linux/drivers/cpufreq/cpufreq_ondemand.c
Zhao Liu 3e5c04f97c cpufreq: ondemand: Use cpumask_var_t for on-stack cpu mask
A cpumask structure on the stack can cause a warning with
CONFIG_NR_CPUS=8192 (e.g. Ubuntu 22.04 uses this):

drivers/cpufreq/cpufreq_ondemand.c: In function 'od_set_powersave_bias':
drivers/cpufreq/cpufreq_ondemand.c:449:1: warning: the frame size of
	1032 bytes is larger than 1024 bytes [-Wframe-larger-than=]
  449 | }
      | ^

CONFIG_CPUMASK_OFFSTACK=y is enabled by default for most distros, and
hence we can work around the warning by using cpumask_var_t.

Signed-off-by: Zhao Liu <zhao1.liu@linux.intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2022-07-25 20:38:01 +02:00

487 lines
13 KiB
C

// SPDX-License-Identifier: GPL-2.0-only
/*
* drivers/cpufreq/cpufreq_ondemand.c
*
* Copyright (C) 2001 Russell King
* (C) 2003 Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>.
* Jun Nakajima <jun.nakajima@intel.com>
*/
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
#include <linux/cpu.h>
#include <linux/percpu-defs.h>
#include <linux/slab.h>
#include <linux/tick.h>
#include <linux/sched/cpufreq.h>
#include "cpufreq_ondemand.h"
/* On-demand governor macros */
#define DEF_FREQUENCY_UP_THRESHOLD (80)
#define DEF_SAMPLING_DOWN_FACTOR (1)
#define MAX_SAMPLING_DOWN_FACTOR (100000)
#define MICRO_FREQUENCY_UP_THRESHOLD (95)
#define MICRO_FREQUENCY_MIN_SAMPLE_RATE (10000)
#define MIN_FREQUENCY_UP_THRESHOLD (1)
#define MAX_FREQUENCY_UP_THRESHOLD (100)
static struct od_ops od_ops;
static unsigned int default_powersave_bias;
/*
* Not all CPUs want IO time to be accounted as busy; this depends on how
* efficient idling at a higher frequency/voltage is.
* Pavel Machek says this is not so for various generations of AMD and old
* Intel systems.
* Mike Chan (android.com) claims this is also not true for ARM.
* Because of this, whitelist specific known (series) of CPUs by default, and
* leave all others up to the user.
*/
static int should_io_be_busy(void)
{
#if defined(CONFIG_X86)
/*
* For Intel, Core 2 (model 15) and later have an efficient idle.
*/
if (boot_cpu_data.x86_vendor == X86_VENDOR_INTEL &&
boot_cpu_data.x86 == 6 &&
boot_cpu_data.x86_model >= 15)
return 1;
#endif
return 0;
}
/*
* Find right freq to be set now with powersave_bias on.
* Returns the freq_hi to be used right now and will set freq_hi_delay_us,
* freq_lo, and freq_lo_delay_us in percpu area for averaging freqs.
*/
static unsigned int generic_powersave_bias_target(struct cpufreq_policy *policy,
unsigned int freq_next, unsigned int relation)
{
unsigned int freq_req, freq_reduc, freq_avg;
unsigned int freq_hi, freq_lo;
unsigned int index;
unsigned int delay_hi_us;
struct policy_dbs_info *policy_dbs = policy->governor_data;
struct od_policy_dbs_info *dbs_info = to_dbs_info(policy_dbs);
struct dbs_data *dbs_data = policy_dbs->dbs_data;
struct od_dbs_tuners *od_tuners = dbs_data->tuners;
struct cpufreq_frequency_table *freq_table = policy->freq_table;
if (!freq_table) {
dbs_info->freq_lo = 0;
dbs_info->freq_lo_delay_us = 0;
return freq_next;
}
index = cpufreq_frequency_table_target(policy, freq_next, relation);
freq_req = freq_table[index].frequency;
freq_reduc = freq_req * od_tuners->powersave_bias / 1000;
freq_avg = freq_req - freq_reduc;
/* Find freq bounds for freq_avg in freq_table */
index = cpufreq_table_find_index_h(policy, freq_avg,
relation & CPUFREQ_RELATION_E);
freq_lo = freq_table[index].frequency;
index = cpufreq_table_find_index_l(policy, freq_avg,
relation & CPUFREQ_RELATION_E);
freq_hi = freq_table[index].frequency;
/* Find out how long we have to be in hi and lo freqs */
if (freq_hi == freq_lo) {
dbs_info->freq_lo = 0;
dbs_info->freq_lo_delay_us = 0;
return freq_lo;
}
delay_hi_us = (freq_avg - freq_lo) * dbs_data->sampling_rate;
delay_hi_us += (freq_hi - freq_lo) / 2;
delay_hi_us /= freq_hi - freq_lo;
dbs_info->freq_hi_delay_us = delay_hi_us;
dbs_info->freq_lo = freq_lo;
dbs_info->freq_lo_delay_us = dbs_data->sampling_rate - delay_hi_us;
return freq_hi;
}
static void ondemand_powersave_bias_init(struct cpufreq_policy *policy)
{
struct od_policy_dbs_info *dbs_info = to_dbs_info(policy->governor_data);
dbs_info->freq_lo = 0;
}
static void dbs_freq_increase(struct cpufreq_policy *policy, unsigned int freq)
{
struct policy_dbs_info *policy_dbs = policy->governor_data;
struct dbs_data *dbs_data = policy_dbs->dbs_data;
struct od_dbs_tuners *od_tuners = dbs_data->tuners;
if (od_tuners->powersave_bias)
freq = od_ops.powersave_bias_target(policy, freq,
CPUFREQ_RELATION_HE);
else if (policy->cur == policy->max)
return;
__cpufreq_driver_target(policy, freq, od_tuners->powersave_bias ?
CPUFREQ_RELATION_LE : CPUFREQ_RELATION_HE);
}
/*
* Every sampling_rate, we check, if current idle time is less than 20%
* (default), then we try to increase frequency. Else, we adjust the frequency
* proportional to load.
*/
static void od_update(struct cpufreq_policy *policy)
{
struct policy_dbs_info *policy_dbs = policy->governor_data;
struct od_policy_dbs_info *dbs_info = to_dbs_info(policy_dbs);
struct dbs_data *dbs_data = policy_dbs->dbs_data;
struct od_dbs_tuners *od_tuners = dbs_data->tuners;
unsigned int load = dbs_update(policy);
dbs_info->freq_lo = 0;
/* Check for frequency increase */
if (load > dbs_data->up_threshold) {
/* If switching to max speed, apply sampling_down_factor */
if (policy->cur < policy->max)
policy_dbs->rate_mult = dbs_data->sampling_down_factor;
dbs_freq_increase(policy, policy->max);
} else {
/* Calculate the next frequency proportional to load */
unsigned int freq_next, min_f, max_f;
min_f = policy->cpuinfo.min_freq;
max_f = policy->cpuinfo.max_freq;
freq_next = min_f + load * (max_f - min_f) / 100;
/* No longer fully busy, reset rate_mult */
policy_dbs->rate_mult = 1;
if (od_tuners->powersave_bias)
freq_next = od_ops.powersave_bias_target(policy,
freq_next,
CPUFREQ_RELATION_LE);
__cpufreq_driver_target(policy, freq_next, CPUFREQ_RELATION_CE);
}
}
static unsigned int od_dbs_update(struct cpufreq_policy *policy)
{
struct policy_dbs_info *policy_dbs = policy->governor_data;
struct dbs_data *dbs_data = policy_dbs->dbs_data;
struct od_policy_dbs_info *dbs_info = to_dbs_info(policy_dbs);
int sample_type = dbs_info->sample_type;
/* Common NORMAL_SAMPLE setup */
dbs_info->sample_type = OD_NORMAL_SAMPLE;
/*
* OD_SUB_SAMPLE doesn't make sense if sample_delay_ns is 0, so ignore
* it then.
*/
if (sample_type == OD_SUB_SAMPLE && policy_dbs->sample_delay_ns > 0) {
__cpufreq_driver_target(policy, dbs_info->freq_lo,
CPUFREQ_RELATION_HE);
return dbs_info->freq_lo_delay_us;
}
od_update(policy);
if (dbs_info->freq_lo) {
/* Setup SUB_SAMPLE */
dbs_info->sample_type = OD_SUB_SAMPLE;
return dbs_info->freq_hi_delay_us;
}
return dbs_data->sampling_rate * policy_dbs->rate_mult;
}
/************************** sysfs interface ************************/
static struct dbs_governor od_dbs_gov;
static ssize_t io_is_busy_store(struct gov_attr_set *attr_set, const char *buf,
size_t count)
{
struct dbs_data *dbs_data = to_dbs_data(attr_set);
unsigned int input;
int ret;
ret = sscanf(buf, "%u", &input);
if (ret != 1)
return -EINVAL;
dbs_data->io_is_busy = !!input;
/* we need to re-evaluate prev_cpu_idle */
gov_update_cpu_data(dbs_data);
return count;
}
static ssize_t up_threshold_store(struct gov_attr_set *attr_set,
const char *buf, size_t count)
{
struct dbs_data *dbs_data = to_dbs_data(attr_set);
unsigned int input;
int ret;
ret = sscanf(buf, "%u", &input);
if (ret != 1 || input > MAX_FREQUENCY_UP_THRESHOLD ||
input < MIN_FREQUENCY_UP_THRESHOLD) {
return -EINVAL;
}
dbs_data->up_threshold = input;
return count;
}
static ssize_t sampling_down_factor_store(struct gov_attr_set *attr_set,
const char *buf, size_t count)
{
struct dbs_data *dbs_data = to_dbs_data(attr_set);
struct policy_dbs_info *policy_dbs;
unsigned int input;
int ret;
ret = sscanf(buf, "%u", &input);
if (ret != 1 || input > MAX_SAMPLING_DOWN_FACTOR || input < 1)
return -EINVAL;
dbs_data->sampling_down_factor = input;
/* Reset down sampling multiplier in case it was active */
list_for_each_entry(policy_dbs, &attr_set->policy_list, list) {
/*
* Doing this without locking might lead to using different
* rate_mult values in od_update() and od_dbs_update().
*/
mutex_lock(&policy_dbs->update_mutex);
policy_dbs->rate_mult = 1;
mutex_unlock(&policy_dbs->update_mutex);
}
return count;
}
static ssize_t ignore_nice_load_store(struct gov_attr_set *attr_set,
const char *buf, size_t count)
{
struct dbs_data *dbs_data = to_dbs_data(attr_set);
unsigned int input;
int ret;
ret = sscanf(buf, "%u", &input);
if (ret != 1)
return -EINVAL;
if (input > 1)
input = 1;
if (input == dbs_data->ignore_nice_load) { /* nothing to do */
return count;
}
dbs_data->ignore_nice_load = input;
/* we need to re-evaluate prev_cpu_idle */
gov_update_cpu_data(dbs_data);
return count;
}
static ssize_t powersave_bias_store(struct gov_attr_set *attr_set,
const char *buf, size_t count)
{
struct dbs_data *dbs_data = to_dbs_data(attr_set);
struct od_dbs_tuners *od_tuners = dbs_data->tuners;
struct policy_dbs_info *policy_dbs;
unsigned int input;
int ret;
ret = sscanf(buf, "%u", &input);
if (ret != 1)
return -EINVAL;
if (input > 1000)
input = 1000;
od_tuners->powersave_bias = input;
list_for_each_entry(policy_dbs, &attr_set->policy_list, list)
ondemand_powersave_bias_init(policy_dbs->policy);
return count;
}
gov_show_one_common(sampling_rate);
gov_show_one_common(up_threshold);
gov_show_one_common(sampling_down_factor);
gov_show_one_common(ignore_nice_load);
gov_show_one_common(io_is_busy);
gov_show_one(od, powersave_bias);
gov_attr_rw(sampling_rate);
gov_attr_rw(io_is_busy);
gov_attr_rw(up_threshold);
gov_attr_rw(sampling_down_factor);
gov_attr_rw(ignore_nice_load);
gov_attr_rw(powersave_bias);
static struct attribute *od_attrs[] = {
&sampling_rate.attr,
&up_threshold.attr,
&sampling_down_factor.attr,
&ignore_nice_load.attr,
&powersave_bias.attr,
&io_is_busy.attr,
NULL
};
ATTRIBUTE_GROUPS(od);
/************************** sysfs end ************************/
static struct policy_dbs_info *od_alloc(void)
{
struct od_policy_dbs_info *dbs_info;
dbs_info = kzalloc(sizeof(*dbs_info), GFP_KERNEL);
return dbs_info ? &dbs_info->policy_dbs : NULL;
}
static void od_free(struct policy_dbs_info *policy_dbs)
{
kfree(to_dbs_info(policy_dbs));
}
static int od_init(struct dbs_data *dbs_data)
{
struct od_dbs_tuners *tuners;
u64 idle_time;
int cpu;
tuners = kzalloc(sizeof(*tuners), GFP_KERNEL);
if (!tuners)
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 */
dbs_data->up_threshold = MICRO_FREQUENCY_UP_THRESHOLD;
} else {
dbs_data->up_threshold = DEF_FREQUENCY_UP_THRESHOLD;
}
dbs_data->sampling_down_factor = DEF_SAMPLING_DOWN_FACTOR;
dbs_data->ignore_nice_load = 0;
tuners->powersave_bias = default_powersave_bias;
dbs_data->io_is_busy = should_io_be_busy();
dbs_data->tuners = tuners;
return 0;
}
static void od_exit(struct dbs_data *dbs_data)
{
kfree(dbs_data->tuners);
}
static void od_start(struct cpufreq_policy *policy)
{
struct od_policy_dbs_info *dbs_info = to_dbs_info(policy->governor_data);
dbs_info->sample_type = OD_NORMAL_SAMPLE;
ondemand_powersave_bias_init(policy);
}
static struct od_ops od_ops = {
.powersave_bias_target = generic_powersave_bias_target,
};
static struct dbs_governor od_dbs_gov = {
.gov = CPUFREQ_DBS_GOVERNOR_INITIALIZER("ondemand"),
.kobj_type = { .default_groups = od_groups },
.gov_dbs_update = od_dbs_update,
.alloc = od_alloc,
.free = od_free,
.init = od_init,
.exit = od_exit,
.start = od_start,
};
#define CPU_FREQ_GOV_ONDEMAND (od_dbs_gov.gov)
static void od_set_powersave_bias(unsigned int powersave_bias)
{
unsigned int cpu;
cpumask_var_t done;
if (!alloc_cpumask_var(&done, GFP_KERNEL))
return;
default_powersave_bias = powersave_bias;
cpumask_clear(done);
cpus_read_lock();
for_each_online_cpu(cpu) {
struct cpufreq_policy *policy;
struct policy_dbs_info *policy_dbs;
struct dbs_data *dbs_data;
struct od_dbs_tuners *od_tuners;
if (cpumask_test_cpu(cpu, done))
continue;
policy = cpufreq_cpu_get_raw(cpu);
if (!policy || policy->governor != &CPU_FREQ_GOV_ONDEMAND)
continue;
policy_dbs = policy->governor_data;
if (!policy_dbs)
continue;
cpumask_or(done, done, policy->cpus);
dbs_data = policy_dbs->dbs_data;
od_tuners = dbs_data->tuners;
od_tuners->powersave_bias = default_powersave_bias;
}
cpus_read_unlock();
free_cpumask_var(done);
}
void od_register_powersave_bias_handler(unsigned int (*f)
(struct cpufreq_policy *, unsigned int, unsigned int),
unsigned int powersave_bias)
{
od_ops.powersave_bias_target = f;
od_set_powersave_bias(powersave_bias);
}
EXPORT_SYMBOL_GPL(od_register_powersave_bias_handler);
void od_unregister_powersave_bias_handler(void)
{
od_ops.powersave_bias_target = generic_powersave_bias_target;
od_set_powersave_bias(0);
}
EXPORT_SYMBOL_GPL(od_unregister_powersave_bias_handler);
MODULE_AUTHOR("Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>");
MODULE_AUTHOR("Alexey Starikovskiy <alexey.y.starikovskiy@intel.com>");
MODULE_DESCRIPTION("'cpufreq_ondemand' - A dynamic cpufreq governor for "
"Low Latency Frequency Transition capable processors");
MODULE_LICENSE("GPL");
#ifdef CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND
struct cpufreq_governor *cpufreq_default_governor(void)
{
return &CPU_FREQ_GOV_ONDEMAND;
}
#endif
cpufreq_governor_init(CPU_FREQ_GOV_ONDEMAND);
cpufreq_governor_exit(CPU_FREQ_GOV_ONDEMAND);