1 // SPDX-License-Identifier: GPL-2.0-only
2 /*
3 * This file is part of UBIFS.
4 *
5 * Copyright (C) 2006-2008 Nokia Corporation.
6 *
7 * Authors: Adrian Hunter
8 * Artem Bityutskiy (Битюцкий Артём)
9 */
10
11 /*
12 * This file contains journal replay code. It runs when the file-system is being
13 * mounted and requires no locking.
14 *
15 * The larger is the journal, the longer it takes to scan it, so the longer it
16 * takes to mount UBIFS. This is why the journal has limited size which may be
17 * changed depending on the system requirements. But a larger journal gives
18 * faster I/O speed because it writes the index less frequently. So this is a
19 * trade-off. Also, the journal is indexed by the in-memory index (TNC), so the
20 * larger is the journal, the more memory its index may consume.
21 */
22
23 #include "ubifs.h"
24 #include <linux/list_sort.h>
25 #include <crypto/hash.h>
26
27 /**
28 * struct replay_entry - replay list entry.
29 * @lnum: logical eraseblock number of the node
30 * @offs: node offset
31 * @len: node length
32 * @hash: node hash
33 * @deletion: non-zero if this entry corresponds to a node deletion
34 * @sqnum: node sequence number
35 * @list: links the replay list
36 * @key: node key
37 * @nm: directory entry name
38 * @old_size: truncation old size
39 * @new_size: truncation new size
40 *
41 * The replay process first scans all buds and builds the replay list, then
42 * sorts the replay list in nodes sequence number order, and then inserts all
43 * the replay entries to the TNC.
44 */
45 struct replay_entry {
46 int lnum;
47 int offs;
48 int len;
49 u8 hash[UBIFS_HASH_ARR_SZ];
50 unsigned int deletion:1;
51 unsigned long long sqnum;
52 struct list_head list;
53 union ubifs_key key;
54 union {
55 struct fscrypt_name nm;
56 struct {
57 loff_t old_size;
58 loff_t new_size;
59 };
60 };
61 };
62
63 /**
64 * struct bud_entry - entry in the list of buds to replay.
65 * @list: next bud in the list
66 * @bud: bud description object
67 * @sqnum: reference node sequence number
68 * @free: free bytes in the bud
69 * @dirty: dirty bytes in the bud
70 */
71 struct bud_entry {
72 struct list_head list;
73 struct ubifs_bud *bud;
74 unsigned long long sqnum;
75 int free;
76 int dirty;
77 };
78
79 /**
80 * set_bud_lprops - set free and dirty space used by a bud.
81 * @c: UBIFS file-system description object
82 * @b: bud entry which describes the bud
83 *
84 * This function makes sure the LEB properties of bud @b are set correctly
85 * after the replay. Returns zero in case of success and a negative error code
86 * in case of failure.
87 */
set_bud_lprops(struct ubifs_info * c,struct bud_entry * b)88 static int set_bud_lprops(struct ubifs_info *c, struct bud_entry *b)
89 {
90 const struct ubifs_lprops *lp;
91 int err = 0, dirty;
92
93 ubifs_get_lprops(c);
94
95 lp = ubifs_lpt_lookup_dirty(c, b->bud->lnum);
96 if (IS_ERR(lp)) {
97 err = PTR_ERR(lp);
98 goto out;
99 }
100
101 dirty = lp->dirty;
102 if (b->bud->start == 0 && (lp->free != c->leb_size || lp->dirty != 0)) {
103 /*
104 * The LEB was added to the journal with a starting offset of
105 * zero which means the LEB must have been empty. The LEB
106 * property values should be @lp->free == @c->leb_size and
107 * @lp->dirty == 0, but that is not the case. The reason is that
108 * the LEB had been garbage collected before it became the bud,
109 * and there was no commit in between. The garbage collector
110 * resets the free and dirty space without recording it
111 * anywhere except lprops, so if there was no commit then
112 * lprops does not have that information.
113 *
114 * We do not need to adjust free space because the scan has told
115 * us the exact value which is recorded in the replay entry as
116 * @b->free.
117 *
118 * However we do need to subtract from the dirty space the
119 * amount of space that the garbage collector reclaimed, which
120 * is the whole LEB minus the amount of space that was free.
121 */
122 dbg_mnt("bud LEB %d was GC'd (%d free, %d dirty)", b->bud->lnum,
123 lp->free, lp->dirty);
124 dbg_gc("bud LEB %d was GC'd (%d free, %d dirty)", b->bud->lnum,
125 lp->free, lp->dirty);
126 dirty -= c->leb_size - lp->free;
127 /*
128 * If the replay order was perfect the dirty space would now be
129 * zero. The order is not perfect because the journal heads
130 * race with each other. This is not a problem but is does mean
131 * that the dirty space may temporarily exceed c->leb_size
132 * during the replay.
133 */
134 if (dirty != 0)
135 dbg_mnt("LEB %d lp: %d free %d dirty replay: %d free %d dirty",
136 b->bud->lnum, lp->free, lp->dirty, b->free,
137 b->dirty);
138 }
139 lp = ubifs_change_lp(c, lp, b->free, dirty + b->dirty,
140 lp->flags | LPROPS_TAKEN, 0);
141 if (IS_ERR(lp)) {
142 err = PTR_ERR(lp);
143 goto out;
144 }
145
146 /* Make sure the journal head points to the latest bud */
147 err = ubifs_wbuf_seek_nolock(&c->jheads[b->bud->jhead].wbuf,
148 b->bud->lnum, c->leb_size - b->free);
149
150 out:
151 ubifs_release_lprops(c);
152 return err;
153 }
154
155 /**
156 * set_buds_lprops - set free and dirty space for all replayed buds.
157 * @c: UBIFS file-system description object
158 *
159 * This function sets LEB properties for all replayed buds. Returns zero in
160 * case of success and a negative error code in case of failure.
161 */
set_buds_lprops(struct ubifs_info * c)162 static int set_buds_lprops(struct ubifs_info *c)
163 {
164 struct bud_entry *b;
165 int err;
166
167 list_for_each_entry(b, &c->replay_buds, list) {
168 err = set_bud_lprops(c, b);
169 if (err)
170 return err;
171 }
172
173 return 0;
174 }
175
176 /**
177 * trun_remove_range - apply a replay entry for a truncation to the TNC.
178 * @c: UBIFS file-system description object
179 * @r: replay entry of truncation
180 */
trun_remove_range(struct ubifs_info * c,struct replay_entry * r)181 static int trun_remove_range(struct ubifs_info *c, struct replay_entry *r)
182 {
183 unsigned min_blk, max_blk;
184 union ubifs_key min_key, max_key;
185 ino_t ino;
186
187 min_blk = r->new_size / UBIFS_BLOCK_SIZE;
188 if (r->new_size & (UBIFS_BLOCK_SIZE - 1))
189 min_blk += 1;
190
191 max_blk = r->old_size / UBIFS_BLOCK_SIZE;
192 if ((r->old_size & (UBIFS_BLOCK_SIZE - 1)) == 0)
193 max_blk -= 1;
194
195 ino = key_inum(c, &r->key);
196
197 data_key_init(c, &min_key, ino, min_blk);
198 data_key_init(c, &max_key, ino, max_blk);
199
200 return ubifs_tnc_remove_range(c, &min_key, &max_key);
201 }
202
203 /**
204 * inode_still_linked - check whether inode in question will be re-linked.
205 * @c: UBIFS file-system description object
206 * @rino: replay entry to test
207 *
208 * O_TMPFILE files can be re-linked, this means link count goes from 0 to 1.
209 * This case needs special care, otherwise all references to the inode will
210 * be removed upon the first replay entry of an inode with link count 0
211 * is found.
212 */
inode_still_linked(struct ubifs_info * c,struct replay_entry * rino)213 static bool inode_still_linked(struct ubifs_info *c, struct replay_entry *rino)
214 {
215 struct replay_entry *r;
216
217 ubifs_assert(c, rino->deletion);
218 ubifs_assert(c, key_type(c, &rino->key) == UBIFS_INO_KEY);
219
220 /*
221 * Find the most recent entry for the inode behind @rino and check
222 * whether it is a deletion.
223 */
224 list_for_each_entry_reverse(r, &c->replay_list, list) {
225 ubifs_assert(c, r->sqnum >= rino->sqnum);
226 if (key_inum(c, &r->key) == key_inum(c, &rino->key) &&
227 key_type(c, &r->key) == UBIFS_INO_KEY)
228 return r->deletion == 0;
229
230 }
231
232 ubifs_assert(c, 0);
233 return false;
234 }
235
236 /**
237 * apply_replay_entry - apply a replay entry to the TNC.
238 * @c: UBIFS file-system description object
239 * @r: replay entry to apply
240 *
241 * Apply a replay entry to the TNC.
242 */
apply_replay_entry(struct ubifs_info * c,struct replay_entry * r)243 static int apply_replay_entry(struct ubifs_info *c, struct replay_entry *r)
244 {
245 int err;
246
247 dbg_mntk(&r->key, "LEB %d:%d len %d deletion %d sqnum %llu key ",
248 r->lnum, r->offs, r->len, r->deletion, r->sqnum);
249
250 if (is_hash_key(c, &r->key)) {
251 if (r->deletion)
252 err = ubifs_tnc_remove_nm(c, &r->key, &r->nm);
253 else
254 err = ubifs_tnc_add_nm(c, &r->key, r->lnum, r->offs,
255 r->len, r->hash, &r->nm);
256 } else {
257 if (r->deletion)
258 switch (key_type(c, &r->key)) {
259 case UBIFS_INO_KEY:
260 {
261 ino_t inum = key_inum(c, &r->key);
262
263 if (inode_still_linked(c, r)) {
264 err = 0;
265 break;
266 }
267
268 err = ubifs_tnc_remove_ino(c, inum);
269 break;
270 }
271 case UBIFS_TRUN_KEY:
272 err = trun_remove_range(c, r);
273 break;
274 default:
275 err = ubifs_tnc_remove(c, &r->key);
276 break;
277 }
278 else
279 err = ubifs_tnc_add(c, &r->key, r->lnum, r->offs,
280 r->len, r->hash);
281 if (err)
282 return err;
283
284 if (c->need_recovery)
285 err = ubifs_recover_size_accum(c, &r->key, r->deletion,
286 r->new_size);
287 }
288
289 return err;
290 }
291
292 /**
293 * replay_entries_cmp - compare 2 replay entries.
294 * @priv: UBIFS file-system description object
295 * @a: first replay entry
296 * @b: second replay entry
297 *
298 * This is a comparios function for 'list_sort()' which compares 2 replay
299 * entries @a and @b by comparing their sequence number. Returns %1 if @a has
300 * greater sequence number and %-1 otherwise.
301 */
replay_entries_cmp(void * priv,const struct list_head * a,const struct list_head * b)302 static int replay_entries_cmp(void *priv, const struct list_head *a,
303 const struct list_head *b)
304 {
305 struct ubifs_info *c = priv;
306 struct replay_entry *ra, *rb;
307
308 cond_resched();
309 if (a == b)
310 return 0;
311
312 ra = list_entry(a, struct replay_entry, list);
313 rb = list_entry(b, struct replay_entry, list);
314 ubifs_assert(c, ra->sqnum != rb->sqnum);
315 if (ra->sqnum > rb->sqnum)
316 return 1;
317 return -1;
318 }
319
320 /**
321 * apply_replay_list - apply the replay list to the TNC.
322 * @c: UBIFS file-system description object
323 *
324 * Apply all entries in the replay list to the TNC. Returns zero in case of
325 * success and a negative error code in case of failure.
326 */
apply_replay_list(struct ubifs_info * c)327 static int apply_replay_list(struct ubifs_info *c)
328 {
329 struct replay_entry *r;
330 int err;
331
332 list_sort(c, &c->replay_list, &replay_entries_cmp);
333
334 list_for_each_entry(r, &c->replay_list, list) {
335 cond_resched();
336
337 err = apply_replay_entry(c, r);
338 if (err)
339 return err;
340 }
341
342 return 0;
343 }
344
345 /**
346 * destroy_replay_list - destroy the replay.
347 * @c: UBIFS file-system description object
348 *
349 * Destroy the replay list.
350 */
destroy_replay_list(struct ubifs_info * c)351 static void destroy_replay_list(struct ubifs_info *c)
352 {
353 struct replay_entry *r, *tmp;
354
355 list_for_each_entry_safe(r, tmp, &c->replay_list, list) {
356 if (is_hash_key(c, &r->key))
357 kfree(fname_name(&r->nm));
358 list_del(&r->list);
359 kfree(r);
360 }
361 }
362
363 /**
364 * insert_node - insert a node to the replay list
365 * @c: UBIFS file-system description object
366 * @lnum: node logical eraseblock number
367 * @offs: node offset
368 * @len: node length
369 * @hash: node hash
370 * @key: node key
371 * @sqnum: sequence number
372 * @deletion: non-zero if this is a deletion
373 * @used: number of bytes in use in a LEB
374 * @old_size: truncation old size
375 * @new_size: truncation new size
376 *
377 * This function inserts a scanned non-direntry node to the replay list. The
378 * replay list contains @struct replay_entry elements, and we sort this list in
379 * sequence number order before applying it. The replay list is applied at the
380 * very end of the replay process. Since the list is sorted in sequence number
381 * order, the older modifications are applied first. This function returns zero
382 * in case of success and a negative error code in case of failure.
383 */
insert_node(struct ubifs_info * c,int lnum,int offs,int len,const u8 * hash,union ubifs_key * key,unsigned long long sqnum,int deletion,int * used,loff_t old_size,loff_t new_size)384 static int insert_node(struct ubifs_info *c, int lnum, int offs, int len,
385 const u8 *hash, union ubifs_key *key,
386 unsigned long long sqnum, int deletion, int *used,
387 loff_t old_size, loff_t new_size)
388 {
389 struct replay_entry *r;
390
391 dbg_mntk(key, "add LEB %d:%d, key ", lnum, offs);
392
393 if (key_inum(c, key) >= c->highest_inum)
394 c->highest_inum = key_inum(c, key);
395
396 r = kzalloc(sizeof(struct replay_entry), GFP_KERNEL);
397 if (!r)
398 return -ENOMEM;
399
400 if (!deletion)
401 *used += ALIGN(len, 8);
402 r->lnum = lnum;
403 r->offs = offs;
404 r->len = len;
405 ubifs_copy_hash(c, hash, r->hash);
406 r->deletion = !!deletion;
407 r->sqnum = sqnum;
408 key_copy(c, key, &r->key);
409 r->old_size = old_size;
410 r->new_size = new_size;
411
412 list_add_tail(&r->list, &c->replay_list);
413 return 0;
414 }
415
416 /**
417 * insert_dent - insert a directory entry node into the replay list.
418 * @c: UBIFS file-system description object
419 * @lnum: node logical eraseblock number
420 * @offs: node offset
421 * @len: node length
422 * @hash: node hash
423 * @key: node key
424 * @name: directory entry name
425 * @nlen: directory entry name length
426 * @sqnum: sequence number
427 * @deletion: non-zero if this is a deletion
428 * @used: number of bytes in use in a LEB
429 *
430 * This function inserts a scanned directory entry node or an extended
431 * attribute entry to the replay list. Returns zero in case of success and a
432 * negative error code in case of failure.
433 */
insert_dent(struct ubifs_info * c,int lnum,int offs,int len,const u8 * hash,union ubifs_key * key,const char * name,int nlen,unsigned long long sqnum,int deletion,int * used)434 static int insert_dent(struct ubifs_info *c, int lnum, int offs, int len,
435 const u8 *hash, union ubifs_key *key,
436 const char *name, int nlen, unsigned long long sqnum,
437 int deletion, int *used)
438 {
439 struct replay_entry *r;
440 char *nbuf;
441
442 dbg_mntk(key, "add LEB %d:%d, key ", lnum, offs);
443 if (key_inum(c, key) >= c->highest_inum)
444 c->highest_inum = key_inum(c, key);
445
446 r = kzalloc(sizeof(struct replay_entry), GFP_KERNEL);
447 if (!r)
448 return -ENOMEM;
449
450 nbuf = kmalloc(nlen + 1, GFP_KERNEL);
451 if (!nbuf) {
452 kfree(r);
453 return -ENOMEM;
454 }
455
456 if (!deletion)
457 *used += ALIGN(len, 8);
458 r->lnum = lnum;
459 r->offs = offs;
460 r->len = len;
461 ubifs_copy_hash(c, hash, r->hash);
462 r->deletion = !!deletion;
463 r->sqnum = sqnum;
464 key_copy(c, key, &r->key);
465 fname_len(&r->nm) = nlen;
466 memcpy(nbuf, name, nlen);
467 nbuf[nlen] = '\0';
468 fname_name(&r->nm) = nbuf;
469
470 list_add_tail(&r->list, &c->replay_list);
471 return 0;
472 }
473
474 /**
475 * ubifs_validate_entry - validate directory or extended attribute entry node.
476 * @c: UBIFS file-system description object
477 * @dent: the node to validate
478 *
479 * This function validates directory or extended attribute entry node @dent.
480 * Returns zero if the node is all right and a %-EINVAL if not.
481 */
ubifs_validate_entry(struct ubifs_info * c,const struct ubifs_dent_node * dent)482 int ubifs_validate_entry(struct ubifs_info *c,
483 const struct ubifs_dent_node *dent)
484 {
485 int key_type = key_type_flash(c, dent->key);
486 int nlen = le16_to_cpu(dent->nlen);
487
488 if (le32_to_cpu(dent->ch.len) != nlen + UBIFS_DENT_NODE_SZ + 1 ||
489 dent->type >= UBIFS_ITYPES_CNT ||
490 nlen > UBIFS_MAX_NLEN || dent->name[nlen] != 0 ||
491 (key_type == UBIFS_XENT_KEY && strnlen(dent->name, nlen) != nlen) ||
492 le64_to_cpu(dent->inum) > MAX_INUM) {
493 ubifs_err(c, "bad %s node", key_type == UBIFS_DENT_KEY ?
494 "directory entry" : "extended attribute entry");
495 return -EINVAL;
496 }
497
498 if (key_type != UBIFS_DENT_KEY && key_type != UBIFS_XENT_KEY) {
499 ubifs_err(c, "bad key type %d", key_type);
500 return -EINVAL;
501 }
502
503 return 0;
504 }
505
506 /**
507 * is_last_bud - check if the bud is the last in the journal head.
508 * @c: UBIFS file-system description object
509 * @bud: bud description object
510 *
511 * This function checks if bud @bud is the last bud in its journal head. This
512 * information is then used by 'replay_bud()' to decide whether the bud can
513 * have corruptions or not. Indeed, only last buds can be corrupted by power
514 * cuts. Returns %1 if this is the last bud, and %0 if not.
515 */
is_last_bud(struct ubifs_info * c,struct ubifs_bud * bud)516 static int is_last_bud(struct ubifs_info *c, struct ubifs_bud *bud)
517 {
518 struct ubifs_jhead *jh = &c->jheads[bud->jhead];
519 struct ubifs_bud *next;
520 uint32_t data;
521 int err;
522
523 if (list_is_last(&bud->list, &jh->buds_list))
524 return 1;
525
526 /*
527 * The following is a quirk to make sure we work correctly with UBIFS
528 * images used with older UBIFS.
529 *
530 * Normally, the last bud will be the last in the journal head's list
531 * of bud. However, there is one exception if the UBIFS image belongs
532 * to older UBIFS. This is fairly unlikely: one would need to use old
533 * UBIFS, then have a power cut exactly at the right point, and then
534 * try to mount this image with new UBIFS.
535 *
536 * The exception is: it is possible to have 2 buds A and B, A goes
537 * before B, and B is the last, bud B is contains no data, and bud A is
538 * corrupted at the end. The reason is that in older versions when the
539 * journal code switched the next bud (from A to B), it first added a
540 * log reference node for the new bud (B), and only after this it
541 * synchronized the write-buffer of current bud (A). But later this was
542 * changed and UBIFS started to always synchronize the write-buffer of
543 * the bud (A) before writing the log reference for the new bud (B).
544 *
545 * But because older UBIFS always synchronized A's write-buffer before
546 * writing to B, we can recognize this exceptional situation but
547 * checking the contents of bud B - if it is empty, then A can be
548 * treated as the last and we can recover it.
549 *
550 * TODO: remove this piece of code in a couple of years (today it is
551 * 16.05.2011).
552 */
553 next = list_entry(bud->list.next, struct ubifs_bud, list);
554 if (!list_is_last(&next->list, &jh->buds_list))
555 return 0;
556
557 err = ubifs_leb_read(c, next->lnum, (char *)&data, next->start, 4, 1);
558 if (err)
559 return 0;
560
561 return data == 0xFFFFFFFF;
562 }
563
564 /* authenticate_sleb_hash is split out for stack usage */
565 static int noinline_for_stack
authenticate_sleb_hash(struct ubifs_info * c,struct shash_desc * log_hash,u8 * hash)566 authenticate_sleb_hash(struct ubifs_info *c,
567 struct shash_desc *log_hash, u8 *hash)
568 {
569 SHASH_DESC_ON_STACK(hash_desc, c->hash_tfm);
570
571 hash_desc->tfm = c->hash_tfm;
572
573 ubifs_shash_copy_state(c, log_hash, hash_desc);
574 return crypto_shash_final(hash_desc, hash);
575 }
576
577 /**
578 * authenticate_sleb - authenticate one scan LEB
579 * @c: UBIFS file-system description object
580 * @sleb: the scan LEB to authenticate
581 * @log_hash:
582 * @is_last: if true, this is the last LEB
583 *
584 * This function iterates over the buds of a single LEB authenticating all buds
585 * with the authentication nodes on this LEB. Authentication nodes are written
586 * after some buds and contain a HMAC covering the authentication node itself
587 * and the buds between the last authentication node and the current
588 * authentication node. It can happen that the last buds cannot be authenticated
589 * because a powercut happened when some nodes were written but not the
590 * corresponding authentication node. This function returns the number of nodes
591 * that could be authenticated or a negative error code.
592 */
authenticate_sleb(struct ubifs_info * c,struct ubifs_scan_leb * sleb,struct shash_desc * log_hash,int is_last)593 static int authenticate_sleb(struct ubifs_info *c, struct ubifs_scan_leb *sleb,
594 struct shash_desc *log_hash, int is_last)
595 {
596 int n_not_auth = 0;
597 struct ubifs_scan_node *snod;
598 int n_nodes = 0;
599 int err;
600 u8 hash[UBIFS_HASH_ARR_SZ];
601 u8 hmac[UBIFS_HMAC_ARR_SZ];
602
603 if (!ubifs_authenticated(c))
604 return sleb->nodes_cnt;
605
606 list_for_each_entry(snod, &sleb->nodes, list) {
607
608 n_nodes++;
609
610 if (snod->type == UBIFS_AUTH_NODE) {
611 struct ubifs_auth_node *auth = snod->node;
612
613 err = authenticate_sleb_hash(c, log_hash, hash);
614 if (err)
615 goto out;
616
617 err = crypto_shash_tfm_digest(c->hmac_tfm, hash,
618 c->hash_len, hmac);
619 if (err)
620 goto out;
621
622 err = ubifs_check_hmac(c, auth->hmac, hmac);
623 if (err) {
624 err = -EPERM;
625 goto out;
626 }
627 n_not_auth = 0;
628 } else {
629 err = crypto_shash_update(log_hash, snod->node,
630 snod->len);
631 if (err)
632 goto out;
633 n_not_auth++;
634 }
635 }
636
637 /*
638 * A powercut can happen when some nodes were written, but not yet
639 * the corresponding authentication node. This may only happen on
640 * the last bud though.
641 */
642 if (n_not_auth) {
643 if (is_last) {
644 dbg_mnt("%d unauthenticated nodes found on LEB %d, Ignoring them",
645 n_not_auth, sleb->lnum);
646 err = 0;
647 } else {
648 dbg_mnt("%d unauthenticated nodes found on non-last LEB %d",
649 n_not_auth, sleb->lnum);
650 err = -EPERM;
651 }
652 } else {
653 err = 0;
654 }
655 out:
656 return err ? err : n_nodes - n_not_auth;
657 }
658
659 /**
660 * replay_bud - replay a bud logical eraseblock.
661 * @c: UBIFS file-system description object
662 * @b: bud entry which describes the bud
663 *
664 * This function replays bud @bud, recovers it if needed, and adds all nodes
665 * from this bud to the replay list. Returns zero in case of success and a
666 * negative error code in case of failure.
667 */
replay_bud(struct ubifs_info * c,struct bud_entry * b)668 static int replay_bud(struct ubifs_info *c, struct bud_entry *b)
669 {
670 int is_last = is_last_bud(c, b->bud);
671 int err = 0, used = 0, lnum = b->bud->lnum, offs = b->bud->start;
672 int n_nodes, n = 0;
673 struct ubifs_scan_leb *sleb;
674 struct ubifs_scan_node *snod;
675
676 dbg_mnt("replay bud LEB %d, head %d, offs %d, is_last %d",
677 lnum, b->bud->jhead, offs, is_last);
678
679 if (c->need_recovery && is_last)
680 /*
681 * Recover only last LEBs in the journal heads, because power
682 * cuts may cause corruptions only in these LEBs, because only
683 * these LEBs could possibly be written to at the power cut
684 * time.
685 */
686 sleb = ubifs_recover_leb(c, lnum, offs, c->sbuf, b->bud->jhead);
687 else
688 sleb = ubifs_scan(c, lnum, offs, c->sbuf, 0);
689 if (IS_ERR(sleb))
690 return PTR_ERR(sleb);
691
692 n_nodes = authenticate_sleb(c, sleb, b->bud->log_hash, is_last);
693 if (n_nodes < 0) {
694 err = n_nodes;
695 goto out;
696 }
697
698 ubifs_shash_copy_state(c, b->bud->log_hash,
699 c->jheads[b->bud->jhead].log_hash);
700
701 /*
702 * The bud does not have to start from offset zero - the beginning of
703 * the 'lnum' LEB may contain previously committed data. One of the
704 * things we have to do in replay is to correctly update lprops with
705 * newer information about this LEB.
706 *
707 * At this point lprops thinks that this LEB has 'c->leb_size - offs'
708 * bytes of free space because it only contain information about
709 * committed data.
710 *
711 * But we know that real amount of free space is 'c->leb_size -
712 * sleb->endpt', and the space in the 'lnum' LEB between 'offs' and
713 * 'sleb->endpt' is used by bud data. We have to correctly calculate
714 * how much of these data are dirty and update lprops with this
715 * information.
716 *
717 * The dirt in that LEB region is comprised of padding nodes, deletion
718 * nodes, truncation nodes and nodes which are obsoleted by subsequent
719 * nodes in this LEB. So instead of calculating clean space, we
720 * calculate used space ('used' variable).
721 */
722
723 list_for_each_entry(snod, &sleb->nodes, list) {
724 u8 hash[UBIFS_HASH_ARR_SZ];
725 int deletion = 0;
726
727 cond_resched();
728
729 if (snod->sqnum >= SQNUM_WATERMARK) {
730 ubifs_err(c, "file system's life ended");
731 goto out_dump;
732 }
733
734 ubifs_node_calc_hash(c, snod->node, hash);
735
736 if (snod->sqnum > c->max_sqnum)
737 c->max_sqnum = snod->sqnum;
738
739 switch (snod->type) {
740 case UBIFS_INO_NODE:
741 {
742 struct ubifs_ino_node *ino = snod->node;
743 loff_t new_size = le64_to_cpu(ino->size);
744
745 if (le32_to_cpu(ino->nlink) == 0)
746 deletion = 1;
747 err = insert_node(c, lnum, snod->offs, snod->len, hash,
748 &snod->key, snod->sqnum, deletion,
749 &used, 0, new_size);
750 break;
751 }
752 case UBIFS_DATA_NODE:
753 {
754 struct ubifs_data_node *dn = snod->node;
755 loff_t new_size = le32_to_cpu(dn->size) +
756 key_block(c, &snod->key) *
757 UBIFS_BLOCK_SIZE;
758
759 err = insert_node(c, lnum, snod->offs, snod->len, hash,
760 &snod->key, snod->sqnum, deletion,
761 &used, 0, new_size);
762 break;
763 }
764 case UBIFS_DENT_NODE:
765 case UBIFS_XENT_NODE:
766 {
767 struct ubifs_dent_node *dent = snod->node;
768
769 err = ubifs_validate_entry(c, dent);
770 if (err)
771 goto out_dump;
772
773 err = insert_dent(c, lnum, snod->offs, snod->len, hash,
774 &snod->key, dent->name,
775 le16_to_cpu(dent->nlen), snod->sqnum,
776 !le64_to_cpu(dent->inum), &used);
777 break;
778 }
779 case UBIFS_TRUN_NODE:
780 {
781 struct ubifs_trun_node *trun = snod->node;
782 loff_t old_size = le64_to_cpu(trun->old_size);
783 loff_t new_size = le64_to_cpu(trun->new_size);
784 union ubifs_key key;
785
786 /* Validate truncation node */
787 if (old_size < 0 || old_size > c->max_inode_sz ||
788 new_size < 0 || new_size > c->max_inode_sz ||
789 old_size <= new_size) {
790 ubifs_err(c, "bad truncation node");
791 goto out_dump;
792 }
793
794 /*
795 * Create a fake truncation key just to use the same
796 * functions which expect nodes to have keys.
797 */
798 trun_key_init(c, &key, le32_to_cpu(trun->inum));
799 err = insert_node(c, lnum, snod->offs, snod->len, hash,
800 &key, snod->sqnum, 1, &used,
801 old_size, new_size);
802 break;
803 }
804 case UBIFS_AUTH_NODE:
805 break;
806 default:
807 ubifs_err(c, "unexpected node type %d in bud LEB %d:%d",
808 snod->type, lnum, snod->offs);
809 err = -EINVAL;
810 goto out_dump;
811 }
812 if (err)
813 goto out;
814
815 n++;
816 if (n == n_nodes)
817 break;
818 }
819
820 ubifs_assert(c, ubifs_search_bud(c, lnum));
821 ubifs_assert(c, sleb->endpt - offs >= used);
822 ubifs_assert(c, sleb->endpt % c->min_io_size == 0);
823
824 b->dirty = sleb->endpt - offs - used;
825 b->free = c->leb_size - sleb->endpt;
826 dbg_mnt("bud LEB %d replied: dirty %d, free %d",
827 lnum, b->dirty, b->free);
828
829 out:
830 ubifs_scan_destroy(sleb);
831 return err;
832
833 out_dump:
834 ubifs_err(c, "bad node is at LEB %d:%d", lnum, snod->offs);
835 ubifs_dump_node(c, snod->node, c->leb_size - snod->offs);
836 ubifs_scan_destroy(sleb);
837 return -EINVAL;
838 }
839
840 /**
841 * replay_buds - replay all buds.
842 * @c: UBIFS file-system description object
843 *
844 * This function returns zero in case of success and a negative error code in
845 * case of failure.
846 */
replay_buds(struct ubifs_info * c)847 static int replay_buds(struct ubifs_info *c)
848 {
849 struct bud_entry *b;
850 int err;
851 unsigned long long prev_sqnum = 0;
852
853 list_for_each_entry(b, &c->replay_buds, list) {
854 err = replay_bud(c, b);
855 if (err)
856 return err;
857
858 ubifs_assert(c, b->sqnum > prev_sqnum);
859 prev_sqnum = b->sqnum;
860 }
861
862 return 0;
863 }
864
865 /**
866 * destroy_bud_list - destroy the list of buds to replay.
867 * @c: UBIFS file-system description object
868 */
destroy_bud_list(struct ubifs_info * c)869 static void destroy_bud_list(struct ubifs_info *c)
870 {
871 struct bud_entry *b;
872
873 while (!list_empty(&c->replay_buds)) {
874 b = list_entry(c->replay_buds.next, struct bud_entry, list);
875 list_del(&b->list);
876 kfree(b);
877 }
878 }
879
880 /**
881 * add_replay_bud - add a bud to the list of buds to replay.
882 * @c: UBIFS file-system description object
883 * @lnum: bud logical eraseblock number to replay
884 * @offs: bud start offset
885 * @jhead: journal head to which this bud belongs
886 * @sqnum: reference node sequence number
887 *
888 * This function returns zero in case of success and a negative error code in
889 * case of failure.
890 */
add_replay_bud(struct ubifs_info * c,int lnum,int offs,int jhead,unsigned long long sqnum)891 static int add_replay_bud(struct ubifs_info *c, int lnum, int offs, int jhead,
892 unsigned long long sqnum)
893 {
894 struct ubifs_bud *bud;
895 struct bud_entry *b;
896 int err;
897
898 dbg_mnt("add replay bud LEB %d:%d, head %d", lnum, offs, jhead);
899
900 bud = kmalloc(sizeof(struct ubifs_bud), GFP_KERNEL);
901 if (!bud)
902 return -ENOMEM;
903
904 b = kmalloc(sizeof(struct bud_entry), GFP_KERNEL);
905 if (!b) {
906 err = -ENOMEM;
907 goto out;
908 }
909
910 bud->lnum = lnum;
911 bud->start = offs;
912 bud->jhead = jhead;
913 bud->log_hash = ubifs_hash_get_desc(c);
914 if (IS_ERR(bud->log_hash)) {
915 err = PTR_ERR(bud->log_hash);
916 goto out;
917 }
918
919 ubifs_shash_copy_state(c, c->log_hash, bud->log_hash);
920
921 ubifs_add_bud(c, bud);
922
923 b->bud = bud;
924 b->sqnum = sqnum;
925 list_add_tail(&b->list, &c->replay_buds);
926
927 return 0;
928 out:
929 kfree(bud);
930 kfree(b);
931
932 return err;
933 }
934
935 /**
936 * validate_ref - validate a reference node.
937 * @c: UBIFS file-system description object
938 * @ref: the reference node to validate
939 *
940 * This function returns %1 if a bud reference already exists for the LEB. %0 is
941 * returned if the reference node is new, otherwise %-EINVAL is returned if
942 * validation failed.
943 */
validate_ref(struct ubifs_info * c,const struct ubifs_ref_node * ref)944 static int validate_ref(struct ubifs_info *c, const struct ubifs_ref_node *ref)
945 {
946 struct ubifs_bud *bud;
947 int lnum = le32_to_cpu(ref->lnum);
948 unsigned int offs = le32_to_cpu(ref->offs);
949 unsigned int jhead = le32_to_cpu(ref->jhead);
950
951 /*
952 * ref->offs may point to the end of LEB when the journal head points
953 * to the end of LEB and we write reference node for it during commit.
954 * So this is why we require 'offs > c->leb_size'.
955 */
956 if (jhead >= c->jhead_cnt || lnum >= c->leb_cnt ||
957 lnum < c->main_first || offs > c->leb_size ||
958 offs & (c->min_io_size - 1))
959 return -EINVAL;
960
961 /* Make sure we have not already looked at this bud */
962 bud = ubifs_search_bud(c, lnum);
963 if (bud) {
964 if (bud->jhead == jhead && bud->start <= offs)
965 return 1;
966 ubifs_err(c, "bud at LEB %d:%d was already referred", lnum, offs);
967 return -EINVAL;
968 }
969
970 return 0;
971 }
972
973 /**
974 * replay_log_leb - replay a log logical eraseblock.
975 * @c: UBIFS file-system description object
976 * @lnum: log logical eraseblock to replay
977 * @offs: offset to start replaying from
978 * @sbuf: scan buffer
979 *
980 * This function replays a log LEB and returns zero in case of success, %1 if
981 * this is the last LEB in the log, and a negative error code in case of
982 * failure.
983 */
replay_log_leb(struct ubifs_info * c,int lnum,int offs,void * sbuf)984 static int replay_log_leb(struct ubifs_info *c, int lnum, int offs, void *sbuf)
985 {
986 int err;
987 struct ubifs_scan_leb *sleb;
988 struct ubifs_scan_node *snod;
989 const struct ubifs_cs_node *node;
990
991 dbg_mnt("replay log LEB %d:%d", lnum, offs);
992 sleb = ubifs_scan(c, lnum, offs, sbuf, c->need_recovery);
993 if (IS_ERR(sleb)) {
994 if (PTR_ERR(sleb) != -EUCLEAN || !c->need_recovery)
995 return PTR_ERR(sleb);
996 /*
997 * Note, the below function will recover this log LEB only if
998 * it is the last, because unclean reboots can possibly corrupt
999 * only the tail of the log.
1000 */
1001 sleb = ubifs_recover_log_leb(c, lnum, offs, sbuf);
1002 if (IS_ERR(sleb))
1003 return PTR_ERR(sleb);
1004 }
1005
1006 if (sleb->nodes_cnt == 0) {
1007 err = 1;
1008 goto out;
1009 }
1010
1011 node = sleb->buf;
1012 snod = list_entry(sleb->nodes.next, struct ubifs_scan_node, list);
1013 if (c->cs_sqnum == 0) {
1014 /*
1015 * This is the first log LEB we are looking at, make sure that
1016 * the first node is a commit start node. Also record its
1017 * sequence number so that UBIFS can determine where the log
1018 * ends, because all nodes which were have higher sequence
1019 * numbers.
1020 */
1021 if (snod->type != UBIFS_CS_NODE) {
1022 ubifs_err(c, "first log node at LEB %d:%d is not CS node",
1023 lnum, offs);
1024 goto out_dump;
1025 }
1026 if (le64_to_cpu(node->cmt_no) != c->cmt_no) {
1027 ubifs_err(c, "first CS node at LEB %d:%d has wrong commit number %llu expected %llu",
1028 lnum, offs,
1029 (unsigned long long)le64_to_cpu(node->cmt_no),
1030 c->cmt_no);
1031 goto out_dump;
1032 }
1033
1034 c->cs_sqnum = le64_to_cpu(node->ch.sqnum);
1035 dbg_mnt("commit start sqnum %llu", c->cs_sqnum);
1036
1037 err = ubifs_shash_init(c, c->log_hash);
1038 if (err)
1039 goto out;
1040
1041 err = ubifs_shash_update(c, c->log_hash, node, UBIFS_CS_NODE_SZ);
1042 if (err < 0)
1043 goto out;
1044 }
1045
1046 if (snod->sqnum < c->cs_sqnum) {
1047 /*
1048 * This means that we reached end of log and now
1049 * look to the older log data, which was already
1050 * committed but the eraseblock was not erased (UBIFS
1051 * only un-maps it). So this basically means we have to
1052 * exit with "end of log" code.
1053 */
1054 err = 1;
1055 goto out;
1056 }
1057
1058 /* Make sure the first node sits at offset zero of the LEB */
1059 if (snod->offs != 0) {
1060 ubifs_err(c, "first node is not at zero offset");
1061 goto out_dump;
1062 }
1063
1064 list_for_each_entry(snod, &sleb->nodes, list) {
1065 cond_resched();
1066
1067 if (snod->sqnum >= SQNUM_WATERMARK) {
1068 ubifs_err(c, "file system's life ended");
1069 goto out_dump;
1070 }
1071
1072 if (snod->sqnum < c->cs_sqnum) {
1073 ubifs_err(c, "bad sqnum %llu, commit sqnum %llu",
1074 snod->sqnum, c->cs_sqnum);
1075 goto out_dump;
1076 }
1077
1078 if (snod->sqnum > c->max_sqnum)
1079 c->max_sqnum = snod->sqnum;
1080
1081 switch (snod->type) {
1082 case UBIFS_REF_NODE: {
1083 const struct ubifs_ref_node *ref = snod->node;
1084
1085 err = validate_ref(c, ref);
1086 if (err == 1)
1087 break; /* Already have this bud */
1088 if (err)
1089 goto out_dump;
1090
1091 err = ubifs_shash_update(c, c->log_hash, ref,
1092 UBIFS_REF_NODE_SZ);
1093 if (err)
1094 goto out;
1095
1096 err = add_replay_bud(c, le32_to_cpu(ref->lnum),
1097 le32_to_cpu(ref->offs),
1098 le32_to_cpu(ref->jhead),
1099 snod->sqnum);
1100 if (err)
1101 goto out;
1102
1103 break;
1104 }
1105 case UBIFS_CS_NODE:
1106 /* Make sure it sits at the beginning of LEB */
1107 if (snod->offs != 0) {
1108 ubifs_err(c, "unexpected node in log");
1109 goto out_dump;
1110 }
1111 break;
1112 default:
1113 ubifs_err(c, "unexpected node in log");
1114 goto out_dump;
1115 }
1116 }
1117
1118 if (sleb->endpt || c->lhead_offs >= c->leb_size) {
1119 c->lhead_lnum = lnum;
1120 c->lhead_offs = sleb->endpt;
1121 }
1122
1123 err = !sleb->endpt;
1124 out:
1125 ubifs_scan_destroy(sleb);
1126 return err;
1127
1128 out_dump:
1129 ubifs_err(c, "log error detected while replaying the log at LEB %d:%d",
1130 lnum, offs + snod->offs);
1131 ubifs_dump_node(c, snod->node, c->leb_size - snod->offs);
1132 ubifs_scan_destroy(sleb);
1133 return -EINVAL;
1134 }
1135
1136 /**
1137 * take_ihead - update the status of the index head in lprops to 'taken'.
1138 * @c: UBIFS file-system description object
1139 *
1140 * This function returns the amount of free space in the index head LEB or a
1141 * negative error code.
1142 */
take_ihead(struct ubifs_info * c)1143 static int take_ihead(struct ubifs_info *c)
1144 {
1145 const struct ubifs_lprops *lp;
1146 int err, free;
1147
1148 ubifs_get_lprops(c);
1149
1150 lp = ubifs_lpt_lookup_dirty(c, c->ihead_lnum);
1151 if (IS_ERR(lp)) {
1152 err = PTR_ERR(lp);
1153 goto out;
1154 }
1155
1156 free = lp->free;
1157
1158 lp = ubifs_change_lp(c, lp, LPROPS_NC, LPROPS_NC,
1159 lp->flags | LPROPS_TAKEN, 0);
1160 if (IS_ERR(lp)) {
1161 err = PTR_ERR(lp);
1162 goto out;
1163 }
1164
1165 err = free;
1166 out:
1167 ubifs_release_lprops(c);
1168 return err;
1169 }
1170
1171 /**
1172 * ubifs_replay_journal - replay journal.
1173 * @c: UBIFS file-system description object
1174 *
1175 * This function scans the journal, replays and cleans it up. It makes sure all
1176 * memory data structures related to uncommitted journal are built (dirty TNC
1177 * tree, tree of buds, modified lprops, etc).
1178 */
ubifs_replay_journal(struct ubifs_info * c)1179 int ubifs_replay_journal(struct ubifs_info *c)
1180 {
1181 int err, lnum, free;
1182
1183 BUILD_BUG_ON(UBIFS_TRUN_KEY > 5);
1184
1185 /* Update the status of the index head in lprops to 'taken' */
1186 free = take_ihead(c);
1187 if (free < 0)
1188 return free; /* Error code */
1189
1190 if (c->ihead_offs != c->leb_size - free) {
1191 ubifs_err(c, "bad index head LEB %d:%d", c->ihead_lnum,
1192 c->ihead_offs);
1193 return -EINVAL;
1194 }
1195
1196 dbg_mnt("start replaying the journal");
1197 c->replaying = 1;
1198 lnum = c->ltail_lnum = c->lhead_lnum;
1199
1200 do {
1201 err = replay_log_leb(c, lnum, 0, c->sbuf);
1202 if (err == 1) {
1203 if (lnum != c->lhead_lnum)
1204 /* We hit the end of the log */
1205 break;
1206
1207 /*
1208 * The head of the log must always start with the
1209 * "commit start" node on a properly formatted UBIFS.
1210 * But we found no nodes at all, which means that
1211 * something went wrong and we cannot proceed mounting
1212 * the file-system.
1213 */
1214 ubifs_err(c, "no UBIFS nodes found at the log head LEB %d:%d, possibly corrupted",
1215 lnum, 0);
1216 err = -EINVAL;
1217 }
1218 if (err)
1219 goto out;
1220 lnum = ubifs_next_log_lnum(c, lnum);
1221 } while (lnum != c->ltail_lnum);
1222
1223 err = replay_buds(c);
1224 if (err)
1225 goto out;
1226
1227 err = apply_replay_list(c);
1228 if (err)
1229 goto out;
1230
1231 err = set_buds_lprops(c);
1232 if (err)
1233 goto out;
1234
1235 /*
1236 * UBIFS budgeting calculations use @c->bi.uncommitted_idx variable
1237 * to roughly estimate index growth. Things like @c->bi.min_idx_lebs
1238 * depend on it. This means we have to initialize it to make sure
1239 * budgeting works properly.
1240 */
1241 c->bi.uncommitted_idx = atomic_long_read(&c->dirty_zn_cnt);
1242 c->bi.uncommitted_idx *= c->max_idx_node_sz;
1243
1244 ubifs_assert(c, c->bud_bytes <= c->max_bud_bytes || c->need_recovery);
1245 dbg_mnt("finished, log head LEB %d:%d, max_sqnum %llu, highest_inum %lu",
1246 c->lhead_lnum, c->lhead_offs, c->max_sqnum,
1247 (unsigned long)c->highest_inum);
1248 out:
1249 destroy_replay_list(c);
1250 destroy_bud_list(c);
1251 c->replaying = 0;
1252 return err;
1253 }
1254