Re: [PATCH v8 5/9] sched/fair: Take into account latency priority at wakeup

From: Vincent Guittot
Date: Tue Nov 15 2022 - 10:40:50 EST


On Mon, 14 Nov 2022 at 17:20, Patrick Bellasi
<patrick.bellasi@xxxxxxxxxx> wrote:
>
> Hi Vincent,
>
> On 10-Nov 18:50, Vincent Guittot wrote:
>
> [...]
>
> > diff --git a/init/init_task.c b/init/init_task.c
> > index 7dd71dd2d261..b8ddf403bc62 100644
> > --- a/init/init_task.c
> > +++ b/init/init_task.c
> > @@ -78,7 +78,7 @@ struct task_struct init_task
> > .prio = MAX_PRIO - 20,
> > .static_prio = MAX_PRIO - 20,
> > .normal_prio = MAX_PRIO - 20,
> > - .latency_nice = DEFAULT_LATENCY_NICE,
> > + .latency_prio = NICE_WIDTH - 20,
> ^^^^^^^^^^
>
> For robustness/consistency, shoudln't this be LATENCY_NICE_WIDTH?

yes, good catch

>
> [...]
>
> > diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> > index b2accc9da4fe..caf54e54a74f 100644
> > --- a/kernel/sched/core.c
> > +++ b/kernel/sched/core.c
> > @@ -1284,6 +1284,16 @@ static void set_load_weight(struct task_struct *p, bool update_load)
> > }
> > }
> >
> > +static void set_latency_offset(struct task_struct *p)
> > +{
> > + long weight = sched_latency_to_weight[p->latency_prio];
> > + s64 offset;
> > +
> > + offset = weight * get_sched_latency(false);
> ^^^^^^^^^^^^^^^^^^^^^^^^
> As per my comment in patch 1, we almost always (but one time) call this with
> "false" and that's not returning the sysctl_sched_latency but a possibly
> discounted value in case of feat(GENTLE_FAIR_SLEEPERS).
>
> Just to avoid confusion (this could be not the sched_latency) and to better
> document the code, what about using a accessor define something like e.g.
>
> #define max_wakeup_latency get_wakeup_latency(false)
>
> ?
>
> [...]
>
> Best,
> Patrick
>
> --
> #include <best/regards.h>
>
> Patrick Bellasi