diff options
Diffstat (limited to 'fs/btrfs/inode.c')
| -rw-r--r-- | fs/btrfs/inode.c | 726 |
1 files changed, 440 insertions, 286 deletions
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 83baec24946d..85a1e5053fe6 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -230,7 +230,6 @@ static noinline int cow_file_range_inline(struct btrfs_trans_handle *trans, u64 inline_len = actual_end - start; u64 aligned_end = (end + root->sectorsize - 1) & ~((u64)root->sectorsize - 1); - u64 hint_byte; u64 data_len = inline_len; int ret; @@ -247,8 +246,7 @@ static noinline int cow_file_range_inline(struct btrfs_trans_handle *trans, return 1; } - ret = btrfs_drop_extents(trans, inode, start, aligned_end, - &hint_byte, 1); + ret = btrfs_drop_extents(trans, root, inode, start, aligned_end, 1); if (ret) return ret; @@ -324,7 +322,8 @@ static noinline int add_async_extent(struct async_cow *cow, * If this code finds it can't get good compression, it puts an * entry onto the work queue to write the uncompressed bytes. This * makes sure that both compressed inodes and uncompressed inodes - * are written in the same order that pdflush sent them down. + * are written in the same order that the flusher thread sent them + * down. */ static noinline int compress_file_range(struct inode *inode, struct page *locked_page, @@ -663,7 +662,7 @@ retry: async_extent->compressed_size, async_extent->compressed_size, 0, alloc_hint, &ins, 1); - if (ret) + if (ret && ret != -ENOSPC) btrfs_abort_transaction(trans, root, ret); btrfs_end_transaction(trans, root); } @@ -1007,9 +1006,7 @@ static noinline void async_cow_submit(struct btrfs_work *work) nr_pages = (async_cow->end - async_cow->start + PAGE_CACHE_SIZE) >> PAGE_CACHE_SHIFT; - atomic_sub(nr_pages, &root->fs_info->async_delalloc_pages); - - if (atomic_read(&root->fs_info->async_delalloc_pages) < + if (atomic_sub_return(nr_pages, &root->fs_info->async_delalloc_pages) < 5 * 1024 * 1024 && waitqueue_active(&root->fs_info->async_submit_wait)) wake_up(&root->fs_info->async_submit_wait); @@ -1309,6 +1306,7 @@ out_check: em->block_start = disk_bytenr; em->bdev = root->fs_info->fs_devices->latest_bdev; set_bit(EXTENT_FLAG_PINNED, &em->flags); + set_bit(EXTENT_FLAG_PREALLOC, &em->flags); while (1) { write_lock(&em_tree->lock); ret = add_extent_mapping(em_tree, em); @@ -1365,11 +1363,7 @@ out_check: } error: - if (nolock) { - err = btrfs_end_transaction_nolock(trans, root); - } else { - err = btrfs_end_transaction(trans, root); - } + err = btrfs_end_transaction(trans, root); if (!ret) ret = err; @@ -1786,7 +1780,6 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans, struct btrfs_path *path; struct extent_buffer *leaf; struct btrfs_key ins; - u64 hint; int ret; path = btrfs_alloc_path(); @@ -1804,8 +1797,8 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans, * the caller is expected to unpin it and allow it to be merged * with the others. */ - ret = btrfs_drop_extents(trans, inode, file_pos, file_pos + num_bytes, - &hint, 0); + ret = btrfs_drop_extents(trans, root, inode, file_pos, + file_pos + num_bytes, 0); if (ret) goto out; @@ -1829,10 +1822,8 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans, btrfs_set_file_extent_encryption(leaf, fi, encryption); btrfs_set_file_extent_other_encoding(leaf, fi, other_encoding); - btrfs_unlock_up_safe(path, 1); - btrfs_set_lock_blocking(leaf); - btrfs_mark_buffer_dirty(leaf); + btrfs_release_path(path); inode_add_bytes(inode, num_bytes); @@ -1884,8 +1875,11 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent) trans = btrfs_join_transaction_nolock(root); else trans = btrfs_join_transaction(root); - if (IS_ERR(trans)) - return PTR_ERR(trans); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); + trans = NULL; + goto out; + } trans->block_rsv = &root->fs_info->delalloc_block_rsv; ret = btrfs_update_inode_fallback(trans, root, inode); if (ret) /* -ENOMEM or corruption */ @@ -1927,11 +1921,10 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent) ordered_extent->len, compress_type, 0, 0, BTRFS_FILE_EXTENT_REG); - unpin_extent_cache(&BTRFS_I(inode)->extent_tree, - ordered_extent->file_offset, - ordered_extent->len); } - + unpin_extent_cache(&BTRFS_I(inode)->extent_tree, + ordered_extent->file_offset, ordered_extent->len, + trans->transid); if (ret < 0) { btrfs_abort_transaction(trans, root, ret); goto out_unlock; @@ -1947,6 +1940,8 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent) btrfs_abort_transaction(trans, root, ret); goto out_unlock; } + } else { + btrfs_set_inode_last_trans(trans, inode); } ret = 0; out_unlock: @@ -1956,12 +1951,8 @@ out_unlock: out: if (root != root->fs_info->tree_root) btrfs_delalloc_release_metadata(inode, ordered_extent->len); - if (trans) { - if (nolock) - btrfs_end_transaction_nolock(trans, root); - else - btrfs_end_transaction(trans, root); - } + if (trans) + btrfs_end_transaction(trans, root); if (ret) clear_extent_uptodate(io_tree, ordered_extent->file_offset, @@ -1969,8 +1960,8 @@ out: ordered_extent->len - 1, NULL, GFP_NOFS); /* - * This needs to be dont to make sure anybody waiting knows we are done - * upating everything for this ordered extent. + * This needs to be done to make sure anybody waiting knows we are done + * updating everything for this ordered extent. */ btrfs_remove_ordered_extent(inode, ordered_extent); @@ -2117,7 +2108,6 @@ void btrfs_run_delayed_iputs(struct btrfs_root *root) if (empty) return; - down_read(&root->fs_info->cleanup_work_sem); spin_lock(&fs_info->delayed_iput_lock); list_splice_init(&fs_info->delayed_iputs, &list); spin_unlock(&fs_info->delayed_iput_lock); @@ -2128,7 +2118,6 @@ void btrfs_run_delayed_iputs(struct btrfs_root *root) iput(delayed->inode); kfree(delayed); } - up_read(&root->fs_info->cleanup_work_sem); } enum btrfs_orphan_cleanup_state { @@ -2196,7 +2185,7 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode) int ret; if (!root->orphan_block_rsv) { - block_rsv = btrfs_alloc_block_rsv(root); + block_rsv = btrfs_alloc_block_rsv(root, BTRFS_BLOCK_RSV_TEMP); if (!block_rsv) return -ENOMEM; } @@ -2223,7 +2212,7 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode) insert = 1; #endif insert = 1; - atomic_dec(&root->orphan_inodes); + atomic_inc(&root->orphan_inodes); } if (!test_and_set_bit(BTRFS_INODE_ORPHAN_META_RESERVED, @@ -2570,8 +2559,8 @@ static void btrfs_read_locked_inode(struct inode *inode) struct btrfs_inode_item); inode->i_mode = btrfs_inode_mode(leaf, inode_item); set_nlink(inode, btrfs_inode_nlink(leaf, inode_item)); - inode->i_uid = btrfs_inode_uid(leaf, inode_item); - inode->i_gid = btrfs_inode_gid(leaf, inode_item); + i_uid_write(inode, btrfs_inode_uid(leaf, inode_item)); + i_gid_write(inode, btrfs_inode_gid(leaf, inode_item)); btrfs_i_size_write(inode, btrfs_inode_size(leaf, inode_item)); tspec = btrfs_inode_atime(inode_item); @@ -2588,6 +2577,18 @@ static void btrfs_read_locked_inode(struct inode *inode) inode_set_bytes(inode, btrfs_inode_nbytes(leaf, inode_item)); BTRFS_I(inode)->generation = btrfs_inode_generation(leaf, inode_item); + BTRFS_I(inode)->last_trans = btrfs_inode_transid(leaf, inode_item); + + /* + * If we were modified in the current generation and evicted from memory + * and then re-read we need to do a full sync since we don't have any + * idea about which extents were modified before we were evicted from + * cache. + */ + if (BTRFS_I(inode)->last_trans == root->fs_info->generation) + set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, + &BTRFS_I(inode)->runtime_flags); + inode->i_version = btrfs_inode_sequence(leaf, inode_item); inode->i_generation = BTRFS_I(inode)->generation; inode->i_rdev = 0; @@ -2649,8 +2650,8 @@ static void fill_inode_item(struct btrfs_trans_handle *trans, struct btrfs_inode_item *item, struct inode *inode) { - btrfs_set_inode_uid(leaf, item, inode->i_uid); - btrfs_set_inode_gid(leaf, item, inode->i_gid); + btrfs_set_inode_uid(leaf, item, i_uid_read(inode)); + btrfs_set_inode_gid(leaf, item, i_gid_read(inode)); btrfs_set_inode_size(leaf, item, BTRFS_I(inode)->disk_i_size); btrfs_set_inode_mode(leaf, item, inode->i_mode); btrfs_set_inode_nlink(leaf, item, inode->i_nlink); @@ -2892,7 +2893,6 @@ static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir, struct btrfs_trans_handle *trans; struct btrfs_root *root = BTRFS_I(dir)->root; struct btrfs_path *path; - struct btrfs_inode_ref *ref; struct btrfs_dir_item *di; struct inode *inode = dentry->d_inode; u64 index; @@ -3006,17 +3006,17 @@ static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir, } btrfs_release_path(path); - ref = btrfs_lookup_inode_ref(trans, root, path, - dentry->d_name.name, dentry->d_name.len, - ino, dir_ino, 0); - if (IS_ERR(ref)) { - err = PTR_ERR(ref); + ret = btrfs_get_inode_ref_index(trans, root, path, dentry->d_name.name, + dentry->d_name.len, ino, dir_ino, 0, + &index); + if (ret) { + err = ret; goto out; } - BUG_ON(!ref); /* Logic error */ + if (check_path_shared(root, path)) goto out; - index = btrfs_inode_ref_index(path->nodes[0], ref); + btrfs_release_path(path); /* @@ -3059,7 +3059,7 @@ out: static void __unlink_end_trans(struct btrfs_trans_handle *trans, struct btrfs_root *root) { - if (trans->block_rsv == &root->fs_info->global_block_rsv) { + if (trans->block_rsv->type == BTRFS_BLOCK_RSV_GLOBAL) { btrfs_block_rsv_release(root, trans->block_rsv, trans->bytes_reserved); trans->block_rsv = &root->fs_info->trans_block_rsv; @@ -3173,7 +3173,7 @@ int btrfs_unlink_subvol(struct btrfs_trans_handle *trans, btrfs_i_size_write(dir, dir->i_size - name_len * 2); inode_inc_iversion(dir); dir->i_mtime = dir->i_ctime = CURRENT_TIME; - ret = btrfs_update_inode(trans, root, dir); + ret = btrfs_update_inode_fallback(trans, root, dir); if (ret) btrfs_abort_transaction(trans, root, ret); out: @@ -3189,9 +3189,10 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry) struct btrfs_trans_handle *trans; unsigned long nr = 0; - if (inode->i_size > BTRFS_EMPTY_DIR_SIZE || - btrfs_ino(inode) == BTRFS_FIRST_FREE_OBJECTID) + if (inode->i_size > BTRFS_EMPTY_DIR_SIZE) return -ENOTEMPTY; + if (btrfs_ino(inode) == BTRFS_FIRST_FREE_OBJECTID) + return -EPERM; trans = __unlink_start_trans(dir, dentry); if (IS_ERR(trans)) @@ -3265,8 +3266,13 @@ int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans, return -ENOMEM; path->reada = -1; + /* + * We want to drop from the next block forward in case this new size is + * not block aligned since we will be keeping the last block of the + * extent just the way it is. + */ if (root->ref_cows || root == root->fs_info->tree_root) - btrfs_drop_extent_cache(inode, new_size & (~mask), (u64)-1, 0); + btrfs_drop_extent_cache(inode, (new_size + mask) & (~mask), (u64)-1, 0); /* * This function is also used to drop the items in the log tree before @@ -3427,12 +3433,6 @@ delete: if (path->slots[0] == 0 || path->slots[0] != pending_del_slot) { - if (root->ref_cows && - BTRFS_I(inode)->location.objectid != - BTRFS_FREE_INO_OBJECTID) { - err = -EAGAIN; - goto out; - } if (pending_del_nr) { ret = btrfs_del_items(trans, root, path, pending_del_slot, @@ -3463,12 +3463,20 @@ error: } /* - * taken from block_truncate_page, but does cow as it zeros out - * any bytes left in the last page in the file. + * btrfs_truncate_page - read, zero a chunk and write a page + * @inode - inode that we're zeroing + * @from - the offset to start zeroing + * @len - the length to zero, 0 to zero the entire range respective to the + * offset + * @front - zero up to the offset instead of from the offset on + * + * This will find the page for the "from" offset and cow the page and zero the + * part we want to zero. This is used with truncate and hole punching. */ -static int btrfs_truncate_page(struct address_space *mapping, loff_t from) +int btrfs_truncate_page(struct inode *inode, loff_t from, loff_t len, + int front) { - struct inode *inode = mapping->host; + struct address_space *mapping = inode->i_mapping; struct btrfs_root *root = BTRFS_I(inode)->root; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct btrfs_ordered_extent *ordered; @@ -3483,7 +3491,8 @@ static int btrfs_truncate_page(struct address_space *mapping, loff_t from) u64 page_start; u64 page_end; - if ((offset & (blocksize - 1)) == 0) + if ((offset & (blocksize - 1)) == 0 && + (!len || ((len & (blocksize - 1)) == 0))) goto out; ret = btrfs_delalloc_reserve_space(inode, PAGE_CACHE_SIZE); if (ret) @@ -3530,7 +3539,8 @@ again: } clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, page_end, - EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING, + EXTENT_DIRTY | EXTENT_DELALLOC | + EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 0, 0, &cached_state, GFP_NOFS); ret = btrfs_set_extent_delalloc(inode, page_start, page_end, @@ -3543,8 +3553,13 @@ again: ret = 0; if (offset != PAGE_CACHE_SIZE) { + if (!len) + len = PAGE_CACHE_SIZE - offset; kaddr = kmap(page); - memset(kaddr + offset, 0, PAGE_CACHE_SIZE - offset); + if (front) + memset(kaddr, 0, offset); + else + memset(kaddr + offset, 0, len); flush_dcache_page(page); kunmap(page); } @@ -3575,6 +3590,7 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size) struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct extent_map *em = NULL; struct extent_state *cached_state = NULL; + struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; u64 mask = root->sectorsize - 1; u64 hole_start = (oldsize + mask) & ~mask; u64 block_end = (size + mask) & ~mask; @@ -3611,7 +3627,7 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size) last_byte = min(extent_map_end(em), block_end); last_byte = (last_byte + mask) & ~mask; if (!test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) { - u64 hint_byte = 0; + struct extent_map *hole_em; hole_size = last_byte - cur_offset; trans = btrfs_start_transaction(root, 3); @@ -3620,9 +3636,9 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size) break; } - err = btrfs_drop_extents(trans, inode, cur_offset, - cur_offset + hole_size, - &hint_byte, 1); + err = btrfs_drop_extents(trans, root, inode, + cur_offset, + cur_offset + hole_size, 1); if (err) { btrfs_abort_transaction(trans, root, err); btrfs_end_transaction(trans, root); @@ -3639,9 +3655,39 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size) break; } - btrfs_drop_extent_cache(inode, hole_start, - last_byte - 1, 0); + btrfs_drop_extent_cache(inode, cur_offset, + cur_offset + hole_size - 1, 0); + hole_em = alloc_extent_map(); + if (!hole_em) { + set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, + &BTRFS_I(inode)->runtime_flags); + goto next; + } + hole_em->start = cur_offset; + hole_em->len = hole_size; + hole_em->orig_start = cur_offset; + + hole_em->block_start = EXTENT_MAP_HOLE; + hole_em->block_len = 0; + hole_em->bdev = root->fs_info->fs_devices->latest_bdev; + hole_em->compress_type = BTRFS_COMPRESS_NONE; + hole_em->generation = trans->transid; + while (1) { + write_lock(&em_tree->lock); + err = add_extent_mapping(em_tree, hole_em); + if (!err) + list_move(&hole_em->list, + &em_tree->modified_extents); + write_unlock(&em_tree->lock); + if (err != -EEXIST) + break; + btrfs_drop_extent_cache(inode, cur_offset, + cur_offset + + hole_size - 1, 0); + } + free_extent_map(hole_em); +next: btrfs_update_inode(trans, root, inode); btrfs_end_transaction(trans, root); } @@ -3766,26 +3812,22 @@ void btrfs_evict_inode(struct inode *inode) goto no_delete; } - rsv = btrfs_alloc_block_rsv(root); + rsv = btrfs_alloc_block_rsv(root, BTRFS_BLOCK_RSV_TEMP); if (!rsv) { btrfs_orphan_del(NULL, inode); goto no_delete; } rsv->size = min_size; + rsv->failfast = 1; global_rsv = &root->fs_info->global_block_rsv; btrfs_i_size_write(inode, 0); /* - * This is a bit simpler than btrfs_truncate since - * - * 1) We've already reserved our space for our orphan item in the - * unlink. - * 2) We're going to delete the inode item, so we don't need to update - * it at all. - * - * So we just need to reserve some slack space in case we add bytes when - * doing the truncate. + * This is a bit simpler than btrfs_truncate since we've already + * reserved our space for our orphan item in the unlink, so we just + * need to reserve some slack space in case we add bytes and update + * inode item when doing the truncate. */ while (1) { ret = btrfs_block_rsv_refill_noflush(root, rsv, min_size); @@ -3806,7 +3848,7 @@ void btrfs_evict_inode(struct inode *inode) goto no_delete; } - trans = btrfs_start_transaction(root, 0); + trans = btrfs_start_transaction_noflush(root, 1); if (IS_ERR(trans)) { btrfs_orphan_del(NULL, inode); btrfs_free_block_rsv(root, rsv); @@ -3816,9 +3858,13 @@ void btrfs_evict_inode(struct inode *inode) trans->block_rsv = rsv; ret = btrfs_truncate_inode_items(trans, root, inode, 0, 0); - if (ret != -EAGAIN) + if (ret != -ENOSPC) break; + trans->block_rsv = &root->fs_info->trans_block_rsv; + ret = btrfs_update_inode(trans, root, inode); + BUG_ON(ret); + nr = trans->blocks_used; btrfs_end_transaction(trans, root); trans = NULL; @@ -4468,10 +4514,7 @@ int btrfs_write_inode(struct inode *inode, struct writeback_control *wbc) trans = btrfs_join_transaction(root); if (IS_ERR(trans)) return PTR_ERR(trans); - if (nolock) - ret = btrfs_end_transaction_nolock(trans, root); - else - ret = btrfs_commit_transaction(trans, root); + ret = btrfs_commit_transaction(trans, root); } return ret; } @@ -4669,6 +4712,14 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, BTRFS_I(inode)->generation = trans->transid; inode->i_generation = BTRFS_I(inode)->generation; + /* + * We could have gotten an inode number from somebody who was fsynced + * and then removed in this same transaction, so let's just set full + * sync since it will be a full sync anyway and this will blow away the + * old info in the log. + */ + set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &BTRFS_I(inode)->runtime_flags); + if (S_ISDIR(mode)) owner = 0; else @@ -4678,6 +4729,12 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, btrfs_set_key_type(&key[0], BTRFS_INODE_ITEM_KEY); key[0].offset = 0; + /* + * Start new inodes with an inode_ref. This is slightly more + * efficient for small numbers of hard links since they will + * be packed into one item. Extended refs will kick in if we + * add more hard links than can fit in the ref item. + */ key[1].objectid = objectid; btrfs_set_key_type(&key[1], BTRFS_INODE_REF_KEY); key[1].offset = ref_objectid; @@ -4984,7 +5041,7 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir, if (root->objectid != BTRFS_I(inode)->root->objectid) return -EXDEV; - if (inode->i_nlink == ~0U) + if (inode->i_nlink >= BTRFS_LINK_MAX) return -EMLINK; err = btrfs_set_inode_index(dir, &index); @@ -5448,7 +5505,8 @@ insert: write_unlock(&em_tree->lock); out: - trace_btrfs_get_extent(root, em); + if (em) + trace_btrfs_get_extent(root, em); if (path) btrfs_free_path(path); @@ -5773,18 +5831,154 @@ out: return ret; } +static int lock_extent_direct(struct inode *inode, u64 lockstart, u64 lockend, + struct extent_state **cached_state, int writing) +{ + struct btrfs_ordered_extent *ordered; + int ret = 0; + + while (1) { + lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend, + 0, cached_state); + /* + * We're concerned with the entire range that we're going to be + * doing DIO to, so we need to make sure theres no ordered + * extents in this range. + */ + ordered = btrfs_lookup_ordered_range(inode, lockstart, + lockend - lockstart + 1); + + /* + * We need to make sure there are no buffered pages in this + * range either, we could have raced between the invalidate in + * generic_file_direct_write and locking the extent. The + * invalidate needs to happen so that reads after a write do not + * get stale data. + */ + if (!ordered && (!writing || + !test_range_bit(&BTRFS_I(inode)->io_tree, + lockstart, lockend, EXTENT_UPTODATE, 0, + *cached_state))) + break; + + unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend, + cached_state, GFP_NOFS); + + if (ordered) { + btrfs_start_ordered_extent(inode, ordered, 1); + btrfs_put_ordered_extent(ordered); + } else { + /* Screw you mmap */ + ret = filemap_write_and_wait_range(inode->i_mapping, + lockstart, + lockend); + if (ret) + break; + + /* + * If we found a page that couldn't be invalidated just + * fall back to buffered. + */ + ret = invalidate_inode_pages2_range(inode->i_mapping, + lockstart >> PAGE_CACHE_SHIFT, + lockend >> PAGE_CACHE_SHIFT); + if (ret) + break; + } + + cond_resched(); + } + + return ret; +} + +static struct extent_map *create_pinned_em(struct inode *inode, u64 start, + u64 len, u64 orig_start, + u64 block_start, u64 block_len, + int type) +{ + struct extent_map_tree *em_tree; + struct extent_map *em; + struct btrfs_root *root = BTRFS_I(inode)->root; + int ret; + + em_tree = &BTRFS_I(inode)->extent_tree; + em = alloc_extent_map(); + if (!em) + return ERR_PTR(-ENOMEM); + + em->start = start; + em->orig_start = orig_start; + em->len = len; + em->block_len = block_len; + em->block_start = block_start; + em->bdev = root->fs_info->fs_devices->latest_bdev; + set_bit(EXTENT_FLAG_PINNED, &em->flags); + if (type == BTRFS_ORDERED_PREALLOC) + set_bit(EXTENT_FLAG_PREALLOC, &em->flags); + + do { + btrfs_drop_extent_cache(inode, em->start, + em->start + em->len - 1, 0); + write_lock(&em_tree->lock); + ret = add_extent_mapping(em_tree, em); + write_unlock(&em_tree->lock); + } while (ret == -EEXIST); + + if (ret) { + free_extent_map(em); + return ERR_PTR(ret); + } + + return em; +} + + static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock, struct buffer_head *bh_result, int create) { struct extent_map *em; struct btrfs_root *root = BTRFS_I(inode)->root; + struct extent_state *cached_state = NULL; u64 start = iblock << inode->i_blkbits; + u64 lockstart, lockend; u64 len = bh_result->b_size; struct btrfs_trans_handle *trans; + int unlock_bits = EXTENT_LOCKED; + int ret; + + if (create) { + ret = btrfs_delalloc_reserve_space(inode, len); + if (ret) + return ret; + unlock_bits |= EXTENT_DELALLOC | EXTENT_DIRTY; + } else { + len = min_t(u64, len, root->sectorsize); + } + + lockstart = start; + lockend = start + len - 1; + + /* + * If this errors out it's because we couldn't invalidate pagecache for + * this range and we need to fallback to buffered. + */ + if (lock_extent_direct(inode, lockstart, lockend, &cached_state, create)) + return -ENOTBLK; + + if (create) { + ret = set_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, + lockend, EXTENT_DELALLOC, NULL, + &cached_state, GFP_NOFS); + if (ret) + goto unlock_err; + } em = btrfs_get_extent(inode, NULL, 0, start, len, 0); - if (IS_ERR(em)) - return PTR_ERR(em); + if (IS_ERR(em)) { + ret = PTR_ERR(em); + goto unlock_err; + } /* * Ok for INLINE and COMPRESSED extents we need to fallback on buffered @@ -5803,17 +5997,16 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock, if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags) || em->block_start == EXTENT_MAP_INLINE) { free_extent_map(em); - return -ENOTBLK; + ret = -ENOTBLK; + goto unlock_err; } /* Just a good old fashioned hole, return */ if (!create && (em->block_start == EXTENT_MAP_HOLE || test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) { free_extent_map(em); - /* DIO will do one hole at a time, so just unlock a sector */ - unlock_extent(&BTRFS_I(inode)->io_tree, start, - start + root->sectorsize - 1); - return 0; + ret = 0; + goto unlock_err; } /* @@ -5826,8 +6019,9 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock, * */ if (!create) { - len = em->len - (start - em->start); - goto map; + len = min(len, em->len - (start - em->start)); + lockstart = start + len; + goto unlock; } if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags) || @@ -5854,12 +6048,25 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock, goto must_cow; if (can_nocow_odirect(trans, inode, start, len) == 1) { + u64 orig_start = em->start; + + if (type == BTRFS_ORDERED_PREALLOC) { + free_extent_map(em); + em = create_pinned_em(inode, start, len, + orig_start, + block_start, len, type); + if (IS_ERR(em)) { + btrfs_end_transaction(trans, root); + goto unlock_err; + } + } + ret = btrfs_add_ordered_extent_dio(inode, start, block_start, len, len, type); btrfs_end_transaction(trans, root); if (ret) { free_extent_map(em); - return ret; + goto unlock_err; } goto unlock; } @@ -5872,14 +6079,12 @@ must_cow: */ len = bh_result->b_size; em = btrfs_new_extent_direct(inode, em, start, len); - if (IS_ERR(em)) - return PTR_ERR(em); + if (IS_ERR(em)) { + ret = PTR_ERR(em); + goto unlock_err; + } len = min(len, em->len - (start - em->start)); unlock: - clear_extent_bit(&BTRFS_I(inode)->io_tree, start, start + len - 1, - EXTENT_LOCKED | EXTENT_DELALLOC | EXTENT_DIRTY, 1, - 0, NULL, GFP_NOFS); -map: bh_result->b_blocknr = (em->block_start + (start - em->start)) >> inode->i_blkbits; bh_result->b_size = len; @@ -5897,9 +6102,45 @@ map: i_size_write(inode, start + len); } + /* + * In the case of write we need to clear and unlock the entire range, + * in the case of read we need to unlock only the end area that we + * aren't using if there is any left over space. + */ + if (lockstart < lockend) { + if (create && len < lockend - lockstart) { + clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, + lockstart + len - 1, + unlock_bits | EXTENT_DEFRAG, 1, 0, + &cached_state, GFP_NOFS); + /* + * Beside unlock, we also need to cleanup reserved space + * for the left range by attaching EXTENT_DO_ACCOUNTING. + */ + clear_extent_bit(&BTRFS_I(inode)->io_tree, + lockstart + len, lockend, + unlock_bits | EXTENT_DO_ACCOUNTING | + EXTENT_DEFRAG, 1, 0, NULL, GFP_NOFS); + } else { + clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, + lockend, unlock_bits, 1, 0, + &cached_state, GFP_NOFS); + } + } else { + free_extent_state(cached_state); + } + free_extent_map(em); return 0; + +unlock_err: + if (create) + unlock_bits |= EXTENT_DO_ACCOUNTING; + + clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, lockend, + unlock_bits, 1, 0, &cached_state, GFP_NOFS); + return ret; } struct btrfs_dio_private { @@ -5907,7 +6148,6 @@ struct btrfs_dio_private { u64 logical_offset; u64 disk_bytenr; u64 bytes; - u32 *csums; void *private; /* number of bios pending for this dio */ @@ -5927,7 +6167,6 @@ static void btrfs_endio_direct_read(struct bio *bio, int err) struct inode *inode = dip->inode; struct btrfs_root *root = BTRFS_I(inode)->root; u64 start; - u32 *private = dip->csums; start = dip->logical_offset; do { @@ -5935,8 +6174,12 @@ static void btrfs_endio_direct_read(struct bio *bio, int err) struct page *page = bvec->bv_page; char *kaddr; u32 csum = ~(u32)0; + u64 private = ~(u32)0; unsigned long flags; + if (get_state_private(&BTRFS_I(inode)->io_tree, + start, &private)) + goto failed; local_irq_save(flags); kaddr = kmap_atomic(page); csum = btrfs_csum_data(root, kaddr + bvec->bv_offset, @@ -5946,18 +6189,18 @@ static void btrfs_endio_direct_read(struct bio *bio, int err) local_irq_restore(flags); flush_dcache_page(bvec->bv_page); - if (csum != *private) { + if (csum != private) { +failed: printk(KERN_ERR "btrfs csum failed ino %llu off" " %llu csum %u private %u\n", (unsigned long long)btrfs_ino(inode), (unsigned long long)start, - csum, *private); + csum, (unsigned)private); err = -EIO; } } start += bvec->bv_len; - private++; bvec++; } while (bvec <= bvec_end); @@ -5965,7 +6208,6 @@ static void btrfs_endio_direct_read(struct bio *bio, int err) dip->logical_offset + dip->bytes - 1); bio->bi_private = dip->private; - kfree(dip->csums); kfree(dip); /* If we had a csum failure make sure to clear the uptodate flag */ @@ -6071,7 +6313,7 @@ static struct bio *btrfs_dio_bio_alloc(struct block_device *bdev, static inline int __btrfs_submit_dio_bio(struct bio *bio, struct inode *inode, int rw, u64 file_offset, int skip_sum, - u32 *csums, int async_submit) + int async_submit) { int write = rw & REQ_WRITE; struct btrfs_root *root = BTRFS_I(inode)->root; @@ -6104,8 +6346,7 @@ static inline int __btrfs_submit_dio_bio(struct bio *bio, struct inode *inode, if (ret) goto err; } else if (!skip_sum) { - ret = btrfs_lookup_bio_sums_dio(root, inode, bio, - file_offset, csums); + ret = btrfs_lookup_bio_sums_dio(root, inode, bio, file_offset); if (ret) goto err; } @@ -6131,10 +6372,8 @@ static int btrfs_submit_direct_hook(int rw, struct btrfs_dio_private *dip, u64 submit_len = 0; u64 map_length; int nr_pages = 0; - u32 *csums = dip->csums; int ret = 0; int async_submit = 0; - int write = rw & REQ_WRITE; map_length = orig_bio->bi_size; ret = btrfs_map_block(map_tree, READ, start_sector << 9, @@ -6170,16 +6409,13 @@ static int btrfs_submit_direct_hook(int rw, struct btrfs_dio_private *dip, atomic_inc(&dip->pending_bios); ret = __btrfs_submit_dio_bio(bio, inode, rw, file_offset, skip_sum, - csums, async_submit); + async_submit); if (ret) { bio_put(bio); atomic_dec(&dip->pending_bios); goto out_err; } - /* Write's use the ordered csums */ - if (!write && !skip_sum) - csums = csums + nr_pages; start_sector += submit_len >> 9; file_offset += submit_len; @@ -6209,7 +6445,7 @@ static int btrfs_submit_direct_hook(int rw, struct btrfs_dio_private *dip, submit: ret = __btrfs_submit_dio_bio(bio, inode, rw, file_offset, skip_sum, - csums, async_submit); + async_submit); if (!ret) return 0; @@ -6245,17 +6481,6 @@ static void btrfs_submit_direct(int rw, struct bio *bio, struct inode *inode, ret = -ENOMEM; goto free_ordered; } - dip->csums = NULL; - - /* Write's use the ordered csum stuff, so we don't need dip->csums */ - if (!write && !skip_sum) { - dip->csums = kmalloc(sizeof(u32) * bio->bi_vcnt, GFP_NOFS); - if (!dip->csums) { - kfree(dip); - ret = -ENOMEM; - goto free_ordered; - } - } dip->private = bio->bi_private; dip->inode = inode; @@ -6340,132 +6565,22 @@ static ssize_t check_direct_IO(struct btrfs_root *root, int rw, struct kiocb *io out: return retval; } + static ssize_t btrfs_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov, loff_t offset, unsigned long nr_segs) { struct file *file = iocb->ki_filp; struct inode *inode = file->f_mapping->host; - struct btrfs_ordered_extent *ordered; - struct extent_state *cached_state = NULL; - u64 lockstart, lockend; - ssize_t ret; - int writing = rw & WRITE; - int write_bits = 0; - size_t count = iov_length(iov, nr_segs); if (check_direct_IO(BTRFS_I(inode)->root, rw, iocb, iov, - offset, nr_segs)) { + offset, nr_segs)) return 0; - } - - lockstart = offset; - lockend = offset + count - 1; - - if (writing) { - ret = btrfs_delalloc_reserve_space(inode, count); - if (ret) - goto out; - } - - while (1) { - lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend, - 0, &cached_state); - /* - * We're concerned with the entire range that we're going to be - * doing DIO to, so we need to make sure theres no ordered - * extents in this range. - */ - ordered = btrfs_lookup_ordered_range(inode, lockstart, - lockend - lockstart + 1); - - /* - * We need to make sure there are no buffered pages in this - * range either, we could have raced between the invalidate in - * generic_file_direct_write and locking the extent. The - * invalidate needs to happen so that reads after a write do not - * get stale data. - */ - if (!ordered && (!writing || - !test_range_bit(&BTRFS_I(inode)->io_tree, - lockstart, lockend, EXTENT_UPTODATE, 0, - cached_state))) - break; - - unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend, - &cached_state, GFP_NOFS); - - if (ordered) { - btrfs_start_ordered_extent(inode, ordered, 1); - btrfs_put_ordered_extent(ordered); - } else { - /* Screw you mmap */ - ret = filemap_write_and_wait_range(file->f_mapping, - lockstart, - lockend); - if (ret) - goto out; - /* - * If we found a page that couldn't be invalidated just - * fall back to buffered. - */ - ret = invalidate_inode_pages2_range(file->f_mapping, - lockstart >> PAGE_CACHE_SHIFT, - lockend >> PAGE_CACHE_SHIFT); - if (ret) { - if (ret == -EBUSY) - ret = 0; - goto out; - } - } - - cond_resched(); - } - - /* - * we don't use btrfs_set_extent_delalloc because we don't want - * the dirty or uptodate bits - */ - if (writing) { - write_bits = EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING; - ret = set_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, lockend, - EXTENT_DELALLOC, NULL, &cached_state, - GFP_NOFS); - if (ret) { - clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, - lockend, EXTENT_LOCKED | write_bits, - 1, 0, &cached_state, GFP_NOFS); - goto out; - } - } - - free_extent_state(cached_state); - cached_state = NULL; - - ret = __blockdev_direct_IO(rw, iocb, inode, + return __blockdev_direct_IO(rw, iocb, inode, BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev, iov, offset, nr_segs, btrfs_get_blocks_direct, NULL, btrfs_submit_direct, 0); - - if (ret < 0 && ret != -EIOCBQUEUED) { - clear_extent_bit(&BTRFS_I(inode)->io_tree, offset, - offset + iov_length(iov, nr_segs) - 1, - EXTENT_LOCKED | write_bits, 1, 0, - &cached_state, GFP_NOFS); - } else if (ret >= 0 && ret < iov_length(iov, nr_segs)) { - /* - * We're falling back to buffered, unlock the section we didn't - * do IO on. - */ - clear_extent_bit(&BTRFS_I(inode)->io_tree, offset + ret, - offset + iov_length(iov, nr_segs) - 1, - EXTENT_LOCKED | write_bits, 1, 0, - &cached_state, GFP_NOFS); - } -out: - free_extent_state(cached_state); - return ret; } static int btrfs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, @@ -6570,8 +6685,8 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) */ clear_extent_bit(tree, page_start, page_end, EXTENT_DIRTY | EXTENT_DELALLOC | - EXTENT_LOCKED | EXTENT_DO_ACCOUNTING, 1, 0, - &cached_state, GFP_NOFS); + EXTENT_LOCKED | EXTENT_DO_ACCOUNTING | + EXTENT_DEFRAG, 1, 0, &cached_state, GFP_NOFS); /* * whoever cleared the private bit is responsible * for the finish_ordered_io @@ -6587,7 +6702,8 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) } clear_extent_bit(tree, page_start, page_end, EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC | - EXTENT_DO_ACCOUNTING, 1, 1, &cached_state, GFP_NOFS); + EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 1, 1, + &cached_state, GFP_NOFS); __btrfs_releasepage(page, GFP_NOFS); ClearPageChecked(page); @@ -6684,7 +6800,8 @@ again: * prepare_pages in the normal write path. */ clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, page_end, - EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING, + EXTENT_DIRTY | EXTENT_DELALLOC | + EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 0, 0, &cached_state, GFP_NOFS); ret = btrfs_set_extent_delalloc(inode, page_start, page_end, @@ -6715,6 +6832,7 @@ again: BTRFS_I(inode)->last_trans = root->fs_info->generation; BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid; + BTRFS_I(inode)->last_log_commit = BTRFS_I(inode)->root->last_log_commit; unlock_extent_cached(io_tree, page_start, page_end, &cached_state, GFP_NOFS); @@ -6742,7 +6860,7 @@ static int btrfs_truncate(struct inode *inode) u64 mask = root->sectorsize - 1; u64 min_size = btrfs_calc_trunc_metadata_size(root, 1); - ret = btrfs_truncate_page(inode->i_mapping, inode->i_size); + ret = btrfs_truncate_page(inode, inode->i_size, 0, 0); if (ret) return ret; @@ -6785,10 +6903,11 @@ static int btrfs_truncate(struct inode *inode) * 3) fs_info->trans_block_rsv - this will have 1 items worth left for * updating the inode. */ - rsv = btrfs_alloc_block_rsv(root); + rsv = btrfs_alloc_block_rsv(root, BTRFS_BLOCK_RSV_TEMP); if (!rsv) return -ENOMEM; rsv->size = min_size; + rsv->failfast = 1; /* * 1 for the truncate slack space @@ -6834,36 +6953,21 @@ static int btrfs_truncate(struct inode *inode) &BTRFS_I(inode)->runtime_flags)) btrfs_add_ordered_operation(trans, root, inode); - while (1) { - ret = btrfs_block_rsv_refill(root, rsv, min_size); - if (ret) { - /* - * This can only happen with the original transaction we - * started above, every other time we shouldn't have a - * transaction started yet. - */ - if (ret == -EAGAIN) - goto end_trans; - err = ret; - break; - } - - if (!trans) { - /* Just need the 1 for updating the inode */ - trans = btrfs_start_transaction(root, 1); - if (IS_ERR(trans)) { - ret = err = PTR_ERR(trans); - trans = NULL; - break; - } - } - - trans->block_rsv = rsv; + /* + * So if we truncate and then write and fsync we normally would just + * write the extents that changed, which is a problem if we need to + * first truncate that entire inode. So set this flag so we write out + * all of the extents in the inode to the sync log so we're completely + * safe. + */ + set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &BTRFS_I(inode)->runtime_flags); + trans->block_rsv = rsv; + while (1) { ret = btrfs_truncate_inode_items(trans, root, inode, inode->i_size, BTRFS_EXTENT_DATA_KEY); - if (ret != -EAGAIN) { + if (ret != -ENOSPC) { err = ret; break; } @@ -6874,11 +6978,22 @@ static int btrfs_truncate(struct inode *inode) err = ret; break; } -end_trans: + nr = trans->blocks_used; btrfs_end_transaction(trans, root); - trans = NULL; btrfs_btree_balance_dirty(root, nr); + + trans = btrfs_start_transaction(root, 2); + if (IS_ERR(trans)) { + ret = err = PTR_ERR(trans); + trans = NULL; + break; + } + + ret = btrfs_block_rsv_migrate(&root->fs_info->trans_block_rsv, + rsv, min_size); + BUG_ON(ret); /* shouldn't happen */ + trans->block_rsv = rsv; } if (ret == 0 && inode->i_nlink > 0) { @@ -6962,6 +7077,7 @@ struct inode *btrfs_alloc_inode(struct super_block *sb) ei->csum_bytes = 0; ei->index_cnt = (u64)-1; ei->last_unlink_trans = 0; + ei->last_log_commit = 0; spin_lock_init(&ei->lock); ei->outstanding_extents = 0; @@ -7073,6 +7189,11 @@ static void init_once(void *foo) void btrfs_destroy_cachep(void) { + /* + * Make sure all delayed rcu free inodes are flushed before we + * destroy cache. + */ + rcu_barrier(); if (btrfs_inode_cachep) kmem_cache_destroy(btrfs_inode_cachep); if (btrfs_trans_handle_cachep) @@ -7087,31 +7208,31 @@ void btrfs_destroy_cachep(void) int btrfs_init_cachep(void) { - btrfs_inode_cachep = kmem_cache_create("btrfs_inode_cache", + btrfs_inode_cachep = kmem_cache_create("btrfs_inode", sizeof(struct btrfs_inode), 0, SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, init_once); if (!btrfs_inode_cachep) goto fail; - btrfs_trans_handle_cachep = kmem_cache_create("btrfs_trans_handle_cache", + btrfs_trans_handle_cachep = kmem_cache_create("btrfs_trans_handle", sizeof(struct btrfs_trans_handle), 0, SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL); if (!btrfs_trans_handle_cachep) goto fail; - btrfs_transaction_cachep = kmem_cache_create("btrfs_transaction_cache", + btrfs_transaction_cachep = kmem_cache_create("btrfs_transaction", sizeof(struct btrfs_transaction), 0, SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL); if (!btrfs_transaction_cachep) goto fail; - btrfs_path_cachep = kmem_cache_create("btrfs_path_cache", + btrfs_path_cachep = kmem_cache_create("btrfs_path", sizeof(struct btrfs_path), 0, SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL); if (!btrfs_path_cachep) goto fail; - btrfs_free_space_cachep = kmem_cache_create("btrfs_free_space_cache", + btrfs_free_space_cachep = kmem_cache_create("btrfs_free_space", sizeof(struct btrfs_free_space), 0, SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL); if (!btrfs_free_space_cachep) @@ -7505,6 +7626,8 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode, loff_t actual_len, u64 *alloc_hint, struct btrfs_trans_handle *trans) { + struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + struct extent_map *em; struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_key ins; u64 cur_offset = start; @@ -7545,6 +7668,37 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode, btrfs_drop_extent_cache(inode, cur_offset, cur_offset + ins.offset -1, 0); + em = alloc_extent_map(); + if (!em) { + set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, + &BTRFS_I(inode)->runtime_flags); + goto next; + } + + em->start = cur_offset; + em->orig_start = cur_offset; + em->len = ins.offset; + em->block_start = ins.objectid; + em->block_len = ins.offset; + em->bdev = root->fs_info->fs_devices->latest_bdev; + set_bit(EXTENT_FLAG_PREALLOC, &em->flags); + em->generation = trans->transid; + + while (1) { + write_lock(&em_tree->lock); + ret = add_extent_mapping(em_tree, em); + if (!ret) + list_move(&em->list, + &em_tree->modified_extents); + write_unlock(&em_tree->lock); + if (ret != -EEXIST) + break; + btrfs_drop_extent_cache(inode, cur_offset, + cur_offset + ins.offset - 1, + 0); + } + free_extent_map(em); +next: num_bytes -= ins.offset; cur_offset += ins.offset; *alloc_hint = ins.objectid + ins.offset; |