Merge remote branch 'fsnotify/for-next'

26 files changed, 1186 insertions, 883 deletions

diff --git a/fs/compat.c b/fs/compat.c
index 6c19040ffeef..9ebfea0033ec 100644
--- a/fs/compat.c
+++ b/fs/compat.c
@@ -1170,11 +1170,10 @@ out:
 	if (iov != iovstack)
 		kfree(iov);
 	if ((ret + (type == READ)) > 0) {
-		struct dentry *dentry = file->f_path.dentry;
 		if (type == READ)
-			fsnotify_access(dentry);
+			fsnotify_access(file);
 		else
-			fsnotify_modify(dentry);
+			fsnotify_modify(file);
 	}
 	return ret;
 }
diff --git a/fs/exec.c b/fs/exec.c
index d164342c2b69..d391886fb9b0 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -129,7 +129,7 @@ SYSCALL_DEFINE1(uselib, const char __user *, library)
 	if (file->f_path.mnt->mnt_flags & MNT_NOEXEC)
 		goto exit;
 
-	fsnotify_open(file->f_path.dentry);
+	fsnotify_open(file);
 
 	error = -ENOEXEC;
 	if(file->f_op) {
@@ -664,7 +664,7 @@ struct file *open_exec(const char *name)
 	if (file->f_path.mnt->mnt_flags & MNT_NOEXEC)
 		goto exit;
 
-	fsnotify_open(file->f_path.dentry);
+	fsnotify_open(file);
 
 	err = deny_write_access(file);
 	if (err)
diff --git a/fs/inode.c b/fs/inode.c
index 06c1f02de611..b440974221bd 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -270,7 +270,7 @@ void inode_init_once(struct inode *inode)
 	mutex_init(&inode->inotify_mutex);
 #endif
 #ifdef CONFIG_FSNOTIFY
-	INIT_HLIST_HEAD(&inode->i_fsnotify_mark_entries);
+	INIT_HLIST_HEAD(&inode->i_fsnotify_marks);
 #endif
 }
 EXPORT_SYMBOL(inode_init_once);
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 638573968dcf..098c4b4dfa1f 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -970,7 +970,7 @@ nfsd_vfs_read(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
 		nfsdstats.io_read += host_err;
 		*count = host_err;
 		err = 0;
-		fsnotify_access(file->f_path.dentry);
+		fsnotify_access(file);
 	} else 
 		err = nfserrno(host_err);
 out:
@@ -1081,7 +1081,7 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
 		goto out_nfserr;
 	*cnt = host_err;
 	nfsdstats.io_write += host_err;
-	fsnotify_modify(file->f_path.dentry);
+	fsnotify_modify(file);
 
 	/* clear setuid/setgid flag after write */
 	if (inode->i_mode & (S_ISUID | S_ISGID))
diff --git a/fs/notify/dnotify/dnotify.c b/fs/notify/dnotify/dnotify.c
index 7e54e52964dd..cac2eb896639 100644
--- a/fs/notify/dnotify/dnotify.c
+++ b/fs/notify/dnotify/dnotify.c
@@ -29,17 +29,17 @@
 int dir_notify_enable __read_mostly = 1;
 
 static struct kmem_cache *dnotify_struct_cache __read_mostly;
-static struct kmem_cache *dnotify_mark_entry_cache __read_mostly;
+static struct kmem_cache *dnotify_mark_cache __read_mostly;
 static struct fsnotify_group *dnotify_group __read_mostly;
 static DEFINE_MUTEX(dnotify_mark_mutex);
 
 /*
- * dnotify will attach one of these to each inode (i_fsnotify_mark_entries) which
+ * dnotify will attach one of these to each inode (i_fsnotify_marks) which
  * is being watched by dnotify.  If multiple userspace applications are watching
  * the same directory with dnotify their information is chained in dn
  */
-struct dnotify_mark_entry {
-	struct fsnotify_mark_entry fsn_entry;
+struct dnotify_mark {
+	struct fsnotify_mark fsn_mark;
 	struct dnotify_struct *dn;
 };
 
@@ -51,27 +51,27 @@ struct dnotify_mark_entry {
  * it calls the fsnotify function so it can update the set of all events relevant
  * to this inode.
  */
-static void dnotify_recalc_inode_mask(struct fsnotify_mark_entry *entry)
+static void dnotify_recalc_inode_mask(struct fsnotify_mark *fsn_mark)
 {
 	__u32 new_mask, old_mask;
 	struct dnotify_struct *dn;
-	struct dnotify_mark_entry *dnentry  = container_of(entry,
-							   struct dnotify_mark_entry,
-							   fsn_entry);
+	struct dnotify_mark *dn_mark  = container_of(fsn_mark,
+						     struct dnotify_mark,
+						     fsn_mark);
 
-	assert_spin_locked(&entry->lock);
+	assert_spin_locked(&fsn_mark->lock);
 
-	old_mask = entry->mask;
+	old_mask = fsn_mark->mask;
 	new_mask = 0;
-	for (dn = dnentry->dn; dn != NULL; dn = dn->dn_next)
+	for (dn = dn_mark->dn; dn != NULL; dn = dn->dn_next)
 		new_mask |= (dn->dn_mask & ~FS_DN_MULTISHOT);
-	entry->mask = new_mask;
+	fsn_mark->mask = new_mask;
 
 	if (old_mask == new_mask)
 		return;
 
-	if (entry->inode)
-		fsnotify_recalc_inode_mask(entry->inode);
+	if (fsn_mark->i.inode)
+		fsnotify_recalc_inode_mask(fsn_mark->i.inode);
 }
 
 /*
@@ -85,8 +85,8 @@ static void dnotify_recalc_inode_mask(struct fsnotify_mark_entry *entry)
 static int dnotify_handle_event(struct fsnotify_group *group,
 				struct fsnotify_event *event)
 {
-	struct fsnotify_mark_entry *entry = NULL;
-	struct dnotify_mark_entry *dnentry;
+	struct fsnotify_mark *fsn_mark = NULL;
+	struct dnotify_mark *dn_mark;
 	struct inode *to_tell;
 	struct dnotify_struct *dn;
 	struct dnotify_struct **prev;
@@ -95,17 +95,13 @@ static int dnotify_handle_event(struct fsnotify_group *group,
 
 	to_tell = event->to_tell;
 
-	spin_lock(&to_tell->i_lock);
-	entry = fsnotify_find_mark_entry(group, to_tell);
-	spin_unlock(&to_tell->i_lock);
-
-	/* unlikely since we alreay passed dnotify_should_send_event() */
-	if (unlikely(!entry))
+	fsn_mark = fsnotify_find_mark(group, to_tell);
+	if (unlikely(!fsn_mark))
 		return 0;
-	dnentry = container_of(entry, struct dnotify_mark_entry, fsn_entry);
+	dn_mark = container_of(fsn_mark, struct dnotify_mark, fsn_mark);
 
-	spin_lock(&entry->lock);
-	prev = &dnentry->dn;
+	spin_lock(&fsn_mark->lock);
+	prev = &dn_mark->dn;
 	while ((dn = *prev) != NULL) {
 		if ((dn->dn_mask & test_mask) == 0) {
 			prev = &dn->dn_next;
@@ -118,12 +114,12 @@ static int dnotify_handle_event(struct fsnotify_group *group,
 		else {
 			*prev = dn->dn_next;
 			kmem_cache_free(dnotify_struct_cache, dn);
-			dnotify_recalc_inode_mask(entry);
+			dnotify_recalc_inode_mask(fsn_mark);
 		}
 	}
 
-	spin_unlock(&entry->lock);
-	fsnotify_put_mark(entry);
+	spin_unlock(&fsn_mark->lock);
+	fsnotify_put_mark(fsn_mark);
 
 	return 0;
 }
@@ -133,9 +129,10 @@ static int dnotify_handle_event(struct fsnotify_group *group,
  * userspace notification for that pair.
  */
 static bool dnotify_should_send_event(struct fsnotify_group *group,
-				      struct inode *inode, __u32 mask)
+				      struct inode *inode, struct vfsmount *mnt,
+				      __u32 mask, void *data, int data_type)
 {
-	struct fsnotify_mark_entry *entry;
+	struct fsnotify_mark *fsn_mark;
 	bool send;
 
 	/* !dir_notify_enable should never get here, don't waste time checking
@@ -146,31 +143,27 @@ static bool dnotify_should_send_event(struct fsnotify_group *group,
 	if (!S_ISDIR(inode->i_mode))
 		return false;
 
-	spin_lock(&inode->i_lock);
-	entry = fsnotify_find_mark_entry(group, inode);
-	spin_unlock(&inode->i_lock);
-
-	/* no mark means no dnotify watch */
-	if (!entry)
+	fsn_mark = fsnotify_find_mark(group, inode);
+	if (!fsn_mark)
 		return false;
 
 	mask = (mask & ~FS_EVENT_ON_CHILD);
-	send = (mask & entry->mask);
+	send = (mask & fsn_mark->mask);
 
-	fsnotify_put_mark(entry); /* matches fsnotify_find_mark_entry */
+	fsnotify_put_mark(fsn_mark); /* matches fsnotify_find_mark */
 
 	return send;
 }
 
-static void dnotify_free_mark(struct fsnotify_mark_entry *entry)
+static void dnotify_free_mark(struct fsnotify_mark *fsn_mark)
 {
-	struct dnotify_mark_entry *dnentry = container_of(entry,
-							  struct dnotify_mark_entry,
-							  fsn_entry);
+	struct dnotify_mark *dn_mark = container_of(fsn_mark,
+						    struct dnotify_mark,
+						    fsn_mark);
 
-	BUG_ON(dnentry->dn);
+	BUG_ON(dn_mark->dn);
 
-	kmem_cache_free(dnotify_mark_entry_cache, dnentry);
+	kmem_cache_free(dnotify_mark_cache, dn_mark);
 }
 
 static struct fsnotify_ops dnotify_fsnotify_ops = {
@@ -183,15 +176,15 @@ static struct fsnotify_ops dnotify_fsnotify_ops = {
 
 /*
  * Called every time a file is closed.  Looks first for a dnotify mark on the
- * inode.  If one is found run all of the ->dn entries attached to that
+ * inode.  If one is found run all of the ->dn structures attached to that
  * mark for one relevant to this process closing the file and remove that
  * dnotify_struct.  If that was the last dnotify_struct also remove the
- * fsnotify_mark_entry.
+ * fsnotify_mark.
  */
 void dnotify_flush(struct file *filp, fl_owner_t id)
 {
-	struct fsnotify_mark_entry *entry;
-	struct dnotify_mark_entry *dnentry;
+	struct fsnotify_mark *fsn_mark;
+	struct dnotify_mark *dn_mark;
 	struct dnotify_struct *dn;
 	struct dnotify_struct **prev;
 	struct inode *inode;
@@ -200,38 +193,36 @@ void dnotify_flush(struct file *filp, fl_owner_t id)
 	if (!S_ISDIR(inode->i_mode))
 		return;
 
-	spin_lock(&inode->i_lock);
-	entry = fsnotify_find_mark_entry(dnotify_group, inode);
-	spin_unlock(&inode->i_lock);
-	if (!entry)
+	fsn_mark = fsnotify_find_mark(dnotify_group, inode);
+	if (!fsn_mark)
 		return;
-	dnentry = container_of(entry, struct dnotify_mark_entry, fsn_entry);
+	dn_mark = container_of(fsn_mark, struct dnotify_mark, fsn_mark);
 
 	mutex_lock(&dnotify_mark_mutex);
 
-	spin_lock(&entry->lock);
-	prev = &dnentry->dn;
+	spin_lock(&fsn_mark->lock);
+	prev = &dn_mark->dn;
 	while ((dn = *prev) != NULL) {
 		if ((dn->dn_owner == id) && (dn->dn_filp == filp)) {
 			*prev = dn->dn_next;
 			kmem_cache_free(dnotify_struct_cache, dn);
-			dnotify_recalc_inode_mask(entry);
+			dnotify_recalc_inode_mask(fsn_mark);
 			break;
 		}
 		prev = &dn->dn_next;
 	}
 
-	spin_unlock(&entry->lock);
+	spin_unlock(&fsn_mark->lock);
 
 	/* nothing else could have found us thanks to the dnotify_mark_mutex */
-	if (dnentry->dn == NULL)
-		fsnotify_destroy_mark_by_entry(entry);
+	if (dn_mark->dn == NULL)
+		fsnotify_destroy_mark(fsn_mark);
 
 	fsnotify_recalc_group_mask(dnotify_group);
 
 	mutex_unlock(&dnotify_mark_mutex);
 
-	fsnotify_put_mark(entry);
+	fsnotify_put_mark(fsn_mark);
 }
 
 /* this conversion is done only at watch creation */
@@ -259,16 +250,16 @@ static __u32 convert_arg(unsigned long arg)
 
 /*
  * If multiple processes watch the same inode with dnotify there is only one
- * dnotify mark in inode->i_fsnotify_mark_entries but we chain a dnotify_struct
+ * dnotify mark in inode->i_fsnotify_marks but we chain a dnotify_struct
  * onto that mark.  This function either attaches the new dnotify_struct onto
  * that list, or it |= the mask onto an existing dnofiy_struct.
  */
-static int attach_dn(struct dnotify_struct *dn, struct dnotify_mark_entry *dnentry,
+static int attach_dn(struct dnotify_struct *dn, struct dnotify_mark *dn_mark,
 		     fl_owner_t id, int fd, struct file *filp, __u32 mask)
 {
 	struct dnotify_struct *odn;
 
-	odn = dnentry->dn;
+	odn = dn_mark->dn;
 	while (odn != NULL) {
 		/* adding more events to existing dnofiy_struct? */
 		if ((odn->dn_owner == id) && (odn->dn_filp == filp)) {
@@ -283,8 +274,8 @@ static int attach_dn(struct dnotify_struct *dn, struct dnotify_mark_entry *dnent
 	dn->dn_fd = fd;
 	dn->dn_filp = filp;
 	dn->dn_owner = id;
-	dn->dn_next = dnentry->dn;
-	dnentry->dn = dn;
+	dn->dn_next = dn_mark->dn;
+	dn_mark->dn = dn;
 
 	return 0;
 }
@@ -296,8 +287,8 @@ static int attach_dn(struct dnotify_struct *dn, struct dnotify_mark_entry *dnent
  */
 int fcntl_dirnotify(int fd, struct file *filp, unsigned long arg)
 {
-	struct dnotify_mark_entry *new_dnentry, *dnentry;
-	struct fsnotify_mark_entry *new_entry, *entry;
+	struct dnotify_mark *new_dn_mark, *dn_mark;
+	struct fsnotify_mark *new_fsn_mark, *fsn_mark;
 	struct dnotify_struct *dn;
 	struct inode *inode;
 	fl_owner_t id = current->files;
@@ -306,7 +297,7 @@ int fcntl_dirnotify(int fd, struct file *filp, unsigned long arg)
 	__u32 mask;
 
 	/* we use these to tell if we need to kfree */
-	new_entry = NULL;
+	new_fsn_mark = NULL;
 	dn = NULL;
 
 	if (!dir_notify_enable) {
@@ -336,8 +327,8 @@ int fcntl_dirnotify(int fd, struct file *filp, unsigned long arg)
 	}
 
 	/* new fsnotify mark, we expect most fcntl calls to add a new mark */
-	new_dnentry = kmem_cache_alloc(dnotify_mark_entry_cache, GFP_KERNEL);
-	if (!new_dnentry) {
+	new_dn_mark = kmem_cache_alloc(dnotify_mark_cache, GFP_KERNEL);
+	if (!new_dn_mark) {
 		error = -ENOMEM;
 		goto out_err;
 	}
@@ -345,29 +336,27 @@ int fcntl_dirnotify(int fd, struct file *filp, unsigned long arg)
 	/* convert the userspace DN_* "arg" to the internal FS_* defines in fsnotify */
 	mask = convert_arg(arg);
 
-	/* set up the new_entry and new_dnentry */
-	new_entry = &new_dnentry->fsn_entry;
-	fsnotify_init_mark(new_entry, dnotify_free_mark);
-	new_entry->mask = mask;
-	new_dnentry->dn = NULL;
+	/* set up the new_fsn_mark and new_dn_mark */
+	new_fsn_mark = &new_dn_mark->fsn_mark;
+	fsnotify_init_mark(new_fsn_mark, dnotify_free_mark);
+	new_fsn_mark->mask = mask;
+	new_dn_mark->dn = NULL;
 
 	/* this is needed to prevent the fcntl/close race described below */
 	mutex_lock(&dnotify_mark_mutex);
 
-	/* add the new_entry or find an old one. */
-	spin_lock(&inode->i_lock);
-	entry = fsnotify_find_mark_entry(dnotify_group, inode);
-	spin_unlock(&inode->i_lock);
-	if (entry) {
-		dnentry = container_of(entry, struct dnotify_mark_entry, fsn_entry);
-		spin_lock(&entry->lock);
+	/* add the new_fsn_mark or find an old one. */
+	fsn_mark = fsnotify_find_mark(dnotify_group, inode);
+	if (fsn_mark) {
+		dn_mark = container_of(fsn_mark, struct dnotify_mark, fsn_mark);
+		spin_lock(&fsn_mark->lock);
 	} else {
-		fsnotify_add_mark(new_entry, dnotify_group, inode);
-		spin_lock(&new_entry->lock);
-		entry = new_entry;
-		dnentry = new_dnentry;
-		/* we used new_entry, so don't free it */
-		new_entry = NULL;
+		fsnotify_add_mark(new_fsn_mark, dnotify_group, inode, 0);
+		spin_lock(&new_fsn_mark->lock);
+		fsn_mark = new_fsn_mark;
+		dn_mark = new_dn_mark;
+		/* we used new_fsn_mark, so don't free it */
+		new_fsn_mark = NULL;
 	}
 
 	rcu_read_lock();
@@ -376,17 +365,17 @@ int fcntl_dirnotify(int fd, struct file *filp, unsigned long arg)
 
 	/* if (f != filp) means that we lost a race and another task/thread
 	 * actually closed the fd we are still playing with before we grabbed
-	 * the dnotify_mark_mutex and entry->lock.  Since closing the fd is the
-	 * only time we clean up the mark entries we need to get our mark off
+	 * the dnotify_mark_mutex and fsn_mark->lock.  Since closing the fd is the
+	 * only time we clean up the marks we need to get our mark off
 	 * the list. */
 	if (f != filp) {
 		/* if we added ourselves, shoot ourselves, it's possible that
-		 * the flush actually did shoot this entry.  That's fine too
+		 * the flush actually did shoot this fsn_mark.  That's fine too
 		 * since multiple calls to destroy_mark is perfectly safe, if
-		 * we found a dnentry already attached to the inode, just sod
+		 * we found a dn_mark already attached to the inode, just sod
 		 * off silently as the flush at close time dealt with it.
 		 */
-		if (dnentry == new_dnentry)
+		if (dn_mark == new_dn_mark)
 			destroy = 1;
 		goto out;
 	}
@@ -394,13 +383,13 @@ int fcntl_dirnotify(int fd, struct file *filp, unsigned long arg)
 	error = __f_setown(filp, task_pid(current), PIDTYPE_PID, 0);
 	if (error) {
 		/* if we added, we must shoot */
-		if (dnentry == new_dnentry)
+		if (dn_mark == new_dn_mark)
 			destroy = 1;
 		goto out;
 	}
 
-	error = attach_dn(dn, dnentry, id, fd, filp, mask);
-	/* !error means that we attached the dn to the dnentry, so don't free it */
+	error = attach_dn(dn, dn_mark, id, fd, filp, mask);
+	/* !error means that we attached the dn to the dn_mark, so don't free it */
 	if (!error)
 		dn = NULL;
 	/* -EEXIST means that we didn't add this new dn and used an old one.
@@ -408,20 +397,20 @@ int fcntl_dirnotify(int fd, struct file *filp, unsigned long arg)
 	else if (error == -EEXIST)
 		error = 0;
 
-	dnotify_recalc_inode_mask(entry);
+	dnotify_recalc_inode_mask(fsn_mark);
 out:
-	spin_unlock(&entry->lock);
+	spin_unlock(&fsn_mark->lock);
 
 	if (destroy)
-		fsnotify_destroy_mark_by_entry(entry);
+		fsnotify_destroy_mark(fsn_mark);
 
 	fsnotify_recalc_group_mask(dnotify_group);
 
 	mutex_unlock(&dnotify_mark_mutex);
-	fsnotify_put_mark(entry);
+	fsnotify_put_mark(fsn_mark);
 out_err:
-	if (new_entry)
-		fsnotify_put_mark(new_entry);
+	if (new_fsn_mark)
+		fsnotify_put_mark(new_fsn_mark);
 	if (dn)
 		kmem_cache_free(dnotify_struct_cache, dn);
 	return error;
@@ -430,10 +419,9 @@ out_err:
 static int __init dnotify_init(void)
 {
 	dnotify_struct_cache = KMEM_CACHE(dnotify_struct, SLAB_PANIC);
-	dnotify_mark_entry_cache = KMEM_CACHE(dnotify_mark_entry, SLAB_PANIC);
+	dnotify_mark_cache = KMEM_CACHE(dnotify_mark, SLAB_PANIC);
 
-	dnotify_group = fsnotify_obtain_group(DNOTIFY_GROUP_NUM,
-					      0, &dnotify_fsnotify_ops);
+	dnotify_group = fsnotify_alloc_group(&dnotify_fsnotify_ops);
 	if (IS_ERR(dnotify_group))
 		panic("unable to allocate fsnotify group for dnotify\n");
 	return 0;
diff --git a/fs/notify/fsnotify.c b/fs/notify/fsnotify.c
index 037e878e03fc..b076b062ce9b 100644
--- a/fs/notify/fsnotify.c
+++ b/fs/notify/fsnotify.c
@@ -20,6 +20,7 @@
 #include <linux/fs.h>
 #include <linux/init.h>
 #include <linux/module.h>
+#include <linux/mount.h>
 #include <linux/srcu.h>
 
 #include <linux/fsnotify_backend.h>
@@ -77,13 +78,16 @@ void __fsnotify_update_child_dentry_flags(struct inode *inode)
 }
 
 /* Notify this dentry's parent about a child's events. */
-void __fsnotify_parent(struct dentry *dentry, __u32 mask)
+void __fsnotify_parent(struct path *path, struct dentry *dentry, __u32 mask)
 {
 	struct dentry *parent;
 	struct inode *p_inode;
 	bool send = false;
 	bool should_update_children = false;
 
+	if (!dentry)
+		dentry = path->dentry;
+
 	if (!(dentry->d_flags & DCACHE_FSNOTIFY_PARENT_WATCHED))
 		return;
 
@@ -114,8 +118,12 @@ void __fsnotify_parent(struct dentry *dentry, __u32 mask)
 		 * specifies these are events which came from a child. */
 		mask |= FS_EVENT_ON_CHILD;
 
-		fsnotify(p_inode, mask, dentry->d_inode, FSNOTIFY_EVENT_INODE,
-			 dentry->d_name.name, 0);
+		if (path)
+			fsnotify(p_inode, mask, path, FSNOTIFY_EVENT_PATH,
+				 dentry->d_name.name, 0);
+		else
+			fsnotify(p_inode, mask, dentry->d_inode, FSNOTIFY_EVENT_INODE,
+				 dentry->d_name.name, 0);
 		dput(parent);
 	}
 
@@ -126,6 +134,38 @@ void __fsnotify_parent(struct dentry *dentry, __u32 mask)
 }
 EXPORT_SYMBOL_GPL(__fsnotify_parent);
 
+static void send_to_group(struct fsnotify_group *group, struct inode *to_tell,
+			  struct vfsmount *mnt, __u32 mask, void *data,
+			  int data_is, u32 cookie, const char *file_name,
+			  struct fsnotify_event **event)
+{
+	if (!group->ops->should_send_event(group, to_tell, mnt, mask,
+					   data, data_is))
+		return;
+	if (!*event) {
+		*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
+		 */
+		if (!*event)
+			return;
+	}
+	group->ops->handle_event(group, *event);
+}
+
+static bool needed_by_vfsmount(__u32 test_mask, struct vfsmount *mnt)
+{
+	if (!mnt)
+		return false;
+
+	/* hook in this when mnt->mnt_fsnotify_mask is defined */
+	/* return (test_mask & path->mnt->mnt_fsnotify_mask); */
+	return false;
+}
 /*
  * This is the main call to fsnotify.  The VFS calls into hook specific functions
  * in linux/fsnotify.h.  Those functions then in turn call here.  Here will call
@@ -136,41 +176,54 @@ void fsnotify(struct inode *to_tell, __u32 mask, void *data, int data_is, const
 {
 	struct fsnotify_group *group;
 	struct fsnotify_event *event = NULL;
+	struct vfsmount *mnt = NULL;
 	int idx;
 	/* global tests shouldn't care about events on child only the specific event */
 	__u32 test_mask = (mask & ~FS_EVENT_ON_CHILD);
 
-	if (list_empty(&fsnotify_groups))
-		return;
-
-	if (!(test_mask & fsnotify_mask))
-		return;
+	/* if no fsnotify listeners, nothing to do */
+	if (list_empty(&fsnotify_inode_groups) &&
+	    list_empty(&fsnotify_vfsmount_groups))
+                return;
+ 
+	/* if none of the directed listeners or vfsmount listeners care */
+	if (!(test_mask & fsnotify_inode_mask) &&
+	    !(test_mask & fsnotify_vfsmount_mask))
+                return;
+ 
+	if (data_is == FSNOTIFY_EVENT_PATH)
+		mnt = ((struct path *)data)->mnt;
+
+	/* if this inode's directed listeners don't care and nothing on the vfsmount
+	 * listeners list cares, nothing to do */
+	if (!(test_mask & to_tell->i_fsnotify_mask) &&
+	    !needed_by_vfsmount(test_mask, mnt))
+                return;
 
-	if (!(test_mask & to_tell->i_fsnotify_mask))
-		return;
 	/*
 	 * SRCU!!  the groups list is very very much read only and the path is
 	 * very hot.  The VAST majority of events are not going to need to do
 	 * anything other than walk the list so it's crazy to pre-allocate.
 	 */
 	idx = srcu_read_lock(&fsnotify_grp_srcu);
-	list_for_each_entry_rcu(group, &fsnotify_groups, group_list) {
-		if (test_mask & group->mask) {
-			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,
-							      GFP_KERNEL);
-				/* shit, we OOM'd and now we can't tell, maybe
-				 * someday someone else will want to do something
-				 * here */
-				if (!event)
-					break;
+
+	if (test_mask & to_tell->i_fsnotify_mask) {
+		list_for_each_entry_rcu(group, &fsnotify_inode_groups, inode_group_list) {
+			if (test_mask & group->mask) {
+				send_to_group(group, to_tell, NULL, mask, data, data_is,
+					      cookie, file_name, &event);
 			}
-			group->ops->handle_event(group, event);
 		}
 	}
+	if (needed_by_vfsmount(test_mask, mnt)) {
+		list_for_each_entry_rcu(group, &fsnotify_vfsmount_groups, vfsmount_group_list) {
+			if (test_mask & group->mask) {
+				send_to_group(group, to_tell, mnt, mask, data, data_is,
+					      cookie, file_name, &event);
+			}
+		}
+	}
+
 	srcu_read_unlock(&fsnotify_grp_srcu, idx);
 	/*
 	 * fsnotify_create_event() took a reference so the event can't be cleaned
diff --git a/fs/notify/fsnotify.h b/fs/notify/fsnotify.h
index 4dc240824b2d..2ba59158969f 100644
--- a/fs/notify/fsnotify.h
+++ b/fs/notify/fsnotify.h
@@ -8,14 +8,22 @@
 
 /* protects reads of fsnotify_groups */
 extern struct srcu_struct fsnotify_grp_srcu;
-/* all groups which receive fsnotify events */
-extern struct list_head fsnotify_groups;
-/* all bitwise OR of all event types (FS_*) for all fsnotify_groups */
-extern __u32 fsnotify_mask;
+/* all groups which receive inode fsnotify events */
+extern struct list_head fsnotify_inode_groups;
+/* all groups which receive vfsmount fsnotify events */
+extern struct list_head fsnotify_vfsmount_groups;
+/* all bitwise OR of all event types (FS_*) for all fsnotify_inode_groups */
+extern __u32 fsnotify_inode_mask;
+/* all bitwise OR of all event types (FS_*) for all fsnotify_vfsmount_groups */
+extern __u32 fsnotify_vfsmount_mask;
 
 /* destroy all events sitting in this groups notification queue */
 extern void fsnotify_flush_notify(struct fsnotify_group *group);
 
+/* add a group to the inode group list */
+extern void fsnotify_add_inode_group(struct fsnotify_group *group);
+/* add a group to the vfsmount group list */
+extern void fsnotify_add_vfsmount_group(struct fsnotify_group *group);
 /* final kfree of a group */
 extern void fsnotify_final_destroy_group(struct fsnotify_group *group);
 
diff --git a/fs/notify/group.c b/fs/notify/group.c
index 0e1677144bc5..9e9eb406afdd 100644
--- a/fs/notify/group.c
+++ b/fs/notify/group.c
@@ -32,10 +32,14 @@
 static DEFINE_MUTEX(fsnotify_grp_mutex);
 /* protects reads while running the fsnotify_groups list */
 struct srcu_struct fsnotify_grp_srcu;
-/* all groups registered to receive filesystem notifications */
-LIST_HEAD(fsnotify_groups);
+/* all groups registered to receive inode filesystem notifications */
+LIST_HEAD(fsnotify_inode_groups);
+/* all groups registered to receive mount point filesystem notifications */
+LIST_HEAD(fsnotify_vfsmount_groups);
 /* bitwise OR of all events (FS_*) interesting to some group on this system */
-__u32 fsnotify_mask;
+__u32 fsnotify_inode_mask;
+/* bitwise OR of all events (FS_*) interesting to some group on this system */
+__u32 fsnotify_vfsmount_mask;
 
 /*
  * When a new group registers or changes it's set of interesting events
@@ -44,14 +48,20 @@ __u32 fsnotify_mask;
 void fsnotify_recalc_global_mask(void)
 {
 	struct fsnotify_group *group;
-	__u32 mask = 0;
+	__u32 inode_mask = 0;
+	__u32 vfsmount_mask = 0;
 	int idx;
 
 	idx = srcu_read_lock(&fsnotify_grp_srcu);
-	list_for_each_entry_rcu(group, &fsnotify_groups, group_list)
-		mask |= group->mask;
+	list_for_each_entry_rcu(group, &fsnotify_inode_groups, inode_group_list)
+		inode_mask |= group->mask;
+	list_for_each_entry_rcu(group, &fsnotify_vfsmount_groups, vfsmount_group_list)
+		vfsmount_mask |= group->mask;
+		
 	srcu_read_unlock(&fsnotify_grp_srcu, idx);
-	fsnotify_mask = mask;
+
+	fsnotify_inode_mask = inode_mask;
+	fsnotify_vfsmount_mask = vfsmount_mask;
 }
 
 /*
@@ -64,11 +74,11 @@ void fsnotify_recalc_group_mask(struct fsnotify_group *group)
 {
 	__u32 mask = 0;
 	__u32 old_mask = group->mask;
-	struct fsnotify_mark_entry *entry;
+	struct fsnotify_mark *mark;
 
 	spin_lock(&group->mark_lock);
-	list_for_each_entry(entry, &group->mark_entries, g_list)
-		mask |= entry->mask;
+	list_for_each_entry(mark, &group->marks_list, g_list)
+		mask |= mark->mask;
 	spin_unlock(&group->mark_lock);
 
 	group->mask = mask;
@@ -77,13 +87,26 @@ void fsnotify_recalc_group_mask(struct fsnotify_group *group)
 		fsnotify_recalc_global_mask();
 }
 
-/*
- * Take a reference to a group so things found under the fsnotify_grp_mutex
- * can't get freed under us
- */
-static void fsnotify_get_group(struct fsnotify_group *group)
+void fsnotify_add_vfsmount_group(struct fsnotify_group *group)
 {
-	atomic_inc(&group->refcnt);
+	mutex_lock(&fsnotify_grp_mutex);
+
+	if (!group->on_vfsmount_group_list)
+		list_add_tail_rcu(&group->vfsmount_group_list, &fsnotify_vfsmount_groups);
+	group->on_vfsmount_group_list = 1;
+
+	mutex_unlock(&fsnotify_grp_mutex);
+}
+
+void fsnotify_add_inode_group(struct fsnotify_group *group)
+{
+	mutex_lock(&fsnotify_grp_mutex);
+
+	if (!group->on_inode_group_list)
+		list_add_tail_rcu(&group->inode_group_list, &fsnotify_inode_groups);
+	group->on_inode_group_list = 1;
+
+	mutex_unlock(&fsnotify_grp_mutex);
 }
 
 /*
@@ -110,7 +133,7 @@ void fsnotify_final_destroy_group(struct fsnotify_group *group)
  */
 static void fsnotify_destroy_group(struct fsnotify_group *group)
 {
-	/* clear all inode mark entries for this group */
+	/* clear all inode marks for this group */
 	fsnotify_clear_marks_by_group(group);
 
 	/* past the point of no return, matches the initial value of 1 */
@@ -127,9 +150,12 @@ static void __fsnotify_evict_group(struct fsnotify_group *group)
 {
 	BUG_ON(!mutex_is_locked(&fsnotify_grp_mutex));
 
-	if (group->on_group_list)
-		list_del_rcu(&group->group_list);
-	group->on_group_list = 0;
+	if (group->on_inode_group_list)
+		list_del_rcu(&group->inode_group_list);
+	group->on_inode_group_list = 0;
+	if (group->on_vfsmount_group_list)
+		list_del_rcu(&group->vfsmount_group_list);
+	group->on_vfsmount_group_list = 0;
 }
 
 /*
@@ -171,84 +197,36 @@ void fsnotify_put_group(struct fsnotify_group *group)
 }
 
 /*
- * Simply run the fsnotify_groups list and find a group which matches
- * the given parameters.  If a group is found we take a reference to that
- * group.
+ * Create a new fsnotify_group and hold a reference for the group returned.
  */
-static struct fsnotify_group *fsnotify_find_group(unsigned int group_num, __u32 mask,
-						  const struct fsnotify_ops *ops)
+struct fsnotify_group *fsnotify_alloc_group(const struct fsnotify_ops *ops)
 {
-	struct fsnotify_group *group_iter;
-	struct fsnotify_group *group = NULL;
-
-	BUG_ON(!mutex_is_locked(&fsnotify_grp_mutex));
-
-	list_for_each_entry_rcu(group_iter, &fsnotify_groups, group_list) {
-		if (group_iter->group_num == group_num) {
-			if ((group_iter->mask == mask) &&
-			    (group_iter->ops == ops)) {
-				fsnotify_get_group(group_iter);
-				group = group_iter;
-			} else
-				group = ERR_PTR(-EEXIST);
-		}
-	}
-	return group;
-}
-
-/*
- * Either finds an existing group which matches the group_num, mask, and ops or
- * creates a new group and adds it to the global group list.  In either case we
- * take a reference for the group returned.
- */
-struct fsnotify_group *fsnotify_obtain_group(unsigned int group_num, __u32 mask,
-					     const struct fsnotify_ops *ops)
-{
-	struct fsnotify_group *group, *tgroup;
+	struct fsnotify_group *group;
 
-	/* very low use, simpler locking if we just always alloc */
-	group = kmalloc(sizeof(struct fsnotify_group), GFP_KERNEL);
+	group = kzalloc(sizeof(struct fsnotify_group), GFP_KERNEL);
 	if (!group)
 		return ERR_PTR(-ENOMEM);
 
+	/* set to 0 when there a no external references to this group */
 	atomic_set(&group->refcnt, 1);
-
-	group->on_group_list = 0;
-	group->group_num = group_num;
-	group->mask = mask;
+	/*
+	 * hits 0 when there are no external references AND no marks for
+	 * this group
+	 */
+	atomic_set(&group->num_marks, 1);
 
 	mutex_init(&group->notification_mutex);
 	INIT_LIST_HEAD(&group->notification_list);
 	init_waitqueue_head(&group->notification_waitq);
-	group->q_len = 0;
 	group->max_events = UINT_MAX;
 
+	INIT_LIST_HEAD(&group->inode_group_list);
+	INIT_LIST_HEAD(&group->vfsmount_group_list);
+
 	spin_lock_init(&group->mark_lock);
-	atomic_set(&group->num_marks, 0);
-	INIT_LIST_HEAD(&group->mark_entries);
+	INIT_LIST_HEAD(&group->marks_list);
 
 	group->ops = ops;
 
-	mutex_lock(&fsnotify_grp_mutex);
-	tgroup = fsnotify_find_group(group_num, mask, ops);
-	if (tgroup) {
-		/* group already exists */
-		mutex_unlock(&fsnotify_grp_mutex);
-		/* destroy the new one we made */
-		fsnotify_put_group(group);
-		return tgroup;
-	}
-
-	/* group not found, add a new one */
-	list_add_rcu(&group->group_list, &fsnotify_groups);
-	group->on_group_list = 1;
-	/* being on the fsnotify_groups list holds one num_marks */
-	atomic_inc(&group->num_marks);
-
-	mutex_unlock(&fsnotify_grp_mutex);
-
-	if (mask)
-		fsnotify_recalc_global_mask();
-
 	return group;
 }
diff --git a/fs/notify/inode_mark.c b/fs/notify/inode_mark.c
index 3165d85aada2..7468af10825f 100644
--- a/fs/notify/inode_mark.c
+++ b/fs/notify/inode_mark.c
@@ -30,21 +30,21 @@
  * There are 3 spinlocks involved with fsnotify inode marks and they MUST
  * be taken in order as follows:
  *
- * entry->lock
+ * mark->lock
  * group->mark_lock
  * inode->i_lock
  *
- * entry->lock protects 2 things, entry->group and entry->inode.  You must hold
+ * mark->lock protects 2 things, mark->group and mark->inode.  You must hold
  * that lock to dereference either of these things (they could be NULL even with
  * the lock)
  *
- * group->mark_lock protects the mark_entries list anchored inside a given group
- * and each entry is hooked via the g_list.  It also sorta protects the
+ * group->mark_lock protects the marks_list anchored inside a given group
+ * and each mark is hooked via the g_list.  It also sorta protects the
  * free_g_list, which when used is anchored by a private list on the stack of the
  * task which held the group->mark_lock.
  *
- * inode->i_lock protects the i_fsnotify_mark_entries list anchored inside a
- * given inode and each entry is hooked via the i_list. (and sorta the
+ * inode->i_lock protects the i_fsnotify_marks list anchored inside a
+ * given inode and each mark is hooked via the i_list. (and sorta the
  * free_i_list)
  *
  *
@@ -56,12 +56,12 @@
  * - The inode is unlinked for the last time.  (fsnotify_inode_remove)
  * - The inode is being evicted from cache. (fsnotify_inode_delete)
  * - The fs the inode is on is unmounted.  (fsnotify_inode_delete/fsnotify_unmount_inodes)
- * - Something explicitly requests that it be removed.  (fsnotify_destroy_mark_by_entry)
+ * - Something explicitly requests that it be removed.  (fsnotify_destroy_mark)
  * - The fsnotify_group associated with the mark is going away and all such marks
  *   need to be cleaned up. (fsnotify_clear_marks_by_group)
  *
  * Worst case we are given an inode and need to clean up all the marks on that
- * inode.  We take i_lock and walk the i_fsnotify_mark_entries safely.  For each
+ * inode.  We take i_lock and walk the i_fsnotify_marks safely.  For each
  * mark on the list we take a reference (so the mark can't disappear under us).
  * We remove that mark form the inode's list of marks and we add this mark to a
  * private list anchored on the stack using i_free_list;  At this point we no
@@ -96,15 +96,15 @@
 #include <linux/fsnotify_backend.h>
 #include "fsnotify.h"
 
-void fsnotify_get_mark(struct fsnotify_mark_entry *entry)
+void fsnotify_get_mark(struct fsnotify_mark *mark)
 {
-	atomic_inc(&entry->refcnt);
+	atomic_inc(&mark->refcnt);
 }
 
-void fsnotify_put_mark(struct fsnotify_mark_entry *entry)
+void fsnotify_put_mark(struct fsnotify_mark *mark)
 {
-	if (atomic_dec_and_test(&entry->refcnt))
-		entry->free_mark(entry);
+	if (atomic_dec_and_test(&mark->refcnt))
+		mark->free_mark(mark);
 }
 
 /*
@@ -112,14 +112,14 @@ void fsnotify_put_mark(struct fsnotify_mark_entry *entry)
  */
 static void fsnotify_recalc_inode_mask_locked(struct inode *inode)
 {
-	struct fsnotify_mark_entry *entry;
+	struct fsnotify_mark *mark;
 	struct hlist_node *pos;
 	__u32 new_mask = 0;
 
 	assert_spin_locked(&inode->i_lock);
 
-	hlist_for_each_entry(entry, pos, &inode->i_fsnotify_mark_entries, i_list)
-		new_mask |= entry->mask;
+	hlist_for_each_entry(mark, pos, &inode->i_fsnotify_marks, i.i_list)
+		new_mask |= mark->mask;
 	inode->i_fsnotify_mask = new_mask;
 }
 
@@ -139,43 +139,43 @@ void fsnotify_recalc_inode_mask(struct inode *inode)
 /*
  * Any time a mark is getting freed we end up here.
  * The caller had better be holding a reference to this mark so we don't actually
- * do the final put under the entry->lock
+ * do the final put under the mark->lock
  */
-void fsnotify_destroy_mark_by_entry(struct fsnotify_mark_entry *entry)
+void fsnotify_destroy_mark(struct fsnotify_mark *mark)
 {
 	struct fsnotify_group *group;
 	struct inode *inode;
 
-	spin_lock(&entry->lock);
+	spin_lock(&mark->lock);
 
-	group = entry->group;
-	inode = entry->inode;
+	group = mark->group;
+	inode = mark->i.inode;
 
 	BUG_ON(group && !inode);
 	BUG_ON(!group && inode);
 
 	/* if !group something else already marked this to die */
 	if (!group) {
-		spin_unlock(&entry->lock);
+		spin_unlock(&mark->lock);
 		return;
 	}
 
 	/* 1 from caller and 1 for being on i_list/g_list */
-	BUG_ON(atomic_read(&entry->refcnt) < 2);
+	BUG_ON(atomic_read(&mark->refcnt) < 2);
 
 	spin_lock(&group->mark_lock);
 	spin_lock(&inode->i_lock);
 
-	hlist_del_init(&entry->i_list);
-	entry->inode = NULL;
+	hlist_del_init(&mark->i.i_list);
+	mark->i.inode = NULL;
 
-	list_del_init(&entry->g_list);
-	entry->group = NULL;
+	list_del_init(&mark->g_list);
+	mark->group = NULL;
 
-	fsnotify_put_mark(entry); /* for i_list and g_list */
+	fsnotify_put_mark(mark); /* for i_list and g_list */
 
 	/*
-	 * this mark is now off the inode->i_fsnotify_mark_entries list and we
+	 * this mark is now off the inode->i_fsnotify_marks list and we
 	 * hold the inode->i_lock, so this is the perfect time to update the
 	 * inode->i_fsnotify_mask
 	 */
@@ -183,21 +183,21 @@ void fsnotify_destroy_mark_by_entry(struct fsnotify_mark_entry *entry)
 
 	spin_unlock(&inode->i_lock);
 	spin_unlock(&group->mark_lock);
-	spin_unlock(&entry->lock);
+	spin_unlock(&mark->lock);
 
 	/*
 	 * Some groups like to know that marks are being freed.  This is a
-	 * callback to the group function to let it know that this entry
+	 * callback to the group function to let it know that this mark
 	 * is being freed.
 	 */
 	if (group->ops->freeing_mark)
-		group->ops->freeing_mark(entry, group);
+		group->ops->freeing_mark(mark, group);
 
 	/*
 	 * __fsnotify_update_child_dentry_flags(inode);
 	 *
 	 * I really want to call that, but we can't, we have no idea if the inode
-	 * still exists the second we drop the entry->lock.
+	 * still exists the second we drop the mark->lock.
 	 *
 	 * The next time an event arrive to this inode from one of it's children
 	 * __fsnotify_parent will see that the inode doesn't care about it's
@@ -222,20 +222,20 @@ void fsnotify_destroy_mark_by_entry(struct fsnotify_mark_entry *entry)
  */
 void fsnotify_clear_marks_by_group(struct fsnotify_group *group)
 {
-	struct fsnotify_mark_entry *lentry, *entry;
+	struct fsnotify_mark *lmark, *mark;
 	LIST_HEAD(free_list);
 
 	spin_lock(&group->mark_lock);
-	list_for_each_entry_safe(entry, lentry, &group->mark_entries, g_list) {
-		list_add(&entry->free_g_list, &free_list);
-		list_del_init(&entry->g_list);
-		fsnotify_get_mark(entry);
+	list_for_each_entry_safe(mark, lmark, &group->marks_list, g_list) {
+		list_add(&mark->free_g_list, &free_list);
+		list_del_init(&mark->g_list);
+		fsnotify_get_mark(mark);
 	}
 	spin_unlock(&group->mark_lock);
 
-	list_for_each_entry_safe(entry, lentry, &free_list, free_g_list) {
-		fsnotify_destroy_mark_by_entry(entry);
-		fsnotify_put_mark(entry);
+	list_for_each_entry_safe(mark, lmark, &free_list, free_g_list) {
+		fsnotify_destroy_mark(mark);
+		fsnotify_put_mark(mark);
 	}
 }
 
@@ -244,95 +244,133 @@ void fsnotify_clear_marks_by_group(struct fsnotify_group *group)
  */
 void fsnotify_clear_marks_by_inode(struct inode *inode)
 {
-	struct fsnotify_mark_entry *entry, *lentry;
+	struct fsnotify_mark *mark, *lmark;
 	struct hlist_node *pos, *n;
 	LIST_HEAD(free_list);
 
 	spin_lock(&inode->i_lock);
-	hlist_for_each_entry_safe(entry, pos, n, &inode->i_fsnotify_mark_entries, i_list) {
-		list_add(&entry->free_i_list, &free_list);
-		hlist_del_init(&entry->i_list);
-		fsnotify_get_mark(entry);
+	hlist_for_each_entry_safe(mark, pos, n, &inode->i_fsnotify_marks, i.i_list) {
+		list_add(&mark->i.free_i_list, &free_list);
+		hlist_del_init(&mark->i.i_list);
+		fsnotify_get_mark(mark);
 	}
 	spin_unlock(&inode->i_lock);
 
-	list_for_each_entry_safe(entry, lentry, &free_list, free_i_list) {
-		fsnotify_destroy_mark_by_entry(entry);
-		fsnotify_put_mark(entry);
+	list_for_each_entry_safe(mark, lmark, &free_list, i.free_i_list) {
+		fsnotify_destroy_mark(mark);
+		fsnotify_put_mark(mark);
 	}
 }
 
-/*
- * given a group and inode, find the mark associated with that combination.
- * if found take a reference to that mark and return it, else return NULL
- */
-struct fsnotify_mark_entry *fsnotify_find_mark_entry(struct fsnotify_group *group,
-						     struct inode *inode)
+static struct fsnotify_mark *fsnotify_find_mark_locked(struct fsnotify_group *group,
+						       struct inode *inode)
 {
-	struct fsnotify_mark_entry *entry;
+	struct fsnotify_mark *mark;
 	struct hlist_node *pos;
 
 	assert_spin_locked(&inode->i_lock);
 
-	hlist_for_each_entry(entry, pos, &inode->i_fsnotify_mark_entries, i_list) {
-		if (entry->group == group) {
-			fsnotify_get_mark(entry);
-			return entry;
+	hlist_for_each_entry(mark, pos, &inode->i_fsnotify_marks, i.i_list) {
+		if (mark->group == group) {
+			fsnotify_get_mark(mark);
+			return mark;
 		}
 	}
 	return NULL;
 }
 
 /*
- * Nothing fancy, just initialize lists and locks and counters.
+ * given a group and inode, find the mark associated with that combination.
+ * if found take a reference to that mark and return it, else return NULL
  */
-void fsnotify_init_mark(struct fsnotify_mark_entry *entry,
-			void (*free_mark)(struct fsnotify_mark_entry *entry))
+struct fsnotify_mark *fsnotify_find_mark(struct fsnotify_group *group,
+					 struct inode *inode)
+{
+	struct fsnotify_mark *mark;
+
+	spin_lock(&inode->i_lock);
+	mark = fsnotify_find_mark_locked(group, inode);
+	spin_unlock(&inode->i_lock);
+
+	return mark;
+}
 
+void fsnotify_duplicate_mark(struct fsnotify_mark *new, struct fsnotify_mark *old)
 {
-	spin_lock_init(&entry->lock);
-	atomic_set(&entry->refcnt, 1);
-	INIT_HLIST_NODE(&entry->i_list);
-	entry->group = NULL;
-	entry->mask = 0;
-	entry->inode = NULL;
-	entry->free_mark = free_mark;
+	assert_spin_locked(&old->lock);
+	new->i.inode = old->i.inode;
+	new->group = old->group;
+	new->mask = old->mask;
+	new->free_mark = old->free_mark;
 }
 
 /*
- * Attach an initialized mark entry to a given group and inode.
+ * Nothing fancy, just initialize lists and locks and counters.
+ */
+void fsnotify_init_mark(struct fsnotify_mark *mark,
+			void (*free_mark)(struct fsnotify_mark *mark))
+{
+	spin_lock_init(&mark->lock);
+	atomic_set(&mark->refcnt, 1);
+	INIT_HLIST_NODE(&mark->i.i_list);
+	mark->group = NULL;
+	mark->mask = 0;
+	mark->i.inode = NULL;
+	mark->free_mark = free_mark;
+}
+
+/*
+ * Attach an initialized mark mark to a given group and inode.
  * These marks may be used for the fsnotify backend to determine which
  * event types should be delivered to which group and for which inodes.
  */
-int fsnotify_add_mark(struct fsnotify_mark_entry *entry,
-		      struct fsnotify_group *group, struct inode *inode)
+int fsnotify_add_mark(struct fsnotify_mark *mark,
+		      struct fsnotify_group *group, struct inode *inode,
+		      int allow_dups)
 {
-	struct fsnotify_mark_entry *lentry;
+	struct fsnotify_mark *lmark = NULL;
 	int ret = 0;
 
 	inode = igrab(inode);
 	if (unlikely(!inode))
 		return -EINVAL;
 
+	mark->flags = FSNOTIFY_MARK_FLAG_INODE;
+
+	/*
+	 * if this group isn't being testing for inode type events we need
+	 * to start testing
+	 */
+	if (unlikely(list_empty(&group->inode_group_list)))
+		fsnotify_add_inode_group(group);
+	/*
+	 * XXX This is where we could also do the fsnotify_add_vfsmount_group
+	 * if we are setting and vfsmount mark....
+
+	if (unlikely(list_empty(&group->vfsmount_group_list)))
+		fsnotify_add_vfsmount_group(group);
+	 */
+
 	/*
 	 * LOCKING ORDER!!!!
-	 * entry->lock
+	 * mark->lock
 	 * group->mark_lock
 	 * inode->i_lock
 	 */
-	spin_lock(&entry->lock);
+	spin_lock(&mark->lock);
 	spin_lock(&group->mark_lock);
 	spin_lock(&inode->i_lock);
 
-	lentry = fsnotify_find_mark_entry(group, inode);
-	if (!lentry) {
-		entry->group = group;
-		entry->inode = inode;
+	if (!allow_dups)
+		lmark = fsnotify_find_mark_locked(group, inode);
+	if (!lmark) {
+		mark->group = group;
+		mark->i.inode = inode;
 
-		hlist_add_head(&entry->i_list, &inode->i_fsnotify_mark_entries);
-		list_add(&entry->g_list, &group->mark_entries);
+		hlist_add_head(&mark->i.i_list, &inode->i_fsnotify_marks);
+		list_add(&mark->g_list, &group->marks_list);
 
-		fsnotify_get_mark(entry); /* for i_list and g_list */
+		fsnotify_get_mark(mark); /* for i_list and g_list */
 
 		atomic_inc(&group->num_marks);
 
@@ -341,12 +379,12 @@ int fsnotify_add_mark(struct fsnotify_mark_entry *entry,
 
 	spin_unlock(&inode->i_lock);
 	spin_unlock(&group->mark_lock);
-	spin_unlock(&entry->lock);
+	spin_unlock(&mark->lock);
 
-	if (lentry) {
+	if (lmark) {
 		ret = -EEXIST;
 		iput(inode);
-		fsnotify_put_mark(lentry);
+		fsnotify_put_mark(lmark);
 	} else {
 		__fsnotify_update_child_dentry_flags(inode);
 	}
diff --git a/fs/notify/inotify/inotify.h b/fs/notify/inotify/inotify.h
index f234f3a4c8ca..b6642e4de4bf 100644
--- a/fs/notify/inotify/inotify.h
+++ b/fs/notify/inotify/inotify.h
@@ -9,13 +9,12 @@ struct inotify_event_private_data {
 	int wd;
 };
 
-struct inotify_inode_mark_entry {
-	/* fsnotify_mark_entry MUST be the first thing */
-	struct fsnotify_mark_entry fsn_entry;
+struct inotify_inode_mark {
+	struct fsnotify_mark fsn_mark;
 	int wd;
 };
 
-extern void inotify_ignored_and_remove_idr(struct fsnotify_mark_entry *entry,
+extern void inotify_ignored_and_remove_idr(struct fsnotify_mark *fsn_mark,
 					   struct fsnotify_group *group);
 extern void inotify_free_event_priv(struct fsnotify_event_private_data *event_priv);
 
diff --git a/fs/notify/inotify/inotify_fsnotify.c b/fs/notify/inotify/inotify_fsnotify.c
index c9ee67b442e1..3a20053e4b48 100644
--- a/fs/notify/inotify/inotify_fsnotify.c
+++ b/fs/notify/inotify/inotify_fsnotify.c
@@ -31,10 +31,64 @@
 
 #include "inotify.h"
 
+/*
+ * Check if 2 events contain the same information.  We do not compare private data
+ * but at this moment that isn't a problem for any know fsnotify listeners.
+ */
+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->name_len == new->name_len)) {
+		switch (old->data_type) {
+		case (FSNOTIFY_EVENT_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):
+			if (old->mask & FS_Q_OVERFLOW)
+				return true;
+			else if (old->mask & FS_IN_IGNORED)
+				return false;
+			return true;
+		};
+	}
+	return false;
+}
+
+static int inotify_merge(struct list_head *list, struct fsnotify_event *event)
+{
+	struct fsnotify_event_holder *last_holder;
+	struct fsnotify_event *last_event;
+	int ret = 0;
+
+	/* and the list better be locked by something too */
+	spin_lock(&event->lock);
+
+	last_holder = list_entry(list->prev, struct fsnotify_event_holder, event_list);
+	last_event = last_holder->event;
+	if (event_compare(last_event, event))
+		ret = -EEXIST;
+
+	spin_unlock(&event->lock);
+
+	return ret;
+}
+
 static int inotify_handle_event(struct fsnotify_group *group, struct fsnotify_event *event)
 {
-	struct fsnotify_mark_entry *entry;
-	struct inotify_inode_mark_entry *ientry;
+	struct fsnotify_mark *fsn_mark;
+	struct inotify_inode_mark *i_mark;
 	struct inode *to_tell;
 	struct inotify_event_private_data *event_priv;
 	struct fsnotify_event_private_data *fsn_event_priv;
@@ -42,15 +96,13 @@ static int inotify_handle_event(struct fsnotify_group *group, struct fsnotify_ev
 
 	to_tell = event->to_tell;
 
-	spin_lock(&to_tell->i_lock);
-	entry = fsnotify_find_mark_entry(group, to_tell);
-	spin_unlock(&to_tell->i_lock);
+	fsn_mark = fsnotify_find_mark(group, to_tell);
 	/* race with watch removal?  We already passes should_send */
-	if (unlikely(!entry))
+	if (unlikely(!fsn_mark))
 		return 0;
-	ientry = container_of(entry, struct inotify_inode_mark_entry,
-			      fsn_entry);
-	wd = ientry->wd;
+	i_mark = container_of(fsn_mark, struct inotify_inode_mark,
+			      fsn_mark);
+	wd = i_mark->wd;
 
 	event_priv = kmem_cache_alloc(event_priv_cachep, GFP_KERNEL);
 	if (unlikely(!event_priv))
@@ -61,7 +113,7 @@ static int inotify_handle_event(struct fsnotify_group *group, struct fsnotify_ev
 	fsn_event_priv->group = group;
 	event_priv->wd = wd;
 
-	ret = fsnotify_add_notify_event(group, event, fsn_event_priv);
+	ret = fsnotify_add_notify_event(group, event, fsn_event_priv, inotify_merge);
 	if (ret) {
 		inotify_free_event_priv(fsn_event_priv);
 		/* EEXIST says we tail matched, EOVERFLOW isn't something
@@ -72,35 +124,35 @@ static int inotify_handle_event(struct fsnotify_group *group, struct fsnotify_ev
 	}
 
 	/*
-	 * If we hold the entry until after the event is on the queue
+	 * If we hold the fsn_mark until after the event is on the queue
 	 * IN_IGNORED won't be able to pass this event in the queue
 	 */
-	fsnotify_put_mark(entry);
+	fsnotify_put_mark(fsn_mark);
 
 	return ret;
 }
 
-static void inotify_freeing_mark(struct fsnotify_mark_entry *entry, struct fsnotify_group *group)
+static void inotify_freeing_mark(struct fsnotify_mark *fsn_mark, struct fsnotify_group *group)
 {
-	inotify_ignored_and_remove_idr(entry, group);
+	inotify_ignored_and_remove_idr(fsn_mark, group);
 }
 
-static bool inotify_should_send_event(struct fsnotify_group *group, struct inode *inode, __u32 mask)
+static bool inotify_should_send_event(struct fsnotify_group *group, struct inode *inode,
+				      struct vfsmount *mnt, __u32 mask, void *data,
+				      int data_type)
 {
-	struct fsnotify_mark_entry *entry;
+	struct fsnotify_mark *fsn_mark;
 	bool send;
 
-	spin_lock(&inode->i_lock);
-	entry = fsnotify_find_mark_entry(group, inode);
-	spin_unlock(&inode->i_lock);
-	if (!entry)
+	fsn_mark = fsnotify_find_mark(group, inode);
+	if (!fsn_mark)
 		return false;
 
 	mask = (mask & ~FS_EVENT_ON_CHILD);
-	send = (entry->mask & mask);
+	send = (fsn_mark->mask & mask);
 
 	/* find took a reference */
-	fsnotify_put_mark(entry);
+	fsnotify_put_mark(fsn_mark);
 
 	return send;
 }
@@ -114,18 +166,18 @@ static bool inotify_should_send_event(struct fsnotify_group *group, struct inode
  */
 static int idr_callback(int id, void *p, void *data)
 {
-	struct fsnotify_mark_entry *entry;
-	struct inotify_inode_mark_entry *ientry;
+	struct fsnotify_mark *fsn_mark;
+	struct inotify_inode_mark *i_mark;
 	static bool warned = false;
 
 	if (warned)
 		return 0;
 
 	warned = false;
-	entry = p;
-	ientry = container_of(entry, struct inotify_inode_mark_entry, fsn_entry);
+	fsn_mark = p;
+	i_mark = container_of(fsn_mark, struct inotify_inode_mark, fsn_mark);
 
-	WARN(1, "inotify closing but id=%d for entry=%p in group=%p still in "
+	WARN(1, "inotify closing but id=%d for fsn_mark=%p in group=%p still in "
 		"idr.  Probably leaking memory\n", id, p, data);
 
 	/*
@@ -134,9 +186,9 @@ static int idr_callback(int id, void *p, void *data)
 	 * 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);
+	if (fsn_mark)
+		printk(KERN_WARNING "fsn_mark->group=%p inode=%p wd=%d\n",
+			fsn_mark->group, fsn_mark->i.inode, i_mark->wd);
 	return 0;
 }
 
diff --git a/fs/notify/inotify/inotify_user.c b/fs/notify/inotify/inotify_user.c
index dcd2040d330c..e5191cce2607 100644
--- a/fs/notify/inotify/inotify_user.c
+++ b/fs/notify/inotify/inotify_user.c
@@ -55,12 +55,6 @@ 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;
 
-/*
- * When inotify registers a new group it increments this and uses that
- * value as an offset to set the fsnotify group "name" and priority.
- */
-static atomic_t inotify_grp_num;
-
 #ifdef CONFIG_SYSCTL
 
 #include <linux/sysctl.h>
@@ -367,58 +361,159 @@ static int inotify_find_inode(const char __user *dirname, struct path *path, uns
 	return error;
 }
 
+static int inotify_add_to_idr(struct idr *idr, spinlock_t *idr_lock,
+			      int last_wd,
+			      struct inotify_inode_mark *i_mark)
+{
+	int ret;
+
+	do {
+		if (unlikely(!idr_pre_get(idr, GFP_KERNEL)))
+			return -ENOMEM;
+
+		spin_lock(idr_lock);
+		ret = idr_get_new_above(idr, i_mark, last_wd + 1,
+					&i_mark->wd);
+		/* we added the mark to the idr, take a reference */
+		if (!ret)
+			fsnotify_get_mark(&i_mark->fsn_mark);
+		spin_unlock(idr_lock);
+	} while (ret == -EAGAIN);
+
+	return ret;
+}
+
+static struct inotify_inode_mark *inotify_idr_find_locked(struct fsnotify_group *group,
+								int wd)
+{
+	struct idr *idr = &group->inotify_data.idr;
+	spinlock_t *idr_lock = &group->inotify_data.idr_lock;
+	struct inotify_inode_mark *i_mark;
+
+	assert_spin_locked(idr_lock);
+
+	i_mark = idr_find(idr, wd);
+	if (i_mark) {
+		struct fsnotify_mark *fsn_mark = &i_mark->fsn_mark;
+
+		fsnotify_get_mark(fsn_mark);
+		/* One ref for being in the idr, one ref we just took */
+		BUG_ON(atomic_read(&fsn_mark->refcnt) < 2);
+	}
+
+	return i_mark;
+}
+
+static struct inotify_inode_mark *inotify_idr_find(struct fsnotify_group *group,
+							 int wd)
+{
+	struct inotify_inode_mark *i_mark;
+	spinlock_t *idr_lock = &group->inotify_data.idr_lock;
+
+	spin_lock(idr_lock);
+	i_mark = inotify_idr_find_locked(group, wd);
+	spin_unlock(idr_lock);
+
+	return i_mark;
+}
+
+static void do_inotify_remove_from_idr(struct fsnotify_group *group,
+				       struct inotify_inode_mark *i_mark)
+{
+	struct idr *idr = &group->inotify_data.idr;
+	spinlock_t *idr_lock = &group->inotify_data.idr_lock;
+	int wd = i_mark->wd;
+
+	assert_spin_locked(idr_lock);
+
+	idr_remove(idr, wd);
+
+	/* removed from the idr, drop that ref */
+	fsnotify_put_mark(&i_mark->fsn_mark);
+}
+
 /*
  * 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 inotify_inode_mark *i_mark)
 {
-	struct idr *idr;
-	struct fsnotify_mark_entry *entry;
-	struct inotify_inode_mark_entry *found_ientry;
+	spinlock_t *idr_lock = &group->inotify_data.idr_lock;
+	struct inotify_inode_mark *found_i_mark = NULL;
 	int wd;
 
-	spin_lock(&group->inotify_data.idr_lock);
-	idr = &group->inotify_data.idr;
-	wd = ientry->wd;
+	spin_lock(idr_lock);
+	wd = i_mark->wd;
 
-	if (wd == -1)
+	/*
+	 * does this i_mark think it is in the idr?  we shouldn't get called
+	 * if it wasn't....
+	 */
+	if (wd == -1) {
+		printk(KERN_WARNING "%s: i_mark=%p i_mark->wd=%d i_mark->group=%p"
+			" i_mark->inode=%p\n", __func__, i_mark, i_mark->wd,
+			i_mark->fsn_mark.group, i_mark->fsn_mark.i.inode);
+		WARN_ON(1);
 		goto out;
+	}
 
-	entry = idr_find(&group->inotify_data.idr, wd);
-	if (unlikely(!entry))
+	/* Lets look in the idr to see if we find it */
+	found_i_mark = inotify_idr_find_locked(group, wd);
+	if (unlikely(!found_i_mark)) {
+		printk(KERN_WARNING "%s: i_mark=%p i_mark->wd=%d i_mark->group=%p"
+			" i_mark->inode=%p\n", __func__, i_mark, i_mark->wd,
+			i_mark->fsn_mark.group, i_mark->fsn_mark.i.inode);
+		WARN_ON(1);
 		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. */
+	/*
+	 * We found an mark in the idr at the right wd, but it's
+	 * not the mark we were told to remove.  eparis seriously
+	 * fucked up somewhere.
+	 */
+	if (unlikely(found_i_mark != i_mark)) {
 		WARN_ON(1);
-		ientry->wd = -1;
+		printk(KERN_WARNING "%s: i_mark=%p i_mark->wd=%d i_mark->group=%p "
+			"mark->inode=%p found_i_mark=%p found_i_mark->wd=%d "
+			"found_i_mark->group=%p found_i_mark->inode=%p\n",
+			__func__, i_mark, i_mark->wd, i_mark->fsn_mark.group,
+			i_mark->fsn_mark.i.inode, found_i_mark, found_i_mark->wd,
+			found_i_mark->fsn_mark.group,
+			found_i_mark->fsn_mark.i.inode);
 		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;
+	/*
+	 * One ref for being in the idr
+	 * one ref held by the caller trying to kill us
+	 * one ref grabbed by inotify_idr_find
+	 */
+	if (unlikely(atomic_read(&i_mark->fsn_mark.refcnt) < 3)) {
+		printk(KERN_WARNING "%s: i_mark=%p i_mark->wd=%d i_mark->group=%p"
+			" i_mark->inode=%p\n", __func__, i_mark, i_mark->wd,
+			i_mark->fsn_mark.group, i_mark->fsn_mark.i.inode);
+		/* we can't really recover with bad ref cnting.. */
+		BUG();
+	}
 
-	/* removed from the idr, drop that ref */
-	fsnotify_put_mark(entry);
+	do_inotify_remove_from_idr(group, i_mark);
 out:
-	spin_unlock(&group->inotify_data.idr_lock);
+	/* match the ref taken by inotify_idr_find_locked() */
+	if (found_i_mark)
+		fsnotify_put_mark(&found_i_mark->fsn_mark);
+	i_mark->wd = -1;
+	spin_unlock(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,
+void inotify_ignored_and_remove_idr(struct fsnotify_mark *fsn_mark,
 				    struct fsnotify_group *group)
 {
-	struct inotify_inode_mark_entry *ientry;
+	struct inotify_inode_mark *i_mark;
 	struct fsnotify_event *ignored_event;
 	struct inotify_event_private_data *event_priv;
 	struct fsnotify_event_private_data *fsn_event_priv;
@@ -430,7 +525,7 @@ void inotify_ignored_and_remove_idr(struct fsnotify_mark_entry *entry,
 	if (!ignored_event)
 		return;
 
-	ientry = container_of(entry, struct inotify_inode_mark_entry, fsn_entry);
+	i_mark = container_of(fsn_mark, struct inotify_inode_mark, fsn_mark);
 
 	event_priv = kmem_cache_alloc(event_priv_cachep, GFP_NOFS);
 	if (unlikely(!event_priv))
@@ -439,9 +534,9 @@ void inotify_ignored_and_remove_idr(struct fsnotify_mark_entry *entry,
 	fsn_event_priv = &event_priv->fsnotify_event_priv_data;
 
 	fsn_event_priv->group = group;
-	event_priv->wd = ientry->wd;
+	event_priv->wd = i_mark->wd;
 
-	ret = fsnotify_add_notify_event(group, ignored_event, fsn_event_priv);
+	ret = fsnotify_add_notify_event(group, ignored_event, fsn_event_priv, NULL);
 	if (ret)
 		inotify_free_event_priv(fsn_event_priv);
 
@@ -450,26 +545,28 @@ skip_send_ignore:
 	/* matches the reference taken when the event was created */
 	fsnotify_put_event(ignored_event);
 
-	/* remove this entry from the idr */
-	inotify_remove_from_idr(group, ientry);
+	/* remove this mark from the idr */
+	inotify_remove_from_idr(group, i_mark);
 
 	atomic_dec(&group->inotify_data.user->inotify_watches);
 }
 
 /* ding dong the mark is dead */
-static void inotify_free_mark(struct fsnotify_mark_entry *entry)
+static void inotify_free_mark(struct fsnotify_mark *fsn_mark)
 {
-	struct inotify_inode_mark_entry *ientry = (struct inotify_inode_mark_entry *)entry;
+	struct inotify_inode_mark *i_mark;
+
+	i_mark = container_of(fsn_mark, struct inotify_inode_mark, fsn_mark);
 
-	kmem_cache_free(inotify_inode_mark_cachep, ientry);
+	kmem_cache_free(inotify_inode_mark_cachep, i_mark);
 }
 
 static int inotify_update_existing_watch(struct fsnotify_group *group,
 					 struct inode *inode,
 					 u32 arg)
 {
-	struct fsnotify_mark_entry *entry;
-	struct inotify_inode_mark_entry *ientry;
+	struct fsnotify_mark *fsn_mark;
+	struct inotify_inode_mark *i_mark;
 	__u32 old_mask, new_mask;
 	__u32 mask;
 	int add = (arg & IN_MASK_ADD);
@@ -480,36 +577,34 @@ static int inotify_update_existing_watch(struct fsnotify_group *group,
 	if (unlikely(!mask))
 		return -EINVAL;
 
-	spin_lock(&inode->i_lock);
-	entry = fsnotify_find_mark_entry(group, inode);
-	spin_unlock(&inode->i_lock);
-	if (!entry)
+	fsn_mark = fsnotify_find_mark(group, inode);
+	if (!fsn_mark)
 		return -ENOENT;
 
-	ientry = container_of(entry, struct inotify_inode_mark_entry, fsn_entry);
+	i_mark = container_of(fsn_mark, struct inotify_inode_mark, fsn_mark);
 
-	spin_lock(&entry->lock);
+	spin_lock(&fsn_mark->lock);
 
-	old_mask = entry->mask;
+	old_mask = fsn_mark->mask;
 	if (add) {
-		entry->mask |= mask;
-		new_mask = entry->mask;
+		fsn_mark->mask |= mask;
+		new_mask = fsn_mark->mask;
 	} else {
-		entry->mask = mask;
-		new_mask = entry->mask;
+		fsn_mark->mask = mask;
+		new_mask = fsn_mark->mask;
 	}
 
-	spin_unlock(&entry->lock);
+	spin_unlock(&fsn_mark->lock);
 
 	if (old_mask != new_mask) {
 		/* more bits in old than in new? */
 		int dropped = (old_mask & ~new_mask);
-		/* more bits in this entry than the inode's mask? */
+		/* more bits in this fsn_mark than the inode's mask? */
 		int do_inode = (new_mask & ~inode->i_fsnotify_mask);
-		/* more bits in this entry than the group? */
+		/* more bits in this fsn_mark than the group? */
 		int do_group = (new_mask & ~group->mask);
 
-		/* update the inode with this new entry */
+		/* update the inode with this new fsn_mark */
 		if (dropped || do_inode)
 			fsnotify_recalc_inode_mask(inode);
 
@@ -519,10 +614,10 @@ static int inotify_update_existing_watch(struct fsnotify_group *group,
 	}
 
 	/* return the wd */
-	ret = ientry->wd;
+	ret = i_mark->wd;
 
-	/* match the get from fsnotify_find_mark_entry() */
-	fsnotify_put_mark(entry);
+	/* match the get from fsnotify_find_mark() */
+	fsnotify_put_mark(fsn_mark);
 
 	return ret;
 }
@@ -531,73 +626,58 @@ static int inotify_new_watch(struct fsnotify_group *group,
 			     struct inode *inode,
 			     u32 arg)
 {
-	struct inotify_inode_mark_entry *tmp_ientry;
+	struct inotify_inode_mark *tmp_i_mark;
 	__u32 mask;
 	int ret;
+	struct idr *idr = &group->inotify_data.idr;
+	spinlock_t *idr_lock = &group->inotify_data.idr_lock;
 
 	/* 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))
+	tmp_i_mark = kmem_cache_alloc(inotify_inode_mark_cachep, GFP_KERNEL);
+	if (unlikely(!tmp_i_mark))
 		return -ENOMEM;
 
-	fsnotify_init_mark(&tmp_ientry->fsn_entry, inotify_free_mark);
-	tmp_ientry->fsn_entry.mask = mask;
-	tmp_ientry->wd = -1;
+	fsnotify_init_mark(&tmp_i_mark->fsn_mark, inotify_free_mark);
+	tmp_i_mark->fsn_mark.mask = mask;
+	tmp_i_mark->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;
+	ret = inotify_add_to_idr(idr, idr_lock, group->inotify_data.last_wd,
+				 tmp_i_mark);
+	if (ret)
 		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);
+	ret = fsnotify_add_mark(&tmp_i_mark->fsn_mark, group, inode, 0);
 	if (ret) {
 		/* we failed to get on the inode, get off the idr */
-		inotify_remove_from_idr(group, tmp_ientry);
+		inotify_remove_from_idr(group, tmp_i_mark);
 		goto out_err;
 	}
 
 	/* update the idr hint, who cares about races, it's just a hint */
-	group->inotify_data.last_wd = tmp_ientry->wd;
+	group->inotify_data.last_wd = tmp_i_mark->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);
+	/* return the watch descriptor for this new mark */
+	ret = tmp_i_mark->wd;
 
 	/* if this mark added a new event update the group mask */
 	if (mask & ~group->mask)
 		fsnotify_recalc_group_mask(group);
 
 out_err:
-	if (ret < 0)
-		kmem_cache_free(inotify_inode_mark_cachep, tmp_ientry);
+	/* match the ref from fsnotify_init_mark() */
+	fsnotify_put_mark(&tmp_i_mark->fsn_mark);
 
 	return ret;
 }
@@ -626,11 +706,8 @@ retry:
 static struct fsnotify_group *inotify_new_group(struct user_struct *user, unsigned int max_events)
 {
 	struct fsnotify_group *group;
-	unsigned int grp_num;
 
-	/* fsnotify_obtain_group took a reference to group, we put this when we kill the file in the end */
-	grp_num = (INOTIFY_GROUP_NUM - atomic_inc_return(&inotify_grp_num));
-	group = fsnotify_obtain_group(grp_num, 0, &inotify_fsnotify_ops);
+	group = fsnotify_alloc_group(&inotify_fsnotify_ops);
 	if (IS_ERR(group))
 		return group;
 
@@ -638,7 +715,7 @@ 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 = 1;
+	group->inotify_data.last_wd = 0;
 	group->inotify_data.user = user;
 	group->inotify_data.fa = NULL;
 
@@ -760,7 +837,7 @@ fput_and_out:
 SYSCALL_DEFINE2(inotify_rm_watch, int, fd, __s32, wd)
 {
 	struct fsnotify_group *group;
-	struct fsnotify_mark_entry *entry;
+	struct inotify_inode_mark *i_mark;
 	struct file *filp;
 	int ret = 0, fput_needed;
 
@@ -769,25 +846,23 @@ SYSCALL_DEFINE2(inotify_rm_watch, int, fd, __s32, wd)
 		return -EBADF;
 
 	/* verify that this is indeed an inotify instance */
-	if (unlikely(filp->f_op != &inotify_fops)) {
-		ret = -EINVAL;
+	ret = -EINVAL;
+	if (unlikely(filp->f_op != &inotify_fops))
 		goto out;
-	}
 
 	group = filp->private_data;
 
-	spin_lock(&group->inotify_data.idr_lock);
-	entry = idr_find(&group->inotify_data.idr, wd);
-	if (unlikely(!entry)) {
-		spin_unlock(&group->inotify_data.idr_lock);
-		ret = -EINVAL;
+	ret = -EINVAL;
+	i_mark = inotify_idr_find(group, wd);
+	if (unlikely(!i_mark))
 		goto out;
-	}
-	fsnotify_get_mark(entry);
-	spin_unlock(&group->inotify_data.idr_lock);
 
-	fsnotify_destroy_mark_by_entry(entry);
-	fsnotify_put_mark(entry);
+	ret = 0;
+
+	fsnotify_destroy_mark(&i_mark->fsn_mark);
+
+	/* match ref taken by inotify_idr_find */
+	fsnotify_put_mark(&i_mark->fsn_mark);
 
 out:
 	fput_light(filp, fput_needed);
@@ -825,7 +900,7 @@ static int __init inotify_user_setup(void)
 	if (IS_ERR(inotify_mnt))
 		panic("inotify: kern_mount ret %ld!\n", PTR_ERR(inotify_mnt));
 
-	inotify_inode_mark_cachep = KMEM_CACHE(inotify_inode_mark_entry, SLAB_PANIC);
+	inotify_inode_mark_cachep = KMEM_CACHE(inotify_inode_mark, SLAB_PANIC);
 	event_priv_cachep = KMEM_CACHE(inotify_event_private_data, SLAB_PANIC);
 
 	inotify_max_queued_events = 16384;
diff --git a/fs/notify/notification.c b/fs/notify/notification.c
index b8bf53b4c108..066f1f988bac 100644
--- a/fs/notify/notification.c
+++ b/fs/notify/notification.c
@@ -56,7 +56,7 @@ static struct kmem_cache *fsnotify_event_holder_cachep;
  * it is needed.  It's refcnt is set 1 at kernel init time and will never
  * get set to 0 so it will never get 'freed'
  */
-static struct fsnotify_event q_overflow_event;
+static struct fsnotify_event *q_overflow_event;
 static atomic_t fsnotify_sync_cookie = ATOMIC_INIT(0);
 
 /**
@@ -104,7 +104,8 @@ struct fsnotify_event_holder *fsnotify_alloc_event_holder(void)
 
 void fsnotify_destroy_event_holder(struct fsnotify_event_holder *holder)
 {
-	kmem_cache_free(fsnotify_event_holder_cachep, holder);
+	if (holder)
+		kmem_cache_free(fsnotify_event_holder_cachep, holder);
 }
 
 /*
@@ -129,53 +130,17 @@ struct fsnotify_event_private_data *fsnotify_remove_priv_from_event(struct fsnot
 }
 
 /*
- * Check if 2 events contain the same information.  We do not compare private data
- * but at this moment that isn't a problem for any know fsnotify listeners.
- */
-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->name_len == new->name_len)) {
-		switch (old->data_type) {
-		case (FSNOTIFY_EVENT_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):
-			if (old->mask & FS_Q_OVERFLOW)
-				return true;
-			else if (old->mask & FS_IN_IGNORED)
-				return false;
-			return false;
-		};
-	}
-	return false;
-}
-
-/*
  * Add an event to the group notification queue.  The group can later pull this
  * event off the queue to deal with.  If the event is successfully added to the
  * group's notification queue, a reference is taken on event.
  */
 int fsnotify_add_notify_event(struct fsnotify_group *group, struct fsnotify_event *event,
-			      struct fsnotify_event_private_data *priv)
+			      struct fsnotify_event_private_data *priv,
+			      int (*merge)(struct list_head *, struct fsnotify_event *))
 {
 	struct fsnotify_event_holder *holder = NULL;
 	struct list_head *list = &group->notification_list;
-	struct fsnotify_event_holder *last_holder;
-	struct fsnotify_event *last_event;
-	int ret = 0;
+	int rc = 0;
 
 	/*
 	 * There is one fsnotify_event_holder embedded inside each fsnotify_event.
@@ -195,12 +160,24 @@ alloc_holder:
 	mutex_lock(&group->notification_mutex);
 
 	if (group->q_len >= group->max_events) {
-		event = &q_overflow_event;
-		ret = -EOVERFLOW;
+		event = q_overflow_event;
+		rc = -EOVERFLOW;
 		/* sorry, no private data on the overflow event */
 		priv = NULL;
 	}
 
+	if (!list_empty(list) && merge) {
+		int ret;
+
+		ret = merge(list, event);
+		if (ret) {
+			mutex_unlock(&group->notification_mutex);
+			if (holder != &event->holder)
+				fsnotify_destroy_event_holder(holder);
+			return ret;
+		}
+	}
+
 	spin_lock(&event->lock);
 
 	if (list_empty(&event->holder.event_list)) {
@@ -215,18 +192,6 @@ alloc_holder:
 		goto alloc_holder;
 	}
 
-	if (!list_empty(list)) {
-		last_holder = list_entry(list->prev, struct fsnotify_event_holder, event_list);
-		last_event = last_holder->event;
-		if (event_compare(last_event, event)) {
-			spin_unlock(&event->lock);
-			mutex_unlock(&group->notification_mutex);
-			if (holder != &event->holder)
-				fsnotify_destroy_event_holder(holder);
-			return -EEXIST;
-		}
-	}
-
 	group->q_len++;
 	holder->event = event;
 
@@ -238,7 +203,7 @@ alloc_holder:
 	mutex_unlock(&group->notification_mutex);
 
 	wake_up(&group->notification_waitq);
-	return ret;
+	return rc;
 }
 
 /*
@@ -314,25 +279,77 @@ void fsnotify_flush_notify(struct fsnotify_group *group)
 
 static void initialize_event(struct fsnotify_event *event)
 {
-	event->holder.event = NULL;
 	INIT_LIST_HEAD(&event->holder.event_list);
 	atomic_set(&event->refcnt, 1);
 
 	spin_lock_init(&event->lock);
 
-	event->path.dentry = NULL;
-	event->path.mnt = NULL;
-	event->inode = NULL;
-	event->data_type = FSNOTIFY_EVENT_NONE;
-
 	INIT_LIST_HEAD(&event->private_data_list);
+}
+
+/*
+ * Caller damn well better be holding whatever mutex is protecting the
+ * old_holder->event_list and the new_event must be a clean event which
+ * cannot be found anywhere else in the kernel.
+ */
+int fsnotify_replace_event(struct fsnotify_event_holder *old_holder,
+			   struct fsnotify_event *new_event)
+{
+	struct fsnotify_event *old_event = old_holder->event;
+	struct fsnotify_event_holder *new_holder = &new_event->holder;
+
+	enum event_spinlock_class {
+		SPINLOCK_OLD,
+		SPINLOCK_NEW,
+	};
+
+	/*
+	 * if the new_event's embedded holder is in use someone
+	 * screwed up and didn't give us a clean new event.
+	 */
+	BUG_ON(!list_empty(&new_holder->event_list));
+
+	spin_lock_nested(&old_event->lock, SPINLOCK_OLD);
+	spin_lock_nested(&new_event->lock, SPINLOCK_NEW);
+
+	new_holder->event = new_event;
+	list_replace_init(&old_holder->event_list, &new_holder->event_list);
+
+	spin_unlock(&new_event->lock);
+	spin_unlock(&old_event->lock);
+
+	/* event == holder means we are referenced through the in event holder */
+	if (old_holder != &old_event->holder)
+		fsnotify_destroy_event_holder(old_holder);
+
+	fsnotify_get_event(new_event); /* on the list take reference */
+	fsnotify_put_event(old_event); /* off the list, drop reference */
+
+	return 0;
+}
+
+struct fsnotify_event *fsnotify_clone_event(struct fsnotify_event *old_event)
+{
+	struct fsnotify_event *event;
+
+	event = kmem_cache_alloc(fsnotify_event_cachep, GFP_KERNEL);
+	if (!event)
+		return NULL;
 
-	event->to_tell = NULL;
+	memcpy(event, old_event, sizeof(*event));
+	initialize_event(event);
 
-	event->file_name = NULL;
-	event->name_len = 0;
+	if (event->name_len) {
+		event->file_name = kstrdup(old_event->file_name, GFP_KERNEL);
+		if (!event->file_name) {
+			kmem_cache_free(fsnotify_event_cachep, event);
+			return NULL;
+		}
+	}
+	if (event->data_type == FSNOTIFY_EVENT_PATH)
+		path_get(&event->path);
 
-	event->sync_cookie = 0;
+	return event;
 }
 
 /*
@@ -353,7 +370,7 @@ struct fsnotify_event *fsnotify_create_event(struct inode *to_tell, __u32 mask,
 {
 	struct fsnotify_event *event;
 
-	event = kmem_cache_alloc(fsnotify_event_cachep, gfp);
+	event = kmem_cache_zalloc(fsnotify_event_cachep, gfp);
 	if (!event)
 		return NULL;
 
@@ -370,28 +387,18 @@ struct fsnotify_event *fsnotify_create_event(struct inode *to_tell, __u32 mask,
 
 	event->sync_cookie = cookie;
 	event->to_tell = to_tell;
+	event->data_type = data_type;
 
 	switch (data_type) {
-	case FSNOTIFY_EVENT_FILE: {
-		struct file *file = data;
-		struct path *path = &file->f_path;
-		event->path.dentry = path->dentry;
-		event->path.mnt = path->mnt;
-		path_get(&event->path);
-		event->data_type = FSNOTIFY_EVENT_PATH;
-		break;
-	}
 	case FSNOTIFY_EVENT_PATH: {
 		struct path *path = data;
 		event->path.dentry = path->dentry;
 		event->path.mnt = path->mnt;
 		path_get(&event->path);
-		event->data_type = FSNOTIFY_EVENT_PATH;
 		break;
 	}
 	case FSNOTIFY_EVENT_INODE:
 		event->inode = data;
-		event->data_type = FSNOTIFY_EVENT_INODE;
 		break;
 	case FSNOTIFY_EVENT_NONE:
 		event->inode = NULL;
@@ -412,8 +419,11 @@ __init int fsnotify_notification_init(void)
 	fsnotify_event_cachep = KMEM_CACHE(fsnotify_event, SLAB_PANIC);
 	fsnotify_event_holder_cachep = KMEM_CACHE(fsnotify_event_holder, SLAB_PANIC);
 
-	initialize_event(&q_overflow_event);
-	q_overflow_event.mask = FS_Q_OVERFLOW;
+	q_overflow_event = fsnotify_create_event(NULL, FS_Q_OVERFLOW, NULL,
+						 FSNOTIFY_EVENT_NONE, NULL, 0,
+						 GFP_KERNEL);
+	if (!q_overflow_event)
+		panic("unable to allocate fsnotify q_overflow_event\n");
 
 	return 0;
 }
diff --git a/fs/open.c b/fs/open.c
index 201041dfca57..ce737b3e4c6f 100644
--- a/fs/open.c
+++ b/fs/open.c
@@ -1060,7 +1060,7 @@ long do_sys_open(int dfd, const char __user *filename, int flags, int mode)
 				put_unused_fd(fd);
 				fd = PTR_ERR(f);
 			} else {
-				fsnotify_open(f->f_path.dentry);
+				fsnotify_open(f);
 				fd_install(fd, f);
 			}
 		}
diff --git a/fs/read_write.c b/fs/read_write.c
index b7f4a1f94d48..1963debee646 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -293,7 +293,7 @@ ssize_t vfs_read(struct file *file, char __user *buf, size_t count, loff_t *pos)
 		else
 			ret = do_sync_read(file, buf, count, pos);
 		if (ret > 0) {
-			fsnotify_access(file->f_path.dentry);
+			fsnotify_access(file);
 			add_rchar(current, ret);
 		}
 		inc_syscr(current);
@@ -348,7 +348,7 @@ ssize_t vfs_write(struct file *file, const char __user *buf, size_t count, loff_
 		else
 			ret = do_sync_write(file, buf, count, pos);
 		if (ret > 0) {
-			fsnotify_modify(file->f_path.dentry);
+			fsnotify_modify(file);
 			add_wchar(current, ret);
 		}
 		inc_syscw(current);
@@ -656,9 +656,9 @@ out:
 		kfree(iov);
 	if ((ret + (type == READ)) > 0) {
 		if (type == READ)
-			fsnotify_access(file->f_path.dentry);
+			fsnotify_access(file);
 		else
-			fsnotify_modify(file->f_path.dentry);
+			fsnotify_modify(file);
 	}
 	return ret;
 }
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 93131e193d0e..ab92f1321abd 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -760,7 +760,7 @@ struct inode {
 
 #ifdef CONFIG_FSNOTIFY
 	__u32			i_fsnotify_mask; /* all events this inode cares about */
-	struct hlist_head	i_fsnotify_mark_entries; /* fsnotify mark entries */
+	struct hlist_head	i_fsnotify_marks;
 #endif
 
 #ifdef CONFIG_INOTIFY
diff --git a/include/linux/fsnotify.h b/include/linux/fsnotify.h
index 936f9aa8bb97..294f488b7b10 100644
--- a/include/linux/fsnotify.h
+++ b/include/linux/fsnotify.h
@@ -20,35 +20,38 @@
  * fsnotify_d_instantiate - instantiate a dentry for inode
  * Called with dcache_lock held.
  */
-static inline void fsnotify_d_instantiate(struct dentry *entry,
+static inline void fsnotify_d_instantiate(struct dentry *dentry,
 						struct inode *inode)
 {
-	__fsnotify_d_instantiate(entry, inode);
+	__fsnotify_d_instantiate(dentry, inode);
 
-	inotify_d_instantiate(entry, inode);
+	inotify_d_instantiate(dentry, inode);
 }
 
 /* Notify this dentry's parent about a child's events. */
-static inline void fsnotify_parent(struct dentry *dentry, __u32 mask)
+static inline void fsnotify_parent(struct path *path, struct dentry *dentry, __u32 mask)
 {
-	__fsnotify_parent(dentry, mask);
+	if (!dentry)
+		dentry = path->dentry;
+
+	__fsnotify_parent(path, dentry, mask);
 
 	inotify_dentry_parent_queue_event(dentry, mask, 0, dentry->d_name.name);
 }
 
 /*
- * fsnotify_d_move - entry has been moved
- * Called with dcache_lock and entry->d_lock held.
+ * fsnotify_d_move - dentry has been moved
+ * Called with dcache_lock and dentry->d_lock held.
  */
-static inline void fsnotify_d_move(struct dentry *entry)
+static inline void fsnotify_d_move(struct dentry *dentry)
 {
 	/*
-	 * On move we need to update entry->d_flags to indicate if the new parent
-	 * cares about events from this entry.
+	 * On move we need to update dentry->d_flags to indicate if the new parent
+	 * cares about events from this dentry.
 	 */
-	__fsnotify_update_dcache_flags(entry);
+	__fsnotify_update_dcache_flags(dentry);
 
-	inotify_d_move(entry);
+	inotify_d_move(dentry);
 }
 
 /*
@@ -124,7 +127,7 @@ static inline void fsnotify_nameremove(struct dentry *dentry, int isdir)
 	if (isdir)
 		mask |= FS_IN_ISDIR;
 
-	fsnotify_parent(dentry, mask);
+	fsnotify_parent(NULL, dentry, mask);
 }
 
 /*
@@ -183,9 +186,10 @@ static inline void fsnotify_mkdir(struct inode *inode, struct dentry *dentry)
 /*
  * fsnotify_access - file was read
  */
-static inline void fsnotify_access(struct dentry *dentry)
+static inline void fsnotify_access(struct file *file)
 {
-	struct inode *inode = dentry->d_inode;
+	struct path *path = &file->f_path;
+	struct inode *inode = path->dentry->d_inode;
 	__u32 mask = FS_ACCESS;
 
 	if (S_ISDIR(inode->i_mode))
@@ -193,16 +197,17 @@ static inline void fsnotify_access(struct dentry *dentry)
 
 	inotify_inode_queue_event(inode, mask, 0, NULL, NULL);
 
-	fsnotify_parent(dentry, mask);
-	fsnotify(inode, mask, inode, FSNOTIFY_EVENT_INODE, NULL, 0);
+	fsnotify_parent(path, NULL, mask);
+	fsnotify(inode, mask, path, FSNOTIFY_EVENT_PATH, NULL, 0);
 }
 
 /*
  * fsnotify_modify - file was modified
  */
-static inline void fsnotify_modify(struct dentry *dentry)
+static inline void fsnotify_modify(struct file *file)
 {
-	struct inode *inode = dentry->d_inode;
+	struct path *path = &file->f_path;
+	struct inode *inode = path->dentry->d_inode;
 	__u32 mask = FS_MODIFY;
 
 	if (S_ISDIR(inode->i_mode))
@@ -210,16 +215,17 @@ static inline void fsnotify_modify(struct dentry *dentry)
 
 	inotify_inode_queue_event(inode, mask, 0, NULL, NULL);
 
-	fsnotify_parent(dentry, mask);
-	fsnotify(inode, mask, inode, FSNOTIFY_EVENT_INODE, NULL, 0);
+	fsnotify_parent(path, NULL, mask);
+	fsnotify(inode, mask, path, FSNOTIFY_EVENT_PATH, NULL, 0);
 }
 
 /*
  * fsnotify_open - file was opened
  */
-static inline void fsnotify_open(struct dentry *dentry)
+static inline void fsnotify_open(struct file *file)
 {
-	struct inode *inode = dentry->d_inode;
+	struct path *path = &file->f_path;
+	struct inode *inode = path->dentry->d_inode;
 	__u32 mask = FS_OPEN;
 
 	if (S_ISDIR(inode->i_mode))
@@ -227,8 +233,8 @@ static inline void fsnotify_open(struct dentry *dentry)
 
 	inotify_inode_queue_event(inode, mask, 0, NULL, NULL);
 
-	fsnotify_parent(dentry, mask);
-	fsnotify(inode, mask, inode, FSNOTIFY_EVENT_INODE, NULL, 0);
+	fsnotify_parent(path, NULL, mask);
+	fsnotify(inode, mask, path, FSNOTIFY_EVENT_PATH, NULL, 0);
 }
 
 /*
@@ -236,8 +242,8 @@ static inline void fsnotify_open(struct dentry *dentry)
  */
 static inline void fsnotify_close(struct file *file)
 {
-	struct dentry *dentry = file->f_path.dentry;
-	struct inode *inode = dentry->d_inode;
+	struct path *path = &file->f_path;
+	struct inode *inode = file->f_path.dentry->d_inode;
 	fmode_t mode = file->f_mode;
 	__u32 mask = (mode & FMODE_WRITE) ? FS_CLOSE_WRITE : FS_CLOSE_NOWRITE;
 
@@ -246,8 +252,8 @@ static inline void fsnotify_close(struct file *file)
 
 	inotify_inode_queue_event(inode, mask, 0, NULL, NULL);
 
-	fsnotify_parent(dentry, mask);
-	fsnotify(inode, mask, file, FSNOTIFY_EVENT_FILE, NULL, 0);
+	fsnotify_parent(path, NULL, mask);
+	fsnotify(inode, mask, path, FSNOTIFY_EVENT_PATH, NULL, 0);
 }
 
 /*
@@ -263,7 +269,7 @@ static inline void fsnotify_xattr(struct dentry *dentry)
 
 	inotify_inode_queue_event(inode, mask, 0, NULL, NULL);
 
-	fsnotify_parent(dentry, mask);
+	fsnotify_parent(NULL, dentry, mask);
 	fsnotify(inode, mask, inode, FSNOTIFY_EVENT_INODE, NULL, 0);
 }
 
@@ -299,7 +305,7 @@ static inline void fsnotify_change(struct dentry *dentry, unsigned int ia_valid)
 			mask |= FS_IN_ISDIR;
 		inotify_inode_queue_event(inode, mask, 0, NULL, NULL);
 
-		fsnotify_parent(dentry, mask);
+		fsnotify_parent(NULL, dentry, mask);
 		fsnotify(inode, mask, inode, FSNOTIFY_EVENT_INODE, NULL, 0);
 	}
 }
diff --git a/include/linux/fsnotify_backend.h b/include/linux/fsnotify_backend.h
index 4d6f47b51189..ff654c1932f2 100644
--- a/include/linux/fsnotify_backend.h
+++ b/include/linux/fsnotify_backend.h
@@ -58,13 +58,11 @@
 				   FS_MOVED_FROM | FS_MOVED_TO | FS_CREATE |\
 				   FS_DELETE)
 
-/* listeners that hard code group numbers near the top */
-#define DNOTIFY_GROUP_NUM	UINT_MAX
-#define INOTIFY_GROUP_NUM	(DNOTIFY_GROUP_NUM-1)
+#define FS_MOVE			(FS_MOVED_FROM | FS_MOVED_TO)
 
 struct fsnotify_group;
 struct fsnotify_event;
-struct fsnotify_mark_entry;
+struct fsnotify_mark;
 struct fsnotify_event_private_data;
 
 /*
@@ -80,10 +78,12 @@ struct fsnotify_event_private_data;
  *		valid group and inode to use to clean up.
  */
 struct fsnotify_ops {
-	bool (*should_send_event)(struct fsnotify_group *group, struct inode *inode, __u32 mask);
+	bool (*should_send_event)(struct fsnotify_group *group, struct inode *inode,
+				  struct vfsmount *mnt, __u32 mask, void *data,
+				  int data_type);
 	int (*handle_event)(struct fsnotify_group *group, struct fsnotify_event *event);
 	void (*free_group_priv)(struct fsnotify_group *group);
-	void (*freeing_mark)(struct fsnotify_mark_entry *entry, struct fsnotify_group *group);
+	void (*freeing_mark)(struct fsnotify_mark *mark, struct fsnotify_group *group);
 	void (*free_event_priv)(struct fsnotify_event_private_data *priv);
 };
 
@@ -96,10 +96,14 @@ struct fsnotify_ops {
 struct fsnotify_group {
 	/*
 	 * global list of all groups receiving events from fsnotify.
-	 * anchored by fsnotify_groups and protected by either fsnotify_grp_mutex
+	 * anchored by fsnotify_inode_groups and protected by either fsnotify_grp_mutex
 	 * or fsnotify_grp_srcu depending on write vs read.
 	 */
-	struct list_head group_list;
+	struct list_head inode_group_list;
+	/*
+	 * same as above except anchored by fsnotify_vfsmount_groups
+	 */
+	struct list_head vfsmount_group_list;
 
 	/*
 	 * Defines all of the event types in which this group is interested.
@@ -119,7 +123,6 @@ struct fsnotify_group {
 	 * closed.
 	 */
 	atomic_t refcnt;		/* things with interest in this group */
-	unsigned int group_num;		/* simply prevents accidental group collision */
 
 	const struct fsnotify_ops *ops;	/* how this group handles things */
 
@@ -130,15 +133,16 @@ struct fsnotify_group {
 	unsigned int q_len;			/* events on the queue */
 	unsigned int max_events;		/* maximum events allowed on the list */
 
-	/* stores all fastapth entries assoc with this group so they can be cleaned on unregister */
-	spinlock_t mark_lock;		/* protect mark_entries list */
-	atomic_t num_marks;		/* 1 for each mark entry and 1 for not being
+	/* stores all fastpath marks assoc with this group so they can be cleaned on unregister */
+	spinlock_t mark_lock;		/* protect marks_list */
+	atomic_t num_marks;		/* 1 for each mark and 1 for not being
 					 * past the point of no return when freeing
 					 * a group */
-	struct list_head mark_entries;	/* all inode mark entries for this group */
+	struct list_head marks_list;	/* all inode marks for this group */
 
 	/* prevents double list_del of group_list.  protected by global fsnotify_grp_mutex */
-	bool on_group_list;
+	bool on_inode_group_list;
+	bool on_vfsmount_group_list;
 
 	/* groups can define private fields here or use the void *private */
 	union {
@@ -210,7 +214,6 @@ struct fsnotify_event {
 #define FSNOTIFY_EVENT_NONE	0
 #define FSNOTIFY_EVENT_PATH	1
 #define FSNOTIFY_EVENT_INODE	2
-#define FSNOTIFY_EVENT_FILE	3
 	int data_type;		/* which of the above union we have */
 	atomic_t refcnt;	/* how many groups still are using/need to send this event */
 	__u32 mask;		/* the type of access, bitwise OR for FS_* event types */
@@ -223,7 +226,25 @@ struct fsnotify_event {
 };
 
 /*
- * a mark is simply an entry attached to an in core inode which allows an
+ * Inode specific fields in an fsnotify_mark
+ */
+struct fsnotify_inode_mark {
+	struct inode *inode;		/* inode this mark is associated with */
+	struct hlist_node i_list;	/* list of marks by inode->i_fsnotify_marks */
+	struct list_head free_i_list;	/* tmp list used when freeing this mark */
+};
+
+/*
+ * Mount point specific fields in an fsnotify_mark
+ */
+struct fsnotify_vfsmount_mark {
+	struct vfsmount *mnt;		/* vfsmount this mark is associated with */
+	struct hlist_node m_list;	/* list of marks by inode->i_fsnotify_marks */
+	struct list_head free_m_list;	/* tmp list used when freeing this mark */
+};
+
+/*
+ * a mark is simply an object attached to an in core inode which allows an
  * fsnotify listener to indicate they are either no longer interested in events
  * of a type matching mask or only interested in those events.
  *
@@ -232,19 +253,23 @@ struct fsnotify_event {
  * (such as dnotify) will flush these when the open fd is closed and not at
  * inode eviction or modification.
  */
-struct fsnotify_mark_entry {
-	__u32 mask;			/* mask this mark entry is for */
+struct fsnotify_mark {
+	__u32 mask;			/* mask this mark is for */
 	/* we hold ref for each i_list and g_list.  also one ref for each 'thing'
 	 * in kernel that found and may be using this mark. */
 	atomic_t refcnt;		/* active things looking at this mark */
-	struct inode *inode;		/* inode this entry is associated with */
-	struct fsnotify_group *group;	/* group this mark entry is for */
-	struct hlist_node i_list;	/* list of mark_entries by inode->i_fsnotify_mark_entries */
-	struct list_head g_list;	/* list of mark_entries by group->i_fsnotify_mark_entries */
-	spinlock_t lock;		/* protect group, inode, and killme */
-	struct list_head free_i_list;	/* tmp list used when freeing this mark */
+	struct fsnotify_group *group;	/* group this mark is for */
+	struct list_head g_list;	/* list of marks by group->i_fsnotify_marks */
+	spinlock_t lock;		/* protect group and inode */
+	union {
+		struct fsnotify_inode_mark i;
+		struct fsnotify_vfsmount_mark m;
+	};
 	struct list_head free_g_list;	/* tmp list used when freeing this mark */
-	void (*free_mark)(struct fsnotify_mark_entry *entry); /* called on final put+free */
+#define FSNOTIFY_MARK_FLAG_INODE	0x01
+#define FSNOTIFY_MARK_FLAG_VFSMOUNT	0x02
+	unsigned int flags;		/* vfsmount or inode mark? */
+	void (*free_mark)(struct fsnotify_mark *mark); /* called on final put+free */
 };
 
 #ifdef CONFIG_FSNOTIFY
@@ -254,7 +279,7 @@ struct fsnotify_mark_entry {
 /* main fsnotify call to send events */
 extern void fsnotify(struct inode *to_tell, __u32 mask, void *data, int data_is,
 		     const char *name, u32 cookie);
-extern void __fsnotify_parent(struct dentry *dentry, __u32 mask);
+extern void __fsnotify_parent(struct path *path, struct dentry *dentry, __u32 mask);
 extern void __fsnotify_inode_delete(struct inode *inode);
 extern u32 fsnotify_get_cookie(void);
 
@@ -307,12 +332,10 @@ static inline void __fsnotify_d_instantiate(struct dentry *dentry, struct inode
 /* must call when a group changes its ->mask */
 extern void fsnotify_recalc_global_mask(void);
 /* get a reference to an existing or create a new group */
-extern struct fsnotify_group *fsnotify_obtain_group(unsigned int group_num,
-						    __u32 mask,
-						    const struct fsnotify_ops *ops);
+extern struct fsnotify_group *fsnotify_alloc_group(const struct fsnotify_ops *ops);
 /* run all marks associated with this group and update group->mask */
 extern void fsnotify_recalc_group_mask(struct fsnotify_group *group);
-/* drop reference on a group from fsnotify_obtain_group */
+/* drop reference on a group from fsnotify_alloc_group */
 extern void fsnotify_put_group(struct fsnotify_group *group);
 
 /* take a reference to an event */
@@ -323,8 +346,10 @@ extern struct fsnotify_event_private_data *fsnotify_remove_priv_from_event(struc
 									   struct fsnotify_event *event);
 
 /* attach the event to the group notification queue */
-extern int fsnotify_add_notify_event(struct fsnotify_group *group, struct fsnotify_event *event,
-				     struct fsnotify_event_private_data *priv);
+extern int fsnotify_add_notify_event(struct fsnotify_group *group,
+				     struct fsnotify_event *event,
+				     struct fsnotify_event_private_data *priv,
+				     int (*merge)(struct list_head *, struct fsnotify_event *));
 /* true if the group notification queue is empty */
 extern bool fsnotify_notify_queue_is_empty(struct fsnotify_group *group);
 /* return, but do not dequeue the first event on the notification queue */
@@ -336,17 +361,19 @@ extern struct fsnotify_event *fsnotify_remove_notify_event(struct fsnotify_group
 
 /* run all marks associated with an inode and update inode->i_fsnotify_mask */
 extern void fsnotify_recalc_inode_mask(struct inode *inode);
-extern void fsnotify_init_mark(struct fsnotify_mark_entry *entry, void (*free_mark)(struct fsnotify_mark_entry *entry));
+extern void fsnotify_init_mark(struct fsnotify_mark *mark, void (*free_mark)(struct fsnotify_mark *mark));
 /* find (and take a reference) to a mark associated with group and inode */
-extern struct fsnotify_mark_entry *fsnotify_find_mark_entry(struct fsnotify_group *group, struct inode *inode);
+extern struct fsnotify_mark *fsnotify_find_mark(struct fsnotify_group *group, struct inode *inode);
+/* copy the values from old into new */
+extern void fsnotify_duplicate_mark(struct fsnotify_mark *new, struct fsnotify_mark *old);
 /* attach the mark to both the group and the inode */
-extern int fsnotify_add_mark(struct fsnotify_mark_entry *entry, struct fsnotify_group *group, struct inode *inode);
+extern int fsnotify_add_mark(struct fsnotify_mark *mark, struct fsnotify_group *group, struct inode *inode, int allow_dups);
 /* given a mark, flag it to be freed when all references are dropped */
-extern void fsnotify_destroy_mark_by_entry(struct fsnotify_mark_entry *entry);
+extern void fsnotify_destroy_mark(struct fsnotify_mark *mark);
 /* run all the marks in a group, and flag them to be freed */
 extern void fsnotify_clear_marks_by_group(struct fsnotify_group *group);
-extern void fsnotify_get_mark(struct fsnotify_mark_entry *entry);
-extern void fsnotify_put_mark(struct fsnotify_mark_entry *entry);
+extern void fsnotify_get_mark(struct fsnotify_mark *mark);
+extern void fsnotify_put_mark(struct fsnotify_mark *mark);
 extern void fsnotify_unmount_inodes(struct list_head *list);
 
 /* put here because inotify does some weird stuff when destroying watches */
@@ -354,13 +381,18 @@ extern struct fsnotify_event *fsnotify_create_event(struct inode *to_tell, __u32
 						    void *data, int data_is, const char *name,
 						    u32 cookie, gfp_t gfp);
 
+/* fanotify likes to change events after they are on lists... */
+extern struct fsnotify_event *fsnotify_clone_event(struct fsnotify_event *old_event);
+extern int fsnotify_replace_event(struct fsnotify_event_holder *old_holder,
+				  struct fsnotify_event *new_event);
+
 #else
 
 static inline void fsnotify(struct inode *to_tell, __u32 mask, void *data, int data_is,
 			    const char *name, u32 cookie)
 {}
 
-static inline void __fsnotify_parent(struct dentry *dentry, __u32 mask)
+static inline void __fsnotify_parent(struct path *path, struct dentry *dentry, __u32 mask)
 {}
 
 static inline void __fsnotify_inode_delete(struct inode *inode)
diff --git a/init/Kconfig b/init/Kconfig
index 3fbaee7b3ad2..9047f8f03479 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -302,13 +302,17 @@ config AUDITSYSCALL
 	help
 	  Enable low-overhead system-call auditing infrastructure that
 	  can be used independently or with another kernel subsystem,
-	  such as SELinux.  To use audit's filesystem watch feature, please
-	  ensure that INOTIFY is configured.
+	  such as SELinux.
+
+config AUDIT_WATCH
+	def_bool y
+	depends on AUDITSYSCALL
+	select FSNOTIFY
 
 config AUDIT_TREE
 	def_bool y
 	depends on AUDITSYSCALL
-	select INOTIFY
+	select FSNOTIFY
 
 menu "RCU Subsystem"
 
diff --git a/kernel/Makefile b/kernel/Makefile
index 0ae57a83d481..1620454b8c52 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -68,10 +68,11 @@ obj-$(CONFIG_IKCONFIG) += configs.o
 obj-$(CONFIG_RESOURCE_COUNTERS) += res_counter.o
 obj-$(CONFIG_STOP_MACHINE) += stop_machine.o
 obj-$(CONFIG_KPROBES_SANITY_TEST) += test_kprobes.o
-obj-$(CONFIG_AUDIT) += audit.o auditfilter.o audit_watch.o
+obj-$(CONFIG_AUDIT) += audit.o auditfilter.o
 obj-$(CONFIG_AUDITSYSCALL) += auditsc.o
-obj-$(CONFIG_GCOV_KERNEL) += gcov/
+obj-$(CONFIG_AUDIT_WATCH) += audit_watch.o
 obj-$(CONFIG_AUDIT_TREE) += audit_tree.o
+obj-$(CONFIG_GCOV_KERNEL) += gcov/
 obj-$(CONFIG_KPROBES) += kprobes.o
 obj-$(CONFIG_KGDB) += kgdb.o
 obj-$(CONFIG_DETECT_SOFTLOCKUP) += softlockup.o
diff --git a/kernel/audit.c b/kernel/audit.c
index 5feed232be9d..057a3a4dbfa5 100644
--- a/kernel/audit.c
+++ b/kernel/audit.c
@@ -55,7 +55,6 @@
 #include <net/netlink.h>
 #include <linux/skbuff.h>
 #include <linux/netlink.h>
-#include <linux/inotify.h>
 #include <linux/freezer.h>
 #include <linux/tty.h>
 
diff --git a/kernel/audit.h b/kernel/audit.h
index 208687be4f30..f7206db4e13d 100644
--- a/kernel/audit.h
+++ b/kernel/audit.h
@@ -103,21 +103,27 @@ extern struct mutex audit_filter_mutex;
 extern void audit_free_rule_rcu(struct rcu_head *);
 extern struct list_head audit_filter_list[];
 
+extern struct audit_entry *audit_dupe_rule(struct audit_krule *old);
+
 /* audit watch functions */
-extern unsigned long audit_watch_inode(struct audit_watch *watch);
-extern dev_t audit_watch_dev(struct audit_watch *watch);
+#ifdef CONFIG_AUDIT_WATCH
 extern void audit_put_watch(struct audit_watch *watch);
 extern void audit_get_watch(struct audit_watch *watch);
 extern int audit_to_watch(struct audit_krule *krule, char *path, int len, u32 op);
-extern int audit_add_watch(struct audit_krule *krule);
-extern void audit_remove_watch(struct audit_watch *watch);
-extern void audit_remove_watch_rule(struct audit_krule *krule, struct list_head *list);
-extern void audit_inotify_unregister(struct list_head *in_list);
+extern int audit_add_watch(struct audit_krule *krule, struct list_head **list);
+extern void audit_remove_watch_rule(struct audit_krule *krule);
 extern char *audit_watch_path(struct audit_watch *watch);
-extern struct list_head *audit_watch_rules(struct audit_watch *watch);
-
-extern struct audit_entry *audit_dupe_rule(struct audit_krule *old,
-					   struct audit_watch *watch);
+extern int audit_watch_compare(struct audit_watch *watch, unsigned long ino, dev_t dev);
+#else
+#define audit_put_watch(w) {}
+#define audit_get_watch(w) {}
+#define audit_to_watch(k, p, l, o) (-EINVAL)
+#define audit_add_watch(k, l) (-EINVAL)
+#define audit_remove_watch_rule(k) BUG()
+#define audit_watch_path(w) ""
+#define audit_watch_compare(w, i, d) 0
+
+#endif /* CONFIG_AUDIT_WATCH */
 
 #ifdef CONFIG_AUDIT_TREE
 extern struct audit_chunk *audit_tree_lookup(const struct inode *);
diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c
index 2451dc6f3282..001f4e75d80d 100644
--- a/kernel/audit_tree.c
+++ b/kernel/audit_tree.c
@@ -1,5 +1,5 @@
 #include "audit.h"
-#include <linux/inotify.h>
+#include <linux/fsnotify_backend.h>
 #include <linux/namei.h>
 #include <linux/mount.h>
 #include <linux/kthread.h>
@@ -21,7 +21,7 @@ struct audit_tree {
 
 struct audit_chunk {
 	struct list_head hash;
-	struct inotify_watch watch;
+	struct fsnotify_mark mark;
 	struct list_head trees;		/* with root here */
 	int dead;
 	int count;
@@ -67,7 +67,7 @@ static LIST_HEAD(prune_list);
  * that makes a difference.  Some.
  */
 
-static struct inotify_handle *rtree_ih;
+static struct fsnotify_group *audit_tree_group;
 
 static struct audit_tree *alloc_tree(const char *s)
 {
@@ -110,29 +110,6 @@ const char *audit_tree_path(struct audit_tree *tree)
 	return tree->pathname;
 }
 
-static struct audit_chunk *alloc_chunk(int count)
-{
-	struct audit_chunk *chunk;
-	size_t size;
-	int i;
-
-	size = offsetof(struct audit_chunk, owners) + count * sizeof(struct node);
-	chunk = kzalloc(size, GFP_KERNEL);
-	if (!chunk)
-		return NULL;
-
-	INIT_LIST_HEAD(&chunk->hash);
-	INIT_LIST_HEAD(&chunk->trees);
-	chunk->count = count;
-	atomic_long_set(&chunk->refs, 1);
-	for (i = 0; i < count; i++) {
-		INIT_LIST_HEAD(&chunk->owners[i].list);
-		chunk->owners[i].index = i;
-	}
-	inotify_init_watch(&chunk->watch);
-	return chunk;
-}
-
 static void free_chunk(struct audit_chunk *chunk)
 {
 	int i;
@@ -156,6 +133,35 @@ static void __put_chunk(struct rcu_head *rcu)
 	audit_put_chunk(chunk);
 }
 
+static void audit_tree_destroy_watch(struct fsnotify_mark *entry)
+{
+	struct audit_chunk *chunk = container_of(entry, struct audit_chunk, mark);
+	call_rcu(&chunk->head, __put_chunk);
+}
+
+static struct audit_chunk *alloc_chunk(int count)
+{
+	struct audit_chunk *chunk;
+	size_t size;
+	int i;
+
+	size = offsetof(struct audit_chunk, owners) + count * sizeof(struct node);
+	chunk = kzalloc(size, GFP_KERNEL);
+	if (!chunk)
+		return NULL;
+
+	INIT_LIST_HEAD(&chunk->hash);
+	INIT_LIST_HEAD(&chunk->trees);
+	chunk->count = count;
+	atomic_long_set(&chunk->refs, 1);
+	for (i = 0; i < count; i++) {
+		INIT_LIST_HEAD(&chunk->owners[i].list);
+		chunk->owners[i].index = i;
+	}
+	fsnotify_init_mark(&chunk->mark, audit_tree_destroy_watch);
+	return chunk;
+}
+
 enum {HASH_SIZE = 128};
 static struct list_head chunk_hash_heads[HASH_SIZE];
 static __cacheline_aligned_in_smp DEFINE_SPINLOCK(hash_lock);
@@ -166,10 +172,15 @@ static inline struct list_head *chunk_hash(const struct inode *inode)
 	return chunk_hash_heads + n % HASH_SIZE;
 }
 
-/* hash_lock is held by caller */
+/* hash_lock & entry->lock is held by caller */
 static void insert_hash(struct audit_chunk *chunk)
 {
-	struct list_head *list = chunk_hash(chunk->watch.inode);
+	struct fsnotify_mark *entry = &chunk->mark;
+	struct list_head *list;
+
+	if (!entry->i.inode)
+		return;
+	list = chunk_hash(entry->i.inode);
 	list_add_rcu(&chunk->hash, list);
 }
 
@@ -180,7 +191,8 @@ struct audit_chunk *audit_tree_lookup(const struct inode *inode)
 	struct audit_chunk *p;
 
 	list_for_each_entry_rcu(p, list, hash) {
-		if (p->watch.inode == inode) {
+		/* mark.inode may have gone NULL, but who cares? */
+		if (p->mark.i.inode == inode) {
 			atomic_long_inc(&p->refs);
 			return p;
 		}
@@ -209,38 +221,19 @@ static struct audit_chunk *find_chunk(struct node *p)
 static void untag_chunk(struct node *p)
 {
 	struct audit_chunk *chunk = find_chunk(p);
+	struct fsnotify_mark *entry = &chunk->mark;
 	struct audit_chunk *new;
 	struct audit_tree *owner;
 	int size = chunk->count - 1;
 	int i, j;
 
-	if (!pin_inotify_watch(&chunk->watch)) {
-		/*
-		 * Filesystem is shutting down; all watches are getting
-		 * evicted, just take it off the node list for this
-		 * tree and let the eviction logics take care of the
-		 * rest.
-		 */
-		owner = p->owner;
-		if (owner->root == chunk) {
-			list_del_init(&owner->same_root);
-			owner->root = NULL;
-		}
-		list_del_init(&p->list);
-		p->owner = NULL;
-		put_tree(owner);
-		return;
-	}
+	fsnotify_get_mark(entry);
 
 	spin_unlock(&hash_lock);
 
-	/*
-	 * pin_inotify_watch() succeeded, so the watch won't go away
-	 * from under us.
-	 */
-	mutex_lock(&chunk->watch.inode->inotify_mutex);
-	if (chunk->dead) {
-		mutex_unlock(&chunk->watch.inode->inotify_mutex);
+	spin_lock(&entry->lock);
+	if (chunk->dead || !entry->i.inode) {
+		spin_unlock(&entry->lock);
 		goto out;
 	}
 
@@ -255,16 +248,17 @@ static void untag_chunk(struct node *p)
 		list_del_init(&p->list);
 		list_del_rcu(&chunk->hash);
 		spin_unlock(&hash_lock);
-		inotify_evict_watch(&chunk->watch);
-		mutex_unlock(&chunk->watch.inode->inotify_mutex);
-		put_inotify_watch(&chunk->watch);
+		spin_unlock(&entry->lock);
+		fsnotify_destroy_mark(entry);
+		fsnotify_put_mark(entry);
 		goto out;
 	}
 
 	new = alloc_chunk(size);
 	if (!new)
 		goto Fallback;
-	if (inotify_clone_watch(&chunk->watch, &new->watch) < 0) {
+	fsnotify_duplicate_mark(&new->mark, entry);
+	if (fsnotify_add_mark(&new->mark, new->mark.group, new->mark.i.inode, 1)) {
 		free_chunk(new);
 		goto Fallback;
 	}
@@ -297,9 +291,9 @@ static void untag_chunk(struct node *p)
 	list_for_each_entry(owner, &new->trees, same_root)
 		owner->root = new;
 	spin_unlock(&hash_lock);
-	inotify_evict_watch(&chunk->watch);
-	mutex_unlock(&chunk->watch.inode->inotify_mutex);
-	put_inotify_watch(&chunk->watch);
+	spin_unlock(&entry->lock);
+	fsnotify_destroy_mark(entry);
+	fsnotify_put_mark(entry);
 	goto out;
 
 Fallback:
@@ -313,31 +307,33 @@ Fallback:
 	p->owner = NULL;
 	put_tree(owner);
 	spin_unlock(&hash_lock);
-	mutex_unlock(&chunk->watch.inode->inotify_mutex);
+	spin_unlock(&entry->lock);
 out:
-	unpin_inotify_watch(&chunk->watch);
+	fsnotify_put_mark(entry);
 	spin_lock(&hash_lock);
 }
 
 static int create_chunk(struct inode *inode, struct audit_tree *tree)
 {
+	struct fsnotify_mark *entry;
 	struct audit_chunk *chunk = alloc_chunk(1);
 	if (!chunk)
 		return -ENOMEM;
 
-	if (inotify_add_watch(rtree_ih, &chunk->watch, inode, IN_IGNORED | IN_DELETE_SELF) < 0) {
+	entry = &chunk->mark;
+	if (fsnotify_add_mark(entry, audit_tree_group, inode, 0)) {
 		free_chunk(chunk);
 		return -ENOSPC;
 	}
 
-	mutex_lock(&inode->inotify_mutex);
+	spin_lock(&entry->lock);
 	spin_lock(&hash_lock);
 	if (tree->goner) {
 		spin_unlock(&hash_lock);
 		chunk->dead = 1;
-		inotify_evict_watch(&chunk->watch);
-		mutex_unlock(&inode->inotify_mutex);
-		put_inotify_watch(&chunk->watch);
+		spin_unlock(&entry->lock);
+		fsnotify_destroy_mark(entry);
+		fsnotify_put_mark(entry);
 		return 0;
 	}
 	chunk->owners[0].index = (1U << 31);
@@ -350,30 +346,31 @@ static int create_chunk(struct inode *inode, struct audit_tree *tree)
 	}
 	insert_hash(chunk);
 	spin_unlock(&hash_lock);
-	mutex_unlock(&inode->inotify_mutex);
+	spin_unlock(&entry->lock);
 	return 0;
 }
 
 /* the first tagged inode becomes root of tree */
 static int tag_chunk(struct inode *inode, struct audit_tree *tree)
 {
-	struct inotify_watch *watch;
+	struct fsnotify_mark *old_entry, *chunk_entry;
 	struct audit_tree *owner;
 	struct audit_chunk *chunk, *old;
 	struct node *p;
 	int n;
 
-	if (inotify_find_watch(rtree_ih, inode, &watch) < 0)
+	old_entry = fsnotify_find_mark(audit_tree_group, inode);
+	if (!old_entry)
 		return create_chunk(inode, tree);
 
-	old = container_of(watch, struct audit_chunk, watch);
+	old = container_of(old_entry, struct audit_chunk, mark);
 
 	/* are we already there? */
 	spin_lock(&hash_lock);
 	for (n = 0; n < old->count; n++) {
 		if (old->owners[n].owner == tree) {
 			spin_unlock(&hash_lock);
-			put_inotify_watch(watch);
+			fsnotify_put_mark(old_entry);
 			return 0;
 		}
 	}
@@ -382,22 +379,40 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree)
 	chunk = alloc_chunk(old->count + 1);
 	if (!chunk)
 		return -ENOMEM;
+	chunk_entry = &chunk->mark;
 
-	mutex_lock(&inode->inotify_mutex);
-	if (inotify_clone_watch(&old->watch, &chunk->watch) < 0) {
-		mutex_unlock(&inode->inotify_mutex);
-		put_inotify_watch(&old->watch);
+	spin_lock(&old_entry->lock);
+	if (!old_entry->i.inode) {
+		/* old_entry is being shot, lets just lie */
+		spin_unlock(&old_entry->lock);
+		fsnotify_put_mark(old_entry);
 		free_chunk(chunk);
+		return -ENOENT;
+	}
+
+	fsnotify_duplicate_mark(chunk_entry, old_entry);
+	if (fsnotify_add_mark(chunk_entry, chunk_entry->group, chunk_entry->i.inode, 1)) {
+		spin_unlock(&old_entry->lock);
+		free_chunk(chunk);
+		fsnotify_put_mark(old_entry);
 		return -ENOSPC;
 	}
+
+	/* even though we hold old_entry->lock, this is safe since chunk_entry->lock could NEVER have been grabbed before */
+	spin_lock(&chunk_entry->lock);
 	spin_lock(&hash_lock);
+
+	/* we now hold old_entry->lock, chunk_entry->lock, and hash_lock */
 	if (tree->goner) {
 		spin_unlock(&hash_lock);
 		chunk->dead = 1;
-		inotify_evict_watch(&chunk->watch);
-		mutex_unlock(&inode->inotify_mutex);
-		put_inotify_watch(&old->watch);
-		put_inotify_watch(&chunk->watch);
+		spin_unlock(&chunk_entry->lock);
+		spin_unlock(&old_entry->lock);
+
+		fsnotify_destroy_mark(chunk_entry);
+
+		fsnotify_put_mark(chunk_entry);
+		fsnotify_put_mark(old_entry);
 		return 0;
 	}
 	list_replace_init(&old->trees, &chunk->trees);
@@ -423,9 +438,10 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree)
 		list_add(&tree->same_root, &chunk->trees);
 	}
 	spin_unlock(&hash_lock);
-	inotify_evict_watch(&old->watch);
-	mutex_unlock(&inode->inotify_mutex);
-	put_inotify_watch(&old->watch);
+	spin_unlock(&chunk_entry->lock);
+	spin_unlock(&old_entry->lock);
+	fsnotify_destroy_mark(old_entry);
+	fsnotify_put_mark(old_entry);
 	return 0;
 }
 
@@ -578,7 +594,8 @@ void audit_trim_trees(void)
 		spin_lock(&hash_lock);
 		list_for_each_entry(node, &tree->chunks, list) {
 			struct audit_chunk *chunk = find_chunk(node);
-			struct inode *inode = chunk->watch.inode;
+			/* this could be NULL if the watch is dieing else where... */
+			struct inode *inode = chunk->mark.i.inode;
 			struct vfsmount *mnt;
 			node->index |= 1U<<31;
 			list_for_each_entry(mnt, &list, mnt_list) {
@@ -925,34 +942,41 @@ static void evict_chunk(struct audit_chunk *chunk)
 	mutex_unlock(&audit_filter_mutex);
 }
 
-static void handle_event(struct inotify_watch *watch, u32 wd, u32 mask,
-                         u32 cookie, const char *dname, struct inode *inode)
+static int audit_tree_handle_event(struct fsnotify_group *group, struct fsnotify_event *event)
 {
-	struct audit_chunk *chunk = container_of(watch, struct audit_chunk, watch);
+	BUG();
+	return -EOPNOTSUPP;
+}
 
-	if (mask & IN_IGNORED) {
-		evict_chunk(chunk);
-		put_inotify_watch(watch);
-	}
+static void audit_tree_freeing_mark(struct fsnotify_mark *entry, struct fsnotify_group *group)
+{
+	struct audit_chunk *chunk = container_of(entry, struct audit_chunk, mark);
+
+	evict_chunk(chunk);
+	fsnotify_put_mark(entry);
 }
 
-static void destroy_watch(struct inotify_watch *watch)
+static bool audit_tree_send_event(struct fsnotify_group *group, struct inode *inode,
+				  struct vfsmount *mnt, __u32 mask, void *data,
+				  int data_type)
 {
-	struct audit_chunk *chunk = container_of(watch, struct audit_chunk, watch);
-	call_rcu(&chunk->head, __put_chunk);
+	return 0;
 }
 
-static const struct inotify_operations rtree_inotify_ops = {
-	.handle_event	= handle_event,
-	.destroy_watch	= destroy_watch,
+static const struct fsnotify_ops audit_tree_ops = {
+	.handle_event = audit_tree_handle_event,
+	.should_send_event = audit_tree_send_event,
+	.free_group_priv = NULL,
+	.free_event_priv = NULL,
+	.freeing_mark = audit_tree_freeing_mark,
 };
 
 static int __init audit_tree_init(void)
 {
 	int i;
 
-	rtree_ih = inotify_init(&rtree_inotify_ops);
-	if (IS_ERR(rtree_ih))
+	audit_tree_group = fsnotify_alloc_group(&audit_tree_ops);
+	if (IS_ERR(audit_tree_group))
 		audit_panic("cannot initialize inotify handle for rectree watches");
 
 	for (i = 0; i < HASH_SIZE; i++)
diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c
index cc7e87936cbc..e8023ad8ecf9 100644
--- a/kernel/audit_watch.c
+++ b/kernel/audit_watch.c
@@ -24,17 +24,17 @@
 #include <linux/kthread.h>
 #include <linux/mutex.h>
 #include <linux/fs.h>
+#include <linux/fsnotify_backend.h>
 #include <linux/namei.h>
 #include <linux/netlink.h>
 #include <linux/sched.h>
-#include <linux/inotify.h>
 #include <linux/security.h>
 #include "audit.h"
 
 /*
  * Reference counting:
  *
- * audit_parent: lifetime is from audit_init_parent() to receipt of an IN_IGNORED
+ * audit_parent: lifetime is from audit_init_parent() to receipt of an FS_IGNORED
  * 	event.  Each audit_watch holds a reference to its associated parent.
  *
  * audit_watch: if added to lists, lifetime is from audit_init_watch() to
@@ -50,40 +50,61 @@ struct audit_watch {
 	unsigned long		ino;	/* associated inode number */
 	struct audit_parent	*parent; /* associated parent */
 	struct list_head	wlist;	/* entry in parent->watches list */
-	struct list_head	rules;	/* associated rules */
+	struct list_head	rules;	/* anchor for krule->rlist */
 };
 
 struct audit_parent {
-	struct list_head	ilist;	/* entry in inotify registration list */
-	struct list_head	watches; /* associated watches */
-	struct inotify_watch	wdata;	/* inotify watch data */
-	unsigned		flags;	/* status flags */
+	struct list_head	watches; /* anchor for audit_watch->wlist */
+	struct fsnotify_mark mark; /* fsnotify mark on the inode */
 };
 
-/* Inotify handle. */
-struct inotify_handle *audit_ih;
+/* fsnotify handle. */
+struct fsnotify_group *audit_watch_group;
 
-/*
- * audit_parent status flags:
- *
- * AUDIT_PARENT_INVALID - set anytime rules/watches are auto-removed due to
- * a filesystem event to ensure we're adding audit watches to a valid parent.
- * Technically not needed for IN_DELETE_SELF or IN_UNMOUNT events, as we cannot
- * receive them while we have nameidata, but must be used for IN_MOVE_SELF which
- * we can receive while holding nameidata.
- */
-#define AUDIT_PARENT_INVALID	0x001
+/* fsnotify events we care about. */
+#define AUDIT_FS_WATCH (FS_MOVE | FS_CREATE | FS_DELETE | FS_DELETE_SELF |\
+			FS_MOVE_SELF | FS_EVENT_ON_CHILD)
 
-/* Inotify events we care about. */
-#define AUDIT_IN_WATCH IN_MOVE|IN_CREATE|IN_DELETE|IN_DELETE_SELF|IN_MOVE_SELF
+static void audit_free_parent(struct audit_parent *parent)
+{
+	WARN_ON(!list_empty(&parent->watches));
+	kfree(parent);
+}
 
-static void audit_free_parent(struct inotify_watch *i_watch)
+static void audit_watch_free_mark(struct fsnotify_mark *entry)
 {
 	struct audit_parent *parent;
 
-	parent = container_of(i_watch, struct audit_parent, wdata);
-	WARN_ON(!list_empty(&parent->watches));
-	kfree(parent);
+	parent = container_of(entry, struct audit_parent, mark);
+	audit_free_parent(parent);
+}
+
+static void audit_get_parent(struct audit_parent *parent)
+{
+	if (likely(parent))
+		fsnotify_get_mark(&parent->mark);
+}
+
+static void audit_put_parent(struct audit_parent *parent)
+{
+	if (likely(parent))
+		fsnotify_put_mark(&parent->mark);
+}
+
+/*
+ * Find and return the audit_parent on the given inode.  If found a reference
+ * is taken on this parent.
+ */
+static inline struct audit_parent *audit_find_parent(struct inode *inode)
+{
+	struct audit_parent *parent = NULL;
+	struct fsnotify_mark *entry;
+
+	entry = fsnotify_find_mark(audit_watch_group, inode);
+	if (entry)
+		parent = container_of(entry, struct audit_parent, mark);
+
+	return parent;
 }
 
 void audit_get_watch(struct audit_watch *watch)
@@ -104,7 +125,7 @@ void audit_put_watch(struct audit_watch *watch)
 void audit_remove_watch(struct audit_watch *watch)
 {
 	list_del(&watch->wlist);
-	put_inotify_watch(&watch->parent->wdata);
+	audit_put_parent(watch->parent);
 	watch->parent = NULL;
 	audit_put_watch(watch); /* match initial get */
 }
@@ -114,44 +135,36 @@ char *audit_watch_path(struct audit_watch *watch)
 	return watch->path;
 }
 
-struct list_head *audit_watch_rules(struct audit_watch *watch)
+int audit_watch_compare(struct audit_watch *watch, unsigned long ino, dev_t dev)
 {
-	return &watch->rules;
-}
-
-unsigned long audit_watch_inode(struct audit_watch *watch)
-{
-	return watch->ino;
-}
-
-dev_t audit_watch_dev(struct audit_watch *watch)
-{
-	return watch->dev;
+	return (watch->ino != (unsigned long)-1) &&
+		(watch->ino == ino) &&
+		(watch->dev == dev);
 }
 
 /* Initialize a parent watch entry. */
 static struct audit_parent *audit_init_parent(struct nameidata *ndp)
 {
+	struct inode *inode = ndp->path.dentry->d_inode;
 	struct audit_parent *parent;
-	s32 wd;
+	int ret;
 
 	parent = kzalloc(sizeof(*parent), GFP_KERNEL);
 	if (unlikely(!parent))
 		return ERR_PTR(-ENOMEM);
 
 	INIT_LIST_HEAD(&parent->watches);
-	parent->flags = 0;
-
-	inotify_init_watch(&parent->wdata);
-	/* grab a ref so inotify watch hangs around until we take audit_filter_mutex */
-	get_inotify_watch(&parent->wdata);
-	wd = inotify_add_watch(audit_ih, &parent->wdata,
-			       ndp->path.dentry->d_inode, AUDIT_IN_WATCH);
-	if (wd < 0) {
-		audit_free_parent(&parent->wdata);
-		return ERR_PTR(wd);
+
+	fsnotify_init_mark(&parent->mark, audit_watch_free_mark);
+	parent->mark.mask = AUDIT_FS_WATCH;
+	ret = fsnotify_add_mark(&parent->mark, audit_watch_group, inode, 0);
+	if (ret < 0) {
+		audit_free_parent(parent);
+		return ERR_PTR(ret);
 	}
 
+	fsnotify_recalc_group_mask(audit_watch_group);
+
 	return parent;
 }
 
@@ -178,7 +191,7 @@ int audit_to_watch(struct audit_krule *krule, char *path, int len, u32 op)
 {
 	struct audit_watch *watch;
 
-	if (!audit_ih)
+	if (!audit_watch_group)
 		return -EOPNOTSUPP;
 
 	if (path[0] != '/' || path[len-1] == '/' ||
@@ -216,7 +229,7 @@ static struct audit_watch *audit_dupe_watch(struct audit_watch *old)
 
 	new->dev = old->dev;
 	new->ino = old->ino;
-	get_inotify_watch(&old->parent->wdata);
+	audit_get_parent(old->parent);
 	new->parent = old->parent;
 
 out:
@@ -250,15 +263,19 @@ static void audit_update_watch(struct audit_parent *parent,
 	struct audit_entry *oentry, *nentry;
 
 	mutex_lock(&audit_filter_mutex);
+	/* Run all of the watches on this parent looking for the one that
+	 * matches the given dname */
 	list_for_each_entry_safe(owatch, nextw, &parent->watches, wlist) {
 		if (audit_compare_dname_path(dname, owatch->path, NULL))
 			continue;
 
 		/* If the update involves invalidating rules, do the inode-based
 		 * filtering now, so we don't omit records. */
-		if (invalidating && current->audit_context)
+		if (invalidating && !audit_dummy_context())
 			audit_filter_inodes(current, current->audit_context);
 
+		/* updating ino will likely change which audit_hash_list we
+		 * are on so we need a new watch for the new list */
 		nwatch = audit_dupe_watch(owatch);
 		if (IS_ERR(nwatch)) {
 			mutex_unlock(&audit_filter_mutex);
@@ -274,12 +291,21 @@ static void audit_update_watch(struct audit_parent *parent,
 			list_del(&oentry->rule.rlist);
 			list_del_rcu(&oentry->list);
 
-			nentry = audit_dupe_rule(&oentry->rule, nwatch);
+			nentry = audit_dupe_rule(&oentry->rule);
 			if (IS_ERR(nentry)) {
 				list_del(&oentry->rule.list);
 				audit_panic("error updating watch, removing");
 			} else {
 				int h = audit_hash_ino((u32)ino);
+
+				/*
+				 * nentry->rule.watch == oentry->rule.watch so
+				 * we must drop that reference and set it to our
+				 * new watch.
+				 */
+				audit_put_watch(nentry->rule.watch);
+				audit_get_watch(nwatch);
+				nentry->rule.watch = nwatch;
 				list_add(&nentry->rule.rlist, &nwatch->rules);
 				list_add_rcu(&nentry->list, &audit_inode_hash[h]);
 				list_replace(&oentry->rule.list,
@@ -311,7 +337,6 @@ static void audit_remove_parent_watches(struct audit_parent *parent)
 	struct audit_entry *e;
 
 	mutex_lock(&audit_filter_mutex);
-	parent->flags |= AUDIT_PARENT_INVALID;
 	list_for_each_entry_safe(w, nextw, &parent->watches, wlist) {
 		list_for_each_entry_safe(r, nextr, &w->rules, rlist) {
 			e = container_of(r, struct audit_entry, rule);
@@ -324,20 +349,11 @@ static void audit_remove_parent_watches(struct audit_parent *parent)
 		audit_remove_watch(w);
 	}
 	mutex_unlock(&audit_filter_mutex);
-}
 
-/* Unregister inotify watches for parents on in_list.
- * Generates an IN_IGNORED event. */
-void audit_inotify_unregister(struct list_head *in_list)
-{
-	struct audit_parent *p, *n;
+	fsnotify_destroy_mark(&parent->mark);
+
+	fsnotify_recalc_group_mask(audit_watch_group);
 
-	list_for_each_entry_safe(p, n, in_list, ilist) {
-		list_del(&p->ilist);
-		inotify_rm_watch(audit_ih, &p->wdata);
-		/* the unpin matching the pin in audit_do_del_rule() */
-		unpin_inotify_watch(&p->wdata);
-	}
 }
 
 /* Get path information necessary for adding watches. */
@@ -388,7 +404,7 @@ static void audit_put_nd(struct nameidata *ndp, struct nameidata *ndw)
 	}
 }
 
-/* Associate the given rule with an existing parent inotify_watch.
+/* Associate the given rule with an existing parent.
  * Caller must hold audit_filter_mutex. */
 static void audit_add_to_parent(struct audit_krule *krule,
 				struct audit_parent *parent)
@@ -396,6 +412,8 @@ static void audit_add_to_parent(struct audit_krule *krule,
 	struct audit_watch *w, *watch = krule->watch;
 	int watch_found = 0;
 
+	BUG_ON(!mutex_is_locked(&audit_filter_mutex));
+
 	list_for_each_entry(w, &parent->watches, wlist) {
 		if (strcmp(watch->path, w->path))
 			continue;
@@ -412,7 +430,7 @@ static void audit_add_to_parent(struct audit_krule *krule,
 	}
 
 	if (!watch_found) {
-		get_inotify_watch(&parent->wdata);
+		audit_get_parent(parent);
 		watch->parent = parent;
 
 		list_add(&watch->wlist, &parent->watches);
@@ -422,13 +440,12 @@ static void audit_add_to_parent(struct audit_krule *krule,
 
 /* Find a matching watch entry, or add this one.
  * Caller must hold audit_filter_mutex. */
-int audit_add_watch(struct audit_krule *krule)
+int audit_add_watch(struct audit_krule *krule, struct list_head **list)
 {
 	struct audit_watch *watch = krule->watch;
-	struct inotify_watch *i_watch;
 	struct audit_parent *parent;
 	struct nameidata *ndp = NULL, *ndw = NULL;
-	int ret = 0;
+	int h, ret = 0;
 
 	mutex_unlock(&audit_filter_mutex);
 
@@ -440,47 +457,38 @@ int audit_add_watch(struct audit_krule *krule)
 		goto error;
 	}
 
+	mutex_lock(&audit_filter_mutex);
+
 	/* update watch filter fields */
 	if (ndw) {
 		watch->dev = ndw->path.dentry->d_inode->i_sb->s_dev;
 		watch->ino = ndw->path.dentry->d_inode->i_ino;
 	}
 
-	/* The audit_filter_mutex must not be held during inotify calls because
-	 * we hold it during inotify event callback processing.  If an existing
-	 * inotify watch is found, inotify_find_watch() grabs a reference before
-	 * returning.
-	 */
-	if (inotify_find_watch(audit_ih, ndp->path.dentry->d_inode,
-			       &i_watch) < 0) {
+	/* either find an old parent or attach a new one */
+	parent = audit_find_parent(ndp->path.dentry->d_inode);
+	if (!parent) {
 		parent = audit_init_parent(ndp);
 		if (IS_ERR(parent)) {
-			/* caller expects mutex locked */
-			mutex_lock(&audit_filter_mutex);
 			ret = PTR_ERR(parent);
 			goto error;
 		}
-	} else
-		parent = container_of(i_watch, struct audit_parent, wdata);
-
-	mutex_lock(&audit_filter_mutex);
+	}
 
-	/* parent was moved before we took audit_filter_mutex */
-	if (parent->flags & AUDIT_PARENT_INVALID)
-		ret = -ENOENT;
-	else
-		audit_add_to_parent(krule, parent);
+	audit_add_to_parent(krule, parent);
 
-	/* match get in audit_init_parent or inotify_find_watch */
-	put_inotify_watch(&parent->wdata);
+	/* match get in audit_find_parent or audit_init_parent */
+	audit_put_parent(parent);
 
+	h = audit_hash_ino((u32)watch->ino);
+	*list = &audit_inode_hash[h];
 error:
 	audit_put_nd(ndp, ndw);		/* NULL args OK */
 	return ret;
 
 }
 
-void audit_remove_watch_rule(struct audit_krule *krule, struct list_head *list)
+void audit_remove_watch_rule(struct audit_krule *krule)
 {
 	struct audit_watch *watch = krule->watch;
 	struct audit_parent *parent = watch->parent;
@@ -491,53 +499,92 @@ void audit_remove_watch_rule(struct audit_krule *krule, struct list_head *list)
 		audit_remove_watch(watch);
 
 		if (list_empty(&parent->watches)) {
-			/* Put parent on the inotify un-registration
-			 * list.  Grab a reference before releasing
-			 * audit_filter_mutex, to be released in
-			 * audit_inotify_unregister().
-			 * If filesystem is going away, just leave
-			 * the sucker alone, eviction will take
-			 * care of it. */
-			if (pin_inotify_watch(&parent->wdata))
-				list_add(&parent->ilist, list);
+			audit_get_parent(parent);
+			fsnotify_destroy_mark(&parent->mark);
+			audit_put_parent(parent);
 		}
 	}
+
+	fsnotify_recalc_group_mask(audit_watch_group);
+
 }
 
-/* Update watch data in audit rules based on inotify events. */
-static void audit_handle_ievent(struct inotify_watch *i_watch, u32 wd, u32 mask,
-			 u32 cookie, const char *dname, struct inode *inode)
+static bool audit_watch_should_send_event(struct fsnotify_group *group, struct inode *inode,
+					  struct vfsmount *mnt, __u32 mask, void *data,
+					  int data_type)
 {
+	struct fsnotify_mark *entry;
+	bool send;
+
+	entry = fsnotify_find_mark(group, inode);
+	if (!entry)
+		return false;
+
+	mask = (mask & ~FS_EVENT_ON_CHILD);
+	send = (entry->mask & mask);
+
+	/* find took a reference */
+	fsnotify_put_mark(entry);
+
+	return send;
+}
+
+/* Update watch data in audit rules based on fsnotify events. */
+static int audit_watch_handle_event(struct fsnotify_group *group, struct fsnotify_event *event)
+{
+	struct inode *inode;
+	__u32 mask = event->mask;
+	const char *dname = event->file_name;
 	struct audit_parent *parent;
 
-	parent = container_of(i_watch, struct audit_parent, wdata);
+	BUG_ON(group != audit_watch_group);
 
-	if (mask & (IN_CREATE|IN_MOVED_TO) && inode)
-		audit_update_watch(parent, dname, inode->i_sb->s_dev,
-				   inode->i_ino, 0);
-	else if (mask & (IN_DELETE|IN_MOVED_FROM))
+	parent = audit_find_parent(event->to_tell);
+	if (unlikely(!parent))
+		return 0;
+
+	switch (event->data_type) {
+	case (FSNOTIFY_EVENT_PATH):
+		inode = event->path.dentry->d_inode;
+		break;
+	case (FSNOTIFY_EVENT_INODE):
+		inode = event->inode;
+		break;
+	default:
+		BUG();
+		inode = NULL;
+		break;
+	};
+
+	if (mask & (FS_CREATE|FS_MOVED_TO) && inode)
+		audit_update_watch(parent, dname, inode->i_sb->s_dev, inode->i_ino, 0);
+	else if (mask & (FS_DELETE|FS_MOVED_FROM))
 		audit_update_watch(parent, dname, (dev_t)-1, (unsigned long)-1, 1);
-	/* inotify automatically removes the watch and sends IN_IGNORED */
-	else if (mask & (IN_DELETE_SELF|IN_UNMOUNT))
+	else if (mask & (FS_DELETE_SELF|FS_UNMOUNT|FS_MOVE_SELF))
 		audit_remove_parent_watches(parent);
-	/* inotify does not remove the watch, so remove it manually */
-	else if(mask & IN_MOVE_SELF) {
-		audit_remove_parent_watches(parent);
-		inotify_remove_watch_locked(audit_ih, i_watch);
-	} else if (mask & IN_IGNORED)
-		put_inotify_watch(i_watch);
+	/* moved put_inotify_watch to freeing mark */
+
+	/* matched the ref taken by audit_find_parent */
+	audit_put_parent(parent);
+
+	return 0;
 }
 
-static const struct inotify_operations audit_inotify_ops = {
-	.handle_event   = audit_handle_ievent,
-	.destroy_watch  = audit_free_parent,
+static const struct fsnotify_ops audit_watch_fsnotify_ops = {
+	.should_send_event = 	audit_watch_should_send_event,
+	.handle_event = 	audit_watch_handle_event,
+	.free_group_priv = 	NULL,
+	.freeing_mark = 	NULL,
+	.free_event_priv = 	NULL,
 };
 
 static int __init audit_watch_init(void)
 {
-	audit_ih = inotify_init(&audit_inotify_ops);
-	if (IS_ERR(audit_ih))
-		audit_panic("cannot initialize inotify handle");
+	audit_watch_group = fsnotify_alloc_group(&audit_watch_fsnotify_ops);
+	if (IS_ERR(audit_watch_group)) {
+		audit_watch_group = NULL;
+		audit_panic("cannot create audit fsnotify group");
+	}
 	return 0;
 }
-subsys_initcall(audit_watch_init);
+device_initcall(audit_watch_init);
diff --git a/kernel/auditfilter.c b/kernel/auditfilter.c
index a70604047f3c..f5e4cae5ad82 100644
--- a/kernel/auditfilter.c
+++ b/kernel/auditfilter.c
@@ -70,6 +70,7 @@ static inline void audit_free_rule(struct audit_entry *e)
 {
 	int i;
 	struct audit_krule *erule = &e->rule;
+
 	/* some rules don't have associated watches */
 	if (erule->watch)
 		audit_put_watch(erule->watch);
@@ -745,8 +746,7 @@ static inline int audit_dupe_lsm_field(struct audit_field *df,
  * rule with the new rule in the filterlist, then free the old rule.
  * The rlist element is undefined; list manipulations are handled apart from
  * the initial copy. */
-struct audit_entry *audit_dupe_rule(struct audit_krule *old,
-				    struct audit_watch *watch)
+struct audit_entry *audit_dupe_rule(struct audit_krule *old)
 {
 	u32 fcount = old->field_count;
 	struct audit_entry *entry;
@@ -768,8 +768,8 @@ struct audit_entry *audit_dupe_rule(struct audit_krule *old,
 	new->prio = old->prio;
 	new->buflen = old->buflen;
 	new->inode_f = old->inode_f;
-	new->watch = NULL;
 	new->field_count = old->field_count;
+
 	/*
 	 * note that we are OK with not refcounting here; audit_match_tree()
 	 * never dereferences tree and we can't get false positives there
@@ -810,9 +810,9 @@ struct audit_entry *audit_dupe_rule(struct audit_krule *old,
 		}
 	}
 
-	if (watch) {
-		audit_get_watch(watch);
-		new->watch = watch;
+	if (old->watch) {
+		audit_get_watch(old->watch);
+		new->watch = old->watch;
 	}
 
 	return entry;
@@ -865,7 +865,7 @@ static inline int audit_add_rule(struct audit_entry *entry)
 	struct audit_watch *watch = entry->rule.watch;
 	struct audit_tree *tree = entry->rule.tree;
 	struct list_head *list;
-	int h, err;
+	int err;
 #ifdef CONFIG_AUDITSYSCALL
 	int dont_count = 0;
 
@@ -888,15 +888,11 @@ static inline int audit_add_rule(struct audit_entry *entry)
 
 	if (watch) {
 		/* audit_filter_mutex is dropped and re-taken during this call */
-		err = audit_add_watch(&entry->rule);
+		err = audit_add_watch(&entry->rule, &list);
 		if (err) {
 			mutex_unlock(&audit_filter_mutex);
 			goto error;
 		}
-		/* entry->rule.watch may have changed during audit_add_watch() */
-		watch = entry->rule.watch;
-		h = audit_hash_ino((u32)audit_watch_inode(watch));
-		list = &audit_inode_hash[h];
 	}
 	if (tree) {
 		err = audit_add_tree_rule(&entry->rule);
@@ -948,7 +944,6 @@ static inline int audit_del_rule(struct audit_entry *entry)
 	struct audit_watch *watch = entry->rule.watch;
 	struct audit_tree *tree = entry->rule.tree;
 	struct list_head *list;
-	LIST_HEAD(inotify_list);
 	int ret = 0;
 #ifdef CONFIG_AUDITSYSCALL
 	int dont_count = 0;
@@ -968,7 +963,7 @@ static inline int audit_del_rule(struct audit_entry *entry)
 	}
 
 	if (e->rule.watch)
-		audit_remove_watch_rule(&e->rule, &inotify_list);
+		audit_remove_watch_rule(&e->rule);
 
 	if (e->rule.tree)
 		audit_remove_tree_rule(&e->rule);
@@ -986,9 +981,6 @@ static inline int audit_del_rule(struct audit_entry *entry)
 #endif
 	mutex_unlock(&audit_filter_mutex);
 
-	if (!list_empty(&inotify_list))
-		audit_inotify_unregister(&inotify_list);
-
 out:
 	if (watch)
 		audit_put_watch(watch); /* match initial get */
@@ -1322,30 +1314,23 @@ static int update_lsm_rule(struct audit_krule *r)
 {
 	struct audit_entry *entry = container_of(r, struct audit_entry, rule);
 	struct audit_entry *nentry;
-	struct audit_watch *watch;
-	struct audit_tree *tree;
 	int err = 0;
 
 	if (!security_audit_rule_known(r))
 		return 0;
 
-	watch = r->watch;
-	tree = r->tree;
-	nentry = audit_dupe_rule(r, watch);
+	nentry = audit_dupe_rule(r);
 	if (IS_ERR(nentry)) {
 		/* save the first error encountered for the
 		 * return value */
 		err = PTR_ERR(nentry);
 		audit_panic("error updating LSM filters");
-		if (watch)
+		if (r->watch)
 			list_del(&r->rlist);
 		list_del_rcu(&entry->list);
 		list_del(&r->list);
 	} else {
-		if (watch) {
-			list_add(&nentry->rule.rlist, audit_watch_rules(watch));
-			list_del(&r->rlist);
-		} else if (tree)
+		if (r->watch || r->tree)
 			list_replace_init(&r->rlist, &nentry->rule.rlist);
 		list_replace_rcu(&entry->list, &nentry->list);
 		list_replace(&r->list, &nentry->rule.list);
diff --git a/kernel/auditsc.c b/kernel/auditsc.c
index 267e484f0198..880120c98332 100644
--- a/kernel/auditsc.c
+++ b/kernel/auditsc.c
@@ -549,9 +549,8 @@ static int audit_filter_rules(struct task_struct *tsk,
 			}
 			break;
 		case AUDIT_WATCH:
-			if (name && audit_watch_inode(rule->watch) != (unsigned long)-1)
-				result = (name->dev == audit_watch_dev(rule->watch) &&
-					  name->ino == audit_watch_inode(rule->watch));
+			if (name)
+				result = audit_watch_compare(rule->watch, name->ino, name->dev);
 			break;
 		case AUDIT_DIR:
 			if (ctx)
@@ -1726,7 +1725,7 @@ static inline void handle_one(const struct inode *inode)
 	struct audit_tree_refs *p;
 	struct audit_chunk *chunk;
 	int count;
-	if (likely(list_empty(&inode->inotify_watches)))
+	if (likely(hlist_empty(&inode->i_fsnotify_marks)))
 		return;
 	context = current->audit_context;
 	p = context->trees;
@@ -1769,7 +1768,7 @@ retry:
 	seq = read_seqbegin(&rename_lock);
 	for(;;) {
 		struct inode *inode = d->d_inode;
-		if (inode && unlikely(!list_empty(&inode->inotify_watches))) {
+		if (inode && unlikely(!hlist_empty(&inode->i_fsnotify_marks))) {
 			struct audit_chunk *chunk;
 			chunk = audit_tree_lookup(inode);
 			if (chunk) {