diff options
author | Viresh Kumar <viresh.kumar@linaro.org> | 2013-03-27 15:58:58 (GMT) |
---|---|---|
committer | Rafael J. Wysocki <rafael.j.wysocki@intel.com> | 2013-03-31 23:11:34 (GMT) |
commit | 4d5dcc4211f9def4281eafb54b8ed483862e8135 (patch) | |
tree | 7f3c725675ce3042d2d2eb86b0b40f93cef73de3 /drivers/cpufreq/cpufreq_ondemand.c | |
parent | 7bd353a995d9049262661d85811d6109140582a3 (diff) | |
download | linux-fsl-qoriq-4d5dcc4211f9def4281eafb54b8ed483862e8135.tar.xz |
cpufreq: governor: Implement per policy instances of governors
Currently, there can't be multiple instances of single governor_type.
If we have a multi-package system, where we have multiple instances
of struct policy (per package), we can't have multiple instances of
same governor. i.e. We can't have multiple instances of ondemand
governor for multiple packages.
Governors directory in sysfs is created at /sys/devices/system/cpu/cpufreq/
governor-name/. Which again reflects that there can be only one
instance of a governor_type in the system.
This is a bottleneck for multicluster system, where we want different
packages to use same governor type, but with different tunables.
This patch uses the infrastructure provided by earlier patch and
implements init/exit routines for ondemand and conservative
governors.
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Diffstat (limited to 'drivers/cpufreq/cpufreq_ondemand.c')
-rw-r--r-- | drivers/cpufreq/cpufreq_ondemand.c | 263 |
1 files changed, 157 insertions, 106 deletions
diff --git a/drivers/cpufreq/cpufreq_ondemand.c b/drivers/cpufreq/cpufreq_ondemand.c index f3eb26c..15e80ee 100644 --- a/drivers/cpufreq/cpufreq_ondemand.c +++ b/drivers/cpufreq/cpufreq_ondemand.c @@ -20,6 +20,7 @@ #include <linux/module.h> #include <linux/mutex.h> #include <linux/percpu-defs.h> +#include <linux/slab.h> #include <linux/sysfs.h> #include <linux/tick.h> #include <linux/types.h> @@ -37,22 +38,12 @@ #define MIN_FREQUENCY_UP_THRESHOLD (11) #define MAX_FREQUENCY_UP_THRESHOLD (100) -static struct dbs_data od_dbs_data; static DEFINE_PER_CPU(struct od_cpu_dbs_info_s, od_cpu_dbs_info); #ifndef CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND static struct cpufreq_governor cpufreq_gov_ondemand; #endif -static struct od_dbs_tuners od_tuners = { - .up_threshold = DEF_FREQUENCY_UP_THRESHOLD, - .sampling_down_factor = DEF_SAMPLING_DOWN_FACTOR, - .adj_up_threshold = DEF_FREQUENCY_UP_THRESHOLD - - DEF_FREQUENCY_DOWN_DIFFERENTIAL, - .ignore_nice = 0, - .powersave_bias = 0, -}; - static void ondemand_powersave_bias_init_cpu(int cpu) { struct od_cpu_dbs_info_s *dbs_info = &per_cpu(od_cpu_dbs_info, cpu); @@ -98,6 +89,8 @@ static unsigned int powersave_bias_target(struct cpufreq_policy *policy, unsigned int jiffies_total, jiffies_hi, jiffies_lo; struct od_cpu_dbs_info_s *dbs_info = &per_cpu(od_cpu_dbs_info, policy->cpu); + struct dbs_data *dbs_data = policy->governor_data; + struct od_dbs_tuners *od_tuners = dbs_data->tuners; if (!dbs_info->freq_table) { dbs_info->freq_lo = 0; @@ -108,7 +101,7 @@ static unsigned int powersave_bias_target(struct cpufreq_policy *policy, cpufreq_frequency_table_target(policy, dbs_info->freq_table, freq_next, relation, &index); freq_req = dbs_info->freq_table[index].frequency; - freq_reduc = freq_req * od_tuners.powersave_bias / 1000; + freq_reduc = freq_req * od_tuners->powersave_bias / 1000; freq_avg = freq_req - freq_reduc; /* Find freq bounds for freq_avg in freq_table */ @@ -127,7 +120,7 @@ static unsigned int powersave_bias_target(struct cpufreq_policy *policy, dbs_info->freq_lo_jiffies = 0; return freq_lo; } - jiffies_total = usecs_to_jiffies(od_tuners.sampling_rate); + jiffies_total = usecs_to_jiffies(od_tuners->sampling_rate); jiffies_hi = (freq_avg - freq_lo) * jiffies_total; jiffies_hi += ((freq_hi - freq_lo) / 2); jiffies_hi /= (freq_hi - freq_lo); @@ -148,12 +141,15 @@ static void ondemand_powersave_bias_init(void) static void dbs_freq_increase(struct cpufreq_policy *p, unsigned int freq) { - if (od_tuners.powersave_bias) + struct dbs_data *dbs_data = p->governor_data; + struct od_dbs_tuners *od_tuners = dbs_data->tuners; + + if (od_tuners->powersave_bias) freq = powersave_bias_target(p, freq, CPUFREQ_RELATION_H); else if (p->cur == p->max) return; - __cpufreq_driver_target(p, freq, od_tuners.powersave_bias ? + __cpufreq_driver_target(p, freq, od_tuners->powersave_bias ? CPUFREQ_RELATION_L : CPUFREQ_RELATION_H); } @@ -170,15 +166,17 @@ static void od_check_cpu(int cpu, unsigned int load_freq) { struct od_cpu_dbs_info_s *dbs_info = &per_cpu(od_cpu_dbs_info, cpu); struct cpufreq_policy *policy = dbs_info->cdbs.cur_policy; + struct dbs_data *dbs_data = policy->governor_data; + struct od_dbs_tuners *od_tuners = dbs_data->tuners; dbs_info->freq_lo = 0; /* Check for frequency increase */ - if (load_freq > od_tuners.up_threshold * policy->cur) { + if (load_freq > od_tuners->up_threshold * policy->cur) { /* If switching to max speed, apply sampling_down_factor */ if (policy->cur < policy->max) dbs_info->rate_mult = - od_tuners.sampling_down_factor; + od_tuners->sampling_down_factor; dbs_freq_increase(policy, policy->max); return; } @@ -193,9 +191,10 @@ static void od_check_cpu(int cpu, unsigned int load_freq) * support the current CPU usage without triggering the up policy. To be * safe, we focus 10 points under the threshold. */ - if (load_freq < od_tuners.adj_up_threshold * policy->cur) { + if (load_freq < od_tuners->adj_up_threshold + * policy->cur) { unsigned int freq_next; - freq_next = load_freq / od_tuners.adj_up_threshold; + freq_next = load_freq / od_tuners->adj_up_threshold; /* No longer fully busy, reset rate_mult */ dbs_info->rate_mult = 1; @@ -203,7 +202,7 @@ static void od_check_cpu(int cpu, unsigned int load_freq) if (freq_next < policy->min) freq_next = policy->min; - if (!od_tuners.powersave_bias) { + if (!od_tuners->powersave_bias) { __cpufreq_driver_target(policy, freq_next, CPUFREQ_RELATION_L); } else { @@ -223,12 +222,14 @@ static void od_dbs_timer(struct work_struct *work) unsigned int cpu = dbs_info->cdbs.cur_policy->cpu; struct od_cpu_dbs_info_s *core_dbs_info = &per_cpu(od_cpu_dbs_info, cpu); + struct dbs_data *dbs_data = dbs_info->cdbs.cur_policy->governor_data; + struct od_dbs_tuners *od_tuners = dbs_data->tuners; int delay, sample_type = core_dbs_info->sample_type; bool eval_load; mutex_lock(&core_dbs_info->cdbs.timer_mutex); eval_load = need_load_eval(&core_dbs_info->cdbs, - od_tuners.sampling_rate); + od_tuners->sampling_rate); /* Common NORMAL_SAMPLE setup */ core_dbs_info->sample_type = OD_NORMAL_SAMPLE; @@ -240,13 +241,13 @@ static void od_dbs_timer(struct work_struct *work) CPUFREQ_RELATION_H); } else { if (eval_load) - dbs_check_cpu(&od_dbs_data, cpu); + dbs_check_cpu(dbs_data, cpu); if (core_dbs_info->freq_lo) { /* Setup timer for SUB_SAMPLE */ core_dbs_info->sample_type = OD_SUB_SAMPLE; delay = core_dbs_info->freq_hi_jiffies; } else { - delay = delay_for_sampling_rate(od_tuners.sampling_rate + delay = delay_for_sampling_rate(od_tuners->sampling_rate * core_dbs_info->rate_mult); } } @@ -256,12 +257,7 @@ static void od_dbs_timer(struct work_struct *work) } /************************** sysfs interface ************************/ - -static ssize_t show_sampling_rate_min(struct kobject *kobj, - struct attribute *attr, char *buf) -{ - return sprintf(buf, "%u\n", od_dbs_data.min_sampling_rate); -} +static struct common_dbs_data od_dbs_cdata; /** * update_sampling_rate - update sampling rate effective immediately if needed. @@ -276,12 +272,14 @@ static ssize_t show_sampling_rate_min(struct kobject *kobj, * reducing the sampling rate, we need to make the new value effective * immediately. */ -static void update_sampling_rate(unsigned int new_rate) +static void update_sampling_rate(struct dbs_data *dbs_data, + unsigned int new_rate) { + struct od_dbs_tuners *od_tuners = dbs_data->tuners; int cpu; - od_tuners.sampling_rate = new_rate = max(new_rate, - od_dbs_data.min_sampling_rate); + od_tuners->sampling_rate = new_rate = max(new_rate, + dbs_data->min_sampling_rate); for_each_online_cpu(cpu) { struct cpufreq_policy *policy; @@ -322,34 +320,37 @@ static void update_sampling_rate(unsigned int new_rate) } } -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) { unsigned int input; int ret; ret = sscanf(buf, "%u", &input); if (ret != 1) return -EINVAL; - update_sampling_rate(input); + + update_sampling_rate(dbs_data, input); return count; } -static ssize_t store_io_is_busy(struct kobject *a, struct attribute *b, - const char *buf, size_t count) +static ssize_t store_io_is_busy(struct dbs_data *dbs_data, const char *buf, + size_t count) { + struct od_dbs_tuners *od_tuners = dbs_data->tuners; unsigned int input; int ret; ret = sscanf(buf, "%u", &input); if (ret != 1) return -EINVAL; - od_tuners.io_is_busy = !!input; + od_tuners->io_is_busy = !!input; 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 od_dbs_tuners *od_tuners = dbs_data->tuners; unsigned int input; int ret; ret = sscanf(buf, "%u", &input); @@ -359,23 +360,24 @@ static ssize_t store_up_threshold(struct kobject *a, struct attribute *b, return -EINVAL; } /* Calculate the new adj_up_threshold */ - od_tuners.adj_up_threshold += input; - od_tuners.adj_up_threshold -= od_tuners.up_threshold; + od_tuners->adj_up_threshold += input; + od_tuners->adj_up_threshold -= od_tuners->up_threshold; - od_tuners.up_threshold = input; + od_tuners->up_threshold = input; return count; } -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 od_dbs_tuners *od_tuners = dbs_data->tuners; unsigned int input, j; int ret; ret = sscanf(buf, "%u", &input); if (ret != 1 || input > MAX_SAMPLING_DOWN_FACTOR || input < 1) return -EINVAL; - od_tuners.sampling_down_factor = input; + od_tuners->sampling_down_factor = input; /* Reset down sampling multiplier in case it was active */ for_each_online_cpu(j) { @@ -386,9 +388,10 @@ static ssize_t store_sampling_down_factor(struct kobject *a, 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(struct dbs_data *dbs_data, const char *buf, + size_t count) { + struct od_dbs_tuners *od_tuners = dbs_data->tuners; unsigned int input; int ret; @@ -401,10 +404,10 @@ static ssize_t store_ignore_nice_load(struct kobject *a, struct attribute *b, if (input > 1) input = 1; - if (input == od_tuners.ignore_nice) { /* nothing to do */ + if (input == od_tuners->ignore_nice) { /* nothing to do */ return count; } - od_tuners.ignore_nice = input; + od_tuners->ignore_nice = input; /* we need to re-evaluate prev_cpu_idle */ for_each_online_cpu(j) { @@ -412,7 +415,7 @@ static ssize_t store_ignore_nice_load(struct kobject *a, struct attribute *b, dbs_info = &per_cpu(od_cpu_dbs_info, j); dbs_info->cdbs.prev_cpu_idle = get_cpu_idle_time(j, &dbs_info->cdbs.prev_cpu_wall); - if (od_tuners.ignore_nice) + if (od_tuners->ignore_nice) dbs_info->cdbs.prev_cpu_nice = kcpustat_cpu(j).cpustat[CPUTIME_NICE]; @@ -420,9 +423,10 @@ static ssize_t store_ignore_nice_load(struct kobject *a, struct attribute *b, return count; } -static ssize_t store_powersave_bias(struct kobject *a, struct attribute *b, - const char *buf, size_t count) +static ssize_t store_powersave_bias(struct dbs_data *dbs_data, const char *buf, + size_t count) { + struct od_dbs_tuners *od_tuners = dbs_data->tuners; unsigned int input; int ret; ret = sscanf(buf, "%u", &input); @@ -433,68 +437,138 @@ static ssize_t store_powersave_bias(struct kobject *a, struct attribute *b, if (input > 1000) input = 1000; - od_tuners.powersave_bias = input; + od_tuners->powersave_bias = input; ondemand_powersave_bias_init(); return count; } -show_one(od, sampling_rate, sampling_rate); -show_one(od, io_is_busy, io_is_busy); -show_one(od, up_threshold, up_threshold); -show_one(od, sampling_down_factor, sampling_down_factor); -show_one(od, ignore_nice_load, ignore_nice); -show_one(od, powersave_bias, powersave_bias); - -define_one_global_rw(sampling_rate); -define_one_global_rw(io_is_busy); -define_one_global_rw(up_threshold); -define_one_global_rw(sampling_down_factor); -define_one_global_rw(ignore_nice_load); -define_one_global_rw(powersave_bias); -define_one_global_ro(sampling_rate_min); - -static struct attribute *dbs_attributes[] = { - &sampling_rate_min.attr, - &sampling_rate.attr, - &up_threshold.attr, - &sampling_down_factor.attr, - &ignore_nice_load.attr, - &powersave_bias.attr, - &io_is_busy.attr, +show_store_one(od, sampling_rate); +show_store_one(od, io_is_busy); +show_store_one(od, up_threshold); +show_store_one(od, sampling_down_factor); +show_store_one(od, ignore_nice); +show_store_one(od, powersave_bias); +declare_show_sampling_rate_min(od); + +gov_sys_pol_attr_rw(sampling_rate); +gov_sys_pol_attr_rw(io_is_busy); +gov_sys_pol_attr_rw(up_threshold); +gov_sys_pol_attr_rw(sampling_down_factor); +gov_sys_pol_attr_rw(ignore_nice); +gov_sys_pol_attr_rw(powersave_bias); +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, + &up_threshold_gov_sys.attr, + &sampling_down_factor_gov_sys.attr, + &ignore_nice_gov_sys.attr, + &powersave_bias_gov_sys.attr, + &io_is_busy_gov_sys.attr, NULL }; -static struct attribute_group od_attr_group = { - .attrs = dbs_attributes, +static struct attribute_group od_attr_group_gov_sys = { + .attrs = dbs_attributes_gov_sys, + .name = "ondemand", +}; + +static struct attribute *dbs_attributes_gov_pol[] = { + &sampling_rate_min_gov_pol.attr, + &sampling_rate_gov_pol.attr, + &up_threshold_gov_pol.attr, + &sampling_down_factor_gov_pol.attr, + &ignore_nice_gov_pol.attr, + &powersave_bias_gov_pol.attr, + &io_is_busy_gov_pol.attr, + NULL +}; + +static struct attribute_group od_attr_group_gov_pol = { + .attrs = dbs_attributes_gov_pol, .name = "ondemand", }; /************************** sysfs end ************************/ +static int od_init(struct dbs_data *dbs_data) +{ + struct od_dbs_tuners *tuners; + u64 idle_time; + int cpu; + + tuners = kzalloc(sizeof(struct od_dbs_tuners), GFP_KERNEL); + if (!tuners) { + pr_err("%s: kzalloc failed\n", __func__); + return -ENOMEM; + } + + cpu = get_cpu(); + idle_time = get_cpu_idle_time_us(cpu, NULL); + put_cpu(); + if (idle_time != -1ULL) { + /* Idle micro accounting is supported. Use finer thresholds */ + tuners->up_threshold = MICRO_FREQUENCY_UP_THRESHOLD; + tuners->adj_up_threshold = MICRO_FREQUENCY_UP_THRESHOLD - + MICRO_FREQUENCY_DOWN_DIFFERENTIAL; + /* + * In nohz/micro accounting case we set the minimum frequency + * not depending on HZ, but fixed (very low). The deferred + * timer might skip some samples if idle/sleeping as needed. + */ + dbs_data->min_sampling_rate = MICRO_FREQUENCY_MIN_SAMPLE_RATE; + } else { + tuners->up_threshold = DEF_FREQUENCY_UP_THRESHOLD; + tuners->adj_up_threshold = DEF_FREQUENCY_UP_THRESHOLD - + DEF_FREQUENCY_DOWN_DIFFERENTIAL; + + /* For correct statistics, we need 10 ticks for each measure */ + dbs_data->min_sampling_rate = MIN_SAMPLING_RATE_RATIO * + jiffies_to_usecs(10); + } + + tuners->sampling_down_factor = DEF_SAMPLING_DOWN_FACTOR; + tuners->ignore_nice = 0; + tuners->powersave_bias = 0; + tuners->io_is_busy = should_io_be_busy(); + + dbs_data->tuners = tuners; + pr_info("%s: tuners %p\n", __func__, tuners); + mutex_init(&dbs_data->mutex); + return 0; +} + +static void od_exit(struct dbs_data *dbs_data) +{ + kfree(dbs_data->tuners); +} + define_get_cpu_dbs_routines(od_cpu_dbs_info); static struct od_ops od_ops = { - .io_busy = should_io_be_busy, .powersave_bias_init_cpu = ondemand_powersave_bias_init_cpu, .powersave_bias_target = powersave_bias_target, .freq_increase = dbs_freq_increase, }; -static struct dbs_data od_dbs_data = { +static struct common_dbs_data od_dbs_cdata = { .governor = GOV_ONDEMAND, - .attr_group = &od_attr_group, - .tuners = &od_tuners, + .attr_group_gov_sys = &od_attr_group_gov_sys, + .attr_group_gov_pol = &od_attr_group_gov_pol, .get_cpu_cdbs = get_cpu_cdbs, .get_cpu_dbs_info_s = get_cpu_dbs_info_s, .gov_dbs_timer = od_dbs_timer, .gov_check_cpu = od_check_cpu, .gov_ops = &od_ops, + .init = od_init, + .exit = od_exit, }; static int od_cpufreq_governor_dbs(struct cpufreq_policy *policy, unsigned int event) { - return cpufreq_governor_dbs(&od_dbs_data, policy, event); + return cpufreq_governor_dbs(policy, &od_dbs_cdata, event); } #ifndef CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND @@ -509,29 +583,6 @@ struct cpufreq_governor cpufreq_gov_ondemand = { static int __init cpufreq_gov_dbs_init(void) { - u64 idle_time; - int cpu = get_cpu(); - - mutex_init(&od_dbs_data.mutex); - idle_time = get_cpu_idle_time_us(cpu, NULL); - put_cpu(); - if (idle_time != -1ULL) { - /* Idle micro accounting is supported. Use finer thresholds */ - od_tuners.up_threshold = MICRO_FREQUENCY_UP_THRESHOLD; - od_tuners.adj_up_threshold = MICRO_FREQUENCY_UP_THRESHOLD - - MICRO_FREQUENCY_DOWN_DIFFERENTIAL; - /* - * In nohz/micro accounting case we set the minimum frequency - * not depending on HZ, but fixed (very low). The deferred - * timer might skip some samples if idle/sleeping as needed. - */ - od_dbs_data.min_sampling_rate = MICRO_FREQUENCY_MIN_SAMPLE_RATE; - } else { - /* For correct statistics, we need 10 ticks for each measure */ - od_dbs_data.min_sampling_rate = MIN_SAMPLING_RATE_RATIO * - jiffies_to_usecs(10); - } - return cpufreq_register_governor(&cpufreq_gov_ondemand); } |