// SPDX-License-Identifier: GPL-2.0
#include "bcachefs.h"
+#include "backpointers.h"
#include "bkey_buf.h"
#include "alloc_background.h"
#include "btree_gc.h"
#include "buckets.h"
#include "dirent.h"
#include "ec.h"
+#include "errcode.h"
#include "error.h"
#include "fs-common.h"
#include "fsck.h"
return keys->d + idx_to_pos(keys, idx);
}
-static size_t bch2_journal_key_search(struct journal_keys *keys,
- enum btree_id id, unsigned level,
- struct bpos pos)
+static size_t __bch2_journal_key_search(struct journal_keys *keys,
+ enum btree_id id, unsigned level,
+ struct bpos pos)
{
size_t l = 0, r = keys->nr, m;
BUG_ON(l &&
__journal_key_cmp(id, level, pos, idx_to_key(keys, l - 1)) <= 0);
- return idx_to_pos(keys, l);
+ return l;
+}
+
+static size_t bch2_journal_key_search(struct journal_keys *keys,
+ enum btree_id id, unsigned level,
+ struct bpos pos)
+{
+ return idx_to_pos(keys, __bch2_journal_key_search(keys, id, level, pos));
}
struct bkey_i *bch2_journal_keys_peek_upto(struct bch_fs *c, enum btree_id btree_id,
{
struct journal_keys *keys = &c->journal_keys;
unsigned iters = 0;
+ struct journal_key *k;
search:
if (!*idx)
- *idx = bch2_journal_key_search(keys, btree_id, level, pos);
+ *idx = __bch2_journal_key_search(keys, btree_id, level, pos);
- while (*idx < keys->size &&
- keys->d[*idx].btree_id == btree_id &&
- keys->d[*idx].level == level &&
- bpos_cmp(keys->d[*idx].k->k.p, end_pos) <= 0) {
- if (bpos_cmp(keys->d[*idx].k->k.p, pos) >= 0 &&
- !keys->d[*idx].overwritten)
- return keys->d[*idx].k;
+ while (*idx < keys->nr &&
+ (k = idx_to_key(keys, *idx),
+ k->btree_id == btree_id &&
+ k->level == level &&
+ bpos_le(k->k->k.p, end_pos))) {
+ if (bpos_ge(k->k->k.p, pos) && !k->overwritten)
+ return k->k;
(*idx)++;
- if (*idx == keys->gap)
- *idx += keys->size - keys->nr;
-
iters++;
if (iters == 10) {
*idx = 0;
struct journal_keys new_keys = {
.nr = keys->nr,
.size = max_t(size_t, keys->size, 8) * 2,
- .journal_seq_base = keys->journal_seq_base,
};
- new_keys.d = kvmalloc(sizeof(new_keys.d[0]) * new_keys.size, GFP_KERNEL);
+ new_keys.d = kvmalloc_array(new_keys.size, sizeof(new_keys.d[0]), GFP_KERNEL);
if (!new_keys.d) {
bch_err(c, "%s: error allocating new key array (size %zu)",
__func__, new_keys.size);
if (idx < keys->size &&
keys->d[idx].btree_id == btree &&
keys->d[idx].level == level &&
- !bpos_cmp(keys->d[idx].k->k.p, pos))
+ bpos_eq(keys->d[idx].k->k.p, pos))
keys->d[idx].overwritten = true;
}
void bch2_btree_and_journal_iter_advance(struct btree_and_journal_iter *iter)
{
- if (!bpos_cmp(iter->pos, SPOS_MAX))
+ if (bpos_eq(iter->pos, SPOS_MAX))
iter->at_end = true;
else
iter->pos = bpos_successor(iter->pos);
return bkey_s_c_null;
while ((btree_k = bch2_journal_iter_peek_btree(iter)).k &&
- bpos_cmp(btree_k.k->p, iter->pos) < 0)
+ bpos_lt(btree_k.k->p, iter->pos))
bch2_journal_iter_advance_btree(iter);
while ((journal_k = bch2_journal_iter_peek(&iter->journal)).k &&
- bpos_cmp(journal_k.k->p, iter->pos) < 0)
+ bpos_lt(journal_k.k->p, iter->pos))
bch2_journal_iter_advance(&iter->journal);
ret = journal_k.k &&
- (!btree_k.k || bpos_cmp(journal_k.k->p, btree_k.k->p) <= 0)
+ (!btree_k.k || bpos_le(journal_k.k->p, btree_k.k->p))
? journal_k
: btree_k;
- if (ret.k && iter->b && bpos_cmp(ret.k->p, iter->b->data->max_key) > 0)
+ if (ret.k && iter->b && bpos_gt(ret.k->p, iter->b->data->max_key))
ret = bkey_s_c_null;
if (ret.k) {
if (!i || i->ignore)
continue;
- if (!keys->journal_seq_base)
- keys->journal_seq_base = le64_to_cpu(i->j.seq);
-
for_each_jset_key(k, _n, entry, &i->j)
nr_keys++;
}
keys->size = roundup_pow_of_two(nr_keys);
- keys->d = kvmalloc(sizeof(keys->d[0]) * keys->size, GFP_KERNEL);
+ keys->d = kvmalloc_array(keys->size, sizeof(keys->d[0]), GFP_KERNEL);
if (!keys->d)
return -ENOMEM;
if (!i || i->ignore)
continue;
- BUG_ON(le64_to_cpu(i->j.seq) - keys->journal_seq_base > U32_MAX);
-
for_each_jset_key(k, _n, entry, &i->j)
keys->d[keys->nr++] = (struct journal_key) {
.btree_id = entry->btree_id,
.level = entry->level,
.k = k,
- .journal_seq = le64_to_cpu(i->j.seq) -
- keys->journal_seq_base,
+ .journal_seq = le64_to_cpu(i->j.seq),
.journal_offset = k->_data - i->j._data,
};
}
while (src + 1 < keys->d + keys->nr &&
src[0].btree_id == src[1].btree_id &&
src[0].level == src[1].level &&
- !bpos_cmp(src[0].k->k.p, src[1].k->k.p))
+ bpos_eq(src[0].k->k.p, src[1].k->k.p))
src++;
*dst++ = *src++;
sizeof(keys_sorted[0]),
journal_sort_seq_cmp, NULL);
- if (keys->nr)
- replay_now_at(j, keys->journal_seq_base);
-
for (i = 0; i < keys->nr; i++) {
k = keys_sorted[i];
cond_resched();
- replay_now_at(j, keys->journal_seq_base + k->journal_seq);
+ replay_now_at(j, k->journal_seq);
ret = bch2_trans_do(c, NULL, NULL,
BTREE_INSERT_LAZY_RW|
if (k1)
bch2_bkey_val_to_text(&buf1, c, bkey_i_to_s_c(k1));
else
- pr_buf(&buf1, "(none)");
+ prt_printf(&buf1, "(none)");
if (k2)
bch2_bkey_val_to_text(&buf2, c, bkey_i_to_s_c(k2));
else
- pr_buf(&buf2, "(none)");
+ prt_printf(&buf2, "(none)");
mustfix_fsck_err_on(!k1 || !k2 ||
IS_ERR(k1) ||
return ERR_PTR(ret);
}
+static bool btree_id_is_alloc(enum btree_id id)
+{
+ switch (id) {
+ case BTREE_ID_alloc:
+ case BTREE_ID_backpointers:
+ case BTREE_ID_need_discard:
+ case BTREE_ID_freespace:
+ return true;
+ default:
+ return false;
+ }
+}
+
static int read_btree_roots(struct bch_fs *c)
{
unsigned i;
if (!r->alive)
continue;
- if (i == BTREE_ID_alloc &&
+ if (btree_id_is_alloc(i) &&
c->opts.reconstruct_alloc) {
c->sb.compat &= ~(1ULL << BCH_COMPAT_alloc_info);
continue;
}
if (r->error) {
- __fsck_err(c, i == BTREE_ID_alloc
+ __fsck_err(c, btree_id_is_alloc(i)
? FSCK_CAN_IGNORE : 0,
"invalid btree root %s",
bch2_btree_ids[i]);
ret = bch2_btree_root_read(c, i, &r->key, r->level);
if (ret) {
- __fsck_err(c, i == BTREE_ID_alloc
+ __fsck_err(c,
+ btree_id_is_alloc(i)
? FSCK_CAN_IGNORE : 0,
"error reading btree root %s",
bch2_btree_ids[i]);
}
if (!c->opts.nochanges) {
- if (c->sb.version < bcachefs_metadata_version_new_data_types) {
- bch_info(c, "version prior to new_data_types, upgrade and fsck required");
+ if (c->sb.version < bcachefs_metadata_version_backpointers) {
+ bch_info(c, "version prior to backpointers, upgrade and fsck required");
c->opts.version_upgrade = true;
c->opts.fsck = true;
c->opts.fix_errors = FSCK_OPT_YES;
+ } else if (c->sb.version < bcachefs_metadata_version_inode_v3) {
+ bch_info(c, "version prior to inode_v3, upgrade required");
+ c->opts.version_upgrade = true;
}
}
use_clean:
if (!clean) {
bch_err(c, "no superblock clean section found");
- ret = BCH_FSCK_REPAIR_IMPOSSIBLE;
+ ret = -BCH_ERR_fsck_repair_impossible;
goto err;
}
goto err;
bch_verbose(c, "done checking need_discard and freespace btrees");
+ if (c->sb.version < bcachefs_metadata_version_snapshot_2) {
+ err = "error creating root snapshot node";
+ ret = bch2_fs_initialize_subvolumes(c);
+ if (ret)
+ goto err;
+ }
+
+ bch_verbose(c, "reading snapshots table");
+ err = "error reading snapshots table";
+ ret = bch2_fs_snapshots_start(c);
+ if (ret)
+ goto err;
+ bch_verbose(c, "reading snapshots done");
+
set_bit(BCH_FS_MAY_GO_RW, &c->flags);
bch_info(c, "starting journal replay, %zu keys", c->journal_keys.nr);
bch_verbose(c, "done checking lrus");
set_bit(BCH_FS_CHECK_LRUS_DONE, &c->flags);
+ bch_info(c, "checking backpointers to alloc keys");
+ err = "error checking backpointers to alloc keys";
+ ret = bch2_check_btree_backpointers(c);
+ if (ret)
+ goto err;
+ bch_verbose(c, "done checking backpointers to alloc keys");
+
+ bch_info(c, "checking backpointers to extents");
+ err = "error checking backpointers to extents";
+ ret = bch2_check_backpointers_to_extents(c);
+ if (ret)
+ goto err;
+ bch_verbose(c, "done checking backpointers to extents");
+
+ bch_info(c, "checking extents to backpointers");
+ err = "error checking extents to backpointers";
+ ret = bch2_check_extents_to_backpointers(c);
+ if (ret)
+ goto err;
+ bch_verbose(c, "done checking extents to backpointers");
+ set_bit(BCH_FS_CHECK_BACKPOINTERS_DONE, &c->flags);
+
bch_info(c, "checking alloc to lru refs");
err = "error checking alloc to lru refs";
ret = bch2_check_alloc_to_lru_refs(c);
bch_verbose(c, "done checking alloc to lru refs");
set_bit(BCH_FS_CHECK_ALLOC_TO_LRU_REFS_DONE, &c->flags);
} else {
- set_bit(BCH_FS_MAY_GO_RW, &c->flags);
set_bit(BCH_FS_INITIAL_GC_DONE, &c->flags);
set_bit(BCH_FS_CHECK_LRUS_DONE, &c->flags);
+ set_bit(BCH_FS_CHECK_BACKPOINTERS_DONE, &c->flags);
set_bit(BCH_FS_CHECK_ALLOC_TO_LRU_REFS_DONE, &c->flags);
set_bit(BCH_FS_FSCK_DONE, &c->flags);
if (c->opts.norecovery)
goto out;
+ if (c->sb.version < bcachefs_metadata_version_snapshot_2) {
+ err = "error creating root snapshot node";
+ ret = bch2_fs_initialize_subvolumes(c);
+ if (ret)
+ goto err;
+ }
+
+ bch_verbose(c, "reading snapshots table");
+ err = "error reading snapshots table";
+ ret = bch2_fs_snapshots_start(c);
+ if (ret)
+ goto err;
+ bch_verbose(c, "reading snapshots done");
+
+ set_bit(BCH_FS_MAY_GO_RW, &c->flags);
+
bch_verbose(c, "starting journal replay, %zu keys", c->journal_keys.nr);
err = "journal replay failed";
ret = bch2_journal_replay(c);
if (ret)
goto err;
- if (c->sb.version < bcachefs_metadata_version_snapshot_2) {
- bch2_fs_lazy_rw(c);
-
- err = "error creating root snapshot node";
- ret = bch2_fs_initialize_subvolumes(c);
- if (ret)
- goto err;
- }
-
- bch_verbose(c, "reading snapshots table");
- err = "error reading snapshots table";
- ret = bch2_fs_snapshots_start(c);
- if (ret)
- goto err;
- bch_verbose(c, "reading snapshots done");
-
if (c->sb.version < bcachefs_metadata_version_snapshot_2) {
/* set bi_subvol on root inode */
err = "error upgrade root inode for subvolumes";
le16_to_cpu(c->sb.version_min) < bcachefs_metadata_version_btree_ptr_sectors_written) {
struct bch_move_stats stats;
- bch_move_stats_init(&stats, "recovery");
+ bch2_move_stats_init(&stats, "recovery");
bch_info(c, "scanning for old btree nodes");
ret = bch2_fs_read_write(c);
set_bit(BCH_FS_FSCK_DONE, &c->flags);
bch2_flush_fsck_errs(c);
- if (!c->opts.keep_journal) {
+ if (!c->opts.keep_journal &&
+ test_bit(JOURNAL_REPLAY_DONE, &c->journal.flags)) {
bch2_journal_keys_free(&c->journal_keys);
bch2_journal_entries_free(c);
}
kfree(clean);
+
+ if (!ret && test_bit(BCH_FS_HAVE_DELETED_SNAPSHOTS, &c->flags)) {
+ bch2_fs_read_write_early(c);
+ bch2_delete_dead_snapshots_async(c);
+ }
+
if (ret)
- bch_err(c, "Error in recovery: %s (%i)", err, ret);
+ bch_err(c, "Error in recovery: %s (%s)", err, bch2_err_str(ret));
else
- bch_verbose(c, "ret %i", ret);
+ bch_verbose(c, "ret %s", bch2_err_str(ret));
return ret;
err:
fsck_err:
c->disk_sb.sb->compat[0] |= cpu_to_le64(1ULL << BCH_COMPAT_extents_above_btree_updates_done);
c->disk_sb.sb->compat[0] |= cpu_to_le64(1ULL << BCH_COMPAT_bformat_overflow_done);
+ if (c->sb.version < bcachefs_metadata_version_inode_v3)
+ c->opts.version_upgrade = true;
+
if (c->opts.version_upgrade) {
c->disk_sb.sb->version = cpu_to_le16(bcachefs_metadata_version_current);
c->disk_sb.sb->features[0] |= cpu_to_le64(BCH_SB_FEATURES_ALL);
mutex_unlock(&c->sb_lock);
set_bit(BCH_FS_INITIAL_GC_DONE, &c->flags);
+ set_bit(BCH_FS_CHECK_LRUS_DONE, &c->flags);
+ set_bit(BCH_FS_CHECK_BACKPOINTERS_DONE, &c->flags);
+ set_bit(BCH_FS_CHECK_ALLOC_TO_LRU_REFS_DONE, &c->flags);
set_bit(BCH_FS_MAY_GO_RW, &c->flags);
set_bit(BCH_FS_FSCK_DONE, &c->flags);
goto err;
bch_verbose(c, "reading snapshots done");
- bch2_inode_init(c, &root_inode, 0, 0,
- S_IFDIR|S_IRWXU|S_IRUGO|S_IXUGO, 0, NULL);
+ bch2_inode_init(c, &root_inode, 0, 0, S_IFDIR|0755, 0, NULL);
root_inode.bi_inum = BCACHEFS_ROOT_INO;
root_inode.bi_subvol = BCACHEFS_ROOT_SUBVOL;
- bch2_inode_pack(c, &packed_inode, &root_inode);
+ bch2_inode_pack(&packed_inode, &root_inode);
packed_inode.inode.k.p.snapshot = U32_MAX;
err = "error creating root directory";
projects / bcachefs-tools-debian / blobdiff