min_vruntime update when a task is sleeping/migrating

min_vruntime update when a task is sleeping/migrating

[Pavan Kondeti]

Hi Peter,

The comment and the code around 2nd update_min_vruntime() call in
dequeue_entity() are not matching. If I understand commit b60205c7c558
("sched/fair: Fix min_vruntime tracking") correctly, the check is
inverted there. We want to update min_vruntime when a task is
sleeping/migrating. is my understanding right?

static void
dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
{

 <snip>
  /*
   * Now advance min_vruntime if @se was the entity holding it back,
   * except when: DEQUEUE_SAVE && !DEQUEUE_MOVE, in this case we'll be
   * put back on, and if we advance min_vruntime, we'll be placed back
   * further than we started -- ie. we'll be penalized.
   */
   if ((flags & (DEQUEUE_SAVE | DEQUEUE_MOVE)) == DEQUEUE_SAVE)
      update_min_vruntime(cfs_rq);
}

Thanks,
Pavan

-- 
Qualcomm India Private Limited, on behalf of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a
Linux Foundation Collaborative Project

Re: min_vruntime update when a task is sleeping/migrating

[Peter Zijlstra]

On Thu, Feb 23, 2017 at 11:45:26AM +0530, Pavan Kondeti wrote:
> Hi Peter,
> 
> The comment and the code around 2nd update_min_vruntime() call in
> dequeue_entity() are not matching. If I understand commit b60205c7c558
> ("sched/fair: Fix min_vruntime tracking") correctly, the check is
> inverted there. We want to update min_vruntime when a task is
> sleeping/migrating. is my understanding right?

Hurm, yes that comment and the code are not in agreement :/

Having gone over the code again, I think the comment is right and the
code is wrong, but it would be good to double check that.

> static void
> dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
> {
> 
>  <snip>
>   /*
>    * Now advance min_vruntime if @se was the entity holding it back,
>    * except when: DEQUEUE_SAVE && !DEQUEUE_MOVE, in this case we'll be
>    * put back on, and if we advance min_vruntime, we'll be placed back
>    * further than we started -- ie. we'll be penalized.
>    */
>    if ((flags & (DEQUEUE_SAVE | DEQUEUE_MOVE)) == DEQUEUE_SAVE)
>       update_min_vruntime(cfs_rq);
> }