cpuidle: Single/Global registration of idle states
This patch makes the cpuidle_states structure global (single copy)
instead of per-cpu. The statistics needed on per-cpu basis
by the governor are kept per-cpu. This simplifies the cpuidle
subsystem as state registration is done by single cpu only.
Having single copy of cpuidle_states saves memory. Rare case
of asymmetric C-states can be handled within the cpuidle driver
and architectures such as POWER do not have asymmetric C-states.
Having single/global registration of all the idle states,
dynamic C-state transitions on x86 are handled by
the boot cpu. Here, the boot cpu would disable all the devices,
re-populate the states and later enable all the devices,
irrespective of the cpu that would receive the notification first.
Reference:
https://lkml.org/lkml/2011/4/25/83
Signed-off-by: Deepthi Dharwar <deepthi@linux.vnet.ibm.com>
Signed-off-by: Trinabh Gupta <g.trinabh@gmail.com>
Tested-by: Jean Pihet <j-pihet@ti.com>
Reviewed-by: Kevin Hilman <khilman@ti.com>
Acked-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Kevin Hilman <khilman@ti.com>
Signed-off-by: Len Brown <len.brown@intel.com>
diff --git a/drivers/cpuidle/cpuidle.c b/drivers/cpuidle/cpuidle.c
index 7127e92..7a57b11 100644
--- a/drivers/cpuidle/cpuidle.c
+++ b/drivers/cpuidle/cpuidle.c
@@ -61,6 +61,7 @@
int cpuidle_idle_call(void)
{
struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
+ struct cpuidle_driver *drv = cpuidle_get_driver();
struct cpuidle_state *target_state;
int next_state, entered_state;
@@ -84,18 +85,18 @@
#endif
/* ask the governor for the next state */
- next_state = cpuidle_curr_governor->select(dev);
+ next_state = cpuidle_curr_governor->select(drv, dev);
if (need_resched()) {
local_irq_enable();
return 0;
}
- target_state = &dev->states[next_state];
+ target_state = &drv->states[next_state];
trace_power_start(POWER_CSTATE, next_state, dev->cpu);
trace_cpu_idle(next_state, dev->cpu);
- entered_state = target_state->enter(dev, next_state);
+ entered_state = target_state->enter(dev, drv, next_state);
trace_power_end(dev->cpu);
trace_cpu_idle(PWR_EVENT_EXIT, dev->cpu);
@@ -163,7 +164,8 @@
EXPORT_SYMBOL_GPL(cpuidle_resume_and_unlock);
#ifdef CONFIG_ARCH_HAS_CPU_RELAX
-static int poll_idle(struct cpuidle_device *dev, int index)
+static int poll_idle(struct cpuidle_device *dev,
+ struct cpuidle_driver *drv, int index)
{
ktime_t t1, t2;
s64 diff;
@@ -183,12 +185,9 @@
return index;
}
-static void poll_idle_init(struct cpuidle_device *dev)
+static void poll_idle_init(struct cpuidle_driver *drv)
{
- struct cpuidle_state *state = &dev->states[0];
- struct cpuidle_state_usage *state_usage = &dev->states_usage[0];
-
- cpuidle_set_statedata(state_usage, NULL);
+ struct cpuidle_state *state = &drv->states[0];
snprintf(state->name, CPUIDLE_NAME_LEN, "POLL");
snprintf(state->desc, CPUIDLE_DESC_LEN, "CPUIDLE CORE POLL IDLE");
@@ -199,7 +198,7 @@
state->enter = poll_idle;
}
#else
-static void poll_idle_init(struct cpuidle_device *dev) {}
+static void poll_idle_init(struct cpuidle_driver *drv) {}
#endif /* CONFIG_ARCH_HAS_CPU_RELAX */
/**
@@ -226,13 +225,13 @@
return ret;
}
- poll_idle_init(dev);
+ poll_idle_init(cpuidle_get_driver());
if ((ret = cpuidle_add_state_sysfs(dev)))
return ret;
if (cpuidle_curr_governor->enable &&
- (ret = cpuidle_curr_governor->enable(dev)))
+ (ret = cpuidle_curr_governor->enable(cpuidle_get_driver(), dev)))
goto fail_sysfs;
for (i = 0; i < dev->state_count; i++) {
@@ -273,7 +272,7 @@
dev->enabled = 0;
if (cpuidle_curr_governor->disable)
- cpuidle_curr_governor->disable(dev);
+ cpuidle_curr_governor->disable(cpuidle_get_driver(), dev);
cpuidle_remove_state_sysfs(dev);
enabled_devices--;
@@ -301,26 +300,6 @@
init_completion(&dev->kobj_unregister);
- /*
- * cpuidle driver should set the dev->power_specified bit
- * before registering the device if the driver provides
- * power_usage numbers.
- *
- * For those devices whose ->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 (!dev->power_specified) {
- int i;
- for (i = CPUIDLE_DRIVER_STATE_START; i < dev->state_count; i++)
- dev->states[i].power_usage = -1 - i;
- }
-
per_cpu(cpuidle_devices, dev->cpu) = dev;
list_add(&dev->device_list, &cpuidle_detected_devices);
if ((ret = cpuidle_add_sysfs(sys_dev))) {
diff --git a/drivers/cpuidle/driver.c b/drivers/cpuidle/driver.c
index 3f7e3ce..284d7af 100644
--- a/drivers/cpuidle/driver.c
+++ b/drivers/cpuidle/driver.c
@@ -17,6 +17,30 @@
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
@@ -34,6 +58,7 @@
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 6a686a7..ef6b9e4 100644
--- a/drivers/cpuidle/governors/ladder.c
+++ b/drivers/cpuidle/governors/ladder.c
@@ -60,9 +60,11 @@
/**
* ladder_select_state - selects the next state to enter
+ * @drv: cpuidle driver
* @dev: the CPU
*/
-static int ladder_select_state(struct cpuidle_device *dev)
+static int ladder_select_state(struct cpuidle_driver *drv,
+ struct cpuidle_device *dev)
{
struct ladder_device *ldev = &__get_cpu_var(ladder_devices);
struct ladder_device_state *last_state;
@@ -77,15 +79,17 @@
last_state = &ldev->states[last_idx];
- if (dev->states[last_idx].flags & CPUIDLE_FLAG_TIME_VALID)
- last_residency = cpuidle_get_last_residency(dev) - dev->states[last_idx].exit_latency;
+ if (drv->states[last_idx].flags & CPUIDLE_FLAG_TIME_VALID) {
+ last_residency = cpuidle_get_last_residency(dev) - \
+ drv->states[last_idx].exit_latency;
+ }
else
last_residency = last_state->threshold.promotion_time + 1;
/* consider promotion */
- if (last_idx < dev->state_count - 1 &&
+ if (last_idx < drv->state_count - 1 &&
last_residency > last_state->threshold.promotion_time &&
- dev->states[last_idx + 1].exit_latency <= latency_req) {
+ drv->states[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) {
@@ -96,11 +100,11 @@
/* consider demotion */
if (last_idx > CPUIDLE_DRIVER_STATE_START &&
- dev->states[last_idx].exit_latency > latency_req) {
+ drv->states[last_idx].exit_latency > latency_req) {
int i;
for (i = last_idx - 1; i > CPUIDLE_DRIVER_STATE_START; i--) {
- if (dev->states[i].exit_latency <= latency_req)
+ if (drv->states[i].exit_latency <= latency_req)
break;
}
ladder_do_selection(ldev, last_idx, i);
@@ -123,9 +127,11 @@
/**
* ladder_enable_device - setup for the governor
+ * @drv: cpuidle driver
* @dev: the CPU
*/
-static int ladder_enable_device(struct cpuidle_device *dev)
+static int ladder_enable_device(struct cpuidle_driver *drv,
+ struct cpuidle_device *dev)
{
int i;
struct ladder_device *ldev = &per_cpu(ladder_devices, dev->cpu);
@@ -134,8 +140,8 @@
ldev->last_state_idx = CPUIDLE_DRIVER_STATE_START;
- for (i = 0; i < dev->state_count; i++) {
- state = &dev->states[i];
+ for (i = 0; i < drv->state_count; i++) {
+ state = &drv->states[i];
lstate = &ldev->states[i];
lstate->stats.promotion_count = 0;
@@ -144,7 +150,7 @@
lstate->threshold.promotion_count = PROMOTION_COUNT;
lstate->threshold.demotion_count = DEMOTION_COUNT;
- if (i < dev->state_count - 1)
+ if (i < drv->state_count - 1)
lstate->threshold.promotion_time = state->exit_latency;
if (i > 0)
lstate->threshold.demotion_time = state->exit_latency;
diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c
index af724e8..bcbe881 100644
--- a/drivers/cpuidle/governors/menu.c
+++ b/drivers/cpuidle/governors/menu.c
@@ -182,7 +182,7 @@
static DEFINE_PER_CPU(struct menu_device, menu_devices);
-static void menu_update(struct cpuidle_device *dev);
+static void menu_update(struct cpuidle_driver *drv, struct cpuidle_device *dev);
/* This implements DIV_ROUND_CLOSEST but avoids 64 bit division */
static u64 div_round64(u64 dividend, u32 divisor)
@@ -228,9 +228,10 @@
/**
* menu_select - selects the next idle state to enter
+ * @drv: cpuidle driver containing state data
* @dev: the CPU
*/
-static int menu_select(struct cpuidle_device *dev)
+static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
{
struct menu_device *data = &__get_cpu_var(menu_devices);
int latency_req = pm_qos_request(PM_QOS_CPU_DMA_LATENCY);
@@ -240,7 +241,7 @@
struct timespec t;
if (data->needs_update) {
- menu_update(dev);
+ menu_update(drv, dev);
data->needs_update = 0;
}
@@ -285,8 +286,8 @@
* Find the idle state with the lowest power while satisfying
* our constraints.
*/
- for (i = CPUIDLE_DRIVER_STATE_START; i < dev->state_count; i++) {
- struct cpuidle_state *s = &dev->states[i];
+ for (i = CPUIDLE_DRIVER_STATE_START; i < drv->state_count; i++) {
+ struct cpuidle_state *s = &drv->states[i];
if (s->target_residency > data->predicted_us)
continue;
@@ -323,14 +324,15 @@
/**
* menu_update - attempts to guess what happened after entry
+ * @drv: cpuidle driver containing state data
* @dev: the CPU
*/
-static void menu_update(struct cpuidle_device *dev)
+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 = &dev->states[last_idx];
+ struct cpuidle_state *target = &drv->states[last_idx];
unsigned int measured_us;
u64 new_factor;
@@ -384,9 +386,11 @@
/**
* menu_enable_device - scans a CPU's states and does setup
+ * @drv: cpuidle driver
* @dev: the CPU
*/
-static int menu_enable_device(struct cpuidle_device *dev)
+static int menu_enable_device(struct cpuidle_driver *drv,
+ struct cpuidle_device *dev)
{
struct menu_device *data = &per_cpu(menu_devices, dev->cpu);
diff --git a/drivers/cpuidle/sysfs.c b/drivers/cpuidle/sysfs.c
index 8a1ace1..1e756e1 100644
--- a/drivers/cpuidle/sysfs.c
+++ b/drivers/cpuidle/sysfs.c
@@ -322,13 +322,14 @@
{
int i, ret = -ENOMEM;
struct cpuidle_state_kobj *kobj;
+ struct cpuidle_driver *drv = cpuidle_get_driver();
/* state statistics */
for (i = 0; i < device->state_count; i++) {
kobj = kzalloc(sizeof(struct cpuidle_state_kobj), GFP_KERNEL);
if (!kobj)
goto error_state;
- kobj->state = &device->states[i];
+ kobj->state = &drv->states[i];
kobj->state_usage = &device->states_usage[i];
init_completion(&kobj->kobj_unregister);