cpuidle: Move dev->last_residency update to driver enter routine; remove dev->last_state

Cpuidle governor only suggests the state to enter using the
governor->select() interface, but allows the low level driver to
override the recommended state. The actual entered state
may be different because of software or hardware demotion. Software
demotion is done by the back-end cpuidle driver and can be accounted
correctly. Current cpuidle code uses last_state field to capture the
actual state entered and based on that updates the statistics for the
state entered.

Ideally the driver enter routine should update the counters,
and it should return the state actually entered rather than the time
spent there. The generic cpuidle code should simply handle where
the counters live in the sysfs namespace, not updating the counters.

Reference:
https://lkml.org/lkml/2011/3/25/52

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>

arch/arm/mach-at91/cpuidle.c

@@ -33,7 +33,7 @@ static struct cpuidle_driver at91_idle_driver = {
 
 /* Actual code that puts the SoC in different idle states */
 static int at91_enter_idle(struct cpuidle_device *dev,
-			       struct cpuidle_state *state)
+			       int index)
 {
 	struct timeval before, after;
 	int idle_time;
@@ -41,10 +41,10 @@ static int at91_enter_idle(struct cpuidle_device *dev,
 
 	local_irq_disable();
 	do_gettimeofday(&before);
-	if (state == &dev->states[0])
+	if (index == 0)
 		/* Wait for interrupt state */
 		cpu_do_idle();
-	else if (state == &dev->states[1]) {
+	else if (index == 1) {
 		asm("b 1f; .align 5; 1:");
 		asm("mcr p15, 0, r0, c7, c10, 4");	/* drain write buffer */
 		saved_lpr = sdram_selfrefresh_enable();
@@ -55,7 +55,9 @@ static int at91_enter_idle(struct cpuidle_device *dev,
 	local_irq_enable();
 	idle_time = (after.tv_sec - before.tv_sec) * USEC_PER_SEC +
 			(after.tv_usec - before.tv_usec);
-	return idle_time;
+
+	dev->last_residency = idle_time;
+	return index;
 }
 
 /* Initialize CPU idle by registering the idle states */

arch/arm/mach-davinci/cpuidle.c

@@ -78,9 +78,9 @@ static struct davinci_ops davinci_states[DAVINCI_CPUIDLE_MAX_STATES] = {
 
 /* Actual code that puts the SoC in different idle states */
 static int davinci_enter_idle(struct cpuidle_device *dev,
-						struct cpuidle_state *state)
+						int index)
 {
-	struct davinci_ops *ops = cpuidle_get_statedata(state);
+	struct davinci_ops *ops = cpuidle_get_statedata(&dev->states[index]);
 	struct timeval before, after;
 	int idle_time;
 
@@ -98,7 +98,10 @@ static int davinci_enter_idle(struct cpuidle_device *dev,
 	local_irq_enable();
 	idle_time = (after.tv_sec - before.tv_sec) * USEC_PER_SEC +
 			(after.tv_usec - before.tv_usec);
-	return idle_time;
+
+	dev->last_residency = idle_time;
+
+	return index;
 }
 
 static int __init davinci_cpuidle_probe(struct platform_device *pdev)

arch/arm/mach-exynos4/cpuidle.c

@@ -16,7 +16,7 @@
 #include <asm/proc-fns.h>
 
 static int exynos4_enter_idle(struct cpuidle_device *dev,
-			      struct cpuidle_state *state);
+			      int index);
 
 static struct cpuidle_state exynos4_cpuidle_set[] = {
 	[0] = {
@@ -37,7 +37,7 @@ static struct cpuidle_driver exynos4_idle_driver = {
 };
 
 static int exynos4_enter_idle(struct cpuidle_device *dev,
-			      struct cpuidle_state *state)
+			      int index)
 {
 	struct timeval before, after;
 	int idle_time;
@@ -52,7 +52,8 @@ static int exynos4_enter_idle(struct cpuidle_device *dev,
 	idle_time = (after.tv_sec - before.tv_sec) * USEC_PER_SEC +
 		    (after.tv_usec - before.tv_usec);
 
-	return idle_time;
+	dev->last_residency = idle_time;
+	return index;
 }
 
 static int __init exynos4_init_cpuidle(void)

arch/arm/mach-kirkwood/cpuidle.c

@@ -32,17 +32,17 @@ static DEFINE_PER_CPU(struct cpuidle_device, kirkwood_cpuidle_device);
 
 /* Actual code that puts the SoC in different idle states */
 static int kirkwood_enter_idle(struct cpuidle_device *dev,
-			       struct cpuidle_state *state)
+			       int index)
 {
 	struct timeval before, after;
 	int idle_time;
 
 	local_irq_disable();
 	do_gettimeofday(&before);
-	if (state == &dev->states[0])
+	if (index == 0)
 		/* Wait for interrupt state */
 		cpu_do_idle();
-	else if (state == &dev->states[1]) {
+	else if (index == 1) {
 		/*
 		 * Following write will put DDR in self refresh.
 		 * Note that we have 256 cycles before DDR puts it
@@ -57,7 +57,11 @@ static int kirkwood_enter_idle(struct cpuidle_device *dev,
 	local_irq_enable();
 	idle_time = (after.tv_sec - before.tv_sec) * USEC_PER_SEC +
 			(after.tv_usec - before.tv_usec);
-	return idle_time;
+
+	/* Update last residency */
+	dev->last_residency = idle_time;
+
+	return index;
 }
 
 /* Initialize CPU idle by registering the idle states */

arch/arm/mach-omap2/cpuidle34xx.c

@@ -88,17 +88,19 @@ static int _cpuidle_deny_idle(struct powerdomain *pwrdm,
 /**
  * omap3_enter_idle - Programs OMAP3 to enter the specified state
  * @dev: cpuidle device
- * @state: The target state to be programmed
+ * @index: the index of state to be entered
  *
  * Called from the CPUidle framework to program the device to the
  * specified target state selected by the governor.
  */
 static int omap3_enter_idle(struct cpuidle_device *dev,
-			struct cpuidle_state *state)
+				int index)
 {
-	struct omap3_idle_statedata *cx = cpuidle_get_statedata(state);
+	struct omap3_idle_statedata *cx =
+			cpuidle_get_statedata(&dev->states[index]);
 	struct timespec ts_preidle, ts_postidle, ts_idle;
 	u32 mpu_state = cx->mpu_state, core_state = cx->core_state;
+	int idle_time;
 
 	/* Used to keep track of the total time in idle */
 	getnstimeofday(&ts_preidle);
@@ -113,7 +115,7 @@ static int omap3_enter_idle(struct cpuidle_device *dev,
 		goto return_sleep_time;
 
 	/* Deny idle for C1 */
-	if (state == &dev->states[0]) {
+	if (index == 0) {
 		pwrdm_for_each_clkdm(mpu_pd, _cpuidle_deny_idle);
 		pwrdm_for_each_clkdm(core_pd, _cpuidle_deny_idle);
 	}
@@ -122,7 +124,7 @@ static int omap3_enter_idle(struct cpuidle_device *dev,
 	omap_sram_idle();
 
 	/* Re-allow idle for C1 */
-	if (state == &dev->states[0]) {
+	if (index == 0) {
 		pwrdm_for_each_clkdm(mpu_pd, _cpuidle_allow_idle);
 		pwrdm_for_each_clkdm(core_pd, _cpuidle_allow_idle);
 	}
@@ -134,28 +136,35 @@ return_sleep_time:
 	local_irq_enable();
 	local_fiq_enable();
 
-	return ts_idle.tv_nsec / NSEC_PER_USEC + ts_idle.tv_sec * USEC_PER_SEC;
+	idle_time = ts_idle.tv_nsec / NSEC_PER_USEC + ts_idle.tv_sec * \
+								USEC_PER_SEC;
+
+	/* Update cpuidle counters */
+	dev->last_residency = idle_time;
+
+	return index;
 }
 
 /**
  * next_valid_state - Find next valid C-state
  * @dev: cpuidle device
- * @state: Currently selected C-state
+ * @index: Index of currently selected c-state
  *
- * If the current state is valid, it is returned back to the caller.
- * Else, this function searches for a lower c-state which is still
- * valid.
+ * If the state corresponding to index is valid, index is returned back
+ * to the caller. Else, this function searches for a lower c-state which is
+ * still valid (as defined in omap3_power_states[]) and returns its index.
  *
  * A state is valid if the 'valid' field is enabled and
  * if it satisfies the enable_off_mode condition.
  */
-static struct cpuidle_state *next_valid_state(struct cpuidle_device *dev,
-					      struct cpuidle_state *curr)
+static int next_valid_state(struct cpuidle_device *dev,
+				int index)
 {
-	struct cpuidle_state *next = NULL;
+	struct cpuidle_state *curr = &dev->states[index];
 	struct omap3_idle_statedata *cx = cpuidle_get_statedata(curr);
 	u32 mpu_deepest_state = PWRDM_POWER_RET;
 	u32 core_deepest_state = PWRDM_POWER_RET;
+	int next_index = -1;
 
 	if (enable_off_mode) {
 		mpu_deepest_state = PWRDM_POWER_OFF;
@@ -172,20 +181,20 @@ static struct cpuidle_state *next_valid_state(struct cpuidle_device *dev,
 	if ((cx->valid) &&
 	    (cx->mpu_state >= mpu_deepest_state) &&
 	    (cx->core_state >= core_deepest_state)) {
-		return curr;
+		return index;
 	} else {
 		int idx = OMAP3_NUM_STATES - 1;
 
 		/* Reach the current state starting at highest C-state */
 		for (; idx >= 0; idx--) {
 			if (&dev->states[idx] == curr) {
-				next = &dev->states[idx];
+				next_index = idx;
 				break;
 			}
 		}
 
 		/* Should never hit this condition */
-		WARN_ON(next == NULL);
+		WARN_ON(next_index == -1);
 
 		/*
 		 * Drop to next valid state.
@@ -197,37 +206,39 @@ static struct cpuidle_state *next_valid_state(struct cpuidle_device *dev,
 			if ((cx->valid) &&
 			    (cx->mpu_state >= mpu_deepest_state) &&
 			    (cx->core_state >= core_deepest_state)) {
-				next = &dev->states[idx];
+				next_index = idx;
 				break;
 			}
 		}
 		/*
 		 * C1 is always valid.
-		 * So, no need to check for 'next==NULL' outside this loop.
+		 * So, no need to check for 'next_index == -1' outside
+		 * this loop.
 		 */
 	}
 
-	return next;
+	return next_index;
 }
 
 /**
  * omap3_enter_idle_bm - Checks for any bus activity
  * @dev: cpuidle device
- * @state: The target state to be programmed
+ * @index: array index of target state to be programmed
  *
  * This function checks for any pending activity and then programs
  * the device to the specified or a safer state.
  */
 static int omap3_enter_idle_bm(struct cpuidle_device *dev,
-			       struct cpuidle_state *state)
+			       int index)
 {
-	struct cpuidle_state *new_state;
+	struct cpuidle_state *state = &dev->states[index];
+	int new_state_idx;
 	u32 core_next_state, per_next_state = 0, per_saved_state = 0, cam_state;
 	struct omap3_idle_statedata *cx;
 	int ret;
 
 	if (!omap3_can_sleep()) {
-		new_state = dev->safe_state;
+		new_state_idx = dev->safe_state_index;
 		goto select_state;
 	}
 
@@ -237,7 +248,7 @@ static int omap3_enter_idle_bm(struct cpuidle_device *dev,
 	 */
 	cam_state = pwrdm_read_pwrst(cam_pd);
 	if (cam_state == PWRDM_POWER_ON) {
-		new_state = dev->safe_state;
+		new_state_idx = dev->safe_state_index;
 		goto select_state;
 	}
 
@@ -264,11 +275,10 @@ static int omap3_enter_idle_bm(struct cpuidle_device *dev,
 	if (per_next_state != per_saved_state)
 		pwrdm_set_next_pwrst(per_pd, per_next_state);
 
-	new_state = next_valid_state(dev, state);
+	new_state_idx = next_valid_state(dev, index);
 
 select_state:
-	dev->last_state = new_state;
-	ret = omap3_enter_idle(dev, new_state);
+	ret = omap3_enter_idle(dev, new_state_idx);
 
 	/* Restore original PER state if it was modified */
 	if (per_next_state != per_saved_state)
@@ -339,11 +349,12 @@ int __init omap3_idle_init(void)
 
 	cpuidle_register_driver(&omap3_idle_driver);
 	dev = &per_cpu(omap3_idle_dev, smp_processor_id());
+	dev->safe_state_index = -1;
 
 	/* C1 . MPU WFI + Core active */
 	cx = _fill_cstate(dev, 0, "MPU ON + CORE ON");
 	(&dev->states[0])->enter = omap3_enter_idle;
-	dev->safe_state = &dev->states[0];
+	dev->safe_state_index = 0;
 	cx->valid = 1;	/* C1 is always valid */
 	cx->mpu_state = PWRDM_POWER_ON;
 	cx->core_state = PWRDM_POWER_ON;

arch/sh/kernel/cpu/shmobile/cpuidle.c

@@ -25,11 +25,11 @@ static unsigned long cpuidle_mode[] = {
 };
 
 static int cpuidle_sleep_enter(struct cpuidle_device *dev,
-			       struct cpuidle_state *state)
+				int index)
 {
 	unsigned long allowed_mode = arch_hwblk_sleep_mode();
 	ktime_t before, after;
-	int requested_state = state - &dev->states[0];
+	int requested_state = index;
 	int allowed_state;
 	int k;
 
@@ -46,11 +46,13 @@ static int cpuidle_sleep_enter(struct cpuidle_device *dev,
 	 */
 	k = min_t(int, allowed_state, requested_state);
 
-	dev->last_state = &dev->states[k];
 	before = ktime_get();
 	sh_mobile_call_standby(cpuidle_mode[k]);
 	after = ktime_get();
-	return ktime_to_ns(ktime_sub(after, before)) >> 10;
+
+	dev->last_residency = (int)ktime_to_ns(ktime_sub(after, before)) >> 10;
+
+	return k;
 }
 
 static struct cpuidle_device cpuidle_dev;
@@ -84,7 +86,7 @@ void sh_mobile_setup_cpuidle(void)
 	state->flags |= CPUIDLE_FLAG_TIME_VALID;
 	state->enter = cpuidle_sleep_enter;
 
-	dev->safe_state = state;
+	dev->safe_state_index = i-1;
 
 	if (sh_mobile_sleep_supported & SUSP_SH_SF) {
 		state = &dev->states[i++];

drivers/acpi/processor_idle.c

@@ -741,22 +741,24 @@ static inline void acpi_idle_do_entry(struct acpi_processor_cx *cx)
 /**
  * acpi_idle_enter_c1 - enters an ACPI C1 state-type
  * @dev: the target CPU
- * @state: the state data
+ * @index: index of target state
  *
  * This is equivalent to the HALT instruction.
  */
 static int acpi_idle_enter_c1(struct cpuidle_device *dev,
-			      struct cpuidle_state *state)
+				int index)
 {
 	ktime_t  kt1, kt2;
 	s64 idle_time;
 	struct acpi_processor *pr;
+	struct cpuidle_state *state = &dev->states[index];
 	struct acpi_processor_cx *cx = cpuidle_get_statedata(state);
 
 	pr = __this_cpu_read(processors);
+	dev->last_residency = 0;
 
 	if (unlikely(!pr))
-		return 0;
+		return -EINVAL;
 
 	local_irq_disable();
 
@@ -764,7 +766,7 @@ static int acpi_idle_enter_c1(struct cpuidle_device *dev,
 	if (acpi_idle_suspend) {
 		local_irq_enable();
 		cpu_relax();
-		return 0;
+		return -EINVAL;
 	}
 
 	lapic_timer_state_broadcast(pr, cx, 1);
@@ -773,37 +775,46 @@ static int acpi_idle_enter_c1(struct cpuidle_device *dev,
 	kt2 = ktime_get_real();
 	idle_time =  ktime_to_us(ktime_sub(kt2, kt1));
 
+	/* Update device last_residency*/
+	dev->last_residency = (int)idle_time;
+
 	local_irq_enable();
 	cx->usage++;
 	lapic_timer_state_broadcast(pr, cx, 0);
 
-	return idle_time;
+	return index;
 }
 
 /**
  * acpi_idle_enter_simple - enters an ACPI state without BM handling
  * @dev: the target CPU
- * @state: the state data
+ * @index: the index of suggested state
  */
 static int acpi_idle_enter_simple(struct cpuidle_device *dev,
-				  struct cpuidle_state *state)
+				int index)
 {
 	struct acpi_processor *pr;
+	struct cpuidle_state *state = &dev->states[index];
 	struct acpi_processor_cx *cx = cpuidle_get_statedata(state);
 	ktime_t  kt1, kt2;
 	s64 idle_time_ns;
 	s64 idle_time;
 
 	pr = __this_cpu_read(processors);
+	dev->last_residency = 0;
 
 	if (unlikely(!pr))
-		return 0;
-
-	if (acpi_idle_suspend)
-		return(acpi_idle_enter_c1(dev, state));
+		return -EINVAL;
 
 	local_irq_disable();
 
+	if (acpi_idle_suspend) {
+		local_irq_enable();
+		cpu_relax();
+		return -EINVAL;
+	}
+
+
 	if (cx->entry_method != ACPI_CSTATE_FFH) {
 		current_thread_info()->status &= ~TS_POLLING;
 		/*
@@ -815,7 +826,7 @@ static int acpi_idle_enter_simple(struct cpuidle_device *dev,
 		if (unlikely(need_resched())) {
 			current_thread_info()->status |= TS_POLLING;
 			local_irq_enable();
-			return 0;
+			return -EINVAL;
 		}
 	}
 
@@ -837,6 +848,9 @@ static int acpi_idle_enter_simple(struct cpuidle_device *dev,
 	idle_time = idle_time_ns;
 	do_div(idle_time, NSEC_PER_USEC);
 
+	/* Update device last_residency*/
+	dev->last_residency = (int)idle_time;
+
 	/* Tell the scheduler how much we idled: */
 	sched_clock_idle_wakeup_event(idle_time_ns);
 
@@ -848,7 +862,7 @@ static int acpi_idle_enter_simple(struct cpuidle_device *dev,
 
 	lapic_timer_state_broadcast(pr, cx, 0);
 	cx->time += idle_time;
-	return idle_time;
+	return index;
 }
 
 static int c3_cpu_count;
@@ -857,14 +871,15 @@ static DEFINE_SPINLOCK(c3_lock);
 /**
  * acpi_idle_enter_bm - enters C3 with proper BM handling
  * @dev: the target CPU
- * @state: the state data
+ * @index: the index of suggested state
  *
  * If BM is detected, the deepest non-C3 idle state is entered instead.
  */
 static int acpi_idle_enter_bm(struct cpuidle_device *dev,
-			      struct cpuidle_state *state)
+				int index)
 {
 	struct acpi_processor *pr;
+	struct cpuidle_state *state = &dev->states[index];
 	struct acpi_processor_cx *cx = cpuidle_get_statedata(state);
 	ktime_t  kt1, kt2;
 	s64 idle_time_ns;
@@ -872,22 +887,26 @@ static int acpi_idle_enter_bm(struct cpuidle_device *dev,
 
 
 	pr = __this_cpu_read(processors);
+	dev->last_residency = 0;
 
 	if (unlikely(!pr))
-		return 0;
+		return -EINVAL;
 
-	if (acpi_idle_suspend)
-		return(acpi_idle_enter_c1(dev, state));
+
+	if (acpi_idle_suspend) {
+		cpu_relax();
+		return -EINVAL;
+	}
 
 	if (!cx->bm_sts_skip && acpi_idle_bm_check()) {
-		if (dev->safe_state) {
-			dev->last_state = dev->safe_state;
-			return dev->safe_state->enter(dev, dev->safe_state);
+		if (dev->safe_state_index >= 0) {
+			return dev->states[dev->safe_state_index].enter(dev,
+						dev->safe_state_index);
 		} else {
 			local_irq_disable();
 			acpi_safe_halt();
 			local_irq_enable();
-			return 0;
+			return -EINVAL;
 		}
 	}
 
@@ -904,7 +923,7 @@ static int acpi_idle_enter_bm(struct cpuidle_device *dev,
 		if (unlikely(need_resched())) {
 			current_thread_info()->status |= TS_POLLING;
 			local_irq_enable();
-			return 0;
+			return -EINVAL;
 		}
 	}
 
@@ -954,6 +973,9 @@ static int acpi_idle_enter_bm(struct cpuidle_device *dev,
 	idle_time = idle_time_ns;
 	do_div(idle_time, NSEC_PER_USEC);
 
+	/* Update device last_residency*/
+	dev->last_residency = (int)idle_time;
+
 	/* Tell the scheduler how much we idled: */
 	sched_clock_idle_wakeup_event(idle_time_ns);
 
@@ -965,7 +987,7 @@ static int acpi_idle_enter_bm(struct cpuidle_device *dev,
 
 	lapic_timer_state_broadcast(pr, cx, 0);
 	cx->time += idle_time;
-	return idle_time;
+	return index;
 }
 
 struct cpuidle_driver acpi_idle_driver = {
@@ -992,6 +1014,7 @@ static int acpi_processor_setup_cpuidle(struct acpi_processor *pr)
 	}
 
 	dev->cpu = pr->id;
+	dev->safe_state_index = -1;
 	for (i = 0; i < CPUIDLE_STATE_MAX; i++) {
 		dev->states[i].name[0] = '\0';
 		dev->states[i].desc[0] = '\0';
@@ -1027,13 +1050,13 @@ static int acpi_processor_setup_cpuidle(struct acpi_processor *pr)
 				state->flags |= CPUIDLE_FLAG_TIME_VALID;
 
 			state->enter = acpi_idle_enter_c1;
-			dev->safe_state = state;
+			dev->safe_state_index = count;
 			break;
 
 			case ACPI_STATE_C2:
 			state->flags |= CPUIDLE_FLAG_TIME_VALID;
 			state->enter = acpi_idle_enter_simple;
-			dev->safe_state = state;
+			dev->safe_state_index = count;
 			break;
 
 			case ACPI_STATE_C3:

drivers/cpuidle/cpuidle.c

@@ -62,7 +62,7 @@ int cpuidle_idle_call(void)
 {
 	struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
 	struct cpuidle_state *target_state;
-	int next_state;
+	int next_state, entered_state;
 
 	if (off)
 		return -ENODEV;
@@ -102,26 +102,27 @@ int cpuidle_idle_call(void)
 
 	target_state = &dev->states[next_state];
 
-	/* enter the state and update stats */
-	dev->last_state = target_state;
-
 	trace_power_start(POWER_CSTATE, next_state, dev->cpu);
 	trace_cpu_idle(next_state, dev->cpu);
 
-	dev->last_residency = target_state->enter(dev, target_state);
+	entered_state = target_state->enter(dev, next_state);
 
 	trace_power_end(dev->cpu);
 	trace_cpu_idle(PWR_EVENT_EXIT, dev->cpu);
 
-	if (dev->last_state)
-		target_state = dev->last_state;
-
-	target_state->time += (unsigned long long)dev->last_residency;
-	target_state->usage++;
+	if (entered_state >= 0) {
+		/* Update cpuidle counters */
+		/* This can be moved to within driver enter routine
+		 * but that results in multiple copies of same code.
+		 */
+		dev->states[entered_state].time +=
+				(unsigned long long)dev->last_residency;
+		dev->states[entered_state].usage++;
+	}
 
 	/* give the governor an opportunity to reflect on the outcome */
 	if (cpuidle_curr_governor->reflect)
-		cpuidle_curr_governor->reflect(dev);
+		cpuidle_curr_governor->reflect(dev, entered_state);
 
 	return 0;
 }
@@ -172,11 +173,10 @@ void cpuidle_resume_and_unlock(void)
 EXPORT_SYMBOL_GPL(cpuidle_resume_and_unlock);
 
 #ifdef CONFIG_ARCH_HAS_CPU_RELAX
-static int poll_idle(struct cpuidle_device *dev, struct cpuidle_state *st)
+static int poll_idle(struct cpuidle_device *dev, int index)
 {
 	ktime_t	t1, t2;
 	s64 diff;
-	int ret;
 
 	t1 = ktime_get();
 	local_irq_enable();
@@ -188,8 +188,9 @@ static int poll_idle(struct cpuidle_device *dev, struct cpuidle_state *st)
 	if (diff > INT_MAX)
 		diff = INT_MAX;
 
-	ret = (int) diff;
-	return ret;
+	dev->last_residency = (int) diff;
+
+	return index;
 }
 
 static void poll_idle_init(struct cpuidle_device *dev)
@@ -248,7 +249,6 @@ int cpuidle_enable_device(struct cpuidle_device *dev)
 		dev->states[i].time = 0;
 	}
 	dev->last_residency = 0;
-	dev->last_state = NULL;
 
 	smp_wmb();
 

drivers/cpuidle/governors/ladder.c

@@ -153,11 +153,24 @@ static int ladder_enable_device(struct cpuidle_device *dev)
 	return 0;
 }
 
+/**
+ * ladder_reflect - update the correct last_state_idx
+ * @dev: the CPU
+ * @index: the index of actual state entered
+ */
+static void ladder_reflect(struct cpuidle_device *dev, int index)
+{
+	struct ladder_device *ldev = &__get_cpu_var(ladder_devices);
+	if (index > 0)
+		ldev->last_state_idx = index;
+}
+
 static struct cpuidle_governor ladder_governor = {
 	.name =		"ladder",
 	.rating =	10,
 	.enable =	ladder_enable_device,
 	.select =	ladder_select_state,
+	.reflect =	ladder_reflect,
 	.owner =	THIS_MODULE,
 };
 

drivers/cpuidle/governors/menu.c

@@ -310,14 +310,17 @@ static int menu_select(struct cpuidle_device *dev)
 /**
  * menu_reflect - records that data structures need update
  * @dev: the CPU
+ * @index: the index of actual entered state
  *
  * NOTE: it's important to be fast here because this operation will add to
  *       the overall exit latency.
  */
-static void menu_reflect(struct cpuidle_device *dev)
+static void menu_reflect(struct cpuidle_device *dev, int index)
 {
 	struct menu_device *data = &__get_cpu_var(menu_devices);
-	data->needs_update = 1;
+	data->last_state_idx = index;
+	if (index >= 0)
+		data->needs_update = 1;
 }
 
 /**

drivers/idle/intel_idle.c

@@ -81,7 +81,7 @@ static unsigned int mwait_substates;
 static unsigned int lapic_timer_reliable_states = (1 << 1);	 /* Default to only C1 */
 
 static struct cpuidle_device __percpu *intel_idle_cpuidle_devices;
-static int intel_idle(struct cpuidle_device *dev, struct cpuidle_state *state);
+static int intel_idle(struct cpuidle_device *dev, int index);
 
 static struct cpuidle_state *cpuidle_state_table;
 
@@ -209,12 +209,13 @@ static struct cpuidle_state atom_cstates[MWAIT_MAX_NUM_CSTATES] = {
 /**
  * intel_idle
  * @dev: cpuidle_device
- * @state: cpuidle state
+ * @index: index of cpuidle state
  *
  */
-static int intel_idle(struct cpuidle_device *dev, struct cpuidle_state *state)
+static int intel_idle(struct cpuidle_device *dev, int index)
 {
 	unsigned long ecx = 1; /* break on interrupt flag */
+	struct cpuidle_state *state = &dev->states[index];
 	unsigned long eax = (unsigned long)cpuidle_get_statedata(state);
 	unsigned int cstate;
 	ktime_t kt_before, kt_after;
@@ -256,7 +257,10 @@ static int intel_idle(struct cpuidle_device *dev, struct cpuidle_state *state)
 	if (!(lapic_timer_reliable_states & (1 << (cstate))))
 		clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_EXIT, &cpu);
 
-	return usec_delta;
+	/* Update cpuidle counters */
+	dev->last_residency = (int)usec_delta;
+
+	return index;
 }
 
 static void __setup_broadcast_timer(void *arg)

include/linux/cpuidle.h

@@ -42,7 +42,7 @@ struct cpuidle_state {
 	unsigned long long	time; /* in US */
 
 	int (*enter)	(struct cpuidle_device *dev,
-			 struct cpuidle_state *state);
+			int index);
 };
 
 /* Idle State Flags */
@@ -87,13 +87,12 @@ struct cpuidle_device {
 	int			state_count;
 	struct cpuidle_state	states[CPUIDLE_STATE_MAX];
 	struct cpuidle_state_kobj *kobjs[CPUIDLE_STATE_MAX];
-	struct cpuidle_state	*last_state;
 
 	struct list_head 	device_list;
 	struct kobject		kobj;
 	struct completion	kobj_unregister;
 	void			*governor_data;
-	struct cpuidle_state	*safe_state;
+	int			safe_state_index;
 
 	int (*prepare)		(struct cpuidle_device *dev);
 };
@@ -169,7 +168,7 @@ struct cpuidle_governor {
 	void (*disable)		(struct cpuidle_device *dev);
 
 	int  (*select)		(struct cpuidle_device *dev);
-	void (*reflect)		(struct cpuidle_device *dev);
+	void (*reflect)		(struct cpuidle_device *dev, int index);
 
 	struct module 		*owner;
 };