Re: [PATCH 3/3] hotplug: Optimize cpu_hotplug_{begin,done}() using rcu_sync

[Oleg Nesterov <oleg@redhat.com>]

On 10/03, Peter Zijlstra wrote:
>
> On Thu, Oct 03, 2013 at 09:48:38AM -0700, Paul E. McKenney wrote:
> > > -enum { readers_fast = 0, readers_slow, readers_block };
> > > +enum { readers_slow, readers_block };
> >
> > It took me a bit to realize that readers_fast is obsoleted by the
> > rcu_sync_is_idle() above.  ;-)
>
> Yeah.. I pondered changing/adding to the rcu_sync interface to allow
> using gp_count like status to avoid the extra variable, but decided
> against it for now.

Agreed, it looks simple enough even if get/put has to read ->gp_state
or/and cpuhp_state.

But, just is case, this is one example of why it probably makes sense
to rcu_sync_struct->state_changed(new_state, void *data) callback. In
this case it could simply do

	static void cpuhp_rcu_sync_state_cb(state, ...)
	{
		switch (state) {
		case GP_IDLE:
			cpuhp_state = readers_fast;
			break;
		case GP_PENDING:
			cpuhp_state = readers_slow;
			break;
		}
	}

Doesn't make sense in this particular case, but perhaps in general.

Oleg.