1 files changed, 98 insertions, 133 deletions

diff --git a/kernel/futex.c b/kernel/futex.c
index fe90164aa6ec..688b6fcb79cb 100644
--- a/kernel/futex.c
+++ b/kernel/futex.c
@@ -240,7 +240,7 @@ struct futex_q {
 	struct plist_node list;
 
 	struct task_struct *task;
-	spinlock_t *lock_ptr;
+	raw_spinlock_t *lock_ptr;
 	union futex_key key;
 	struct futex_pi_state *pi_state;
 	struct rt_mutex_waiter *rt_waiter;
@@ -261,7 +261,7 @@ static const struct futex_q futex_q_init = {
  */
 struct futex_hash_bucket {
 	atomic_t waiters;
-	spinlock_t lock;
+	raw_spinlock_t lock;
 	struct plist_head chain;
 } ____cacheline_aligned_in_smp;
 
@@ -822,13 +822,13 @@ static void get_pi_state(struct futex_pi_state *pi_state)
  * Drops a reference to the pi_state object and frees or caches it
  * when the last reference is gone.
  */
-static void put_pi_state(struct futex_pi_state *pi_state)
+static struct futex_pi_state *__put_pi_state(struct futex_pi_state *pi_state)
 {
 	if (!pi_state)
-		return;
+		return NULL;
 
 	if (!atomic_dec_and_test(&pi_state->refcount))
-		return;
+		return NULL;
 
 	/*
 	 * If pi_state->owner is NULL, the owner is most probably dying
@@ -848,9 +848,7 @@ static void put_pi_state(struct futex_pi_state *pi_state)
 		raw_spin_unlock_irq(&pi_state->pi_mutex.wait_lock);
 	}
 
-	if (current->pi_state_cache) {
-		kfree(pi_state);
-	} else {
+	if (!current->pi_state_cache) {
 		/*
 		 * pi_state->list is already empty.
 		 * clear pi_state->owner.
@@ -859,6 +857,30 @@ static void put_pi_state(struct futex_pi_state *pi_state)
 		pi_state->owner = NULL;
 		atomic_set(&pi_state->refcount, 1);
 		current->pi_state_cache = pi_state;
+		pi_state = NULL;
+	}
+	return pi_state;
+}
+
+static void put_pi_state(struct futex_pi_state *pi_state)
+{
+	kfree(__put_pi_state(pi_state));
+}
+
+static void put_pi_state_atomic(struct futex_pi_state *pi_state,
+				struct list_head *to_free)
+{
+	if (__put_pi_state(pi_state))
+		list_add(&pi_state->list, to_free);
+}
+
+static void free_pi_state_list(struct list_head *to_free)
+{
+	struct futex_pi_state *p, *next;
+
+	list_for_each_entry_safe(p, next, to_free, list) {
+		list_del(&p->list);
+		kfree(p);
 	}
 }
 
@@ -875,6 +897,7 @@ void exit_pi_state_list(struct task_struct *curr)
 	struct futex_pi_state *pi_state;
 	struct futex_hash_bucket *hb;
 	union futex_key key = FUTEX_KEY_INIT;
+	LIST_HEAD(to_free);
 
 	if (!futex_cmpxchg_enabled)
 		return;
@@ -908,7 +931,7 @@ void exit_pi_state_list(struct task_struct *curr)
 		}
 		raw_spin_unlock_irq(&curr->pi_lock);
 
-		spin_lock(&hb->lock);
+		raw_spin_lock(&hb->lock);
 		raw_spin_lock_irq(&pi_state->pi_mutex.wait_lock);
 		raw_spin_lock(&curr->pi_lock);
 		/*
@@ -918,10 +941,8 @@ void exit_pi_state_list(struct task_struct *curr)
 		if (head->next != next) {
 			/* retain curr->pi_lock for the loop invariant */
 			raw_spin_unlock(&pi_state->pi_mutex.wait_lock);
-			raw_spin_unlock_irq(&curr->pi_lock);
-			spin_unlock(&hb->lock);
-			raw_spin_lock_irq(&curr->pi_lock);
-			put_pi_state(pi_state);
+			raw_spin_unlock(&hb->lock);
+			put_pi_state_atomic(pi_state, &to_free);
 			continue;
 		}
 
@@ -932,7 +953,7 @@ void exit_pi_state_list(struct task_struct *curr)
 
 		raw_spin_unlock(&curr->pi_lock);
 		raw_spin_unlock_irq(&pi_state->pi_mutex.wait_lock);
-		spin_unlock(&hb->lock);
+		raw_spin_unlock(&hb->lock);
 
 		rt_mutex_futex_unlock(&pi_state->pi_mutex);
 		put_pi_state(pi_state);
@@ -940,6 +961,8 @@ void exit_pi_state_list(struct task_struct *curr)
 		raw_spin_lock_irq(&curr->pi_lock);
 	}
 	raw_spin_unlock_irq(&curr->pi_lock);
+
+	free_pi_state_list(&to_free);
 }
 
 #endif
@@ -1426,7 +1449,7 @@ static void __unqueue_futex(struct futex_q *q)
 {
 	struct futex_hash_bucket *hb;
 
-	if (WARN_ON_SMP(!q->lock_ptr || !spin_is_locked(q->lock_ptr))
+	if (WARN_ON_SMP(!q->lock_ptr || !raw_spin_is_locked(q->lock_ptr))
 	    || WARN_ON(plist_node_empty(&q->list)))
 		return;
 
@@ -1554,21 +1577,21 @@ static inline void
 double_lock_hb(struct futex_hash_bucket *hb1, struct futex_hash_bucket *hb2)
 {
 	if (hb1 <= hb2) {
-		spin_lock(&hb1->lock);
+		raw_spin_lock(&hb1->lock);
 		if (hb1 < hb2)
-			spin_lock_nested(&hb2->lock, SINGLE_DEPTH_NESTING);
+			raw_spin_lock_nested(&hb2->lock, SINGLE_DEPTH_NESTING);
 	} else { /* hb1 > hb2 */
-		spin_lock(&hb2->lock);
-		spin_lock_nested(&hb1->lock, SINGLE_DEPTH_NESTING);
+		raw_spin_lock(&hb2->lock);
+		raw_spin_lock_nested(&hb1->lock, SINGLE_DEPTH_NESTING);
 	}
 }
 
 static inline void
 double_unlock_hb(struct futex_hash_bucket *hb1, struct futex_hash_bucket *hb2)
 {
-	spin_unlock(&hb1->lock);
+	raw_spin_unlock(&hb1->lock);
 	if (hb1 != hb2)
-		spin_unlock(&hb2->lock);
+		raw_spin_unlock(&hb2->lock);
 }
 
 /*
@@ -1596,7 +1619,7 @@ futex_wake(u32 __user *uaddr, unsigned int flags, int nr_wake, u32 bitset)
 	if (!hb_waiters_pending(hb))
 		goto out_put_key;
 
-	spin_lock(&hb->lock);
+	raw_spin_lock(&hb->lock);
 
 	plist_for_each_entry_safe(this, next, &hb->chain, list) {
 		if (match_futex (&this->key, &key)) {
@@ -1615,7 +1638,7 @@ futex_wake(u32 __user *uaddr, unsigned int flags, int nr_wake, u32 bitset)
 		}
 	}
 
-	spin_unlock(&hb->lock);
+	raw_spin_unlock(&hb->lock);
 	wake_up_q(&wake_q);
 out_put_key:
 	put_futex_key(&key);
@@ -1922,6 +1945,7 @@ static int futex_requeue(u32 __user *uaddr1, unsigned int flags,
 	struct futex_hash_bucket *hb1, *hb2;
 	struct futex_q *this, *next;
 	DEFINE_WAKE_Q(wake_q);
+	LIST_HEAD(to_free);
 
 	if (nr_wake < 0 || nr_requeue < 0)
 		return -EINVAL;
@@ -2149,16 +2173,6 @@ retry_private:
 				requeue_pi_wake_futex(this, &key2, hb2);
 				drop_count++;
 				continue;
-			} else if (ret == -EAGAIN) {
-				/*
-				 * Waiter was woken by timeout or
-				 * signal and has set pi_blocked_on to
-				 * PI_WAKEUP_INPROGRESS before we
-				 * tried to enqueue it on the rtmutex.
-				 */
-				this->pi_state = NULL;
-				put_pi_state(pi_state);
-				continue;
 			} else if (ret) {
 				/*
 				 * rt_mutex_start_proxy_lock() detected a
@@ -2169,7 +2183,7 @@ retry_private:
 				 * object.
 				 */
 				this->pi_state = NULL;
-				put_pi_state(pi_state);
+				put_pi_state_atomic(pi_state, &to_free);
 				/*
 				 * We stop queueing more waiters and let user
 				 * space deal with the mess.
@@ -2186,7 +2200,7 @@ retry_private:
 	 * in futex_proxy_trylock_atomic() or in lookup_pi_state(). We
 	 * need to drop it here again.
 	 */
-	put_pi_state(pi_state);
+	put_pi_state_atomic(pi_state, &to_free);
 
 out_unlock:
 	double_unlock_hb(hb1, hb2);
@@ -2207,6 +2221,7 @@ out_put_keys:
 out_put_key1:
 	put_futex_key(&key1);
 out:
+	free_pi_state_list(&to_free);
 	return ret ? ret : task_count;
 }
 
@@ -2230,7 +2245,7 @@ static inline struct futex_hash_bucket *queue_lock(struct futex_q *q)
 
 	q->lock_ptr = &hb->lock;
 
-	spin_lock(&hb->lock); /* implies smp_mb(); (A) */
+	raw_spin_lock(&hb->lock); /* implies smp_mb(); (A) */
 	return hb;
 }
 
@@ -2238,7 +2253,7 @@ static inline void
 queue_unlock(struct futex_hash_bucket *hb)
 	__releases(&hb->lock)
 {
-	spin_unlock(&hb->lock);
+	raw_spin_unlock(&hb->lock);
 	hb_waiters_dec(hb);
 }
 
@@ -2277,7 +2292,7 @@ static inline void queue_me(struct futex_q *q, struct futex_hash_bucket *hb)
 	__releases(&hb->lock)
 {
 	__queue_me(q, hb);
-	spin_unlock(&hb->lock);
+	raw_spin_unlock(&hb->lock);
 }
 
 /**
@@ -2293,41 +2308,41 @@ static inline void queue_me(struct futex_q *q, struct futex_hash_bucket *hb)
  */
 static int unqueue_me(struct futex_q *q)
 {
-	spinlock_t *lock_ptr;
+	raw_spinlock_t *lock_ptr;
 	int ret = 0;
 
 	/* In the common case we don't take the spinlock, which is nice. */
 retry:
 	/*
-	 * q->lock_ptr can change between this read and the following spin_lock.
-	 * Use READ_ONCE to forbid the compiler from reloading q->lock_ptr and
-	 * optimizing lock_ptr out of the logic below.
+	 * q->lock_ptr can change between this read and the following
+	 * raw_spin_lock. Use READ_ONCE to forbid the compiler from reloading
+	 * q->lock_ptr and optimizing lock_ptr out of the logic below.
 	 */
 	lock_ptr = READ_ONCE(q->lock_ptr);
 	if (lock_ptr != NULL) {
-		spin_lock(lock_ptr);
+		raw_spin_lock(lock_ptr);
 		/*
 		 * q->lock_ptr can change between reading it and
-		 * spin_lock(), causing us to take the wrong lock.  This
+		 * raw_spin_lock(), causing us to take the wrong lock.  This
 		 * corrects the race condition.
 		 *
 		 * Reasoning goes like this: if we have the wrong lock,
 		 * q->lock_ptr must have changed (maybe several times)
-		 * between reading it and the spin_lock().  It can
-		 * change again after the spin_lock() but only if it was
-		 * already changed before the spin_lock().  It cannot,
+		 * between reading it and the raw_spin_lock().  It can
+		 * change again after the raw_spin_lock() but only if it was
+		 * already changed before the raw_spin_lock().  It cannot,
 		 * however, change back to the original value.  Therefore
 		 * we can detect whether we acquired the correct lock.
 		 */
 		if (unlikely(lock_ptr != q->lock_ptr)) {
-			spin_unlock(lock_ptr);
+			raw_spin_unlock(lock_ptr);
 			goto retry;
 		}
 		__unqueue_futex(q);
 
 		BUG_ON(q->pi_state);
 
-		spin_unlock(lock_ptr);
+		raw_spin_unlock(lock_ptr);
 		ret = 1;
 	}
 
@@ -2343,13 +2358,16 @@ retry:
 static void unqueue_me_pi(struct futex_q *q)
 	__releases(q->lock_ptr)
 {
+	struct futex_pi_state *ps;
+
 	__unqueue_futex(q);
 
 	BUG_ON(!q->pi_state);
-	put_pi_state(q->pi_state);
+	ps = __put_pi_state(q->pi_state);
 	q->pi_state = NULL;
 
-	spin_unlock(q->lock_ptr);
+	raw_spin_unlock(q->lock_ptr);
+	kfree(ps);
 }
 
 static int fixup_pi_state_owner(u32 __user *uaddr, struct futex_q *q,
@@ -2482,7 +2500,7 @@ retry:
 	 */
 handle_err:
 	raw_spin_unlock_irq(&pi_state->pi_mutex.wait_lock);
-	spin_unlock(q->lock_ptr);
+	raw_spin_unlock(q->lock_ptr);
 
 	switch (err) {
 	case -EFAULT:
@@ -2500,7 +2518,7 @@ handle_err:
 		break;
 	}
 
-	spin_lock(q->lock_ptr);
+	raw_spin_lock(q->lock_ptr);
 	raw_spin_lock_irq(&pi_state->pi_mutex.wait_lock);
 
 	/*
@@ -2596,7 +2614,7 @@ static void futex_wait_queue_me(struct futex_hash_bucket *hb, struct futex_q *q,
 	/*
 	 * The task state is guaranteed to be set before another task can
 	 * wake it. set_current_state() is implemented using smp_store_mb() and
-	 * queue_me() calls spin_unlock() upon completion, both serializing
+	 * queue_me() calls raw_spin_unlock() upon completion, both serializing
 	 * access to the hash list and forcing another memory barrier.
 	 */
 	set_current_state(TASK_INTERRUPTIBLE);
@@ -2887,15 +2905,7 @@ retry_private:
 	 * before __rt_mutex_start_proxy_lock() is done.
 	 */
 	raw_spin_lock_irq(&q.pi_state->pi_mutex.wait_lock);
-	/*
-	 * the migrate_disable() here disables migration in the in_atomic() fast
-	 * path which is enabled again in the following spin_unlock(). We have
-	 * one migrate_disable() pending in the slow-path which is reversed
-	 * after the raw_spin_unlock_irq() where we leave the atomic context.
-	 */
-	migrate_disable();
-
-	spin_unlock(q.lock_ptr);
+	raw_spin_unlock(q.lock_ptr);
 	/*
 	 * __rt_mutex_start_proxy_lock() unconditionally enqueues the @rt_waiter
 	 * such that futex_unlock_pi() is guaranteed to observe the waiter when
@@ -2903,7 +2913,6 @@ retry_private:
 	 */
 	ret = __rt_mutex_start_proxy_lock(&q.pi_state->pi_mutex, &rt_waiter, current);
 	raw_spin_unlock_irq(&q.pi_state->pi_mutex.wait_lock);
-	migrate_enable();
 
 	if (ret) {
 		if (ret == 1)
@@ -2917,7 +2926,7 @@ retry_private:
 	ret = rt_mutex_wait_proxy_lock(&q.pi_state->pi_mutex, to, &rt_waiter);
 
 cleanup:
-	spin_lock(q.lock_ptr);
+	raw_spin_lock(q.lock_ptr);
 	/*
 	 * If we failed to acquire the lock (deadlock/signal/timeout), we must
 	 * first acquire the hb->lock before removing the lock from the
@@ -3018,7 +3027,7 @@ retry:
 		return ret;
 
 	hb = hash_futex(&key);
-	spin_lock(&hb->lock);
+	raw_spin_lock(&hb->lock);
 
 	/*
 	 * Check waiters first. We do not trust user space values at
@@ -3052,21 +3061,11 @@ retry:
 		 * rt_waiter. Also see the WARN in wake_futex_pi().
 		 */
 		raw_spin_lock_irq(&pi_state->pi_mutex.wait_lock);
-		/*
-		 * Magic trickery for now to make the RT migrate disable
-		 * logic happy. The following spin_unlock() happens with
-		 * interrupts disabled so the internal migrate_enable()
-		 * won't undo the migrate_disable() which was issued when
-		 * locking hb->lock.
-		 */
-		migrate_disable();
-		spin_unlock(&hb->lock);
+		raw_spin_unlock(&hb->lock);
 
 		/* drops pi_state->pi_mutex.wait_lock */
 		ret = wake_futex_pi(uaddr, uval, pi_state);
 
-		migrate_enable();
-
 		put_pi_state(pi_state);
 
 		/*
@@ -3101,7 +3100,7 @@ retry:
 	 * owner.
 	 */
 	if ((ret = cmpxchg_futex_value_locked(&curval, uaddr, uval, 0))) {
-		spin_unlock(&hb->lock);
+		raw_spin_unlock(&hb->lock);
 		switch (ret) {
 		case -EFAULT:
 			goto pi_faulted;
@@ -3121,7 +3120,7 @@ retry:
 	ret = (curval == uval) ? 0 : -EAGAIN;
 
 out_unlock:
-	spin_unlock(&hb->lock);
+	raw_spin_unlock(&hb->lock);
 out_putkey:
 	put_futex_key(&key);
 	return ret;
@@ -3237,7 +3236,7 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, unsigned int flags,
 	struct hrtimer_sleeper timeout, *to = NULL;
 	struct futex_pi_state *pi_state = NULL;
 	struct rt_mutex_waiter rt_waiter;
-	struct futex_hash_bucket *hb, *hb2;
+	struct futex_hash_bucket *hb;
 	union futex_key key2 = FUTEX_KEY_INIT;
 	struct futex_q q = futex_q_init;
 	int res, ret;
@@ -3295,55 +3294,20 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, unsigned int flags,
 	/* Queue the futex_q, drop the hb lock, wait for wakeup. */
 	futex_wait_queue_me(hb, &q, to);
 
-	/*
-	 * On RT we must avoid races with requeue and trying to block
-	 * on two mutexes (hb->lock and uaddr2's rtmutex) by
-	 * serializing access to pi_blocked_on with pi_lock.
-	 */
-	raw_spin_lock_irq(&current->pi_lock);
-	if (current->pi_blocked_on) {
-		/*
-		 * We have been requeued or are in the process of
-		 * being requeued.
-		 */
-		raw_spin_unlock_irq(&current->pi_lock);
-	} else {
-		/*
-		 * Setting pi_blocked_on to PI_WAKEUP_INPROGRESS
-		 * prevents a concurrent requeue from moving us to the
-		 * uaddr2 rtmutex. After that we can safely acquire
-		 * (and possibly block on) hb->lock.
-		 */
-		current->pi_blocked_on = PI_WAKEUP_INPROGRESS;
-		raw_spin_unlock_irq(&current->pi_lock);
-
-		spin_lock(&hb->lock);
-
-		/*
-		 * Clean up pi_blocked_on. We might leak it otherwise
-		 * when we succeeded with the hb->lock in the fast
-		 * path.
-		 */
-		raw_spin_lock_irq(&current->pi_lock);
-		current->pi_blocked_on = NULL;
-		raw_spin_unlock_irq(&current->pi_lock);
-
-		ret = handle_early_requeue_pi_wakeup(hb, &q, &key2, to);
-		spin_unlock(&hb->lock);
-		if (ret)
-			goto out_put_keys;
-	}
+	raw_spin_lock(&hb->lock);
+	ret = handle_early_requeue_pi_wakeup(hb, &q, &key2, to);
+	raw_spin_unlock(&hb->lock);
+	if (ret)
+		goto out_put_keys;
 
 	/*
-	 * In order to be here, we have either been requeued, are in
-	 * the process of being requeued, or requeue successfully
-	 * acquired uaddr2 on our behalf.  If pi_blocked_on was
-	 * non-null above, we may be racing with a requeue.  Do not
-	 * rely on q->lock_ptr to be hb2->lock until after blocking on
-	 * hb->lock or hb2->lock. The futex_requeue dropped our key1
-	 * reference and incremented our key2 reference count.
+	 * In order for us to be here, we know our q.key == key2, and since
+	 * we took the hb->lock above, we also know that futex_requeue() has
+	 * completed and we no longer have to concern ourselves with a wakeup
+	 * race with the atomic proxy lock acquisition by the requeue code. The
+	 * futex_requeue dropped our key1 reference and incremented our key2
+	 * reference count.
 	 */
-	hb2 = hash_futex(&key2);
 
 	/* Check if the requeue code acquired the second futex for us. */
 	if (!q.rt_waiter) {
@@ -3352,8 +3316,9 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, unsigned int flags,
 		 * did a lock-steal - fix up the PI-state in that case.
 		 */
 		if (q.pi_state && (q.pi_state->owner != current)) {
-			spin_lock(&hb2->lock);
-			BUG_ON(&hb2->lock != q.lock_ptr);
+			struct futex_pi_state *ps_free;
+
+			raw_spin_lock(q.lock_ptr);
 			ret = fixup_pi_state_owner(uaddr2, &q, current);
 			if (ret && rt_mutex_owner(&q.pi_state->pi_mutex) == current) {
 				pi_state = q.pi_state;
@@ -3363,8 +3328,9 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, unsigned int flags,
 			 * Drop the reference to the pi state which
 			 * the requeue_pi() code acquired for us.
 			 */
-			put_pi_state(q.pi_state);
-			spin_unlock(&hb2->lock);
+			ps_free = __put_pi_state(q.pi_state);
+			raw_spin_unlock(q.lock_ptr);
+			kfree(ps_free);
 		}
 	} else {
 		struct rt_mutex *pi_mutex;
@@ -3378,8 +3344,7 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, unsigned int flags,
 		pi_mutex = &q.pi_state->pi_mutex;
 		ret = rt_mutex_wait_proxy_lock(pi_mutex, to, &rt_waiter);
 
-		spin_lock(&hb2->lock);
-		BUG_ON(&hb2->lock != q.lock_ptr);
+		raw_spin_lock(q.lock_ptr);
 		if (ret && !rt_mutex_cleanup_proxy_lock(pi_mutex, &rt_waiter))
 			ret = 0;
 
@@ -3816,7 +3781,7 @@ static int __init futex_init(void)
 	for (i = 0; i < futex_hashsize; i++) {
 		atomic_set(&futex_queues[i].waiters, 0);
 		plist_head_init(&futex_queues[i].chain);
-		spin_lock_init(&futex_queues[i].lock);
+		raw_spin_lock_init(&futex_queues[i].lock);
 	}
 
 	return 0;