Merge tag 'pm-5.7-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull power management updates from Rafael Wysocki:
 "These clean up and rework the PM QoS API, address a suspend-to-idle
  wakeup regression on some ACPI-based platforms, clean up and extend a
  few cpuidle drivers, update multiple cpufreq drivers and cpufreq
  documentation, and fix a number of issues in devfreq and several other
  things all over.

  Specifics:

   - Clean up and rework the PM QoS API to simplify the code and reduce
     the size of it (Rafael Wysocki).

   - Fix a suspend-to-idle wakeup regression on Dell XPS13 9370 and
     similar platforms where the USB plug/unplug events are handled by
     the EC (Rafael Wysocki).

   - CLean up the intel_idle and PSCI cpuidle drivers (Rafael Wysocki,
     Ulf Hansson).

   - Extend the haltpoll cpuidle driver so that it can be forced to run
     on some systems where it refused to load (Maciej Szmigiero).

   - Convert several cpufreq documents to the .rst format and move the
     legacy driver documentation into one common file (Mauro Carvalho
     Chehab, Rafael Wysocki).

   - Update several cpufreq drivers:

        * Extend and fix the imx-cpufreq-dt driver (Anson Huang).

        * Improve the -EPROBE_DEFER handling and fix unwanted CPU
          overclocking on i.MX6ULL in imx6q-cpufreq (Anson Huang,
          Christoph Niedermaier).

        * Add support for Krait based SoCs to the qcom driver (Ansuel
          Smith).

        * Add support for OPP_PLUS to ti-cpufreq (Lokesh Vutla).

        * Add platform specific intermediate callbacks support to
          cpufreq-dt and update the imx6q driver (Peng Fan).

        * Simplify and consolidate some pieces of the intel_pstate
          driver and update its documentation (Rafael Wysocki, Alex
          Hung).

   - Fix several devfreq issues:

        * Remove unneeded extern keyword from a devfreq header file and
          use the DEVFREQ_GOV_UPDATE_INTERNAL event name instead of
          DEVFREQ_GOV_INTERNAL (Chanwoo Choi).

        * Fix the handling of dev_pm_qos_remove_request() result
          (Leonard Crestez).

        * Use constant name for userspace governor (Pierre Kuo).

        * Get rid of doc warnings and fix a typo (Christophe JAILLET).

   - Use built-in RCU list checking in some places in the PM core to
     avoid false-positive RCU usage warnings (Madhuparna Bhowmik).

   - Add explicit READ_ONCE()/WRITE_ONCE() annotations to low-level PM
     QoS routines (Qian Cai).

   - Fix removal of wakeup sources to avoid NULL pointer dereferences in
     a corner case (Neeraj Upadhyay).

   - Clean up the handling of hibernate compat ioctls and fix the
     related documentation (Eric Biggers).

   - Update the idle_inject power capping driver to use variable-length
     arrays instead of zero-length arrays (Gustavo Silva).

   - Fix list format in a PM QoS document (Randy Dunlap).

   - Make the cpufreq stats module use scnprintf() to avoid potential
     buffer overflows (Takashi Iwai).

   - Add pm_runtime_get_if_active() to PM-runtime API (Sakari Ailus).

   - Allow no domain-idle-states DT property in generic PM domains (Ulf
     Hansson).

   - Fix a broken y-axis scale in the intel_pstate_tracer utility (Doug
     Smythies)"

* tag 'pm-5.7-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (78 commits)
  cpufreq: intel_pstate: Simplify intel_pstate_cpu_init()
  tools/power/x86/intel_pstate_tracer: fix a broken y-axis scale
  ACPI: PM: s2idle: Refine active GPEs check
  ACPICA: Allow acpi_any_gpe_status_set() to skip one GPE
  PM: sleep: wakeup: Skip wakeup_source_sysfs_remove() if device is not there
  PM / devfreq: Get rid of some doc warnings
  PM / devfreq: Fix handling dev_pm_qos_remove_request result
  PM / devfreq: Fix a typo in a comment
  PM / devfreq: Change to DEVFREQ_GOV_UPDATE_INTERVAL event name
  PM / devfreq: Remove unneeded extern keyword
  PM / devfreq: Use constant name of userspace governor
  ACPI: PM: s2idle: Fix comment in acpi_s2idle_prepare_late()
  cpufreq: qcom: Add support for krait based socs
  cpufreq: imx6q-cpufreq: Improve the logic of -EPROBE_DEFER handling
  cpufreq: Use scnprintf() for avoiding potential buffer overflow
  cpuidle: psci: Split psci_dt_cpu_init_idle()
  PM / Domains: Allow no domain-idle-states DT property in genpd when parsing
  PM / hibernate: Remove unnecessary compat ioctl overrides
  PM: hibernate: fix docs for ioctls that return loff_t via pointer
  Documentation: intel_pstate: update links for references
  ...

4 files changed, 39 insertions, 61 deletions

diff --git a/drivers/cpuidle/cpuidle-haltpoll.c b/drivers/cpuidle/cpuidle-haltpoll.c
index b0ce9bc78113..db124bc1ca2c 100644
--- a/drivers/cpuidle/cpuidle-haltpoll.c
+++ b/drivers/cpuidle/cpuidle-haltpoll.c
@@ -18,6 +18,10 @@
 #include <linux/kvm_para.h>
 #include <linux/cpuidle_haltpoll.h>
 
+static bool force __read_mostly;
+module_param(force, bool, 0444);
+MODULE_PARM_DESC(force, "Load unconditionally");
+
 static struct cpuidle_device __percpu *haltpoll_cpuidle_devices;
 static enum cpuhp_state haltpoll_hp_state;
 
@@ -90,6 +94,11 @@ static void haltpoll_uninit(void)
 	haltpoll_cpuidle_devices = NULL;
 }
 
+static bool haltpool_want(void)
+{
+	return kvm_para_has_hint(KVM_HINTS_REALTIME) || force;
+}
+
 static int __init haltpoll_init(void)
 {
 	int ret;
@@ -101,8 +110,7 @@ static int __init haltpoll_init(void)
 
 	cpuidle_poll_state_init(drv);
 
-	if (!kvm_para_available() ||
-		!kvm_para_has_hint(KVM_HINTS_REALTIME))
+	if (!kvm_para_available() || !haltpool_want())
 		return -ENODEV;
 
 	ret = cpuidle_register_driver(drv);
diff --git a/drivers/cpuidle/cpuidle-psci.c b/drivers/cpuidle/cpuidle-psci.c
index edd7a54ef0d3..bae9140a65a5 100644
--- a/drivers/cpuidle/cpuidle-psci.c
+++ b/drivers/cpuidle/cpuidle-psci.c
@@ -160,6 +160,29 @@ int __init psci_dt_parse_state_node(struct device_node *np, u32 *state)
 	return 0;
 }
 
+static int __init psci_dt_cpu_init_topology(struct cpuidle_driver *drv,
+					    struct psci_cpuidle_data *data,
+					    unsigned int state_count, int cpu)
+{
+	/* Currently limit the hierarchical topology to be used in OSI mode. */
+	if (!psci_has_osi_support())
+		return 0;
+
+	data->dev = psci_dt_attach_cpu(cpu);
+	if (IS_ERR_OR_NULL(data->dev))
+		return PTR_ERR_OR_ZERO(data->dev);
+
+	/*
+	 * Using the deepest state for the CPU to trigger a potential selection
+	 * of a shared state for the domain, assumes the domain states are all
+	 * deeper states.
+	 */
+	drv->states[state_count - 1].enter = psci_enter_domain_idle_state;
+	psci_cpuidle_use_cpuhp = true;
+
+	return 0;
+}
+
 static int __init psci_dt_cpu_init_idle(struct cpuidle_driver *drv,
 					struct device_node *cpu_node,
 					unsigned int state_count, int cpu)
@@ -193,25 +216,10 @@ static int __init psci_dt_cpu_init_idle(struct cpuidle_driver *drv,
 		goto free_mem;
 	}
 
-	/* Currently limit the hierarchical topology to be used in OSI mode. */
-	if (psci_has_osi_support()) {
-		data->dev = psci_dt_attach_cpu(cpu);
-		if (IS_ERR(data->dev)) {
-			ret = PTR_ERR(data->dev);
-			goto free_mem;
-		}
-
-		/*
-		 * Using the deepest state for the CPU to trigger a potential
-		 * selection of a shared state for the domain, assumes the
-		 * domain states are all deeper states.
-		 */
-		if (data->dev) {
-			drv->states[state_count - 1].enter =
-				psci_enter_domain_idle_state;
-			psci_cpuidle_use_cpuhp = true;
-		}
-	}
+	/* Initialize optional data, used for the hierarchical topology. */
+	ret = psci_dt_cpu_init_topology(drv, data, state_count, cpu);
+	if (ret < 0)
+		goto free_mem;
 
 	/* Idle states parsed correctly, store them in the per-cpu struct. */
 	data->psci_states = psci_states;
diff --git a/drivers/cpuidle/cpuidle.c b/drivers/cpuidle/cpuidle.c
index de81298051b3..c149d9e20dfd 100644
--- a/drivers/cpuidle/cpuidle.c
+++ b/drivers/cpuidle/cpuidle.c
@@ -736,53 +736,15 @@ int cpuidle_register(struct cpuidle_driver *drv,
 }
 EXPORT_SYMBOL_GPL(cpuidle_register);
 
-#ifdef CONFIG_SMP
-
-/*
- * This function gets called when a part of the kernel has a new latency
- * requirement.  This means we need to get all processors out of their C-state,
- * and then recalculate a new suitable C-state. Just do a cross-cpu IPI; that
- * wakes them all right up.
- */
-static int cpuidle_latency_notify(struct notifier_block *b,
-		unsigned long l, void *v)
-{
-	wake_up_all_idle_cpus();
-	return NOTIFY_OK;
-}
-
-static struct notifier_block cpuidle_latency_notifier = {
-	.notifier_call = cpuidle_latency_notify,
-};
-
-static inline void latency_notifier_init(struct notifier_block *n)
-{
-	pm_qos_add_notifier(PM_QOS_CPU_DMA_LATENCY, n);
-}
-
-#else /* CONFIG_SMP */
-
-#define latency_notifier_init(x) do { } while (0)
-
-#endif /* CONFIG_SMP */
-
 /**
  * cpuidle_init - core initializer
  */
 static int __init cpuidle_init(void)
 {
-	int ret;
-
 	if (cpuidle_disabled())
 		return -ENODEV;
 
-	ret = cpuidle_add_interface(cpu_subsys.dev_root);
-	if (ret)
-		return ret;
-
-	latency_notifier_init(&cpuidle_latency_notifier);
-
-	return 0;
+	return cpuidle_add_interface(cpu_subsys.dev_root);
 }
 
 module_param(off, int, 0444);
diff --git a/drivers/cpuidle/governor.c b/drivers/cpuidle/governor.c
index e48271e117a3..29acaf48e575 100644
--- a/drivers/cpuidle/governor.c
+++ b/drivers/cpuidle/governor.c
@@ -109,9 +109,9 @@ int cpuidle_register_governor(struct cpuidle_governor *gov)
  */
 s64 cpuidle_governor_latency_req(unsigned int cpu)
 {
-	int global_req = pm_qos_request(PM_QOS_CPU_DMA_LATENCY);
 	struct device *device = get_cpu_device(cpu);
 	int device_req = dev_pm_qos_raw_resume_latency(device);
+	int global_req = cpu_latency_qos_limit();
 
 	if (device_req > global_req)
 		device_req = global_req;