xref: /openbmc/linux/fs/notify/mark.c (revision 4f6cce39)
1 /*
2  *  Copyright (C) 2008 Red Hat, Inc., Eric Paris <eparis@redhat.com>
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:
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.
29  *
30  * LOCKING:
31  * There are 3 locks involved with fsnotify inode marks and they MUST be taken
32  * in order as follows:
33  *
34  * group->mark_mutex
35  * mark->lock
36  * inode->i_lock
37  *
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.
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
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 
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>
82 #include <linux/kthread.h>
83 #include <linux/module.h>
84 #include <linux/mutex.h>
85 #include <linux/slab.h>
86 #include <linux/spinlock.h>
87 #include <linux/srcu.h>
88 
89 #include <linux/atomic.h>
90 
91 #include <linux/fsnotify_backend.h>
92 #include "fsnotify.h"
93 
94 #define FSNOTIFY_REAPER_DELAY	(1)	/* 1 jiffy */
95 
96 struct srcu_struct fsnotify_mark_srcu;
97 static DEFINE_SPINLOCK(destroy_lock);
98 static LIST_HEAD(destroy_list);
99 
100 static void fsnotify_mark_destroy_workfn(struct work_struct *work);
101 static DECLARE_DELAYED_WORK(reaper_work, fsnotify_mark_destroy_workfn);
102 
103 void fsnotify_get_mark(struct fsnotify_mark *mark)
104 {
105 	atomic_inc(&mark->refcnt);
106 }
107 
108 void fsnotify_put_mark(struct fsnotify_mark *mark)
109 {
110 	if (atomic_dec_and_test(&mark->refcnt)) {
111 		if (mark->group)
112 			fsnotify_put_group(mark->group);
113 		mark->free_mark(mark);
114 	}
115 }
116 
117 /* Calculate mask of events for a list of marks */
118 u32 fsnotify_recalc_mask(struct hlist_head *head)
119 {
120 	u32 new_mask = 0;
121 	struct fsnotify_mark *mark;
122 
123 	hlist_for_each_entry(mark, head, obj_list)
124 		new_mask |= mark->mask;
125 	return new_mask;
126 }
127 
128 /*
129  * Remove mark from inode / vfsmount list, group list, drop inode reference
130  * if we got one.
131  *
132  * Must be called with group->mark_mutex held.
133  */
134 void fsnotify_detach_mark(struct fsnotify_mark *mark)
135 {
136 	struct inode *inode = NULL;
137 	struct fsnotify_group *group = mark->group;
138 
139 	BUG_ON(!mutex_is_locked(&group->mark_mutex));
140 
141 	spin_lock(&mark->lock);
142 
143 	/* something else already called this function on this mark */
144 	if (!(mark->flags & FSNOTIFY_MARK_FLAG_ATTACHED)) {
145 		spin_unlock(&mark->lock);
146 		return;
147 	}
148 
149 	mark->flags &= ~FSNOTIFY_MARK_FLAG_ATTACHED;
150 
151 	if (mark->flags & FSNOTIFY_MARK_FLAG_INODE) {
152 		inode = mark->inode;
153 		fsnotify_destroy_inode_mark(mark);
154 	} else if (mark->flags & FSNOTIFY_MARK_FLAG_VFSMOUNT)
155 		fsnotify_destroy_vfsmount_mark(mark);
156 	else
157 		BUG();
158 	/*
159 	 * Note that we didn't update flags telling whether inode cares about
160 	 * what's happening with children. We update these flags from
161 	 * __fsnotify_parent() lazily when next event happens on one of our
162 	 * children.
163 	 */
164 
165 	list_del_init(&mark->g_list);
166 
167 	spin_unlock(&mark->lock);
168 
169 	if (inode && (mark->flags & FSNOTIFY_MARK_FLAG_OBJECT_PINNED))
170 		iput(inode);
171 
172 	atomic_dec(&group->num_marks);
173 }
174 
175 /*
176  * Prepare mark for freeing and add it to the list of marks prepared for
177  * freeing. The actual freeing must happen after SRCU period ends and the
178  * caller is responsible for this.
179  *
180  * The function returns true if the mark was added to the list of marks for
181  * freeing. The function returns false if someone else has already called
182  * __fsnotify_free_mark() for the mark.
183  */
184 static bool __fsnotify_free_mark(struct fsnotify_mark *mark)
185 {
186 	struct fsnotify_group *group = mark->group;
187 
188 	spin_lock(&mark->lock);
189 	/* something else already called this function on this mark */
190 	if (!(mark->flags & FSNOTIFY_MARK_FLAG_ALIVE)) {
191 		spin_unlock(&mark->lock);
192 		return false;
193 	}
194 	mark->flags &= ~FSNOTIFY_MARK_FLAG_ALIVE;
195 	spin_unlock(&mark->lock);
196 
197 	/*
198 	 * Some groups like to know that marks are being freed.  This is a
199 	 * callback to the group function to let it know that this mark
200 	 * is being freed.
201 	 */
202 	if (group->ops->freeing_mark)
203 		group->ops->freeing_mark(mark, group);
204 
205 	spin_lock(&destroy_lock);
206 	list_add(&mark->g_list, &destroy_list);
207 	spin_unlock(&destroy_lock);
208 
209 	return true;
210 }
211 
212 /*
213  * Free fsnotify mark. The freeing is actually happening from a workqueue which
214  * first waits for srcu period end. Caller must have a reference to the mark
215  * or be protected by fsnotify_mark_srcu.
216  */
217 void fsnotify_free_mark(struct fsnotify_mark *mark)
218 {
219 	if (__fsnotify_free_mark(mark)) {
220 		queue_delayed_work(system_unbound_wq, &reaper_work,
221 				   FSNOTIFY_REAPER_DELAY);
222 	}
223 }
224 
225 void fsnotify_destroy_mark(struct fsnotify_mark *mark,
226 			   struct fsnotify_group *group)
227 {
228 	mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING);
229 	fsnotify_detach_mark(mark);
230 	mutex_unlock(&group->mark_mutex);
231 	fsnotify_free_mark(mark);
232 }
233 
234 void fsnotify_destroy_marks(struct hlist_head *head, spinlock_t *lock)
235 {
236 	struct fsnotify_mark *mark;
237 
238 	while (1) {
239 		/*
240 		 * We have to be careful since we can race with e.g.
241 		 * fsnotify_clear_marks_by_group() and once we drop 'lock',
242 		 * mark can get removed from the obj_list and destroyed. But
243 		 * we are holding mark reference so mark cannot be freed and
244 		 * calling fsnotify_destroy_mark() more than once is fine.
245 		 */
246 		spin_lock(lock);
247 		if (hlist_empty(head)) {
248 			spin_unlock(lock);
249 			break;
250 		}
251 		mark = hlist_entry(head->first, struct fsnotify_mark, obj_list);
252 		/*
253 		 * We don't update i_fsnotify_mask / mnt_fsnotify_mask here
254 		 * since inode / mount is going away anyway. So just remove
255 		 * mark from the list.
256 		 */
257 		hlist_del_init_rcu(&mark->obj_list);
258 		fsnotify_get_mark(mark);
259 		spin_unlock(lock);
260 		fsnotify_destroy_mark(mark, mark->group);
261 		fsnotify_put_mark(mark);
262 	}
263 }
264 
265 void fsnotify_set_mark_mask_locked(struct fsnotify_mark *mark, __u32 mask)
266 {
267 	assert_spin_locked(&mark->lock);
268 
269 	mark->mask = mask;
270 
271 	if (mark->flags & FSNOTIFY_MARK_FLAG_INODE)
272 		fsnotify_set_inode_mark_mask_locked(mark, mask);
273 }
274 
275 void fsnotify_set_mark_ignored_mask_locked(struct fsnotify_mark *mark, __u32 mask)
276 {
277 	assert_spin_locked(&mark->lock);
278 
279 	mark->ignored_mask = mask;
280 }
281 
282 /*
283  * Sorting function for lists of fsnotify marks.
284  *
285  * Fanotify supports different notification classes (reflected as priority of
286  * notification group). Events shall be passed to notification groups in
287  * decreasing priority order. To achieve this marks in notification lists for
288  * inodes and vfsmounts are sorted so that priorities of corresponding groups
289  * are descending.
290  *
291  * Furthermore correct handling of the ignore mask requires processing inode
292  * and vfsmount marks of each group together. Using the group address as
293  * further sort criterion provides a unique sorting order and thus we can
294  * merge inode and vfsmount lists of marks in linear time and find groups
295  * present in both lists.
296  *
297  * A return value of 1 signifies that b has priority over a.
298  * A return value of 0 signifies that the two marks have to be handled together.
299  * A return value of -1 signifies that a has priority over b.
300  */
301 int fsnotify_compare_groups(struct fsnotify_group *a, struct fsnotify_group *b)
302 {
303 	if (a == b)
304 		return 0;
305 	if (!a)
306 		return 1;
307 	if (!b)
308 		return -1;
309 	if (a->priority < b->priority)
310 		return 1;
311 	if (a->priority > b->priority)
312 		return -1;
313 	if (a < b)
314 		return 1;
315 	return -1;
316 }
317 
318 /* Add mark into proper place in given list of marks */
319 int fsnotify_add_mark_list(struct hlist_head *head, struct fsnotify_mark *mark,
320 			   int allow_dups)
321 {
322 	struct fsnotify_mark *lmark, *last = NULL;
323 	int cmp;
324 
325 	/* is mark the first mark? */
326 	if (hlist_empty(head)) {
327 		hlist_add_head_rcu(&mark->obj_list, head);
328 		return 0;
329 	}
330 
331 	/* should mark be in the middle of the current list? */
332 	hlist_for_each_entry(lmark, head, obj_list) {
333 		last = lmark;
334 
335 		if ((lmark->group == mark->group) && !allow_dups)
336 			return -EEXIST;
337 
338 		cmp = fsnotify_compare_groups(lmark->group, mark->group);
339 		if (cmp >= 0) {
340 			hlist_add_before_rcu(&mark->obj_list, &lmark->obj_list);
341 			return 0;
342 		}
343 	}
344 
345 	BUG_ON(last == NULL);
346 	/* mark should be the last entry.  last is the current last entry */
347 	hlist_add_behind_rcu(&mark->obj_list, &last->obj_list);
348 	return 0;
349 }
350 
351 /*
352  * Attach an initialized mark to a given group and fs object.
353  * These marks may be used for the fsnotify backend to determine which
354  * event types should be delivered to which group.
355  */
356 int fsnotify_add_mark_locked(struct fsnotify_mark *mark,
357 			     struct fsnotify_group *group, struct inode *inode,
358 			     struct vfsmount *mnt, int allow_dups)
359 {
360 	int ret = 0;
361 
362 	BUG_ON(inode && mnt);
363 	BUG_ON(!inode && !mnt);
364 	BUG_ON(!mutex_is_locked(&group->mark_mutex));
365 
366 	/*
367 	 * LOCKING ORDER!!!!
368 	 * group->mark_mutex
369 	 * mark->lock
370 	 * inode->i_lock
371 	 */
372 	spin_lock(&mark->lock);
373 	mark->flags |= FSNOTIFY_MARK_FLAG_ALIVE | FSNOTIFY_MARK_FLAG_ATTACHED;
374 
375 	fsnotify_get_group(group);
376 	mark->group = group;
377 	list_add(&mark->g_list, &group->marks_list);
378 	atomic_inc(&group->num_marks);
379 	fsnotify_get_mark(mark); /* for i_list and g_list */
380 
381 	if (inode) {
382 		ret = fsnotify_add_inode_mark(mark, group, inode, allow_dups);
383 		if (ret)
384 			goto err;
385 	} else if (mnt) {
386 		ret = fsnotify_add_vfsmount_mark(mark, group, mnt, allow_dups);
387 		if (ret)
388 			goto err;
389 	} else {
390 		BUG();
391 	}
392 
393 	/* this will pin the object if appropriate */
394 	fsnotify_set_mark_mask_locked(mark, mark->mask);
395 	spin_unlock(&mark->lock);
396 
397 	if (inode)
398 		__fsnotify_update_child_dentry_flags(inode);
399 
400 	return ret;
401 err:
402 	mark->flags &= ~FSNOTIFY_MARK_FLAG_ALIVE;
403 	list_del_init(&mark->g_list);
404 	fsnotify_put_group(group);
405 	mark->group = NULL;
406 	atomic_dec(&group->num_marks);
407 
408 	spin_unlock(&mark->lock);
409 
410 	spin_lock(&destroy_lock);
411 	list_add(&mark->g_list, &destroy_list);
412 	spin_unlock(&destroy_lock);
413 	queue_delayed_work(system_unbound_wq, &reaper_work,
414 				FSNOTIFY_REAPER_DELAY);
415 
416 	return ret;
417 }
418 
419 int fsnotify_add_mark(struct fsnotify_mark *mark, struct fsnotify_group *group,
420 		      struct inode *inode, struct vfsmount *mnt, int allow_dups)
421 {
422 	int ret;
423 	mutex_lock(&group->mark_mutex);
424 	ret = fsnotify_add_mark_locked(mark, group, inode, mnt, allow_dups);
425 	mutex_unlock(&group->mark_mutex);
426 	return ret;
427 }
428 
429 /*
430  * Given a list of marks, find the mark associated with given group. If found
431  * take a reference to that mark and return it, else return NULL.
432  */
433 struct fsnotify_mark *fsnotify_find_mark(struct hlist_head *head,
434 					 struct fsnotify_group *group)
435 {
436 	struct fsnotify_mark *mark;
437 
438 	hlist_for_each_entry(mark, head, obj_list) {
439 		if (mark->group == group) {
440 			fsnotify_get_mark(mark);
441 			return mark;
442 		}
443 	}
444 	return NULL;
445 }
446 
447 /*
448  * clear any marks in a group in which mark->flags & flags is true
449  */
450 void fsnotify_clear_marks_by_group_flags(struct fsnotify_group *group,
451 					 unsigned int flags)
452 {
453 	struct fsnotify_mark *lmark, *mark;
454 	LIST_HEAD(to_free);
455 
456 	/*
457 	 * We have to be really careful here. Anytime we drop mark_mutex, e.g.
458 	 * fsnotify_clear_marks_by_inode() can come and free marks. Even in our
459 	 * to_free list so we have to use mark_mutex even when accessing that
460 	 * list. And freeing mark requires us to drop mark_mutex. So we can
461 	 * reliably free only the first mark in the list. That's why we first
462 	 * move marks to free to to_free list in one go and then free marks in
463 	 * to_free list one by one.
464 	 */
465 	mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING);
466 	list_for_each_entry_safe(mark, lmark, &group->marks_list, g_list) {
467 		if (mark->flags & flags)
468 			list_move(&mark->g_list, &to_free);
469 	}
470 	mutex_unlock(&group->mark_mutex);
471 
472 	while (1) {
473 		mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING);
474 		if (list_empty(&to_free)) {
475 			mutex_unlock(&group->mark_mutex);
476 			break;
477 		}
478 		mark = list_first_entry(&to_free, struct fsnotify_mark, g_list);
479 		fsnotify_get_mark(mark);
480 		fsnotify_detach_mark(mark);
481 		mutex_unlock(&group->mark_mutex);
482 		fsnotify_free_mark(mark);
483 		fsnotify_put_mark(mark);
484 	}
485 }
486 
487 /*
488  * Given a group, prepare for freeing all the marks associated with that group.
489  * The marks are attached to the list of marks prepared for destruction, the
490  * caller is responsible for freeing marks in that list after SRCU period has
491  * ended.
492  */
493 void fsnotify_detach_group_marks(struct fsnotify_group *group)
494 {
495 	struct fsnotify_mark *mark;
496 
497 	while (1) {
498 		mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING);
499 		if (list_empty(&group->marks_list)) {
500 			mutex_unlock(&group->mark_mutex);
501 			break;
502 		}
503 		mark = list_first_entry(&group->marks_list,
504 					struct fsnotify_mark, g_list);
505 		fsnotify_get_mark(mark);
506 		fsnotify_detach_mark(mark);
507 		mutex_unlock(&group->mark_mutex);
508 		__fsnotify_free_mark(mark);
509 		fsnotify_put_mark(mark);
510 	}
511 }
512 
513 /*
514  * Nothing fancy, just initialize lists and locks and counters.
515  */
516 void fsnotify_init_mark(struct fsnotify_mark *mark,
517 			void (*free_mark)(struct fsnotify_mark *mark))
518 {
519 	memset(mark, 0, sizeof(*mark));
520 	spin_lock_init(&mark->lock);
521 	atomic_set(&mark->refcnt, 1);
522 	mark->free_mark = free_mark;
523 }
524 
525 /*
526  * Destroy all marks in destroy_list, waits for SRCU period to finish before
527  * actually freeing marks.
528  */
529 void fsnotify_mark_destroy_list(void)
530 {
531 	struct fsnotify_mark *mark, *next;
532 	struct list_head private_destroy_list;
533 
534 	spin_lock(&destroy_lock);
535 	/* exchange the list head */
536 	list_replace_init(&destroy_list, &private_destroy_list);
537 	spin_unlock(&destroy_lock);
538 
539 	synchronize_srcu(&fsnotify_mark_srcu);
540 
541 	list_for_each_entry_safe(mark, next, &private_destroy_list, g_list) {
542 		list_del_init(&mark->g_list);
543 		fsnotify_put_mark(mark);
544 	}
545 }
546 
547 static void fsnotify_mark_destroy_workfn(struct work_struct *work)
548 {
549 	fsnotify_mark_destroy_list();
550 }
551