rcu/rcuscale: Use call_rcu_hurry() for async reader test

rcuscale uses call_rcu() to queue async readers. With recent changes to
save power, the test will have fewer async readers in flight. Use the
call_rcu_hurry() API instead to revert to the old behavior.

[ paulmck: Apply s/call_rcu_flush/call_rcu_hurry/ feedback from Tejun Heo. ]

Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>

kernel/rcu/rcuscale.c

@@ -176,7 +176,7 @@ static struct rcu_scale_ops rcu_ops = {
 	.get_gp_seq	= rcu_get_gp_seq,
 	.gp_diff	= rcu_seq_diff,
 	.exp_completed	= rcu_exp_batches_completed,
-	.async		= call_rcu,
+	.async		= call_rcu_hurry,
 	.gp_barrier	= rcu_barrier,
 	.sync		= synchronize_rcu,
 	.exp_sync	= synchronize_rcu_expedited,