diff options
Diffstat (limited to 'fs/btrfs/free-space-cache.c')
| -rw-r--r-- | fs/btrfs/free-space-cache.c | 454 |
1 files changed, 293 insertions, 161 deletions
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c index b4f9904c4c6..2b0a627cb5f 100644 --- a/fs/btrfs/free-space-cache.c +++ b/fs/btrfs/free-space-cache.c @@ -218,7 +218,6 @@ int btrfs_check_trunc_cache_free_space(struct btrfs_root *root, int btrfs_truncate_free_space_cache(struct btrfs_root *root, struct btrfs_trans_handle *trans, - struct btrfs_path *path, struct inode *inode) { int ret = 0; @@ -275,18 +274,32 @@ struct io_ctl { }; static int io_ctl_init(struct io_ctl *io_ctl, struct inode *inode, - struct btrfs_root *root) + struct btrfs_root *root, int write) { + int num_pages; + int check_crcs = 0; + + num_pages = (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >> + PAGE_CACHE_SHIFT; + + if (btrfs_ino(inode) != BTRFS_FREE_INO_OBJECTID) + check_crcs = 1; + + /* Make sure we can fit our crcs into the first page */ + if (write && check_crcs && + (num_pages * sizeof(u32)) >= PAGE_CACHE_SIZE) + return -ENOSPC; + memset(io_ctl, 0, sizeof(struct io_ctl)); - io_ctl->num_pages = (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >> - PAGE_CACHE_SHIFT; - io_ctl->pages = kzalloc(sizeof(struct page *) * io_ctl->num_pages, - GFP_NOFS); + + io_ctl->pages = kzalloc(sizeof(struct page *) * num_pages, GFP_NOFS); if (!io_ctl->pages) return -ENOMEM; + + io_ctl->num_pages = num_pages; io_ctl->root = root; - if (btrfs_ino(inode) != BTRFS_FREE_INO_OBJECTID) - io_ctl->check_crcs = 1; + io_ctl->check_crcs = check_crcs; + return 0; } @@ -348,8 +361,8 @@ static int io_ctl_prepare_pages(struct io_ctl *io_ctl, struct inode *inode, btrfs_readpage(NULL, page); lock_page(page); if (!PageUptodate(page)) { - printk(KERN_ERR "btrfs: error reading free " - "space cache\n"); + btrfs_err(BTRFS_I(inode)->root->fs_info, + "error reading free space cache"); io_ctl_drop_pages(io_ctl); return -EIO; } @@ -406,7 +419,7 @@ static int io_ctl_check_generation(struct io_ctl *io_ctl, u64 generation) gen = io_ctl->cur; if (le64_to_cpu(*gen) != generation) { - printk_ratelimited(KERN_ERR "btrfs: space cache generation " + printk_ratelimited(KERN_ERR "BTRFS: space cache generation " "(%Lu) does not match inode (%Lu)\n", *gen, generation); io_ctl_unmap_page(io_ctl); @@ -464,7 +477,7 @@ static int io_ctl_check_crc(struct io_ctl *io_ctl, int index) PAGE_CACHE_SIZE - offset); btrfs_csum_final(crc, (char *)&crc); if (val != crc) { - printk_ratelimited(KERN_ERR "btrfs: csum mismatch on free " + printk_ratelimited(KERN_ERR "BTRFS: csum mismatch on free " "space cache\n"); io_ctl_unmap_page(io_ctl); return -EIO; @@ -667,6 +680,13 @@ static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode, generation = btrfs_free_space_generation(leaf, header); btrfs_release_path(path); + if (!BTRFS_I(inode)->generation) { + btrfs_info(root->fs_info, + "The free space cache file (%llu) is invalid. skip it\n", + offset); + return 0; + } + if (BTRFS_I(inode)->generation != generation) { btrfs_err(root->fs_info, "free space inode generation (%llu) " @@ -678,7 +698,7 @@ static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode, if (!num_entries) return 0; - ret = io_ctl_init(&io_ctl, inode, root); + ret = io_ctl_init(&io_ctl, inode, root, 0); if (ret) return ret; @@ -832,7 +852,7 @@ int load_free_space_cache(struct btrfs_fs_info *fs_info, if (!matched) { __btrfs_remove_free_space_cache(ctl); - btrfs_err(fs_info, "block group %llu has wrong amount of free space", + btrfs_warn(fs_info, "block group %llu has wrong amount of free space", block_group->key.objectid); ret = -1; } @@ -844,7 +864,7 @@ out: spin_unlock(&block_group->lock); ret = 0; - btrfs_err(fs_info, "failed to load free space cache for block group %llu", + btrfs_warn(fs_info, "failed to load free space cache for block group %llu, rebuild it now", block_group->key.objectid); } @@ -852,90 +872,44 @@ out: return ret; } -/** - * __btrfs_write_out_cache - write out cached info to an inode - * @root - the root the inode belongs to - * @ctl - the free space cache we are going to write out - * @block_group - the block_group for this cache if it belongs to a block_group - * @trans - the trans handle - * @path - the path to use - * @offset - the offset for the key we'll insert - * - * This function writes out a free space cache struct to disk for quick recovery - * on mount. This will return 0 if it was successfull in writing the cache out, - * and -1 if it was not. - */ -static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode, - struct btrfs_free_space_ctl *ctl, - struct btrfs_block_group_cache *block_group, - struct btrfs_trans_handle *trans, - struct btrfs_path *path, u64 offset) +static noinline_for_stack +int write_cache_extent_entries(struct io_ctl *io_ctl, + struct btrfs_free_space_ctl *ctl, + struct btrfs_block_group_cache *block_group, + int *entries, int *bitmaps, + struct list_head *bitmap_list) { - struct btrfs_free_space_header *header; - struct extent_buffer *leaf; - struct rb_node *node; - struct list_head *pos, *n; - struct extent_state *cached_state = NULL; - struct btrfs_free_cluster *cluster = NULL; - struct extent_io_tree *unpin = NULL; - struct io_ctl io_ctl; - struct list_head bitmap_list; - struct btrfs_key key; - u64 start, extent_start, extent_end, len; - int entries = 0; - int bitmaps = 0; int ret; - int err = -1; - - INIT_LIST_HEAD(&bitmap_list); - - if (!i_size_read(inode)) - return -1; - - ret = io_ctl_init(&io_ctl, inode, root); - if (ret) - return -1; + struct btrfs_free_cluster *cluster = NULL; + struct rb_node *node = rb_first(&ctl->free_space_offset); /* Get the cluster for this block_group if it exists */ - if (block_group && !list_empty(&block_group->cluster_list)) + if (block_group && !list_empty(&block_group->cluster_list)) { cluster = list_entry(block_group->cluster_list.next, struct btrfs_free_cluster, block_group_list); + } - /* Lock all pages first so we can lock the extent safely. */ - io_ctl_prepare_pages(&io_ctl, inode, 0); - - lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1, - 0, &cached_state); - - node = rb_first(&ctl->free_space_offset); if (!node && cluster) { node = rb_first(&cluster->root); cluster = NULL; } - /* Make sure we can fit our crcs into the first page */ - if (io_ctl.check_crcs && - (io_ctl.num_pages * sizeof(u32)) >= PAGE_CACHE_SIZE) - goto out_nospc; - - io_ctl_set_generation(&io_ctl, trans->transid); - /* Write out the extent entries */ while (node) { struct btrfs_free_space *e; e = rb_entry(node, struct btrfs_free_space, offset_index); - entries++; + *entries += 1; - ret = io_ctl_add_entry(&io_ctl, e->offset, e->bytes, + ret = io_ctl_add_entry(io_ctl, e->offset, e->bytes, e->bitmap); if (ret) - goto out_nospc; + goto fail; if (e->bitmap) { - list_add_tail(&e->list, &bitmap_list); - bitmaps++; + list_add_tail(&e->list, bitmap_list); + *bitmaps += 1; } node = rb_next(node); if (!node && cluster) { @@ -943,131 +917,289 @@ static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode, cluster = NULL; } } + return 0; +fail: + return -ENOSPC; +} + +static noinline_for_stack int +update_cache_item(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct inode *inode, + struct btrfs_path *path, u64 offset, + int entries, int bitmaps) +{ + struct btrfs_key key; + struct btrfs_free_space_header *header; + struct extent_buffer *leaf; + int ret; + + key.objectid = BTRFS_FREE_SPACE_OBJECTID; + key.offset = offset; + key.type = 0; + + ret = btrfs_search_slot(trans, root, &key, path, 0, 1); + if (ret < 0) { + clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1, + EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL, + GFP_NOFS); + goto fail; + } + leaf = path->nodes[0]; + if (ret > 0) { + struct btrfs_key found_key; + ASSERT(path->slots[0]); + path->slots[0]--; + btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); + if (found_key.objectid != BTRFS_FREE_SPACE_OBJECTID || + found_key.offset != offset) { + clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, + inode->i_size - 1, + EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, + NULL, GFP_NOFS); + btrfs_release_path(path); + goto fail; + } + } + + BTRFS_I(inode)->generation = trans->transid; + header = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_free_space_header); + btrfs_set_free_space_entries(leaf, header, entries); + btrfs_set_free_space_bitmaps(leaf, header, bitmaps); + btrfs_set_free_space_generation(leaf, header, trans->transid); + btrfs_mark_buffer_dirty(leaf); + btrfs_release_path(path); + + return 0; + +fail: + return -1; +} + +static noinline_for_stack int +write_pinned_extent_entries(struct btrfs_root *root, + struct btrfs_block_group_cache *block_group, + struct io_ctl *io_ctl, + int *entries) +{ + u64 start, extent_start, extent_end, len; + struct extent_io_tree *unpin = NULL; + int ret; + + if (!block_group) + return 0; /* * We want to add any pinned extents to our free space cache * so we don't leak the space - */ - - /* + * * We shouldn't have switched the pinned extents yet so this is the * right one */ unpin = root->fs_info->pinned_extents; - if (block_group) - start = block_group->key.objectid; + start = block_group->key.objectid; - while (block_group && (start < block_group->key.objectid + - block_group->key.offset)) { + while (start < block_group->key.objectid + block_group->key.offset) { ret = find_first_extent_bit(unpin, start, &extent_start, &extent_end, EXTENT_DIRTY, NULL); - if (ret) { - ret = 0; - break; - } + if (ret) + return 0; /* This pinned extent is out of our range */ if (extent_start >= block_group->key.objectid + block_group->key.offset) - break; + return 0; extent_start = max(extent_start, start); extent_end = min(block_group->key.objectid + block_group->key.offset, extent_end + 1); len = extent_end - extent_start; - entries++; - ret = io_ctl_add_entry(&io_ctl, extent_start, len, NULL); + *entries += 1; + ret = io_ctl_add_entry(io_ctl, extent_start, len, NULL); if (ret) - goto out_nospc; + return -ENOSPC; start = extent_end; } + return 0; +} + +static noinline_for_stack int +write_bitmap_entries(struct io_ctl *io_ctl, struct list_head *bitmap_list) +{ + struct list_head *pos, *n; + int ret; + /* Write out the bitmaps */ - list_for_each_safe(pos, n, &bitmap_list) { + list_for_each_safe(pos, n, bitmap_list) { struct btrfs_free_space *entry = list_entry(pos, struct btrfs_free_space, list); - ret = io_ctl_add_bitmap(&io_ctl, entry->bitmap); + ret = io_ctl_add_bitmap(io_ctl, entry->bitmap); if (ret) - goto out_nospc; + return -ENOSPC; list_del_init(&entry->list); } - /* Zero out the rest of the pages just to make sure */ - io_ctl_zero_remaining_pages(&io_ctl); + return 0; +} - ret = btrfs_dirty_pages(root, inode, io_ctl.pages, io_ctl.num_pages, - 0, i_size_read(inode), &cached_state); - io_ctl_drop_pages(&io_ctl); - unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0, - i_size_read(inode) - 1, &cached_state, GFP_NOFS); +static int flush_dirty_cache(struct inode *inode) +{ + int ret; + ret = btrfs_wait_ordered_range(inode, 0, (u64)-1); if (ret) - goto out; - - - btrfs_wait_ordered_range(inode, 0, (u64)-1); - - key.objectid = BTRFS_FREE_SPACE_OBJECTID; - key.offset = offset; - key.type = 0; - - ret = btrfs_search_slot(trans, root, &key, path, 0, 1); - if (ret < 0) { clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1, EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL, GFP_NOFS); - goto out; + + return ret; +} + +static void noinline_for_stack +cleanup_write_cache_enospc(struct inode *inode, + struct io_ctl *io_ctl, + struct extent_state **cached_state, + struct list_head *bitmap_list) +{ + struct list_head *pos, *n; + + list_for_each_safe(pos, n, bitmap_list) { + struct btrfs_free_space *entry = + list_entry(pos, struct btrfs_free_space, list); + list_del_init(&entry->list); } - leaf = path->nodes[0]; - if (ret > 0) { - struct btrfs_key found_key; - ASSERT(path->slots[0]); - path->slots[0]--; - btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); - if (found_key.objectid != BTRFS_FREE_SPACE_OBJECTID || - found_key.offset != offset) { - clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, - inode->i_size - 1, - EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, - NULL, GFP_NOFS); - btrfs_release_path(path); + io_ctl_drop_pages(io_ctl); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0, + i_size_read(inode) - 1, cached_state, + GFP_NOFS); +} + +/** + * __btrfs_write_out_cache - write out cached info to an inode + * @root - the root the inode belongs to + * @ctl - the free space cache we are going to write out + * @block_group - the block_group for this cache if it belongs to a block_group + * @trans - the trans handle + * @path - the path to use + * @offset - the offset for the key we'll insert + * + * This function writes out a free space cache struct to disk for quick recovery + * on mount. This will return 0 if it was successfull in writing the cache out, + * and -1 if it was not. + */ +static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode, + struct btrfs_free_space_ctl *ctl, + struct btrfs_block_group_cache *block_group, + struct btrfs_trans_handle *trans, + struct btrfs_path *path, u64 offset) +{ + struct extent_state *cached_state = NULL; + struct io_ctl io_ctl; + LIST_HEAD(bitmap_list); + int entries = 0; + int bitmaps = 0; + int ret; + + if (!i_size_read(inode)) + return -1; + + ret = io_ctl_init(&io_ctl, inode, root, 1); + if (ret) + return -1; + + if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA)) { + down_write(&block_group->data_rwsem); + spin_lock(&block_group->lock); + if (block_group->delalloc_bytes) { + block_group->disk_cache_state = BTRFS_DC_WRITTEN; + spin_unlock(&block_group->lock); + up_write(&block_group->data_rwsem); + BTRFS_I(inode)->generation = 0; + ret = 0; goto out; } + spin_unlock(&block_group->lock); } - BTRFS_I(inode)->generation = trans->transid; - header = btrfs_item_ptr(leaf, path->slots[0], - struct btrfs_free_space_header); - btrfs_set_free_space_entries(leaf, header, entries); - btrfs_set_free_space_bitmaps(leaf, header, bitmaps); - btrfs_set_free_space_generation(leaf, header, trans->transid); - btrfs_mark_buffer_dirty(leaf); - btrfs_release_path(path); + /* Lock all pages first so we can lock the extent safely. */ + io_ctl_prepare_pages(&io_ctl, inode, 0); + + lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1, + 0, &cached_state); - err = 0; + io_ctl_set_generation(&io_ctl, trans->transid); + + /* Write out the extent entries in the free space cache */ + ret = write_cache_extent_entries(&io_ctl, ctl, + block_group, &entries, &bitmaps, + &bitmap_list); + if (ret) + goto out_nospc; + + /* + * Some spaces that are freed in the current transaction are pinned, + * they will be added into free space cache after the transaction is + * committed, we shouldn't lose them. + */ + ret = write_pinned_extent_entries(root, block_group, &io_ctl, &entries); + if (ret) + goto out_nospc; + + /* At last, we write out all the bitmaps. */ + ret = write_bitmap_entries(&io_ctl, &bitmap_list); + if (ret) + goto out_nospc; + + /* Zero out the rest of the pages just to make sure */ + io_ctl_zero_remaining_pages(&io_ctl); + + /* Everything is written out, now we dirty the pages in the file. */ + ret = btrfs_dirty_pages(root, inode, io_ctl.pages, io_ctl.num_pages, + 0, i_size_read(inode), &cached_state); + if (ret) + goto out_nospc; + + if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA)) + up_write(&block_group->data_rwsem); + /* + * Release the pages and unlock the extent, we will flush + * them out later + */ + io_ctl_drop_pages(&io_ctl); + + unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0, + i_size_read(inode) - 1, &cached_state, GFP_NOFS); + + /* Flush the dirty pages in the cache file. */ + ret = flush_dirty_cache(inode); + if (ret) + goto out; + + /* Update the cache item to tell everyone this cache file is valid. */ + ret = update_cache_item(trans, root, inode, path, offset, + entries, bitmaps); out: io_ctl_free(&io_ctl); - if (err) { + if (ret) { invalidate_inode_pages2(inode->i_mapping); BTRFS_I(inode)->generation = 0; } btrfs_update_inode(trans, root, inode); - return err; + return ret; out_nospc: - list_for_each_safe(pos, n, &bitmap_list) { - struct btrfs_free_space *entry = - list_entry(pos, struct btrfs_free_space, list); - list_del_init(&entry->list); - } - io_ctl_drop_pages(&io_ctl); - unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0, - i_size_read(inode) - 1, &cached_state, GFP_NOFS); + cleanup_write_cache_enospc(inode, &io_ctl, &cached_state, &bitmap_list); + + if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA)) + up_write(&block_group->data_rwsem); + goto out; } @@ -1087,6 +1219,12 @@ int btrfs_write_out_cache(struct btrfs_root *root, spin_unlock(&block_group->lock); return 0; } + + if (block_group->delalloc_bytes) { + block_group->disk_cache_state = BTRFS_DC_WRITTEN; + spin_unlock(&block_group->lock); + return 0; + } spin_unlock(&block_group->lock); inode = lookup_free_space_inode(root, block_group, path); @@ -1898,7 +2036,7 @@ out: spin_unlock(&ctl->tree_lock); if (ret) { - printk(KERN_CRIT "btrfs: unable to add free space :%d\n", ret); + printk(KERN_CRIT "BTRFS: unable to add free space :%d\n", ret); ASSERT(ret != -EEXIST); } @@ -2007,14 +2145,15 @@ void btrfs_dump_free_space(struct btrfs_block_group_cache *block_group, info = rb_entry(n, struct btrfs_free_space, offset_index); if (info->bytes >= bytes && !block_group->ro) count++; - printk(KERN_CRIT "entry offset %llu, bytes %llu, bitmap %s\n", - info->offset, info->bytes, + btrfs_crit(block_group->fs_info, + "entry offset %llu, bytes %llu, bitmap %s", + info->offset, info->bytes, (info->bitmap) ? "yes" : "no"); } - printk(KERN_INFO "block group has cluster?: %s\n", + btrfs_info(block_group->fs_info, "block group has cluster?: %s", list_empty(&block_group->cluster_list) ? "no" : "yes"); - printk(KERN_INFO "%d blocks of free space at or bigger than bytes is" - "\n", count); + btrfs_info(block_group->fs_info, + "%d blocks of free space at or bigger than bytes is", count); } void btrfs_init_free_space_ctl(struct btrfs_block_group_cache *block_group) @@ -2276,7 +2415,7 @@ u64 btrfs_alloc_from_cluster(struct btrfs_block_group_cache *block_group, goto out; entry = rb_entry(node, struct btrfs_free_space, offset_index); - while(1) { + while (1) { if (entry->bytes < bytes && entry->bytes > *max_extent_size) *max_extent_size = entry->bytes; @@ -2417,7 +2556,6 @@ setup_cluster_no_bitmap(struct btrfs_block_group_cache *block_group, struct btrfs_free_space *entry = NULL; struct btrfs_free_space *last; struct rb_node *node; - u64 window_start; u64 window_free; u64 max_extent; u64 total_size = 0; @@ -2439,7 +2577,6 @@ setup_cluster_no_bitmap(struct btrfs_block_group_cache *block_group, entry = rb_entry(node, struct btrfs_free_space, offset_index); } - window_start = entry->offset; window_free = entry->bytes; max_extent = entry->bytes; first = entry; @@ -2967,19 +3104,15 @@ out: int btrfs_write_out_ino_cache(struct btrfs_root *root, struct btrfs_trans_handle *trans, - struct btrfs_path *path) + struct btrfs_path *path, + struct inode *inode) { struct btrfs_free_space_ctl *ctl = root->free_ino_ctl; - struct inode *inode; int ret; if (!btrfs_test_opt(root, INODE_MAP_CACHE)) return 0; - inode = lookup_free_ino_inode(root, path); - if (IS_ERR(inode)) - return 0; - ret = __btrfs_write_out_cache(root, inode, ctl, NULL, trans, path, 0); if (ret) { btrfs_delalloc_release_metadata(inode, inode->i_size); @@ -2990,7 +3123,6 @@ int btrfs_write_out_ino_cache(struct btrfs_root *root, #endif } - iput(inode); return ret; } |
