Merge tag 'pm-4.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull power management updates from Rafael Wysocki:
 "These make hibernation on 32-bit x86 systems work in all of the cases
  in which it works on 64-bit x86 ones, update the menu cpuidle governor
  and the "polling" state to make them more efficient, add more hardware
  support to cpufreq drivers and fix issues with some of them, fix a bug
  in the conservative cpufreq governor, fix the operating performance
  points (OPP) framework and make it more stable, update the devfreq
  subsystem to take changes in the APIs used by into account and clean
  up some things all over.

  Specifics:

   - Backport hibernation bug fixes from x86-64 to x86-32 and
     consolidate hibernation handling on x86 to allow 32-bit systems to
     work in all of the cases in which 64-bit ones work (Zhimin Gu, Chen
     Yu).

   - Fix hibernation documentation (Vladimir D. Seleznev).

   - Update the menu cpuidle governor to fix a couple of issues with it,
     make it more efficient in some cases and clean it up (Rafael
     Wysocki).

   - Rework the cpuidle polling state implementation to make it more
     efficient (Rafael Wysocki).

   - Clean up the cpuidle core somewhat (Fieah Lim).

   - Fix the cpufreq conservative governor to take policy limits into
     account properly in some cases (Rafael Wysocki).

   - Add support for retrieving guaranteed performance information to
     the ACPI CPPC library and make the intel_pstate driver use it to
     expose the CPU base frequency via sysfs on systems with the
     hardware-managed P-states (HWP) feature enabled (Srinivas
     Pandruvada).

   - Fix clang warning in the CPPC cpufreq driver (Nathan Chancellor).

   - Get rid of device_node.name printing from cpufreq (Rob Herring).

   - Remove unnecessary unlikely() from the cpufreq core (Igor Stoppa).

   - Add support for the r8a7744 SoC to the cpufreq-dt driver (Biju
     Das).

   - Update the dt-platdev cpufreq driver to allow RK3399 to have
     separate tunables per cluster (Dmitry Torokhov).

   - Fix the dma_alloc_coherent() usage in the tegra186 cpufreq driver
     (Christoph Hellwig).

   - Make the imx6q cpufreq driver read OCOTP through nvmem for
     imx6ul/imx6ull (Anson Huang).

   - Fix several bugs in the operating performance points (OPP)
     framework and make it more stable (Viresh Kumar, Dave Gerlach).

   - Update the devfreq subsystem to take changes in the APIs used by
     into account, fix some issues with it and make it stop print
     device_node.name directly (Bjorn Andersson, Enric Balletbo i Serra,
     Matthias Kaehlcke, Rob Herring, Vincent Donnefort, zhong jiang).

   - Prepare the generic power domains (genpd) framework for dealing
     with domains containing CPUs (Ulf Hansson).

   - Prevent sysfs attributes representing low-power S0 residency
     counters from being exposed if low-power S0 support is not
     indicated in ACPI FADT (Rajneesh Bhardwaj).

   - Get rid of custom CPU features macros for Intel CPUs from the
     intel_idle and RAPL drivers (Andy Shevchenko).

   - Update the tasks freezer to list tasks that refused to freeze and
     caused a system transition to a sleep state to be aborted (Todd
     Brandt).

   - Update the pm-graph set of tools to v5.2 (Todd Brandt).

   - Fix some issues in the cpupower utility (Anders Roxell, Prarit
     Bhargava)"

* tag 'pm-4.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (73 commits)
  PM / Domains: Document flags for genpd
  PM / Domains: Deal with multiple states but no governor in genpd
  PM / Domains: Don't treat zero found compatible idle states as an error
  cpuidle: menu: Avoid computations when result will be discarded
  cpuidle: menu: Drop redundant comparison
  cpufreq: tegra186: don't pass GFP_DMA32 to dma_alloc_coherent()
  cpufreq: conservative: Take limits changes into account properly
  Documentation: intel_pstate: Add base_frequency information
  cpufreq: intel_pstate: Add base_frequency attribute
  ACPI / CPPC: Add support for guaranteed performance
  cpuidle: menu: Simplify checks related to the polling state
  PM / tools: sleepgraph and bootgraph: upgrade to v5.2
  PM / tools: sleepgraph: first batch of v5.2 changes
  cpupower: Fix coredump on VMWare
  cpupower: Fix AMD Family 0x17 msr_pstate size
  cpufreq: imx6q: read OCOTP through nvmem for imx6ul/imx6ull
  cpufreq: dt-platdev: allow RK3399 to have separate tunables per cluster
  cpuidle: poll_state: Revise loop termination condition
  cpuidle: menu: Move the latency_req == 0 special case check
  cpuidle: menu: Avoid computations for very close timers
  ...

1 files changed, 65 insertions, 49 deletions

diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c
index e26a409..575a68f 100644
--- a/drivers/cpuidle/governors/menu.c
+++ b/drivers/cpuidle/governors/menu.c
@@ -124,7 +124,6 @@ struct menu_device {
 	int             tick_wakeup;
 
 	unsigned int	next_timer_us;
-	unsigned int	predicted_us;
 	unsigned int	bucket;
 	unsigned int	correction_factor[BUCKETS];
 	unsigned int	intervals[INTERVALS];
@@ -197,10 +196,11 @@ static void menu_update(struct cpuidle_driver *drv, struct cpuidle_device *dev);
  * of points is below a threshold. If it is... then use the
  * average of these 8 points as the estimated value.
  */
-static unsigned int get_typical_interval(struct menu_device *data)
+static unsigned int get_typical_interval(struct menu_device *data,
+					 unsigned int predicted_us)
 {
 	int i, divisor;
-	unsigned int max, thresh, avg;
+	unsigned int min, max, thresh, avg;
 	uint64_t sum, variance;
 
 	thresh = UINT_MAX; /* Discard outliers above this value */
@@ -208,6 +208,7 @@ static unsigned int get_typical_interval(struct menu_device *data)
 again:
 
 	/* First calculate the average of past intervals */
+	min = UINT_MAX;
 	max = 0;
 	sum = 0;
 	divisor = 0;
@@ -218,8 +219,19 @@ again:
 			divisor++;
 			if (value > max)
 				max = value;
+
+			if (value < min)
+				min = value;
 		}
 	}
+
+	/*
+	 * If the result of the computation is going to be discarded anyway,
+	 * avoid the computation altogether.
+	 */
+	if (min >= predicted_us)
+		return UINT_MAX;
+
 	if (divisor == INTERVALS)
 		avg = sum >> INTERVAL_SHIFT;
 	else
@@ -286,10 +298,9 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 	struct menu_device *data = this_cpu_ptr(&menu_devices);
 	int latency_req = cpuidle_governor_latency_req(dev->cpu);
 	int i;
-	int first_idx;
 	int idx;
 	unsigned int interactivity_req;
-	unsigned int expected_interval;
+	unsigned int predicted_us;
 	unsigned long nr_iowaiters, cpu_load;
 	ktime_t delta_next;
 
@@ -298,50 +309,36 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 		data->needs_update = 0;
 	}
 
-	/* Special case when user has set very strict latency requirement */
-	if (unlikely(latency_req == 0)) {
-		*stop_tick = false;
-		return 0;
-	}
-
 	/* determine the expected residency time, round up */
 	data->next_timer_us = ktime_to_us(tick_nohz_get_sleep_length(&delta_next));
 
 	get_iowait_load(&nr_iowaiters, &cpu_load);
 	data->bucket = which_bucket(data->next_timer_us, nr_iowaiters);
 
+	if (unlikely(drv->state_count <= 1 || latency_req == 0) ||
+	    ((data->next_timer_us < drv->states[1].target_residency ||
+	      latency_req < drv->states[1].exit_latency) &&
+	     !drv->states[0].disabled && !dev->states_usage[0].disable)) {
+		/*
+		 * In this case state[0] will be used no matter what, so return
+		 * it right away and keep the tick running.
+		 */
+		*stop_tick = false;
+		return 0;
+	}
+
 	/*
 	 * Force the result of multiplication to be 64 bits even if both
 	 * operands are 32 bits.
 	 * Make sure to round up for half microseconds.
 	 */
-	data->predicted_us = DIV_ROUND_CLOSEST_ULL((uint64_t)data->next_timer_us *
+	predicted_us = DIV_ROUND_CLOSEST_ULL((uint64_t)data->next_timer_us *
 					 data->correction_factor[data->bucket],
 					 RESOLUTION * DECAY);
-
-	expected_interval = get_typical_interval(data);
-	expected_interval = min(expected_interval, data->next_timer_us);
-
-	first_idx = 0;
-	if (drv->states[0].flags & CPUIDLE_FLAG_POLLING) {
-		struct cpuidle_state *s = &drv->states[1];
-		unsigned int polling_threshold;
-
-		/*
-		 * Default to a physical idle state, not to busy polling, unless
-		 * a timer is going to trigger really really soon.
-		 */
-		polling_threshold = max_t(unsigned int, 20, s->target_residency);
-		if (data->next_timer_us > polling_threshold &&
-		    latency_req > s->exit_latency && !s->disabled &&
-		    !dev->states_usage[1].disable)
-			first_idx = 1;
-	}
-
 	/*
 	 * Use the lowest expected idle interval to pick the idle state.
 	 */
-	data->predicted_us = min(data->predicted_us, expected_interval);
+	predicted_us = min(predicted_us, get_typical_interval(data, predicted_us));
 
 	if (tick_nohz_tick_stopped()) {
 		/*
@@ -352,34 +349,46 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 		 * the known time till the closest timer event for the idle
 		 * state selection.
 		 */
-		if (data->predicted_us < TICK_USEC)
-			data->predicted_us = ktime_to_us(delta_next);
+		if (predicted_us < TICK_USEC)
+			predicted_us = ktime_to_us(delta_next);
 	} else {
 		/*
 		 * Use the performance multiplier and the user-configurable
 		 * latency_req to determine the maximum exit latency.
 		 */
-		interactivity_req = data->predicted_us / performance_multiplier(nr_iowaiters, cpu_load);
+		interactivity_req = predicted_us / performance_multiplier(nr_iowaiters, cpu_load);
 		if (latency_req > interactivity_req)
 			latency_req = interactivity_req;
 	}
 
-	expected_interval = data->predicted_us;
 	/*
 	 * Find the idle state with the lowest power while satisfying
 	 * our constraints.
 	 */
 	idx = -1;
-	for (i = first_idx; i < drv->state_count; i++) {
+	for (i = 0; i < drv->state_count; i++) {
 		struct cpuidle_state *s = &drv->states[i];
 		struct cpuidle_state_usage *su = &dev->states_usage[i];
 
 		if (s->disabled || su->disable)
 			continue;
+
 		if (idx == -1)
 			idx = i; /* first enabled state */
-		if (s->target_residency > data->predicted_us) {
-			if (data->predicted_us < TICK_USEC)
+
+		if (s->target_residency > predicted_us) {
+			/*
+			 * Use a physical idle state, not busy polling, unless
+			 * a timer is going to trigger soon enough.
+			 */
+			if ((drv->states[idx].flags & CPUIDLE_FLAG_POLLING) &&
+			    s->exit_latency <= latency_req &&
+			    s->target_residency <= data->next_timer_us) {
+				predicted_us = s->target_residency;
+				idx = i;
+				break;
+			}
+			if (predicted_us < TICK_USEC)
 				break;
 
 			if (!tick_nohz_tick_stopped()) {
@@ -389,7 +398,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 				 * tick in that case and let the governor run
 				 * again in the next iteration of the loop.
 				 */
-				expected_interval = drv->states[idx].target_residency;
+				predicted_us = drv->states[idx].target_residency;
 				break;
 			}
 
@@ -403,7 +412,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 			    s->target_residency <= ktime_to_us(delta_next))
 				idx = i;
 
-			goto out;
+			return idx;
 		}
 		if (s->exit_latency > latency_req) {
 			/*
@@ -412,7 +421,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 			 * expected idle duration so that the tick is retained
 			 * as long as that target residency is low enough.
 			 */
-			expected_interval = drv->states[idx].target_residency;
+			predicted_us = drv->states[idx].target_residency;
 			break;
 		}
 		idx = i;
@@ -426,7 +435,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 	 * expected idle duration is shorter than the tick period length.
 	 */
 	if (((drv->states[idx].flags & CPUIDLE_FLAG_POLLING) ||
-	     expected_interval < TICK_USEC) && !tick_nohz_tick_stopped()) {
+	     predicted_us < TICK_USEC) && !tick_nohz_tick_stopped()) {
 		unsigned int delta_next_us = ktime_to_us(delta_next);
 
 		*stop_tick = false;
@@ -450,10 +459,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 		}
 	}
 
-out:
-	data->last_state_idx = idx;
-
-	return data->last_state_idx;
+	return idx;
 }
 
 /**
@@ -512,9 +518,19 @@ static void menu_update(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 		 * duration predictor do a better job next time.
 		 */
 		measured_us = 9 * MAX_INTERESTING / 10;
+	} else if ((drv->states[last_idx].flags & CPUIDLE_FLAG_POLLING) &&
+		   dev->poll_time_limit) {
+		/*
+		 * The CPU exited the "polling" state due to a time limit, so
+		 * the idle duration prediction leading to the selection of that
+		 * state was inaccurate.  If a better prediction had been made,
+		 * the CPU might have been woken up from idle by the next timer.
+		 * Assume that to be the case.
+		 */
+		measured_us = data->next_timer_us;
 	} else {
 		/* measured value */
-		measured_us = cpuidle_get_last_residency(dev);
+		measured_us = dev->last_residency;
 
 		/* Deduct exit latency */
 		if (measured_us > 2 * target->exit_latency)