On Fri, May 14, 2021 at 4:42 AM Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
>
> On Thu, May 13, 2021 at 10:53:49AM -0700, Suren Baghdasaryan wrote:
>
> > +bool cgroup_psi_enabled(void)
> > +{
> > + return (cgroup_feature_disable_mask & (1 << OPT_FEATURE_PRESSURE)) == 0;
> > +}
>
> > diff --git a/kernel/sched/psi.c b/kernel/sched/psi.c
> > index cc25a3cff41f..c73efd7d4fba 100644
> > --- a/kernel/sched/psi.c
> > +++ b/kernel/sched/psi.c
> > @@ -747,9 +747,12 @@ static struct psi_group *iterate_groups(struct task_struct *task, void **iter)
> > #ifdef CONFIG_CGROUPS
> > struct cgroup *cgroup = NULL;
> >
> > - if (!*iter)
> > + if (!*iter) {
> > + /* Skip to psi_system if per-cgroup accounting is disabled */
> > + if (!cgroup_psi_enabled())
> > + goto update_sys;
> > cgroup = task->cgroups->dfl_cgrp;
> > - else if (*iter == &psi_system)
> > + } else if (*iter == &psi_system)
> > return NULL;
> > else
> > cgroup = cgroup_parent(*iter);
> > @@ -758,6 +761,7 @@ static struct psi_group *iterate_groups(struct task_struct *task, void **iter)
> > *iter = cgroup;
> > return cgroup_psi(cgroup);
> > }
> > +update_sys:
> > #else
> > if (*iter)
> > return NULL;
>
> I'm confused; shouldn't that do the same as that #else branch?
Correct, for this function CONFIG_CGROUPS=n and
cgroup_disable=pressure are treated the same. True, from the code it's
not very obvious. Do you have some refactoring in mind that would make
it more explicit?
>Also, can you pretty please make cgroup_psi_enabled() a static_key ?
Certainly, will post an update on Monday.
Thanks for the feedback, Peter!
>
> --
> To unsubscribe from this group and stop receiving emails from it, send an email to kernel-team+unsubscribe@xxxxxxxxxxx.
>
