From 102a0743326a03cd1a1202ceda21e175b7d3575c Mon Sep 17 00:00:00 2001 From: hc <hc@nodka.com> Date: Tue, 20 Feb 2024 01:20:52 +0000 Subject: [PATCH] add new system file --- kernel/fs/btrfs/transaction.c | 678 +++++++++++++++++++++++++++++++++---------------------- 1 files changed, 404 insertions(+), 274 deletions(-) diff --git a/kernel/fs/btrfs/transaction.c b/kernel/fs/btrfs/transaction.c index 0495351..d23047b 100644 --- a/kernel/fs/btrfs/transaction.c +++ b/kernel/fs/btrfs/transaction.c @@ -10,6 +10,7 @@ #include <linux/pagemap.h> #include <linux/blkdev.h> #include <linux/uuid.h> +#include "misc.h" #include "ctree.h" #include "disk-io.h" #include "transaction.h" @@ -19,12 +20,84 @@ #include "volumes.h" #include "dev-replace.h" #include "qgroup.h" +#include "block-group.h" +#include "space-info.h" #define BTRFS_ROOT_TRANS_TAG 0 +/* + * Transaction states and transitions + * + * No running transaction (fs tree blocks are not modified) + * | + * | To next stage: + * | Call start_transaction() variants. Except btrfs_join_transaction_nostart(). + * V + * Transaction N [[TRANS_STATE_RUNNING]] + * | + * | New trans handles can be attached to transaction N by calling all + * | start_transaction() variants. + * | + * | To next stage: + * | Call btrfs_commit_transaction() on any trans handle attached to + * | transaction N + * V + * Transaction N [[TRANS_STATE_COMMIT_START]] + * | + * | Will wait for previous running transaction to completely finish if there + * | is one + * | + * | Then one of the following happes: + * | - Wait for all other trans handle holders to release. + * | The btrfs_commit_transaction() caller will do the commit work. + * | - Wait for current transaction to be committed by others. + * | Other btrfs_commit_transaction() caller will do the commit work. + * | + * | At this stage, only btrfs_join_transaction*() variants can attach + * | to this running transaction. + * | All other variants will wait for current one to finish and attach to + * | transaction N+1. + * | + * | To next stage: + * | Caller is chosen to commit transaction N, and all other trans handle + * | haven been released. + * V + * Transaction N [[TRANS_STATE_COMMIT_DOING]] + * | + * | The heavy lifting transaction work is started. + * | From running delayed refs (modifying extent tree) to creating pending + * | snapshots, running qgroups. + * | In short, modify supporting trees to reflect modifications of subvolume + * | trees. + * | + * | At this stage, all start_transaction() calls will wait for this + * | transaction to finish and attach to transaction N+1. + * | + * | To next stage: + * | Until all supporting trees are updated. + * V + * Transaction N [[TRANS_STATE_UNBLOCKED]] + * | Transaction N+1 + * | All needed trees are modified, thus we only [[TRANS_STATE_RUNNING]] + * | need to write them back to disk and update | + * | super blocks. | + * | | + * | At this stage, new transaction is allowed to | + * | start. | + * | All new start_transaction() calls will be | + * | attached to transid N+1. | + * | | + * | To next stage: | + * | Until all tree blocks are super blocks are | + * | written to block devices | + * V | + * Transaction N [[TRANS_STATE_COMPLETED]] V + * All tree blocks and super blocks are written. Transaction N+1 + * This transaction is finished and all its [[TRANS_STATE_COMMIT_START]] + * data structures will be cleaned up. | Life goes on + */ static const unsigned int btrfs_blocked_trans_types[TRANS_STATE_MAX] = { [TRANS_STATE_RUNNING] = 0U, - [TRANS_STATE_BLOCKED] = __TRANS_START, [TRANS_STATE_COMMIT_START] = (__TRANS_START | __TRANS_ATTACH), [TRANS_STATE_COMMIT_DOING] = (__TRANS_START | __TRANS_ATTACH | @@ -47,19 +120,14 @@ WARN_ON(refcount_read(&transaction->use_count) == 0); if (refcount_dec_and_test(&transaction->use_count)) { BUG_ON(!list_empty(&transaction->list)); - WARN_ON(!RB_EMPTY_ROOT(&transaction->delayed_refs.href_root)); + WARN_ON(!RB_EMPTY_ROOT( + &transaction->delayed_refs.href_root.rb_root)); + WARN_ON(!RB_EMPTY_ROOT( + &transaction->delayed_refs.dirty_extent_root)); if (transaction->delayed_refs.pending_csums) btrfs_err(transaction->fs_info, "pending csums is %llu", transaction->delayed_refs.pending_csums); - while (!list_empty(&transaction->pending_chunks)) { - struct extent_map *em; - - em = list_first_entry(&transaction->pending_chunks, - struct extent_map, list); - list_del_init(&em->list); - free_extent_map(em); - } /* * If any block groups are found in ->deleted_bgs then it's * because the transaction was aborted and a commit did not @@ -68,75 +136,90 @@ * discard the physical locations of the block groups. */ while (!list_empty(&transaction->deleted_bgs)) { - struct btrfs_block_group_cache *cache; + struct btrfs_block_group *cache; cache = list_first_entry(&transaction->deleted_bgs, - struct btrfs_block_group_cache, + struct btrfs_block_group, bg_list); list_del_init(&cache->bg_list); - btrfs_put_block_group_trimming(cache); + btrfs_unfreeze_block_group(cache); btrfs_put_block_group(cache); } + WARN_ON(!list_empty(&transaction->dev_update_list)); kfree(transaction); } } -static void clear_btree_io_tree(struct extent_io_tree *tree) +static noinline void switch_commit_roots(struct btrfs_trans_handle *trans) { - spin_lock(&tree->lock); - /* - * Do a single barrier for the waitqueue_active check here, the state - * of the waitqueue should not change once clear_btree_io_tree is - * called. - */ - smp_mb(); - while (!RB_EMPTY_ROOT(&tree->state)) { - struct rb_node *node; - struct extent_state *state; - - node = rb_first(&tree->state); - state = rb_entry(node, struct extent_state, rb_node); - rb_erase(&state->rb_node, &tree->state); - RB_CLEAR_NODE(&state->rb_node); - /* - * btree io trees aren't supposed to have tasks waiting for - * changes in the flags of extent states ever. - */ - ASSERT(!waitqueue_active(&state->wq)); - free_extent_state(state); - - cond_resched_lock(&tree->lock); - } - spin_unlock(&tree->lock); -} - -static noinline void switch_commit_roots(struct btrfs_transaction *trans) -{ + struct btrfs_transaction *cur_trans = trans->transaction; struct btrfs_fs_info *fs_info = trans->fs_info; struct btrfs_root *root, *tmp; + struct btrfs_caching_control *caching_ctl, *next; down_write(&fs_info->commit_root_sem); - list_for_each_entry_safe(root, tmp, &trans->switch_commits, + list_for_each_entry_safe(root, tmp, &cur_trans->switch_commits, dirty_list) { list_del_init(&root->dirty_list); free_extent_buffer(root->commit_root); root->commit_root = btrfs_root_node(root); - if (is_fstree(root->objectid)) + if (is_fstree(root->root_key.objectid)) btrfs_unpin_free_ino(root); - clear_btree_io_tree(&root->dirty_log_pages); + extent_io_tree_release(&root->dirty_log_pages); + btrfs_qgroup_clean_swapped_blocks(root); } /* We can free old roots now. */ - spin_lock(&trans->dropped_roots_lock); - while (!list_empty(&trans->dropped_roots)) { - root = list_first_entry(&trans->dropped_roots, + spin_lock(&cur_trans->dropped_roots_lock); + while (!list_empty(&cur_trans->dropped_roots)) { + root = list_first_entry(&cur_trans->dropped_roots, struct btrfs_root, root_list); list_del_init(&root->root_list); - spin_unlock(&trans->dropped_roots_lock); + spin_unlock(&cur_trans->dropped_roots_lock); + btrfs_free_log(trans, root); btrfs_drop_and_free_fs_root(fs_info, root); - spin_lock(&trans->dropped_roots_lock); + spin_lock(&cur_trans->dropped_roots_lock); } - spin_unlock(&trans->dropped_roots_lock); + spin_unlock(&cur_trans->dropped_roots_lock); + + /* + * We have to update the last_byte_to_unpin under the commit_root_sem, + * at the same time we swap out the commit roots. + * + * This is because we must have a real view of the last spot the caching + * kthreads were while caching. Consider the following views of the + * extent tree for a block group + * + * commit root + * +----+----+----+----+----+----+----+ + * |\\\\| |\\\\|\\\\| |\\\\|\\\\| + * +----+----+----+----+----+----+----+ + * 0 1 2 3 4 5 6 7 + * + * new commit root + * +----+----+----+----+----+----+----+ + * | | | |\\\\| | |\\\\| + * +----+----+----+----+----+----+----+ + * 0 1 2 3 4 5 6 7 + * + * If the cache_ctl->progress was at 3, then we are only allowed to + * unpin [0,1) and [2,3], because the caching thread has already + * processed those extents. We are not allowed to unpin [5,6), because + * the caching thread will re-start it's search from 3, and thus find + * the hole from [4,6) to add to the free space cache. + */ + list_for_each_entry_safe(caching_ctl, next, + &fs_info->caching_block_groups, list) { + struct btrfs_block_group *cache = caching_ctl->block_group; + + if (btrfs_block_group_done(cache)) { + cache->last_byte_to_unpin = (u64)-1; + list_del_init(&caching_ctl->list); + btrfs_put_caching_control(caching_ctl); + } else { + cache->last_byte_to_unpin = caching_ctl->progress; + } + } up_write(&fs_info->commit_root_sem); } @@ -166,6 +249,24 @@ } /* + * To be called after all the new block groups attached to the transaction + * handle have been created (btrfs_create_pending_block_groups()). + */ +void btrfs_trans_release_chunk_metadata(struct btrfs_trans_handle *trans) +{ + struct btrfs_fs_info *fs_info = trans->fs_info; + + if (!trans->chunk_bytes_reserved) + return; + + WARN_ON_ONCE(!list_empty(&trans->new_bgs)); + + btrfs_block_rsv_release(fs_info, &fs_info->chunk_block_rsv, + trans->chunk_bytes_reserved, NULL); + trans->chunk_bytes_reserved = 0; +} + +/* * either allocate a new transaction or hop into the existing one */ static noinline int join_transaction(struct btrfs_fs_info *fs_info, @@ -183,7 +284,7 @@ cur_trans = fs_info->running_transaction; if (cur_trans) { - if (cur_trans->aborted) { + if (TRANS_ABORTED(cur_trans)) { spin_unlock(&fs_info->trans_lock); return cur_trans->aborted; } @@ -200,10 +301,11 @@ spin_unlock(&fs_info->trans_lock); /* - * If we are ATTACH, we just want to catch the current transaction, - * and commit it. If there is no transaction, just return ENOENT. + * If we are ATTACH or TRANS_JOIN_NOSTART, we just want to catch the + * current transaction, and commit it. If there is no transaction, just + * return ENOENT. */ - if (type == TRANS_ATTACH) + if (type == TRANS_ATTACH || type == TRANS_JOIN_NOSTART) return -ENOENT; /* @@ -231,24 +333,24 @@ } cur_trans->fs_info = fs_info; + atomic_set(&cur_trans->pending_ordered, 0); + init_waitqueue_head(&cur_trans->pending_wait); atomic_set(&cur_trans->num_writers, 1); extwriter_counter_init(cur_trans, type); init_waitqueue_head(&cur_trans->writer_wait); init_waitqueue_head(&cur_trans->commit_wait); - init_waitqueue_head(&cur_trans->pending_wait); cur_trans->state = TRANS_STATE_RUNNING; /* * One for this trans handle, one so it will live on until we * commit the transaction. */ refcount_set(&cur_trans->use_count, 2); - atomic_set(&cur_trans->pending_ordered, 0); cur_trans->flags = 0; cur_trans->start_time = ktime_get_seconds(); memset(&cur_trans->delayed_refs, 0, sizeof(cur_trans->delayed_refs)); - cur_trans->delayed_refs.href_root = RB_ROOT; + cur_trans->delayed_refs.href_root = RB_ROOT_CACHED; cur_trans->delayed_refs.dirty_extent_root = RB_ROOT; atomic_set(&cur_trans->delayed_refs.num_entries, 0); @@ -266,19 +368,20 @@ spin_lock_init(&cur_trans->delayed_refs.lock); INIT_LIST_HEAD(&cur_trans->pending_snapshots); - INIT_LIST_HEAD(&cur_trans->pending_chunks); + INIT_LIST_HEAD(&cur_trans->dev_update_list); INIT_LIST_HEAD(&cur_trans->switch_commits); INIT_LIST_HEAD(&cur_trans->dirty_bgs); INIT_LIST_HEAD(&cur_trans->io_bgs); INIT_LIST_HEAD(&cur_trans->dropped_roots); mutex_init(&cur_trans->cache_write_mutex); - cur_trans->num_dirty_bgs = 0; spin_lock_init(&cur_trans->dirty_bgs_lock); INIT_LIST_HEAD(&cur_trans->deleted_bgs); spin_lock_init(&cur_trans->dropped_roots_lock); list_add_tail(&cur_trans->list, &fs_info->trans_list); - extent_io_tree_init(&cur_trans->dirty_pages, - fs_info->btree_inode); + extent_io_tree_init(fs_info, &cur_trans->dirty_pages, + IO_TREE_TRANS_DIRTY_PAGES, fs_info->btree_inode); + extent_io_tree_init(fs_info, &cur_trans->pinned_extents, + IO_TREE_FS_PINNED_EXTENTS, NULL); fs_info->generation++; cur_trans->transid = fs_info->generation; fs_info->running_transaction = cur_trans; @@ -289,10 +392,10 @@ } /* - * this does all the record keeping required to make sure that a reference - * counted root is properly recorded in a given transaction. This is required - * to make sure the old root from before we joined the transaction is deleted - * when the transaction commits + * This does all the record keeping required to make sure that a shareable root + * is properly recorded in a given transaction. This is required to make sure + * the old root from before we joined the transaction is deleted when the + * transaction commits. */ static int record_root_in_trans(struct btrfs_trans_handle *trans, struct btrfs_root *root, @@ -300,7 +403,7 @@ { struct btrfs_fs_info *fs_info = root->fs_info; - if ((test_bit(BTRFS_ROOT_REF_COWS, &root->state) && + if ((test_bit(BTRFS_ROOT_SHAREABLE, &root->state) && root->last_trans < trans->transid) || force) { WARN_ON(root == fs_info->extent_root); WARN_ON(!force && root->commit_root != root->node); @@ -379,7 +482,7 @@ { struct btrfs_fs_info *fs_info = root->fs_info; - if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state)) + if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) return 0; /* @@ -400,9 +503,9 @@ static inline int is_transaction_blocked(struct btrfs_transaction *trans) { - return (trans->state >= TRANS_STATE_BLOCKED && + return (trans->state >= TRANS_STATE_COMMIT_START && trans->state < TRANS_STATE_UNBLOCKED && - !trans->aborted); + !TRANS_ABORTED(trans)); } /* wait for commit against the current transaction to become unblocked @@ -421,7 +524,7 @@ wait_event(fs_info->transaction_wait, cur_trans->state >= TRANS_STATE_UNBLOCKED || - cur_trans->aborted); + TRANS_ABORTED(cur_trans)); btrfs_put_transaction(cur_trans); } else { spin_unlock(&fs_info->trans_lock); @@ -444,7 +547,7 @@ struct btrfs_fs_info *fs_info = root->fs_info; if (!fs_info->reloc_ctl || - !test_bit(BTRFS_ROOT_REF_COWS, &root->state) || + !test_bit(BTRFS_ROOT_SHAREABLE, &root->state) || root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID || root->reloc_root) return false; @@ -458,12 +561,13 @@ bool enforce_qgroups) { struct btrfs_fs_info *fs_info = root->fs_info; - + struct btrfs_block_rsv *delayed_refs_rsv = &fs_info->delayed_refs_rsv; struct btrfs_trans_handle *h; struct btrfs_transaction *cur_trans; u64 num_bytes = 0; u64 qgroup_reserved = 0; bool reloc_reserved = false; + bool do_chunk_alloc = false; int ret; /* Send isn't supposed to start transactions. */ @@ -487,13 +591,29 @@ * the appropriate flushing if need be. */ if (num_items && root != fs_info->chunk_root) { + struct btrfs_block_rsv *rsv = &fs_info->trans_block_rsv; + u64 delayed_refs_bytes = 0; + qgroup_reserved = num_items * fs_info->nodesize; ret = btrfs_qgroup_reserve_meta_pertrans(root, qgroup_reserved, enforce_qgroups); if (ret) return ERR_PTR(ret); - num_bytes = btrfs_calc_trans_metadata_size(fs_info, num_items); + /* + * We want to reserve all the bytes we may need all at once, so + * we only do 1 enospc flushing cycle per transaction start. We + * accomplish this by simply assuming we'll do 2 x num_items + * worth of delayed refs updates in this trans handle, and + * refill that amount for whatever is missing in the reserve. + */ + num_bytes = btrfs_calc_insert_metadata_size(fs_info, num_items); + if (flush == BTRFS_RESERVE_FLUSH_ALL && + delayed_refs_rsv->full == 0) { + delayed_refs_bytes = num_bytes; + num_bytes <<= 1; + } + /* * Do the reservation for the relocation root creation */ @@ -502,8 +622,27 @@ reloc_reserved = true; } - ret = btrfs_block_rsv_add(root, &fs_info->trans_block_rsv, - num_bytes, flush); + ret = btrfs_block_rsv_add(root, rsv, num_bytes, flush); + if (ret) + goto reserve_fail; + if (delayed_refs_bytes) { + btrfs_migrate_to_delayed_refs_rsv(fs_info, rsv, + delayed_refs_bytes); + num_bytes -= delayed_refs_bytes; + } + + if (rsv->space_info->force_alloc) + do_chunk_alloc = true; + } else if (num_items == 0 && flush == BTRFS_RESERVE_FLUSH_ALL && + !delayed_refs_rsv->full) { + /* + * Some people call with btrfs_start_transaction(root, 0) + * because they can be throttled, but have some other mechanism + * for reserving space. We still want these guys to refill the + * delayed block_rsv so just add 1 items worth of reservation + * here. + */ + ret = btrfs_delayed_refs_rsv_refill(fs_info, flush); if (ret) goto reserve_fail; } @@ -556,7 +695,7 @@ INIT_LIST_HEAD(&h->new_bgs); smp_mb(); - if (cur_trans->state >= TRANS_STATE_BLOCKED && + if (cur_trans->state >= TRANS_STATE_COMMIT_START && may_wait_transaction(fs_info, type)) { current->journal_info = h; btrfs_commit_transaction(h); @@ -574,6 +713,19 @@ got_it: if (!current->journal_info) current->journal_info = h; + + /* + * If the space_info is marked ALLOC_FORCE then we'll get upgraded to + * ALLOC_FORCE the first run through, and then we won't allocate for + * anybody else who races in later. We don't care about the return + * value here. + */ + if (do_chunk_alloc && num_bytes) { + u64 flags = h->block_rsv->space_info->flags; + + btrfs_chunk_alloc(h, btrfs_get_alloc_profile(fs_info, flags), + CHUNK_ALLOC_NO_FORCE); + } /* * btrfs_record_root_in_trans() needs to alloc new extents, and may @@ -594,7 +746,7 @@ alloc_fail: if (num_bytes) btrfs_block_rsv_release(fs_info, &fs_info->trans_block_rsv, - num_bytes); + num_bytes, NULL); reserve_fail: btrfs_qgroup_free_meta_pertrans(root, qgroup_reserved); return ERR_PTR(ret); @@ -609,43 +761,10 @@ struct btrfs_trans_handle *btrfs_start_transaction_fallback_global_rsv( struct btrfs_root *root, - unsigned int num_items, - int min_factor) + unsigned int num_items) { - struct btrfs_fs_info *fs_info = root->fs_info; - struct btrfs_trans_handle *trans; - u64 num_bytes; - int ret; - - /* - * We have two callers: unlink and block group removal. The - * former should succeed even if we will temporarily exceed - * quota and the latter operates on the extent root so - * qgroup enforcement is ignored anyway. - */ - trans = start_transaction(root, num_items, TRANS_START, - BTRFS_RESERVE_FLUSH_ALL, false); - if (!IS_ERR(trans) || PTR_ERR(trans) != -ENOSPC) - return trans; - - trans = btrfs_start_transaction(root, 0); - if (IS_ERR(trans)) - return trans; - - num_bytes = btrfs_calc_trans_metadata_size(fs_info, num_items); - ret = btrfs_cond_migrate_bytes(fs_info, &fs_info->trans_block_rsv, - num_bytes, min_factor); - if (ret) { - btrfs_end_transaction(trans); - return ERR_PTR(ret); - } - - trans->block_rsv = &fs_info->trans_block_rsv; - trans->bytes_reserved = num_bytes; - trace_btrfs_space_reservation(fs_info, "transaction", - trans->transid, num_bytes, 1); - - return trans; + return start_transaction(root, num_items, TRANS_START, + BTRFS_RESERVE_FLUSH_ALL_STEAL, false); } struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root) @@ -654,7 +773,7 @@ true); } -struct btrfs_trans_handle *btrfs_join_transaction_nolock(struct btrfs_root *root) +struct btrfs_trans_handle *btrfs_join_transaction_spacecache(struct btrfs_root *root) { return start_transaction(root, 0, TRANS_JOIN_NOLOCK, BTRFS_RESERVE_NO_FLUSH, true); @@ -692,7 +811,7 @@ /* * btrfs_attach_transaction_barrier() - catch the running transaction * - * It is similar to the above function, the differentia is this one + * It is similar to the above function, the difference is this one * will wait for all the inactive transactions until they fully * complete. */ @@ -703,8 +822,13 @@ trans = start_transaction(root, 0, TRANS_ATTACH, BTRFS_RESERVE_NO_FLUSH, true); - if (trans == ERR_PTR(-ENOENT)) - btrfs_wait_for_commit(root->fs_info, 0); + if (trans == ERR_PTR(-ENOENT)) { + int ret; + + ret = btrfs_wait_for_commit(root->fs_info, 0); + if (ret) + return ERR_PTR(ret); + } return trans; } @@ -768,6 +892,7 @@ } wait_for_commit(cur_trans); + ret = cur_trans->aborted; btrfs_put_transaction(cur_trans); out: return ret; @@ -782,7 +907,7 @@ { struct btrfs_fs_info *fs_info = trans->fs_info; - if (btrfs_check_space_for_delayed_refs(trans, fs_info)) + if (btrfs_check_space_for_delayed_refs(fs_info)) return 1; return !!btrfs_block_rsv_check(&fs_info->global_block_rsv, 5); @@ -791,21 +916,11 @@ int btrfs_should_end_transaction(struct btrfs_trans_handle *trans) { struct btrfs_transaction *cur_trans = trans->transaction; - int updates; - int err; smp_mb(); - if (cur_trans->state >= TRANS_STATE_BLOCKED || + if (cur_trans->state >= TRANS_STATE_COMMIT_START || cur_trans->delayed_refs.flushing) return 1; - - updates = trans->delayed_ref_updates; - trans->delayed_ref_updates = 0; - if (updates) { - err = btrfs_run_delayed_refs(trans, updates * 2); - if (err) /* Error code will also eval true */ - return err; - } return should_end_transaction(trans); } @@ -827,7 +942,7 @@ trace_btrfs_space_reservation(fs_info, "transaction", trans->transid, trans->bytes_reserved, 0); btrfs_block_rsv_release(fs_info, trans->block_rsv, - trans->bytes_reserved); + trans->bytes_reserved, NULL); trans->bytes_reserved = 0; } @@ -836,11 +951,7 @@ { struct btrfs_fs_info *info = trans->fs_info; struct btrfs_transaction *cur_trans = trans->transaction; - u64 transid = trans->transid; - unsigned long cur = trans->delayed_ref_updates; - int lock = (trans->type != TRANS_JOIN_NOLOCK); int err = 0; - int must_run_delayed_refs = 0; if (refcount_read(&trans->use_count) > 1) { refcount_dec(&trans->use_count); @@ -851,46 +962,9 @@ btrfs_trans_release_metadata(trans); trans->block_rsv = NULL; - if (!list_empty(&trans->new_bgs)) - btrfs_create_pending_block_groups(trans); - - trans->delayed_ref_updates = 0; - if (!trans->sync) { - must_run_delayed_refs = - btrfs_should_throttle_delayed_refs(trans, info); - cur = max_t(unsigned long, cur, 32); - - /* - * don't make the caller wait if they are from a NOLOCK - * or ATTACH transaction, it will deadlock with commit - */ - if (must_run_delayed_refs == 1 && - (trans->type & (__TRANS_JOIN_NOLOCK | __TRANS_ATTACH))) - must_run_delayed_refs = 2; - } - - btrfs_trans_release_metadata(trans); - trans->block_rsv = NULL; - - if (!list_empty(&trans->new_bgs)) - btrfs_create_pending_block_groups(trans); + btrfs_create_pending_block_groups(trans); btrfs_trans_release_chunk_metadata(trans); - - if (lock && should_end_transaction(trans) && - READ_ONCE(cur_trans->state) == TRANS_STATE_RUNNING) { - spin_lock(&info->trans_lock); - if (cur_trans->state == TRANS_STATE_RUNNING) - cur_trans->state = TRANS_STATE_BLOCKED; - spin_unlock(&info->trans_lock); - } - - if (lock && READ_ONCE(cur_trans->state) == TRANS_STATE_BLOCKED) { - if (throttle) - return btrfs_commit_transaction(trans); - else - wake_up_process(info->transaction_kthread); - } if (trans->type & __TRANS_FREEZABLE) sb_end_intwrite(info->sb); @@ -909,17 +983,16 @@ if (throttle) btrfs_run_delayed_iputs(info); - if (trans->aborted || + if (TRANS_ABORTED(trans) || test_bit(BTRFS_FS_STATE_ERROR, &info->fs_state)) { wake_up_process(info->transaction_kthread); - err = -EIO; + if (TRANS_ABORTED(trans)) + err = trans->aborted; + else + err = -EROFS; } kmem_cache_free(btrfs_trans_handle_cachep, trans); - if (must_run_delayed_refs) { - btrfs_async_run_delayed_refs(info, cur, transid, - must_run_delayed_refs == 1); - } return err; } @@ -967,7 +1040,7 @@ * superblock that points to btree nodes/leafs for which * writeback hasn't finished yet (and without errors). * We cleanup any entries left in the io tree when committing - * the transaction (through clear_btree_io_tree()). + * the transaction (through extent_io_tree_release()). */ if (err == -ENOMEM) { err = 0; @@ -1012,7 +1085,7 @@ * left in the io tree. For a log commit, we don't remove them * after committing the log because the tree can be accessed * concurrently - we do it only at transaction commit time when - * it's safe to do it (through clear_btree_io_tree()). + * it's safe to do it (through extent_io_tree_release()). */ err = clear_extent_bit(dirty_pages, start, end, EXTENT_NEED_WAIT, 0, 0, &cached_state); @@ -1032,7 +1105,7 @@ return werr; } -int btrfs_wait_extents(struct btrfs_fs_info *fs_info, +static int btrfs_wait_extents(struct btrfs_fs_info *fs_info, struct extent_io_tree *dirty_pages) { bool errors = false; @@ -1090,7 +1163,7 @@ blk_finish_plug(&plug); ret2 = btrfs_wait_extents(fs_info, dirty_pages); - clear_btree_io_tree(&trans->transaction->dirty_pages); + extent_io_tree_release(&trans->transaction->dirty_pages); if (ret) return ret; @@ -1158,7 +1231,7 @@ eb = btrfs_lock_root_node(fs_info->tree_root); ret = btrfs_cow_block(trans, fs_info->tree_root, eb, NULL, - 0, &eb); + 0, &eb, BTRFS_NESTING_COW); btrfs_tree_unlock(eb); free_extent_buffer(eb); @@ -1169,17 +1242,17 @@ if (ret) return ret; - ret = btrfs_run_dev_stats(trans, fs_info); + ret = btrfs_run_dev_stats(trans); if (ret) return ret; - ret = btrfs_run_dev_replace(trans, fs_info); + ret = btrfs_run_dev_replace(trans); if (ret) return ret; ret = btrfs_run_qgroups(trans); if (ret) return ret; - ret = btrfs_setup_space_cache(trans, fs_info); + ret = btrfs_setup_space_cache(trans); if (ret) return ret; @@ -1207,7 +1280,7 @@ } while (!list_empty(dirty_bgs) || !list_empty(io_bgs)) { - ret = btrfs_write_dirty_block_groups(trans, fs_info); + ret = btrfs_write_dirty_block_groups(trans); if (ret) return ret; ret = btrfs_run_delayed_refs(trans, (unsigned long)-1); @@ -1220,7 +1293,10 @@ list_add_tail(&fs_info->extent_root->dirty_list, &trans->transaction->switch_commits); - btrfs_after_dev_replace_commit(fs_info); + + /* Update dev-replace pointer once everything is committed */ + fs_info->dev_replace.committed_cursor_left = + fs_info->dev_replace.cursor_left_last_write_of_item; return 0; } @@ -1235,8 +1311,10 @@ struct btrfs_fs_info *fs_info = root->fs_info; spin_lock(&fs_info->trans_lock); - if (list_empty(&root->root_list)) + if (list_empty(&root->root_list)) { + btrfs_grab_root(root); list_add_tail(&root->root_list, &fs_info->dead_roots); + } spin_unlock(&fs_info->trans_lock); } @@ -1360,7 +1438,7 @@ return 0; /* - * Ensure dirty @src will be commited. Or, after comming + * Ensure dirty @src will be committed. Or, after coming * commit_fs_roots() and switch_commit_roots(), any dirty but not * recorded root will never be updated again, causing an outdated root * item. @@ -1401,7 +1479,7 @@ ret = commit_cowonly_roots(trans); if (ret) goto out; - switch_commit_roots(trans->transaction); + switch_commit_roots(trans); ret = btrfs_write_and_wait_transaction(trans); if (ret) btrfs_handle_fs_error(fs_info, ret, @@ -1453,7 +1531,6 @@ u64 index = 0; u64 objectid; u64 root_flags; - uuid_le new_uuid; ASSERT(pending->path); path = pending->path; @@ -1546,8 +1623,7 @@ btrfs_set_root_generation_v2(new_root_item, trans->transid); - uuid_le_gen(&new_uuid); - memcpy(new_root_item->uuid, new_uuid.b, BTRFS_UUID_SIZE); + generate_random_guid(new_root_item->uuid); memcpy(new_root_item->parent_uuid, root->root_item.uuid, BTRFS_UUID_SIZE); if (!(root_flags & BTRFS_ROOT_SUBVOL_RDONLY)) { @@ -1563,7 +1639,8 @@ btrfs_set_root_otransid(new_root_item, trans->transid); old = btrfs_lock_root_node(root); - ret = btrfs_cow_block(trans, root, old, NULL, 0, &old); + ret = btrfs_cow_block(trans, root, old, NULL, 0, &old, + BTRFS_NESTING_COW); if (ret) { btrfs_tree_unlock(old); free_extent_buffer(old); @@ -1571,7 +1648,7 @@ goto fail; } - btrfs_set_lock_blocking(old); + btrfs_set_lock_blocking_write(old); ret = btrfs_copy_root(trans, root, old, &tmp, objectid); /* clean up in any case */ @@ -1609,9 +1686,10 @@ } key.offset = (u64)-1; - pending->snap = btrfs_read_fs_root_no_name(fs_info, &key); + pending->snap = btrfs_get_new_fs_root(fs_info, objectid, pending->anon_dev); if (IS_ERR(pending->snap)) { ret = PTR_ERR(pending->snap); + pending->snap = NULL; btrfs_abort_transaction(trans, ret); goto fail; } @@ -1639,10 +1717,9 @@ if (ret < 0) goto fail; - ret = btrfs_insert_dir_item(trans, parent_root, - dentry->d_name.name, dentry->d_name.len, - BTRFS_I(parent_inode), &key, - BTRFS_FT_DIR, index); + ret = btrfs_insert_dir_item(trans, dentry->d_name.name, + dentry->d_name.len, BTRFS_I(parent_inode), + &key, BTRFS_FT_DIR, index); /* We have check then name at the beginning, so it is impossible. */ BUG_ON(ret == -EEXIST || ret == -EOVERFLOW); if (ret) { @@ -1659,7 +1736,8 @@ btrfs_abort_transaction(trans, ret); goto fail; } - ret = btrfs_uuid_tree_add(trans, new_uuid.b, BTRFS_UUID_KEY_SUBVOL, + ret = btrfs_uuid_tree_add(trans, new_root_item->uuid, + BTRFS_UUID_KEY_SUBVOL, objectid); if (ret) { btrfs_abort_transaction(trans, ret); @@ -1771,7 +1849,8 @@ struct btrfs_transaction *trans) { wait_event(fs_info->transaction_blocked_wait, - trans->state >= TRANS_STATE_COMMIT_START || trans->aborted); + trans->state >= TRANS_STATE_COMMIT_START || + TRANS_ABORTED(trans)); } /* @@ -1783,7 +1862,8 @@ struct btrfs_transaction *trans) { wait_event(fs_info->transaction_wait, - trans->state >= TRANS_STATE_UNBLOCKED || trans->aborted); + trans->state >= TRANS_STATE_UNBLOCKED || + TRANS_ABORTED(trans)); } /* @@ -1865,7 +1945,6 @@ { struct btrfs_fs_info *fs_info = trans->fs_info; struct btrfs_transaction *cur_trans = trans->transaction; - DEFINE_WAIT(wait); WARN_ON(refcount_read(&trans->use_count) > 1); @@ -1880,7 +1959,6 @@ */ BUG_ON(list_empty(&cur_trans->list)); - list_del_init(&cur_trans->list); if (cur_trans == fs_info->running_transaction) { cur_trans->state = TRANS_STATE_COMMIT_DOING; spin_unlock(&fs_info->trans_lock); @@ -1889,6 +1967,17 @@ spin_lock(&fs_info->trans_lock); } + + /* + * Now that we know no one else is still using the transaction we can + * remove the transaction from the list of transactions. This avoids + * the transaction kthread from cleaning up the transaction while some + * other task is still using it, which could result in a use-after-free + * on things like log trees, as it forces the transaction kthread to + * wait for this transaction to be cleaned up by us. + */ + list_del_init(&cur_trans->list); + spin_unlock(&fs_info->trans_lock); btrfs_cleanup_one_transaction(trans->transaction, fs_info); @@ -1912,8 +2001,25 @@ kmem_cache_free(btrfs_trans_handle_cachep, trans); } -static inline int btrfs_start_delalloc_flush(struct btrfs_fs_info *fs_info) +/* + * Release reserved delayed ref space of all pending block groups of the + * transaction and remove them from the list + */ +static void btrfs_cleanup_pending_block_groups(struct btrfs_trans_handle *trans) { + struct btrfs_fs_info *fs_info = trans->fs_info; + struct btrfs_block_group *block_group, *tmp; + + list_for_each_entry_safe(block_group, tmp, &trans->new_bgs, bg_list) { + btrfs_delayed_refs_rsv_release(fs_info, 1); + list_del_init(&block_group->bg_list); + } +} + +static inline int btrfs_start_delalloc_flush(struct btrfs_trans_handle *trans) +{ + struct btrfs_fs_info *fs_info = trans->fs_info; + /* * We use writeback_inodes_sb here because if we used * btrfs_start_delalloc_roots we would deadlock with fs freeze. @@ -1923,22 +2029,50 @@ * from already being in a transaction and our join_transaction doesn't * have to re-take the fs freeze lock. */ - if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) + if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) { writeback_inodes_sb(fs_info->sb, WB_REASON_SYNC); + } else { + struct btrfs_pending_snapshot *pending; + struct list_head *head = &trans->transaction->pending_snapshots; + + /* + * Flush dellaloc for any root that is going to be snapshotted. + * This is done to avoid a corrupted version of files, in the + * snapshots, that had both buffered and direct IO writes (even + * if they were done sequentially) due to an unordered update of + * the inode's size on disk. + */ + list_for_each_entry(pending, head, list) { + int ret; + + ret = btrfs_start_delalloc_snapshot(pending->root); + if (ret) + return ret; + } + } return 0; } -static inline void btrfs_wait_delalloc_flush(struct btrfs_fs_info *fs_info) +static inline void btrfs_wait_delalloc_flush(struct btrfs_trans_handle *trans) { - if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) - btrfs_wait_ordered_roots(fs_info, U64_MAX, 0, (u64)-1); -} + struct btrfs_fs_info *fs_info = trans->fs_info; -static inline void -btrfs_wait_pending_ordered(struct btrfs_transaction *cur_trans) -{ - wait_event(cur_trans->pending_wait, - atomic_read(&cur_trans->pending_ordered) == 0); + if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) { + btrfs_wait_ordered_roots(fs_info, U64_MAX, 0, (u64)-1); + } else { + struct btrfs_pending_snapshot *pending; + struct list_head *head = &trans->transaction->pending_snapshots; + + /* + * Wait for any dellaloc that we started previously for the roots + * that are going to be snapshotted. This is to avoid a corrupted + * version of files in the snapshots that had both buffered and + * direct IO writes (even if they were done sequentially). + */ + list_for_each_entry(pending, head, list) + btrfs_wait_ordered_extents(pending->root, + U64_MAX, 0, U64_MAX); + } } int btrfs_commit_transaction(struct btrfs_trans_handle *trans) @@ -1947,6 +2081,8 @@ struct btrfs_transaction *cur_trans = trans->transaction; struct btrfs_transaction *prev_trans = NULL; int ret; + + ASSERT(refcount_read(&trans->use_count) == 1); /* * Some places just start a transaction to commit it. We need to make @@ -1957,7 +2093,7 @@ trans->dirty = true; /* Stop the commit early if ->aborted is set */ - if (unlikely(READ_ONCE(cur_trans->aborted))) { + if (TRANS_ABORTED(cur_trans)) { ret = cur_trans->aborted; btrfs_end_transaction(trans); return ret; @@ -1984,8 +2120,7 @@ cur_trans->delayed_refs.flushing = 1; smp_wmb(); - if (!list_empty(&trans->new_bgs)) - btrfs_create_pending_block_groups(trans); + btrfs_create_pending_block_groups(trans); ret = btrfs_run_delayed_refs(trans, 0); if (ret) { @@ -2032,7 +2167,7 @@ wait_for_commit(cur_trans); - if (unlikely(cur_trans->aborted)) + if (TRANS_ABORTED(cur_trans)) ret = cur_trans->aborted; btrfs_put_transaction(cur_trans); @@ -2051,7 +2186,7 @@ spin_unlock(&fs_info->trans_lock); wait_for_commit(prev_trans); - ret = prev_trans->aborted; + ret = READ_ONCE(prev_trans->aborted); btrfs_put_transaction(prev_trans); if (ret) @@ -2075,7 +2210,7 @@ extwriter_counter_dec(cur_trans, trans->type); - ret = btrfs_start_delalloc_flush(fs_info); + ret = btrfs_start_delalloc_flush(trans); if (ret) goto cleanup_transaction; @@ -2091,9 +2226,15 @@ if (ret) goto cleanup_transaction; - btrfs_wait_delalloc_flush(fs_info); + btrfs_wait_delalloc_flush(trans); - btrfs_wait_pending_ordered(cur_trans); + /* + * Wait for all ordered extents started by a fast fsync that joined this + * transaction. Otherwise if this transaction commits before the ordered + * extents complete we lose logged data after a power failure. + */ + wait_event(cur_trans->pending_wait, + atomic_read(&cur_trans->pending_ordered) == 0); btrfs_scrub_pause(fs_info); /* @@ -2107,8 +2248,7 @@ wait_event(cur_trans->writer_wait, atomic_read(&cur_trans->num_writers) == 1); - /* ->aborted might be set after the previous check, so check it */ - if (unlikely(READ_ONCE(cur_trans->aborted))) { + if (TRANS_ABORTED(cur_trans)) { ret = cur_trans->aborted; goto scrub_continue; } @@ -2125,10 +2265,8 @@ * core function of the snapshot creation. */ ret = create_pending_snapshots(trans); - if (ret) { - mutex_unlock(&fs_info->reloc_mutex); - goto scrub_continue; - } + if (ret) + goto unlock_reloc; /* * We insert the dir indexes of the snapshots and update the inode @@ -2141,16 +2279,12 @@ * the nodes and leaves. */ ret = btrfs_run_delayed_items(trans); - if (ret) { - mutex_unlock(&fs_info->reloc_mutex); - goto scrub_continue; - } + if (ret) + goto unlock_reloc; ret = btrfs_run_delayed_refs(trans, (unsigned long)-1); - if (ret) { - mutex_unlock(&fs_info->reloc_mutex); - goto scrub_continue; - } + if (ret) + goto unlock_reloc; /* * make sure none of the code above managed to slip in a @@ -2176,11 +2310,8 @@ mutex_lock(&fs_info->tree_log_mutex); ret = commit_fs_roots(trans); - if (ret) { - mutex_unlock(&fs_info->tree_log_mutex); - mutex_unlock(&fs_info->reloc_mutex); - goto scrub_continue; - } + if (ret) + goto unlock_tree_log; /* * Since the transaction is done, we can apply the pending changes @@ -2198,42 +2329,29 @@ * new delayed refs. Must handle them or qgroup can be wrong. */ ret = btrfs_run_delayed_refs(trans, (unsigned long)-1); - if (ret) { - mutex_unlock(&fs_info->tree_log_mutex); - mutex_unlock(&fs_info->reloc_mutex); - goto scrub_continue; - } + if (ret) + goto unlock_tree_log; /* * Since fs roots are all committed, we can get a quite accurate * new_roots. So let's do quota accounting. */ ret = btrfs_qgroup_account_extents(trans); - if (ret < 0) { - mutex_unlock(&fs_info->tree_log_mutex); - mutex_unlock(&fs_info->reloc_mutex); - goto scrub_continue; - } + if (ret < 0) + goto unlock_tree_log; ret = commit_cowonly_roots(trans); - if (ret) { - mutex_unlock(&fs_info->tree_log_mutex); - mutex_unlock(&fs_info->reloc_mutex); - goto scrub_continue; - } + if (ret) + goto unlock_tree_log; /* * The tasks which save the space cache and inode cache may also * update ->aborted, check it. */ - if (unlikely(READ_ONCE(cur_trans->aborted))) { + if (TRANS_ABORTED(cur_trans)) { ret = cur_trans->aborted; - mutex_unlock(&fs_info->tree_log_mutex); - mutex_unlock(&fs_info->reloc_mutex); - goto scrub_continue; + goto unlock_tree_log; } - - btrfs_prepare_extent_commit(fs_info); cur_trans = fs_info->running_transaction; @@ -2247,7 +2365,7 @@ list_add_tail(&fs_info->chunk_root->dirty_list, &cur_trans->switch_commits); - switch_commit_roots(cur_trans); + switch_commit_roots(trans); ASSERT(list_empty(&cur_trans->dirty_bgs)); ASSERT(list_empty(&cur_trans->io_bgs)); @@ -2258,8 +2376,7 @@ memcpy(fs_info->super_for_commit, fs_info->super_copy, sizeof(*fs_info->super_copy)); - btrfs_update_commit_device_size(fs_info); - btrfs_update_commit_device_bytes_used(cur_trans); + btrfs_commit_device_sizes(cur_trans); clear_bit(BTRFS_FS_LOG1_ERR, &fs_info->flags); clear_bit(BTRFS_FS_LOG2_ERR, &fs_info->flags); @@ -2278,6 +2395,10 @@ if (ret) { btrfs_handle_fs_error(fs_info, ret, "Error while writing out transaction"); + /* + * reloc_mutex has been unlocked, tree_log_mutex is still held + * but we can't jump to unlock_tree_log causing double unlock + */ mutex_unlock(&fs_info->tree_log_mutex); goto scrub_continue; } @@ -2303,7 +2424,6 @@ */ cur_trans->state = TRANS_STATE_COMPLETED; wake_up(&cur_trans->commit_wait); - clear_bit(BTRFS_FS_NEED_ASYNC_COMMIT, &fs_info->flags); spin_lock(&fs_info->trans_lock); list_del_init(&cur_trans->list); @@ -2326,10 +2446,15 @@ return ret; +unlock_tree_log: + mutex_unlock(&fs_info->tree_log_mutex); +unlock_reloc: + mutex_unlock(&fs_info->reloc_mutex); scrub_continue: btrfs_scrub_continue(fs_info); cleanup_transaction: btrfs_trans_release_metadata(trans); + btrfs_cleanup_pending_block_groups(trans); btrfs_trans_release_chunk_metadata(trans); trans->block_rsv = NULL; btrfs_warn(fs_info, "Skipping commit of aborted transaction."); @@ -2365,16 +2490,21 @@ list_del_init(&root->root_list); spin_unlock(&fs_info->trans_lock); - btrfs_debug(fs_info, "cleaner removing %llu", root->objectid); + btrfs_debug(fs_info, "cleaner removing %llu", root->root_key.objectid); btrfs_kill_all_delayed_nodes(root); + if (root->ino_cache_inode) { + iput(root->ino_cache_inode); + root->ino_cache_inode = NULL; + } if (btrfs_header_backref_rev(root->node) < BTRFS_MIXED_BACKREF_REV) - ret = btrfs_drop_snapshot(root, NULL, 0, 0); + ret = btrfs_drop_snapshot(root, 0, 0); else - ret = btrfs_drop_snapshot(root, NULL, 1, 0); + ret = btrfs_drop_snapshot(root, 1, 0); + btrfs_put_root(root); return (ret < 0) ? 0 : 1; } -- Gitblit v1.6.2