xref: /openbmc/linux/fs/ubifs/orphan.c (revision d4fd6347)
1 /*
2  * This file is part of UBIFS.
3  *
4  * Copyright (C) 2006-2008 Nokia Corporation.
5  *
6  * This program is free software; you can redistribute it and/or modify it
7  * under the terms of the GNU General Public License version 2 as published by
8  * the Free Software Foundation.
9  *
10  * This program is distributed in the hope that it will be useful, but WITHOUT
11  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
12  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
13  * more details.
14  *
15  * You should have received a copy of the GNU General Public License along with
16  * this program; if not, write to the Free Software Foundation, Inc., 51
17  * Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
18  *
19  * Author: Adrian Hunter
20  */
21 
22 #include "ubifs.h"
23 
24 /*
25  * An orphan is an inode number whose inode node has been committed to the index
26  * with a link count of zero. That happens when an open file is deleted
27  * (unlinked) and then a commit is run. In the normal course of events the inode
28  * would be deleted when the file is closed. However in the case of an unclean
29  * unmount, orphans need to be accounted for. After an unclean unmount, the
30  * orphans' inodes must be deleted which means either scanning the entire index
31  * looking for them, or keeping a list on flash somewhere. This unit implements
32  * the latter approach.
33  *
34  * The orphan area is a fixed number of LEBs situated between the LPT area and
35  * the main area. The number of orphan area LEBs is specified when the file
36  * system is created. The minimum number is 1. The size of the orphan area
37  * should be so that it can hold the maximum number of orphans that are expected
38  * to ever exist at one time.
39  *
40  * The number of orphans that can fit in a LEB is:
41  *
42  *         (c->leb_size - UBIFS_ORPH_NODE_SZ) / sizeof(__le64)
43  *
44  * For example: a 15872 byte LEB can fit 1980 orphans so 1 LEB may be enough.
45  *
46  * Orphans are accumulated in a rb-tree. When an inode's link count drops to
47  * zero, the inode number is added to the rb-tree. It is removed from the tree
48  * when the inode is deleted.  Any new orphans that are in the orphan tree when
49  * the commit is run, are written to the orphan area in 1 or more orphan nodes.
50  * If the orphan area is full, it is consolidated to make space.  There is
51  * always enough space because validation prevents the user from creating more
52  * than the maximum number of orphans allowed.
53  */
54 
55 static int dbg_check_orphans(struct ubifs_info *c);
56 
57 static struct ubifs_orphan *orphan_add(struct ubifs_info *c, ino_t inum,
58 				       struct ubifs_orphan *parent_orphan)
59 {
60 	struct ubifs_orphan *orphan, *o;
61 	struct rb_node **p, *parent = NULL;
62 
63 	orphan = kzalloc(sizeof(struct ubifs_orphan), GFP_NOFS);
64 	if (!orphan)
65 		return ERR_PTR(-ENOMEM);
66 	orphan->inum = inum;
67 	orphan->new = 1;
68 	INIT_LIST_HEAD(&orphan->child_list);
69 
70 	spin_lock(&c->orphan_lock);
71 	if (c->tot_orphans >= c->max_orphans) {
72 		spin_unlock(&c->orphan_lock);
73 		kfree(orphan);
74 		return ERR_PTR(-ENFILE);
75 	}
76 	p = &c->orph_tree.rb_node;
77 	while (*p) {
78 		parent = *p;
79 		o = rb_entry(parent, struct ubifs_orphan, rb);
80 		if (inum < o->inum)
81 			p = &(*p)->rb_left;
82 		else if (inum > o->inum)
83 			p = &(*p)->rb_right;
84 		else {
85 			ubifs_err(c, "orphaned twice");
86 			spin_unlock(&c->orphan_lock);
87 			kfree(orphan);
88 			return ERR_PTR(-EINVAL);
89 		}
90 	}
91 	c->tot_orphans += 1;
92 	c->new_orphans += 1;
93 	rb_link_node(&orphan->rb, parent, p);
94 	rb_insert_color(&orphan->rb, &c->orph_tree);
95 	list_add_tail(&orphan->list, &c->orph_list);
96 	list_add_tail(&orphan->new_list, &c->orph_new);
97 
98 	if (parent_orphan) {
99 		list_add_tail(&orphan->child_list,
100 			      &parent_orphan->child_list);
101 	}
102 
103 	spin_unlock(&c->orphan_lock);
104 	dbg_gen("ino %lu", (unsigned long)inum);
105 	return orphan;
106 }
107 
108 static struct ubifs_orphan *lookup_orphan(struct ubifs_info *c, ino_t inum)
109 {
110 	struct ubifs_orphan *o;
111 	struct rb_node *p;
112 
113 	p = c->orph_tree.rb_node;
114 	while (p) {
115 		o = rb_entry(p, struct ubifs_orphan, rb);
116 		if (inum < o->inum)
117 			p = p->rb_left;
118 		else if (inum > o->inum)
119 			p = p->rb_right;
120 		else {
121 			return o;
122 		}
123 	}
124 	return NULL;
125 }
126 
127 static void __orphan_drop(struct ubifs_info *c, struct ubifs_orphan *o)
128 {
129 	rb_erase(&o->rb, &c->orph_tree);
130 	list_del(&o->list);
131 	c->tot_orphans -= 1;
132 
133 	if (o->new) {
134 		list_del(&o->new_list);
135 		c->new_orphans -= 1;
136 	}
137 
138 	kfree(o);
139 }
140 
141 static void orphan_delete(struct ubifs_info *c, ino_t inum)
142 {
143 	struct ubifs_orphan *orph, *child_orph, *tmp_o;
144 
145 	spin_lock(&c->orphan_lock);
146 
147 	orph = lookup_orphan(c, inum);
148 	if (!orph) {
149 		spin_unlock(&c->orphan_lock);
150 		ubifs_err(c, "missing orphan ino %lu", (unsigned long)inum);
151 		dump_stack();
152 
153 		return;
154 	}
155 
156 	if (orph->del) {
157 		spin_unlock(&c->orphan_lock);
158 		dbg_gen("deleted twice ino %lu",
159 			(unsigned long)inum);
160 		return;
161 	}
162 
163 	if (orph->cmt) {
164 		orph->del = 1;
165 		orph->dnext = c->orph_dnext;
166 		c->orph_dnext = orph;
167 		spin_unlock(&c->orphan_lock);
168 		dbg_gen("delete later ino %lu",
169 			(unsigned long)inum);
170 		return;
171 	}
172 
173 	list_for_each_entry_safe(child_orph, tmp_o, &orph->child_list, child_list) {
174 		list_del(&child_orph->child_list);
175 		__orphan_drop(c, child_orph);
176 	}
177 
178 	__orphan_drop(c, orph);
179 
180 	spin_unlock(&c->orphan_lock);
181 }
182 
183 /**
184  * ubifs_add_orphan - add an orphan.
185  * @c: UBIFS file-system description object
186  * @inum: orphan inode number
187  *
188  * Add an orphan. This function is called when an inodes link count drops to
189  * zero.
190  */
191 int ubifs_add_orphan(struct ubifs_info *c, ino_t inum)
192 {
193 	int err = 0;
194 	ino_t xattr_inum;
195 	union ubifs_key key;
196 	struct ubifs_dent_node *xent;
197 	struct fscrypt_name nm = {0};
198 	struct ubifs_orphan *xattr_orphan;
199 	struct ubifs_orphan *orphan;
200 
201 	orphan = orphan_add(c, inum, NULL);
202 	if (IS_ERR(orphan))
203 		return PTR_ERR(orphan);
204 
205 	lowest_xent_key(c, &key, inum);
206 	while (1) {
207 		xent = ubifs_tnc_next_ent(c, &key, &nm);
208 		if (IS_ERR(xent)) {
209 			err = PTR_ERR(xent);
210 			if (err == -ENOENT)
211 				break;
212 			return err;
213 		}
214 
215 		fname_name(&nm) = xent->name;
216 		fname_len(&nm) = le16_to_cpu(xent->nlen);
217 		xattr_inum = le64_to_cpu(xent->inum);
218 
219 		xattr_orphan = orphan_add(c, xattr_inum, orphan);
220 		if (IS_ERR(xattr_orphan))
221 			return PTR_ERR(xattr_orphan);
222 
223 		key_read(c, &xent->key, &key);
224 	}
225 
226 	return 0;
227 }
228 
229 /**
230  * ubifs_delete_orphan - delete an orphan.
231  * @c: UBIFS file-system description object
232  * @inum: orphan inode number
233  *
234  * Delete an orphan. This function is called when an inode is deleted.
235  */
236 void ubifs_delete_orphan(struct ubifs_info *c, ino_t inum)
237 {
238 	orphan_delete(c, inum);
239 }
240 
241 /**
242  * ubifs_orphan_start_commit - start commit of orphans.
243  * @c: UBIFS file-system description object
244  *
245  * Start commit of orphans.
246  */
247 int ubifs_orphan_start_commit(struct ubifs_info *c)
248 {
249 	struct ubifs_orphan *orphan, **last;
250 
251 	spin_lock(&c->orphan_lock);
252 	last = &c->orph_cnext;
253 	list_for_each_entry(orphan, &c->orph_new, new_list) {
254 		ubifs_assert(c, orphan->new);
255 		ubifs_assert(c, !orphan->cmt);
256 		orphan->new = 0;
257 		orphan->cmt = 1;
258 		*last = orphan;
259 		last = &orphan->cnext;
260 	}
261 	*last = NULL;
262 	c->cmt_orphans = c->new_orphans;
263 	c->new_orphans = 0;
264 	dbg_cmt("%d orphans to commit", c->cmt_orphans);
265 	INIT_LIST_HEAD(&c->orph_new);
266 	if (c->tot_orphans == 0)
267 		c->no_orphs = 1;
268 	else
269 		c->no_orphs = 0;
270 	spin_unlock(&c->orphan_lock);
271 	return 0;
272 }
273 
274 /**
275  * avail_orphs - calculate available space.
276  * @c: UBIFS file-system description object
277  *
278  * This function returns the number of orphans that can be written in the
279  * available space.
280  */
281 static int avail_orphs(struct ubifs_info *c)
282 {
283 	int avail_lebs, avail, gap;
284 
285 	avail_lebs = c->orph_lebs - (c->ohead_lnum - c->orph_first) - 1;
286 	avail = avail_lebs *
287 	       ((c->leb_size - UBIFS_ORPH_NODE_SZ) / sizeof(__le64));
288 	gap = c->leb_size - c->ohead_offs;
289 	if (gap >= UBIFS_ORPH_NODE_SZ + sizeof(__le64))
290 		avail += (gap - UBIFS_ORPH_NODE_SZ) / sizeof(__le64);
291 	return avail;
292 }
293 
294 /**
295  * tot_avail_orphs - calculate total space.
296  * @c: UBIFS file-system description object
297  *
298  * This function returns the number of orphans that can be written in half
299  * the total space. That leaves half the space for adding new orphans.
300  */
301 static int tot_avail_orphs(struct ubifs_info *c)
302 {
303 	int avail_lebs, avail;
304 
305 	avail_lebs = c->orph_lebs;
306 	avail = avail_lebs *
307 	       ((c->leb_size - UBIFS_ORPH_NODE_SZ) / sizeof(__le64));
308 	return avail / 2;
309 }
310 
311 /**
312  * do_write_orph_node - write a node to the orphan head.
313  * @c: UBIFS file-system description object
314  * @len: length of node
315  * @atomic: write atomically
316  *
317  * This function writes a node to the orphan head from the orphan buffer. If
318  * %atomic is not zero, then the write is done atomically. On success, %0 is
319  * returned, otherwise a negative error code is returned.
320  */
321 static int do_write_orph_node(struct ubifs_info *c, int len, int atomic)
322 {
323 	int err = 0;
324 
325 	if (atomic) {
326 		ubifs_assert(c, c->ohead_offs == 0);
327 		ubifs_prepare_node(c, c->orph_buf, len, 1);
328 		len = ALIGN(len, c->min_io_size);
329 		err = ubifs_leb_change(c, c->ohead_lnum, c->orph_buf, len);
330 	} else {
331 		if (c->ohead_offs == 0) {
332 			/* Ensure LEB has been unmapped */
333 			err = ubifs_leb_unmap(c, c->ohead_lnum);
334 			if (err)
335 				return err;
336 		}
337 		err = ubifs_write_node(c, c->orph_buf, len, c->ohead_lnum,
338 				       c->ohead_offs);
339 	}
340 	return err;
341 }
342 
343 /**
344  * write_orph_node - write an orphan node.
345  * @c: UBIFS file-system description object
346  * @atomic: write atomically
347  *
348  * This function builds an orphan node from the cnext list and writes it to the
349  * orphan head. On success, %0 is returned, otherwise a negative error code
350  * is returned.
351  */
352 static int write_orph_node(struct ubifs_info *c, int atomic)
353 {
354 	struct ubifs_orphan *orphan, *cnext;
355 	struct ubifs_orph_node *orph;
356 	int gap, err, len, cnt, i;
357 
358 	ubifs_assert(c, c->cmt_orphans > 0);
359 	gap = c->leb_size - c->ohead_offs;
360 	if (gap < UBIFS_ORPH_NODE_SZ + sizeof(__le64)) {
361 		c->ohead_lnum += 1;
362 		c->ohead_offs = 0;
363 		gap = c->leb_size;
364 		if (c->ohead_lnum > c->orph_last) {
365 			/*
366 			 * We limit the number of orphans so that this should
367 			 * never happen.
368 			 */
369 			ubifs_err(c, "out of space in orphan area");
370 			return -EINVAL;
371 		}
372 	}
373 	cnt = (gap - UBIFS_ORPH_NODE_SZ) / sizeof(__le64);
374 	if (cnt > c->cmt_orphans)
375 		cnt = c->cmt_orphans;
376 	len = UBIFS_ORPH_NODE_SZ + cnt * sizeof(__le64);
377 	ubifs_assert(c, c->orph_buf);
378 	orph = c->orph_buf;
379 	orph->ch.node_type = UBIFS_ORPH_NODE;
380 	spin_lock(&c->orphan_lock);
381 	cnext = c->orph_cnext;
382 	for (i = 0; i < cnt; i++) {
383 		orphan = cnext;
384 		ubifs_assert(c, orphan->cmt);
385 		orph->inos[i] = cpu_to_le64(orphan->inum);
386 		orphan->cmt = 0;
387 		cnext = orphan->cnext;
388 		orphan->cnext = NULL;
389 	}
390 	c->orph_cnext = cnext;
391 	c->cmt_orphans -= cnt;
392 	spin_unlock(&c->orphan_lock);
393 	if (c->cmt_orphans)
394 		orph->cmt_no = cpu_to_le64(c->cmt_no);
395 	else
396 		/* Mark the last node of the commit */
397 		orph->cmt_no = cpu_to_le64((c->cmt_no) | (1ULL << 63));
398 	ubifs_assert(c, c->ohead_offs + len <= c->leb_size);
399 	ubifs_assert(c, c->ohead_lnum >= c->orph_first);
400 	ubifs_assert(c, c->ohead_lnum <= c->orph_last);
401 	err = do_write_orph_node(c, len, atomic);
402 	c->ohead_offs += ALIGN(len, c->min_io_size);
403 	c->ohead_offs = ALIGN(c->ohead_offs, 8);
404 	return err;
405 }
406 
407 /**
408  * write_orph_nodes - write orphan nodes until there are no more to commit.
409  * @c: UBIFS file-system description object
410  * @atomic: write atomically
411  *
412  * This function writes orphan nodes for all the orphans to commit. On success,
413  * %0 is returned, otherwise a negative error code is returned.
414  */
415 static int write_orph_nodes(struct ubifs_info *c, int atomic)
416 {
417 	int err;
418 
419 	while (c->cmt_orphans > 0) {
420 		err = write_orph_node(c, atomic);
421 		if (err)
422 			return err;
423 	}
424 	if (atomic) {
425 		int lnum;
426 
427 		/* Unmap any unused LEBs after consolidation */
428 		for (lnum = c->ohead_lnum + 1; lnum <= c->orph_last; lnum++) {
429 			err = ubifs_leb_unmap(c, lnum);
430 			if (err)
431 				return err;
432 		}
433 	}
434 	return 0;
435 }
436 
437 /**
438  * consolidate - consolidate the orphan area.
439  * @c: UBIFS file-system description object
440  *
441  * This function enables consolidation by putting all the orphans into the list
442  * to commit. The list is in the order that the orphans were added, and the
443  * LEBs are written atomically in order, so at no time can orphans be lost by
444  * an unclean unmount.
445  *
446  * This function returns %0 on success and a negative error code on failure.
447  */
448 static int consolidate(struct ubifs_info *c)
449 {
450 	int tot_avail = tot_avail_orphs(c), err = 0;
451 
452 	spin_lock(&c->orphan_lock);
453 	dbg_cmt("there is space for %d orphans and there are %d",
454 		tot_avail, c->tot_orphans);
455 	if (c->tot_orphans - c->new_orphans <= tot_avail) {
456 		struct ubifs_orphan *orphan, **last;
457 		int cnt = 0;
458 
459 		/* Change the cnext list to include all non-new orphans */
460 		last = &c->orph_cnext;
461 		list_for_each_entry(orphan, &c->orph_list, list) {
462 			if (orphan->new)
463 				continue;
464 			orphan->cmt = 1;
465 			*last = orphan;
466 			last = &orphan->cnext;
467 			cnt += 1;
468 		}
469 		*last = NULL;
470 		ubifs_assert(c, cnt == c->tot_orphans - c->new_orphans);
471 		c->cmt_orphans = cnt;
472 		c->ohead_lnum = c->orph_first;
473 		c->ohead_offs = 0;
474 	} else {
475 		/*
476 		 * We limit the number of orphans so that this should
477 		 * never happen.
478 		 */
479 		ubifs_err(c, "out of space in orphan area");
480 		err = -EINVAL;
481 	}
482 	spin_unlock(&c->orphan_lock);
483 	return err;
484 }
485 
486 /**
487  * commit_orphans - commit orphans.
488  * @c: UBIFS file-system description object
489  *
490  * This function commits orphans to flash. On success, %0 is returned,
491  * otherwise a negative error code is returned.
492  */
493 static int commit_orphans(struct ubifs_info *c)
494 {
495 	int avail, atomic = 0, err;
496 
497 	ubifs_assert(c, c->cmt_orphans > 0);
498 	avail = avail_orphs(c);
499 	if (avail < c->cmt_orphans) {
500 		/* Not enough space to write new orphans, so consolidate */
501 		err = consolidate(c);
502 		if (err)
503 			return err;
504 		atomic = 1;
505 	}
506 	err = write_orph_nodes(c, atomic);
507 	return err;
508 }
509 
510 /**
511  * erase_deleted - erase the orphans marked for deletion.
512  * @c: UBIFS file-system description object
513  *
514  * During commit, the orphans being committed cannot be deleted, so they are
515  * marked for deletion and deleted by this function. Also, the recovery
516  * adds killed orphans to the deletion list, and therefore they are deleted
517  * here too.
518  */
519 static void erase_deleted(struct ubifs_info *c)
520 {
521 	struct ubifs_orphan *orphan, *dnext;
522 
523 	spin_lock(&c->orphan_lock);
524 	dnext = c->orph_dnext;
525 	while (dnext) {
526 		orphan = dnext;
527 		dnext = orphan->dnext;
528 		ubifs_assert(c, !orphan->new);
529 		ubifs_assert(c, orphan->del);
530 		rb_erase(&orphan->rb, &c->orph_tree);
531 		list_del(&orphan->list);
532 		c->tot_orphans -= 1;
533 		dbg_gen("deleting orphan ino %lu", (unsigned long)orphan->inum);
534 		kfree(orphan);
535 	}
536 	c->orph_dnext = NULL;
537 	spin_unlock(&c->orphan_lock);
538 }
539 
540 /**
541  * ubifs_orphan_end_commit - end commit of orphans.
542  * @c: UBIFS file-system description object
543  *
544  * End commit of orphans.
545  */
546 int ubifs_orphan_end_commit(struct ubifs_info *c)
547 {
548 	int err;
549 
550 	if (c->cmt_orphans != 0) {
551 		err = commit_orphans(c);
552 		if (err)
553 			return err;
554 	}
555 	erase_deleted(c);
556 	err = dbg_check_orphans(c);
557 	return err;
558 }
559 
560 /**
561  * ubifs_clear_orphans - erase all LEBs used for orphans.
562  * @c: UBIFS file-system description object
563  *
564  * If recovery is not required, then the orphans from the previous session
565  * are not needed. This function locates the LEBs used to record
566  * orphans, and un-maps them.
567  */
568 int ubifs_clear_orphans(struct ubifs_info *c)
569 {
570 	int lnum, err;
571 
572 	for (lnum = c->orph_first; lnum <= c->orph_last; lnum++) {
573 		err = ubifs_leb_unmap(c, lnum);
574 		if (err)
575 			return err;
576 	}
577 	c->ohead_lnum = c->orph_first;
578 	c->ohead_offs = 0;
579 	return 0;
580 }
581 
582 /**
583  * insert_dead_orphan - insert an orphan.
584  * @c: UBIFS file-system description object
585  * @inum: orphan inode number
586  *
587  * This function is a helper to the 'do_kill_orphans()' function. The orphan
588  * must be kept until the next commit, so it is added to the rb-tree and the
589  * deletion list.
590  */
591 static int insert_dead_orphan(struct ubifs_info *c, ino_t inum)
592 {
593 	struct ubifs_orphan *orphan, *o;
594 	struct rb_node **p, *parent = NULL;
595 
596 	orphan = kzalloc(sizeof(struct ubifs_orphan), GFP_KERNEL);
597 	if (!orphan)
598 		return -ENOMEM;
599 	orphan->inum = inum;
600 
601 	p = &c->orph_tree.rb_node;
602 	while (*p) {
603 		parent = *p;
604 		o = rb_entry(parent, struct ubifs_orphan, rb);
605 		if (inum < o->inum)
606 			p = &(*p)->rb_left;
607 		else if (inum > o->inum)
608 			p = &(*p)->rb_right;
609 		else {
610 			/* Already added - no problem */
611 			kfree(orphan);
612 			return 0;
613 		}
614 	}
615 	c->tot_orphans += 1;
616 	rb_link_node(&orphan->rb, parent, p);
617 	rb_insert_color(&orphan->rb, &c->orph_tree);
618 	list_add_tail(&orphan->list, &c->orph_list);
619 	orphan->del = 1;
620 	orphan->dnext = c->orph_dnext;
621 	c->orph_dnext = orphan;
622 	dbg_mnt("ino %lu, new %d, tot %d", (unsigned long)inum,
623 		c->new_orphans, c->tot_orphans);
624 	return 0;
625 }
626 
627 /**
628  * do_kill_orphans - remove orphan inodes from the index.
629  * @c: UBIFS file-system description object
630  * @sleb: scanned LEB
631  * @last_cmt_no: cmt_no of last orphan node read is passed and returned here
632  * @outofdate: whether the LEB is out of date is returned here
633  * @last_flagged: whether the end orphan node is encountered
634  *
635  * This function is a helper to the 'kill_orphans()' function. It goes through
636  * every orphan node in a LEB and for every inode number recorded, removes
637  * all keys for that inode from the TNC.
638  */
639 static int do_kill_orphans(struct ubifs_info *c, struct ubifs_scan_leb *sleb,
640 			   unsigned long long *last_cmt_no, int *outofdate,
641 			   int *last_flagged)
642 {
643 	struct ubifs_scan_node *snod;
644 	struct ubifs_orph_node *orph;
645 	unsigned long long cmt_no;
646 	ino_t inum;
647 	int i, n, err, first = 1;
648 
649 	list_for_each_entry(snod, &sleb->nodes, list) {
650 		if (snod->type != UBIFS_ORPH_NODE) {
651 			ubifs_err(c, "invalid node type %d in orphan area at %d:%d",
652 				  snod->type, sleb->lnum, snod->offs);
653 			ubifs_dump_node(c, snod->node);
654 			return -EINVAL;
655 		}
656 
657 		orph = snod->node;
658 
659 		/* Check commit number */
660 		cmt_no = le64_to_cpu(orph->cmt_no) & LLONG_MAX;
661 		/*
662 		 * The commit number on the master node may be less, because
663 		 * of a failed commit. If there are several failed commits in a
664 		 * row, the commit number written on orphan nodes will continue
665 		 * to increase (because the commit number is adjusted here) even
666 		 * though the commit number on the master node stays the same
667 		 * because the master node has not been re-written.
668 		 */
669 		if (cmt_no > c->cmt_no)
670 			c->cmt_no = cmt_no;
671 		if (cmt_no < *last_cmt_no && *last_flagged) {
672 			/*
673 			 * The last orphan node had a higher commit number and
674 			 * was flagged as the last written for that commit
675 			 * number. That makes this orphan node, out of date.
676 			 */
677 			if (!first) {
678 				ubifs_err(c, "out of order commit number %llu in orphan node at %d:%d",
679 					  cmt_no, sleb->lnum, snod->offs);
680 				ubifs_dump_node(c, snod->node);
681 				return -EINVAL;
682 			}
683 			dbg_rcvry("out of date LEB %d", sleb->lnum);
684 			*outofdate = 1;
685 			return 0;
686 		}
687 
688 		if (first)
689 			first = 0;
690 
691 		n = (le32_to_cpu(orph->ch.len) - UBIFS_ORPH_NODE_SZ) >> 3;
692 		for (i = 0; i < n; i++) {
693 			union ubifs_key key1, key2;
694 
695 			inum = le64_to_cpu(orph->inos[i]);
696 			dbg_rcvry("deleting orphaned inode %lu",
697 				  (unsigned long)inum);
698 
699 			lowest_ino_key(c, &key1, inum);
700 			highest_ino_key(c, &key2, inum);
701 
702 			err = ubifs_tnc_remove_range(c, &key1, &key2);
703 			if (err)
704 				return err;
705 			err = insert_dead_orphan(c, inum);
706 			if (err)
707 				return err;
708 		}
709 
710 		*last_cmt_no = cmt_no;
711 		if (le64_to_cpu(orph->cmt_no) & (1ULL << 63)) {
712 			dbg_rcvry("last orph node for commit %llu at %d:%d",
713 				  cmt_no, sleb->lnum, snod->offs);
714 			*last_flagged = 1;
715 		} else
716 			*last_flagged = 0;
717 	}
718 
719 	return 0;
720 }
721 
722 /**
723  * kill_orphans - remove all orphan inodes from the index.
724  * @c: UBIFS file-system description object
725  *
726  * If recovery is required, then orphan inodes recorded during the previous
727  * session (which ended with an unclean unmount) must be deleted from the index.
728  * This is done by updating the TNC, but since the index is not updated until
729  * the next commit, the LEBs where the orphan information is recorded are not
730  * erased until the next commit.
731  */
732 static int kill_orphans(struct ubifs_info *c)
733 {
734 	unsigned long long last_cmt_no = 0;
735 	int lnum, err = 0, outofdate = 0, last_flagged = 0;
736 
737 	c->ohead_lnum = c->orph_first;
738 	c->ohead_offs = 0;
739 	/* Check no-orphans flag and skip this if no orphans */
740 	if (c->no_orphs) {
741 		dbg_rcvry("no orphans");
742 		return 0;
743 	}
744 	/*
745 	 * Orph nodes always start at c->orph_first and are written to each
746 	 * successive LEB in turn. Generally unused LEBs will have been unmapped
747 	 * but may contain out of date orphan nodes if the unmap didn't go
748 	 * through. In addition, the last orphan node written for each commit is
749 	 * marked (top bit of orph->cmt_no is set to 1). It is possible that
750 	 * there are orphan nodes from the next commit (i.e. the commit did not
751 	 * complete successfully). In that case, no orphans will have been lost
752 	 * due to the way that orphans are written, and any orphans added will
753 	 * be valid orphans anyway and so can be deleted.
754 	 */
755 	for (lnum = c->orph_first; lnum <= c->orph_last; lnum++) {
756 		struct ubifs_scan_leb *sleb;
757 
758 		dbg_rcvry("LEB %d", lnum);
759 		sleb = ubifs_scan(c, lnum, 0, c->sbuf, 1);
760 		if (IS_ERR(sleb)) {
761 			if (PTR_ERR(sleb) == -EUCLEAN)
762 				sleb = ubifs_recover_leb(c, lnum, 0,
763 							 c->sbuf, -1);
764 			if (IS_ERR(sleb)) {
765 				err = PTR_ERR(sleb);
766 				break;
767 			}
768 		}
769 		err = do_kill_orphans(c, sleb, &last_cmt_no, &outofdate,
770 				      &last_flagged);
771 		if (err || outofdate) {
772 			ubifs_scan_destroy(sleb);
773 			break;
774 		}
775 		if (sleb->endpt) {
776 			c->ohead_lnum = lnum;
777 			c->ohead_offs = sleb->endpt;
778 		}
779 		ubifs_scan_destroy(sleb);
780 	}
781 	return err;
782 }
783 
784 /**
785  * ubifs_mount_orphans - delete orphan inodes and erase LEBs that recorded them.
786  * @c: UBIFS file-system description object
787  * @unclean: indicates recovery from unclean unmount
788  * @read_only: indicates read only mount
789  *
790  * This function is called when mounting to erase orphans from the previous
791  * session. If UBIFS was not unmounted cleanly, then the inodes recorded as
792  * orphans are deleted.
793  */
794 int ubifs_mount_orphans(struct ubifs_info *c, int unclean, int read_only)
795 {
796 	int err = 0;
797 
798 	c->max_orphans = tot_avail_orphs(c);
799 
800 	if (!read_only) {
801 		c->orph_buf = vmalloc(c->leb_size);
802 		if (!c->orph_buf)
803 			return -ENOMEM;
804 	}
805 
806 	if (unclean)
807 		err = kill_orphans(c);
808 	else if (!read_only)
809 		err = ubifs_clear_orphans(c);
810 
811 	return err;
812 }
813 
814 /*
815  * Everything below is related to debugging.
816  */
817 
818 struct check_orphan {
819 	struct rb_node rb;
820 	ino_t inum;
821 };
822 
823 struct check_info {
824 	unsigned long last_ino;
825 	unsigned long tot_inos;
826 	unsigned long missing;
827 	unsigned long long leaf_cnt;
828 	struct ubifs_ino_node *node;
829 	struct rb_root root;
830 };
831 
832 static bool dbg_find_orphan(struct ubifs_info *c, ino_t inum)
833 {
834 	bool found = false;
835 
836 	spin_lock(&c->orphan_lock);
837 	found = !!lookup_orphan(c, inum);
838 	spin_unlock(&c->orphan_lock);
839 
840 	return found;
841 }
842 
843 static int dbg_ins_check_orphan(struct rb_root *root, ino_t inum)
844 {
845 	struct check_orphan *orphan, *o;
846 	struct rb_node **p, *parent = NULL;
847 
848 	orphan = kzalloc(sizeof(struct check_orphan), GFP_NOFS);
849 	if (!orphan)
850 		return -ENOMEM;
851 	orphan->inum = inum;
852 
853 	p = &root->rb_node;
854 	while (*p) {
855 		parent = *p;
856 		o = rb_entry(parent, struct check_orphan, rb);
857 		if (inum < o->inum)
858 			p = &(*p)->rb_left;
859 		else if (inum > o->inum)
860 			p = &(*p)->rb_right;
861 		else {
862 			kfree(orphan);
863 			return 0;
864 		}
865 	}
866 	rb_link_node(&orphan->rb, parent, p);
867 	rb_insert_color(&orphan->rb, root);
868 	return 0;
869 }
870 
871 static int dbg_find_check_orphan(struct rb_root *root, ino_t inum)
872 {
873 	struct check_orphan *o;
874 	struct rb_node *p;
875 
876 	p = root->rb_node;
877 	while (p) {
878 		o = rb_entry(p, struct check_orphan, rb);
879 		if (inum < o->inum)
880 			p = p->rb_left;
881 		else if (inum > o->inum)
882 			p = p->rb_right;
883 		else
884 			return 1;
885 	}
886 	return 0;
887 }
888 
889 static void dbg_free_check_tree(struct rb_root *root)
890 {
891 	struct check_orphan *o, *n;
892 
893 	rbtree_postorder_for_each_entry_safe(o, n, root, rb)
894 		kfree(o);
895 }
896 
897 static int dbg_orphan_check(struct ubifs_info *c, struct ubifs_zbranch *zbr,
898 			    void *priv)
899 {
900 	struct check_info *ci = priv;
901 	ino_t inum;
902 	int err;
903 
904 	inum = key_inum(c, &zbr->key);
905 	if (inum != ci->last_ino) {
906 		/* Lowest node type is the inode node, so it comes first */
907 		if (key_type(c, &zbr->key) != UBIFS_INO_KEY)
908 			ubifs_err(c, "found orphan node ino %lu, type %d",
909 				  (unsigned long)inum, key_type(c, &zbr->key));
910 		ci->last_ino = inum;
911 		ci->tot_inos += 1;
912 		err = ubifs_tnc_read_node(c, zbr, ci->node);
913 		if (err) {
914 			ubifs_err(c, "node read failed, error %d", err);
915 			return err;
916 		}
917 		if (ci->node->nlink == 0)
918 			/* Must be recorded as an orphan */
919 			if (!dbg_find_check_orphan(&ci->root, inum) &&
920 			    !dbg_find_orphan(c, inum)) {
921 				ubifs_err(c, "missing orphan, ino %lu",
922 					  (unsigned long)inum);
923 				ci->missing += 1;
924 			}
925 	}
926 	ci->leaf_cnt += 1;
927 	return 0;
928 }
929 
930 static int dbg_read_orphans(struct check_info *ci, struct ubifs_scan_leb *sleb)
931 {
932 	struct ubifs_scan_node *snod;
933 	struct ubifs_orph_node *orph;
934 	ino_t inum;
935 	int i, n, err;
936 
937 	list_for_each_entry(snod, &sleb->nodes, list) {
938 		cond_resched();
939 		if (snod->type != UBIFS_ORPH_NODE)
940 			continue;
941 		orph = snod->node;
942 		n = (le32_to_cpu(orph->ch.len) - UBIFS_ORPH_NODE_SZ) >> 3;
943 		for (i = 0; i < n; i++) {
944 			inum = le64_to_cpu(orph->inos[i]);
945 			err = dbg_ins_check_orphan(&ci->root, inum);
946 			if (err)
947 				return err;
948 		}
949 	}
950 	return 0;
951 }
952 
953 static int dbg_scan_orphans(struct ubifs_info *c, struct check_info *ci)
954 {
955 	int lnum, err = 0;
956 	void *buf;
957 
958 	/* Check no-orphans flag and skip this if no orphans */
959 	if (c->no_orphs)
960 		return 0;
961 
962 	buf = __vmalloc(c->leb_size, GFP_NOFS, PAGE_KERNEL);
963 	if (!buf) {
964 		ubifs_err(c, "cannot allocate memory to check orphans");
965 		return 0;
966 	}
967 
968 	for (lnum = c->orph_first; lnum <= c->orph_last; lnum++) {
969 		struct ubifs_scan_leb *sleb;
970 
971 		sleb = ubifs_scan(c, lnum, 0, buf, 0);
972 		if (IS_ERR(sleb)) {
973 			err = PTR_ERR(sleb);
974 			break;
975 		}
976 
977 		err = dbg_read_orphans(ci, sleb);
978 		ubifs_scan_destroy(sleb);
979 		if (err)
980 			break;
981 	}
982 
983 	vfree(buf);
984 	return err;
985 }
986 
987 static int dbg_check_orphans(struct ubifs_info *c)
988 {
989 	struct check_info ci;
990 	int err;
991 
992 	if (!dbg_is_chk_orph(c))
993 		return 0;
994 
995 	ci.last_ino = 0;
996 	ci.tot_inos = 0;
997 	ci.missing  = 0;
998 	ci.leaf_cnt = 0;
999 	ci.root = RB_ROOT;
1000 	ci.node = kmalloc(UBIFS_MAX_INO_NODE_SZ, GFP_NOFS);
1001 	if (!ci.node) {
1002 		ubifs_err(c, "out of memory");
1003 		return -ENOMEM;
1004 	}
1005 
1006 	err = dbg_scan_orphans(c, &ci);
1007 	if (err)
1008 		goto out;
1009 
1010 	err = dbg_walk_index(c, &dbg_orphan_check, NULL, &ci);
1011 	if (err) {
1012 		ubifs_err(c, "cannot scan TNC, error %d", err);
1013 		goto out;
1014 	}
1015 
1016 	if (ci.missing) {
1017 		ubifs_err(c, "%lu missing orphan(s)", ci.missing);
1018 		err = -EINVAL;
1019 		goto out;
1020 	}
1021 
1022 	dbg_cmt("last inode number is %lu", ci.last_ino);
1023 	dbg_cmt("total number of inodes is %lu", ci.tot_inos);
1024 	dbg_cmt("total number of leaf nodes is %llu", ci.leaf_cnt);
1025 
1026 out:
1027 	dbg_free_check_tree(&ci.root);
1028 	kfree(ci.node);
1029 	return err;
1030 }
1031