Merge branch 'for-linus' of git://git.infradead.org/users/eparis/notify

* 'for-linus' of git://git.infradead.org/users/eparis/notify: (132 commits)
  fanotify: use both marks when possible
  fsnotify: pass both the vfsmount mark and inode mark
  fsnotify: walk the inode and vfsmount lists simultaneously
  fsnotify: rework ignored mark flushing
  fsnotify: remove global fsnotify groups lists
  fsnotify: remove group->mask
  fsnotify: remove the global masks
  fsnotify: cleanup should_send_event
  fanotify: use the mark in handler functions
  audit: use the mark in handler functions
  dnotify: use the mark in handler functions
  inotify: use the mark in handler functions
  fsnotify: send fsnotify_mark to groups in event handling functions
  fsnotify: Exchange list heads instead of moving elements
  fsnotify: srcu to protect read side of inode and vfsmount locks
  fsnotify: use an explicit flag to indicate fsnotify_destroy_mark has been called
  fsnotify: use _rcu functions for mark list traversal
  fsnotify: place marks on object in order of group memory address
  vfs/fsnotify: fsnotify_close can delay the final work in fput
  fsnotify: store struct file not struct path
  ...

Fix up trivial delete/modify conflict in fs/notify/inotify/inotify.c.

9 files changed, 325 insertions, 278 deletions

diff --git a/kernel/Makefile b/kernel/Makefile
index c53e491..0b72d1a 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -70,10 +70,11 @@ obj-$(CONFIG_IKCONFIG) += configs.o
 obj-$(CONFIG_RESOURCE_COUNTERS) += res_counter.o
 obj-$(CONFIG_SMP) += stop_machine.o
 obj-$(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-$(CONFIG_KGDB) += debug/
 obj-$(CONFIG_DETECT_HUNG_TASK) += hung_task.o
diff --git a/kernel/audit.c b/kernel/audit.c
index 8296aa5..d960457 100644
--- a/kernel/audit.c
+++ b/kernel/audit.c
@@ -56,7 +56,6 @@
 #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 208687b..f7206db 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 46a57b5..7f18d3a 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>
@@ -22,7 +22,7 @@ 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;
@@ -59,7 +59,7 @@ static LIST_HEAD(prune_list);
  * tree is refcounted; one reference for "some rules on rules_list refer to
  * it", one for each chunk with pointer to it.
  *
- * chunk is refcounted by embedded inotify_watch + .refs (non-zero refcount
+ * chunk is refcounted by embedded fsnotify_mark + .refs (non-zero refcount
  * of watch contributes 1 to .refs).
  *
  * node.index allows to get from node.list to containing chunk.
@@ -68,7 +68,7 @@ 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)
 {
@@ -111,29 +111,6 @@ 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;
@@ -157,6 +134,35 @@ 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);
@@ -167,10 +173,15 @@ 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);
 }
 
@@ -181,7 +192,8 @@ 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;
 		}
@@ -210,38 +222,19 @@ 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;
 	}
 
@@ -256,16 +249,17 @@ 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, NULL, 1)) {
 		free_chunk(new);
 		goto Fallback;
 	}
@@ -298,9 +292,9 @@ 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:
@@ -314,31 +308,33 @@ 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, NULL, 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);
@@ -351,30 +347,31 @@ 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_inode_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(&old->watch);
+			fsnotify_put_mark(old_entry);
 			return 0;
 		}
 	}
@@ -382,25 +379,44 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree)
 
 	chunk = alloc_chunk(old->count + 1);
 	if (!chunk) {
-		put_inotify_watch(&old->watch);
+		fsnotify_put_mark(old_entry);
 		return -ENOMEM;
 	}
 
-	mutex_lock(&inode->inotify_mutex);
-	if (inotify_clone_watch(&old->watch, &chunk->watch) < 0) {
-		mutex_unlock(&inode->inotify_mutex);
-		put_inotify_watch(&old->watch);
+	chunk_entry = &chunk->mark;
+
+	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, NULL, 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);
@@ -426,10 +442,11 @@ 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); /* pair to inotify_find_watch */
-	put_inotify_watch(&old->watch); /* and kill it */
+	spin_unlock(&chunk_entry->lock);
+	spin_unlock(&old_entry->lock);
+	fsnotify_destroy_mark(old_entry);
+	fsnotify_put_mark(old_entry); /* pair to fsnotify_find mark_entry */
+	fsnotify_put_mark(old_entry); /* and kill it */
 	return 0;
 }
 
@@ -584,7 +601,9 @@ void audit_trim_trees(void)
 
 		spin_lock(&hash_lock);
 		list_for_each_entry(node, &tree->chunks, list) {
-			struct inode *inode = find_chunk(node)->watch.inode;
+			struct audit_chunk *chunk = find_chunk(node);
+			/* this could be NULL if the watch is dieing else where... */
+			struct inode *inode = chunk->mark.i.inode;
 			node->index |= 1U<<31;
 			if (iterate_mounts(compare_root, inode, root_mnt))
 				node->index &= ~(1U<<31);
@@ -846,7 +865,6 @@ void audit_kill_trees(struct list_head *list)
  *  Here comes the stuff asynchronous to auditctl operations
  */
 
-/* inode->inotify_mutex is locked */
 static void evict_chunk(struct audit_chunk *chunk)
 {
 	struct audit_tree *owner;
@@ -885,35 +903,46 @@ 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_mark *inode_mark,
+				   struct fsnotify_mark *vfsmonut_mark,
+				   struct fsnotify_event *event)
+{
+	BUG();
+	return -EOPNOTSUPP;
+}
+
+static void audit_tree_freeing_mark(struct fsnotify_mark *entry, struct fsnotify_group *group)
 {
-	struct audit_chunk *chunk = container_of(watch, struct audit_chunk, watch);
+	struct audit_chunk *chunk = container_of(entry, struct audit_chunk, mark);
 
-	if (mask & IN_IGNORED) {
-		evict_chunk(chunk);
-		put_inotify_watch(watch);
-	}
+	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 fsnotify_mark *inode_mark,
+				  struct fsnotify_mark *vfsmount_mark,
+				  __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 false;
 }
 
-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_panic("cannot initialize inotify handle for rectree watches");
+	audit_tree_group = fsnotify_alloc_group(&audit_tree_ops);
+	if (IS_ERR(audit_tree_group))
+		audit_panic("cannot initialize fsnotify group for rectree watches");
 
 	for (i = 0; i < HASH_SIZE; i++)
 		INIT_LIST_HEAD(&chunk_hash_heads[i]);
diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c
index 8df4369..6bf2306 100644
--- a/kernel/audit_watch.c
+++ b/kernel/audit_watch.c
@@ -24,18 +24,18 @@
 #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/slab.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
@@ -51,40 +51,61 @@ 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_inode_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)
@@ -105,7 +126,7 @@ 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 */
 }
@@ -115,42 +136,32 @@ char *audit_watch_path(struct audit_watch *watch)
 	return watch->path;
 }
 
-struct list_head *audit_watch_rules(struct audit_watch *watch)
-{
-	return &watch->rules;
-}
-
-unsigned long audit_watch_inode(struct audit_watch *watch)
+int audit_watch_compare(struct audit_watch *watch, unsigned long ino, dev_t dev)
 {
-	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, NULL, 0);
+	if (ret < 0) {
+		audit_free_parent(parent);
+		return ERR_PTR(ret);
 	}
 
 	return parent;
@@ -179,7 +190,7 @@ 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] == '/' ||
@@ -217,7 +228,7 @@ 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:
@@ -251,15 +262,19 @@ 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);
@@ -275,12 +290,21 @@ 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,
@@ -312,7 +336,6 @@ 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);
@@ -325,20 +348,8 @@ 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;
 
-	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);
-	}
+	fsnotify_destroy_mark(&parent->mark);
 }
 
 /* Get path information necessary for adding watches. */
@@ -389,7 +400,7 @@ 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)
@@ -397,6 +408,8 @@ 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;
@@ -413,7 +426,7 @@ 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);
@@ -423,13 +436,12 @@ 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);
 
@@ -441,47 +453,38 @@ 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;
@@ -492,53 +495,74 @@ 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);
 		}
 	}
 }
 
-/* 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 fsnotify_mark *inode_mark,
+					  struct fsnotify_mark *vfsmount_mark,
+					  __u32 mask, void *data, int data_type)
+{
+       return true;
+}
+
+/* Update watch data in audit rules based on fsnotify events. */
+static int audit_watch_handle_event(struct fsnotify_group *group,
+				    struct fsnotify_mark *inode_mark,
+				    struct fsnotify_mark *vfsmount_mark,
+				    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);
+	parent = container_of(inode_mark, struct audit_parent, mark);
 
-	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))
+	BUG_ON(group != audit_watch_group);
+
+	switch (event->data_type) {
+	case (FSNOTIFY_EVENT_FILE):
+		inode = event->file->f_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))
-		audit_remove_parent_watches(parent);
-	/* inotify does not remove the watch, so remove it manually */
-	else if(mask & IN_MOVE_SELF) {
+	else if (mask & (FS_DELETE_SELF|FS_UNMOUNT|FS_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);
+
+	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 ce08041..eb76754 100644
--- a/kernel/auditfilter.c
+++ b/kernel/auditfilter.c
@@ -71,6 +71,7 @@ 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);
@@ -746,8 +747,7 @@ 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;
@@ -769,8 +769,8 @@ 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
@@ -811,9 +811,9 @@ 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;
@@ -866,7 +866,7 @@ 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;
 
@@ -889,15 +889,11 @@ 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);
@@ -949,7 +945,6 @@ 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;
@@ -969,7 +964,7 @@ 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);
@@ -987,9 +982,6 @@ 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 */
@@ -1323,30 +1315,23 @@ 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 3828ad5..b87a63b 100644
--- a/kernel/auditsc.c
+++ b/kernel/auditsc.c
@@ -65,7 +65,6 @@
 #include <linux/binfmts.h>
 #include <linux/highmem.h>
 #include <linux/syscalls.h>
-#include <linux/inotify.h>
 #include <linux/capability.h>
 #include <linux/fs_struct.h>
 
@@ -549,9 +548,8 @@ 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 +1724,7 @@ 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 +1767,7 @@ 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) {
diff --git a/kernel/sys_ni.c b/kernel/sys_ni.c
index 70f2ea7..bad369e 100644
--- a/kernel/sys_ni.c
+++ b/kernel/sys_ni.c
@@ -181,3 +181,7 @@ cond_syscall(sys_eventfd2);
 
 /* performance counters: */
 cond_syscall(sys_perf_event_open);
+
+/* fanotify! */
+cond_syscall(sys_fanotify_init);
+cond_syscall(sys_fanotify_mark);
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 6b005e4..ca38e8e 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -44,6 +44,7 @@
 #include <linux/times.h>
 #include <linux/limits.h>
 #include <linux/dcache.h>
+#include <linux/dnotify.h>
 #include <linux/syscalls.h>
 #include <linux/vmstat.h>
 #include <linux/nfs_fs.h>
@@ -131,6 +132,9 @@ static int min_percpu_pagelist_fract = 8;
 
 static int ngroups_max = NGROUPS_MAX;
 
+#ifdef CONFIG_INOTIFY_USER
+#include <linux/inotify.h>
+#endif
 #ifdef CONFIG_SPARC
 #include <asm/system.h>
 #endif
@@ -207,9 +211,6 @@ static struct ctl_table fs_table[];
 static struct ctl_table debug_table[];
 static struct ctl_table dev_table[];
 extern struct ctl_table random_table[];
-#ifdef CONFIG_INOTIFY_USER
-extern struct ctl_table inotify_table[];
-#endif
 #ifdef CONFIG_EPOLL
 extern struct ctl_table epoll_table[];
 #endif