Now that we can verify all qgroups, we can write the corrected qgroups out to disk when '--repair' is specified. The qgroup status item is also updated to clear any out-of-date state. The repair_ functions were modeled after the inode repair code in cmds-check.c.
I also renamed the 'scan' member of qgroup_status_item to 'rescan' in order to keep consistency with the kernel. Testing this was easy, I just reproduced qgroup inconsistencies via the usual routes and had btrfsck fix them. Signed-off-by: Mark Fasheh <[email protected]> --- cmds-check.c | 15 +++-- ctree.h | 10 ++-- print-tree.c | 2 +- qgroup-verify.c | 178 ++++++++++++++++++++++++++++++++++++++++++++++++++++---- qgroup-verify.h | 3 +- repair.c | 2 + repair.h | 2 + 7 files changed, 189 insertions(+), 23 deletions(-) diff --git a/cmds-check.c b/cmds-check.c index 7b65f89..b7f4bd5 100644 --- a/cmds-check.c +++ b/cmds-check.c @@ -67,7 +67,6 @@ 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; @@ -9543,6 +9542,7 @@ int cmd_check(int argc, char **argv) int init_csum_tree = 0; int readonly = 0; int qgroup_report = 0; + int qgroups_repaired = 0; enum btrfs_open_ctree_flags ctree_flags = OPEN_CTREE_EXCLUSIVE; while(1) { @@ -9698,7 +9698,7 @@ int cmd_check(int argc, char **argv) uuidbuf); ret = qgroup_verify_all(info); if (ret == 0) - ret = report_qgroups(1); + report_qgroups(1); goto close_out; } if (subvolid) { @@ -9852,6 +9852,10 @@ int cmd_check(int argc, char **argv) err = qgroup_verify_all(info); if (err) goto out; + report_qgroups(0); + err = repair_qgroups(info, &qgroups_repaired); + if (err) + goto out; } if (!list_empty(&root->fs_info->recow_ebs)) { @@ -9860,10 +9864,9 @@ int cmd_check(int argc, char **argv) } out: /* Don't override original ret */ - if (ret) - report_qgroups(0); - else - ret = report_qgroups(0); + if (!ret && qgroups_repaired) + ret = qgroups_repaired; + if (found_old_backref) { /* * there was a disk format change when mixed * backref was in testing tree. The old format diff --git a/ctree.h b/ctree.h index 86227f8..34c6b73 100644 --- a/ctree.h +++ b/ctree.h @@ -897,7 +897,7 @@ struct btrfs_qgroup_status_item { __le64 version; __le64 generation; __le64 flags; - __le64 scan; /* progress during scanning */ + __le64 rescan; /* progress during scanning */ } __attribute__ ((__packed__)); struct btrfs_block_group_item { @@ -2130,8 +2130,8 @@ BTRFS_SETGET_FUNCS(qgroup_status_generation, struct btrfs_qgroup_status_item, generation, 64); BTRFS_SETGET_FUNCS(qgroup_status_flags, struct btrfs_qgroup_status_item, flags, 64); -BTRFS_SETGET_FUNCS(qgroup_status_scan, struct btrfs_qgroup_status_item, - scan, 64); +BTRFS_SETGET_FUNCS(qgroup_status_rescan, struct btrfs_qgroup_status_item, + rescan, 64); BTRFS_SETGET_STACK_FUNCS(stack_qgroup_status_version, struct btrfs_qgroup_status_item, version, 64); @@ -2139,8 +2139,8 @@ BTRFS_SETGET_STACK_FUNCS(stack_qgroup_status_generation, struct btrfs_qgroup_status_item, generation, 64); BTRFS_SETGET_STACK_FUNCS(stack_qgroup_status_flags, struct btrfs_qgroup_status_item, flags, 64); -BTRFS_SETGET_STACK_FUNCS(stack_qgroup_status_scan, - struct btrfs_qgroup_status_item, scan, 64); +BTRFS_SETGET_STACK_FUNCS(stack_qgroup_status_rescan, + struct btrfs_qgroup_status_item, rescan, 64); /* btrfs_qgroup_info_item */ BTRFS_SETGET_FUNCS(qgroup_info_generation, struct btrfs_qgroup_info_item, diff --git a/print-tree.c b/print-tree.c index 746f25b..9f9e11e 100644 --- a/print-tree.c +++ b/print-tree.c @@ -1037,7 +1037,7 @@ void btrfs_print_leaf(struct btrfs_root *root, struct extent_buffer *l) btrfs_qgroup_status_generation(l, qg_status), flags_str, (unsigned long long) - btrfs_qgroup_status_scan(l, qg_status)); + btrfs_qgroup_status_rescan(l, qg_status)); break; case BTRFS_QGROUP_RELATION_KEY: break; diff --git a/qgroup-verify.c b/qgroup-verify.c index 23f2961..12921ad 100644 --- a/qgroup-verify.c +++ b/qgroup-verify.c @@ -29,6 +29,8 @@ #include "utils.h" #include "ulist.h" #include "rbtree-utils.h" +#include "repair.h" +#include "transaction.h" #include "qgroup-verify.h" @@ -65,6 +67,8 @@ struct qgroup_count { struct list_head members; u64 cur_refcnt; + + struct list_head bad_list; }; static struct counts_tree { @@ -74,6 +78,8 @@ static struct counts_tree { unsigned int qgroup_inconsist:1; } counts = { .root = RB_ROOT }; +static LIST_HEAD(bad_qgroups); + static struct rb_root by_bytenr = RB_ROOT; /* @@ -817,6 +823,7 @@ static struct qgroup_count *alloc_count(struct btrfs_disk_key *key, btrfs_qgroup_info_exclusive_compressed(leaf, disk); INIT_LIST_HEAD(&c->groups); INIT_LIST_HEAD(&c->members); + INIT_LIST_HEAD(&c->bad_list); if (insert_count(c)) { free(c); @@ -1243,34 +1250,36 @@ static int report_qgroup_difference(struct qgroup_count *count, int verbose) print_fields_signed(excl_diff, excl_diff, "diff:", "exclusive"); } - return (is_different && count->subvol_exists); + + return is_different; } -int report_qgroups(int all) +void report_qgroups(int all) { struct rb_node *node; struct qgroup_count *c; - int ret = 0; - if (counts.rescan_running) { + if (!repair && counts.rescan_running) { if (all) { printf( - "Qgroup rescan is running, qgroup counts difference is expected\n"); + "Qgroup rescan is running, a difference in qgroup counts is expected\n"); } else { printf( - "Qgroup rescan is running, ignore qgroup check\n"); - return ret; + "Qgroup rescan is running, qgroups will not be printed.\n"); + return; } } if (counts.qgroup_inconsist && !counts.rescan_running) - fprintf(stderr, "Qgroup is already inconsistent before checking\n"); + fprintf(stderr, "Qgroup are marked as inconsistent.\n"); node = rb_first(&counts.root); while (node) { c = rb_entry(node, struct qgroup_count, rb_node); - ret |= report_qgroup_difference(c, all); + + if (report_qgroup_difference(c, all)) + list_add_tail(&c->bad_list, &bad_qgroups); + node = rb_next(node); } - return ret; } void free_qgroup_counts(void) @@ -1283,6 +1292,8 @@ void free_qgroup_counts(void) while (node) { c = rb_entry(node, struct qgroup_count, rb_node); + list_del(&c->bad_list); + list_for_each_entry_safe(glist, tmpglist, &c->groups, next_group) { list_del(&glist->next_group); @@ -1418,3 +1429,150 @@ out: return ret; } +static int repair_qgroup_info(struct btrfs_fs_info *info, + struct qgroup_count *count) +{ + int ret; + struct btrfs_root *root = info->quota_root; + struct btrfs_trans_handle *trans; + struct btrfs_path *path; + struct btrfs_qgroup_info_item *info_item; + struct btrfs_key key; + + printf("Repair qgroup %llu/%llu\n", btrfs_qgroup_level(count->qgroupid), + btrfs_qgroup_subvid(count->qgroupid)); + + 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); + } + + key.objectid = 0; + key.type = BTRFS_QGROUP_INFO_KEY; + key.offset = count->qgroupid; + ret = btrfs_search_slot(trans, root, &key, path, 0, 1); + if (ret) { + error("Could not find disk item for qgroup %llu/%llu.\n", + btrfs_qgroup_level(count->qgroupid), + btrfs_qgroup_subvid(count->qgroupid)); + if (ret > 0) + ret = -ENOENT; + goto out; + } + + info_item = btrfs_item_ptr(path->nodes[0], path->slots[0], + struct btrfs_qgroup_info_item); + + btrfs_set_qgroup_info_generation(path->nodes[0], info_item, + trans->transid); + + btrfs_set_qgroup_info_referenced(path->nodes[0], info_item, + count->info.referenced); + btrfs_set_qgroup_info_referenced_compressed(path->nodes[0], info_item, + count->info.referenced_compressed); + + btrfs_set_qgroup_info_exclusive(path->nodes[0], info_item, + count->info.exclusive); + btrfs_set_qgroup_info_exclusive_compressed(path->nodes[0], info_item, + count->info.exclusive_compressed); + + btrfs_mark_buffer_dirty(path->nodes[0]); + +out: + btrfs_commit_transaction(trans, root); + btrfs_free_path(path); + + return ret; +} + +static int repair_qgroup_status(struct btrfs_fs_info *info) +{ + int ret; + struct btrfs_root *root = info->quota_root; + struct btrfs_trans_handle *trans; + struct btrfs_path *path; + struct btrfs_key key; + struct btrfs_qgroup_status_item *status_item; + + printf("Repair qgroup status item\n"); + + 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); + } + + key.objectid = 0; + key.type = BTRFS_QGROUP_STATUS_KEY; + key.offset = 0; + ret = btrfs_search_slot(trans, root, &key, path, 0, 1); + if (ret) { + error("Could not find qgroup status item\n"); + if (ret > 0) + ret = -ENOENT; + goto out; + } + + status_item = btrfs_item_ptr(path->nodes[0], path->slots[0], + struct btrfs_qgroup_status_item); + btrfs_set_qgroup_status_flags(path->nodes[0], status_item, + BTRFS_QGROUP_STATUS_FLAG_ON); + btrfs_set_qgroup_status_rescan(path->nodes[0], status_item, 0); + btrfs_set_qgroup_status_generation(path->nodes[0], status_item, + trans->transid); + + btrfs_mark_buffer_dirty(path->nodes[0]); + +out: + btrfs_commit_transaction(trans, root); + btrfs_free_path(path); + + return ret; +} + +int repair_qgroups(struct btrfs_fs_info *info, int *repaired) +{ + int ret; + struct qgroup_count *count, *tmpcount; + + *repaired = 0; + + if (!repair) + return 0; + + list_for_each_entry_safe(count, tmpcount, &bad_qgroups, bad_list) { + ret = repair_qgroup_info(info, count); + if (ret) { + goto out; + } + + (*repaired)++; + + list_del_init(&count->bad_list); + } + + /* + * Do this step last as we want the latest transaction id on + * our qgroup status to avoid a (useless) warning after + * mount. + */ + if (*repaired || counts.qgroup_inconsist || counts.rescan_running) { + ret = repair_qgroup_status(info); + if (ret) + goto out; + + (*repaired)++; + } + +out: + return ret; +} diff --git a/qgroup-verify.h b/qgroup-verify.h index 0f8ff9b..d7d83a4 100644 --- a/qgroup-verify.h +++ b/qgroup-verify.h @@ -23,7 +23,8 @@ #include "ctree.h" int qgroup_verify_all(struct btrfs_fs_info *info); -int report_qgroups(int all); +void report_qgroups(int all); +int repair_qgroups(struct btrfs_fs_info *info, int *repaired); int print_extent_state(struct btrfs_fs_info *info, u64 subvol); diff --git a/repair.c b/repair.c index 4f74742..07a1232 100644 --- a/repair.c +++ b/repair.c @@ -21,6 +21,8 @@ #include "utils.h" #include "repair.h" +int repair = 0; + int btrfs_add_corrupt_extent_record(struct btrfs_fs_info *info, struct btrfs_key *first_key, u64 start, u64 len, int level) diff --git a/repair.h b/repair.h index 3fc0e8b..355bbf2 100644 --- a/repair.h +++ b/repair.h @@ -21,6 +21,8 @@ #include "ctree.h" +extern int repair; /* repair mode */ + struct btrfs_corrupt_block { struct cache_extent cache; struct btrfs_key key; -- 2.1.4 -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
