[linux-block.git] / drivers / cpufreq / cpufreq_ondemand.c

/*
 *  drivers/cpufreq/cpufreq_ondemand.c
 *
 *  Copyright (C)  2001 Russell King
 *            (C)  2003 Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>.
 *                      Jun Nakajima <jun.nakajima@intel.com>
 *
 * 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/smp.h>
#include <linux/init.h>
#include <linux/interrupt.h>
#include <linux/ctype.h>
#include <linux/cpufreq.h>
#include <linux/sysctl.h>
#include <linux/types.h>
#include <linux/fs.h>
#include <linux/sysfs.h>
#include <linux/sched.h>
#include <linux/kmod.h>
#include <linux/workqueue.h>
#include <linux/jiffies.h>
#include <linux/kernel_stat.h>
#include <linux/percpu.h>
#include <linux/mutex.h>

/*
 * dbs is used in this file as a shortform for demandbased switching
 * It helps to keep variable names smaller, simpler
 */

#define DEF_FREQUENCY_UP_THRESHOLD		(80)
#define MIN_FREQUENCY_UP_THRESHOLD		(11)
#define MAX_FREQUENCY_UP_THRESHOLD		(100)

/*
 * The polling frequency of this governor depends on the capability of
 * the processor. Default polling frequency is 1000 times the transition
 * latency of the processor. The governor will work on any processor with
 * transition latency <= 10mS, using appropriate sampling
 * rate.
 * For CPUs with transition latency > 10mS (mostly drivers with CPUFREQ_ETERNAL)
 * this governor will not work.
 * All times here are in uS.
 */
static unsigned int def_sampling_rate;
#define MIN_SAMPLING_RATE_RATIO			(2)
/* for correct statistics, we need at least 10 ticks between each measure */
#define MIN_STAT_SAMPLING_RATE			(MIN_SAMPLING_RATE_RATIO * jiffies_to_usecs(10))
#define MIN_SAMPLING_RATE			(def_sampling_rate / MIN_SAMPLING_RATE_RATIO)
#define MAX_SAMPLING_RATE			(500 * def_sampling_rate)
#define DEF_SAMPLING_RATE_LATENCY_MULTIPLIER	(1000)
#define DEF_SAMPLING_DOWN_FACTOR		(1)
#define MAX_SAMPLING_DOWN_FACTOR		(10)
#define TRANSITION_LATENCY_LIMIT		(10 * 1000)

static void do_dbs_timer(void *data);

struct cpu_dbs_info_s {
	struct cpufreq_policy *cur_policy;
	unsigned int prev_cpu_idle_up;
	unsigned int prev_cpu_idle_down;
	unsigned int enable;
};
static DEFINE_PER_CPU(struct cpu_dbs_info_s, cpu_dbs_info);

static unsigned int dbs_enable;	/* number of CPUs using this policy */

static DEFINE_MUTEX (dbs_mutex);
static DECLARE_WORK	(dbs_work, do_dbs_timer, NULL);

struct dbs_tuners {
	unsigned int sampling_rate;
	unsigned int sampling_down_factor;
	unsigned int up_threshold;
	unsigned int ignore_nice;
};

static struct dbs_tuners dbs_tuners_ins = {
	.up_threshold = DEF_FREQUENCY_UP_THRESHOLD,
	.sampling_down_factor = DEF_SAMPLING_DOWN_FACTOR,
	.ignore_nice = 0,
};

static inline unsigned int get_cpu_idle_time(unsigned int cpu)
{
	return	kstat_cpu(cpu).cpustat.idle +
		kstat_cpu(cpu).cpustat.iowait +
		( dbs_tuners_ins.ignore_nice ?
		  kstat_cpu(cpu).cpustat.nice :
		  0);
}

/************************** sysfs interface ************************/
static ssize_t show_sampling_rate_max(struct cpufreq_policy *policy, char *buf)
{
	return sprintf (buf, "%u\n", MAX_SAMPLING_RATE);
}

static ssize_t show_sampling_rate_min(struct cpufreq_policy *policy, char *buf)
{
	return sprintf (buf, "%u\n", MIN_SAMPLING_RATE);
}

#define define_one_ro(_name)		\
static struct freq_attr _name =		\
__ATTR(_name, 0444, show_##_name, NULL)

define_one_ro(sampling_rate_max);
define_one_ro(sampling_rate_min);

/* cpufreq_ondemand Governor Tunables */
#define show_one(file_name, object)					\
static ssize_t show_##file_name						\
(struct cpufreq_policy *unused, char *buf)				\
{									\
	return sprintf(buf, "%u\n", dbs_tuners_ins.object);		\
}
show_one(sampling_rate, sampling_rate);
show_one(sampling_down_factor, sampling_down_factor);
show_one(up_threshold, up_threshold);
show_one(ignore_nice_load, ignore_nice);

static ssize_t store_sampling_down_factor(struct cpufreq_policy *unused,
		const char *buf, size_t count)
{
	unsigned int input;
	int ret;
	ret = sscanf (buf, "%u", &input);
	if (ret != 1 )
		return -EINVAL;

	if (input > MAX_SAMPLING_DOWN_FACTOR || input < 1)
		return -EINVAL;

	mutex_lock(&dbs_mutex);
	dbs_tuners_ins.sampling_down_factor = input;
	mutex_unlock(&dbs_mutex);

	return count;
}

static ssize_t store_sampling_rate(struct cpufreq_policy *unused,
		const char *buf, size_t count)
{
	unsigned int input;
	int ret;
	ret = sscanf (buf, "%u", &input);

	mutex_lock(&dbs_mutex);
	if (ret != 1 || input > MAX_SAMPLING_RATE || input < MIN_SAMPLING_RATE) {
		mutex_unlock(&dbs_mutex);
		return -EINVAL;
	}

	dbs_tuners_ins.sampling_rate = input;
	mutex_unlock(&dbs_mutex);

	return count;
}

static ssize_t store_up_threshold(struct cpufreq_policy *unused,
		const char *buf, size_t count)
{
	unsigned int input;
	int ret;
	ret = sscanf (buf, "%u", &input);

	mutex_lock(&dbs_mutex);
	if (ret != 1 || input > MAX_FREQUENCY_UP_THRESHOLD ||
			input < MIN_FREQUENCY_UP_THRESHOLD) {
		mutex_unlock(&dbs_mutex);
		return -EINVAL;
	}

	dbs_tuners_ins.up_threshold = input;
	mutex_unlock(&dbs_mutex);

	return count;
}

static ssize_t store_ignore_nice_load(struct cpufreq_policy *policy,
		const char *buf, size_t count)
{
	unsigned int input;
	int ret;

	unsigned int j;

	ret = sscanf (buf, "%u", &input);
	if ( ret != 1 )
		return -EINVAL;

	if ( input > 1 )
		input = 1;

	mutex_lock(&dbs_mutex);
	if ( input == dbs_tuners_ins.ignore_nice ) { /* nothing to do */
		mutex_unlock(&dbs_mutex);
		return count;
	}
	dbs_tuners_ins.ignore_nice = input;

	/* we need to re-evaluate prev_cpu_idle_up and prev_cpu_idle_down */
	for_each_online_cpu(j) {
		struct cpu_dbs_info_s *j_dbs_info;
		j_dbs_info = &per_cpu(cpu_dbs_info, j);
		j_dbs_info->prev_cpu_idle_up = get_cpu_idle_time(j);
		j_dbs_info->prev_cpu_idle_down = j_dbs_info->prev_cpu_idle_up;
	}
	mutex_unlock(&dbs_mutex);

	return count;
}

#define define_one_rw(_name) \
static struct freq_attr _name = \
__ATTR(_name, 0644, show_##_name, store_##_name)

define_one_rw(sampling_rate);
define_one_rw(sampling_down_factor);
define_one_rw(up_threshold);
define_one_rw(ignore_nice_load);

static struct attribute * dbs_attributes[] = {
	&sampling_rate_max.attr,
	&sampling_rate_min.attr,
	&sampling_rate.attr,
	&sampling_down_factor.attr,
	&up_threshold.attr,
	&ignore_nice_load.attr,
	NULL
};

static struct attribute_group dbs_attr_group = {
	.attrs = dbs_attributes,
	.name = "ondemand",
};

/************************** sysfs end ************************/

static void dbs_check_cpu(int cpu)
{
	unsigned int idle_ticks, up_idle_ticks, total_ticks;
	unsigned int freq_next;
	unsigned int freq_down_sampling_rate;
	static int down_skip[NR_CPUS];
	struct cpu_dbs_info_s *this_dbs_info;

	struct cpufreq_policy *policy;
	unsigned int j;

	this_dbs_info = &per_cpu(cpu_dbs_info, cpu);
	if (!this_dbs_info->enable)
		return;

	policy = this_dbs_info->cur_policy;
	/*
	 * Every sampling_rate, we check, if current idle time is less
	 * than 20% (default), then we try to increase frequency
	 * Every sampling_rate*sampling_down_factor, we look for a the lowest
	 * frequency which can sustain the load while keeping idle time over
	 * 30%. If such a frequency exist, we try to decrease to this frequency.
	 *
	 * Any frequency increase takes it to the maximum frequency.
	 * Frequency reduction happens at minimum steps of
	 * 5% (default) of current frequency
	 */

	/* Check for frequency increase */
	idle_ticks = UINT_MAX;
	for_each_cpu_mask(j, policy->cpus) {
		unsigned int tmp_idle_ticks, total_idle_ticks;
		struct cpu_dbs_info_s *j_dbs_info;

		j_dbs_info = &per_cpu(cpu_dbs_info, j);
		total_idle_ticks = get_cpu_idle_time(j);
		tmp_idle_ticks = total_idle_ticks -
			j_dbs_info->prev_cpu_idle_up;
		j_dbs_info->prev_cpu_idle_up = total_idle_ticks;

		if (tmp_idle_ticks < idle_ticks)
			idle_ticks = tmp_idle_ticks;
	}

	/* Scale idle ticks by 100 and compare with up and down ticks */
	idle_ticks *= 100;
	up_idle_ticks = (100 - dbs_tuners_ins.up_threshold) *
			usecs_to_jiffies(dbs_tuners_ins.sampling_rate);

	if (idle_ticks < up_idle_ticks) {
		down_skip[cpu] = 0;
		for_each_cpu_mask(j, policy->cpus) {
			struct cpu_dbs_info_s *j_dbs_info;

			j_dbs_info = &per_cpu(cpu_dbs_info, j);
			j_dbs_info->prev_cpu_idle_down =
					j_dbs_info->prev_cpu_idle_up;
		}
		/* if we are already at full speed then break out early */
		if (policy->cur == policy->max)
			return;

		__cpufreq_driver_target(policy, policy->max,
			CPUFREQ_RELATION_H);
		return;
	}

	/* Check for frequency decrease */
	down_skip[cpu]++;
	if (down_skip[cpu] < dbs_tuners_ins.sampling_down_factor)
		return;

	idle_ticks = UINT_MAX;
	for_each_cpu_mask(j, policy->cpus) {
		unsigned int tmp_idle_ticks, total_idle_ticks;
		struct cpu_dbs_info_s *j_dbs_info;

		j_dbs_info = &per_cpu(cpu_dbs_info, j);
		/* Check for frequency decrease */
		total_idle_ticks = j_dbs_info->prev_cpu_idle_up;
		tmp_idle_ticks = total_idle_ticks -
			j_dbs_info->prev_cpu_idle_down;
		j_dbs_info->prev_cpu_idle_down = total_idle_ticks;

		if (tmp_idle_ticks < idle_ticks)
			idle_ticks = tmp_idle_ticks;
	}

	down_skip[cpu] = 0;
	/* if we cannot reduce the frequency anymore, break out early */
	if (policy->cur == policy->min)
		return;

	/* Compute how many ticks there are between two measurements */
	freq_down_sampling_rate = dbs_tuners_ins.sampling_rate *
		dbs_tuners_ins.sampling_down_factor;
	total_ticks = usecs_to_jiffies(freq_down_sampling_rate);

	/*
	 * The optimal frequency is the frequency that is the lowest that
	 * can support the current CPU usage without triggering the up
	 * policy. To be safe, we focus 10 points under the threshold.
	 */
	freq_next = ((total_ticks - idle_ticks) * 100) / total_ticks;
	freq_next = (freq_next * policy->cur) /
			(dbs_tuners_ins.up_threshold - 10);

	if (freq_next < policy->min)
		freq_next = policy->min;

	if (freq_next <= ((policy->cur * 95) / 100))
		__cpufreq_driver_target(policy, freq_next, CPUFREQ_RELATION_L);
}

static void do_dbs_timer(void *data)
{
	int i;
	mutex_lock(&dbs_mutex);
	for_each_online_cpu(i)
		dbs_check_cpu(i);
	schedule_delayed_work(&dbs_work,
			usecs_to_jiffies(dbs_tuners_ins.sampling_rate));
	mutex_unlock(&dbs_mutex);
}

static inline void dbs_timer_init(void)
{
	INIT_WORK(&dbs_work, do_dbs_timer, NULL);
	schedule_delayed_work(&dbs_work,
			usecs_to_jiffies(dbs_tuners_ins.sampling_rate));
	return;
}

static inline void dbs_timer_exit(void)
{
	cancel_delayed_work(&dbs_work);
	return;
}

static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
				   unsigned int event)
{
	unsigned int cpu = policy->cpu;
	struct cpu_dbs_info_s *this_dbs_info;
	unsigned int j;

	this_dbs_info = &per_cpu(cpu_dbs_info, cpu);

	switch (event) {
	case CPUFREQ_GOV_START:
		if ((!cpu_online(cpu)) ||
		    (!policy->cur))
			return -EINVAL;

		if (policy->cpuinfo.transition_latency >
				(TRANSITION_LATENCY_LIMIT * 1000)) {
			printk(KERN_WARNING "ondemand governor failed to load "
			       "due to too long transition latency\n");
			return -EINVAL;
		}
		if (this_dbs_info->enable) /* Already enabled */
			break;

		mutex_lock(&dbs_mutex);
		for_each_cpu_mask(j, policy->cpus) {
			struct cpu_dbs_info_s *j_dbs_info;
			j_dbs_info = &per_cpu(cpu_dbs_info, j);
			j_dbs_info->cur_policy = policy;

			j_dbs_info->prev_cpu_idle_up = get_cpu_idle_time(j);
			j_dbs_info->prev_cpu_idle_down
				= j_dbs_info->prev_cpu_idle_up;
		}
		this_dbs_info->enable = 1;
		sysfs_create_group(&policy->kobj, &dbs_attr_group);
		dbs_enable++;
		/*
		 * Start the timerschedule work, when this governor
		 * is used for first time
		 */
		if (dbs_enable == 1) {
			unsigned int latency;
			/* policy latency is in nS. Convert it to uS first */
			latency = policy->cpuinfo.transition_latency / 1000;
			if (latency == 0)
				latency = 1;

			def_sampling_rate = latency *
					DEF_SAMPLING_RATE_LATENCY_MULTIPLIER;

			if (def_sampling_rate < MIN_STAT_SAMPLING_RATE)
				def_sampling_rate = MIN_STAT_SAMPLING_RATE;

			dbs_tuners_ins.sampling_rate = def_sampling_rate;
			dbs_timer_init();
		}

		mutex_unlock(&dbs_mutex);
		break;

	case CPUFREQ_GOV_STOP:
		mutex_lock(&dbs_mutex);
		this_dbs_info->enable = 0;
		sysfs_remove_group(&policy->kobj, &dbs_attr_group);
		dbs_enable--;
		/*
		 * Stop the timerschedule work, when this governor
		 * is used for first time
		 */
		if (dbs_enable == 0)
			dbs_timer_exit();

		mutex_unlock(&dbs_mutex);

		break;

	case CPUFREQ_GOV_LIMITS:
		mutex_lock(&dbs_mutex);
		if (policy->max < this_dbs_info->cur_policy->cur)
			__cpufreq_driver_target(
					this_dbs_info->cur_policy,
					policy->max, CPUFREQ_RELATION_H);
		else if (policy->min > this_dbs_info->cur_policy->cur)
			__cpufreq_driver_target(
					this_dbs_info->cur_policy,
					policy->min, CPUFREQ_RELATION_L);
		mutex_unlock(&dbs_mutex);
		break;
	}
	return 0;
}

static struct cpufreq_governor cpufreq_gov_dbs = {
	.name		= "ondemand",
	.governor	= cpufreq_governor_dbs,
	.owner		= THIS_MODULE,
};

static int __init cpufreq_gov_dbs_init(void)
{
	return cpufreq_register_governor(&cpufreq_gov_dbs);
}

static void __exit cpufreq_gov_dbs_exit(void)
{
	/* Make sure that the scheduled work is indeed not running */
	flush_scheduled_work();

	cpufreq_unregister_governor(&cpufreq_gov_dbs);
}


MODULE_AUTHOR ("Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>");
MODULE_DESCRIPTION ("'cpufreq_ondemand' - A dynamic cpufreq governor for "
		"Low Latency Frequency Transition capable processors");
MODULE_LICENSE ("GPL");

module_init(cpufreq_gov_dbs_init);
module_exit(cpufreq_gov_dbs_exit);
Commit	Line	Data
1da177e4 LT	1	/*
	2	* drivers/cpufreq/cpufreq_ondemand.c
	3	*
	4	* Copyright (C) 2001 Russell King
	5	* (C) 2003 Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>.
	6	* Jun Nakajima <jun.nakajima@intel.com>
	7	*
	8	* This program is free software; you can redistribute it and/or modify
	9	* it under the terms of the GNU General Public License version 2 as
	10	* published by the Free Software Foundation.
	11	*/
	12
	13	#include <linux/kernel.h>
	14	#include <linux/module.h>
	15	#include <linux/smp.h>
	16	#include <linux/init.h>
	17	#include <linux/interrupt.h>
	18	#include <linux/ctype.h>
	19	#include <linux/cpufreq.h>
	20	#include <linux/sysctl.h>
	21	#include <linux/types.h>
	22	#include <linux/fs.h>
	23	#include <linux/sysfs.h>
	24	#include <linux/sched.h>
	25	#include <linux/kmod.h>
	26	#include <linux/workqueue.h>
	27	#include <linux/jiffies.h>
	28	#include <linux/kernel_stat.h>
	29	#include <linux/percpu.h>
3fc54d37	30	#include <linux/mutex.h>
1da177e4 LT	31
	32	/*
	33	* dbs is used in this file as a shortform for demandbased switching
	34	* It helps to keep variable names smaller, simpler
	35	*/
	36
	37	#define DEF_FREQUENCY_UP_THRESHOLD (80)
c29f1403	38	#define MIN_FREQUENCY_UP_THRESHOLD (11)
1da177e4 LT	39	#define MAX_FREQUENCY_UP_THRESHOLD (100)
1da177e4 LT	40
32ee8c3e DJ	41	/*
32ee8c3e DJ	42	* The polling frequency of this governor depends on the capability of
1da177e4	43	* the processor. Default polling frequency is 1000 times the transition
32ee8c3e DJ	44	* latency of the processor. The governor will work on any processor with
32ee8c3e DJ	45	* transition latency <= 10mS, using appropriate sampling
1da177e4 LT	46	* rate.
	47	* For CPUs with transition latency > 10mS (mostly drivers with CPUFREQ_ETERNAL)
	48	* this governor will not work.
	49	* All times here are in uS.
	50	*/
32ee8c3e	51	static unsigned int def_sampling_rate;
df8b59be DJ	52	#define MIN_SAMPLING_RATE_RATIO (2)
	53	/* for correct statistics, we need at least 10 ticks between each measure */
	54	#define MIN_STAT_SAMPLING_RATE (MIN_SAMPLING_RATE_RATIO * jiffies_to_usecs(10))
	55	#define MIN_SAMPLING_RATE (def_sampling_rate / MIN_SAMPLING_RATE_RATIO)
1da177e4 LT	56	#define MAX_SAMPLING_RATE (500 * def_sampling_rate)
1da177e4 LT	57	#define DEF_SAMPLING_RATE_LATENCY_MULTIPLIER (1000)
e131832c DJ	58	#define DEF_SAMPLING_DOWN_FACTOR (1)
e131832c DJ	59	#define MAX_SAMPLING_DOWN_FACTOR (10)
1da177e4	60	#define TRANSITION_LATENCY_LIMIT (10 * 1000)
1da177e4 LT	61
	62	static void do_dbs_timer(void *data);
	63
	64	struct cpu_dbs_info_s {
32ee8c3e DJ	65	struct cpufreq_policy *cur_policy;
	66	unsigned int prev_cpu_idle_up;
	67	unsigned int prev_cpu_idle_down;
	68	unsigned int enable;
1da177e4 LT	69	};
	70	static DEFINE_PER_CPU(struct cpu_dbs_info_s, cpu_dbs_info);
	71
	72	static unsigned int dbs_enable; /* number of CPUs using this policy */
	73
32ee8c3e	74	static DEFINE_MUTEX (dbs_mutex);
1da177e4 LT	75	static DECLARE_WORK (dbs_work, do_dbs_timer, NULL);
	76
	77	struct dbs_tuners {
32ee8c3e DJ	78	unsigned int sampling_rate;
	79	unsigned int sampling_down_factor;
	80	unsigned int up_threshold;
	81	unsigned int ignore_nice;
1da177e4 LT	82	};
	83
	84	static struct dbs_tuners dbs_tuners_ins = {
32ee8c3e DJ	85	.up_threshold = DEF_FREQUENCY_UP_THRESHOLD,
32ee8c3e DJ	86	.sampling_down_factor = DEF_SAMPLING_DOWN_FACTOR,
9cbad61b	87	.ignore_nice = 0,
1da177e4 LT	88	};
1da177e4 LT	89
dac1c1a5 DJ	90	static inline unsigned int get_cpu_idle_time(unsigned int cpu)
	91	{
	92	return kstat_cpu(cpu).cpustat.idle +
	93	kstat_cpu(cpu).cpustat.iowait +
001893cd	94	( dbs_tuners_ins.ignore_nice ?
dac1c1a5 DJ	95	kstat_cpu(cpu).cpustat.nice :
	96	0);
	97	}
	98
1da177e4 LT	99	/************************ sysfs interface **********************/
	100	static ssize_t show_sampling_rate_max(struct cpufreq_policy policy, char buf)
	101	{
	102	return sprintf (buf, "%u\n", MAX_SAMPLING_RATE);
	103	}
	104
	105	static ssize_t show_sampling_rate_min(struct cpufreq_policy policy, char buf)
	106	{
	107	return sprintf (buf, "%u\n", MIN_SAMPLING_RATE);
	108	}
	109
32ee8c3e DJ	110	#define define_one_ro(_name) \
32ee8c3e DJ	111	static struct freq_attr _name = \
1da177e4 LT	112	__ATTR(_name, 0444, show_##_name, NULL)
	113
	114	define_one_ro(sampling_rate_max);
	115	define_one_ro(sampling_rate_min);
	116
	117	/* cpufreq_ondemand Governor Tunables */
	118	#define show_one(file_name, object) \
	119	static ssize_t show_##file_name \
	120	(struct cpufreq_policy unused, char buf) \
	121	{ \
	122	return sprintf(buf, "%u\n", dbs_tuners_ins.object); \
	123	}
	124	show_one(sampling_rate, sampling_rate);
	125	show_one(sampling_down_factor, sampling_down_factor);
	126	show_one(up_threshold, up_threshold);
001893cd	127	show_one(ignore_nice_load, ignore_nice);
1da177e4	128
32ee8c3e	129	static ssize_t store_sampling_down_factor(struct cpufreq_policy *unused,
1da177e4 LT	130	const char *buf, size_t count)
	131	{
	132	unsigned int input;
	133	int ret;
	134	ret = sscanf (buf, "%u", &input);
	135	if (ret != 1 )
	136	return -EINVAL;
	137
e131832c DJ	138	if (input > MAX_SAMPLING_DOWN_FACTOR \|\| input < 1)
	139	return -EINVAL;
	140
3fc54d37	141	mutex_lock(&dbs_mutex);
1da177e4	142	dbs_tuners_ins.sampling_down_factor = input;
3fc54d37	143	mutex_unlock(&dbs_mutex);
1da177e4 LT	144
	145	return count;
	146	}
	147
32ee8c3e	148	static ssize_t store_sampling_rate(struct cpufreq_policy *unused,
1da177e4 LT	149	const char *buf, size_t count)
	150	{
	151	unsigned int input;
	152	int ret;
	153	ret = sscanf (buf, "%u", &input);
	154
3fc54d37	155	mutex_lock(&dbs_mutex);
1da177e4	156	if (ret != 1 \|\| input > MAX_SAMPLING_RATE \|\| input < MIN_SAMPLING_RATE) {
3fc54d37	157	mutex_unlock(&dbs_mutex);
1da177e4 LT	158	return -EINVAL;
	159	}
	160
	161	dbs_tuners_ins.sampling_rate = input;
3fc54d37	162	mutex_unlock(&dbs_mutex);
1da177e4 LT	163
	164	return count;
	165	}
	166
32ee8c3e	167	static ssize_t store_up_threshold(struct cpufreq_policy *unused,
1da177e4 LT	168	const char *buf, size_t count)
	169	{
	170	unsigned int input;
	171	int ret;
	172	ret = sscanf (buf, "%u", &input);
	173
3fc54d37	174	mutex_lock(&dbs_mutex);
32ee8c3e	175	if (ret != 1 \|\| input > MAX_FREQUENCY_UP_THRESHOLD \|\|
c29f1403	176	input < MIN_FREQUENCY_UP_THRESHOLD) {
3fc54d37	177	mutex_unlock(&dbs_mutex);
1da177e4 LT	178	return -EINVAL;
	179	}
	180
	181	dbs_tuners_ins.up_threshold = input;
3fc54d37	182	mutex_unlock(&dbs_mutex);
1da177e4 LT	183
	184	return count;
	185	}
	186
001893cd	187	static ssize_t store_ignore_nice_load(struct cpufreq_policy *policy,
3d5ee9e5 DJ	188	const char *buf, size_t count)
	189	{
	190	unsigned int input;
	191	int ret;
	192
	193	unsigned int j;
32ee8c3e	194
3d5ee9e5 DJ	195	ret = sscanf (buf, "%u", &input);
	196	if ( ret != 1 )
	197	return -EINVAL;
	198
	199	if ( input > 1 )
	200	input = 1;
32ee8c3e	201
3fc54d37	202	mutex_lock(&dbs_mutex);
3d5ee9e5	203	if ( input == dbs_tuners_ins.ignore_nice ) { /* nothing to do */
3fc54d37	204	mutex_unlock(&dbs_mutex);
3d5ee9e5 DJ	205	return count;
	206	}
	207	dbs_tuners_ins.ignore_nice = input;
	208
	209	/* we need to re-evaluate prev_cpu_idle_up and prev_cpu_idle_down */
dac1c1a5	210	for_each_online_cpu(j) {
3d5ee9e5 DJ	211	struct cpu_dbs_info_s *j_dbs_info;
3d5ee9e5 DJ	212	j_dbs_info = &per_cpu(cpu_dbs_info, j);
dac1c1a5	213	j_dbs_info->prev_cpu_idle_up = get_cpu_idle_time(j);
3d5ee9e5 DJ	214	j_dbs_info->prev_cpu_idle_down = j_dbs_info->prev_cpu_idle_up;
3d5ee9e5 DJ	215	}
3fc54d37	216	mutex_unlock(&dbs_mutex);
3d5ee9e5 DJ	217
	218	return count;
	219	}
	220
1da177e4 LT	221	#define define_one_rw(_name) \
	222	static struct freq_attr _name = \
	223	__ATTR(_name, 0644, show_##_name, store_##_name)
	224
	225	define_one_rw(sampling_rate);
	226	define_one_rw(sampling_down_factor);
	227	define_one_rw(up_threshold);
001893cd	228	define_one_rw(ignore_nice_load);
1da177e4 LT	229
	230	static struct attribute * dbs_attributes[] = {
	231	&sampling_rate_max.attr,
	232	&sampling_rate_min.attr,
	233	&sampling_rate.attr,
	234	&sampling_down_factor.attr,
	235	&up_threshold.attr,
001893cd	236	&ignore_nice_load.attr,
1da177e4 LT	237	NULL
	238	};
	239
	240	static struct attribute_group dbs_attr_group = {
	241	.attrs = dbs_attributes,
	242	.name = "ondemand",
	243	};
	244
	245	/************************ sysfs end **********************/
	246
	247	static void dbs_check_cpu(int cpu)
	248	{
c29f1403 DJ	249	unsigned int idle_ticks, up_idle_ticks, total_ticks;
c29f1403 DJ	250	unsigned int freq_next;
1da177e4 LT	251	unsigned int freq_down_sampling_rate;
	252	static int down_skip[NR_CPUS];
	253	struct cpu_dbs_info_s *this_dbs_info;
	254
	255	struct cpufreq_policy *policy;
	256	unsigned int j;
	257
	258	this_dbs_info = &per_cpu(cpu_dbs_info, cpu);
	259	if (!this_dbs_info->enable)
	260	return;
	261
	262	policy = this_dbs_info->cur_policy;
32ee8c3e	263	/*
c29f1403 DJ	264	* Every sampling_rate, we check, if current idle time is less
	265	* than 20% (default), then we try to increase frequency
	266	* Every sampling_rate*sampling_down_factor, we look for a the lowest
	267	* frequency which can sustain the load while keeping idle time over
	268	* 30%. If such a frequency exist, we try to decrease to this frequency.
1da177e4	269	*
32ee8c3e DJ	270	* Any frequency increase takes it to the maximum frequency.
	271	* Frequency reduction happens at minimum steps of
	272	* 5% (default) of current frequency
1da177e4 LT	273	*/
	274
	275	/* Check for frequency increase */
9c7d269b	276	idle_ticks = UINT_MAX;
1da177e4	277	for_each_cpu_mask(j, policy->cpus) {
9c7d269b	278	unsigned int tmp_idle_ticks, total_idle_ticks;
1da177e4 LT	279	struct cpu_dbs_info_s *j_dbs_info;
1da177e4 LT	280
1da177e4	281	j_dbs_info = &per_cpu(cpu_dbs_info, j);
dac1c1a5	282	total_idle_ticks = get_cpu_idle_time(j);
1da177e4 LT	283	tmp_idle_ticks = total_idle_ticks -
	284	j_dbs_info->prev_cpu_idle_up;
	285	j_dbs_info->prev_cpu_idle_up = total_idle_ticks;
	286
	287	if (tmp_idle_ticks < idle_ticks)
	288	idle_ticks = tmp_idle_ticks;
	289	}
	290
	291	/* Scale idle ticks by 100 and compare with up and down ticks */
	292	idle_ticks *= 100;
	293	up_idle_ticks = (100 - dbs_tuners_ins.up_threshold) *
6fe71165	294	usecs_to_jiffies(dbs_tuners_ins.sampling_rate);
1da177e4 LT	295
1da177e4 LT	296	if (idle_ticks < up_idle_ticks) {
dac1c1a5	297	down_skip[cpu] = 0;
790d76fa DJ	298	for_each_cpu_mask(j, policy->cpus) {
	299	struct cpu_dbs_info_s *j_dbs_info;
	300
	301	j_dbs_info = &per_cpu(cpu_dbs_info, j);
32ee8c3e	302	j_dbs_info->prev_cpu_idle_down =
790d76fa DJ	303	j_dbs_info->prev_cpu_idle_up;
790d76fa DJ	304	}
c11420a6 DJ	305	/* if we are already at full speed then break out early */
	306	if (policy->cur == policy->max)
	307	return;
32ee8c3e DJ	308
32ee8c3e DJ	309	__cpufreq_driver_target(policy, policy->max,
1da177e4	310	CPUFREQ_RELATION_H);
1da177e4 LT	311	return;
	312	}
	313
	314	/* Check for frequency decrease */
	315	down_skip[cpu]++;
	316	if (down_skip[cpu] < dbs_tuners_ins.sampling_down_factor)
	317	return;
	318
9c7d269b	319	idle_ticks = UINT_MAX;
1da177e4	320	for_each_cpu_mask(j, policy->cpus) {
9c7d269b	321	unsigned int tmp_idle_ticks, total_idle_ticks;
1da177e4 LT	322	struct cpu_dbs_info_s *j_dbs_info;
1da177e4 LT	323
1da177e4	324	j_dbs_info = &per_cpu(cpu_dbs_info, j);
dac1c1a5 DJ	325	/* Check for frequency decrease */
dac1c1a5 DJ	326	total_idle_ticks = j_dbs_info->prev_cpu_idle_up;
1da177e4 LT	327	tmp_idle_ticks = total_idle_ticks -
	328	j_dbs_info->prev_cpu_idle_down;
	329	j_dbs_info->prev_cpu_idle_down = total_idle_ticks;
	330
	331	if (tmp_idle_ticks < idle_ticks)
	332	idle_ticks = tmp_idle_ticks;
	333	}
	334
1da177e4	335	down_skip[cpu] = 0;
c29f1403 DJ	336	/* if we cannot reduce the frequency anymore, break out early */
	337	if (policy->cur == policy->min)
	338	return;
1da177e4	339
c29f1403	340	/* Compute how many ticks there are between two measurements */
1da177e4 LT	341	freq_down_sampling_rate = dbs_tuners_ins.sampling_rate *
1da177e4 LT	342	dbs_tuners_ins.sampling_down_factor;
c29f1403	343	total_ticks = usecs_to_jiffies(freq_down_sampling_rate);
1206aaac	344
c29f1403 DJ	345	/*
	346	* The optimal frequency is the frequency that is the lowest that
	347	* can support the current CPU usage without triggering the up
	348	* policy. To be safe, we focus 10 points under the threshold.
	349	*/
	350	freq_next = ((total_ticks - idle_ticks) * 100) / total_ticks;
32ee8c3e	351	freq_next = (freq_next * policy->cur) /
c29f1403	352	(dbs_tuners_ins.up_threshold - 10);
1da177e4	353
7c9d8c0e DB	354	if (freq_next < policy->min)
	355	freq_next = policy->min;
	356
c29f1403 DJ	357	if (freq_next <= ((policy->cur * 95) / 100))
c29f1403 DJ	358	__cpufreq_driver_target(policy, freq_next, CPUFREQ_RELATION_L);
1da177e4 LT	359	}
	360
	361	static void do_dbs_timer(void *data)
32ee8c3e	362	{
1da177e4	363	int i;
3fc54d37	364	mutex_lock(&dbs_mutex);
6fe71165 DJ	365	for_each_online_cpu(i)
6fe71165 DJ	366	dbs_check_cpu(i);
32ee8c3e	367	schedule_delayed_work(&dbs_work,
6fe71165	368	usecs_to_jiffies(dbs_tuners_ins.sampling_rate));
3fc54d37	369	mutex_unlock(&dbs_mutex);
32ee8c3e	370	}
1da177e4 LT	371
	372	static inline void dbs_timer_init(void)
	373	{
	374	INIT_WORK(&dbs_work, do_dbs_timer, NULL);
	375	schedule_delayed_work(&dbs_work,
6fe71165	376	usecs_to_jiffies(dbs_tuners_ins.sampling_rate));
1da177e4 LT	377	return;
	378	}
	379
	380	static inline void dbs_timer_exit(void)
	381	{
	382	cancel_delayed_work(&dbs_work);
	383	return;
	384	}
	385
	386	static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
	387	unsigned int event)
	388	{
	389	unsigned int cpu = policy->cpu;
	390	struct cpu_dbs_info_s *this_dbs_info;
	391	unsigned int j;
	392
	393	this_dbs_info = &per_cpu(cpu_dbs_info, cpu);
	394
	395	switch (event) {
	396	case CPUFREQ_GOV_START:
32ee8c3e	397	if ((!cpu_online(cpu)) \|\|
1da177e4 LT	398	(!policy->cur))
	399	return -EINVAL;
	400
	401	if (policy->cpuinfo.transition_latency >
ff8c288d EP	402	(TRANSITION_LATENCY_LIMIT * 1000)) {
	403	printk(KERN_WARNING "ondemand governor failed to load "
	404	"due to too long transition latency\n");
1da177e4	405	return -EINVAL;
ff8c288d	406	}
1da177e4 LT	407	if (this_dbs_info->enable) /* Already enabled */
1da177e4 LT	408	break;
32ee8c3e	409
3fc54d37	410	mutex_lock(&dbs_mutex);
1da177e4 LT	411	for_each_cpu_mask(j, policy->cpus) {
	412	struct cpu_dbs_info_s *j_dbs_info;
	413	j_dbs_info = &per_cpu(cpu_dbs_info, j);
	414	j_dbs_info->cur_policy = policy;
32ee8c3e	415
dac1c1a5	416	j_dbs_info->prev_cpu_idle_up = get_cpu_idle_time(j);
3d5ee9e5 DJ	417	j_dbs_info->prev_cpu_idle_down
3d5ee9e5 DJ	418	= j_dbs_info->prev_cpu_idle_up;
1da177e4 LT	419	}
	420	this_dbs_info->enable = 1;
	421	sysfs_create_group(&policy->kobj, &dbs_attr_group);
	422	dbs_enable++;
	423	/*
	424	* Start the timerschedule work, when this governor
	425	* is used for first time
	426	*/
	427	if (dbs_enable == 1) {
	428	unsigned int latency;
	429	/* policy latency is in nS. Convert it to uS first */
df8b59be DJ	430	latency = policy->cpuinfo.transition_latency / 1000;
	431	if (latency == 0)
	432	latency = 1;
1da177e4	433
df8b59be	434	def_sampling_rate = latency *
1da177e4	435	DEF_SAMPLING_RATE_LATENCY_MULTIPLIER;
df8b59be DJ	436
	437	if (def_sampling_rate < MIN_STAT_SAMPLING_RATE)
	438	def_sampling_rate = MIN_STAT_SAMPLING_RATE;
	439
1da177e4	440	dbs_tuners_ins.sampling_rate = def_sampling_rate;
1da177e4 LT	441	dbs_timer_init();
1da177e4 LT	442	}
32ee8c3e	443
3fc54d37	444	mutex_unlock(&dbs_mutex);
1da177e4 LT	445	break;
	446
	447	case CPUFREQ_GOV_STOP:
3fc54d37	448	mutex_lock(&dbs_mutex);
1da177e4 LT	449	this_dbs_info->enable = 0;
	450	sysfs_remove_group(&policy->kobj, &dbs_attr_group);
	451	dbs_enable--;
	452	/*
	453	* Stop the timerschedule work, when this governor
	454	* is used for first time
	455	*/
32ee8c3e	456	if (dbs_enable == 0)
1da177e4	457	dbs_timer_exit();
32ee8c3e	458
3fc54d37	459	mutex_unlock(&dbs_mutex);
1da177e4 LT	460
	461	break;
	462
	463	case CPUFREQ_GOV_LIMITS:
3fc54d37	464	mutex_lock(&dbs_mutex);
1da177e4 LT	465	if (policy->max < this_dbs_info->cur_policy->cur)
	466	__cpufreq_driver_target(
	467	this_dbs_info->cur_policy,
32ee8c3e	468	policy->max, CPUFREQ_RELATION_H);
1da177e4 LT	469	else if (policy->min > this_dbs_info->cur_policy->cur)
	470	__cpufreq_driver_target(
	471	this_dbs_info->cur_policy,
32ee8c3e	472	policy->min, CPUFREQ_RELATION_L);
3fc54d37	473	mutex_unlock(&dbs_mutex);
1da177e4 LT	474	break;
	475	}
	476	return 0;
	477	}
	478
7f335d4e	479	static struct cpufreq_governor cpufreq_gov_dbs = {
1da177e4 LT	480	.name = "ondemand",
	481	.governor = cpufreq_governor_dbs,
	482	.owner = THIS_MODULE,
	483	};
1da177e4 LT	484
	485	static int __init cpufreq_gov_dbs_init(void)
	486	{
	487	return cpufreq_register_governor(&cpufreq_gov_dbs);
	488	}
	489
	490	static void __exit cpufreq_gov_dbs_exit(void)
	491	{
	492	/* Make sure that the scheduled work is indeed not running */
	493	flush_scheduled_work();
	494
	495	cpufreq_unregister_governor(&cpufreq_gov_dbs);
	496	}
	497
	498
	499	MODULE_AUTHOR ("Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>");
	500	MODULE_DESCRIPTION ("'cpufreq_ondemand' - A dynamic cpufreq governor for "
	501	"Low Latency Frequency Transition capable processors");
	502	MODULE_LICENSE ("GPL");
	503
	504	module_init(cpufreq_gov_dbs_init);
	505	module_exit(cpufreq_gov_dbs_exit);