Re: [PATCH v2 07/13] sched/fair: Let asymmetric cpu configurations balance at wake-up

[Morten Rasmussen]

On Wed, Jul 13, 2016 at 02:56:41PM +0200, Vincent Guittot wrote:
> On 22 June 2016 at 19:03, Morten Rasmussen <morten.rasmussen@xxxxxxx> wrote:
> > Currently, SD_WAKE_AFFINE always takes priority over wakeup balancing if
> > SD_BALANCE_WAKE is set on the sched_domains. For asymmetric
> > configurations SD_WAKE_AFFINE is only desirable if the waking task's
> > compute demand (utilization) is suitable for all the cpu capacities
> > available within the SD_WAKE_AFFINE sched_domain. If not, let wakeup
> 
> instead of "suitable for all the cpu capacities available within the
> SD_WAKE_AFFINE sched_domain", should it be "suitable for local cpu and
> prev cpu" becasue you only check the capacity of these 2 CPUs.

Good point. I currently make the implicit assumption that capacity of local cpu
and prev cpu represent the capacity for all cpus their SD_WAKE_AFFINE
domains. It breaks if you should choose to have SD_WAKE_AFFINE on a
domain that spans both little and big cpus, as if local/prev cpu happens
to be big we assume that they are all big and let select_idle_sibling()
handle the task placement even for big tasks if local/prev cpu are both
big.

I don't see why anybody would want that kind of setup, but I think the
assumption should still be written down somewhere, either here or in a
comment in wake_cap() or both.

The next paragraph in the commit message mentions that we actually only
check waker cpu and prev_cpu capacity. Would it be more clear if we
extend that to something like:

    This patch makes affine wake-ups conditional on whether both the waker
    cpu and prev_cpu has sufficient capacity for the waking task, or
    not, assuming that the cpu capacities within an SD_WAKE_AFFINE
    domain are homogeneous.

Thoughts?

> 
> Other than this comment for the commit message, the patch looks good to me
> Acked-by: Vincent Guittot <vincent.guittot@xxxxxxxxxx>

Thanks,
Morten


> 
> > balancing take over (find_idlest_{group, cpu}()).
> >
> > This patch makes affine wake-ups conditional on whether both the waker
> > cpu and prev_cpu has sufficient capacity for the waking task, or not.
> >
> > It is assumed that the sched_group(s) containing the waker cpu and
> > prev_cpu only contain cpu with the same capacity (homogeneous).
> >
> > Ideally, we shouldn't set 'want_affine' in the first place, but we don't
> > know if SD_BALANCE_WAKE is enabled on the sched_domain(s) until we start
> > traversing them.
> >
> > cc: Ingo Molnar <mingo@xxxxxxxxxx>
> > cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> >
> > Signed-off-by: Morten Rasmussen <morten.rasmussen@xxxxxxx>
> > ---
> >  kernel/sched/fair.c | 28 +++++++++++++++++++++++++++-
> >  1 file changed, 27 insertions(+), 1 deletion(-)
> >
> > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > index 216db302e87d..dba02c7b57b3 100644
> > --- a/kernel/sched/fair.c
> > +++ b/kernel/sched/fair.c
> > @@ -114,6 +114,12 @@ unsigned int __read_mostly sysctl_sched_shares_window = 10000000UL;
> >  unsigned int sysctl_sched_cfs_bandwidth_slice = 5000UL;
> >  #endif
> >
> > +/*
> > + * The margin used when comparing utilization with cpu capacity:
> > + * util * 1024 < capacity * margin
> > + */
> > +unsigned int capacity_margin = 1280; /* ~20% */
> > +
> >  static inline void update_load_add(struct load_weight *lw, unsigned long inc)
> >  {
> >         lw->weight += inc;
> > @@ -5260,6 +5266,25 @@ static int cpu_util(int cpu)
> >         return (util >= capacity) ? capacity : util;
> >  }
> >
> > +static inline int task_util(struct task_struct *p)
> > +{
> > +       return p->se.avg.util_avg;
> > +}
> > +
> > +static int wake_cap(struct task_struct *p, int cpu, int prev_cpu)
> > +{
> > +       long min_cap, max_cap;
> > +
> > +       min_cap = min(capacity_orig_of(prev_cpu), capacity_orig_of(cpu));
> > +       max_cap = cpu_rq(cpu)->rd->max_cpu_capacity;
> > +
> > +       /* Minimum capacity is close to max, no need to abort wake_affine */
> > +       if (max_cap - min_cap < max_cap >> 3)
> > +               return 0;
> > +
> > +       return min_cap * 1024 < task_util(p) * capacity_margin;
> > +}
> > +
> >  /*
> >   * select_task_rq_fair: Select target runqueue for the waking task in domains
> >   * that have the 'sd_flag' flag set. In practice, this is SD_BALANCE_WAKE,
> > @@ -5283,7 +5308,8 @@ select_task_rq_fair(struct task_struct *p, int prev_cpu, int sd_flag, int wake_f
> >
> >         if (sd_flag & SD_BALANCE_WAKE) {
> >                 record_wakee(p);
> > -               want_affine = !wake_wide(p) && cpumask_test_cpu(cpu, tsk_cpus_allowed(p));
> > +               want_affine = !wake_wide(p) && !wake_cap(p, cpu, prev_cpu)
> > +                             && cpumask_test_cpu(cpu, tsk_cpus_allowed(p));
> >         }
> >
> >         rcu_read_lock();
> > --
> > 1.9.1
> >