xref: /openbmc/linux/fs/ubifs/orphan.c (revision 63dc02bd)
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 #ifdef CONFIG_UBIFS_FS_DEBUG
56 static int dbg_check_orphans(struct ubifs_info *c);
57 #else
58 #define dbg_check_orphans(c) 0
59 #endif
60 
61 /**
62  * ubifs_add_orphan - add an orphan.
63  * @c: UBIFS file-system description object
64  * @inum: orphan inode number
65  *
66  * Add an orphan. This function is called when an inodes link count drops to
67  * zero.
68  */
69 int ubifs_add_orphan(struct ubifs_info *c, ino_t inum)
70 {
71 	struct ubifs_orphan *orphan, *o;
72 	struct rb_node **p, *parent = NULL;
73 
74 	orphan = kzalloc(sizeof(struct ubifs_orphan), GFP_NOFS);
75 	if (!orphan)
76 		return -ENOMEM;
77 	orphan->inum = inum;
78 	orphan->new = 1;
79 
80 	spin_lock(&c->orphan_lock);
81 	if (c->tot_orphans >= c->max_orphans) {
82 		spin_unlock(&c->orphan_lock);
83 		kfree(orphan);
84 		return -ENFILE;
85 	}
86 	p = &c->orph_tree.rb_node;
87 	while (*p) {
88 		parent = *p;
89 		o = rb_entry(parent, struct ubifs_orphan, rb);
90 		if (inum < o->inum)
91 			p = &(*p)->rb_left;
92 		else if (inum > o->inum)
93 			p = &(*p)->rb_right;
94 		else {
95 			dbg_err("orphaned twice");
96 			spin_unlock(&c->orphan_lock);
97 			kfree(orphan);
98 			return 0;
99 		}
100 	}
101 	c->tot_orphans += 1;
102 	c->new_orphans += 1;
103 	rb_link_node(&orphan->rb, parent, p);
104 	rb_insert_color(&orphan->rb, &c->orph_tree);
105 	list_add_tail(&orphan->list, &c->orph_list);
106 	list_add_tail(&orphan->new_list, &c->orph_new);
107 	spin_unlock(&c->orphan_lock);
108 	dbg_gen("ino %lu", (unsigned long)inum);
109 	return 0;
110 }
111 
112 /**
113  * ubifs_delete_orphan - delete an orphan.
114  * @c: UBIFS file-system description object
115  * @inum: orphan inode number
116  *
117  * Delete an orphan. This function is called when an inode is deleted.
118  */
119 void ubifs_delete_orphan(struct ubifs_info *c, ino_t inum)
120 {
121 	struct ubifs_orphan *o;
122 	struct rb_node *p;
123 
124 	spin_lock(&c->orphan_lock);
125 	p = c->orph_tree.rb_node;
126 	while (p) {
127 		o = rb_entry(p, struct ubifs_orphan, rb);
128 		if (inum < o->inum)
129 			p = p->rb_left;
130 		else if (inum > o->inum)
131 			p = p->rb_right;
132 		else {
133 			if (o->dnext) {
134 				spin_unlock(&c->orphan_lock);
135 				dbg_gen("deleted twice ino %lu",
136 					(unsigned long)inum);
137 				return;
138 			}
139 			if (o->cnext) {
140 				o->dnext = c->orph_dnext;
141 				c->orph_dnext = o;
142 				spin_unlock(&c->orphan_lock);
143 				dbg_gen("delete later ino %lu",
144 					(unsigned long)inum);
145 				return;
146 			}
147 			rb_erase(p, &c->orph_tree);
148 			list_del(&o->list);
149 			c->tot_orphans -= 1;
150 			if (o->new) {
151 				list_del(&o->new_list);
152 				c->new_orphans -= 1;
153 			}
154 			spin_unlock(&c->orphan_lock);
155 			kfree(o);
156 			dbg_gen("inum %lu", (unsigned long)inum);
157 			return;
158 		}
159 	}
160 	spin_unlock(&c->orphan_lock);
161 	dbg_err("missing orphan ino %lu", (unsigned long)inum);
162 	dbg_dump_stack();
163 }
164 
165 /**
166  * ubifs_orphan_start_commit - start commit of orphans.
167  * @c: UBIFS file-system description object
168  *
169  * Start commit of orphans.
170  */
171 int ubifs_orphan_start_commit(struct ubifs_info *c)
172 {
173 	struct ubifs_orphan *orphan, **last;
174 
175 	spin_lock(&c->orphan_lock);
176 	last = &c->orph_cnext;
177 	list_for_each_entry(orphan, &c->orph_new, new_list) {
178 		ubifs_assert(orphan->new);
179 		orphan->new = 0;
180 		*last = orphan;
181 		last = &orphan->cnext;
182 	}
183 	*last = orphan->cnext;
184 	c->cmt_orphans = c->new_orphans;
185 	c->new_orphans = 0;
186 	dbg_cmt("%d orphans to commit", c->cmt_orphans);
187 	INIT_LIST_HEAD(&c->orph_new);
188 	if (c->tot_orphans == 0)
189 		c->no_orphs = 1;
190 	else
191 		c->no_orphs = 0;
192 	spin_unlock(&c->orphan_lock);
193 	return 0;
194 }
195 
196 /**
197  * avail_orphs - calculate available space.
198  * @c: UBIFS file-system description object
199  *
200  * This function returns the number of orphans that can be written in the
201  * available space.
202  */
203 static int avail_orphs(struct ubifs_info *c)
204 {
205 	int avail_lebs, avail, gap;
206 
207 	avail_lebs = c->orph_lebs - (c->ohead_lnum - c->orph_first) - 1;
208 	avail = avail_lebs *
209 	       ((c->leb_size - UBIFS_ORPH_NODE_SZ) / sizeof(__le64));
210 	gap = c->leb_size - c->ohead_offs;
211 	if (gap >= UBIFS_ORPH_NODE_SZ + sizeof(__le64))
212 		avail += (gap - UBIFS_ORPH_NODE_SZ) / sizeof(__le64);
213 	return avail;
214 }
215 
216 /**
217  * tot_avail_orphs - calculate total space.
218  * @c: UBIFS file-system description object
219  *
220  * This function returns the number of orphans that can be written in half
221  * the total space. That leaves half the space for adding new orphans.
222  */
223 static int tot_avail_orphs(struct ubifs_info *c)
224 {
225 	int avail_lebs, avail;
226 
227 	avail_lebs = c->orph_lebs;
228 	avail = avail_lebs *
229 	       ((c->leb_size - UBIFS_ORPH_NODE_SZ) / sizeof(__le64));
230 	return avail / 2;
231 }
232 
233 /**
234  * do_write_orph_node - write a node to the orphan head.
235  * @c: UBIFS file-system description object
236  * @len: length of node
237  * @atomic: write atomically
238  *
239  * This function writes a node to the orphan head from the orphan buffer. If
240  * %atomic is not zero, then the write is done atomically. On success, %0 is
241  * returned, otherwise a negative error code is returned.
242  */
243 static int do_write_orph_node(struct ubifs_info *c, int len, int atomic)
244 {
245 	int err = 0;
246 
247 	if (atomic) {
248 		ubifs_assert(c->ohead_offs == 0);
249 		ubifs_prepare_node(c, c->orph_buf, len, 1);
250 		len = ALIGN(len, c->min_io_size);
251 		err = ubifs_leb_change(c, c->ohead_lnum, c->orph_buf, len,
252 				       UBI_SHORTTERM);
253 	} else {
254 		if (c->ohead_offs == 0) {
255 			/* Ensure LEB has been unmapped */
256 			err = ubifs_leb_unmap(c, c->ohead_lnum);
257 			if (err)
258 				return err;
259 		}
260 		err = ubifs_write_node(c, c->orph_buf, len, c->ohead_lnum,
261 				       c->ohead_offs, UBI_SHORTTERM);
262 	}
263 	return err;
264 }
265 
266 /**
267  * write_orph_node - write an orphan node.
268  * @c: UBIFS file-system description object
269  * @atomic: write atomically
270  *
271  * This function builds an orphan node from the cnext list and writes it to the
272  * orphan head. On success, %0 is returned, otherwise a negative error code
273  * is returned.
274  */
275 static int write_orph_node(struct ubifs_info *c, int atomic)
276 {
277 	struct ubifs_orphan *orphan, *cnext;
278 	struct ubifs_orph_node *orph;
279 	int gap, err, len, cnt, i;
280 
281 	ubifs_assert(c->cmt_orphans > 0);
282 	gap = c->leb_size - c->ohead_offs;
283 	if (gap < UBIFS_ORPH_NODE_SZ + sizeof(__le64)) {
284 		c->ohead_lnum += 1;
285 		c->ohead_offs = 0;
286 		gap = c->leb_size;
287 		if (c->ohead_lnum > c->orph_last) {
288 			/*
289 			 * We limit the number of orphans so that this should
290 			 * never happen.
291 			 */
292 			ubifs_err("out of space in orphan area");
293 			return -EINVAL;
294 		}
295 	}
296 	cnt = (gap - UBIFS_ORPH_NODE_SZ) / sizeof(__le64);
297 	if (cnt > c->cmt_orphans)
298 		cnt = c->cmt_orphans;
299 	len = UBIFS_ORPH_NODE_SZ + cnt * sizeof(__le64);
300 	ubifs_assert(c->orph_buf);
301 	orph = c->orph_buf;
302 	orph->ch.node_type = UBIFS_ORPH_NODE;
303 	spin_lock(&c->orphan_lock);
304 	cnext = c->orph_cnext;
305 	for (i = 0; i < cnt; i++) {
306 		orphan = cnext;
307 		orph->inos[i] = cpu_to_le64(orphan->inum);
308 		cnext = orphan->cnext;
309 		orphan->cnext = NULL;
310 	}
311 	c->orph_cnext = cnext;
312 	c->cmt_orphans -= cnt;
313 	spin_unlock(&c->orphan_lock);
314 	if (c->cmt_orphans)
315 		orph->cmt_no = cpu_to_le64(c->cmt_no);
316 	else
317 		/* Mark the last node of the commit */
318 		orph->cmt_no = cpu_to_le64((c->cmt_no) | (1ULL << 63));
319 	ubifs_assert(c->ohead_offs + len <= c->leb_size);
320 	ubifs_assert(c->ohead_lnum >= c->orph_first);
321 	ubifs_assert(c->ohead_lnum <= c->orph_last);
322 	err = do_write_orph_node(c, len, atomic);
323 	c->ohead_offs += ALIGN(len, c->min_io_size);
324 	c->ohead_offs = ALIGN(c->ohead_offs, 8);
325 	return err;
326 }
327 
328 /**
329  * write_orph_nodes - write orphan nodes until there are no more to commit.
330  * @c: UBIFS file-system description object
331  * @atomic: write atomically
332  *
333  * This function writes orphan nodes for all the orphans to commit. On success,
334  * %0 is returned, otherwise a negative error code is returned.
335  */
336 static int write_orph_nodes(struct ubifs_info *c, int atomic)
337 {
338 	int err;
339 
340 	while (c->cmt_orphans > 0) {
341 		err = write_orph_node(c, atomic);
342 		if (err)
343 			return err;
344 	}
345 	if (atomic) {
346 		int lnum;
347 
348 		/* Unmap any unused LEBs after consolidation */
349 		lnum = c->ohead_lnum + 1;
350 		for (lnum = c->ohead_lnum + 1; lnum <= c->orph_last; lnum++) {
351 			err = ubifs_leb_unmap(c, lnum);
352 			if (err)
353 				return err;
354 		}
355 	}
356 	return 0;
357 }
358 
359 /**
360  * consolidate - consolidate the orphan area.
361  * @c: UBIFS file-system description object
362  *
363  * This function enables consolidation by putting all the orphans into the list
364  * to commit. The list is in the order that the orphans were added, and the
365  * LEBs are written atomically in order, so at no time can orphans be lost by
366  * an unclean unmount.
367  *
368  * This function returns %0 on success and a negative error code on failure.
369  */
370 static int consolidate(struct ubifs_info *c)
371 {
372 	int tot_avail = tot_avail_orphs(c), err = 0;
373 
374 	spin_lock(&c->orphan_lock);
375 	dbg_cmt("there is space for %d orphans and there are %d",
376 		tot_avail, c->tot_orphans);
377 	if (c->tot_orphans - c->new_orphans <= tot_avail) {
378 		struct ubifs_orphan *orphan, **last;
379 		int cnt = 0;
380 
381 		/* Change the cnext list to include all non-new orphans */
382 		last = &c->orph_cnext;
383 		list_for_each_entry(orphan, &c->orph_list, list) {
384 			if (orphan->new)
385 				continue;
386 			*last = orphan;
387 			last = &orphan->cnext;
388 			cnt += 1;
389 		}
390 		*last = orphan->cnext;
391 		ubifs_assert(cnt == c->tot_orphans - c->new_orphans);
392 		c->cmt_orphans = cnt;
393 		c->ohead_lnum = c->orph_first;
394 		c->ohead_offs = 0;
395 	} else {
396 		/*
397 		 * We limit the number of orphans so that this should
398 		 * never happen.
399 		 */
400 		ubifs_err("out of space in orphan area");
401 		err = -EINVAL;
402 	}
403 	spin_unlock(&c->orphan_lock);
404 	return err;
405 }
406 
407 /**
408  * commit_orphans - commit orphans.
409  * @c: UBIFS file-system description object
410  *
411  * This function commits orphans to flash. On success, %0 is returned,
412  * otherwise a negative error code is returned.
413  */
414 static int commit_orphans(struct ubifs_info *c)
415 {
416 	int avail, atomic = 0, err;
417 
418 	ubifs_assert(c->cmt_orphans > 0);
419 	avail = avail_orphs(c);
420 	if (avail < c->cmt_orphans) {
421 		/* Not enough space to write new orphans, so consolidate */
422 		err = consolidate(c);
423 		if (err)
424 			return err;
425 		atomic = 1;
426 	}
427 	err = write_orph_nodes(c, atomic);
428 	return err;
429 }
430 
431 /**
432  * erase_deleted - erase the orphans marked for deletion.
433  * @c: UBIFS file-system description object
434  *
435  * During commit, the orphans being committed cannot be deleted, so they are
436  * marked for deletion and deleted by this function. Also, the recovery
437  * adds killed orphans to the deletion list, and therefore they are deleted
438  * here too.
439  */
440 static void erase_deleted(struct ubifs_info *c)
441 {
442 	struct ubifs_orphan *orphan, *dnext;
443 
444 	spin_lock(&c->orphan_lock);
445 	dnext = c->orph_dnext;
446 	while (dnext) {
447 		orphan = dnext;
448 		dnext = orphan->dnext;
449 		ubifs_assert(!orphan->new);
450 		rb_erase(&orphan->rb, &c->orph_tree);
451 		list_del(&orphan->list);
452 		c->tot_orphans -= 1;
453 		dbg_gen("deleting orphan ino %lu", (unsigned long)orphan->inum);
454 		kfree(orphan);
455 	}
456 	c->orph_dnext = NULL;
457 	spin_unlock(&c->orphan_lock);
458 }
459 
460 /**
461  * ubifs_orphan_end_commit - end commit of orphans.
462  * @c: UBIFS file-system description object
463  *
464  * End commit of orphans.
465  */
466 int ubifs_orphan_end_commit(struct ubifs_info *c)
467 {
468 	int err;
469 
470 	if (c->cmt_orphans != 0) {
471 		err = commit_orphans(c);
472 		if (err)
473 			return err;
474 	}
475 	erase_deleted(c);
476 	err = dbg_check_orphans(c);
477 	return err;
478 }
479 
480 /**
481  * ubifs_clear_orphans - erase all LEBs used for orphans.
482  * @c: UBIFS file-system description object
483  *
484  * If recovery is not required, then the orphans from the previous session
485  * are not needed. This function locates the LEBs used to record
486  * orphans, and un-maps them.
487  */
488 int ubifs_clear_orphans(struct ubifs_info *c)
489 {
490 	int lnum, err;
491 
492 	for (lnum = c->orph_first; lnum <= c->orph_last; lnum++) {
493 		err = ubifs_leb_unmap(c, lnum);
494 		if (err)
495 			return err;
496 	}
497 	c->ohead_lnum = c->orph_first;
498 	c->ohead_offs = 0;
499 	return 0;
500 }
501 
502 /**
503  * insert_dead_orphan - insert an orphan.
504  * @c: UBIFS file-system description object
505  * @inum: orphan inode number
506  *
507  * This function is a helper to the 'do_kill_orphans()' function. The orphan
508  * must be kept until the next commit, so it is added to the rb-tree and the
509  * deletion list.
510  */
511 static int insert_dead_orphan(struct ubifs_info *c, ino_t inum)
512 {
513 	struct ubifs_orphan *orphan, *o;
514 	struct rb_node **p, *parent = NULL;
515 
516 	orphan = kzalloc(sizeof(struct ubifs_orphan), GFP_KERNEL);
517 	if (!orphan)
518 		return -ENOMEM;
519 	orphan->inum = inum;
520 
521 	p = &c->orph_tree.rb_node;
522 	while (*p) {
523 		parent = *p;
524 		o = rb_entry(parent, struct ubifs_orphan, rb);
525 		if (inum < o->inum)
526 			p = &(*p)->rb_left;
527 		else if (inum > o->inum)
528 			p = &(*p)->rb_right;
529 		else {
530 			/* Already added - no problem */
531 			kfree(orphan);
532 			return 0;
533 		}
534 	}
535 	c->tot_orphans += 1;
536 	rb_link_node(&orphan->rb, parent, p);
537 	rb_insert_color(&orphan->rb, &c->orph_tree);
538 	list_add_tail(&orphan->list, &c->orph_list);
539 	orphan->dnext = c->orph_dnext;
540 	c->orph_dnext = orphan;
541 	dbg_mnt("ino %lu, new %d, tot %d", (unsigned long)inum,
542 		c->new_orphans, c->tot_orphans);
543 	return 0;
544 }
545 
546 /**
547  * do_kill_orphans - remove orphan inodes from the index.
548  * @c: UBIFS file-system description object
549  * @sleb: scanned LEB
550  * @last_cmt_no: cmt_no of last orphan node read is passed and returned here
551  * @outofdate: whether the LEB is out of date is returned here
552  * @last_flagged: whether the end orphan node is encountered
553  *
554  * This function is a helper to the 'kill_orphans()' function. It goes through
555  * every orphan node in a LEB and for every inode number recorded, removes
556  * all keys for that inode from the TNC.
557  */
558 static int do_kill_orphans(struct ubifs_info *c, struct ubifs_scan_leb *sleb,
559 			   unsigned long long *last_cmt_no, int *outofdate,
560 			   int *last_flagged)
561 {
562 	struct ubifs_scan_node *snod;
563 	struct ubifs_orph_node *orph;
564 	unsigned long long cmt_no;
565 	ino_t inum;
566 	int i, n, err, first = 1;
567 
568 	list_for_each_entry(snod, &sleb->nodes, list) {
569 		if (snod->type != UBIFS_ORPH_NODE) {
570 			ubifs_err("invalid node type %d in orphan area at "
571 				  "%d:%d", snod->type, sleb->lnum, snod->offs);
572 			dbg_dump_node(c, snod->node);
573 			return -EINVAL;
574 		}
575 
576 		orph = snod->node;
577 
578 		/* Check commit number */
579 		cmt_no = le64_to_cpu(orph->cmt_no) & LLONG_MAX;
580 		/*
581 		 * The commit number on the master node may be less, because
582 		 * of a failed commit. If there are several failed commits in a
583 		 * row, the commit number written on orphan nodes will continue
584 		 * to increase (because the commit number is adjusted here) even
585 		 * though the commit number on the master node stays the same
586 		 * because the master node has not been re-written.
587 		 */
588 		if (cmt_no > c->cmt_no)
589 			c->cmt_no = cmt_no;
590 		if (cmt_no < *last_cmt_no && *last_flagged) {
591 			/*
592 			 * The last orphan node had a higher commit number and
593 			 * was flagged as the last written for that commit
594 			 * number. That makes this orphan node, out of date.
595 			 */
596 			if (!first) {
597 				ubifs_err("out of order commit number %llu in "
598 					  "orphan node at %d:%d",
599 					  cmt_no, sleb->lnum, snod->offs);
600 				dbg_dump_node(c, snod->node);
601 				return -EINVAL;
602 			}
603 			dbg_rcvry("out of date LEB %d", sleb->lnum);
604 			*outofdate = 1;
605 			return 0;
606 		}
607 
608 		if (first)
609 			first = 0;
610 
611 		n = (le32_to_cpu(orph->ch.len) - UBIFS_ORPH_NODE_SZ) >> 3;
612 		for (i = 0; i < n; i++) {
613 			inum = le64_to_cpu(orph->inos[i]);
614 			dbg_rcvry("deleting orphaned inode %lu",
615 				  (unsigned long)inum);
616 			err = ubifs_tnc_remove_ino(c, inum);
617 			if (err)
618 				return err;
619 			err = insert_dead_orphan(c, inum);
620 			if (err)
621 				return err;
622 		}
623 
624 		*last_cmt_no = cmt_no;
625 		if (le64_to_cpu(orph->cmt_no) & (1ULL << 63)) {
626 			dbg_rcvry("last orph node for commit %llu at %d:%d",
627 				  cmt_no, sleb->lnum, snod->offs);
628 			*last_flagged = 1;
629 		} else
630 			*last_flagged = 0;
631 	}
632 
633 	return 0;
634 }
635 
636 /**
637  * kill_orphans - remove all orphan inodes from the index.
638  * @c: UBIFS file-system description object
639  *
640  * If recovery is required, then orphan inodes recorded during the previous
641  * session (which ended with an unclean unmount) must be deleted from the index.
642  * This is done by updating the TNC, but since the index is not updated until
643  * the next commit, the LEBs where the orphan information is recorded are not
644  * erased until the next commit.
645  */
646 static int kill_orphans(struct ubifs_info *c)
647 {
648 	unsigned long long last_cmt_no = 0;
649 	int lnum, err = 0, outofdate = 0, last_flagged = 0;
650 
651 	c->ohead_lnum = c->orph_first;
652 	c->ohead_offs = 0;
653 	/* Check no-orphans flag and skip this if no orphans */
654 	if (c->no_orphs) {
655 		dbg_rcvry("no orphans");
656 		return 0;
657 	}
658 	/*
659 	 * Orph nodes always start at c->orph_first and are written to each
660 	 * successive LEB in turn. Generally unused LEBs will have been unmapped
661 	 * but may contain out of date orphan nodes if the unmap didn't go
662 	 * through. In addition, the last orphan node written for each commit is
663 	 * marked (top bit of orph->cmt_no is set to 1). It is possible that
664 	 * there are orphan nodes from the next commit (i.e. the commit did not
665 	 * complete successfully). In that case, no orphans will have been lost
666 	 * due to the way that orphans are written, and any orphans added will
667 	 * be valid orphans anyway and so can be deleted.
668 	 */
669 	for (lnum = c->orph_first; lnum <= c->orph_last; lnum++) {
670 		struct ubifs_scan_leb *sleb;
671 
672 		dbg_rcvry("LEB %d", lnum);
673 		sleb = ubifs_scan(c, lnum, 0, c->sbuf, 1);
674 		if (IS_ERR(sleb)) {
675 			if (PTR_ERR(sleb) == -EUCLEAN)
676 				sleb = ubifs_recover_leb(c, lnum, 0,
677 							 c->sbuf, -1);
678 			if (IS_ERR(sleb)) {
679 				err = PTR_ERR(sleb);
680 				break;
681 			}
682 		}
683 		err = do_kill_orphans(c, sleb, &last_cmt_no, &outofdate,
684 				      &last_flagged);
685 		if (err || outofdate) {
686 			ubifs_scan_destroy(sleb);
687 			break;
688 		}
689 		if (sleb->endpt) {
690 			c->ohead_lnum = lnum;
691 			c->ohead_offs = sleb->endpt;
692 		}
693 		ubifs_scan_destroy(sleb);
694 	}
695 	return err;
696 }
697 
698 /**
699  * ubifs_mount_orphans - delete orphan inodes and erase LEBs that recorded them.
700  * @c: UBIFS file-system description object
701  * @unclean: indicates recovery from unclean unmount
702  * @read_only: indicates read only mount
703  *
704  * This function is called when mounting to erase orphans from the previous
705  * session. If UBIFS was not unmounted cleanly, then the inodes recorded as
706  * orphans are deleted.
707  */
708 int ubifs_mount_orphans(struct ubifs_info *c, int unclean, int read_only)
709 {
710 	int err = 0;
711 
712 	c->max_orphans = tot_avail_orphs(c);
713 
714 	if (!read_only) {
715 		c->orph_buf = vmalloc(c->leb_size);
716 		if (!c->orph_buf)
717 			return -ENOMEM;
718 	}
719 
720 	if (unclean)
721 		err = kill_orphans(c);
722 	else if (!read_only)
723 		err = ubifs_clear_orphans(c);
724 
725 	return err;
726 }
727 
728 #ifdef CONFIG_UBIFS_FS_DEBUG
729 
730 struct check_orphan {
731 	struct rb_node rb;
732 	ino_t inum;
733 };
734 
735 struct check_info {
736 	unsigned long last_ino;
737 	unsigned long tot_inos;
738 	unsigned long missing;
739 	unsigned long long leaf_cnt;
740 	struct ubifs_ino_node *node;
741 	struct rb_root root;
742 };
743 
744 static int dbg_find_orphan(struct ubifs_info *c, ino_t inum)
745 {
746 	struct ubifs_orphan *o;
747 	struct rb_node *p;
748 
749 	spin_lock(&c->orphan_lock);
750 	p = c->orph_tree.rb_node;
751 	while (p) {
752 		o = rb_entry(p, struct ubifs_orphan, rb);
753 		if (inum < o->inum)
754 			p = p->rb_left;
755 		else if (inum > o->inum)
756 			p = p->rb_right;
757 		else {
758 			spin_unlock(&c->orphan_lock);
759 			return 1;
760 		}
761 	}
762 	spin_unlock(&c->orphan_lock);
763 	return 0;
764 }
765 
766 static int dbg_ins_check_orphan(struct rb_root *root, ino_t inum)
767 {
768 	struct check_orphan *orphan, *o;
769 	struct rb_node **p, *parent = NULL;
770 
771 	orphan = kzalloc(sizeof(struct check_orphan), GFP_NOFS);
772 	if (!orphan)
773 		return -ENOMEM;
774 	orphan->inum = inum;
775 
776 	p = &root->rb_node;
777 	while (*p) {
778 		parent = *p;
779 		o = rb_entry(parent, struct check_orphan, rb);
780 		if (inum < o->inum)
781 			p = &(*p)->rb_left;
782 		else if (inum > o->inum)
783 			p = &(*p)->rb_right;
784 		else {
785 			kfree(orphan);
786 			return 0;
787 		}
788 	}
789 	rb_link_node(&orphan->rb, parent, p);
790 	rb_insert_color(&orphan->rb, root);
791 	return 0;
792 }
793 
794 static int dbg_find_check_orphan(struct rb_root *root, ino_t inum)
795 {
796 	struct check_orphan *o;
797 	struct rb_node *p;
798 
799 	p = root->rb_node;
800 	while (p) {
801 		o = rb_entry(p, struct check_orphan, rb);
802 		if (inum < o->inum)
803 			p = p->rb_left;
804 		else if (inum > o->inum)
805 			p = p->rb_right;
806 		else
807 			return 1;
808 	}
809 	return 0;
810 }
811 
812 static void dbg_free_check_tree(struct rb_root *root)
813 {
814 	struct rb_node *this = root->rb_node;
815 	struct check_orphan *o;
816 
817 	while (this) {
818 		if (this->rb_left) {
819 			this = this->rb_left;
820 			continue;
821 		} else if (this->rb_right) {
822 			this = this->rb_right;
823 			continue;
824 		}
825 		o = rb_entry(this, struct check_orphan, rb);
826 		this = rb_parent(this);
827 		if (this) {
828 			if (this->rb_left == &o->rb)
829 				this->rb_left = NULL;
830 			else
831 				this->rb_right = NULL;
832 		}
833 		kfree(o);
834 	}
835 }
836 
837 static int dbg_orphan_check(struct ubifs_info *c, struct ubifs_zbranch *zbr,
838 			    void *priv)
839 {
840 	struct check_info *ci = priv;
841 	ino_t inum;
842 	int err;
843 
844 	inum = key_inum(c, &zbr->key);
845 	if (inum != ci->last_ino) {
846 		/* Lowest node type is the inode node, so it comes first */
847 		if (key_type(c, &zbr->key) != UBIFS_INO_KEY)
848 			ubifs_err("found orphan node ino %lu, type %d",
849 				  (unsigned long)inum, key_type(c, &zbr->key));
850 		ci->last_ino = inum;
851 		ci->tot_inos += 1;
852 		err = ubifs_tnc_read_node(c, zbr, ci->node);
853 		if (err) {
854 			ubifs_err("node read failed, error %d", err);
855 			return err;
856 		}
857 		if (ci->node->nlink == 0)
858 			/* Must be recorded as an orphan */
859 			if (!dbg_find_check_orphan(&ci->root, inum) &&
860 			    !dbg_find_orphan(c, inum)) {
861 				ubifs_err("missing orphan, ino %lu",
862 					  (unsigned long)inum);
863 				ci->missing += 1;
864 			}
865 	}
866 	ci->leaf_cnt += 1;
867 	return 0;
868 }
869 
870 static int dbg_read_orphans(struct check_info *ci, struct ubifs_scan_leb *sleb)
871 {
872 	struct ubifs_scan_node *snod;
873 	struct ubifs_orph_node *orph;
874 	ino_t inum;
875 	int i, n, err;
876 
877 	list_for_each_entry(snod, &sleb->nodes, list) {
878 		cond_resched();
879 		if (snod->type != UBIFS_ORPH_NODE)
880 			continue;
881 		orph = snod->node;
882 		n = (le32_to_cpu(orph->ch.len) - UBIFS_ORPH_NODE_SZ) >> 3;
883 		for (i = 0; i < n; i++) {
884 			inum = le64_to_cpu(orph->inos[i]);
885 			err = dbg_ins_check_orphan(&ci->root, inum);
886 			if (err)
887 				return err;
888 		}
889 	}
890 	return 0;
891 }
892 
893 static int dbg_scan_orphans(struct ubifs_info *c, struct check_info *ci)
894 {
895 	int lnum, err = 0;
896 	void *buf;
897 
898 	/* Check no-orphans flag and skip this if no orphans */
899 	if (c->no_orphs)
900 		return 0;
901 
902 	buf = __vmalloc(c->leb_size, GFP_NOFS, PAGE_KERNEL);
903 	if (!buf) {
904 		ubifs_err("cannot allocate memory to check orphans");
905 		return 0;
906 	}
907 
908 	for (lnum = c->orph_first; lnum <= c->orph_last; lnum++) {
909 		struct ubifs_scan_leb *sleb;
910 
911 		sleb = ubifs_scan(c, lnum, 0, buf, 0);
912 		if (IS_ERR(sleb)) {
913 			err = PTR_ERR(sleb);
914 			break;
915 		}
916 
917 		err = dbg_read_orphans(ci, sleb);
918 		ubifs_scan_destroy(sleb);
919 		if (err)
920 			break;
921 	}
922 
923 	vfree(buf);
924 	return err;
925 }
926 
927 static int dbg_check_orphans(struct ubifs_info *c)
928 {
929 	struct check_info ci;
930 	int err;
931 
932 	if (!dbg_is_chk_orph(c))
933 		return 0;
934 
935 	ci.last_ino = 0;
936 	ci.tot_inos = 0;
937 	ci.missing  = 0;
938 	ci.leaf_cnt = 0;
939 	ci.root = RB_ROOT;
940 	ci.node = kmalloc(UBIFS_MAX_INO_NODE_SZ, GFP_NOFS);
941 	if (!ci.node) {
942 		ubifs_err("out of memory");
943 		return -ENOMEM;
944 	}
945 
946 	err = dbg_scan_orphans(c, &ci);
947 	if (err)
948 		goto out;
949 
950 	err = dbg_walk_index(c, &dbg_orphan_check, NULL, &ci);
951 	if (err) {
952 		ubifs_err("cannot scan TNC, error %d", err);
953 		goto out;
954 	}
955 
956 	if (ci.missing) {
957 		ubifs_err("%lu missing orphan(s)", ci.missing);
958 		err = -EINVAL;
959 		goto out;
960 	}
961 
962 	dbg_cmt("last inode number is %lu", ci.last_ino);
963 	dbg_cmt("total number of inodes is %lu", ci.tot_inos);
964 	dbg_cmt("total number of leaf nodes is %llu", ci.leaf_cnt);
965 
966 out:
967 	dbg_free_check_tree(&ci.root);
968 	kfree(ci.node);
969 	return err;
970 }
971 
972 #endif /* CONFIG_UBIFS_FS_DEBUG */
973