X-Git-Url: http://review.tizen.org/git/?a=blobdiff_plain;f=cmds-check.c;h=75b984f22d92bc8aae0a8126325a520a14d8795c;hb=187b85436b8d2209bb0df1d67a97be711663615e;hp=288b36e5ef3929296f82b8ea4b8402bba6073b1f;hpb=e0ac34931aedc0df48171f47a2b2b77a503fbc41;p=platform%2Fupstream%2Fbtrfs-progs.git diff --git a/cmds-check.c b/cmds-check.c index 288b36e..75b984f 100644 --- a/cmds-check.c +++ b/cmds-check.c @@ -27,18 +27,19 @@ #include #include #include -#include "kerncompat.h" #include "ctree.h" #include "volumes.h" #include "repair.h" #include "disk-io.h" #include "print-tree.h" #include "transaction.h" -#include "list.h" #include "version.h" #include "utils.h" #include "commands.h" #include "free-space-cache.h" +#include "btrfsck.h" +#include "qgroup-verify.h" +#include "rbtree-utils.h" static u64 bytes_used = 0; static u64 total_csum_bytes = 0; @@ -50,6 +51,11 @@ static u64 data_bytes_allocated = 0; static u64 data_bytes_referenced = 0; static int found_old_backref = 0; static LIST_HEAD(duplicate_extents); +static LIST_HEAD(delete_items); +static int repair = 0; +static int no_holes = 0; +static int init_extent_tree = 0; +static int check_data_csum = 0; struct extent_backref { struct list_head list; @@ -57,6 +63,7 @@ struct extent_backref { unsigned int found_extent_tree:1; unsigned int full_backref:1; unsigned int found_ref:1; + unsigned int broken:1; }; struct data_backref { @@ -94,14 +101,15 @@ struct extent_record { u64 refs; u64 extent_item_refs; u64 generation; + u64 parent_generation; u64 info_objectid; + u32 num_duplicates; u8 info_level; + unsigned int found_rec:1; unsigned int content_checked:1; unsigned int owner_ref_checked:1; unsigned int is_root:1; unsigned int metadata:1; - unsigned int found_rec:1; - unsigned int has_duplicate:1; }; struct inode_backref { @@ -118,6 +126,12 @@ struct inode_backref { char name[0]; }; +struct dropping_root_item_record { + struct list_head list; + struct btrfs_root_item ri; + struct btrfs_key found_key; +}; + #define REF_ERR_NO_DIR_ITEM (1 << 0) #define REF_ERR_NO_DIR_INDEX (1 << 1) #define REF_ERR_NO_INODE_REF (1 << 2) @@ -222,6 +236,25 @@ struct walk_control { int root_level; }; +struct bad_item { + struct btrfs_key key; + u64 root_id; + struct list_head list; +}; + +static void reset_cached_block_groups(struct btrfs_fs_info *fs_info); + +static void record_root_in_trans(struct btrfs_trans_handle *trans, + struct btrfs_root *root) +{ + if (root->last_trans != trans->transid) { + root->track_dirty = 1; + root->last_trans = trans->transid; + root->commit_root = root->node; + extent_buffer_get(root->node); + } +} + static u8 imode_to_type(u32 imode) { #define S_SHIFT 12 @@ -239,6 +272,21 @@ static u8 imode_to_type(u32 imode) #undef S_SHIFT } +static int device_record_compare(struct rb_node *node1, struct rb_node *node2) +{ + struct device_record *rec1; + struct device_record *rec2; + + rec1 = rb_entry(node1, struct device_record, node); + rec2 = rb_entry(node2, struct device_record, node); + if (rec1->devid > rec2->devid) + return -1; + else if (rec1->devid < rec2->devid) + return 1; + else + return 0; +} + static struct inode_record *clone_inode_rec(struct inode_record *orig_rec) { struct inode_record *rec; @@ -260,6 +308,84 @@ static struct inode_record *clone_inode_rec(struct inode_record *orig_rec) return rec; } +static void print_inode_error(struct btrfs_root *root, struct inode_record *rec) +{ + u64 root_objectid = root->root_key.objectid; + int errors = rec->errors; + + if (!errors) + return; + /* reloc root errors, we print its corresponding fs root objectid*/ + if (root_objectid == BTRFS_TREE_RELOC_OBJECTID) { + root_objectid = root->root_key.offset; + fprintf(stderr, "reloc"); + } + fprintf(stderr, "root %llu inode %llu errors %x", + (unsigned long long) root_objectid, + (unsigned long long) rec->ino, rec->errors); + + if (errors & I_ERR_NO_INODE_ITEM) + fprintf(stderr, ", no inode item"); + if (errors & I_ERR_NO_ORPHAN_ITEM) + fprintf(stderr, ", no orphan item"); + if (errors & I_ERR_DUP_INODE_ITEM) + fprintf(stderr, ", dup inode item"); + if (errors & I_ERR_DUP_DIR_INDEX) + fprintf(stderr, ", dup dir index"); + if (errors & I_ERR_ODD_DIR_ITEM) + fprintf(stderr, ", odd dir item"); + if (errors & I_ERR_ODD_FILE_EXTENT) + fprintf(stderr, ", odd file extent"); + if (errors & I_ERR_BAD_FILE_EXTENT) + fprintf(stderr, ", bad file extent"); + if (errors & I_ERR_FILE_EXTENT_OVERLAP) + fprintf(stderr, ", file extent overlap"); + if (errors & I_ERR_FILE_EXTENT_DISCOUNT) + fprintf(stderr, ", file extent discount"); + if (errors & I_ERR_DIR_ISIZE_WRONG) + fprintf(stderr, ", dir isize wrong"); + if (errors & I_ERR_FILE_NBYTES_WRONG) + fprintf(stderr, ", nbytes wrong"); + if (errors & I_ERR_ODD_CSUM_ITEM) + fprintf(stderr, ", odd csum item"); + if (errors & I_ERR_SOME_CSUM_MISSING) + fprintf(stderr, ", some csum missing"); + if (errors & I_ERR_LINK_COUNT_WRONG) + fprintf(stderr, ", link count wrong"); + fprintf(stderr, "\n"); +} + +static void print_ref_error(int errors) +{ + if (errors & REF_ERR_NO_DIR_ITEM) + fprintf(stderr, ", no dir item"); + if (errors & REF_ERR_NO_DIR_INDEX) + fprintf(stderr, ", no dir index"); + if (errors & REF_ERR_NO_INODE_REF) + fprintf(stderr, ", no inode ref"); + if (errors & REF_ERR_DUP_DIR_ITEM) + fprintf(stderr, ", dup dir item"); + if (errors & REF_ERR_DUP_DIR_INDEX) + fprintf(stderr, ", dup dir index"); + if (errors & REF_ERR_DUP_INODE_REF) + fprintf(stderr, ", dup inode ref"); + if (errors & REF_ERR_INDEX_UNMATCH) + fprintf(stderr, ", index unmatch"); + if (errors & REF_ERR_FILETYPE_UNMATCH) + fprintf(stderr, ", filetype unmatch"); + if (errors & REF_ERR_NAME_TOO_LONG) + fprintf(stderr, ", name too long"); + if (errors & REF_ERR_NO_ROOT_REF) + fprintf(stderr, ", no root ref"); + if (errors & REF_ERR_NO_ROOT_BACKREF) + fprintf(stderr, ", no root backref"); + if (errors & REF_ERR_DUP_ROOT_REF) + fprintf(stderr, ", dup root ref"); + if (errors & REF_ERR_DUP_ROOT_BACKREF) + fprintf(stderr, ", dup root backref"); + fprintf(stderr, "\n"); +} + static struct inode_record *get_inode_rec(struct cache_tree *inode_cache, u64 ino, int mod) { @@ -268,7 +394,7 @@ static struct inode_record *get_inode_rec(struct cache_tree *inode_cache, struct inode_record *rec = NULL; int ret; - cache = find_cache_extent(inode_cache, ino, 1); + cache = lookup_cache_extent(inode_cache, ino, 1); if (cache) { node = container_of(cache, struct ptr_node, cache); rec = node->data; @@ -293,7 +419,7 @@ static struct inode_record *get_inode_rec(struct cache_tree *inode_cache, if (ino == BTRFS_FREE_INO_OBJECTID) rec->found_link = 1; - ret = insert_existing_cache_extent(inode_cache, &node->cache); + ret = insert_cache_extent(inode_cache, &node->cache); BUG_ON(ret); } return rec; @@ -361,8 +487,9 @@ static void maybe_free_inode_rec(struct cache_tree *inode_cache, rec->errors |= I_ERR_FILE_NBYTES_WRONG; if (rec->extent_start == (u64)-1 || rec->extent_start > 0) rec->first_extent_gap = 0; - if (rec->nlink > 0 && (rec->extent_end < rec->isize || - rec->first_extent_gap < rec->isize)) + if (rec->nlink > 0 && !no_holes && + (rec->extent_end < rec->isize || + rec->first_extent_gap < rec->isize)) rec->errors |= I_ERR_FILE_EXTENT_DISCOUNT; } @@ -375,7 +502,7 @@ static void maybe_free_inode_rec(struct cache_tree *inode_cache, BUG_ON(rec->refs != 1); if (can_free_inode_rec(rec)) { - cache = find_cache_extent(inode_cache, rec->ino, 1); + cache = lookup_cache_extent(inode_cache, rec->ino, 1); node = container_of(cache, struct ptr_node, cache); BUG_ON(node->data != rec); remove_cache_extent(inode_cache, &node->cache); @@ -396,7 +523,7 @@ static int check_orphan_item(struct btrfs_root *root, u64 ino) btrfs_init_path(&path); ret = btrfs_search_slot(NULL, root, &key, &path, 0, 0); - btrfs_release_path(root, &path); + btrfs_release_path(&path); if (ret > 0) ret = -ENOENT; return ret; @@ -436,6 +563,8 @@ static struct inode_backref *get_inode_backref(struct inode_record *rec, struct inode_backref *backref; list_for_each_entry(backref, &rec->backrefs, list) { + if (rec->ino == BTRFS_MULTIPLE_OBJECTIDS) + break; if (backref->dir != dir || backref->namelen != namelen) continue; if (memcmp(name, backref->name, namelen)) @@ -491,9 +620,10 @@ static int add_inode_backref(struct cache_tree *inode_cache, backref->errors |= REF_ERR_DUP_INODE_REF; if (backref->found_dir_index && backref->index != index) backref->errors |= REF_ERR_INDEX_UNMATCH; + else + backref->index = index; backref->ref_type = itemtype; - backref->index = index; backref->found_inode_ref = 1; } else { BUG_ON(1); @@ -598,7 +728,7 @@ static int splice_shared_node(struct shared_node *src_node, src = &src_node->root_cache; dst = &dst_node->root_cache; again: - cache = find_first_cache_extent(src, 0); + cache = search_cache_extent(src, 0); while (cache) { node = container_of(cache, struct ptr_node, cache); rec = node->data; @@ -614,7 +744,7 @@ again: ins->data = rec; rec->refs++; } - ret = insert_existing_cache_extent(dst, &ins->cache); + ret = insert_cache_extent(dst, &ins->cache); if (ret == -EEXIST) { conflict = get_inode_rec(dst, rec->ino, 1); merge_inode_recs(rec, conflict, dst); @@ -648,31 +778,26 @@ again: return 0; } -static void free_inode_recs(struct cache_tree *inode_cache) +static void free_inode_ptr(struct cache_extent *cache) { - struct cache_extent *cache; struct ptr_node *node; struct inode_record *rec; - while (1) { - cache = find_first_cache_extent(inode_cache, 0); - if (!cache) - break; - node = container_of(cache, struct ptr_node, cache); - rec = node->data; - remove_cache_extent(inode_cache, &node->cache); - free(node); - free_inode_rec(rec); - } + node = container_of(cache, struct ptr_node, cache); + rec = node->data; + free_inode_rec(rec); + free(node); } +FREE_EXTENT_CACHE_BASED_TREE(inode_recs, free_inode_ptr); + static struct shared_node *find_shared_node(struct cache_tree *shared, u64 bytenr) { struct cache_extent *cache; struct shared_node *node; - cache = find_cache_extent(shared, bytenr, 1); + cache = lookup_cache_extent(shared, bytenr, 1); if (cache) { node = container_of(cache, struct shared_node, cache); return node; @@ -692,7 +817,7 @@ static int add_shared_node(struct cache_tree *shared, u64 bytenr, u32 refs) cache_tree_init(&node->inode_cache); node->refs = refs; - ret = insert_existing_cache_extent(shared, &node->cache); + ret = insert_cache_extent(shared, &node->cache); BUG_ON(ret); return 0; } @@ -719,8 +844,8 @@ static int enter_shared_node(struct btrfs_root *root, u64 bytenr, u32 refs, if (wc->root_level == wc->active_node && btrfs_root_refs(&root->root_item) == 0) { if (--node->refs == 0) { - free_inode_recs(&node->root_cache); - free_inode_recs(&node->inode_cache); + free_inode_recs_tree(&node->root_cache); + free_inode_recs_tree(&node->inode_cache); remove_cache_extent(&wc->shared, &node->cache); free(node); } @@ -784,8 +909,9 @@ static int is_child_root(struct btrfs_root *root, u64 parent_root_id, key.offset = child_root_id; ret = btrfs_search_slot(NULL, root->fs_info->tree_root, &key, &path, 0, 0); - BUG_ON(ret < 0); - btrfs_release_path(root, &path); + if (ret < 0) + return ret; + btrfs_release_path(&path); if (!ret) return 1; @@ -794,15 +920,14 @@ static int is_child_root(struct btrfs_root *root, u64 parent_root_id, key.offset = 0; ret = btrfs_search_slot(NULL, root->fs_info->tree_root, &key, &path, 0, 0); - BUG_ON(ret <= 0); + if (ret < 0) + goto out; while (1) { leaf = path.nodes[0]; if (path.slots[0] >= btrfs_header_nritems(leaf)) { ret = btrfs_next_leaf(root->fs_info->tree_root, &path); - BUG_ON(ret < 0); - - if (ret > 0) + if (ret) break; leaf = path.nodes[0]; } @@ -815,14 +940,16 @@ static int is_child_root(struct btrfs_root *root, u64 parent_root_id, has_parent = 1; if (key.offset == parent_root_id) { - btrfs_release_path(root, &path); + btrfs_release_path(&path); return 1; } path.slots[0]++; } - - btrfs_release_path(root, &path); +out: + btrfs_release_path(&path); + if (ret < 0) + return ret; return has_parent? 0 : -1; } @@ -880,7 +1007,11 @@ static int process_dir_item(struct btrfs_root *root, namebuf, len, filetype, key->type, error); } else { - fprintf(stderr, "warning line %d\n", __LINE__); + fprintf(stderr, "invalid location in dir item %u\n", + location.type); + add_inode_backref(inode_cache, BTRFS_MULTIPLE_OBJECTIDS, + key->objectid, key->offset, namebuf, + len, filetype, key->type, error); } len = sizeof(*di) + name_len + data_len; @@ -975,14 +1106,15 @@ static int process_inode_extref(struct extent_buffer *eb, } -static u64 count_csum_range(struct btrfs_root *root, u64 start, u64 len) +static int count_csum_range(struct btrfs_root *root, u64 start, + u64 len, u64 *found) { struct btrfs_key key; struct btrfs_path path; struct extent_buffer *leaf; - int ret ; + int ret; size_t size; - u64 found = 0; + *found = 0; u64 csum_end; u16 csum_size = btrfs_super_csum_size(root->fs_info->super_copy); @@ -994,7 +1126,8 @@ static u64 count_csum_range(struct btrfs_root *root, u64 start, u64 len) ret = btrfs_search_slot(NULL, root->fs_info->csum_root, &key, &path, 0, 0); - BUG_ON(ret < 0); + if (ret < 0) + goto out; if (ret > 0 && path.slots[0] > 0) { leaf = path.nodes[0]; btrfs_item_key_to_cpu(leaf, &key, path.slots[0] - 1); @@ -1007,9 +1140,10 @@ static u64 count_csum_range(struct btrfs_root *root, u64 start, u64 len) leaf = path.nodes[0]; if (path.slots[0] >= btrfs_header_nritems(leaf)) { ret = btrfs_next_leaf(root->fs_info->csum_root, &path); - BUG_ON(ret < 0); if (ret > 0) break; + else if (ret < 0) + goto out; leaf = path.nodes[0]; } @@ -1031,13 +1165,16 @@ static u64 count_csum_range(struct btrfs_root *root, u64 start, u64 len) size = min(csum_end - start, len); len -= size; start += size; - found += size; + *found += size; } path.slots[0]++; } - btrfs_release_path(root->fs_info->csum_root, &path); - return found; +out: + if (ret < 0) + return ret; + btrfs_release_path(&path); + return 0; } static int process_file_extent(struct btrfs_root *root, @@ -1052,6 +1189,7 @@ static int process_file_extent(struct btrfs_root *root, u64 extent_offset = 0; u64 mask = root->sectorsize - 1; int extent_type; + int ret; rec = active_node->current; BUG_ON(rec->ino != key->objectid || rec->refs > 1); @@ -1072,7 +1210,7 @@ static int process_file_extent(struct btrfs_root *root, extent_type = btrfs_file_extent_type(eb, fi); if (extent_type == BTRFS_FILE_EXTENT_INLINE) { - num_bytes = btrfs_file_extent_inline_len(eb, fi); + num_bytes = btrfs_file_extent_inline_len(eb, slot, fi); if (num_bytes == 0) rec->errors |= I_ERR_BAD_FILE_EXTENT; rec->found_size += num_bytes; @@ -1106,7 +1244,9 @@ static int process_file_extent(struct btrfs_root *root, else disk_bytenr += extent_offset; - found = count_csum_range(root, disk_bytenr, num_bytes); + ret = count_csum_range(root, disk_bytenr, num_bytes, &found); + if (ret < 0) + return ret; if (extent_type == BTRFS_FILE_EXTENT_REG) { if (found > 0) rec->found_csum_item = 1; @@ -1142,6 +1282,8 @@ static int process_one_leaf(struct btrfs_root *root, struct extent_buffer *eb, if (key.objectid == BTRFS_FREE_SPACE_OBJECTID) continue; + if (key.type == BTRFS_ORPHAN_ITEM_KEY) + continue; if (active_node->current == NULL || active_node->current->ino < key.objectid) { @@ -1186,7 +1328,6 @@ static void reada_walk_down(struct btrfs_root *root, u32 nritems; u32 blocksize; int i; - int ret; int level; level = btrfs_header_level(node); @@ -1198,10 +1339,58 @@ static void reada_walk_down(struct btrfs_root *root, for (i = slot; i < nritems; i++) { bytenr = btrfs_node_blockptr(node, i); ptr_gen = btrfs_node_ptr_generation(node, i); - ret = readahead_tree_block(root, bytenr, blocksize, ptr_gen); - if (ret) - break; + readahead_tree_block(root, bytenr, blocksize, ptr_gen); + } +} + +/* + * Check the child node/leaf by the following condition: + * 1. the first item key of the node/leaf should be the same with the one + * in parent. + * 2. block in parent node should match the child node/leaf. + * 3. generation of parent node and child's header should be consistent. + * + * Or the child node/leaf pointed by the key in parent is not valid. + * + * We hope to check leaf owner too, but since subvol may share leaves, + * which makes leaf owner check not so strong, key check should be + * sufficient enough for that case. + */ +static int check_child_node(struct btrfs_root *root, + struct extent_buffer *parent, int slot, + struct extent_buffer *child) +{ + struct btrfs_key parent_key; + struct btrfs_key child_key; + int ret = 0; + + btrfs_node_key_to_cpu(parent, &parent_key, slot); + if (btrfs_header_level(child) == 0) + btrfs_item_key_to_cpu(child, &child_key, 0); + else + btrfs_node_key_to_cpu(child, &child_key, 0); + + if (memcmp(&parent_key, &child_key, sizeof(parent_key))) { + ret = -EINVAL; + fprintf(stderr, + "Wrong key of child node/leaf, wanted: (%llu, %u, %llu), have: (%llu, %u, %llu)\n", + parent_key.objectid, parent_key.type, parent_key.offset, + child_key.objectid, child_key.type, child_key.offset); + } + if (btrfs_header_bytenr(child) != btrfs_node_blockptr(parent, slot)) { + ret = -EINVAL; + fprintf(stderr, "Wrong block of child node/leaf, wanted: %llu, have: %llu\n", + btrfs_node_blockptr(parent, slot), + btrfs_header_bytenr(child)); + } + if (btrfs_node_ptr_generation(parent, slot) != + btrfs_header_generation(child)) { + ret = -EINVAL; + fprintf(stderr, "Wrong generation of child node/leaf, wanted: %llu, have: %llu\n", + btrfs_header_generation(child), + btrfs_node_ptr_generation(parent, slot)); } + return ret; } static int walk_down_tree(struct btrfs_root *root, struct btrfs_path *path, @@ -1212,7 +1401,7 @@ static int walk_down_tree(struct btrfs_root *root, struct btrfs_path *path, struct extent_buffer *next; struct extent_buffer *cur; u32 blocksize; - int ret; + int ret, err = 0; u64 refs; WARN_ON(*level < 0); @@ -1220,14 +1409,18 @@ static int walk_down_tree(struct btrfs_root *root, struct btrfs_path *path, ret = btrfs_lookup_extent_info(NULL, root, path->nodes[*level]->start, *level, 1, &refs, NULL); - if (ret < 0) + if (ret < 0) { + err = ret; goto out; + } if (refs > 1) { ret = enter_shared_node(root, path->nodes[*level]->start, refs, wc, *level); - if (ret > 0) + if (ret > 0) { + err = ret; goto out; + } } while (*level >= 0) { @@ -1242,6 +1435,8 @@ static int walk_down_tree(struct btrfs_root *root, struct btrfs_path *path, break; if (*level == 0) { ret = process_one_leaf(root, cur, wc); + if (ret < 0) + err = ret; break; } bytenr = btrfs_node_blockptr(cur, path->slots[*level]); @@ -1267,8 +1462,17 @@ static int walk_down_tree(struct btrfs_root *root, struct btrfs_path *path, reada_walk_down(root, cur, path->slots[*level]); next = read_tree_block(root, bytenr, blocksize, ptr_gen); + if (!next) { + err = -EIO; + goto out; + } } + ret = check_child_node(root, cur, path->slots[*level], next); + if (ret) { + err = ret; + goto out; + } *level = *level - 1; free_extent_buffer(path->nodes[*level]); path->nodes[*level] = next; @@ -1276,7 +1480,7 @@ static int walk_down_tree(struct btrfs_root *root, struct btrfs_path *path, } out: path->slots[*level] = btrfs_header_nritems(path->nodes[*level]); - return 0; + return err; } static int walk_up_tree(struct btrfs_root *root, struct btrfs_path *path, @@ -1327,162 +1531,469 @@ out: return ret; } -static int check_inode_recs(struct btrfs_root *root, - struct cache_tree *inode_cache) +static int repair_inode_isize(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct btrfs_path *path, + struct inode_record *rec) { - struct cache_extent *cache; - struct ptr_node *node; - struct inode_record *rec; - struct inode_backref *backref; + struct btrfs_inode_item *ei; + struct btrfs_key key; int ret; - u64 error = 0; - u64 root_dirid = btrfs_root_dirid(&root->root_item); - if (btrfs_root_refs(&root->root_item) == 0) { - if (!cache_tree_empty(inode_cache)) - fprintf(stderr, "warning line %d\n", __LINE__); - return 0; - } + key.objectid = rec->ino; + key.type = BTRFS_INODE_ITEM_KEY; + key.offset = (u64)-1; - rec = get_inode_rec(inode_cache, root_dirid, 0); - if (rec) { - ret = check_root_dir(rec); - if (ret) { - fprintf(stderr, "root %llu root dir %llu error\n", - (unsigned long long)root->root_key.objectid, - (unsigned long long)root_dirid); - error++; + ret = btrfs_search_slot(trans, root, &key, path, 0, 1); + if (ret < 0) + goto out; + if (ret) { + if (!path->slots[0]) { + ret = -ENOENT; + goto out; } - } else { - fprintf(stderr, "root %llu root dir %llu not found\n", - (unsigned long long)root->root_key.objectid, - (unsigned long long)root_dirid); + path->slots[0]--; + ret = 0; } - - while (1) { - cache = find_first_cache_extent(inode_cache, 0); - if (!cache) - break; - node = container_of(cache, struct ptr_node, cache); - rec = node->data; - remove_cache_extent(inode_cache, &node->cache); - free(node); - if (rec->ino == root_dirid || - rec->ino == BTRFS_ORPHAN_OBJECTID) { - free_inode_rec(rec); - continue; - } - - if (rec->errors & I_ERR_NO_ORPHAN_ITEM) { - ret = check_orphan_item(root, rec->ino); - if (ret == 0) - rec->errors &= ~I_ERR_NO_ORPHAN_ITEM; - if (can_free_inode_rec(rec)) { - free_inode_rec(rec); - continue; - } - } - - error++; - if (!rec->found_inode_item) - rec->errors |= I_ERR_NO_INODE_ITEM; - if (rec->found_link != rec->nlink) - rec->errors |= I_ERR_LINK_COUNT_WRONG; - fprintf(stderr, "root %llu inode %llu errors %x\n", - (unsigned long long) root->root_key.objectid, - (unsigned long long) rec->ino, rec->errors); - list_for_each_entry(backref, &rec->backrefs, list) { - if (!backref->found_dir_item) - backref->errors |= REF_ERR_NO_DIR_ITEM; - if (!backref->found_dir_index) - backref->errors |= REF_ERR_NO_DIR_INDEX; - if (!backref->found_inode_ref) - backref->errors |= REF_ERR_NO_INODE_REF; - fprintf(stderr, "\tunresolved ref dir %llu index %llu" - " namelen %u name %s filetype %d error %x\n", - (unsigned long long)backref->dir, - (unsigned long long)backref->index, - backref->namelen, backref->name, - backref->filetype, backref->errors); - } - free_inode_rec(rec); + btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]); + if (key.objectid != rec->ino) { + ret = -ENOENT; + goto out; } - return (error > 0) ? -1 : 0; + + ei = btrfs_item_ptr(path->nodes[0], path->slots[0], + struct btrfs_inode_item); + btrfs_set_inode_size(path->nodes[0], ei, rec->found_size); + btrfs_mark_buffer_dirty(path->nodes[0]); + rec->errors &= ~I_ERR_DIR_ISIZE_WRONG; + printf("reset isize for dir %Lu root %Lu\n", rec->ino, + root->root_key.objectid); +out: + btrfs_release_path(path); + return ret; } -static struct root_record *get_root_rec(struct cache_tree *root_cache, - u64 objectid) +static int repair_inode_orphan_item(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, + struct inode_record *rec) { - struct cache_extent *cache; - struct root_record *rec = NULL; + struct btrfs_key key; int ret; - cache = find_cache_extent(root_cache, objectid, 1); - if (cache) { - rec = container_of(cache, struct root_record, cache); - } else { - rec = calloc(1, sizeof(*rec)); - rec->objectid = objectid; - INIT_LIST_HEAD(&rec->backrefs); - rec->cache.start = objectid; - rec->cache.size = 1; + key.objectid = BTRFS_ORPHAN_OBJECTID; + key.type = BTRFS_ORPHAN_ITEM_KEY; + key.offset = rec->ino; - ret = insert_existing_cache_extent(root_cache, &rec->cache); - BUG_ON(ret); - } - return rec; + ret = btrfs_insert_empty_item(trans, root, path, &key, 0); + btrfs_release_path(path); + if (!ret) + rec->errors &= ~I_ERR_NO_ORPHAN_ITEM; + return ret; } -static struct root_backref *get_root_backref(struct root_record *rec, - u64 ref_root, u64 dir, u64 index, - const char *name, int namelen) +static int add_missing_dir_index(struct btrfs_root *root, + struct cache_tree *inode_cache, + struct inode_record *rec, + struct inode_backref *backref) { - struct root_backref *backref; + struct btrfs_path *path; + struct btrfs_trans_handle *trans; + struct btrfs_dir_item *dir_item; + struct extent_buffer *leaf; + struct btrfs_key key; + struct btrfs_disk_key disk_key; + struct inode_record *dir_rec; + unsigned long name_ptr; + u32 data_size = sizeof(*dir_item) + backref->namelen; + int ret; - list_for_each_entry(backref, &rec->backrefs, list) { - if (backref->ref_root != ref_root || backref->dir != dir || - backref->namelen != namelen) - continue; - if (memcmp(name, backref->name, namelen)) - continue; - return backref; + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + + trans = btrfs_start_transaction(root, 1); + if (IS_ERR(trans)) { + btrfs_free_path(path); + return PTR_ERR(trans); } - backref = malloc(sizeof(*backref) + namelen + 1); - memset(backref, 0, sizeof(*backref)); - backref->ref_root = ref_root; - backref->dir = dir; - backref->index = index; - backref->namelen = namelen; - memcpy(backref->name, name, namelen); - backref->name[namelen] = '\0'; - list_add_tail(&backref->list, &rec->backrefs); - return backref; -} + fprintf(stderr, "repairing missing dir index item for inode %llu\n", + (unsigned long long)rec->ino); + key.objectid = backref->dir; + key.type = BTRFS_DIR_INDEX_KEY; + key.offset = backref->index; -static void free_root_recs(struct cache_tree *root_cache) -{ - struct cache_extent *cache; - struct root_record *rec; - struct root_backref *backref; + ret = btrfs_insert_empty_item(trans, root, path, &key, data_size); + BUG_ON(ret); + + leaf = path->nodes[0]; + dir_item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dir_item); + + disk_key.objectid = cpu_to_le64(rec->ino); + disk_key.type = BTRFS_INODE_ITEM_KEY; + disk_key.offset = 0; + + btrfs_set_dir_item_key(leaf, dir_item, &disk_key); + btrfs_set_dir_type(leaf, dir_item, imode_to_type(rec->imode)); + btrfs_set_dir_data_len(leaf, dir_item, 0); + btrfs_set_dir_name_len(leaf, dir_item, backref->namelen); + name_ptr = (unsigned long)(dir_item + 1); + write_extent_buffer(leaf, backref->name, name_ptr, backref->namelen); + btrfs_mark_buffer_dirty(leaf); + btrfs_free_path(path); + btrfs_commit_transaction(trans, root); + + backref->found_dir_index = 1; + dir_rec = get_inode_rec(inode_cache, backref->dir, 0); + if (!dir_rec) + return 0; + dir_rec->found_size += backref->namelen; + if (dir_rec->found_size == dir_rec->isize && + (dir_rec->errors & I_ERR_DIR_ISIZE_WRONG)) + dir_rec->errors &= ~I_ERR_DIR_ISIZE_WRONG; + if (dir_rec->found_size != dir_rec->isize) + dir_rec->errors |= I_ERR_DIR_ISIZE_WRONG; + + return 0; +} + +static int delete_dir_index(struct btrfs_root *root, + struct cache_tree *inode_cache, + struct inode_record *rec, + struct inode_backref *backref) +{ + struct btrfs_trans_handle *trans; + struct btrfs_dir_item *di; + struct btrfs_path *path; + int ret = 0; + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + + trans = btrfs_start_transaction(root, 1); + if (IS_ERR(trans)) { + btrfs_free_path(path); + return PTR_ERR(trans); + } + + + fprintf(stderr, "Deleting bad dir index [%llu,%u,%llu] root %llu\n", + (unsigned long long)backref->dir, + BTRFS_DIR_INDEX_KEY, (unsigned long long)backref->index, + (unsigned long long)root->objectid); + + di = btrfs_lookup_dir_index(trans, root, path, backref->dir, + backref->name, backref->namelen, + backref->index, -1); + if (IS_ERR(di)) { + ret = PTR_ERR(di); + btrfs_free_path(path); + btrfs_commit_transaction(trans, root); + if (ret == -ENOENT) + return 0; + return ret; + } + + if (!di) + ret = btrfs_del_item(trans, root, path); + else + ret = btrfs_delete_one_dir_name(trans, root, path, di); + BUG_ON(ret); + btrfs_free_path(path); + btrfs_commit_transaction(trans, root); + return ret; +} + +static int repair_inode_backrefs(struct btrfs_root *root, + struct inode_record *rec, + struct cache_tree *inode_cache, + int delete) +{ + struct inode_backref *tmp, *backref; + u64 root_dirid = btrfs_root_dirid(&root->root_item); + int ret = 0; + int repaired = 0; + + list_for_each_entry_safe(backref, tmp, &rec->backrefs, list) { + /* Index 0 for root dir's are special, don't mess with it */ + if (rec->ino == root_dirid && backref->index == 0) + continue; + + if (delete && + ((backref->found_dir_index && !backref->found_inode_ref) || + (backref->found_dir_index && backref->found_inode_ref && + (backref->errors & REF_ERR_INDEX_UNMATCH)))) { + ret = delete_dir_index(root, inode_cache, rec, backref); + if (ret) + break; + repaired++; + list_del(&backref->list); + free(backref); + } + + if (!delete && !backref->found_dir_index && + backref->found_dir_item && backref->found_inode_ref) { + ret = add_missing_dir_index(root, inode_cache, rec, + backref); + if (ret) + break; + repaired++; + if (backref->found_dir_item && + backref->found_dir_index && + backref->found_dir_index) { + if (!backref->errors && + backref->found_inode_ref) { + list_del(&backref->list); + free(backref); + } + } + } + + } + return ret ? ret : repaired; +} + +static int try_repair_inode(struct btrfs_root *root, struct inode_record *rec) +{ + struct btrfs_trans_handle *trans; + struct btrfs_path *path; + int ret = 0; + + if (!(rec->errors & (I_ERR_DIR_ISIZE_WRONG | I_ERR_NO_ORPHAN_ITEM))) + return rec->errors; + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + + trans = btrfs_start_transaction(root, 1); + if (IS_ERR(trans)) { + btrfs_free_path(path); + return PTR_ERR(trans); + } + + if (rec->errors & I_ERR_DIR_ISIZE_WRONG) + ret = repair_inode_isize(trans, root, path, rec); + if (!ret && rec->errors & I_ERR_NO_ORPHAN_ITEM) + ret = repair_inode_orphan_item(trans, root, path, rec); + btrfs_commit_transaction(trans, root); + btrfs_free_path(path); + return ret; +} + +static int check_inode_recs(struct btrfs_root *root, + struct cache_tree *inode_cache) +{ + struct cache_extent *cache; + struct ptr_node *node; + struct inode_record *rec; + struct inode_backref *backref; + int stage = 0; + int ret; + int err = 0; + u64 error = 0; + u64 root_dirid = btrfs_root_dirid(&root->root_item); + + if (btrfs_root_refs(&root->root_item) == 0) { + if (!cache_tree_empty(inode_cache)) + fprintf(stderr, "warning line %d\n", __LINE__); + return 0; + } + + /* + * We need to repair backrefs first because we could change some of the + * errors in the inode recs. + * + * We also need to go through and delete invalid backrefs first and then + * add the correct ones second. We do this because we may get EEXIST + * when adding back the correct index because we hadn't yet deleted the + * invalid index. + * + * For example, if we were missing a dir index then the directories + * isize would be wrong, so if we fixed the isize to what we thought it + * would be and then fixed the backref we'd still have a invalid fs, so + * we need to add back the dir index and then check to see if the isize + * is still wrong. + */ + while (stage < 3) { + stage++; + if (stage == 3 && !err) + break; + + cache = search_cache_extent(inode_cache, 0); + while (repair && cache) { + node = container_of(cache, struct ptr_node, cache); + rec = node->data; + cache = next_cache_extent(cache); + + /* Need to free everything up and rescan */ + if (stage == 3) { + remove_cache_extent(inode_cache, &node->cache); + free(node); + free_inode_rec(rec); + continue; + } + + if (list_empty(&rec->backrefs)) + continue; + + ret = repair_inode_backrefs(root, rec, inode_cache, + stage == 1); + if (ret < 0) { + err = ret; + stage = 2; + break; + } if (ret > 0) { + err = -EAGAIN; + } + } + } + if (err) + return err; + + rec = get_inode_rec(inode_cache, root_dirid, 0); + if (rec) { + ret = check_root_dir(rec); + if (ret) { + fprintf(stderr, "root %llu root dir %llu error\n", + (unsigned long long)root->root_key.objectid, + (unsigned long long)root_dirid); + error++; + } + } else { + fprintf(stderr, "root %llu root dir %llu not found\n", + (unsigned long long)root->root_key.objectid, + (unsigned long long)root_dirid); + } while (1) { - cache = find_first_cache_extent(root_cache, 0); + cache = search_cache_extent(inode_cache, 0); if (!cache) break; - rec = container_of(cache, struct root_record, cache); - remove_cache_extent(root_cache, &rec->cache); + node = container_of(cache, struct ptr_node, cache); + rec = node->data; + remove_cache_extent(inode_cache, &node->cache); + free(node); + if (rec->ino == root_dirid || + rec->ino == BTRFS_ORPHAN_OBJECTID) { + free_inode_rec(rec); + continue; + } - while (!list_empty(&rec->backrefs)) { - backref = list_entry(rec->backrefs.next, - struct root_backref, list); - list_del(&backref->list); - free(backref); + if (rec->errors & I_ERR_NO_ORPHAN_ITEM) { + ret = check_orphan_item(root, rec->ino); + if (ret == 0) + rec->errors &= ~I_ERR_NO_ORPHAN_ITEM; + if (can_free_inode_rec(rec)) { + free_inode_rec(rec); + continue; + } + } + + if (repair) { + ret = try_repair_inode(root, rec); + if (ret == 0 && can_free_inode_rec(rec)) { + free_inode_rec(rec); + continue; + } + ret = 0; + } + + error++; + if (!rec->found_inode_item) + rec->errors |= I_ERR_NO_INODE_ITEM; + if (rec->found_link != rec->nlink) + rec->errors |= I_ERR_LINK_COUNT_WRONG; + print_inode_error(root, rec); + list_for_each_entry(backref, &rec->backrefs, list) { + if (!backref->found_dir_item) + backref->errors |= REF_ERR_NO_DIR_ITEM; + if (!backref->found_dir_index) + backref->errors |= REF_ERR_NO_DIR_INDEX; + if (!backref->found_inode_ref) + backref->errors |= REF_ERR_NO_INODE_REF; + fprintf(stderr, "\tunresolved ref dir %llu index %llu" + " namelen %u name %s filetype %d errors %x", + (unsigned long long)backref->dir, + (unsigned long long)backref->index, + backref->namelen, backref->name, + backref->filetype, backref->errors); + print_ref_error(backref->errors); } - kfree(rec); + free_inode_rec(rec); } + return (error > 0) ? -1 : 0; } +static struct root_record *get_root_rec(struct cache_tree *root_cache, + u64 objectid) +{ + struct cache_extent *cache; + struct root_record *rec = NULL; + int ret; + + cache = lookup_cache_extent(root_cache, objectid, 1); + if (cache) { + rec = container_of(cache, struct root_record, cache); + } else { + rec = calloc(1, sizeof(*rec)); + rec->objectid = objectid; + INIT_LIST_HEAD(&rec->backrefs); + rec->cache.start = objectid; + rec->cache.size = 1; + + ret = insert_cache_extent(root_cache, &rec->cache); + BUG_ON(ret); + } + return rec; +} + +static struct root_backref *get_root_backref(struct root_record *rec, + u64 ref_root, u64 dir, u64 index, + const char *name, int namelen) +{ + struct root_backref *backref; + + list_for_each_entry(backref, &rec->backrefs, list) { + if (backref->ref_root != ref_root || backref->dir != dir || + backref->namelen != namelen) + continue; + if (memcmp(name, backref->name, namelen)) + continue; + return backref; + } + + backref = malloc(sizeof(*backref) + namelen + 1); + memset(backref, 0, sizeof(*backref)); + backref->ref_root = ref_root; + backref->dir = dir; + backref->index = index; + backref->namelen = namelen; + memcpy(backref->name, name, namelen); + backref->name[namelen] = '\0'; + list_add_tail(&backref->list, &rec->backrefs); + return backref; +} + +static void free_root_record(struct cache_extent *cache) +{ + struct root_record *rec; + struct root_backref *backref; + + rec = container_of(cache, struct root_record, cache); + while (!list_empty(&rec->backrefs)) { + backref = list_entry(rec->backrefs.next, + struct root_backref, list); + list_del(&backref->list); + free(backref); + } + + kfree(rec); +} + +FREE_EXTENT_CACHE_BASED_TREE(root_recs, free_root_record); + static int add_root_backref(struct cache_tree *root_cache, u64 root_id, u64 ref_root, u64 dir, u64 index, const char *name, int namelen, @@ -1507,14 +2018,16 @@ static int add_root_backref(struct cache_tree *root_cache, } if (item_type == BTRFS_DIR_ITEM_KEY) { + if (backref->found_forward_ref) + rec->found_ref++; backref->found_dir_item = 1; - backref->reachable = 1; - rec->found_ref++; } else if (item_type == BTRFS_DIR_INDEX_KEY) { backref->found_dir_index = 1; } else if (item_type == BTRFS_ROOT_REF_KEY) { if (backref->found_forward_ref) backref->errors |= REF_ERR_DUP_ROOT_REF; + else if (backref->found_dir_item) + rec->found_ref++; backref->found_forward_ref = 1; } else if (item_type == BTRFS_ROOT_BACKREF_KEY) { if (backref->found_back_ref) @@ -1524,6 +2037,8 @@ static int add_root_backref(struct cache_tree *root_cache, BUG_ON(1); } + if (backref->found_forward_ref && backref->found_dir_item) + backref->reachable = 1; return 0; } @@ -1535,14 +2050,15 @@ static int merge_root_recs(struct btrfs_root *root, struct ptr_node *node; struct inode_record *rec; struct inode_backref *backref; + int ret = 0; if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) { - free_inode_recs(src_cache); + free_inode_recs_tree(src_cache); return 0; } while (1) { - cache = find_first_cache_extent(src_cache, 0); + cache = search_cache_extent(src_cache, 0); if (!cache) break; node = container_of(cache, struct ptr_node, cache); @@ -1550,7 +2066,10 @@ static int merge_root_recs(struct btrfs_root *root, remove_cache_extent(src_cache, &node->cache); free(node); - if (!is_child_root(root, root->objectid, rec->ino)) + ret = is_child_root(root, root->objectid, rec->ino); + if (ret < 0) + break; + else if (ret == 0) goto skip; list_for_each_entry(backref, &rec->backrefs, list) { @@ -1571,6 +2090,8 @@ static int merge_root_recs(struct btrfs_root *root, skip: free_inode_rec(rec); } + if (ret < 0) + return ret; return 0; } @@ -1592,7 +2113,7 @@ static int check_root_refs(struct btrfs_root *root, /* fixme: this can not detect circular references */ while (loop) { loop = 0; - cache = find_first_cache_extent(root_cache, 0); + cache = search_cache_extent(root_cache, 0); while (1) { if (!cache) break; @@ -1619,7 +2140,7 @@ static int check_root_refs(struct btrfs_root *root, } } - cache = find_first_cache_extent(root_cache, 0); + cache = search_cache_extent(root_cache, 0); while (1) { if (!cache) break; @@ -1633,6 +2154,14 @@ static int check_root_refs(struct btrfs_root *root, rec->objectid); if (ret == 0) continue; + + /* + * If we don't have a root item then we likely just have + * a dir item in a snapshot for this root but no actual + * ref key or anything so it's meaningless. + */ + if (!rec->found_root_item) + continue; errors++; fprintf(stderr, "fs tree %llu not referenced\n", (unsigned long long)rec->objectid); @@ -1667,12 +2196,13 @@ static int check_root_refs(struct btrfs_root *root, if (!backref->errors && rec->found_root_item) continue; fprintf(stderr, "\tunresolved ref root %llu dir %llu" - " index %llu namelen %u name %s error %x\n", + " index %llu namelen %u name %s errors %x\n", (unsigned long long)backref->ref_root, (unsigned long long)backref->dir, (unsigned long long)backref->index, backref->namelen, backref->name, backref->errors); + print_ref_error(backref->errors); } } return errors > 0 ? 1 : 0; @@ -1720,6 +2250,7 @@ static int check_fs_root(struct btrfs_root *root, struct walk_control *wc) { int ret = 0; + int err = 0; int wret; int level; struct btrfs_path path; @@ -1757,7 +2288,8 @@ static int check_fs_root(struct btrfs_root *root, level = root_item->drop_level; path.lowest_level = level; wret = btrfs_search_slot(NULL, root, &key, &path, 0, 0); - BUG_ON(wret < 0); + if (wret < 0) + goto skip_walking; btrfs_node_key(path.nodes[level], &found_key, path.slots[level]); WARN_ON(memcmp(&found_key, &root_item->drop_progress, @@ -1777,9 +2309,12 @@ static int check_fs_root(struct btrfs_root *root, if (wret != 0) break; } - btrfs_release_path(root, &path); +skip_walking: + btrfs_release_path(&path); - merge_root_recs(root, &root_node.root_cache, root_cache); + err = merge_root_recs(root, &root_node.root_cache, root_cache); + if (err < 0) + ret = err; if (root_node.current) { root_node.current->checked = 1; @@ -1787,19 +2322,18 @@ static int check_fs_root(struct btrfs_root *root, root_node.current); } - ret = check_inode_recs(root, &root_node.inode_cache); + err = check_inode_recs(root, &root_node.inode_cache); + if (!ret) + ret = err; return ret; } static int fs_root_objectid(u64 objectid) { - if (objectid == BTRFS_FS_TREE_OBJECTID || - objectid == BTRFS_TREE_RELOC_OBJECTID || - objectid == BTRFS_DATA_RELOC_TREE_OBJECTID || - (objectid >= BTRFS_FIRST_FREE_OBJECTID && - objectid <= BTRFS_LAST_FREE_OBJECTID)) + if (objectid == BTRFS_TREE_RELOC_OBJECTID || + objectid == BTRFS_DATA_RELOC_TREE_OBJECTID) return 1; - return 0; + return is_fstree(objectid); } static int check_fs_roots(struct btrfs_root *root, @@ -1808,42 +2342,73 @@ static int check_fs_roots(struct btrfs_root *root, struct btrfs_path path; struct btrfs_key key; struct walk_control wc; - struct extent_buffer *leaf; + struct extent_buffer *leaf, *tree_node; struct btrfs_root *tmp_root; struct btrfs_root *tree_root = root->fs_info->tree_root; int ret; int err = 0; + /* + * Just in case we made any changes to the extent tree that weren't + * reflected into the free space cache yet. + */ + if (repair) + reset_cached_block_groups(root->fs_info); memset(&wc, 0, sizeof(wc)); cache_tree_init(&wc.shared); btrfs_init_path(&path); +again: key.offset = 0; key.objectid = 0; key.type = BTRFS_ROOT_ITEM_KEY; ret = btrfs_search_slot(NULL, tree_root, &key, &path, 0, 0); - BUG_ON(ret < 0); + if (ret < 0) { + err = 1; + goto out; + } + tree_node = tree_root->node; while (1) { + if (tree_node != tree_root->node) { + free_root_recs_tree(root_cache); + btrfs_release_path(&path); + goto again; + } leaf = path.nodes[0]; if (path.slots[0] >= btrfs_header_nritems(leaf)) { ret = btrfs_next_leaf(tree_root, &path); - if (ret != 0) + if (ret) { + if (ret < 0) + err = 1; break; + } leaf = path.nodes[0]; } btrfs_item_key_to_cpu(leaf, &key, path.slots[0]); if (key.type == BTRFS_ROOT_ITEM_KEY && fs_root_objectid(key.objectid)) { - tmp_root = btrfs_read_fs_root_no_cache(root->fs_info, - &key); + if (key.objectid == BTRFS_TREE_RELOC_OBJECTID) { + tmp_root = btrfs_read_fs_root_no_cache( + root->fs_info, &key); + } else { + key.offset = (u64)-1; + tmp_root = btrfs_read_fs_root( + root->fs_info, &key); + } if (IS_ERR(tmp_root)) { err = 1; goto next; } ret = check_fs_root(tmp_root, root_cache, &wc); + if (ret == -EAGAIN) { + free_root_recs_tree(root_cache); + btrfs_release_path(&path); + goto again; + } if (ret) err = 1; - btrfs_free_fs_root(root->fs_info, tmp_root); + if (key.objectid == BTRFS_TREE_RELOC_OBJECTID) + btrfs_free_fs_root(tmp_root); } else if (key.type == BTRFS_ROOT_REF_KEY || key.type == BTRFS_ROOT_BACKREF_KEY) { process_root_ref(leaf, path.slots[0], &key, @@ -1852,8 +2417,10 @@ static int check_fs_roots(struct btrfs_root *root, next: path.slots[0]++; } - btrfs_release_path(tree_root, &path); - +out: + btrfs_release_path(&path); + if (err) + free_extent_cache_tree(&wc.shared); if (!cache_tree_empty(&wc.shared)) fprintf(stderr, "warning line %d\n", __LINE__); @@ -1923,7 +2490,7 @@ static int all_backpointers_checked(struct extent_record *rec, int print_errs) (unsigned long long)rec->start, back->full_backref ? "parent" : "root", - back->full_backref ? + back->full_backref ? (unsigned long long)dback->parent: (unsigned long long)dback->root, (unsigned long long)dback->owner, @@ -1987,14 +2554,14 @@ static int free_all_extent_backrefs(struct extent_record *rec) return 0; } -static void free_extent_cache(struct btrfs_fs_info *fs_info, - struct cache_tree *extent_cache) +static void free_extent_record_cache(struct btrfs_fs_info *fs_info, + struct cache_tree *extent_cache) { struct cache_extent *cache; struct extent_record *rec; while (1) { - cache = find_first_cache_extent(extent_cache, 0); + cache = first_cache_extent(extent_cache); if (!cache) break; rec = container_of(cache, struct extent_record, cache); @@ -2010,9 +2577,10 @@ static int maybe_free_extent_rec(struct cache_tree *extent_cache, { if (rec->content_checked && rec->owner_ref_checked && rec->extent_item_refs == rec->refs && rec->refs > 0 && - !all_backpointers_checked(rec, 0)) { + rec->num_duplicates == 0 && !all_backpointers_checked(rec, 0)) { remove_cache_extent(extent_cache, &rec->cache); free_all_extent_backrefs(rec); + list_del_init(&rec->list); free(rec); } return 0; @@ -2071,7 +2639,7 @@ static int check_owner_ref(struct btrfs_root *root, path.slots[level + 1])) found = 1; - btrfs_release_path(ref_root, &path); + btrfs_release_path(&path); return found ? 0 : 1; } @@ -2105,7 +2673,7 @@ static int record_bad_block_io(struct btrfs_fs_info *info, struct cache_extent *cache; struct btrfs_key key; - cache = find_cache_extent(extent_cache, start, len); + cache = lookup_cache_extent(extent_cache, start, len); if (!cache) return 0; @@ -2117,17 +2685,252 @@ static int record_bad_block_io(struct btrfs_fs_info *info, return btrfs_add_corrupt_extent_record(info, &key, start, len, 0); } -static int check_block(struct btrfs_root *root, +static int swap_values(struct btrfs_root *root, struct btrfs_path *path, + struct extent_buffer *buf, int slot) +{ + if (btrfs_header_level(buf)) { + struct btrfs_key_ptr ptr1, ptr2; + + read_extent_buffer(buf, &ptr1, btrfs_node_key_ptr_offset(slot), + sizeof(struct btrfs_key_ptr)); + read_extent_buffer(buf, &ptr2, + btrfs_node_key_ptr_offset(slot + 1), + sizeof(struct btrfs_key_ptr)); + write_extent_buffer(buf, &ptr1, + btrfs_node_key_ptr_offset(slot + 1), + sizeof(struct btrfs_key_ptr)); + write_extent_buffer(buf, &ptr2, + btrfs_node_key_ptr_offset(slot), + sizeof(struct btrfs_key_ptr)); + if (slot == 0) { + struct btrfs_disk_key key; + btrfs_node_key(buf, &key, 0); + btrfs_fixup_low_keys(root, path, &key, + btrfs_header_level(buf) + 1); + } + } else { + struct btrfs_item *item1, *item2; + struct btrfs_key k1, k2; + char *item1_data, *item2_data; + u32 item1_offset, item2_offset, item1_size, item2_size; + + item1 = btrfs_item_nr(slot); + item2 = btrfs_item_nr(slot + 1); + btrfs_item_key_to_cpu(buf, &k1, slot); + btrfs_item_key_to_cpu(buf, &k2, slot + 1); + item1_offset = btrfs_item_offset(buf, item1); + item2_offset = btrfs_item_offset(buf, item2); + item1_size = btrfs_item_size(buf, item1); + item2_size = btrfs_item_size(buf, item2); + + item1_data = malloc(item1_size); + if (!item1_data) + return -ENOMEM; + item2_data = malloc(item2_size); + if (!item2_data) { + free(item1_data); + return -ENOMEM; + } + + read_extent_buffer(buf, item1_data, item1_offset, item1_size); + read_extent_buffer(buf, item2_data, item2_offset, item2_size); + + write_extent_buffer(buf, item1_data, item2_offset, item2_size); + write_extent_buffer(buf, item2_data, item1_offset, item1_size); + free(item1_data); + free(item2_data); + + btrfs_set_item_offset(buf, item1, item2_offset); + btrfs_set_item_offset(buf, item2, item1_offset); + btrfs_set_item_size(buf, item1, item2_size); + btrfs_set_item_size(buf, item2, item1_size); + + path->slots[0] = slot; + btrfs_set_item_key_unsafe(root, path, &k2); + path->slots[0] = slot + 1; + btrfs_set_item_key_unsafe(root, path, &k1); + } + return 0; +} + +static int fix_key_order(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct extent_buffer *buf) +{ + struct btrfs_path *path; + struct btrfs_key k1, k2; + int i; + int level; + int ret; + + k1.objectid = btrfs_header_owner(buf); + k1.type = BTRFS_ROOT_ITEM_KEY; + k1.offset = (u64)-1; + + root = btrfs_read_fs_root(root->fs_info, &k1); + if (IS_ERR(root)) + return -EIO; + + record_root_in_trans(trans, root); + + path = btrfs_alloc_path(); + if (!path) + return -EIO; + + level = btrfs_header_level(buf); + path->lowest_level = level; + path->skip_check_block = 1; + if (level) + btrfs_node_key_to_cpu(buf, &k1, 0); + else + btrfs_item_key_to_cpu(buf, &k1, 0); + + ret = btrfs_search_slot(trans, root, &k1, path, 0, 1); + if (ret) { + btrfs_free_path(path); + return -EIO; + } + + buf = path->nodes[level]; + for (i = 0; i < btrfs_header_nritems(buf) - 1; i++) { + if (level) { + btrfs_node_key_to_cpu(buf, &k1, i); + btrfs_node_key_to_cpu(buf, &k2, i + 1); + } else { + btrfs_item_key_to_cpu(buf, &k1, i); + btrfs_item_key_to_cpu(buf, &k2, i + 1); + } + if (btrfs_comp_cpu_keys(&k1, &k2) < 0) + continue; + ret = swap_values(root, path, buf, i); + if (ret) + break; + btrfs_mark_buffer_dirty(buf); + i = 0; + } + + btrfs_free_path(path); + return ret; +} + +static int fix_item_offset(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct extent_buffer *buf) +{ + struct btrfs_path *path; + struct btrfs_key k1; + int i; + int level; + int ret; + + k1.objectid = btrfs_header_owner(buf); + k1.type = BTRFS_ROOT_ITEM_KEY; + k1.offset = (u64)-1; + + root = btrfs_read_fs_root(root->fs_info, &k1); + if (IS_ERR(root)) + return -EIO; + + record_root_in_trans(trans, root); + + path = btrfs_alloc_path(); + if (!path) + return -EIO; + + level = btrfs_header_level(buf); + path->lowest_level = level; + path->skip_check_block = 1; + if (level) + btrfs_node_key_to_cpu(buf, &k1, 0); + else + btrfs_item_key_to_cpu(buf, &k1, 0); + + ret = btrfs_search_slot(trans, root, &k1, path, 0, 1); + if (ret) { + btrfs_free_path(path); + return -EIO; + } + + buf = path->nodes[level]; + for (i = 0; i < btrfs_header_nritems(buf); i++) { + unsigned int shift = 0, offset; + + if (i == 0 && btrfs_item_end_nr(buf, i) != + BTRFS_LEAF_DATA_SIZE(root)) { + if (btrfs_item_end_nr(buf, i) > + BTRFS_LEAF_DATA_SIZE(root)) { + fprintf(stderr, "item is off the end of the " + "leaf, can't fix\n"); + ret = -EIO; + break; + } + shift = BTRFS_LEAF_DATA_SIZE(root) - + btrfs_item_end_nr(buf, i); + } else if (i > 0 && btrfs_item_end_nr(buf, i) != + btrfs_item_offset_nr(buf, i - 1)) { + if (btrfs_item_end_nr(buf, i) > + btrfs_item_offset_nr(buf, i - 1)) { + fprintf(stderr, "items overlap, can't fix\n"); + ret = -EIO; + break; + } + shift = btrfs_item_offset_nr(buf, i - 1) - + btrfs_item_end_nr(buf, i); + } + if (!shift) + continue; + + printf("Shifting item nr %d by %u bytes in block %llu\n", + i, shift, (unsigned long long)buf->start); + offset = btrfs_item_offset_nr(buf, i); + memmove_extent_buffer(buf, + btrfs_leaf_data(buf) + offset + shift, + btrfs_leaf_data(buf) + offset, + btrfs_item_size_nr(buf, i)); + btrfs_set_item_offset(buf, btrfs_item_nr(i), + offset + shift); + btrfs_mark_buffer_dirty(buf); + } + + /* + * We may have moved things, in which case we want to exit so we don't + * write those changes out. Once we have proper abort functionality in + * progs this can be changed to something nicer. + */ + BUG_ON(ret); + btrfs_free_path(path); + return ret; +} + +/* + * Attempt to fix basic block failures. If we can't fix it for whatever reason + * then just return -EIO. + */ +static int try_to_fix_bad_block(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct extent_buffer *buf, + enum btrfs_tree_block_status status) +{ + if (status == BTRFS_TREE_BLOCK_BAD_KEY_ORDER) + return fix_key_order(trans, root, buf); + if (status == BTRFS_TREE_BLOCK_INVALID_OFFSETS) + return fix_item_offset(trans, root, buf); + return -EIO; +} + +static int check_block(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct cache_tree *extent_cache, struct extent_buffer *buf, u64 flags) { struct extent_record *rec; struct cache_extent *cache; struct btrfs_key key; - int ret = 1; + enum btrfs_tree_block_status status; + int ret = 0; int level; - cache = find_cache_extent(extent_cache, buf->start, buf->len); + cache = lookup_cache_extent(extent_cache, buf->start, buf->len); if (!cache) return 1; rec = container_of(cache, struct extent_record, cache); @@ -2146,13 +2949,25 @@ static int check_block(struct btrfs_root *root, rec->info_level = level; if (btrfs_is_leaf(buf)) - ret = btrfs_check_leaf(root, &rec->parent_key, buf); + status = btrfs_check_leaf(root, &rec->parent_key, buf); else - ret = btrfs_check_node(root, &rec->parent_key, buf); - - if (ret) { - fprintf(stderr, "bad block %llu\n", - (unsigned long long)buf->start); + status = btrfs_check_node(root, &rec->parent_key, buf); + + if (status != BTRFS_TREE_BLOCK_CLEAN) { + if (repair) + status = try_to_fix_bad_block(trans, root, buf, + status); + if (status != BTRFS_TREE_BLOCK_CLEAN) { + ret = -EIO; + fprintf(stderr, "bad block %llu\n", + (unsigned long long)buf->start); + } else { + /* + * Signal to callers we need to start the scan over + * again since we'll have cow'ed blocks. + */ + ret = -EAGAIN; + } } else { rec->content_checked = 1; if (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF) @@ -2280,7 +3095,7 @@ static struct data_backref *alloc_data_backref(struct extent_record *rec, } static int add_extent_rec(struct cache_tree *extent_cache, - struct btrfs_key *parent_key, + struct btrfs_key *parent_key, u64 parent_gen, u64 start, u64 nr, u64 extent_item_refs, int is_root, int inc_ref, int set_checked, int metadata, int extent_rec, u64 max_size) @@ -2290,7 +3105,7 @@ static int add_extent_rec(struct cache_tree *extent_cache, int ret = 0; int dup = 0; - cache = find_cache_extent(extent_cache, start, nr); + cache = lookup_cache_extent(extent_cache, start, nr); if (cache) { rec = container_of(cache, struct extent_record, cache); if (inc_ref) @@ -2304,7 +3119,7 @@ static int add_extent_rec(struct cache_tree *extent_cache, * the backrefs. */ if (extent_rec) { - if (rec->found_rec) { + if (start != rec->start || rec->found_rec) { struct extent_record *tmp; dup = 1; @@ -2326,13 +3141,14 @@ static int add_extent_rec(struct cache_tree *extent_cache, tmp->nr = nr; tmp->found_rec = 1; tmp->metadata = metadata; + tmp->extent_item_refs = extent_item_refs; INIT_LIST_HEAD(&tmp->list); list_add_tail(&tmp->list, &rec->dups); - rec->has_duplicate = 1; + rec->num_duplicates++; } else { rec->nr = nr; + rec->found_rec = 1; } - rec->found_rec = 1; } if (extent_item_refs && !dup) { @@ -2355,6 +3171,8 @@ static int add_extent_rec(struct cache_tree *extent_cache, if (parent_key) btrfs_cpu_key_to_disk(&rec->parent_key, parent_key); + if (parent_gen) + rec->parent_generation = parent_gen; if (rec->max_size < max_size) rec->max_size = max_size; @@ -2366,10 +3184,10 @@ static int add_extent_rec(struct cache_tree *extent_cache, rec->start = start; rec->max_size = max_size; rec->nr = max(nr, max_size); - rec->found_rec = extent_rec; + rec->found_rec = !!extent_rec; rec->content_checked = 0; rec->owner_ref_checked = 0; - rec->has_duplicate = 0; + rec->num_duplicates = 0; rec->metadata = metadata; INIT_LIST_HEAD(&rec->backrefs); INIT_LIST_HEAD(&rec->dups); @@ -2395,9 +3213,14 @@ static int add_extent_rec(struct cache_tree *extent_cache, else memset(&rec->parent_key, 0, sizeof(*parent_key)); + if (parent_gen) + rec->parent_generation = parent_gen; + else + rec->parent_generation = 0; + rec->cache.start = start; rec->cache.size = nr; - ret = insert_existing_cache_extent(extent_cache, &rec->cache); + ret = insert_cache_extent(extent_cache, &rec->cache); BUG_ON(ret); bytes_used += nr; if (set_checked) { @@ -2414,11 +3237,11 @@ static int add_tree_backref(struct cache_tree *extent_cache, u64 bytenr, struct tree_backref *back; struct cache_extent *cache; - cache = find_cache_extent(extent_cache, bytenr, 1); + cache = lookup_cache_extent(extent_cache, bytenr, 1); if (!cache) { - add_extent_rec(extent_cache, NULL, bytenr, + add_extent_rec(extent_cache, NULL, 0, bytenr, 1, 0, 0, 0, 0, 1, 0, 0); - cache = find_cache_extent(extent_cache, bytenr, 1); + cache = lookup_cache_extent(extent_cache, bytenr, 1); if (!cache) abort(); } @@ -2451,6 +3274,7 @@ static int add_tree_backref(struct cache_tree *extent_cache, u64 bytenr, } back->node.found_extent_tree = 1; } + maybe_free_extent_rec(extent_cache, rec); return 0; } @@ -2462,11 +3286,11 @@ static int add_data_backref(struct cache_tree *extent_cache, u64 bytenr, struct data_backref *back; struct cache_extent *cache; - cache = find_cache_extent(extent_cache, bytenr, 1); + cache = lookup_cache_extent(extent_cache, bytenr, 1); if (!cache) { - add_extent_rec(extent_cache, NULL, bytenr, 1, 0, 0, 0, 0, + add_extent_rec(extent_cache, NULL, 0, bytenr, 1, 0, 0, 0, 0, 0, 0, max_size); - cache = find_cache_extent(extent_cache, bytenr, 1); + cache = lookup_cache_extent(extent_cache, bytenr, 1); if (!cache) abort(); } @@ -2505,7 +3329,7 @@ static int add_data_backref(struct cache_tree *extent_cache, u64 bytenr, } else { if (back->node.found_extent_tree) { fprintf(stderr, "Extent back ref already exists " - "for %llu parent %llu root %llu" + "for %llu parent %llu root %llu " "owner %llu offset %llu num_refs %lu\n", (unsigned long long)bytenr, (unsigned long long)parent, @@ -2517,6 +3341,7 @@ static int add_data_backref(struct cache_tree *extent_cache, u64 bytenr, back->num_refs = num_refs; back->node.found_extent_tree = 1; } + maybe_free_extent_rec(extent_cache, rec); return 0; } @@ -2524,10 +3349,10 @@ static int add_pending(struct cache_tree *pending, struct cache_tree *seen, u64 bytenr, u32 size) { int ret; - ret = insert_cache_extent(seen, bytenr, size); + ret = add_cache_extent(seen, bytenr, size); if (ret) return ret; - insert_cache_extent(pending, bytenr, size); + add_cache_extent(pending, bytenr, size); return 0; } @@ -2541,10 +3366,10 @@ static int pick_next_pending(struct cache_tree *pending, struct cache_extent *cache; int ret; - cache = find_first_cache_extent(reada, 0); + cache = search_cache_extent(reada, 0); if (cache) { bits[0].start = cache->start; - bits[1].size = cache->size; + bits[0].size = cache->size; *reada_bits = 1; return 1; } @@ -2552,12 +3377,12 @@ static int pick_next_pending(struct cache_tree *pending, if (node_start > 32768) node_start -= 32768; - cache = find_first_cache_extent(nodes, node_start); + cache = search_cache_extent(nodes, node_start); if (!cache) - cache = find_first_cache_extent(nodes, 0); + cache = search_cache_extent(nodes, 0); if (!cache) { - cache = find_first_cache_extent(pending, 0); + cache = search_cache_extent(pending, 0); if (!cache) return 0; ret = 0; @@ -2581,7 +3406,7 @@ static int pick_next_pending(struct cache_tree *pending, if (bits_nr - ret > 8) { u64 lookup = bits[0].start + bits[0].size; struct cache_extent *next; - next = find_first_cache_extent(pending, lookup); + next = search_cache_extent(pending, lookup); while(next) { if (next->start - lookup > 32768) break; @@ -2599,6 +3424,95 @@ static int pick_next_pending(struct cache_tree *pending, return ret; } +static void free_chunk_record(struct cache_extent *cache) +{ + struct chunk_record *rec; + + rec = container_of(cache, struct chunk_record, cache); + list_del_init(&rec->list); + list_del_init(&rec->dextents); + free(rec); +} + +void free_chunk_cache_tree(struct cache_tree *chunk_cache) +{ + cache_tree_free_extents(chunk_cache, free_chunk_record); +} + +static void free_device_record(struct rb_node *node) +{ + struct device_record *rec; + + rec = container_of(node, struct device_record, node); + free(rec); +} + +FREE_RB_BASED_TREE(device_cache, free_device_record); + +int insert_block_group_record(struct block_group_tree *tree, + struct block_group_record *bg_rec) +{ + int ret; + + ret = insert_cache_extent(&tree->tree, &bg_rec->cache); + if (ret) + return ret; + + list_add_tail(&bg_rec->list, &tree->block_groups); + return 0; +} + +static void free_block_group_record(struct cache_extent *cache) +{ + struct block_group_record *rec; + + rec = container_of(cache, struct block_group_record, cache); + list_del_init(&rec->list); + free(rec); +} + +void free_block_group_tree(struct block_group_tree *tree) +{ + cache_tree_free_extents(&tree->tree, free_block_group_record); +} + +int insert_device_extent_record(struct device_extent_tree *tree, + struct device_extent_record *de_rec) +{ + int ret; + + /* + * Device extent is a bit different from the other extents, because + * the extents which belong to the different devices may have the + * same start and size, so we need use the special extent cache + * search/insert functions. + */ + ret = insert_cache_extent2(&tree->tree, &de_rec->cache); + if (ret) + return ret; + + list_add_tail(&de_rec->chunk_list, &tree->no_chunk_orphans); + list_add_tail(&de_rec->device_list, &tree->no_device_orphans); + return 0; +} + +static void free_device_extent_record(struct cache_extent *cache) +{ + struct device_extent_record *rec; + + rec = container_of(cache, struct device_extent_record, cache); + if (!list_empty(&rec->chunk_list)) + list_del_init(&rec->chunk_list); + if (!list_empty(&rec->device_list)) + list_del_init(&rec->device_list); + free(rec); +} + +void free_device_extent_tree(struct device_extent_tree *tree) +{ + cache_tree_free_extents(&tree->tree, free_device_extent_record); +} + #ifdef BTRFS_COMPAT_EXTENT_TREE_V0 static int process_extent_ref_v0(struct cache_tree *extent_cache, struct extent_buffer *leaf, int slot) @@ -2618,6 +3532,221 @@ static int process_extent_ref_v0(struct cache_tree *extent_cache, } #endif +struct chunk_record *btrfs_new_chunk_record(struct extent_buffer *leaf, + struct btrfs_key *key, + int slot) +{ + struct btrfs_chunk *ptr; + struct chunk_record *rec; + int num_stripes, i; + + ptr = btrfs_item_ptr(leaf, slot, struct btrfs_chunk); + num_stripes = btrfs_chunk_num_stripes(leaf, ptr); + + rec = malloc(btrfs_chunk_record_size(num_stripes)); + if (!rec) { + fprintf(stderr, "memory allocation failed\n"); + exit(-1); + } + + memset(rec, 0, btrfs_chunk_record_size(num_stripes)); + + INIT_LIST_HEAD(&rec->list); + INIT_LIST_HEAD(&rec->dextents); + rec->bg_rec = NULL; + + rec->cache.start = key->offset; + rec->cache.size = btrfs_chunk_length(leaf, ptr); + + rec->generation = btrfs_header_generation(leaf); + + rec->objectid = key->objectid; + rec->type = key->type; + rec->offset = key->offset; + + rec->length = rec->cache.size; + rec->owner = btrfs_chunk_owner(leaf, ptr); + rec->stripe_len = btrfs_chunk_stripe_len(leaf, ptr); + rec->type_flags = btrfs_chunk_type(leaf, ptr); + rec->io_width = btrfs_chunk_io_width(leaf, ptr); + rec->io_align = btrfs_chunk_io_align(leaf, ptr); + rec->sector_size = btrfs_chunk_sector_size(leaf, ptr); + rec->num_stripes = num_stripes; + rec->sub_stripes = btrfs_chunk_sub_stripes(leaf, ptr); + + for (i = 0; i < rec->num_stripes; ++i) { + rec->stripes[i].devid = + btrfs_stripe_devid_nr(leaf, ptr, i); + rec->stripes[i].offset = + btrfs_stripe_offset_nr(leaf, ptr, i); + read_extent_buffer(leaf, rec->stripes[i].dev_uuid, + (unsigned long)btrfs_stripe_dev_uuid_nr(ptr, i), + BTRFS_UUID_SIZE); + } + + return rec; +} + +static int process_chunk_item(struct cache_tree *chunk_cache, + struct btrfs_key *key, struct extent_buffer *eb, + int slot) +{ + struct chunk_record *rec; + int ret = 0; + + rec = btrfs_new_chunk_record(eb, key, slot); + ret = insert_cache_extent(chunk_cache, &rec->cache); + if (ret) { + fprintf(stderr, "Chunk[%llu, %llu] existed.\n", + rec->offset, rec->length); + free(rec); + } + + return ret; +} + +static int process_device_item(struct rb_root *dev_cache, + struct btrfs_key *key, struct extent_buffer *eb, int slot) +{ + struct btrfs_dev_item *ptr; + struct device_record *rec; + int ret = 0; + + ptr = btrfs_item_ptr(eb, + slot, struct btrfs_dev_item); + + rec = malloc(sizeof(*rec)); + if (!rec) { + fprintf(stderr, "memory allocation failed\n"); + return -ENOMEM; + } + + rec->devid = key->offset; + rec->generation = btrfs_header_generation(eb); + + rec->objectid = key->objectid; + rec->type = key->type; + rec->offset = key->offset; + + rec->devid = btrfs_device_id(eb, ptr); + rec->total_byte = btrfs_device_total_bytes(eb, ptr); + rec->byte_used = btrfs_device_bytes_used(eb, ptr); + + ret = rb_insert(dev_cache, &rec->node, device_record_compare); + if (ret) { + fprintf(stderr, "Device[%llu] existed.\n", rec->devid); + free(rec); + } + + return ret; +} + +struct block_group_record * +btrfs_new_block_group_record(struct extent_buffer *leaf, struct btrfs_key *key, + int slot) +{ + struct btrfs_block_group_item *ptr; + struct block_group_record *rec; + + rec = malloc(sizeof(*rec)); + if (!rec) { + fprintf(stderr, "memory allocation failed\n"); + exit(-1); + } + memset(rec, 0, sizeof(*rec)); + + rec->cache.start = key->objectid; + rec->cache.size = key->offset; + + rec->generation = btrfs_header_generation(leaf); + + rec->objectid = key->objectid; + rec->type = key->type; + rec->offset = key->offset; + + ptr = btrfs_item_ptr(leaf, slot, struct btrfs_block_group_item); + rec->flags = btrfs_disk_block_group_flags(leaf, ptr); + + INIT_LIST_HEAD(&rec->list); + + return rec; +} + +static int process_block_group_item(struct block_group_tree *block_group_cache, + struct btrfs_key *key, + struct extent_buffer *eb, int slot) +{ + struct block_group_record *rec; + int ret = 0; + + rec = btrfs_new_block_group_record(eb, key, slot); + ret = insert_block_group_record(block_group_cache, rec); + if (ret) { + fprintf(stderr, "Block Group[%llu, %llu] existed.\n", + rec->objectid, rec->offset); + free(rec); + } + + return ret; +} + +struct device_extent_record * +btrfs_new_device_extent_record(struct extent_buffer *leaf, + struct btrfs_key *key, int slot) +{ + struct device_extent_record *rec; + struct btrfs_dev_extent *ptr; + + rec = malloc(sizeof(*rec)); + if (!rec) { + fprintf(stderr, "memory allocation failed\n"); + exit(-1); + } + memset(rec, 0, sizeof(*rec)); + + rec->cache.objectid = key->objectid; + rec->cache.start = key->offset; + + rec->generation = btrfs_header_generation(leaf); + + rec->objectid = key->objectid; + rec->type = key->type; + rec->offset = key->offset; + + ptr = btrfs_item_ptr(leaf, slot, struct btrfs_dev_extent); + rec->chunk_objecteid = + btrfs_dev_extent_chunk_objectid(leaf, ptr); + rec->chunk_offset = + btrfs_dev_extent_chunk_offset(leaf, ptr); + rec->length = btrfs_dev_extent_length(leaf, ptr); + rec->cache.size = rec->length; + + INIT_LIST_HEAD(&rec->chunk_list); + INIT_LIST_HEAD(&rec->device_list); + + return rec; +} + +static int +process_device_extent_item(struct device_extent_tree *dev_extent_cache, + struct btrfs_key *key, struct extent_buffer *eb, + int slot) +{ + struct device_extent_record *rec; + int ret; + + rec = btrfs_new_device_extent_record(eb, key, slot); + ret = insert_device_extent_record(dev_extent_cache, rec); + if (ret) { + fprintf(stderr, + "Device extent[%llu, %llu, %llu] existed.\n", + rec->objectid, rec->offset, rec->length); + free(rec); + } + + return ret; +} + static int process_extent_item(struct btrfs_root *root, struct cache_tree *extent_cache, struct extent_buffer *eb, int slot) @@ -2654,7 +3783,7 @@ static int process_extent_item(struct btrfs_root *root, #else BUG(); #endif - return add_extent_rec(extent_cache, NULL, key.objectid, + return add_extent_rec(extent_cache, NULL, 0, key.objectid, num_bytes, refs, 0, 0, 0, metadata, 1, num_bytes); } @@ -2662,7 +3791,7 @@ static int process_extent_item(struct btrfs_root *root, ei = btrfs_item_ptr(eb, slot, struct btrfs_extent_item); refs = btrfs_extent_refs(eb, ei); - add_extent_rec(extent_cache, NULL, key.objectid, num_bytes, + add_extent_rec(extent_cache, NULL, 0, key.objectid, num_bytes, refs, 0, 0, 0, metadata, 1, num_bytes); ptr = (unsigned long)(ei + 1); @@ -2830,13 +3959,13 @@ static int verify_space_cache(struct btrfs_root *root, ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); if (ret < 0) - return ret; + goto out; ret = 0; while (1) { if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) { ret = btrfs_next_leaf(root, path); if (ret < 0) - return ret; + goto out; if (ret > 0) { ret = 0; break; @@ -2876,6 +4005,8 @@ static int verify_space_cache(struct btrfs_root *root, ret = check_cache_range(root, cache, last, cache->key.objectid + cache->key.offset - last); + +out: btrfs_free_path(path); if (!ret && @@ -2895,7 +4026,8 @@ static int check_space_cache(struct btrfs_root *root) int ret; int error = 0; - if (btrfs_super_generation(root->fs_info->super_copy) != + if (btrfs_super_cache_generation(root->fs_info->super_copy) != -1ULL && + btrfs_super_generation(root->fs_info->super_copy) != btrfs_super_cache_generation(root->fs_info->super_copy)) { printf("cache and super generation don't match, space cache " "will be invalidated\n"); @@ -2933,6 +4065,109 @@ static int check_space_cache(struct btrfs_root *root) return error ? -EINVAL : 0; } +static int read_extent_data(struct btrfs_root *root, char *data, + u64 logical, u64 *len, int mirror) +{ + u64 offset = 0; + struct btrfs_multi_bio *multi = NULL; + struct btrfs_fs_info *info = root->fs_info; + struct btrfs_device *device; + int ret = 0; + u64 max_len = *len; + + ret = btrfs_map_block(&info->mapping_tree, READ, logical, len, + &multi, mirror, NULL); + if (ret) { + fprintf(stderr, "Couldn't map the block %llu\n", + logical + offset); + goto err; + } + device = multi->stripes[0].dev; + + if (device->fd == 0) + goto err; + if (*len > max_len) + *len = max_len; + + ret = pread64(device->fd, data, *len, multi->stripes[0].physical); + if (ret != *len) + ret = -EIO; + else + ret = 0; +err: + kfree(multi); + return ret; +} + +static int check_extent_csums(struct btrfs_root *root, u64 bytenr, + u64 num_bytes, unsigned long leaf_offset, + struct extent_buffer *eb) { + + u64 offset = 0; + u16 csum_size = btrfs_super_csum_size(root->fs_info->super_copy); + char *data; + unsigned long csum_offset; + u32 csum; + u32 csum_expected; + u64 read_len; + u64 data_checked = 0; + u64 tmp; + int ret = 0; + int mirror; + int num_copies; + + if (num_bytes % root->sectorsize) + return -EINVAL; + + data = malloc(num_bytes); + if (!data) + return -ENOMEM; + + while (offset < num_bytes) { + mirror = 0; +again: + read_len = num_bytes - offset; + /* read as much space once a time */ + ret = read_extent_data(root, data + offset, + bytenr + offset, &read_len, mirror); + if (ret) + goto out; + data_checked = 0; + /* verify every 4k data's checksum */ + while (data_checked < read_len) { + csum = ~(u32)0; + tmp = offset + data_checked; + + csum = btrfs_csum_data(NULL, (char *)data + tmp, + csum, root->sectorsize); + btrfs_csum_final(csum, (char *)&csum); + + csum_offset = leaf_offset + + tmp / root->sectorsize * csum_size; + read_extent_buffer(eb, (char *)&csum_expected, + csum_offset, csum_size); + /* try another mirror */ + if (csum != csum_expected) { + fprintf(stderr, "mirror %d bytenr %llu csum %u expected csum %u\n", + mirror, bytenr + tmp, + csum, csum_expected); + num_copies = btrfs_num_copies( + &root->fs_info->mapping_tree, + bytenr, num_bytes); + if (mirror < num_copies - 1) { + mirror += 1; + goto again; + } + } + data_checked += root->sectorsize; + } + offset += read_len; + } +out: + free(data); + return ret; +} + static int check_extent_exists(struct btrfs_root *root, u64 bytenr, u64 num_bytes) { @@ -2949,8 +4184,7 @@ static int check_extent_exists(struct btrfs_root *root, u64 bytenr, key.objectid = bytenr; key.type = BTRFS_EXTENT_ITEM_KEY; - key.offset = 0; - + key.offset = (u64)-1; again: ret = btrfs_search_slot(NULL, root->fs_info->extent_root, &key, path, @@ -2960,10 +4194,17 @@ again: btrfs_free_path(path); return ret; } else if (ret) { - if (path->slots[0]) + if (path->slots[0] > 0) { path->slots[0]--; - else - btrfs_prev_leaf(root, path); + } else { + ret = btrfs_prev_leaf(root, path); + if (ret < 0) { + goto out; + } else if (ret > 0) { + ret = 0; + goto out; + } + } } btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]); @@ -2973,13 +4214,22 @@ again: * bytenr, so walk back one more just in case. Dear future traveler, * first congrats on mastering time travel. Now if it's not too much * trouble could you go back to 2006 and tell Chris to make the - * BLOCK_GROUP_ITEM_KEY lower than the EXTENT_ITEM_KEY please? + * BLOCK_GROUP_ITEM_KEY (and BTRFS_*_REF_KEY) lower than the + * EXTENT_ITEM_KEY please? */ - if (key.type == BTRFS_BLOCK_GROUP_ITEM_KEY) { - if (path->slots[0]) + while (key.type > BTRFS_EXTENT_ITEM_KEY) { + if (path->slots[0] > 0) { path->slots[0]--; - else - btrfs_prev_leaf(root, path); + } else { + ret = btrfs_prev_leaf(root, path); + if (ret < 0) { + goto out; + } else if (ret > 0) { + ret = 0; + goto out; + } + } + btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]); } while (num_bytes) { @@ -3034,7 +4284,7 @@ again: * in real life, but no harm in coding it up * anyway just in case. */ - btrfs_release_path(root, path); + btrfs_release_path(path); ret = check_extent_exists(root, new_start, new_bytes); if (ret) { @@ -3051,7 +4301,8 @@ again: } ret = 0; - if (num_bytes) { +out: + if (num_bytes && !ret) { fprintf(stderr, "There are no extents for csum range " "%Lu-%Lu\n", bytenr, bytenr+num_bytes); ret = 1; @@ -3070,8 +4321,14 @@ static int check_csums(struct btrfs_root *root) u16 csum_size = btrfs_super_csum_size(root->fs_info->super_copy); int errors = 0; int ret; + u64 data_len; + unsigned long leaf_offset; root = root->fs_info->csum_root; + if (!extent_buffer_uptodate(root->node)) { + fprintf(stderr, "No valid csum tree found\n"); + return -ENOENT; + } key.objectid = BTRFS_EXTENT_CSUM_OBJECTID; key.type = BTRFS_EXTENT_CSUM_KEY; @@ -3111,6 +4368,16 @@ static int check_csums(struct btrfs_root *root) continue; } + data_len = (btrfs_item_size_nr(leaf, path->slots[0]) / + csum_size) * root->sectorsize; + if (!check_data_csum) + goto skip_csum_check; + leaf_offset = btrfs_item_ptr_offset(leaf, path->slots[0]); + ret = check_extent_csums(root, key.offset, data_len, + leaf_offset, leaf); + if (ret) + break; +skip_csum_check: if (!num_bytes) { offset = key.offset; } else if (key.offset != offset + num_bytes) { @@ -3124,9 +4391,7 @@ static int check_csums(struct btrfs_root *root) offset = key.offset; num_bytes = 0; } - - num_bytes += (btrfs_item_size_nr(leaf, path->slots[0]) / - csum_size) * root->sectorsize; + num_bytes += data_len; path->slots[0]++; } @@ -3134,7 +4399,23 @@ static int check_csums(struct btrfs_root *root) return errors; } -static int run_next_block(struct btrfs_root *root, +static int is_dropped_key(struct btrfs_key *key, + struct btrfs_key *drop_key) { + if (key->objectid < drop_key->objectid) + return 1; + else if (key->objectid == drop_key->objectid) { + if (key->type < drop_key->type) + return 1; + else if (key->type == drop_key->type) { + if (key->offset < drop_key->offset) + return 1; + } + } + return 0; +} + +static int run_next_block(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct block_info *bits, int bits_nr, u64 *last, @@ -3142,7 +4423,12 @@ static int run_next_block(struct btrfs_root *root, struct cache_tree *seen, struct cache_tree *reada, struct cache_tree *nodes, - struct cache_tree *extent_cache) + struct cache_tree *extent_cache, + struct cache_tree *chunk_cache, + struct rb_root *dev_cache, + struct block_group_tree *block_group_cache, + struct device_extent_tree *dev_extent_cache, + struct btrfs_root_item *ri) { struct extent_buffer *buf; u64 bytenr; @@ -3150,22 +4436,26 @@ static int run_next_block(struct btrfs_root *root, u64 parent; u64 owner; u64 flags; - int ret; + u64 ptr; + u64 gen = 0; + int ret = 0; int i; int nritems; struct btrfs_key key; struct cache_extent *cache; int reada_bits; - ret = pick_next_pending(pending, reada, nodes, *last, bits, - bits_nr, &reada_bits); - if (ret == 0) { + nritems = pick_next_pending(pending, reada, nodes, *last, bits, + bits_nr, &reada_bits); + if (nritems == 0) return 1; - } + if (!reada_bits) { - for(i = 0; i < ret; i++) { - insert_cache_extent(reada, bits[i].start, - bits[i].size); + for(i = 0; i < nritems; i++) { + ret = add_cache_extent(reada, bits[i].start, + bits[i].size); + if (ret == -EEXIST) + continue; /* fixme, get the parent transid */ readahead_tree_block(root, bits[i].start, @@ -3176,24 +4466,31 @@ static int run_next_block(struct btrfs_root *root, bytenr = bits[0].start; size = bits[0].size; - cache = find_cache_extent(pending, bytenr, size); + cache = lookup_cache_extent(pending, bytenr, size); if (cache) { remove_cache_extent(pending, cache); free(cache); } - cache = find_cache_extent(reada, bytenr, size); + cache = lookup_cache_extent(reada, bytenr, size); if (cache) { remove_cache_extent(reada, cache); free(cache); } - cache = find_cache_extent(nodes, bytenr, size); + cache = lookup_cache_extent(nodes, bytenr, size); if (cache) { remove_cache_extent(nodes, cache); free(cache); } + cache = lookup_cache_extent(extent_cache, bytenr, size); + if (cache) { + struct extent_record *rec; + + rec = container_of(cache, struct extent_record, cache); + gen = rec->parent_generation; + } /* fixme, get the real parent transid */ - buf = read_tree_block(root, bytenr, size, 0); + buf = read_tree_block(root, bytenr, size, gen); if (!extent_buffer_uptodate(buf)) { record_bad_block_io(root->fs_info, extent_cache, bytenr, size); @@ -3202,11 +4499,19 @@ static int run_next_block(struct btrfs_root *root, nritems = btrfs_header_nritems(buf); - ret = btrfs_lookup_extent_info(NULL, root, bytenr, + /* + * FIXME, this only works only if we don't have any full + * backref mode. + */ + if (!init_extent_tree) { + ret = btrfs_lookup_extent_info(NULL, root, bytenr, btrfs_header_level(buf), 1, NULL, &flags); - if (ret < 0) - flags = BTRFS_BLOCK_FLAG_FULL_BACKREF; + if (ret < 0) + flags = 0; + } else { + flags = 0; + } if (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF) { parent = bytenr; @@ -3216,7 +4521,7 @@ static int run_next_block(struct btrfs_root *root, owner = btrfs_header_owner(buf); } - ret = check_block(root, extent_cache, buf, flags); + ret = check_block(trans, root, extent_cache, buf, flags); if (ret) goto out; @@ -3240,8 +4545,24 @@ static int run_next_block(struct btrfs_root *root, btrfs_item_size_nr(buf, i); continue; } + if (key.type == BTRFS_CHUNK_ITEM_KEY) { + process_chunk_item(chunk_cache, &key, buf, i); + continue; + } + if (key.type == BTRFS_DEV_ITEM_KEY) { + process_device_item(dev_cache, &key, buf, i); + continue; + } if (key.type == BTRFS_BLOCK_GROUP_ITEM_KEY) { + process_block_group_item(block_group_cache, + &key, buf, i); + continue; + } + if (key.type == BTRFS_DEV_EXTENT_KEY) { + process_device_extent_item(dev_extent_cache, + &key, buf, i); continue; + } if (key.type == BTRFS_EXTENT_REF_V0_KEY) { #ifdef BTRFS_COMPAT_EXTENT_TREE_V0 @@ -3281,11 +4602,28 @@ static int run_next_block(struct btrfs_root *root, ref = btrfs_item_ptr(buf, i, struct btrfs_shared_data_ref); add_data_backref(extent_cache, - key.objectid, key.offset, 0, 0, 0, + key.objectid, key.offset, 0, 0, 0, btrfs_shared_data_ref_count(buf, ref), 0, root->sectorsize); continue; } + if (key.type == BTRFS_ORPHAN_ITEM_KEY) { + struct bad_item *bad; + + if (key.objectid == BTRFS_ORPHAN_OBJECTID) + continue; + if (!owner) + continue; + bad = malloc(sizeof(struct bad_item)); + if (!bad) + continue; + INIT_LIST_HEAD(&bad->list); + memcpy(&bad->key, &key, + sizeof(struct btrfs_key)); + bad->root_id = owner; + list_add_tail(&bad->list, &delete_items); + continue; + } if (key.type != BTRFS_EXTENT_DATA_KEY) continue; fi = btrfs_item_ptr(buf, i, @@ -3308,7 +4646,6 @@ static int run_next_block(struct btrfs_root *root, parent, owner, key.objectid, key.offset - btrfs_file_extent_offset(buf, fi), 1, 1, btrfs_file_extent_disk_num_bytes(buf, fi)); - BUG_ON(ret); } } else { int level; @@ -3320,10 +4657,20 @@ static int run_next_block(struct btrfs_root *root, btrfs_item_key_to_cpu(buf, &first_key, 0); level = btrfs_header_level(buf); for (i = 0; i < nritems; i++) { - u64 ptr = btrfs_node_blockptr(buf, i); - u32 size = btrfs_level_size(root, level - 1); + ptr = btrfs_node_blockptr(buf, i); + size = btrfs_level_size(root, level - 1); btrfs_node_key_to_cpu(buf, &key, i); + if (ri != NULL) { + struct btrfs_key drop_key; + btrfs_disk_key_to_cpu(&drop_key, + &ri->drop_progress); + if ((level == ri->drop_level) + && is_dropped_key(&key, &drop_key)) { + continue; + } + } ret = add_extent_rec(extent_cache, &key, + btrfs_node_ptr_generation(buf, i), ptr, size, 0, 0, 1, 0, 1, 0, size); BUG_ON(ret); @@ -3351,7 +4698,7 @@ static int run_next_block(struct btrfs_root *root, found_old_backref = 1; out: free_extent_buffer(buf); - return 0; + return ret; } static int add_root_to_pending(struct extent_buffer *buf, @@ -3365,7 +4712,7 @@ static int add_root_to_pending(struct extent_buffer *buf, add_pending(nodes, seen, buf->start, buf->len); else add_pending(pending, seen, buf->start, buf->len); - add_extent_rec(extent_cache, NULL, buf->start, buf->len, + add_extent_rec(extent_cache, NULL, 0, buf->start, buf->len, 0, 1, 1, 0, 1, 0, buf->len); if (root_key->objectid == BTRFS_TREE_RELOC_OBJECTID || @@ -3394,7 +4741,7 @@ static int free_extent_hook(struct btrfs_trans_handle *trans, struct cache_tree *extent_cache = root->fs_info->fsck_extent_cache; is_data = owner >= BTRFS_FIRST_FREE_OBJECTID; - cache = find_cache_extent(extent_cache, bytenr, num_bytes); + cache = lookup_cache_extent(extent_cache, bytenr, num_bytes); if (!cache) return 0; @@ -3493,7 +4840,7 @@ static int delete_extent_records(struct btrfs_trans_handle *trans, found_key.type != BTRFS_EXTENT_REF_V0_KEY && found_key.type != BTRFS_SHARED_BLOCK_REF_KEY && found_key.type != BTRFS_SHARED_DATA_REF_KEY) { - btrfs_release_path(NULL, path); + btrfs_release_path(path); if (found_key.type == 0) { if (found_key.offset == 0) break; @@ -3511,7 +4858,7 @@ static int delete_extent_records(struct btrfs_trans_handle *trans, ret = btrfs_del_item(trans, root->fs_info->extent_root, path); if (ret) break; - btrfs_release_path(NULL, path); + btrfs_release_path(path); if (found_key.type == BTRFS_EXTENT_ITEM_KEY || found_key.type == BTRFS_METADATA_ITEM_KEY) { @@ -3525,7 +4872,7 @@ static int delete_extent_records(struct btrfs_trans_handle *trans, } } - btrfs_release_path(NULL, path); + btrfs_release_path(path); return ret; } @@ -3585,9 +4932,12 @@ static int record_extent(struct btrfs_trans_handle *trans, bi = (struct btrfs_tree_block_info *)(ei + 1); memset_extent_buffer(leaf, 0, (unsigned long)bi, sizeof(*bi)); - memset(©_key, 0, sizeof(copy_key)); - copy_key.objectid = le64_to_cpu(rec->info_objectid); + btrfs_set_disk_key_objectid(©_key, + rec->info_objectid); + btrfs_set_disk_key_type(©_key, 0); + btrfs_set_disk_key_offset(©_key, 0); + btrfs_set_tree_block_level(leaf, bi, rec->info_level); btrfs_set_tree_block_key(leaf, bi, ©_key); @@ -3600,7 +4950,7 @@ static int record_extent(struct btrfs_trans_handle *trans, rec->max_size, 1, 0); if (ret) goto fail; - btrfs_release_path(NULL, path); + btrfs_release_path(path); } if (back->is_data) { @@ -3661,7 +5011,7 @@ static int record_extent(struct btrfs_trans_handle *trans, if (ret) goto fail; fail: - btrfs_release_path(NULL, path); + btrfs_release_path(path); return ret; } @@ -3669,6 +5019,7 @@ struct extent_entry { u64 bytenr; u64 bytes; int count; + int broken; struct list_head list; }; @@ -3696,6 +5047,13 @@ static struct extent_entry *find_most_right_entry(struct list_head *entries) } /* + * If there are as many broken entries as entries then we know + * not to trust this particular entry. + */ + if (entry->broken == entry->count) + continue; + + /* * If our current entry == best then we can't be sure our best * is really the best, so we need to keep searching. */ @@ -3706,7 +5064,7 @@ static struct extent_entry *find_most_right_entry(struct list_head *entries) } /* Prev == entry, not good enough, have to keep searching */ - if (prev->count == entry->count) + if (!prev->broken && prev->count == entry->count) continue; if (!best) @@ -3778,18 +5136,13 @@ static int repair_ref(struct btrfs_trans_handle *trans, path->slots[0]++; } - btrfs_release_path(root, path); + btrfs_release_path(path); /* * Have to make sure that this root gets updated when we commit the * transaction */ - root->track_dirty = 1; - if (root->last_trans != trans->transid) { - root->last_trans = trans->transid; - root->commit_root = root->node; - extent_buffer_get(root->node); - } + record_root_in_trans(trans, root); /* * Ok we have the key of the file extent we want to fix, now we can cow @@ -3821,7 +5174,9 @@ static int repair_ref(struct btrfs_trans_handle *trans, return -EINVAL; } - if (dback->disk_bytenr > entry->bytenr) { + if (dback->node.broken && dback->disk_bytenr != entry->bytenr) { + btrfs_set_file_extent_disk_bytenr(leaf, fi, entry->bytenr); + } else if (dback->disk_bytenr > entry->bytenr) { u64 off_diff, offset; off_diff = dback->disk_bytenr - entry->bytenr; @@ -3868,7 +5223,7 @@ static int repair_ref(struct btrfs_trans_handle *trans, else printf("ram bytes may be wrong?\n"); btrfs_mark_buffer_dirty(leaf); - btrfs_release_path(root, path); + btrfs_release_path(path); return 0; } @@ -3881,7 +5236,9 @@ static int verify_backrefs(struct btrfs_trans_handle *trans, struct extent_entry *entry, *best = NULL; LIST_HEAD(entries); int nr_entries = 0; + int broken_entries = 0; int ret = 0; + short mismatch = 0; /* * Metadata is easy and the backrefs should always agree on bytenr and @@ -3891,15 +5248,17 @@ static int verify_backrefs(struct btrfs_trans_handle *trans, return 0; list_for_each_entry(back, &rec->backrefs, list) { + if (back->full_backref || !back->is_data) + continue; + dback = (struct data_backref *)back; + /* * We only pay attention to backrefs that we found a real * backref for. */ if (dback->found_ref == 0) continue; - if (back->full_backref) - continue; /* * For now we only catch when the bytes don't match, not the @@ -3923,11 +5282,25 @@ static int verify_backrefs(struct btrfs_trans_handle *trans, list_add_tail(&entry->list, &entries); nr_entries++; } + + /* + * If we only have on entry we may think the entries agree when + * in reality they don't so we have to do some extra checking. + */ + if (dback->disk_bytenr != rec->start || + dback->bytes != rec->nr || back->broken) + mismatch = 1; + + if (back->broken) { + entry->broken++; + broken_entries++; + } + entry->count++; } /* Yay all the backrefs agree, carry on good sir */ - if (nr_entries <= 1) + if (nr_entries <= 1 && !mismatch) goto out; fprintf(stderr, "attempting to repair backref discrepency for bytenr " @@ -3946,13 +5319,28 @@ static int verify_backrefs(struct btrfs_trans_handle *trans, */ if (!best) { entry = find_entry(&entries, rec->start, rec->nr); - if (!entry) { - fprintf(stderr, "Backrefs don't agree with eachother " + if (!entry && (!broken_entries || !rec->found_rec)) { + fprintf(stderr, "Backrefs don't agree with each other " "and extent record doesn't agree with anybody," " so we can't fix bytenr %Lu bytes %Lu\n", rec->start, rec->nr); ret = -EINVAL; goto out; + } else if (!entry) { + /* + * Ok our backrefs were broken, we'll assume this is the + * correct value and add an entry for this range. + */ + entry = malloc(sizeof(struct extent_entry)); + if (!entry) { + ret = -ENOMEM; + goto out; + } + memset(entry, 0, sizeof(*entry)); + entry->bytenr = rec->start; + entry->bytes = rec->nr; + list_add_tail(&entry->list, &entries); + nr_entries++; } entry->count++; best = find_most_right_entry(&entries); @@ -3986,6 +5374,9 @@ static int verify_backrefs(struct btrfs_trans_handle *trans, * references and fix up the ones that don't match. */ list_for_each_entry(back, &rec->backrefs, list) { + if (back->full_backref || !back->is_data) + continue; + dback = (struct data_backref *)back; /* @@ -3994,8 +5385,6 @@ static int verify_backrefs(struct btrfs_trans_handle *trans, */ if (dback->found_ref == 0) continue; - if (back->full_backref) - continue; if (dback->bytes == best->bytes && dback->disk_bytenr == best->bytenr) @@ -4023,6 +5412,85 @@ out: return ret; } +static int process_duplicates(struct btrfs_root *root, + struct cache_tree *extent_cache, + struct extent_record *rec) +{ + struct extent_record *good, *tmp; + struct cache_extent *cache; + int ret; + + /* + * If we found a extent record for this extent then return, or if we + * have more than one duplicate we are likely going to need to delete + * something. + */ + if (rec->found_rec || rec->num_duplicates > 1) + return 0; + + /* Shouldn't happen but just in case */ + BUG_ON(!rec->num_duplicates); + + /* + * So this happens if we end up with a backref that doesn't match the + * actual extent entry. So either the backref is bad or the extent + * entry is bad. Either way we want to have the extent_record actually + * reflect what we found in the extent_tree, so we need to take the + * duplicate out and use that as the extent_record since the only way we + * get a duplicate is if we find a real life BTRFS_EXTENT_ITEM_KEY. + */ + remove_cache_extent(extent_cache, &rec->cache); + + good = list_entry(rec->dups.next, struct extent_record, list); + list_del_init(&good->list); + INIT_LIST_HEAD(&good->backrefs); + INIT_LIST_HEAD(&good->dups); + good->cache.start = good->start; + good->cache.size = good->nr; + good->content_checked = 0; + good->owner_ref_checked = 0; + good->num_duplicates = 0; + good->refs = rec->refs; + list_splice_init(&rec->backrefs, &good->backrefs); + while (1) { + cache = lookup_cache_extent(extent_cache, good->start, + good->nr); + if (!cache) + break; + tmp = container_of(cache, struct extent_record, cache); + + /* + * If we find another overlapping extent and it's found_rec is + * set then it's a duplicate and we need to try and delete + * something. + */ + if (tmp->found_rec || tmp->num_duplicates > 0) { + if (list_empty(&good->list)) + list_add_tail(&good->list, + &duplicate_extents); + good->num_duplicates += tmp->num_duplicates + 1; + list_splice_init(&tmp->dups, &good->dups); + list_del_init(&tmp->list); + list_add_tail(&tmp->list, &good->dups); + remove_cache_extent(extent_cache, &tmp->cache); + continue; + } + + /* + * Ok we have another non extent item backed extent rec, so lets + * just add it to this extent and carry on like we did above. + */ + good->refs += tmp->refs; + list_splice_init(&tmp->backrefs, &good->backrefs); + remove_cache_extent(extent_cache, &tmp->cache); + free(tmp); + } + ret = insert_cache_extent(extent_cache, &good->cache); + BUG_ON(ret); + free(rec); + return good->num_duplicates ? 0 : 1; +} + static int delete_duplicate_records(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_record *rec) @@ -4030,6 +5498,7 @@ static int delete_duplicate_records(struct btrfs_trans_handle *trans, LIST_HEAD(delete_list); struct btrfs_path *path; struct extent_record *tmp, *good, *n; + int nr_del = 0; int ret = 0; struct btrfs_key key; @@ -4069,6 +5538,8 @@ static int delete_duplicate_records(struct btrfs_trans_handle *trans, root = root->fs_info->extent_root; list_for_each_entry(tmp, &delete_list, list) { + if (tmp->found_rec == 0) + continue; key.objectid = tmp->start; key.type = BTRFS_EXTENT_ITEM_KEY; key.offset = tmp->nr; @@ -4090,7 +5561,8 @@ static int delete_duplicate_records(struct btrfs_trans_handle *trans, ret = btrfs_del_item(trans, root, path); if (ret) goto out; - btrfs_release_path(root, path); + btrfs_release_path(path); + nr_del++; } out: @@ -4110,7 +5582,98 @@ out: btrfs_free_path(path); - return ret; + if (!ret && !nr_del) + rec->num_duplicates = 0; + + return ret ? ret : nr_del; +} + +static int find_possible_backrefs(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *info, + struct btrfs_path *path, + struct cache_tree *extent_cache, + struct extent_record *rec) +{ + struct btrfs_root *root; + struct extent_backref *back; + struct data_backref *dback; + struct cache_extent *cache; + struct btrfs_file_extent_item *fi; + struct btrfs_key key; + u64 bytenr, bytes; + int ret; + + list_for_each_entry(back, &rec->backrefs, list) { + /* Don't care about full backrefs (poor unloved backrefs) */ + if (back->full_backref || !back->is_data) + continue; + + dback = (struct data_backref *)back; + + /* We found this one, we don't need to do a lookup */ + if (dback->found_ref) + continue; + + key.objectid = dback->root; + key.type = BTRFS_ROOT_ITEM_KEY; + key.offset = (u64)-1; + + root = btrfs_read_fs_root(info, &key); + + /* No root, definitely a bad ref, skip */ + if (IS_ERR(root) && PTR_ERR(root) == -ENOENT) + continue; + /* Other err, exit */ + if (IS_ERR(root)) + return PTR_ERR(root); + + key.objectid = dback->owner; + key.type = BTRFS_EXTENT_DATA_KEY; + key.offset = dback->offset; + ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); + if (ret) { + btrfs_release_path(path); + if (ret < 0) + return ret; + /* Didn't find it, we can carry on */ + ret = 0; + continue; + } + + fi = btrfs_item_ptr(path->nodes[0], path->slots[0], + struct btrfs_file_extent_item); + bytenr = btrfs_file_extent_disk_bytenr(path->nodes[0], fi); + bytes = btrfs_file_extent_disk_num_bytes(path->nodes[0], fi); + btrfs_release_path(path); + cache = lookup_cache_extent(extent_cache, bytenr, 1); + if (cache) { + struct extent_record *tmp; + tmp = container_of(cache, struct extent_record, cache); + + /* + * If we found an extent record for the bytenr for this + * particular backref then we can't add it to our + * current extent record. We only want to add backrefs + * that don't have a corresponding extent item in the + * extent tree since they likely belong to this record + * and we need to fix it if it doesn't match bytenrs. + */ + if (tmp->found_rec) + continue; + } + + dback->found_ref += 1; + dback->disk_bytenr = bytenr; + dback->bytes = bytes; + + /* + * Set this so the verify backref code knows not to trust the + * values in this backref. + */ + back->broken = 1; + } + + return 0; } /* @@ -4120,6 +5683,7 @@ out: */ static int fixup_extent_refs(struct btrfs_trans_handle *trans, struct btrfs_fs_info *info, + struct cache_tree *extent_cache, struct extent_record *rec) { int ret; @@ -4130,14 +5694,38 @@ static int fixup_extent_refs(struct btrfs_trans_handle *trans, int allocated = 0; u64 flags = 0; - /* remember our flags for recreating the extent */ - ret = btrfs_lookup_extent_info(NULL, info->extent_root, rec->start, - rec->max_size, rec->metadata, NULL, - &flags); - if (ret < 0) - flags = BTRFS_BLOCK_FLAG_FULL_BACKREF; + /* + * remember our flags for recreating the extent. + * FIXME, if we have cleared extent tree, we can not + * lookup extent info in extent tree. + */ + if (!init_extent_tree) { + ret = btrfs_lookup_extent_info(NULL, info->extent_root, + rec->start, rec->max_size, + rec->metadata, NULL, &flags); + if (ret < 0) + flags = 0; + } else { + flags = 0; + } path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + + if (rec->refs != rec->extent_item_refs && !rec->metadata) { + /* + * Sometimes the backrefs themselves are so broken they don't + * get attached to any meaningful rec, so first go back and + * check any of our backrefs that we couldn't find and throw + * them into the list if we find the backref so that + * verify_backrefs can figure out what to do. + */ + ret = find_possible_backrefs(trans, info, path, extent_cache, + rec); + if (ret < 0) + goto out; + } /* step one, make sure all of the backrefs agree */ ret = verify_backrefs(trans, info, path, rec); @@ -4152,7 +5740,8 @@ static int fixup_extent_refs(struct btrfs_trans_handle *trans, goto out; /* was this block corrupt? If so, don't add references to it */ - cache = find_cache_extent(info->corrupt_blocks, rec->start, rec->max_size); + cache = lookup_cache_extent(info->corrupt_blocks, + rec->start, rec->max_size); if (cache) { ret = 0; goto out; @@ -4237,7 +5826,7 @@ again: goto out; } else { level++; - btrfs_release_path(NULL, &path); + btrfs_release_path(&path); goto again; } @@ -4246,7 +5835,7 @@ del_ptr: ret = btrfs_del_ptr(trans, info->extent_root, &path, level, slot); out: - btrfs_release_path(NULL, &path); + btrfs_release_path(&path); return ret; } @@ -4256,7 +5845,7 @@ static int prune_corrupt_blocks(struct btrfs_trans_handle *trans, struct cache_extent *cache; struct btrfs_corrupt_block *corrupt; - cache = find_first_cache_extent(info->corrupt_blocks, 0); + cache = search_cache_extent(info->corrupt_blocks, 0); while (1) { if (!cache) break; @@ -4267,69 +5856,15 @@ static int prune_corrupt_blocks(struct btrfs_trans_handle *trans, return 0; } -static void free_corrupt_blocks(struct btrfs_fs_info *info) +static void free_corrupt_block(struct cache_extent *cache) { - struct cache_extent *cache; struct btrfs_corrupt_block *corrupt; - while (1) { - cache = find_first_cache_extent(info->corrupt_blocks, 0); - if (!cache) - break; - corrupt = container_of(cache, struct btrfs_corrupt_block, cache); - remove_cache_extent(info->corrupt_blocks, cache); - free(corrupt); - } -} - -static int check_block_group(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *info, - struct map_lookup *map, - int *reinit) -{ - struct btrfs_key key; - struct btrfs_path path; - int ret; - - key.objectid = map->ce.start; - key.offset = map->ce.size; - key.type = BTRFS_BLOCK_GROUP_ITEM_KEY; - - btrfs_init_path(&path); - ret = btrfs_search_slot(NULL, info->extent_root, - &key, &path, 0, 0); - btrfs_release_path(NULL, &path); - if (ret <= 0) - goto out; - - ret = btrfs_make_block_group(trans, info->extent_root, 0, map->type, - BTRFS_FIRST_CHUNK_TREE_OBJECTID, - key.objectid, key.offset); - *reinit = 1; -out: - return ret; + corrupt = container_of(cache, struct btrfs_corrupt_block, cache); + free(corrupt); } -static int check_block_groups(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *info, int *reinit) -{ - struct cache_extent *ce; - struct map_lookup *map; - struct btrfs_mapping_tree *map_tree = &info->mapping_tree; - - /* this isn't quite working */ - return 0; - - ce = find_first_cache_extent(&map_tree->cache_tree, 0); - while (1) { - if (!ce) - break; - map = container_of(ce, struct map_lookup, ce); - check_block_group(trans, info, map, reinit); - ce = next_cache_extent(ce); - } - return 0; -} +FREE_EXTENT_CACHE_BASED_TREE(corrupt_blocks, free_corrupt_block); static void reset_cached_block_groups(struct btrfs_fs_info *fs_info) { @@ -4359,14 +5894,13 @@ static void reset_cached_block_groups(struct btrfs_fs_info *fs_info) static int check_extent_refs(struct btrfs_trans_handle *trans, struct btrfs_root *root, - struct cache_tree *extent_cache, int repair) + struct cache_tree *extent_cache) { - struct extent_record *rec, *n; + struct extent_record *rec; struct cache_extent *cache; int err = 0; int ret = 0; int fixed = 0; - int reinit = 0; int had_dups = 0; if (repair) { @@ -4376,7 +5910,7 @@ static int check_extent_refs(struct btrfs_trans_handle *trans, * In the worst case, this will be all the * extents in the FS */ - cache = find_first_cache_extent(extent_cache, 0); + cache = search_cache_extent(extent_cache, 0); while(cache) { rec = container_of(cache, struct extent_record, cache); btrfs_pin_extent(root->fs_info, @@ -4385,17 +5919,13 @@ static int check_extent_refs(struct btrfs_trans_handle *trans, } /* pin down all the corrupted blocks too */ - cache = find_first_cache_extent(root->fs_info->corrupt_blocks, 0); + cache = search_cache_extent(root->fs_info->corrupt_blocks, 0); while(cache) { - rec = container_of(cache, struct extent_record, cache); btrfs_pin_extent(root->fs_info, - rec->start, rec->max_size); + cache->start, cache->size); cache = next_cache_extent(cache); } prune_corrupt_blocks(trans, root->fs_info); - check_block_groups(trans, root->fs_info, &reinit); - if (reinit) - btrfs_read_block_groups(root->fs_info->extent_root); reset_cached_block_groups(root->fs_info); } @@ -4404,14 +5934,30 @@ static int check_extent_refs(struct btrfs_trans_handle *trans, * could mess up the extent tree when we have backrefs that actually * belong to a different extent item and not the weird duplicate one. */ - list_for_each_entry_safe(rec, n, &duplicate_extents, list) { - if (!repair) - break; + while (repair && !list_empty(&duplicate_extents)) { + rec = list_entry(duplicate_extents.next, struct extent_record, + list); list_del_init(&rec->list); - had_dups = 1; + + /* Sometimes we can find a backref before we find an actual + * extent, so we need to process it a little bit to see if there + * truly are multiple EXTENT_ITEM_KEY's for the same range, or + * if this is a backref screwup. If we need to delete stuff + * process_duplicates() will return 0, otherwise it will return + * 1 and we + */ + if (process_duplicates(root, extent_cache, rec)) + continue; ret = delete_duplicate_records(trans, root, rec); - if (ret) + if (ret < 0) return ret; + /* + * delete_duplicate_records will return the number of entries + * deleted, so if it's greater than 0 then we know we actually + * did something and we need to remove. + */ + if (ret) + had_dups = 1; } if (had_dups) @@ -4419,11 +5965,11 @@ static int check_extent_refs(struct btrfs_trans_handle *trans, while(1) { fixed = 0; - cache = find_first_cache_extent(extent_cache, 0); + cache = search_cache_extent(extent_cache, 0); if (!cache) break; rec = container_of(cache, struct extent_record, cache); - if (rec->has_duplicate) { + if (rec->num_duplicates) { fprintf(stderr, "extent item %llu has multiple extent " "items\n", (unsigned long long)rec->start); err = 1; @@ -4437,7 +5983,8 @@ static int check_extent_refs(struct btrfs_trans_handle *trans, (unsigned long long)rec->extent_item_refs, (unsigned long long)rec->refs); if (!fixed && repair) { - ret = fixup_extent_refs(trans, root->fs_info, rec); + ret = fixup_extent_refs(trans, root->fs_info, + extent_cache, rec); if (ret) goto repair_abort; fixed = 1; @@ -4451,7 +5998,8 @@ static int check_extent_refs(struct btrfs_trans_handle *trans, (unsigned long long)rec->nr); if (!fixed && repair) { - ret = fixup_extent_refs(trans, root->fs_info, rec); + ret = fixup_extent_refs(trans, root->fs_info, + extent_cache, rec); if (ret) goto repair_abort; fixed = 1; @@ -4464,7 +6012,8 @@ static int check_extent_refs(struct btrfs_trans_handle *trans, (unsigned long long)rec->start, (unsigned long long)rec->nr); if (!fixed && repair) { - ret = fixup_extent_refs(trans, root->fs_info, rec); + ret = fixup_extent_refs(trans, root->fs_info, + extent_cache, rec); if (ret) goto repair_abort; fixed = 1; @@ -4491,21 +6040,256 @@ repair_abort: return err; } -static void free_cache_tree(struct cache_tree *tree) +u64 calc_stripe_length(u64 type, u64 length, int num_stripes) +{ + u64 stripe_size; + + if (type & BTRFS_BLOCK_GROUP_RAID0) { + stripe_size = length; + stripe_size /= num_stripes; + } else if (type & BTRFS_BLOCK_GROUP_RAID10) { + stripe_size = length * 2; + stripe_size /= num_stripes; + } else if (type & BTRFS_BLOCK_GROUP_RAID5) { + stripe_size = length; + stripe_size /= (num_stripes - 1); + } else if (type & BTRFS_BLOCK_GROUP_RAID6) { + stripe_size = length; + stripe_size /= (num_stripes - 2); + } else { + stripe_size = length; + } + return stripe_size; +} + +static int check_chunk_refs(struct chunk_record *chunk_rec, + struct block_group_tree *block_group_cache, + struct device_extent_tree *dev_extent_cache, + int silent) +{ + struct cache_extent *block_group_item; + struct block_group_record *block_group_rec; + struct cache_extent *dev_extent_item; + struct device_extent_record *dev_extent_rec; + u64 devid; + u64 offset; + u64 length; + int i; + int ret = 0; + + block_group_item = lookup_cache_extent(&block_group_cache->tree, + chunk_rec->offset, + chunk_rec->length); + if (block_group_item) { + block_group_rec = container_of(block_group_item, + struct block_group_record, + cache); + if (chunk_rec->length != block_group_rec->offset || + chunk_rec->offset != block_group_rec->objectid || + chunk_rec->type_flags != block_group_rec->flags) { + if (!silent) + fprintf(stderr, + "Chunk[%llu, %u, %llu]: length(%llu), offset(%llu), type(%llu) mismatch with block group[%llu, %u, %llu]: offset(%llu), objectid(%llu), flags(%llu)\n", + chunk_rec->objectid, + chunk_rec->type, + chunk_rec->offset, + chunk_rec->length, + chunk_rec->offset, + chunk_rec->type_flags, + block_group_rec->objectid, + block_group_rec->type, + block_group_rec->offset, + block_group_rec->offset, + block_group_rec->objectid, + block_group_rec->flags); + ret = -1; + } else { + list_del_init(&block_group_rec->list); + chunk_rec->bg_rec = block_group_rec; + } + } else { + if (!silent) + fprintf(stderr, + "Chunk[%llu, %u, %llu]: length(%llu), offset(%llu), type(%llu) is not found in block group\n", + chunk_rec->objectid, + chunk_rec->type, + chunk_rec->offset, + chunk_rec->length, + chunk_rec->offset, + chunk_rec->type_flags); + ret = -1; + } + + length = calc_stripe_length(chunk_rec->type_flags, chunk_rec->length, + chunk_rec->num_stripes); + for (i = 0; i < chunk_rec->num_stripes; ++i) { + devid = chunk_rec->stripes[i].devid; + offset = chunk_rec->stripes[i].offset; + dev_extent_item = lookup_cache_extent2(&dev_extent_cache->tree, + devid, offset, length); + if (dev_extent_item) { + dev_extent_rec = container_of(dev_extent_item, + struct device_extent_record, + cache); + if (dev_extent_rec->objectid != devid || + dev_extent_rec->offset != offset || + dev_extent_rec->chunk_offset != chunk_rec->offset || + dev_extent_rec->length != length) { + if (!silent) + fprintf(stderr, + "Chunk[%llu, %u, %llu] stripe[%llu, %llu] dismatch dev extent[%llu, %llu, %llu]\n", + chunk_rec->objectid, + chunk_rec->type, + chunk_rec->offset, + chunk_rec->stripes[i].devid, + chunk_rec->stripes[i].offset, + dev_extent_rec->objectid, + dev_extent_rec->offset, + dev_extent_rec->length); + ret = -1; + } else { + list_move(&dev_extent_rec->chunk_list, + &chunk_rec->dextents); + } + } else { + if (!silent) + fprintf(stderr, + "Chunk[%llu, %u, %llu] stripe[%llu, %llu] is not found in dev extent\n", + chunk_rec->objectid, + chunk_rec->type, + chunk_rec->offset, + chunk_rec->stripes[i].devid, + chunk_rec->stripes[i].offset); + ret = -1; + } + } + return ret; +} + +/* check btrfs_chunk -> btrfs_dev_extent / btrfs_block_group_item */ +int check_chunks(struct cache_tree *chunk_cache, + struct block_group_tree *block_group_cache, + struct device_extent_tree *dev_extent_cache, + struct list_head *good, struct list_head *bad, int silent) +{ + struct cache_extent *chunk_item; + struct chunk_record *chunk_rec; + struct block_group_record *bg_rec; + struct device_extent_record *dext_rec; + int err; + int ret = 0; + + chunk_item = first_cache_extent(chunk_cache); + while (chunk_item) { + chunk_rec = container_of(chunk_item, struct chunk_record, + cache); + err = check_chunk_refs(chunk_rec, block_group_cache, + dev_extent_cache, silent); + if (err) { + ret = err; + if (bad) + list_add_tail(&chunk_rec->list, bad); + } else { + if (good) + list_add_tail(&chunk_rec->list, good); + } + + chunk_item = next_cache_extent(chunk_item); + } + + list_for_each_entry(bg_rec, &block_group_cache->block_groups, list) { + if (!silent) + fprintf(stderr, + "Block group[%llu, %llu] (flags = %llu) didn't find the relative chunk.\n", + bg_rec->objectid, + bg_rec->offset, + bg_rec->flags); + if (!ret) + ret = 1; + } + + list_for_each_entry(dext_rec, &dev_extent_cache->no_chunk_orphans, + chunk_list) { + if (!silent) + fprintf(stderr, + "Device extent[%llu, %llu, %llu] didn't find the relative chunk.\n", + dext_rec->objectid, + dext_rec->offset, + dext_rec->length); + if (!ret) + ret = 1; + } + return ret; +} + + +static int check_device_used(struct device_record *dev_rec, + struct device_extent_tree *dext_cache) { struct cache_extent *cache; + struct device_extent_record *dev_extent_rec; + u64 total_byte = 0; - while (1) { - cache = find_first_cache_extent(tree, 0); - if (!cache) + cache = search_cache_extent2(&dext_cache->tree, dev_rec->devid, 0); + while (cache) { + dev_extent_rec = container_of(cache, + struct device_extent_record, + cache); + if (dev_extent_rec->objectid != dev_rec->devid) break; - remove_cache_extent(tree, cache); - free(cache); + + list_del_init(&dev_extent_rec->device_list); + total_byte += dev_extent_rec->length; + cache = next_cache_extent(cache); + } + + if (total_byte != dev_rec->byte_used) { + fprintf(stderr, + "Dev extent's total-byte(%llu) is not equal to byte-used(%llu) in dev[%llu, %u, %llu]\n", + total_byte, dev_rec->byte_used, dev_rec->objectid, + dev_rec->type, dev_rec->offset); + return -1; + } else { + return 0; + } +} + +/* check btrfs_dev_item -> btrfs_dev_extent */ +static int check_devices(struct rb_root *dev_cache, + struct device_extent_tree *dev_extent_cache) +{ + struct rb_node *dev_node; + struct device_record *dev_rec; + struct device_extent_record *dext_rec; + int err; + int ret = 0; + + dev_node = rb_first(dev_cache); + while (dev_node) { + dev_rec = container_of(dev_node, struct device_record, node); + err = check_device_used(dev_rec, dev_extent_cache); + if (err) + ret = err; + + dev_node = rb_next(dev_node); + } + list_for_each_entry(dext_rec, &dev_extent_cache->no_device_orphans, + device_list) { + fprintf(stderr, + "Device extent[%llu, %llu, %llu] didn't find its device.\n", + dext_rec->objectid, dext_rec->offset, dext_rec->length); + if (!ret) + ret = 1; } + return ret; } -static int check_extents(struct btrfs_root *root, int repair) +static int check_chunks_and_extents(struct btrfs_root *root) { + struct rb_root dev_cache; + struct cache_tree chunk_cache; + struct block_group_tree block_group_cache; + struct device_extent_tree dev_extent_cache; struct cache_tree extent_cache; struct cache_tree seen; struct cache_tree pending; @@ -4515,7 +6299,7 @@ static int check_extents(struct btrfs_root *root, int repair) struct btrfs_path path; struct btrfs_key key; struct btrfs_key found_key; - int ret; + int ret, err = 0; u64 last = 0; struct block_info *bits; int bits_nr; @@ -4523,6 +6307,12 @@ static int check_extents(struct btrfs_root *root, int repair) struct btrfs_trans_handle *trans = NULL; int slot; struct btrfs_root_item ri; + struct list_head dropping_trees; + + dev_cache = RB_ROOT; + cache_tree_init(&chunk_cache); + block_group_tree_init(&block_group_cache); + device_extent_tree_init(&dev_extent_cache); cache_tree_init(&extent_cache); cache_tree_init(&seen); @@ -4530,6 +6320,7 @@ static int check_extents(struct btrfs_root *root, int repair) cache_tree_init(&nodes); cache_tree_init(&reada); cache_tree_init(&corrupt_blocks); + INIT_LIST_HEAD(&dropping_trees); if (repair) { trans = btrfs_start_transaction(root, 1); @@ -4564,7 +6355,8 @@ again: btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY); ret = btrfs_search_slot(NULL, root->fs_info->tree_root, &key, &path, 0, 0); - BUG_ON(ret < 0); + if (ret < 0) + goto out; while(1) { leaf = path.nodes[0]; slot = path.slots[0]; @@ -4582,25 +6374,84 @@ again: offset = btrfs_item_ptr_offset(leaf, path.slots[0]); read_extent_buffer(leaf, &ri, offset, sizeof(ri)); - buf = read_tree_block(root->fs_info->tree_root, - btrfs_root_bytenr(&ri), - btrfs_level_size(root, - btrfs_root_level(&ri)), 0); - add_root_to_pending(buf, &extent_cache, &pending, - &seen, &nodes, &found_key); - free_extent_buffer(buf); + if (btrfs_disk_key_objectid(&ri.drop_progress) == 0) { + buf = read_tree_block(root->fs_info->tree_root, + btrfs_root_bytenr(&ri), + btrfs_level_size(root, + btrfs_root_level(&ri)), + 0); + if (!buf) { + ret = -EIO; + goto out; + } + add_root_to_pending(buf, &extent_cache, + &pending, &seen, &nodes, + &found_key); + free_extent_buffer(buf); + } else { + struct dropping_root_item_record *dri_rec; + dri_rec = malloc(sizeof(*dri_rec)); + if (!dri_rec) { + perror("malloc"); + exit(1); + } + memcpy(&dri_rec->ri, &ri, sizeof(ri)); + memcpy(&dri_rec->found_key, &found_key, + sizeof(found_key)); + list_add_tail(&dri_rec->list, &dropping_trees); + } } path.slots[0]++; } - btrfs_release_path(root, &path); - while(1) { - ret = run_next_block(root, bits, bits_nr, &last, &pending, - &seen, &reada, &nodes, &extent_cache); + btrfs_release_path(&path); + while (1) { + ret = run_next_block(trans, root, bits, bits_nr, &last, + &pending, &seen, &reada, &nodes, + &extent_cache, &chunk_cache, &dev_cache, + &block_group_cache, &dev_extent_cache, + NULL); if (ret != 0) break; } - ret = check_extent_refs(trans, root, &extent_cache, repair); + while (!list_empty(&dropping_trees)) { + struct dropping_root_item_record *rec; + struct extent_buffer *buf; + rec = list_entry(dropping_trees.next, + struct dropping_root_item_record, list); + last = 0; + if (!bits) { + perror("realloc"); + exit(1); + } + buf = read_tree_block(root->fs_info->tree_root, + btrfs_root_bytenr(&rec->ri), + btrfs_level_size(root, + btrfs_root_level(&rec->ri)), 0); + if (!buf) { + ret = -EIO; + goto out; + } + add_root_to_pending(buf, &extent_cache, &pending, + &seen, &nodes, &rec->found_key); + while (1) { + ret = run_next_block(trans, root, bits, bits_nr, &last, + &pending, &seen, &reada, + &nodes, &extent_cache, + &chunk_cache, &dev_cache, + &block_group_cache, + &dev_extent_cache, + &rec->ri); + if (ret != 0) + break; + } + free_extent_buffer(buf); + list_del(&rec->list); + free(rec); + } + + if (ret >= 0) + ret = check_extent_refs(trans, root, &extent_cache); if (ret == -EAGAIN) { ret = btrfs_commit_transaction(trans, root); if (ret) @@ -4612,31 +6463,117 @@ again: goto out; } - free_corrupt_blocks(root->fs_info); - free_cache_tree(&seen); - free_cache_tree(&pending); - free_cache_tree(&reada); - free_cache_tree(&nodes); - free_extent_cache(root->fs_info, &extent_cache); + free_corrupt_blocks_tree(root->fs_info->corrupt_blocks); + free_extent_cache_tree(&seen); + free_extent_cache_tree(&pending); + free_extent_cache_tree(&reada); + free_extent_cache_tree(&nodes); + free_chunk_cache_tree(&chunk_cache); + free_block_group_tree(&block_group_cache); + free_device_cache_tree(&dev_cache); + free_device_extent_tree(&dev_extent_cache); + free_extent_record_cache(root->fs_info, &extent_cache); goto again; } - if (trans) { - int err; + err = check_chunks(&chunk_cache, &block_group_cache, + &dev_extent_cache, NULL, NULL, 0); + if (err && !ret) + ret = err; + + err = check_devices(&dev_cache, &dev_extent_cache); + if (err && !ret) + ret = err; +out: + if (trans) { err = btrfs_commit_transaction(trans, root); if (!ret) ret = err; } -out: if (repair) { - free_corrupt_blocks(root->fs_info); + free_corrupt_blocks_tree(root->fs_info->corrupt_blocks); root->fs_info->fsck_extent_cache = NULL; root->fs_info->free_extent_hook = NULL; root->fs_info->corrupt_blocks = NULL; } - free(bits); - return ret; + free(bits); + free_chunk_cache_tree(&chunk_cache); + free_device_cache_tree(&dev_cache); + free_block_group_tree(&block_group_cache); + free_device_extent_tree(&dev_extent_cache); + free_extent_cache_tree(&seen); + free_extent_cache_tree(&pending); + free_extent_cache_tree(&reada); + free_extent_cache_tree(&nodes); + return ret; +} + +static int btrfs_fsck_reinit_root(struct btrfs_trans_handle *trans, + struct btrfs_root *root, int overwrite) +{ + struct extent_buffer *c; + struct extent_buffer *old = root->node; + int level; + int ret; + struct btrfs_disk_key disk_key = {0,0,0}; + + level = 0; + + if (overwrite) { + c = old; + extent_buffer_get(c); + goto init; + } + c = btrfs_alloc_free_block(trans, root, + btrfs_level_size(root, 0), + root->root_key.objectid, + &disk_key, level, 0, 0); + if (IS_ERR(c)) { + c = old; + extent_buffer_get(c); + overwrite = 1; + } +init: + memset_extent_buffer(c, 0, 0, sizeof(struct btrfs_header)); + btrfs_set_header_level(c, level); + btrfs_set_header_bytenr(c, c->start); + btrfs_set_header_generation(c, trans->transid); + btrfs_set_header_backref_rev(c, BTRFS_MIXED_BACKREF_REV); + btrfs_set_header_owner(c, root->root_key.objectid); + + write_extent_buffer(c, root->fs_info->fsid, + btrfs_header_fsid(), BTRFS_FSID_SIZE); + + write_extent_buffer(c, root->fs_info->chunk_tree_uuid, + btrfs_header_chunk_tree_uuid(c), + BTRFS_UUID_SIZE); + + btrfs_mark_buffer_dirty(c); + /* + * this case can happen in the following case: + * + * 1.overwrite previous root. + * + * 2.reinit reloc data root, this is because we skip pin + * down reloc data tree before which means we can allocate + * same block bytenr here. + */ + if (old->start == c->start) { + btrfs_set_root_generation(&root->root_item, + trans->transid); + root->root_item.level = btrfs_header_level(root->node); + ret = btrfs_update_root(trans, root->fs_info->tree_root, + &root->root_key, &root->root_item); + if (ret) { + free_extent_buffer(c); + return ret; + } + } + free_extent_buffer(old); + root->node = c; + add_root_to_dirty_list(root); + return 0; } static int pin_down_tree_blocks(struct btrfs_fs_info *fs_info, @@ -4652,6 +6589,15 @@ static int pin_down_tree_blocks(struct btrfs_fs_info *fs_info, int ret; int i; + /* + * If we have pinned this block before, don't pin it again. + * This can not only avoid forever loop with broken filesystem + * but also give us some speedups. + */ + if (test_range_bit(&fs_info->pinned_extents, eb->start, + eb->start + eb->len - 1, EXTENT_DIRTY, 0)) + return 0; + btrfs_pin_extent(fs_info, eb->start, eb->len); leafsize = btrfs_super_leafsize(fs_info->super_copy); @@ -4723,11 +6669,13 @@ static int pin_metadata_blocks(struct btrfs_fs_info *fs_info) static int reset_block_groups(struct btrfs_fs_info *fs_info) { + struct btrfs_block_group_cache *cache; struct btrfs_path *path; struct extent_buffer *leaf; struct btrfs_chunk *chunk; struct btrfs_key key; int ret; + u64 start; path = btrfs_alloc_path(); if (!path) @@ -4778,8 +6726,19 @@ static int reset_block_groups(struct btrfs_fs_info *fs_info) btrfs_chunk_type(leaf, chunk), key.objectid, key.offset, btrfs_chunk_length(leaf, chunk)); + set_extent_dirty(&fs_info->free_space_cache, key.offset, + key.offset + btrfs_chunk_length(leaf, chunk), + GFP_NOFS); path->slots[0]++; } + start = 0; + while (1) { + cache = btrfs_lookup_first_block_group(fs_info, start); + if (!cache) + break; + cache->cached = 1; + start = cache->key.objectid + cache->key.offset; + } btrfs_free_path(path); return 0; @@ -4808,13 +6767,16 @@ static int reset_balance(struct btrfs_trans_handle *trans, if (ret) { if (ret > 0) ret = 0; - goto out; + if (!ret) + goto reinit_data_reloc; + else + goto out; } ret = btrfs_del_item(trans, root, path); if (ret) goto out; - btrfs_release_path(root, path); + btrfs_release_path(path); key.objectid = BTRFS_TREE_RELOC_OBJECTID; key.type = BTRFS_ROOT_ITEM_KEY; @@ -4836,7 +6798,7 @@ static int reset_balance(struct btrfs_trans_handle *trans, goto out; } key.offset++; - btrfs_release_path(root, path); + btrfs_release_path(path); found = 0; ret = btrfs_search_slot(trans, root, &key, path, @@ -4868,8 +6830,9 @@ static int reset_balance(struct btrfs_trans_handle *trans, if (ret) goto out; } - btrfs_release_path(root, path); + btrfs_release_path(path); +reinit_data_reloc: key.objectid = BTRFS_DATA_RELOC_TREE_OBJECTID; key.type = BTRFS_ROOT_ITEM_KEY; key.offset = (u64)-1; @@ -4878,21 +6841,19 @@ static int reset_balance(struct btrfs_trans_handle *trans, fprintf(stderr, "Error reading data reloc tree\n"); return PTR_ERR(root); } - root->track_dirty = 1; - if (root->last_trans != trans->transid) { - root->last_trans = trans->transid; - root->commit_root = root->node; - extent_buffer_get(root->node); - } + record_root_in_trans(trans, root); ret = btrfs_fsck_reinit_root(trans, root, 0); + if (ret) + goto out; + ret = btrfs_make_root_dir(trans, root, BTRFS_FIRST_FREE_OBJECTID); out: btrfs_free_path(path); return ret; } -static int reinit_extent_tree(struct btrfs_fs_info *fs_info) +static int reinit_extent_tree(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info) { - struct btrfs_trans_handle *trans; u64 start = 0; int ret; @@ -4911,12 +6872,6 @@ static int reinit_extent_tree(struct btrfs_fs_info *fs_info) return -EINVAL; } - trans = btrfs_start_transaction(fs_info->extent_root, 1); - if (IS_ERR(trans)) { - fprintf(stderr, "Error starting transaction\n"); - return PTR_ERR(trans); - } - /* * first we need to walk all of the trees except the extent tree and pin * down the bytes that are in use so we don't overwrite any existing @@ -4940,7 +6895,7 @@ static int reinit_extent_tree(struct btrfs_fs_info *fs_info) } /* Ok we can allocate now, reinit the extent root */ - ret = btrfs_fsck_reinit_root(trans, fs_info->extent_root, 1); + ret = btrfs_fsck_reinit_root(trans, fs_info->extent_root, 0); if (ret) { fprintf(stderr, "extent root initialization failed\n"); /* @@ -4951,12 +6906,6 @@ static int reinit_extent_tree(struct btrfs_fs_info *fs_info) return ret; } - ret = reset_balance(trans, fs_info); - if (ret) { - fprintf(stderr, "error reseting the pending balance\n"); - return ret; - } - /* * Now we have all the in-memory block groups setup so we can make * allocations properly, and the metadata we care about is safe since we @@ -4979,11 +6928,203 @@ static int reinit_extent_tree(struct btrfs_fs_info *fs_info) btrfs_extent_post_op(trans, fs_info->extent_root); } - /* - * Ok now we commit and run the normal fsck, which will add extent - * entries for all of the items it finds. - */ - return btrfs_commit_transaction(trans, fs_info->extent_root); + ret = reset_balance(trans, fs_info); + if (ret) + fprintf(stderr, "error reseting the pending balance\n"); + + return ret; +} + +static int recow_extent_buffer(struct btrfs_root *root, struct extent_buffer *eb) +{ + struct btrfs_path *path; + struct btrfs_trans_handle *trans; + struct btrfs_key key; + int ret; + + printf("Recowing metadata block %llu\n", eb->start); + key.objectid = btrfs_header_owner(eb); + key.type = BTRFS_ROOT_ITEM_KEY; + key.offset = (u64)-1; + + root = btrfs_read_fs_root(root->fs_info, &key); + if (IS_ERR(root)) { + fprintf(stderr, "Couldn't find owner root %llu\n", + key.objectid); + return PTR_ERR(root); + } + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + + trans = btrfs_start_transaction(root, 1); + if (IS_ERR(trans)) { + btrfs_free_path(path); + return PTR_ERR(trans); + } + + path->lowest_level = btrfs_header_level(eb); + if (path->lowest_level) + btrfs_node_key_to_cpu(eb, &key, 0); + else + btrfs_item_key_to_cpu(eb, &key, 0); + + ret = btrfs_search_slot(trans, root, &key, path, 0, 1); + btrfs_commit_transaction(trans, root); + btrfs_free_path(path); + return ret; +} + +static int delete_bad_item(struct btrfs_root *root, struct bad_item *bad) +{ + struct btrfs_path *path; + struct btrfs_trans_handle *trans; + struct btrfs_key key; + int ret; + + printf("Deleting bad item [%llu,%u,%llu]\n", bad->key.objectid, + bad->key.type, bad->key.offset); + key.objectid = bad->root_id; + key.type = BTRFS_ROOT_ITEM_KEY; + key.offset = (u64)-1; + + root = btrfs_read_fs_root(root->fs_info, &key); + if (IS_ERR(root)) { + fprintf(stderr, "Couldn't find owner root %llu\n", + key.objectid); + return PTR_ERR(root); + } + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + + trans = btrfs_start_transaction(root, 1); + if (IS_ERR(trans)) { + btrfs_free_path(path); + return PTR_ERR(trans); + } + + ret = btrfs_search_slot(trans, root, &bad->key, path, -1, 1); + if (ret) { + if (ret > 0) + ret = 0; + goto out; + } + ret = btrfs_del_item(trans, root, path); +out: + btrfs_commit_transaction(trans, root); + btrfs_free_path(path); + return ret; +} + +static int zero_log_tree(struct btrfs_root *root) +{ + struct btrfs_trans_handle *trans; + int ret; + + trans = btrfs_start_transaction(root, 1); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); + return ret; + } + btrfs_set_super_log_root(root->fs_info->super_copy, 0); + btrfs_set_super_log_root_level(root->fs_info->super_copy, 0); + ret = btrfs_commit_transaction(trans, root); + return ret; +} + +static int populate_csum(struct btrfs_trans_handle *trans, + struct btrfs_root *csum_root, char *buf, u64 start, + u64 len) +{ + u64 offset = 0; + u64 sectorsize; + int ret = 0; + + while (offset < len) { + sectorsize = csum_root->sectorsize; + ret = read_extent_data(csum_root, buf, start + offset, + §orsize, 0); + if (ret) + break; + ret = btrfs_csum_file_block(trans, csum_root, start + len, + start + offset, buf, sectorsize); + if (ret) + break; + offset += sectorsize; + } + return ret; +} + +static int fill_csum_tree(struct btrfs_trans_handle *trans, + struct btrfs_root *csum_root) +{ + struct btrfs_root *extent_root = csum_root->fs_info->extent_root; + struct btrfs_path *path; + struct btrfs_extent_item *ei; + struct extent_buffer *leaf; + char *buf; + struct btrfs_key key; + int ret; + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + + key.objectid = 0; + key.type = BTRFS_EXTENT_ITEM_KEY; + key.offset = 0; + + ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0); + if (ret < 0) { + btrfs_free_path(path); + return ret; + } + + buf = malloc(csum_root->sectorsize); + if (!buf) { + btrfs_free_path(path); + return -ENOMEM; + } + + while (1) { + if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) { + ret = btrfs_next_leaf(extent_root, path); + if (ret < 0) + break; + if (ret) { + ret = 0; + break; + } + } + leaf = path->nodes[0]; + + btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); + if (key.type != BTRFS_EXTENT_ITEM_KEY) { + path->slots[0]++; + continue; + } + + ei = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_extent_item); + if (!(btrfs_extent_flags(leaf, ei) & + BTRFS_EXTENT_FLAG_DATA)) { + path->slots[0]++; + continue; + } + + ret = populate_csum(trans, csum_root, buf, key.objectid, + key.offset); + if (ret) + break; + path->slots[0]++; + } + + btrfs_free_path(path); + free(buf); + return ret; } static struct option long_options[] = { @@ -4991,7 +7132,11 @@ static struct option long_options[] = { { "repair", 0, NULL, 0 }, { "init-csum-tree", 0, NULL, 0 }, { "init-extent-tree", 0, NULL, 0 }, - { 0, 0, 0, 0} + { "check-data-csum", 0, NULL, 0 }, + { "backup", 0, NULL, 0 }, + { "subvol-extents", no_argument, NULL, 'E' }, + { "qgroup-report", 0, NULL, 'Q' }, + { NULL, 0, NULL, 0} }; const char * const cmd_check_usage[] = { @@ -4999,9 +7144,13 @@ const char * const cmd_check_usage[] = { "Check an unmounted btrfs filesystem.", "", "-s|--super use this superblock copy", + "-b|--backup use the backup root copy", "--repair try to repair the filesystem", "--init-csum-tree create a new CRC tree", "--init-extent-tree create a new extent tree", + "--check-data-csum verify checkums of data blocks", + "--qgroup-report print a report on qgroup consistency", + "--subvol-extents print subvolume extents and sharing state", NULL }; @@ -5011,29 +7160,44 @@ int cmd_check(int argc, char **argv) struct btrfs_root *root; struct btrfs_fs_info *info; u64 bytenr = 0; - char uuidbuf[37]; + u64 subvolid = 0; + char uuidbuf[BTRFS_UUID_UNPARSED_SIZE]; int ret; - int num; - int repair = 0; + u64 num; int option_index = 0; int init_csum_tree = 0; - int init_extent_tree = 0; - int rw = 0; + int qgroup_report = 0; + enum btrfs_open_ctree_flags ctree_flags = OPEN_CTREE_EXCLUSIVE; while(1) { int c; - c = getopt_long(argc, argv, "as:", long_options, + c = getopt_long(argc, argv, "as:b", long_options, &option_index); if (c < 0) break; switch(c) { case 'a': /* ignored */ break; + case 'b': + ctree_flags |= OPEN_CTREE_BACKUP_ROOT; + break; case 's': - num = atol(optarg); - bytenr = btrfs_sb_offset(num); - printf("using SB copy %d, bytenr %llu\n", num, + num = arg_strtou64(optarg); + if (num >= BTRFS_SUPER_MIRROR_MAX) { + fprintf(stderr, + "ERROR: super mirror should be less than: %d\n", + BTRFS_SUPER_MIRROR_MAX); + exit(1); + } + bytenr = btrfs_sb_offset(((int)num)); + printf("using SB copy %llu, bytenr %llu\n", num, (unsigned long long)bytenr); break; + case 'Q': + qgroup_report = 1; + break; + case 'E': + subvolid = arg_strtou64(optarg); + break; case '?': case 'h': usage(cmd_check_usage); @@ -5041,21 +7205,24 @@ int cmd_check(int argc, char **argv) if (option_index == 1) { printf("enabling repair mode\n"); repair = 1; - rw = 1; + ctree_flags |= OPEN_CTREE_WRITES; } else if (option_index == 2) { printf("Creating a new CRC tree\n"); init_csum_tree = 1; - rw = 1; + repair = 1; + ctree_flags |= OPEN_CTREE_WRITES; } else if (option_index == 3) { init_extent_tree = 1; - rw = 1; + ctree_flags |= (OPEN_CTREE_WRITES | + OPEN_CTREE_NO_BLOCK_GROUPS); repair = 1; + } else if (option_index == 4) { + check_data_csum = 1; } - } argc = argc - optind; - if (argc != 1) + if (check_argc_exact(argc, 1)) usage(cmd_check_usage); radix_tree_init(); @@ -5063,68 +7230,136 @@ int cmd_check(int argc, char **argv) if((ret = check_mounted(argv[optind])) < 0) { fprintf(stderr, "Could not check mount status: %s\n", strerror(-ret)); - return ret; + goto err_out; } else if(ret) { fprintf(stderr, "%s is currently mounted. Aborting.\n", argv[optind]); - return -EBUSY; + ret = -EBUSY; + goto err_out; } - info = open_ctree_fs_info(argv[optind], bytenr, 0, rw, 1); + /* only allow partial opening under repair mode */ + if (repair) + ctree_flags |= OPEN_CTREE_PARTIAL; + + info = open_ctree_fs_info(argv[optind], bytenr, 0, ctree_flags); if (!info) { fprintf(stderr, "Couldn't open file system\n"); - return -EIO; + ret = -EIO; + goto err_out; + } + + root = info->fs_root; + /* + * repair mode will force us to commit transaction which + * will make us fail to load log tree when mounting. + */ + if (repair && btrfs_super_log_root(info->super_copy)) { + ret = ask_user("repair mode will force to clear out log tree, Are you sure?"); + if (!ret) { + ret = 1; + goto close_out; + } + ret = zero_log_tree(root); + if (ret) { + fprintf(stderr, "fail to zero log tree\n"); + goto close_out; + } } uuid_unparse(info->super_copy->fsid, uuidbuf); + if (qgroup_report) { + printf("Print quota groups for %s\nUUID: %s\n", argv[optind], + uuidbuf); + ret = qgroup_verify_all(info); + if (ret == 0) + print_qgroup_report(1); + goto close_out; + } + if (subvolid) { + printf("Print extent state for subvolume %llu on %s\nUUID: %s\n", + subvolid, argv[optind], uuidbuf); + ret = print_extent_state(info, subvolid); + goto close_out; + } printf("Checking filesystem on %s\nUUID: %s\n", argv[optind], uuidbuf); if (!extent_buffer_uptodate(info->tree_root->node) || !extent_buffer_uptodate(info->dev_root->node) || - !extent_buffer_uptodate(info->extent_root->node) || !extent_buffer_uptodate(info->chunk_root->node)) { fprintf(stderr, "Critical roots corrupted, unable to fsck the FS\n"); - return -EIO; + ret = -EIO; + goto close_out; } - root = info->fs_root; - - if (init_extent_tree) { - printf("Creating a new extent tree\n"); - ret = reinit_extent_tree(info); - if (ret) - return ret; - } - fprintf(stderr, "checking extents\n"); - if (init_csum_tree) { + if (init_extent_tree || init_csum_tree) { struct btrfs_trans_handle *trans; - fprintf(stderr, "Reinit crc root\n"); - trans = btrfs_start_transaction(info->csum_root, 1); + trans = btrfs_start_transaction(info->extent_root, 0); if (IS_ERR(trans)) { fprintf(stderr, "Error starting transaction\n"); - return PTR_ERR(trans); + ret = PTR_ERR(trans); + goto close_out; } - ret = btrfs_fsck_reinit_root(trans, info->csum_root, 0); - if (ret) { - fprintf(stderr, "crc root initialization failed\n"); - return -EIO; + if (init_extent_tree) { + printf("Creating a new extent tree\n"); + ret = reinit_extent_tree(trans, info); + if (ret) + goto close_out; } - ret = btrfs_commit_transaction(trans, root); + if (init_csum_tree) { + fprintf(stderr, "Reinit crc root\n"); + ret = btrfs_fsck_reinit_root(trans, info->csum_root, 0); + if (ret) { + fprintf(stderr, "crc root initialization failed\n"); + ret = -EIO; + goto close_out; + } + + ret = fill_csum_tree(trans, info->csum_root); + if (ret) { + fprintf(stderr, "crc refilling failed\n"); + return -EIO; + } + } + /* + * Ok now we commit and run the normal fsck, which will add + * extent entries for all of the items it finds. + */ + ret = btrfs_commit_transaction(trans, info->extent_root); if (ret) - exit(1); - goto out; + goto close_out; + } + if (!extent_buffer_uptodate(info->extent_root->node)) { + fprintf(stderr, "Critical roots corrupted, unable to fsck the FS\n"); + ret = -EIO; + goto close_out; + } + if (!extent_buffer_uptodate(info->csum_root->node)) { + fprintf(stderr, "Checksum root corrupted, rerun with --init-csum-tree option\n"); + ret = -EIO; + goto close_out; } - ret = check_extents(root, repair); + + fprintf(stderr, "checking extents\n"); + ret = check_chunks_and_extents(root); if (ret) - fprintf(stderr, "Errors found in extent allocation tree\n"); + fprintf(stderr, "Errors found in extent allocation tree or chunk allocation\n"); fprintf(stderr, "checking free space cache\n"); ret = check_space_cache(root); if (ret) goto out; + /* + * We used to have to have these hole extents in between our real + * extents so if we don't have this flag set we need to make sure there + * are no gaps in the file extents for inodes, otherwise we can just + * ignore it when this happens. + */ + no_holes = btrfs_fs_incompat(root->fs_info, + BTRFS_FEATURE_INCOMPAT_NO_HOLES); fprintf(stderr, "checking fs roots\n"); ret = check_fs_roots(root, &root_cache); if (ret) @@ -5137,10 +7372,44 @@ int cmd_check(int argc, char **argv) fprintf(stderr, "checking root refs\n"); ret = check_root_refs(root, &root_cache); -out: - free_root_recs(&root_cache); - close_ctree(root); + if (ret) + goto out; + + while (repair && !list_empty(&root->fs_info->recow_ebs)) { + struct extent_buffer *eb; + + eb = list_first_entry(&root->fs_info->recow_ebs, + struct extent_buffer, recow); + list_del_init(&eb->recow); + ret = recow_extent_buffer(root, eb); + if (ret) + break; + } + + while (!list_empty(&delete_items)) { + struct bad_item *bad; + + bad = list_first_entry(&delete_items, struct bad_item, list); + list_del_init(&bad->list); + if (repair) + ret = delete_bad_item(root, bad); + free(bad); + } + + if (info->quota_enabled) { + int err; + fprintf(stderr, "checking quota groups\n"); + err = qgroup_verify_all(info); + if (err) + goto out; + } + if (!list_empty(&root->fs_info->recow_ebs)) { + fprintf(stderr, "Transid errors in file system\n"); + ret = 1; + } +out: + print_qgroup_report(0); if (found_old_backref) { /* * there was a disk format change when mixed * backref was in testing tree. The old format @@ -5167,5 +7436,10 @@ out: (unsigned long long)data_bytes_allocated, (unsigned long long)data_bytes_referenced); printf("%s\n", BTRFS_BUILD_VERSION); + + free_root_recs_tree(&root_cache); +close_out: + close_ctree(root); +err_out: return ret; }