Re: [PATCH 2/4] sched/fair: Move avg_scan_cost calculations under SIS_PROP

[Mel Gorman]

On Tue, Dec 08, 2020 at 05:03:21PM +0100, Vincent Guittot wrote:
> On Tue, 8 Dec 2020 at 16:35, Mel Gorman <mgorman@xxxxxxxxxxxxxxxxxxx> wrote:
> >
> > As noted by Vincent Guittot, avg_scan_costs are calculated for SIS_PROP
> > even if SIS_PROP is disabled. Move the time calculations under a SIS_PROP
> > check and while we are at it, exclude the cost of initialising the CPU
> > mask from the average scan cost.
> >
> > Signed-off-by: Mel Gorman <mgorman@xxxxxxxxxxxxxxxxxxx>
> > ---
> >  kernel/sched/fair.c | 14 ++++++++------
> >  1 file changed, 8 insertions(+), 6 deletions(-)
> >
> > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > index ac7b34e7372b..5c41875aec23 100644
> > --- a/kernel/sched/fair.c
> > +++ b/kernel/sched/fair.c
> > @@ -6153,6 +6153,8 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, int t
> >         if (!this_sd)
> >                 return -1;
> 
> Just noticed while reviewing the patch that the above related to
> this_sd can also go under sched_feat(SIS_PROP)
> 

Technically yes but I also decided against it. It's a functional difference
depending on whether SIS_PROP is set in the highly unlikely case that
this_sd == NULL. I was also thinking in terms of what happens if SIS_PROP
was disabled and enabled while a search is in progress even if it's very
unlikely. In that case, this_sd would be uninitialised. That might be
impossible in practice depending on how static branching is implemented
but I don't think we should rely on the internals of jump labels and play
it safe. I can move it in if you feel strongly about it but I think the
disable/enable race is enough of a concern to leave it alone.

-- 
Mel Gorman
SUSE Labs