Merge branches 'pm-cpufreq' and 'pm-cpuidle'

Merge cpufreq and cpuidle updates for 5.16-rc1:

 - Fix cpu->pstate.turbo_freq initialization in intel_pstate (Zhang
   Rui).

 - Make intel_pstate process HWP Guaranteed change notifications from
   the processor (Srinivas Pandruvada).

 - Fix typo in cpufreq.h (Rafael Wysocki).

 - Fix tegra driver to handle BPMP errors properly (Mikko Perttunen).

 - Fix the parameter usage of the newly added perf-domain API (Hector
   Yuan).

 - Minor cleanups to cppc, vexpress and s3c244x drivers (Han Wang,
   Guenter Roeck, and Arnd Bergmann).

 - Fix kobject memory leaks in cpuidle error paths (Anel Orazgaliyeva).

 - Make intel_idle enable interrupts before entering C1 on some Xeon
   processor models (Artem Bityutskiy).

* pm-cpufreq:
  cpufreq: Fix parameter in parse_perf_domain()
  cpufreq: intel_pstate: Fix cpu->pstate.turbo_freq initialization
  cpufreq: Fix typo in cpufreq.h
  cpufreq: intel_pstate: Process HWP Guaranteed change notification
  cpufreq: tegra186/tegra194: Handle errors in BPMP response
  cpufreq: remove useless INIT_LIST_HEAD()
  cpufreq: s3c244x: add fallthrough comments for switch
  cpufreq: vexpress: Drop unused variable

* pm-cpuidle:
  cpuidle: Fix kobject memory leaks in error paths
  intel_idle: enable interrupts before C1 on Xeons

drivers/cpufreq/cppc_cpufreq.c

@@ -741,8 +741,6 @@ static int __init cppc_cpufreq_init(void)
 	if ((acpi_disabled) || !acpi_cpc_valid())
 		return -ENODEV;
 
-	INIT_LIST_HEAD(&cpu_data_list);
-
 	cppc_check_hisi_workaround();
 	cppc_freq_invariance_init();
 

drivers/cpufreq/intel_pstate.c

@@ -32,6 +32,7 @@
 #include <asm/cpu_device_id.h>
 #include <asm/cpufeature.h>
 #include <asm/intel-family.h>
+#include "../drivers/thermal/intel/thermal_interrupt.h"
 
 #define INTEL_PSTATE_SAMPLING_INTERVAL	(10 * NSEC_PER_MSEC)
 
@@ -219,6 +220,7 @@ struct global_params {
  * @sched_flags:	Store scheduler flags for possible cross CPU update
  * @hwp_boost_min:	Last HWP boosted min performance
  * @suspended:		Whether or not the driver has been suspended.
+ * @hwp_notify_work:	workqueue for HWP notifications.
  *
  * This structure stores per CPU instance data for all CPUs.
  */
@@ -257,6 +259,7 @@ struct cpudata {
 	unsigned int sched_flags;
 	u32 hwp_boost_min;
 	bool suspended;
+	struct delayed_work hwp_notify_work;
 };
 
 static struct cpudata **all_cpu_data;
@@ -537,7 +540,8 @@ static void intel_pstate_hybrid_hwp_adjust(struct cpudata *cpu)
 	 * scaling factor is too high, so recompute it to make the HWP_CAP
 	 * highest performance correspond to the maximum turbo frequency.
 	 */
-	if (turbo_freq < cpu->pstate.turbo_pstate * scaling) {
+	cpu->pstate.turbo_freq = cpu->pstate.turbo_pstate * scaling;
+	if (turbo_freq < cpu->pstate.turbo_freq) {
 		cpu->pstate.turbo_freq = turbo_freq;
 		scaling = DIV_ROUND_UP(turbo_freq, cpu->pstate.turbo_pstate);
 		cpu->pstate.scaling = scaling;
@@ -985,11 +989,15 @@ static void intel_pstate_hwp_set(unsigned int cpu)
 	wrmsrl_on_cpu(cpu, MSR_HWP_REQUEST, value);
 }
 
+static void intel_pstate_disable_hwp_interrupt(struct cpudata *cpudata);
+
 static void intel_pstate_hwp_offline(struct cpudata *cpu)
 {
 	u64 value = READ_ONCE(cpu->hwp_req_cached);
 	int min_perf;
 
+	intel_pstate_disable_hwp_interrupt(cpu);
+
 	if (boot_cpu_has(X86_FEATURE_HWP_EPP)) {
 		/*
 		 * In case the EPP has been set to "performance" by the
@@ -1053,6 +1061,9 @@ static int intel_pstate_suspend(struct cpufreq_policy *policy)
 
 	cpu->suspended = true;
 
+	/* disable HWP interrupt and cancel any pending work */
+	intel_pstate_disable_hwp_interrupt(cpu);
+
 	return 0;
 }
 
@@ -1546,15 +1557,105 @@ static void intel_pstate_sysfs_hide_hwp_dynamic_boost(void)
 
 /************************** sysfs end ************************/
 
+static void intel_pstate_notify_work(struct work_struct *work)
+{
+	struct cpudata *cpudata =
+		container_of(to_delayed_work(work), struct cpudata, hwp_notify_work);
+
+	cpufreq_update_policy(cpudata->cpu);
+	wrmsrl_on_cpu(cpudata->cpu, MSR_HWP_STATUS, 0);
+}
+
+static DEFINE_SPINLOCK(hwp_notify_lock);
+static cpumask_t hwp_intr_enable_mask;
+
+void notify_hwp_interrupt(void)
+{
+	unsigned int this_cpu = smp_processor_id();
+	struct cpudata *cpudata;
+	unsigned long flags;
+	u64 value;
+
+	if (!READ_ONCE(hwp_active) || !boot_cpu_has(X86_FEATURE_HWP_NOTIFY))
+		return;
+
+	rdmsrl_safe(MSR_HWP_STATUS, &value);
+	if (!(value & 0x01))
+		return;
+
+	spin_lock_irqsave(&hwp_notify_lock, flags);
+
+	if (!cpumask_test_cpu(this_cpu, &hwp_intr_enable_mask))
+		goto ack_intr;
+
+	/*
+	 * Currently we never free all_cpu_data. And we can't reach here
+	 * without this allocated. But for safety for future changes, added
+	 * check.
+	 */
+	if (unlikely(!READ_ONCE(all_cpu_data)))
+		goto ack_intr;
+
+	/*
+	 * The free is done during cleanup, when cpufreq registry is failed.
+	 * We wouldn't be here if it fails on init or switch status. But for
+	 * future changes, added check.
+	 */
+	cpudata = READ_ONCE(all_cpu_data[this_cpu]);
+	if (unlikely(!cpudata))
+		goto ack_intr;
+
+	schedule_delayed_work(&cpudata->hwp_notify_work, msecs_to_jiffies(10));
+
+	spin_unlock_irqrestore(&hwp_notify_lock, flags);
+
+	return;
+
+ack_intr:
+	wrmsrl_safe(MSR_HWP_STATUS, 0);
+	spin_unlock_irqrestore(&hwp_notify_lock, flags);
+}
+
+static void intel_pstate_disable_hwp_interrupt(struct cpudata *cpudata)
+{
+	unsigned long flags;
+
+	/* wrmsrl_on_cpu has to be outside spinlock as this can result in IPC */
+	wrmsrl_on_cpu(cpudata->cpu, MSR_HWP_INTERRUPT, 0x00);
+
+	spin_lock_irqsave(&hwp_notify_lock, flags);
+	if (cpumask_test_and_clear_cpu(cpudata->cpu, &hwp_intr_enable_mask))
+		cancel_delayed_work(&cpudata->hwp_notify_work);
+	spin_unlock_irqrestore(&hwp_notify_lock, flags);
+}
+
+static void intel_pstate_enable_hwp_interrupt(struct cpudata *cpudata)
+{
+	/* Enable HWP notification interrupt for guaranteed performance change */
+	if (boot_cpu_has(X86_FEATURE_HWP_NOTIFY)) {
+		unsigned long flags;
+
+		spin_lock_irqsave(&hwp_notify_lock, flags);
+		INIT_DELAYED_WORK(&cpudata->hwp_notify_work, intel_pstate_notify_work);
+		cpumask_set_cpu(cpudata->cpu, &hwp_intr_enable_mask);
+		spin_unlock_irqrestore(&hwp_notify_lock, flags);
+
+		/* wrmsrl_on_cpu has to be outside spinlock as this can result in IPC */
+		wrmsrl_on_cpu(cpudata->cpu, MSR_HWP_INTERRUPT, 0x01);
+	}
+}
+
 static void intel_pstate_hwp_enable(struct cpudata *cpudata)
 {
-	/* First disable HWP notification interrupt as we don't process them */
+	/* First disable HWP notification interrupt till we activate again */
 	if (boot_cpu_has(X86_FEATURE_HWP_NOTIFY))
 		wrmsrl_on_cpu(cpudata->cpu, MSR_HWP_INTERRUPT, 0x00);
 
 	wrmsrl_on_cpu(cpudata->cpu, MSR_PM_ENABLE, 0x1);
 	if (cpudata->epp_default == -EINVAL)
 		cpudata->epp_default = intel_pstate_get_epp(cpudata, 0);
+
+	intel_pstate_enable_hwp_interrupt(cpudata);
 }
 
 static int atom_get_min_pstate(void)
@@ -2266,7 +2367,7 @@ static int intel_pstate_init_cpu(unsigned int cpunum)
 		if (!cpu)
 			return -ENOMEM;
 
-		all_cpu_data[cpunum] = cpu;
+		WRITE_ONCE(all_cpu_data[cpunum], cpu);
 
 		cpu->cpu = cpunum;
 
@@ -2929,8 +3030,10 @@ static void intel_pstate_driver_cleanup(void)
 			if (intel_pstate_driver == &intel_pstate)
 				intel_pstate_clear_update_util_hook(cpu);
 
+			spin_lock(&hwp_notify_lock);
 			kfree(all_cpu_data[cpu]);
-			all_cpu_data[cpu] = NULL;
+			WRITE_ONCE(all_cpu_data[cpu], NULL);
+			spin_unlock(&hwp_notify_lock);
 		}
 	}
 	cpus_read_unlock();
@@ -3199,6 +3302,7 @@ static bool intel_pstate_hwp_is_enabled(void)
 
 static int __init intel_pstate_init(void)
 {
+	static struct cpudata **_all_cpu_data;
 	const struct x86_cpu_id *id;
 	int rc;
 
@@ -3224,7 +3328,7 @@ static int __init intel_pstate_init(void)
 		 * deal with it.
 		 */
 		if ((!no_hwp && boot_cpu_has(X86_FEATURE_HWP_EPP)) || hwp_forced) {
-			hwp_active++;
+			WRITE_ONCE(hwp_active, 1);
 			hwp_mode_bdw = id->driver_data;
 			intel_pstate.attr = hwp_cpufreq_attrs;
 			intel_cpufreq.attr = hwp_cpufreq_attrs;
@@ -3275,10 +3379,12 @@ static int __init intel_pstate_init(void)
 
 	pr_info("Intel P-state driver initializing\n");
 
-	all_cpu_data = vzalloc(array_size(sizeof(void *), num_possible_cpus()));
-	if (!all_cpu_data)
+	_all_cpu_data = vzalloc(array_size(sizeof(void *), num_possible_cpus()));
+	if (!_all_cpu_data)
 		return -ENOMEM;
 
+	WRITE_ONCE(all_cpu_data, _all_cpu_data);
+
 	intel_pstate_request_control_from_smm();
 
 	intel_pstate_sysfs_expose_params();

drivers/cpufreq/s3c2440-cpufreq.c

@@ -173,12 +173,14 @@ static void s3c2440_cpufreq_setdivs(struct s3c_cpufreq_config *cfg)
 
 	case 6:
 		camdiv |= S3C2440_CAMDIVN_HCLK3_HALF;
+		fallthrough;
 	case 3:
 		clkdiv |= S3C2440_CLKDIVN_HDIVN_3_6;
 		break;
 
 	case 8:
 		camdiv |= S3C2440_CAMDIVN_HCLK4_HALF;
+		fallthrough;
 	case 4:
 		clkdiv |= S3C2440_CLKDIVN_HDIVN_4_8;
 		break;

drivers/cpufreq/tegra186-cpufreq.c

@@ -159,6 +159,10 @@ static struct cpufreq_frequency_table *init_vhint_table(
 		table = ERR_PTR(err);
 		goto free;
 	}
+	if (msg.rx.ret) {
+		table = ERR_PTR(-EINVAL);
+		goto free;
+	}
 
 	for (i = data->vfloor; i <= data->vceil; i++) {
 		u16 ndiv = data->ndiv[i];

drivers/cpufreq/tegra194-cpufreq.c

@@ -242,7 +242,7 @@ static int tegra194_cpufreq_init(struct cpufreq_policy *policy)
 
 	smp_call_function_single(policy->cpu, get_cpu_cluster, &cl, true);
 
-	if (cl >= data->num_clusters)
+	if (cl >= data->num_clusters || !data->tables[cl])
 		return -EINVAL;
 
 	/* set same policy for all cpus in a cluster */
@@ -310,6 +310,12 @@ init_freq_table(struct platform_device *pdev, struct tegra_bpmp *bpmp,
 	err = tegra_bpmp_transfer(bpmp, &msg);
 	if (err)
 		return ERR_PTR(err);
+	if (msg.rx.ret == -BPMP_EINVAL) {
+		/* Cluster not available */
+		return NULL;
+	}
+	if (msg.rx.ret)
+		return ERR_PTR(-EINVAL);
 
 	/*
 	 * Make sure frequency table step is a multiple of mdiv to match

drivers/cpuidle/sysfs.c

@@ -488,6 +488,7 @@ static int cpuidle_add_state_sysfs(struct cpuidle_device *device)
 					   &kdev->kobj, "state%d", i);
 		if (ret) {
 			kobject_put(&kobj->kobj);
+			kfree(kobj);
 			goto error_state;
 		}
 		cpuidle_add_s2idle_attr_group(kobj);
@@ -619,6 +620,7 @@ static int cpuidle_add_driver_sysfs(struct cpuidle_device *dev)
 				   &kdev->kobj, "driver");
 	if (ret) {
 		kobject_put(&kdrv->kobj);
+		kfree(kdrv);
 		return ret;
 	}
 
@@ -705,7 +707,6 @@ int cpuidle_add_sysfs(struct cpuidle_device *dev)
 	if (!kdev)
 		return -ENOMEM;
 	kdev->dev = dev;
-	dev->kobj_dev = kdev;
 
 	init_completion(&kdev->kobj_unregister);
 
@@ -713,9 +714,11 @@ int cpuidle_add_sysfs(struct cpuidle_device *dev)
 				   "cpuidle");
 	if (error) {
 		kobject_put(&kdev->kobj);
+		kfree(kdev);
 		return error;
 	}
 
+	dev->kobj_dev = kdev;
 	kobject_uevent(&kdev->kobj, KOBJ_ADD);
 
 	return 0;

drivers/idle/intel_idle.c

@@ -88,6 +88,12 @@ static struct cpuidle_state *cpuidle_state_table __initdata;
 
 static unsigned int mwait_substates __initdata;
 
+/*
+ * Enable interrupts before entering the C-state. On some platforms and for
+ * some C-states, this may measurably decrease interrupt latency.
+ */
+#define CPUIDLE_FLAG_IRQ_ENABLE		BIT(14)
+
 /*
  * Enable this state by default even if the ACPI _CST does not list it.
  */
@@ -127,6 +133,9 @@ static __cpuidle int intel_idle(struct cpuidle_device *dev,
 	unsigned long eax = flg2MWAIT(state->flags);
 	unsigned long ecx = 1; /* break on interrupt flag */
 
+	if (state->flags & CPUIDLE_FLAG_IRQ_ENABLE)
+		local_irq_enable();
+
 	mwait_idle_with_hints(eax, ecx);
 
 	return index;
@@ -698,7 +707,7 @@ static struct cpuidle_state skx_cstates[] __initdata = {
 	{
 		.name = "C1",
 		.desc = "MWAIT 0x00",
-		.flags = MWAIT2flg(0x00),
+		.flags = MWAIT2flg(0x00) | CPUIDLE_FLAG_IRQ_ENABLE,
 		.exit_latency = 2,
 		.target_residency = 2,
 		.enter = &intel_idle,
@@ -727,7 +736,7 @@ static struct cpuidle_state icx_cstates[] __initdata = {
 	{
 		.name = "C1",
 		.desc = "MWAIT 0x00",
-		.flags = MWAIT2flg(0x00),
+		.flags = MWAIT2flg(0x00) | CPUIDLE_FLAG_IRQ_ENABLE,
 		.exit_latency = 1,
 		.target_residency = 1,
 		.enter = &intel_idle,

include/linux/cpufreq.h

@@ -385,7 +385,7 @@ struct cpufreq_driver {
 /* flags */
 
 /*
- * Set by drivers that need to update internale upper and lower boundaries along
+ * Set by drivers that need to update internal upper and lower boundaries along
  * with the target frequency and so the core and governors should also invoke
  * the diver if the target frequency does not change, but the policy min or max
  * may have changed.
@@ -1041,7 +1041,7 @@ static inline int of_perf_domain_get_sharing_cpumask(int pcpu, const char *list_
 		if (cpu == pcpu)
 			continue;
 
-		ret = parse_perf_domain(pcpu, list_name, cell_name);
+		ret = parse_perf_domain(cpu, list_name, cell_name);
 		if (ret < 0)
 			continue;