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);
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_eq(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;
}
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];
journal_sort_seq_cmp, NULL);
if (keys->nr) {
- ret = bch2_fs_log_msg(c, "Starting journal replay (%zu keys in entries %llu-%llu)",
- keys->nr, start_seq, end_seq);
+ 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;
}
ret = bch2_journal_error(j);
if (keys->nr && !ret)
- bch2_fs_log_msg(c, "journal replay finished");
+ bch2_journal_log_msg(c, "journal replay finished");
err:
kvfree(keys_sorted);
+
+ if (ret)
+ bch_err_fn(c, ret);
return ret;
}
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);
? 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 = &c->btree_roots[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)
{
- 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.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;
+}
+
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 last_seq, blacklist_seq, journal_seq;
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_lru_v2) {
- bch_info(c, "version prior to backpointers, upgrade and fsck required");
+ if (c->sb.version < bcachefs_metadata_required_upgrade_below) {
+ bch_info(c, "version %s (%u) prior to %s (%u), upgrade and fsck required",
+ bch2_metadata_versions[c->sb.version],
+ c->sb.version,
+ bch2_metadata_versions[bcachefs_metadata_required_upgrade_below],
+ bcachefs_metadata_required_upgrade_below);
c->opts.version_upgrade = true;
c->opts.fsck = true;
c->opts.fix_errors = FSCK_OPT_YES;
journal_seq += 8;
if (blacklist_seq != journal_seq) {
- ret = bch2_fs_log_msg(c, "blacklisting entries %llu-%llu",
- blacklist_seq, journal_seq) ?:
+ 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) {
}
}
- ret = bch2_fs_log_msg(c, "starting journal at entry %llu, replaying %llu-%llu",
- journal_seq, last_seq, blacklist_seq - 1) ?:
+ 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:
goto err;
bch_verbose(c, "starting alloc read");
- err = "error reading allocation information";
-
down_read(&c->gc_lock);
ret = c->sb.version < bcachefs_metadata_version_bucket_gens
? bch2_alloc_read(c)
: bch2_bucket_gens_read(c);
up_read(&c->gc_lock);
-
if (ret)
goto err;
bch_verbose(c, "alloc read done");
bch_verbose(c, "starting stripes_read");
- err = "error reading stripes";
ret = bch2_stripes_read(c);
if (ret)
goto err;
bch_verbose(c, "stripes_read done");
- bch2_stripes_heap_start(c);
+ if (c->sb.version < bcachefs_metadata_version_snapshot_2) {
+ ret = bch2_fs_initialize_subvolumes(c);
+ if (ret)
+ goto err;
+ }
+
+ bch_verbose(c, "reading snapshots table");
+ ret = bch2_fs_snapshots_start(c);
+ if (ret)
+ goto err;
+ bch_verbose(c, "reading snapshots done");
if (c->opts.fsck) {
bool metadata_only = c->opts.norecovery;
bch_info(c, "checking allocations");
- err = "error checking allocations";
ret = bch2_gc(c, true, metadata_only);
if (ret)
goto err;
set_bit(BCH_FS_INITIAL_GC_DONE, &c->flags);
- 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);
- err = "journal replay failed";
ret = bch2_journal_replay(c, last_seq, blacklist_seq - 1);
if (ret)
goto err;
bch_info(c, "journal replay done");
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;
set_bit(BCH_FS_CHECK_ALLOC_DONE, &c->flags);
bch_info(c, "checking lrus");
- err = "error checking lrus";
ret = bch2_check_lrus(c);
if (ret)
goto err;
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;
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;
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, last_seq, blacklist_seq - 1);
if (ret)
goto err;
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_bucket_gens &&
c->opts.version_upgrade) {
bch_info(c, "initializing bucket_gens");
- err = "error initializing bucket gens";
ret = bch2_bucket_gens_init(c);
if (ret)
goto err;
}
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));
+ ret = bch2_fs_upgrade_for_subvolumes(c);
if (ret)
goto err;
}
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;
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");
}
if (ret)
- bch_err(c, "Error in recovery: %s (%s)", err, bch2_err_str(ret));
+ bch_err_fn(c, ret);
else
bch_verbose(c, "ret %s", bch2_err_str(ret));
return ret;
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;
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);
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;
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