rcu: Mark __rcu_irq_enter_check_tick() ->rcu_urgent_qs load
authorPaul E. McKenney <paulmck@kernel.org>
Tue, 27 Jun 2023 01:37:05 +0000 (18:37 -0700)
committerPaul E. McKenney <paulmck@kernel.org>
Wed, 16 Aug 2023 21:27:41 +0000 (14:27 -0700)
The rcu_request_urgent_qs_task() function does a cross-CPU store
to ->rcu_urgent_qs, so this commit therefore marks the load in
__rcu_irq_enter_check_tick() with READ_ONCE().

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

index f788981272f8b5682eaa66e92fdb831f30c437af..cb1caefa8bd0708181c0793018cf13b46cb683b4 100644 (file)
@@ -632,7 +632,7 @@ void __rcu_irq_enter_check_tick(void)
        // prevents self-deadlock.  So we can safely recheck under the lock.
        // Note that the nohz_full state currently cannot change.
        raw_spin_lock_rcu_node(rdp->mynode);
-       if (rdp->rcu_urgent_qs && !rdp->rcu_forced_tick) {
+       if (READ_ONCE(rdp->rcu_urgent_qs) && !rdp->rcu_forced_tick) {
                // A nohz_full CPU is in the kernel and RCU needs a
                // quiescent state.  Turn on the tick!
                WRITE_ONCE(rdp->rcu_forced_tick, true);