Merge branch 'pm-cpufreq' into pm-opp

17 files changed, 638 insertions, 404 deletions

diff --git a/arch/powerpc/include/asm/opal-api.h b/arch/powerpc/include/asm/opal-api.h
index e9e4c52f3685..64dc9f547fb6 100644
--- a/arch/powerpc/include/asm/opal-api.h
+++ b/arch/powerpc/include/asm/opal-api.h
@@ -361,6 +361,7 @@ enum opal_msg_type {
 	OPAL_MSG_HMI_EVT,
 	OPAL_MSG_DPO,
 	OPAL_MSG_PRD,
+	OPAL_MSG_OCC,
 	OPAL_MSG_TYPE_MAX,
 };
 
@@ -700,6 +701,17 @@ struct opal_prd_msg_header {
 
 struct opal_prd_msg;
 
+#define OCC_RESET                       0
+#define OCC_LOAD                        1
+#define OCC_THROTTLE                    2
+#define OCC_MAX_THROTTLE_STATUS         5
+
+struct opal_occ_msg {
+	__be64 type;
+	__be64 chip;
+	__be64 throttle_status;
+};
+
 /*
  * SG entries
  *
diff --git a/drivers/acpi/processor_perflib.c b/drivers/acpi/processor_perflib.c
index cfc8aba72f86..36b6da2918a6 100644
--- a/drivers/acpi/processor_perflib.c
+++ b/drivers/acpi/processor_perflib.c
@@ -784,9 +784,7 @@ acpi_processor_register_performance(struct acpi_processor_performance
 
 EXPORT_SYMBOL(acpi_processor_register_performance);
 
-void
-acpi_processor_unregister_performance(struct acpi_processor_performance
-				      *performance, unsigned int cpu)
+void acpi_processor_unregister_performance(unsigned int cpu)
 {
 	struct acpi_processor *pr;
 
diff --git a/drivers/cpufreq/acpi-cpufreq.c b/drivers/cpufreq/acpi-cpufreq.c
index 0136dfcdabf0..15b921a9248c 100644
--- a/drivers/cpufreq/acpi-cpufreq.c
+++ b/drivers/cpufreq/acpi-cpufreq.c
@@ -65,18 +65,21 @@ enum {
 #define MSR_K7_HWCR_CPB_DIS	(1ULL << 25)
 
 struct acpi_cpufreq_data {
-	struct acpi_processor_performance *acpi_data;
 	struct cpufreq_frequency_table *freq_table;
 	unsigned int resume;
 	unsigned int cpu_feature;
+	unsigned int acpi_perf_cpu;
 	cpumask_var_t freqdomain_cpus;
 };
 
-static DEFINE_PER_CPU(struct acpi_cpufreq_data *, acfreq_data);
-
 /* acpi_perf_data is a pointer to percpu data. */
 static struct acpi_processor_performance __percpu *acpi_perf_data;
 
+static inline struct acpi_processor_performance *to_perf_data(struct acpi_cpufreq_data *data)
+{
+	return per_cpu_ptr(acpi_perf_data, data->acpi_perf_cpu);
+}
+
 static struct cpufreq_driver acpi_cpufreq_driver;
 
 static unsigned int acpi_pstate_strict;
@@ -144,7 +147,7 @@ static int _store_boost(int val)
 
 static ssize_t show_freqdomain_cpus(struct cpufreq_policy *policy, char *buf)
 {
-	struct acpi_cpufreq_data *data = per_cpu(acfreq_data, policy->cpu);
+	struct acpi_cpufreq_data *data = policy->driver_data;
 
 	return cpufreq_show_cpus(data->freqdomain_cpus, buf);
 }
@@ -202,7 +205,7 @@ static unsigned extract_io(u32 value, struct acpi_cpufreq_data *data)
 	struct acpi_processor_performance *perf;
 	int i;
 
-	perf = data->acpi_data;
+	perf = to_perf_data(data);
 
 	for (i = 0; i < perf->state_count; i++) {
 		if (value == perf->states[i].status)
@@ -221,7 +224,7 @@ static unsigned extract_msr(u32 msr, struct acpi_cpufreq_data *data)
 	else
 		msr &= INTEL_MSR_RANGE;
 
-	perf = data->acpi_data;
+	perf = to_perf_data(data);
 
 	cpufreq_for_each_entry(pos, data->freq_table)
 		if (msr == perf->states[pos->driver_data].status)
@@ -327,7 +330,8 @@ static void drv_write(struct drv_cmd *cmd)
 	put_cpu();
 }
 
-static u32 get_cur_val(const struct cpumask *mask)
+static u32
+get_cur_val(const struct cpumask *mask, struct acpi_cpufreq_data *data)
 {
 	struct acpi_processor_performance *perf;
 	struct drv_cmd cmd;
@@ -335,7 +339,7 @@ static u32 get_cur_val(const struct cpumask *mask)
 	if (unlikely(cpumask_empty(mask)))
 		return 0;
 
-	switch (per_cpu(acfreq_data, cpumask_first(mask))->cpu_feature) {
+	switch (data->cpu_feature) {
 	case SYSTEM_INTEL_MSR_CAPABLE:
 		cmd.type = SYSTEM_INTEL_MSR_CAPABLE;
 		cmd.addr.msr.reg = MSR_IA32_PERF_CTL;
@@ -346,7 +350,7 @@ static u32 get_cur_val(const struct cpumask *mask)
 		break;
 	case SYSTEM_IO_CAPABLE:
 		cmd.type = SYSTEM_IO_CAPABLE;
-		perf = per_cpu(acfreq_data, cpumask_first(mask))->acpi_data;
+		perf = to_perf_data(data);
 		cmd.addr.io.port = perf->control_register.address;
 		cmd.addr.io.bit_width = perf->control_register.bit_width;
 		break;
@@ -364,19 +368,24 @@ static u32 get_cur_val(const struct cpumask *mask)
 
 static unsigned int get_cur_freq_on_cpu(unsigned int cpu)
 {
-	struct acpi_cpufreq_data *data = per_cpu(acfreq_data, cpu);
+	struct acpi_cpufreq_data *data;
+	struct cpufreq_policy *policy;
 	unsigned int freq;
 	unsigned int cached_freq;
 
 	pr_debug("get_cur_freq_on_cpu (%d)\n", cpu);
 
-	if (unlikely(data == NULL ||
-		     data->acpi_data == NULL || data->freq_table == NULL)) {
+	policy = cpufreq_cpu_get(cpu);
+	if (unlikely(!policy))
 		return 0;
-	}
 
-	cached_freq = data->freq_table[data->acpi_data->state].frequency;
-	freq = extract_freq(get_cur_val(cpumask_of(cpu)), data);
+	data = policy->driver_data;
+	cpufreq_cpu_put(policy);
+	if (unlikely(!data || !data->freq_table))
+		return 0;
+
+	cached_freq = data->freq_table[to_perf_data(data)->state].frequency;
+	freq = extract_freq(get_cur_val(cpumask_of(cpu), data), data);
 	if (freq != cached_freq) {
 		/*
 		 * The dreaded BIOS frequency change behind our back.
@@ -397,7 +406,7 @@ static unsigned int check_freqs(const struct cpumask *mask, unsigned int freq,
 	unsigned int i;
 
 	for (i = 0; i < 100; i++) {
-		cur_freq = extract_freq(get_cur_val(mask), data);
+		cur_freq = extract_freq(get_cur_val(mask, data), data);
 		if (cur_freq == freq)
 			return 1;
 		udelay(10);
@@ -408,18 +417,17 @@ static unsigned int check_freqs(const struct cpumask *mask, unsigned int freq,
 static int acpi_cpufreq_target(struct cpufreq_policy *policy,
 			       unsigned int index)
 {
-	struct acpi_cpufreq_data *data = per_cpu(acfreq_data, policy->cpu);
+	struct acpi_cpufreq_data *data = policy->driver_data;
 	struct acpi_processor_performance *perf;
 	struct drv_cmd cmd;
 	unsigned int next_perf_state = 0; /* Index into perf table */
 	int result = 0;
 
-	if (unlikely(data == NULL ||
-	     data->acpi_data == NULL || data->freq_table == NULL)) {
+	if (unlikely(data == NULL || data->freq_table == NULL)) {
 		return -ENODEV;
 	}
 
-	perf = data->acpi_data;
+	perf = to_perf_data(data);
 	next_perf_state = data->freq_table[index].driver_data;
 	if (perf->state == next_perf_state) {
 		if (unlikely(data->resume)) {
@@ -482,8 +490,9 @@ out:
 static unsigned long
 acpi_cpufreq_guess_freq(struct acpi_cpufreq_data *data, unsigned int cpu)
 {
-	struct acpi_processor_performance *perf = data->acpi_data;
+	struct acpi_processor_performance *perf;
 
+	perf = to_perf_data(data);
 	if (cpu_khz) {
 		/* search the closest match to cpu_khz */
 		unsigned int i;
@@ -672,17 +681,17 @@ static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy)
 		goto err_free;
 	}
 
-	data->acpi_data = per_cpu_ptr(acpi_perf_data, cpu);
-	per_cpu(acfreq_data, cpu) = data;
+	perf = per_cpu_ptr(acpi_perf_data, cpu);
+	data->acpi_perf_cpu = cpu;
+	policy->driver_data = data;
 
 	if (cpu_has(c, X86_FEATURE_CONSTANT_TSC))
 		acpi_cpufreq_driver.flags |= CPUFREQ_CONST_LOOPS;
 
-	result = acpi_processor_register_performance(data->acpi_data, cpu);
+	result = acpi_processor_register_performance(perf, cpu);
 	if (result)
 		goto err_free_mask;
 
-	perf = data->acpi_data;
 	policy->shared_type = perf->shared_type;
 
 	/*
@@ -838,26 +847,25 @@ static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy)
 err_freqfree:
 	kfree(data->freq_table);
 err_unreg:
-	acpi_processor_unregister_performance(perf, cpu);
+	acpi_processor_unregister_performance(cpu);
 err_free_mask:
 	free_cpumask_var(data->freqdomain_cpus);
 err_free:
 	kfree(data);
-	per_cpu(acfreq_data, cpu) = NULL;
+	policy->driver_data = NULL;
 
 	return result;
 }
 
 static int acpi_cpufreq_cpu_exit(struct cpufreq_policy *policy)
 {
-	struct acpi_cpufreq_data *data = per_cpu(acfreq_data, policy->cpu);
+	struct acpi_cpufreq_data *data = policy->driver_data;
 
 	pr_debug("acpi_cpufreq_cpu_exit\n");
 
 	if (data) {
-		per_cpu(acfreq_data, policy->cpu) = NULL;
-		acpi_processor_unregister_performance(data->acpi_data,
-						      policy->cpu);
+		policy->driver_data = NULL;
+		acpi_processor_unregister_performance(data->acpi_perf_cpu);
 		free_cpumask_var(data->freqdomain_cpus);
 		kfree(data->freq_table);
 		kfree(data);
@@ -868,7 +876,7 @@ static int acpi_cpufreq_cpu_exit(struct cpufreq_policy *policy)
 
 static int acpi_cpufreq_resume(struct cpufreq_policy *policy)
 {
-	struct acpi_cpufreq_data *data = per_cpu(acfreq_data, policy->cpu);
+	struct acpi_cpufreq_data *data = policy->driver_data;
 
 	pr_debug("acpi_cpufreq_resume\n");
 
@@ -880,7 +888,9 @@ static int acpi_cpufreq_resume(struct cpufreq_policy *policy)
 static struct freq_attr *acpi_cpufreq_attr[] = {
 	&cpufreq_freq_attr_scaling_available_freqs,
 	&freqdomain_cpus,
-	NULL,	/* this is a placeholder for cpb, do not remove */
+#ifdef CONFIG_X86_ACPI_CPUFREQ_CPB
+	&cpb,
+#endif
 	NULL,
 };
 
@@ -953,17 +963,16 @@ static int __init acpi_cpufreq_init(void)
 	 * only if configured. This is considered legacy code, which
 	 * will probably be removed at some point in the future.
 	 */
-	if (check_amd_hwpstate_cpu(0)) {
-		struct freq_attr **iter;
-
-		pr_debug("adding sysfs entry for cpb\n");
+	if (!check_amd_hwpstate_cpu(0)) {
+		struct freq_attr **attr;
 
-		for (iter = acpi_cpufreq_attr; *iter != NULL; iter++)
-			;
+		pr_debug("CPB unsupported, do not expose it\n");
 
-		/* make sure there is a terminator behind it */
-		if (iter[1] == NULL)
-			*iter = &cpb;
+		for (attr = acpi_cpufreq_attr; *attr; attr++)
+			if (*attr == &cpb) {
+				*attr = NULL;
+				break;
+			}
 	}
 #endif
 	acpi_cpufreq_boost_init();
diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
index 7a3c30c4336f..76a26609d96b 100644
--- a/drivers/cpufreq/cpufreq.c
+++ b/drivers/cpufreq/cpufreq.c
@@ -112,12 +112,6 @@ static inline bool has_target(void)
 	return cpufreq_driver->target_index || cpufreq_driver->target;
 }
 
-/*
- * rwsem to guarantee that cpufreq driver module doesn't unload during critical
- * sections
- */
-static DECLARE_RWSEM(cpufreq_rwsem);
-
 /* internal prototypes */
 static int __cpufreq_governor(struct cpufreq_policy *policy,
 		unsigned int event);
@@ -277,10 +271,6 @@ EXPORT_SYMBOL_GPL(cpufreq_generic_get);
  * If corresponding call cpufreq_cpu_put() isn't made, the policy wouldn't be
  * freed as that depends on the kobj count.
  *
- * It also takes a read-lock of 'cpufreq_rwsem' and doesn't put it back if a
- * valid policy is found. This is done to make sure the driver doesn't get
- * unregistered while the policy is being used.
- *
  * Return: A valid policy on success, otherwise NULL on failure.
  */
 struct cpufreq_policy *cpufreq_cpu_get(unsigned int cpu)
@@ -291,9 +281,6 @@ struct cpufreq_policy *cpufreq_cpu_get(unsigned int cpu)
 	if (WARN_ON(cpu >= nr_cpu_ids))
 		return NULL;
 
-	if (!down_read_trylock(&cpufreq_rwsem))
-		return NULL;
-
 	/* get the cpufreq driver */
 	read_lock_irqsave(&cpufreq_driver_lock, flags);
 
@@ -306,9 +293,6 @@ struct cpufreq_policy *cpufreq_cpu_get(unsigned int cpu)
 
 	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
 
-	if (!policy)
-		up_read(&cpufreq_rwsem);
-
 	return policy;
 }
 EXPORT_SYMBOL_GPL(cpufreq_cpu_get);
@@ -320,13 +304,10 @@ EXPORT_SYMBOL_GPL(cpufreq_cpu_get);
  *
  * This decrements the kobject reference count incremented earlier by calling
  * cpufreq_cpu_get().
- *
- * It also drops the read-lock of 'cpufreq_rwsem' taken at cpufreq_cpu_get().
  */
 void cpufreq_cpu_put(struct cpufreq_policy *policy)
 {
 	kobject_put(&policy->kobj);
-	up_read(&cpufreq_rwsem);
 }
 EXPORT_SYMBOL_GPL(cpufreq_cpu_put);
 
@@ -851,9 +832,6 @@ static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
 	struct freq_attr *fattr = to_attr(attr);
 	ssize_t ret;
 
-	if (!down_read_trylock(&cpufreq_rwsem))
-		return -EINVAL;
-
 	down_read(&policy->rwsem);
 
 	if (fattr->show)
@@ -862,7 +840,6 @@ static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
 		ret = -EIO;
 
 	up_read(&policy->rwsem);
-	up_read(&cpufreq_rwsem);
 
 	return ret;
 }
@@ -879,9 +856,6 @@ static ssize_t store(struct kobject *kobj, struct attribute *attr,
 	if (!cpu_online(policy->cpu))
 		goto unlock;
 
-	if (!down_read_trylock(&cpufreq_rwsem))
-		goto unlock;
-
 	down_write(&policy->rwsem);
 
 	/* Updating inactive policies is invalid, so avoid doing that. */
@@ -897,8 +871,6 @@ static ssize_t store(struct kobject *kobj, struct attribute *attr,
 
 unlock_policy_rwsem:
 	up_write(&policy->rwsem);
-
-	up_read(&cpufreq_rwsem);
 unlock:
 	put_online_cpus();
 
@@ -1027,8 +999,7 @@ static void cpufreq_remove_dev_symlink(struct cpufreq_policy *policy)
 	}
 }
 
-static int cpufreq_add_dev_interface(struct cpufreq_policy *policy,
-				     struct device *dev)
+static int cpufreq_add_dev_interface(struct cpufreq_policy *policy)
 {
 	struct freq_attr **drv_attr;
 	int ret = 0;
@@ -1060,11 +1031,10 @@ static int cpufreq_add_dev_interface(struct cpufreq_policy *policy,
 	return cpufreq_add_dev_symlink(policy);
 }
 
-static void cpufreq_init_policy(struct cpufreq_policy *policy)
+static int cpufreq_init_policy(struct cpufreq_policy *policy)
 {
 	struct cpufreq_governor *gov = NULL;
 	struct cpufreq_policy new_policy;
-	int ret = 0;
 
 	memcpy(&new_policy, policy, sizeof(*policy));
 
@@ -1083,16 +1053,10 @@ static void cpufreq_init_policy(struct cpufreq_policy *policy)
 		cpufreq_parse_governor(gov->name, &new_policy.policy, NULL);
 
 	/* set default policy */
-	ret = cpufreq_set_policy(policy, &new_policy);
-	if (ret) {
-		pr_debug("setting policy failed\n");
-		if (cpufreq_driver->exit)
-			cpufreq_driver->exit(policy);
-	}
+	return cpufreq_set_policy(policy, &new_policy);
 }
 
-static int cpufreq_add_policy_cpu(struct cpufreq_policy *policy,
-				  unsigned int cpu, struct device *dev)
+static int cpufreq_add_policy_cpu(struct cpufreq_policy *policy, unsigned int cpu)
 {
 	int ret = 0;
 
@@ -1126,33 +1090,15 @@ static int cpufreq_add_policy_cpu(struct cpufreq_policy *policy,
 	return 0;
 }
 
-static struct cpufreq_policy *cpufreq_policy_restore(unsigned int cpu)
-{
-	struct cpufreq_policy *policy;
-	unsigned long flags;
-
-	read_lock_irqsave(&cpufreq_driver_lock, flags);
-	policy = per_cpu(cpufreq_cpu_data, cpu);
-	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
-
-	if (likely(policy)) {
-		/* Policy should be inactive here */
-		WARN_ON(!policy_is_inactive(policy));
-
-		down_write(&policy->rwsem);
-		policy->cpu = cpu;
-		policy->governor = NULL;
-		up_write(&policy->rwsem);
-	}
-
-	return policy;
-}
-
-static struct cpufreq_policy *cpufreq_policy_alloc(struct device *dev)
+static struct cpufreq_policy *cpufreq_policy_alloc(unsigned int cpu)
 {
+	struct device *dev = get_cpu_device(cpu);
 	struct cpufreq_policy *policy;
 	int ret;
 
+	if (WARN_ON(!dev))
+		return NULL;
+
 	policy = kzalloc(sizeof(*policy), GFP_KERNEL);
 	if (!policy)
 		return NULL;
@@ -1180,10 +1126,10 @@ static struct cpufreq_policy *cpufreq_policy_alloc(struct device *dev)
 	init_completion(&policy->kobj_unregister);
 	INIT_WORK(&policy->update, handle_update);
 
-	policy->cpu = dev->id;
+	policy->cpu = cpu;
 
 	/* Set this once on allocation */
-	policy->kobj_cpu = dev->id;
+	policy->kobj_cpu = cpu;
 
 	return policy;
 
@@ -1245,59 +1191,34 @@ static void cpufreq_policy_free(struct cpufreq_policy *policy, bool notify)
 	kfree(policy);
 }
 
-/**
- * cpufreq_add_dev - add a CPU device
- *
- * Adds the cpufreq interface for a CPU device.
- *
- * The Oracle says: try running cpufreq registration/unregistration concurrently
- * with with cpu hotplugging and all hell will break loose. Tried to clean this
- * mess up, but more thorough testing is needed. - Mathieu
- */
-static int cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
+static int cpufreq_online(unsigned int cpu)
 {
-	unsigned int j, cpu = dev->id;
-	int ret = -ENOMEM;
 	struct cpufreq_policy *policy;
+	bool new_policy;
 	unsigned long flags;
-	bool recover_policy = !sif;
-
-	pr_debug("adding CPU %u\n", cpu);
-
-	if (cpu_is_offline(cpu)) {
-		/*
-		 * Only possible if we are here from the subsys_interface add
-		 * callback.  A hotplug notifier will follow and we will handle
-		 * it as CPU online then.  For now, just create the sysfs link,
-		 * unless there is no policy or the link is already present.
-		 */
-		policy = per_cpu(cpufreq_cpu_data, cpu);
-		return policy && !cpumask_test_and_set_cpu(cpu, policy->real_cpus)
-			? add_cpu_dev_symlink(policy, cpu) : 0;
-	}
+	unsigned int j;
+	int ret;
 
-	if (!down_read_trylock(&cpufreq_rwsem))
-		return 0;
+	pr_debug("%s: bringing CPU%u online\n", __func__, cpu);
 
 	/* Check if this CPU already has a policy to manage it */
 	policy = per_cpu(cpufreq_cpu_data, cpu);
-	if (policy && !policy_is_inactive(policy)) {
+	if (policy) {
 		WARN_ON(!cpumask_test_cpu(cpu, policy->related_cpus));
-		ret = cpufreq_add_policy_cpu(policy, cpu, dev);
-		up_read(&cpufreq_rwsem);
-		return ret;
-	}
+		if (!policy_is_inactive(policy))
+			return cpufreq_add_policy_cpu(policy, cpu);
 
-	/*
-	 * Restore the saved policy when doing light-weight init and fall back
-	 * to the full init if that fails.
-	 */
-	policy = recover_policy ? cpufreq_policy_restore(cpu) : NULL;
-	if (!policy) {
-		recover_policy = false;
-		policy = cpufreq_policy_alloc(dev);
+		/* This is the only online CPU for the policy.  Start over. */
+		new_policy = false;
+		down_write(&policy->rwsem);
+		policy->cpu = cpu;
+		policy->governor = NULL;
+		up_write(&policy->rwsem);
+	} else {
+		new_policy = true;
+		policy = cpufreq_policy_alloc(cpu);
 		if (!policy)
-			goto nomem_out;
+			return -ENOMEM;
 	}
 
 	cpumask_copy(policy->cpus, cpumask_of(cpu));
@@ -1308,17 +1229,17 @@ static int cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
 	ret = cpufreq_driver->init(policy);
 	if (ret) {
 		pr_debug("initialization failed\n");
-		goto err_set_policy_cpu;
+		goto out_free_policy;
 	}
 
 	down_write(&policy->rwsem);
 
-	/* related cpus should atleast have policy->cpus */
-	cpumask_or(policy->related_cpus, policy->related_cpus, policy->cpus);
-
-	/* Remember which CPUs have been present at the policy creation time. */
-	if (!recover_policy)
+	if (new_policy) {
+		/* related_cpus should at least include policy->cpus. */
+		cpumask_or(policy->related_cpus, policy->related_cpus, policy->cpus);
+		/* Remember CPUs present at the policy creation time. */
 		cpumask_and(policy->real_cpus, policy->cpus, cpu_present_mask);
+	}
 
 	/*
 	 * affected cpus must always be the one, which are online. We aren't
@@ -1326,7 +1247,7 @@ static int cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
 	 */
 	cpumask_and(policy->cpus, policy->cpus, cpu_online_mask);
 
-	if (!recover_policy) {
+	if (new_policy) {
 		policy->user_policy.min = policy->min;
 		policy->user_policy.max = policy->max;
 
@@ -1340,7 +1261,7 @@ static int cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
 		policy->cur = cpufreq_driver->get(policy->cpu);
 		if (!policy->cur) {
 			pr_err("%s: ->get() failed\n", __func__);
-			goto err_get_freq;
+			goto out_exit_policy;
 		}
 	}
 
@@ -1387,10 +1308,10 @@ static int cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
 	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
 				     CPUFREQ_START, policy);
 
-	if (!recover_policy) {
-		ret = cpufreq_add_dev_interface(policy, dev);
+	if (new_policy) {
+		ret = cpufreq_add_dev_interface(policy);
 		if (ret)
-			goto err_out_unregister;
+			goto out_exit_policy;
 		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
 				CPUFREQ_CREATE_POLICY, policy);
 
@@ -1399,9 +1320,16 @@ static int cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
 		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
 	}
 
-	cpufreq_init_policy(policy);
+	ret = cpufreq_init_policy(policy);
+	if (ret) {
+		pr_err("%s: Failed to initialize policy for cpu: %d (%d)\n",
+		       __func__, cpu, ret);
+		/* cpufreq_policy_free() will notify based on this */
+		new_policy = false;
+		goto out_exit_policy;
+	}
 
-	if (!recover_policy) {
+	if (new_policy) {
 		policy->user_policy.policy = policy->policy;
 		policy->user_policy.governor = policy->governor;
 	}
@@ -1409,8 +1337,6 @@ static int cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
 
 	kobject_uevent(&policy->kobj, KOBJ_ADD);
 
-	up_read(&cpufreq_rwsem);
-
 	/* Callback for handling stuff after policy is ready */
 	if (cpufreq_driver->ready)
 		cpufreq_driver->ready(policy);
@@ -1419,24 +1345,47 @@ static int cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
 
 	return 0;
 
-err_out_unregister:
-err_get_freq:
+out_exit_policy:
 	up_write(&policy->rwsem);
 
 	if (cpufreq_driver->exit)
 		cpufreq_driver->exit(policy);
-err_set_policy_cpu:
-	cpufreq_policy_free(policy, recover_policy);
-nomem_out:
-	up_read(&cpufreq_rwsem);
+out_free_policy:
+	cpufreq_policy_free(policy, !new_policy);
+	return ret;
+}
+
+/**
+ * cpufreq_add_dev - the cpufreq interface for a CPU device.
+ * @dev: CPU device.
+ * @sif: Subsystem interface structure pointer (not used)
+ */
+static int cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
+{
+	unsigned cpu = dev->id;
+	int ret;
+
+	dev_dbg(dev, "%s: adding CPU%u\n", __func__, cpu);
+
+	if (cpu_online(cpu)) {
+		ret = cpufreq_online(cpu);
+	} else {
+		/*
+		 * A hotplug notifier will follow and we will handle it as CPU
+		 * online then.  For now, just create the sysfs link, unless
+		 * there is no policy or the link is already present.
+		 */
+		struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);
+
+		ret = policy && !cpumask_test_and_set_cpu(cpu, policy->real_cpus)
+			? add_cpu_dev_symlink(policy, cpu) : 0;
+	}
 
 	return ret;
 }
 
-static int __cpufreq_remove_dev_prepare(struct device *dev)
+static void cpufreq_offline_prepare(unsigned int cpu)
 {
-	unsigned int cpu = dev->id;
-	int ret = 0;
 	struct cpufreq_policy *policy;
 
 	pr_debug("%s: unregistering CPU %u\n", __func__, cpu);
@@ -1444,11 +1393,11 @@ static int __cpufreq_remove_dev_prepare(struct device *dev)
 	policy = cpufreq_cpu_get_raw(cpu);
 	if (!policy) {
 		pr_debug("%s: No cpu_data found\n", __func__);
-		return -EINVAL;
+		return;
 	}
 
 	if (has_target()) {
-		ret = __cpufreq_governor(policy, CPUFREQ_GOV_STOP);
+		int ret = __cpufreq_governor(policy, CPUFREQ_GOV_STOP);
 		if (ret)
 			pr_err("%s: Failed to stop governor\n", __func__);
 	}
@@ -1469,7 +1418,7 @@ static int __cpufreq_remove_dev_prepare(struct device *dev)
 	/* Start governor again for active policy */
 	if (!policy_is_inactive(policy)) {
 		if (has_target()) {
-			ret = __cpufreq_governor(policy, CPUFREQ_GOV_START);
+			int ret = __cpufreq_governor(policy, CPUFREQ_GOV_START);
 			if (!ret)
 				ret = __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);
 
@@ -1479,28 +1428,24 @@ static int __cpufreq_remove_dev_prepare(struct device *dev)
 	} else if (cpufreq_driver->stop_cpu) {
 		cpufreq_driver->stop_cpu(policy);
 	}
-
-	return ret;
 }
 
-static int __cpufreq_remove_dev_finish(struct device *dev)
+static void cpufreq_offline_finish(unsigned int cpu)
 {
-	unsigned int cpu = dev->id;
-	int ret;
 	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);
 
 	if (!policy) {
 		pr_debug("%s: No cpu_data found\n", __func__);
-		return -EINVAL;
+		return;
 	}
 
 	/* Only proceed for inactive policies */
 	if (!policy_is_inactive(policy))
-		return 0;
+		return;
 
 	/* If cpu is last user of policy, free policy */
 	if (has_target()) {
-		ret = __cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
+		int ret = __cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
 		if (ret)
 			pr_err("%s: Failed to exit governor\n", __func__);
 	}
@@ -1512,8 +1457,6 @@ static int __cpufreq_remove_dev_finish(struct device *dev)
 	 */
 	if (cpufreq_driver->exit)
 		cpufreq_driver->exit(policy);
-
-	return 0;
 }
 
 /**
@@ -1530,8 +1473,8 @@ static int cpufreq_remove_dev(struct device *dev, struct subsys_interface *sif)
 		return 0;
 
 	if (cpu_online(cpu)) {
-		__cpufreq_remove_dev_prepare(dev);
-		__cpufreq_remove_dev_finish(dev);
+		cpufreq_offline_prepare(cpu);
+		cpufreq_offline_finish(cpu);
 	}
 
 	cpumask_clear_cpu(cpu, policy->real_cpus);
@@ -2247,7 +2190,11 @@ static int cpufreq_set_policy(struct cpufreq_policy *policy,
 
 	memcpy(&new_policy->cpuinfo, &policy->cpuinfo, sizeof(policy->cpuinfo));
 
-	if (new_policy->min > policy->max || new_policy->max < policy->min)
+	/*
+	* This check works well when we store new min/max freq attributes,
+	* because new_policy is a copy of policy with one field updated.
+	*/
+	if (new_policy->min > new_policy->max)
 		return -EINVAL;
 
 	/* verify the cpu speed can be set within this limit */
@@ -2296,16 +2243,31 @@ static int cpufreq_set_policy(struct cpufreq_policy *policy,
 	old_gov = policy->governor;
 	/* end old governor */
 	if (old_gov) {
-		__cpufreq_governor(policy, CPUFREQ_GOV_STOP);
+		ret = __cpufreq_governor(policy, CPUFREQ_GOV_STOP);
+		if (ret) {
+			/* This can happen due to race with other operations */
+			pr_debug("%s: Failed to Stop Governor: %s (%d)\n",
+				 __func__, old_gov->name, ret);
+			return ret;
+		}
+
 		up_write(&policy->rwsem);
-		__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
+		ret = __cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
 		down_write(&policy->rwsem);
+
+		if (ret) {
+			pr_err("%s: Failed to Exit Governor: %s (%d)\n",
+			       __func__, old_gov->name, ret);
+			return ret;
+		}
 	}
 
 	/* start new governor */
 	policy->governor = new_policy->governor;
-	if (!__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_INIT)) {
-		if (!__cpufreq_governor(policy, CPUFREQ_GOV_START))
+	ret = __cpufreq_governor(policy, CPUFREQ_GOV_POLICY_INIT);
+	if (!ret) {
+		ret = __cpufreq_governor(policy, CPUFREQ_GOV_START);
+		if (!ret)
 			goto out;
 
 		up_write(&policy->rwsem);
@@ -2317,11 +2279,13 @@ static int cpufreq_set_policy(struct cpufreq_policy *policy,
 	pr_debug("starting governor %s failed\n", policy->governor->name);
 	if (old_gov) {
 		policy->governor = old_gov;
-		__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_INIT);
-		__cpufreq_governor(policy, CPUFREQ_GOV_START);
+		if (__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_INIT))
+			policy->governor = NULL;
+		else
+			__cpufreq_governor(policy, CPUFREQ_GOV_START);
 	}
 
-	return -EINVAL;
+	return ret;
 
  out:
 	pr_debug("governor: change or update limits\n");
@@ -2387,27 +2351,23 @@ static int cpufreq_cpu_callback(struct notifier_block *nfb,
 					unsigned long action, void *hcpu)
 {
 	unsigned int cpu = (unsigned long)hcpu;
-	struct device *dev;
 
-	dev = get_cpu_device(cpu);
-	if (dev) {
-		switch (action & ~CPU_TASKS_FROZEN) {
-		case CPU_ONLINE:
-			cpufreq_add_dev(dev, NULL);
-			break;
+	switch (action & ~CPU_TASKS_FROZEN) {
+	case CPU_ONLINE:
+		cpufreq_online(cpu);
+		break;
 
-		case CPU_DOWN_PREPARE:
-			__cpufreq_remove_dev_prepare(dev);
-			break;
+	case CPU_DOWN_PREPARE:
+		cpufreq_offline_prepare(cpu);
+		break;
 
-		case CPU_POST_DEAD:
-			__cpufreq_remove_dev_finish(dev);
-			break;
+	case CPU_POST_DEAD:
+		cpufreq_offline_finish(cpu);
+		break;
 
-		case CPU_DOWN_FAILED:
-			cpufreq_add_dev(dev, NULL);
-			break;
-		}
+	case CPU_DOWN_FAILED:
+		cpufreq_online(cpu);
+		break;
 	}
 	return NOTIFY_OK;
 }
@@ -2515,10 +2475,14 @@ int cpufreq_register_driver(struct cpufreq_driver *driver_data)
 
 	pr_debug("trying to register driver %s\n", driver_data->name);
 
+	/* Protect against concurrent CPU online/offline. */
+	get_online_cpus();
+
 	write_lock_irqsave(&cpufreq_driver_lock, flags);
 	if (cpufreq_driver) {
 		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
-		return -EEXIST;
+		ret = -EEXIST;
+		goto out;
 	}
 	cpufreq_driver = driver_data;
 	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
@@ -2557,7 +2521,10 @@ int cpufreq_register_driver(struct cpufreq_driver *driver_data)
 	register_hotcpu_notifier(&cpufreq_cpu_notifier);
 	pr_debug("driver %s up and running\n", driver_data->name);
 
-	return 0;
+out:
+	put_online_cpus();
+	return ret;
+
 err_if_unreg:
 	subsys_interface_unregister(&cpufreq_interface);
 err_boost_unreg:
@@ -2567,7 +2534,7 @@ err_null_driver:
 	write_lock_irqsave(&cpufreq_driver_lock, flags);
 	cpufreq_driver = NULL;
 	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
-	return ret;
+	goto out;
 }
 EXPORT_SYMBOL_GPL(cpufreq_register_driver);
 
@@ -2588,19 +2555,20 @@ int cpufreq_unregister_driver(struct cpufre