Commit 677ab23b authored by Frederic Weisbecker's avatar Frederic Weisbecker Committed by Paul E. McKenney
Browse files

rcu/exp: Remove redundant full memory barrier at the end of GP 



A full memory barrier is necessary at the end of the expedited grace
period to order:

1) The grace period completion (pictured by the GP sequence
   number) with all preceding accesses. This pairs with rcu_seq_end()
   performed by the concurrent kworker.

2) The grace period completion and subsequent post-GP update side
   accesses. Pairs again against rcu_seq_end().

This full barrier is already provided by the final sync_exp_work_done()
test, making the subsequent explicit one redundant. Remove it and
improve comments.

Signed-off-by: default avatarFrederic Weisbecker <frederic@kernel.org>
Signed-off-by: default avatarPaul E. McKenney <paulmck@kernel.org>
Reviewed-by: default avatarBoqun Feng <boqun.feng@gmail.com>
Reviewed-by: default avatarNeeraj Upadhyay <neeraj.upadhyay@kernel.org>
parent 55911a9f

kernel/rcu/tree_exp.h

+6 −2
Original line number Diff line number Diff line
@@ -265,7 +265,12 @@ static bool sync_exp_work_done(unsigned long s) 
{

	if (rcu_exp_gp_seq_done(s)) {

		trace_rcu_exp_grace_period(rcu_state.name, s, TPS("done"));

		smp_mb(); /* Ensure test happens before caller kfree(). */

		/*

		 * Order GP completion with preceding accesses. Order also GP

		 * completion with post GP update side accesses. Pairs with

		 * rcu_seq_end().

		 */

		smp_mb();

		return true;

	}

	return false;
@@ -967,7 +972,6 @@ void synchronize_rcu_expedited(void) 
	rnp = rcu_get_root();

	wait_event(rnp->exp_wq[rcu_seq_ctr(s) & 0x3],

		   sync_exp_work_done(s));

	smp_mb(); /* Work actions happen before return. */



	/* Let the next expedited grace period start. */

	mutex_unlock(&rcu_state.exp_mutex);