Merge tag 'pm-5.11-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull more power management updates from Rafael Wysocki:
 "These update the CPPC cpufreq driver and intel_pstate (which involves
  updating the cpufreq core and the schedutil governor) and make
  janitorial changes in the ACPI code handling processor objects.

  Specifics:

   - Rework the passive-mode "fast switch" path in the intel_pstate
     driver to allow it receive the minimum (required) and target
     (desired) performance information from the schedutil governor so as
     to avoid running some workloads too fast (Rafael Wysocki).

   - Make the intel_pstate driver allow the policy max limit to be
     increased after the guaranteed performance value for the given CPU
     has increased (Rafael Wysocki).

   - Clean up the handling of CPU coordination types in the CPPC cpufreq
     driver and make it export frequency domains information to user
     space via sysfs (Ionela Voinescu).

   - Fix the ACPI code handling processor objects to use a correct
     coordination type when it fails to map frequency domains and drop a
     redundant CPU map initialization from it (Ionela Voinescu, Punit
     Agrawal)"

* tag 'pm-5.11-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  cpufreq: intel_pstate: Use most recent guaranteed performance values
  cpufreq: intel_pstate: Implement the ->adjust_perf() callback
  cpufreq: Add special-purpose fast-switching callback for drivers
  cpufreq: schedutil: Add util to struct sg_cpu
  cppc_cpufreq: replace per-cpu data array with a list
  cppc_cpufreq: expose information on frequency domains
  cppc_cpufreq: clarify support for coordination types
  cppc_cpufreq: use policy->cpu as driver of frequency setting
  ACPI: processor: fix NONE coordination for domain mapping failure

Documentation/ABI/testing/sysfs-devices-system-cpu

@@ -264,7 +264,8 @@ Description:	Discover CPUs in the same CPU frequency coordination domain
 		attribute is useful for user space DVFS controllers to get better
 		power/performance results for platforms using acpi-cpufreq.
 
-		This file is only present if the acpi-cpufreq driver is in use.
+		This file is only present if the acpi-cpufreq or the cppc-cpufreq
+		drivers are in use.
 
 
 What:		/sys/devices/system/cpu/cpu*/cache/index3/cache_disable_{0,1}

drivers/acpi/cppc_acpi.c

@@ -414,109 +414,88 @@ static int acpi_get_psd(struct cpc_desc *cpc_ptr, acpi_handle handle)
 	return result;
 }
 
+bool acpi_cpc_valid(void)
+{
+	struct cpc_desc *cpc_ptr;
+	int cpu;
+
+	for_each_possible_cpu(cpu) {
+		cpc_ptr = per_cpu(cpc_desc_ptr, cpu);
+		if (!cpc_ptr)
+			return false;
+	}
+
+	return true;
+}
+EXPORT_SYMBOL_GPL(acpi_cpc_valid);
+
 /**
- * acpi_get_psd_map - Map the CPUs in a common freq domain.
- * @all_cpu_data: Ptrs to CPU specific CPPC data including PSD info.
+ * acpi_get_psd_map - Map the CPUs in the freq domain of a given cpu
+ * @cpu: Find all CPUs that share a domain with cpu.
+ * @cpu_data: Pointer to CPU specific CPPC data including PSD info.
  *
  *	Return: 0 for success or negative value for err.
  */
-int acpi_get_psd_map(struct cppc_cpudata **all_cpu_data)
+int acpi_get_psd_map(unsigned int cpu, struct cppc_cpudata *cpu_data)
 {
-	int count_target;
-	int retval = 0;
-	unsigned int i, j;
-	cpumask_var_t covered_cpus;
-	struct cppc_cpudata *pr, *match_pr;
-	struct acpi_psd_package *pdomain;
-	struct acpi_psd_package *match_pdomain;
 	struct cpc_desc *cpc_ptr, *match_cpc_ptr;
-
-	if (!zalloc_cpumask_var(&covered_cpus, GFP_KERNEL))
-		return -ENOMEM;
+	struct acpi_psd_package *match_pdomain;
+	struct acpi_psd_package *pdomain;
+	int count_target, i;
 
 	/*
 	 * Now that we have _PSD data from all CPUs, let's setup P-state
 	 * domain info.
 	 */
-	for_each_possible_cpu(i) {
-		if (cpumask_test_cpu(i, covered_cpus))
-			continue;
-
-		pr = all_cpu_data[i];
-		cpc_ptr = per_cpu(cpc_desc_ptr, i);
-		if (!cpc_ptr) {
-			retval = -EFAULT;
-			goto err_ret;
-		}
+	cpc_ptr = per_cpu(cpc_desc_ptr, cpu);
+	if (!cpc_ptr)
+		return -EFAULT;
 
-		pdomain = &(cpc_ptr->domain_info);
-		cpumask_set_cpu(i, pr->shared_cpu_map);
-		cpumask_set_cpu(i, covered_cpus);
-		if (pdomain->num_processors <= 1)
-			continue;
+	pdomain = &(cpc_ptr->domain_info);
+	cpumask_set_cpu(cpu, cpu_data->shared_cpu_map);
+	if (pdomain->num_processors <= 1)
+		return 0;
 
-		/* Validate the Domain info */
-		count_target = pdomain->num_processors;
-		if (pdomain->coord_type == DOMAIN_COORD_TYPE_SW_ALL)
-			pr->shared_type = CPUFREQ_SHARED_TYPE_ALL;
-		else if (pdomain->coord_type == DOMAIN_COORD_TYPE_HW_ALL)
-			pr->shared_type = CPUFREQ_SHARED_TYPE_HW;
-		else if (pdomain->coord_type == DOMAIN_COORD_TYPE_SW_ANY)
-			pr->shared_type = CPUFREQ_SHARED_TYPE_ANY;
-
-		for_each_possible_cpu(j) {
-			if (i == j)
-				continue;
-
-			match_cpc_ptr = per_cpu(cpc_desc_ptr, j);
-			if (!match_cpc_ptr) {
-				retval = -EFAULT;
-				goto err_ret;
-			}
+	/* Validate the Domain info */
+	count_target = pdomain->num_processors;
+	if (pdomain->coord_type == DOMAIN_COORD_TYPE_SW_ALL)
+		cpu_data->shared_type = CPUFREQ_SHARED_TYPE_ALL;
+	else if (pdomain->coord_type == DOMAIN_COORD_TYPE_HW_ALL)
+		cpu_data->shared_type = CPUFREQ_SHARED_TYPE_HW;
+	else if (pdomain->coord_type == DOMAIN_COORD_TYPE_SW_ANY)
+		cpu_data->shared_type = CPUFREQ_SHARED_TYPE_ANY;
 
-			match_pdomain = &(match_cpc_ptr->domain_info);
-			if (match_pdomain->domain != pdomain->domain)
-				continue;
+	for_each_possible_cpu(i) {
+		if (i == cpu)
+			continue;
 
-			/* Here i and j are in the same domain */
-			if (match_pdomain->num_processors != count_target) {
-				retval = -EFAULT;
-				goto err_ret;
-			}
+		match_cpc_ptr = per_cpu(cpc_desc_ptr, i);
+		if (!match_cpc_ptr)
+			goto err_fault;
 
-			if (pdomain->coord_type != match_pdomain->coord_type) {
-				retval = -EFAULT;
-				goto err_ret;
-			}
+		match_pdomain = &(match_cpc_ptr->domain_info);
+		if (match_pdomain->domain != pdomain->domain)
+			continue;
 
-			cpumask_set_cpu(j, covered_cpus);
-			cpumask_set_cpu(j, pr->shared_cpu_map);
-		}
+		/* Here i and cpu are in the same domain */
+		if (match_pdomain->num_processors != count_target)
+			goto err_fault;
 
-		for_each_cpu(j, pr->shared_cpu_map) {
-			if (i == j)
-				continue;
+		if (pdomain->coord_type != match_pdomain->coord_type)
+			goto err_fault;
 
-			match_pr = all_cpu_data[j];
-			match_pr->shared_type = pr->shared_type;
-			cpumask_copy(match_pr->shared_cpu_map,
-				     pr->shared_cpu_map);
-		}
+		cpumask_set_cpu(i, cpu_data->shared_cpu_map);
 	}
-	goto out;
 
-err_ret:
-	for_each_possible_cpu(i) {
-		pr = all_cpu_data[i];
+	return 0;
 
-		/* Assume no coordination on any error parsing domain info */
-		cpumask_clear(pr->shared_cpu_map);
-		cpumask_set_cpu(i, pr->shared_cpu_map);
-		pr->shared_type = CPUFREQ_SHARED_TYPE_ALL;
-	}
-out:
-	free_cpumask_var(covered_cpus);
-	return retval;
+err_fault:
+	/* Assume no coordination on any error parsing domain info */
+	cpumask_clear(cpu_data->shared_cpu_map);
+	cpumask_set_cpu(cpu, cpu_data->shared_cpu_map);
+	cpu_data->shared_type = CPUFREQ_SHARED_TYPE_NONE;
+
+	return -EFAULT;
 }
 EXPORT_SYMBOL_GPL(acpi_get_psd_map);
 

drivers/acpi/processor_perflib.c

@@ -708,7 +708,7 @@ int acpi_processor_preregister_performance(
 		if (retval) {
 			cpumask_clear(pr->performance->shared_cpu_map);
 			cpumask_set_cpu(i, pr->performance->shared_cpu_map);
-			pr->performance->shared_type = CPUFREQ_SHARED_TYPE_ALL;
+			pr->performance->shared_type = CPUFREQ_SHARED_TYPE_NONE;
 		}
 		pr->performance = NULL; /* Will be set for real in register */
 	}

drivers/cpufreq/cppc_cpufreq.c

@@ -30,13 +30,13 @@
 #define DMI_PROCESSOR_MAX_SPEED		0x14
 
 /*
- * These structs contain information parsed from per CPU
- * ACPI _CPC structures.
- * e.g. For each CPU the highest, lowest supported
- * performance capabilities, desired performance level
- * requested etc.
+ * This list contains information parsed from per CPU ACPI _CPC and _PSD
+ * structures: e.g. the highest and lowest supported performance, capabilities,
+ * desired performance, level requested etc. Depending on the share_type, not
+ * all CPUs will have an entry in the list.
  */
-static struct cppc_cpudata **all_cpu_data;
+static LIST_HEAD(cpu_data_list);
+
 static bool boost_supported;
 
 struct cppc_workaround_oem_info {
@@ -148,8 +148,10 @@ static unsigned int cppc_cpufreq_khz_to_perf(struct cppc_cpudata *cpu_data,
 static int cppc_cpufreq_set_target(struct cpufreq_policy *policy,
 				   unsigned int target_freq,
 				   unsigned int relation)
+
 {
-	struct cppc_cpudata *cpu_data = all_cpu_data[policy->cpu];
+	struct cppc_cpudata *cpu_data = policy->driver_data;
+	unsigned int cpu = policy->cpu;
 	struct cpufreq_freqs freqs;
 	u32 desired_perf;
 	int ret = 0;
@@ -164,12 +166,12 @@ static int cppc_cpufreq_set_target(struct cpufreq_policy *policy,
 	freqs.new = target_freq;
 
 	cpufreq_freq_transition_begin(policy, &freqs);
-	ret = cppc_set_perf(cpu_data->cpu, &cpu_data->perf_ctrls);
+	ret = cppc_set_perf(cpu, &cpu_data->perf_ctrls);
 	cpufreq_freq_transition_end(policy, &freqs, ret != 0);
 
 	if (ret)
 		pr_debug("Failed to set target on CPU:%d. ret:%d\n",
-			 cpu_data->cpu, ret);
+			 cpu, ret);
 
 	return ret;
 }
@@ -182,7 +184,7 @@ static int cppc_verify_policy(struct cpufreq_policy_data *policy)
 
 static void cppc_cpufreq_stop_cpu(struct cpufreq_policy *policy)
 {
-	struct cppc_cpudata *cpu_data = all_cpu_data[policy->cpu];
+	struct cppc_cpudata *cpu_data = policy->driver_data;
 	struct cppc_perf_caps *caps = &cpu_data->perf_caps;
 	unsigned int cpu = policy->cpu;
 	int ret;
@@ -193,6 +195,12 @@ static void cppc_cpufreq_stop_cpu(struct cpufreq_policy *policy)
 	if (ret)
 		pr_debug("Err setting perf value:%d on CPU:%d. ret:%d\n",
 			 caps->lowest_perf, cpu, ret);
+
+	/* Remove CPU node from list and free driver data for policy */
+	free_cpumask_var(cpu_data->shared_cpu_map);
+	list_del(&cpu_data->node);
+	kfree(policy->driver_data);
+	policy->driver_data = NULL;
 }
 
 /*
@@ -238,25 +246,61 @@ static unsigned int cppc_cpufreq_get_transition_delay_us(unsigned int cpu)
 }
 #endif
 
-static int cppc_cpufreq_cpu_init(struct cpufreq_policy *policy)
+
+static struct cppc_cpudata *cppc_cpufreq_get_cpu_data(unsigned int cpu)
 {
-	struct cppc_cpudata *cpu_data = all_cpu_data[policy->cpu];
-	struct cppc_perf_caps *caps = &cpu_data->perf_caps;
-	unsigned int cpu = policy->cpu;
-	int ret = 0;
+	struct cppc_cpudata *cpu_data;
+	int ret;
 
-	cpu_data->cpu = cpu;
-	ret = cppc_get_perf_caps(cpu, caps);
+	cpu_data = kzalloc(sizeof(struct cppc_cpudata), GFP_KERNEL);
+	if (!cpu_data)
+		goto out;
 
+	if (!zalloc_cpumask_var(&cpu_data->shared_cpu_map, GFP_KERNEL))
+		goto free_cpu;
+
+	ret = acpi_get_psd_map(cpu, cpu_data);
 	if (ret) {
-		pr_debug("Err reading CPU%d perf capabilities. ret:%d\n",
-			 cpu, ret);
-		return ret;
+		pr_debug("Err parsing CPU%d PSD data: ret:%d\n", cpu, ret);
+		goto free_mask;
+	}
+
+	ret = cppc_get_perf_caps(cpu, &cpu_data->perf_caps);
+	if (ret) {
+		pr_debug("Err reading CPU%d perf caps: ret:%d\n", cpu, ret);
+		goto free_mask;
 	}
 
 	/* Convert the lowest and nominal freq from MHz to KHz */
-	caps->lowest_freq *= 1000;
-	caps->nominal_freq *= 1000;
+	cpu_data->perf_caps.lowest_freq *= 1000;
+	cpu_data->perf_caps.nominal_freq *= 1000;
+
+	list_add(&cpu_data->node, &cpu_data_list);
+
+	return cpu_data;
+
+free_mask:
+	free_cpumask_var(cpu_data->shared_cpu_map);
+free_cpu:
+	kfree(cpu_data);
+out:
+	return NULL;
+}
+
+static int cppc_cpufreq_cpu_init(struct cpufreq_policy *policy)
+{
+	unsigned int cpu = policy->cpu;
+	struct cppc_cpudata *cpu_data;
+	struct cppc_perf_caps *caps;
+	int ret;
+
+	cpu_data = cppc_cpufreq_get_cpu_data(cpu);
+	if (!cpu_data) {
+		pr_err("Error in acquiring _CPC/_PSD data for CPU%d.\n", cpu);
+		return -ENODEV;
+	}
+	caps = &cpu_data->perf_caps;
+	policy->driver_data = cpu_data;
 
 	/*
 	 * Set min to lowest nonlinear perf to avoid any efficiency penalty (see
@@ -280,26 +324,25 @@ static int cppc_cpufreq_cpu_init(struct cpufreq_policy *policy)
 	policy->transition_delay_us = cppc_cpufreq_get_transition_delay_us(cpu);
 	policy->shared_type = cpu_data->shared_type;
 
-	if (policy->shared_type == CPUFREQ_SHARED_TYPE_ANY) {
-		int i;
-
+	switch (policy->shared_type) {
+	case CPUFREQ_SHARED_TYPE_HW:
+	case CPUFREQ_SHARED_TYPE_NONE:
+		/* Nothing to be done - we'll have a policy for each CPU */
+		break;
+	case CPUFREQ_SHARED_TYPE_ANY:
+		/*
+		 * All CPUs in the domain will share a policy and all cpufreq
+		 * operations will use a single cppc_cpudata structure stored
+		 * in policy->driver_data.
+		 */
 		cpumask_copy(policy->cpus, cpu_data->shared_cpu_map);
-
-		for_each_cpu(i, policy->cpus) {
-			if (unlikely(i == cpu))
-				continue;
-
-			memcpy(&all_cpu_data[i]->perf_caps, caps,
-			       sizeof(cpu_data->perf_caps));
-		}
-	} else if (policy->shared_type == CPUFREQ_SHARED_TYPE_ALL) {
-		/* Support only SW_ANY for now. */
-		pr_debug("Unsupported CPU co-ord type\n");
+		break;
+	default:
+		pr_debug("Unsupported CPU co-ord type: %d\n",
+			 policy->shared_type);
 		return -EFAULT;
 	}
 
-	cpu_data->cur_policy = policy;
-
 	/*
 	 * If 'highest_perf' is greater than 'nominal_perf', we assume CPU Boost
 	 * is supported.
@@ -354,9 +397,12 @@ static int cppc_get_rate_from_fbctrs(struct cppc_cpudata *cpu_data,
 static unsigned int cppc_cpufreq_get_rate(unsigned int cpu)
 {
 	struct cppc_perf_fb_ctrs fb_ctrs_t0 = {0}, fb_ctrs_t1 = {0};
-	struct cppc_cpudata *cpu_data = all_cpu_data[cpu];
+	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
+	struct cppc_cpudata *cpu_data = policy->driver_data;
 	int ret;
 
+	cpufreq_cpu_put(policy);
+
 	ret = cppc_get_perf_ctrs(cpu, &fb_ctrs_t0);
 	if (ret)
 		return ret;
@@ -372,7 +418,7 @@ static unsigned int cppc_cpufreq_get_rate(unsigned int cpu)
 
 static int cppc_cpufreq_set_boost(struct cpufreq_policy *policy, int state)
 {
-	struct cppc_cpudata *cpu_data = all_cpu_data[policy->cpu];
+	struct cppc_cpudata *cpu_data = policy->driver_data;
 	struct cppc_perf_caps *caps = &cpu_data->perf_caps;
 	int ret;
 
@@ -396,6 +442,19 @@ static int cppc_cpufreq_set_boost(struct cpufreq_policy *policy, int state)
 	return 0;
 }
 
+static ssize_t show_freqdomain_cpus(struct cpufreq_policy *policy, char *buf)
+{
+	struct cppc_cpudata *cpu_data = policy->driver_data;
+
+	return cpufreq_show_cpus(cpu_data->shared_cpu_map, buf);
+}
+cpufreq_freq_attr_ro(freqdomain_cpus);
+
+static struct freq_attr *cppc_cpufreq_attr[] = {
+	&freqdomain_cpus,
+	NULL,
+};
+
 static struct cpufreq_driver cppc_cpufreq_driver = {
 	.flags = CPUFREQ_CONST_LOOPS,
 	.verify = cppc_verify_policy,
@@ -404,6 +463,7 @@ static struct cpufreq_driver cppc_cpufreq_driver = {
 	.init = cppc_cpufreq_cpu_init,
 	.stop_cpu = cppc_cpufreq_stop_cpu,
 	.set_boost = cppc_cpufreq_set_boost,
+	.attr = cppc_cpufreq_attr,
 	.name = "cppc_cpufreq",
 };
 
@@ -415,10 +475,13 @@ static struct cpufreq_driver cppc_cpufreq_driver = {
  */
 static unsigned int hisi_cppc_cpufreq_get_rate(unsigned int cpu)
 {
-	struct cppc_cpudata *cpu_data = all_cpu_data[cpu];
+	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
+	struct cppc_cpudata *cpu_data = policy->driver_data;
 	u64 desired_perf;
 	int ret;
 
+	cpufreq_cpu_put(policy);
+
 	ret = cppc_get_desired_perf(cpu, &desired_perf);
 	if (ret < 0)
 		return -EIO;
@@ -451,68 +514,33 @@ static void cppc_check_hisi_workaround(void)
 
 static int __init cppc_cpufreq_init(void)
 {
-	struct cppc_cpudata *cpu_data;
-	int i, ret = 0;
-
-	if (acpi_disabled)
+	if ((acpi_disabled) || !acpi_cpc_valid())
 		return -ENODEV;
 
-	all_cpu_data = kcalloc(num_possible_cpus(), sizeof(void *),
-			       GFP_KERNEL);
-	if (!all_cpu_data)
-		return -ENOMEM;
-
-	for_each_possible_cpu(i) {
-		all_cpu_data[i] = kzalloc(sizeof(struct cppc_cpudata), GFP_KERNEL);
-		if (!all_cpu_data[i])
-			goto out;
-
-		cpu_data = all_cpu_data[i];
-		if (!zalloc_cpumask_var(&cpu_data->shared_cpu_map, GFP_KERNEL))
-			goto out;
-	}
-
-	ret = acpi_get_psd_map(all_cpu_data);
-	if (ret) {
-		pr_debug("Error parsing PSD data. Aborting cpufreq registration.\n");
-		goto out;
-	}
+	INIT_LIST_HEAD(&cpu_data_list);
 
 	cppc_check_hisi_workaround();
 
-	ret = cpufreq_register_driver(&cppc_cpufreq_driver);
-	if (ret)
-		goto out;
+	return cpufreq_register_driver(&cppc_cpufreq_driver);
+}
 
-	return ret;
+static inline void free_cpu_data(void)
+{
+	struct cppc_cpudata *iter, *tmp;
 
-out:
-	for_each_possible_cpu(i) {
-		cpu_data = all_cpu_data[i];
-		if (!cpu_data)
-			break;
-		free_cpumask_var(cpu_data->shared_cpu_map);
-		kfree(cpu_data);
+	list_for_each_entry_safe(iter, tmp, &cpu_data_list, node) {
+		free_cpumask_var(iter->shared_cpu_map);
+		list_del(&iter->node);
+		kfree(iter);
 	}
 
-	kfree(all_cpu_data);
-	return -ENODEV;
 }
 
 static void __exit cppc_cpufreq_exit(void)
 {
-	struct cppc_cpudata *cpu_data;
-	int i;
-
 	cpufreq_unregister_driver(&cppc_cpufreq_driver);
 
-	for_each_possible_cpu(i) {
-		cpu_data = all_cpu_data[i];
-		free_cpumask_var(cpu_data->shared_cpu_map);
-		kfree(cpu_data);
-	}
-
-	kfree(all_cpu_data);
+	free_cpu_data();
 }
 
 module_exit(cppc_cpufreq_exit);

drivers/cpufreq/cpufreq.c

@@ -2097,6 +2097,46 @@ unsigned int cpufreq_driver_fast_switch(struct cpufreq_policy *policy,
 }
 EXPORT_SYMBOL_GPL(cpufreq_driver_fast_switch);
 
+/**
+ * cpufreq_driver_adjust_perf - Adjust CPU performance level in one go.
+ * @cpu: Target CPU.
+ * @min_perf: Minimum (required) performance level (units of @capacity).
+ * @target_perf: Terget (desired) performance level (units of @capacity).
+ * @capacity: Capacity of the target CPU.
+ *
+ * Carry out a fast performance level switch of @cpu without sleeping.
+ *
+ * The driver's ->adjust_perf() callback invoked by this function must be
+ * suitable for being called from within RCU-sched read-side critical sections
+ * and it is expected to select a suitable performance level equal to or above
+ * @min_perf and preferably equal to or below @target_perf.
+ *
+ * This function must not be called if policy->fast_switch_enabled is unset.
+ *
+ * Governors calling this function must guarantee that it will never be invoked
+ * twice in parallel for the same CPU and that it will never be called in
+ * parallel with either ->target() or ->target_index() or ->fast_switch() for
+ * the same CPU.
+ */
+void cpufreq_driver_adjust_perf(unsigned int cpu,
+				 unsigned long min_perf,
+				 unsigned long target_perf,
+				 unsigned long capacity)
+{
+	cpufreq_driver->adjust_perf(cpu, min_perf, target_perf, capacity);
+}
+
+/**
+ * cpufreq_driver_has_adjust_perf - Check "direct fast switch" callback.
+ *
+ * Return 'true' if the ->adjust_perf callback is present for the
+ * current driver or 'false' otherwise.
+ */
+bool cpufreq_driver_has_adjust_perf(void)
+{
+	return !!cpufreq_driver->adjust_perf;
+}
+
 /* Must set freqs->new to intermediate frequency */
 static int __target_intermediate(struct cpufreq_policy *policy,
 				 struct cpufreq_freqs *freqs, int index)

drivers/cpufreq/intel_pstate.c

@@ -2207,9 +2207,9 @@ static void intel_pstate_update_perf_limits(struct cpudata *cpu,
 					    unsigned int policy_min,
 					    unsigned int policy_max)
 {
-	int max_freq = intel_pstate_get_max_freq(cpu);
 	int32_t max_policy_perf, min_policy_perf;
 	int max_state, turbo_max;
+	int max_freq;
 
 	/*
 	 * HWP needs some special consideration, because on BDX the
@@ -2223,6 +2223,7 @@ static void intel_pstate_update_perf_limits(struct cpudata *cpu,
 			cpu->pstate.max_pstate : cpu->pstate.turbo_pstate;
 		turbo_max = cpu->pstate.turbo_pstate;
 	}
+	max_freq = max_state * cpu->pstate.scaling;
 
 	max_policy_perf = max_state * policy_max / max_freq;
 	if (policy_max == policy_min) {
@@ -2325,9 +2326,18 @@ static void intel_pstate_adjust_policy_max(struct cpudata *cpu,
 static void intel_pstate_verify_cpu_policy(struct cpudata *cpu,
 					   struct cpufreq_policy_data *policy)
 {
+	int max_freq;
+
 	update_turbo_state();
-	cpufreq_verify_within_limits(policy, policy->cpuinfo.min_freq,
-				     intel_pstate_get_max_freq(cpu));
+	if (hwp_active) {
+		int max_state, turbo_max;
+
+		intel_pstate_get_hwp_max(cpu->cpu, &turbo_max, &max_state);
+		max_freq = max_state * cpu->pstate.scaling;
+	} else {
+		max_freq = intel_pstate_get_max_freq(cpu);
+	}
+	cpufreq_verify_within_limits(policy, policy->cpuinfo.min_freq, max_freq);
 
 	intel_pstate_adjust_policy_max(cpu, policy);
 }
@@ -2526,20 +2536,19 @@ static void intel_cpufreq_trace(struct cpudata *cpu, unsigned int trace_type, in
 		fp_toint(cpu->iowait_boost * 100));
 }
 
-static void intel_cpufreq_adjust_hwp(struct cpudata *cpu, u32 target_pstate,
-				     bool strict, bool fast_switch)
+static void intel_cpufreq_adjust_hwp(struct cpudata *cpu, u32 min, u32 max,
+				     u32 desired, bool fast_switch)
 {
 	u64 prev = READ_ONCE(cpu->hwp_req_cached), value = prev;
 
 	value &= ~HWP_MIN_PERF(~0L);
-	value |= HWP_MIN_PERF(target_pstate);
+	value |= HWP_MIN_PERF(min);
 
-	/*
-	 * The entire MSR needs to be updated in order to update the HWP min
-	 * field in it, so opportunistically update the max too if needed.
-	 */
 	value &= ~HWP_MAX_PERF(~0L);
-	value |= HWP_MAX_PERF(strict ? target_pstate : cpu->max_perf_ratio);
+	value |= HWP_MAX_PERF(max);
+
+	value &= ~HWP_DESIRED_PERF(~0L);
+	value |= HWP_DESIRED_PERF(desired);
 
 	if (value == prev)
 		return;
@@ -2569,11 +2578,15 @@ static int intel_cpufreq_update_pstate(struct cpufreq_policy *policy,
 	int old_pstate = cpu->pstate.current_pstate;
 
 	target_pstate = intel_pstate_prepare_request(cpu, target_pstate);
-	if (hwp_active)
-		intel_cpufreq_adjust_hwp(cpu, target_pstate,
-					 policy->strict_target, fast_switch);
-	else if (target_pstate != old_pstate)
+	if (hwp_active) {
+		int max_pstate = policy->strict_target ?
+					target_pstate : cpu->max_perf_ratio;
+
+		intel_cpufreq_adjust_hwp(cpu, target_pstate, max_pstate, 0,
+					 fast_switch);
+	} else if (target_pstate != old_pstate) {
 		intel_cpufreq_adjust_perf_ctl(cpu, target_pstate, fast_switch);
+	}
 
 	cpu->pstate.current_pstate = target_pstate;
 
@@ -2634,6 +2647,47 @@ static unsigned int intel_cpufreq_fast_switch(struct cpufreq_policy *policy,
 	return target_pstate * cpu->pstate.scaling;
 }
 
+static void intel_cpufreq_adjust_perf(unsigned int cpunum,
+				      unsigned long min_perf,
+				      unsigned long target_perf,
+				      unsigned long capacity)
+{
+	struct cpudata *cpu = all_cpu_data[cpunum];
+	int old_pstate = cpu->pstate.current_pstate;
+	int cap_pstate, min_pstate, max_pstate, target_pstate;
+
+	update_turbo_state();
+	cap_pstate = global.turbo_disabled ? cpu->pstate.max_pstate :
+					     cpu->pstate.turbo_pstate;
+
+	/* Optimization: Avoid unnecessary divisions. */
+
+	target_pstate = cap_pstate;
+	if (target_perf < capacity)
+		target_pstate = DIV_ROUND_UP(cap_pstate * target_perf, capacity);
+
+	min_pstate = cap_pstate;
+	if (min_perf < capacity)
+		min_pstate = DIV_ROUND_UP(cap_pstate * min_perf, capacity);
+
+	if (min_pstate < cpu->pstate.min_pstate)
+		min_pstate = cpu->pstate.min_pstate;
+
+	if (min_pstate < cpu->min_perf_ratio)
+		min_pstate = cpu->min_perf_ratio;
+
+	max_pstate = min(cap_pstate, cpu->max_perf_ratio);
+	if (max_pstate < min_pstate)
+		max_pstate = min_pstate;
+
+	target_pstate = clamp_t(int, target_pstate, min_pstate, max_pstate);
+
+	intel_cpufreq_adjust_hwp(cpu, min_pstate, max_pstate, target_pstate, true);
+
+	cpu->pstate.current_pstate = target_pstate;
+	intel_cpufreq_trace(cpu, INTEL_PSTATE_TRACE_FAST_SWITCH, old_pstate);
+}
+
 static int intel_cpufreq_cpu_init(struct cpufreq_policy *policy)
 {
 	int max_state, turbo_max, min_freq, max_freq, ret;
@@ -3032,6 +3086,8 @@ static int __init intel_pstate_init(void)
 			intel_pstate.attr = hwp_cpufreq_attrs;
 			intel_cpufreq.attr = hwp_cpufreq_attrs;
 			intel_cpufreq.flags |= CPUFREQ_NEED_UPDATE_LIMITS;
+			intel_cpufreq.fast_switch = NULL;
+			intel_cpufreq.adjust_perf = intel_cpufreq_adjust_perf;
 			if (!default_driver)
 				default_driver = &intel_pstate;
 

include/acpi/cppc_acpi.h

@@ -124,11 +124,10 @@ struct cppc_perf_fb_ctrs {
 
 /* Per CPU container for runtime CPPC management. */
 struct cppc_cpudata {
-	int cpu;
+	struct list_head node;
 	struct cppc_perf_caps perf_caps;
 	struct cppc_perf_ctrls perf_ctrls;
 	struct cppc_perf_fb_ctrs perf_fb_ctrs;
-	struct cpufreq_policy *cur_policy;
 	unsigned int shared_type;
 	cpumask_var_t shared_cpu_map;
 };
@@ -137,7 +136,8 @@ extern int cppc_get_desired_perf(int cpunum, u64 *desired_perf);
 extern int cppc_get_perf_ctrs(int cpu, struct cppc_perf_fb_ctrs *perf_fb_ctrs);
 extern int cppc_set_perf(int cpu, struct cppc_perf_ctrls *perf_ctrls);
 extern int cppc_get_perf_caps(int cpu, struct cppc_perf_caps *caps);
-extern int acpi_get_psd_map(struct cppc_cpudata **);
+extern bool acpi_cpc_valid(void);
+extern int acpi_get_psd_map(unsigned int cpu, struct cppc_cpudata *cpu_data);
 extern unsigned int cppc_get_transition_latency(int cpu);
 extern bool cpc_ffh_supported(void);
 extern int cpc_read_ffh(int cpunum, struct cpc_reg *reg, u64 *val);

include/linux/cpufreq.h

@@ -320,6 +320,15 @@ struct cpufreq_driver {
 					unsigned int index);
 	unsigned int	(*fast_switch)(struct cpufreq_policy *policy,
 				       unsigned int target_freq);
+	/*
+	 * ->fast_switch() replacement for drivers that use an internal
+	 * representation of performance levels and can pass hints other than
+	 * the target performance level to the hardware.
+	 */
+	void		(*adjust_perf)(unsigned int cpu,
+				       unsigned long min_perf,
+				       unsigned long target_perf,
+				       unsigned long capacity);
 
 	/*
 	 * Caches and returns the lowest driver-supported frequency greater than
@@ -588,6 +597,11 @@ struct cpufreq_governor {
 /* Pass a target to the cpufreq driver */
 unsigned int cpufreq_driver_fast_switch(struct cpufreq_policy *policy,
 					unsigned int target_freq);
+void cpufreq_driver_adjust_perf(unsigned int cpu,
+				unsigned long min_perf,
+				unsigned long target_perf,
+				unsigned long capacity);
+bool cpufreq_driver_has_adjust_perf(void);
 int cpufreq_driver_target(struct cpufreq_policy *policy,
 				 unsigned int target_freq,
 				 unsigned int relation);

include/linux/sched/cpufreq.h

@@ -28,6 +28,11 @@ static inline unsigned long map_util_freq(unsigned long util,
 {
 	return (freq + (freq >> 2)) * util / cap;
 }
+
+static inline unsigned long map_util_perf(unsigned long util)
+{
+	return util + (util >> 2);
+}
 #endif /* CONFIG_CPU_FREQ */
 
 #endif /* _LINUX_SCHED_CPUFREQ_H */

kernel/sched/cpufreq_schedutil.c

@@ -53,6 +53,7 @@ struct sugov_cpu {
 	unsigned int		iowait_boost;
 	u64			last_update;
 
+	unsigned long		util;
 	unsigned long		bw_dl;
 	unsigned long		max;
 
@@ -276,16 +277,15 @@ unsigned long schedutil_cpu_util(int cpu, unsigned long util_cfs,
 	return min(max, util);
 }
 
-static unsigned long sugov_get_util(struct sugov_cpu *sg_cpu)
+static void sugov_get_util(struct sugov_cpu *sg_cpu)
 {
 	struct rq *rq = cpu_rq(sg_cpu->cpu);
-	unsigned long util = cpu_util_cfs(rq);
 	unsigned long max = arch_scale_cpu_capacity(sg_cpu->cpu);
 
 	sg_cpu->max = max;
 	sg_cpu->bw_dl = cpu_bw_dl(rq);
-
-	return schedutil_cpu_util(sg_cpu->cpu, util, max, FREQUENCY_UTIL, NULL);
+	sg_cpu->util = schedutil_cpu_util(sg_cpu->cpu, cpu_util_cfs(rq), max,
+					  FREQUENCY_UTIL, NULL);
 }
 
 /**
@@ -362,8 +362,6 @@ static void sugov_iowait_boost(struct sugov_cpu *sg_cpu, u64 time,
  * sugov_iowait_apply() - Apply the IO boost to a CPU.
  * @sg_cpu: the sugov data for the cpu to boost
  * @time: the update time from the caller
- * @util: the utilization to (eventually) boost
- * @max: the maximum value the utilization can be boosted to
  *
  * A CPU running a task which woken up after an IO operation can have its
  * utilization boosted to speed up the completion of those IO operations.
@@ -377,18 +375,17 @@ static void sugov_iowait_boost(struct sugov_cpu *sg_cpu, u64 time,
  * This mechanism is designed to boost high frequently IO waiting tasks, while
  * being more conservative on tasks which does sporadic IO operations.
  */
-static unsigned long sugov_iowait_apply(struct sugov_cpu *sg_cpu, u64 time,
-					unsigned long util, unsigned long max)
+static void sugov_iowait_apply(struct sugov_cpu *sg_cpu, u64 time)
 {
 	unsigned long boost;
 
 	/* No boost currently required */
 	if (!sg_cpu->iowait_boost)
-		return util;
+		return;
 
 	/* Reset boost if the CPU appears to have been idle enough */
 	if (sugov_iowait_reset(sg_cpu, time, false))
-		return util;
+		return;
 
 	if (!sg_cpu->iowait_boost_pending) {
 		/*
@@ -397,18 +394,19 @@ static unsigned long sugov_iowait_apply(struct sugov_cpu *sg_cpu, u64 time,
 		sg_cpu->iowait_boost >>= 1;
 		if (sg_cpu->iowait_boost < IOWAIT_BOOST_MIN) {
 			sg_cpu->iowait_boost = 0;
-			return util;
+			return;
 		}
 	}
 
 	sg_cpu->iowait_boost_pending = false;
 
 	/*
-	 * @util is already in capacity scale; convert iowait_boost
+	 * sg_cpu->util is already in capacity scale; convert iowait_boost
 	 * into the same scale so we can compare.
 	 */
-	boost = (sg_cpu->iowait_boost * max) >> SCHED_CAPACITY_SHIFT;
-	return max(boost, util);
+	boost = (sg_cpu->iowait_boost * sg_cpu->max) >> SCHED_CAPACITY_SHIFT;
+	if (sg_cpu->util < boost)
+		sg_cpu->util = boost;
 }
 
 #ifdef CONFIG_NO_HZ_COMMON
@@ -434,14 +432,10 @@ static inline void ignore_dl_rate_limit(struct sugov_cpu *sg_cpu, struct sugov_p
 		sg_policy->limits_changed = true;
 }
 
-static void sugov_update_single(struct update_util_data *hook, u64 time,
-				unsigned int flags)
+static inline bool sugov_update_single_common(struct sugov_cpu *sg_cpu,
+					      u64 time, unsigned int flags)
 {
-	struct sugov_cpu *sg_cpu = container_of(hook, struct sugov_cpu, update_util);
 	struct sugov_policy *sg_policy = sg_cpu->sg_policy;
-	unsigned long util, max;
-	unsigned int next_f;
-	unsigned int cached_freq = sg_policy->cached_raw_freq;
 
 	sugov_iowait_boost(sg_cpu, time, flags);
 	sg_cpu->last_update = time;
@@ -449,12 +443,26 @@ static void sugov_update_single(struct update_util_data *hook, u64 time,
 	ignore_dl_rate_limit(sg_cpu, sg_policy);
 
 	if (!sugov_should_update_freq(sg_policy, time))
+		return false;
+
+	sugov_get_util(sg_cpu);
+	sugov_iowait_apply(sg_cpu, time);
+
+	return true;
+}
+
+static void sugov_update_single_freq(struct update_util_data *hook, u64 time,
+				     unsigned int flags)
+{
+	struct sugov_cpu *sg_cpu = container_of(hook, struct sugov_cpu, update_util);
+	struct sugov_policy *sg_policy = sg_cpu->sg_policy;
+	unsigned int cached_freq = sg_policy->cached_raw_freq;
+	unsigned int next_f;
+
+	if (!sugov_update_single_common(sg_cpu, time, flags))
 		return;
 
-	util = sugov_get_util(sg_cpu);
-	max = sg_cpu->max;
-	util = sugov_iowait_apply(sg_cpu, time, util, max);
-	next_f = get_next_freq(sg_policy, util, max);
+	next_f = get_next_freq(sg_policy, sg_cpu->util, sg_cpu->max);
 	/*
 	 * Do not reduce the frequency if the CPU has not been idle
 	 * recently, as the reduction is likely to be premature then.
@@ -480,6 +488,38 @@ static void sugov_update_single(struct update_util_data *hook, u64 time,
 	}
 }
 
+static void sugov_update_single_perf(struct update_util_data *hook, u64 time,
+				     unsigned int flags)
+{
+	struct sugov_cpu *sg_cpu = container_of(hook, struct sugov_cpu, update_util);
+	unsigned long prev_util = sg_cpu->util;
+
+	/*
+	 * Fall back to the "frequency" path if frequency invariance is not
+	 * supported, because the direct mapping between the utilization and
+	 * the performance levels depends on the frequency invariance.
+	 */
+	if (!arch_scale_freq_invariant()) {
+		sugov_update_single_freq(hook, time, flags);
+		return;
+	}
+
+	if (!sugov_update_single_common(sg_cpu, time, flags))
+		return;
+
+	/*
+	 * Do not reduce the target performance level if the CPU has not been
+	 * idle recently, as the reduction is likely to be premature then.
+	 */
+	if (sugov_cpu_is_busy(sg_cpu) && sg_cpu->util < prev_util)
+		sg_cpu->util = prev_util;
+
+	cpufreq_driver_adjust_perf(sg_cpu->cpu, map_util_perf(sg_cpu->bw_dl),
+				   map_util_perf(sg_cpu->util), sg_cpu->max);
+
+	sg_cpu->sg_policy->last_freq_update_time = time;
+}
+
 static unsigned int sugov_next_freq_shared(struct sugov_cpu *sg_cpu, u64 time)
 {
 	struct sugov_policy *sg_policy = sg_cpu->sg_policy;
@@ -491,9 +531,10 @@ static unsigned int sugov_next_freq_shared(struct sugov_cpu *sg_cpu, u64 time)
 		struct sugov_cpu *j_sg_cpu = &per_cpu(sugov_cpu, j);
 		unsigned long j_util, j_max;
 
-		j_util = sugov_get_util(j_sg_cpu);
+		sugov_get_util(j_sg_cpu);
+		sugov_iowait_apply(j_sg_cpu, time);
+		j_util = j_sg_cpu->util;
 		j_max = j_sg_cpu->max;
-		j_util = sugov_iowait_apply(j_sg_cpu, time, j_util, j_max);
 
 		if (j_util * max > j_max * util) {
 			util = j_util;
@@ -817,6 +858,7 @@ static void sugov_exit(struct cpufreq_policy *policy)
 static int sugov_start(struct cpufreq_policy *policy)
 {
 	struct sugov_policy *sg_policy = policy->governor_data;
+	void (*uu)(struct update_util_data *data, u64 time, unsigned int flags);
 	unsigned int cpu;
 
 	sg_policy->freq_update_delay_ns	= sg_policy->tunables->rate_limit_us * NSEC_PER_USEC;
@@ -836,13 +878,17 @@ static int sugov_start(struct cpufreq_policy *policy)
 		sg_cpu->sg_policy		= sg_policy;
 	}
 
+	if (policy_is_shared(policy))
+		uu = sugov_update_shared;
+	else if (policy->fast_switch_enabled && cpufreq_driver_has_adjust_perf())
+		uu = sugov_update_single_perf;
+	else
+		uu = sugov_update_single_freq;
+
 	for_each_cpu(cpu, policy->cpus) {
 		struct sugov_cpu *sg_cpu = &per_cpu(sugov_cpu, cpu);
 
-		cpufreq_add_update_util_hook(cpu, &sg_cpu->update_util,
-					     policy_is_shared(policy) ?
-							sugov_update_shared :
-							sugov_update_single);
+		cpufreq_add_update_util_hook(cpu, &sg_cpu->update_util, uu);
 	}
 	return 0;
 }