locking/mutex: Remove wakeups from under mutex::wait_lock

In preparation to nest mutex::wait_lock under rq::lock we need
to remove wakeups from under it.

Do this by utilizing wake_qs to defer the wakeup until after the
lock is dropped.

[Heavily changed after 55f036c ("locking: WW mutex cleanup") and
08295b3 ("locking: Implement an algorithm choice for Wound-Wait
mutexes")]
[jstultz: rebased to mainline, added extra wake_up_q & init
 to avoid hangs, similar to Connor's rework of this patch]

Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Signed-off-by: Juri Lelli <[email protected]>
Signed-off-by: John Stultz <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Reviewed-by: Metin Kaya <[email protected]>
Acked-by: Davidlohr Bueso <[email protected]>
Tested-by: K Prateek Nayak <[email protected]>
Tested-by: Metin Kaya <[email protected]>
Link: https://lore.kernel.org/r/[email protected]

Author: Peter Zijlstra

kernel/futex/pi.c

@@ -922,6 +922,7 @@ int futex_lock_pi(u32 __user *uaddr, unsigned int flags, ktime_t *time, int tryl
 	struct rt_mutex_waiter rt_waiter;
 	struct futex_hash_bucket *hb;
 	struct futex_q q = futex_q_init;
+	DEFINE_WAKE_Q(wake_q);
 	int res, ret;
 
 	if (!IS_ENABLED(CONFIG_FUTEX_PI))
@@ -1018,8 +1019,11 @@ int futex_lock_pi(u32 __user *uaddr, unsigned int flags, ktime_t *time, int tryl
 	 * such that futex_unlock_pi() is guaranteed to observe the waiter when
 	 * it sees the futex_q::pi_state.
 	 */
-	ret = __rt_mutex_start_proxy_lock(&q.pi_state->pi_mutex, &rt_waiter, current);
+	ret = __rt_mutex_start_proxy_lock(&q.pi_state->pi_mutex, &rt_waiter, current, &wake_q);
+	preempt_disable();
 	raw_spin_unlock_irq(&q.pi_state->pi_mutex.wait_lock);
+	wake_up_q(&wake_q);
+	preempt_enable();
 
 	if (ret) {
 		if (ret == 1)

kernel/locking/mutex.c

@@ -575,6 +575,7 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
 		    struct lockdep_map *nest_lock, unsigned long ip,
 		    struct ww_acquire_ctx *ww_ctx, const bool use_ww_ctx)
 {
+	DEFINE_WAKE_Q(wake_q);
 	struct mutex_waiter waiter;
 	struct ww_mutex *ww;
 	int ret;
@@ -625,7 +626,7 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
 	 */
 	if (__mutex_trylock(lock)) {
 		if (ww_ctx)
-			__ww_mutex_check_waiters(lock, ww_ctx);
+			__ww_mutex_check_waiters(lock, ww_ctx, &wake_q);
 
 		goto skip_wait;
 	}
@@ -645,7 +646,7 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
 		 * Add in stamp order, waking up waiters that must kill
 		 * themselves.
 		 */
-		ret = __ww_mutex_add_waiter(&waiter, lock, ww_ctx);
+		ret = __ww_mutex_add_waiter(&waiter, lock, ww_ctx, &wake_q);
 		if (ret)
 			goto err_early_kill;
 	}
@@ -681,6 +682,10 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
 		}
 
 		raw_spin_unlock(&lock->wait_lock);
+		/* Make sure we do wakeups before calling schedule */
+		wake_up_q(&wake_q);
+		wake_q_init(&wake_q);
+
 		schedule_preempt_disabled();
 
 		first = __mutex_waiter_is_first(lock, &waiter);
@@ -714,7 +719,7 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
 		 */
 		if (!ww_ctx->is_wait_die &&
 		    !__mutex_waiter_is_first(lock, &waiter))
-			__ww_mutex_check_waiters(lock, ww_ctx);
+			__ww_mutex_check_waiters(lock, ww_ctx, &wake_q);
 	}
 
 	__mutex_remove_waiter(lock, &waiter);
@@ -730,6 +735,7 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
 		ww_mutex_lock_acquired(ww, ww_ctx);
 
 	raw_spin_unlock(&lock->wait_lock);
+	wake_up_q(&wake_q);
 	preempt_enable();
 	return 0;
 
@@ -741,6 +747,7 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
 	raw_spin_unlock(&lock->wait_lock);
 	debug_mutex_free_waiter(&waiter);
 	mutex_release(&lock->dep_map, ip);
+	wake_up_q(&wake_q);
 	preempt_enable();
 	return ret;
 }
@@ -951,9 +958,10 @@ static noinline void __sched __mutex_unlock_slowpath(struct mutex *lock, unsigne
 	if (owner & MUTEX_FLAG_HANDOFF)
 		__mutex_handoff(lock, next);
 
+	preempt_disable();
 	raw_spin_unlock(&lock->wait_lock);
-
 	wake_up_q(&wake_q);
+	preempt_enable();
 }
 
 #ifndef CONFIG_DEBUG_LOCK_ALLOC

kernel/locking/rtmutex.c

@@ -34,13 +34,15 @@
 
 static inline int __ww_mutex_add_waiter(struct rt_mutex_waiter *waiter,
 					struct rt_mutex *lock,
-					struct ww_acquire_ctx *ww_ctx)
+					struct ww_acquire_ctx *ww_ctx,
+					struct wake_q_head *wake_q)
 {
 	return 0;
 }
 
 static inline void __ww_mutex_check_waiters(struct rt_mutex *lock,
-					    struct ww_acquire_ctx *ww_ctx)
+					    struct ww_acquire_ctx *ww_ctx,
+					    struct wake_q_head *wake_q)
 {
 }
 
@@ -1201,7 +1203,8 @@ static int __sched task_blocks_on_rt_mutex(struct rt_mutex_base *lock,
 					   struct rt_mutex_waiter *waiter,
 					   struct task_struct *task,
 					   struct ww_acquire_ctx *ww_ctx,
-					   enum rtmutex_chainwalk chwalk)
+					   enum rtmutex_chainwalk chwalk,
+					   struct wake_q_head *wake_q)
 {
 	struct task_struct *owner = rt_mutex_owner(lock);
 	struct rt_mutex_waiter *top_waiter = waiter;
@@ -1245,7 +1248,10 @@ static int __sched task_blocks_on_rt_mutex(struct rt_mutex_base *lock,
 
 		/* Check whether the waiter should back out immediately */
 		rtm = container_of(lock, struct rt_mutex, rtmutex);
-		res = __ww_mutex_add_waiter(waiter, rtm, ww_ctx);
+		preempt_disable();
+		res = __ww_mutex_add_waiter(waiter, rtm, ww_ctx, wake_q);
+		wake_up_q(wake_q);
+		preempt_enable();
 		if (res) {
 			raw_spin_lock(&task->pi_lock);
 			rt_mutex_dequeue(lock, waiter);
@@ -1674,12 +1680,14 @@ static void __sched rt_mutex_handle_deadlock(int res, int detect_deadlock,
  * @state:	The task state for sleeping
  * @chwalk:	Indicator whether full or partial chainwalk is requested
  * @waiter:	Initializer waiter for blocking
+ * @wake_q:	The wake_q to wake tasks after we release the wait_lock
  */
 static int __sched __rt_mutex_slowlock(struct rt_mutex_base *lock,
 				       struct ww_acquire_ctx *ww_ctx,
 				       unsigned int state,
 				       enum rtmutex_chainwalk chwalk,
-				       struct rt_mutex_waiter *waiter)
+				       struct rt_mutex_waiter *waiter,
+				       struct wake_q_head *wake_q)
 {
 	struct rt_mutex *rtm = container_of(lock, struct rt_mutex, rtmutex);
 	struct ww_mutex *ww = ww_container_of(rtm);
@@ -1690,7 +1698,7 @@ static int __sched __rt_mutex_slowlock(struct rt_mutex_base *lock,
 	/* Try to acquire the lock again: */
 	if (try_to_take_rt_mutex(lock, current, NULL)) {
 		if (build_ww_mutex() && ww_ctx) {
-			__ww_mutex_check_waiters(rtm, ww_ctx);
+			__ww_mutex_check_waiters(rtm, ww_ctx, wake_q);
 			ww_mutex_lock_acquired(ww, ww_ctx);
 		}
 		return 0;
@@ -1700,15 +1708,15 @@ static int __sched __rt_mutex_slowlock(struct rt_mutex_base *lock,
 
 	trace_contention_begin(lock, LCB_F_RT);
 
-	ret = task_blocks_on_rt_mutex(lock, waiter, current, ww_ctx, chwalk);
+	ret = task_blocks_on_rt_mutex(lock, waiter, current, ww_ctx, chwalk, wake_q);
 	if (likely(!ret))
 		ret = rt_mutex_slowlock_block(lock, ww_ctx, state, NULL, waiter);
 
 	if (likely(!ret)) {
 		/* acquired the lock */
 		if (build_ww_mutex() && ww_ctx) {
 			if (!ww_ctx->is_wait_die)
-				__ww_mutex_check_waiters(rtm, ww_ctx);
+				__ww_mutex_check_waiters(rtm, ww_ctx, wake_q);
 			ww_mutex_lock_acquired(ww, ww_ctx);
 		}
 	} else {
@@ -1730,7 +1738,8 @@ static int __sched __rt_mutex_slowlock(struct rt_mutex_base *lock,
 
 static inline int __rt_mutex_slowlock_locked(struct rt_mutex_base *lock,
 					     struct ww_acquire_ctx *ww_ctx,
-					     unsigned int state)
+					     unsigned int state,
+					     struct wake_q_head *wake_q)
 {
 	struct rt_mutex_waiter waiter;
 	int ret;
@@ -1739,7 +1748,7 @@ static inline int __rt_mutex_slowlock_locked(struct rt_mutex_base *lock,
 	waiter.ww_ctx = ww_ctx;
 
 	ret = __rt_mutex_slowlock(lock, ww_ctx, state, RT_MUTEX_MIN_CHAINWALK,
-				  &waiter);
+				  &waiter, wake_q);
 
 	debug_rt_mutex_free_waiter(&waiter);
 	return ret;
@@ -1755,6 +1764,7 @@ static int __sched rt_mutex_slowlock(struct rt_mutex_base *lock,
 				     struct ww_acquire_ctx *ww_ctx,
 				     unsigned int state)
 {
+	DEFINE_WAKE_Q(wake_q);
 	unsigned long flags;
 	int ret;
 
@@ -1776,8 +1786,11 @@ static int __sched rt_mutex_slowlock(struct rt_mutex_base *lock,
 	 * irqsave/restore variants.
 	 */
 	raw_spin_lock_irqsave(&lock->wait_lock, flags);
-	ret = __rt_mutex_slowlock_locked(lock, ww_ctx, state);
+	ret = __rt_mutex_slowlock_locked(lock, ww_ctx, state, &wake_q);
+	preempt_disable();
 	raw_spin_unlock_irqrestore(&lock->wait_lock, flags);
+	wake_up_q(&wake_q);
+	preempt_enable();
 	rt_mutex_post_schedule();
 
 	return ret;
@@ -1803,8 +1816,10 @@ static __always_inline int __rt_mutex_lock(struct rt_mutex_base *lock,
 /**
  * rtlock_slowlock_locked - Slow path lock acquisition for RT locks
  * @lock:	The underlying RT mutex
+ * @wake_q:	The wake_q to wake tasks after we release the wait_lock
  */
-static void __sched rtlock_slowlock_locked(struct rt_mutex_base *lock)
+static void __sched rtlock_slowlock_locked(struct rt_mutex_base *lock,
+					   struct wake_q_head *wake_q)
 {
 	struct rt_mutex_waiter waiter;
 	struct task_struct *owner;
@@ -1821,7 +1836,7 @@ static void __sched rtlock_slowlock_locked(struct rt_mutex_base *lock)
 
 	trace_contention_begin(lock, LCB_F_RT);
 
-	task_blocks_on_rt_mutex(lock, &waiter, current, NULL, RT_MUTEX_MIN_CHAINWALK);
+	task_blocks_on_rt_mutex(lock, &waiter, current, NULL, RT_MUTEX_MIN_CHAINWALK, wake_q);
 
 	for (;;) {
 		/* Try to acquire the lock again */
@@ -1832,7 +1847,11 @@ static void __sched rtlock_slowlock_locked(struct rt_mutex_base *lock)
 			owner = rt_mutex_owner(lock);
 		else
 			owner = NULL;
+		preempt_disable();
 		raw_spin_unlock_irq(&lock->wait_lock);
+		wake_up_q(wake_q);
+		wake_q_init(wake_q);
+		preempt_enable();
 
 		if (!owner || !rtmutex_spin_on_owner(lock, &waiter, owner))
 			schedule_rtlock();
@@ -1857,10 +1876,14 @@ static void __sched rtlock_slowlock_locked(struct rt_mutex_base *lock)
 static __always_inline void __sched rtlock_slowlock(struct rt_mutex_base *lock)
 {
 	unsigned long flags;
+	DEFINE_WAKE_Q(wake_q);
 
 	raw_spin_lock_irqsave(&lock->wait_lock, flags);
-	rtlock_slowlock_locked(lock);
+	rtlock_slowlock_locked(lock, &wake_q);
+	preempt_disable();
 	raw_spin_unlock_irqrestore(&lock->wait_lock, flags);
+	wake_up_q(&wake_q);
+	preempt_enable();
 }
 
 #endif /* RT_MUTEX_BUILD_SPINLOCKS */

kernel/locking/rtmutex_api.c

@@ -275,6 +275,7 @@ void __sched rt_mutex_proxy_unlock(struct rt_mutex_base *lock)
  * @lock:		the rt_mutex to take
  * @waiter:		the pre-initialized rt_mutex_waiter
  * @task:		the task to prepare
+ * @wake_q:		the wake_q to wake tasks after we release the wait_lock
  *
  * Starts the rt_mutex acquire; it enqueues the @waiter and does deadlock
  * detection. It does not wait, see rt_mutex_wait_proxy_lock() for that.
@@ -291,7 +292,8 @@ void __sched rt_mutex_proxy_unlock(struct rt_mutex_base *lock)
  */
 int __sched __rt_mutex_start_proxy_lock(struct rt_mutex_base *lock,
 					struct rt_mutex_waiter *waiter,
-					struct task_struct *task)
+					struct task_struct *task,
+					struct wake_q_head *wake_q)
 {
 	int ret;
 
@@ -302,7 +304,7 @@ int __sched __rt_mutex_start_proxy_lock(struct rt_mutex_base *lock,
 
 	/* We enforce deadlock detection for futexes */
 	ret = task_blocks_on_rt_mutex(lock, waiter, task, NULL,
-				      RT_MUTEX_FULL_CHAINWALK);
+				      RT_MUTEX_FULL_CHAINWALK, wake_q);
 
 	if (ret && !rt_mutex_owner(lock)) {
 		/*
@@ -341,12 +343,16 @@ int __sched rt_mutex_start_proxy_lock(struct rt_mutex_base *lock,
 				      struct task_struct *task)
 {
 	int ret;
+	DEFINE_WAKE_Q(wake_q);
 
 	raw_spin_lock_irq(&lock->wait_lock);
-	ret = __rt_mutex_start_proxy_lock(lock, waiter, task);
+	ret = __rt_mutex_start_proxy_lock(lock, waiter, task, &wake_q);
 	if (unlikely(ret))
 		remove_waiter(lock, waiter);
+	preempt_disable();
 	raw_spin_unlock_irq(&lock->wait_lock);
+	wake_up_q(&wake_q);
+	preempt_enable();
 
 	return ret;
 }

kernel/locking/rtmutex_common.h

@@ -83,7 +83,8 @@ extern void rt_mutex_init_proxy_locked(struct rt_mutex_base *lock,
 extern void rt_mutex_proxy_unlock(struct rt_mutex_base *lock);
 extern int __rt_mutex_start_proxy_lock(struct rt_mutex_base *lock,
 				     struct rt_mutex_waiter *waiter,
-				     struct task_struct *task);
+				     struct task_struct *task,
+				     struct wake_q_head *);
 extern int rt_mutex_start_proxy_lock(struct rt_mutex_base *lock,
 				     struct rt_mutex_waiter *waiter,
 				     struct task_struct *task);

kernel/locking/rwbase_rt.c

@@ -69,6 +69,7 @@ static int __sched __rwbase_read_lock(struct rwbase_rt *rwb,
 				      unsigned int state)
 {
 	struct rt_mutex_base *rtm = &rwb->rtmutex;
+	DEFINE_WAKE_Q(wake_q);
 	int ret;
 
 	rwbase_pre_schedule();
@@ -110,7 +111,7 @@ static int __sched __rwbase_read_lock(struct rwbase_rt *rwb,
 	 * For rwlocks this returns 0 unconditionally, so the below
 	 * !ret conditionals are optimized out.
 	 */
-	ret = rwbase_rtmutex_slowlock_locked(rtm, state);
+	ret = rwbase_rtmutex_slowlock_locked(rtm, state, &wake_q);
 
 	/*
 	 * On success the rtmutex is held, so there can't be a writer
@@ -121,7 +122,12 @@ static int __sched __rwbase_read_lock(struct rwbase_rt *rwb,
 	 */
 	if (!ret)
 		atomic_inc(&rwb->readers);
+
+	preempt_disable();
 	raw_spin_unlock_irq(&rtm->wait_lock);
+	wake_up_q(&wake_q);
+	preempt_enable();
+
 	if (!ret)
 		rwbase_rtmutex_unlock(rtm);
 

kernel/locking/rwsem.c

@@ -1413,8 +1413,8 @@ static inline void __downgrade_write(struct rw_semaphore *sem)
 #define rwbase_rtmutex_lock_state(rtm, state)		\
 	__rt_mutex_lock(rtm, state)
 
-#define rwbase_rtmutex_slowlock_locked(rtm, state)	\
-	__rt_mutex_slowlock_locked(rtm, NULL, state)
+#define rwbase_rtmutex_slowlock_locked(rtm, state, wq)	\
+	__rt_mutex_slowlock_locked(rtm, NULL, state, wq)
 
 #define rwbase_rtmutex_unlock(rtm)			\
 	__rt_mutex_unlock(rtm)

kernel/locking/spinlock_rt.c

@@ -162,9 +162,10 @@ rwbase_rtmutex_lock_state(struct rt_mutex_base *rtm, unsigned int state)
 }
 
 static __always_inline int
-rwbase_rtmutex_slowlock_locked(struct rt_mutex_base *rtm, unsigned int state)
+rwbase_rtmutex_slowlock_locked(struct rt_mutex_base *rtm, unsigned int state,
+			       struct wake_q_head *wake_q)
 {
-	rtlock_slowlock_locked(rtm);
+	rtlock_slowlock_locked(rtm, wake_q);
 	return 0;
 }