cpufreq.c

来自「linux 内核源代码」· C语言 代码 · 共 1,883 行 · 第 1/4 页

/*
 *  linux/drivers/cpufreq/cpufreq.c
 *
 *  Copyright (C) 2001 Russell King
 *            (C) 2002 - 2003 Dominik Brodowski <[e-mail]>
 *
 *  Oct 2005 - Ashok Raj <[e-mail]>
 *	Added handling for CPU hotplug
 *  Feb 2006 - Jacob Shin <[e-mail]>
 *	Fix handling for CPU hotplug -- affected CPUs
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 *
 */

#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/notifier.h>
#include <linux/cpufreq.h>
#include <linux/delay.h>
#include <linux/interrupt.h>
#include <linux/spinlock.h>
#include <linux/device.h>
#include <linux/slab.h>
#include <linux/cpu.h>
#include <linux/completion.h>
#include <linux/mutex.h>

#define dprintk(msg...) cpufreq_debug_printk(CPUFREQ_DEBUG_CORE, \
						"cpufreq-core", msg)

/**
 * The "cpufreq driver" - the arch- or hardware-dependent low
 * level driver of CPUFreq support, and its spinlock. This lock
 * also protects the cpufreq_cpu_data array.
 */
static struct cpufreq_driver *cpufreq_driver;
static struct cpufreq_policy *cpufreq_cpu_data[NR_CPUS];
#ifdef CONFIG_HOTPLUG_CPU
/* This one keeps track of the previously set governor of a removed CPU */
static struct cpufreq_governor *cpufreq_cpu_governor[NR_CPUS];
#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 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
 * validation process for a new CPU frequency policy; the
 * "transition" list for kernel code that needs to handle
 * changes to devices when the CPU clock speed changes.
 * The mutex locks both lists.
 */
static BLOCKING_NOTIFIER_HEAD(cpufreq_policy_notifier_list);
static struct srcu_notifier_head cpufreq_transition_notifier_list;

static int __init init_cpufreq_transition_notifier_list(void)
{
	srcu_init_notifier_head(&cpufreq_transition_notifier_list);
	return 0;
}
pure_initcall(init_cpufreq_transition_notifier_list);

static LIST_HEAD(cpufreq_governor_list);
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)
		goto err_out;

	/* get the cpufreq driver */
	spin_lock_irqsave(&cpufreq_driver_lock, flags);

	if (!cpufreq_driver)
		goto err_out_unlock;

	if (!try_module_get(cpufreq_driver->owner))
		goto err_out_unlock;


	/* get the CPU */
	data = cpufreq_cpu_data[cpu];

	if (!data)
		goto err_out_put_module;

	if (!kobject_get(&data->kobj))
		goto err_out_put_module;

	spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
	return data;

err_out_put_module:
	module_put(cpufreq_driver->owner);
err_out_unlock:
	spin_unlock_irqrestore(&cpufreq_driver_lock, flags);
err_out:
	return NULL;
}
EXPORT_SYMBOL_GPL(cpufreq_cpu_get);


void cpufreq_cpu_put(struct cpufreq_policy *data)
{
	kobject_put(&data->kobj);
	module_put(cpufreq_driver->owner);
}
EXPORT_SYMBOL_GPL(cpufreq_cpu_put);


/*********************************************************************
 *                     UNIFIED DEBUG HELPERS                         *
 *********************************************************************/
#ifdef CONFIG_CPU_FREQ_DEBUG

/* what part(s) of the CPUfreq subsystem are debugged? */
static unsigned int debug;

/* is the debug output ratelimit'ed using printk_ratelimit? User can
 * set or modify this value.
 */
static unsigned int debug_ratelimit = 1;

/* is the printk_ratelimit'ing enabled? It's enabled after a successful
 * loading of a cpufreq driver, temporarily disabled when a new policy
 * is set, and disabled upon cpufreq driver removal
 */
static unsigned int disable_ratelimit = 1;
static DEFINE_SPINLOCK(disable_ratelimit_lock);

static void cpufreq_debug_enable_ratelimit(void)
{
	unsigned long flags;

	spin_lock_irqsave(&disable_ratelimit_lock, flags);
	if (disable_ratelimit)
		disable_ratelimit--;
	spin_unlock_irqrestore(&disable_ratelimit_lock, flags);
}

static void cpufreq_debug_disable_ratelimit(void)
{
	unsigned long flags;

	spin_lock_irqsave(&disable_ratelimit_lock, flags);
	disable_ratelimit++;
	spin_unlock_irqrestore(&disable_ratelimit_lock, flags);
}

void cpufreq_debug_printk(unsigned int type, const char *prefix,
							const char *fmt, ...)
{
	char s[256];
	va_list args;
	unsigned int len;
	unsigned long flags;

	WARN_ON(!prefix);
	if (type & debug) {
		spin_lock_irqsave(&disable_ratelimit_lock, flags);
		if (!disable_ratelimit && debug_ratelimit
					&& !printk_ratelimit()) {
			spin_unlock_irqrestore(&disable_ratelimit_lock, flags);
			return;
		}
		spin_unlock_irqrestore(&disable_ratelimit_lock, flags);

		len = snprintf(s, 256, KERN_DEBUG "%s: ", prefix);

		va_start(args, fmt);
		len += vsnprintf(&s[len], (256 - len), fmt, args);
		va_end(args);

		printk(s);

		WARN_ON(len < 5);
	}
}
EXPORT_SYMBOL(cpufreq_debug_printk);


module_param(debug, uint, 0644);
MODULE_PARM_DESC(debug, "CPUfreq debugging: add 1 to debug core,"
			" 2 to debug drivers, and 4 to debug governors.");

module_param(debug_ratelimit, uint, 0644);
MODULE_PARM_DESC(debug_ratelimit, "CPUfreq debugging:"
					" set to 0 to disable ratelimiting.");

#else /* !CONFIG_CPU_FREQ_DEBUG */

static inline void cpufreq_debug_enable_ratelimit(void) { return; }
static inline void cpufreq_debug_disable_ratelimit(void) { return; }

#endif /* CONFIG_CPU_FREQ_DEBUG */


/*********************************************************************
 *            EXTERNALLY AFFECTING FREQUENCY CHANGES                 *
 *********************************************************************/

/**
 * adjust_jiffies - adjust the system "loops_per_jiffy"
 *
 * This function alters the system "loops_per_jiffy" for the clock
 * speed change. Note that loops_per_jiffy cannot be updated on SMP
 * systems as each CPU might be scaled differently. So, use the arch
 * per-CPU loops_per_jiffy value wherever possible.
 */
#ifndef CONFIG_SMP
static unsigned long l_p_j_ref;
static unsigned int  l_p_j_ref_freq;

static void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci)
{
	if (ci->flags & CPUFREQ_CONST_LOOPS)
		return;

	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;"
			"freq is %u kHz\n", l_p_j_ref, l_p_j_ref_freq);
	}
	if ((val == CPUFREQ_PRECHANGE  && ci->old < ci->new) ||
	    (val == CPUFREQ_POSTCHANGE && ci->old > ci->new) ||
	    (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"
			"for frequency %u kHz\n", loops_per_jiffy, ci->new);
	}
}
#else
static inline void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci)
{
	return;
}
#endif


/**
 * cpufreq_notify_transition - call notifier chain and adjust_jiffies
 * on frequency transition.
 *
 * This function calls the transition notifiers and the "adjust_jiffies"
 * function. It is called twice on all CPU frequency changes that have
 * external effects.
 */
void cpufreq_notify_transition(struct cpufreq_freqs *freqs, unsigned int state)
{
	struct cpufreq_policy *policy;

	BUG_ON(irqs_disabled());

	freqs->flags = cpufreq_driver->flags;
	dprintk("notification %u of frequency transition to %u kHz\n",
		state, freqs->new);

	policy = cpufreq_cpu_data[freqs->cpu];
	switch (state) {

	case CPUFREQ_PRECHANGE:
		/* detect if the driver reported a value as "old frequency"
		 * which is not equal to what the cpufreq core thinks is
		 * "old frequency".
		 */
		if (!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
			if ((policy) && (policy->cpu == freqs->cpu) &&
			    (policy->cur) && (policy->cur != freqs->old)) {
				dprintk("Warning: CPU frequency is"
					" %u, cpufreq assumed %u kHz.\n",
					freqs->old, policy->cur);
				freqs->old = policy->cur;
			}
		}
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
				CPUFREQ_PRECHANGE, freqs);
		adjust_jiffies(CPUFREQ_PRECHANGE, freqs);
		break;

	case CPUFREQ_POSTCHANGE:
		adjust_jiffies(CPUFREQ_POSTCHANGE, freqs);
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
				CPUFREQ_POSTCHANGE, freqs);
		if (likely(policy) && likely(policy->cpu == freqs->cpu))
			policy->cur = freqs->new;
		break;
	}
}
EXPORT_SYMBOL_GPL(cpufreq_notify_transition);



/*********************************************************************
 *                          SYSFS INTERFACE                          *
 *********************************************************************/

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))
			return t;

	return NULL;
}

/**
 * cpufreq_parse_governor - parse a governor string
 */
static int cpufreq_parse_governor (char *str_governor, unsigned int *policy,
				struct cpufreq_governor **governor)
{
	int err = -EINVAL;

	if (!cpufreq_driver)
		goto out;

	if (cpufreq_driver->setpolicy) {
		if (!strnicmp(str_governor, "performance", CPUFREQ_NAME_LEN)) {
			*policy = CPUFREQ_POLICY_PERFORMANCE;
			err = 0;
		} else if (!strnicmp(str_governor, "powersave",
						CPUFREQ_NAME_LEN)) {
			*policy = CPUFREQ_POLICY_POWERSAVE;
			err = 0;
		}
	} else if (cpufreq_driver->target) {
		struct cpufreq_governor *t;

		mutex_lock(&cpufreq_governor_mutex);

		t = __find_governor(str_governor);

		if (t == NULL) {
			char *name = kasprintf(GFP_KERNEL, "cpufreq_%s",
								str_governor);

			if (name) {
				int ret;

				mutex_unlock(&cpufreq_governor_mutex);
				ret = request_module(name);
				mutex_lock(&cpufreq_governor_mutex);

				if (ret == 0)
					t = __find_governor(str_governor);
			}

			kfree(name);
		}

		if (t != NULL) {
			*governor = t;
			err = 0;
		}

		mutex_unlock(&cpufreq_governor_mutex);
	}
  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
 *
 * Write out information from cpufreq_driver->policy[cpu]; object must be
 * "unsigned int".
 */

#define show_one(file_name, object)			\
static ssize_t show_##file_name				\
(struct cpufreq_policy * policy, char *buf)		\
{							\
	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(scaling_min_freq, min);
show_one(scaling_max_freq, max);
show_one(scaling_cur_freq, cur);

static int __cpufreq_set_policy(struct cpufreq_policy *data,
				struct cpufreq_policy *policy);

/**
 * cpufreq_per_cpu_attr_write() / store_##file_name() - sysfs write access
 */
#define store_one(file_name, object)			\
static ssize_t store_##file_name					\
(struct cpufreq_policy * policy, const char *buf, size_t count)		\
{									\
	unsigned int ret = -EINVAL;					\
	struct cpufreq_policy new_policy;				\