Re: [Xen-devel] [PATCH v3 33/47] xen/sched: add a percpu resource index

[Jan Beulich <jbeulich@xxxxxxxx>]

On 14.09.2019 10:52, Juergen Gross wrote:
> --- a/xen/common/schedule.c
> +++ b/xen/common/schedule.c
> @@ -71,6 +71,7 @@ static void poll_timer_fn(void *data);
>  /* This is global for now so that private implementations can reach it */
>  DEFINE_PER_CPU(struct scheduler *, scheduler);
>  DEFINE_PER_CPU_READ_MOSTLY(struct sched_resource *, sched_res);
> +static DEFINE_PER_CPU_READ_MOSTLY(unsigned int, sched_res_idx);

It's of course not very helpful that this variable (right here) doesn't
ever get written to, and hence one can't judge where / how this is to
be done (without going look elsewhere). But I guess that calculation
can't be moved into this patch (accepting that it would always yield
zero for now)?

> @@ -135,6 +136,12 @@ static struct scheduler sched_idle_ops = {
>      .switch_sched   = sched_idle_switch_sched,
>  };
>  
> +static inline struct vcpu *sched_unit2vcpu_cpu(struct sched_unit *unit,

const (on the parameter)?

> +                                               unsigned int cpu)
> +{
> +    return unit->domain->vcpu[unit->unit_id + per_cpu(sched_res_idx, cpu)];
> +}
> +
>  static inline struct scheduler *dom_scheduler(const struct domain *d)
>  {
>      if ( likely(d->cpupool != NULL) )
> @@ -2008,7 +2015,7 @@ static void sched_slave(void)
>  
>      pcpu_schedule_unlock_irq(lock, cpu);
>  
> -    sched_context_switch(vprev, next->vcpu_list, now);
> +    sched_context_switch(vprev, sched_unit2vcpu_cpu(next, cpu), now);
>  }

By the end of the series this will be

    sched_context_switch(vprev, sched_unit2vcpu_cpu(next, cpu),
                         is_idle_unit(next) && !is_idle_unit(prev), now);

clarifying (I think) that "next" can indeed be an idle unit. I'm having
difficulty seeing how can produce the correct result in all three cases
- all vCPU-s in the unit belong to a real domain
- all vCPU-s in the unit belong to the idle domain
- there's a mix of "real" and (filler) "idle" vCPU-s
My main question is what "next" is in the last of the 3 cases above.
Considering that scheduling happens in terms of units, I'd expect it to
be a real domain's unit, yet then I don't see how you'd get at the
correct (idle) vCPU when calling sched_unit2vcpu_cpu(next, cpu).

Jan

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel