rcu: Add READ_ONCE and data_race() to rcu_node ->boost_tasks

The rcu_node structure's ->boost_tasks field is read locklessly, so this
commit adds the READ_ONCE() to one load in order to avoid destructive
compiler optimizations.  The other load is from a diagnostic print,
so data_race() suffices.

This data race was reported by KCSAN.  Not appropriate for backporting
due to failure being unlikely.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>

kernel/rcu/tree_plugin.h

@@ -760,7 +760,7 @@ dump_blkd_tasks(struct rcu_node *rnp, int ncheck)
 		pr_info("%s: %d:%d ->qsmask %#lx ->qsmaskinit %#lx ->qsmaskinitnext %#lx\n",
 			__func__, rnp1->grplo, rnp1->grphi, rnp1->qsmask, rnp1->qsmaskinit, rnp1->qsmaskinitnext);
 	pr_info("%s: ->gp_tasks %p ->boost_tasks %p ->exp_tasks %p\n",
-		__func__, READ_ONCE(rnp->gp_tasks), rnp->boost_tasks,
+		__func__, READ_ONCE(rnp->gp_tasks), data_race(rnp->boost_tasks),
 		READ_ONCE(rnp->exp_tasks));
 	pr_info("%s: ->blkd_tasks", __func__);
 	i = 0;
@@ -1036,7 +1036,8 @@ static int rcu_boost_kthread(void *arg)
 	for (;;) {
 		WRITE_ONCE(rnp->boost_kthread_status, RCU_KTHREAD_WAITING);
 		trace_rcu_utilization(TPS("End boost kthread@rcu_wait"));
-		rcu_wait(rnp->boost_tasks || READ_ONCE(rnp->exp_tasks));
+		rcu_wait(READ_ONCE(rnp->boost_tasks) ||
+			 READ_ONCE(rnp->exp_tasks));
 		trace_rcu_utilization(TPS("Start boost kthread@rcu_wait"));
 		WRITE_ONCE(rnp->boost_kthread_status, RCU_KTHREAD_RUNNING);
 		more2boost = rcu_boost(rnp);