1 // SPDX-License-Identifier: GPL-2.0+
3 * This file is part of UBIFS.
5 * Copyright (C) 2006-2008 Nokia Corporation.
7 * Authors: Adrian Hunter
8 * Artem Bityutskiy (Битюцкий Артём)
12 * This file contains journal replay code. It runs when the file-system is being
13 * mounted and requires no locking.
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.
25 #include <dm/devres.h>
26 #include <linux/compat.h>
27 #include <linux/err.h>
30 #include <linux/bug.h>
31 #include <linux/list_sort.h>
34 * struct replay_entry - replay list entry.
35 * @lnum: logical eraseblock number of the node
38 * @deletion: non-zero if this entry corresponds to a node deletion
39 * @sqnum: node sequence number
40 * @list: links the replay list
42 * @nm: directory entry name
43 * @old_size: truncation old size
44 * @new_size: truncation new size
46 * The replay process first scans all buds and builds the replay list, then
47 * sorts the replay list in nodes sequence number order, and then inserts all
48 * the replay entries to the TNC.
54 unsigned int deletion:1;
55 unsigned long long sqnum;
56 struct list_head list;
68 * struct bud_entry - entry in the list of buds to replay.
69 * @list: next bud in the list
70 * @bud: bud description object
71 * @sqnum: reference node sequence number
72 * @free: free bytes in the bud
73 * @dirty: dirty bytes in the bud
76 struct list_head list;
77 struct ubifs_bud *bud;
78 unsigned long long sqnum;
84 * set_bud_lprops - set free and dirty space used by a bud.
85 * @c: UBIFS file-system description object
86 * @b: bud entry which describes the bud
88 * This function makes sure the LEB properties of bud @b are set correctly
89 * after the replay. Returns zero in case of success and a negative error code
92 static int set_bud_lprops(struct ubifs_info *c, struct bud_entry *b)
94 const struct ubifs_lprops *lp;
99 lp = ubifs_lpt_lookup_dirty(c, b->bud->lnum);
106 if (b->bud->start == 0 && (lp->free != c->leb_size || lp->dirty != 0)) {
108 * The LEB was added to the journal with a starting offset of
109 * zero which means the LEB must have been empty. The LEB
110 * property values should be @lp->free == @c->leb_size and
111 * @lp->dirty == 0, but that is not the case. The reason is that
112 * the LEB had been garbage collected before it became the bud,
113 * and there was not commit inbetween. The garbage collector
114 * resets the free and dirty space without recording it
115 * anywhere except lprops, so if there was no commit then
116 * lprops does not have that information.
118 * We do not need to adjust free space because the scan has told
119 * us the exact value which is recorded in the replay entry as
122 * However we do need to subtract from the dirty space the
123 * amount of space that the garbage collector reclaimed, which
124 * is the whole LEB minus the amount of space that was free.
126 dbg_mnt("bud LEB %d was GC'd (%d free, %d dirty)", b->bud->lnum,
127 lp->free, lp->dirty);
128 dbg_gc("bud LEB %d was GC'd (%d free, %d dirty)", b->bud->lnum,
129 lp->free, lp->dirty);
130 dirty -= c->leb_size - lp->free;
132 * If the replay order was perfect the dirty space would now be
133 * zero. The order is not perfect because the journal heads
134 * race with each other. This is not a problem but is does mean
135 * that the dirty space may temporarily exceed c->leb_size
139 dbg_mnt("LEB %d lp: %d free %d dirty replay: %d free %d dirty",
140 b->bud->lnum, lp->free, lp->dirty, b->free,
143 lp = ubifs_change_lp(c, lp, b->free, dirty + b->dirty,
144 lp->flags | LPROPS_TAKEN, 0);
150 /* Make sure the journal head points to the latest bud */
151 err = ubifs_wbuf_seek_nolock(&c->jheads[b->bud->jhead].wbuf,
152 b->bud->lnum, c->leb_size - b->free);
155 ubifs_release_lprops(c);
160 * set_buds_lprops - set free and dirty space for all replayed buds.
161 * @c: UBIFS file-system description object
163 * This function sets LEB properties for all replayed buds. Returns zero in
164 * case of success and a negative error code in case of failure.
166 static int set_buds_lprops(struct ubifs_info *c)
171 list_for_each_entry(b, &c->replay_buds, list) {
172 err = set_bud_lprops(c, b);
181 * trun_remove_range - apply a replay entry for a truncation to the TNC.
182 * @c: UBIFS file-system description object
183 * @r: replay entry of truncation
185 static int trun_remove_range(struct ubifs_info *c, struct replay_entry *r)
187 unsigned min_blk, max_blk;
188 union ubifs_key min_key, max_key;
191 min_blk = r->new_size / UBIFS_BLOCK_SIZE;
192 if (r->new_size & (UBIFS_BLOCK_SIZE - 1))
195 max_blk = r->old_size / UBIFS_BLOCK_SIZE;
196 if ((r->old_size & (UBIFS_BLOCK_SIZE - 1)) == 0)
199 ino = key_inum(c, &r->key);
201 data_key_init(c, &min_key, ino, min_blk);
202 data_key_init(c, &max_key, ino, max_blk);
204 return ubifs_tnc_remove_range(c, &min_key, &max_key);
208 * apply_replay_entry - apply a replay entry to the TNC.
209 * @c: UBIFS file-system description object
210 * @r: replay entry to apply
212 * Apply a replay entry to the TNC.
214 static int apply_replay_entry(struct ubifs_info *c, struct replay_entry *r)
218 dbg_mntk(&r->key, "LEB %d:%d len %d deletion %d sqnum %llu key ",
219 r->lnum, r->offs, r->len, r->deletion, r->sqnum);
221 /* Set c->replay_sqnum to help deal with dangling branches. */
222 c->replay_sqnum = r->sqnum;
224 if (is_hash_key(c, &r->key)) {
226 err = ubifs_tnc_remove_nm(c, &r->key, &r->nm);
228 err = ubifs_tnc_add_nm(c, &r->key, r->lnum, r->offs,
232 switch (key_type(c, &r->key)) {
235 ino_t inum = key_inum(c, &r->key);
237 err = ubifs_tnc_remove_ino(c, inum);
241 err = trun_remove_range(c, r);
244 err = ubifs_tnc_remove(c, &r->key);
248 err = ubifs_tnc_add(c, &r->key, r->lnum, r->offs,
253 if (c->need_recovery)
254 err = ubifs_recover_size_accum(c, &r->key, r->deletion,
262 * replay_entries_cmp - compare 2 replay entries.
263 * @priv: UBIFS file-system description object
264 * @a: first replay entry
265 * @a: second replay entry
267 * This is a comparios function for 'list_sort()' which compares 2 replay
268 * entries @a and @b by comparing their sequence numer. Returns %1 if @a has
269 * greater sequence number and %-1 otherwise.
271 static int replay_entries_cmp(void *priv, struct list_head *a,
274 struct replay_entry *ra, *rb;
280 ra = list_entry(a, struct replay_entry, list);
281 rb = list_entry(b, struct replay_entry, list);
282 ubifs_assert(ra->sqnum != rb->sqnum);
283 if (ra->sqnum > rb->sqnum)
289 * apply_replay_list - apply the replay list to the TNC.
290 * @c: UBIFS file-system description object
292 * Apply all entries in the replay list to the TNC. Returns zero in case of
293 * success and a negative error code in case of failure.
295 static int apply_replay_list(struct ubifs_info *c)
297 struct replay_entry *r;
300 list_sort(c, &c->replay_list, &replay_entries_cmp);
302 list_for_each_entry(r, &c->replay_list, list) {
305 err = apply_replay_entry(c, r);
314 * destroy_replay_list - destroy the replay.
315 * @c: UBIFS file-system description object
317 * Destroy the replay list.
319 static void destroy_replay_list(struct ubifs_info *c)
321 struct replay_entry *r, *tmp;
323 list_for_each_entry_safe(r, tmp, &c->replay_list, list) {
324 if (is_hash_key(c, &r->key))
332 * insert_node - insert a node to the replay list
333 * @c: UBIFS file-system description object
334 * @lnum: node logical eraseblock number
338 * @sqnum: sequence number
339 * @deletion: non-zero if this is a deletion
340 * @used: number of bytes in use in a LEB
341 * @old_size: truncation old size
342 * @new_size: truncation new size
344 * This function inserts a scanned non-direntry node to the replay list. The
345 * replay list contains @struct replay_entry elements, and we sort this list in
346 * sequence number order before applying it. The replay list is applied at the
347 * very end of the replay process. Since the list is sorted in sequence number
348 * order, the older modifications are applied first. This function returns zero
349 * in case of success and a negative error code in case of failure.
351 static int insert_node(struct ubifs_info *c, int lnum, int offs, int len,
352 union ubifs_key *key, unsigned long long sqnum,
353 int deletion, int *used, loff_t old_size,
356 struct replay_entry *r;
358 dbg_mntk(key, "add LEB %d:%d, key ", lnum, offs);
360 if (key_inum(c, key) >= c->highest_inum)
361 c->highest_inum = key_inum(c, key);
363 r = kzalloc(sizeof(struct replay_entry), GFP_KERNEL);
368 *used += ALIGN(len, 8);
372 r->deletion = !!deletion;
374 key_copy(c, key, &r->key);
375 r->old_size = old_size;
376 r->new_size = new_size;
378 list_add_tail(&r->list, &c->replay_list);
383 * insert_dent - insert a directory entry node into the replay list.
384 * @c: UBIFS file-system description object
385 * @lnum: node logical eraseblock number
389 * @name: directory entry name
390 * @nlen: directory entry name length
391 * @sqnum: sequence number
392 * @deletion: non-zero if this is a deletion
393 * @used: number of bytes in use in a LEB
395 * This function inserts a scanned directory entry node or an extended
396 * attribute entry to the replay list. Returns zero in case of success and a
397 * negative error code in case of failure.
399 static int insert_dent(struct ubifs_info *c, int lnum, int offs, int len,
400 union ubifs_key *key, const char *name, int nlen,
401 unsigned long long sqnum, int deletion, int *used)
403 struct replay_entry *r;
406 dbg_mntk(key, "add LEB %d:%d, key ", lnum, offs);
407 if (key_inum(c, key) >= c->highest_inum)
408 c->highest_inum = key_inum(c, key);
410 r = kzalloc(sizeof(struct replay_entry), GFP_KERNEL);
414 nbuf = kmalloc(nlen + 1, GFP_KERNEL);
421 *used += ALIGN(len, 8);
425 r->deletion = !!deletion;
427 key_copy(c, key, &r->key);
429 memcpy(nbuf, name, nlen);
433 list_add_tail(&r->list, &c->replay_list);
438 * ubifs_validate_entry - validate directory or extended attribute entry node.
439 * @c: UBIFS file-system description object
440 * @dent: the node to validate
442 * This function validates directory or extended attribute entry node @dent.
443 * Returns zero if the node is all right and a %-EINVAL if not.
445 int ubifs_validate_entry(struct ubifs_info *c,
446 const struct ubifs_dent_node *dent)
448 int key_type = key_type_flash(c, dent->key);
449 int nlen = le16_to_cpu(dent->nlen);
451 if (le32_to_cpu(dent->ch.len) != nlen + UBIFS_DENT_NODE_SZ + 1 ||
452 dent->type >= UBIFS_ITYPES_CNT ||
453 nlen > UBIFS_MAX_NLEN || dent->name[nlen] != 0 ||
454 (key_type == UBIFS_XENT_KEY && strnlen(dent->name, nlen) != nlen) ||
455 le64_to_cpu(dent->inum) > MAX_INUM) {
456 ubifs_err(c, "bad %s node", key_type == UBIFS_DENT_KEY ?
457 "directory entry" : "extended attribute entry");
461 if (key_type != UBIFS_DENT_KEY && key_type != UBIFS_XENT_KEY) {
462 ubifs_err(c, "bad key type %d", key_type);
470 * is_last_bud - check if the bud is the last in the journal head.
471 * @c: UBIFS file-system description object
472 * @bud: bud description object
474 * This function checks if bud @bud is the last bud in its journal head. This
475 * information is then used by 'replay_bud()' to decide whether the bud can
476 * have corruptions or not. Indeed, only last buds can be corrupted by power
477 * cuts. Returns %1 if this is the last bud, and %0 if not.
479 static int is_last_bud(struct ubifs_info *c, struct ubifs_bud *bud)
481 struct ubifs_jhead *jh = &c->jheads[bud->jhead];
482 struct ubifs_bud *next;
486 if (list_is_last(&bud->list, &jh->buds_list))
490 * The following is a quirk to make sure we work correctly with UBIFS
491 * images used with older UBIFS.
493 * Normally, the last bud will be the last in the journal head's list
494 * of bud. However, there is one exception if the UBIFS image belongs
495 * to older UBIFS. This is fairly unlikely: one would need to use old
496 * UBIFS, then have a power cut exactly at the right point, and then
497 * try to mount this image with new UBIFS.
499 * The exception is: it is possible to have 2 buds A and B, A goes
500 * before B, and B is the last, bud B is contains no data, and bud A is
501 * corrupted at the end. The reason is that in older versions when the
502 * journal code switched the next bud (from A to B), it first added a
503 * log reference node for the new bud (B), and only after this it
504 * synchronized the write-buffer of current bud (A). But later this was
505 * changed and UBIFS started to always synchronize the write-buffer of
506 * the bud (A) before writing the log reference for the new bud (B).
508 * But because older UBIFS always synchronized A's write-buffer before
509 * writing to B, we can recognize this exceptional situation but
510 * checking the contents of bud B - if it is empty, then A can be
511 * treated as the last and we can recover it.
513 * TODO: remove this piece of code in a couple of years (today it is
516 next = list_entry(bud->list.next, struct ubifs_bud, list);
517 if (!list_is_last(&next->list, &jh->buds_list))
520 err = ubifs_leb_read(c, next->lnum, (char *)&data, next->start, 4, 1);
524 return data == 0xFFFFFFFF;
528 * replay_bud - replay a bud logical eraseblock.
529 * @c: UBIFS file-system description object
530 * @b: bud entry which describes the bud
532 * This function replays bud @bud, recovers it if needed, and adds all nodes
533 * from this bud to the replay list. Returns zero in case of success and a
534 * negative error code in case of failure.
536 static int replay_bud(struct ubifs_info *c, struct bud_entry *b)
538 int is_last = is_last_bud(c, b->bud);
539 int err = 0, used = 0, lnum = b->bud->lnum, offs = b->bud->start;
540 struct ubifs_scan_leb *sleb;
541 struct ubifs_scan_node *snod;
543 dbg_mnt("replay bud LEB %d, head %d, offs %d, is_last %d",
544 lnum, b->bud->jhead, offs, is_last);
546 if (c->need_recovery && is_last)
548 * Recover only last LEBs in the journal heads, because power
549 * cuts may cause corruptions only in these LEBs, because only
550 * these LEBs could possibly be written to at the power cut
553 sleb = ubifs_recover_leb(c, lnum, offs, c->sbuf, b->bud->jhead);
555 sleb = ubifs_scan(c, lnum, offs, c->sbuf, 0);
557 return PTR_ERR(sleb);
560 * The bud does not have to start from offset zero - the beginning of
561 * the 'lnum' LEB may contain previously committed data. One of the
562 * things we have to do in replay is to correctly update lprops with
563 * newer information about this LEB.
565 * At this point lprops thinks that this LEB has 'c->leb_size - offs'
566 * bytes of free space because it only contain information about
569 * But we know that real amount of free space is 'c->leb_size -
570 * sleb->endpt', and the space in the 'lnum' LEB between 'offs' and
571 * 'sleb->endpt' is used by bud data. We have to correctly calculate
572 * how much of these data are dirty and update lprops with this
575 * The dirt in that LEB region is comprised of padding nodes, deletion
576 * nodes, truncation nodes and nodes which are obsoleted by subsequent
577 * nodes in this LEB. So instead of calculating clean space, we
578 * calculate used space ('used' variable).
581 list_for_each_entry(snod, &sleb->nodes, list) {
586 if (snod->sqnum >= SQNUM_WATERMARK) {
587 ubifs_err(c, "file system's life ended");
591 if (snod->sqnum > c->max_sqnum)
592 c->max_sqnum = snod->sqnum;
594 switch (snod->type) {
597 struct ubifs_ino_node *ino = snod->node;
598 loff_t new_size = le64_to_cpu(ino->size);
600 if (le32_to_cpu(ino->nlink) == 0)
602 err = insert_node(c, lnum, snod->offs, snod->len,
603 &snod->key, snod->sqnum, deletion,
607 case UBIFS_DATA_NODE:
609 struct ubifs_data_node *dn = snod->node;
610 loff_t new_size = le32_to_cpu(dn->size) +
611 key_block(c, &snod->key) *
614 err = insert_node(c, lnum, snod->offs, snod->len,
615 &snod->key, snod->sqnum, deletion,
619 case UBIFS_DENT_NODE:
620 case UBIFS_XENT_NODE:
622 struct ubifs_dent_node *dent = snod->node;
624 err = ubifs_validate_entry(c, dent);
628 err = insert_dent(c, lnum, snod->offs, snod->len,
629 &snod->key, dent->name,
630 le16_to_cpu(dent->nlen), snod->sqnum,
631 !le64_to_cpu(dent->inum), &used);
634 case UBIFS_TRUN_NODE:
636 struct ubifs_trun_node *trun = snod->node;
637 loff_t old_size = le64_to_cpu(trun->old_size);
638 loff_t new_size = le64_to_cpu(trun->new_size);
641 /* Validate truncation node */
642 if (old_size < 0 || old_size > c->max_inode_sz ||
643 new_size < 0 || new_size > c->max_inode_sz ||
644 old_size <= new_size) {
645 ubifs_err(c, "bad truncation node");
650 * Create a fake truncation key just to use the same
651 * functions which expect nodes to have keys.
653 trun_key_init(c, &key, le32_to_cpu(trun->inum));
654 err = insert_node(c, lnum, snod->offs, snod->len,
655 &key, snod->sqnum, 1, &used,
660 ubifs_err(c, "unexpected node type %d in bud LEB %d:%d",
661 snod->type, lnum, snod->offs);
669 ubifs_assert(ubifs_search_bud(c, lnum));
670 ubifs_assert(sleb->endpt - offs >= used);
671 ubifs_assert(sleb->endpt % c->min_io_size == 0);
673 b->dirty = sleb->endpt - offs - used;
674 b->free = c->leb_size - sleb->endpt;
675 dbg_mnt("bud LEB %d replied: dirty %d, free %d",
676 lnum, b->dirty, b->free);
679 ubifs_scan_destroy(sleb);
683 ubifs_err(c, "bad node is at LEB %d:%d", lnum, snod->offs);
684 ubifs_dump_node(c, snod->node);
685 ubifs_scan_destroy(sleb);
690 * replay_buds - replay all buds.
691 * @c: UBIFS file-system description object
693 * This function returns zero in case of success and a negative error code in
696 static int replay_buds(struct ubifs_info *c)
700 unsigned long long prev_sqnum = 0;
702 list_for_each_entry(b, &c->replay_buds, list) {
703 err = replay_bud(c, b);
707 ubifs_assert(b->sqnum > prev_sqnum);
708 prev_sqnum = b->sqnum;
715 * destroy_bud_list - destroy the list of buds to replay.
716 * @c: UBIFS file-system description object
718 static void destroy_bud_list(struct ubifs_info *c)
722 while (!list_empty(&c->replay_buds)) {
723 b = list_entry(c->replay_buds.next, struct bud_entry, list);
730 * add_replay_bud - add a bud to the list of buds to replay.
731 * @c: UBIFS file-system description object
732 * @lnum: bud logical eraseblock number to replay
733 * @offs: bud start offset
734 * @jhead: journal head to which this bud belongs
735 * @sqnum: reference node sequence number
737 * This function returns zero in case of success and a negative error code in
740 static int add_replay_bud(struct ubifs_info *c, int lnum, int offs, int jhead,
741 unsigned long long sqnum)
743 struct ubifs_bud *bud;
746 dbg_mnt("add replay bud LEB %d:%d, head %d", lnum, offs, jhead);
748 bud = kmalloc(sizeof(struct ubifs_bud), GFP_KERNEL);
752 b = kmalloc(sizeof(struct bud_entry), GFP_KERNEL);
761 ubifs_add_bud(c, bud);
765 list_add_tail(&b->list, &c->replay_buds);
771 * validate_ref - validate a reference node.
772 * @c: UBIFS file-system description object
773 * @ref: the reference node to validate
774 * @ref_lnum: LEB number of the reference node
775 * @ref_offs: reference node offset
777 * This function returns %1 if a bud reference already exists for the LEB. %0 is
778 * returned if the reference node is new, otherwise %-EINVAL is returned if
781 static int validate_ref(struct ubifs_info *c, const struct ubifs_ref_node *ref)
783 struct ubifs_bud *bud;
784 int lnum = le32_to_cpu(ref->lnum);
785 unsigned int offs = le32_to_cpu(ref->offs);
786 unsigned int jhead = le32_to_cpu(ref->jhead);
789 * ref->offs may point to the end of LEB when the journal head points
790 * to the end of LEB and we write reference node for it during commit.
791 * So this is why we require 'offs > c->leb_size'.
793 if (jhead >= c->jhead_cnt || lnum >= c->leb_cnt ||
794 lnum < c->main_first || offs > c->leb_size ||
795 offs & (c->min_io_size - 1))
798 /* Make sure we have not already looked at this bud */
799 bud = ubifs_search_bud(c, lnum);
801 if (bud->jhead == jhead && bud->start <= offs)
803 ubifs_err(c, "bud at LEB %d:%d was already referred", lnum, offs);
811 * replay_log_leb - replay a log logical eraseblock.
812 * @c: UBIFS file-system description object
813 * @lnum: log logical eraseblock to replay
814 * @offs: offset to start replaying from
817 * This function replays a log LEB and returns zero in case of success, %1 if
818 * this is the last LEB in the log, and a negative error code in case of
821 static int replay_log_leb(struct ubifs_info *c, int lnum, int offs, void *sbuf)
824 struct ubifs_scan_leb *sleb;
825 struct ubifs_scan_node *snod;
826 const struct ubifs_cs_node *node;
828 dbg_mnt("replay log LEB %d:%d", lnum, offs);
829 sleb = ubifs_scan(c, lnum, offs, sbuf, c->need_recovery);
831 if (PTR_ERR(sleb) != -EUCLEAN || !c->need_recovery)
832 return PTR_ERR(sleb);
834 * Note, the below function will recover this log LEB only if
835 * it is the last, because unclean reboots can possibly corrupt
836 * only the tail of the log.
838 sleb = ubifs_recover_log_leb(c, lnum, offs, sbuf);
840 return PTR_ERR(sleb);
843 if (sleb->nodes_cnt == 0) {
849 snod = list_entry(sleb->nodes.next, struct ubifs_scan_node, list);
850 if (c->cs_sqnum == 0) {
852 * This is the first log LEB we are looking at, make sure that
853 * the first node is a commit start node. Also record its
854 * sequence number so that UBIFS can determine where the log
855 * ends, because all nodes which were have higher sequence
858 if (snod->type != UBIFS_CS_NODE) {
859 ubifs_err(c, "first log node at LEB %d:%d is not CS node",
863 if (le64_to_cpu(node->cmt_no) != c->cmt_no) {
864 ubifs_err(c, "first CS node at LEB %d:%d has wrong commit number %llu expected %llu",
866 (unsigned long long)le64_to_cpu(node->cmt_no),
871 c->cs_sqnum = le64_to_cpu(node->ch.sqnum);
872 dbg_mnt("commit start sqnum %llu", c->cs_sqnum);
875 if (snod->sqnum < c->cs_sqnum) {
877 * This means that we reached end of log and now
878 * look to the older log data, which was already
879 * committed but the eraseblock was not erased (UBIFS
880 * only un-maps it). So this basically means we have to
881 * exit with "end of log" code.
887 /* Make sure the first node sits at offset zero of the LEB */
888 if (snod->offs != 0) {
889 ubifs_err(c, "first node is not at zero offset");
893 list_for_each_entry(snod, &sleb->nodes, list) {
896 if (snod->sqnum >= SQNUM_WATERMARK) {
897 ubifs_err(c, "file system's life ended");
901 if (snod->sqnum < c->cs_sqnum) {
902 ubifs_err(c, "bad sqnum %llu, commit sqnum %llu",
903 snod->sqnum, c->cs_sqnum);
907 if (snod->sqnum > c->max_sqnum)
908 c->max_sqnum = snod->sqnum;
910 switch (snod->type) {
911 case UBIFS_REF_NODE: {
912 const struct ubifs_ref_node *ref = snod->node;
914 err = validate_ref(c, ref);
916 break; /* Already have this bud */
920 err = add_replay_bud(c, le32_to_cpu(ref->lnum),
921 le32_to_cpu(ref->offs),
922 le32_to_cpu(ref->jhead),
930 /* Make sure it sits at the beginning of LEB */
931 if (snod->offs != 0) {
932 ubifs_err(c, "unexpected node in log");
937 ubifs_err(c, "unexpected node in log");
942 if (sleb->endpt || c->lhead_offs >= c->leb_size) {
943 c->lhead_lnum = lnum;
944 c->lhead_offs = sleb->endpt;
949 ubifs_scan_destroy(sleb);
953 ubifs_err(c, "log error detected while replaying the log at LEB %d:%d",
954 lnum, offs + snod->offs);
955 ubifs_dump_node(c, snod->node);
956 ubifs_scan_destroy(sleb);
961 * take_ihead - update the status of the index head in lprops to 'taken'.
962 * @c: UBIFS file-system description object
964 * This function returns the amount of free space in the index head LEB or a
965 * negative error code.
967 static int take_ihead(struct ubifs_info *c)
969 const struct ubifs_lprops *lp;
974 lp = ubifs_lpt_lookup_dirty(c, c->ihead_lnum);
982 lp = ubifs_change_lp(c, lp, LPROPS_NC, LPROPS_NC,
983 lp->flags | LPROPS_TAKEN, 0);
991 ubifs_release_lprops(c);
996 * ubifs_replay_journal - replay journal.
997 * @c: UBIFS file-system description object
999 * This function scans the journal, replays and cleans it up. It makes sure all
1000 * memory data structures related to uncommitted journal are built (dirty TNC
1001 * tree, tree of buds, modified lprops, etc).
1003 int ubifs_replay_journal(struct ubifs_info *c)
1005 int err, lnum, free;
1007 BUILD_BUG_ON(UBIFS_TRUN_KEY > 5);
1009 /* Update the status of the index head in lprops to 'taken' */
1010 free = take_ihead(c);
1012 return free; /* Error code */
1014 if (c->ihead_offs != c->leb_size - free) {
1015 ubifs_err(c, "bad index head LEB %d:%d", c->ihead_lnum,
1020 dbg_mnt("start replaying the journal");
1022 lnum = c->ltail_lnum = c->lhead_lnum;
1025 err = replay_log_leb(c, lnum, 0, c->sbuf);
1027 if (lnum != c->lhead_lnum)
1028 /* We hit the end of the log */
1032 * The head of the log must always start with the
1033 * "commit start" node on a properly formatted UBIFS.
1034 * But we found no nodes at all, which means that
1035 * someting went wrong and we cannot proceed mounting
1038 ubifs_err(c, "no UBIFS nodes found at the log head LEB %d:%d, possibly corrupted",
1044 lnum = ubifs_next_log_lnum(c, lnum);
1045 } while (lnum != c->ltail_lnum);
1047 err = replay_buds(c);
1051 err = apply_replay_list(c);
1055 err = set_buds_lprops(c);
1060 * UBIFS budgeting calculations use @c->bi.uncommitted_idx variable
1061 * to roughly estimate index growth. Things like @c->bi.min_idx_lebs
1062 * depend on it. This means we have to initialize it to make sure
1063 * budgeting works properly.
1065 c->bi.uncommitted_idx = atomic_long_read(&c->dirty_zn_cnt);
1066 c->bi.uncommitted_idx *= c->max_idx_node_sz;
1068 ubifs_assert(c->bud_bytes <= c->max_bud_bytes || c->need_recovery);
1069 dbg_mnt("finished, log head LEB %d:%d, max_sqnum %llu, highest_inum %lu",
1070 c->lhead_lnum, c->lhead_offs, c->max_sqnum,
1071 (unsigned long)c->highest_inum);
1073 destroy_replay_list(c);
1074 destroy_bud_list(c);