[linux.git] / fs / notify / fsnotify.c

/*
 *  Copyright (C) 2008 Red Hat, Inc., Eric Paris <[email protected]>
 *
 *  This program is free software; you can redistribute it and/or modify
 *  it under the terms of the GNU General Public License as published by
 *  the Free Software Foundation; either version 2, or (at your option)
 *  any later version.
 *
 *  This program is distributed in the hope that it will be useful,
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *  GNU General Public License for more details.
 *
 *  You should have received a copy of the GNU General Public License
 *  along with this program; see the file COPYING.  If not, write to
 *  the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
 */

#include <linux/dcache.h>
#include <linux/fs.h>
#include <linux/gfp.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/mount.h>
#include <linux/srcu.h>

#include <linux/fsnotify_backend.h>
#include "fsnotify.h"

/*
 * Clear all of the marks on an inode when it is being evicted from core
 */
void __fsnotify_inode_delete(struct inode *inode)
{
	fsnotify_clear_marks_by_inode(inode);
}
EXPORT_SYMBOL_GPL(__fsnotify_inode_delete);

void __fsnotify_vfsmount_delete(struct vfsmount *mnt)
{
	fsnotify_clear_marks_by_mount(mnt);
}

/**
 * fsnotify_unmount_inodes - an sb is unmounting.  handle any watched inodes.
 * @sb: superblock being unmounted.
 *
 * Called during unmount with no locks held, so needs to be safe against
 * concurrent modifiers. We temporarily drop sb->s_inode_list_lock and CAN block.
 */
static void fsnotify_unmount_inodes(struct super_block *sb)
{
	struct inode *inode, *iput_inode = NULL;

	spin_lock(&sb->s_inode_list_lock);
	list_for_each_entry(inode, &sb->s_inodes, i_sb_list) {
		/*
		 * We cannot __iget() an inode in state I_FREEING,
		 * I_WILL_FREE, or I_NEW which is fine because by that point
		 * the inode cannot have any associated watches.
		 */
		spin_lock(&inode->i_lock);
		if (inode->i_state & (I_FREEING|I_WILL_FREE|I_NEW)) {
			spin_unlock(&inode->i_lock);
			continue;
		}

		/*
		 * If i_count is zero, the inode cannot have any watches and
		 * doing an __iget/iput with SB_ACTIVE clear would actually
		 * evict all inodes with zero i_count from icache which is
		 * unnecessarily violent and may in fact be illegal to do.
		 */
		if (!atomic_read(&inode->i_count)) {
			spin_unlock(&inode->i_lock);
			continue;
		}

		__iget(inode);
		spin_unlock(&inode->i_lock);
		spin_unlock(&sb->s_inode_list_lock);

		if (iput_inode)
			iput(iput_inode);

		/* for each watch, send FS_UNMOUNT and then remove it */
		fsnotify(inode, FS_UNMOUNT, inode, FSNOTIFY_EVENT_INODE, NULL, 0);

		fsnotify_inode_delete(inode);

		iput_inode = inode;

		spin_lock(&sb->s_inode_list_lock);
	}
	spin_unlock(&sb->s_inode_list_lock);

	if (iput_inode)
		iput(iput_inode);
	/* Wait for outstanding inode references from connectors */
	wait_var_event(&sb->s_fsnotify_inode_refs,
		       !atomic_long_read(&sb->s_fsnotify_inode_refs));
}

void fsnotify_sb_delete(struct super_block *sb)
{
	fsnotify_unmount_inodes(sb);
	fsnotify_clear_marks_by_sb(sb);
}

/*
 * Given an inode, first check if we care what happens to our children.  Inotify
 * and dnotify both tell their parents about events.  If we care about any event
 * on a child we run all of our children and set a dentry flag saying that the
 * parent cares.  Thus when an event happens on a child it can quickly tell if
 * if there is a need to find a parent and send the event to the parent.
 */
void __fsnotify_update_child_dentry_flags(struct inode *inode)
{
	struct dentry *alias;
	int watched;

	if (!S_ISDIR(inode->i_mode))
		return;

	/* determine if the children should tell inode about their events */
	watched = fsnotify_inode_watches_children(inode);

	spin_lock(&inode->i_lock);
	/* run all of the dentries associated with this inode.  Since this is a
	 * directory, there damn well better only be one item on this list */
	hlist_for_each_entry(alias, &inode->i_dentry, d_u.d_alias) {
		struct dentry *child;

		/* run all of the children of the original inode and fix their
		 * d_flags to indicate parental interest (their parent is the
		 * original inode) */
		spin_lock(&alias->d_lock);
		list_for_each_entry(child, &alias->d_subdirs, d_child) {
			if (!child->d_inode)
				continue;

			spin_lock_nested(&child->d_lock, DENTRY_D_LOCK_NESTED);
			if (watched)
				child->d_flags |= DCACHE_FSNOTIFY_PARENT_WATCHED;
			else
				child->d_flags &= ~DCACHE_FSNOTIFY_PARENT_WATCHED;
			spin_unlock(&child->d_lock);
		}
		spin_unlock(&alias->d_lock);
	}
	spin_unlock(&inode->i_lock);
}

/* Notify this dentry's parent about a child's events. */
int __fsnotify_parent(const struct path *path, struct dentry *dentry, __u32 mask)
{
	struct dentry *parent;
	struct inode *p_inode;
	int ret = 0;

	if (!dentry)
		dentry = path->dentry;

	if (!(dentry->d_flags & DCACHE_FSNOTIFY_PARENT_WATCHED))
		return 0;

	parent = dget_parent(dentry);
	p_inode = parent->d_inode;

	if (unlikely(!fsnotify_inode_watches_children(p_inode))) {
		__fsnotify_update_child_dentry_flags(p_inode);
	} else if (p_inode->i_fsnotify_mask & mask & ALL_FSNOTIFY_EVENTS) {
		struct name_snapshot name;

		/* we are notifying a parent so come up with the new mask which
		 * specifies these are events which came from a child. */
		mask |= FS_EVENT_ON_CHILD;

		take_dentry_name_snapshot(&name, dentry);
		if (path)
			ret = fsnotify(p_inode, mask, path, FSNOTIFY_EVENT_PATH,
				       name.name, 0);
		else
			ret = fsnotify(p_inode, mask, dentry->d_inode, FSNOTIFY_EVENT_INODE,
				       name.name, 0);
		release_dentry_name_snapshot(&name);
	}

	dput(parent);

	return ret;
}
EXPORT_SYMBOL_GPL(__fsnotify_parent);

static int send_to_group(struct inode *to_tell,
			 __u32 mask, const void *data,
			 int data_is, u32 cookie,
			 const unsigned char *file_name,
			 struct fsnotify_iter_info *iter_info)
{
	struct fsnotify_group *group = NULL;
	__u32 test_mask = (mask & ALL_FSNOTIFY_EVENTS);
	__u32 marks_mask = 0;
	__u32 marks_ignored_mask = 0;
	struct fsnotify_mark *mark;
	int type;

	if (WARN_ON(!iter_info->report_mask))
		return 0;

	/* clear ignored on inode modification */
	if (mask & FS_MODIFY) {
		fsnotify_foreach_obj_type(type) {
			if (!fsnotify_iter_should_report_type(iter_info, type))
				continue;
			mark = iter_info->marks[type];
			if (mark &&
			    !(mark->flags & FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY))
				mark->ignored_mask = 0;
		}
	}

	fsnotify_foreach_obj_type(type) {
		if (!fsnotify_iter_should_report_type(iter_info, type))
			continue;
		mark = iter_info->marks[type];
		/* does the object mark tell us to do something? */
		if (mark) {
			group = mark->group;
			marks_mask |= mark->mask;
			marks_ignored_mask |= mark->ignored_mask;
		}
	}

	pr_debug("%s: group=%p to_tell=%p mask=%x marks_mask=%x marks_ignored_mask=%x"
		 " data=%p data_is=%d cookie=%d\n",
		 __func__, group, to_tell, mask, marks_mask, marks_ignored_mask,
		 data, data_is, cookie);

	if (!(test_mask & marks_mask & ~marks_ignored_mask))
		return 0;

	return group->ops->handle_event(group, to_tell, mask, data, data_is,
					file_name, cookie, iter_info);
}

static struct fsnotify_mark *fsnotify_first_mark(struct fsnotify_mark_connector **connp)
{
	struct fsnotify_mark_connector *conn;
	struct hlist_node *node = NULL;

	conn = srcu_dereference(*connp, &fsnotify_mark_srcu);
	if (conn)
		node = srcu_dereference(conn->list.first, &fsnotify_mark_srcu);

	return hlist_entry_safe(node, struct fsnotify_mark, obj_list);
}

static struct fsnotify_mark *fsnotify_next_mark(struct fsnotify_mark *mark)
{
	struct hlist_node *node = NULL;

	if (mark)
		node = srcu_dereference(mark->obj_list.next,
					&fsnotify_mark_srcu);

	return hlist_entry_safe(node, struct fsnotify_mark, obj_list);
}

/*
 * iter_info is a multi head priority queue of marks.
 * Pick a subset of marks from queue heads, all with the
 * same group and set the report_mask for selected subset.
 * Returns the report_mask of the selected subset.
 */
static unsigned int fsnotify_iter_select_report_types(
		struct fsnotify_iter_info *iter_info)
{
	struct fsnotify_group *max_prio_group = NULL;
	struct fsnotify_mark *mark;
	int type;

	/* Choose max prio group among groups of all queue heads */
	fsnotify_foreach_obj_type(type) {
		mark = iter_info->marks[type];
		if (mark &&
		    fsnotify_compare_groups(max_prio_group, mark->group) > 0)
			max_prio_group = mark->group;
	}

	if (!max_prio_group)
		return 0;

	/* Set the report mask for marks from same group as max prio group */
	iter_info->report_mask = 0;
	fsnotify_foreach_obj_type(type) {
		mark = iter_info->marks[type];
		if (mark &&
		    fsnotify_compare_groups(max_prio_group, mark->group) == 0)
			fsnotify_iter_set_report_type(iter_info, type);
	}

	return iter_info->report_mask;
}

/*
 * Pop from iter_info multi head queue, the marks that were iterated in the
 * current iteration step.
 */
static void fsnotify_iter_next(struct fsnotify_iter_info *iter_info)
{
	int type;

	fsnotify_foreach_obj_type(type) {
		if (fsnotify_iter_should_report_type(iter_info, type))
			iter_info->marks[type] =
				fsnotify_next_mark(iter_info->marks[type]);
	}
}

/*
 * 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
 * out to all of the registered fsnotify_group.  Those groups can then use the
 * notification event in whatever means they feel necessary.
 */
int fsnotify(struct inode *to_tell, __u32 mask, const void *data, int data_is,
	     const unsigned char *file_name, u32 cookie)
{
	struct fsnotify_iter_info iter_info = {};
	struct super_block *sb = NULL;
	struct mount *mnt = NULL;
	__u32 mnt_or_sb_mask = 0;
	int ret = 0;
	__u32 test_mask = (mask & ALL_FSNOTIFY_EVENTS);

	if (data_is == FSNOTIFY_EVENT_PATH) {
		mnt = real_mount(((const struct path *)data)->mnt);
		sb = mnt->mnt.mnt_sb;
		mnt_or_sb_mask = mnt->mnt_fsnotify_mask | sb->s_fsnotify_mask;
	}
	/* An event "on child" is not intended for a mount/sb mark */
	if (mask & FS_EVENT_ON_CHILD)
		mnt_or_sb_mask = 0;

	/*
	 * Optimization: srcu_read_lock() has a memory barrier which can
	 * be expensive.  It protects walking the *_fsnotify_marks lists.
	 * However, if we do not walk the lists, we do not have to do
	 * SRCU because we have no references to any objects and do not
	 * need SRCU to keep them "alive".
	 */
	if (!to_tell->i_fsnotify_marks &&
	    (!mnt || (!mnt->mnt_fsnotify_marks && !sb->s_fsnotify_marks)))
		return 0;
	/*
	 * if this is a modify event we may need to clear the ignored masks
	 * otherwise return if neither the inode nor the vfsmount/sb care about
	 * this type of event.
	 */
	if (!(mask & FS_MODIFY) &&
	    !(test_mask & (to_tell->i_fsnotify_mask | mnt_or_sb_mask)))
		return 0;

	iter_info.srcu_idx = srcu_read_lock(&fsnotify_mark_srcu);

	iter_info.marks[FSNOTIFY_OBJ_TYPE_INODE] =
		fsnotify_first_mark(&to_tell->i_fsnotify_marks);
	if (mnt) {
		iter_info.marks[FSNOTIFY_OBJ_TYPE_VFSMOUNT] =
			fsnotify_first_mark(&mnt->mnt_fsnotify_marks);
		iter_info.marks[FSNOTIFY_OBJ_TYPE_SB] =
			fsnotify_first_mark(&sb->s_fsnotify_marks);
	}

	/*
	 * We need to merge inode/vfsmount/sb mark lists so that e.g. inode mark
	 * ignore masks are properly reflected for mount/sb mark notifications.
	 * That's why this traversal is so complicated...
	 */
	while (fsnotify_iter_select_report_types(&iter_info)) {
		ret = send_to_group(to_tell, mask, data, data_is, cookie,
				    file_name, &iter_info);

		if (ret && (mask & ALL_FSNOTIFY_PERM_EVENTS))
			goto out;

		fsnotify_iter_next(&iter_info);
	}
	ret = 0;
out:
	srcu_read_unlock(&fsnotify_mark_srcu, iter_info.srcu_idx);

	return ret;
}
EXPORT_SYMBOL_GPL(fsnotify);

extern struct kmem_cache *fsnotify_mark_connector_cachep;

static __init int fsnotify_init(void)
{
	int ret;

	BUILD_BUG_ON(HWEIGHT32(ALL_FSNOTIFY_BITS) != 23);

	ret = init_srcu_struct(&fsnotify_mark_srcu);
	if (ret)
		panic("initializing fsnotify_mark_srcu");

	fsnotify_mark_connector_cachep = KMEM_CACHE(fsnotify_mark_connector,
						    SLAB_PANIC);

	return 0;
}
core_initcall(fsnotify_init);
Commit	Line	Data
90586523 EP	1	/*
	2	* Copyright (C) 2008 Red Hat, Inc., Eric Paris <[email protected]>
	3	*
	4	* This program is free software; you can redistribute it and/or modify
	5	* it under the terms of the GNU General Public License as published by
	6	* the Free Software Foundation; either version 2, or (at your option)
	7	* any later version.
	8	*
	9	* This program is distributed in the hope that it will be useful,
	10	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	11	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	12	* GNU General Public License for more details.
	13	*
	14	* You should have received a copy of the GNU General Public License
	15	* along with this program; see the file COPYING. If not, write to
	16	* the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
	17	*/
	18
	19	#include <linux/dcache.h>
	20	#include <linux/fs.h>
5a0e3ad6	21	#include <linux/gfp.h>
90586523 EP	22	#include <linux/init.h>
90586523 EP	23	#include <linux/module.h>
7131485a	24	#include <linux/mount.h>
90586523 EP	25	#include <linux/srcu.h>
	26
	27	#include <linux/fsnotify_backend.h>
	28	#include "fsnotify.h"
	29
3be25f49 EP	30	/*
	31	* Clear all of the marks on an inode when it is being evicted from core
	32	*/
	33	void __fsnotify_inode_delete(struct inode *inode)
	34	{
	35	fsnotify_clear_marks_by_inode(inode);
	36	}
	37	EXPORT_SYMBOL_GPL(__fsnotify_inode_delete);
	38
ca9c726e AG	39	void __fsnotify_vfsmount_delete(struct vfsmount *mnt)
	40	{
	41	fsnotify_clear_marks_by_mount(mnt);
	42	}
	43
ebb3b47e JK	44	/**
	45	* fsnotify_unmount_inodes - an sb is unmounting. handle any watched inodes.
	46	* @sb: superblock being unmounted.
	47	*
	48	* Called during unmount with no locks held, so needs to be safe against
	49	* concurrent modifiers. We temporarily drop sb->s_inode_list_lock and CAN block.
	50	*/
1e6cb723	51	static void fsnotify_unmount_inodes(struct super_block *sb)
ebb3b47e JK	52	{
	53	struct inode inode, iput_inode = NULL;
	54
	55	spin_lock(&sb->s_inode_list_lock);
	56	list_for_each_entry(inode, &sb->s_inodes, i_sb_list) {
	57	/*
	58	* We cannot __iget() an inode in state I_FREEING,
	59	* I_WILL_FREE, or I_NEW which is fine because by that point
	60	* the inode cannot have any associated watches.
	61	*/
	62	spin_lock(&inode->i_lock);
	63	if (inode->i_state & (I_FREEING\|I_WILL_FREE\|I_NEW)) {
	64	spin_unlock(&inode->i_lock);
	65	continue;
	66	}
	67
	68	/*
	69	* If i_count is zero, the inode cannot have any watches and
1751e8a6	70	* doing an __iget/iput with SB_ACTIVE clear would actually
ebb3b47e JK	71	* evict all inodes with zero i_count from icache which is
	72	* unnecessarily violent and may in fact be illegal to do.
	73	*/
	74	if (!atomic_read(&inode->i_count)) {
	75	spin_unlock(&inode->i_lock);
	76	continue;
	77	}
	78
	79	__iget(inode);
	80	spin_unlock(&inode->i_lock);
	81	spin_unlock(&sb->s_inode_list_lock);
	82
	83	if (iput_inode)
	84	iput(iput_inode);
	85
	86	/* for each watch, send FS_UNMOUNT and then remove it */
	87	fsnotify(inode, FS_UNMOUNT, inode, FSNOTIFY_EVENT_INODE, NULL, 0);
	88
	89	fsnotify_inode_delete(inode);
	90
	91	iput_inode = inode;
	92
	93	spin_lock(&sb->s_inode_list_lock);
	94	}
	95	spin_unlock(&sb->s_inode_list_lock);
	96
	97	if (iput_inode)
	98	iput(iput_inode);
721fb6fb JK	99	/* Wait for outstanding inode references from connectors */
	100	wait_var_event(&sb->s_fsnotify_inode_refs,
	101	!atomic_long_read(&sb->s_fsnotify_inode_refs));
ebb3b47e JK	102	}
ebb3b47e JK	103
1e6cb723 AG	104	void fsnotify_sb_delete(struct super_block *sb)
	105	{
	106	fsnotify_unmount_inodes(sb);
	107	fsnotify_clear_marks_by_sb(sb);
	108	}
	109
c28f7e56 EP	110	/*
	111	* Given an inode, first check if we care what happens to our children. Inotify
	112	* and dnotify both tell their parents about events. If we care about any event
	113	* on a child we run all of our children and set a dentry flag saying that the
	114	* parent cares. Thus when an event happens on a child it can quickly tell if
	115	* if there is a need to find a parent and send the event to the parent.
	116	*/
	117	void __fsnotify_update_child_dentry_flags(struct inode *inode)
	118	{
	119	struct dentry *alias;
	120	int watched;
	121
	122	if (!S_ISDIR(inode->i_mode))
	123	return;
	124
	125	/* determine if the children should tell inode about their events */
	126	watched = fsnotify_inode_watches_children(inode);
	127
873feea0	128	spin_lock(&inode->i_lock);
c28f7e56 EP	129	/* run all of the dentries associated with this inode. Since this is a
c28f7e56 EP	130	* directory, there damn well better only be one item on this list */
946e51f2	131	hlist_for_each_entry(alias, &inode->i_dentry, d_u.d_alias) {
c28f7e56 EP	132	struct dentry *child;
	133
	134	/* run all of the children of the original inode and fix their
	135	* d_flags to indicate parental interest (their parent is the
	136	* original inode) */
2fd6b7f5	137	spin_lock(&alias->d_lock);
946e51f2	138	list_for_each_entry(child, &alias->d_subdirs, d_child) {
c28f7e56 EP	139	if (!child->d_inode)
	140	continue;
	141
2fd6b7f5	142	spin_lock_nested(&child->d_lock, DENTRY_D_LOCK_NESTED);
c28f7e56 EP	143	if (watched)
	144	child->d_flags \|= DCACHE_FSNOTIFY_PARENT_WATCHED;
	145	else
	146	child->d_flags &= ~DCACHE_FSNOTIFY_PARENT_WATCHED;
	147	spin_unlock(&child->d_lock);
	148	}
2fd6b7f5	149	spin_unlock(&alias->d_lock);
c28f7e56	150	}
873feea0	151	spin_unlock(&inode->i_lock);
c28f7e56 EP	152	}
	153
	154	/* Notify this dentry's parent about a child's events. */
12c7f9dc	155	int __fsnotify_parent(const struct path path, struct dentry dentry, __u32 mask)
c28f7e56 EP	156	{
	157	struct dentry *parent;
	158	struct inode *p_inode;
52420392	159	int ret = 0;
c28f7e56	160
72acc854	161	if (!dentry)
2069601b	162	dentry = path->dentry;
28c60e37	163
c28f7e56	164	if (!(dentry->d_flags & DCACHE_FSNOTIFY_PARENT_WATCHED))
52420392	165	return 0;
c28f7e56	166
4d4eb366	167	parent = dget_parent(dentry);
c28f7e56 EP	168	p_inode = parent->d_inode;
c28f7e56 EP	169
b469e7e4	170	if (unlikely(!fsnotify_inode_watches_children(p_inode))) {
4d4eb366	171	__fsnotify_update_child_dentry_flags(p_inode);
b469e7e4	172	} else if (p_inode->i_fsnotify_mask & mask & ALL_FSNOTIFY_EVENTS) {
49d31c2f AV	173	struct name_snapshot name;
49d31c2f AV	174
c28f7e56 EP	175	/* we are notifying a parent so come up with the new mask which
	176	* specifies these are events which came from a child. */
	177	mask \|= FS_EVENT_ON_CHILD;
	178
49d31c2f	179	take_dentry_name_snapshot(&name, dentry);
2069601b	180	if (path)
52420392	181	ret = fsnotify(p_inode, mask, path, FSNOTIFY_EVENT_PATH,
49d31c2f	182	name.name, 0);
28c60e37	183	else
52420392	184	ret = fsnotify(p_inode, mask, dentry->d_inode, FSNOTIFY_EVENT_INODE,
49d31c2f AV	185	name.name, 0);
49d31c2f AV	186	release_dentry_name_snapshot(&name);
c28f7e56 EP	187	}
c28f7e56 EP	188
4d4eb366	189	dput(parent);
52420392 EP	190
52420392 EP	191	return ret;
c28f7e56 EP	192	}
	193	EXPORT_SYMBOL_GPL(__fsnotify_parent);
	194
fd657170	195	static int send_to_group(struct inode *to_tell,
e637835e	196	__u32 mask, const void *data,
613a807f	197	int data_is, u32 cookie,
9385a84d JK	198	const unsigned char *file_name,
9385a84d JK	199	struct fsnotify_iter_info *iter_info)
7131485a	200	{
faa9560a	201	struct fsnotify_group *group = NULL;
007d1e83	202	__u32 test_mask = (mask & ALL_FSNOTIFY_EVENTS);
92183a42 AG	203	__u32 marks_mask = 0;
92183a42 AG	204	__u32 marks_ignored_mask = 0;
3dca1a74 AG	205	struct fsnotify_mark *mark;
3dca1a74 AG	206	int type;
613a807f	207
5b0457ad	208	if (WARN_ON(!iter_info->report_mask))
faa9560a	209	return 0;
ce8f76fb EP	210
	211	/* clear ignored on inode modification */
	212	if (mask & FS_MODIFY) {
3dca1a74 AG	213	fsnotify_foreach_obj_type(type) {
	214	if (!fsnotify_iter_should_report_type(iter_info, type))
	215	continue;
	216	mark = iter_info->marks[type];
	217	if (mark &&
	218	!(mark->flags & FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY))
	219	mark->ignored_mask = 0;
	220	}
ce8f76fb	221	}
613a807f	222
3dca1a74 AG	223	fsnotify_foreach_obj_type(type) {
	224	if (!fsnotify_iter_should_report_type(iter_info, type))
	225	continue;
	226	mark = iter_info->marks[type];
	227	/* does the object mark tell us to do something? */
	228	if (mark) {
	229	group = mark->group;
	230	marks_mask \|= mark->mask;
	231	marks_ignored_mask \|= mark->ignored_mask;
	232	}
ce8f76fb EP	233	}
ce8f76fb EP	234
3dca1a74	235	pr_debug("%s: group=%p to_tell=%p mask=%x marks_mask=%x marks_ignored_mask=%x"
7053aee2	236	" data=%p data_is=%d cookie=%d\n",
3dca1a74 AG	237	__func__, group, to_tell, mask, marks_mask, marks_ignored_mask,
3dca1a74 AG	238	data, data_is, cookie);
faa9560a	239
92183a42	240	if (!(test_mask & marks_mask & ~marks_ignored_mask))
613a807f EP	241	return 0;
613a807f EP	242
5b0457ad	243	return group->ops->handle_event(group, to_tell, mask, data, data_is,
9385a84d	244	file_name, cookie, iter_info);
7131485a EP	245	}
7131485a EP	246
3427ce71 MS	247	static struct fsnotify_mark fsnotify_first_mark(struct fsnotify_mark_connector *connp)
	248	{
	249	struct fsnotify_mark_connector *conn;
	250	struct hlist_node *node = NULL;
	251
	252	conn = srcu_dereference(*connp, &fsnotify_mark_srcu);
	253	if (conn)
	254	node = srcu_dereference(conn->list.first, &fsnotify_mark_srcu);
	255
	256	return hlist_entry_safe(node, struct fsnotify_mark, obj_list);
	257	}
	258
	259	static struct fsnotify_mark fsnotify_next_mark(struct fsnotify_mark mark)
	260	{
	261	struct hlist_node *node = NULL;
	262
	263	if (mark)
	264	node = srcu_dereference(mark->obj_list.next,
	265	&fsnotify_mark_srcu);
	266
	267	return hlist_entry_safe(node, struct fsnotify_mark, obj_list);
	268	}
	269
d9a6f30b AG	270	/*
	271	* iter_info is a multi head priority queue of marks.
	272	* Pick a subset of marks from queue heads, all with the
	273	* same group and set the report_mask for selected subset.
	274	* Returns the report_mask of the selected subset.
	275	*/
	276	static unsigned int fsnotify_iter_select_report_types(
	277	struct fsnotify_iter_info *iter_info)
	278	{
47d9c7cc AG	279	struct fsnotify_group *max_prio_group = NULL;
	280	struct fsnotify_mark *mark;
	281	int type;
	282
	283	/* Choose max prio group among groups of all queue heads */
	284	fsnotify_foreach_obj_type(type) {
	285	mark = iter_info->marks[type];
	286	if (mark &&
	287	fsnotify_compare_groups(max_prio_group, mark->group) > 0)
	288	max_prio_group = mark->group;
	289	}
d9a6f30b	290
47d9c7cc	291	if (!max_prio_group)
d9a6f30b AG	292	return 0;
d9a6f30b AG	293
47d9c7cc	294	/* Set the report mask for marks from same group as max prio group */
d9a6f30b	295	iter_info->report_mask = 0;
47d9c7cc AG	296	fsnotify_foreach_obj_type(type) {
	297	mark = iter_info->marks[type];
	298	if (mark &&
	299	fsnotify_compare_groups(max_prio_group, mark->group) == 0)
	300	fsnotify_iter_set_report_type(iter_info, type);
	301	}
d9a6f30b AG	302
	303	return iter_info->report_mask;
	304	}
	305
	306	/*
	307	* Pop from iter_info multi head queue, the marks that were iterated in the
	308	* current iteration step.
	309	*/
	310	static void fsnotify_iter_next(struct fsnotify_iter_info *iter_info)
	311	{
47d9c7cc	312	int type;
d9a6f30b	313
47d9c7cc AG	314	fsnotify_foreach_obj_type(type) {
	315	if (fsnotify_iter_should_report_type(iter_info, type))
	316	iter_info->marks[type] =
	317	fsnotify_next_mark(iter_info->marks[type]);
	318	}
d9a6f30b AG	319	}
d9a6f30b AG	320
90586523 EP	321	/*
	322	* This is the main call to fsnotify. The VFS calls into hook specific functions
	323	* in linux/fsnotify.h. Those functions then in turn call here. Here will call
	324	* out to all of the registered fsnotify_group. Those groups can then use the
	325	* notification event in whatever means they feel necessary.
	326	*/
e637835e	327	int fsnotify(struct inode to_tell, __u32 mask, const void data, int data_is,
c4ec54b4	328	const unsigned char *file_name, u32 cookie)
90586523	329	{
3427ce71	330	struct fsnotify_iter_info iter_info = {};
60f7ed8c AG	331	struct super_block *sb = NULL;
	332	struct mount *mnt = NULL;
	333	__u32 mnt_or_sb_mask = 0;
9385a84d	334	int ret = 0;
007d1e83	335	__u32 test_mask = (mask & ALL_FSNOTIFY_EVENTS);
90586523	336
60f7ed8c	337	if (data_is == FSNOTIFY_EVENT_PATH) {
e637835e	338	mnt = real_mount(((const struct path *)data)->mnt);
60f7ed8c AG	339	sb = mnt->mnt.mnt_sb;
	340	mnt_or_sb_mask = mnt->mnt_fsnotify_mask \| sb->s_fsnotify_mask;
	341	}
b469e7e4 AG	342	/* An event "on child" is not intended for a mount/sb mark */
	343	if (mask & FS_EVENT_ON_CHILD)
	344	mnt_or_sb_mask = 0;
613a807f	345
7c49b861 DH	346	/*
	347	* Optimization: srcu_read_lock() has a memory barrier which can
	348	* be expensive. It protects walking the *_fsnotify_marks lists.
	349	* However, if we do not walk the lists, we do not have to do
	350	* SRCU because we have no references to any objects and do not
	351	* need SRCU to keep them "alive".
	352	*/
9dd813c1	353	if (!to_tell->i_fsnotify_marks &&
60f7ed8c	354	(!mnt \|\| (!mnt->mnt_fsnotify_marks && !sb->s_fsnotify_marks)))
7c49b861	355	return 0;
613a807f EP	356	/*
613a807f EP	357	* if this is a modify event we may need to clear the ignored masks
60f7ed8c	358	* otherwise return if neither the inode nor the vfsmount/sb care about
613a807f EP	359	* this type of event.
	360	*/
	361	if (!(mask & FS_MODIFY) &&
60f7ed8c	362	!(test_mask & (to_tell->i_fsnotify_mask \| mnt_or_sb_mask)))
613a807f	363	return 0;
3a9fb89f	364
9385a84d	365	iter_info.srcu_idx = srcu_read_lock(&fsnotify_mark_srcu);
7131485a	366
9bdda4e9 AG	367	iter_info.marks[FSNOTIFY_OBJ_TYPE_INODE] =
	368	fsnotify_first_mark(&to_tell->i_fsnotify_marks);
	369	if (mnt) {
47d9c7cc	370	iter_info.marks[FSNOTIFY_OBJ_TYPE_VFSMOUNT] =
3427ce71	371	fsnotify_first_mark(&mnt->mnt_fsnotify_marks);
60f7ed8c AG	372	iter_info.marks[FSNOTIFY_OBJ_TYPE_SB] =
60f7ed8c AG	373	fsnotify_first_mark(&sb->s_fsnotify_marks);
90586523	374	}
75c1be48	375
8edc6e16	376	/*
60f7ed8c AG	377	* We need to merge inode/vfsmount/sb mark lists so that e.g. inode mark
60f7ed8c AG	378	* ignore masks are properly reflected for mount/sb mark notifications.
8edc6e16 JK	379	* That's why this traversal is so complicated...
8edc6e16 JK	380	*/
d9a6f30b	381	while (fsnotify_iter_select_report_types(&iter_info)) {
5b0457ad AG	382	ret = send_to_group(to_tell, mask, data, data_is, cookie,
5b0457ad AG	383	file_name, &iter_info);
613a807f	384
ff8bcbd0 EP	385	if (ret && (mask & ALL_FSNOTIFY_PERM_EVENTS))
	386	goto out;
	387
d9a6f30b	388	fsnotify_iter_next(&iter_info);
7131485a	389	}
ff8bcbd0 EP	390	ret = 0;
ff8bcbd0 EP	391	out:
9385a84d	392	srcu_read_unlock(&fsnotify_mark_srcu, iter_info.srcu_idx);
c4ec54b4	393
98b5c10d	394	return ret;
90586523 EP	395	}
	396	EXPORT_SYMBOL_GPL(fsnotify);
	397
9dd813c1 JK	398	extern struct kmem_cache *fsnotify_mark_connector_cachep;
9dd813c1 JK	399
90586523 EP	400	static __init int fsnotify_init(void)
90586523 EP	401	{
75c1be48 EP	402	int ret;
75c1be48 EP	403
a39f7ec4	404	BUILD_BUG_ON(HWEIGHT32(ALL_FSNOTIFY_BITS) != 23);
20dee624	405
75c1be48 EP	406	ret = init_srcu_struct(&fsnotify_mark_srcu);
	407	if (ret)
	408	panic("initializing fsnotify_mark_srcu");
	409
9dd813c1 JK	410	fsnotify_mark_connector_cachep = KMEM_CACHE(fsnotify_mark_connector,
	411	SLAB_PANIC);
	412
75c1be48	413	return 0;
90586523	414	}
75c1be48	415	core_initcall(fsnotify_init);