rcu: apply TINY_PREEMPT_RCU read-side speedup to TREE_PREEMPT_RCU (80dcf60e) · Commits · 戴 / test

Replace one of the ACCESS_ONCE() calls in each of __rcu_read_lock() and __rcu_read_unlock() with barrier() as suggested by Steve Rostedt in order to avoid the potential compiler-optimization-induced bug noted by Mathieu Desnoyers. Located-by:

Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca> Suggested-by:

Steven Rostedt <rostedt@goodmis.org> Signed-off-by:

Paul E. McKenney <paulmck@linux.vnet.ibm.com>

kernel/rcutree_plugin.h

+4 −2

Original line number	Diff line number	Diff line
		@@ -201,7 +201,7 @@ static void rcu_preempt_note_context_switch(int cpu)
		*/
		void __rcu_read_lock(void)
		{
		ACCESS_ONCE(current->rcu_read_lock_nesting)++;
		current->rcu_read_lock_nesting++;
		barrier(); /* needed if we ever invoke rcu_read_lock in rcutree.c */
		}
		EXPORT_SYMBOL_GPL(__rcu_read_lock);
		@@ -344,7 +344,9 @@ void __rcu_read_unlock(void)
		struct task_struct *t = current;

		barrier(); /* needed if we ever invoke rcu_read_unlock in rcutree.c */
		if (--ACCESS_ONCE(t->rcu_read_lock_nesting) == 0 &&
		--t->rcu_read_lock_nesting;
		barrier(); /* decrement before load of ->rcu_read_unlock_special */
		if (t->rcu_read_lock_nesting == 0 &&
		unlikely(ACCESS_ONCE(t->rcu_read_unlock_special)))
		rcu_read_unlock_special(t);
		#ifdef CONFIG_PROVE_LOCKING

Admin message