xref: /openbmc/linux/kernel/audit_tree.c (revision 93d90ad7)
1 #include "audit.h"
2 #include <linux/fsnotify_backend.h>
3 #include <linux/namei.h>
4 #include <linux/mount.h>
5 #include <linux/kthread.h>
6 #include <linux/slab.h>
7 
8 struct audit_tree;
9 struct audit_chunk;
10 
11 struct audit_tree {
12 	atomic_t count;
13 	int goner;
14 	struct audit_chunk *root;
15 	struct list_head chunks;
16 	struct list_head rules;
17 	struct list_head list;
18 	struct list_head same_root;
19 	struct rcu_head head;
20 	char pathname[];
21 };
22 
23 struct audit_chunk {
24 	struct list_head hash;
25 	struct fsnotify_mark mark;
26 	struct list_head trees;		/* with root here */
27 	int dead;
28 	int count;
29 	atomic_long_t refs;
30 	struct rcu_head head;
31 	struct node {
32 		struct list_head list;
33 		struct audit_tree *owner;
34 		unsigned index;		/* index; upper bit indicates 'will prune' */
35 	} owners[];
36 };
37 
38 static LIST_HEAD(tree_list);
39 static LIST_HEAD(prune_list);
40 
41 /*
42  * One struct chunk is attached to each inode of interest.
43  * We replace struct chunk on tagging/untagging.
44  * Rules have pointer to struct audit_tree.
45  * Rules have struct list_head rlist forming a list of rules over
46  * the same tree.
47  * References to struct chunk are collected at audit_inode{,_child}()
48  * time and used in AUDIT_TREE rule matching.
49  * These references are dropped at the same time we are calling
50  * audit_free_names(), etc.
51  *
52  * Cyclic lists galore:
53  * tree.chunks anchors chunk.owners[].list			hash_lock
54  * tree.rules anchors rule.rlist				audit_filter_mutex
55  * chunk.trees anchors tree.same_root				hash_lock
56  * chunk.hash is a hash with middle bits of watch.inode as
57  * a hash function.						RCU, hash_lock
58  *
59  * tree is refcounted; one reference for "some rules on rules_list refer to
60  * it", one for each chunk with pointer to it.
61  *
62  * chunk is refcounted by embedded fsnotify_mark + .refs (non-zero refcount
63  * of watch contributes 1 to .refs).
64  *
65  * node.index allows to get from node.list to containing chunk.
66  * MSB of that sucker is stolen to mark taggings that we might have to
67  * revert - several operations have very unpleasant cleanup logics and
68  * that makes a difference.  Some.
69  */
70 
71 static struct fsnotify_group *audit_tree_group;
72 
73 static struct audit_tree *alloc_tree(const char *s)
74 {
75 	struct audit_tree *tree;
76 
77 	tree = kmalloc(sizeof(struct audit_tree) + strlen(s) + 1, GFP_KERNEL);
78 	if (tree) {
79 		atomic_set(&tree->count, 1);
80 		tree->goner = 0;
81 		INIT_LIST_HEAD(&tree->chunks);
82 		INIT_LIST_HEAD(&tree->rules);
83 		INIT_LIST_HEAD(&tree->list);
84 		INIT_LIST_HEAD(&tree->same_root);
85 		tree->root = NULL;
86 		strcpy(tree->pathname, s);
87 	}
88 	return tree;
89 }
90 
91 static inline void get_tree(struct audit_tree *tree)
92 {
93 	atomic_inc(&tree->count);
94 }
95 
96 static inline void put_tree(struct audit_tree *tree)
97 {
98 	if (atomic_dec_and_test(&tree->count))
99 		kfree_rcu(tree, head);
100 }
101 
102 /* to avoid bringing the entire thing in audit.h */
103 const char *audit_tree_path(struct audit_tree *tree)
104 {
105 	return tree->pathname;
106 }
107 
108 static void free_chunk(struct audit_chunk *chunk)
109 {
110 	int i;
111 
112 	for (i = 0; i < chunk->count; i++) {
113 		if (chunk->owners[i].owner)
114 			put_tree(chunk->owners[i].owner);
115 	}
116 	kfree(chunk);
117 }
118 
119 void audit_put_chunk(struct audit_chunk *chunk)
120 {
121 	if (atomic_long_dec_and_test(&chunk->refs))
122 		free_chunk(chunk);
123 }
124 
125 static void __put_chunk(struct rcu_head *rcu)
126 {
127 	struct audit_chunk *chunk = container_of(rcu, struct audit_chunk, head);
128 	audit_put_chunk(chunk);
129 }
130 
131 static void audit_tree_destroy_watch(struct fsnotify_mark *entry)
132 {
133 	struct audit_chunk *chunk = container_of(entry, struct audit_chunk, mark);
134 	call_rcu(&chunk->head, __put_chunk);
135 }
136 
137 static struct audit_chunk *alloc_chunk(int count)
138 {
139 	struct audit_chunk *chunk;
140 	size_t size;
141 	int i;
142 
143 	size = offsetof(struct audit_chunk, owners) + count * sizeof(struct node);
144 	chunk = kzalloc(size, GFP_KERNEL);
145 	if (!chunk)
146 		return NULL;
147 
148 	INIT_LIST_HEAD(&chunk->hash);
149 	INIT_LIST_HEAD(&chunk->trees);
150 	chunk->count = count;
151 	atomic_long_set(&chunk->refs, 1);
152 	for (i = 0; i < count; i++) {
153 		INIT_LIST_HEAD(&chunk->owners[i].list);
154 		chunk->owners[i].index = i;
155 	}
156 	fsnotify_init_mark(&chunk->mark, audit_tree_destroy_watch);
157 	chunk->mark.mask = FS_IN_IGNORED;
158 	return chunk;
159 }
160 
161 enum {HASH_SIZE = 128};
162 static struct list_head chunk_hash_heads[HASH_SIZE];
163 static __cacheline_aligned_in_smp DEFINE_SPINLOCK(hash_lock);
164 
165 static inline struct list_head *chunk_hash(const struct inode *inode)
166 {
167 	unsigned long n = (unsigned long)inode / L1_CACHE_BYTES;
168 	return chunk_hash_heads + n % HASH_SIZE;
169 }
170 
171 /* hash_lock & entry->lock is held by caller */
172 static void insert_hash(struct audit_chunk *chunk)
173 {
174 	struct fsnotify_mark *entry = &chunk->mark;
175 	struct list_head *list;
176 
177 	if (!entry->inode)
178 		return;
179 	list = chunk_hash(entry->inode);
180 	list_add_rcu(&chunk->hash, list);
181 }
182 
183 /* called under rcu_read_lock */
184 struct audit_chunk *audit_tree_lookup(const struct inode *inode)
185 {
186 	struct list_head *list = chunk_hash(inode);
187 	struct audit_chunk *p;
188 
189 	list_for_each_entry_rcu(p, list, hash) {
190 		/* mark.inode may have gone NULL, but who cares? */
191 		if (p->mark.inode == inode) {
192 			atomic_long_inc(&p->refs);
193 			return p;
194 		}
195 	}
196 	return NULL;
197 }
198 
199 int audit_tree_match(struct audit_chunk *chunk, struct audit_tree *tree)
200 {
201 	int n;
202 	for (n = 0; n < chunk->count; n++)
203 		if (chunk->owners[n].owner == tree)
204 			return 1;
205 	return 0;
206 }
207 
208 /* tagging and untagging inodes with trees */
209 
210 static struct audit_chunk *find_chunk(struct node *p)
211 {
212 	int index = p->index & ~(1U<<31);
213 	p -= index;
214 	return container_of(p, struct audit_chunk, owners[0]);
215 }
216 
217 static void untag_chunk(struct node *p)
218 {
219 	struct audit_chunk *chunk = find_chunk(p);
220 	struct fsnotify_mark *entry = &chunk->mark;
221 	struct audit_chunk *new = NULL;
222 	struct audit_tree *owner;
223 	int size = chunk->count - 1;
224 	int i, j;
225 
226 	fsnotify_get_mark(entry);
227 
228 	spin_unlock(&hash_lock);
229 
230 	if (size)
231 		new = alloc_chunk(size);
232 
233 	spin_lock(&entry->lock);
234 	if (chunk->dead || !entry->inode) {
235 		spin_unlock(&entry->lock);
236 		if (new)
237 			free_chunk(new);
238 		goto out;
239 	}
240 
241 	owner = p->owner;
242 
243 	if (!size) {
244 		chunk->dead = 1;
245 		spin_lock(&hash_lock);
246 		list_del_init(&chunk->trees);
247 		if (owner->root == chunk)
248 			owner->root = NULL;
249 		list_del_init(&p->list);
250 		list_del_rcu(&chunk->hash);
251 		spin_unlock(&hash_lock);
252 		spin_unlock(&entry->lock);
253 		fsnotify_destroy_mark(entry, audit_tree_group);
254 		goto out;
255 	}
256 
257 	if (!new)
258 		goto Fallback;
259 
260 	fsnotify_duplicate_mark(&new->mark, entry);
261 	if (fsnotify_add_mark(&new->mark, new->mark.group, new->mark.inode, NULL, 1)) {
262 		fsnotify_put_mark(&new->mark);
263 		goto Fallback;
264 	}
265 
266 	chunk->dead = 1;
267 	spin_lock(&hash_lock);
268 	list_replace_init(&chunk->trees, &new->trees);
269 	if (owner->root == chunk) {
270 		list_del_init(&owner->same_root);
271 		owner->root = NULL;
272 	}
273 
274 	for (i = j = 0; j <= size; i++, j++) {
275 		struct audit_tree *s;
276 		if (&chunk->owners[j] == p) {
277 			list_del_init(&p->list);
278 			i--;
279 			continue;
280 		}
281 		s = chunk->owners[j].owner;
282 		new->owners[i].owner = s;
283 		new->owners[i].index = chunk->owners[j].index - j + i;
284 		if (!s) /* result of earlier fallback */
285 			continue;
286 		get_tree(s);
287 		list_replace_init(&chunk->owners[j].list, &new->owners[i].list);
288 	}
289 
290 	list_replace_rcu(&chunk->hash, &new->hash);
291 	list_for_each_entry(owner, &new->trees, same_root)
292 		owner->root = new;
293 	spin_unlock(&hash_lock);
294 	spin_unlock(&entry->lock);
295 	fsnotify_destroy_mark(entry, audit_tree_group);
296 	fsnotify_put_mark(&new->mark);	/* drop initial reference */
297 	goto out;
298 
299 Fallback:
300 	// do the best we can
301 	spin_lock(&hash_lock);
302 	if (owner->root == chunk) {
303 		list_del_init(&owner->same_root);
304 		owner->root = NULL;
305 	}
306 	list_del_init(&p->list);
307 	p->owner = NULL;
308 	put_tree(owner);
309 	spin_unlock(&hash_lock);
310 	spin_unlock(&entry->lock);
311 out:
312 	fsnotify_put_mark(entry);
313 	spin_lock(&hash_lock);
314 }
315 
316 static int create_chunk(struct inode *inode, struct audit_tree *tree)
317 {
318 	struct fsnotify_mark *entry;
319 	struct audit_chunk *chunk = alloc_chunk(1);
320 	if (!chunk)
321 		return -ENOMEM;
322 
323 	entry = &chunk->mark;
324 	if (fsnotify_add_mark(entry, audit_tree_group, inode, NULL, 0)) {
325 		fsnotify_put_mark(entry);
326 		return -ENOSPC;
327 	}
328 
329 	spin_lock(&entry->lock);
330 	spin_lock(&hash_lock);
331 	if (tree->goner) {
332 		spin_unlock(&hash_lock);
333 		chunk->dead = 1;
334 		spin_unlock(&entry->lock);
335 		fsnotify_destroy_mark(entry, audit_tree_group);
336 		fsnotify_put_mark(entry);
337 		return 0;
338 	}
339 	chunk->owners[0].index = (1U << 31);
340 	chunk->owners[0].owner = tree;
341 	get_tree(tree);
342 	list_add(&chunk->owners[0].list, &tree->chunks);
343 	if (!tree->root) {
344 		tree->root = chunk;
345 		list_add(&tree->same_root, &chunk->trees);
346 	}
347 	insert_hash(chunk);
348 	spin_unlock(&hash_lock);
349 	spin_unlock(&entry->lock);
350 	fsnotify_put_mark(entry);	/* drop initial reference */
351 	return 0;
352 }
353 
354 /* the first tagged inode becomes root of tree */
355 static int tag_chunk(struct inode *inode, struct audit_tree *tree)
356 {
357 	struct fsnotify_mark *old_entry, *chunk_entry;
358 	struct audit_tree *owner;
359 	struct audit_chunk *chunk, *old;
360 	struct node *p;
361 	int n;
362 
363 	old_entry = fsnotify_find_inode_mark(audit_tree_group, inode);
364 	if (!old_entry)
365 		return create_chunk(inode, tree);
366 
367 	old = container_of(old_entry, struct audit_chunk, mark);
368 
369 	/* are we already there? */
370 	spin_lock(&hash_lock);
371 	for (n = 0; n < old->count; n++) {
372 		if (old->owners[n].owner == tree) {
373 			spin_unlock(&hash_lock);
374 			fsnotify_put_mark(old_entry);
375 			return 0;
376 		}
377 	}
378 	spin_unlock(&hash_lock);
379 
380 	chunk = alloc_chunk(old->count + 1);
381 	if (!chunk) {
382 		fsnotify_put_mark(old_entry);
383 		return -ENOMEM;
384 	}
385 
386 	chunk_entry = &chunk->mark;
387 
388 	spin_lock(&old_entry->lock);
389 	if (!old_entry->inode) {
390 		/* old_entry is being shot, lets just lie */
391 		spin_unlock(&old_entry->lock);
392 		fsnotify_put_mark(old_entry);
393 		free_chunk(chunk);
394 		return -ENOENT;
395 	}
396 
397 	fsnotify_duplicate_mark(chunk_entry, old_entry);
398 	if (fsnotify_add_mark(chunk_entry, chunk_entry->group, chunk_entry->inode, NULL, 1)) {
399 		spin_unlock(&old_entry->lock);
400 		fsnotify_put_mark(chunk_entry);
401 		fsnotify_put_mark(old_entry);
402 		return -ENOSPC;
403 	}
404 
405 	/* even though we hold old_entry->lock, this is safe since chunk_entry->lock could NEVER have been grabbed before */
406 	spin_lock(&chunk_entry->lock);
407 	spin_lock(&hash_lock);
408 
409 	/* we now hold old_entry->lock, chunk_entry->lock, and hash_lock */
410 	if (tree->goner) {
411 		spin_unlock(&hash_lock);
412 		chunk->dead = 1;
413 		spin_unlock(&chunk_entry->lock);
414 		spin_unlock(&old_entry->lock);
415 
416 		fsnotify_destroy_mark(chunk_entry, audit_tree_group);
417 
418 		fsnotify_put_mark(chunk_entry);
419 		fsnotify_put_mark(old_entry);
420 		return 0;
421 	}
422 	list_replace_init(&old->trees, &chunk->trees);
423 	for (n = 0, p = chunk->owners; n < old->count; n++, p++) {
424 		struct audit_tree *s = old->owners[n].owner;
425 		p->owner = s;
426 		p->index = old->owners[n].index;
427 		if (!s) /* result of fallback in untag */
428 			continue;
429 		get_tree(s);
430 		list_replace_init(&old->owners[n].list, &p->list);
431 	}
432 	p->index = (chunk->count - 1) | (1U<<31);
433 	p->owner = tree;
434 	get_tree(tree);
435 	list_add(&p->list, &tree->chunks);
436 	list_replace_rcu(&old->hash, &chunk->hash);
437 	list_for_each_entry(owner, &chunk->trees, same_root)
438 		owner->root = chunk;
439 	old->dead = 1;
440 	if (!tree->root) {
441 		tree->root = chunk;
442 		list_add(&tree->same_root, &chunk->trees);
443 	}
444 	spin_unlock(&hash_lock);
445 	spin_unlock(&chunk_entry->lock);
446 	spin_unlock(&old_entry->lock);
447 	fsnotify_destroy_mark(old_entry, audit_tree_group);
448 	fsnotify_put_mark(chunk_entry);	/* drop initial reference */
449 	fsnotify_put_mark(old_entry); /* pair to fsnotify_find mark_entry */
450 	return 0;
451 }
452 
453 static void audit_tree_log_remove_rule(struct audit_krule *rule)
454 {
455 	struct audit_buffer *ab;
456 
457 	ab = audit_log_start(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE);
458 	if (unlikely(!ab))
459 		return;
460 	audit_log_format(ab, "op=");
461 	audit_log_string(ab, "remove_rule");
462 	audit_log_format(ab, " dir=");
463 	audit_log_untrustedstring(ab, rule->tree->pathname);
464 	audit_log_key(ab, rule->filterkey);
465 	audit_log_format(ab, " list=%d res=1", rule->listnr);
466 	audit_log_end(ab);
467 }
468 
469 static void kill_rules(struct audit_tree *tree)
470 {
471 	struct audit_krule *rule, *next;
472 	struct audit_entry *entry;
473 
474 	list_for_each_entry_safe(rule, next, &tree->rules, rlist) {
475 		entry = container_of(rule, struct audit_entry, rule);
476 
477 		list_del_init(&rule->rlist);
478 		if (rule->tree) {
479 			/* not a half-baked one */
480 			audit_tree_log_remove_rule(rule);
481 			rule->tree = NULL;
482 			list_del_rcu(&entry->list);
483 			list_del(&entry->rule.list);
484 			call_rcu(&entry->rcu, audit_free_rule_rcu);
485 		}
486 	}
487 }
488 
489 /*
490  * finish killing struct audit_tree
491  */
492 static void prune_one(struct audit_tree *victim)
493 {
494 	spin_lock(&hash_lock);
495 	while (!list_empty(&victim->chunks)) {
496 		struct node *p;
497 
498 		p = list_entry(victim->chunks.next, struct node, list);
499 
500 		untag_chunk(p);
501 	}
502 	spin_unlock(&hash_lock);
503 	put_tree(victim);
504 }
505 
506 /* trim the uncommitted chunks from tree */
507 
508 static void trim_marked(struct audit_tree *tree)
509 {
510 	struct list_head *p, *q;
511 	spin_lock(&hash_lock);
512 	if (tree->goner) {
513 		spin_unlock(&hash_lock);
514 		return;
515 	}
516 	/* reorder */
517 	for (p = tree->chunks.next; p != &tree->chunks; p = q) {
518 		struct node *node = list_entry(p, struct node, list);
519 		q = p->next;
520 		if (node->index & (1U<<31)) {
521 			list_del_init(p);
522 			list_add(p, &tree->chunks);
523 		}
524 	}
525 
526 	while (!list_empty(&tree->chunks)) {
527 		struct node *node;
528 
529 		node = list_entry(tree->chunks.next, struct node, list);
530 
531 		/* have we run out of marked? */
532 		if (!(node->index & (1U<<31)))
533 			break;
534 
535 		untag_chunk(node);
536 	}
537 	if (!tree->root && !tree->goner) {
538 		tree->goner = 1;
539 		spin_unlock(&hash_lock);
540 		mutex_lock(&audit_filter_mutex);
541 		kill_rules(tree);
542 		list_del_init(&tree->list);
543 		mutex_unlock(&audit_filter_mutex);
544 		prune_one(tree);
545 	} else {
546 		spin_unlock(&hash_lock);
547 	}
548 }
549 
550 static void audit_schedule_prune(void);
551 
552 /* called with audit_filter_mutex */
553 int audit_remove_tree_rule(struct audit_krule *rule)
554 {
555 	struct audit_tree *tree;
556 	tree = rule->tree;
557 	if (tree) {
558 		spin_lock(&hash_lock);
559 		list_del_init(&rule->rlist);
560 		if (list_empty(&tree->rules) && !tree->goner) {
561 			tree->root = NULL;
562 			list_del_init(&tree->same_root);
563 			tree->goner = 1;
564 			list_move(&tree->list, &prune_list);
565 			rule->tree = NULL;
566 			spin_unlock(&hash_lock);
567 			audit_schedule_prune();
568 			return 1;
569 		}
570 		rule->tree = NULL;
571 		spin_unlock(&hash_lock);
572 		return 1;
573 	}
574 	return 0;
575 }
576 
577 static int compare_root(struct vfsmount *mnt, void *arg)
578 {
579 	return mnt->mnt_root->d_inode == arg;
580 }
581 
582 void audit_trim_trees(void)
583 {
584 	struct list_head cursor;
585 
586 	mutex_lock(&audit_filter_mutex);
587 	list_add(&cursor, &tree_list);
588 	while (cursor.next != &tree_list) {
589 		struct audit_tree *tree;
590 		struct path path;
591 		struct vfsmount *root_mnt;
592 		struct node *node;
593 		int err;
594 
595 		tree = container_of(cursor.next, struct audit_tree, list);
596 		get_tree(tree);
597 		list_del(&cursor);
598 		list_add(&cursor, &tree->list);
599 		mutex_unlock(&audit_filter_mutex);
600 
601 		err = kern_path(tree->pathname, 0, &path);
602 		if (err)
603 			goto skip_it;
604 
605 		root_mnt = collect_mounts(&path);
606 		path_put(&path);
607 		if (IS_ERR(root_mnt))
608 			goto skip_it;
609 
610 		spin_lock(&hash_lock);
611 		list_for_each_entry(node, &tree->chunks, list) {
612 			struct audit_chunk *chunk = find_chunk(node);
613 			/* this could be NULL if the watch is dying else where... */
614 			struct inode *inode = chunk->mark.inode;
615 			node->index |= 1U<<31;
616 			if (iterate_mounts(compare_root, inode, root_mnt))
617 				node->index &= ~(1U<<31);
618 		}
619 		spin_unlock(&hash_lock);
620 		trim_marked(tree);
621 		drop_collected_mounts(root_mnt);
622 skip_it:
623 		put_tree(tree);
624 		mutex_lock(&audit_filter_mutex);
625 	}
626 	list_del(&cursor);
627 	mutex_unlock(&audit_filter_mutex);
628 }
629 
630 int audit_make_tree(struct audit_krule *rule, char *pathname, u32 op)
631 {
632 
633 	if (pathname[0] != '/' ||
634 	    rule->listnr != AUDIT_FILTER_EXIT ||
635 	    op != Audit_equal ||
636 	    rule->inode_f || rule->watch || rule->tree)
637 		return -EINVAL;
638 	rule->tree = alloc_tree(pathname);
639 	if (!rule->tree)
640 		return -ENOMEM;
641 	return 0;
642 }
643 
644 void audit_put_tree(struct audit_tree *tree)
645 {
646 	put_tree(tree);
647 }
648 
649 static int tag_mount(struct vfsmount *mnt, void *arg)
650 {
651 	return tag_chunk(mnt->mnt_root->d_inode, arg);
652 }
653 
654 /* called with audit_filter_mutex */
655 int audit_add_tree_rule(struct audit_krule *rule)
656 {
657 	struct audit_tree *seed = rule->tree, *tree;
658 	struct path path;
659 	struct vfsmount *mnt;
660 	int err;
661 
662 	rule->tree = NULL;
663 	list_for_each_entry(tree, &tree_list, list) {
664 		if (!strcmp(seed->pathname, tree->pathname)) {
665 			put_tree(seed);
666 			rule->tree = tree;
667 			list_add(&rule->rlist, &tree->rules);
668 			return 0;
669 		}
670 	}
671 	tree = seed;
672 	list_add(&tree->list, &tree_list);
673 	list_add(&rule->rlist, &tree->rules);
674 	/* do not set rule->tree yet */
675 	mutex_unlock(&audit_filter_mutex);
676 
677 	err = kern_path(tree->pathname, 0, &path);
678 	if (err)
679 		goto Err;
680 	mnt = collect_mounts(&path);
681 	path_put(&path);
682 	if (IS_ERR(mnt)) {
683 		err = PTR_ERR(mnt);
684 		goto Err;
685 	}
686 
687 	get_tree(tree);
688 	err = iterate_mounts(tag_mount, tree, mnt);
689 	drop_collected_mounts(mnt);
690 
691 	if (!err) {
692 		struct node *node;
693 		spin_lock(&hash_lock);
694 		list_for_each_entry(node, &tree->chunks, list)
695 			node->index &= ~(1U<<31);
696 		spin_unlock(&hash_lock);
697 	} else {
698 		trim_marked(tree);
699 		goto Err;
700 	}
701 
702 	mutex_lock(&audit_filter_mutex);
703 	if (list_empty(&rule->rlist)) {
704 		put_tree(tree);
705 		return -ENOENT;
706 	}
707 	rule->tree = tree;
708 	put_tree(tree);
709 
710 	return 0;
711 Err:
712 	mutex_lock(&audit_filter_mutex);
713 	list_del_init(&tree->list);
714 	list_del_init(&tree->rules);
715 	put_tree(tree);
716 	return err;
717 }
718 
719 int audit_tag_tree(char *old, char *new)
720 {
721 	struct list_head cursor, barrier;
722 	int failed = 0;
723 	struct path path1, path2;
724 	struct vfsmount *tagged;
725 	int err;
726 
727 	err = kern_path(new, 0, &path2);
728 	if (err)
729 		return err;
730 	tagged = collect_mounts(&path2);
731 	path_put(&path2);
732 	if (IS_ERR(tagged))
733 		return PTR_ERR(tagged);
734 
735 	err = kern_path(old, 0, &path1);
736 	if (err) {
737 		drop_collected_mounts(tagged);
738 		return err;
739 	}
740 
741 	mutex_lock(&audit_filter_mutex);
742 	list_add(&barrier, &tree_list);
743 	list_add(&cursor, &barrier);
744 
745 	while (cursor.next != &tree_list) {
746 		struct audit_tree *tree;
747 		int good_one = 0;
748 
749 		tree = container_of(cursor.next, struct audit_tree, list);
750 		get_tree(tree);
751 		list_del(&cursor);
752 		list_add(&cursor, &tree->list);
753 		mutex_unlock(&audit_filter_mutex);
754 
755 		err = kern_path(tree->pathname, 0, &path2);
756 		if (!err) {
757 			good_one = path_is_under(&path1, &path2);
758 			path_put(&path2);
759 		}
760 
761 		if (!good_one) {
762 			put_tree(tree);
763 			mutex_lock(&audit_filter_mutex);
764 			continue;
765 		}
766 
767 		failed = iterate_mounts(tag_mount, tree, tagged);
768 		if (failed) {
769 			put_tree(tree);
770 			mutex_lock(&audit_filter_mutex);
771 			break;
772 		}
773 
774 		mutex_lock(&audit_filter_mutex);
775 		spin_lock(&hash_lock);
776 		if (!tree->goner) {
777 			list_del(&tree->list);
778 			list_add(&tree->list, &tree_list);
779 		}
780 		spin_unlock(&hash_lock);
781 		put_tree(tree);
782 	}
783 
784 	while (barrier.prev != &tree_list) {
785 		struct audit_tree *tree;
786 
787 		tree = container_of(barrier.prev, struct audit_tree, list);
788 		get_tree(tree);
789 		list_del(&tree->list);
790 		list_add(&tree->list, &barrier);
791 		mutex_unlock(&audit_filter_mutex);
792 
793 		if (!failed) {
794 			struct node *node;
795 			spin_lock(&hash_lock);
796 			list_for_each_entry(node, &tree->chunks, list)
797 				node->index &= ~(1U<<31);
798 			spin_unlock(&hash_lock);
799 		} else {
800 			trim_marked(tree);
801 		}
802 
803 		put_tree(tree);
804 		mutex_lock(&audit_filter_mutex);
805 	}
806 	list_del(&barrier);
807 	list_del(&cursor);
808 	mutex_unlock(&audit_filter_mutex);
809 	path_put(&path1);
810 	drop_collected_mounts(tagged);
811 	return failed;
812 }
813 
814 /*
815  * That gets run when evict_chunk() ends up needing to kill audit_tree.
816  * Runs from a separate thread.
817  */
818 static int prune_tree_thread(void *unused)
819 {
820 	mutex_lock(&audit_cmd_mutex);
821 	mutex_lock(&audit_filter_mutex);
822 
823 	while (!list_empty(&prune_list)) {
824 		struct audit_tree *victim;
825 
826 		victim = list_entry(prune_list.next, struct audit_tree, list);
827 		list_del_init(&victim->list);
828 
829 		mutex_unlock(&audit_filter_mutex);
830 
831 		prune_one(victim);
832 
833 		mutex_lock(&audit_filter_mutex);
834 	}
835 
836 	mutex_unlock(&audit_filter_mutex);
837 	mutex_unlock(&audit_cmd_mutex);
838 	return 0;
839 }
840 
841 static void audit_schedule_prune(void)
842 {
843 	kthread_run(prune_tree_thread, NULL, "audit_prune_tree");
844 }
845 
846 /*
847  * ... and that one is done if evict_chunk() decides to delay until the end
848  * of syscall.  Runs synchronously.
849  */
850 void audit_kill_trees(struct list_head *list)
851 {
852 	mutex_lock(&audit_cmd_mutex);
853 	mutex_lock(&audit_filter_mutex);
854 
855 	while (!list_empty(list)) {
856 		struct audit_tree *victim;
857 
858 		victim = list_entry(list->next, struct audit_tree, list);
859 		kill_rules(victim);
860 		list_del_init(&victim->list);
861 
862 		mutex_unlock(&audit_filter_mutex);
863 
864 		prune_one(victim);
865 
866 		mutex_lock(&audit_filter_mutex);
867 	}
868 
869 	mutex_unlock(&audit_filter_mutex);
870 	mutex_unlock(&audit_cmd_mutex);
871 }
872 
873 /*
874  *  Here comes the stuff asynchronous to auditctl operations
875  */
876 
877 static void evict_chunk(struct audit_chunk *chunk)
878 {
879 	struct audit_tree *owner;
880 	struct list_head *postponed = audit_killed_trees();
881 	int need_prune = 0;
882 	int n;
883 
884 	if (chunk->dead)
885 		return;
886 
887 	chunk->dead = 1;
888 	mutex_lock(&audit_filter_mutex);
889 	spin_lock(&hash_lock);
890 	while (!list_empty(&chunk->trees)) {
891 		owner = list_entry(chunk->trees.next,
892 				   struct audit_tree, same_root);
893 		owner->goner = 1;
894 		owner->root = NULL;
895 		list_del_init(&owner->same_root);
896 		spin_unlock(&hash_lock);
897 		if (!postponed) {
898 			kill_rules(owner);
899 			list_move(&owner->list, &prune_list);
900 			need_prune = 1;
901 		} else {
902 			list_move(&owner->list, postponed);
903 		}
904 		spin_lock(&hash_lock);
905 	}
906 	list_del_rcu(&chunk->hash);
907 	for (n = 0; n < chunk->count; n++)
908 		list_del_init(&chunk->owners[n].list);
909 	spin_unlock(&hash_lock);
910 	if (need_prune)
911 		audit_schedule_prune();
912 	mutex_unlock(&audit_filter_mutex);
913 }
914 
915 static int audit_tree_handle_event(struct fsnotify_group *group,
916 				   struct inode *to_tell,
917 				   struct fsnotify_mark *inode_mark,
918 				   struct fsnotify_mark *vfsmount_mark,
919 				   u32 mask, void *data, int data_type,
920 				   const unsigned char *file_name, u32 cookie)
921 {
922 	return 0;
923 }
924 
925 static void audit_tree_freeing_mark(struct fsnotify_mark *entry, struct fsnotify_group *group)
926 {
927 	struct audit_chunk *chunk = container_of(entry, struct audit_chunk, mark);
928 
929 	evict_chunk(chunk);
930 
931 	/*
932 	 * We are guaranteed to have at least one reference to the mark from
933 	 * either the inode or the caller of fsnotify_destroy_mark().
934 	 */
935 	BUG_ON(atomic_read(&entry->refcnt) < 1);
936 }
937 
938 static const struct fsnotify_ops audit_tree_ops = {
939 	.handle_event = audit_tree_handle_event,
940 	.freeing_mark = audit_tree_freeing_mark,
941 };
942 
943 static int __init audit_tree_init(void)
944 {
945 	int i;
946 
947 	audit_tree_group = fsnotify_alloc_group(&audit_tree_ops);
948 	if (IS_ERR(audit_tree_group))
949 		audit_panic("cannot initialize fsnotify group for rectree watches");
950 
951 	for (i = 0; i < HASH_SIZE; i++)
952 		INIT_LIST_HEAD(&chunk_hash_heads[i]);
953 
954 	return 0;
955 }
956 __initcall(audit_tree_init);
957