When we turn on CONFIG_SCHEDSTATS, per-task cpu runtime is accumulated
twice. Once in task->se.sum_exec_runtime and once in sched_info.cpu_time.
These two stats are exactly the same.
Given that task->se.sum_exec_runtime is always accumulated by the core
scheduler, sched_info can reuse that data instead of duplicate the accounting.
Signed-off-by: Ken Chen <[email protected]>
diff --git a/fs/proc/base.c b/fs/proc/base.c
index d467760..4d745ba 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -347,7 +347,7 @@
static int proc_pid_schedstat(struct task_struct *task, char *buffer)
{
return sprintf(buffer, "%llu %llu %lu\n",
- task->sched_info.cpu_time,
+ task->se.sum_exec_runtime,
task->sched_info.run_delay,
task->sched_info.pcount);
}
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 55e30d1..a0fa340 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -672,8 +672,7 @@ struct reclaim_state;
struct sched_info {
/* cumulative counters */
unsigned long pcount; /* # of times run on this cpu */
- unsigned long long cpu_time, /* time spent on the cpu */
- run_delay; /* time spent waiting on a runqueue */
+ unsigned long long run_delay; /* time spent waiting on a runqueue */
/* timestamps */
unsigned long long last_arrival,/* when we last ran on a cpu */
diff --git a/kernel/delayacct.c b/kernel/delayacct.c
index b3179da..abb6e17 100644
--- a/kernel/delayacct.c
+++ b/kernel/delayacct.c
@@ -127,7 +127,7 @@ int __delayacct_add_tsk
*/
t1 = tsk->sched_info.pcount;
t2 = tsk->sched_info.run_delay;
- t3 = tsk->sched_info.cpu_time;
+ t3 = tsk->se.sum_exec_runtime;
d->cpu_count += t1;
diff --git a/kernel/sched.c b/kernel/sched.c
index 4d1920d..1ed1809 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -586,6 +586,8 @@ struct rq {
#ifdef CONFIG_SCHEDSTATS
/* latency stats */
struct sched_info rq_sched_info;
+ unsigned long long rq_cpu_time;
+ /* could above be rq->cfs_rq.exec_clock + rq->rt_rq.rt_runtime ? */
/* sys_sched_yield() stats */
unsigned int yld_exp_empty;
diff --git a/kernel/sched_stats.h b/kernel/sched_stats.h
index 4272594..b07eee8 100644
--- a/kernel/sched_stats.h
+++ b/kernel/sched_stats.h
@@ -31,7 +31,7 @@ static int show_schedstat(struct seq_file *seq, void *v)
rq->yld_act_empty, rq->yld_exp_empty, rq->yld_count,
rq->sched_switch, rq->sched_count, rq->sched_goidle,
rq->ttwu_count, rq->ttwu_local,
- rq->rq_sched_info.cpu_time,
+ rq->rq_cpu_time,
rq->rq_sched_info.run_delay, rq->rq_sched_info.pcount);
seq_printf(seq, "\n");
@@ -123,7 +123,7 @@ static inline void
rq_sched_info_depart(struct rq *rq, unsigned long long delta)
{
if (rq)
- rq->rq_sched_info.cpu_time += delta;
+ rq->rq_cpu_time += delta;
}
static inline void
@@ -238,7 +238,6 @@ static inline void sched_info_depart
unsigned long long delta = task_rq(t)->clock -
t->sched_info.last_arrival;
- t->sched_info.cpu_time += delta;
rq_sched_info_depart(task_rq(t), delta);
if (t->state == TASK_RUNNING)
On Tue, 2008-12-16 at 23:41 -0800, Ken Chen wrote:
> When we turn on CONFIG_SCHEDSTATS, per-task cpu runtime is accumulated
> twice. Once in task->se.sum_exec_runtime and once in sched_info.cpu_time.
> These two stats are exactly the same.
>
> Given that task->se.sum_exec_runtime is always accumulated by the core
> scheduler, sched_info can reuse that data instead of duplicate the accounting.
>
> Signed-off-by: Ken Chen <[email protected]>
Looks good, thanks!
> diff --git a/fs/proc/base.c b/fs/proc/base.c
> index d467760..4d745ba 100644
> --- a/fs/proc/base.c
> +++ b/fs/proc/base.c
> @@ -347,7 +347,7 @@
> static int proc_pid_schedstat(struct task_struct *task, char *buffer)
> {
> return sprintf(buffer, "%llu %llu %lu\n",
> - task->sched_info.cpu_time,
> + task->se.sum_exec_runtime,
> task->sched_info.run_delay,
> task->sched_info.pcount);
> }
> diff --git a/include/linux/sched.h b/include/linux/sched.h
> index 55e30d1..a0fa340 100644
> --- a/include/linux/sched.h
> +++ b/include/linux/sched.h
> @@ -672,8 +672,7 @@ struct reclaim_state;
> struct sched_info {
> /* cumulative counters */
> unsigned long pcount; /* # of times run on this cpu */
> - unsigned long long cpu_time, /* time spent on the cpu */
> - run_delay; /* time spent waiting on a runqueue */
> + unsigned long long run_delay; /* time spent waiting on a runqueue */
>
> /* timestamps */
> unsigned long long last_arrival,/* when we last ran on a cpu */
> diff --git a/kernel/delayacct.c b/kernel/delayacct.c
> index b3179da..abb6e17 100644
> --- a/kernel/delayacct.c
> +++ b/kernel/delayacct.c
> @@ -127,7 +127,7 @@ int __delayacct_add_tsk
> */
> t1 = tsk->sched_info.pcount;
> t2 = tsk->sched_info.run_delay;
> - t3 = tsk->sched_info.cpu_time;
> + t3 = tsk->se.sum_exec_runtime;
>
> d->cpu_count += t1;
>
> diff --git a/kernel/sched.c b/kernel/sched.c
> index 4d1920d..1ed1809 100644
> --- a/kernel/sched.c
> +++ b/kernel/sched.c
> @@ -586,6 +586,8 @@ struct rq {
> #ifdef CONFIG_SCHEDSTATS
> /* latency stats */
> struct sched_info rq_sched_info;
> + unsigned long long rq_cpu_time;
> + /* could above be rq->cfs_rq.exec_clock + rq->rt_rq.rt_runtime ? */
>
> /* sys_sched_yield() stats */
> unsigned int yld_exp_empty;
> diff --git a/kernel/sched_stats.h b/kernel/sched_stats.h
> index 4272594..b07eee8 100644
> --- a/kernel/sched_stats.h
> +++ b/kernel/sched_stats.h
> @@ -31,7 +31,7 @@ static int show_schedstat(struct seq_file *seq, void *v)
> rq->yld_act_empty, rq->yld_exp_empty, rq->yld_count,
> rq->sched_switch, rq->sched_count, rq->sched_goidle,
> rq->ttwu_count, rq->ttwu_local,
> - rq->rq_sched_info.cpu_time,
> + rq->rq_cpu_time,
> rq->rq_sched_info.run_delay, rq->rq_sched_info.pcount);
>
> seq_printf(seq, "\n");
> @@ -123,7 +123,7 @@ static inline void
> rq_sched_info_depart(struct rq *rq, unsigned long long delta)
> {
> if (rq)
> - rq->rq_sched_info.cpu_time += delta;
> + rq->rq_cpu_time += delta;
> }
>
> static inline void
> @@ -238,7 +238,6 @@ static inline void sched_info_depart
> unsigned long long delta = task_rq(t)->clock -
> t->sched_info.last_arrival;
>
> - t->sched_info.cpu_time += delta;
> rq_sched_info_depart(task_rq(t), delta);
>
> if (t->state == TASK_RUNNING)