[net-next-2.6.git] / drivers / cpufreq / cpufreq_conservative.c

/*
 *  drivers/cpufreq/cpufreq_conservative.c
 *
 *  Copyright (C)  2001 Russell King
 *            (C)  2003 Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>.
 *                      Jun Nakajima <jun.nakajima@intel.com>
 *            (C)  2004 Alexander Clouter <alex-kernel@digriz.org.uk>
 *
 * 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 DEF_FREQUENCY_DOWN_THRESHOLD		(20)

/* 
 * 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;
	unsigned int		down_skip;
	unsigned int		requested_freq;
};
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		down_threshold;
	unsigned int		ignore_nice;
	unsigned int		freq_step;
};

static struct dbs_tuners dbs_tuners_ins = {
	.up_threshold 		= DEF_FREQUENCY_UP_THRESHOLD,
	.down_threshold 	= DEF_FREQUENCY_DOWN_THRESHOLD,
	.sampling_down_factor 	= DEF_SAMPLING_DOWN_FACTOR,
	.ignore_nice		= 0,
	.freq_step		= 5,
};

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_conservative 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(down_threshold, down_threshold);
show_one(ignore_nice_load, ignore_nice);
show_one(freq_step, freq_step);

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 || 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 > 100 || input <= dbs_tuners_ins.down_threshold) {
		mutex_unlock(&dbs_mutex);
		return -EINVAL;
	}

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

	return count;
}

static ssize_t store_down_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 > 100 || input >= dbs_tuners_ins.up_threshold) {
		mutex_unlock(&dbs_mutex);
		return -EINVAL;
	}

	dbs_tuners_ins.down_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;
}

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

	ret = sscanf (buf, "%u", &input);

	if ( ret != 1 )
		return -EINVAL;

	if ( input > 100 )
		input = 100;
	
	/* no need to test here if freq_step is zero as the user might actually
	 * want this, they would be crazy though :) */
	mutex_lock(&dbs_mutex);
	dbs_tuners_ins.freq_step = input;
	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(down_threshold);
define_one_rw(ignore_nice_load);
define_one_rw(freq_step);

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

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

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

static void dbs_check_cpu(int cpu)
{
	unsigned int idle_ticks, up_idle_ticks, down_idle_ticks;
	unsigned int tmp_idle_ticks, total_idle_ticks;
	unsigned int freq_step;
	unsigned int freq_down_sampling_rate;
	struct cpu_dbs_info_s *this_dbs_info = &per_cpu(cpu_dbs_info, cpu);
	struct cpufreq_policy *policy;

	if (!this_dbs_info->enable)
		return;

	policy = this_dbs_info->cur_policy;

	/* 
	 * The default safe range is 20% to 80% 
	 * Every sampling_rate, we check
	 * 	- If current idle time is less than 20%, then we try to 
	 * 	  increase frequency
	 * Every sampling_rate*sampling_down_factor, we check
	 * 	- If current idle time is more than 80%, then we try to
	 * 	  decrease frequency
	 *
	 * Any frequency increase takes it to the maximum frequency. 
	 * Frequency reduction happens at minimum steps of 
	 * 5% (default) of max_frequency 
	 */

	/* Check for frequency increase */
	idle_ticks = UINT_MAX;

	/* Check for frequency increase */
	total_idle_ticks = get_cpu_idle_time(cpu);
	tmp_idle_ticks = total_idle_ticks -
		this_dbs_info->prev_cpu_idle_up;
	this_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) {
		this_dbs_info->down_skip = 0;
		this_dbs_info->prev_cpu_idle_down =
			this_dbs_info->prev_cpu_idle_up;

		/* if we are already at full speed then break out early */
		if (this_dbs_info->requested_freq == policy->max)
			return;
		
		freq_step = (dbs_tuners_ins.freq_step * policy->max) / 100;

		/* max freq cannot be less than 100. But who knows.... */
		if (unlikely(freq_step == 0))
			freq_step = 5;
		
		this_dbs_info->requested_freq += freq_step;
		if (this_dbs_info->requested_freq > policy->max)
			this_dbs_info->requested_freq = policy->max;

		__cpufreq_driver_target(policy, this_dbs_info->requested_freq,
			CPUFREQ_RELATION_H);
		return;
	}

	/* Check for frequency decrease */
	this_dbs_info->down_skip++;
	if (this_dbs_info->down_skip < dbs_tuners_ins.sampling_down_factor)
		return;

	/* Check for frequency decrease */
	total_idle_ticks = this_dbs_info->prev_cpu_idle_up;
	tmp_idle_ticks = total_idle_ticks -
		this_dbs_info->prev_cpu_idle_down;
	this_dbs_info->prev_cpu_idle_down = 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;
	this_dbs_info->down_skip = 0;

	freq_down_sampling_rate = dbs_tuners_ins.sampling_rate *
		dbs_tuners_ins.sampling_down_factor;
	down_idle_ticks = (100 - dbs_tuners_ins.down_threshold) *
		usecs_to_jiffies(freq_down_sampling_rate);

	if (idle_ticks > down_idle_ticks) {
		/*
		 * if we are already at the lowest speed then break out early
		 * or if we 'cannot' reduce the speed as the user might want
		 * freq_step to be zero
		 */
		if (this_dbs_info->requested_freq == policy->min
				|| dbs_tuners_ins.freq_step == 0)
			return;

		freq_step = (dbs_tuners_ins.freq_step * policy->max) / 100;

		/* max freq cannot be less than 100. But who knows.... */
		if (unlikely(freq_step == 0))
			freq_step = 5;

		this_dbs_info->requested_freq -= freq_step;
		if (this_dbs_info->requested_freq < policy->min)
			this_dbs_info->requested_freq = policy->min;

		__cpufreq_driver_target(policy, this_dbs_info->requested_freq,
				CPUFREQ_RELATION_H);
		return;
	}
}

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))
			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(cpu);
			j_dbs_info->prev_cpu_idle_down
				= j_dbs_info->prev_cpu_idle_up;
		}
		this_dbs_info->enable = 1;
		this_dbs_info->down_skip = 0;
		this_dbs_info->requested_freq = policy->cur;
		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 = 10 * 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		= "conservative",
	.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 ("Alexander Clouter <alex-kernel@digriz.org.uk>");
MODULE_DESCRIPTION ("'cpufreq_conservative' - A dynamic cpufreq governor for "
		"Low Latency Frequency Transition capable processors "
		"optimised for use in a battery environment");
MODULE_LICENSE ("GPL");

module_init(cpufreq_gov_dbs_init);
module_exit(cpufreq_gov_dbs_exit);
Commit	Line	Data
b9170836 DJ	1	/*
	2	* drivers/cpufreq/cpufreq_conservative.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	* (C) 2004 Alexander Clouter <alex-kernel@digriz.org.uk>
	8	*
	9	* This program is free software; you can redistribute it and/or modify
	10	* it under the terms of the GNU General Public License version 2 as
	11	* published by the Free Software Foundation.
	12	*/
	13
	14	#include <linux/kernel.h>
	15	#include <linux/module.h>
	16	#include <linux/smp.h>
	17	#include <linux/init.h>
	18	#include <linux/interrupt.h>
	19	#include <linux/ctype.h>
	20	#include <linux/cpufreq.h>
	21	#include <linux/sysctl.h>
	22	#include <linux/types.h>
	23	#include <linux/fs.h>
	24	#include <linux/sysfs.h>
	25	#include <linux/sched.h>
	26	#include <linux/kmod.h>
	27	#include <linux/workqueue.h>
	28	#include <linux/jiffies.h>
	29	#include <linux/kernel_stat.h>
	30	#include <linux/percpu.h>
3fc54d37	31	#include <linux/mutex.h>
b9170836 DJ	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)
b9170836	38	#define DEF_FREQUENCY_DOWN_THRESHOLD (20)
b9170836 DJ	39
	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;
2c906b31 AC	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)
b9170836	55	#define MAX_SAMPLING_RATE (500 * def_sampling_rate)
2c906b31 AC	56	#define DEF_SAMPLING_RATE_LATENCY_MULTIPLIER (1000)
	57	#define DEF_SAMPLING_DOWN_FACTOR (1)
	58	#define MAX_SAMPLING_DOWN_FACTOR (10)
b9170836 DJ	59	#define TRANSITION_LATENCY_LIMIT (10 * 1000)
	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;
a159b827 AC	68	unsigned int down_skip;
a159b827 AC	69	unsigned int requested_freq;
b9170836 DJ	70	};
	71	static DEFINE_PER_CPU(struct cpu_dbs_info_s, cpu_dbs_info);
	72
	73	static unsigned int dbs_enable; /* number of CPUs using this policy */
	74
3fc54d37	75	static DEFINE_MUTEX (dbs_mutex);
b9170836 DJ	76	static DECLARE_WORK (dbs_work, do_dbs_timer, NULL);
	77
	78	struct dbs_tuners {
	79	unsigned int sampling_rate;
	80	unsigned int sampling_down_factor;
	81	unsigned int up_threshold;
	82	unsigned int down_threshold;
	83	unsigned int ignore_nice;
	84	unsigned int freq_step;
	85	};
	86
	87	static struct dbs_tuners dbs_tuners_ins = {
	88	.up_threshold = DEF_FREQUENCY_UP_THRESHOLD,
	89	.down_threshold = DEF_FREQUENCY_DOWN_THRESHOLD,
	90	.sampling_down_factor = DEF_SAMPLING_DOWN_FACTOR,
c326e27e MD	91	.ignore_nice = 0,
c326e27e MD	92	.freq_step = 5,
b9170836 DJ	93	};
b9170836 DJ	94
dac1c1a5 DJ	95	static inline unsigned int get_cpu_idle_time(unsigned int cpu)
	96	{
	97	return kstat_cpu(cpu).cpustat.idle +
	98	kstat_cpu(cpu).cpustat.iowait +
001893cd	99	( dbs_tuners_ins.ignore_nice ?
dac1c1a5 DJ	100	kstat_cpu(cpu).cpustat.nice :
	101	0);
	102	}
	103
b9170836 DJ	104	/************************ sysfs interface **********************/
	105	static ssize_t show_sampling_rate_max(struct cpufreq_policy policy, char buf)
	106	{
	107	return sprintf (buf, "%u\n", MAX_SAMPLING_RATE);
	108	}
	109
	110	static ssize_t show_sampling_rate_min(struct cpufreq_policy policy, char buf)
	111	{
	112	return sprintf (buf, "%u\n", MIN_SAMPLING_RATE);
	113	}
	114
	115	#define define_one_ro(_name) \
	116	static struct freq_attr _name = \
	117	__ATTR(_name, 0444, show_##_name, NULL)
	118
	119	define_one_ro(sampling_rate_max);
	120	define_one_ro(sampling_rate_min);
	121
	122	/* cpufreq_conservative Governor Tunables */
	123	#define show_one(file_name, object) \
	124	static ssize_t show_##file_name \
	125	(struct cpufreq_policy unused, char buf) \
	126	{ \
	127	return sprintf(buf, "%u\n", dbs_tuners_ins.object); \
	128	}
	129	show_one(sampling_rate, sampling_rate);
	130	show_one(sampling_down_factor, sampling_down_factor);
	131	show_one(up_threshold, up_threshold);
	132	show_one(down_threshold, down_threshold);
001893cd	133	show_one(ignore_nice_load, ignore_nice);
b9170836 DJ	134	show_one(freq_step, freq_step);
	135
	136	static ssize_t store_sampling_down_factor(struct cpufreq_policy *unused,
	137	const char *buf, size_t count)
	138	{
	139	unsigned int input;
	140	int ret;
	141	ret = sscanf (buf, "%u", &input);
2c906b31	142	if (ret != 1 \|\| input > MAX_SAMPLING_DOWN_FACTOR \|\| input < 1)
b9170836 DJ	143	return -EINVAL;
b9170836 DJ	144
3fc54d37	145	mutex_lock(&dbs_mutex);
b9170836	146	dbs_tuners_ins.sampling_down_factor = input;
3fc54d37	147	mutex_unlock(&dbs_mutex);
b9170836 DJ	148
	149	return count;
	150	}
	151
	152	static ssize_t store_sampling_rate(struct cpufreq_policy *unused,
	153	const char *buf, size_t count)
	154	{
	155	unsigned int input;
	156	int ret;
	157	ret = sscanf (buf, "%u", &input);
	158
3fc54d37	159	mutex_lock(&dbs_mutex);
b9170836	160	if (ret != 1 \|\| input > MAX_SAMPLING_RATE \|\| input < MIN_SAMPLING_RATE) {
3fc54d37	161	mutex_unlock(&dbs_mutex);
b9170836 DJ	162	return -EINVAL;
	163	}
	164
	165	dbs_tuners_ins.sampling_rate = input;
3fc54d37	166	mutex_unlock(&dbs_mutex);
b9170836 DJ	167
	168	return count;
	169	}
	170
	171	static ssize_t store_up_threshold(struct cpufreq_policy *unused,
	172	const char *buf, size_t count)
	173	{
	174	unsigned int input;
	175	int ret;
	176	ret = sscanf (buf, "%u", &input);
	177
3fc54d37	178	mutex_lock(&dbs_mutex);
b82fbe6c	179	if (ret != 1 \|\| input > 100 \|\| input <= dbs_tuners_ins.down_threshold) {
3fc54d37	180	mutex_unlock(&dbs_mutex);
b9170836 DJ	181	return -EINVAL;
	182	}
	183
	184	dbs_tuners_ins.up_threshold = input;
3fc54d37	185	mutex_unlock(&dbs_mutex);
b9170836 DJ	186
	187	return count;
	188	}
	189
	190	static ssize_t store_down_threshold(struct cpufreq_policy *unused,
	191	const char *buf, size_t count)
	192	{
	193	unsigned int input;
	194	int ret;
	195	ret = sscanf (buf, "%u", &input);
	196
3fc54d37	197	mutex_lock(&dbs_mutex);
b82fbe6c	198	if (ret != 1 \|\| input > 100 \|\| input >= dbs_tuners_ins.up_threshold) {
3fc54d37	199	mutex_unlock(&dbs_mutex);
b9170836 DJ	200	return -EINVAL;
	201	}
	202
	203	dbs_tuners_ins.down_threshold = input;
3fc54d37	204	mutex_unlock(&dbs_mutex);
b9170836 DJ	205
	206	return count;
	207	}
	208
001893cd	209	static ssize_t store_ignore_nice_load(struct cpufreq_policy *policy,
b9170836 DJ	210	const char *buf, size_t count)
	211	{
	212	unsigned int input;
	213	int ret;
	214
	215	unsigned int j;
	216
	217	ret = sscanf (buf, "%u", &input);
	218	if ( ret != 1 )
	219	return -EINVAL;
	220
	221	if ( input > 1 )
	222	input = 1;
	223
3fc54d37	224	mutex_lock(&dbs_mutex);
b9170836	225	if ( input == dbs_tuners_ins.ignore_nice ) { /* nothing to do */
3fc54d37	226	mutex_unlock(&dbs_mutex);
b9170836 DJ	227	return count;
	228	}
	229	dbs_tuners_ins.ignore_nice = input;
	230
	231	/* we need to re-evaluate prev_cpu_idle_up and prev_cpu_idle_down */
dac1c1a5	232	for_each_online_cpu(j) {
b9170836 DJ	233	struct cpu_dbs_info_s *j_dbs_info;
b9170836 DJ	234	j_dbs_info = &per_cpu(cpu_dbs_info, j);
dac1c1a5	235	j_dbs_info->prev_cpu_idle_up = get_cpu_idle_time(j);
b9170836 DJ	236	j_dbs_info->prev_cpu_idle_down = j_dbs_info->prev_cpu_idle_up;
b9170836 DJ	237	}
3fc54d37	238	mutex_unlock(&dbs_mutex);
b9170836 DJ	239
	240	return count;
	241	}
	242
	243	static ssize_t store_freq_step(struct cpufreq_policy *policy,
	244	const char *buf, size_t count)
	245	{
	246	unsigned int input;
	247	int ret;
	248
	249	ret = sscanf (buf, "%u", &input);
	250
	251	if ( ret != 1 )
	252	return -EINVAL;
	253
	254	if ( input > 100 )
	255	input = 100;
	256
	257	/* no need to test here if freq_step is zero as the user might actually
	258	* want this, they would be crazy though :) */
3fc54d37	259	mutex_lock(&dbs_mutex);
b9170836	260	dbs_tuners_ins.freq_step = input;
3fc54d37	261	mutex_unlock(&dbs_mutex);
b9170836 DJ	262
	263	return count;
	264	}
	265
	266	#define define_one_rw(_name) \
	267	static struct freq_attr _name = \
	268	__ATTR(_name, 0644, show_##_name, store_##_name)
	269
	270	define_one_rw(sampling_rate);
	271	define_one_rw(sampling_down_factor);
	272	define_one_rw(up_threshold);
	273	define_one_rw(down_threshold);
001893cd	274	define_one_rw(ignore_nice_load);
b9170836 DJ	275	define_one_rw(freq_step);
	276
	277	static struct attribute * dbs_attributes[] = {
	278	&sampling_rate_max.attr,
	279	&sampling_rate_min.attr,
	280	&sampling_rate.attr,
	281	&sampling_down_factor.attr,
	282	&up_threshold.attr,
	283	&down_threshold.attr,
001893cd	284	&ignore_nice_load.attr,
b9170836 DJ	285	&freq_step.attr,
	286	NULL
	287	};
	288
	289	static struct attribute_group dbs_attr_group = {
	290	.attrs = dbs_attributes,
	291	.name = "conservative",
	292	};
	293
	294	/************************ sysfs end **********************/
	295
	296	static void dbs_check_cpu(int cpu)
	297	{
	298	unsigned int idle_ticks, up_idle_ticks, down_idle_ticks;
08a28e2e	299	unsigned int tmp_idle_ticks, total_idle_ticks;
b9170836 DJ	300	unsigned int freq_step;
b9170836 DJ	301	unsigned int freq_down_sampling_rate;
08a28e2e	302	struct cpu_dbs_info_s *this_dbs_info = &per_cpu(cpu_dbs_info, cpu);
b9170836	303	struct cpufreq_policy *policy;
b9170836	304
b9170836 DJ	305	if (!this_dbs_info->enable)
	306	return;
	307
08a28e2e AC	308	policy = this_dbs_info->cur_policy;
08a28e2e AC	309
b9170836 DJ	310	/*
	311	* The default safe range is 20% to 80%
	312	* Every sampling_rate, we check
	313	* - If current idle time is less than 20%, then we try to
	314	* increase frequency
	315	* Every sampling_rate*sampling_down_factor, we check
	316	* - If current idle time is more than 80%, then we try to
	317	* decrease frequency
	318	*
	319	* Any frequency increase takes it to the maximum frequency.
	320	* Frequency reduction happens at minimum steps of
	321	* 5% (default) of max_frequency
	322	*/
	323
	324	/* Check for frequency increase */
9c7d269b	325	idle_ticks = UINT_MAX;
b9170836	326
08a28e2e AC	327	/* Check for frequency increase */
	328	total_idle_ticks = get_cpu_idle_time(cpu);
	329	tmp_idle_ticks = total_idle_ticks -
	330	this_dbs_info->prev_cpu_idle_up;
	331	this_dbs_info->prev_cpu_idle_up = total_idle_ticks;
	332
	333	if (tmp_idle_ticks < idle_ticks)
	334	idle_ticks = tmp_idle_ticks;
b9170836 DJ	335
	336	/* Scale idle ticks by 100 and compare with up and down ticks */
	337	idle_ticks *= 100;
	338	up_idle_ticks = (100 - dbs_tuners_ins.up_threshold) *
2c906b31	339	usecs_to_jiffies(dbs_tuners_ins.sampling_rate);
b9170836 DJ	340
b9170836 DJ	341	if (idle_ticks < up_idle_ticks) {
a159b827	342	this_dbs_info->down_skip = 0;
08a28e2e AC	343	this_dbs_info->prev_cpu_idle_down =
08a28e2e AC	344	this_dbs_info->prev_cpu_idle_up;
790d76fa	345
b9170836	346	/* if we are already at full speed then break out early */
a159b827	347	if (this_dbs_info->requested_freq == policy->max)
b9170836 DJ	348	return;
	349
	350	freq_step = (dbs_tuners_ins.freq_step * policy->max) / 100;
	351
	352	/* max freq cannot be less than 100. But who knows.... */
	353	if (unlikely(freq_step == 0))
	354	freq_step = 5;
	355
a159b827 AC	356	this_dbs_info->requested_freq += freq_step;
	357	if (this_dbs_info->requested_freq > policy->max)
	358	this_dbs_info->requested_freq = policy->max;
b9170836	359
a159b827	360	__cpufreq_driver_target(policy, this_dbs_info->requested_freq,
b9170836	361	CPUFREQ_RELATION_H);
b9170836 DJ	362	return;
	363	}
	364
	365	/* Check for frequency decrease */
a159b827 AC	366	this_dbs_info->down_skip++;
a159b827 AC	367	if (this_dbs_info->down_skip < dbs_tuners_ins.sampling_down_factor)
b9170836 DJ	368	return;
b9170836 DJ	369
08a28e2e AC	370	/* Check for frequency decrease */
	371	total_idle_ticks = this_dbs_info->prev_cpu_idle_up;
	372	tmp_idle_ticks = total_idle_ticks -
	373	this_dbs_info->prev_cpu_idle_down;
	374	this_dbs_info->prev_cpu_idle_down = total_idle_ticks;
b9170836	375
08a28e2e AC	376	if (tmp_idle_ticks < idle_ticks)
08a28e2e AC	377	idle_ticks = tmp_idle_ticks;
b9170836 DJ	378
	379	/* Scale idle ticks by 100 and compare with up and down ticks */
	380	idle_ticks *= 100;
a159b827	381	this_dbs_info->down_skip = 0;
b9170836 DJ	382
	383	freq_down_sampling_rate = dbs_tuners_ins.sampling_rate *
	384	dbs_tuners_ins.sampling_down_factor;
	385	down_idle_ticks = (100 - dbs_tuners_ins.down_threshold) *
2c906b31	386	usecs_to_jiffies(freq_down_sampling_rate);
b9170836	387
9c7d269b	388	if (idle_ticks > down_idle_ticks) {
2c906b31 AC	389	/*
2c906b31 AC	390	* if we are already at the lowest speed then break out early
b9170836	391	* or if we 'cannot' reduce the speed as the user might want
2c906b31 AC	392	* freq_step to be zero
2c906b31 AC	393	*/
a159b827	394	if (this_dbs_info->requested_freq == policy->min
b9170836 DJ	395	\|\| dbs_tuners_ins.freq_step == 0)
	396	return;
	397
	398	freq_step = (dbs_tuners_ins.freq_step * policy->max) / 100;
	399
	400	/* max freq cannot be less than 100. But who knows.... */
	401	if (unlikely(freq_step == 0))
	402	freq_step = 5;
	403
a159b827 AC	404	this_dbs_info->requested_freq -= freq_step;
	405	if (this_dbs_info->requested_freq < policy->min)
	406	this_dbs_info->requested_freq = policy->min;
b9170836	407
a159b827	408	__cpufreq_driver_target(policy, this_dbs_info->requested_freq,
2c906b31	409	CPUFREQ_RELATION_H);
b9170836 DJ	410	return;
	411	}
	412	}
	413
	414	static void do_dbs_timer(void *data)
	415	{
	416	int i;
3fc54d37	417	mutex_lock(&dbs_mutex);
b9170836 DJ	418	for_each_online_cpu(i)
	419	dbs_check_cpu(i);
	420	schedule_delayed_work(&dbs_work,
	421	usecs_to_jiffies(dbs_tuners_ins.sampling_rate));
3fc54d37	422	mutex_unlock(&dbs_mutex);
b9170836 DJ	423	}
	424
	425	static inline void dbs_timer_init(void)
	426	{
	427	INIT_WORK(&dbs_work, do_dbs_timer, NULL);
	428	schedule_delayed_work(&dbs_work,
	429	usecs_to_jiffies(dbs_tuners_ins.sampling_rate));
	430	return;
	431	}
	432
	433	static inline void dbs_timer_exit(void)
	434	{
	435	cancel_delayed_work(&dbs_work);
	436	return;
	437	}
	438
	439	static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
	440	unsigned int event)
	441	{
	442	unsigned int cpu = policy->cpu;
	443	struct cpu_dbs_info_s *this_dbs_info;
	444	unsigned int j;
	445
	446	this_dbs_info = &per_cpu(cpu_dbs_info, cpu);
	447
	448	switch (event) {
	449	case CPUFREQ_GOV_START:
	450	if ((!cpu_online(cpu)) \|\|
	451	(!policy->cur))
	452	return -EINVAL;
	453
	454	if (policy->cpuinfo.transition_latency >
	455	(TRANSITION_LATENCY_LIMIT * 1000))
	456	return -EINVAL;
	457	if (this_dbs_info->enable) /* Already enabled */
	458	break;
	459
3fc54d37	460	mutex_lock(&dbs_mutex);
b9170836 DJ	461	for_each_cpu_mask(j, policy->cpus) {
	462	struct cpu_dbs_info_s *j_dbs_info;
	463	j_dbs_info = &per_cpu(cpu_dbs_info, j);
	464	j_dbs_info->cur_policy = policy;
	465
08a28e2e	466	j_dbs_info->prev_cpu_idle_up = get_cpu_idle_time(cpu);
b9170836 DJ	467	j_dbs_info->prev_cpu_idle_down
	468	= j_dbs_info->prev_cpu_idle_up;
	469	}
	470	this_dbs_info->enable = 1;
a159b827 AC	471	this_dbs_info->down_skip = 0;
a159b827 AC	472	this_dbs_info->requested_freq = policy->cur;
b9170836 DJ	473	sysfs_create_group(&policy->kobj, &dbs_attr_group);
	474	dbs_enable++;
	475	/*
	476	* Start the timerschedule work, when this governor
	477	* is used for first time
	478	*/
	479	if (dbs_enable == 1) {
	480	unsigned int latency;
	481	/* policy latency is in nS. Convert it to uS first */
2c906b31 AC	482	latency = policy->cpuinfo.transition_latency / 1000;
	483	if (latency == 0)
	484	latency = 1;
b9170836	485
e8a02572	486	def_sampling_rate = 10 * latency *
b9170836	487	DEF_SAMPLING_RATE_LATENCY_MULTIPLIER;
2c906b31 AC	488
	489	if (def_sampling_rate < MIN_STAT_SAMPLING_RATE)
	490	def_sampling_rate = MIN_STAT_SAMPLING_RATE;
	491
b9170836	492	dbs_tuners_ins.sampling_rate = def_sampling_rate;
b9170836 DJ	493
	494	dbs_timer_init();
	495	}
	496
3fc54d37	497	mutex_unlock(&dbs_mutex);
b9170836 DJ	498	break;
	499
	500	case CPUFREQ_GOV_STOP:
3fc54d37	501	mutex_lock(&dbs_mutex);
b9170836 DJ	502	this_dbs_info->enable = 0;
	503	sysfs_remove_group(&policy->kobj, &dbs_attr_group);
	504	dbs_enable--;
	505	/*
	506	* Stop the timerschedule work, when this governor
	507	* is used for first time
	508	*/
	509	if (dbs_enable == 0)
	510	dbs_timer_exit();
	511
3fc54d37	512	mutex_unlock(&dbs_mutex);
b9170836 DJ	513
	514	break;
	515
	516	case CPUFREQ_GOV_LIMITS:
3fc54d37	517	mutex_lock(&dbs_mutex);
b9170836 DJ	518	if (policy->max < this_dbs_info->cur_policy->cur)
	519	__cpufreq_driver_target(
	520	this_dbs_info->cur_policy,
	521	policy->max, CPUFREQ_RELATION_H);
	522	else if (policy->min > this_dbs_info->cur_policy->cur)
	523	__cpufreq_driver_target(
	524	this_dbs_info->cur_policy,
	525	policy->min, CPUFREQ_RELATION_L);
3fc54d37	526	mutex_unlock(&dbs_mutex);
b9170836 DJ	527	break;
	528	}
	529	return 0;
	530	}
	531
	532	static struct cpufreq_governor cpufreq_gov_dbs = {
	533	.name = "conservative",
	534	.governor = cpufreq_governor_dbs,
	535	.owner = THIS_MODULE,
	536	};
	537
	538	static int __init cpufreq_gov_dbs_init(void)
	539	{
	540	return cpufreq_register_governor(&cpufreq_gov_dbs);
	541	}
	542
	543	static void __exit cpufreq_gov_dbs_exit(void)
	544	{
	545	/* Make sure that the scheduled work is indeed not running */
	546	flush_scheduled_work();
	547
	548	cpufreq_unregister_governor(&cpufreq_gov_dbs);
	549	}
	550
	551
	552	MODULE_AUTHOR ("Alexander Clouter <alex-kernel@digriz.org.uk>");
	553	MODULE_DESCRIPTION ("'cpufreq_conservative' - A dynamic cpufreq governor for "
	554	"Low Latency Frequency Transition capable processors "
	555	"optimised for use in a battery environment");
	556	MODULE_LICENSE ("GPL");
	557
	558	module_init(cpufreq_gov_dbs_init);
	559	module_exit(cpufreq_gov_dbs_exit);