[last_idx].exit_latency;
 	}
 	else
 		last_residency = last_state->threshold.promotion_time + 1;
@@ -89,7 +89,7 @@ static int ladder_select_state(struct cpuidle_driver *drv,
 	/* consider promotion */
 	if (last_idx < drv->state_count - 1 &&
 	    last_residency > last_state->threshold.promotion_time &&
-	    drv->states[last_idx + 1].exit_latency <= latency_req) {
+	    dev->state_parameters[last_idx + 1].exit_latency <= latency_req) {
 		last_state->stats.promotion_count++;
 		last_state->stats.demotion_count = 0;
 		if (last_state->stats.promotion_count >= last_state->threshold.promotion_count) {
@@ -100,11 +100,12 @@ static int ladder_select_state(struct cpuidle_driver *drv,
 
 	/* consider demotion */
 	if (last_idx > CPUIDLE_DRIVER_STATE_START &&
-	    drv->states[last_idx].exit_latency > latency_req) {
+	    dev->state_parameters[last_idx].exit_latency > latency_req) {
 		int i;
 
 		for (i = last_idx - 1; i > CPUIDLE_DRIVER_STATE_START; i--) {
-			if (drv->states[i].exit_latency <= latency_req)
+			if (dev->state_parameters[i].exit_latency <=
+					latency_req)
 				break;
 		}
 		ladder_do_selection(ldev, last_idx, i);
@@ -136,12 +137,12 @@ static int ladder_enable_device(struct cpuidle_driver *drv,
 	int i;
 	struct ladder_device *ldev = &per_cpu(ladder_devices, dev->cpu);
 	struct ladder_device_state *lstate;
-	struct cpuidle_state *state;
+	struct cpuidle_state_parameters *state_parameters;
 
 	ldev->last_state_idx = CPUIDLE_DRIVER_STATE_START;
 
 	for (i = 0; i < drv->state_count; i++) {
-		state = &drv->states[i];
+		state_parameters = &dev->state_parameters[i];
 		lstate = &ldev->states[i];
 
 		lstate->stats.promotion_count = 0;
@@ -151,9 +152,11 @@ static int ladder_enable_device(struct cpuidle_driver *drv,
 		lstate->threshold.demotion_count = DEMOTION_COUNT;
 
 		if (i < drv->state_count - 1)
-			lstate->threshold.promotion_time = state->exit_latency;
+			lstate->threshold.promotion_time =
+				state_parameters->exit_latency;
 		if (i > 0)
-			lstate->threshold.demotion_time = state->exit_latency;
+			lstate->threshold.demotion_time =
+				state_parameters->exit_latency;
 	}
 
 	return 0;
diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c
index 0633575..e9bbc20 100644
--- a/drivers/cpuidle/governors/menu.c
+++ b/drivers/cpuidle/governors/menu.c
@@ -281,7 +281,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 	 * unless the timer is happening really really soon.
 	 */
 	if (data->expected_us > 5 &&
-		drv->states[CPUIDLE_DRIVER_STATE_START].disable == 0)
+		dev->state_parameters[CPUIDLE_DRIVER_STATE_START].disable == 0)
 		data->last_state_idx = CPUIDLE_DRIVER_STATE_START;
 
 	/*
@@ -289,7 +289,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 	 * our constraints.
 	 */
 	for (i = CPUIDLE_DRIVER_STATE_START; i < drv->state_count; i++) {
-		struct cpuidle_state *s = &drv->states[i];
+		struct cpuidle_state_parameters *s = &dev->state_parameters[i];
 
 		if (s->disable)
 			continue;
@@ -336,7 +336,8 @@ static void menu_update(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 	struct menu_device *data = &__get_cpu_var(menu_devices);
 	int last_idx = data->last_state_idx;
 	unsigned int last_idle_us = cpuidle_get_last_residency(dev);
-	struct cpuidle_state *target = &drv->states[last_idx];
+	struct cpuidle_state_parameters *target =
+			&dev->state_parameters[last_idx];
 	unsigned int measured_us;
 	u64 new_factor;
 
diff --git a/drivers/cpuidle/sysfs.c b/drivers/cpuidle/sysfs.c
index 88032b4..7b17413 100644
--- a/drivers/cpuidle/sysfs.c
+++ b/drivers/cpuidle/sysfs.c
@@ -215,9 +215,10 @@ static struct kobj_type ktype_cpuidle = {
 
 struct cpuidle_state_attr {
 	struct attribute attr;
-	ssize_t (*show)(struct cpuidle_state *, \
-					struct cpuidle_state_usage *, char *);
-	ssize_t (*store)(struct cpuidle_state *, const char *, size_t);
+	ssize_t (*show)(struct cpuidle_state *, struct cpuidle_state_usage *,
+				struct cpuidle_state_parameters *, char *);
+	ssize_t (*store)(struct cpuidle_state_parameters *, const char *,
+				size_t);
 };
 
 #define define_one_state_ro(_name, show) \
@@ -228,13 +229,15 @@ static struct cpuidle_state_attr attr_##_name = __ATTR(_name, 0644, show, store)
 
 #define define_show_state_function(_name) \
 static ssize_t show_state_##_name(struct cpuidle_state *state, \
-			 struct cpuidle_state_usage *state_usage, char *buf) \
+			 struct cpuidle_state_usage *state_usage, \
+			 struct cpuidle_state_parameters *state_parameters, \
+			 char *buf) \
 { \
-	return sprintf(buf, "%u\n", state->_name);\
+	return sprintf(buf, "%u\n", state_parameters->_name);\
 }
 
 #define define_store_state_function(_name) \
-static ssize_t store_state_##_name(struct cpuidle_state *state, \
+static ssize_t store_state_##_name(struct cpuidle_state_parameters *state, \
 		const char *buf, size_t size) \
 { \
 	long value; \
@@ -253,14 +256,18 @@ static ssize_t store_state_##_name(struct cpuidle_state *state, \
 
 #define define_show_state_ull_function(_name) \
 static ssize_t show_state_##_name(struct cpuidle_state *state, \
-			struct cpuidle_state_usage *state_usage, char *buf) \
+			struct cpuidle_state_usage *state_usage, \
+			struct cpuidle_state_parameters *state_parameters, \
+			char *buf) \
 { \
 	return sprintf(buf, "%llu\n", state_usage->_name);\
 }
 
 #define define_show_state_str_function(_name) \
 static ssize_t show_state_##_name(struct cpuidle_state *state, \
-			struct cpuidle_state_usage *state_usage, char *buf) \
+			struct cpuidle_state_usage *state_usage, \
+			struct cpuidle_state_parameters *state_parameters, \
+			char *buf) \
 { \
 	if (state->_name[0] == '\0')\
 		return sprintf(buf, "<null>\n");\
@@ -298,6 +305,7 @@ static struct attribute *cpuidle_state_default_attrs[] = {
 #define kobj_to_state_obj(k) container_of(k, struct cpuidle_state_kobj, kobj)
 #define kobj_to_state(k) (kobj_to_state_obj(k)->state)
 #define kobj_to_state_usage(k) (kobj_to_state_obj(k)->state_usage)
+#define kobj_to_state_parameters(k) (kobj_to_state_obj(k)->state_parameters)
 #define attr_to_stateattr(a) container_of(a, struct cpuidle_state_attr, attr)
 static ssize_t cpuidle_state_show(struct kobject * kobj,
 	struct attribute * attr ,char * buf)
@@ -305,10 +313,12 @@ static ssize_t cpuidle_state_show(struct kobject * kobj,
 	int ret = -EIO;
 	struct cpuidle_state *state = kobj_to_state(kobj);
 	struct cpuidle_state_usage *state_usage = kobj_to_state_usage(kobj);
+	struct cpuidle_state_parameters *state_parameters =
+				kobj_to_state_parameters(kobj);
 	struct cpuidle_state_attr * cattr = attr_to_stateattr(attr);
 
 	if (cattr->show)
-		ret = cattr->show(state, state_usage, buf);
+		ret = cattr->show(state, state_usage, state_parameters, buf);
 
 	return ret;
 }
@@ -317,11 +327,12 @@ static ssize_t cpuidle_state_store(struct kobject *kobj,
 	struct attribute *attr, const char *buf, size_t size)
 {
 	int ret = -EIO;
-	struct cpuidle_state *state = kobj_to_state(kobj);
+	struct cpuidle_state_parameters *state_parameters =
+				kobj_to_state_parameters(kobj);
 	struct cpuidle_state_attr *cattr = attr_to_stateattr(attr);
 
 	if (cattr->store)
-		ret = cattr->store(state, buf, size);
+		ret = cattr->store(state_parameters, buf, size);
 
 	return ret;
 }
@@ -369,6 +380,7 @@ int cpuidle_add_state_sysfs(struct cpuidle_device *device)
 			goto error_state;
 		kobj->state = &drv->states[i];
 		kobj->state_usage = &device->states_usage[i];
+		kobj->state_parameters = &device->state_parameters[i];
 		init_completion(&kobj->kobj_unregister);
 
 		ret = kobject_init_and_add(&kobj->kobj, &ktype_state_cpuidle, &device->kobj,
diff --git a/include/linux/cpuidle.h b/include/linux/cpuidle.h
index 6c26a3d..cddbd34 100644
--- a/include/linux/cpuidle.h
+++ b/include/linux/cpuidle.h
@@ -42,12 +42,6 @@ struct cpuidle_state {
 	char		name[CPUIDLE_NAME_LEN];
 	char		desc[CPUIDLE_DESC_LEN];
 
-	unsigned int	flags;
-	unsigned int	exit_latency; /* in US */
-	int		power_usage; /* in mW */
-	unsigned int	target_residency; /* in US */
-	unsigned int    disable;
-
 	int (*enter)	(struct cpuidle_device *dev,
 			struct cpuidle_driver *drv,
 			int index);
@@ -55,6 +49,14 @@ struct cpuidle_state {
 	int (*enter_dead) (struct cpuidle_device *dev, int index);
 };
 
+struct cpuidle_state_parameters {
+	unsigned int	flags;
+	unsigned int	exit_latency; /* in US */
+	int		power_usage; /* in mW */
+	unsigned int	target_residency; /* in US */
+	unsigned int    disable;
+};
+
 /* Idle State Flags */
 #define CPUIDLE_FLAG_TIME_VALID	(0x01) /* is residency time measurable? */
 
@@ -83,6 +85,7 @@ cpuidle_set_statedata(struct cpuidle_state_usage *st_usage, void *data)
 struct cpuidle_state_kobj {
 	struct cpuidle_state *state;
 	struct cpuidle_state_usage *state_usage;
+	struct cpuidle_state_parameters *state_parameters;
 	struct completion kobj_unregister;
 	struct kobject kobj;
 };
@@ -96,7 +99,7 @@ struct cpuidle_device {
 	int			state_count;
 	struct cpuidle_state_usage	states_usage[CPUIDLE_STATE_MAX];
 	struct cpuidle_state_kobj *kobjs[CPUIDLE_STATE_MAX];
-
+	struct cpuidle_state_parameters state_parameters[CPUIDLE_STATE_MAX];
 	struct list_head 	device_list;
 	struct kobject		kobj;
 	struct completion	kobj_unregister;
-- 
1.7.7.rc0.72.g4b5ea.dirty

Re: [RFC PATCH] cpuidle: allow per cpu latencies

[Arjan van de Ven]

On 4/5/2012 2:53 AM, Peter De Schrijver wrote:
> This patch doesn't update all cpuidle device registrations. I will do that

question is if you want to do per cpu latencies, or if you want to have
both types of C state in one big table, and have each of the tegra cpyu
types pick half of them...

Re: [RFC PATCH] cpuidle: allow per cpu latencies

[Shilimkar, Santosh]

Peter,

On Thu, Apr 5, 2012 at 7:07 PM, Arjan van de Ven <arjan@linux.intel.com> wrote:
> On 4/5/2012 2:53 AM, Peter De Schrijver wrote:
>> This patch doesn't update all cpuidle device registrations. I will do that
>
> question is if you want to do per cpu latencies, or if you want to have
> both types of C state in one big table, and have each of the tegra cpyu
> types pick half of them...
>
>
Indeed !! That should work.
I thought the C-states are always per CPU based and during the
cpuidle registration you can register C-state accordingly based on the
specific CPU types with different latencies if needed.

Am I missing something ?

Regards
Santosh

Re: [RFC PATCH] cpuidle: allow per cpu latencies

[Daniel Lezcano]

On 04/06/2012 12:32 PM, Shilimkar, Santosh wrote:
> Peter,
>
> On Thu, Apr 5, 2012 at 7:07 PM, Arjan van de Ven<arjan@linux.intel.com>  wrote:
>> On 4/5/2012 2:53 AM, Peter De Schrijver wrote:
>>> This patch doesn't update all cpuidle device registrations. I will do that
>>
>> question is if you want to do per cpu latencies, or if you want to have
>> both types of C state in one big table, and have each of the tegra cpyu
>> types pick half of them...
>>
>>
> Indeed !! That should work.
> I thought the C-states are always per CPU based and during the
> cpuidle registration you can register C-state accordingly based on the
> specific CPU types with different latencies if needed.
>
> Am I missing something ?

That was the case before the cpuidle_state were moved from the 
cpuidle_device to the cpuidle_driver structure [1].

That had the benefit of using a single latencies array instead of 
multiple copy of the same array, which was the case until today.

I looked at the white paper for the tegra3 and understand this is no 
longer true because of the 4-plus-1 architecture [2].

With the increasing number of SoCs, we have a lot of new cpuidle drivers 
and each time we modify something in the cpuidle core, that impacts all 
the cpuidle drivers.

My feeling is we are going back and forth when patching the cpuidle core 
and may be it is time to define a clear semantic before patching again 
the cpuidle, no ?

What could nice is to have:

  * in case of the same latencies for all cpus, use a single array

  * in case of different latencies, group the same latencies into a 
single array (I assume this is the case for 4-plus-1, right ?)

May be we can move the cpuidle_state to a per_cpu pointer like 
cpuidle_devices in cpuidle.c and then add:

register_latencies(struct cpuidle_latencies l, int cpu);

If we have the same latencies for all the cpus, then we can register the 
same array, which is only a pointer.

Thanks
   -- Daniel


[1] https://lkml.org/lkml/2011/10/3/57
[2] 
http://www.nvidia.com/content/PDF/tegra_white_papers/Variable-SMP-A-Multi-Core-CPU-Architecture-for-Low-Power-and-High-Performance.pdf


-- 
  <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog

Re: [RFC PATCH] cpuidle: allow per cpu latencies

[Peter De Schrijver]

On Fri, Apr 06, 2012 at 05:35:46PM +0200, Daniel Lezcano wrote:
> On 04/06/2012 12:32 PM, Shilimkar, Santosh wrote:
> > Peter,
> >
> > On Thu, Apr 5, 2012 at 7:07 PM, Arjan van de Ven<arjan@linux.intel.com>  wrote:
> >> On 4/5/2012 2:53 AM, Peter De Schrijver wrote:
> >>> This patch doesn't update all cpuidle device registrations. I will do that
> >>
> >> question is if you want to do per cpu latencies, or if you want to have
> >> both types of C state in one big table, and have each of the tegra cpyu
> >> types pick half of them...
> >>
> >>
> > Indeed !! That should work.
> > I thought the C-states are always per CPU based and during the
> > cpuidle registration you can register C-state accordingly based on the
> > specific CPU types with different latencies if needed.
> >
> > Am I missing something ?
> 
> That was the case before the cpuidle_state were moved from the 
> cpuidle_device to the cpuidle_driver structure [1].
> 
> That had the benefit of using a single latencies array instead of 
> multiple copy of the same array, which was the case until today.
> 
> I looked at the white paper for the tegra3 and understand this is no 
> longer true because of the 4-plus-1 architecture [2].
> 

The reason is not so much 4-plus-1, but in 4 CPU mode, only CPUs 1 - 3 can
be powergated individually. To turn off CPU0, the external regulator for
the entire cluster is turned off. This means latencies for CPU0 are different
from the other CPUs.

> With the increasing number of SoCs, we have a lot of new cpuidle drivers 
> and each time we modify something in the cpuidle core, that impacts all 
> the cpuidle drivers.
> 
> My feeling is we are going back and forth when patching the cpuidle core 
> and may be it is time to define a clear semantic before patching again 
> the cpuidle, no ?
> 
> What could nice is to have:
> 
>   * in case of the same latencies for all cpus, use a single array
> 
>   * in case of different latencies, group the same latencies into a 
> single array (I assume this is the case for 4-plus-1, right ?)
> 
> May be we can move the cpuidle_state to a per_cpu pointer like 
> cpuidle_devices in cpuidle.c and then add:
> 
> register_latencies(struct cpuidle_latencies l, int cpu);
> 
> If we have the same latencies for all the cpus, then we can register the 
> same array, which is only a pointer.

Maybe we also want to make the 'disabled' flag per CPU then or provide some
other way the number of C states can be different per CPU?

Cheers,

Peter.

Re: [RFC PATCH] cpuidle: allow per cpu latencies

[Peter De Schrijver]

On Thu, Apr 05, 2012 at 03:37:13PM +0200, Arjan van de Ven wrote:
> On 4/5/2012 2:53 AM, Peter De Schrijver wrote:
> > This patch doesn't update all cpuidle device registrations. I will do that
> 
> question is if you want to do per cpu latencies, or if you want to have
> both types of C state in one big table, and have each of the tegra cpyu
> types pick half of them...
> 

How would the governor then know which C states belong to which CPUs?

Cheers,

Peter.

Re: [RFC PATCH] cpuidle: allow per cpu latencies

[Peter De Schrijver]

> 
> Maybe we also want to make the 'disabled' flag per CPU then or provide some
> other way the number of C states can be different per CPU?

What do you think about this? Do we also want to make the disabled flag per
CPU? Or how should we deal with a different number of C states per CPU?

Thanks,

Peter.

Re: [RFC PATCH] cpuidle: allow per cpu latencies

[Daniel Lezcano]

On 04/16/2012 05:34 PM, Peter De Schrijver wrote:
>>
>> Maybe we also want to make the 'disabled' flag per CPU then or provide some
>> other way the number of C states can be different per CPU?
>
> What do you think about this? Do we also want to make the disabled flag per
> CPU? Or how should we deal with a different number of C states per CPU?

Hi Peter,

yes, that could makes sense. But in most of the architecture, this is 
not needed, so duplicating the state's array and latencies is unneeded 
memory consumption.

Maybe we can look for a COW approach, similar to what is done for the 
nsproxy structure, no ?



-- 
  <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog

Re: [RFC PATCH] cpuidle: allow per cpu latencies

[Peter De Schrijver]

On Thu, Apr 19, 2012 at 11:14:27AM +0200, Daniel Lezcano wrote:
> On 04/16/2012 05:34 PM, Peter De Schrijver wrote:
> >>
> >> Maybe we also want to make the 'disabled' flag per CPU then or provide some
> >> other way the number of C states can be different per CPU?
> >
> > What do you think about this? Do we also want to make the disabled flag per
> > CPU? Or how should we deal with a different number of C states per CPU?
> 
> Hi Peter,
> 
> yes, that could makes sense. But in most of the architecture, this is 
> not needed, so duplicating the state's array and latencies is unneeded 
> memory consumption.
> 
> Maybe we can look for a COW approach, similar to what is done for the 
> nsproxy structure, no ?
> 

That could be easily solved by just having a pointer to the state table in the
per CPU datastructure I think?


Cheers,

Peter.

Re: [RFC PATCH] cpuidle: allow per cpu latencies

[Peter 'p2' De Schrijver]

On 2012-04-05 14:25:23 (+0300), Peter De Schrijver <pdeschrijver@nvidia.com> wrote:
> On some systems (eg Tegra30) latencies for a given C state are not equal for
> all CPUs. Therefore we introduce a new per CPU structure which contains
> those parameters.

Sorry for sending this several times.

Cheers,

Peter.

[RFC PATCH] cpuidle: allow per cpu latencies

[Peter De Schrijver]

On some systems (eg Tegra30) latencies for a given C state are not equal for
all CPUs. Therefore we introduce a new per CPU structure which contains
those parameters.

--

This patch doesn't update all cpuidle device registrations. I will do that
in a next version after agreement on the exact data structure to be
introduced.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
---
 arch/arm/mach-tegra/cpuidle.c      |   15 +++++++++++----
 drivers/cpuidle/cpuidle.c          |   27 ++++++++++++++++++++++++---
 drivers/cpuidle/driver.c           |   25 -------------------------
 drivers/cpuidle/governors/ladder.c |   21 ++++++++++++---------
 drivers/cpuidle/governors/menu.c   |    7 ++++---
 drivers/cpuidle/sysfs.c            |   34 +++++++++++++++++++++++-----------
 include/linux/cpuidle.h            |   17 ++++++++++-------
 7 files changed, 84 insertions(+), 62 deletions(-)

diff --git a/arch/arm/mach-tegra/cpuidle.c b/arch/arm/mach-tegra/cpuidle.c
index d83a8c0..b8fdf02 100644
--- a/arch/arm/mach-tegra/cpuidle.c
+++ b/arch/arm/mach-tegra/cpuidle.c
@@ -41,14 +41,20 @@ struct cpuidle_driver tegra_idle_driver = {
 	.states = {
 		[0] = {
 			.enter			= tegra_idle_enter_lp3,
-			.exit_latency		= 10,
-			.target_residency	= 10,
-			.power_usage		= 600,
-			.flags			= CPUIDLE_FLAG_TIME_VALID,
 			.name			= "LP3",
 			.desc			= "CPU flow-controlled",
 		},
 	},
+	.power_specified = 1,
+};
+
+static struct cpuidle_state_parameters __initdata tegra_idle_parameters[] = {
+	[0] = {
+		.exit_latency		= 10,
+		.target_residency	= 10,
+		.power_usage		= 600,
+		.flags			= CPUIDLE_FLAG_TIME_VALID,
+	},
 };
 
 static DEFINE_PER_CPU(struct cpuidle_device, tegra_idle_device);
@@ -95,6 +101,7 @@ static int __init tegra_cpuidle_init(void)
 		dev->cpu = cpu;
 
 		dev->state_count = drv->state_count;
+		dev->state_parameters[0] =  tegra_idle_parameters[0];
 		ret = cpuidle_register_device(dev);
 		if (ret) {
 			pr_err("CPU%u: CPUidle device registration failed\n",
diff --git a/drivers/cpuidle/cpuidle.c b/drivers/cpuidle/cpuidle.c
index 87411ce..b992d00 100644
--- a/drivers/cpuidle/cpuidle.c
+++ b/drivers/cpuidle/cpuidle.c
@@ -87,8 +87,9 @@ int cpuidle_play_dead(void)
 	for (i = CPUIDLE_DRIVER_STATE_START; i < drv->state_count; i++) {
 		struct cpuidle_state *s = &drv->states[i];
 
-		if (s->power_usage < power_usage && s->enter_dead) {
-			power_usage = s->power_usage;
+		if (dev->state_parameters[i].power_usage < power_usage
+				&& s->enter_dead) {
+			power_usage = dev->state_parameters[i].power_usage;
 			dead_state = i;
 		}
 	}
@@ -371,7 +372,7 @@ EXPORT_SYMBOL_GPL(cpuidle_disable_device);
  */
 static int __cpuidle_register_device(struct cpuidle_device *dev)
 {
-	int ret;
+	int ret, i;
 	struct device *cpu_dev = get_cpu_device((unsigned long)dev->cpu);
 	struct cpuidle_driver *cpuidle_driver = cpuidle_get_driver();
 
@@ -389,6 +390,26 @@ static int __cpuidle_register_device(struct cpuidle_device *dev)
 		return ret;
 	}
 
+	/*
+	 * cpuidle driver should set the drv->power_specified bit
+	 * before registering if the driver provides
+	 * power_usage numbers.
+	 *
+	 * If power_specified is not set,
+	 * we fill in power_usage with decreasing values as the
+	 * cpuidle code has an implicit assumption that state Cn
+	 * uses less power than C(n-1).
+	 *
+	 * With CONFIG_ARCH_HAS_CPU_RELAX, C0 is already assigned
+	 * an power value of -1.  So we use -2, -3, etc, for other
+	 * c-states.
+	 */
+	if (!cpuidle_driver->power_specified) {
+		for (i = CPUIDLE_DRIVER_STATE_START;
+				i < cpuidle_driver->state_count; i++)
+			dev->state_parameters[i].power_usage = -1 - i;
+	}
+
 	dev->registered = 1;
 	return 0;
 }
diff --git a/drivers/cpuidle/driver.c b/drivers/cpuidle/driver.c
index 40cd3f3..d81c6db 100644
--- a/drivers/cpuidle/driver.c
+++ b/drivers/cpuidle/driver.c
@@ -17,30 +17,6 @@
 static struct cpuidle_driver *cpuidle_curr_driver;
 DEFINE_SPINLOCK(cpuidle_driver_lock);
 
-static void __cpuidle_register_driver(struct cpuidle_driver *drv)
-{
-	int i;
-	/*
-	 * cpuidle driver should set the drv->power_specified bit
-	 * before registering if the driver provides
-	 * power_usage numbers.
-	 *
-	 * If power_specified is not set,
-	 * we fill in power_usage with decreasing values as the
-	 * cpuidle code has an implicit assumption that state Cn
-	 * uses less power than C(n-1).
-	 *
-	 * With CONFIG_ARCH_HAS_CPU_RELAX, C0 is already assigned
-	 * an power value of -1.  So we use -2, -3, etc, for other
-	 * c-states.
-	 */
-	if (!drv->power_specified) {
-		for (i = CPUIDLE_DRIVER_STATE_START; i < drv->state_count; i++)
-			drv->states[i].power_usage = -1 - i;
-	}
-}
-
-
 /**
  * cpuidle_register_driver - registers a driver
  * @drv: the driver
@@ -58,7 +34,6 @@ int cpuidle_register_driver(struct cpuidle_driver *drv)
 		spin_unlock(&cpuidle_driver_lock);
 		return -EBUSY;
 	}
-	__cpuidle_register_driver(drv);
 	cpuidle_curr_driver = drv;
 	spin_unlock(&cpuidle_driver_lock);
 
diff --git a/drivers/cpuidle/governors/ladder.c b/drivers/cpuidle/governors/ladder.c
index b6a09ea..2e7ea8c 100644
--- a/drivers/cpuidle/governors/ladder.c
+++ b/drivers/cpuidle/governors/ladder.c
@@ -79,9 +79,9 @@ static int ladder_select_state(struct cpuidle_driver *drv,
 
 	last_state = &ldev->states[last_idx];
 
-	if (drv->states[last_idx].flags & CPUIDLE_FLAG_TIME_VALID) {
+	if (dev->state_parameters[last_idx].flags & CPUIDLE_FLAG_TIME_VALID) {
 		last_residency = cpuidle_get_last_residency(dev) - \
-					 drv->states[last_idx].exit_latency;
+				 dev->state_parameters[last_idx].exit_latency;
 	}
 	else
 		last_residency = last_state->threshold.promotion_time + 1;
@@ -89,7 +89,7 @@ static int ladder_select_state(struct cpuidle_driver *drv,
 	/* consider promotion */
 	if (last_idx < drv->state_count - 1 &&
 	    last_residency > last_state->threshold.promotion_time &&
-	    drv->states[last_idx + 1].exit_latency <= latency_req) {
+	    dev->state_parameters[last_idx + 1].exit_latency <= latency_req) {
 		last_state->stats.promotion_count++;
 		last_state->stats.demotion_count = 0;
 		if (last_state->stats.promotion_count >= last_state->threshold.promotion_count) {
@@ -100,11 +100,12 @@ static int ladder_select_state(struct cpuidle_driver *drv,
 
 	/* consider demotion */
 	if (last_idx > CPUIDLE_DRIVER_STATE_START &&
-	    drv->states[last_idx].exit_latency > latency_req) {
+	    dev->state_parameters[last_idx].exit_latency > latency_req) {
 		int i;
 
 		for (i = last_idx - 1; i > CPUIDLE_DRIVER_STATE_START; i--) {
-			if (drv->states[i].exit_latency <= latency_req)
+			if (dev->state_parameters[i].exit_latency <=
+					latency_req)
 				break;
 		}
 		ladder_do_selection(ldev, last_idx, i);
@@ -136,12 +137,12 @@ static int ladder_enable_device(struct cpuidle_driver *drv,
 	int i;
 	struct ladder_device *ldev = &per_cpu(ladder_devices, dev->cpu);
 	struct ladder_device_state *lstate;
-	struct cpuidle_state *state;
+	struct cpuidle_state_parameters *state_parameters;
 
 	ldev->last_state_idx = CPUIDLE_DRIVER_STATE_START;
 
 	for (i = 0; i < drv->state_count; i++) {
-		state = &drv->states[i];
+		state_parameters = &dev->state_parameters[i];
 		lstate = &ldev->states[i];
 
 		lstate->stats.promotion_count = 0;
@@ -151,9 +152,11 @@ static int ladder_enable_device(struct cpuidle_driver *drv,
 		lstate->threshold.demotion_count = DEMOTION_COUNT;
 
 		if (i < drv->state_count - 1)
-			lstate->threshold.promotion_time = state->exit_latency;
+			lstate->threshold.promotion_time =
+				state_parameters->exit_latency;
 		if (i > 0)
-			lstate->threshold.demotion_time = state->exit_latency;
+			lstate->threshold.demotion_time =
+				state_parameters->exit_latency;
 	}
 
 	return 0;
diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c
index 0633575..e9bbc20 100644
--- a/drivers/cpuidle/governors/menu.c
+++ b/drivers/cpuidle/governors/menu.c
@@ -281,7 +281,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 	 * unless the timer is happening really really soon.
 	 */
 	if (data->expected_us > 5 &&
-		drv->states[CPUIDLE_DRIVER_STATE_START].disable == 0)
+		dev->state_parameters[CPUIDLE_DRIVER_STATE_START].disable == 0)
 		data->last_state_idx = CPUIDLE_DRIVER_STATE_START;
 
 	/*
@@ -289,7 +289,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 	 * our constraints.
 	 */
 	for (i = CPUIDLE_DRIVER_STATE_START; i < drv->state_count; i++) {
-		struct cpuidle_state *s = &drv->states[i];
+		struct cpuidle_state_parameters *s = &dev->state_parameters[i];
 
 		if (s->disable)
 			continue;
@@ -336,7 +336,8 @@ static void menu_update(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 	struct menu_device *data = &__get_cpu_var(menu_devices);
 	int last_idx = data->last_state_idx;
 	unsigned int last_idle_us = cpuidle_get_last_residency(dev);
-	struct cpuidle_state *target = &drv->states[last_idx];
+	struct cpuidle_state_parameters *target =
+			&dev->state_parameters[last_idx];
 	unsigned int measured_us;
 	u64 new_factor;
 
diff --git a/drivers/cpuidle/sysfs.c b/drivers/cpuidle/sysfs.c
index 88032b4..7b17413 100644
--- a/drivers/cpuidle/sysfs.c
+++ b/drivers/cpuidle/sysfs.c
@@ -215,9 +215,10 @@ static struct kobj_type ktype_cpuidle = {
 
 struct cpuidle_state_attr {
 	struct attribute attr;
-	ssize_t (*show)(struct cpuidle_state *, \
-					struct cpuidle_state_usage *, char *);
-	ssize_t (*store)(struct cpuidle_state *, const char *, size_t);
+	ssize_t (*show)(struct cpuidle_state *, struct cpuidle_state_usage *,
+				struct cpuidle_state_parameters *, char *);
+	ssize_t (*store)(struct cpuidle_state_parameters *, const char *,
+				size_t);
 };
 
 #define define_one_state_ro(_name, show) \
@@ -228,13 +229,15 @@ static struct cpuidle_state_attr attr_##_name = __ATTR(_name, 0644, show, store)
 
 #define define_show_state_function(_name) \
 static ssize_t show_state_##_name(struct cpuidle_state *state, \
-			 struct cpuidle_state_usage *state_usage, char *buf) \
+			 struct cpuidle_state_usage *state_usage, \
+			 struct cpuidle_state_parameters *state_parameters, \
+			 char *buf) \
 { \
-	return sprintf(buf, "%u\n", state->_name);\
+	return sprintf(buf, "%u\n", state_parameters->_name);\
 }
 
 #define define_store_state_function(_name) \
-static ssize_t store_state_##_name(struct cpuidle_state *state, \
+static ssize_t store_state_##_name(struct cpuidle_state_parameters *state, \
 		const char *buf, size_t size) \
 { \
 	long value; \
@@ -253,14 +256,18 @@ static ssize_t store_state_##_name(struct cpuidle_state *state, \
 
 #define define_show_state_ull_function(_name) \
 static ssize_t show_state_##_name(struct cpuidle_state *state, \
-			struct cpuidle_state_usage *state_usage, char *buf) \
+			struct cpuidle_state_usage *state_usage, \
+			struct cpuidle_state_parameters *state_parameters, \
+			char *buf) \
 { \
 	return sprintf(buf, "%llu\n", state_usage->_name);\
 }
 
 #define define_show_state_str_function(_name) \
 static ssize_t show_state_##_name(struct cpuidle_state *state, \
-			struct cpuidle_state_usage *state_usage, char *buf) \
+			struct cpuidle_state_usage *state_usage, \
+			struct cpuidle_state_parameters *state_parameters, \
+			char *buf) \
 { \
 	if (state->_name[0] == '\0')\
 		return sprintf(buf, "<null>\n");\
@@ -298,6 +305,7 @@ static struct attribute *cpuidle_state_default_attrs[] = {
 #define kobj_to_state_obj(k) container_of(k, struct cpuidle_state_kobj, kobj)
 #define kobj_to_state(k) (kobj_to_state_obj(k)->state)
 #define kobj_to_state_usage(k) (kobj_to_state_obj(k)->state_usage)
+#define kobj_to_state_parameters(k) (kobj_to_state_obj(k)->state_parameters)
 #define attr_to_stateattr(a) container_of(a, struct cpuidle_state_attr, attr)
 static ssize_t cpuidle_state_show(struct kobject * kobj,
 	struct attribute * attr ,char * buf)
@@ -305,10 +313,12 @@ static ssize_t cpuidle_state_show(struct kobject * kobj,
 	int ret = -EIO;
 	struct cpuidle_state *state = kobj_to_state(kobj);
 	struct cpuidle_state_usage *state_usage = kobj_to_state_usage(kobj);
+	struct cpuidle_state_parameters *state_parameters =
+				kobj_to_state_parameters(kobj);
 	struct cpuidle_state_attr * cattr = attr_to_stateattr(attr);
 
 	if (cattr->show)
-		ret = cattr->show(state, state_usage, buf);
+		ret = cattr->show(state, state_usage, state_parameters, buf);
 
 	return ret;
 }
@@ -317,11 +327,12 @@ static ssize_t cpuidle_state_store(struct kobject *kobj,
 	struct attribute *attr, const char *buf, size_t size)
 {
 	int ret = -EIO;
-	struct cpuidle_state *state = kobj_to_state(kobj);
+	struct cpuidle_state_parameters *state_parameters =
+				kobj_to_state_parameters(kobj);
 	struct cpuidle_state_attr *cattr = attr_to_stateattr(attr);
 
 	if (cattr->store)
-		ret = cattr->store(state, buf, size);
+		ret = cattr->store(state_parameters, buf, size);
 
 	return ret;
 }
@@ -369,6 +380,7 @@ int cpuidle_add_state_sysfs(struct cpuidle_device *device)
 			goto error_state;
 		kobj->state = &drv->states[i];
 		kobj->state_usage = &device->states_usage[i];
+		kobj->state_parameters = &device->state_parameters[i];
 		init_completion(&kobj->kobj_unregister);
 
 		ret = kobject_init_and_add(&kobj->kobj, &ktype_state_cpuidle, &device->kobj,
diff --git a/include/linux/cpuidle.h b/include/linux/cpuidle.h
index 6c26a3d..cddbd34 100644
--- a/include/linux/cpuidle.h
+++ b/include/linux/cpuidle.h
@@ -42,12 +42,6 @@ struct cpuidle_state {
 	char		name[CPUIDLE_NAME_LEN];
 	char		desc[CPUIDLE_DESC_LEN];
 
-	unsigned int	flags;
-	unsigned int	exit_latency; /* in US */
-	int		power_usage; /* in mW */
-	unsigned int	target_residency; /* in US */
-	unsigned int    disable;
-
 	int (*enter)	(struct cpuidle_device *dev,
 			struct cpuidle_driver *drv,
 			int index);
@@ -55,6 +49,14 @@ struct cpuidle_state {
 	int (*enter_dead) (struct cpuidle_device *dev, int index);
 };
 
+struct cpuidle_state_parameters {
+	unsigned int	flags;
+	unsigned int	exit_latency; /* in US */
+	int		power_usage; /* in mW */
+	unsigned int	target_residency; /* in US */
+	unsigned int    disable;
+};
+
 /* Idle State Flags */
 #define CPUIDLE_FLAG_TIME_VALID	(0x01) /* is residency time measurable? */
 
@@ -83,6 +85,7 @@ cpuidle_set_statedata(struct cpuidle_state_usage *st_usage, void *data)
 struct cpuidle_state_kobj {
 	struct cpuidle_state *state;
 	struct cpuidle_state_usage *state_usage;
+	struct cpuidle_state_parameters *state_parameters;
 	struct completion kobj_unregister;
 	struct kobject kobj;
 };
@@ -96,7 +99,7 @@ struct cpuidle_device {
 	int			state_count;
 	struct cpuidle_state_usage	states_usage[CPUIDLE_STATE_MAX];
 	struct cpuidle_state_kobj *kobjs[CPUIDLE_STATE_MAX];
-
+	struct cpuidle_state_parameters state_parameters[CPUIDLE_STATE_MAX];
 	struct list_head 	device_list;
 	struct kobject		kobj;
 	struct completion	kobj_unregister;
-- 
1.7.7.rc0.72.g4b5ea.dirty

[RFC PATCH] cpuidle: allow per cpu latencies

[Peter De Schrijver]

On some systems (eg Tegra30) latencies for a given C state are not equal for
all CPUs. Therefore we introduce a new per CPU structure which contains
those parameters.

--

This patch doesn't update all cpuidle device registrations. I will do that
in a next version after agreement on the exact data structure to be
introduced.

Signed-off-by: Peter De Schrijver <pdeschrijver@nvidia.com>
---
 arch/arm/mach-tegra/cpuidle.c      |   15 +++++++++++----
 drivers/cpuidle/cpuidle.c          |   27 ++++++++++++++++++++++++---
 drivers/cpuidle/driver.c           |   25 -------------------------
 drivers/cpuidle/governors/ladder.c |   21 ++++++++++++---------
 drivers/cpuidle/governors/menu.c   |    7 ++++---
 drivers/cpuidle/sysfs.c            |   34 +++++++++++++++++++++++-----------
 include/linux/cpuidle.h            |   17 ++++++++++-------
 7 files changed, 84 insertions(+), 62 deletions(-)

diff --git a/arch/arm/mach-tegra/cpuidle.c b/arch/arm/mach-tegra/cpuidle.c
index d83a8c0..b8fdf02 100644
--- a/arch/arm/mach-tegra/cpuidle.c
+++ b/arch/arm/mach-tegra/cpuidle.c
@@ -41,14 +41,20 @@ struct cpuidle_driver tegra_idle_driver = {
 	.states = {
 		[0] = {
 			.enter			= tegra_idle_enter_lp3,
-			.exit_latency		= 10,
-			.target_residency	= 10,
-			.power_usage		= 600,
-			.flags			= CPUIDLE_FLAG_TIME_VALID,
 			.name			= "LP3",
 			.desc			= "CPU flow-controlled",
 		},
 	},
+	.power_specified = 1,
+};
+
+static struct cpuidle_state_parameters __initdata tegra_idle_parameters[] = {
+	[0] = {
+		.exit_latency		= 10,
+		.target_residency	= 10,
+		.power_usage		= 600,
+		.flags			= CPUIDLE_FLAG_TIME_VALID,
+	},
 };
 
 static DEFINE_PER_CPU(struct cpuidle_device, tegra_idle_device);
@@ -95,6 +101,7 @@ static int __init tegra_cpuidle_init(void)
 		dev->cpu = cpu;
 
 		dev->state_count = drv->state_count;
+		dev->state_parameters[0] =  tegra_idle_parameters[0];
 		ret = cpuidle_register_device(dev);
 		if (ret) {
 			pr_err("CPU%u: CPUidle device registration failed\n",
diff --git a/drivers/cpuidle/cpuidle.c b/drivers/cpuidle/cpuidle.c
index 87411ce..b992d00 100644
--- a/drivers/cpuidle/cpuidle.c
+++ b/drivers/cpuidle/cpuidle.c
@@ -87,8 +87,9 @@ int cpuidle_play_dead(void)
 	for (i = CPUIDLE_DRIVER_STATE_START; i < drv->state_count; i++) {
 		struct cpuidle_state *s = &drv->states[i];
 
-		if (s->power_usage < power_usage && s->enter_dead) {
-			power_usage = s->power_usage;
+		if (dev->state_parameters[i].power_usage < power_usage
+				&& s->enter_dead) {
+			power_usage = dev->state_parameters[i].power_usage;
 			dead_state = i;
 		}
 	}
@@ -371,7 +372,7 @@ EXPORT_SYMBOL_GPL(cpuidle_disable_device);
  */
 static int __cpuidle_register_device(struct cpuidle_device *dev)
 {
-	int ret;
+	int ret, i;
 	struct device *cpu_dev = get_cpu_device((unsigned long)dev->cpu);
 	struct cpuidle_driver *cpuidle_driver = cpuidle_get_driver();
 
@@ -389,6 +390,26 @@ static int __cpuidle_register_device(struct cpuidle_device *dev)
 		return ret;
 	}
 
+	/*
+	 * cpuidle driver should set the drv->power_specified bit
+	 * before registering if the driver provides
+	 * power_usage numbers.
+	 *
+	 * If power_specified is not set,
+	 * we fill in power_usage with decreasing values as the
+	 * cpuidle code has an implicit assumption that state Cn
+	 * uses less power than C(n-1).
+	 *
+	 * With CONFIG_ARCH_HAS_CPU_RELAX, C0 is already assigned
+	 * an power value of -1.  So we use -2, -3, etc, for other
+	 * c-states.
+	 */
+	if (!cpuidle_driver->power_specified) {
+		for (i = CPUIDLE_DRIVER_STATE_START;
+				i < cpuidle_driver->state_count; i++)
+			dev->state_parameters[i].power_usage = -1 - i;
+	}
+
 	dev->registered = 1;
 	return 0;
 }
diff --git a/drivers/cpuidle/driver.c b/drivers/cpuidle/driver.c
index 40cd3f3..d81c6db 100644
--- a/drivers/cpuidle/driver.c
+++ b/drivers/cpuidle/driver.c
@@ -17,30 +17,6 @@
 static struct cpuidle_driver *cpuidle_curr_driver;
 DEFINE_SPINLOCK(cpuidle_driver_lock);
 
-static void __cpuidle_register_driver(struct cpuidle_driver *drv)
-{
-	int i;
-	/*
-	 * cpuidle driver should set the drv->power_specified bit
-	 * before registering if the driver provides
-	 * power_usage numbers.
-	 *
-	 * If power_specified is not set,
-	 * we fill in power_usage with decreasing values as the
-	 * cpuidle code has an implicit assumption that state Cn
-	 * uses less power than C(n-1).
-	 *
-	 * With CONFIG_ARCH_HAS_CPU_RELAX, C0 is already assigned
-	 * an power value of -1.  So we use -2, -3, etc, for other
-	 * c-states.
-	 */
-	if (!drv->power_specified) {
-		for (i = CPUIDLE_DRIVER_STATE_START; i < drv->state_count; i++)
-			drv->states[i].power_usage = -1 - i;
-	}
-}
-
-
 /**
  * cpuidle_register_driver - registers a driver
  * @drv: the driver
@@ -58,7 +34,6 @@ int cpuidle_register_driver(struct cpuidle_driver *drv)
 		spin_unlock(&cpuidle_driver_lock);
 		return -EBUSY;
 	}
-	__cpuidle_register_driver(drv);
 	cpuidle_curr_driver = drv;
 	spin_unlock(&cpuidle_driver_lock);
 
diff --git a/drivers/cpuidle/governors/ladder.c b/drivers/cpuidle/governors/ladder.c
index b6a09ea..2e7ea8c 100644
--- a/drivers/cpuidle/governors/ladder.c
+++ b/drivers/cpuidle/governors/ladder.c
@@ -79,9 +79,9 @@ static int ladder_select_state(struct cpuidle_driver *drv,
 
 	last_state = &ldev->states[last_idx];
 
-	if (drv->states[last_idx].flags & CPUIDLE_FLAG_TIME_VALID) {
+	if (dev->state_parameters[last_idx].flags & CPUIDLE_FLAG_TIME_VALID) {
 		last_residency = cpuidle_get_last_residency(dev) - \
-					 drv->states[last_idx].exit_latency;
+				 dev->state_parameters[last_idx].exit_latency;
 	}
 	else
 		last_residency = last_state->threshold.promotion_time + 1;
@@ -89,7 +89,7 @@ static int ladder_select_state(struct cpuidle_driver *drv,
 	/* consider promotion */
 	if (last_idx < drv->state_count - 1 &&
 	    last_residency > last_state->threshold.promotion_time &&
-	    drv->states[last_idx + 1].exit_latency <= latency_req) {
+	    dev->state_parameters[last_idx + 1].exit_latency <= latency_req) {
 		last_state->stats.promotion_count++;
 		last_state->stats.demotion_count = 0;
 		if (last_state->stats.promotion_count >= last_state->threshold.promotion_count) {
@@ -100,11 +100,12 @@ static int ladder_select_state(struct cpuidle_driver *drv,
 
 	/* consider demotion */
 	if (last_idx > CPUIDLE_DRIVER_STATE_START &&
-	    drv->states[last_idx].exit_latency > latency_req) {
+	    dev->state_parameters[last_idx].exit_latency > latency_req) {
 		int i;
 
 		for (i = last_idx - 1; i > CPUIDLE_DRIVER_STATE_START; i--) {
-			if (drv->states[i].exit_latency <= latency_req)
+			if (dev->state_parameters[i].exit_latency <=
+					latency_req)
 				break;
 		}
 		ladder_do_selection(ldev, last_idx, i);
@@ -136,12 +137,12 @@ static int ladder_enable_device(struct cpuidle_driver *drv,
 	int i;
 	struct ladder_device *ldev = &per_cpu(ladder_devices, dev->cpu);
 	struct ladder_device_state *lstate;
-	struct cpuidle_state *state;
+	struct cpuidle_state_parameters *state_parameters;
 
 	ldev->last_state_idx = CPUIDLE_DRIVER_STATE_START;
 
 	for (i = 0; i < drv->state_count; i++) {
-		state = &drv->states[i];
+		state_parameters = &dev->state_parameters[i];
 		lstate = &ldev->states[i];
 
 		lstate->stats.promotion_count = 0;
@@ -151,9 +152,11 @@ static int ladder_enable_device(struct cpuidle_driver *drv,
 		lstate->threshold.demotion_count = DEMOTION_COUNT;
 
 		if (i < drv->state_count - 1)
-			lstate->threshold.promotion_time = state->exit_latency;
+			lstate->threshold.promotion_time =
+				state_parameters->exit_latency;
 		if (i > 0)
-			lstate->threshold.demotion_time = state->exit_latency;
+			lstate->threshold.demotion_time =
+				state_parameters->exit_latency;
 	}
 
 	return 0;
diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c
index 0633575..e9bbc20 100644
--- a/drivers/cpuidle/governors/menu.c
+++ b/drivers/cpuidle/governors/menu.c
@@ -281,7 +281,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 	 * unless the timer is happening really really soon.
 	 */
 	if (data->expected_us > 5 &&
-		drv->states[CPUIDLE_DRIVER_STATE_START].disable == 0)
+		dev->state_parameters[CPUIDLE_DRIVER_STATE_START].disable == 0)
 		data->last_state_idx = CPUIDLE_DRIVER_STATE_START;
 
 	/*
@@ -289,7 +289,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 	 * our constraints.
 	 */
 	for (i = CPUIDLE_DRIVER_STATE_START; i < drv->state_count; i++) {
-		struct cpuidle_state *s = &drv->states[i];
+		struct cpuidle_state_parameters *s = &dev->state_parameters[i];
 
 		if (s->disable)
 			continue;
@@ -336,7 +336,8 @@ static void menu_update(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 	struct menu_device *data = &__get_cpu_var(menu_devices);
 	int last_idx = data->last_state_idx;
 	unsigned int last_idle_us = cpuidle_get_last_residency(dev);
-	struct cpuidle_state *target = &drv->states[last_idx];
+	struct cpuidle_state_parameters *target =
+			&dev->state_parameters[last_idx];
 	unsigned int measured_us;
 	u64 new_factor;
 
diff --git a/drivers/cpuidle/sysfs.c b/drivers/cpuidle/sysfs.c
index 88032b4..7b17413 100644
--- a/drivers/cpuidle/sysfs.c
+++ b/drivers/cpuidle/sysfs.c
@@ -215,9 +215,10 @@ static struct kobj_type ktype_cpuidle = {
 
 struct cpuidle_state_attr {
 	struct attribute attr;
-	ssize_t (*show)(struct cpuidle_state *, \
-					struct cpuidle_state_usage *, char *);
-	ssize_t (*store)(struct cpuidle_state *, const char *, size_t);
+	ssize_t (*show)(struct cpuidle_state *, struct cpuidle_state_usage *,
+				struct cpuidle_state_parameters *, char *);
+	ssize_t (*store)(struct cpuidle_state_parameters *, const char *,
+				size_t);
 };
 
 #define define_one_state_ro(_name, show) \
@@ -228,13 +229,15 @@ static struct cpuidle_state_attr attr_##_name = __ATTR(_name, 0644, show, store)
 
 #define define_show_state_function(_name) \
 static ssize_t show_state_##_name(struct cpuidle_state *state, \
-			 struct cpuidle_state_usage *state_usage, char *buf) \
+			 struct cpuidle_state_usage *state_usage, \
+			 struct cpuidle_state_parameters *state_parameters, \
+			 char *buf) \
 { \
-	return sprintf(buf, "%u\n", state->_name);\
+	return sprintf(buf, "%u\n", state_parameters->_name);\
 }
 
 #define define_store_state_function(_name) \
-static ssize_t store_state_##_name(struct cpuidle_state *state, \
+static ssize_t store_state_##_name(struct cpuidle_state_parameters *state, \
 		const char *buf, size_t size) \
 { \
 	long value; \
@@ -253,14 +256,18 @@ static ssize_t store_state_##_name(struct cpuidle_state *state, \
 
 #define define_show_state_ull_function(_name) \
 static ssize_t show_state_##_name(struct cpuidle_state *state, \
-			struct cpuidle_state_usage *state_usage, char *buf) \
+			struct cpuidle_state_usage *state_usage, \
+			struct cpuidle_state_parameters *state_parameters, \
+			char *buf) \
 { \
 	return sprintf(buf, "%llu\n", state_usage->_name);\
 }
 
 #define define_show_state_str_function(_name) \
 static ssize_t show_state_##_name(struct cpuidle_state *state, \
-			struct cpuidle_state_usage *state_usage, char *buf) \
+			struct cpuidle_state_usage *state_usage, \
+			struct cpuidle_state_parameters *state_parameters, \
+			char *buf) \
 { \
 	if (state->_name[0] == '\0')\
 		return sprintf(buf, "<null>\n");\
@@ -298,6 +305,7 @@ static struct attribute *cpuidle_state_default_attrs[] = {
 #define kobj_to_state_obj(k) container_of(k, struct cpuidle_state_kobj, kobj)
 #define kobj_to_state(k) (kobj_to_state_obj(k)->state)
 #define kobj_to_state_usage(k) (kobj_to_state_obj(k)->state_usage)
+#define kobj_to_state_parameters(k) (kobj_to_state_obj(k)->state_parameters)
 #define attr_to_stateattr(a) container_of(a, struct cpuidle_state_attr, attr)
 static ssize_t cpuidle_state_show(struct kobject * kobj,
 	struct attribute * attr ,char * buf)
@@ -305,10 +313,12 @@ static ssize_t cpuidle_state_show(struct kobject * kobj,
 	int ret = -EIO;
 	struct cpuidle_state *state = kobj_to_state(kobj);
 	struct cpuidle_state_usage *state_usage = kobj_to_state_usage(kobj);
+	struct cpuidle_state_parameters *state_parameters =
+				kobj_to_state_parameters(kobj);
 	struct cpuidle_state_attr * cattr = attr_to_stateattr(attr);
 
 	if (cattr->show)
-		ret = cattr->show(state, state_usage, buf);
+		ret = cattr->show(state, state_usage, state_parameters, buf);
 
 	return ret;
 }
@@ -317,11 +327,12 @@ static ssize_t cpuidle_state_store(struct kobject *kobj,
 	struct attribute *attr, const char *buf, size_t size)
 {
 	int ret = -EIO;
-	struct cpuidle_state *state = kobj_to_state(kobj);
+	struct cpuidle_state_parameters *state_parameters =
+				kobj_to_state_parameters(kobj);
 	struct cpuidle_state_attr *cattr = attr_to_stateattr(attr);
 
 	if (cattr->store)
-		ret = cattr->store(state, buf, size);
+		ret = cattr->store(state_parameters, buf, size);
 
 	return ret;
 }
@@ -369,6 +380,7 @@ int cpuidle_add_state_sysfs(struct cpuidle_device *device)
 			goto error_state;
 		kobj->state = &drv->states[i];
 		kobj->state_usage = &device->states_usage[i];
+		kobj->state_parameters = &device->state_parameters[i];
 		init_completion(&kobj->kobj_unregister);
 
 		ret = kobject_init_and_add(&kobj->kobj, &ktype_state_cpuidle, &device->kobj,
diff --git a/include/linux/cpuidle.h b/include/linux/cpuidle.h
index 6c26a3d..cddbd34 100644
--- a/include/linux/cpuidle.h
+++ b/include/linux/cpuidle.h
@@ -42,12 +42,6 @@ struct cpuidle_state {
 	char		name[CPUIDLE_NAME_LEN];
 	char		desc[CPUIDLE_DESC_LEN];
 
-	unsigned int	flags;
-	unsigned int	exit_latency; /* in US */
-	int		power_usage; /* in mW */
-	unsigned int	target_residency; /* in US */
-	unsigned int    disable;
-
 	int (*enter)	(struct cpuidle_device *dev,
 			struct cpuidle_driver *drv,
 			int index);
@@ -55,6 +49,14 @@ struct cpuidle_state {
 	int (*enter_dead) (struct cpuidle_device *dev, int index);
 };
 
+struct cpuidle_state_parameters {
+	unsigned int	flags;
+	unsigned int	exit_latency; /* in US */
+	int		power_usage; /* in mW */
+	unsigned int	target_residency; /* in US */
+	unsigned int    disable;
+};
+
 /* Idle State Flags */
 #define CPUIDLE_FLAG_TIME_VALID	(0x01) /* is residency time measurable? */
 
@@ -83,6 +85,7 @@ cpuidle_set_statedata(struct cpuidle_state_usage *st_usage, void *data)
 struct cpuidle_state_kobj {
 	struct cpuidle_state *state;
 	struct cpuidle_state_usage *state_usage;
+	struct cpuidle_state_parameters *state_parameters;
 	struct completion kobj_unregister;
 	struct kobject kobj;
 };
@@ -96,7 +99,7 @@ struct cpuidle_device {
 	int			state_count;
 	struct cpuidle_state_usage	states_usage[CPUIDLE_STATE_MAX];
 	struct cpuidle_state_kobj *kobjs[CPUIDLE_STATE_MAX];
-
+	struct cpuidle_state_parameters state_parameters[CPUIDLE_STATE_MAX];
 	struct list_head 	device_list;
 	struct kobject		kobj;
 	struct completion	kobj_unregister;
-- 
1.7.7.rc0.72.g4b5ea.dirty