2022-07-13 04:06:54

by Chengming Zhou

Subject: [PATCH v2 06/10] sched/fair: delete superfluous SKIP_AGE_LOAD

There are three types of attach_entity_cfs_rq():

1. task migrate to CPU
2. task move to cgroup
3. task switched to fair from !fair

The case1 and case2 already have sched_avg last_update_time
reset to 0 when attach_entity_cfs_rq().

We will make case3 also have last_update_time reset to 0 when
attach_entity_cfs_rq() in the following patches.

So it makes no difference whether SKIP_AGE_LOAD is set or not.

Signed-off-by: Chengming Zhou <[email protected]>
---
kernel/sched/fair.c | 10 ++++------
1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 171bc22bc142..29811869c1fe 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -4001,9 +4001,8 @@ static void detach_entity_load_avg(struct cfs_rq *cfs_rq, struct sched_entity *s
* Optional action to be done while updating the load average
*/
#define UPDATE_TG 0x1
-#define SKIP_AGE_LOAD 0x2
-#define DO_ATTACH 0x4
-#define DO_DETACH 0x8
+#define DO_ATTACH 0x2
+#define DO_DETACH 0x4

/* Update task and its cfs_rq load average */
static inline void update_load_avg(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
@@ -4015,7 +4014,7 @@ static inline void update_load_avg(struct cfs_rq *cfs_rq, struct sched_entity *s
* Track task load average for carrying it to new CPU after migrated, and
* track group sched_entity load average for task_h_load calc in migration
*/
- if (se->avg.last_update_time && !(flags & SKIP_AGE_LOAD))
+ if (se->avg.last_update_time)
__update_load_avg_se(now, cfs_rq, se);

decayed = update_cfs_rq_load_avg(now, cfs_rq);
@@ -4298,7 +4297,6 @@ static inline bool cfs_rq_is_decayed(struct cfs_rq *cfs_rq)
}

#define UPDATE_TG 0x0
-#define SKIP_AGE_LOAD 0x0
#define DO_ATTACH 0x0
#define DO_DETACH 0x0

@@ -11540,7 +11538,7 @@ static void attach_entity_cfs_rq(struct sched_entity *se)
struct cfs_rq *cfs_rq = cfs_rq_of(se);

/* Synchronize entity with its cfs_rq */
- update_load_avg(cfs_rq, se, sched_feat(ATTACH_AGE_LOAD) ? 0 : SKIP_AGE_LOAD);
+ update_load_avg(cfs_rq, se, 0);
attach_entity_load_avg(cfs_rq, se);
update_tg_load_avg(cfs_rq);
propagate_entity_cfs_rq(se);
--
2.36.1

2022-07-14 13:04:40

by Dietmar Eggemann

Subject: Re: [PATCH v2 06/10] sched/fair: delete superfluous SKIP_AGE_LOAD

On 13/07/2022 06:04, Chengming Zhou wrote:
> There are three types of attach_entity_cfs_rq():
>
> 1. task migrate to CPU
> 2. task move to cgroup
> 3. task switched to fair from !fair
>
> The case1 and case2 already have sched_avg last_update_time
> reset to 0 when attach_entity_cfs_rq().
>
> We will make case3 also have last_update_time reset to 0 when
> attach_entity_cfs_rq() in the following patches.
>
> So it makes no difference whether SKIP_AGE_LOAD is set or not.

Wouldn't this patch make more sense after 09/10?

[...]

2022-07-14 14:11:01

by Chengming Zhou

Subject: Re: [External] Re: [PATCH v2 06/10] sched/fair: delete superfluous SKIP_AGE_LOAD

On 2022/7/14 20:33, Dietmar Eggemann wrote:
> On 13/07/2022 06:04, Chengming Zhou wrote:
>> There are three types of attach_entity_cfs_rq():
>>
>> 1. task migrate to CPU
>> 2. task move to cgroup
>> 3. task switched to fair from !fair
>>
>> The case1 and case2 already have sched_avg last_update_time
>> reset to 0 when attach_entity_cfs_rq().
>>
>> We will make case3 also have last_update_time reset to 0 when
>> attach_entity_cfs_rq() in the following patches.
>>
>> So it makes no difference whether SKIP_AGE_LOAD is set or not.
>
> Wouldn't this patch make more sense after 09/10?
>

Yes, this patch was put at last in the v1.

In this v2, the following patch 07/10 make attach_entity_cfs_rq() to use
update_load_avg(DO_ATTACH) to do conditional attach.

So I want to get rid of "sched_feat(ATTACH_AGE_LOAD) ? 0 : SKIP_AGE_LOAD"
earlier in this patch, or have to write like this:


int flags = UPDATE_TG | DO_ATTACH;

update_load_avg(cfs_rq, se, sched_feat(ATTACH_AGE_LOAD) ? flags : SKIP_AGE_LOAD | flags);


Thanks.

> [...]