@@ -297,6 +297,36 @@ static __always_inline u32  __pv_wait_head_or_lock(struct qspinlock *lock,
 #define queued_spin_lock_slowpath	native_queued_spin_lock_slowpath
 #endif
 
+/*
+ * set_locked_empty_mcs - Try to set the spinlock value to _Q_LOCKED_VAL,
+ * and by doing that unlock the MCS lock when its waiting queue is empty
+ * @lock: Pointer to queued spinlock structure
+ * @val: Current value of the lock
+ * @node: Pointer to the MCS node of the lock holder
+ *
+ * *,*,* -> 0,0,1
+ */
+static __always_inline bool __set_locked_empty_mcs(struct qspinlock *lock,
+						   u32 val,
+						   struct mcs_spinlock *node)
+{
+	return atomic_try_cmpxchg_relaxed(&lock->val, &val, _Q_LOCKED_VAL);
+}
+
+/*
+ * pass_mcs_lock - pass the MCS lock to the next waiter
+ * @node: Pointer to the MCS node of the lock holder
+ * @next: Pointer to the MCS node of the first waiter in the MCS queue
+ */
+static __always_inline void __pass_mcs_lock(struct mcs_spinlock *node,
+					    struct mcs_spinlock *next)
+{
+	arch_mcs_spin_unlock_contended(&next->locked, 1);
+}
+
+#define set_locked_empty_mcs	__set_locked_empty_mcs
+#define pass_mcs_lock		__pass_mcs_lock
+
 #endif /* _GEN_PV_LOCK_SLOWPATH */
 
 /**

@@ -541,7 +571,7 @@ void queued_spin_lock_slowpath(struct qspinlock *lock, u32 val)
 	 *       PENDING will make the uncontended transition fail.
 	 */
 	if ((val & _Q_TAIL_MASK) == tail) {
-		if (atomic_try_cmpxchg_relaxed(&lock->val, &val, _Q_LOCKED_VAL))
+		if (set_locked_empty_mcs(lock, val, node))
 			goto release; /* No contention */
 	}
 

@@ -558,7 +588,7 @@ void queued_spin_lock_slowpath(struct qspinlock *lock, u32 val)
 	if (!next)
 		next = smp_cond_load_relaxed(&node->next, (VAL));
 
-	arch_mcs_spin_unlock_contended(&next->locked, 1);
+	pass_mcs_lock(node, next);
 	pv_kick_node(lock, next);
 
 release:

@@ -583,6 +613,12 @@ EXPORT_SYMBOL(queued_spin_lock_slowpath);
 #undef pv_kick_node
 #undef pv_wait_head_or_lock
 
+#undef set_locked_empty_mcs
+#define set_locked_empty_mcs		__set_locked_empty_mcs
+
+#undef pass_mcs_lock
+#define pass_mcs_lock			__pass_mcs_lock
+
 #undef  queued_spin_lock_slowpath
 #define queued_spin_lock_slowpath	__pv_queued_spin_lock_slowpath