RCU: Privatize rcu_node::lock

In patch:

"rcu: Add transitivity to remaining rcu_node ->lock acquisitions"

All locking operations on rcu_node::lock are replaced with the wrappers
because of the need of transitivity, which indicates we should never
write code using LOCK primitives alone(i.e. without a proper barrier
following) on rcu_node::lock outside those wrappers. We could detect
this kind of misuses on rcu_node::lock in the future by adding __private
modifier on rcu_node::lock.

To privatize rcu_node::lock, unlock wrappers are also needed. Replacing
spinlock unlocks with these wrappers not only privatizes rcu_node::lock
but also makes it easier to figure out critical sections of rcu_node.

This patch adds __private modifier to rcu_node::lock and makes every
access to it wrapped by ACCESS_PRIVATE(). Besides, unlock wrappers are
added and raw_spin_unlock(&rnp->lock) and its friends are replaced with
those wrappers.
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>

kernel/rcu/tree.h

@@ -149,8 +149,9 @@ struct rcu_dynticks {
  * Definition for node within the RCU grace-period-detection hierarchy.
  */
 struct rcu_node {
-	raw_spinlock_t lock;	/* Root rcu_node's lock protects some */
-				/*  rcu_state fields as well as following. */
+	raw_spinlock_t __private lock;	/* Root rcu_node's lock protects */
+					/*  some rcu_state fields as well as */
+					/*  following. */
 	unsigned long gpnum;	/* Current grace period for this node. */
 				/*  This will either be equal to or one */
 				/*  behind the root rcu_node's gpnum. */
@@ -680,7 +681,7 @@ static inline void rcu_nocb_q_lengths(struct rcu_data *rdp, long *ql, long *qll)
 #endif /* #else #ifdef CONFIG_PPC */
 
 /*
- * Wrappers for the rcu_node::lock acquire.
+ * Wrappers for the rcu_node::lock acquire and release.
  *
  * Because the rcu_nodes form a tree, the tree traversal locking will observe
  * different lock values, this in turn means that an UNLOCK of one level
@@ -689,29 +690,48 @@ static inline void rcu_nocb_q_lengths(struct rcu_data *rdp, long *ql, long *qll)
  *
  * In order to restore full ordering between tree levels, augment the regular
  * lock acquire functions with smp_mb__after_unlock_lock().
+ *
+ * As ->lock of struct rcu_node is a __private field, therefore one should use
+ * these wrappers rather than directly call raw_spin_{lock,unlock}* on ->lock.
  */
 static inline void raw_spin_lock_rcu_node(struct rcu_node *rnp)
 {
-	raw_spin_lock(&rnp->lock);
+	raw_spin_lock(&ACCESS_PRIVATE(rnp, lock));
 	smp_mb__after_unlock_lock();
 }
 
+static inline void raw_spin_unlock_rcu_node(struct rcu_node *rnp)
+{
+	raw_spin_unlock(&ACCESS_PRIVATE(rnp, lock));
+}
+
 static inline void raw_spin_lock_irq_rcu_node(struct rcu_node *rnp)
 {
-	raw_spin_lock_irq(&rnp->lock);
+	raw_spin_lock_irq(&ACCESS_PRIVATE(rnp, lock));
 	smp_mb__after_unlock_lock();
 }
 
+static inline void raw_spin_unlock_irq_rcu_node(struct rcu_node *rnp)
+{
+	raw_spin_unlock_irq(&ACCESS_PRIVATE(rnp, lock));
+}
+
 #define raw_spin_lock_irqsave_rcu_node(rnp, flags)			\
 do {									\
 	typecheck(unsigned long, flags);				\
-	raw_spin_lock_irqsave(&(rnp)->lock, flags);	\
+	raw_spin_lock_irqsave(&ACCESS_PRIVATE(rnp, lock), flags);	\
 	smp_mb__after_unlock_lock();					\
 } while (0)
 
+#define raw_spin_unlock_irqrestore_rcu_node(rnp, flags)			\
+do {									\
+	typecheck(unsigned long, flags);				\
+	raw_spin_unlock_irqrestore(&ACCESS_PRIVATE(rnp, lock), flags);	\
+} while (0)
+
 static inline bool raw_spin_trylock_rcu_node(struct rcu_node *rnp)
 {
-	bool locked = raw_spin_trylock(&rnp->lock);
+	bool locked = raw_spin_trylock(&ACCESS_PRIVATE(rnp, lock));
 
 	if (locked)
 		smp_mb__after_unlock_lock();

kernel/rcu/tree_plugin.h

@@ -235,7 +235,7 @@ static void rcu_preempt_ctxt_queue(struct rcu_node *rnp, struct rcu_data *rdp)
 		rnp->gp_tasks = &t->rcu_node_entry;
 	if (!rnp->exp_tasks && (blkd_state & RCU_EXP_BLKD))
 		rnp->exp_tasks = &t->rcu_node_entry;
-	raw_spin_unlock(&rnp->lock); /* rrupts remain disabled. */
+	raw_spin_unlock_rcu_node(rnp); /* interrupts remain disabled. */
 
 	/*
 	 * Report the quiescent state for the expedited GP.  This expedited
@@ -489,7 +489,7 @@ void rcu_read_unlock_special(struct task_struct *t)
 							 !!rnp->gp_tasks);
 			rcu_report_unblock_qs_rnp(rcu_state_p, rnp, flags);
 		} else {
-			raw_spin_unlock_irqrestore(&rnp->lock, flags);
+			raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 		}
 
 		/* Unboost if we were boosted. */
@@ -518,14 +518,14 @@ static void rcu_print_detail_task_stall_rnp(struct rcu_node *rnp)
 
 	raw_spin_lock_irqsave_rcu_node(rnp, flags);
 	if (!rcu_preempt_blocked_readers_cgp(rnp)) {
-		raw_spin_unlock_irqrestore(&rnp->lock, flags);
+		raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 		return;
 	}
 	t = list_entry(rnp->gp_tasks->prev,
 		       struct task_struct, rcu_node_entry);
 	list_for_each_entry_continue(t, &rnp->blkd_tasks, rcu_node_entry)
 		sched_show_task(t);
-	raw_spin_unlock_irqrestore(&rnp->lock, flags);
+	raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 }
 
 /*
@@ -990,7 +990,7 @@ static int rcu_boost(struct rcu_node *rnp)
 	 * might exit their RCU read-side critical sections on their own.
 	 */
 	if (rnp->exp_tasks == NULL && rnp->boost_tasks == NULL) {
-		raw_spin_unlock_irqrestore(&rnp->lock, flags);
+		raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 		return 0;
 	}
 
@@ -1027,7 +1027,7 @@ static int rcu_boost(struct rcu_node *rnp)
 	 */
 	t = container_of(tb, struct task_struct, rcu_node_entry);
 	rt_mutex_init_proxy_locked(&rnp->boost_mtx, t);
-	raw_spin_unlock_irqrestore(&rnp->lock, flags);
+	raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 	/* Lock only for side effect: boosts task t's priority. */
 	rt_mutex_lock(&rnp->boost_mtx);
 	rt_mutex_unlock(&rnp->boost_mtx);  /* Then keep lockdep happy. */
@@ -1087,7 +1087,7 @@ static void rcu_initiate_boost(struct rcu_node *rnp, unsigned long flags)
 
 	if (!rcu_preempt_blocked_readers_cgp(rnp) && rnp->exp_tasks == NULL) {
 		rnp->n_balk_exp_gp_tasks++;
-		raw_spin_unlock_irqrestore(&rnp->lock, flags);
+		raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 		return;
 	}
 	if (rnp->exp_tasks != NULL ||
@@ -1097,13 +1097,13 @@ static void rcu_initiate_boost(struct rcu_node *rnp, unsigned long flags)
 	     ULONG_CMP_GE(jiffies, rnp->boost_time))) {
 		if (rnp->exp_tasks == NULL)
 			rnp->boost_tasks = rnp->gp_tasks;
-		raw_spin_unlock_irqrestore(&rnp->lock, flags);
+		raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 		t = rnp->boost_kthread_task;
 		if (t)
 			rcu_wake_cond(t, rnp->boost_kthread_status);
 	} else {
 		rcu_initiate_boost_trace(rnp);
-		raw_spin_unlock_irqrestore(&rnp->lock, flags);
+		raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 	}
 }
 
@@ -1171,7 +1171,7 @@ static int rcu_spawn_one_boost_kthread(struct rcu_state *rsp,
 		return PTR_ERR(t);
 	raw_spin_lock_irqsave_rcu_node(rnp, flags);
 	rnp->boost_kthread_task = t;
-	raw_spin_unlock_irqrestore(&rnp->lock, flags);
+	raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 	sp.sched_priority = kthread_prio;
 	sched_setscheduler_nocheck(t, SCHED_FIFO, &sp);
 	wake_up_process(t); /* get to TASK_INTERRUPTIBLE quickly. */
@@ -1307,7 +1307,7 @@ static void rcu_prepare_kthreads(int cpu)
 static void rcu_initiate_boost(struct rcu_node *rnp, unsigned long flags)
 	__releases(rnp->lock)
 {
-	raw_spin_unlock_irqrestore(&rnp->lock, flags);
+	raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 }
 
 static void invoke_rcu_callbacks_kthread(void)
@@ -1558,7 +1558,7 @@ static void rcu_prepare_for_idle(void)
 		rnp = rdp->mynode;
 		raw_spin_lock_rcu_node(rnp); /* irqs already disabled. */
 		needwake = rcu_accelerate_cbs(rsp, rnp, rdp);
-		raw_spin_unlock(&rnp->lock); /* irqs remain disabled. */
+		raw_spin_unlock_rcu_node(rnp); /* irqs remain disabled. */
 		if (needwake)
 			rcu_gp_kthread_wake(rsp);
 	}
@@ -2058,7 +2058,7 @@ static void rcu_nocb_wait_gp(struct rcu_data *rdp)
 
 	raw_spin_lock_irqsave_rcu_node(rnp, flags);
 	needwake = rcu_start_future_gp(rnp, rdp, &c);
-	raw_spin_unlock_irqrestore(&rnp->lock, flags);
+	raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
 	if (needwake)
 		rcu_gp_kthread_wake(rdp->rsp);