[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>

/*
 * 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 DECLARE_MUTEX 	(dbs_sem);
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, 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;

	down(&dbs_sem);
	dbs_tuners_ins.sampling_down_factor = input;
	up(&dbs_sem);

	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);

	down(&dbs_sem);
	if (ret != 1 || input > MAX_SAMPLING_RATE || input < MIN_SAMPLING_RATE) {
		up(&dbs_sem);
		return -EINVAL;
	}

	dbs_tuners_ins.sampling_rate = input;
	up(&dbs_sem);

	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);

	down(&dbs_sem);
	if (ret != 1 || input > MAX_FREQUENCY_UP_THRESHOLD || 
			input < MIN_FREQUENCY_UP_THRESHOLD) {
		up(&dbs_sem);
		return -EINVAL;
	}

	dbs_tuners_ins.up_threshold = input;
	up(&dbs_sem);

	return count;
}

static ssize_t store_ignore_nice(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;
	
	down(&dbs_sem);
	if ( input == dbs_tuners_ins.ignore_nice ) { /* nothing to do */
		up(&dbs_sem);
		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;
	}
	up(&dbs_sem);

	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);

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.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;
	down(&dbs_sem);
	for_each_online_cpu(i)
		dbs_check_cpu(i);
	schedule_delayed_work(&dbs_work, 
			usecs_to_jiffies(dbs_tuners_ins.sampling_rate));
	up(&dbs_sem);
} 

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))
			return -EINVAL;
		if (this_dbs_info->enable) /* Already enabled */
			break;
		 
		down(&dbs_sem);
		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();
		}
		
		up(&dbs_sem);
		break;

	case CPUFREQ_GOV_STOP:
		down(&dbs_sem);
		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();
		
		up(&dbs_sem);

		break;

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