Re: [PATCH 04/19] sched: Prepare for Core-wide rq->lock

[Paul E. McKenney]

On Wed, Apr 28, 2021 at 01:03:29PM +0200, Peter Zijlstra wrote:
> On Wed, Apr 28, 2021 at 06:35:36PM +0800, Aubrey Li wrote:
> > On Wed, Apr 28, 2021 at 5:14 PM Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> 
> > > Ah, indeed so.. rq_lockp() could do with an assertion, not sure how to
> > > sanely do that. Anyway, double_rq_unlock() is simple enough to fix, we
> > > can simply flip the unlock()s.
> > >
> > > ( I'm suffering a cold and am really quite slow atm )
> > >
> > > How's this then?
> > >
> > > ---
> > > diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> > > index f732642e3e09..3a534c0c1c46 100644
> > > --- a/kernel/sched/core.c
> > > +++ b/kernel/sched/core.c
> > > @@ -290,6 +290,10 @@ static void sched_core_assert_empty(void)
> > >  static void __sched_core_enable(void)
> > >  {
> > >         static_branch_enable(&__sched_core_enabled);
> > > +       /*
> > > +        * Ensure raw_spin_rq_*lock*() have completed before flipping.
> > > +        */
> > > +       synchronize_sched();
> > 
> > synchronize_sched() seems no longer exist...
> 
> Bah.. Paul, why did that go away? I realize RCU merged in the sched and
> bh flavours, but I still find it expressive to use sync_sched() vs
> preempt_disable().

I could have made synchronize_sched() a synonym for synchronize_rcu(),
but that would be more likely to mislead than to help.

> Anyway, just use sync_rcu().

And yes, just use synchronize_rcu().

							Thanx, Paul