summaryrefslogtreecommitdiff
path: root/fs/notify/inode_mark.c
blob: 7e69f6b08d4e375813dd37f43e83578a0034880f (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
/*
 *  Copyright (C) 2008 Red Hat, Inc., Eric Paris <eparis@redhat.com>
 *
 *  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 mark->refcnt tells how many "things" in the kernel currently are
 * referencing this object.  The object typically will live inside the kernel
 * with a refcnt of 2, one for each list it is on (i_list, g_list).  Any task
 * which can find this object holding the appropriete locks, can take a reference
 * and the object itself is guarenteed to survive until the reference is dropped.
 *
 * LOCKING:
 * There are 3 spinlocks involved with fsnotify inode marks and they MUST
 * be taken in order as follows:
 *
 * entry->lock
 * group->mark_lock
 * inode->i_lock
 *
 * entry->lock protects 2 things, entry->group and entry->inode.  You must hold
 * that lock to dereference either of these things (they could be NULL even with
 * the lock)
 *
 * group->mark_lock protects the marks_list anchored inside a given group
 * and each entry is hooked via the g_list.  It also sorta protects the
 * free_g_list, which when used is anchored by a private list on the stack of the
 * task which held the group->mark_lock.
 *
 * inode->i_lock protects the i_fsnotify_marks list anchored inside a
 * given inode and each entry 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_by_entry)
 * - 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;  At this point we no
 * longer fear anything finding the mark using the inode's list of marks.
 *
 * We can safely and locklessly run the private list on the stack of everything
 * we just unattached from the original inode.  For each mark on the private list
 * we grab the mark-> and can thus dereference mark->group and mark->inode.  If
 * we see the group and inode are not NULL we take those locks.  Now holding all
 * 3 locks we can completely remove the mark from other tasks finding it in the
 * future.  Remember, 10 things might already be referencing this mark, but they
 * better be holding a ref.  We drop our reference we took before we unhooked it
 * from the inode.  When the ref hits 0 we can free the mark.
 *
 * 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/module.h>
#include <linux/mutex.h>
#include <linux/spinlock.h>
#include <linux/writeback.h> /* for inode_lock */

#include <asm/atomic.h>

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

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

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

/*
 * Recalculate the mask of events relevant to a given inode locked.
 */
static void fsnotify_recalc_inode_mask_locked(struct inode *inode)
{
	struct fsnotify_mark *entry;
	struct hlist_node *pos;
	__u32 new_mask = 0;

	assert_spin_locked(&inode->i_lock);

	hlist_for_each_entry(entry, pos, &inode->i_fsnotify_marks, i.i_list)
		new_mask |= entry->mask;
	inode->i_fsnotify_mask = new_mask;
}

/*
 * Recalculate the inode->i_fsnotify_mask, or the mask of all FS_* event types
 * any notifier is interested in hearing for this inode.
 */
void fsnotify_recalc_inode_mask(struct inode *inode)
{
	spin_lock(&inode->i_lock);
	fsnotify_recalc_inode_mask_locked(inode);
	spin_unlock(&inode->i_lock);

	__fsnotify_update_child_dentry_flags(inode);
}

/*
 * 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 entry->lock
 */
void fsnotify_destroy_mark_by_entry(struct fsnotify_mark *entry)
{
	struct fsnotify_group *group;
	struct inode *inode;

	spin_lock(&entry->lock);

	group = entry->group;
	inode = entry->i.inode;

	BUG_ON(group && !inode);
	BUG_ON(!group && inode);

	/* if !group something else already marked this to die */
	if (!group) {
		spin_unlock(&entry->lock);
		return;
	}

	/* 1 from caller and 1 for being on i_list/g_list */
	BUG_ON(atomic_read(&entry->refcnt) < 2);

	spin_lock(&group->mark_lock);
	spin_lock(&inode->i_lock);

	hlist_del_init(&entry->i.i_list);
	entry->i.inode = NULL;

	list_del_init(&entry->g_list);
	entry->group = NULL;

	fsnotify_put_mark(entry); /* for i_list and g_list */

	/*
	 * this mark is now off the inode->i_fsnotify_marks list and we
	 * hold the inode->i_lock, so this is the perfect time to update the
	 * inode->i_fsnotify_mask
	 */
	fsnotify_recalc_inode_mask_locked(inode);

	spin_unlock(&inode->i_lock);
	spin_unlock(&group->mark_lock);
	spin_unlock(&entry->lock);

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


	iput(inode);

	/*
	 * it's possible that this group tried to destroy itself, but this
	 * this mark was simultaneously being freed by inode.  If that's the
	 * case, we finish freeing the group here.
	 */
	if (unlikely(atomic_dec_and_test(&group->num_marks)))
		fsnotify_final_destroy_group(group);
}

/*
 * Given a group, destroy all of the marks associated with that group.
 */
void fsnotify_clear_marks_by_group(struct fsnotify_group *group)
{
	struct fsnotify_mark *lentry, *entry;
	LIST_HEAD(free_list);

	spin_lock(&group->mark_lock);
	list_for_each_entry_safe(entry, lentry, &group->marks_list, g_list) {
		list_add(&entry->free_g_list, &free_list);
		list_del_init(&entry->g_list);
		fsnotify_get_mark(entry);
	}
	spin_unlock(&group->mark_lock);

	list_for_each_entry_safe(entry, lentry, &free_list, free_g_list) {
		fsnotify_destroy_mark_by_entry(entry);
		fsnotify_put_mark(entry);
	}
}

/*
 * Given an inode, destroy all of the marks associated with that inode.
 */
void fsnotify_clear_marks_by_inode(struct inode *inode)
{
	struct fsnotify_mark *entry, *lentry;
	struct hlist_node *pos, *n;
	LIST_HEAD(free_list);

	spin_lock(&inode->i_lock);
	hlist_for_each_entry_safe(entry, pos, n, &inode->i_fsnotify_marks, i.i_list) {
		list_add(&entry->i.free_i_list, &free_list);
		hlist_del_init(&entry->i.i_list);
		fsnotify_get_mark(entry);
	}
	spin_unlock(&inode->i_lock);

	list_for_each_entry_safe(entry, lentry, &free_list, i.free_i_list) {
		fsnotify_destroy_mark_by_entry(entry);
		fsnotify_put_mark(entry);
	}
}

/*
 * given a group and inode, find the mark associated with that combination.
 * if found take a reference to that mark and return it, else return NULL
 */
struct fsnotify_mark *fsnotify_find_mark_entry(struct fsnotify_group *group,
					       struct inode *inode)
{
	struct fsnotify_mark *entry;
	struct hlist_node *pos;

	assert_spin_locked(&inode->i_lock);

	hlist_for_each_entry(entry, pos, &inode->i_fsnotify_marks, i.i_list) {
		if (entry->group == group) {
			fsnotify_get_mark(entry);
			return entry;
		}
	}
	return NULL;
}

void fsnotify_duplicate_mark(struct fsnotify_mark *new, struct fsnotify_mark *old)
{
	assert_spin_locked(&old->lock);
	new->i.inode = old->i.inode;
	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 *entry,
			void (*free_mark)(struct fsnotify_mark *entry))
{
	spin_lock_init(&entry->lock);
	atomic_set(&entry->refcnt, 1);
	INIT_HLIST_NODE(&entry->i.i_list);
	entry->group = NULL;
	entry->mask = 0;
	entry->i.inode = NULL;
	entry->free_mark = free_mark;
}

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

	inode = igrab(inode);
	if (unlikely(!inode))
		return -EINVAL;

	entry->flags = FSNOTIFY_MARK_FLAG_INODE;

	/*
	 * if this group isn't being testing for inode type events we need
	 * to start testing
	 */
	if (unlikely(list_empty(&group->inode_group_list)))
		fsnotify_add_inode_group(group);
	/*
	 * XXX This is where we could also do the fsnotify_add_vfsmount_group
	 * if we are setting and vfsmount mark....

	if (unlikely(list_empty(&group->vfsmount_group_list)))
		fsnotify_add_vfsmount_group(group);
	 */

	/*
	 * LOCKING ORDER!!!!
	 * entry->lock
	 * group->mark_lock
	 * inode->i_lock
	 */
	spin_lock(&entry->lock);
	spin_lock(&group->mark_lock);
	spin_lock(&inode->i_lock);

	if (!allow_dups)
		lentry = fsnotify_find_mark_entry(group, inode);
	if (!lentry) {
		entry->group = group;
		entry->i.inode = inode;

		hlist_add_head(&entry->i.i_list, &inode->i_fsnotify_marks);
		list_add(&entry->g_list, &group->marks_list);

		fsnotify_get_mark(entry); /* for i_list and g_list */

		atomic_inc(&group->num_marks);

		fsnotify_recalc_inode_mask_locked(inode);
	}

	spin_unlock(&inode->i_lock);
	spin_unlock(&group->mark_lock);
	spin_unlock(&entry->lock);

	if (lentry) {
		ret = -EEXIST;
		iput(inode);
		fsnotify_put_mark(lentry);
	} else {
		__fsnotify_update_child_dentry_flags(inode);
	}

	return ret;
}

/**
 * fsnotify_unmount_inodes - an sb is unmounting.  handle any watched inodes.
 * @list: list of inodes being unmounted (sb->s_inodes)
 *
 * Called with inode_lock held, protecting the unmounting super block's list
 * of inodes, and with iprune_mutex held, keeping shrink_icache_memory() at bay.
 * We temporarily drop inode_lock, however, and CAN block.
 */
void fsnotify_unmount_inodes(struct list_head *list)
{
	struct inode *inode, *next_i, *need_iput = NULL;

	list_for_each_entry_safe(inode, next_i, list, i_sb_list) {
		struct inode *need_iput_tmp;

		/*
		 * We cannot __iget() an inode in state I_CLEAR, I_FREEING,
		 * I_WILL_FREE, or I_NEW which is fine because by that point
		 * the inode cannot have any associated watches.
		 */
		if (inode->i_state & (I_CLEAR|I_FREEING|I_WILL_FREE|I_NEW))
			continue;

		/*
		 * If i_count is zero, the inode cannot have any watches and
		 * doing an __iget/iput with MS_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))
			continue;

		need_iput_tmp = need_iput;
		need_iput = NULL;

		/* In case fsnotify_inode_delete() drops a reference. */
		if (inode != need_iput_tmp)
			__iget(inode);
		else
			need_iput_tmp = NULL;

		/* In case the dropping of a reference would nuke next_i. */
		if ((&next_i->i_sb_list != list) &&
		    atomic_read(&next_i->i_count) &&
		    !(next_i->i_state & (I_CLEAR | I_FREEING | I_WILL_FREE))) {
			__iget(next_i);
			need_iput = next_i;
		}

		/*
		 * We can safely drop inode_lock here because we hold
		 * references on both inode and next_i.  Also no new inodes
		 * will be added since the umount has begun.  Finally,
		 * iprune_mutex keeps shrink_icache_memory() away.
		 */
		spin_unlock(&inode_lock);

		if (need_iput_tmp)
			iput(need_iput_tmp);

		/* 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);

		spin_lock(&inode_lock);
	}
}