2018-02-06 02:00:03

by Wen Yang

Subject: [PATCH] sched/deadline: Make update_curr_dl() more accurate

rq->clock_task may be updated between the two calls of
rq_clock_task() in update_curr_dl(). Calling rq_clock_task() only
once makes it more accurate and efficient, taking update_curr() as
reference.

Signed-off-by: Wen Yang <[email protected]>
Reviewed-by: Jiang Biao <[email protected]>
Suggested-by: Peter Zijlstra <[email protected]>
---
kernel/sched/deadline.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
index 2473736..a32209c 100644
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -1115,6 +1115,7 @@ static void update_curr_dl(struct rq *rq)
struct task_struct *curr = rq->curr;
struct sched_dl_entity *dl_se = &curr->dl;
u64 delta_exec;
+ u64 now;

if (!dl_task(curr) || !on_dl_rq(dl_se))
return;
@@ -1127,7 +1128,8 @@ static void update_curr_dl(struct rq *rq)
* natural solution, but the full ramifications of this
* approach need further study.
*/
- delta_exec = rq_clock_task(rq) - curr->se.exec_start;
+ now = rq_clock_task(rq);
+ delta_exec = now - curr->se.exec_start;
if (unlikely((s64)delta_exec <= 0)) {
if (unlikely(dl_se->dl_yielded))
goto throttle;
@@ -1143,7 +1145,7 @@ static void update_curr_dl(struct rq *rq)
curr->se.sum_exec_runtime += delta_exec;
account_group_exec_runtime(curr, delta_exec);

- curr->se.exec_start = rq_clock_task(rq);
+ curr->se.exec_start = now;
cgroup_account_cputime(curr, delta_exec);

sched_rt_avg_update(rq, delta_exec);
--
1.8.3.1

2018-02-06 07:28:02

by Juri Lelli

Subject: Re: [PATCH] sched/deadline: Make update_curr_dl() more accurate

Hi,

On 06/02/18 09:55, Wen Yang wrote:
> rq->clock_task may be updated between the two calls of
> rq_clock_task() in update_curr_dl(). Calling rq_clock_task() only
> once makes it more accurate and efficient, taking update_curr() as
> reference.
>
> Signed-off-by: Wen Yang <[email protected]>
> Reviewed-by: Jiang Biao <[email protected]>
> Suggested-by: Peter Zijlstra <[email protected]>

Acked-by: Juri Lelli <[email protected]>

Thanks!

- Juri

2018-02-13 11:29:41

by tip-bot for Vasyl Gomonovych

Subject: [tip:sched/urgent] sched/deadline: Make update_curr_dl() more accurate

Commit-ID: 6fe0ce1eb04f99a1eb1eb6e7f775666966cf6c80
Gitweb: https://git.kernel.org/tip/6fe0ce1eb04f99a1eb1eb6e7f775666966cf6c80
Author: Wen Yang <[email protected]>
AuthorDate: Tue, 6 Feb 2018 09:55:48 +0800
Committer: Ingo Molnar <[email protected]>
CommitDate: Tue, 13 Feb 2018 11:44:40 +0100

sched/deadline: Make update_curr_dl() more accurate

rq->clock_task may be updated between the two calls of
rq_clock_task() in update_curr_dl(). Calling rq_clock_task() only
once makes it more accurate and efficient, taking update_curr() as
reference.

Suggested-by: Peter Zijlstra <[email protected]>
Signed-off-by: Wen Yang <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Reviewed-by: Jiang Biao <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Cc: [email protected]
Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Ingo Molnar <[email protected]>
---
kernel/sched/deadline.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
index 9bb0e0c..9df0978 100644
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -1153,6 +1153,7 @@ static void update_curr_dl(struct rq *rq)
struct sched_dl_entity *dl_se = &curr->dl;
u64 delta_exec, scaled_delta_exec;
int cpu = cpu_of(rq);
+ u64 now;

if (!dl_task(curr) || !on_dl_rq(dl_se))
return;
@@ -1165,7 +1166,8 @@ static void update_curr_dl(struct rq *rq)
* natural solution, but the full ramifications of this
* approach need further study.
*/
- delta_exec = rq_clock_task(rq) - curr->se.exec_start;
+ now = rq_clock_task(rq);
+ delta_exec = now - curr->se.exec_start;
if (unlikely((s64)delta_exec <= 0)) {
if (unlikely(dl_se->dl_yielded))
goto throttle;
@@ -1178,7 +1180,7 @@ static void update_curr_dl(struct rq *rq)
curr->se.sum_exec_runtime += delta_exec;
account_group_exec_runtime(curr, delta_exec);

- curr->se.exec_start = rq_clock_task(rq);
+ curr->se.exec_start = now;
cgroup_account_cputime(curr, delta_exec);

sched_rt_avg_update(rq, delta_exec);