2005-04-17 05:20:36 +07:00
|
|
|
/*
|
|
|
|
* drivers/cpufreq/cpufreq_ondemand.c
|
|
|
|
*
|
|
|
|
* Copyright (C) 2001 Russell King
|
|
|
|
* (C) 2003 Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>.
|
|
|
|
* Jun Nakajima <jun.nakajima@intel.com>
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or modify
|
|
|
|
* it under the terms of the GNU General Public License version 2 as
|
|
|
|
* published by the Free Software Foundation.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/kernel.h>
|
|
|
|
#include <linux/module.h>
|
|
|
|
#include <linux/smp.h>
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/interrupt.h>
|
|
|
|
#include <linux/ctype.h>
|
|
|
|
#include <linux/cpufreq.h>
|
|
|
|
#include <linux/sysctl.h>
|
|
|
|
#include <linux/types.h>
|
|
|
|
#include <linux/fs.h>
|
|
|
|
#include <linux/sysfs.h>
|
|
|
|
#include <linux/sched.h>
|
|
|
|
#include <linux/kmod.h>
|
|
|
|
#include <linux/workqueue.h>
|
|
|
|
#include <linux/jiffies.h>
|
|
|
|
#include <linux/kernel_stat.h>
|
|
|
|
#include <linux/percpu.h>
|
2006-01-14 06:54:22 +07:00
|
|
|
#include <linux/mutex.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* 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)
|
2005-06-01 09:03:50 +07:00
|
|
|
#define MIN_FREQUENCY_UP_THRESHOLD (11)
|
2005-04-17 05:20:36 +07:00
|
|
|
#define MAX_FREQUENCY_UP_THRESHOLD (100)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The polling frequency of this governor depends on the capability of
|
|
|
|
* the processor. Default polling frequency is 1000 times the transition
|
|
|
|
* latency of the processor. The governor will work on any processor with
|
|
|
|
* transition latency <= 10mS, using appropriate sampling
|
|
|
|
* rate.
|
|
|
|
* For CPUs with transition latency > 10mS (mostly drivers with CPUFREQ_ETERNAL)
|
|
|
|
* this governor will not work.
|
|
|
|
* All times here are in uS.
|
|
|
|
*/
|
|
|
|
static unsigned int def_sampling_rate;
|
[CPUFREQ] Avoid the ondemand cpufreq governor to use a too high frequency for stats.
The problem is in the ondemand governor, there is a periodic measurement
of the CPU usage. This CPU usage is updated by the scheduler after every
tick (basically, by adding 1 either to "idle" or to "user" or to
"system"). So if the frequency of the governor is too high, the stat
will be meaningless (as mostly no number have changed).
So this patch checks that the measurements are separated by at least 10
ticks. It means that by default, stats will have about 5% error (20
ticks). Of course those numbers can be argued but, IMHO, they look sane.
The patch also includes a small clean-up to check more explictly the
result of the conversion from ns to µs being null.
Let's note that (on x86) this has never been really needed before 2.6.13
because HZ was always 1000. Now that HZ can be 100, some CPU might be
affected by this problem. For instance when HZ=100, the centrino ,which
has a 10µs transition latency, would lead to the governor allowing to
read stats every tick (10ms)!
Signed-off-by: Eric Piel <eric.piel@tremplin-utc.net>
Signed-off-by: Dave Jones <davej@redhat.com>
2005-09-21 02:39:35 +07:00
|
|
|
#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)
|
2005-04-17 05:20:36 +07:00
|
|
|
#define MAX_SAMPLING_RATE (500 * def_sampling_rate)
|
|
|
|
#define DEF_SAMPLING_RATE_LATENCY_MULTIPLIER (1000)
|
2005-06-01 09:03:50 +07:00
|
|
|
#define DEF_SAMPLING_DOWN_FACTOR (1)
|
|
|
|
#define MAX_SAMPLING_DOWN_FACTOR (10)
|
2005-04-17 05:20:36 +07:00
|
|
|
#define TRANSITION_LATENCY_LIMIT (10 * 1000)
|
|
|
|
|
|
|
|
static void do_dbs_timer(void *data);
|
|
|
|
|
|
|
|
struct cpu_dbs_info_s {
|
|
|
|
struct cpufreq_policy *cur_policy;
|
|
|
|
unsigned int prev_cpu_idle_up;
|
|
|
|
unsigned int prev_cpu_idle_down;
|
|
|
|
unsigned int enable;
|
|
|
|
};
|
|
|
|
static DEFINE_PER_CPU(struct cpu_dbs_info_s, cpu_dbs_info);
|
|
|
|
|
|
|
|
static unsigned int dbs_enable; /* number of CPUs using this policy */
|
|
|
|
|
2006-01-14 06:54:22 +07:00
|
|
|
static DEFINE_MUTEX (dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
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;
|
2005-06-01 09:03:47 +07:00
|
|
|
unsigned int ignore_nice;
|
2005-04-17 05:20:36 +07:00
|
|
|
};
|
|
|
|
|
|
|
|
static struct dbs_tuners dbs_tuners_ins = {
|
|
|
|
.up_threshold = DEF_FREQUENCY_UP_THRESHOLD,
|
|
|
|
.sampling_down_factor = DEF_SAMPLING_DOWN_FACTOR,
|
|
|
|
};
|
|
|
|
|
2005-06-01 09:03:49 +07:00
|
|
|
static inline unsigned int get_cpu_idle_time(unsigned int cpu)
|
|
|
|
{
|
|
|
|
return kstat_cpu(cpu).cpustat.idle +
|
|
|
|
kstat_cpu(cpu).cpustat.iowait +
|
2005-12-01 16:09:25 +07:00
|
|
|
( dbs_tuners_ins.ignore_nice ?
|
2005-06-01 09:03:49 +07:00
|
|
|
kstat_cpu(cpu).cpustat.nice :
|
|
|
|
0);
|
|
|
|
}
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/************************** sysfs interface ************************/
|
|
|
|
static ssize_t show_sampling_rate_max(struct cpufreq_policy *policy, char *buf)
|
|
|
|
{
|
|
|
|
return sprintf (buf, "%u\n", MAX_SAMPLING_RATE);
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t show_sampling_rate_min(struct cpufreq_policy *policy, char *buf)
|
|
|
|
{
|
|
|
|
return sprintf (buf, "%u\n", MIN_SAMPLING_RATE);
|
|
|
|
}
|
|
|
|
|
|
|
|
#define define_one_ro(_name) \
|
|
|
|
static struct freq_attr _name = \
|
|
|
|
__ATTR(_name, 0444, show_##_name, NULL)
|
|
|
|
|
|
|
|
define_one_ro(sampling_rate_max);
|
|
|
|
define_one_ro(sampling_rate_min);
|
|
|
|
|
|
|
|
/* cpufreq_ondemand Governor Tunables */
|
|
|
|
#define show_one(file_name, object) \
|
|
|
|
static ssize_t show_##file_name \
|
|
|
|
(struct cpufreq_policy *unused, char *buf) \
|
|
|
|
{ \
|
|
|
|
return sprintf(buf, "%u\n", dbs_tuners_ins.object); \
|
|
|
|
}
|
|
|
|
show_one(sampling_rate, sampling_rate);
|
|
|
|
show_one(sampling_down_factor, sampling_down_factor);
|
|
|
|
show_one(up_threshold, up_threshold);
|
2005-12-01 16:09:25 +07:00
|
|
|
show_one(ignore_nice_load, ignore_nice);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
static ssize_t store_sampling_down_factor(struct cpufreq_policy *unused,
|
|
|
|
const char *buf, size_t count)
|
|
|
|
{
|
|
|
|
unsigned int input;
|
|
|
|
int ret;
|
|
|
|
ret = sscanf (buf, "%u", &input);
|
|
|
|
if (ret != 1 )
|
|
|
|
return -EINVAL;
|
|
|
|
|
2005-06-01 09:03:50 +07:00
|
|
|
if (input > MAX_SAMPLING_DOWN_FACTOR || input < 1)
|
|
|
|
return -EINVAL;
|
|
|
|
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_lock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
dbs_tuners_ins.sampling_down_factor = input;
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_unlock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
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);
|
|
|
|
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_lock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
if (ret != 1 || input > MAX_SAMPLING_RATE || input < MIN_SAMPLING_RATE) {
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_unlock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
dbs_tuners_ins.sampling_rate = input;
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_unlock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
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);
|
|
|
|
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_lock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
if (ret != 1 || input > MAX_FREQUENCY_UP_THRESHOLD ||
|
2005-06-01 09:03:50 +07:00
|
|
|
input < MIN_FREQUENCY_UP_THRESHOLD) {
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_unlock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
dbs_tuners_ins.up_threshold = input;
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_unlock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
return count;
|
|
|
|
}
|
|
|
|
|
2005-12-01 16:09:25 +07:00
|
|
|
static ssize_t store_ignore_nice_load(struct cpufreq_policy *policy,
|
2005-06-01 09:03:47 +07:00
|
|
|
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;
|
|
|
|
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_lock(&dbs_mutex);
|
2005-06-01 09:03:47 +07:00
|
|
|
if ( input == dbs_tuners_ins.ignore_nice ) { /* nothing to do */
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_unlock(&dbs_mutex);
|
2005-06-01 09:03:47 +07:00
|
|
|
return count;
|
|
|
|
}
|
|
|
|
dbs_tuners_ins.ignore_nice = input;
|
|
|
|
|
|
|
|
/* we need to re-evaluate prev_cpu_idle_up and prev_cpu_idle_down */
|
2005-06-01 09:03:49 +07:00
|
|
|
for_each_online_cpu(j) {
|
2005-06-01 09:03:47 +07:00
|
|
|
struct cpu_dbs_info_s *j_dbs_info;
|
|
|
|
j_dbs_info = &per_cpu(cpu_dbs_info, j);
|
2005-06-01 09:03:49 +07:00
|
|
|
j_dbs_info->prev_cpu_idle_up = get_cpu_idle_time(j);
|
2005-06-01 09:03:47 +07:00
|
|
|
j_dbs_info->prev_cpu_idle_down = j_dbs_info->prev_cpu_idle_up;
|
|
|
|
}
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_unlock(&dbs_mutex);
|
2005-06-01 09:03:47 +07:00
|
|
|
|
|
|
|
return count;
|
|
|
|
}
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
#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);
|
2005-12-01 16:09:25 +07:00
|
|
|
define_one_rw(ignore_nice_load);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
static struct attribute * dbs_attributes[] = {
|
|
|
|
&sampling_rate_max.attr,
|
|
|
|
&sampling_rate_min.attr,
|
|
|
|
&sampling_rate.attr,
|
|
|
|
&sampling_down_factor.attr,
|
|
|
|
&up_threshold.attr,
|
2005-12-01 16:09:25 +07:00
|
|
|
&ignore_nice_load.attr,
|
2005-04-17 05:20:36 +07:00
|
|
|
NULL
|
|
|
|
};
|
|
|
|
|
|
|
|
static struct attribute_group dbs_attr_group = {
|
|
|
|
.attrs = dbs_attributes,
|
|
|
|
.name = "ondemand",
|
|
|
|
};
|
|
|
|
|
|
|
|
/************************** sysfs end ************************/
|
|
|
|
|
|
|
|
static void dbs_check_cpu(int cpu)
|
|
|
|
{
|
2005-06-01 09:03:50 +07:00
|
|
|
unsigned int idle_ticks, up_idle_ticks, total_ticks;
|
|
|
|
unsigned int freq_next;
|
2005-04-17 05:20:36 +07:00
|
|
|
unsigned int freq_down_sampling_rate;
|
|
|
|
static int down_skip[NR_CPUS];
|
|
|
|
struct cpu_dbs_info_s *this_dbs_info;
|
|
|
|
|
|
|
|
struct cpufreq_policy *policy;
|
|
|
|
unsigned int j;
|
|
|
|
|
|
|
|
this_dbs_info = &per_cpu(cpu_dbs_info, cpu);
|
|
|
|
if (!this_dbs_info->enable)
|
|
|
|
return;
|
|
|
|
|
|
|
|
policy = this_dbs_info->cur_policy;
|
|
|
|
/*
|
2005-06-01 09:03:50 +07:00
|
|
|
* Every sampling_rate, we check, if current idle time is less
|
|
|
|
* than 20% (default), then we try to increase frequency
|
|
|
|
* Every sampling_rate*sampling_down_factor, we look for a the lowest
|
|
|
|
* frequency which can sustain the load while keeping idle time over
|
|
|
|
* 30%. If such a frequency exist, we try to decrease to this frequency.
|
2005-04-17 05:20:36 +07:00
|
|
|
*
|
|
|
|
* Any frequency increase takes it to the maximum frequency.
|
|
|
|
* Frequency reduction happens at minimum steps of
|
2005-06-01 09:03:50 +07:00
|
|
|
* 5% (default) of current frequency
|
2005-04-17 05:20:36 +07:00
|
|
|
*/
|
|
|
|
|
|
|
|
/* Check for frequency increase */
|
2005-06-01 09:03:49 +07:00
|
|
|
idle_ticks = UINT_MAX;
|
2005-04-17 05:20:36 +07:00
|
|
|
for_each_cpu_mask(j, policy->cpus) {
|
2005-06-01 09:03:49 +07:00
|
|
|
unsigned int tmp_idle_ticks, total_idle_ticks;
|
2005-04-17 05:20:36 +07:00
|
|
|
struct cpu_dbs_info_s *j_dbs_info;
|
|
|
|
|
|
|
|
j_dbs_info = &per_cpu(cpu_dbs_info, j);
|
2005-06-01 09:03:49 +07:00
|
|
|
total_idle_ticks = get_cpu_idle_time(j);
|
2005-04-17 05:20:36 +07:00
|
|
|
tmp_idle_ticks = total_idle_ticks -
|
|
|
|
j_dbs_info->prev_cpu_idle_up;
|
|
|
|
j_dbs_info->prev_cpu_idle_up = total_idle_ticks;
|
|
|
|
|
|
|
|
if (tmp_idle_ticks < idle_ticks)
|
|
|
|
idle_ticks = tmp_idle_ticks;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Scale idle ticks by 100 and compare with up and down ticks */
|
|
|
|
idle_ticks *= 100;
|
|
|
|
up_idle_ticks = (100 - dbs_tuners_ins.up_threshold) *
|
2005-06-01 09:03:44 +07:00
|
|
|
usecs_to_jiffies(dbs_tuners_ins.sampling_rate);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
if (idle_ticks < up_idle_ticks) {
|
2005-06-01 09:03:49 +07:00
|
|
|
down_skip[cpu] = 0;
|
2005-06-01 09:03:49 +07:00
|
|
|
for_each_cpu_mask(j, policy->cpus) {
|
|
|
|
struct cpu_dbs_info_s *j_dbs_info;
|
|
|
|
|
|
|
|
j_dbs_info = &per_cpu(cpu_dbs_info, j);
|
|
|
|
j_dbs_info->prev_cpu_idle_down =
|
|
|
|
j_dbs_info->prev_cpu_idle_up;
|
|
|
|
}
|
2005-06-01 09:03:48 +07:00
|
|
|
/* if we are already at full speed then break out early */
|
|
|
|
if (policy->cur == policy->max)
|
|
|
|
return;
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
__cpufreq_driver_target(policy, policy->max,
|
|
|
|
CPUFREQ_RELATION_H);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Check for frequency decrease */
|
|
|
|
down_skip[cpu]++;
|
|
|
|
if (down_skip[cpu] < dbs_tuners_ins.sampling_down_factor)
|
|
|
|
return;
|
|
|
|
|
2005-06-01 09:03:49 +07:00
|
|
|
idle_ticks = UINT_MAX;
|
2005-04-17 05:20:36 +07:00
|
|
|
for_each_cpu_mask(j, policy->cpus) {
|
2005-06-01 09:03:49 +07:00
|
|
|
unsigned int tmp_idle_ticks, total_idle_ticks;
|
2005-04-17 05:20:36 +07:00
|
|
|
struct cpu_dbs_info_s *j_dbs_info;
|
|
|
|
|
|
|
|
j_dbs_info = &per_cpu(cpu_dbs_info, j);
|
2005-06-01 09:03:49 +07:00
|
|
|
/* Check for frequency decrease */
|
|
|
|
total_idle_ticks = j_dbs_info->prev_cpu_idle_up;
|
2005-04-17 05:20:36 +07:00
|
|
|
tmp_idle_ticks = total_idle_ticks -
|
|
|
|
j_dbs_info->prev_cpu_idle_down;
|
|
|
|
j_dbs_info->prev_cpu_idle_down = total_idle_ticks;
|
|
|
|
|
|
|
|
if (tmp_idle_ticks < idle_ticks)
|
|
|
|
idle_ticks = tmp_idle_ticks;
|
|
|
|
}
|
|
|
|
|
|
|
|
down_skip[cpu] = 0;
|
2005-06-01 09:03:50 +07:00
|
|
|
/* if we cannot reduce the frequency anymore, break out early */
|
|
|
|
if (policy->cur == policy->min)
|
|
|
|
return;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2005-06-01 09:03:50 +07:00
|
|
|
/* Compute how many ticks there are between two measurements */
|
2005-04-17 05:20:36 +07:00
|
|
|
freq_down_sampling_rate = dbs_tuners_ins.sampling_rate *
|
|
|
|
dbs_tuners_ins.sampling_down_factor;
|
2005-06-01 09:03:50 +07:00
|
|
|
total_ticks = usecs_to_jiffies(freq_down_sampling_rate);
|
2005-06-01 09:03:48 +07:00
|
|
|
|
2005-06-01 09:03:50 +07:00
|
|
|
/*
|
|
|
|
* The optimal frequency is the frequency that is the lowest that
|
|
|
|
* can support the current CPU usage without triggering the up
|
|
|
|
* policy. To be safe, we focus 10 points under the threshold.
|
|
|
|
*/
|
|
|
|
freq_next = ((total_ticks - idle_ticks) * 100) / total_ticks;
|
|
|
|
freq_next = (freq_next * policy->cur) /
|
|
|
|
(dbs_tuners_ins.up_threshold - 10);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2005-06-01 09:03:50 +07:00
|
|
|
if (freq_next <= ((policy->cur * 95) / 100))
|
|
|
|
__cpufreq_driver_target(policy, freq_next, CPUFREQ_RELATION_L);
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static void do_dbs_timer(void *data)
|
|
|
|
{
|
|
|
|
int i;
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_lock(&dbs_mutex);
|
2005-06-01 09:03:44 +07:00
|
|
|
for_each_online_cpu(i)
|
|
|
|
dbs_check_cpu(i);
|
2005-04-17 05:20:36 +07:00
|
|
|
schedule_delayed_work(&dbs_work,
|
2005-06-01 09:03:44 +07:00
|
|
|
usecs_to_jiffies(dbs_tuners_ins.sampling_rate));
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_unlock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline void dbs_timer_init(void)
|
|
|
|
{
|
|
|
|
INIT_WORK(&dbs_work, do_dbs_timer, NULL);
|
|
|
|
schedule_delayed_work(&dbs_work,
|
2005-06-01 09:03:44 +07:00
|
|
|
usecs_to_jiffies(dbs_tuners_ins.sampling_rate));
|
2005-04-17 05:20:36 +07:00
|
|
|
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;
|
|
|
|
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_lock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
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;
|
|
|
|
|
2005-06-01 09:03:49 +07:00
|
|
|
j_dbs_info->prev_cpu_idle_up = get_cpu_idle_time(j);
|
2005-06-01 09:03:47 +07:00
|
|
|
j_dbs_info->prev_cpu_idle_down
|
|
|
|
= j_dbs_info->prev_cpu_idle_up;
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
this_dbs_info->enable = 1;
|
|
|
|
sysfs_create_group(&policy->kobj, &dbs_attr_group);
|
|
|
|
dbs_enable++;
|
|
|
|
/*
|
|
|
|
* Start the timerschedule work, when this governor
|
|
|
|
* is used for first time
|
|
|
|
*/
|
|
|
|
if (dbs_enable == 1) {
|
|
|
|
unsigned int latency;
|
|
|
|
/* policy latency is in nS. Convert it to uS first */
|
[CPUFREQ] Avoid the ondemand cpufreq governor to use a too high frequency for stats.
The problem is in the ondemand governor, there is a periodic measurement
of the CPU usage. This CPU usage is updated by the scheduler after every
tick (basically, by adding 1 either to "idle" or to "user" or to
"system"). So if the frequency of the governor is too high, the stat
will be meaningless (as mostly no number have changed).
So this patch checks that the measurements are separated by at least 10
ticks. It means that by default, stats will have about 5% error (20
ticks). Of course those numbers can be argued but, IMHO, they look sane.
The patch also includes a small clean-up to check more explictly the
result of the conversion from ns to µs being null.
Let's note that (on x86) this has never been really needed before 2.6.13
because HZ was always 1000. Now that HZ can be 100, some CPU might be
affected by this problem. For instance when HZ=100, the centrino ,which
has a 10µs transition latency, would lead to the governor allowing to
read stats every tick (10ms)!
Signed-off-by: Eric Piel <eric.piel@tremplin-utc.net>
Signed-off-by: Dave Jones <davej@redhat.com>
2005-09-21 02:39:35 +07:00
|
|
|
latency = policy->cpuinfo.transition_latency / 1000;
|
|
|
|
if (latency == 0)
|
|
|
|
latency = 1;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
[CPUFREQ] Avoid the ondemand cpufreq governor to use a too high frequency for stats.
The problem is in the ondemand governor, there is a periodic measurement
of the CPU usage. This CPU usage is updated by the scheduler after every
tick (basically, by adding 1 either to "idle" or to "user" or to
"system"). So if the frequency of the governor is too high, the stat
will be meaningless (as mostly no number have changed).
So this patch checks that the measurements are separated by at least 10
ticks. It means that by default, stats will have about 5% error (20
ticks). Of course those numbers can be argued but, IMHO, they look sane.
The patch also includes a small clean-up to check more explictly the
result of the conversion from ns to µs being null.
Let's note that (on x86) this has never been really needed before 2.6.13
because HZ was always 1000. Now that HZ can be 100, some CPU might be
affected by this problem. For instance when HZ=100, the centrino ,which
has a 10µs transition latency, would lead to the governor allowing to
read stats every tick (10ms)!
Signed-off-by: Eric Piel <eric.piel@tremplin-utc.net>
Signed-off-by: Dave Jones <davej@redhat.com>
2005-09-21 02:39:35 +07:00
|
|
|
def_sampling_rate = latency *
|
2005-04-17 05:20:36 +07:00
|
|
|
DEF_SAMPLING_RATE_LATENCY_MULTIPLIER;
|
[CPUFREQ] Avoid the ondemand cpufreq governor to use a too high frequency for stats.
The problem is in the ondemand governor, there is a periodic measurement
of the CPU usage. This CPU usage is updated by the scheduler after every
tick (basically, by adding 1 either to "idle" or to "user" or to
"system"). So if the frequency of the governor is too high, the stat
will be meaningless (as mostly no number have changed).
So this patch checks that the measurements are separated by at least 10
ticks. It means that by default, stats will have about 5% error (20
ticks). Of course those numbers can be argued but, IMHO, they look sane.
The patch also includes a small clean-up to check more explictly the
result of the conversion from ns to µs being null.
Let's note that (on x86) this has never been really needed before 2.6.13
because HZ was always 1000. Now that HZ can be 100, some CPU might be
affected by this problem. For instance when HZ=100, the centrino ,which
has a 10µs transition latency, would lead to the governor allowing to
read stats every tick (10ms)!
Signed-off-by: Eric Piel <eric.piel@tremplin-utc.net>
Signed-off-by: Dave Jones <davej@redhat.com>
2005-09-21 02:39:35 +07:00
|
|
|
|
|
|
|
if (def_sampling_rate < MIN_STAT_SAMPLING_RATE)
|
|
|
|
def_sampling_rate = MIN_STAT_SAMPLING_RATE;
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
dbs_tuners_ins.sampling_rate = def_sampling_rate;
|
2005-06-01 09:03:47 +07:00
|
|
|
dbs_tuners_ins.ignore_nice = 0;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
dbs_timer_init();
|
|
|
|
}
|
|
|
|
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_unlock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
break;
|
|
|
|
|
|
|
|
case CPUFREQ_GOV_STOP:
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_lock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
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();
|
|
|
|
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_unlock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CPUFREQ_GOV_LIMITS:
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_lock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
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);
|
2006-01-14 06:54:22 +07:00
|
|
|
mutex_unlock(&dbs_mutex);
|
2005-04-17 05:20:36 +07:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2005-06-01 09:03:46 +07:00
|
|
|
static struct cpufreq_governor cpufreq_gov_dbs = {
|
2005-04-17 05:20:36 +07:00
|
|
|
.name = "ondemand",
|
|
|
|
.governor = cpufreq_governor_dbs,
|
|
|
|
.owner = THIS_MODULE,
|
|
|
|
};
|
|
|
|
|
|
|
|
static int __init cpufreq_gov_dbs_init(void)
|
|
|
|
{
|
|
|
|
return cpufreq_register_governor(&cpufreq_gov_dbs);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void __exit cpufreq_gov_dbs_exit(void)
|
|
|
|
{
|
|
|
|
/* Make sure that the scheduled work is indeed not running */
|
|
|
|
flush_scheduled_work();
|
|
|
|
|
|
|
|
cpufreq_unregister_governor(&cpufreq_gov_dbs);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
MODULE_AUTHOR ("Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>");
|
|
|
|
MODULE_DESCRIPTION ("'cpufreq_ondemand' - A dynamic cpufreq governor for "
|
|
|
|
"Low Latency Frequency Transition capable processors");
|
|
|
|
MODULE_LICENSE ("GPL");
|
|
|
|
|
|
|
|
module_init(cpufreq_gov_dbs_init);
|
|
|
|
module_exit(cpufreq_gov_dbs_exit);
|