Commit 36131fdc authored May 15, 2014 by Kaushal Kumar Committed by Runmin Wang Aug 02, 2016

sched: Remove synchronize rcu/sched calls from _cpu_down



There is no need for sync_sched() in _cpu_down as stop_machine()
provides that barrier implicitly. Removing it also helps improve
hot-unplug latency.

The sync_sched/rcu were earlier removed for the same reason by the
commit 9ee349ad ("sched: Fix set_cpu_active() in cpu_down()"),
but recently got added as part of commit <6acce3ef> ("sched:
Remove get_online_cpus() usage.").

CRs-Fixed: 1048941
Change-Id: I97763004454d082d3cc2d9d9dbef7da923608600
Signed-off-by: Kaushal Kumar <kaushalk@codeaurora.org>
Signed-off-by: Matt Wagantall <mattw@codeaurora.org>
Signed-off-by: Runmin Wang <runminw@codeaurora.org>

parent 2566204c

kernel/cpu.c

+0 −15

Original line number	Diff line number	Diff line
		@@ -365,21 +365,6 @@ static int _cpu_down(unsigned int cpu, int tasks_frozen)
		goto out_release;
		}

		/*
		* By now we've cleared cpu_active_mask, wait for all preempt-disabled
		* and RCU users of this state to go away such that all new such users
		* will observe it.
		*
		* For CONFIG_PREEMPT we have preemptible RCU and its sync_rcu() might
		* not imply sync_sched(), so wait for both.
		*
		* Do sync before park smpboot threads to take care the rcu boost case.
		*/
		if (IS_ENABLED(CONFIG_PREEMPT))
		synchronize_rcu_mult(call_rcu, call_rcu_sched);
		else
		synchronize_rcu();

		smpboot_park_threads(cpu);

		/*