diff options
Diffstat (limited to 'drivers/cpufreq/cpufreq_conservative.c')
| -rw-r--r-- | drivers/cpufreq/cpufreq_conservative.c | 798 |
1 files changed, 221 insertions, 577 deletions
diff --git a/drivers/cpufreq/cpufreq_conservative.c b/drivers/cpufreq/cpufreq_conservative.c index 599a40b25cb..25a70d06c5b 100644 --- a/drivers/cpufreq/cpufreq_conservative.c +++ b/drivers/cpufreq/cpufreq_conservative.c @@ -11,229 +11,150 @@ * published by the Free Software Foundation. */ -#include <linux/kernel.h> -#include <linux/module.h> -#include <linux/init.h> -#include <linux/cpufreq.h> -#include <linux/cpu.h> -#include <linux/jiffies.h> -#include <linux/kernel_stat.h> -#include <linux/mutex.h> -#include <linux/hrtimer.h> -#include <linux/tick.h> -#include <linux/ktime.h> -#include <linux/sched.h> - -/* - * dbs is used in this file as a shortform for demandbased switching - * It helps to keep variable names smaller, simpler - */ +#include <linux/slab.h> +#include "cpufreq_governor.h" +/* Conservative governor macros */ #define DEF_FREQUENCY_UP_THRESHOLD (80) #define DEF_FREQUENCY_DOWN_THRESHOLD (20) +#define DEF_FREQUENCY_STEP (5) +#define DEF_SAMPLING_DOWN_FACTOR (1) +#define MAX_SAMPLING_DOWN_FACTOR (10) + +static DEFINE_PER_CPU(struct cs_cpu_dbs_info_s, cs_cpu_dbs_info); + +static inline unsigned int get_freq_target(struct cs_dbs_tuners *cs_tuners, + struct cpufreq_policy *policy) +{ + unsigned int freq_target = (cs_tuners->freq_step * policy->max) / 100; + + /* max freq cannot be less than 100. But who knows... */ + if (unlikely(freq_target == 0)) + freq_target = DEF_FREQUENCY_STEP; + + return freq_target; +} /* - * 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. + * 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 check, if current idle time is more than 80% + * (default), 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 maximum frequency */ -#define MIN_SAMPLING_RATE_RATIO (2) - -static unsigned int min_sampling_rate; +static void cs_check_cpu(int cpu, unsigned int load) +{ + struct cs_cpu_dbs_info_s *dbs_info = &per_cpu(cs_cpu_dbs_info, cpu); + struct cpufreq_policy *policy = dbs_info->cdbs.cur_policy; + struct dbs_data *dbs_data = policy->governor_data; + struct cs_dbs_tuners *cs_tuners = dbs_data->tuners; -#define LATENCY_MULTIPLIER (1000) -#define MIN_LATENCY_MULTIPLIER (100) -#define DEF_SAMPLING_DOWN_FACTOR (1) -#define MAX_SAMPLING_DOWN_FACTOR (10) -#define TRANSITION_LATENCY_LIMIT (10 * 1000 * 1000) - -static void do_dbs_timer(struct work_struct *work); - -struct cpu_dbs_info_s { - cputime64_t prev_cpu_idle; - cputime64_t prev_cpu_wall; - cputime64_t prev_cpu_nice; - struct cpufreq_policy *cur_policy; - struct delayed_work work; - unsigned int down_skip; - unsigned int requested_freq; - int cpu; - unsigned int enable:1; /* - * percpu mutex that serializes governor limit change with - * do_dbs_timer invocation. We do not want do_dbs_timer to run - * when user is changing the governor or limits. + * break out if we 'cannot' reduce the speed as the user might + * want freq_step to be zero */ - struct mutex timer_mutex; -}; -static DEFINE_PER_CPU(struct cpu_dbs_info_s, cs_cpu_dbs_info); + if (cs_tuners->freq_step == 0) + return; -static unsigned int dbs_enable; /* number of CPUs using this policy */ + /* Check for frequency increase */ + if (load > cs_tuners->up_threshold) { + dbs_info->down_skip = 0; -/* - * dbs_mutex protects data in dbs_tuners_ins from concurrent changes on - * different CPUs. It protects dbs_enable in governor start/stop. - */ -static DEFINE_MUTEX(dbs_mutex); - -static struct workqueue_struct *kconservative_wq; - -static 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; -} 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, -}; + /* if we are already at full speed then break out early */ + if (dbs_info->requested_freq == policy->max) + return; -static inline cputime64_t get_cpu_idle_time_jiffy(unsigned int cpu, - cputime64_t *wall) -{ - cputime64_t idle_time; - cputime64_t cur_wall_time; - cputime64_t busy_time; + dbs_info->requested_freq += get_freq_target(cs_tuners, policy); - cur_wall_time = jiffies64_to_cputime64(get_jiffies_64()); - busy_time = cputime64_add(kstat_cpu(cpu).cpustat.user, - kstat_cpu(cpu).cpustat.system); + if (dbs_info->requested_freq > policy->max) + dbs_info->requested_freq = policy->max; - busy_time = cputime64_add(busy_time, kstat_cpu(cpu).cpustat.irq); - busy_time = cputime64_add(busy_time, kstat_cpu(cpu).cpustat.softirq); - busy_time = cputime64_add(busy_time, kstat_cpu(cpu).cpustat.steal); - busy_time = cputime64_add(busy_time, kstat_cpu(cpu).cpustat.nice); + __cpufreq_driver_target(policy, dbs_info->requested_freq, + CPUFREQ_RELATION_H); + return; + } - idle_time = cputime64_sub(cur_wall_time, busy_time); - if (wall) - *wall = (cputime64_t)jiffies_to_usecs(cur_wall_time); + /* if sampling_down_factor is active break out early */ + if (++dbs_info->down_skip < cs_tuners->sampling_down_factor) + return; + dbs_info->down_skip = 0; - return (cputime64_t)jiffies_to_usecs(idle_time);; -} + /* Check for frequency decrease */ + if (load < cs_tuners->down_threshold) { + unsigned int freq_target; + /* + * if we cannot reduce the frequency anymore, break out early + */ + if (policy->cur == policy->min) + return; -static inline cputime64_t get_cpu_idle_time(unsigned int cpu, cputime64_t *wall) -{ - u64 idle_time = get_cpu_idle_time_us(cpu, wall); + freq_target = get_freq_target(cs_tuners, policy); + if (dbs_info->requested_freq > freq_target) + dbs_info->requested_freq -= freq_target; + else + dbs_info->requested_freq = policy->min; - if (idle_time == -1ULL) - return get_cpu_idle_time_jiffy(cpu, wall); + __cpufreq_driver_target(policy, dbs_info->requested_freq, + CPUFREQ_RELATION_L); + return; + } +} - return idle_time; +static void cs_dbs_timer(struct work_struct *work) +{ + struct cs_cpu_dbs_info_s *dbs_info = container_of(work, + struct cs_cpu_dbs_info_s, cdbs.work.work); + unsigned int cpu = dbs_info->cdbs.cur_policy->cpu; + struct cs_cpu_dbs_info_s *core_dbs_info = &per_cpu(cs_cpu_dbs_info, + cpu); + struct dbs_data *dbs_data = dbs_info->cdbs.cur_policy->governor_data; + struct cs_dbs_tuners *cs_tuners = dbs_data->tuners; + int delay = delay_for_sampling_rate(cs_tuners->sampling_rate); + bool modify_all = true; + + mutex_lock(&core_dbs_info->cdbs.timer_mutex); + if (!need_load_eval(&core_dbs_info->cdbs, cs_tuners->sampling_rate)) + modify_all = false; + else + dbs_check_cpu(dbs_data, cpu); + + gov_queue_work(dbs_data, dbs_info->cdbs.cur_policy, delay, modify_all); + mutex_unlock(&core_dbs_info->cdbs.timer_mutex); } -/* keep track of frequency transitions */ -static int -dbs_cpufreq_notifier(struct notifier_block *nb, unsigned long val, - void *data) +static int dbs_cpufreq_notifier(struct notifier_block *nb, unsigned long val, + void *data) { struct cpufreq_freqs *freq = data; - struct cpu_dbs_info_s *this_dbs_info = &per_cpu(cs_cpu_dbs_info, - freq->cpu); - + struct cs_cpu_dbs_info_s *dbs_info = + &per_cpu(cs_cpu_dbs_info, freq->cpu); struct cpufreq_policy *policy; - if (!this_dbs_info->enable) + if (!dbs_info->enable) return 0; - policy = this_dbs_info->cur_policy; + policy = dbs_info->cdbs.cur_policy; /* - * we only care if our internally tracked freq moves outside - * the 'valid' ranges of freqency available to us otherwise - * we do not change it + * we only care if our internally tracked freq moves outside the 'valid' + * ranges of frequency available to us otherwise we do not change it */ - if (this_dbs_info->requested_freq > policy->max - || this_dbs_info->requested_freq < policy->min) - this_dbs_info->requested_freq = freq->new; + if (dbs_info->requested_freq > policy->max + || dbs_info->requested_freq < policy->min) + dbs_info->requested_freq = freq->new; return 0; } -static struct notifier_block dbs_cpufreq_notifier_block = { - .notifier_call = dbs_cpufreq_notifier -}; - /************************** sysfs interface ************************/ -static ssize_t show_sampling_rate_max(struct kobject *kobj, - struct attribute *attr, char *buf) -{ - printk_once(KERN_INFO "CPUFREQ: conservative sampling_rate_max " - "sysfs file is deprecated - used by: %s\n", current->comm); - return sprintf(buf, "%u\n", -1U); -} +static struct common_dbs_data cs_dbs_cdata; -static ssize_t show_sampling_rate_min(struct kobject *kobj, - struct attribute *attr, char *buf) -{ - return sprintf(buf, "%u\n", min_sampling_rate); -} - -#define define_one_ro(_name) \ -static struct global_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 kobject *kobj, struct attribute *attr, 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); - -/*** delete after deprecation time ***/ -#define DEPRECATION_MSG(file_name) \ - printk_once(KERN_INFO "CPUFREQ: Per core conservative sysfs " \ - "interface is deprecated - " #file_name "\n"); - -#define show_one_old(file_name) \ -static ssize_t show_##file_name##_old \ -(struct cpufreq_policy *unused, char *buf) \ -{ \ - printk_once(KERN_INFO "CPUFREQ: Per core conservative sysfs " \ - "interface is deprecated - " #file_name "\n"); \ - return show_##file_name(NULL, NULL, buf); \ -} -show_one_old(sampling_rate); -show_one_old(sampling_down_factor); -show_one_old(up_threshold); -show_one_old(down_threshold); -show_one_old(ignore_nice_load); -show_one_old(freq_step); -show_one_old(sampling_rate_min); -show_one_old(sampling_rate_max); - -#define define_one_ro_old(object, _name) \ -static struct freq_attr object = \ -__ATTR(_name, 0444, show_##_name##_old, NULL) - -define_one_ro_old(sampling_rate_min_old, sampling_rate_min); -define_one_ro_old(sampling_rate_max_old, sampling_rate_max); - -/*** delete after deprecation time ***/ - -static ssize_t store_sampling_down_factor(struct kobject *a, - struct attribute *b, - const char *buf, size_t count) +static ssize_t store_sampling_down_factor(struct dbs_data *dbs_data, + const char *buf, size_t count) { + struct cs_dbs_tuners *cs_tuners = dbs_data->tuners; unsigned int input; int ret; ret = sscanf(buf, "%u", &input); @@ -241,16 +162,14 @@ static ssize_t store_sampling_down_factor(struct kobject *a, 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); - + cs_tuners->sampling_down_factor = input; return count; } -static ssize_t store_sampling_rate(struct kobject *a, struct attribute *b, - const char *buf, size_t count) +static ssize_t store_sampling_rate(struct dbs_data *dbs_data, const char *buf, + size_t count) { + struct cs_dbs_tuners *cs_tuners = dbs_data->tuners; unsigned int input; int ret; ret = sscanf(buf, "%u", &input); @@ -258,62 +177,49 @@ static ssize_t store_sampling_rate(struct kobject *a, struct attribute *b, if (ret != 1) return -EINVAL; - mutex_lock(&dbs_mutex); - dbs_tuners_ins.sampling_rate = max(input, min_sampling_rate); - mutex_unlock(&dbs_mutex); - + cs_tuners->sampling_rate = max(input, dbs_data->min_sampling_rate); return count; } -static ssize_t store_up_threshold(struct kobject *a, struct attribute *b, - const char *buf, size_t count) +static ssize_t store_up_threshold(struct dbs_data *dbs_data, const char *buf, + size_t count) { + struct cs_dbs_tuners *cs_tuners = dbs_data->tuners; 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); + if (ret != 1 || input > 100 || input <= cs_tuners->down_threshold) return -EINVAL; - } - - dbs_tuners_ins.up_threshold = input; - mutex_unlock(&dbs_mutex); + cs_tuners->up_threshold = input; return count; } -static ssize_t store_down_threshold(struct kobject *a, struct attribute *b, - const char *buf, size_t count) +static ssize_t store_down_threshold(struct dbs_data *dbs_data, const char *buf, + size_t count) { + struct cs_dbs_tuners *cs_tuners = dbs_data->tuners; unsigned int input; int ret; ret = sscanf(buf, "%u", &input); - mutex_lock(&dbs_mutex); /* cannot be lower than 11 otherwise freq will not fall */ if (ret != 1 || input < 11 || input > 100 || - input >= dbs_tuners_ins.up_threshold) { - mutex_unlock(&dbs_mutex); + input >= cs_tuners->up_threshold) return -EINVAL; - } - - dbs_tuners_ins.down_threshold = input; - mutex_unlock(&dbs_mutex); + cs_tuners->down_threshold = input; return count; } -static ssize_t store_ignore_nice_load(struct kobject *a, struct attribute *b, - const char *buf, size_t count) +static ssize_t store_ignore_nice_load(struct dbs_data *dbs_data, + const char *buf, size_t count) { - unsigned int input; + struct cs_dbs_tuners *cs_tuners = dbs_data->tuners; + unsigned int input, j; int ret; - unsigned int j; - ret = sscanf(buf, "%u", &input); if (ret != 1) return -EINVAL; @@ -321,30 +227,28 @@ static ssize_t store_ignore_nice_load(struct kobject *a, struct attribute *b, if (input > 1) input = 1; - mutex_lock(&dbs_mutex); - if (input == dbs_tuners_ins.ignore_nice) { /* nothing to do */ - mutex_unlock(&dbs_mutex); + if (input == cs_tuners->ignore_nice_load) /* nothing to do */ return count; - } - dbs_tuners_ins.ignore_nice = input; + + cs_tuners->ignore_nice_load = input; /* we need to re-evaluate prev_cpu_idle */ for_each_online_cpu(j) { - struct cpu_dbs_info_s *dbs_info; + struct cs_cpu_dbs_info_s *dbs_info; dbs_info = &per_cpu(cs_cpu_dbs_info, j); - dbs_info->prev_cpu_idle = get_cpu_idle_time(j, - &dbs_info->prev_cpu_wall); - if (dbs_tuners_ins.ignore_nice) - dbs_info->prev_cpu_nice = kstat_cpu(j).cpustat.nice; + dbs_info->cdbs.prev_cpu_idle = get_cpu_idle_time(j, + &dbs_info->cdbs.prev_cpu_wall, 0); + if (cs_tuners->ignore_nice_load) + dbs_info->cdbs.prev_cpu_nice = + kcpustat_cpu(j).cpustat[CPUTIME_NICE]; } - mutex_unlock(&dbs_mutex); - return count; } -static ssize_t store_freq_step(struct kobject *a, struct attribute *b, - const char *buf, size_t count) +static ssize_t store_freq_step(struct dbs_data *dbs_data, const char *buf, + size_t count) { + struct cs_dbs_tuners *cs_tuners = dbs_data->tuners; unsigned int input; int ret; ret = sscanf(buf, "%u", &input); @@ -355,365 +259,119 @@ static ssize_t store_freq_step(struct kobject *a, struct attribute *b, 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); - + /* + * no need to test here if freq_step is zero as the user might actually + * want this, they would be crazy though :) + */ + cs_tuners->freq_step = input; return count; } -#define define_one_rw(_name) \ -static struct global_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, +show_store_one(cs, sampling_rate); +show_store_one(cs, sampling_down_factor); +show_store_one(cs, up_threshold); +show_store_one(cs, down_threshold); +show_store_one(cs, ignore_nice_load); +show_store_one(cs, freq_step); +declare_show_sampling_rate_min(cs); + +gov_sys_pol_attr_rw(sampling_rate); +gov_sys_pol_attr_rw(sampling_down_factor); +gov_sys_pol_attr_rw(up_threshold); +gov_sys_pol_attr_rw(down_threshold); +gov_sys_pol_attr_rw(ignore_nice_load); +gov_sys_pol_attr_rw(freq_step); +gov_sys_pol_attr_ro(sampling_rate_min); + +static struct attribute *dbs_attributes_gov_sys[] = { + &sampling_rate_min_gov_sys.attr, + &sampling_rate_gov_sys.attr, + &sampling_down_factor_gov_sys.attr, + &up_threshold_gov_sys.attr, + &down_threshold_gov_sys.attr, + &ignore_nice_load_gov_sys.attr, + &freq_step_gov_sys.attr, NULL }; -static struct attribute_group dbs_attr_group = { - .attrs = dbs_attributes, +static struct attribute_group cs_attr_group_gov_sys = { + .attrs = dbs_attributes_gov_sys, .name = "conservative", }; -/*** delete after deprecation time ***/ - -#define write_one_old(file_name) \ -static ssize_t store_##file_name##_old \ -(struct cpufreq_policy *unused, const char *buf, size_t count) \ -{ \ - printk_once(KERN_INFO "CPUFREQ: Per core conservative sysfs " \ - "interface is deprecated - " #file_name "\n"); \ - return store_##file_name(NULL, NULL, buf, count); \ -} -write_one_old(sampling_rate); -write_one_old(sampling_down_factor); -write_one_old(up_threshold); -write_one_old(down_threshold); -write_one_old(ignore_nice_load); -write_one_old(freq_step); - -#define define_one_rw_old(object, _name) \ -static struct freq_attr object = \ -__ATTR(_name, 0644, show_##_name##_old, store_##_name##_old) - -define_one_rw_old(sampling_rate_old, sampling_rate); -define_one_rw_old(sampling_down_factor_old, sampling_down_factor); -define_one_rw_old(up_threshold_old, up_threshold); -define_one_rw_old(down_threshold_old, down_threshold); -define_one_rw_old(ignore_nice_load_old, ignore_nice_load); -define_one_rw_old(freq_step_old, freq_step); - -static struct attribute *dbs_attributes_old[] = { - &sampling_rate_max_old.attr, - &sampling_rate_min_old.attr, - &sampling_rate_old.attr, - &sampling_down_factor_old.attr, - &up_threshold_old.attr, - &down_threshold_old.attr, - &ignore_nice_load_old.attr, - &freq_step_old.attr, +static struct attribute *dbs_attributes_gov_pol[] = { + &sampling_rate_min_gov_pol.attr, + &sampling_rate_gov_pol.attr, + &sampling_down_factor_gov_pol.attr, + &up_threshold_gov_pol.attr, + &down_threshold_gov_pol.attr, + &ignore_nice_load_gov_pol.attr, + &freq_step_gov_pol.attr, NULL }; -static struct attribute_group dbs_attr_group_old = { - .attrs = dbs_attributes_old, +static struct attribute_group cs_attr_group_gov_pol = { + .attrs = dbs_attributes_gov_pol, .name = "conservative", }; -/*** delete after deprecation time ***/ - /************************** sysfs end ************************/ -static void dbs_check_cpu(struct cpu_dbs_info_s *this_dbs_info) +static int cs_init(struct dbs_data *dbs_data) { - unsigned int load = 0; - unsigned int freq_target; - - struct cpufreq_policy *policy; - unsigned int j; - - 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 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 maximum frequency - */ - - /* Get Absolute Load */ - for_each_cpu(j, policy->cpus) { - struct cpu_dbs_info_s *j_dbs_info; - cputime64_t cur_wall_time, cur_idle_time; - unsigned int idle_time, wall_time; - - j_dbs_info = &per_cpu(cs_cpu_dbs_info, j); + struct cs_dbs_tuners *tuners; - cur_idle_time = get_cpu_idle_time(j, &cur_wall_time); - - wall_time = (unsigned int) cputime64_sub(cur_wall_time, - j_dbs_info->prev_cpu_wall); - j_dbs_info->prev_cpu_wall = cur_wall_time; - - idle_time = (unsigned int) cputime64_sub(cur_idle_time, - j_dbs_info->prev_cpu_idle); - j_dbs_info->prev_cpu_idle = cur_idle_time; - - if (dbs_tuners_ins.ignore_nice) { - cputime64_t cur_nice; - unsigned long cur_nice_jiffies; - - cur_nice = cputime64_sub(kstat_cpu(j).cpustat.nice, - j_dbs_info->prev_cpu_nice); - /* - * Assumption: nice time between sampling periods will - * be less than 2^32 jiffies for 32 bit sys - */ - cur_nice_jiffies = (unsigned long) - cputime64_to_jiffies64(cur_nice); - - j_dbs_info->prev_cpu_nice = kstat_cpu(j).cpustat.nice; - idle_time += jiffies_to_usecs(cur_nice_jiffies); - } - - if (unlikely(!wall_time || wall_time < idle_time)) - continue; - - load = 100 * (wall_time - idle_time) / wall_time; + tuners = kzalloc(sizeof(*tuners), GFP_KERNEL); + if (!tuners) { + pr_err("%s: kzalloc failed\n", __func__); + return -ENOMEM; } - /* - * break out if we 'cannot' reduce the speed as the user might - * want freq_step to be zero - */ - if (dbs_tuners_ins.freq_step == 0) - return; - - /* Check for frequency increase */ - if (load > dbs_tuners_ins.up_threshold) { - this_dbs_info->down_skip = 0; - - /* if we are already at full speed then break out early */ - if (this_dbs_info->requested_freq == policy->max) - return; - - freq_target = (dbs_tuners_ins.freq_step * policy->max) / 100; - - /* max freq cannot be less than 100. But who knows.... */ - if (unlikely(freq_target == 0)) - freq_target = 5; - - this_dbs_info->requested_freq += freq_target; - 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; - } - - /* - * 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. - */ - if (load < (dbs_tuners_ins.down_threshold - 10)) { - freq_target = (dbs_tuners_ins.freq_step * policy->max) / 100; - - this_dbs_info->requested_freq -= freq_target; - if (this_dbs_info->requested_freq < policy->min) - this_dbs_info->requested_freq = policy->min; + tuners->up_threshold = DEF_FREQUENCY_UP_THRESHOLD; + tuners->down_threshold = DEF_FREQUENCY_DOWN_THRESHOLD; + tuners->sampling_down_factor = DEF_SAMPLING_DOWN_FACTOR; + tuners->ignore_nice_load = 0; + tuners->freq_step = DEF_FREQUENCY_STEP; - /* - * if we cannot reduce the frequency anymore, break out early - */ - if (policy->cur == policy->min) - return; - - __cpufreq_driver_target(policy, this_dbs_info->requested_freq, - CPUFREQ_RELATION_H); - return; - } + dbs_data->tuners = tuners; + dbs_data->min_sampling_rate = MIN_SAMPLING_RATE_RATIO * + jiffies_to_usecs(10); + mutex_init(&dbs_data->mutex); + return 0; } -static void do_dbs_timer(struct work_struct *work) +static void cs_exit(struct dbs_data *dbs_data) { - struct cpu_dbs_info_s *dbs_info = - container_of(work, struct cpu_dbs_info_s, work.work); - unsigned int cpu = dbs_info->cpu; - - /* We want all CPUs to do sampling nearly on same jiffy */ - int delay = usecs_to_jiffies(dbs_tuners_ins.sampling_rate); - - delay -= jiffies % delay; - - mutex_lock(&dbs_info->timer_mutex); + kfree(dbs_data->tuners); +} - dbs_check_cpu(dbs_info); +define_get_cpu_dbs_routines(cs_cpu_dbs_info); - queue_delayed_work_on(cpu, kconservative_wq, &dbs_info->work, delay); - mutex_unlock(&dbs_info->timer_mutex); -} +static struct notifier_block cs_cpufreq_notifier_block = { + .notifier_call = dbs_cpufreq_notifier, +}; -static inline void dbs_timer_init(struct cpu_dbs_info_s *dbs_info) -{ - /* We want all CPUs to do sampling nearly on same jiffy */ - int delay = usecs_to_jiffies(dbs_tuners_ins.sampling_rate); - delay -= jiffies % delay; - - dbs_info->enable = 1; - INIT_DELAYED_WORK_DEFERRABLE(&dbs_info->work, do_dbs_timer); - queue_delayed_work_on(dbs_info->cpu, kconservative_wq, &dbs_info->work, - delay); -} +static struct cs_ops cs_ops = { + .notifier_block = &cs_cpufreq_notifier_block, +}; -static inline void dbs_timer_exit(struct cpu_dbs_info_s *dbs_info) -{ - dbs_info->enable = 0; - cancel_delayed_work_sync(&dbs_info->work); -} +static struct common_dbs_data cs_dbs_cdata = { + .governor = GOV_CONSERVATIVE, + .attr_group_gov_sys = &cs_attr_group_gov_sys, + .attr_group_gov_pol = &cs_attr_group_gov_pol, + .get_cpu_cdbs = get_cpu_cdbs, + .get_cpu_dbs_info_s = get_cpu_dbs_info_s, + .gov_dbs_timer = cs_dbs_timer, + .gov_check_cpu = cs_check_cpu, + .gov_ops = &cs_ops, + .init = cs_init, + .exit = cs_exit, +}; -static int cpufreq_governor_dbs(struct cpufreq_policy *policy, +static int cs_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; - int rc; - - this_dbs_info = &per_cpu(cs_cpu_dbs_info, cpu); - - switch (event) { - case CPUFREQ_GOV_START: - if ((!cpu_online(cpu)) || (!policy->cur)) - return -EINVAL; - - mutex_lock(&dbs_mutex); - - rc = sysfs_create_group(&policy->kobj, &dbs_attr_group_old); - if (rc) { - mutex_unlock(&dbs_mutex); - return rc; - } - - for_each_cpu(j, policy->cpus) { - struct cpu_dbs_info_s *j_dbs_info; - j_dbs_info = &per_cpu(cs_cpu_dbs_info, j); - j_dbs_info->cur_policy = policy; - - j_dbs_info->prev_cpu_idle = get_cpu_idle_time(j, - &j_dbs_info->prev_cpu_wall); - if (dbs_tuners_ins.ignore_nice) { - j_dbs_info->prev_cpu_nice = - kstat_cpu(j).cpustat.nice; - } - } - this_dbs_info->down_skip = 0; - this_dbs_info->requested_freq = policy->cur; - - mutex_init(&this_dbs_info->timer_mutex); - 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; - - rc = sysfs_create_group(cpufreq_global_kobject, - &dbs_attr_group); - if (rc) { - mutex_unlock(&dbs_mutex); - return rc; - } - - /* - * conservative does not implement micro like ondemand - * governor, thus we are bound to jiffes/HZ - */ - min_sampling_rate = - MIN_SAMPLING_RATE_RATIO * jiffies_to_usecs(10); - /* Bring kernel and HW constraints together */ - min_sampling_rate = max(min_sampling_rate, - MIN_LATENCY_MULTIPLIER * latency); - dbs_tuners_ins.sampling_rate = - max(min_sampling_rate, - latency * LATENCY_MULTIPLIER); - - cpufreq_register_notifier( - &dbs_cpufreq_notifier_block, - CPUFREQ_TRANSITION_NOTIFIER); - } - mutex_unlock(&dbs_mutex); - - dbs_timer_init(this_dbs_info); - - break; - - case CPUFREQ_GOV_STOP: - dbs_timer_exit(this_dbs_info); - - mutex_lock(&dbs_mutex); - sysfs_remove_group(&policy->kobj, &dbs_attr_group_old); - dbs_enable--; - mutex_destroy(&this_dbs_info->timer_mutex); - - /* - * Stop the timerschedule work, when this governor - * is used for first time - */ - if (dbs_enable == 0) - cpufreq_unregister_notifier( - &dbs_cpufreq_notifier_block, - CPUFREQ_TRANSITION_NOTIFIER); - - mutex_unlock(&dbs_mutex); - if (!dbs_enable) - sysfs_remove_group(cpufreq_global_kobject, - &dbs_attr_group); - - break; - - case CPUFREQ_GOV_LIMITS: - mutex_lock(&this_dbs_info->timer_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(&this_dbs_info->timer_mutex); - - break; - } - return 0; + return cpufreq_governor_dbs(policy, &cs_dbs_cdata, event); } #ifndef CONFIG_CPU_FREQ_DEFAULT_GOV_CONSERVATIVE @@ -721,35 +379,21 @@ static #endif struct cpufreq_governor cpufreq_gov_conservative = { .name = "conservative", - .governor = cpufreq_governor_dbs, + .governor = cs_cpufreq_governor_dbs, .max_transition_latency = TRANSITION_LATENCY_LIMIT, .owner = THIS_MODULE, }; static int __init cpufreq_gov_dbs_init(void) { - int err; - - kconservative_wq = create_workqueue("kconservative"); - if (!kconservative_wq) { - printk(KERN_ERR "Creation of kconservative failed\n"); - return -EFAULT; - } - - err = cpufreq_register_governor(&cpufreq_gov_conservative); - if (err) - destroy_workqueue(kconservative_wq); - - return err; + return cpufreq_register_governor(&cpufreq_gov_conservative); } static void __exit cpufreq_gov_dbs_exit(void) { cpufreq_unregister_governor(&cpufreq_gov_conservative); - destroy_workqueue(kconservative_wq); } - MODULE_AUTHOR("Alexander Clouter <alex@digriz.org.uk>"); MODULE_DESCRIPTION("'cpufreq_conservative' - A dynamic cpufreq governor for " "Low Latency Frequency Transition capable processors " |