7 files changed, 239 insertions, 112 deletions

diff --git a/fs/notify/Kconfig b/fs/notify/Kconfig
index 31dac7e3b0f..dffbb0911d0 100644
--- a/fs/notify/Kconfig
+++ b/fs/notify/Kconfig
@@ -1,15 +1,5 @@
 config FSNOTIFY
-	bool "Filesystem notification backend"
-	default y
-	---help---
-	   fsnotify is a backend for filesystem notification.  fsnotify does
-	   not provide any userspace interface but does provide the basis
-	   needed for other notification schemes such as dnotify, inotify,
-	   and fanotify.
-
-	   Say Y here to enable fsnotify suport.
-
-	   If unsure, say Y.
+	def_bool n
 
 source "fs/notify/dnotify/Kconfig"
 source "fs/notify/inotify/Kconfig"
diff --git a/fs/notify/dnotify/Kconfig b/fs/notify/dnotify/Kconfig
index 904ff8d5405..f9c1ca139d8 100644
--- a/fs/notify/dnotify/Kconfig
+++ b/fs/notify/dnotify/Kconfig
@@ -1,6 +1,6 @@
 config DNOTIFY
 	bool "Dnotify support"
-	depends on FSNOTIFY
+	select FSNOTIFY
 	default y
 	help
 	  Dnotify is a directory-based per-fd file change notification system
diff --git a/fs/notify/fsnotify.c b/fs/notify/fsnotify.c
index ec2f7bd7681..037e878e03f 100644
--- a/fs/notify/fsnotify.c
+++ b/fs/notify/fsnotify.c
@@ -159,7 +159,9 @@ void fsnotify(struct inode *to_tell, __u32 mask, void *data, int data_is, const
 			if (!group->ops->should_send_event(group, to_tell, mask))
 				continue;
 			if (!event) {
-				event = fsnotify_create_event(to_tell, mask, data, data_is, file_name, cookie);
+				event = fsnotify_create_event(to_tell, mask, data,
+							      data_is, file_name, cookie,
+							      GFP_KERNEL);
 				/* shit, we OOM'd and now we can't tell, maybe
 				 * someday someone else will want to do something
 				 * here */
diff --git a/fs/notify/inotify/Kconfig b/fs/notify/inotify/Kconfig
index 5356884289a..3e56dbffe72 100644
--- a/fs/notify/inotify/Kconfig
+++ b/fs/notify/inotify/Kconfig
@@ -15,7 +15,7 @@ config INOTIFY
 
 config INOTIFY_USER
 	bool "Inotify support for userspace"
-	depends on FSNOTIFY
+	select FSNOTIFY
 	default y
 	---help---
 	  Say Y here to enable inotify support for userspace, including the
diff --git a/fs/notify/inotify/inotify_fsnotify.c b/fs/notify/inotify/inotify_fsnotify.c
index 47cd258fd24..c9ee67b442e 100644
--- a/fs/notify/inotify/inotify_fsnotify.c
+++ b/fs/notify/inotify/inotify_fsnotify.c
@@ -62,13 +62,14 @@ static int inotify_handle_event(struct fsnotify_group *group, struct fsnotify_ev
 	event_priv->wd = wd;
 
 	ret = fsnotify_add_notify_event(group, event, fsn_event_priv);
-	/* EEXIST is not an error */
-	if (ret == -EEXIST)
-		ret = 0;
-
-	/* did event_priv get attached? */
-	if (list_empty(&fsn_event_priv->event_list))
+	if (ret) {
 		inotify_free_event_priv(fsn_event_priv);
+		/* EEXIST says we tail matched, EOVERFLOW isn't something
+		 * to report up the stack. */
+		if ((ret == -EEXIST) ||
+		    (ret == -EOVERFLOW))
+			ret = 0;
+	}
 
 	/*
 	 * If we hold the entry until after the event is on the queue
@@ -104,16 +105,45 @@ static bool inotify_should_send_event(struct fsnotify_group *group, struct inode
 	return send;
 }
 
+/*
+ * This is NEVER supposed to be called.  Inotify marks should either have been
+ * removed from the idr when the watch was removed or in the
+ * fsnotify_destroy_mark_by_group() call when the inotify instance was being
+ * torn down.  This is only called if the idr is about to be freed but there
+ * are still marks in it.
+ */
 static int idr_callback(int id, void *p, void *data)
 {
-	BUG();
+	struct fsnotify_mark_entry *entry;
+	struct inotify_inode_mark_entry *ientry;
+	static bool warned = false;
+
+	if (warned)
+		return 0;
+
+	warned = false;
+	entry = p;
+	ientry = container_of(entry, struct inotify_inode_mark_entry, fsn_entry);
+
+	WARN(1, "inotify closing but id=%d for entry=%p in group=%p still in "
+		"idr.  Probably leaking memory\n", id, p, data);
+
+	/*
+	 * I'm taking the liberty of assuming that the mark in question is a
+	 * valid address and I'm dereferencing it.  This might help to figure
+	 * out why we got here and the panic is no worse than the original
+	 * BUG() that was here.
+	 */
+	if (entry)
+		printk(KERN_WARNING "entry->group=%p inode=%p wd=%d\n",
+			entry->group, entry->inode, ientry->wd);
 	return 0;
 }
 
 static void inotify_free_group_priv(struct fsnotify_group *group)
 {
 	/* ideally the idr is empty and we won't hit the BUG in teh callback */
-	idr_for_each(&group->inotify_data.idr, idr_callback, NULL);
+	idr_for_each(&group->inotify_data.idr, idr_callback, group);
 	idr_remove_all(&group->inotify_data.idr);
 	idr_destroy(&group->inotify_data.idr);
 }
diff --git a/fs/notify/inotify/inotify_user.c b/fs/notify/inotify/inotify_user.c
index ff27a296584..0e781bc88d1 100644
--- a/fs/notify/inotify/inotify_user.c
+++ b/fs/notify/inotify/inotify_user.c
@@ -47,9 +47,6 @@
 
 static struct vfsmount *inotify_mnt __read_mostly;
 
-/* this just sits here and wastes global memory.  used to just pad userspace messages with zeros */
-static struct inotify_event nul_inotify_event;
-
 /* these are configurable via /proc/sys/fs/inotify/ */
 static int inotify_max_user_instances __read_mostly;
 static int inotify_max_queued_events __read_mostly;
@@ -57,7 +54,6 @@ int inotify_max_user_watches __read_mostly;
 
 static struct kmem_cache *inotify_inode_mark_cachep __read_mostly;
 struct kmem_cache *event_priv_cachep __read_mostly;
-static struct fsnotify_event *inotify_ignored_event;
 
 /*
  * When inotify registers a new group it increments this and uses that
@@ -200,8 +196,10 @@ static ssize_t copy_event_to_user(struct fsnotify_group *group,
 		inotify_free_event_priv(fsn_priv);
 	}
 
-	/* round up event->name_len so it is a multiple of event_size */
-	name_len = roundup(event->name_len, event_size);
+	/* round up event->name_len so it is a multiple of event_size
+	 * plus an extra byte for the terminating '\0'.
+	 */
+	name_len = roundup(event->name_len + 1, event_size);
 	inotify_event.len = name_len;
 
 	inotify_event.mask = inotify_mask_to_arg(event->mask);
@@ -225,8 +223,8 @@ static ssize_t copy_event_to_user(struct fsnotify_group *group,
 			return -EFAULT;
 		buf += event->name_len;
 
-		/* fill userspace with 0's from nul_inotify_event */
-		if (copy_to_user(buf, &nul_inotify_event, len_to_zero))
+		/* fill userspace with 0's */
+		if (clear_user(buf, len_to_zero))
 			return -EFAULT;
 		buf += len_to_zero;
 		event_size += name_len;
@@ -366,20 +364,71 @@ static int inotify_find_inode(const char __user *dirname, struct path *path, uns
 }
 
 /*
- * Send IN_IGNORED for this wd, remove this wd from the idr, and drop the
- * internal reference help on the mark because it is in the idr.
+ * Remove the mark from the idr (if present) and drop the reference
+ * on the mark because it was in the idr.
+ */
+static void inotify_remove_from_idr(struct fsnotify_group *group,
+				    struct inotify_inode_mark_entry *ientry)
+{
+	struct idr *idr;
+	struct fsnotify_mark_entry *entry;
+	struct inotify_inode_mark_entry *found_ientry;
+	int wd;
+
+	spin_lock(&group->inotify_data.idr_lock);
+	idr = &group->inotify_data.idr;
+	wd = ientry->wd;
+
+	if (wd == -1)
+		goto out;
+
+	entry = idr_find(&group->inotify_data.idr, wd);
+	if (unlikely(!entry))
+		goto out;
+
+	found_ientry = container_of(entry, struct inotify_inode_mark_entry, fsn_entry);
+	if (unlikely(found_ientry != ientry)) {
+		/* We found an entry in the idr with the right wd, but it's
+		 * not the entry we were told to remove.  eparis seriously
+		 * fucked up somewhere. */
+		WARN_ON(1);
+		ientry->wd = -1;
+		goto out;
+	}
+
+	/* One ref for being in the idr, one ref held by the caller */
+	BUG_ON(atomic_read(&entry->refcnt) < 2);
+
+	idr_remove(idr, wd);
+	ientry->wd = -1;
+
+	/* removed from the idr, drop that ref */
+	fsnotify_put_mark(entry);
+out:
+	spin_unlock(&group->inotify_data.idr_lock);
+}
+
+/*
+ * Send IN_IGNORED for this wd, remove this wd from the idr.
  */
 void inotify_ignored_and_remove_idr(struct fsnotify_mark_entry *entry,
 				    struct fsnotify_group *group)
 {
 	struct inotify_inode_mark_entry *ientry;
+	struct fsnotify_event *ignored_event;
 	struct inotify_event_private_data *event_priv;
 	struct fsnotify_event_private_data *fsn_event_priv;
-	struct idr *idr;
+	int ret;
+
+	ignored_event = fsnotify_create_event(NULL, FS_IN_IGNORED, NULL,
+					      FSNOTIFY_EVENT_NONE, NULL, 0,
+					      GFP_NOFS);
+	if (!ignored_event)
+		return;
 
 	ientry = container_of(entry, struct inotify_inode_mark_entry, fsn_entry);
 
-	event_priv = kmem_cache_alloc(event_priv_cachep, GFP_KERNEL);
+	event_priv = kmem_cache_alloc(event_priv_cachep, GFP_NOFS);
 	if (unlikely(!event_priv))
 		goto skip_send_ignore;
 
@@ -388,22 +437,19 @@ void inotify_ignored_and_remove_idr(struct fsnotify_mark_entry *entry,
 	fsn_event_priv->group = group;
 	event_priv->wd = ientry->wd;
 
-	fsnotify_add_notify_event(group, inotify_ignored_event, fsn_event_priv);
-
-	/* did the private data get added? */
-	if (list_empty(&fsn_event_priv->event_list))
+	ret = fsnotify_add_notify_event(group, ignored_event, fsn_event_priv);
+	if (ret)
 		inotify_free_event_priv(fsn_event_priv);
 
 skip_send_ignore:
 
+	/* matches the reference taken when the event was created */
+	fsnotify_put_event(ignored_event);
+
 	/* remove this entry from the idr */
-	spin_lock(&group->inotify_data.idr_lock);
-	idr = &group->inotify_data.idr;
-	idr_remove(idr, ientry->wd);
-	spin_unlock(&group->inotify_data.idr_lock);
+	inotify_remove_from_idr(group, ientry);
 
-	/* removed from idr, drop that reference */
-	fsnotify_put_mark(entry);
+	atomic_dec(&group->inotify_data.user->inotify_watches);
 }
 
 /* ding dong the mark is dead */
@@ -414,67 +460,29 @@ static void inotify_free_mark(struct fsnotify_mark_entry *entry)
 	kmem_cache_free(inotify_inode_mark_cachep, ientry);
 }
 
-static int inotify_update_watch(struct fsnotify_group *group, struct inode *inode, u32 arg)
+static int inotify_update_existing_watch(struct fsnotify_group *group,
+					 struct inode *inode,
+					 u32 arg)
 {
-	struct fsnotify_mark_entry *entry = NULL;
+	struct fsnotify_mark_entry *entry;
 	struct inotify_inode_mark_entry *ientry;
-	int ret = 0;
-	int add = (arg & IN_MASK_ADD);
-	__u32 mask;
 	__u32 old_mask, new_mask;
+	__u32 mask;
+	int add = (arg & IN_MASK_ADD);
+	int ret;
 
 	/* don't allow invalid bits: we don't want flags set */
 	mask = inotify_arg_to_mask(arg);
 	if (unlikely(!mask))
 		return -EINVAL;
 
-	ientry = kmem_cache_alloc(inotify_inode_mark_cachep, GFP_KERNEL);
-	if (unlikely(!ientry))
-		return -ENOMEM;
-	/* we set the mask at the end after attaching it */
-	fsnotify_init_mark(&ientry->fsn_entry, inotify_free_mark);
-	ientry->wd = 0;
-
-find_entry:
 	spin_lock(&inode->i_lock);
 	entry = fsnotify_find_mark_entry(group, inode);
 	spin_unlock(&inode->i_lock);
-	if (entry) {
-		kmem_cache_free(inotify_inode_mark_cachep, ientry);
-		ientry = container_of(entry, struct inotify_inode_mark_entry, fsn_entry);
-	} else {
-		if (atomic_read(&group->inotify_data.user->inotify_watches) >= inotify_max_user_watches) {
-			ret = -ENOSPC;
-			goto out_err;
-		}
-
-		ret = fsnotify_add_mark(&ientry->fsn_entry, group, inode);
-		if (ret == -EEXIST)
-			goto find_entry;
-		else if (ret)
-			goto out_err;
+	if (!entry)
+		return -ENOENT;
 
-		entry = &ientry->fsn_entry;
-retry:
-		ret = -ENOMEM;
-		if (unlikely(!idr_pre_get(&group->inotify_data.idr, GFP_KERNEL)))
-			goto out_err;
-
-		spin_lock(&group->inotify_data.idr_lock);
-		/* if entry is added to the idr we keep the reference obtained
-		 * through fsnotify_mark_add.  remember to drop this reference
-		 * when entry is removed from idr */
-		ret = idr_get_new_above(&group->inotify_data.idr, entry,
-					++group->inotify_data.last_wd,
-					&ientry->wd);
-		spin_unlock(&group->inotify_data.idr_lock);
-		if (ret) {
-			if (ret == -EAGAIN)
-				goto retry;
-			goto out_err;
-		}
-		atomic_inc(&group->inotify_data.user->inotify_watches);
-	}
+	ientry = container_of(entry, struct inotify_inode_mark_entry, fsn_entry);
 
 	spin_lock(&entry->lock);
 
@@ -506,14 +514,104 @@ retry:
 			fsnotify_recalc_group_mask(group);
 	}
 
-	return ientry->wd;
+	/* return the wd */
+	ret = ientry->wd;
 
-out_err:
-	/* see this isn't supposed to happen, just kill the watch */
-	if (entry) {
-		fsnotify_destroy_mark_by_entry(entry);
-		fsnotify_put_mark(entry);
+	/* match the get from fsnotify_find_mark_entry() */
+	fsnotify_put_mark(entry);
+
+	return ret;
+}
+
+static int inotify_new_watch(struct fsnotify_group *group,
+			     struct inode *inode,
+			     u32 arg)
+{
+	struct inotify_inode_mark_entry *tmp_ientry;
+	__u32 mask;
+	int ret;
+
+	/* don't allow invalid bits: we don't want flags set */
+	mask = inotify_arg_to_mask(arg);
+	if (unlikely(!mask))
+		return -EINVAL;
+
+	tmp_ientry = kmem_cache_alloc(inotify_inode_mark_cachep, GFP_KERNEL);
+	if (unlikely(!tmp_ientry))
+		return -ENOMEM;
+
+	fsnotify_init_mark(&tmp_ientry->fsn_entry, inotify_free_mark);
+	tmp_ientry->fsn_entry.mask = mask;
+	tmp_ientry->wd = -1;
+
+	ret = -ENOSPC;
+	if (atomic_read(&group->inotify_data.user->inotify_watches) >= inotify_max_user_watches)
+		goto out_err;
+retry:
+	ret = -ENOMEM;
+	if (unlikely(!idr_pre_get(&group->inotify_data.idr, GFP_KERNEL)))
+		goto out_err;
+
+	spin_lock(&group->inotify_data.idr_lock);
+	ret = idr_get_new_above(&group->inotify_data.idr, &tmp_ientry->fsn_entry,
+				group->inotify_data.last_wd,
+				&tmp_ientry->wd);
+	spin_unlock(&group->inotify_data.idr_lock);
+	if (ret) {
+		/* idr was out of memory allocate and try again */
+		if (ret == -EAGAIN)
+			goto retry;
+		goto out_err;
+	}
+
+	/* we put the mark on the idr, take a reference */
+	fsnotify_get_mark(&tmp_ientry->fsn_entry);
+
+	/* we are on the idr, now get on the inode */
+	ret = fsnotify_add_mark(&tmp_ientry->fsn_entry, group, inode);
+	if (ret) {
+		/* we failed to get on the inode, get off the idr */
+		inotify_remove_from_idr(group, tmp_ientry);
+		goto out_err;
 	}
+
+	/* update the idr hint, who cares about races, it's just a hint */
+	group->inotify_data.last_wd = tmp_ientry->wd;
+
+	/* increment the number of watches the user has */
+	atomic_inc(&group->inotify_data.user->inotify_watches);
+
+	/* return the watch descriptor for this new entry */
+	ret = tmp_ientry->wd;
+
+	/* match the ref from fsnotify_init_markentry() */
+	fsnotify_put_mark(&tmp_ientry->fsn_entry);
+
+out_err:
+	if (ret < 0)
+		kmem_cache_free(inotify_inode_mark_cachep, tmp_ientry);
+
+	return ret;
+}
+
+static int inotify_update_watch(struct fsnotify_group *group, struct inode *inode, u32 arg)
+{
+	int ret = 0;
+
+retry:
+	/* try to update and existing watch with the new arg */
+	ret = inotify_update_existing_watch(group, inode, arg);
+	/* no mark present, try to add a new one */
+	if (ret == -ENOENT)
+		ret = inotify_new_watch(group, inode, arg);
+	/*
+	 * inotify_new_watch could race with another thread which did an
+	 * inotify_new_watch between the update_existing and the add watch
+	 * here, go back and try to update an existing mark again.
+	 */
+	if (ret == -EEXIST)
+		goto retry;
+
 	return ret;
 }
 
@@ -532,7 +630,7 @@ static struct fsnotify_group *inotify_new_group(struct user_struct *user, unsign
 
 	spin_lock_init(&group->inotify_data.idr_lock);
 	idr_init(&group->inotify_data.idr);
-	group->inotify_data.last_wd = 0;
+	group->inotify_data.last_wd = 1;
 	group->inotify_data.user = user;
 	group->inotify_data.fa = NULL;
 
@@ -721,9 +819,6 @@ static int __init inotify_user_setup(void)
 
 	inotify_inode_mark_cachep = KMEM_CACHE(inotify_inode_mark_entry, SLAB_PANIC);
 	event_priv_cachep = KMEM_CACHE(inotify_event_private_data, SLAB_PANIC);
-	inotify_ignored_event = fsnotify_create_event(NULL, FS_IN_IGNORED, NULL, FSNOTIFY_EVENT_NONE, NULL, 0);
-	if (!inotify_ignored_event)
-		panic("unable to allocate the inotify ignored event\n");
 
 	inotify_max_queued_events = 16384;
 	inotify_max_user_instances = 128;
diff --git a/fs/notify/notification.c b/fs/notify/notification.c
index 959b73e756f..3816d5750dd 100644
--- a/fs/notify/notification.c
+++ b/fs/notify/notification.c
@@ -136,18 +136,28 @@ static bool event_compare(struct fsnotify_event *old, struct fsnotify_event *new
 {
 	if ((old->mask == new->mask) &&
 	    (old->to_tell == new->to_tell) &&
-	    (old->data_type == new->data_type)) {
+	    (old->data_type == new->data_type) &&
+	    (old->name_len == new->name_len)) {
 		switch (old->data_type) {
 		case (FSNOTIFY_EVENT_INODE):
-			if (old->inode == new->inode)
+			/* remember, after old was put on the wait_q we aren't
+			 * allowed to look at the inode any more, only thing
+			 * left to check was if the file_name is the same */
+			if (old->name_len &&
+			    !strcmp(old->file_name, new->file_name))
 				return true;
 			break;
 		case (FSNOTIFY_EVENT_PATH):
 			if ((old->path.mnt == new->path.mnt) &&
 			    (old->path.dentry == new->path.dentry))
 				return true;
+			break;
 		case (FSNOTIFY_EVENT_NONE):
-			return true;
+			if (old->mask & FS_Q_OVERFLOW)
+				return true;
+			else if (old->mask & FS_IN_IGNORED)
+				return false;
+			return false;
 		};
 	}
 	return false;
@@ -165,9 +175,7 @@ int fsnotify_add_notify_event(struct fsnotify_group *group, struct fsnotify_even
 	struct list_head *list = &group->notification_list;
 	struct fsnotify_event_holder *last_holder;
 	struct fsnotify_event *last_event;
-
-	/* easy to tell if priv was attached to the event */
-	INIT_LIST_HEAD(&priv->event_list);
+	int ret = 0;
 
 	/*
 	 * There is one fsnotify_event_holder embedded inside each fsnotify_event.
@@ -188,6 +196,7 @@ alloc_holder:
 
 	if (group->q_len >= group->max_events) {
 		event = &q_overflow_event;
+		ret = -EOVERFLOW;
 		/* sorry, no private data on the overflow event */
 		priv = NULL;
 	}
@@ -229,7 +238,7 @@ alloc_holder:
 	mutex_unlock(&group->notification_mutex);
 
 	wake_up(&group->notification_waitq);
-	return 0;
+	return ret;
 }
 
 /*
@@ -339,18 +348,19 @@ static void initialize_event(struct fsnotify_event *event)
  * @name the filename, if available
  */
 struct fsnotify_event *fsnotify_create_event(struct inode *to_tell, __u32 mask, void *data,
-					     int data_type, const char *name, u32 cookie)
+					     int data_type, const char *name, u32 cookie,
+					     gfp_t gfp)
 {
 	struct fsnotify_event *event;
 
-	event = kmem_cache_alloc(fsnotify_event_cachep, GFP_KERNEL);
+	event = kmem_cache_alloc(fsnotify_event_cachep, gfp);
 	if (!event)
 		return NULL;
 
 	initialize_event(event);
 
 	if (name) {
-		event->file_name = kstrdup(name, GFP_KERNEL);
+		event->file_name = kstrdup(name, gfp);
 		if (!event->file_name) {
 			kmem_cache_free(fsnotify_event_cachep, event);
 			return NULL;