[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>

/*
 * 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			(def_sampling_rate / 2)
#define MAX_SAMPLING_RATE			(500 * def_sampling_rate)
#define DEF_SAMPLING_RATE_LATENCY_MULTIPLIER	(1000)
#define DEF_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;

	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;
			if (latency < 1000)
				latency = 1000;

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