This patch collect CPUs load when cpufreq governos is performance.
The collected CPUs load is used for providing data through debugfs file.
- /sys/kernel/debug/cpufreq/cpuX/load_table
And this patch create basic sysfs file as below:
- /sys/devices/system/cpu/cpufreq/performance/ignore_nice (rw)
- /sys/devices/system/cpu/cpufreq/performance/sampling_rate (rw)
- /sys/devices/system/cpu/cpufreq/performance/sampling_rate_min (r)
Signed-off-by: Chanwoo Choi <cw00.choi@xxxxxxxxxxx>
Signed-off-by: Kyungmin Park <kyungmin.park@xxxxxxxxxxx>
Signed-off-by: Myungjoo Ham <myungjoo.ham@xxxxxxxxxxx>
---
drivers/cpufreq/cpufreq_governor.h | 1 +
drivers/cpufreq/cpufreq_performance.c | 156 +++++++++++++++++++++++++++++++++-
2 files changed, 154 insertions(+), 3 deletions(-)
diff --git a/drivers/cpufreq/cpufreq_governor.h b/drivers/cpufreq/cpufreq_governor.h
index 55ef8c6..7ad4d3b 100644
--- a/drivers/cpufreq/cpufreq_governor.h
+++ b/drivers/cpufreq/cpufreq_governor.h
@@ -198,6 +198,7 @@ struct common_dbs_data {
/* Common across governors */
#define GOV_ONDEMAND 0
#define GOV_CONSERVATIVE 1
+ #define GOV_PERFORMANCE 2
int governor;
struct attribute_group *attr_group_gov_sys; /* one governor - system */
struct attribute_group *attr_group_gov_pol; /* one governor - policy */
diff --git a/drivers/cpufreq/cpufreq_performance.c b/drivers/cpufreq/cpufreq_performance.c
index ceee068..15195fc9 100644
--- a/drivers/cpufreq/cpufreq_performance.c
+++ b/drivers/cpufreq/cpufreq_performance.c
@@ -17,10 +17,158 @@
#include <linux/cpufreq.h>
#include <linux/init.h>
+#ifdef CONFIG_CPU_FREQ_STAT
+
+#include <linux/kernel_stat.h>
+#include <linux/slab.h>
+
+#include "cpufreq_governor.h"
+
+#define pf_cpu_dbs_info_s cpu_dbs_info_s
+#define pf_dbs_tuners dbs_tuners
+
+static DEFINE_PER_CPU(struct pf_cpu_dbs_info_s, pf_cpu_dbs_info);
+static struct common_dbs_data pf_dbs_cdata;
+
+static ssize_t store_sampling_rate(struct dbs_data *dbs_data, const char *buf,
+ size_t count)
+{
+ struct pf_dbs_tuners *tuners = dbs_data->tuners;
+ unsigned int input;
+ int ret;
+ ret = sscanf(buf, "%u", &input);
+
+ if (ret != 1)
+ return -EINVAL;
+
+ tuners->sampling_rate = max(input, dbs_data->min_sampling_rate);
+ return count;
+}
+
+static ssize_t store_ignore_nice(struct dbs_data *dbs_data, const char *buf,
+ size_t count)
+{
+ struct pf_dbs_tuners *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 == tuners->ignore_nice) /* nothing to do */
+ return count;
+
+ tuners->ignore_nice = input;
+
+ /* we need to re-evaluate prev_cpu_idle */
+ for_each_online_cpu(j) {
+ struct pf_cpu_dbs_info_s *dbs_info;
+ dbs_info = &per_cpu(pf_cpu_dbs_info, j);
+ dbs_info->cdbs.prev_cpu_idle = get_cpu_idle_time(j,
+ &dbs_info->cdbs.prev_cpu_wall, 0);
+ if (tuners->ignore_nice)
+ dbs_info->cdbs.prev_cpu_nice =
+ kcpustat_cpu(j).cpustat[CPUTIME_NICE];
+ }
+ return count;
+}
+
+show_store_one(pf, sampling_rate);
+show_store_one(pf, ignore_nice);
+declare_show_sampling_rate_min(pf);
+
+gov_sys_pol_attr_rw(sampling_rate);
+gov_sys_pol_attr_rw(ignore_nice);
+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,
+ &ignore_nice_gov_sys.attr,
+ NULL
+};
+
+static struct attribute_group pf_attr_group_gov_sys = {
+ .attrs = dbs_attributes_gov_sys,
+ .name = "performance",
+};
+
+static struct attribute *dbs_attributes_gov_pol[] = {
+ &sampling_rate_min_gov_pol.attr,
+ &sampling_rate_gov_pol.attr,
+ &ignore_nice_gov_pol.attr,
+ NULL
+};
+
+static struct attribute_group pf_attr_group_gov_pol = {
+ .attrs = dbs_attributes_gov_pol,
+ .name = "performance",
+};
+
+static void pf_dbs_timer(struct work_struct *work)
+{
+ struct pf_cpu_dbs_info_s *dbs_info = container_of(work,
+ struct pf_cpu_dbs_info_s, cdbs.work.work);
+ unsigned int cpu = dbs_info->cdbs.cur_policy->cpu;
+ struct dbs_data *dbs_data = dbs_info->cdbs.cur_policy->governor_data;
+ unsigned int delay = 0;
+
+ delay = usecs_to_jiffies(dbs_data->min_sampling_rate);
+
+ mutex_lock(&dbs_info->cdbs.timer_mutex);
+ dbs_check_cpu(dbs_data, cpu);
+ gov_queue_work(dbs_data, dbs_info->cdbs.cur_policy, delay, false);
+ mutex_unlock(&dbs_info->cdbs.timer_mutex);
+}
+
+static int pf_init(struct dbs_data *dbs_data)
+{
+ struct pf_dbs_tuners *tuners;
+
+ tuners = kzalloc(sizeof(struct pf_dbs_tuners), GFP_KERNEL);
+ if (!tuners) {
+ pr_err("%s: kzalloc failed\n", __func__);
+ return -ENOMEM;
+ }
+
+ tuners->ignore_nice = 0;
+
+ dbs_data->tuners = tuners;
+ dbs_data->min_sampling_rate = MIN_SAMPLING_RATE_RATIO *
+ jiffies_to_usecs(100);
+
+ mutex_init(&dbs_data->mutex);
+
+ return 0;
+}
+static void pf_exit(struct dbs_data *dbs_data)
+{
+ kfree(dbs_data->tuners);
+}
+
+define_get_cpu_dbs_routines(pf_cpu_dbs_info);
+
+static struct common_dbs_data pf_dbs_cdata = {
+ .governor = GOV_PERFORMANCE,
+ .attr_group_gov_sys = &pf_attr_group_gov_sys,
+ .attr_group_gov_pol = &pf_attr_group_gov_pol,
+ .get_cpu_cdbs = get_cpu_cdbs,
+ .get_cpu_dbs_info_s = get_cpu_dbs_info_s,
+ .gov_dbs_timer = pf_dbs_timer,
+ .init = pf_init,
+ .exit = pf_exit,
+};
+#endif /* CONFIG_CPU_FREQ_STAT */
static int cpufreq_governor_performance(struct cpufreq_policy *policy,
unsigned int event)
{
+ int ret = 0;
+
switch (event) {
case CPUFREQ_GOV_START:
case CPUFREQ_GOV_LIMITS:
@@ -29,10 +177,12 @@ static int cpufreq_governor_performance(struct cpufreq_policy *policy,
__cpufreq_driver_target(policy, policy->max,
CPUFREQ_RELATION_H);
break;
- default:
- break;
}
- return 0;
+
+#ifdef CONFIG_CPU_FREQ_STAT
+ ret = cpufreq_governor_dbs(policy, &pf_dbs_cdata, event);
+#endif
+ return ret;
}
#ifdef CONFIG_CPU_FREQ_GOV_PERFORMANCE_MODULE
--
1.8.0
--
To unsubscribe from this list: send the line "unsubscribe cpufreq" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
