Merge back earlier cpuidle updates for v5.13.

Author: rafaeljw

drivers/cpuidle/driver.c

@@ -181,9 +181,13 @@ static void __cpuidle_driver_init(struct cpuidle_driver *drv)
 		 */
 		if (s->target_residency > 0)
 			s->target_residency_ns = s->target_residency * NSEC_PER_USEC;
+		else if (s->target_residency_ns < 0)
+			s->target_residency_ns = 0;
 
 		if (s->exit_latency > 0)
 			s->exit_latency_ns = s->exit_latency * NSEC_PER_USEC;
+		else if (s->exit_latency_ns < 0)
+			s->exit_latency_ns =  0;
 	}
 }
 

drivers/cpuidle/governors/menu.c

@@ -271,7 +271,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 	u64 predicted_ns;
 	u64 interactivity_req;
 	unsigned long nr_iowaiters;
-	ktime_t delta_next;
+	ktime_t delta, delta_tick;
 	int i, idx;
 
 	if (data->needs_update) {
@@ -280,7 +280,12 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 	}
 
 	/* determine the expected residency time, round up */
-	data->next_timer_ns = tick_nohz_get_sleep_length(&delta_next);
+	delta = tick_nohz_get_sleep_length(&delta_tick);
+	if (unlikely(delta < 0)) {
+		delta = 0;
+		delta_tick = 0;
+	}
+	data->next_timer_ns = delta;
 
 	nr_iowaiters = nr_iowait_cpu(dev->cpu);
 	data->bucket = which_bucket(data->next_timer_ns, nr_iowaiters);
@@ -318,7 +323,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 		 * state selection.
 		 */
 		if (predicted_ns < TICK_NSEC)
-			predicted_ns = delta_next;
+			predicted_ns = data->next_timer_ns;
 	} else {
 		/*
 		 * Use the performance multiplier and the user-configurable
@@ -377,7 +382,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 			 * stuck in the shallow one for too long.
 			 */
 			if (drv->states[idx].target_residency_ns < TICK_NSEC &&
-			    s->target_residency_ns <= delta_next)
+			    s->target_residency_ns <= delta_tick)
 				idx = i;
 
 			return idx;
@@ -399,7 +404,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 	     predicted_ns < TICK_NSEC) && !tick_nohz_tick_stopped()) {
 		*stop_tick = false;
 
-		if (idx > 0 && drv->states[idx].target_residency_ns > delta_next) {
+		if (idx > 0 && drv->states[idx].target_residency_ns > delta_tick) {
 			/*
 			 * The tick is not going to be stopped and the target
 			 * residency of the state to be returned is not within
@@ -411,7 +416,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 					continue;
 
 				idx = i;
-				if (drv->states[i].target_residency_ns <= delta_next)
+				if (drv->states[i].target_residency_ns <= delta_tick)
 					break;
 			}
 		}

drivers/cpuidle/governors/teo.c

@@ -100,8 +100,8 @@ struct teo_idle_state {
  * @intervals: Saved idle duration values.
  */
 struct teo_cpu {
-	u64 time_span_ns;
-	u64 sleep_length_ns;
+	s64 time_span_ns;
+	s64 sleep_length_ns;
 	struct teo_idle_state states[CPUIDLE_STATE_MAX];
 	int interval_idx;
 	u64 intervals[INTERVALS];
@@ -117,7 +117,8 @@ static DEFINE_PER_CPU(struct teo_cpu, teo_cpus);
 static void teo_update(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 {
 	struct teo_cpu *cpu_data = per_cpu_ptr(&teo_cpus, dev->cpu);
-	int i, idx_hit = -1, idx_timer = -1;
+	int i, idx_hit = 0, idx_timer = 0;
+	unsigned int hits, misses;
 	u64 measured_ns;
 
 	if (cpu_data->time_span_ns >= cpu_data->sleep_length_ns) {
@@ -174,25 +175,22 @@ static void teo_update(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 	 * also increase the "early hits" metric for the state that actually
 	 * matches the measured idle duration.
 	 */
-	if (idx_timer >= 0) {
-		unsigned int hits = cpu_data->states[idx_timer].hits;
-		unsigned int misses = cpu_data->states[idx_timer].misses;
-
-		hits -= hits >> DECAY_SHIFT;
-		misses -= misses >> DECAY_SHIFT;
-
-		if (idx_timer > idx_hit) {
-			misses += PULSE;
-			if (idx_hit >= 0)
-				cpu_data->states[idx_hit].early_hits += PULSE;
-		} else {
-			hits += PULSE;
-		}
+	hits = cpu_data->states[idx_timer].hits;
+	hits -= hits >> DECAY_SHIFT;
+
+	misses = cpu_data->states[idx_timer].misses;
+	misses -= misses >> DECAY_SHIFT;
 
-		cpu_data->states[idx_timer].misses = misses;
-		cpu_data->states[idx_timer].hits = hits;
+	if (idx_timer == idx_hit) {
+		hits += PULSE;
+	} else {
+		misses += PULSE;
+		cpu_data->states[idx_hit].early_hits += PULSE;
 	}
 
+	cpu_data->states[idx_timer].misses = misses;
+	cpu_data->states[idx_timer].hits = hits;
+
 	/*
 	 * Save idle duration values corresponding to non-timer wakeups for
 	 * pattern detection.
@@ -216,7 +214,7 @@ static bool teo_time_ok(u64 interval_ns)
  */
 static int teo_find_shallower_state(struct cpuidle_driver *drv,
 				    struct cpuidle_device *dev, int state_idx,
-				    u64 duration_ns)
+				    s64 duration_ns)
 {
 	int i;
 
@@ -242,10 +240,10 @@ static int teo_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 {
 	struct teo_cpu *cpu_data = per_cpu_ptr(&teo_cpus, dev->cpu);
 	s64 latency_req = cpuidle_governor_latency_req(dev->cpu);
-	u64 duration_ns;
+	int max_early_idx, prev_max_early_idx, constraint_idx, idx0, idx, i;
 	unsigned int hits, misses, early_hits;
-	int max_early_idx, prev_max_early_idx, constraint_idx, idx, i;
 	ktime_t delta_tick;
+	s64 duration_ns;
 
 	if (dev->last_state_idx >= 0) {
 		teo_update(drv, dev);
@@ -264,6 +262,7 @@ static int teo_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 	prev_max_early_idx = -1;
 	constraint_idx = drv->state_count;
 	idx = -1;
+	idx0 = idx;
 
 	for (i = 0; i < drv->state_count; i++) {
 		struct cpuidle_state *s = &drv->states[i];
@@ -324,6 +323,7 @@ static int teo_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 			idx = i; /* first enabled state */
 			hits = cpu_data->states[i].hits;
 			misses = cpu_data->states[i].misses;
+			idx0 = i;
 		}
 
 		if (s->target_residency_ns > duration_ns)
@@ -376,11 +376,16 @@ static int teo_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 
 	if (idx < 0) {
 		idx = 0; /* No states enabled. Must use 0. */
-	} else if (idx > 0) {
+	} else if (idx > idx0) {
 		unsigned int count = 0;
 		u64 sum = 0;
 
 		/*
+		 * The target residencies of at least two different enabled idle
+		 * states are less than or equal to the current expected idle
+		 * duration.  Try to refine the selection using the most recent
+		 * measured idle duration values.
+		 *
 		 * Count and sum the most recent idle duration values less than
 		 * the current expected idle duration value.
 		 */
@@ -428,7 +433,8 @@ static int teo_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 		 * till the closest timer including the tick, try to correct
 		 * that.
 		 */
-		if (idx > 0 && drv->states[idx].target_residency_ns > delta_tick)
+		if (idx > idx0 &&
+		    drv->states[idx].target_residency_ns > delta_tick)
 			idx = teo_find_shallower_state(drv, dev, idx, delta_tick);
 	}
 

drivers/idle/intel_idle.c

@@ -744,8 +744,8 @@ static struct cpuidle_state icx_cstates[] __initdata = {
 		.name = "C6",
 		.desc = "MWAIT 0x20",
 		.flags = MWAIT2flg(0x20) | CPUIDLE_FLAG_TLB_FLUSHED,
-		.exit_latency = 128,
-		.target_residency = 384,
+		.exit_latency = 170,
+		.target_residency = 600,
 		.enter = &intel_idle,
 		.enter_s2idle = intel_idle_s2idle, },
 	{
@@ -1156,6 +1156,7 @@ static const struct x86_cpu_id intel_idle_ids[] __initconst = {
 	X86_MATCH_INTEL_FAM6_MODEL(KABYLAKE,		&idle_cpu_skl),
 	X86_MATCH_INTEL_FAM6_MODEL(SKYLAKE_X,		&idle_cpu_skx),
 	X86_MATCH_INTEL_FAM6_MODEL(ICELAKE_X,		&idle_cpu_icx),
+	X86_MATCH_INTEL_FAM6_MODEL(ICELAKE_D,		&idle_cpu_icx),
 	X86_MATCH_INTEL_FAM6_MODEL(XEON_PHI_KNL,	&idle_cpu_knl),
 	X86_MATCH_INTEL_FAM6_MODEL(XEON_PHI_KNM,	&idle_cpu_knl),
 	X86_MATCH_INTEL_FAM6_MODEL(ATOM_GOLDMONT,	&idle_cpu_bxt),

include/linux/cpuidle.h

@@ -49,8 +49,8 @@ struct cpuidle_state {
 	char		name[CPUIDLE_NAME_LEN];
 	char		desc[CPUIDLE_DESC_LEN];
 
-	u64		exit_latency_ns;
-	u64		target_residency_ns;
+	s64		exit_latency_ns;
+	s64		target_residency_ns;
 	unsigned int	flags;
 	unsigned int	exit_latency; /* in US */
 	int		power_usage; /* in mW */

kernel/time/tick-sched.c

@@ -1124,7 +1124,11 @@ ktime_t tick_nohz_get_next_hrtimer(void)
  * tick_nohz_get_sleep_length - return the expected length of the current sleep
  * @delta_next: duration until the next event if the tick cannot be stopped
  *
- * Called from power state control code with interrupts disabled
+ * Called from power state control code with interrupts disabled.
+ *
+ * The return value of this function and/or the value returned by it through the
+ * @delta_next pointer can be negative which must be taken into account by its
+ * callers.
  */
 ktime_t tick_nohz_get_sleep_length(ktime_t *delta_next)
 {