linux/drivers/cpufreq/cpufreq_conservative.c
Viresh Kumar 732b6d617a cpufreq: governor: Serialize governor callbacks
There are several races reported in cpufreq core around governors (only
ondemand and conservative) by different people.

There are at least two race scenarios present in governor code:
 (a) Concurrent access/updates of governor internal structures.

 It is possible that fields such as 'dbs_data->usage_count', etc.  are
 accessed simultaneously for different policies using same governor
 structure (i.e. CPUFREQ_HAVE_GOVERNOR_PER_POLICY flag unset). And
 because of this we can dereference bad pointers.

 For example consider a system with two CPUs with separate 'struct
 cpufreq_policy' instances. CPU0 governor: ondemand and CPU1: powersave.
 CPU0 switching to powersave and CPU1 to ondemand:
	CPU0				CPU1

	store*				store*

	cpufreq_governor_exit()		cpufreq_governor_init()
					dbs_data = cdata->gdbs_data;

	if (!--dbs_data->usage_count)
		kfree(dbs_data);

					dbs_data->usage_count++;
					*Bad pointer dereference*

 There are other races possible between EXIT and START/STOP/LIMIT as
 well. Its really complicated.

 (b) Switching governor state in bad sequence:

 For example trying to switch a governor to START state, when the
 governor is in EXIT state. There are some checks present in
 __cpufreq_governor() but they aren't sufficient as they compare events
 against 'policy->governor_enabled', where as we need to take governor's
 state into account, which can be used by multiple policies.

These two issues need to be solved separately and the responsibility
should be properly divided between cpufreq and governor core.

The first problem is more about the governor core, as it needs to
protect its structures properly. And the second problem should be fixed
in cpufreq core instead of governor, as its all about sequence of
events.

This patch is trying to solve only the first problem.

There are two types of data we need to protect,
- 'struct common_dbs_data': No matter what, there is going to be a
  single copy of this per governor.
- 'struct dbs_data': With CPUFREQ_HAVE_GOVERNOR_PER_POLICY flag set, we
  will have per-policy copy of this data, otherwise a single copy.

Because of such complexities, the mutex present in 'struct dbs_data' is
insufficient to solve our problem. For example we need to protect
fetching of 'dbs_data' from different structures at the beginning of
cpufreq_governor_dbs(), to make sure it isn't currently being updated.

This can be fixed if we can guarantee serialization of event parsing
code for an individual governor. This is best solved with a mutex per
governor, and the placeholder for that is 'struct common_dbs_data'.

And so this patch moves the mutex from 'struct dbs_data' to 'struct
common_dbs_data' and takes it at the beginning and drops it at the end
of cpufreq_governor_dbs().

Tested with and without following configuration options:

CONFIG_LOCKDEP_SUPPORT=y
CONFIG_DEBUG_RT_MUTEXES=y
CONFIG_DEBUG_PI_LIST=y
CONFIG_DEBUG_SPINLOCK=y
CONFIG_DEBUG_MUTEXES=y
CONFIG_DEBUG_LOCK_ALLOC=y
CONFIG_PROVE_LOCKING=y
CONFIG_LOCKDEP=y
CONFIG_DEBUG_ATOMIC_SLEEP=y

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-06-15 15:42:53 +02:00

413 lines
11 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/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;
}
/*
* 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
*/
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;
/*
* break out if we 'cannot' reduce the speed as the user might
* want freq_step to be zero
*/
if (cs_tuners->freq_step == 0)
return;
/* Check for frequency increase */
if (load > cs_tuners->up_threshold) {
dbs_info->down_skip = 0;
/* if we are already at full speed then break out early */
if (dbs_info->requested_freq == policy->max)
return;
dbs_info->requested_freq += get_freq_target(cs_tuners, policy);
if (dbs_info->requested_freq > policy->max)
dbs_info->requested_freq = policy->max;
__cpufreq_driver_target(policy, dbs_info->requested_freq,
CPUFREQ_RELATION_H);
return;
}
/* 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;
/* 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;
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;
__cpufreq_driver_target(policy, dbs_info->requested_freq,
CPUFREQ_RELATION_L);
return;
}
}
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);
}
static int dbs_cpufreq_notifier(struct notifier_block *nb, unsigned long val,
void *data)
{
struct cpufreq_freqs *freq = data;
struct cs_cpu_dbs_info_s *dbs_info =
&per_cpu(cs_cpu_dbs_info, freq->cpu);
struct cpufreq_policy *policy;
if (!dbs_info->enable)
return 0;
policy = dbs_info->cdbs.cur_policy;
/*
* 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 (dbs_info->requested_freq > policy->max
|| dbs_info->requested_freq < policy->min)
dbs_info->requested_freq = freq->new;
return 0;
}
static struct notifier_block cs_cpufreq_notifier_block = {
.notifier_call = dbs_cpufreq_notifier,
};
/************************** sysfs interface ************************/
static struct common_dbs_data cs_dbs_cdata;
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);
if (ret != 1 || input > MAX_SAMPLING_DOWN_FACTOR || input < 1)
return -EINVAL;
cs_tuners->sampling_down_factor = input;
return 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);
if (ret != 1)
return -EINVAL;
cs_tuners->sampling_rate = max(input, dbs_data->min_sampling_rate);
return 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);
if (ret != 1 || input > 100 || input <= cs_tuners->down_threshold)
return -EINVAL;
cs_tuners->up_threshold = input;
return 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);
/* cannot be lower than 11 otherwise freq will not fall */
if (ret != 1 || input < 11 || input > 100 ||
input >= cs_tuners->up_threshold)
return -EINVAL;
cs_tuners->down_threshold = input;
return count;
}
static ssize_t store_ignore_nice_load(struct dbs_data *dbs_data,
const char *buf, size_t count)
{
struct cs_dbs_tuners *cs_tuners = dbs_data->tuners;
unsigned int input, j;
int ret;
ret = sscanf(buf, "%u", &input);
if (ret != 1)
return -EINVAL;
if (input > 1)
input = 1;
if (input == cs_tuners->ignore_nice_load) /* nothing to do */
return count;
cs_tuners->ignore_nice_load = input;
/* we need to re-evaluate prev_cpu_idle */
for_each_online_cpu(j) {
struct cs_cpu_dbs_info_s *dbs_info;
dbs_info = &per_cpu(cs_cpu_dbs_info, j);
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];
}
return 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);
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 :)
*/
cs_tuners->freq_step = input;
return count;
}
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 cs_attr_group_gov_sys = {
.attrs = dbs_attributes_gov_sys,
.name = "conservative",
};
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 cs_attr_group_gov_pol = {
.attrs = dbs_attributes_gov_pol,
.name = "conservative",
};
/************************** sysfs end ************************/
static int cs_init(struct dbs_data *dbs_data, bool notify)
{
struct cs_dbs_tuners *tuners;
tuners = kzalloc(sizeof(*tuners), GFP_KERNEL);
if (!tuners) {
pr_err("%s: kzalloc failed\n", __func__);
return -ENOMEM;
}
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;
dbs_data->tuners = tuners;
dbs_data->min_sampling_rate = MIN_SAMPLING_RATE_RATIO *
jiffies_to_usecs(10);
if (notify)
cpufreq_register_notifier(&cs_cpufreq_notifier_block,
CPUFREQ_TRANSITION_NOTIFIER);
return 0;
}
static void cs_exit(struct dbs_data *dbs_data, bool notify)
{
if (notify)
cpufreq_unregister_notifier(&cs_cpufreq_notifier_block,
CPUFREQ_TRANSITION_NOTIFIER);
kfree(dbs_data->tuners);
}
define_get_cpu_dbs_routines(cs_cpu_dbs_info);
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,
.init = cs_init,
.exit = cs_exit,
.mutex = __MUTEX_INITIALIZER(cs_dbs_cdata.mutex),
};
static int cs_cpufreq_governor_dbs(struct cpufreq_policy *policy,
unsigned int event)
{
return cpufreq_governor_dbs(policy, &cs_dbs_cdata, event);
}
#ifndef CONFIG_CPU_FREQ_DEFAULT_GOV_CONSERVATIVE
static
#endif
struct cpufreq_governor cpufreq_gov_conservative = {
.name = "conservative",
.governor = cs_cpufreq_governor_dbs,
.max_transition_latency = TRANSITION_LATENCY_LIMIT,
.owner = THIS_MODULE,
};
static int __init cpufreq_gov_dbs_init(void)
{
return cpufreq_register_governor(&cpufreq_gov_conservative);
}
static void __exit cpufreq_gov_dbs_exit(void)
{
cpufreq_unregister_governor(&cpufreq_gov_conservative);
}
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);