1 files changed, 256 insertions, 98 deletions

diff --git a/fs/eventpoll.c b/fs/eventpoll.c
index 9fec1836057..b10b48c2a7a 100644
--- a/fs/eventpoll.c
+++ b/fs/eventpoll.c
@@ -40,6 +40,8 @@
 #include <linux/atomic.h>
 #include <linux/proc_fs.h>
 #include <linux/seq_file.h>
+#include <linux/compat.h>
+#include <linux/rculist.h>
 
 /*
  * LOCKING:
@@ -104,7 +106,7 @@
 struct epoll_filefd {
 	struct file *file;
 	int fd;
-};
+} __packed;
 
 /*
  * Structure used to track possible nested calls, for too deep recursions
@@ -128,10 +130,16 @@ struct nested_calls {
 /*
  * Each file descriptor added to the eventpoll interface will
  * have an entry of this type linked to the "rbr" RB tree.
+ * Avoid increasing the size of this struct, there can be many thousands
+ * of these on a server and we do not want this to take another cache line.
  */
 struct epitem {
-	/* RB tree node used to link this structure to the eventpoll RB tree */
-	struct rb_node rbn;
+	union {
+		/* RB tree node links this structure to the eventpoll RB tree */
+		struct rb_node rbn;
+		/* Used to free the struct epitem */
+		struct rcu_head rcu;
+	};
 
 	/* List header used to link this structure to the eventpoll ready list */
 	struct list_head rdllink;
@@ -158,7 +166,7 @@ struct epitem {
 	struct list_head fllink;
 
 	/* wakeup_source used when EPOLLWAKEUP is set */
-	struct wakeup_source *ws;
+	struct wakeup_source __rcu *ws;
 
 	/* The structure that describe the interested events and the source fd */
 	struct epoll_event event;
@@ -285,7 +293,7 @@ static LIST_HEAD(tfile_check_list);
 static long zero;
 static long long_max = LONG_MAX;
 
-ctl_table epoll_table[] = {
+struct ctl_table epoll_table[] = {
 	{
 		.procname	= "max_user_watches",
 		.data		= &max_user_watches,
@@ -536,6 +544,38 @@ static void ep_unregister_pollwait(struct eventpoll *ep, struct epitem *epi)
 	}
 }
 
+/* call only when ep->mtx is held */
+static inline struct wakeup_source *ep_wakeup_source(struct epitem *epi)
+{
+	return rcu_dereference_check(epi->ws, lockdep_is_held(&epi->ep->mtx));
+}
+
+/* call only when ep->mtx is held */
+static inline void ep_pm_stay_awake(struct epitem *epi)
+{
+	struct wakeup_source *ws = ep_wakeup_source(epi);
+
+	if (ws)
+		__pm_stay_awake(ws);
+}
+
+static inline bool ep_has_wakeup_source(struct epitem *epi)
+{
+	return rcu_access_pointer(epi->ws) ? true : false;
+}
+
+/* call when ep->mtx cannot be held (ep_poll_callback) */
+static inline void ep_pm_stay_awake_rcu(struct epitem *epi)
+{
+	struct wakeup_source *ws;
+
+	rcu_read_lock();
+	ws = rcu_dereference(epi->ws);
+	if (ws)
+		__pm_stay_awake(ws);
+	rcu_read_unlock();
+}
+
 /**
  * ep_scan_ready_list - Scans the ready list in a way that makes possible for
  *                      the scan code, to call f_op->poll(). Also allows for
@@ -545,14 +585,14 @@ static void ep_unregister_pollwait(struct eventpoll *ep, struct epitem *epi)
  * @sproc: Pointer to the scan callback.
  * @priv: Private opaque data passed to the @sproc callback.
  * @depth: The current depth of recursive f_op->poll calls.
+ * @ep_locked: caller already holds ep->mtx
  *
  * Returns: The same integer error code returned by the @sproc callback.
  */
 static int ep_scan_ready_list(struct eventpoll *ep,
 			      int (*sproc)(struct eventpoll *,
 					   struct list_head *, void *),
-			      void *priv,
-			      int depth)
+			      void *priv, int depth, bool ep_locked)
 {
 	int error, pwake = 0;
 	unsigned long flags;
@@ -563,7 +603,9 @@ static int ep_scan_ready_list(struct eventpoll *ep,
 	 * We need to lock this because we could be hit by
 	 * eventpoll_release_file() and epoll_ctl().
 	 */
-	mutex_lock_nested(&ep->mtx, depth);
+
+	if (!ep_locked)
+		mutex_lock_nested(&ep->mtx, depth);
 
 	/*
 	 * Steal the ready list, and re-init the original one to the
@@ -599,7 +641,7 @@ static int ep_scan_ready_list(struct eventpoll *ep,
 		 */
 		if (!ep_is_linked(&epi->rdllink)) {
 			list_add_tail(&epi->rdllink, &ep->rdllist);
-			__pm_stay_awake(epi->ws);
+			ep_pm_stay_awake(epi);
 		}
 	}
 	/*
@@ -627,7 +669,8 @@ static int ep_scan_ready_list(struct eventpoll *ep,
 	}
 	spin_unlock_irqrestore(&ep->lock, flags);
 
-	mutex_unlock(&ep->mtx);
+	if (!ep_locked)
+		mutex_unlock(&ep->mtx);
 
 	/* We have to call this outside the lock */
 	if (pwake)
@@ -636,6 +679,12 @@ static int ep_scan_ready_list(struct eventpoll *ep,
 	return error;
 }
 
+static void epi_rcu_free(struct rcu_head *head)
+{
+	struct epitem *epi = container_of(head, struct epitem, rcu);
+	kmem_cache_free(epi_cache, epi);
+}
+
 /*
  * Removes a "struct epitem" from the eventpoll RB tree and deallocates
  * all the associated resources. Must be called with "mtx" held.
@@ -657,8 +706,7 @@ static int ep_remove(struct eventpoll *ep, struct epitem *epi)
 
 	/* Remove the current item from the list of epoll hooks */
 	spin_lock(&file->f_lock);
-	if (ep_is_linked(&epi->fllink))
-		list_del_init(&epi->fllink);
+	list_del_rcu(&epi->fllink);
 	spin_unlock(&file->f_lock);
 
 	rb_erase(&epi->rbn, &ep->rbr);
@@ -668,10 +716,15 @@ static int ep_remove(struct eventpoll *ep, struct epitem *epi)
 		list_del_init(&epi->rdllink);
 	spin_unlock_irqrestore(&ep->lock, flags);
 
-	wakeup_source_unregister(epi->ws);
-
-	/* At this point it is safe to free the eventpoll item */
-	kmem_cache_free(epi_cache, epi);
+	wakeup_source_unregister(ep_wakeup_source(epi));
+	/*
+	 * At this point it is safe to free the eventpoll item. Use the union
+	 * field epi->rcu, since we are trying to minimize the size of
+	 * 'struct epitem'. The 'rbn' field is no longer in use. Protected by
+	 * ep->mtx. The rcu read side, reverse_path_check_proc(), does not make
+	 * use of the rbn field.
+	 */
+	call_rcu(&epi->rcu, epi_rcu_free);
 
 	atomic_long_dec(&ep->user->epoll_watches);
 
@@ -704,6 +757,7 @@ static void ep_free(struct eventpoll *ep)
 		epi = rb_entry(rbp, struct epitem, rbn);
 
 		ep_unregister_pollwait(ep, epi);
+		cond_resched();
 	}
 
 	/*
@@ -711,11 +765,16 @@ static void ep_free(struct eventpoll *ep)
 	 * point we are sure no poll callbacks will be lingering around, and also by
 	 * holding "epmutex" we can be sure that no file cleanup code will hit
 	 * us during this operation. So we can avoid the lock on "ep->lock".
+	 * We do not need to lock ep->mtx, either, we only do it to prevent
+	 * a lockdep warning.
 	 */
+	mutex_lock(&ep->mtx);
 	while ((rbp = rb_first(&ep->rbr)) != NULL) {
 		epi = rb_entry(rbp, struct epitem, rbn);
 		ep_remove(ep, epi);
+		cond_resched();
 	}
+	mutex_unlock(&ep->mtx);
 
 	mutex_unlock(&epmutex);
 	mutex_destroy(&ep->mtx);
@@ -734,6 +793,13 @@ static int ep_eventpoll_release(struct inode *inode, struct file *file)
 	return 0;
 }
 
+static inline unsigned int ep_item_poll(struct epitem *epi, poll_table *pt)
+{
+	pt->_key = epi->event.events;
+
+	return epi->ffd.file->f_op->poll(epi->ffd.file, pt) & epi->event.events;
+}
+
 static int ep_read_events_proc(struct eventpoll *ep, struct list_head *head,
 			       void *priv)
 {
@@ -741,10 +807,9 @@ static int ep_read_events_proc(struct eventpoll *ep, struct list_head *head,
 	poll_table pt;
 
 	init_poll_funcptr(&pt, NULL);
+
 	list_for_each_entry_safe(epi, tmp, head, rdllink) {
-		pt._key = epi->event.events;
-		if (epi->ffd.file->f_op->poll(epi->ffd.file, &pt) &
-		    epi->event.events)
+		if (ep_item_poll(epi, &pt))
 			return POLLIN | POLLRDNORM;
 		else {
 			/*
@@ -752,7 +817,7 @@ static int ep_read_events_proc(struct eventpoll *ep, struct list_head *head,
 			 * callback, but it's not actually ready, as far as
 			 * caller requested events goes. We can remove it here.
 			 */
-			__pm_relax(epi->ws);
+			__pm_relax(ep_wakeup_source(epi));
 			list_del_init(&epi->rdllink);
 		}
 	}
@@ -760,15 +825,34 @@ static int ep_read_events_proc(struct eventpoll *ep, struct list_head *head,
 	return 0;
 }
 
+static void ep_ptable_queue_proc(struct file *file, wait_queue_head_t *whead,
+				 poll_table *pt);
+
+struct readyevents_arg {
+	struct eventpoll *ep;
+	bool locked;
+};
+
 static int ep_poll_readyevents_proc(void *priv, void *cookie, int call_nests)
 {
-	return ep_scan_ready_list(priv, ep_read_events_proc, NULL, call_nests + 1);
+	struct readyevents_arg *arg = priv;
+
+	return ep_scan_ready_list(arg->ep, ep_read_events_proc, NULL,
+				  call_nests + 1, arg->locked);
 }
 
 static unsigned int ep_eventpoll_poll(struct file *file, poll_table *wait)
 {
 	int pollflags;
 	struct eventpoll *ep = file->private_data;
+	struct readyevents_arg arg;
+
+	/*
+	 * During ep_insert() we already hold the ep->mtx for the tfile.
+	 * Prevent re-aquisition.
+	 */
+	arg.locked = wait && (wait->_qproc == ep_ptable_queue_proc);
+	arg.ep = ep;
 
 	/* Insert inside our poll wait queue */
 	poll_wait(file, &ep->poll_wait, wait);
@@ -780,7 +864,7 @@ static unsigned int ep_eventpoll_poll(struct file *file, poll_table *wait)
 	 * could re-enter here.
 	 */
 	pollflags = ep_call_nested(&poll_readywalk_ncalls, EP_MAX_NESTS,
-				   ep_poll_readyevents_proc, ep, ep, current);
+				   ep_poll_readyevents_proc, &arg, ep, current);
 
 	return pollflags != -1 ? pollflags : 0;
 }
@@ -825,9 +909,8 @@ static const struct file_operations eventpoll_fops = {
  */
 void eventpoll_release_file(struct file *file)
 {
-	struct list_head *lsthead = &file->f_ep_links;
 	struct eventpoll *ep;
-	struct epitem *epi;
+	struct epitem *epi, *next;
 
 	/*
 	 * We don't want to get "file->f_lock" because it is not
@@ -843,17 +926,12 @@ void eventpoll_release_file(struct file *file)
 	 * Besides, ep_remove() acquires the lock, so we can't hold it here.
 	 */
 	mutex_lock(&epmutex);
-
-	while (!list_empty(lsthead)) {
-		epi = list_first_entry(lsthead, struct epitem, fllink);
-
+	list_for_each_entry_safe(epi, next, &file->f_ep_links, fllink) {
 		ep = epi->ep;
-		list_del_init(&epi->fllink);
 		mutex_lock_nested(&ep->mtx, 0);
 		ep_remove(ep, epi);
 		mutex_unlock(&ep->mtx);
 	}
-
 	mutex_unlock(&epmutex);
 }
 
@@ -984,7 +1062,7 @@ static int ep_poll_callback(wait_queue_t *wait, unsigned mode, int sync, void *k
 	/* If this file is already in the ready list we exit soon */
 	if (!ep_is_linked(&epi->rdllink)) {
 		list_add_tail(&epi->rdllink, &ep->rdllist);
-		__pm_stay_awake(epi->ws);
+		ep_pm_stay_awake_rcu(epi);
 	}
 
 	/*
@@ -1091,7 +1169,9 @@ static int reverse_path_check_proc(void *priv, void *cookie, int call_nests)
 	struct file *child_file;
 	struct epitem *epi;
 
-	list_for_each_entry(epi, &file->f_ep_links, fllink) {
+	/* CTL_DEL can remove links here, but that can't increase our count */
+	rcu_read_lock();
+	list_for_each_entry_rcu(epi, &file->f_ep_links, fllink) {
 		child_file = epi->ep->file;
 		if (is_file_epoll(child_file)) {
 			if (list_empty(&child_file->f_ep_links)) {
@@ -1113,6 +1193,7 @@ static int reverse_path_check_proc(void *priv, void *cookie, int call_nests)
 				"file is not an ep!\n");
 		}
 	}
+	rcu_read_unlock();
 	return error;
 }
 
@@ -1146,6 +1227,7 @@ static int reverse_path_check(void)
 static int ep_create_wakeup_source(struct epitem *epi)
 {
 	const char *name;
+	struct wakeup_source *ws;
 
 	if (!epi->ep->ws) {
 		epi->ep->ws = wakeup_source_register("eventpoll");
@@ -1154,24 +1236,36 @@ static int ep_create_wakeup_source(struct epitem *epi)
 	}
 
 	name = epi->ffd.file->f_path.dentry->d_name.name;
-	epi->ws = wakeup_source_register(name);
-	if (!epi->ws)
+	ws = wakeup_source_register(name);
+
+	if (!ws)
 		return -ENOMEM;
+	rcu_assign_pointer(epi->ws, ws);
 
 	return 0;
 }
 
-static void ep_destroy_wakeup_source(struct epitem *epi)
+/* rare code path, only used when EPOLL_CTL_MOD removes a wakeup source */
+static noinline void ep_destroy_wakeup_source(struct epitem *epi)
 {
-	wakeup_source_unregister(epi->ws);
-	epi->ws = NULL;
+	struct wakeup_source *ws = ep_wakeup_source(epi);
+
+	RCU_INIT_POINTER(epi->ws, NULL);
+
+	/*
+	 * wait for ep_pm_stay_awake_rcu to finish, synchronize_rcu is
+	 * used internally by wakeup_source_remove, too (called by
+	 * wakeup_source_unregister), so we cannot use call_rcu
+	 */
+	synchronize_rcu();
+	wakeup_source_unregister(ws);
 }
 
 /*
  * Must be called with "mtx" held.
  */
 static int ep_insert(struct eventpoll *ep, struct epoll_event *event,
-		     struct file *tfile, int fd)
+		     struct file *tfile, int fd, int full_check)
 {
 	int error, revents, pwake = 0;
 	unsigned long flags;
@@ -1199,13 +1293,12 @@ static int ep_insert(struct eventpoll *ep, struct epoll_event *event,
 		if (error)
 			goto error_create_wakeup_source;
 	} else {
-		epi->ws = NULL;
+		RCU_INIT_POINTER(epi->ws, NULL);
 	}
 
 	/* Initialize the poll table using the queue callback */
 	epq.epi = epi;
 	init_poll_funcptr(&epq.pt, ep_ptable_queue_proc);
-	epq.pt._key = event->events;
 
 	/*
 	 * Attach the item to the poll hooks and get current event bits.
@@ -1214,7 +1307,7 @@ static int ep_insert(struct eventpoll *ep, struct epoll_event *event,
 	 * this operation completes, the poll callback can start hitting
 	 * the new item.
 	 */
-	revents = tfile->f_op->poll(tfile, &epq.pt);
+	revents = ep_item_poll(epi, &epq.pt);
 
 	/*
 	 * We have to check if something went wrong during the poll wait queue
@@ -1227,7 +1320,7 @@ static int ep_insert(struct eventpoll *ep, struct epoll_event *event,
 
 	/* Add the current item to the list of active epoll hook for this file */
 	spin_lock(&tfile->f_lock);
-	list_add_tail(&epi->fllink, &tfile->f_ep_links);
+	list_add_tail_rcu(&epi->fllink, &tfile->f_ep_links);
 	spin_unlock(&tfile->f_lock);
 
 	/*
@@ -1238,7 +1331,7 @@ static int ep_insert(struct eventpoll *ep, struct epoll_event *event,
 
 	/* now check if we've created too many backpaths */
 	error = -EINVAL;
-	if (reverse_path_check())
+	if (full_check && reverse_path_check())
 		goto error_remove_epi;
 
 	/* We have to drop the new item inside our item list to keep track of it */
@@ -1247,7 +1340,7 @@ static int ep_insert(struct eventpoll *ep, struct epoll_event *event,
 	/* If the file is already "ready" we drop it inside the ready list */
 	if ((revents & event->events) && !ep_is_linked(&epi->rdllink)) {
 		list_add_tail(&epi->rdllink, &ep->rdllist);
-		__pm_stay_awake(epi->ws);
+		ep_pm_stay_awake(epi);
 
 		/* Notify waiting tasks that events are available */
 		if (waitqueue_active(&ep->wq))
@@ -1268,8 +1361,7 @@ static int ep_insert(struct eventpoll *ep, struct epoll_event *event,
 
 error_remove_epi:
 	spin_lock(&tfile->f_lock);
-	if (ep_is_linked(&epi->fllink))
-		list_del_init(&epi->fllink);
+	list_del_rcu(&epi->fllink);
 	spin_unlock(&tfile->f_lock);
 
 	rb_erase(&epi->rbn, &ep->rbr);
@@ -1288,7 +1380,7 @@ error_unregister:
 		list_del_init(&epi->rdllink);
 	spin_unlock_irqrestore(&ep->lock, flags);
 
-	wakeup_source_unregister(epi->ws);
+	wakeup_source_unregister(ep_wakeup_source(epi));
 
 error_create_wakeup_source:
 	kmem_cache_free(epi_cache, epi);
@@ -1314,12 +1406,11 @@ static int ep_modify(struct eventpoll *ep, struct epitem *epi, struct epoll_even
 	 * f_op->poll() call and the new event set registering.
 	 */
 	epi->event.events = event->events; /* need barrier below */
-	pt._key = event->events;
 	epi->event.data = event->data; /* protected by mtx */
 	if (epi->event.events & EPOLLWAKEUP) {
-		if (!epi->ws)
+		if (!ep_has_wakeup_source(epi))
 			ep_create_wakeup_source(epi);
-	} else if (epi->ws) {
+	} else if (ep_has_wakeup_source(epi)) {
 		ep_destroy_wakeup_source(epi);
 	}
 
@@ -1347,7 +1438,7 @@ static int ep_modify(struct eventpoll *ep, struct epitem *epi, struct epoll_even
 	 * Get current event bits. We can safely use the file* here because
 	 * its usage count has been increased by the caller of this function.
 	 */
-	revents = epi->ffd.file->f_op->poll(epi->ffd.file, &pt);
+	revents = ep_item_poll(epi, &pt);
 
 	/*
 	 * If the item is "hot" and it is not registered inside the ready
@@ -1357,7 +1448,7 @@ static int ep_modify(struct eventpoll *ep, struct epitem *epi, struct epoll_even
 		spin_lock_irq(&ep->lock);
 		if (!ep_is_linked(&epi->rdllink)) {
 			list_add_tail(&epi->rdllink, &ep->rdllist);
-			__pm_stay_awake(epi->ws);
+			ep_pm_stay_awake(epi);
 
 			/* Notify waiting tasks that events are available */
 			if (waitqueue_active(&ep->wq))
@@ -1383,6 +1474,7 @@ static int ep_send_events_proc(struct eventpoll *ep, struct list_head *head,
 	unsigned int revents;
 	struct epitem *epi;
 	struct epoll_event __user *uevent;
+	struct wakeup_source *ws;
 	poll_table pt;
 
 	init_poll_funcptr(&pt, NULL);
@@ -1405,14 +1497,16 @@ static int ep_send_events_proc(struct eventpoll *ep, struct list_head *head,
 		 * instead, but then epi->ws would temporarily be out of sync
 		 * with ep_is_linked().
 		 */
-		if (epi->ws && epi->ws->active)
-			__pm_stay_awake(ep->ws);
-		__pm_relax(epi->ws);
+		ws = ep_wakeup_source(epi);
+		if (ws) {
+			if (ws->active)
+				__pm_stay_awake(ep->ws);
+			__pm_relax(ws);
+		}
+
 		list_del_init(&epi->rdllink);
 
-		pt._key = epi->event.events;
-		revents = epi->ffd.file->f_op->poll(epi->ffd.file, &pt) &
-			epi->event.events;
+		revents = ep_item_poll(epi, &pt);
 
 		/*
 		 * If the event mask intersect the caller-requested one,
@@ -1424,7 +1518,7 @@ static int ep_send_events_proc(struct eventpoll *ep, struct list_head *head,
 			if (__put_user(revents, &uevent->events) ||
 			    __put_user(epi->event.data, &uevent->data)) {
 				list_add(&epi->rdllink, head);
-				__pm_stay_awake(epi->ws);
+				ep_pm_stay_awake(epi);
 				return eventcnt ? eventcnt : -EFAULT;
 			}
 			eventcnt++;
@@ -1444,7 +1538,7 @@ static int ep_send_events_proc(struct eventpoll *ep, struct list_head *head,
 				 * poll callback will queue them in ep->ovflist.
 				 */
 				list_add_tail(&epi->rdllink, &ep->rdllist);
-				__pm_stay_awake(epi->ws);
+				ep_pm_stay_awake(epi);
 			}
 		}
 	}
@@ -1460,7 +1554,7 @@ static int ep_send_events(struct eventpoll *ep,
 	esed.maxevents = maxevents;
 	esed.events = events;
 
-	return ep_scan_ready_list(ep, ep_send_events_proc, &esed, 0);
+	return ep_scan_ready_list(ep, ep_send_events_proc, &esed, 0, false);
 }
 
 static inline struct timespec ep_set_mstimeout(long ms)
@@ -1730,36 +1824,35 @@ SYSCALL_DEFINE4(epoll_ctl, int, epfd, int, op, int, fd,
 		struct epoll_event __user *, event)
 {
 	int error;
-	int did_lock_epmutex = 0;
-	struct file *file, *tfile;
+	int full_check = 0;
+	struct fd f, tf;
 	struct eventpoll *ep;
 	struct epitem *epi;
 	struct epoll_event epds;
+	struct eventpoll *tep = NULL;
 
 	error = -EFAULT;
 	if (ep_op_has_event(op) &&
 	    copy_from_user(&epds, event, sizeof(struct epoll_event)))
 		goto error_return;
 
-	/* Get the "struct file *" for the eventpoll file */
 	error = -EBADF;
-	file = fget(epfd);
-	if (!file)
+	f = fdget(epfd);
+	if (!f.file)
 		goto error_return;
 
 	/* Get the "struct file *" for the target file */
-	tfile = fget(fd);
-	if (!tfile)
+	tf = fdget(fd);
+	if (!tf.file)
 		goto error_fput;
 
 	/* The target file descriptor must support poll */
 	error = -EPERM;
-	if (!tfile->f_op || !tfile->f_op->poll)
+	if (!tf.file->f_op->poll)
 		goto error_tgt_fput;
 
 	/* Check if EPOLLWAKEUP is allowed */
-	if ((epds.events & EPOLLWAKEUP) && !capable(CAP_BLOCK_SUSPEND))
-		epds.events &= ~EPOLLWAKEUP;
+	ep_take_care_of_epollwakeup(&epds);
 
 	/*
 	 * We have to check that the file structure underneath the file descriptor
@@ -1767,14 +1860,14 @@ SYSCALL_DEFINE4(epoll_ctl, int, epfd, int, op, int, fd,
 	 * adding an epoll file descriptor inside itself.
 	 */
 	error = -EINVAL;
-	if (file == tfile || !is_file_epoll(file))
+	if (f.file == tf.file || !is_file_epoll(f.file))
 		goto error_tgt_fput;
 
 	/*
 	 * At this point it is safe to assume that the "private_data" contains
 	 * our own data structure.
 	 */
-	ep = file->private_data;
+	ep = f.file->private_data;
 
 	/*
 	 * When we insert an epoll file descriptor, inside another epoll file
@@ -1784,43 +1877,54 @@ SYSCALL_DEFINE4(epoll_ctl, int, epfd, int, op, int, fd,
 	 * and hang them on the tfile_check_list, so we can check that we
 	 * haven't created too many possible wakeup paths.
 	 *
-	 * We need to hold the epmutex across both ep_insert and ep_remove
-	 * b/c we want to make sure we are looking at a coherent view of
-	 * epoll network.
+	 * We do not need to take the global 'epumutex' on EPOLL_CTL_ADD when
+	 * the epoll file descriptor is attaching directly to a wakeup source,
+	 * unless the epoll file descriptor is nested. The purpose of taking the
+	 * 'epmutex' on add is to prevent complex toplogies such as loops and
+	 * deep wakeup paths from forming in parallel through multiple
+	 * EPOLL_CTL_ADD operations.
 	 */
-	if (op == EPOLL_CTL_ADD || op == EPOLL_CTL_DEL) {
-		mutex_lock(&epmutex);
-		did_lock_epmutex = 1;
-	}
+	mutex_lock_nested(&ep->mtx, 0);
 	if (op == EPOLL_CTL_ADD) {
-		if (is_file_epoll(tfile)) {
-			error = -ELOOP;
-			if (ep_loop_check(ep, tfile) != 0) {
-				clear_tfile_check_list();
-				goto error_tgt_fput;
+		if (!list_empty(&f.file->f_ep_links) ||
+						is_file_epoll(tf.file)) {
+			full_check = 1;
+			mutex_unlock(&ep->mtx);
+			mutex_lock(&epmutex);
+			if (is_file_epoll(tf.file)) {
+				error = -ELOOP;
+				if (ep_loop_check(ep, tf.file) != 0) {
+					clear_tfile_check_list();
+					goto error_tgt_fput;
+				}
+			} else
+				list_add(&tf.file->f_tfile_llink,
+							&tfile_check_list);
+			mutex_lock_nested(&ep->mtx, 0);
+			if (is_file_epoll(tf.file)) {
+				tep = tf.file->private_data;
+				mutex_lock_nested(&tep->mtx, 1);
 			}
-		} else
-			list_add(&tfile->f_tfile_llink, &tfile_check_list);
+		}
 	}
 
-	mutex_lock_nested(&ep->mtx, 0);
-
 	/*
 	 * Try to lookup the file inside our RB tree, Since we grabbed "mtx"
 	 * above, we can be sure to be able to use the item looked up by
 	 * ep_find() till we release the mutex.
 	 */
-	epi = ep_find(ep, tfile, fd);
+	epi = ep_find(ep, tf.file, fd);
 
 	error = -EINVAL;
 	switch (op) {
 	case EPOLL_CTL_ADD:
 		if (!epi) {
 			epds.events |= POLLERR | POLLHUP;
-			error = ep_insert(ep, &epds, tfile, fd);
+			error = ep_insert(ep, &epds, tf.file, fd, full_check);
 		} else
 			error = -EEXIST;
-		clear_tfile_check_list();
+		if (full_check)
+			clear_tfile_check_list();
 		break;
 	case EPOLL_CTL_DEL:
 		if (epi)
@@ -1836,15 +1940,17 @@ SYSCALL_DEFINE4(epoll_ctl, int, epfd, int, op, int, fd,
 			error = -ENOENT;
 		break;
 	}
+	if (tep != NULL)
+		mutex_unlock(&tep->mtx);
 	mutex_unlock(&ep->mtx);
 
 error_tgt_fput:
-	if (did_lock_epmutex)
+	if (full_check)
 		mutex_unlock(&epmutex);
 
-	fput(tfile);
+	fdput(tf);
 error_fput:
-	fput(file);
+	fdput(f);
 error_return:
 
 	return error;
@@ -1916,8 +2022,8 @@ SYSCALL_DEFINE6(epoll_pwait, int, epfd, struct epoll_event __user *, events,
 			return -EINVAL;
 		if (copy_from_user(&ksigmask, sigmask, sizeof(ksigmask)))
 			return -EFAULT;
-		sigdelsetmask(&ksigmask, sigmask(SIGKILL) | sigmask(SIGSTOP));
-		sigprocmask(SIG_SETMASK, &ksigmask, &sigsaved);
+		sigsaved = current->blocked;
+		set_current_blocked(&ksigmask);
 	}
 
 	error = sys_epoll_wait(epfd, events, maxevents, timeout);
@@ -1934,12 +2040,58 @@ SYSCALL_DEFINE6(epoll_pwait, int, epfd, struct epoll_event __user *, events,
 			       sizeof(sigsaved));
 			set_restore_sigmask();
 		} else
-			sigprocmask(SIG_SETMASK, &sigsaved, NULL);
+			set_current_blocked(&sigsaved);
 	}
 
 	return error;
 }
 
+#ifdef CONFIG_COMPAT
+COMPAT_SYSCALL_DEFINE6(epoll_pwait, int, epfd,
+			struct epoll_event __user *, events,
+			int, maxevents, int, timeout,
+			const compat_sigset_t __user *, sigmask,
+			compat_size_t, sigsetsize)
+{
+	long err;
+	compat_sigset_t csigmask;
+	sigset_t ksigmask, sigsaved;
+
+	/*
+	 * If the caller wants a certain signal mask to be set during the wait,
+	 * we apply it here.
+	 */
+	if (sigmask) {
+		if (sigsetsize != sizeof(compat_sigset_t))
+			return -EINVAL;
+		if (copy_from_user(&csigmask, sigmask, sizeof(csigmask)))
+			return -EFAULT;
+		sigset_from_compat(&ksigmask, &csigmask);
+		sigsaved = current->blocked;
+		set_current_blocked(&ksigmask);
+	}
+
+	err = sys_epoll_wait(epfd, events, maxevents, timeout);
+
+	/*
+	 * If we changed the signal mask, we need to restore the original one.
+	 * In case we've got a signal while waiting, we do not restore the
+	 * signal mask yet, and we allow do_signal() to deliver the signal on
+	 * the way back to userspace, before the signal mask is restored.
+	 */
+	if (sigmask) {
+		if (err == -EINTR) {
+			memcpy(&current->saved_sigmask, &sigsaved,
+			       sizeof(sigsaved));
+			set_restore_sigmask();
+		} else
+			set_current_blocked(&sigsaved);
+	}
+
+	return err;
+}
+#endif
+
 static int __init eventpoll_init(void)
 {
 	struct sysinfo si;
@@ -1964,6 +2116,12 @@ static int __init eventpoll_init(void)
 	/* Initialize the structure used to perform file's f_op->poll() calls */
 	ep_nested_calls_init(&poll_readywalk_ncalls);
 
+	/*
+	 * We can have many thousands of epitems, so prevent this from
+	 * using an extra cache line on 64-bit (and smaller) CPUs
+	 */
+	BUILD_BUG_ON(sizeof(void *) <= 8 && sizeof(struct epitem) > 128);
+
 	/* Allocates slab cache used to allocate "struct epitem" items */
 	epi_cache = kmem_cache_create("eventpoll_epi", sizeof(struct epitem),
 			0, SLAB_HWCACHE_ALIGN | SLAB_PANIC, NULL);