linux/drivers/cpufreq/cpufreq_conservative.c
Mathieu Desnoyers b253d2b2d2 [CPUFREQ] fix timer teardown in conservative governor
* Rafael J. Wysocki (rjw@sisk.pl) wrote:
> This message has been generated automatically as a part of a report
> of regressions introduced between 2.6.28 and 2.6.29.
>
> The following bug entry is on the current list of known regressions
> introduced between 2.6.28 and 2.6.29.  Please verify if it still should
> be listed and let me know (either way).
>
>
> Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=13186
> Subject		: cpufreq timer teardown problem
> Submitter	: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
> Date		: 2009-04-23 14:00 (24 days old)
> References	: http://marc.info/?l=linux-kernel&m=124049523515036&w=4
> Handled-By	: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
> Patch		: http://patchwork.kernel.org/patch/19754/
> 		  http://patchwork.kernel.org/patch/19753/
>

(re-send with updated changelog)

cpufreq fix timer teardown in conservative governor

The problem is that dbs_timer_exit() uses cancel_delayed_work() when it should
use cancel_delayed_work_sync(). cancel_delayed_work() does not wait for the
workqueue handler to exit.

The ondemand governor does not seem to be affected because the
"if (!dbs_info->enable)" check at the beginning of the workqueue handler returns
immediately without rescheduling the work. The conservative governor in
2.6.30-rc has the same check as the ondemand governor, which makes things
usually run smoothly. However, if the governor is quickly stopped and then
started, this could lead to the following race :

dbs_enable could be reenabled and multiple do_dbs_timer handlers would run.
This is why a synchronized teardown is required.

Depends on patch
cpufreq: remove rwsem lock from CPUFREQ_GOV_STOP call

The following patch applies to 2.6.30-rc2. Stable kernels have a similar
issue which should also be fixed, but the code changed between 2.6.29
and 2.6.30, so this patch only applies to 2.6.30-rc.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
CC: Andrew Morton <akpm@linux-foundation.org>
CC: gregkh@suse.de
CC: stable@kernel.org
CC: cpufreq@vger.kernel.org
CC: Ingo Molnar <mingo@elte.hu>
CC: rjw@sisk.pl
CC: Ben Slusky <sluskyb@paranoiacs.org>
Signed-off-by: Dave Jones <davej@redhat.com>
2009-05-26 12:04:50 -04:00

701 lines
18 KiB
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) 2009 Alexander Clouter <alex@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/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
*/
#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)
/* Above MIN_SAMPLING_RATE will vanish with its sysfs file soon
* Define the minimal settable sampling rate to the greater of:
* - "HW transition latency" * 100 (same as default sampling / 10)
* - MIN_STAT_SAMPLING_RATE
* To avoid that userspace shoots itself.
*/
static unsigned int minimum_sampling_rate(void)
{
return max(def_sampling_rate / 10, MIN_STAT_SAMPLING_RATE);
}
/* This will also vanish soon with removing sampling_rate_max */
#define MAX_SAMPLING_RATE (500 * def_sampling_rate)
#define LATENCY_MULTIPLIER (1000)
#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;
};
static DEFINE_PER_CPU(struct cpu_dbs_info_s, cpu_dbs_info);
static unsigned int dbs_enable; /* number of CPUs using this policy */
/*
* DEADLOCK ALERT! There is a ordering requirement between cpu_hotplug
* lock and dbs_mutex. cpu_hotplug lock should always be held before
* dbs_mutex. If any function that can potentially take cpu_hotplug lock
* (like __cpufreq_driver_target()) is being called with dbs_mutex taken, then
* cpu_hotplug lock should be taken before that. Note that cpu_hotplug lock
* is recursive for the same process. -Venki
* DEADLOCK ALERT! (2) : do_dbs_timer() must not take the dbs_mutex, because it
* would deadlock with cancel_delayed_work_sync(), which is needed for proper
* raceless workqueue teardown.
*/
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,
};
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;
cur_wall_time = jiffies64_to_cputime64(get_jiffies_64());
busy_time = cputime64_add(kstat_cpu(cpu).cpustat.user,
kstat_cpu(cpu).cpustat.system);
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);
idle_time = cputime64_sub(cur_wall_time, busy_time);
if (wall)
*wall = cur_wall_time;
return idle_time;
}
static inline cputime64_t get_cpu_idle_time(unsigned int cpu, cputime64_t *wall)
{
u64 idle_time = get_cpu_idle_time_us(cpu, wall);
if (idle_time == -1ULL)
return get_cpu_idle_time_jiffy(cpu, wall);
return idle_time;
}
/* keep track of frequency transitions */
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(cpu_dbs_info,
freq->cpu);
struct cpufreq_policy *policy;
if (!this_dbs_info->enable)
return 0;
policy = this_dbs_info->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
*/
if (this_dbs_info->requested_freq > policy->max
|| this_dbs_info->requested_freq < policy->min)
this_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 cpufreq_policy *policy, char *buf)
{
static int print_once;
if (!print_once) {
printk(KERN_INFO "CPUFREQ: conservative sampling_rate_max "
"sysfs file is deprecated - used by: %s\n",
current->comm);
print_once = 1;
}
return sprintf(buf, "%u\n", MAX_SAMPLING_RATE);
}
static ssize_t show_sampling_rate_min(struct cpufreq_policy *policy, char *buf)
{
static int print_once;
if (!print_once) {
printk(KERN_INFO "CPUFREQ: conservative sampling_rate_max "
"sysfs file is deprecated - used by: %s\n", current->comm);
print_once = 1;
}
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);
if (ret != 1)
return -EINVAL;
mutex_lock(&dbs_mutex);
dbs_tuners_ins.sampling_rate = max(input, minimum_sampling_rate());
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);
/* 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);
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 */
for_each_online_cpu(j) {
struct cpu_dbs_info_s *dbs_info;
dbs_info = &per_cpu(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;
}
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(struct cpu_dbs_info_s *this_dbs_info)
{
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(cpu_dbs_info, j);
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;
}
/*
* 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;
/*
* 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;
}
}
static void do_dbs_timer(struct work_struct *work)
{
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;
if (lock_policy_rwsem_write(cpu) < 0)
return;
if (!dbs_info->enable) {
unlock_policy_rwsem_write(cpu);
return;
}
dbs_check_cpu(dbs_info);
queue_delayed_work_on(cpu, kconservative_wq, &dbs_info->work, delay);
unlock_policy_rwsem_write(cpu);
}
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 inline void dbs_timer_exit(struct cpu_dbs_info_s *dbs_info)
{
dbs_info->enable = 0;
cancel_delayed_work_sync(&dbs_info->work);
}
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;
int rc;
this_dbs_info = &per_cpu(cpu_dbs_info, cpu);
switch (event) {
case CPUFREQ_GOV_START:
if ((!cpu_online(cpu)) || (!policy->cur))
return -EINVAL;
if (this_dbs_info->enable) /* Already enabled */
break;
mutex_lock(&dbs_mutex);
rc = sysfs_create_group(&policy->kobj, &dbs_attr_group);
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(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;
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 =
max(latency * LATENCY_MULTIPLIER,
MIN_STAT_SAMPLING_RATE);
dbs_tuners_ins.sampling_rate = def_sampling_rate;
cpufreq_register_notifier(
&dbs_cpufreq_notifier_block,
CPUFREQ_TRANSITION_NOTIFIER);
}
dbs_timer_init(this_dbs_info);
mutex_unlock(&dbs_mutex);
break;
case CPUFREQ_GOV_STOP:
mutex_lock(&dbs_mutex);
dbs_timer_exit(this_dbs_info);
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)
cpufreq_unregister_notifier(
&dbs_cpufreq_notifier_block,
CPUFREQ_TRANSITION_NOTIFIER);
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;
}
#ifndef CONFIG_CPU_FREQ_DEFAULT_GOV_CONSERVATIVE
static
#endif
struct cpufreq_governor cpufreq_gov_conservative = {
.name = "conservative",
.governor = 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;
}
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 "
"optimised for use in a battery environment");
MODULE_LICENSE("GPL");
#ifdef CONFIG_CPU_FREQ_DEFAULT_GOV_CONSERVATIVE
fs_initcall(cpufreq_gov_dbs_init);
#else
module_init(cpufreq_gov_dbs_init);
#endif
module_exit(cpufreq_gov_dbs_exit);