More power management updates for 4.19-rc1

- Make the idle loop handle stopped scheduler tick correctly (Rafael
    Wysocki).
 
  - Prevent the menu cpuidle governor from letting CPUs spend too much
    time in shallow idle states when it is invoked with scheduler tick
    stopped and clean it up somewhat (Rafael Wysocki).
 
  - Avoid invoking the platform firmware to make the platform enter
    the ACPI S3 sleep state with suspended PCIe root ports which may
    confuse the firmware and cause it to crash (Rafael Wysocki).
 
  - Fix sysfs-related race in the ondemand and conservative cpufreq
    governors which may cause the system to crash if the governor
    module is removed during an update of CPU frequency limits (Henry
    Willard).
 
  - Select SRCU when building the system wakeup framework to avoid a
    build issue in it (zhangyi).
 
  - Make the descriptions of ACPI C-states vendor-neutral to avoid
    confusion (Prarit Bhargava).
 -----BEGIN PGP SIGNATURE-----
 Version: GnuPG v2
 
 iQIcBAABCAAGBQJbfSnVAAoJEILEb/54YlRxBn4QAKQ8PqkSYkBby+1hb90ET4dk
 VaLkbCYXuzLK5rIDvnbYOALhVKo4B29Ex5GdCLN7cWkZMkrVKe7oX8QQTnp3/7lF
 URjTKgTNec5uJG652PrE3ESAa3X/kYggj6aeQOxDR4iYKzcpJEQ92ekFW+SoJTNp
 Jc2kZh3qkC2On64GB3ibsZaKnmHfPvLg0t4agwzuYq/Gff8NRJFk7kMwAPzqGzZo
 b2UVRcYFWIRkJjgmU9iInoeHIY8mBdT3IiKwTemZP1dOhb5T1AHOXwGTk6/cS+RH
 A9qx4eg7I3R00KmnYvO8WytYJeOu2qb83GIUx4fIJGOqfvevm5xkxB9F+nfE+ouj
 ROBqO4+X4XfQGPw8slayg0rJjI9JSkXLnLdl0Qw2WRlbc4/fVWntra1C57EeKFBR
 EG9UAF9+7nUUx0bOCLsfFF3+r9R3SDUjk7b4thyhYncyQRsYC+FL7ztlxnMzVtzW
 M5SF2sPrpcQzqmcszdUdbESI10n5X8m/crJW4rsbTxBpAM+coO+uLcvHWOY4MpkW
 BgBsR6bMDAlG/VlTFgeeP/tkCRd5zNlJi7yBFItXuOoVKXpnHCJuxq2WkZ1Rb74M
 Gk1d3TduekHJm8VsLEdCJR/tEk1cMc0zVUD/a1yzI4Z21QxvXUCqMDdws4/Ey184
 qmKgNR9R94vSC5xIPRhM
 =9GrU
 -----END PGP SIGNATURE-----

Merge tag 'pm-4.19-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull more power management updates from Rafael Wysocki:
 "These fix the main idle loop and the menu cpuidle governor, clean up
  the latter, fix a mistake in the PCI bus type's support for system
  suspend and resume, fix the ondemand and conservative cpufreq
  governors, address a build issue in the system wakeup framework and
  make the ACPI C-states desciptions less confusing.

  Specifics:

   - Make the idle loop handle stopped scheduler tick correctly (Rafael
     Wysocki).

   - Prevent the menu cpuidle governor from letting CPUs spend too much
     time in shallow idle states when it is invoked with scheduler tick
     stopped and clean it up somewhat (Rafael Wysocki).

   - Avoid invoking the platform firmware to make the platform enter the
     ACPI S3 sleep state with suspended PCIe root ports which may
     confuse the firmware and cause it to crash (Rafael Wysocki).

   - Fix sysfs-related race in the ondemand and conservative cpufreq
     governors which may cause the system to crash if the governor
     module is removed during an update of CPU frequency limits (Henry
     Willard).

   - Select SRCU when building the system wakeup framework to avoid a
     build issue in it (zhangyi).

   - Make the descriptions of ACPI C-states vendor-neutral to avoid
     confusion (Prarit Bhargava)"

* tag 'pm-4.19-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  cpuidle: menu: Handle stopped tick more aggressively
  sched: idle: Avoid retaining the tick when it has been stopped
  PCI / ACPI / PM: Resume all bridges on suspend-to-RAM
  cpuidle: menu: Update stale polling override comment
  cpufreq: governor: Avoid accessing invalid governor_data
  x86/ACPI/cstate: Make APCI C1 FFH MWAIT C-state description vendor-neutral
  cpuidle: menu: Fix white space
  PM / sleep: wakeup: Fix build error caused by missing SRCU support

arch/x86/kernel/acpi/cstate.c

@@ -108,7 +108,7 @@ static long acpi_processor_ffh_cstate_probe_cpu(void *_cx)
 			cx->type);
 	}
 	snprintf(cx->desc,
-			ACPI_CX_DESC_LEN, "ACPI FFH INTEL MWAIT 0x%x",
+			ACPI_CX_DESC_LEN, "ACPI FFH MWAIT 0x%x",
 			cx->address);
 out:
 	return retval;

drivers/cpufreq/cpufreq_governor.c

@@ -555,12 +555,20 @@ EXPORT_SYMBOL_GPL(cpufreq_dbs_governor_stop);
 
 void cpufreq_dbs_governor_limits(struct cpufreq_policy *policy)
 {
-	struct policy_dbs_info *policy_dbs = policy->governor_data;
+	struct policy_dbs_info *policy_dbs;
+
+	/* Protect gov->gdbs_data against cpufreq_dbs_governor_exit() */
+	mutex_lock(&gov_dbs_data_mutex);
+	policy_dbs = policy->governor_data;
+	if (!policy_dbs)
+		goto out;
 
 	mutex_lock(&policy_dbs->update_mutex);
 	cpufreq_policy_apply_limits(policy);
 	gov_update_sample_delay(policy_dbs, 0);
-
 	mutex_unlock(&policy_dbs->update_mutex);
+
+out:
+	mutex_unlock(&gov_dbs_data_mutex);
 }
 EXPORT_SYMBOL_GPL(cpufreq_dbs_governor_limits);

drivers/cpuidle/governors/menu.c

@@ -328,9 +328,8 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 		unsigned int polling_threshold;
 
 		/*
-		 * We want to default to C1 (hlt), not to busy polling
-		 * unless the timer is happening really really soon, or
-		 * C1's exit latency exceeds the user configured limit.
+		 * 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 &&
@@ -349,14 +348,12 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 		 * If the tick is already stopped, the cost of possible short
 		 * idle duration misprediction is much higher, because the CPU
 		 * may be stuck in a shallow idle state for a long time as a
-		 * result of it.  In that case say we might mispredict and try
-		 * to force the CPU into a state for which we would have stopped
-		 * the tick, unless a timer is going to expire really soon
-		 * anyway.
+		 * result of it.  In that case say we might mispredict and use
+		 * the known time till the closest timer event for the idle
+		 * state selection.
 		 */
 		if (data->predicted_us < TICK_USEC)
-			data->predicted_us = min_t(unsigned int, TICK_USEC,
-						   ktime_to_us(delta_next));
+			data->predicted_us = ktime_to_us(delta_next);
 	} else {
 		/*
 		 * Use the performance multiplier and the user-configurable
@@ -381,8 +378,22 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 			continue;
 		if (idx == -1)
 			idx = i; /* first enabled state */
-		if (s->target_residency > data->predicted_us)
-			break;
+		if (s->target_residency > data->predicted_us) {
+			if (!tick_nohz_tick_stopped())
+				break;
+
+			/*
+			 * If the state selected so far is shallow and this
+			 * state's target residency matches the time till the
+			 * closest timer event, select this one to avoid getting
+			 * stuck in the shallow one for too long.
+			 */
+			if (drv->states[idx].target_residency < TICK_USEC &&
+			    s->target_residency <= ktime_to_us(delta_next))
+				idx = i;
+
+			goto out;
+		}
 		if (s->exit_latency > latency_req) {
 			/*
 			 * If we break out of the loop for latency reasons, use
@@ -403,14 +414,13 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 	 * Don't stop the tick if the selected state is a polling one or if the
 	 * expected idle duration is shorter than the tick period length.
 	 */
-	if ((drv->states[idx].flags & CPUIDLE_FLAG_POLLING) ||
-	    expected_interval < TICK_USEC) {
+	if (((drv->states[idx].flags & CPUIDLE_FLAG_POLLING) ||
+	     expected_interval < TICK_USEC) && !tick_nohz_tick_stopped()) {
 		unsigned int delta_next_us = ktime_to_us(delta_next);
 
 		*stop_tick = false;
 
-		if (!tick_nohz_tick_stopped() && idx > 0 &&
-		    drv->states[idx].target_residency > delta_next_us) {
+		if (idx > 0 && drv->states[idx].target_residency > delta_next_us) {
 			/*
 			 * The tick is not going to be stopped and the target
 			 * residency of the state to be returned is not within
@@ -418,8 +428,8 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 			 * tick, so try to correct that.
 			 */
 			for (i = idx - 1; i >= 0; i--) {
-			    if (drv->states[i].disabled ||
-			        dev->states_usage[i].disable)
+				if (drv->states[i].disabled ||
+				    dev->states_usage[i].disable)
 					continue;
 
 				idx = i;
@@ -429,6 +439,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 		}
 	}
 
+out:
 	data->last_state_idx = idx;
 
 	return data->last_state_idx;

drivers/pci/pci-acpi.c

@@ -615,13 +615,11 @@ static bool acpi_pci_need_resume(struct pci_dev *dev)
 	/*
 	 * In some cases (eg. Samsung 305V4A) leaving a bridge in suspend over
 	 * system-wide suspend/resume confuses the platform firmware, so avoid
-	 * doing that, unless the bridge has a driver that should take care of
-	 * the PM handling.  According to Section 16.1.6 of ACPI 6.2, endpoint
+	 * doing that.  According to Section 16.1.6 of ACPI 6.2, endpoint
 	 * devices are expected to be in D3 before invoking the S3 entry path
 	 * from the firmware, so they should not be affected by this issue.
 	 */
-	if (pci_is_bridge(dev) && !dev->driver &&
-	    acpi_target_system_state() != ACPI_STATE_S0)
+	if (pci_is_bridge(dev) && acpi_target_system_state() != ACPI_STATE_S0)
 		return true;
 
 	if (!adev || !acpi_device_power_manageable(adev))

kernel/power/Kconfig

@@ -105,6 +105,7 @@ config PM_SLEEP
 	def_bool y
 	depends on SUSPEND || HIBERNATE_CALLBACKS
 	select PM
+	select SRCU
 
 config PM_SLEEP_SMP
 	def_bool y

kernel/sched/idle.c

@@ -190,7 +190,7 @@ static void cpuidle_idle_call(void)
 		 */
 		next_state = cpuidle_select(drv, dev, &stop_tick);
 
-		if (stop_tick)
+		if (stop_tick || tick_nohz_tick_stopped())
 			tick_nohz_idle_stop_tick();
 		else
 			tick_nohz_idle_retain_tick();