[deliverable/linux.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,
};

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->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_tuners_ins.ignore_nice = 0;

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