fsnotify: do not share events between notification groups

Currently fsnotify framework creates one event structure for each
notification event and links this event into all interested notification
groups.  This is done so that we save memory when several notification
groups are interested in the event.  However the need for event
structure shared between inotify & fanotify bloats the event structure
so the result is often higher memory consumption.

Another problem is that fsnotify framework keeps path references with
outstanding events so that fanotify can return open file descriptors
with its events.  This has the undesirable effect that filesystem cannot
be unmounted while there are outstanding events - a regression for
inotify compared to a situation before it was converted to fsnotify
framework.  For fanotify this problem is hard to avoid and users of
fanotify should kind of expect this behavior when they ask for file
descriptors from notified files.

This patch changes fsnotify and its users to create separate event
structure for each group.  This allows for much simpler code (~400 lines
removed by this patch) and also smaller event structures.  For example
on 64-bit system original struct fsnotify_event consumes 120 bytes, plus
additional space for file name, additional 24 bytes for second and each
subsequent group linking the event, and additional 32 bytes for each
inotify group for private data.  After the conversion inotify event
consumes 48 bytes plus space for file name which is considerably less
memory unless file names are long and there are several groups
interested in the events (both of which are uncommon).  Fanotify event
fits in 56 bytes after the conversion (fanotify doesn't care about file
names so its events don't have to have it allocated).  A win unless
there are four or more fanotify groups interested in the event.

The conversion also solves the problem with unmount when only inotify is
used as we don't have to grab path references for inotify events.

[hughd@google.com: fanotify: fix corruption preventing startup]
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Eric Paris <eparis@parisplace.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

3 files changed, 86 insertions, 146 deletions

diff --git a/fs/notify/inotify/inotify.h b/fs/notify/inotify/inotify.h
index b6642e4..485eef3 100644
--- a/fs/notify/inotify/inotify.h
+++ b/fs/notify/inotify/inotify.h
@@ -2,11 +2,12 @@
 #include <linux/inotify.h>
 #include <linux/slab.h> /* struct kmem_cache */
 
-extern struct kmem_cache *event_priv_cachep;
-
-struct inotify_event_private_data {
-	struct fsnotify_event_private_data fsnotify_event_priv_data;
+struct inotify_event_info {
+	struct fsnotify_event fse;
 	int wd;
+	u32 sync_cookie;
+	int name_len;
+	char name[];
 };
 
 struct inotify_inode_mark {
@@ -14,8 +15,18 @@ struct inotify_inode_mark {
 	int wd;
 };
 
+static inline struct inotify_event_info *INOTIFY_E(struct fsnotify_event *fse)
+{
+	return container_of(fse, struct inotify_event_info, fse);
+}
+
 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);
+extern int inotify_handle_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,
+				const unsigned char *file_name);
 
 extern const struct fsnotify_ops inotify_fsnotify_ops;
diff --git a/fs/notify/inotify/inotify_fsnotify.c b/fs/notify/inotify/inotify_fsnotify.c
index 4216308..6fabbd1 100644
--- a/fs/notify/inotify/inotify_fsnotify.c
+++ b/fs/notify/inotify/inotify_fsnotify.c
@@ -34,100 +34,80 @@
 #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.
+ * Check if 2 events contain the same information.
  */
-static bool event_compare(struct fsnotify_event *old, struct fsnotify_event *new)
+static bool event_compare(struct fsnotify_event *old_fsn,
+			  struct fsnotify_event *new_fsn)
 {
-	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;
-		};
-	}
+	struct inotify_event_info *old, *new;
+
+	if (old_fsn->mask & FS_IN_IGNORED)
+		return false;
+	old = INOTIFY_E(old_fsn);
+	new = INOTIFY_E(new_fsn);
+	if ((old_fsn->mask == new_fsn->mask) &&
+	    (old_fsn->inode == new_fsn->inode) &&
+	    (old->name_len == new->name_len) &&
+	    (!old->name_len || !strcmp(old->name, new->name)))
+		return true;
 	return false;
 }
 
 static struct fsnotify_event *inotify_merge(struct list_head *list,
 					    struct fsnotify_event *event)
 {
-	struct fsnotify_event_holder *last_holder;
 	struct fsnotify_event *last_event;
 
-	/* 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))
-		fsnotify_get_event(last_event);
-	else
-		last_event = NULL;
-
-	spin_unlock(&event->lock);
-
+	last_event = list_entry(list->prev, struct fsnotify_event, list);
+	if (!event_compare(last_event, event))
+		return NULL;
 	return last_event;
 }
 
-static int inotify_handle_event(struct fsnotify_group *group,
-				struct fsnotify_mark *inode_mark,
-				struct fsnotify_mark *vfsmount_mark,
-				struct fsnotify_event *event)
+int inotify_handle_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,
+			 const unsigned char *file_name)
 {
 	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;
+	struct inotify_event_info *event;
 	struct fsnotify_event *added_event;
-	int wd, ret = 0;
+	struct fsnotify_event *fsn_event;
+	int ret = 0;
+	int len = 0;
+	int alloc_len = sizeof(struct inotify_event_info);
 
 	BUG_ON(vfsmount_mark);
 
-	pr_debug("%s: group=%p event=%p to_tell=%p mask=%x\n", __func__, group,
-		 event, event->to_tell, event->mask);
+	if (file_name) {
+		len = strlen(file_name);
+		alloc_len += len + 1;
+	}
 
-	to_tell = event->to_tell;
+	pr_debug("%s: group=%p inode=%p mask=%x\n", __func__, group, inode,
+		 mask);
 
 	i_mark = container_of(inode_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))
+	event = kmalloc(alloc_len, GFP_KERNEL);
+	if (unlikely(!event))
 		return -ENOMEM;
 
-	fsn_event_priv = &event_priv->fsnotify_event_priv_data;
+	fsn_event = &event->fse;
+	fsnotify_init_event(fsn_event, inode, mask);
+	event->wd = i_mark->wd;
+	event->name_len = len;
+	if (len)
+		strcpy(event->name, file_name);
 
-	fsnotify_get_group(group);
-	fsn_event_priv->group = group;
-	event_priv->wd = wd;
-
-	added_event = fsnotify_add_notify_event(group, event, fsn_event_priv, inotify_merge);
+	added_event = fsnotify_add_notify_event(group, fsn_event, inotify_merge);
 	if (added_event) {
-		inotify_free_event_priv(fsn_event_priv);
-		if (!IS_ERR(added_event))
-			fsnotify_put_event(added_event);
-		else
+		/* Our event wasn't used in the end. Free it. */
+		fsnotify_destroy_event(group, fsn_event);
+		if (IS_ERR(added_event))
 			ret = PTR_ERR(added_event);
 	}
 
@@ -202,22 +182,15 @@ static void inotify_free_group_priv(struct fsnotify_group *group)
 	free_uid(group->inotify_data.user);
 }
 
-void inotify_free_event_priv(struct fsnotify_event_private_data *fsn_event_priv)
+static void inotify_free_event(struct fsnotify_event *fsn_event)
 {
-	struct inotify_event_private_data *event_priv;
-
-
-	event_priv = container_of(fsn_event_priv, struct inotify_event_private_data,
-				  fsnotify_event_priv_data);
-
-	fsnotify_put_group(fsn_event_priv->group);
-	kmem_cache_free(event_priv_cachep, event_priv);
+	kfree(INOTIFY_E(fsn_event));
 }
 
 const struct fsnotify_ops inotify_fsnotify_ops = {
 	.handle_event = inotify_handle_event,
 	.should_send_event = inotify_should_send_event,
 	.free_group_priv = inotify_free_group_priv,
-	.free_event_priv = inotify_free_event_priv,
+	.free_event = inotify_free_event,
 	.freeing_mark = inotify_freeing_mark,
 };
diff --git a/fs/notify/inotify/inotify_user.c b/fs/notify/inotify/inotify_user.c
index 1bb6dc8..497395c 100644
--- a/fs/notify/inotify/inotify_user.c
+++ b/fs/notify/inotify/inotify_user.c
@@ -50,7 +50,6 @@ static int inotify_max_queued_events __read_mostly;
 static 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;
 
 #ifdef CONFIG_SYSCTL
 
@@ -124,8 +123,11 @@ static unsigned int inotify_poll(struct file *file, poll_table *wait)
 	return ret;
 }
 
-static int round_event_name_len(struct fsnotify_event *event)
+static int round_event_name_len(struct fsnotify_event *fsn_event)
 {
+	struct inotify_event_info *event;
+
+	event = INOTIFY_E(fsn_event);
 	if (!event->name_len)
 		return 0;
 	return roundup(event->name_len + 1, sizeof(struct inotify_event));
@@ -169,40 +171,27 @@ static struct fsnotify_event *get_one_event(struct fsnotify_group *group,
  * buffer we had in "get_one_event()" above.
  */
 static ssize_t copy_event_to_user(struct fsnotify_group *group,
-				  struct fsnotify_event *event,
+				  struct fsnotify_event *fsn_event,
 				  char __user *buf)
 {
 	struct inotify_event inotify_event;
-	struct fsnotify_event_private_data *fsn_priv;
-	struct inotify_event_private_data *priv;
+	struct inotify_event_info *event;
 	size_t event_size = sizeof(struct inotify_event);
 	size_t name_len;
 	size_t pad_name_len;
 
-	pr_debug("%s: group=%p event=%p\n", __func__, group, event);
-
-	/* we get the inotify watch descriptor from the event private data */
-	spin_lock(&event->lock);
-	fsn_priv = fsnotify_remove_priv_from_event(group, event);
-	spin_unlock(&event->lock);
-
-	if (!fsn_priv)
-		inotify_event.wd = -1;
-	else {
-		priv = container_of(fsn_priv, struct inotify_event_private_data,
-				    fsnotify_event_priv_data);
-		inotify_event.wd = priv->wd;
-		inotify_free_event_priv(fsn_priv);
-	}
+	pr_debug("%s: group=%p event=%p\n", __func__, group, fsn_event);
 
+	event = INOTIFY_E(fsn_event);
 	name_len = event->name_len;
 	/*
 	 * round up name length so it is a multiple of event_size
 	 * plus an extra byte for the terminating '\0'.
 	 */
-	pad_name_len = round_event_name_len(event);
+	pad_name_len = round_event_name_len(fsn_event);
 	inotify_event.len = pad_name_len;
-	inotify_event.mask = inotify_mask_to_arg(event->mask);
+	inotify_event.mask = inotify_mask_to_arg(fsn_event->mask);
+	inotify_event.wd = event->wd;
 	inotify_event.cookie = event->sync_cookie;
 
 	/* send the main event */
@@ -218,7 +207,7 @@ static ssize_t copy_event_to_user(struct fsnotify_group *group,
 	 */
 	if (pad_name_len) {
 		/* copy the path name */
-		if (copy_to_user(buf, event->file_name, name_len))
+		if (copy_to_user(buf, event->name, name_len))
 			return -EFAULT;
 		buf += name_len;
 
@@ -257,7 +246,7 @@ static ssize_t inotify_read(struct file *file, char __user *buf,
 			if (IS_ERR(kevent))
 				break;
 			ret = copy_event_to_user(group, kevent, buf);
-			fsnotify_put_event(kevent);
+			fsnotify_destroy_event(group, kevent);
 			if (ret < 0)
 				break;
 			buf += ret;
@@ -300,8 +289,7 @@ static long inotify_ioctl(struct file *file, unsigned int cmd,
 			  unsigned long arg)
 {
 	struct fsnotify_group *group;
-	struct fsnotify_event_holder *holder;
-	struct fsnotify_event *event;
+	struct fsnotify_event *fsn_event;
 	void __user *p;
 	int ret = -ENOTTY;
 	size_t send_len = 0;
@@ -314,10 +302,10 @@ static long inotify_ioctl(struct file *file, unsigned int cmd,
 	switch (cmd) {
 	case FIONREAD:
 		mutex_lock(&group->notification_mutex);
-		list_for_each_entry(holder, &group->notification_list, event_list) {
-			event = holder->event;
+		list_for_each_entry(fsn_event, &group->notification_list,
+				    list) {
 			send_len += sizeof(struct inotify_event);
-			send_len += round_event_name_len(event);
+			send_len += round_event_name_len(fsn_event);
 		}
 		mutex_unlock(&group->notification_mutex);
 		ret = put_user(send_len, (int __user *) p);
@@ -504,43 +492,12 @@ void inotify_ignored_and_remove_idr(struct fsnotify_mark *fsn_mark,
 				    struct fsnotify_group *group)
 {
 	struct inotify_inode_mark *i_mark;
-	struct fsnotify_event *ignored_event, *notify_event;
-	struct inotify_event_private_data *event_priv;
-	struct fsnotify_event_private_data *fsn_event_priv;
-	int ret;
 
-	i_mark = container_of(fsn_mark, struct inotify_inode_mark, fsn_mark);
-
-	ignored_event = fsnotify_create_event(NULL, FS_IN_IGNORED, NULL,
-					      FSNOTIFY_EVENT_NONE, NULL, 0,
-					      GFP_NOFS);
-	if (!ignored_event)
-		goto skip_send_ignore;
-
-	event_priv = kmem_cache_alloc(event_priv_cachep, GFP_NOFS);
-	if (unlikely(!event_priv))
-		goto skip_send_ignore;
-
-	fsn_event_priv = &event_priv->fsnotify_event_priv_data;
-
-	fsnotify_get_group(group);
-	fsn_event_priv->group = group;
-	event_priv->wd = i_mark->wd;
-
-	notify_event = fsnotify_add_notify_event(group, ignored_event, fsn_event_priv, NULL);
-	if (notify_event) {
-		if (IS_ERR(notify_event))
-			ret = PTR_ERR(notify_event);
-		else
-			fsnotify_put_event(notify_event);
-		inotify_free_event_priv(fsn_event_priv);
-	}
-
-skip_send_ignore:
-	/* matches the reference taken when the event was created */
-	if (ignored_event)
-		fsnotify_put_event(ignored_event);
+	/* Queue ignore event for the watch */
+	inotify_handle_event(group, NULL, fsn_mark, NULL, FS_IN_IGNORED,
+			     NULL, FSNOTIFY_EVENT_NONE, NULL);
 
+	i_mark = container_of(fsn_mark, struct inotify_inode_mark, fsn_mark);
 	/* remove this mark from the idr */
 	inotify_remove_from_idr(group, i_mark);
 
@@ -837,7 +794,6 @@ static int __init inotify_user_setup(void)
 	BUG_ON(hweight32(ALL_INOTIFY_BITS) != 21);
 
 	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;
 	inotify_max_user_instances = 128;