[mirror_ubuntu-jammy-kernel.git] / fs / notify / fsnotify.c

// SPDX-License-Identifier: GPL-2.0-or-later
/*
 *  Copyright (C) 2008 Red Hat, Inc., Eric Paris <eparis@redhat.com>
 */

#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.
		 * However, we should have been called /after/ evict_inodes
		 * removed all zero refcount inodes, in any case.  Test to
		 * be sure.
		 */
		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;

		cond_resched();
		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 struct qstr *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 struct qstr *file_name, u32 cookie)
{
	struct fsnotify_iter_info iter_info = {};
	struct super_block *sb = to_tell->i_sb;
	struct mount *mnt = NULL;
	__u32 mnt_or_sb_mask = sb->s_fsnotify_mask;
	int ret = 0;
	__u32 test_mask = (mask & ALL_FSNOTIFY_EVENTS);

	if (data_is == FSNOTIFY_EVENT_PATH) {
		mnt = real_mount(((const struct path *)data)->mnt);
		mnt_or_sb_mask |= mnt->mnt_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 && !sb->s_fsnotify_marks &&
	    (!mnt || !mnt->mnt_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);
	iter_info.marks[FSNOTIFY_OBJ_TYPE_SB] =
		fsnotify_first_mark(&sb->s_fsnotify_marks);
	if (mnt) {
		iter_info.marks[FSNOTIFY_OBJ_TYPE_VFSMOUNT] =
			fsnotify_first_mark(&mnt->mnt_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);

static __init int fsnotify_init(void)
{
	int ret;

	BUILD_BUG_ON(HWEIGHT32(ALL_FSNOTIFY_BITS) != 25);

	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
c82ee6d3	1	// SPDX-License-Identifier: GPL-2.0-or-later
90586523 EP	2	/*
90586523 EP	3	* Copyright (C) 2008 Red Hat, Inc., Eric Paris <eparis@redhat.com>
90586523 EP	4	*/
	5
	6	#include <linux/dcache.h>
	7	#include <linux/fs.h>
5a0e3ad6	8	#include <linux/gfp.h>
90586523 EP	9	#include <linux/init.h>
90586523 EP	10	#include <linux/module.h>
7131485a	11	#include <linux/mount.h>
90586523 EP	12	#include <linux/srcu.h>
	13
	14	#include <linux/fsnotify_backend.h>
	15	#include "fsnotify.h"
	16
3be25f49 EP	17	/*
	18	* Clear all of the marks on an inode when it is being evicted from core
	19	*/
	20	void __fsnotify_inode_delete(struct inode *inode)
	21	{
	22	fsnotify_clear_marks_by_inode(inode);
	23	}
	24	EXPORT_SYMBOL_GPL(__fsnotify_inode_delete);
	25
ca9c726e AG	26	void __fsnotify_vfsmount_delete(struct vfsmount *mnt)
	27	{
	28	fsnotify_clear_marks_by_mount(mnt);
	29	}
	30
ebb3b47e JK	31	/**
	32	* fsnotify_unmount_inodes - an sb is unmounting. handle any watched inodes.
	33	* @sb: superblock being unmounted.
	34	*
	35	* Called during unmount with no locks held, so needs to be safe against
	36	* concurrent modifiers. We temporarily drop sb->s_inode_list_lock and CAN block.
	37	*/
1e6cb723	38	static void fsnotify_unmount_inodes(struct super_block *sb)
ebb3b47e JK	39	{
	40	struct inode inode, iput_inode = NULL;
	41
	42	spin_lock(&sb->s_inode_list_lock);
	43	list_for_each_entry(inode, &sb->s_inodes, i_sb_list) {
	44	/*
	45	* We cannot __iget() an inode in state I_FREEING,
	46	* I_WILL_FREE, or I_NEW which is fine because by that point
	47	* the inode cannot have any associated watches.
	48	*/
	49	spin_lock(&inode->i_lock);
	50	if (inode->i_state & (I_FREEING\|I_WILL_FREE\|I_NEW)) {
	51	spin_unlock(&inode->i_lock);
	52	continue;
	53	}
	54
	55	/*
	56	* If i_count is zero, the inode cannot have any watches and
1751e8a6	57	* doing an __iget/iput with SB_ACTIVE clear would actually
ebb3b47e JK	58	* evict all inodes with zero i_count from icache which is
ebb3b47e JK	59	* unnecessarily violent and may in fact be illegal to do.
1edc8eb2 ES	60	* However, we should have been called /after/ evict_inodes
	61	* removed all zero refcount inodes, in any case. Test to
	62	* be sure.
ebb3b47e JK	63	*/
	64	if (!atomic_read(&inode->i_count)) {
	65	spin_unlock(&inode->i_lock);
	66	continue;
	67	}
	68
	69	__iget(inode);
	70	spin_unlock(&inode->i_lock);
	71	spin_unlock(&sb->s_inode_list_lock);
	72
	73	if (iput_inode)
	74	iput(iput_inode);
	75
	76	/* for each watch, send FS_UNMOUNT and then remove it */
	77	fsnotify(inode, FS_UNMOUNT, inode, FSNOTIFY_EVENT_INODE, NULL, 0);
	78
	79	fsnotify_inode_delete(inode);
	80
	81	iput_inode = inode;
	82
04646aeb	83	cond_resched();
ebb3b47e JK	84	spin_lock(&sb->s_inode_list_lock);
	85	}
	86	spin_unlock(&sb->s_inode_list_lock);
	87
	88	if (iput_inode)
	89	iput(iput_inode);
721fb6fb JK	90	/* Wait for outstanding inode references from connectors */
	91	wait_var_event(&sb->s_fsnotify_inode_refs,
	92	!atomic_long_read(&sb->s_fsnotify_inode_refs));
ebb3b47e JK	93	}
ebb3b47e JK	94
1e6cb723 AG	95	void fsnotify_sb_delete(struct super_block *sb)
	96	{
	97	fsnotify_unmount_inodes(sb);
	98	fsnotify_clear_marks_by_sb(sb);
	99	}
	100
c28f7e56 EP	101	/*
	102	* Given an inode, first check if we care what happens to our children. Inotify
	103	* and dnotify both tell their parents about events. If we care about any event
	104	* on a child we run all of our children and set a dentry flag saying that the
	105	* parent cares. Thus when an event happens on a child it can quickly tell if
	106	* if there is a need to find a parent and send the event to the parent.
	107	*/
	108	void __fsnotify_update_child_dentry_flags(struct inode *inode)
	109	{
	110	struct dentry *alias;
	111	int watched;
	112
	113	if (!S_ISDIR(inode->i_mode))
	114	return;
	115
	116	/* determine if the children should tell inode about their events */
	117	watched = fsnotify_inode_watches_children(inode);
	118
873feea0	119	spin_lock(&inode->i_lock);
c28f7e56 EP	120	/* run all of the dentries associated with this inode. Since this is a
c28f7e56 EP	121	* directory, there damn well better only be one item on this list */
946e51f2	122	hlist_for_each_entry(alias, &inode->i_dentry, d_u.d_alias) {
c28f7e56 EP	123	struct dentry *child;
	124
	125	/* run all of the children of the original inode and fix their
	126	* d_flags to indicate parental interest (their parent is the
	127	* original inode) */
2fd6b7f5	128	spin_lock(&alias->d_lock);
946e51f2	129	list_for_each_entry(child, &alias->d_subdirs, d_child) {
c28f7e56 EP	130	if (!child->d_inode)
	131	continue;
	132
2fd6b7f5	133	spin_lock_nested(&child->d_lock, DENTRY_D_LOCK_NESTED);
c28f7e56 EP	134	if (watched)
	135	child->d_flags \|= DCACHE_FSNOTIFY_PARENT_WATCHED;
	136	else
	137	child->d_flags &= ~DCACHE_FSNOTIFY_PARENT_WATCHED;
	138	spin_unlock(&child->d_lock);
	139	}
2fd6b7f5	140	spin_unlock(&alias->d_lock);
c28f7e56	141	}
873feea0	142	spin_unlock(&inode->i_lock);
c28f7e56 EP	143	}
	144
	145	/* Notify this dentry's parent about a child's events. */
12c7f9dc	146	int __fsnotify_parent(const struct path path, struct dentry dentry, __u32 mask)
c28f7e56 EP	147	{
	148	struct dentry *parent;
	149	struct inode *p_inode;
52420392	150	int ret = 0;
c28f7e56	151
72acc854	152	if (!dentry)
2069601b	153	dentry = path->dentry;
28c60e37	154
c28f7e56	155	if (!(dentry->d_flags & DCACHE_FSNOTIFY_PARENT_WATCHED))
52420392	156	return 0;
c28f7e56	157
4d4eb366	158	parent = dget_parent(dentry);
c28f7e56 EP	159	p_inode = parent->d_inode;
c28f7e56 EP	160
b469e7e4	161	if (unlikely(!fsnotify_inode_watches_children(p_inode))) {
4d4eb366	162	__fsnotify_update_child_dentry_flags(p_inode);
b469e7e4	163	} else if (p_inode->i_fsnotify_mask & mask & ALL_FSNOTIFY_EVENTS) {
49d31c2f AV	164	struct name_snapshot name;
49d31c2f AV	165
c28f7e56 EP	166	/* we are notifying a parent so come up with the new mask which
	167	* specifies these are events which came from a child. */
	168	mask \|= FS_EVENT_ON_CHILD;
	169
49d31c2f	170	take_dentry_name_snapshot(&name, dentry);
2069601b	171	if (path)
52420392	172	ret = fsnotify(p_inode, mask, path, FSNOTIFY_EVENT_PATH,
25b229df	173	&name.name, 0);
28c60e37	174	else
52420392	175	ret = fsnotify(p_inode, mask, dentry->d_inode, FSNOTIFY_EVENT_INODE,
25b229df	176	&name.name, 0);
49d31c2f	177	release_dentry_name_snapshot(&name);
c28f7e56 EP	178	}
c28f7e56 EP	179
4d4eb366	180	dput(parent);
52420392 EP	181
52420392 EP	182	return ret;
c28f7e56 EP	183	}
	184	EXPORT_SYMBOL_GPL(__fsnotify_parent);
	185
fd657170	186	static int send_to_group(struct inode *to_tell,
e637835e	187	__u32 mask, const void *data,
613a807f	188	int data_is, u32 cookie,
e43e9c33	189	const struct qstr *file_name,
9385a84d	190	struct fsnotify_iter_info *iter_info)
7131485a	191	{
faa9560a	192	struct fsnotify_group *group = NULL;
007d1e83	193	__u32 test_mask = (mask & ALL_FSNOTIFY_EVENTS);
92183a42 AG	194	__u32 marks_mask = 0;
92183a42 AG	195	__u32 marks_ignored_mask = 0;
3dca1a74 AG	196	struct fsnotify_mark *mark;
3dca1a74 AG	197	int type;
613a807f	198
5b0457ad	199	if (WARN_ON(!iter_info->report_mask))
faa9560a	200	return 0;
ce8f76fb EP	201
	202	/* clear ignored on inode modification */
	203	if (mask & FS_MODIFY) {
3dca1a74 AG	204	fsnotify_foreach_obj_type(type) {
	205	if (!fsnotify_iter_should_report_type(iter_info, type))
	206	continue;
	207	mark = iter_info->marks[type];
	208	if (mark &&
	209	!(mark->flags & FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY))
	210	mark->ignored_mask = 0;
	211	}
ce8f76fb	212	}
613a807f	213
3dca1a74 AG	214	fsnotify_foreach_obj_type(type) {
	215	if (!fsnotify_iter_should_report_type(iter_info, type))
	216	continue;
	217	mark = iter_info->marks[type];
	218	/* does the object mark tell us to do something? */
	219	if (mark) {
	220	group = mark->group;
	221	marks_mask \|= mark->mask;
	222	marks_ignored_mask \|= mark->ignored_mask;
	223	}
ce8f76fb EP	224	}
ce8f76fb EP	225
3dca1a74	226	pr_debug("%s: group=%p to_tell=%p mask=%x marks_mask=%x marks_ignored_mask=%x"
7053aee2	227	" data=%p data_is=%d cookie=%d\n",
3dca1a74 AG	228	__func__, group, to_tell, mask, marks_mask, marks_ignored_mask,
3dca1a74 AG	229	data, data_is, cookie);
faa9560a	230
92183a42	231	if (!(test_mask & marks_mask & ~marks_ignored_mask))
613a807f EP	232	return 0;
613a807f EP	233
5b0457ad	234	return group->ops->handle_event(group, to_tell, mask, data, data_is,
9385a84d	235	file_name, cookie, iter_info);
7131485a EP	236	}
7131485a EP	237
3427ce71 MS	238	static struct fsnotify_mark fsnotify_first_mark(struct fsnotify_mark_connector *connp)
	239	{
	240	struct fsnotify_mark_connector *conn;
	241	struct hlist_node *node = NULL;
	242
	243	conn = srcu_dereference(*connp, &fsnotify_mark_srcu);
	244	if (conn)
	245	node = srcu_dereference(conn->list.first, &fsnotify_mark_srcu);
	246
	247	return hlist_entry_safe(node, struct fsnotify_mark, obj_list);
	248	}
	249
	250	static struct fsnotify_mark fsnotify_next_mark(struct fsnotify_mark mark)
	251	{
	252	struct hlist_node *node = NULL;
	253
	254	if (mark)
	255	node = srcu_dereference(mark->obj_list.next,
	256	&fsnotify_mark_srcu);
	257
	258	return hlist_entry_safe(node, struct fsnotify_mark, obj_list);
	259	}
	260
d9a6f30b AG	261	/*
	262	* iter_info is a multi head priority queue of marks.
	263	* Pick a subset of marks from queue heads, all with the
	264	* same group and set the report_mask for selected subset.
	265	* Returns the report_mask of the selected subset.
	266	*/
	267	static unsigned int fsnotify_iter_select_report_types(
	268	struct fsnotify_iter_info *iter_info)
	269	{
47d9c7cc AG	270	struct fsnotify_group *max_prio_group = NULL;
	271	struct fsnotify_mark *mark;
	272	int type;
	273
	274	/* Choose max prio group among groups of all queue heads */
	275	fsnotify_foreach_obj_type(type) {
	276	mark = iter_info->marks[type];
	277	if (mark &&
	278	fsnotify_compare_groups(max_prio_group, mark->group) > 0)
	279	max_prio_group = mark->group;
	280	}
d9a6f30b	281
47d9c7cc	282	if (!max_prio_group)
d9a6f30b AG	283	return 0;
d9a6f30b AG	284
47d9c7cc	285	/* Set the report mask for marks from same group as max prio group */
d9a6f30b	286	iter_info->report_mask = 0;
47d9c7cc AG	287	fsnotify_foreach_obj_type(type) {
	288	mark = iter_info->marks[type];
	289	if (mark &&
	290	fsnotify_compare_groups(max_prio_group, mark->group) == 0)
	291	fsnotify_iter_set_report_type(iter_info, type);
	292	}
d9a6f30b AG	293
	294	return iter_info->report_mask;
	295	}
	296
	297	/*
	298	* Pop from iter_info multi head queue, the marks that were iterated in the
	299	* current iteration step.
	300	*/
	301	static void fsnotify_iter_next(struct fsnotify_iter_info *iter_info)
	302	{
47d9c7cc	303	int type;
d9a6f30b	304
47d9c7cc AG	305	fsnotify_foreach_obj_type(type) {
	306	if (fsnotify_iter_should_report_type(iter_info, type))
	307	iter_info->marks[type] =
	308	fsnotify_next_mark(iter_info->marks[type]);
	309	}
d9a6f30b AG	310	}
d9a6f30b AG	311
90586523 EP	312	/*
	313	* This is the main call to fsnotify. The VFS calls into hook specific functions
	314	* in linux/fsnotify.h. Those functions then in turn call here. Here will call
	315	* out to all of the registered fsnotify_group. Those groups can then use the
	316	* notification event in whatever means they feel necessary.
	317	*/
e637835e	318	int fsnotify(struct inode to_tell, __u32 mask, const void data, int data_is,
25b229df	319	const struct qstr *file_name, u32 cookie)
90586523	320	{
3427ce71	321	struct fsnotify_iter_info iter_info = {};
45a9fb37	322	struct super_block *sb = to_tell->i_sb;
60f7ed8c	323	struct mount *mnt = NULL;
45a9fb37	324	__u32 mnt_or_sb_mask = sb->s_fsnotify_mask;
9385a84d	325	int ret = 0;
007d1e83	326	__u32 test_mask = (mask & ALL_FSNOTIFY_EVENTS);
90586523	327
60f7ed8c	328	if (data_is == FSNOTIFY_EVENT_PATH) {
e637835e	329	mnt = real_mount(((const struct path *)data)->mnt);
45a9fb37	330	mnt_or_sb_mask \|= mnt->mnt_fsnotify_mask;
60f7ed8c	331	}
b469e7e4 AG	332	/* An event "on child" is not intended for a mount/sb mark */
	333	if (mask & FS_EVENT_ON_CHILD)
	334	mnt_or_sb_mask = 0;
613a807f	335
7c49b861 DH	336	/*
	337	* Optimization: srcu_read_lock() has a memory barrier which can
	338	* be expensive. It protects walking the *_fsnotify_marks lists.
	339	* However, if we do not walk the lists, we do not have to do
	340	* SRCU because we have no references to any objects and do not
	341	* need SRCU to keep them "alive".
	342	*/
45a9fb37 AG	343	if (!to_tell->i_fsnotify_marks && !sb->s_fsnotify_marks &&
45a9fb37 AG	344	(!mnt \|\| !mnt->mnt_fsnotify_marks))
7c49b861	345	return 0;
613a807f EP	346	/*
613a807f EP	347	* if this is a modify event we may need to clear the ignored masks
60f7ed8c	348	* otherwise return if neither the inode nor the vfsmount/sb care about
613a807f EP	349	* this type of event.
	350	*/
	351	if (!(mask & FS_MODIFY) &&
60f7ed8c	352	!(test_mask & (to_tell->i_fsnotify_mask \| mnt_or_sb_mask)))
613a807f	353	return 0;
3a9fb89f	354
9385a84d	355	iter_info.srcu_idx = srcu_read_lock(&fsnotify_mark_srcu);
7131485a	356
9bdda4e9 AG	357	iter_info.marks[FSNOTIFY_OBJ_TYPE_INODE] =
9bdda4e9 AG	358	fsnotify_first_mark(&to_tell->i_fsnotify_marks);
45a9fb37 AG	359	iter_info.marks[FSNOTIFY_OBJ_TYPE_SB] =
45a9fb37 AG	360	fsnotify_first_mark(&sb->s_fsnotify_marks);
9bdda4e9	361	if (mnt) {
47d9c7cc	362	iter_info.marks[FSNOTIFY_OBJ_TYPE_VFSMOUNT] =
3427ce71	363	fsnotify_first_mark(&mnt->mnt_fsnotify_marks);
90586523	364	}
75c1be48	365
8edc6e16	366	/*
60f7ed8c AG	367	* We need to merge inode/vfsmount/sb mark lists so that e.g. inode mark
60f7ed8c AG	368	* ignore masks are properly reflected for mount/sb mark notifications.
8edc6e16 JK	369	* That's why this traversal is so complicated...
8edc6e16 JK	370	*/
d9a6f30b	371	while (fsnotify_iter_select_report_types(&iter_info)) {
5b0457ad	372	ret = send_to_group(to_tell, mask, data, data_is, cookie,
e43e9c33	373	file_name, &iter_info);
613a807f	374
ff8bcbd0 EP	375	if (ret && (mask & ALL_FSNOTIFY_PERM_EVENTS))
	376	goto out;
	377
d9a6f30b	378	fsnotify_iter_next(&iter_info);
7131485a	379	}
ff8bcbd0 EP	380	ret = 0;
ff8bcbd0 EP	381	out:
9385a84d	382	srcu_read_unlock(&fsnotify_mark_srcu, iter_info.srcu_idx);
c4ec54b4	383
98b5c10d	384	return ret;
90586523 EP	385	}
	386	EXPORT_SYMBOL_GPL(fsnotify);
	387
	388	static __init int fsnotify_init(void)
	389	{
75c1be48 EP	390	int ret;
75c1be48 EP	391
66917a31	392	BUILD_BUG_ON(HWEIGHT32(ALL_FSNOTIFY_BITS) != 25);
20dee624	393
75c1be48 EP	394	ret = init_srcu_struct(&fsnotify_mark_srcu);
	395	if (ret)
	396	panic("initializing fsnotify_mark_srcu");
	397
9dd813c1 JK	398	fsnotify_mark_connector_cachep = KMEM_CACHE(fsnotify_mark_connector,
	399	SLAB_PANIC);
	400
75c1be48	401	return 0;
90586523	402	}
75c1be48	403	core_initcall(fsnotify_init);