if (!*idx)
*idx = __bch2_journal_key_search(keys, btree_id, level, pos);
- while (*idx < keys->nr &&
- (k = idx_to_key(keys, *idx),
- k->btree_id == btree_id &&
- k->level == level &&
- bpos_cmp(k->k->k.p, end_pos) <= 0)) {
- if (bpos_cmp(k->k->k.p, pos) >= 0 &&
+ while ((k = *idx < keys->nr ? idx_to_key(keys, *idx) : NULL)) {
+ if (__journal_key_cmp(btree_id, level, end_pos, k) < 0)
+ return NULL;
+
+ if (__journal_key_cmp(btree_id, level, pos, k) <= 0 &&
!k->overwritten)
return k->k;
if (!new_keys.d) {
bch_err(c, "%s: error allocating new key array (size %zu)",
__func__, new_keys.size);
- return -ENOMEM;
+ return -BCH_ERR_ENOMEM_journal_key_insert;
}
/* Since @keys was full, there was no gap: */
n = kmalloc(bkey_bytes(&k->k), GFP_KERNEL);
if (!n)
- return -ENOMEM;
+ return -BCH_ERR_ENOMEM_journal_key_insert;
bkey_copy(n, k);
ret = bch2_journal_key_insert_take(c, id, level, n);
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;
}
}
}
-struct bkey_s_c bch2_journal_iter_peek(struct journal_iter *iter)
+static struct bkey_s_c bch2_journal_iter_peek(struct journal_iter *iter)
{
struct journal_key *k = iter->keys->d + iter->idx;
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) {
keys->nr = keys->gap = keys->size = 0;
}
+static void __journal_keys_sort(struct journal_keys *keys)
+{
+ struct journal_key *src, *dst;
+
+ sort(keys->d, keys->nr, sizeof(keys->d[0]), journal_sort_key_cmp, NULL);
+
+ src = dst = keys->d;
+ while (src < keys->d + keys->nr) {
+ while (src + 1 < keys->d + keys->nr &&
+ src[0].btree_id == src[1].btree_id &&
+ src[0].level == src[1].level &&
+ bpos_eq(src[0].k->k.p, src[1].k->k.p))
+ src++;
+
+ *dst++ = *src++;
+ }
+
+ keys->nr = dst - keys->d;
+}
+
static int journal_keys_sort(struct bch_fs *c)
{
struct genradix_iter iter;
struct journal_replay *i, **_i;
struct jset_entry *entry;
- struct bkey_i *k, *_n;
+ struct bkey_i *k;
struct journal_keys *keys = &c->journal_keys;
- struct journal_key *src, *dst;
- size_t nr_keys = 0;
+ size_t nr_keys = 0, nr_read = 0;
genradix_for_each(&c->journal_entries, iter, _i) {
i = *_i;
if (!i || i->ignore)
continue;
- for_each_jset_key(k, _n, entry, &i->j)
+ for_each_jset_key(k, entry, &i->j)
nr_keys++;
}
keys->size = roundup_pow_of_two(nr_keys);
keys->d = kvmalloc_array(keys->size, sizeof(keys->d[0]), GFP_KERNEL);
- if (!keys->d)
- return -ENOMEM;
+ if (!keys->d) {
+ bch_err(c, "Failed to allocate buffer for sorted journal keys (%zu keys); trying slowpath",
+ nr_keys);
+
+ do {
+ keys->size >>= 1;
+ keys->d = kvmalloc_array(keys->size, sizeof(keys->d[0]), GFP_KERNEL);
+ } while (!keys->d && keys->size > nr_keys / 8);
+
+ if (!keys->d) {
+ bch_err(c, "Failed to allocate %zu size buffer for sorted journal keys; exiting",
+ keys->size);
+ return -BCH_ERR_ENOMEM_journal_keys_sort;
+ }
+ }
genradix_for_each(&c->journal_entries, iter, _i) {
i = *_i;
if (!i || i->ignore)
continue;
- for_each_jset_key(k, _n, entry, &i->j)
+ cond_resched();
+
+ for_each_jset_key(k, entry, &i->j) {
+ if (keys->nr == keys->size) {
+ __journal_keys_sort(keys);
+
+ if (keys->nr > keys->size * 7 / 8) {
+ bch_err(c, "Too many journal keys for slowpath; have %zu compacted, buf size %zu, processed %zu/%zu",
+ keys->nr, keys->size, nr_read, nr_keys);
+ return -BCH_ERR_ENOMEM_journal_keys_sort;
+ }
+ }
+
keys->d[keys->nr++] = (struct journal_key) {
.btree_id = entry->btree_id,
.level = entry->level,
.journal_seq = le64_to_cpu(i->j.seq),
.journal_offset = k->_data - i->j._data,
};
- }
- sort(keys->d, keys->nr, sizeof(keys->d[0]), journal_sort_key_cmp, NULL);
-
- src = dst = keys->d;
- while (src < keys->d + keys->nr) {
- 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))
- src++;
-
- *dst++ = *src++;
+ nr_read++;
+ }
}
- keys->nr = dst - keys->d;
+ __journal_keys_sort(keys);
keys->gap = keys->nr;
+
+ bch_verbose(c, "Journal keys: %zu read, %zu after sorting and compacting", nr_keys, keys->nr);
return 0;
}
unsigned iter_flags =
BTREE_ITER_INTENT|
BTREE_ITER_NOT_EXTENTS;
+ unsigned update_flags = BTREE_TRIGGER_NORUN;
int ret;
+ /*
+ * BTREE_UPDATE_KEY_CACHE_RECLAIM disables key cache lookup/update to
+ * keep the key cache coherent with the underlying btree. Nothing
+ * besides the allocator is doing updates yet so we don't need key cache
+ * coherency for non-alloc btrees, and key cache fills for snapshots
+ * btrees use BTREE_ITER_FILTER_SNAPSHOTS, which isn't available until
+ * the snapshots recovery pass runs.
+ */
if (!k->level && k->btree_id == BTREE_ID_alloc)
iter_flags |= BTREE_ITER_CACHED;
+ else
+ update_flags |= BTREE_UPDATE_KEY_CACHE_RECLAIM;
bch2_trans_node_iter_init(trans, &iter, k->btree_id, k->k->k.p,
BTREE_MAX_DEPTH, k->level,
if (k->overwritten)
goto out;
- ret = bch2_trans_update(trans, &iter, k->k, BTREE_TRIGGER_NORUN);
+ ret = bch2_trans_update(trans, &iter, k->k, update_flags);
out:
bch2_trans_iter_exit(trans, &iter);
return ret;
struct journal_keys *keys = &c->journal_keys;
struct journal_key **keys_sorted, *k;
struct journal *j = &c->journal;
+ u64 start_seq = c->journal_replay_seq_start;
+ u64 end_seq = c->journal_replay_seq_start;
size_t i;
int ret;
keys_sorted = kvmalloc_array(sizeof(*keys_sorted), keys->nr, GFP_KERNEL);
if (!keys_sorted)
- return -ENOMEM;
+ return -BCH_ERR_ENOMEM_journal_replay;
for (i = 0; i < keys->nr; i++)
keys_sorted[i] = &keys->d[i];
sizeof(keys_sorted[0]),
journal_sort_seq_cmp, NULL);
+ if (keys->nr) {
+ ret = bch2_journal_log_msg(c, "Starting journal replay (%zu keys in entries %llu-%llu)",
+ keys->nr, start_seq, end_seq);
+ if (ret)
+ goto err;
+ }
+
for (i = 0; i < keys->nr; i++) {
k = keys_sorted[i];
BTREE_INSERT_LAZY_RW|
BTREE_INSERT_NOFAIL|
(!k->allocated
- ? BTREE_INSERT_JOURNAL_REPLAY|JOURNAL_WATERMARK_reserved
+ ? BTREE_INSERT_JOURNAL_REPLAY|BCH_WATERMARK_reclaim
: 0),
bch2_journal_replay_key(&trans, k));
if (ret) {
- bch_err(c, "journal replay: error %d while replaying key at btree %s level %u",
- ret, bch2_btree_ids[k->btree_id], k->level);
+ bch_err(c, "journal replay: error while replaying key at btree %s level %u: %s",
+ bch2_btree_ids[k->btree_id], k->level, bch2_err_str(ret));
goto err;
}
}
ret = bch2_journal_error(j);
if (keys->nr && !ret)
- bch2_journal_log_msg(&c->journal, "journal replay finished");
+ bch2_journal_log_msg(c, "journal replay finished");
err:
kvfree(keys_sorted);
+
+ if (ret)
+ bch_err_fn(c, ret);
return ret;
}
case BCH_JSET_ENTRY_btree_root: {
struct btree_root *r;
- if (entry->btree_id >= BTREE_ID_NR) {
- bch_err(c, "filesystem has unknown btree type %u",
- entry->btree_id);
- return -EINVAL;
+ while (entry->btree_id >= c->btree_roots_extra.nr + BTREE_ID_NR) {
+ ret = darray_push(&c->btree_roots_extra, (struct btree_root) { NULL });
+ if (ret)
+ return ret;
}
- r = &c->btree_roots[entry->btree_id];
+ r = bch2_btree_id_root(c, entry->btree_id);
if (entry->u64s) {
r->level = entry->level;
IS_ERR(k1) ||
IS_ERR(k2) ||
k1->k.u64s != k2->k.u64s ||
- memcmp(k1, k2, bkey_bytes(k1)) ||
+ memcmp(k1, k2, bkey_bytes(&k1->k)) ||
l1 != l2, c,
"superblock btree root %u doesn't match journal after clean shutdown\n"
"sb: l=%u %s\n"
GFP_KERNEL);
if (!clean) {
mutex_unlock(&c->sb_lock);
- return ERR_PTR(-ENOMEM);
+ return ERR_PTR(-BCH_ERR_ENOMEM_read_superblock_clean);
}
ret = bch2_sb_clean_validate_late(c, clean, READ);
case BTREE_ID_backpointers:
case BTREE_ID_need_discard:
case BTREE_ID_freespace:
+ case BTREE_ID_bucket_gens:
return true;
default:
return false;
unsigned i;
int ret = 0;
- for (i = 0; i < BTREE_ID_NR; i++) {
- struct btree_root *r = &c->btree_roots[i];
+ for (i = 0; i < btree_id_nr_alive(c); i++) {
+ struct btree_root *r = bch2_btree_id_root(c, i);
if (!r->alive)
continue;
? FSCK_CAN_IGNORE : 0,
"error reading btree root %s",
bch2_btree_ids[i]);
- if (i == BTREE_ID_alloc)
+ if (btree_id_is_alloc(i))
c->sb.compat &= ~(1ULL << BCH_COMPAT_alloc_info);
}
}
- for (i = 0; i < BTREE_ID_NR; i++)
- if (!c->btree_roots[i].b)
+ for (i = 0; i < BTREE_ID_NR; i++) {
+ struct btree_root *r = bch2_btree_id_root(c, i);
+
+ if (!r->b) {
+ r->alive = false;
+ r->level = 0;
bch2_btree_root_alloc(c, i);
+ }
+ }
fsck_err:
return ret;
}
-static int bch2_fs_initialize_subvolumes(struct bch_fs *c)
+static int bch2_initialize_subvolumes(struct bch_fs *c)
{
- struct bkey_i_snapshot root_snapshot;
- struct bkey_i_subvolume root_volume;
+ struct bkey_i_snapshot_tree root_tree;
+ struct bkey_i_snapshot root_snapshot;
+ struct bkey_i_subvolume root_volume;
int ret;
+ bkey_snapshot_tree_init(&root_tree.k_i);
+ root_tree.k.p.offset = 1;
+ root_tree.v.master_subvol = cpu_to_le32(1);
+ root_tree.v.root_snapshot = cpu_to_le32(U32_MAX);
+
bkey_snapshot_init(&root_snapshot.k_i);
root_snapshot.k.p.offset = U32_MAX;
root_snapshot.v.flags = 0;
root_snapshot.v.parent = 0;
- root_snapshot.v.subvol = BCACHEFS_ROOT_SUBVOL;
- root_snapshot.v.pad = 0;
+ root_snapshot.v.subvol = cpu_to_le32(BCACHEFS_ROOT_SUBVOL);
+ root_snapshot.v.tree = cpu_to_le32(1);
SET_BCH_SNAPSHOT_SUBVOL(&root_snapshot.v, true);
- ret = bch2_btree_insert(c, BTREE_ID_snapshots,
- &root_snapshot.k_i,
- NULL, NULL, 0);
- if (ret)
- return ret;
-
bkey_subvolume_init(&root_volume.k_i);
root_volume.k.p.offset = BCACHEFS_ROOT_SUBVOL;
root_volume.v.flags = 0;
root_volume.v.snapshot = cpu_to_le32(U32_MAX);
root_volume.v.inode = cpu_to_le64(BCACHEFS_ROOT_INO);
- ret = bch2_btree_insert(c, BTREE_ID_subvolumes,
- &root_volume.k_i,
- NULL, NULL, 0);
+ ret = bch2_btree_insert(c, BTREE_ID_snapshot_trees,
+ &root_tree.k_i,
+ NULL, NULL, 0) ?:
+ bch2_btree_insert(c, BTREE_ID_snapshots,
+ &root_snapshot.k_i,
+ NULL, NULL, 0) ?:
+ bch2_btree_insert(c, BTREE_ID_subvolumes,
+ &root_volume.k_i,
+ NULL, NULL, 0);
if (ret)
- return ret;
-
- return 0;
+ bch_err_fn(c, ret);
+ return ret;
}
-static int bch2_fs_upgrade_for_subvolumes(struct btree_trans *trans)
+static int __bch2_fs_upgrade_for_subvolumes(struct btree_trans *trans)
{
struct btree_iter iter;
struct bkey_s_c k;
struct bch_inode_unpacked inode;
int ret;
- bch2_trans_iter_init(trans, &iter, BTREE_ID_inodes,
- SPOS(0, BCACHEFS_ROOT_INO, U32_MAX), 0);
- k = bch2_btree_iter_peek_slot(&iter);
+ k = bch2_bkey_get_iter(trans, &iter, BTREE_ID_inodes,
+ SPOS(0, BCACHEFS_ROOT_INO, U32_MAX), 0);
ret = bkey_err(k);
if (ret)
- goto err;
+ return ret;
if (!bkey_is_inode(k.k)) {
bch_err(trans->c, "root inode not found");
- ret = -ENOENT;
+ ret = -BCH_ERR_ENOENT_inode;
goto err;
}
return ret;
}
+/* set bi_subvol on root inode */
+noinline_for_stack
+static int bch2_fs_upgrade_for_subvolumes(struct bch_fs *c)
+{
+ int ret = bch2_trans_do(c, NULL, NULL, BTREE_INSERT_LAZY_RW,
+ __bch2_fs_upgrade_for_subvolumes(&trans));
+ if (ret)
+ bch_err_fn(c, ret);
+ return ret;
+}
+
+static void check_version_upgrade(struct bch_fs *c)
+{
+ unsigned latest_compatible = bch2_version_compatible(c->sb.version);
+ unsigned latest_version = bcachefs_metadata_version_current;
+ unsigned old_version = c->sb.version_upgrade_complete ?: c->sb.version;
+ unsigned new_version = 0;
+ u64 recovery_passes;
+
+ if (old_version < bcachefs_metadata_required_upgrade_below) {
+ if (c->opts.version_upgrade == BCH_VERSION_UPGRADE_incompatible ||
+ latest_compatible < bcachefs_metadata_required_upgrade_below)
+ new_version = latest_version;
+ else
+ new_version = latest_compatible;
+ } else {
+ switch (c->opts.version_upgrade) {
+ case BCH_VERSION_UPGRADE_compatible:
+ new_version = latest_compatible;
+ break;
+ case BCH_VERSION_UPGRADE_incompatible:
+ new_version = latest_version;
+ break;
+ case BCH_VERSION_UPGRADE_none:
+ new_version = old_version;
+ break;
+ }
+ }
+
+ if (new_version > old_version) {
+ struct printbuf buf = PRINTBUF;
+
+ if (old_version < bcachefs_metadata_required_upgrade_below)
+ prt_str(&buf, "Version upgrade required:\n");
+
+ if (old_version != c->sb.version) {
+ prt_str(&buf, "Version upgrade from ");
+ bch2_version_to_text(&buf, c->sb.version_upgrade_complete);
+ prt_str(&buf, " to ");
+ bch2_version_to_text(&buf, c->sb.version);
+ prt_str(&buf, " incomplete\n");
+ }
+
+ prt_printf(&buf, "Doing %s version upgrade from ",
+ BCH_VERSION_MAJOR(old_version) != BCH_VERSION_MAJOR(new_version)
+ ? "incompatible" : "compatible");
+ bch2_version_to_text(&buf, old_version);
+ prt_str(&buf, " to ");
+ bch2_version_to_text(&buf, new_version);
+ prt_newline(&buf);
+
+ recovery_passes = bch2_upgrade_recovery_passes(c, old_version, new_version);
+ if (recovery_passes) {
+ prt_str(&buf, "fsck required");
+
+ c->recovery_passes_explicit |= recovery_passes;
+ c->opts.fix_errors = FSCK_FIX_yes;
+ }
+
+ bch_info(c, "%s", buf.buf);
+
+ mutex_lock(&c->sb_lock);
+ bch2_sb_upgrade(c, new_version);
+ mutex_unlock(&c->sb_lock);
+
+ printbuf_exit(&buf);
+ }
+}
+
+static int bch2_check_allocations(struct bch_fs *c)
+{
+ return bch2_gc(c, true, c->opts.norecovery);
+}
+
+static int bch2_set_may_go_rw(struct bch_fs *c)
+{
+ set_bit(BCH_FS_MAY_GO_RW, &c->flags);
+ return 0;
+}
+
+struct recovery_pass_fn {
+ int (*fn)(struct bch_fs *);
+ const char *name;
+ unsigned when;
+};
+
+static struct recovery_pass_fn recovery_passes[] = {
+#define x(_fn, _when) { .fn = bch2_##_fn, .name = #_fn, .when = _when },
+ BCH_RECOVERY_PASSES()
+#undef x
+};
+
+u64 bch2_fsck_recovery_passes(void)
+{
+ u64 ret = 0;
+
+ for (unsigned i = 0; i < ARRAY_SIZE(recovery_passes); i++)
+ if (recovery_passes[i].when & PASS_FSCK)
+ ret |= BIT_ULL(i);
+ return ret;
+}
+
+static bool should_run_recovery_pass(struct bch_fs *c, enum bch_recovery_pass pass)
+{
+ struct recovery_pass_fn *p = recovery_passes + c->curr_recovery_pass;
+
+ if (c->opts.norecovery && pass > BCH_RECOVERY_PASS_snapshots_read)
+ return false;
+ if (c->recovery_passes_explicit & BIT_ULL(pass))
+ return true;
+ if ((p->when & PASS_FSCK) && c->opts.fsck)
+ return true;
+ if ((p->when & PASS_UNCLEAN) && !c->sb.clean)
+ return true;
+ if (p->when & PASS_ALWAYS)
+ return true;
+ return false;
+}
+
+static int bch2_run_recovery_pass(struct bch_fs *c, enum bch_recovery_pass pass)
+{
+ int ret;
+
+ c->curr_recovery_pass = pass;
+
+ if (should_run_recovery_pass(c, pass)) {
+ struct recovery_pass_fn *p = recovery_passes + pass;
+
+ if (!(p->when & PASS_SILENT))
+ printk(KERN_INFO bch2_log_msg(c, "%s..."), p->name);
+ ret = p->fn(c);
+ if (ret)
+ return ret;
+ if (!(p->when & PASS_SILENT))
+ printk(KERN_CONT " done\n");
+ }
+
+ return 0;
+}
+
+static int bch2_run_recovery_passes(struct bch_fs *c)
+{
+ int ret = 0;
+
+ while (c->curr_recovery_pass < ARRAY_SIZE(recovery_passes)) {
+ ret = bch2_run_recovery_pass(c, c->curr_recovery_pass);
+ if (bch2_err_matches(ret, BCH_ERR_restart_recovery))
+ continue;
+ if (ret)
+ break;
+ c->curr_recovery_pass++;
+ }
+
+ return ret;
+}
+
int bch2_fs_recovery(struct bch_fs *c)
{
- const char *err = "cannot allocate memory";
struct bch_sb_field_clean *clean = NULL;
struct jset *last_journal_entry = NULL;
- u64 blacklist_seq, journal_seq;
+ u64 last_seq, blacklist_seq, journal_seq;
bool write_sb = false;
int ret = 0;
goto err;
}
- if (!(c->sb.features & (1ULL << BCH_FEATURE_alloc_v2))) {
- bch_info(c, "alloc_v2 feature bit not set, fsck required");
- c->opts.fsck = true;
- c->opts.fix_errors = FSCK_OPT_YES;
- }
-
- if (!c->opts.nochanges) {
- 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;
- }
- }
+ if (c->opts.fsck || !(c->opts.nochanges && c->opts.norecovery))
+ check_version_upgrade(c);
if (c->opts.fsck && c->opts.norecovery) {
bch_err(c, "cannot select both norecovery and fsck");
struct journal_replay **i;
bch_verbose(c, "starting journal read");
- ret = bch2_journal_read(c, &blacklist_seq, &journal_seq);
+ ret = bch2_journal_read(c, &last_seq, &blacklist_seq, &journal_seq);
if (ret)
goto err;
+ /*
+ * note: cmd_list_journal needs the blacklist table fully up to date so
+ * it can asterisk ignored journal entries:
+ */
+ if (c->opts.read_journal_only)
+ goto out;
+
genradix_for_each_reverse(&c->journal_entries, iter, i)
if (*i && !(*i)->ignore) {
last_journal_entry = &(*i)->j;
if (!last_journal_entry) {
fsck_err_on(!c->sb.clean, c, "no journal entries found");
- goto use_clean;
+ if (clean)
+ goto use_clean;
+
+ genradix_for_each_reverse(&c->journal_entries, iter, i)
+ if (*i) {
+ last_journal_entry = &(*i)->j;
+ (*i)->ignore = false;
+ break;
+ }
}
ret = journal_keys_sort(c);
blacklist_seq = journal_seq = le64_to_cpu(clean->journal_seq) + 1;
}
+ c->journal_replay_seq_start = last_seq;
+ c->journal_replay_seq_end = blacklist_seq - 1;;
+
if (c->opts.reconstruct_alloc) {
c->sb.compat &= ~(1ULL << BCH_COMPAT_alloc_info);
drop_alloc_keys(&c->journal_keys);
journal_seq += 8;
if (blacklist_seq != journal_seq) {
- ret = bch2_journal_seq_blacklist_add(c,
+ ret = bch2_journal_log_msg(c, "blacklisting entries %llu-%llu",
+ blacklist_seq, journal_seq) ?:
+ bch2_journal_seq_blacklist_add(c,
blacklist_seq, journal_seq);
if (ret) {
bch_err(c, "error creating new journal seq blacklist entry");
}
}
- /*
- * note: cmd_list_journal needs the blacklist table fully up to date so
- * it can asterisk ignored journal entries:
- */
- if (c->opts.read_journal_only)
- goto out;
-
- ret = bch2_fs_journal_start(&c->journal, journal_seq);
+ ret = bch2_journal_log_msg(c, "starting journal at entry %llu, replaying %llu-%llu",
+ journal_seq, last_seq, blacklist_seq - 1) ?:
+ bch2_fs_journal_start(&c->journal, journal_seq);
if (ret)
goto err;
+ if (c->opts.reconstruct_alloc)
+ bch2_journal_log_msg(c, "dropping alloc info");
+
/*
* Skip past versions that might have possibly been used (as nonces),
* but hadn't had their pointers written:
if (ret)
goto err;
- bch_verbose(c, "starting alloc read");
- err = "error reading allocation information";
-
- down_read(&c->gc_lock);
- ret = bch2_alloc_read(c);
- up_read(&c->gc_lock);
-
- if (ret)
- goto err;
- bch_verbose(c, "alloc read done");
+ if (c->opts.fsck &&
+ (IS_ENABLED(CONFIG_BCACHEFS_DEBUG) ||
+ BCH_SB_HAS_TOPOLOGY_ERRORS(c->disk_sb.sb)))
+ c->recovery_passes_explicit |= BIT_ULL(BCH_RECOVERY_PASS_check_topology);
- bch_verbose(c, "starting stripes_read");
- err = "error reading stripes";
- ret = bch2_stripes_read(c);
+ ret = bch2_run_recovery_passes(c);
if (ret)
goto err;
- bch_verbose(c, "stripes_read done");
- bch2_stripes_heap_start(c);
+ /* If we fixed errors, verify that fs is actually clean now: */
+ if (IS_ENABLED(CONFIG_BCACHEFS_DEBUG) &&
+ test_bit(BCH_FS_ERRORS_FIXED, &c->flags) &&
+ !test_bit(BCH_FS_ERRORS_NOT_FIXED, &c->flags) &&
+ !test_bit(BCH_FS_ERROR, &c->flags)) {
+ bch_info(c, "Fixed errors, running fsck a second time to verify fs is clean");
+ clear_bit(BCH_FS_ERRORS_FIXED, &c->flags);
- if (c->opts.fsck) {
- bool metadata_only = c->opts.norecovery;
+ c->curr_recovery_pass = BCH_RECOVERY_PASS_check_alloc_info;
- bch_info(c, "checking allocations");
- err = "error checking allocations";
- ret = bch2_gc(c, true, metadata_only);
+ ret = bch2_run_recovery_passes(c);
if (ret)
goto err;
- bch_verbose(c, "done checking allocations");
-
- set_bit(BCH_FS_INITIAL_GC_DONE, &c->flags);
-
- bch_info(c, "checking need_discard and freespace btrees");
- err = "error checking need_discard and freespace btrees";
- ret = bch2_check_alloc_info(c);
- if (ret)
- goto err;
- bch_verbose(c, "done checking need_discard and freespace btrees");
-
- set_bit(BCH_FS_MAY_GO_RW, &c->flags);
- bch_info(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->opts.verbose || !c->sb.clean)
- bch_info(c, "journal replay done");
-
- bch_info(c, "checking lrus");
- err = "error checking lrus";
- ret = bch2_check_lrus(c);
- if (ret)
- goto err;
- 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);
- if (ret)
- goto err;
- 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;
-
- 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->opts.verbose || !c->sb.clean)
- bch_info(c, "journal replay done");
- }
-
- err = "error initializing freespace";
- ret = bch2_fs_freespace_init(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";
- ret = bch2_trans_do(c, NULL, NULL, BTREE_INSERT_LAZY_RW,
- bch2_fs_upgrade_for_subvolumes(&trans));
- if (ret)
- goto err;
- }
+ if (test_bit(BCH_FS_ERRORS_FIXED, &c->flags) ||
+ test_bit(BCH_FS_ERRORS_NOT_FIXED, &c->flags)) {
+ bch_err(c, "Second fsck run was not clean");
+ set_bit(BCH_FS_ERRORS_NOT_FIXED, &c->flags);
+ }
- if (c->opts.fsck) {
- bch_info(c, "starting fsck");
- err = "error in fsck";
- ret = bch2_fsck_full(c);
- if (ret)
- goto err;
- bch_verbose(c, "fsck done");
- } else if (!c->sb.clean) {
- bch_verbose(c, "checking for deleted inodes");
- err = "error in recovery";
- ret = bch2_fsck_walk_inodes_only(c);
- if (ret)
- goto err;
- bch_verbose(c, "check inodes done");
+ set_bit(BCH_FS_ERRORS_FIXED, &c->flags);
}
if (enabled_qtypes(c)) {
}
mutex_lock(&c->sb_lock);
- 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);
+ if (BCH_SB_VERSION_UPGRADE_COMPLETE(c->disk_sb.sb) != c->sb.version) {
+ SET_BCH_SB_VERSION_UPGRADE_COMPLETE(c->disk_sb.sb, c->sb.version);
write_sb = true;
}
if (!(c->sb.compat & (1ULL << BCH_COMPAT_extents_above_btree_updates_done)) ||
!(c->sb.compat & (1ULL << BCH_COMPAT_bformat_overflow_done)) ||
- le16_to_cpu(c->sb.version_min) < bcachefs_metadata_version_btree_ptr_sectors_written) {
+ 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);
- if (ret)
- goto err;
-
- ret = bch2_scan_old_btree_nodes(c, &stats);
+ ret = bch2_fs_read_write(c) ?:
+ bch2_scan_old_btree_nodes(c, &stats);
if (ret)
goto err;
bch_info(c, "scanning for old btree nodes done");
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);
}
}
if (ret)
- bch_err(c, "Error in recovery: %s (%s)", err, bch2_err_str(ret));
- else
- bch_verbose(c, "ret %s", bch2_err_str(ret));
+ bch_err_fn(c, ret);
return ret;
err:
fsck_err:
struct bch_inode_unpacked root_inode, lostfound_inode;
struct bkey_inode_buf packed_inode;
struct qstr lostfound = QSTR("lost+found");
- const char *err = "cannot allocate memory";
struct bch_dev *ca;
unsigned i;
int ret;
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;
+ bch2_sb_maybe_downgrade(c);
- 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);
+ if (c->opts.version_upgrade != BCH_VERSION_UPGRADE_none) {
+ bch2_sb_upgrade(c, bcachefs_metadata_version_current);
+ SET_BCH_SB_VERSION_UPGRADE_COMPLETE(c->disk_sb.sb, bcachefs_metadata_version_current);
bch2_write_super(c);
}
mutex_unlock(&c->sb_lock);
- set_bit(BCH_FS_INITIAL_GC_DONE, &c->flags);
+ c->curr_recovery_pass = ARRAY_SIZE(recovery_passes);
set_bit(BCH_FS_MAY_GO_RW, &c->flags);
set_bit(BCH_FS_FSCK_DONE, &c->flags);
for_each_online_member(ca, c, i)
bch2_dev_usage_init(ca);
- err = "unable to allocate journal buckets";
for_each_online_member(ca, c, i) {
ret = bch2_dev_journal_alloc(ca);
if (ret) {
bch2_fs_journal_start(&c->journal, 1);
bch2_journal_set_replay_done(&c->journal);
- err = "error going read-write";
ret = bch2_fs_read_write_early(c);
if (ret)
goto err;
* btree updates
*/
bch_verbose(c, "marking superblocks");
- err = "error marking superblock and journal";
for_each_member_device(ca, c, i) {
ret = bch2_trans_mark_dev_sb(c, ca);
if (ret) {
ca->new_fs_bucket_idx = 0;
}
- bch_verbose(c, "initializing freespace");
- err = "error initializing freespace";
ret = bch2_fs_freespace_init(c);
if (ret)
goto err;
- err = "error creating root snapshot node";
- ret = bch2_fs_initialize_subvolumes(c);
+ ret = bch2_initialize_subvolumes(c);
if (ret)
goto err;
bch_verbose(c, "reading snapshots table");
- err = "error reading snapshots table";
- ret = bch2_fs_snapshots_start(c);
+ ret = bch2_snapshots_read(c);
if (ret)
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(&packed_inode, &root_inode);
packed_inode.inode.k.p.snapshot = U32_MAX;
- err = "error creating root directory";
ret = bch2_btree_insert(c, BTREE_ID_inodes,
&packed_inode.inode.k_i,
NULL, NULL, 0);
- if (ret)
+ if (ret) {
+ bch_err_msg(c, ret, "creating root directory");
goto err;
+ }
bch2_inode_init_early(c, &lostfound_inode);
- err = "error creating lost+found";
ret = bch2_trans_do(c, NULL, NULL, 0,
bch2_create_trans(&trans,
BCACHEFS_ROOT_SUBVOL_INUM,
0, 0, S_IFDIR|0700, 0,
NULL, NULL, (subvol_inum) { 0 }, 0));
if (ret) {
- bch_err(c, "error creating lost+found");
+ bch_err_msg(c, ret, "creating lost+found");
goto err;
}
goto err;
}
- err = "error writing first journal entry";
ret = bch2_journal_flush(&c->journal);
- if (ret)
+ if (ret) {
+ bch_err_msg(c, ret, "writing first journal entry");
goto err;
+ }
mutex_lock(&c->sb_lock);
SET_BCH_SB_INITIALIZED(c->disk_sb.sb, true);
return 0;
err:
- pr_err("Error initializing new filesystem: %s (%i)", err, ret);
+ bch_err_fn(ca, ret);
return ret;
}
projects / bcachefs-tools-debian / blobdiff