X-Git-Url: http://ftp.safe.ca/?a=blobdiff_plain;f=drivers%2Fcpufreq%2Fcpufreq.c;h=fd69086d08d54f0fb9c2036984cc9c9c88d86446;hb=4bc5d34135039566b8d6efa2de7515b2be505da8;hp=0c18ac2fe7c2c1299d37781c91ce01736fc2d1a5;hpb=e08f5f5bb5dfaaa28d69ffe37eb774533297657f;p=safe%2Fjmp%2Flinux-2.6

diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
index 0c18ac2..fd69086 100644
--- a/drivers/cpufreq/cpufreq.c
+++ b/drivers/cpufreq/cpufreq.c
@@ -38,13 +38,76 @@
  * also protects the cpufreq_cpu_data array.
  */
 static struct cpufreq_driver *cpufreq_driver;
-static struct cpufreq_policy *cpufreq_cpu_data[NR_CPUS];
+static DEFINE_PER_CPU(struct cpufreq_policy *, cpufreq_cpu_data);
+#ifdef CONFIG_HOTPLUG_CPU
+/* This one keeps track of the previously set governor of a removed CPU */
+static DEFINE_PER_CPU(struct cpufreq_governor *, cpufreq_cpu_governor);
+#endif
 static DEFINE_SPINLOCK(cpufreq_driver_lock);
 
+/*
+ * cpu_policy_rwsem is a per CPU reader-writer semaphore designed to cure
+ * all cpufreq/hotplug/workqueue/etc related lock issues.
+ *
+ * The rules for this semaphore:
+ * - Any routine that wants to read from the policy structure will
+ *   do a down_read on this semaphore.
+ * - Any routine that will write to the policy structure and/or may take away
+ *   the policy altogether (eg. CPU hotplug), will hold this lock in write
+ *   mode before doing so.
+ *
+ * Additional rules:
+ * - All holders of the lock should check to make sure that the CPU they
+ *   are concerned with are online after they get the lock.
+ * - Governor routines that can be called in cpufreq hotplug path should not
+ *   take this sem as top level hotplug notifier handler takes this.
+ */
+static DEFINE_PER_CPU(int, policy_cpu);
+static DEFINE_PER_CPU(struct rw_semaphore, cpu_policy_rwsem);
+
+#define lock_policy_rwsem(mode, cpu)					\
+int lock_policy_rwsem_##mode						\
+(int cpu)								\
+{									\
+	int policy_cpu = per_cpu(policy_cpu, cpu);			\
+	BUG_ON(policy_cpu == -1);					\
+	down_##mode(&per_cpu(cpu_policy_rwsem, policy_cpu));		\
+	if (unlikely(!cpu_online(cpu))) {				\
+		up_##mode(&per_cpu(cpu_policy_rwsem, policy_cpu));	\
+		return -1;						\
+	}								\
+									\
+	return 0;							\
+}
+
+lock_policy_rwsem(read, cpu);
+EXPORT_SYMBOL_GPL(lock_policy_rwsem_read);
+
+lock_policy_rwsem(write, cpu);
+EXPORT_SYMBOL_GPL(lock_policy_rwsem_write);
+
+void unlock_policy_rwsem_read(int cpu)
+{
+	int policy_cpu = per_cpu(policy_cpu, cpu);
+	BUG_ON(policy_cpu == -1);
+	up_read(&per_cpu(cpu_policy_rwsem, policy_cpu));
+}
+EXPORT_SYMBOL_GPL(unlock_policy_rwsem_read);
+
+void unlock_policy_rwsem_write(int cpu)
+{
+	int policy_cpu = per_cpu(policy_cpu, cpu);
+	BUG_ON(policy_cpu == -1);
+	up_write(&per_cpu(cpu_policy_rwsem, policy_cpu));
+}
+EXPORT_SYMBOL_GPL(unlock_policy_rwsem_write);
+
+
 /* internal prototypes */
 static int __cpufreq_governor(struct cpufreq_policy *policy,
-						unsigned int event);
-static void handle_update(void *data);
+		unsigned int event);
+static unsigned int __cpufreq_get(unsigned int cpu);
+static void handle_update(struct work_struct *work);
 
 /**
  * Two notifier lists: the "policy" list is involved in the
@@ -56,22 +119,24 @@ static void handle_update(void *data);
 static BLOCKING_NOTIFIER_HEAD(cpufreq_policy_notifier_list);
 static struct srcu_notifier_head cpufreq_transition_notifier_list;
 
+static bool init_cpufreq_transition_notifier_list_called;
 static int __init init_cpufreq_transition_notifier_list(void)
 {
 	srcu_init_notifier_head(&cpufreq_transition_notifier_list);
+	init_cpufreq_transition_notifier_list_called = true;
 	return 0;
 }
-core_initcall(init_cpufreq_transition_notifier_list);
+pure_initcall(init_cpufreq_transition_notifier_list);
 
 static LIST_HEAD(cpufreq_governor_list);
-static DEFINE_MUTEX (cpufreq_governor_mutex);
+static DEFINE_MUTEX(cpufreq_governor_mutex);
 
 struct cpufreq_policy *cpufreq_cpu_get(unsigned int cpu)
 {
 	struct cpufreq_policy *data;
 	unsigned long flags;
 
-	if (cpu >= NR_CPUS)
+	if (cpu >= nr_cpu_ids)
 		goto err_out;
 
 	/* get the cpufreq driver */
@@ -85,7 +150,7 @@ struct cpufreq_policy *cpufreq_cpu_get(unsigned int cpu)
 
 
 	/* get the CPU */
-	data = cpufreq_cpu_data[cpu];
+	data = per_cpu(cpufreq_cpu_data, cpu);
 
 	if (!data)
 		goto err_out_put_module;
@@ -154,7 +219,7 @@ static void cpufreq_debug_disable_ratelimit(void)
 }
 
 void cpufreq_debug_printk(unsigned int type, const char *prefix,
-							const char *fmt, ...)
+			const char *fmt, ...)
 {
 	char s[256];
 	va_list args;
@@ -225,7 +290,7 @@ static void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci)
 	if (!l_p_j_ref_freq) {
 		l_p_j_ref = loops_per_jiffy;
 		l_p_j_ref_freq = ci->old;
-		dprintk("saving %lu as reference value for loops_per_jiffy;"
+		dprintk("saving %lu as reference value for loops_per_jiffy; "
 			"freq is %u kHz\n", l_p_j_ref, l_p_j_ref_freq);
 	}
 	if ((val == CPUFREQ_PRECHANGE  && ci->old < ci->new) ||
@@ -233,7 +298,7 @@ static void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci)
 	    (val == CPUFREQ_RESUMECHANGE || val == CPUFREQ_SUSPENDCHANGE)) {
 		loops_per_jiffy = cpufreq_scale(l_p_j_ref, l_p_j_ref_freq,
 								ci->new);
-		dprintk("scaling loops_per_jiffy to %lu"
+		dprintk("scaling loops_per_jiffy to %lu "
 			"for frequency %u kHz\n", loops_per_jiffy, ci->new);
 	}
 }
@@ -263,7 +328,7 @@ void cpufreq_notify_transition(struct cpufreq_freqs *freqs, unsigned int state)
 	dprintk("notification %u of frequency transition to %u kHz\n",
 		state, freqs->new);
 
-	policy = cpufreq_cpu_data[freqs->cpu];
+	policy = per_cpu(cpufreq_cpu_data, freqs->cpu);
 	switch (state) {
 
 	case CPUFREQ_PRECHANGE:
@@ -307,7 +372,7 @@ static struct cpufreq_governor *__find_governor(const char *str_governor)
 	struct cpufreq_governor *t;
 
 	list_for_each_entry(t, &cpufreq_governor_list, governor_list)
-		if (!strnicmp(str_governor,t->name,CPUFREQ_NAME_LEN))
+		if (!strnicmp(str_governor, t->name, CPUFREQ_NAME_LEN))
 			return t;
 
 	return NULL;
@@ -316,7 +381,7 @@ static struct cpufreq_governor *__find_governor(const char *str_governor)
 /**
  * cpufreq_parse_governor - parse a governor string
  */
-static int cpufreq_parse_governor (char *str_governor, unsigned int *policy,
+static int cpufreq_parse_governor(char *str_governor, unsigned int *policy,
 				struct cpufreq_governor **governor)
 {
 	int err = -EINVAL;
@@ -348,7 +413,7 @@ static int cpufreq_parse_governor (char *str_governor, unsigned int *policy,
 				int ret;
 
 				mutex_unlock(&cpufreq_governor_mutex);
-				ret = request_module(name);
+				ret = request_module("%s", name);
 				mutex_lock(&cpufreq_governor_mutex);
 
 				if (ret == 0)
@@ -365,15 +430,11 @@ static int cpufreq_parse_governor (char *str_governor, unsigned int *policy,
 
 		mutex_unlock(&cpufreq_governor_mutex);
 	}
-  out:
+out:
 	return err;
 }
 
 
-/* drivers/base/cpu.c */
-extern struct sysdev_class cpu_sysdev_class;
-
-
 /**
  * cpufreq_per_cpu_attr_read() / show_##file_name() -
  * print out cpufreq information
@@ -384,13 +445,14 @@ extern struct sysdev_class cpu_sysdev_class;
 
 #define show_one(file_name, object)			\
 static ssize_t show_##file_name				\
-(struct cpufreq_policy * policy, char *buf)		\
+(struct cpufreq_policy *policy, char *buf)		\
 {							\
-	return sprintf (buf, "%u\n", policy->object);	\
+	return sprintf(buf, "%u\n", policy->object);	\
 }
 
 show_one(cpuinfo_min_freq, cpuinfo.min_freq);
 show_one(cpuinfo_max_freq, cpuinfo.max_freq);
+show_one(cpuinfo_transition_latency, cpuinfo.transition_latency);
 show_one(scaling_min_freq, min);
 show_one(scaling_max_freq, max);
 show_one(scaling_cur_freq, cur);
@@ -403,7 +465,7 @@ static int __cpufreq_set_policy(struct cpufreq_policy *data,
  */
 #define store_one(file_name, object)			\
 static ssize_t store_##file_name					\
-(struct cpufreq_policy * policy, const char *buf, size_t count)		\
+(struct cpufreq_policy *policy, const char *buf, size_t count)		\
 {									\
 	unsigned int ret = -EINVAL;					\
 	struct cpufreq_policy new_policy;				\
@@ -412,30 +474,26 @@ static ssize_t store_##file_name					\
 	if (ret)							\
 		return -EINVAL;						\
 									\
-	ret = sscanf (buf, "%u", &new_policy.object);			\
+	ret = sscanf(buf, "%u", &new_policy.object);			\
 	if (ret != 1)							\
 		return -EINVAL;						\
 									\
-	lock_cpu_hotplug();						\
-	mutex_lock(&policy->lock);					\
 	ret = __cpufreq_set_policy(policy, &new_policy);		\
 	policy->user_policy.object = policy->object;			\
-	mutex_unlock(&policy->lock);					\
-	unlock_cpu_hotplug();						\
 									\
 	return ret ? ret : count;					\
 }
 
-store_one(scaling_min_freq,min);
-store_one(scaling_max_freq,max);
+store_one(scaling_min_freq, min);
+store_one(scaling_max_freq, max);
 
 /**
  * show_cpuinfo_cur_freq - current CPU frequency as detected by hardware
  */
-static ssize_t show_cpuinfo_cur_freq (struct cpufreq_policy * policy,
-							char *buf)
+static ssize_t show_cpuinfo_cur_freq(struct cpufreq_policy *policy,
+					char *buf)
 {
-	unsigned int cur_freq = cpufreq_get(policy->cpu);
+	unsigned int cur_freq = __cpufreq_get(policy->cpu);
 	if (!cur_freq)
 		return sprintf(buf, "<unknown>");
 	return sprintf(buf, "%u\n", cur_freq);
@@ -445,15 +503,15 @@ static ssize_t show_cpuinfo_cur_freq (struct cpufreq_policy * policy,
 /**
  * show_scaling_governor - show the current policy for the specified CPU
  */
-static ssize_t show_scaling_governor (struct cpufreq_policy * policy,
-							char *buf)
+static ssize_t show_scaling_governor(struct cpufreq_policy *policy, char *buf)
 {
-	if(policy->policy == CPUFREQ_POLICY_POWERSAVE)
+	if (policy->policy == CPUFREQ_POLICY_POWERSAVE)
 		return sprintf(buf, "powersave\n");
 	else if (policy->policy == CPUFREQ_POLICY_PERFORMANCE)
 		return sprintf(buf, "performance\n");
 	else if (policy->governor)
-		return scnprintf(buf, CPUFREQ_NAME_LEN, "%s\n", policy->governor->name);
+		return scnprintf(buf, CPUFREQ_NAME_LEN, "%s\n",
+				policy->governor->name);
 	return -EINVAL;
 }
 
@@ -461,8 +519,8 @@ static ssize_t show_scaling_governor (struct cpufreq_policy * policy,
 /**
  * store_scaling_governor - store policy for the specified CPU
  */
-static ssize_t store_scaling_governor (struct cpufreq_policy * policy,
-				       const char *buf, size_t count)
+static ssize_t store_scaling_governor(struct cpufreq_policy *policy,
+					const char *buf, size_t count)
 {
 	unsigned int ret = -EINVAL;
 	char	str_governor[16];
@@ -472,7 +530,7 @@ static ssize_t store_scaling_governor (struct cpufreq_policy * policy,
 	if (ret)
 		return ret;
 
-	ret = sscanf (buf, "%15s", str_governor);
+	ret = sscanf(buf, "%15s", str_governor);
 	if (ret != 1)
 		return -EINVAL;
 
@@ -480,18 +538,12 @@ static ssize_t store_scaling_governor (struct cpufreq_policy * policy,
 						&new_policy.governor))
 		return -EINVAL;
 
-	lock_cpu_hotplug();
-
 	/* Do not use cpufreq_set_policy here or the user_policy.max
 	   will be wrongly overridden */
-	mutex_lock(&policy->lock);
 	ret = __cpufreq_set_policy(policy, &new_policy);
 
 	policy->user_policy.policy = policy->policy;
 	policy->user_policy.governor = policy->governor;
-	mutex_unlock(&policy->lock);
-
-	unlock_cpu_hotplug();
 
 	if (ret)
 		return ret;
@@ -502,7 +554,7 @@ static ssize_t store_scaling_governor (struct cpufreq_policy * policy,
 /**
  * show_scaling_driver - show the cpufreq driver currently loaded
  */
-static ssize_t show_scaling_driver (struct cpufreq_policy * policy, char *buf)
+static ssize_t show_scaling_driver(struct cpufreq_policy *policy, char *buf)
 {
 	return scnprintf(buf, CPUFREQ_NAME_LEN, "%s\n", cpufreq_driver->name);
 }
@@ -510,8 +562,8 @@ static ssize_t show_scaling_driver (struct cpufreq_policy * policy, char *buf)
 /**
  * show_scaling_available_governors - show the available CPUfreq governors
  */
-static ssize_t show_scaling_available_governors (struct cpufreq_policy *policy,
-				char *buf)
+static ssize_t show_scaling_available_governors(struct cpufreq_policy *policy,
+						char *buf)
 {
 	ssize_t i = 0;
 	struct cpufreq_governor *t;
@@ -522,7 +574,8 @@ static ssize_t show_scaling_available_governors (struct cpufreq_policy *policy,
 	}
 
 	list_for_each_entry(t, &cpufreq_governor_list, governor_list) {
-		if (i >= (ssize_t) ((PAGE_SIZE / sizeof(char)) - (CPUFREQ_NAME_LEN + 2)))
+		if (i >= (ssize_t) ((PAGE_SIZE / sizeof(char))
+		    - (CPUFREQ_NAME_LEN + 2)))
 			goto out;
 		i += scnprintf(&buf[i], CPUFREQ_NAME_LEN, "%s ", t->name);
 	}
@@ -530,25 +583,67 @@ out:
 	i += sprintf(&buf[i], "\n");
 	return i;
 }
-/**
- * show_affected_cpus - show the CPUs affected by each transition
- */
-static ssize_t show_affected_cpus (struct cpufreq_policy * policy, char *buf)
+
+static ssize_t show_cpus(const struct cpumask *mask, char *buf)
 {
 	ssize_t i = 0;
 	unsigned int cpu;
 
-	for_each_cpu_mask(cpu, policy->cpus) {
+	for_each_cpu(cpu, mask) {
 		if (i)
 			i += scnprintf(&buf[i], (PAGE_SIZE - i - 2), " ");
 		i += scnprintf(&buf[i], (PAGE_SIZE - i - 2), "%u", cpu);
 		if (i >= (PAGE_SIZE - 5))
-		    break;
+			break;
 	}
 	i += sprintf(&buf[i], "\n");
 	return i;
 }
 
+/**
+ * show_related_cpus - show the CPUs affected by each transition even if
+ * hw coordination is in use
+ */
+static ssize_t show_related_cpus(struct cpufreq_policy *policy, char *buf)
+{
+	if (cpumask_empty(policy->related_cpus))
+		return show_cpus(policy->cpus, buf);
+	return show_cpus(policy->related_cpus, buf);
+}
+
+/**
+ * show_affected_cpus - show the CPUs affected by each transition
+ */
+static ssize_t show_affected_cpus(struct cpufreq_policy *policy, char *buf)
+{
+	return show_cpus(policy->cpus, buf);
+}
+
+static ssize_t store_scaling_setspeed(struct cpufreq_policy *policy,
+					const char *buf, size_t count)
+{
+	unsigned int freq = 0;
+	unsigned int ret;
+
+	if (!policy->governor || !policy->governor->store_setspeed)
+		return -EINVAL;
+
+	ret = sscanf(buf, "%u", &freq);
+	if (ret != 1)
+		return -EINVAL;
+
+	policy->governor->store_setspeed(policy, freq);
+
+	return count;
+}
+
+static ssize_t show_scaling_setspeed(struct cpufreq_policy *policy, char *buf)
+{
+	if (!policy->governor || !policy->governor->show_setspeed)
+		return sprintf(buf, "<unsupported>\n");
+
+	return policy->governor->show_setspeed(policy, buf);
+}
 
 #define define_one_ro(_name) \
 static struct freq_attr _name = \
@@ -565,67 +660,87 @@ __ATTR(_name, 0644, show_##_name, store_##_name)
 define_one_ro0400(cpuinfo_cur_freq);
 define_one_ro(cpuinfo_min_freq);
 define_one_ro(cpuinfo_max_freq);
+define_one_ro(cpuinfo_transition_latency);
 define_one_ro(scaling_available_governors);
 define_one_ro(scaling_driver);
 define_one_ro(scaling_cur_freq);
+define_one_ro(related_cpus);
 define_one_ro(affected_cpus);
 define_one_rw(scaling_min_freq);
 define_one_rw(scaling_max_freq);
 define_one_rw(scaling_governor);
+define_one_rw(scaling_setspeed);
 
-static struct attribute * default_attrs[] = {
+static struct attribute *default_attrs[] = {
 	&cpuinfo_min_freq.attr,
 	&cpuinfo_max_freq.attr,
+	&cpuinfo_transition_latency.attr,
 	&scaling_min_freq.attr,
 	&scaling_max_freq.attr,
 	&affected_cpus.attr,
+	&related_cpus.attr,
 	&scaling_governor.attr,
 	&scaling_driver.attr,
 	&scaling_available_governors.attr,
+	&scaling_setspeed.attr,
 	NULL
 };
 
-#define to_policy(k) container_of(k,struct cpufreq_policy,kobj)
-#define to_attr(a) container_of(a,struct freq_attr,attr)
+#define to_policy(k) container_of(k, struct cpufreq_policy, kobj)
+#define to_attr(a) container_of(a, struct freq_attr, attr)
 
-static ssize_t show(struct kobject * kobj, struct attribute * attr ,char * buf)
+static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
 {
-	struct cpufreq_policy * policy = to_policy(kobj);
-	struct freq_attr * fattr = to_attr(attr);
-	ssize_t ret;
+	struct cpufreq_policy *policy = to_policy(kobj);
+	struct freq_attr *fattr = to_attr(attr);
+	ssize_t ret = -EINVAL;
 	policy = cpufreq_cpu_get(policy->cpu);
 	if (!policy)
-		return -EINVAL;
+		goto no_policy;
+
+	if (lock_policy_rwsem_read(policy->cpu) < 0)
+		goto fail;
+
 	if (fattr->show)
 		ret = fattr->show(policy, buf);
 	else
 		ret = -EIO;
 
+	unlock_policy_rwsem_read(policy->cpu);
+fail:
 	cpufreq_cpu_put(policy);
+no_policy:
 	return ret;
 }
 
-static ssize_t store(struct kobject * kobj, struct attribute * attr,
-		     const char * buf, size_t count)
+static ssize_t store(struct kobject *kobj, struct attribute *attr,
+		     const char *buf, size_t count)
 {
-	struct cpufreq_policy * policy = to_policy(kobj);
-	struct freq_attr * fattr = to_attr(attr);
-	ssize_t ret;
+	struct cpufreq_policy *policy = to_policy(kobj);
+	struct freq_attr *fattr = to_attr(attr);
+	ssize_t ret = -EINVAL;
 	policy = cpufreq_cpu_get(policy->cpu);
 	if (!policy)
-		return -EINVAL;
+		goto no_policy;
+
+	if (lock_policy_rwsem_write(policy->cpu) < 0)
+		goto fail;
+
 	if (fattr->store)
 		ret = fattr->store(policy, buf, count);
 	else
 		ret = -EIO;
 
+	unlock_policy_rwsem_write(policy->cpu);
+fail:
 	cpufreq_cpu_put(policy);
+no_policy:
 	return ret;
 }
 
-static void cpufreq_sysfs_release(struct kobject * kobj)
+static void cpufreq_sysfs_release(struct kobject *kobj)
 {
-	struct cpufreq_policy * policy = to_policy(kobj);
+	struct cpufreq_policy *policy = to_policy(kobj);
 	dprintk("last reference is dropped\n");
 	complete(&policy->kobj_unregister);
 }
@@ -646,8 +761,12 @@ static struct kobj_type ktype_cpufreq = {
  * 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 sys_device * sys_dev)
+static int cpufreq_add_dev(struct sys_device *sys_dev)
 {
 	unsigned int cpu = sys_dev->id;
 	int ret = 0;
@@ -657,9 +776,6 @@ static int cpufreq_add_dev (struct sys_device * sys_dev)
 	struct sys_device *cpu_sys_dev;
 	unsigned long flags;
 	unsigned int j;
-#ifdef CONFIG_SMP
-	struct cpufreq_policy *managed_policy;
-#endif
 
 	if (cpu_is_offline(cpu))
 		return 0;
@@ -688,104 +804,171 @@ static int cpufreq_add_dev (struct sys_device * sys_dev)
 		ret = -ENOMEM;
 		goto nomem_out;
 	}
+	if (!alloc_cpumask_var(&policy->cpus, GFP_KERNEL)) {
+		ret = -ENOMEM;
+		goto err_free_policy;
+	}
+	if (!zalloc_cpumask_var(&policy->related_cpus, GFP_KERNEL)) {
+		ret = -ENOMEM;
+		goto err_free_cpumask;
+	}
 
 	policy->cpu = cpu;
-	policy->cpus = cpumask_of_cpu(cpu);
+	cpumask_copy(policy->cpus, cpumask_of(cpu));
+
+	/* Initially set CPU itself as the policy_cpu */
+	per_cpu(policy_cpu, cpu) = cpu;
+	ret = (lock_policy_rwsem_write(cpu) < 0);
+	WARN_ON(ret);
 
-	mutex_init(&policy->lock);
-	mutex_lock(&policy->lock);
 	init_completion(&policy->kobj_unregister);
-	INIT_WORK(&policy->update, handle_update, (void *)(long)cpu);
+	INIT_WORK(&policy->update, handle_update);
 
+	/* Set governor before ->init, so that driver could check it */
+	policy->governor = CPUFREQ_DEFAULT_GOVERNOR;
 	/* call driver. From then on the cpufreq must be able
 	 * to accept all calls to ->verify and ->setpolicy for this CPU
 	 */
 	ret = cpufreq_driver->init(policy);
 	if (ret) {
 		dprintk("initialization failed\n");
-		mutex_unlock(&policy->lock);
-		goto err_out;
+		goto err_unlock_policy;
 	}
+	policy->user_policy.min = policy->min;
+	policy->user_policy.max = policy->max;
+
+	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
+				     CPUFREQ_START, policy);
 
 #ifdef CONFIG_SMP
-	for_each_cpu_mask(j, policy->cpus) {
+
+#ifdef CONFIG_HOTPLUG_CPU
+	if (per_cpu(cpufreq_cpu_governor, cpu)) {
+		policy->governor = per_cpu(cpufreq_cpu_governor, cpu);
+		dprintk("Restoring governor %s for cpu %d\n",
+		       policy->governor->name, cpu);
+	}
+#endif
+
+	for_each_cpu(j, policy->cpus) {
+		struct cpufreq_policy *managed_policy;
+
 		if (cpu == j)
 			continue;
 
-		/* check for existing affected CPUs.  They may not be aware
-		 * of it due to CPU Hotplug.
+		/* Check for existing affected CPUs.
+		 * They may not be aware of it due to CPU Hotplug.
+		 * cpufreq_cpu_put is called when the device is removed
+		 * in __cpufreq_remove_dev()
 		 */
 		managed_policy = cpufreq_cpu_get(j);
 		if (unlikely(managed_policy)) {
+
+			/* Set proper policy_cpu */
+			unlock_policy_rwsem_write(cpu);
+			per_cpu(policy_cpu, cpu) = managed_policy->cpu;
+
+			if (lock_policy_rwsem_write(cpu) < 0) {
+				/* Should not go through policy unlock path */
+				if (cpufreq_driver->exit)
+					cpufreq_driver->exit(policy);
+				ret = -EBUSY;
+				cpufreq_cpu_put(managed_policy);
+				goto err_free_cpumask;
+			}
+
 			spin_lock_irqsave(&cpufreq_driver_lock, flags);
-			managed_policy->cpus = policy->cpus;
-			cpufreq_cpu_data[cpu] = managed_policy;
+			cpumask_copy(managed_policy->cpus, policy->cpus);
+			per_cpu(cpufreq_cpu_data, cpu) = managed_policy;
 			spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
 
 			dprintk("CPU already managed, adding link\n");
-			sysfs_create_link(&sys_dev->kobj,
-					  &managed_policy->kobj, "cpufreq");
-
-			cpufreq_debug_enable_ratelimit();
-			mutex_unlock(&policy->lock);
-			ret = 0;
-			goto err_out_driver_exit; /* call driver->exit() */
+			ret = sysfs_create_link(&sys_dev->kobj,
+						&managed_policy->kobj,
+						"cpufreq");
+			if (ret)
+				cpufreq_cpu_put(managed_policy);
+			/*
+			 * Success. We only needed to be added to the mask.
+			 * Call driver->exit() because only the cpu parent of
+			 * the kobj needed to call init().
+			 */
+			goto out_driver_exit; /* call driver->exit() */
 		}
 	}
 #endif
 	memcpy(&new_policy, policy, sizeof(struct cpufreq_policy));
 
 	/* prepare interface data */
-	policy->kobj.parent = &sys_dev->kobj;
-	policy->kobj.ktype = &ktype_cpufreq;
-	strlcpy(policy->kobj.name, "cpufreq", KOBJ_NAME_LEN);
+	ret = kobject_init_and_add(&policy->kobj, &ktype_cpufreq, &sys_dev->kobj,
+				   "cpufreq");
+	if (ret)
+		goto out_driver_exit;
 
-	ret = kobject_register(&policy->kobj);
-	if (ret) {
-		mutex_unlock(&policy->lock);
-		goto err_out_driver_exit;
-	}
 	/* set up files for this cpu device */
 	drv_attr = cpufreq_driver->attr;
 	while ((drv_attr) && (*drv_attr)) {
-		sysfs_create_file(&policy->kobj, &((*drv_attr)->attr));
+		ret = sysfs_create_file(&policy->kobj, &((*drv_attr)->attr));
+		if (ret)
+			goto err_out_kobj_put;
 		drv_attr++;
 	}
-	if (cpufreq_driver->get)
-		sysfs_create_file(&policy->kobj, &cpuinfo_cur_freq.attr);
-	if (cpufreq_driver->target)
-		sysfs_create_file(&policy->kobj, &scaling_cur_freq.attr);
+	if (cpufreq_driver->get) {
+		ret = sysfs_create_file(&policy->kobj, &cpuinfo_cur_freq.attr);
+		if (ret)
+			goto err_out_kobj_put;
+	}
+	if (cpufreq_driver->target) {
+		ret = sysfs_create_file(&policy->kobj, &scaling_cur_freq.attr);
+		if (ret)
+			goto err_out_kobj_put;
+	}
 
 	spin_lock_irqsave(&cpufreq_driver_lock, flags);
-	for_each_cpu_mask(j, policy->cpus)
-		cpufreq_cpu_data[j] = policy;
+	for_each_cpu(j, policy->cpus) {
+		if (!cpu_online(j))
+			continue;
+		per_cpu(cpufreq_cpu_data, j) = policy;
+		per_cpu(policy_cpu, j) = policy->cpu;
+	}
 	spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
 
 	/* symlink affected CPUs */
-	for_each_cpu_mask(j, policy->cpus) {
+	for_each_cpu(j, policy->cpus) {
+		struct cpufreq_policy *managed_policy;
+
 		if (j == cpu)
 			continue;
 		if (!cpu_online(j))
 			continue;
 
 		dprintk("CPU %u already managed, adding link\n", j);
-		cpufreq_cpu_get(cpu);
+		managed_policy = cpufreq_cpu_get(cpu);
 		cpu_sys_dev = get_cpu_sysdev(j);
-		sysfs_create_link(&cpu_sys_dev->kobj, &policy->kobj,
-				  "cpufreq");
+		ret = sysfs_create_link(&cpu_sys_dev->kobj, &policy->kobj,
+					"cpufreq");
+		if (ret) {
+			cpufreq_cpu_put(managed_policy);
+			goto err_out_unregister;
+		}
 	}
 
 	policy->governor = NULL; /* to assure that the starting sequence is
 				  * run in cpufreq_set_policy */
-	mutex_unlock(&policy->lock);
 
 	/* set default policy */
-	ret = cpufreq_set_policy(&new_policy);
+	ret = __cpufreq_set_policy(policy, &new_policy);
+	policy->user_policy.policy = policy->policy;
+	policy->user_policy.governor = policy->governor;
+
 	if (ret) {
 		dprintk("setting policy failed\n");
 		goto err_out_unregister;
 	}
 
+	unlock_policy_rwsem_write(cpu);
+
+	kobject_uevent(&policy->kobj, KOBJ_ADD);
 	module_put(cpufreq_driver->owner);
 	dprintk("initialization complete\n");
 	cpufreq_debug_enable_ratelimit();
@@ -795,20 +978,24 @@ static int cpufreq_add_dev (struct sys_device * sys_dev)
 
 err_out_unregister:
 	spin_lock_irqsave(&cpufreq_driver_lock, flags);
-	for_each_cpu_mask(j, policy->cpus)
-		cpufreq_cpu_data[j] = NULL;
+	for_each_cpu(j, policy->cpus)
+		per_cpu(cpufreq_cpu_data, j) = NULL;
 	spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
 
-	kobject_unregister(&policy->kobj);
+err_out_kobj_put:
+	kobject_put(&policy->kobj);
 	wait_for_completion(&policy->kobj_unregister);
 
-err_out_driver_exit:
+out_driver_exit:
 	if (cpufreq_driver->exit)
 		cpufreq_driver->exit(policy);
 
-err_out:
+err_unlock_policy:
+	unlock_policy_rwsem_write(cpu);
+err_free_cpumask:
+	free_cpumask_var(policy->cpus);
+err_free_policy:
 	kfree(policy);
-
 nomem_out:
 	module_put(cpufreq_driver->owner);
 module_out:
@@ -818,11 +1005,13 @@ module_out:
 
 
 /**
- * cpufreq_remove_dev - remove a CPU device
+ * __cpufreq_remove_dev - remove a CPU device
  *
  * Removes the cpufreq interface for a CPU device.
+ * Caller should already have policy_rwsem in write mode for this CPU.
+ * This routine frees the rwsem before returning.
  */
-static int cpufreq_remove_dev (struct sys_device * sys_dev)
+static int __cpufreq_remove_dev(struct sys_device *sys_dev)
 {
 	unsigned int cpu = sys_dev->id;
 	unsigned long flags;
@@ -836,14 +1025,15 @@ static int cpufreq_remove_dev (struct sys_device * sys_dev)
 	dprintk("unregistering CPU %u\n", cpu);
 
 	spin_lock_irqsave(&cpufreq_driver_lock, flags);
-	data = cpufreq_cpu_data[cpu];
+	data = per_cpu(cpufreq_cpu_data, cpu);
 
 	if (!data) {
 		spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
 		cpufreq_debug_enable_ratelimit();
+		unlock_policy_rwsem_write(cpu);
 		return -EINVAL;
 	}
-	cpufreq_cpu_data[cpu] = NULL;
+	per_cpu(cpufreq_cpu_data, cpu) = NULL;
 
 
 #ifdef CONFIG_SMP
@@ -852,43 +1042,45 @@ static int cpufreq_remove_dev (struct sys_device * sys_dev)
 	 */
 	if (unlikely(cpu != data->cpu)) {
 		dprintk("removing link\n");
-		cpu_clear(cpu, data->cpus);
+		cpumask_clear_cpu(cpu, data->cpus);
 		spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
 		sysfs_remove_link(&sys_dev->kobj, "cpufreq");
 		cpufreq_cpu_put(data);
 		cpufreq_debug_enable_ratelimit();
+		unlock_policy_rwsem_write(cpu);
 		return 0;
 	}
 #endif
 
+#ifdef CONFIG_SMP
 
-	if (!kobject_get(&data->kobj)) {
-		spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
-		cpufreq_debug_enable_ratelimit();
-		return -EFAULT;
-	}
+#ifdef CONFIG_HOTPLUG_CPU
+	per_cpu(cpufreq_cpu_governor, cpu) = data->governor;
+#endif
 
-#ifdef CONFIG_SMP
 	/* if we have other CPUs still registered, we need to unlink them,
 	 * or else wait_for_completion below will lock up. Clean the
-	 * cpufreq_cpu_data[] while holding the lock, and remove the sysfs
-	 * links afterwards.
+	 * per_cpu(cpufreq_cpu_data) while holding the lock, and remove
+	 * the sysfs links afterwards.
 	 */
-	if (unlikely(cpus_weight(data->cpus) > 1)) {
-		for_each_cpu_mask(j, data->cpus) {
+	if (unlikely(cpumask_weight(data->cpus) > 1)) {
+		for_each_cpu(j, data->cpus) {
 			if (j == cpu)
 				continue;
-			cpufreq_cpu_data[j] = NULL;
+			per_cpu(cpufreq_cpu_data, j) = NULL;
 		}
 	}
 
 	spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
 
-	if (unlikely(cpus_weight(data->cpus) > 1)) {
-		for_each_cpu_mask(j, data->cpus) {
+	if (unlikely(cpumask_weight(data->cpus) > 1)) {
+		for_each_cpu(j, data->cpus) {
 			if (j == cpu)
 				continue;
 			dprintk("removing link for cpu %u\n", j);
+#ifdef CONFIG_HOTPLUG_CPU
+			per_cpu(cpufreq_cpu_governor, j) = data->governor;
+#endif
 			cpu_sys_dev = get_cpu_sysdev(j);
 			sysfs_remove_link(&cpu_sys_dev->kobj, "cpufreq");
 			cpufreq_cpu_put(data);
@@ -898,12 +1090,8 @@ static int cpufreq_remove_dev (struct sys_device * sys_dev)
 	spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
 #endif
 
-	mutex_lock(&data->lock);
 	if (cpufreq_driver->target)
 		__cpufreq_governor(data, CPUFREQ_GOV_STOP);
-	mutex_unlock(&data->lock);
-
-	kobject_unregister(&data->kobj);
 
 	kobject_put(&data->kobj);
 
@@ -918,16 +1106,39 @@ static int cpufreq_remove_dev (struct sys_device * sys_dev)
 	if (cpufreq_driver->exit)
 		cpufreq_driver->exit(data);
 
+	unlock_policy_rwsem_write(cpu);
+
+	free_cpumask_var(data->related_cpus);
+	free_cpumask_var(data->cpus);
 	kfree(data);
+	per_cpu(cpufreq_cpu_data, cpu) = NULL;
 
 	cpufreq_debug_enable_ratelimit();
 	return 0;
 }
 
 
-static void handle_update(void *data)
+static int cpufreq_remove_dev(struct sys_device *sys_dev)
+{
+	unsigned int cpu = sys_dev->id;
+	int retval;
+
+	if (cpu_is_offline(cpu))
+		return 0;
+
+	if (unlikely(lock_policy_rwsem_write(cpu)))
+		BUG();
+
+	retval = __cpufreq_remove_dev(sys_dev);
+	return retval;
+}
+
+
+static void handle_update(struct work_struct *work)
 {
-	unsigned int cpu = (unsigned int)(long)data;
+	struct cpufreq_policy *policy =
+		container_of(work, struct cpufreq_policy, update);
+	unsigned int cpu = policy->cpu;
 	dprintk("handle_update for cpu %u called\n", cpu);
 	cpufreq_update_policy(cpu);
 }
@@ -938,8 +1149,8 @@ static void handle_update(void *data)
  *	@old_freq: CPU frequency the kernel thinks the CPU runs at
  *	@new_freq: CPU frequency the CPU actually runs at
  *
- *	We adjust to current frequency first, and need to clean up later. So either call
- *	to cpufreq_update_policy() or schedule handle_update()).
+ *	We adjust to current frequency first, and need to clean up later.
+ *	So either call to cpufreq_update_policy() or schedule handle_update()).
  */
 static void cpufreq_out_of_sync(unsigned int cpu, unsigned int old_freq,
 				unsigned int new_freq)
@@ -958,7 +1169,7 @@ static void cpufreq_out_of_sync(unsigned int cpu, unsigned int old_freq,
 
 
 /**
- * cpufreq_quick_get - get the CPU frequency (in kHz) frpm policy->cur
+ * cpufreq_quick_get - get the CPU frequency (in kHz) from policy->cur
  * @cpu: CPU number
  *
  * This is the last known freq, without actually getting it from the driver.
@@ -970,35 +1181,22 @@ unsigned int cpufreq_quick_get(unsigned int cpu)
 	unsigned int ret_freq = 0;
 
 	if (policy) {
-		mutex_lock(&policy->lock);
 		ret_freq = policy->cur;
-		mutex_unlock(&policy->lock);
 		cpufreq_cpu_put(policy);
 	}
 
-	return (ret_freq);
+	return ret_freq;
 }
 EXPORT_SYMBOL(cpufreq_quick_get);
 
 
-/**
- * cpufreq_get - get the current CPU frequency (in kHz)
- * @cpu: CPU number
- *
- * Get the CPU current (static) CPU frequency
- */
-unsigned int cpufreq_get(unsigned int cpu)
+static unsigned int __cpufreq_get(unsigned int cpu)
 {
-	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
+	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);
 	unsigned int ret_freq = 0;
 
-	if (!policy)
-		return 0;
-
 	if (!cpufreq_driver->get)
-		goto out;
-
-	mutex_lock(&policy->lock);
+		return ret_freq;
 
 	ret_freq = cpufreq_driver->get(cpu);
 
@@ -1012,12 +1210,34 @@ unsigned int cpufreq_get(unsigned int cpu)
 		}
 	}
 
-	mutex_unlock(&policy->lock);
+	return ret_freq;
+}
 
-out:
-	cpufreq_cpu_put(policy);
+/**
+ * cpufreq_get - get the current CPU frequency (in kHz)
+ * @cpu: CPU number
+ *
+ * Get the CPU current (static) CPU frequency
+ */
+unsigned int cpufreq_get(unsigned int cpu)
+{
+	unsigned int ret_freq = 0;
+	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
+
+	if (!policy)
+		goto out;
+
+	if (unlikely(lock_policy_rwsem_read(cpu)))
+		goto out_policy;
+
+	ret_freq = __cpufreq_get(cpu);
 
-	return (ret_freq);
+	unlock_policy_rwsem_read(cpu);
+
+out_policy:
+	cpufreq_cpu_put(policy);
+out:
+	return ret_freq;
 }
 EXPORT_SYMBOL(cpufreq_get);
 
@@ -1026,15 +1246,24 @@ EXPORT_SYMBOL(cpufreq_get);
  *	cpufreq_suspend - let the low level driver prepare for suspend
  */
 
-static int cpufreq_suspend(struct sys_device * sysdev, pm_message_t pmsg)
+static int cpufreq_suspend(struct sys_device *sysdev, pm_message_t pmsg)
 {
-	int cpu = sysdev->id;
 	int ret = 0;
+
+#ifdef __powerpc__
+	int cpu = sysdev->id;
 	unsigned int cur_freq = 0;
 	struct cpufreq_policy *cpu_policy;
 
 	dprintk("suspending cpu %u\n", cpu);
 
+	/*
+	 * This whole bogosity is here because Powerbooks are made of fail.
+	 * No sane platform should need any of the code below to be run.
+	 * (it's entirely the wrong thing to do, as driver->get may
+	 *  reenable interrupts on some architectures).
+	 */
+
 	if (!cpu_online(cpu))
 		return 0;
 
@@ -1048,22 +1277,18 @@ static int cpufreq_suspend(struct sys_device * sysdev, pm_message_t pmsg)
 		return -EINVAL;
 
 	/* only handle each CPU group once */
-	if (unlikely(cpu_policy->cpu != cpu)) {
-		cpufreq_cpu_put(cpu_policy);
-		return 0;
-	}
+	if (unlikely(cpu_policy->cpu != cpu))
+		goto out;
 
 	if (cpufreq_driver->suspend) {
 		ret = cpufreq_driver->suspend(cpu_policy, pmsg);
 		if (ret) {
 			printk(KERN_ERR "cpufreq: suspend failed in ->suspend "
 					"step on CPU %u\n", cpu_policy->cpu);
-			cpufreq_cpu_put(cpu_policy);
-			return ret;
+			goto out;
 		}
 	}
 
-
 	if (cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)
 		goto out;
 
@@ -1097,7 +1322,8 @@ static int cpufreq_suspend(struct sys_device * sysdev, pm_message_t pmsg)
 
 out:
 	cpufreq_cpu_put(cpu_policy);
-	return 0;
+#endif	/* __powerpc__ */
+	return ret;
 }
 
 /**
@@ -1108,14 +1334,20 @@ out:
  *	3.) schedule call cpufreq_update_policy() ASAP as interrupts are
  *	    restored.
  */
-static int cpufreq_resume(struct sys_device * sysdev)
+static int cpufreq_resume(struct sys_device *sysdev)
 {
-	int cpu = sysdev->id;
 	int ret = 0;
+
+#ifdef __powerpc__
+	int cpu = sysdev->id;
 	struct cpufreq_policy *cpu_policy;
 
 	dprintk("resuming cpu %u\n", cpu);
 
+	/* As with the ->suspend method, all the code below is
+	 * only necessary because Powerbooks suck.
+	 * See commit 42d4dc3f4e1e for jokes. */
+
 	if (!cpu_online(cpu))
 		return 0;
 
@@ -1129,18 +1361,15 @@ static int cpufreq_resume(struct sys_device * sysdev)
 		return -EINVAL;
 
 	/* only handle each CPU group once */
-	if (unlikely(cpu_policy->cpu != cpu)) {
-		cpufreq_cpu_put(cpu_policy);
-		return 0;
-	}
+	if (unlikely(cpu_policy->cpu != cpu))
+		goto fail;
 
 	if (cpufreq_driver->resume) {
 		ret = cpufreq_driver->resume(cpu_policy);
 		if (ret) {
 			printk(KERN_ERR "cpufreq: resume failed in ->resume "
 					"step on CPU %u\n", cpu_policy->cpu);
-			cpufreq_cpu_put(cpu_policy);
-			return ret;
+			goto fail;
 		}
 	}
 
@@ -1161,7 +1390,7 @@ static int cpufreq_resume(struct sys_device * sysdev)
 			struct cpufreq_freqs freqs;
 
 			if (!(cpufreq_driver->flags & CPUFREQ_PM_NO_WARN))
-				dprintk("Warning: CPU frequency"
+				dprintk("Warning: CPU frequency "
 				       "is %u, cpufreq assumed %u kHz.\n",
 				       cur_freq, cpu_policy->cur);
 
@@ -1180,7 +1409,9 @@ static int cpufreq_resume(struct sys_device * sysdev)
 
 out:
 	schedule_work(&cpu_policy->update);
+fail:
 	cpufreq_cpu_put(cpu_policy);
+#endif	/* __powerpc__ */
 	return ret;
 }
 
@@ -1213,6 +1444,8 @@ int cpufreq_register_notifier(struct notifier_block *nb, unsigned int list)
 {
 	int ret;
 
+	WARN_ON(!init_cpufreq_transition_notifier_list_called);
+
 	switch (list) {
 	case CPUFREQ_TRANSITION_NOTIFIER:
 		ret = srcu_notifier_chain_register(
@@ -1268,7 +1501,6 @@ EXPORT_SYMBOL(cpufreq_unregister_notifier);
  *********************************************************************/
 
 
-/* Must be called with lock_cpu_hotplug held */
 int __cpufreq_driver_target(struct cpufreq_policy *policy,
 			    unsigned int target_freq,
 			    unsigned int relation)
@@ -1288,26 +1520,27 @@ int cpufreq_driver_target(struct cpufreq_policy *policy,
 			  unsigned int target_freq,
 			  unsigned int relation)
 {
-	int ret;
+	int ret = -EINVAL;
 
 	policy = cpufreq_cpu_get(policy->cpu);
 	if (!policy)
-		return -EINVAL;
+		goto no_policy;
 
-	lock_cpu_hotplug();
-	mutex_lock(&policy->lock);
+	if (unlikely(lock_policy_rwsem_write(policy->cpu)))
+		goto fail;
 
 	ret = __cpufreq_driver_target(policy, target_freq, relation);
 
-	mutex_unlock(&policy->lock);
-	unlock_cpu_hotplug();
+	unlock_policy_rwsem_write(policy->cpu);
 
+fail:
 	cpufreq_cpu_put(policy);
+no_policy:
 	return ret;
 }
 EXPORT_SYMBOL_GPL(cpufreq_driver_target);
 
-int cpufreq_driver_getavg(struct cpufreq_policy *policy)
+int __cpufreq_driver_getavg(struct cpufreq_policy *policy, unsigned int cpu)
 {
 	int ret = 0;
 
@@ -1315,20 +1548,15 @@ int cpufreq_driver_getavg(struct cpufreq_policy *policy)
 	if (!policy)
 		return -EINVAL;
 
-	mutex_lock(&policy->lock);
-
-	if (cpu_online(policy->cpu) && cpufreq_driver->getavg)
-		ret = cpufreq_driver->getavg(policy->cpu);
-
-	mutex_unlock(&policy->lock);
+	if (cpu_online(cpu) && cpufreq_driver->getavg)
+		ret = cpufreq_driver->getavg(policy, cpu);
 
 	cpufreq_cpu_put(policy);
 	return ret;
 }
-EXPORT_SYMBOL_GPL(cpufreq_driver_getavg);
+EXPORT_SYMBOL_GPL(__cpufreq_driver_getavg);
 
 /*
- * Locking: Must be called with the lock_cpu_hotplug() lock held
  * when "event" is CPUFREQ_GOV_LIMITS
  */
 
@@ -1337,6 +1565,31 @@ static int __cpufreq_governor(struct cpufreq_policy *policy,
 {
 	int ret;
 
+	/* Only must be defined when default governor is known to have latency
+	   restrictions, like e.g. conservative or ondemand.
+	   That this is the case is already ensured in Kconfig
+	*/
+#ifdef CONFIG_CPU_FREQ_GOV_PERFORMANCE
+	struct cpufreq_governor *gov = &cpufreq_gov_performance;
+#else
+	struct cpufreq_governor *gov = NULL;
+#endif
+
+	if (policy->governor->max_transition_latency &&
+	    policy->cpuinfo.transition_latency >
+	    policy->governor->max_transition_latency) {
+		if (!gov)
+			return -EINVAL;
+		else {
+			printk(KERN_WARNING "%s governor failed, too long"
+			       " transition latency of HW, fallback"
+			       " to %s governor\n",
+			       policy->governor->name,
+			       gov->name);
+			policy->governor = gov;
+		}
+	}
+
 	if (!try_module_get(policy->governor->owner))
 		return -EINVAL;
 
@@ -1396,7 +1649,8 @@ EXPORT_SYMBOL_GPL(cpufreq_unregister_governor);
 
 /**
  * cpufreq_get_policy - get the current cpufreq_policy
- * @policy: struct cpufreq_policy into which the current cpufreq_policy is written
+ * @policy: struct cpufreq_policy into which the current cpufreq_policy
+ *	is written
  *
  * Reads the current cpufreq policy.
  */
@@ -1410,9 +1664,7 @@ int cpufreq_get_policy(struct cpufreq_policy *policy, unsigned int cpu)
 	if (!cpu_policy)
 		return -EINVAL;
 
-	mutex_lock(&cpu_policy->lock);
 	memcpy(policy, cpu_policy, sizeof(struct cpufreq_policy));
-	mutex_unlock(&cpu_policy->lock);
 
 	cpufreq_cpu_put(cpu_policy);
 	return 0;
@@ -1423,7 +1675,6 @@ EXPORT_SYMBOL(cpufreq_get_policy);
 /*
  * data   : current policy.
  * policy : policy to be set.
- * Locking: Must be called with the lock_cpu_hotplug() lock held
  */
 static int __cpufreq_set_policy(struct cpufreq_policy *data,
 				struct cpufreq_policy *policy)
@@ -1437,7 +1688,7 @@ static int __cpufreq_set_policy(struct cpufreq_policy *data,
 	memcpy(&policy->cpuinfo, &data->cpuinfo,
 				sizeof(struct cpufreq_cpuinfo));
 
-	if (policy->min > data->min && policy->min > policy->max) {
+	if (policy->min > data->max || policy->max < data->min) {
 		ret = -EINVAL;
 		goto error_out;
 	}
@@ -1512,45 +1763,6 @@ error_out:
 }
 
 /**
- *	cpufreq_set_policy - set a new CPUFreq policy
- *	@policy: policy to be set.
- *
- *	Sets a new CPU frequency and voltage scaling policy.
- */
-int cpufreq_set_policy(struct cpufreq_policy *policy)
-{
-	int ret = 0;
-	struct cpufreq_policy *data;
-
-	if (!policy)
-		return -EINVAL;
-
-	data = cpufreq_cpu_get(policy->cpu);
-	if (!data)
-		return -EINVAL;
-
-	lock_cpu_hotplug();
-
-	/* lock this CPU */
-	mutex_lock(&data->lock);
-
-	ret = __cpufreq_set_policy(data, policy);
-	data->user_policy.min = data->min;
-	data->user_policy.max = data->max;
-	data->user_policy.policy = data->policy;
-	data->user_policy.governor = data->governor;
-
-	mutex_unlock(&data->lock);
-
-	unlock_cpu_hotplug();
-	cpufreq_cpu_put(data);
-
-	return ret;
-}
-EXPORT_SYMBOL(cpufreq_set_policy);
-
-
-/**
  *	cpufreq_update_policy - re-evaluate an existing cpufreq policy
  *	@cpu: CPU which shall be re-evaluated
  *
@@ -1561,13 +1773,17 @@ int cpufreq_update_policy(unsigned int cpu)
 {
 	struct cpufreq_policy *data = cpufreq_cpu_get(cpu);
 	struct cpufreq_policy policy;
-	int ret = 0;
+	int ret;
 
-	if (!data)
-		return -ENODEV;
+	if (!data) {
+		ret = -ENODEV;
+		goto no_policy;
+	}
 
-	lock_cpu_hotplug();
-	mutex_lock(&data->lock);
+	if (unlikely(lock_policy_rwsem_write(cpu))) {
+		ret = -EINVAL;
+		goto fail;
+	}
 
 	dprintk("updating policy for CPU %u\n", cpu);
 	memcpy(&policy, data, sizeof(struct cpufreq_policy));
@@ -1592,54 +1808,48 @@ int cpufreq_update_policy(unsigned int cpu)
 
 	ret = __cpufreq_set_policy(data, &policy);
 
-	mutex_unlock(&data->lock);
-	unlock_cpu_hotplug();
+	unlock_policy_rwsem_write(cpu);
+
+fail:
 	cpufreq_cpu_put(data);
+no_policy:
 	return ret;
 }
 EXPORT_SYMBOL(cpufreq_update_policy);
 
-#ifdef CONFIG_HOTPLUG_CPU
-static int cpufreq_cpu_callback(struct notifier_block *nfb,
+static int __cpuinit cpufreq_cpu_callback(struct notifier_block *nfb,
 					unsigned long action, void *hcpu)
 {
 	unsigned int cpu = (unsigned long)hcpu;
-	struct cpufreq_policy *policy;
 	struct sys_device *sys_dev;
 
 	sys_dev = get_cpu_sysdev(cpu);
-
 	if (sys_dev) {
 		switch (action) {
 		case CPU_ONLINE:
+		case CPU_ONLINE_FROZEN:
 			cpufreq_add_dev(sys_dev);
 			break;
 		case CPU_DOWN_PREPARE:
-			/*
-			 * We attempt to put this cpu in lowest frequency
-			 * possible before going down. This will permit
-			 * hardware-managed P-State to switch other related
-			 * threads to min or higher speeds if possible.
-			 */
-			policy = cpufreq_cpu_data[cpu];
-			if (policy) {
-				cpufreq_driver_target(policy, policy->min,
-						CPUFREQ_RELATION_H);
-			}
+		case CPU_DOWN_PREPARE_FROZEN:
+			if (unlikely(lock_policy_rwsem_write(cpu)))
+				BUG();
+
+			__cpufreq_remove_dev(sys_dev);
 			break;
-		case CPU_DEAD:
-			cpufreq_remove_dev(sys_dev);
+		case CPU_DOWN_FAILED:
+		case CPU_DOWN_FAILED_FROZEN:
+			cpufreq_add_dev(sys_dev);
 			break;
 		}
 	}
 	return NOTIFY_OK;
 }
 
-static struct notifier_block __cpuinitdata cpufreq_cpu_notifier =
+static struct notifier_block __refdata cpufreq_cpu_notifier =
 {
     .notifier_call = cpufreq_cpu_callback,
 };
-#endif /* CONFIG_HOTPLUG_CPU */
 
 /*********************************************************************
  *               REGISTER / UNREGISTER CPUFREQ DRIVER                *
@@ -1677,16 +1887,19 @@ int cpufreq_register_driver(struct cpufreq_driver *driver_data)
 	cpufreq_driver = driver_data;
 	spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
 
-	ret = sysdev_driver_register(&cpu_sysdev_class,&cpufreq_sysdev_driver);
+	ret = sysdev_driver_register(&cpu_sysdev_class,
+					&cpufreq_sysdev_driver);
 
 	if ((!ret) && !(cpufreq_driver->flags & CPUFREQ_STICKY)) {
 		int i;
 		ret = -ENODEV;
 
 		/* check for at least one working CPU */
-		for (i=0; i<NR_CPUS; i++)
-			if (cpufreq_cpu_data[i])
+		for (i = 0; i < nr_cpu_ids; i++)
+			if (cpu_possible(i) && per_cpu(cpufreq_cpu_data, i)) {
 				ret = 0;
+				break;
+			}
 
 		/* if all ->init() calls failed, unregister */
 		if (ret) {
@@ -1707,7 +1920,7 @@ int cpufreq_register_driver(struct cpufreq_driver *driver_data)
 		cpufreq_debug_enable_ratelimit();
 	}
 
-	return (ret);
+	return ret;
 }
 EXPORT_SYMBOL_GPL(cpufreq_register_driver);
 
@@ -1743,3 +1956,16 @@ int cpufreq_unregister_driver(struct cpufreq_driver *driver)
 	return 0;
 }
 EXPORT_SYMBOL_GPL(cpufreq_unregister_driver);
+
+static int __init cpufreq_core_init(void)
+{
+	int cpu;
+
+	for_each_possible_cpu(cpu) {
+		per_cpu(policy_cpu, cpu) = -1;
+		init_rwsem(&per_cpu(cpu_policy_rwsem, cpu));
+	}
+	return 0;
+}
+
+core_initcall(cpufreq_core_init);