[linux.git] / fs / notify / mark.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.
 */

/*
 * fsnotify inode mark locking/lifetime/and refcnting
 *
 * REFCNT:
 * The group->recnt and mark->refcnt tell how many "things" in the kernel
 * currently are referencing the objects. Both kind of objects typically will
 * live inside the kernel with a refcnt of 2, one for its creation and one for
 * the reference a group and a mark hold to each other.
 * If you are holding the appropriate locks, you can take a reference and the
 * object itself is guaranteed to survive until the reference is dropped.
 *
 * LOCKING:
 * There are 3 locks involved with fsnotify inode marks and they MUST be taken
 * in order as follows:
 *
 * group->mark_mutex
 * mark->lock
 * inode->i_lock
 *
 * group->mark_mutex protects the marks_list anchored inside a given group and
 * each mark is hooked via the g_list.  It also protects the groups private
 * data (i.e group limits).

 * mark->lock protects the marks attributes like its masks and flags.
 * Furthermore it protects the access to a reference of the group that the mark
 * is assigned to as well as the access to a reference of the inode/vfsmount
 * that is being watched by the mark.
 *
 * inode->i_lock protects the i_fsnotify_marks list anchored inside a
 * given inode and each mark is hooked via the i_list. (and sorta the
 * free_i_list)
 *
 *
 * LIFETIME:
 * Inode marks survive between when they are added to an inode and when their
 * refcnt==0.
 *
 * The inode mark can be cleared for a number of different reasons including:
 * - The inode is unlinked for the last time.  (fsnotify_inode_remove)
 * - The inode is being evicted from cache. (fsnotify_inode_delete)
 * - The fs the inode is on is unmounted.  (fsnotify_inode_delete/fsnotify_unmount_inodes)
 * - Something explicitly requests that it be removed.  (fsnotify_destroy_mark)
 * - The fsnotify_group associated with the mark is going away and all such marks
 *   need to be cleaned up. (fsnotify_clear_marks_by_group)
 *
 * Worst case we are given an inode and need to clean up all the marks on that
 * inode.  We take i_lock and walk the i_fsnotify_marks safely.  For each
 * mark on the list we take a reference (so the mark can't disappear under us).
 * We remove that mark form the inode's list of marks and we add this mark to a
 * private list anchored on the stack using i_free_list; we walk i_free_list
 * and before we destroy the mark we make sure that we dont race with a
 * concurrent destroy_group by getting a ref to the marks group and taking the
 * groups mutex.

 * Very similarly for freeing by group, except we use free_g_list.
 *
 * This has the very interesting property of being able to run concurrently with
 * any (or all) other directions.
 */

#include <linux/fs.h>
#include <linux/init.h>
#include <linux/kernel.h>
#include <linux/kthread.h>
#include <linux/module.h>
#include <linux/mutex.h>
#include <linux/slab.h>
#include <linux/spinlock.h>
#include <linux/srcu.h>

#include <linux/atomic.h>

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

struct srcu_struct fsnotify_mark_srcu;
static DEFINE_SPINLOCK(destroy_lock);
static LIST_HEAD(destroy_list);
static DECLARE_WAIT_QUEUE_HEAD(destroy_waitq);

void fsnotify_get_mark(struct fsnotify_mark *mark)
{
	atomic_inc(&mark->refcnt);
}

void fsnotify_put_mark(struct fsnotify_mark *mark)
{
	if (atomic_dec_and_test(&mark->refcnt)) {
		if (mark->group)
			fsnotify_put_group(mark->group);
		mark->free_mark(mark);
	}
}

/*
 * Any time a mark is getting freed we end up here.
 * The caller had better be holding a reference to this mark so we don't actually
 * do the final put under the mark->lock
 */
void fsnotify_destroy_mark_locked(struct fsnotify_mark *mark,
				  struct fsnotify_group *group)
{
	struct inode *inode = NULL;

	BUG_ON(!mutex_is_locked(&group->mark_mutex));

	spin_lock(&mark->lock);

	/* something else already called this function on this mark */
	if (!(mark->flags & FSNOTIFY_MARK_FLAG_ALIVE)) {
		spin_unlock(&mark->lock);
		return;
	}

	mark->flags &= ~FSNOTIFY_MARK_FLAG_ALIVE;

	if (mark->flags & FSNOTIFY_MARK_FLAG_INODE) {
		inode = mark->i.inode;
		fsnotify_destroy_inode_mark(mark);
	} else if (mark->flags & FSNOTIFY_MARK_FLAG_VFSMOUNT)
		fsnotify_destroy_vfsmount_mark(mark);
	else
		BUG();

	list_del_init(&mark->g_list);

	spin_unlock(&mark->lock);

	if (inode && (mark->flags & FSNOTIFY_MARK_FLAG_OBJECT_PINNED))
		iput(inode);
	/* release lock temporarily */
	mutex_unlock(&group->mark_mutex);

	spin_lock(&destroy_lock);
	list_add(&mark->destroy_list, &destroy_list);
	spin_unlock(&destroy_lock);
	wake_up(&destroy_waitq);
	/*
	 * We don't necessarily have a ref on mark from caller so the above destroy
	 * may have actually freed it, unless this group provides a 'freeing_mark'
	 * function which must be holding a reference.
	 */

	/*
	 * Some groups like to know that marks are being freed.  This is a
	 * callback to the group function to let it know that this mark
	 * is being freed.
	 */
	if (group->ops->freeing_mark)
		group->ops->freeing_mark(mark, group);

	/*
	 * __fsnotify_update_child_dentry_flags(inode);
	 *
	 * I really want to call that, but we can't, we have no idea if the inode
	 * still exists the second we drop the mark->lock.
	 *
	 * The next time an event arrive to this inode from one of it's children
	 * __fsnotify_parent will see that the inode doesn't care about it's
	 * children and will update all of these flags then.  So really this
	 * is just a lazy update (and could be a perf win...)
	 */

	atomic_dec(&group->num_marks);

	mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING);
}

void fsnotify_destroy_mark(struct fsnotify_mark *mark,
			   struct fsnotify_group *group)
{
	mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING);
	fsnotify_destroy_mark_locked(mark, group);
	mutex_unlock(&group->mark_mutex);
}

void fsnotify_set_mark_mask_locked(struct fsnotify_mark *mark, __u32 mask)
{
	assert_spin_locked(&mark->lock);

	mark->mask = mask;

	if (mark->flags & FSNOTIFY_MARK_FLAG_INODE)
		fsnotify_set_inode_mark_mask_locked(mark, mask);
}

void fsnotify_set_mark_ignored_mask_locked(struct fsnotify_mark *mark, __u32 mask)
{
	assert_spin_locked(&mark->lock);

	mark->ignored_mask = mask;
}

/*
 * Sorting function for lists of fsnotify marks.
 *
 * Fanotify supports different notification classes (reflected as priority of
 * notification group). Events shall be passed to notification groups in
 * decreasing priority order. To achieve this marks in notification lists for
 * inodes and vfsmounts are sorted so that priorities of corresponding groups
 * are descending.
 *
 * Furthermore correct handling of the ignore mask requires processing inode
 * and vfsmount marks of each group together. Using the group address as
 * further sort criterion provides a unique sorting order and thus we can
 * merge inode and vfsmount lists of marks in linear time and find groups
 * present in both lists.
 *
 * A return value of 1 signifies that b has priority over a.
 * A return value of 0 signifies that the two marks have to be handled together.
 * A return value of -1 signifies that a has priority over b.
 */
int fsnotify_compare_groups(struct fsnotify_group *a, struct fsnotify_group *b)
{
	if (a == b)
		return 0;
	if (!a)
		return 1;
	if (!b)
		return -1;
	if (a->priority < b->priority)
		return 1;
	if (a->priority > b->priority)
		return -1;
	if (a < b)
		return 1;
	return -1;
}

/*
 * Attach an initialized mark to a given group and fs object.
 * These marks may be used for the fsnotify backend to determine which
 * event types should be delivered to which group.
 */
int fsnotify_add_mark_locked(struct fsnotify_mark *mark,
			     struct fsnotify_group *group, struct inode *inode,
			     struct vfsmount *mnt, int allow_dups)
{
	int ret = 0;

	BUG_ON(inode && mnt);
	BUG_ON(!inode && !mnt);
	BUG_ON(!mutex_is_locked(&group->mark_mutex));

	/*
	 * LOCKING ORDER!!!!
	 * group->mark_mutex
	 * mark->lock
	 * inode->i_lock
	 */
	spin_lock(&mark->lock);
	mark->flags |= FSNOTIFY_MARK_FLAG_ALIVE;

	fsnotify_get_group(group);
	mark->group = group;
	list_add(&mark->g_list, &group->marks_list);
	atomic_inc(&group->num_marks);
	fsnotify_get_mark(mark); /* for i_list and g_list */

	if (inode) {
		ret = fsnotify_add_inode_mark(mark, group, inode, allow_dups);
		if (ret)
			goto err;
	} else if (mnt) {
		ret = fsnotify_add_vfsmount_mark(mark, group, mnt, allow_dups);
		if (ret)
			goto err;
	} else {
		BUG();
	}

	/* this will pin the object if appropriate */
	fsnotify_set_mark_mask_locked(mark, mark->mask);
	spin_unlock(&mark->lock);

	if (inode)
		__fsnotify_update_child_dentry_flags(inode);

	return ret;
err:
	mark->flags &= ~FSNOTIFY_MARK_FLAG_ALIVE;
	list_del_init(&mark->g_list);
	fsnotify_put_group(group);
	mark->group = NULL;
	atomic_dec(&group->num_marks);

	spin_unlock(&mark->lock);

	spin_lock(&destroy_lock);
	list_add(&mark->destroy_list, &destroy_list);
	spin_unlock(&destroy_lock);
	wake_up(&destroy_waitq);

	return ret;
}

int fsnotify_add_mark(struct fsnotify_mark *mark, struct fsnotify_group *group,
		      struct inode *inode, struct vfsmount *mnt, int allow_dups)
{
	int ret;
	mutex_lock(&group->mark_mutex);
	ret = fsnotify_add_mark_locked(mark, group, inode, mnt, allow_dups);
	mutex_unlock(&group->mark_mutex);
	return ret;
}

/*
 * clear any marks in a group in which mark->flags & flags is true
 */
void fsnotify_clear_marks_by_group_flags(struct fsnotify_group *group,
					 unsigned int flags)
{
	struct fsnotify_mark *lmark, *mark;

	mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING);
	list_for_each_entry_safe(mark, lmark, &group->marks_list, g_list) {
		if (mark->flags & flags) {
			fsnotify_get_mark(mark);
			fsnotify_destroy_mark_locked(mark, group);
			fsnotify_put_mark(mark);
		}
	}
	mutex_unlock(&group->mark_mutex);
}

/*
 * Given a group, destroy all of the marks associated with that group.
 */
void fsnotify_clear_marks_by_group(struct fsnotify_group *group)
{
	fsnotify_clear_marks_by_group_flags(group, (unsigned int)-1);
}

void fsnotify_duplicate_mark(struct fsnotify_mark *new, struct fsnotify_mark *old)
{
	assert_spin_locked(&old->lock);
	new->i.inode = old->i.inode;
	new->m.mnt = old->m.mnt;
	if (old->group)
		fsnotify_get_group(old->group);
	new->group = old->group;
	new->mask = old->mask;
	new->free_mark = old->free_mark;
}

/*
 * Nothing fancy, just initialize lists and locks and counters.
 */
void fsnotify_init_mark(struct fsnotify_mark *mark,
			void (*free_mark)(struct fsnotify_mark *mark))
{
	memset(mark, 0, sizeof(*mark));
	spin_lock_init(&mark->lock);
	atomic_set(&mark->refcnt, 1);
	mark->free_mark = free_mark;
}

static int fsnotify_mark_destroy(void *ignored)
{
	struct fsnotify_mark *mark, *next;
	struct list_head private_destroy_list;

	for (;;) {
		spin_lock(&destroy_lock);
		/* exchange the list head */
		list_replace_init(&destroy_list, &private_destroy_list);
		spin_unlock(&destroy_lock);

		synchronize_srcu(&fsnotify_mark_srcu);

		list_for_each_entry_safe(mark, next, &private_destroy_list, destroy_list) {
			list_del_init(&mark->destroy_list);
			fsnotify_put_mark(mark);
		}

		wait_event_interruptible(destroy_waitq, !list_empty(&destroy_list));
	}

	return 0;
}

static int __init fsnotify_mark_init(void)
{
	struct task_struct *thread;

	thread = kthread_run(fsnotify_mark_destroy, NULL,
			     "fsnotify_mark");
	if (IS_ERR(thread))
		panic("unable to start fsnotify mark destruction thread.");

	return 0;
}
device_initcall(fsnotify_mark_init);
Commit	Line	Data
5444e298 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	/*
	20	* fsnotify inode mark locking/lifetime/and refcnting
	21	*
	22	* REFCNT:
9756b918 LS	23	* The group->recnt and mark->refcnt tell how many "things" in the kernel
	24	* currently are referencing the objects. Both kind of objects typically will
	25	* live inside the kernel with a refcnt of 2, one for its creation and one for
	26	* the reference a group and a mark hold to each other.
	27	* If you are holding the appropriate locks, you can take a reference and the
	28	* object itself is guaranteed to survive until the reference is dropped.
5444e298 EP	29	*
5444e298 EP	30	* LOCKING:
9756b918 LS	31	* There are 3 locks involved with fsnotify inode marks and they MUST be taken
9756b918 LS	32	* in order as follows:
5444e298	33	*
9756b918	34	* group->mark_mutex
5444e298	35	* mark->lock
5444e298 EP	36	* inode->i_lock
5444e298 EP	37	*
9756b918 LS	38	* group->mark_mutex protects the marks_list anchored inside a given group and
	39	* each mark is hooked via the g_list. It also protects the groups private
	40	* data (i.e group limits).
	41
	42	* mark->lock protects the marks attributes like its masks and flags.
	43	* Furthermore it protects the access to a reference of the group that the mark
	44	* is assigned to as well as the access to a reference of the inode/vfsmount
	45	* that is being watched by the mark.
5444e298 EP	46	*
	47	* inode->i_lock protects the i_fsnotify_marks list anchored inside a
	48	* given inode and each mark is hooked via the i_list. (and sorta the
	49	* free_i_list)
	50	*
	51	*
	52	* LIFETIME:
	53	* Inode marks survive between when they are added to an inode and when their
	54	* refcnt==0.
	55	*
	56	* The inode mark can be cleared for a number of different reasons including:
	57	* - The inode is unlinked for the last time. (fsnotify_inode_remove)
	58	* - The inode is being evicted from cache. (fsnotify_inode_delete)
	59	* - The fs the inode is on is unmounted. (fsnotify_inode_delete/fsnotify_unmount_inodes)
	60	* - Something explicitly requests that it be removed. (fsnotify_destroy_mark)
	61	* - The fsnotify_group associated with the mark is going away and all such marks
	62	* need to be cleaned up. (fsnotify_clear_marks_by_group)
	63	*
	64	* Worst case we are given an inode and need to clean up all the marks on that
	65	* inode. We take i_lock and walk the i_fsnotify_marks safely. For each
	66	* mark on the list we take a reference (so the mark can't disappear under us).
	67	* We remove that mark form the inode's list of marks and we add this mark to a
9756b918 LS	68	* private list anchored on the stack using i_free_list; we walk i_free_list
	69	* and before we destroy the mark we make sure that we dont race with a
	70	* concurrent destroy_group by getting a ref to the marks group and taking the
	71	* groups mutex.
	72
5444e298 EP	73	* Very similarly for freeing by group, except we use free_g_list.
	74	*
	75	* This has the very interesting property of being able to run concurrently with
	76	* any (or all) other directions.
	77	*/
	78
	79	#include <linux/fs.h>
	80	#include <linux/init.h>
	81	#include <linux/kernel.h>
75c1be48	82	#include <linux/kthread.h>
5444e298 EP	83	#include <linux/module.h>
	84	#include <linux/mutex.h>
	85	#include <linux/slab.h>
	86	#include <linux/spinlock.h>
75c1be48	87	#include <linux/srcu.h>
5444e298	88
60063497	89	#include <linux/atomic.h>
5444e298 EP	90
	91	#include <linux/fsnotify_backend.h>
	92	#include "fsnotify.h"
	93
75c1be48 EP	94	struct srcu_struct fsnotify_mark_srcu;
	95	static DEFINE_SPINLOCK(destroy_lock);
	96	static LIST_HEAD(destroy_list);
	97	static DECLARE_WAIT_QUEUE_HEAD(destroy_waitq);
	98
5444e298 EP	99	void fsnotify_get_mark(struct fsnotify_mark *mark)
	100	{
	101	atomic_inc(&mark->refcnt);
	102	}
	103
	104	void fsnotify_put_mark(struct fsnotify_mark *mark)
	105	{
23e964c2 LS	106	if (atomic_dec_and_test(&mark->refcnt)) {
	107	if (mark->group)
	108	fsnotify_put_group(mark->group);
5444e298	109	mark->free_mark(mark);
23e964c2	110	}
5444e298 EP	111	}
	112
	113	/*
	114	* Any time a mark is getting freed we end up here.
	115	* The caller had better be holding a reference to this mark so we don't actually
	116	* do the final put under the mark->lock
	117	*/
d5a335b8 LS	118	void fsnotify_destroy_mark_locked(struct fsnotify_mark *mark,
d5a335b8 LS	119	struct fsnotify_group *group)
5444e298	120	{
0d48b7f0	121	struct inode *inode = NULL;
5444e298	122
d5a335b8 LS	123	BUG_ON(!mutex_is_locked(&group->mark_mutex));
d5a335b8 LS	124
104d06f0	125	spin_lock(&mark->lock);
5444e298	126
700307a2 EP	127	/* something else already called this function on this mark */
700307a2 EP	128	if (!(mark->flags & FSNOTIFY_MARK_FLAG_ALIVE)) {
5444e298	129	spin_unlock(&mark->lock);
e2a29943	130	return;
5444e298 EP	131	}
5444e298 EP	132
700307a2 EP	133	mark->flags &= ~FSNOTIFY_MARK_FLAG_ALIVE;
700307a2 EP	134
0d48b7f0	135	if (mark->flags & FSNOTIFY_MARK_FLAG_INODE) {
0d48b7f0	136	inode = mark->i.inode;
b31d397e	137	fsnotify_destroy_inode_mark(mark);
0d48b7f0 EP	138	} else if (mark->flags & FSNOTIFY_MARK_FLAG_VFSMOUNT)
0d48b7f0 EP	139	fsnotify_destroy_vfsmount_mark(mark);
5444e298 EP	140	else
	141	BUG();
	142
	143	list_del_init(&mark->g_list);
5444e298	144
5444e298	145	spin_unlock(&mark->lock);
d5a335b8	146
6960b0d9 LS	147	if (inode && (mark->flags & FSNOTIFY_MARK_FLAG_OBJECT_PINNED))
6960b0d9 LS	148	iput(inode);
d5a335b8	149	/* release lock temporarily */
986ab098	150	mutex_unlock(&group->mark_mutex);
5444e298	151
75c1be48 EP	152	spin_lock(&destroy_lock);
	153	list_add(&mark->destroy_list, &destroy_list);
	154	spin_unlock(&destroy_lock);
	155	wake_up(&destroy_waitq);
6960b0d9 LS	156	/*
	157	* We don't necessarily have a ref on mark from caller so the above destroy
	158	* may have actually freed it, unless this group provides a 'freeing_mark'
	159	* function which must be holding a reference.
	160	*/
75c1be48	161
5444e298 EP	162	/*
	163	* Some groups like to know that marks are being freed. This is a
	164	* callback to the group function to let it know that this mark
	165	* is being freed.
	166	*/
	167	if (group->ops->freeing_mark)
	168	group->ops->freeing_mark(mark, group);
	169
	170	/*
	171	* __fsnotify_update_child_dentry_flags(inode);
	172	*
	173	* I really want to call that, but we can't, we have no idea if the inode
	174	* still exists the second we drop the mark->lock.
	175	*
	176	* The next time an event arrive to this inode from one of it's children
	177	* __fsnotify_parent will see that the inode doesn't care about it's
	178	* children and will update all of these flags then. So really this
	179	* is just a lazy update (and could be a perf win...)
	180	*/
	181
23e964c2	182	atomic_dec(&group->num_marks);
d5a335b8	183
6960b0d9	184	mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING);
d5a335b8 LS	185	}
	186
	187	void fsnotify_destroy_mark(struct fsnotify_mark *mark,
	188	struct fsnotify_group *group)
	189	{
6960b0d9	190	mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING);
d5a335b8 LS	191	fsnotify_destroy_mark_locked(mark, group);
d5a335b8 LS	192	mutex_unlock(&group->mark_mutex);
5444e298 EP	193	}
5444e298 EP	194
90b1e7a5 EP	195	void fsnotify_set_mark_mask_locked(struct fsnotify_mark *mark, __u32 mask)
	196	{
	197	assert_spin_locked(&mark->lock);
	198
	199	mark->mask = mask;
	200
	201	if (mark->flags & FSNOTIFY_MARK_FLAG_INODE)
	202	fsnotify_set_inode_mark_mask_locked(mark, mask);
	203	}
	204
33af5e32 EP	205	void fsnotify_set_mark_ignored_mask_locked(struct fsnotify_mark *mark, __u32 mask)
	206	{
	207	assert_spin_locked(&mark->lock);
	208
	209	mark->ignored_mask = mask;
	210	}
90b1e7a5	211
8edc6e16 JK	212	/*
	213	* Sorting function for lists of fsnotify marks.
	214	*
	215	* Fanotify supports different notification classes (reflected as priority of
	216	* notification group). Events shall be passed to notification groups in
	217	* decreasing priority order. To achieve this marks in notification lists for
	218	* inodes and vfsmounts are sorted so that priorities of corresponding groups
	219	* are descending.
	220	*
	221	* Furthermore correct handling of the ignore mask requires processing inode
	222	* and vfsmount marks of each group together. Using the group address as
	223	* further sort criterion provides a unique sorting order and thus we can
	224	* merge inode and vfsmount lists of marks in linear time and find groups
	225	* present in both lists.
	226	*
	227	* A return value of 1 signifies that b has priority over a.
	228	* A return value of 0 signifies that the two marks have to be handled together.
	229	* A return value of -1 signifies that a has priority over b.
	230	*/
	231	int fsnotify_compare_groups(struct fsnotify_group a, struct fsnotify_group b)
	232	{
	233	if (a == b)
	234	return 0;
	235	if (!a)
	236	return 1;
	237	if (!b)
	238	return -1;
	239	if (a->priority < b->priority)
	240	return 1;
	241	if (a->priority > b->priority)
	242	return -1;
	243	if (a < b)
	244	return 1;
	245	return -1;
	246	}
	247
5444e298 EP	248	/*
	249	* Attach an initialized mark to a given group and fs object.
	250	* These marks may be used for the fsnotify backend to determine which
	251	* event types should be delivered to which group.
	252	*/
d5a335b8 LS	253	int fsnotify_add_mark_locked(struct fsnotify_mark *mark,
	254	struct fsnotify_group group, struct inode inode,
	255	struct vfsmount *mnt, int allow_dups)
5444e298 EP	256	{
	257	int ret = 0;
	258
5444e298 EP	259	BUG_ON(inode && mnt);
5444e298 EP	260	BUG_ON(!inode && !mnt);
d5a335b8	261	BUG_ON(!mutex_is_locked(&group->mark_mutex));
5444e298	262
5444e298 EP	263	/*
5444e298 EP	264	* LOCKING ORDER!!!!
986ab098	265	* group->mark_mutex
104d06f0	266	* mark->lock
5444e298 EP	267	* inode->i_lock
5444e298 EP	268	*/
104d06f0	269	spin_lock(&mark->lock);
700307a2 EP	270	mark->flags \|= FSNOTIFY_MARK_FLAG_ALIVE;
700307a2 EP	271
23e964c2	272	fsnotify_get_group(group);
5444e298 EP	273	mark->group = group;
	274	list_add(&mark->g_list, &group->marks_list);
	275	atomic_inc(&group->num_marks);
	276	fsnotify_get_mark(mark); /* for i_list and g_list */
	277
	278	if (inode) {
	279	ret = fsnotify_add_inode_mark(mark, group, inode, allow_dups);
	280	if (ret)
	281	goto err;
0d48b7f0 EP	282	} else if (mnt) {
	283	ret = fsnotify_add_vfsmount_mark(mark, group, mnt, allow_dups);
	284	if (ret)
	285	goto err;
5444e298 EP	286	} else {
	287	BUG();
	288	}
	289
90b1e7a5 EP	290	/* this will pin the object if appropriate */
90b1e7a5 EP	291	fsnotify_set_mark_mask_locked(mark, mark->mask);
5444e298 EP	292	spin_unlock(&mark->lock);
	293
	294	if (inode)
	295	__fsnotify_update_child_dentry_flags(inode);
	296
	297	return ret;
	298	err:
700307a2	299	mark->flags &= ~FSNOTIFY_MARK_FLAG_ALIVE;
5444e298	300	list_del_init(&mark->g_list);
23e964c2	301	fsnotify_put_group(group);
75c1be48	302	mark->group = NULL;
5444e298	303	atomic_dec(&group->num_marks);
5444e298	304
5444e298 EP	305	spin_unlock(&mark->lock);
5444e298 EP	306
75c1be48 EP	307	spin_lock(&destroy_lock);
	308	list_add(&mark->destroy_list, &destroy_list);
	309	spin_unlock(&destroy_lock);
	310	wake_up(&destroy_waitq);
	311
5444e298 EP	312	return ret;
	313	}
	314
d5a335b8 LS	315	int fsnotify_add_mark(struct fsnotify_mark mark, struct fsnotify_group group,
	316	struct inode inode, struct vfsmount mnt, int allow_dups)
	317	{
	318	int ret;
	319	mutex_lock(&group->mark_mutex);
	320	ret = fsnotify_add_mark_locked(mark, group, inode, mnt, allow_dups);
	321	mutex_unlock(&group->mark_mutex);
	322	return ret;
	323	}
	324
5444e298	325	/*
4d92604c	326	* clear any marks in a group in which mark->flags & flags is true
5444e298	327	*/
4d92604c EP	328	void fsnotify_clear_marks_by_group_flags(struct fsnotify_group *group,
4d92604c EP	329	unsigned int flags)
5444e298 EP	330	{
5444e298 EP	331	struct fsnotify_mark lmark, mark;
5444e298	332
6960b0d9	333	mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING);
5444e298	334	list_for_each_entry_safe(mark, lmark, &group->marks_list, g_list) {
4d92604c	335	if (mark->flags & flags) {
4d92604c	336	fsnotify_get_mark(mark);
64c20d2a LS	337	fsnotify_destroy_mark_locked(mark, group);
64c20d2a LS	338	fsnotify_put_mark(mark);
4d92604c	339	}
5444e298	340	}
986ab098	341	mutex_unlock(&group->mark_mutex);
5444e298 EP	342	}
5444e298 EP	343
4d92604c EP	344	/*
	345	* Given a group, destroy all of the marks associated with that group.
	346	*/
	347	void fsnotify_clear_marks_by_group(struct fsnotify_group *group)
	348	{
	349	fsnotify_clear_marks_by_group_flags(group, (unsigned int)-1);
	350	}
	351
5444e298 EP	352	void fsnotify_duplicate_mark(struct fsnotify_mark new, struct fsnotify_mark old)
	353	{
	354	assert_spin_locked(&old->lock);
	355	new->i.inode = old->i.inode;
	356	new->m.mnt = old->m.mnt;
23e964c2 LS	357	if (old->group)
23e964c2 LS	358	fsnotify_get_group(old->group);
5444e298 EP	359	new->group = old->group;
	360	new->mask = old->mask;
	361	new->free_mark = old->free_mark;
	362	}
	363
	364	/*
	365	* Nothing fancy, just initialize lists and locks and counters.
	366	*/
	367	void fsnotify_init_mark(struct fsnotify_mark *mark,
	368	void (free_mark)(struct fsnotify_mark mark))
	369	{
ba643f04	370	memset(mark, 0, sizeof(*mark));
5444e298 EP	371	spin_lock_init(&mark->lock);
5444e298 EP	372	atomic_set(&mark->refcnt, 1);
5444e298 EP	373	mark->free_mark = free_mark;
5444e298 EP	374	}
75c1be48 EP	375
	376	static int fsnotify_mark_destroy(void *ignored)
	377	{
	378	struct fsnotify_mark mark, next;
efa8f7e5	379	struct list_head private_destroy_list;
75c1be48 EP	380
	381	for (;;) {
	382	spin_lock(&destroy_lock);
8778abb9 AG	383	/* exchange the list head */
8778abb9 AG	384	list_replace_init(&destroy_list, &private_destroy_list);
75c1be48 EP	385	spin_unlock(&destroy_lock);
	386
	387	synchronize_srcu(&fsnotify_mark_srcu);
	388
	389	list_for_each_entry_safe(mark, next, &private_destroy_list, destroy_list) {
	390	list_del_init(&mark->destroy_list);
	391	fsnotify_put_mark(mark);
	392	}
	393
	394	wait_event_interruptible(destroy_waitq, !list_empty(&destroy_list));
	395	}
	396
	397	return 0;
	398	}
	399
	400	static int __init fsnotify_mark_init(void)
	401	{
	402	struct task_struct *thread;
	403
	404	thread = kthread_run(fsnotify_mark_destroy, NULL,
	405	"fsnotify_mark");
	406	if (IS_ERR(thread))
	407	panic("unable to start fsnotify mark destruction thread.");
	408
	409	return 0;
	410	}
	411	device_initcall(fsnotify_mark_init);