diff options
Diffstat (limited to 'fs/btrfs')
33 files changed, 4032 insertions, 1684 deletions
diff --git a/fs/btrfs/acl.c b/fs/btrfs/acl.c index f128427b995b..6df6d6ed74fd 100644 --- a/fs/btrfs/acl.c +++ b/fs/btrfs/acl.c @@ -27,7 +27,7 @@ #include "btrfs_inode.h" #include "xattr.h" -#ifdef CONFIG_FS_POSIX_ACL +#ifdef CONFIG_BTRFS_FS_POSIX_ACL static struct posix_acl *btrfs_get_acl(struct inode *inode, int type) { @@ -73,13 +73,13 @@ static struct posix_acl *btrfs_get_acl(struct inode *inode, int type) return acl; } -static int btrfs_xattr_get_acl(struct inode *inode, int type, - void *value, size_t size) +static int btrfs_xattr_acl_get(struct dentry *dentry, const char *name, + void *value, size_t size, int type) { struct posix_acl *acl; int ret = 0; - acl = btrfs_get_acl(inode, type); + acl = btrfs_get_acl(dentry->d_inode, type); if (IS_ERR(acl)) return PTR_ERR(acl); @@ -94,7 +94,8 @@ static int btrfs_xattr_get_acl(struct inode *inode, int type, /* * Needs to be called with fs_mutex held */ -static int btrfs_set_acl(struct inode *inode, struct posix_acl *acl, int type) +static int btrfs_set_acl(struct btrfs_trans_handle *trans, + struct inode *inode, struct posix_acl *acl, int type) { int ret, size = 0; const char *name; @@ -111,12 +112,14 @@ static int btrfs_set_acl(struct inode *inode, struct posix_acl *acl, int type) switch (type) { case ACL_TYPE_ACCESS: mode = inode->i_mode; - ret = posix_acl_equiv_mode(acl, &mode); - if (ret < 0) - return ret; - ret = 0; - inode->i_mode = mode; name = POSIX_ACL_XATTR_ACCESS; + if (acl) { + ret = posix_acl_equiv_mode(acl, &mode); + if (ret < 0) + return ret; + inode->i_mode = mode; + } + ret = 0; break; case ACL_TYPE_DEFAULT: if (!S_ISDIR(inode->i_mode)) @@ -140,8 +143,7 @@ static int btrfs_set_acl(struct inode *inode, struct posix_acl *acl, int type) goto out; } - ret = __btrfs_setxattr(inode, name, value, size, 0); - + ret = __btrfs_setxattr(trans, inode, name, value, size, 0); out: kfree(value); @@ -151,10 +153,10 @@ out: return ret; } -static int btrfs_xattr_set_acl(struct inode *inode, int type, - const void *value, size_t size) +static int btrfs_xattr_acl_set(struct dentry *dentry, const char *name, + const void *value, size_t size, int flags, int type) { - int ret = 0; + int ret; struct posix_acl *acl = NULL; if (value) { @@ -167,38 +169,13 @@ static int btrfs_xattr_set_acl(struct inode *inode, int type, } } - ret = btrfs_set_acl(inode, acl, type); + ret = btrfs_set_acl(NULL, dentry->d_inode, acl, type); posix_acl_release(acl); return ret; } - -static int btrfs_xattr_acl_access_get(struct inode *inode, const char *name, - void *value, size_t size) -{ - return btrfs_xattr_get_acl(inode, ACL_TYPE_ACCESS, value, size); -} - -static int btrfs_xattr_acl_access_set(struct inode *inode, const char *name, - const void *value, size_t size, int flags) -{ - return btrfs_xattr_set_acl(inode, ACL_TYPE_ACCESS, value, size); -} - -static int btrfs_xattr_acl_default_get(struct inode *inode, const char *name, - void *value, size_t size) -{ - return btrfs_xattr_get_acl(inode, ACL_TYPE_DEFAULT, value, size); -} - -static int btrfs_xattr_acl_default_set(struct inode *inode, const char *name, - const void *value, size_t size, int flags) -{ - return btrfs_xattr_set_acl(inode, ACL_TYPE_DEFAULT, value, size); -} - int btrfs_check_acl(struct inode *inode, int mask) { struct posix_acl *acl; @@ -221,7 +198,8 @@ int btrfs_check_acl(struct inode *inode, int mask) * stuff has been fixed to work with that. If the locking stuff changes, we * need to re-evaluate the acl locking stuff. */ -int btrfs_init_acl(struct inode *inode, struct inode *dir) +int btrfs_init_acl(struct btrfs_trans_handle *trans, + struct inode *inode, struct inode *dir) { struct posix_acl *acl = NULL; int ret = 0; @@ -246,7 +224,8 @@ int btrfs_init_acl(struct inode *inode, struct inode *dir) mode_t mode; if (S_ISDIR(inode->i_mode)) { - ret = btrfs_set_acl(inode, acl, ACL_TYPE_DEFAULT); + ret = btrfs_set_acl(trans, inode, acl, + ACL_TYPE_DEFAULT); if (ret) goto failed; } @@ -261,10 +240,11 @@ int btrfs_init_acl(struct inode *inode, struct inode *dir) inode->i_mode = mode; if (ret > 0) { /* we need an acl */ - ret = btrfs_set_acl(inode, clone, + ret = btrfs_set_acl(trans, inode, clone, ACL_TYPE_ACCESS); } } + posix_acl_release(clone); } failed: posix_acl_release(acl); @@ -294,7 +274,7 @@ int btrfs_acl_chmod(struct inode *inode) ret = posix_acl_chmod_masq(clone, inode->i_mode); if (!ret) - ret = btrfs_set_acl(inode, clone, ACL_TYPE_ACCESS); + ret = btrfs_set_acl(NULL, inode, clone, ACL_TYPE_ACCESS); posix_acl_release(clone); @@ -303,26 +283,29 @@ int btrfs_acl_chmod(struct inode *inode) struct xattr_handler btrfs_xattr_acl_default_handler = { .prefix = POSIX_ACL_XATTR_DEFAULT, - .get = btrfs_xattr_acl_default_get, - .set = btrfs_xattr_acl_default_set, + .flags = ACL_TYPE_DEFAULT, + .get = btrfs_xattr_acl_get, + .set = btrfs_xattr_acl_set, }; struct xattr_handler btrfs_xattr_acl_access_handler = { .prefix = POSIX_ACL_XATTR_ACCESS, - .get = btrfs_xattr_acl_access_get, - .set = btrfs_xattr_acl_access_set, + .flags = ACL_TYPE_ACCESS, + .get = btrfs_xattr_acl_get, + .set = btrfs_xattr_acl_set, }; -#else /* CONFIG_FS_POSIX_ACL */ +#else /* CONFIG_BTRFS_FS_POSIX_ACL */ int btrfs_acl_chmod(struct inode *inode) { return 0; } -int btrfs_init_acl(struct inode *inode, struct inode *dir) +int btrfs_init_acl(struct btrfs_trans_handle *trans, + struct inode *inode, struct inode *dir) { return 0; } -#endif /* CONFIG_FS_POSIX_ACL */ +#endif /* CONFIG_BTRFS_FS_POSIX_ACL */ diff --git a/fs/btrfs/async-thread.c b/fs/btrfs/async-thread.c index 282ca085c2fb..c0861e781cdb 100644 --- a/fs/btrfs/async-thread.c +++ b/fs/btrfs/async-thread.c @@ -64,6 +64,51 @@ struct btrfs_worker_thread { }; /* + * btrfs_start_workers uses kthread_run, which can block waiting for memory + * for a very long time. It will actually throttle on page writeback, + * and so it may not make progress until after our btrfs worker threads + * process all of the pending work structs in their queue + * + * This means we can't use btrfs_start_workers from inside a btrfs worker + * thread that is used as part of cleaning dirty memory, which pretty much + * involves all of the worker threads. + * + * Instead we have a helper queue who never has more than one thread + * where we scheduler thread start operations. This worker_start struct + * is used to contain the work and hold a pointer to the queue that needs + * another worker. + */ +struct worker_start { + struct btrfs_work work; + struct btrfs_workers *queue; +}; + +static void start_new_worker_func(struct btrfs_work *work) +{ + struct worker_start *start; + start = container_of(work, struct worker_start, work); + btrfs_start_workers(start->queue, 1); + kfree(start); +} + +static int start_new_worker(struct btrfs_workers *queue) +{ + struct worker_start *start; + int ret; + + start = kzalloc(sizeof(*start), GFP_NOFS); + if (!start) + return -ENOMEM; + + start->work.func = start_new_worker_func; + start->queue = queue; + ret = btrfs_queue_worker(queue->atomic_worker_start, &start->work); + if (ret) + kfree(start); + return ret; +} + +/* * helper function to move a thread onto the idle list after it * has finished some requests. */ @@ -118,11 +163,13 @@ static void check_pending_worker_creates(struct btrfs_worker_thread *worker) goto out; workers->atomic_start_pending = 0; - if (workers->num_workers >= workers->max_workers) + if (workers->num_workers + workers->num_workers_starting >= + workers->max_workers) goto out; + workers->num_workers_starting += 1; spin_unlock_irqrestore(&workers->lock, flags); - btrfs_start_workers(workers, 1); + start_new_worker(workers); return; out: @@ -390,9 +437,11 @@ int btrfs_stop_workers(struct btrfs_workers *workers) /* * simple init on struct btrfs_workers */ -void btrfs_init_workers(struct btrfs_workers *workers, char *name, int max) +void btrfs_init_workers(struct btrfs_workers *workers, char *name, int max, + struct btrfs_workers *async_helper) { workers->num_workers = 0; + workers->num_workers_starting = 0; INIT_LIST_HEAD(&workers->worker_list); INIT_LIST_HEAD(&workers->idle_list); INIT_LIST_HEAD(&workers->order_list); @@ -404,14 +453,15 @@ void btrfs_init_workers(struct btrfs_workers *workers, char *name, int max) workers->name = name; workers->ordered = 0; workers->atomic_start_pending = 0; - workers->atomic_worker_start = 0; + workers->atomic_worker_start = async_helper; } /* * starts new worker threads. This does not enforce the max worker * count in case you need to temporarily go past it. */ -int btrfs_start_workers(struct btrfs_workers *workers, int num_workers) +static int __btrfs_start_workers(struct btrfs_workers *workers, + int num_workers) { struct btrfs_worker_thread *worker; int ret = 0; @@ -444,6 +494,8 @@ int btrfs_start_workers(struct btrfs_workers *workers, int num_workers) list_add_tail(&worker->worker_list, &workers->idle_list); worker->idle = 1; workers->num_workers++; + workers->num_workers_starting--; + WARN_ON(workers->num_workers_starting < 0); spin_unlock_irq(&workers->lock); } return 0; @@ -452,6 +504,14 @@ fail: return ret; } +int btrfs_start_workers(struct btrfs_workers *workers, int num_workers) +{ + spin_lock_irq(&workers->lock); + workers->num_workers_starting += num_workers; + spin_unlock_irq(&workers->lock); + return __btrfs_start_workers(workers, num_workers); +} + /* * run through the list and find a worker thread that doesn't have a lot * to do right now. This can return null if we aren't yet at the thread @@ -461,7 +521,10 @@ static struct btrfs_worker_thread *next_worker(struct btrfs_workers *workers) { struct btrfs_worker_thread *worker; struct list_head *next; - int enforce_min = workers->num_workers < workers->max_workers; + int enforce_min; + + enforce_min = (workers->num_workers + workers->num_workers_starting) < + workers->max_workers; /* * if we find an idle thread, don't move it to the end of the @@ -509,15 +572,17 @@ again: worker = next_worker(workers); if (!worker) { - if (workers->num_workers >= workers->max_workers) { + if (workers->num_workers + workers->num_workers_starting >= + workers->max_workers) { goto fallback; } else if (workers->atomic_worker_start) { workers->atomic_start_pending = 1; goto fallback; } else { + workers->num_workers_starting++; spin_unlock_irqrestore(&workers->lock, flags); /* we're below the limit, start another worker */ - btrfs_start_workers(workers, 1); + __btrfs_start_workers(workers, 1); goto again; } } diff --git a/fs/btrfs/async-thread.h b/fs/btrfs/async-thread.h index fc089b95ec14..5077746cf85e 100644 --- a/fs/btrfs/async-thread.h +++ b/fs/btrfs/async-thread.h @@ -64,6 +64,8 @@ struct btrfs_workers { /* current number of running workers */ int num_workers; + int num_workers_starting; + /* max number of workers allowed. changed by btrfs_start_workers */ int max_workers; @@ -78,9 +80,10 @@ struct btrfs_workers { /* * are we allowed to sleep while starting workers or are we required - * to start them at a later time? + * to start them at a later time? If we can't sleep, this indicates + * which queue we need to use to schedule thread creation. */ - int atomic_worker_start; + struct btrfs_workers *atomic_worker_start; /* list with all the work threads. The workers on the idle thread * may be actively servicing jobs, but they haven't yet hit the @@ -109,7 +112,8 @@ struct btrfs_workers { int btrfs_queue_worker(struct btrfs_workers *workers, struct btrfs_work *work); int btrfs_start_workers(struct btrfs_workers *workers, int num_workers); int btrfs_stop_workers(struct btrfs_workers *workers); -void btrfs_init_workers(struct btrfs_workers *workers, char *name, int max); +void btrfs_init_workers(struct btrfs_workers *workers, char *name, int max, + struct btrfs_workers *async_starter); int btrfs_requeue_work(struct btrfs_work *work); void btrfs_set_work_high_prio(struct btrfs_work *work); #endif diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h index 82ee56bba299..7a4dee199832 100644 --- a/fs/btrfs/btrfs_inode.h +++ b/fs/btrfs/btrfs_inode.h @@ -44,9 +44,6 @@ struct btrfs_inode { */ struct extent_io_tree io_failure_tree; - /* held while inesrting or deleting extents from files */ - struct mutex extent_mutex; - /* held while logging the inode in tree-log.c */ struct mutex log_mutex; @@ -86,6 +83,12 @@ struct btrfs_inode { * transid of the trans_handle that last modified this inode */ u64 last_trans; + + /* + * log transid when this inode was last modified + */ + u64 last_sub_trans; + /* * transid that last logged this inode */ @@ -128,6 +131,16 @@ struct btrfs_inode { u64 last_unlink_trans; /* + * Counters to keep track of the number of extent item's we may use due + * to delalloc and such. outstanding_extents is the number of extent + * items we think we'll end up using, and reserved_extents is the number + * of extent items we've reserved metadata for. + */ + spinlock_t accounting_lock; + int reserved_extents; + int outstanding_extents; + + /* * ordered_data_close is set by truncate when a file that used * to have good data has been truncated to zero. When it is set * the btrfs file release call will add this inode to the @@ -140,6 +153,11 @@ struct btrfs_inode { unsigned ordered_data_close:1; unsigned dummy_inode:1; + /* + * always compress this one file + */ + unsigned force_compress:1; + struct inode vfs_inode; }; @@ -150,7 +168,7 @@ static inline struct btrfs_inode *BTRFS_I(struct inode *inode) static inline void btrfs_i_size_write(struct inode *inode, u64 size) { - inode->i_size = size; + i_size_write(inode, size); BTRFS_I(inode)->disk_i_size = size; } diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c index a11a32058b50..28b92a7218ab 100644 --- a/fs/btrfs/compression.c +++ b/fs/btrfs/compression.c @@ -478,7 +478,7 @@ static noinline int add_ra_bio_pages(struct inode *inode, goto next; } - page = alloc_page(mapping_gfp_mask(mapping) | GFP_NOFS); + page = alloc_page(mapping_gfp_mask(mapping) & ~__GFP_FS); if (!page) break; diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c index ec96f3a6d536..c4bc570a396e 100644 --- a/fs/btrfs/ctree.c +++ b/fs/btrfs/ctree.c @@ -37,6 +37,11 @@ static int balance_node_right(struct btrfs_trans_handle *trans, struct extent_buffer *src_buf); static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_path *path, int level, int slot); +static int setup_items_for_insert(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct btrfs_path *path, + struct btrfs_key *cpu_key, u32 *data_size, + u32 total_data, u32 total_size, int nr); + struct btrfs_path *btrfs_alloc_path(void) { @@ -451,9 +456,8 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans, extent_buffer_get(cow); spin_unlock(&root->node_lock); - btrfs_free_extent(trans, root, buf->start, buf->len, - parent_start, root->root_key.objectid, - level, 0); + btrfs_free_tree_block(trans, root, buf->start, buf->len, + parent_start, root->root_key.objectid, level); free_extent_buffer(buf); add_root_to_dirty_list(root); } else { @@ -468,9 +472,8 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans, btrfs_set_node_ptr_generation(parent, parent_slot, trans->transid); btrfs_mark_buffer_dirty(parent); - btrfs_free_extent(trans, root, buf->start, buf->len, - parent_start, root->root_key.objectid, - level, 0); + btrfs_free_tree_block(trans, root, buf->start, buf->len, + parent_start, root->root_key.objectid, level); } if (unlock_orig) btrfs_tree_unlock(buf); @@ -1030,8 +1033,8 @@ static noinline int balance_level(struct btrfs_trans_handle *trans, btrfs_tree_unlock(mid); /* once for the path */ free_extent_buffer(mid); - ret = btrfs_free_extent(trans, root, mid->start, mid->len, - 0, root->root_key.objectid, level, 1); + ret = btrfs_free_tree_block(trans, root, mid->start, mid->len, + 0, root->root_key.objectid, level); /* once for the root ptr */ free_extent_buffer(mid); return ret; @@ -1095,10 +1098,10 @@ static noinline int balance_level(struct btrfs_trans_handle *trans, 1); if (wret) ret = wret; - wret = btrfs_free_extent(trans, root, bytenr, - blocksize, 0, - root->root_key.objectid, - level, 0); + wret = btrfs_free_tree_block(trans, root, + bytenr, blocksize, 0, + root->root_key.objectid, + level); if (wret) ret = wret; } else { @@ -1143,9 +1146,8 @@ static noinline int balance_level(struct btrfs_trans_handle *trans, wret = del_ptr(trans, root, path, level + 1, pslot); if (wret) ret = wret; - wret = btrfs_free_extent(trans, root, bytenr, blocksize, - 0, root->root_key.objectid, - level, 0); + wret = btrfs_free_tree_block(trans, root, bytenr, blocksize, + 0, root->root_key.objectid, level); if (wret) ret = wret; } else { @@ -2997,75 +2999,85 @@ again: return ret; } -/* - * This function splits a single item into two items, - * giving 'new_key' to the new item and splitting the - * old one at split_offset (from the start of the item). - * - * The path may be released by this operation. After - * the split, the path is pointing to the old item. The - * new item is going to be in the same node as the old one. - * - * Note, the item being split must be smaller enough to live alone on - * a tree block with room for one extra struct btrfs_item - * - * This allows us to split the item in place, keeping a lock on the - * leaf the entire time. - */ -int btrfs_split_item(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct btrfs_path *path, - struct btrfs_key *new_key, - unsigned long split_offset) +static noinline int setup_leaf_for_split(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, int ins_len) { - u32 item_size; + struct btrfs_key key; struct extent_buffer *leaf; - struct btrfs_key orig_key; - struct btrfs_item *item; - struct btrfs_item *new_item; - int ret = 0; - int slot; - u32 nritems; - u32 orig_offset; - struct btrfs_disk_key disk_key; - char *buf; + struct btrfs_file_extent_item *fi; + u64 extent_len = 0; + u32 item_size; + int ret; leaf = path->nodes[0]; - btrfs_item_key_to_cpu(leaf, &orig_key, path->slots[0]); - if (btrfs_leaf_free_space(root, leaf) >= sizeof(struct btrfs_item)) - goto split; + btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); + + BUG_ON(key.type != BTRFS_EXTENT_DATA_KEY && + key.type != BTRFS_EXTENT_CSUM_KEY); + + if (btrfs_leaf_free_space(root, leaf) >= ins_len) + return 0; item_size = btrfs_item_size_nr(leaf, path->slots[0]); + if (key.type == BTRFS_EXTENT_DATA_KEY) { + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + extent_len = btrfs_file_extent_num_bytes(leaf, fi); + } btrfs_release_path(root, path); - path->search_for_split = 1; path->keep_locks = 1; - - ret = btrfs_search_slot(trans, root, &orig_key, path, 0, 1); + path->search_for_split = 1; + ret = btrfs_search_slot(trans, root, &key, path, 0, 1); path->search_for_split = 0; + if (ret < 0) + goto err; + ret = -EAGAIN; + leaf = path->nodes[0]; /* if our item isn't there or got smaller, return now */ - if (ret != 0 || item_size != btrfs_item_size_nr(path->nodes[0], - path->slots[0])) { - path->keep_locks = 0; - return -EAGAIN; + if (ret > 0 || item_size != btrfs_item_size_nr(leaf, path->slots[0])) + goto err; + + if (key.type == BTRFS_EXTENT_DATA_KEY) { + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + if (extent_len != btrfs_file_extent_num_bytes(leaf, fi)) + goto err; } btrfs_set_path_blocking(path); - ret = split_leaf(trans, root, &orig_key, path, - sizeof(struct btrfs_item), 1); - path->keep_locks = 0; + ret = split_leaf(trans, root, &key, path, ins_len, 1); BUG_ON(ret); + path->keep_locks = 0; btrfs_unlock_up_safe(path, 1); + return 0; +err: + path->keep_locks = 0; + return ret; +} + +static noinline int split_item(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, + struct btrfs_key *new_key, + unsigned long split_offset) +{ + struct extent_buffer *leaf; + struct btrfs_item *item; + struct btrfs_item *new_item; + int slot; + char *buf; + u32 nritems; + u32 item_size; + u32 orig_offset; + struct btrfs_disk_key disk_key; + leaf = path->nodes[0]; BUG_ON(btrfs_leaf_free_space(root, leaf) < sizeof(struct btrfs_item)); -split: - /* - * make sure any changes to the path from split_leaf leave it - * in a blocking state - */ btrfs_set_path_blocking(path); item = btrfs_item_nr(leaf, path->slots[0]); @@ -3073,19 +3085,19 @@ split: item_size = btrfs_item_size(leaf, item); buf = kmalloc(item_size, GFP_NOFS); + if (!buf) + return -ENOMEM; + read_extent_buffer(leaf, buf, btrfs_item_ptr_offset(leaf, path->slots[0]), item_size); - slot = path->slots[0] + 1; - leaf = path->nodes[0]; + slot = path->slots[0] + 1; nritems = btrfs_header_nritems(leaf); - if (slot != nritems) { /* shift the items */ memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + 1), - btrfs_item_nr_offset(slot), - (nritems - slot) * sizeof(struct btrfs_item)); - + btrfs_item_nr_offset(slot), + (nritems - slot) * sizeof(struct btrfs_item)); } btrfs_cpu_key_to_disk(&disk_key, new_key); @@ -3113,16 +3125,81 @@ split: item_size - split_offset); btrfs_mark_buffer_dirty(leaf); - ret = 0; - if (btrfs_leaf_free_space(root, leaf) < 0) { - btrfs_print_leaf(root, leaf); - BUG(); - } + BUG_ON(btrfs_leaf_free_space(root, leaf) < 0); kfree(buf); + return 0; +} + +/* + * This function splits a single item into two items, + * giving 'new_key' to the new item and splitting the + * old one at split_offset (from the start of the item). + * + * The path may be released by this operation. After + * the split, the path is pointing to the old item. The + * new item is going to be in the same node as the old one. + * + * Note, the item being split must be smaller enough to live alone on + * a tree block with room for one extra struct btrfs_item + * + * This allows us to split the item in place, keeping a lock on the + * leaf the entire time. + */ +int btrfs_split_item(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, + struct btrfs_key *new_key, + unsigned long split_offset) +{ + int ret; + ret = setup_leaf_for_split(trans, root, path, + sizeof(struct btrfs_item)); + if (ret) + return ret; + + ret = split_item(trans, root, path, new_key, split_offset); return ret; } /* + * This function duplicate a item, giving 'new_key' to the new item. + * It guarantees both items live in the same tree leaf and the new item + * is contiguous with the original item. + * + * This allows us to split file extent in place, keeping a lock on the + * leaf the entire time. + */ +int btrfs_duplicate_item(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, + struct btrfs_key *new_key) +{ + struct extent_buffer *leaf; + int ret; + u32 item_size; + + leaf = path->nodes[0]; + item_size = btrfs_item_size_nr(leaf, path->slots[0]); + ret = setup_leaf_for_split(trans, root, path, + item_size + sizeof(struct btrfs_item)); + if (ret) + return ret; + + path->slots[0]++; + ret = setup_items_for_insert(trans, root, path, new_key, &item_size, + item_size, item_size + + sizeof(struct btrfs_item), 1); + BUG_ON(ret); + + leaf = path->nodes[0]; + memcpy_extent_buffer(leaf, + btrfs_item_ptr_offset(leaf, path->slots[0]), + btrfs_item_ptr_offset(leaf, path->slots[0] - 1), + item_size); + return 0; +} + +/* * make the item pointed to by the path smaller. new_size indicates * how small to make it, and from_end tells us if we just chop bytes * off the end of the item or if we shift the item to chop bytes off @@ -3714,8 +3791,8 @@ static noinline int btrfs_del_leaf(struct btrfs_trans_handle *trans, */ btrfs_unlock_up_safe(path, 0); - ret = btrfs_free_extent(trans, root, leaf->start, leaf->len, - 0, root->root_key.objectid, 0, 0); + ret = btrfs_free_tree_block(trans, root, leaf->start, leaf->len, + 0, root->root_key.objectid, 0); return ret; } /* diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 80599b4e42bd..0af2e3868573 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -310,6 +310,9 @@ struct btrfs_header { #define BTRFS_MAX_INLINE_DATA_SIZE(r) (BTRFS_LEAF_DATA_SIZE(r) - \ sizeof(struct btrfs_item) - \ sizeof(struct btrfs_file_extent_item)) +#define BTRFS_MAX_XATTR_SIZE(r) (BTRFS_LEAF_DATA_SIZE(r) - \ + sizeof(struct btrfs_item) -\ + sizeof(struct btrfs_dir_item)) /* @@ -370,11 +373,13 @@ struct btrfs_super_block { * ones specified below then we will fail to mount */ #define BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF (1ULL << 0) +#define BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL (2ULL << 0) #define BTRFS_FEATURE_COMPAT_SUPP 0ULL #define BTRFS_FEATURE_COMPAT_RO_SUPP 0ULL #define BTRFS_FEATURE_INCOMPAT_SUPP \ - BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF + (BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF | \ + BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL) /* * A leaf is full of items. offset and size tell us where to find @@ -675,21 +680,28 @@ struct btrfs_space_info { current allocations */ u64 bytes_readonly; /* total bytes that are read only */ u64 bytes_super; /* total bytes reserved for the super blocks */ - - /* delalloc accounting */ - u64 bytes_delalloc; /* number of bytes reserved for allocation, - this space is not necessarily reserved yet - by the allocator */ + u64 bytes_root; /* the number of bytes needed to commit a + transaction */ u64 bytes_may_use; /* number of bytes that may be used for - delalloc */ + delalloc/allocations */ + u64 bytes_delalloc; /* number of bytes currently reserved for + delayed allocation */ int full; /* indicates that we cannot allocate any more chunks for this space */ int force_alloc; /* set if we need to force a chunk alloc for this space */ + int force_delalloc; /* make people start doing filemap_flush until + we're under a threshold */ struct list_head list; + /* for controlling how we free up space for allocations */ + wait_queue_head_t allocate_wait; + wait_queue_head_t flush_wait; + int allocating_chunk; + int flushing; + /* for block groups in our same type */ struct list_head block_groups; spinlock_t lock; @@ -852,8 +864,9 @@ struct btrfs_fs_info { struct mutex ordered_operations_mutex; struct rw_semaphore extent_commit_sem; - struct rw_semaphore subvol_sem; + struct rw_semaphore cleanup_work_sem; + struct rw_semaphore subvol_sem; struct srcu_struct subvol_srcu; struct list_head trans_list; @@ -861,6 +874,9 @@ struct btrfs_fs_info { struct list_head dead_roots; struct list_head caching_block_groups; + spinlock_t delayed_iput_lock; + struct list_head delayed_iputs; + atomic_t nr_async_submits; atomic_t async_submit_draining; atomic_t nr_async_bios; @@ -903,6 +919,7 @@ struct btrfs_fs_info { * A third pool does submit_bio to avoid deadlocking with the other * two */ + struct btrfs_workers generic_worker; struct btrfs_workers workers; struct btrfs_workers delalloc_workers; struct btrfs_workers endio_workers; @@ -910,6 +927,7 @@ struct btrfs_fs_info { struct btrfs_workers endio_meta_write_workers; struct btrfs_workers endio_write_workers; struct btrfs_workers submit_workers; + struct btrfs_workers enospc_workers; /* * fixup workers take dirty pages that didn't properly go through * the cow mechanism and make them safe to write. It happens @@ -1000,7 +1018,10 @@ struct btrfs_root { atomic_t log_writers; atomic_t log_commit[2]; unsigned long log_transid; + unsigned long last_log_commit; unsigned long log_batch; + pid_t log_start_pid; + bool log_multiple_pids; u64 objectid; u64 last_trans; @@ -1022,12 +1043,12 @@ struct btrfs_root { int ref_cows; int track_dirty; int in_radix; + int clean_orphans; u64 defrag_trans_start; struct btrfs_key defrag_progress; struct btrfs_key defrag_max; int defrag_running; - int defrag_level; char *name; int in_sysfs; @@ -1141,6 +1162,8 @@ struct btrfs_root { #define BTRFS_MOUNT_FLUSHONCOMMIT (1 << 7) #define BTRFS_MOUNT_SSD_SPREAD (1 << 8) #define BTRFS_MOUNT_NOSSD (1 << 9) +#define BTRFS_MOUNT_DISCARD (1 << 10) +#define BTRFS_MOUNT_FORCE_COMPRESS (1 << 11) #define btrfs_clear_opt(o, opt) ((o) &= ~BTRFS_MOUNT_##opt) #define btrfs_set_opt(o, opt) ((o) |= BTRFS_MOUNT_##opt) @@ -1161,7 +1184,6 @@ struct btrfs_root { #define BTRFS_INODE_NOATIME (1 << 9) #define BTRFS_INODE_DIRSYNC (1 << 10) - /* some macros to generate set/get funcs for the struct fields. This * assumes there is a lefoo_to_cpu for every type, so lets make a simple * one for u8: @@ -1821,7 +1843,7 @@ BTRFS_SETGET_STACK_FUNCS(super_num_devices, struct btrfs_super_block, BTRFS_SETGET_STACK_FUNCS(super_compat_flags, struct btrfs_super_block, compat_flags, 64); BTRFS_SETGET_STACK_FUNCS(super_compat_ro_flags, struct btrfs_super_block, - compat_flags, 64); + compat_ro_flags, 64); BTRFS_SETGET_STACK_FUNCS(super_incompat_flags, struct btrfs_super_block, incompat_flags, 64); BTRFS_SETGET_STACK_FUNCS(super_csum_type, struct btrfs_super_block, @@ -1962,6 +1984,10 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans, u64 parent, u64 root_objectid, struct btrfs_disk_key *key, int level, u64 hint, u64 empty_size); +int btrfs_free_tree_block(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + u64 bytenr, u32 blocksize, + u64 parent, u64 root_objectid, int level); struct extent_buffer *btrfs_init_new_buffer(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 bytenr, u32 blocksize, @@ -2022,7 +2048,12 @@ u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags); void btrfs_set_inode_space_info(struct btrfs_root *root, struct inode *ionde); void btrfs_clear_space_info_full(struct btrfs_fs_info *info); -int btrfs_check_metadata_free_space(struct btrfs_root *root); +int btrfs_reserve_metadata_space(struct btrfs_root *root, int num_items); +int btrfs_unreserve_metadata_space(struct btrfs_root *root, int num_items); +int btrfs_unreserve_metadata_for_delalloc(struct btrfs_root *root, + struct inode *inode, int num_items); +int btrfs_reserve_metadata_for_delalloc(struct btrfs_root *root, + struct inode *inode, int num_items); int btrfs_check_data_free_space(struct btrfs_root *root, struct inode *inode, u64 bytes); void btrfs_free_reserved_data_space(struct btrfs_root *root, @@ -2071,6 +2102,10 @@ int btrfs_split_item(struct btrfs_trans_handle *trans, struct btrfs_path *path, struct btrfs_key *new_key, unsigned long split_offset); +int btrfs_duplicate_item(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, + struct btrfs_key *new_key); int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_key *key, struct btrfs_path *p, int ins_len, int cow); @@ -2178,9 +2213,10 @@ int btrfs_delete_one_dir_name(struct btrfs_trans_handle *trans, struct btrfs_path *path, struct btrfs_dir_item *di); int btrfs_insert_xattr_item(struct btrfs_trans_handle *trans, - struct btrfs_root *root, const char *name, - u16 name_len, const void *data, u16 data_len, - u64 dir); + struct btrfs_root *root, + struct btrfs_path *path, u64 objectid, + const char *name, u16 name_len, + const void *data, u16 data_len); struct btrfs_dir_item *btrfs_lookup_xattr(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_path *path, u64 dir, @@ -2274,8 +2310,9 @@ int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans, struct inode *inode, u64 new_size, u32 min_type); -int btrfs_start_delalloc_inodes(struct btrfs_root *root); -int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end); +int btrfs_start_delalloc_inodes(struct btrfs_root *root, int delay_iput); +int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end, + struct extent_state **cached_state); int btrfs_writepages(struct address_space *mapping, struct writeback_control *wbc); int btrfs_create_subvol_root(struct btrfs_trans_handle *trans, @@ -2291,7 +2328,7 @@ int btrfs_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); int btrfs_readpage(struct file *file, struct page *page); void btrfs_delete_inode(struct inode *inode); void btrfs_put_inode(struct inode *inode); -int btrfs_write_inode(struct inode *inode, int wait); +int btrfs_write_inode(struct inode *inode, struct writeback_control *wbc); void btrfs_dirty_inode(struct inode *inode); struct inode *btrfs_alloc_inode(struct super_block *sb); void btrfs_destroy_inode(struct inode *inode); @@ -2300,7 +2337,7 @@ int btrfs_init_cachep(void); void btrfs_destroy_cachep(void); long btrfs_ioctl_trans_end(struct file *file); struct inode *btrfs_iget(struct super_block *s, struct btrfs_key *location, - struct btrfs_root *root); + struct btrfs_root *root, int *was_new); int btrfs_commit_write(struct file *file, struct page *page, unsigned from, unsigned to); struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page, @@ -2314,7 +2351,9 @@ int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode); void btrfs_orphan_cleanup(struct btrfs_root *root); int btrfs_cont_expand(struct inode *inode, loff_t size); int btrfs_invalidate_inodes(struct btrfs_root *root); -extern struct dentry_operations btrfs_dentry_operations; +void btrfs_add_delayed_iput(struct inode *inode); +void btrfs_run_delayed_iputs(struct btrfs_root *root); +extern const struct dentry_operations btrfs_dentry_operations; /* ioctl.c */ long btrfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg); @@ -2326,13 +2365,10 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync); int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end, int skip_pinned); int btrfs_check_file(struct btrfs_root *root, struct inode *inode); -extern struct file_operations btrfs_file_operations; -int btrfs_drop_extents(struct btrfs_trans_handle *trans, - struct btrfs_root *root, struct inode *inode, - u64 start, u64 end, u64 locked_end, - u64 inline_limit, u64 *hint_block, int drop_cache); +extern const struct file_operations btrfs_file_operations; +int btrfs_drop_extents(struct btrfs_trans_handle *trans, struct inode *inode, + u64 start, u64 end, u64 *hint_byte, int drop_cache); int btrfs_mark_extent_written(struct btrfs_trans_handle *trans, - struct btrfs_root *root, struct inode *inode, u64 start, u64 end); int btrfs_release_file(struct inode *inode, struct file *file); @@ -2352,17 +2388,17 @@ void btrfs_sysfs_del_super(struct btrfs_fs_info *root); ssize_t btrfs_listxattr(struct dentry *dentry, char *buffer, size_t size); /* super.c */ -u64 btrfs_parse_size(char *str); int btrfs_parse_options(struct btrfs_root *root, char *options); int btrfs_sync_fs(struct super_block *sb, int wait); /* acl.c */ -#ifdef CONFIG_FS_POSIX_ACL +#ifdef CONFIG_BTRFS_FS_POSIX_ACL int btrfs_check_acl(struct inode *inode, int mask); #else #define btrfs_check_acl NULL #endif -int btrfs_init_acl(struct inode *inode, struct inode *dir); +int btrfs_init_acl(struct btrfs_trans_handle *trans, + struct inode *inode, struct inode *dir); int btrfs_acl_chmod(struct inode *inode); /* relocation.c */ diff --git a/fs/btrfs/dir-item.c b/fs/btrfs/dir-item.c index f3a6075519cc..e9103b3baa49 100644 --- a/fs/btrfs/dir-item.c +++ b/fs/btrfs/dir-item.c @@ -68,12 +68,12 @@ static struct btrfs_dir_item *insert_with_overflow(struct btrfs_trans_handle * into the tree */ int btrfs_insert_xattr_item(struct btrfs_trans_handle *trans, - struct btrfs_root *root, const char *name, - u16 name_len, const void *data, u16 data_len, - u64 dir) + struct btrfs_root *root, + struct btrfs_path *path, u64 objectid, + const char *name, u16 name_len, + const void *data, u16 data_len) { int ret = 0; - struct btrfs_path *path; struct btrfs_dir_item *dir_item; unsigned long name_ptr, data_ptr; struct btrfs_key key, location; @@ -81,15 +81,11 @@ int btrfs_insert_xattr_item(struct btrfs_trans_handle *trans, struct extent_buffer *leaf; u32 data_size; - key.objectid = dir; + BUG_ON(name_len + data_len > BTRFS_MAX_XATTR_SIZE(root)); + + key.objectid = objectid; btrfs_set_key_type(&key, BTRFS_XATTR_ITEM_KEY); key.offset = btrfs_name_hash(name, name_len); - path = btrfs_alloc_path(); - if (!path) - return -ENOMEM; - if (name_len + data_len + sizeof(struct btrfs_dir_item) > - BTRFS_LEAF_DATA_SIZE(root) - sizeof(struct btrfs_item)) - return -ENOSPC; data_size = sizeof(*dir_item) + name_len + data_len; dir_item = insert_with_overflow(trans, root, path, &key, data_size, @@ -117,7 +113,6 @@ int btrfs_insert_xattr_item(struct btrfs_trans_handle *trans, write_extent_buffer(leaf, data, data_ptr, data_len); btrfs_mark_buffer_dirty(path->nodes[0]); - btrfs_free_path(path); return ret; } diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 644e796fd643..11d0ad30e203 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -263,13 +263,15 @@ static int csum_tree_block(struct btrfs_root *root, struct extent_buffer *buf, static int verify_parent_transid(struct extent_io_tree *io_tree, struct extent_buffer *eb, u64 parent_transid) { + struct extent_state *cached_state = NULL; int ret; if (!parent_transid || btrfs_header_generation(eb) == parent_transid) return 0; - lock_extent(io_tree, eb->start, eb->start + eb->len - 1, GFP_NOFS); - if (extent_buffer_uptodate(io_tree, eb) && + lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1, + 0, &cached_state, GFP_NOFS); + if (extent_buffer_uptodate(io_tree, eb, cached_state) && btrfs_header_generation(eb) == parent_transid) { ret = 0; goto out; @@ -282,10 +284,10 @@ static int verify_parent_transid(struct extent_io_tree *io_tree, (unsigned long long)btrfs_header_generation(eb)); } ret = 1; - clear_extent_buffer_uptodate(io_tree, eb); + clear_extent_buffer_uptodate(io_tree, eb, &cached_state); out: - unlock_extent(io_tree, eb->start, eb->start + eb->len - 1, - GFP_NOFS); + unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1, + &cached_state, GFP_NOFS); return ret; } @@ -822,14 +824,14 @@ struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root, int btrfs_write_tree_block(struct extent_buffer *buf) { - return btrfs_fdatawrite_range(buf->first_page->mapping, buf->start, - buf->start + buf->len - 1, WB_SYNC_ALL); + return filemap_fdatawrite_range(buf->first_page->mapping, buf->start, + buf->start + buf->len - 1); } int btrfs_wait_tree_block_writeback(struct extent_buffer *buf) { - return btrfs_wait_on_page_writeback_range(buf->first_page->mapping, - buf->start, buf->start + buf->len - 1); + return filemap_fdatawait_range(buf->first_page->mapping, + buf->start, buf->start + buf->len - 1); } struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr, @@ -892,6 +894,8 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, root->stripesize = stripesize; root->ref_cows = 0; root->track_dirty = 0; + root->in_radix = 0; + root->clean_orphans = 0; root->fs_info = fs_info; root->objectid = objectid; @@ -899,7 +903,7 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, root->highest_objectid = 0; root->name = NULL; root->in_sysfs = 0; - root->inode_tree.rb_node = NULL; + root->inode_tree = RB_ROOT; INIT_LIST_HEAD(&root->dirty_list); INIT_LIST_HEAD(&root->orphan_list); @@ -917,6 +921,7 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, atomic_set(&root->log_writers, 0); root->log_batch = 0; root->log_transid = 0; + root->last_log_commit = 0; extent_io_tree_init(&root->dirty_log_pages, fs_info->btree_inode->i_mapping, GFP_NOFS); @@ -927,7 +932,6 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, root->defrag_trans_start = fs_info->generation; init_completion(&root->kobj_unregister); root->defrag_running = 0; - root->defrag_level = 0; root->root_key.objectid = objectid; root->anon_super.s_root = NULL; root->anon_super.s_dev = 0; @@ -979,12 +983,12 @@ int btrfs_free_log_root_tree(struct btrfs_trans_handle *trans, while (1) { ret = find_first_extent_bit(&log_root_tree->dirty_log_pages, - 0, &start, &end, EXTENT_DIRTY); + 0, &start, &end, EXTENT_DIRTY | EXTENT_NEW); if (ret) break; - clear_extent_dirty(&log_root_tree->dirty_log_pages, - start, end, GFP_NOFS); + clear_extent_bits(&log_root_tree->dirty_log_pages, start, end, + EXTENT_DIRTY | EXTENT_NEW, GFP_NOFS); } eb = fs_info->log_root_tree->node; @@ -1087,6 +1091,7 @@ int btrfs_add_log_tree(struct btrfs_trans_handle *trans, WARN_ON(root->log_root); root->log_root = log_root; root->log_transid = 0; + root->last_log_commit = 0; return 0; } @@ -1208,8 +1213,10 @@ again: ret = radix_tree_insert(&fs_info->fs_roots_radix, (unsigned long)root->root_key.objectid, root); - if (ret == 0) + if (ret == 0) { root->in_radix = 1; + root->clean_orphans = 1; + } spin_unlock(&fs_info->fs_roots_radix_lock); radix_tree_preload_end(); if (ret) { @@ -1223,10 +1230,6 @@ again: ret = btrfs_find_dead_roots(fs_info->tree_root, root->root_key.objectid); WARN_ON(ret); - - if (!(fs_info->sb->s_flags & MS_RDONLY)) - btrfs_orphan_cleanup(root); - return root; fail: free_fs_root(root); @@ -1475,6 +1478,7 @@ static int cleaner_kthread(void *arg) if (!(root->fs_info->sb->s_flags & MS_RDONLY) && mutex_trylock(&root->fs_info->cleaner_mutex)) { + btrfs_run_delayed_iputs(root); btrfs_clean_old_snapshots(root); mutex_unlock(&root->fs_info->cleaner_mutex); } @@ -1604,6 +1608,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC); INIT_LIST_HEAD(&fs_info->trans_list); INIT_LIST_HEAD(&fs_info->dead_roots); + INIT_LIST_HEAD(&fs_info->delayed_iputs); INIT_LIST_HEAD(&fs_info->hashers); INIT_LIST_HEAD(&fs_info->delalloc_inodes); INIT_LIST_HEAD(&fs_info->ordered_operations); @@ -1612,6 +1617,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, spin_lock_init(&fs_info->new_trans_lock); spin_lock_init(&fs_info->ref_cache_lock); spin_lock_init(&fs_info->fs_roots_radix_lock); + spin_lock_init(&fs_info->delayed_iput_lock); init_completion(&fs_info->kobj_unregister); fs_info->tree_root = tree_root; @@ -1630,7 +1636,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, fs_info->sb = sb; fs_info->max_extent = (u64)-1; fs_info->max_inline = 8192 * 1024; - fs_info->metadata_ratio = 8; + fs_info->metadata_ratio = 0; fs_info->thread_pool_size = min_t(unsigned long, num_online_cpus() + 2, 8); @@ -1669,7 +1675,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, insert_inode_hash(fs_info->btree_inode); spin_lock_init(&fs_info->block_group_cache_lock); - fs_info->block_group_cache_tree.rb_node = NULL; + fs_info->block_group_cache_tree = RB_ROOT; extent_io_tree_init(&fs_info->freed_extents[0], fs_info->btree_inode->i_mapping, GFP_NOFS); @@ -1687,6 +1693,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, mutex_init(&fs_info->cleaner_mutex); mutex_init(&fs_info->volume_mutex); init_rwsem(&fs_info->extent_commit_sem); + init_rwsem(&fs_info->cleanup_work_sem); init_rwsem(&fs_info->subvol_sem); btrfs_init_free_cluster(&fs_info->meta_alloc_cluster); @@ -1746,21 +1753,25 @@ struct btrfs_root *open_ctree(struct super_block *sb, err = -EINVAL; goto fail_iput; } -printk("thread pool is %d\n", fs_info->thread_pool_size); - /* - * we need to start all the end_io workers up front because the - * queue work function gets called at interrupt time, and so it - * cannot dynamically grow. - */ + + btrfs_init_workers(&fs_info->generic_worker, + "genwork", 1, NULL); + btrfs_init_workers(&fs_info->workers, "worker", - fs_info->thread_pool_size); + fs_info->thread_pool_size, + &fs_info->generic_worker); btrfs_init_workers(&fs_info->delalloc_workers, "delalloc", - fs_info->thread_pool_size); + fs_info->thread_pool_size, + &fs_info->generic_worker); btrfs_init_workers(&fs_info->submit_workers, "submit", min_t(u64, fs_devices->num_devices, - fs_info->thread_pool_size)); + fs_info->thread_pool_size), + &fs_info->generic_worker); + btrfs_init_workers(&fs_info->enospc_workers, "enospc", + fs_info->thread_pool_size, + &fs_info->generic_worker); /* a higher idle thresh on the submit workers makes it much more * likely that bios will be send down in a sane order to the @@ -1774,15 +1785,20 @@ printk("thread pool is %d\n", fs_info->thread_pool_size); fs_info->delalloc_workers.idle_thresh = 2; fs_info->delalloc_workers.ordered = 1; - btrfs_init_workers(&fs_info->fixup_workers, "fixup", 1); + btrfs_init_workers(&fs_info->fixup_workers, "fixup", 1, + &fs_info->generic_worker); btrfs_init_workers(&fs_info->endio_workers, "endio", - fs_info->thread_pool_size); + fs_info->thread_pool_size, + &fs_info->generic_worker); btrfs_init_workers(&fs_info->endio_meta_workers, "endio-meta", - fs_info->thread_pool_size); + fs_info->thread_pool_size, + &fs_info->generic_worker); btrfs_init_workers(&fs_info->endio_meta_write_workers, - "endio-meta-write", fs_info->thread_pool_size); + "endio-meta-write", fs_info->thread_pool_size, + &fs_info->generic_worker); btrfs_init_workers(&fs_info->endio_write_workers, "endio-write", - fs_info->thread_pool_size); + fs_info->thread_pool_size, + &fs_info->generic_worker); /* * endios are largely parallel and should have a very @@ -1794,12 +1810,8 @@ printk("thread pool is %d\n", fs_info->thread_pool_size); fs_info->endio_write_workers.idle_thresh = 2; fs_info->endio_meta_write_workers.idle_thresh = 2; - fs_info->endio_workers.atomic_worker_start = 1; - fs_info->endio_meta_workers.atomic_worker_start = 1; - fs_info->endio_write_workers.atomic_worker_start = 1; - fs_info->endio_meta_write_workers.atomic_worker_start = 1; - btrfs_start_workers(&fs_info->workers, 1); + btrfs_start_workers(&fs_info->generic_worker, 1); btrfs_start_workers(&fs_info->submit_workers, 1); btrfs_start_workers(&fs_info->delalloc_workers, 1); btrfs_start_workers(&fs_info->fixup_workers, 1); @@ -1807,6 +1819,7 @@ printk("thread pool is %d\n", fs_info->thread_pool_size); btrfs_start_workers(&fs_info->endio_meta_workers, 1); btrfs_start_workers(&fs_info->endio_meta_write_workers, 1); btrfs_start_workers(&fs_info->endio_write_workers, 1); + btrfs_start_workers(&fs_info->enospc_workers, 1); fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super); fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages, @@ -1971,7 +1984,12 @@ printk("thread pool is %d\n", fs_info->thread_pool_size); if (!(sb->s_flags & MS_RDONLY)) { ret = btrfs_recover_relocation(tree_root); - BUG_ON(ret); + if (ret < 0) { + printk(KERN_WARNING + "btrfs: failed to recover relocation\n"); + err = -EINVAL; + goto fail_trans_kthread; + } } location.objectid = BTRFS_FS_TREE_OBJECTID; @@ -1982,6 +2000,12 @@ printk("thread pool is %d\n", fs_info->thread_pool_size); if (!fs_info->fs_root) goto fail_trans_kthread; + if (!(sb->s_flags & MS_RDONLY)) { + down_read(&fs_info->cleanup_work_sem); + btrfs_orphan_cleanup(fs_info->fs_root); + up_read(&fs_info->cleanup_work_sem); + } + return tree_root; fail_trans_kthread: @@ -2012,6 +2036,7 @@ fail_chunk_root: free_extent_buffer(chunk_root->node); free_extent_buffer(chunk_root->commit_root); fail_sb_buffer: + btrfs_stop_workers(&fs_info->generic_worker); btrfs_stop_workers(&fs_info->fixup_workers); btrfs_stop_workers(&fs_info->delalloc_workers); btrfs_stop_workers(&fs_info->workers); @@ -2020,6 +2045,7 @@ fail_sb_buffer: btrfs_stop_workers(&fs_info->endio_meta_write_workers); btrfs_stop_workers(&fs_info->endio_write_workers); btrfs_stop_workers(&fs_info->submit_workers); + btrfs_stop_workers(&fs_info->enospc_workers); fail_iput: invalidate_inode_pages2(fs_info->btree_inode->i_mapping); iput(fs_info->btree_inode); @@ -2376,8 +2402,14 @@ int btrfs_commit_super(struct btrfs_root *root) int ret; mutex_lock(&root->fs_info->cleaner_mutex); + btrfs_run_delayed_iputs(root); btrfs_clean_old_snapshots(root); mutex_unlock(&root->fs_info->cleaner_mutex); + + /* wait until ongoing cleanup work done */ + down_write(&root->fs_info->cleanup_work_sem); + up_write(&root->fs_info->cleanup_work_sem); + trans = btrfs_start_transaction(root, 1); ret = btrfs_commit_transaction(trans, root); BUG_ON(ret); @@ -2437,6 +2469,7 @@ int close_ctree(struct btrfs_root *root) iput(fs_info->btree_inode); + btrfs_stop_workers(&fs_info->generic_worker); btrfs_stop_workers(&fs_info->fixup_workers); btrfs_stop_workers(&fs_info->delalloc_workers); btrfs_stop_workers(&fs_info->workers); @@ -2445,6 +2478,7 @@ int close_ctree(struct btrfs_root *root) btrfs_stop_workers(&fs_info->endio_meta_write_workers); btrfs_stop_workers(&fs_info->endio_write_workers); btrfs_stop_workers(&fs_info->submit_workers); + btrfs_stop_workers(&fs_info->enospc_workers); btrfs_close_devices(fs_info->fs_devices); btrfs_mapping_tree_free(&fs_info->mapping_tree); @@ -2465,7 +2499,8 @@ int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid) int ret; struct inode *btree_inode = buf->first_page->mapping->host; - ret = extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf); + ret = extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf, + NULL); if (!ret) return ret; diff --git a/fs/btrfs/export.c b/fs/btrfs/export.c index ba5c3fd5ab8c..951ef09b82f4 100644 --- a/fs/btrfs/export.c +++ b/fs/btrfs/export.c @@ -95,7 +95,7 @@ static struct dentry *btrfs_get_dentry(struct super_block *sb, u64 objectid, btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY); key.offset = 0; - inode = btrfs_iget(sb, &key, root); + inode = btrfs_iget(sb, &key, root, NULL); if (IS_ERR(inode)) { err = PTR_ERR(inode); goto fail; @@ -223,7 +223,7 @@ static struct dentry *btrfs_get_parent(struct dentry *child) key.type = BTRFS_INODE_ITEM_KEY; key.offset = 0; - dentry = d_obtain_alias(btrfs_iget(root->fs_info->sb, &key, root)); + dentry = d_obtain_alias(btrfs_iget(root->fs_info->sb, &key, root, NULL)); if (!IS_ERR(dentry)) dentry->d_op = &btrfs_dentry_operations; return dentry; diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 993f93ff7ba6..1727b26fb194 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -68,6 +68,8 @@ static int pin_down_bytes(struct btrfs_trans_handle *trans, struct extent_buffer **must_clean); static int find_next_key(struct btrfs_path *path, int level, struct btrfs_key *key); +static void dump_space_info(struct btrfs_space_info *info, u64 bytes, + int dump_block_groups); static noinline int block_group_cache_done(struct btrfs_block_group_cache *cache) @@ -81,6 +83,17 @@ static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits) return (cache->flags & bits) == bits; } +void btrfs_get_block_group(struct btrfs_block_group_cache *cache) +{ + atomic_inc(&cache->count); +} + +void btrfs_put_block_group(struct btrfs_block_group_cache *cache) +{ + if (atomic_dec_and_test(&cache->count)) + kfree(cache); +} + /* * this adds the block group to the fs_info rb tree for the block group * cache @@ -154,7 +167,7 @@ block_group_cache_tree_search(struct btrfs_fs_info *info, u64 bytenr, } } if (ret) - atomic_inc(&ret->count); + btrfs_get_block_group(ret); spin_unlock(&info->block_group_cache_lock); return ret; @@ -193,6 +206,14 @@ static int exclude_super_stripes(struct btrfs_root *root, int stripe_len; int i, nr, ret; + if (cache->key.objectid < BTRFS_SUPER_INFO_OFFSET) { + stripe_len = BTRFS_SUPER_INFO_OFFSET - cache->key.objectid; + cache->bytes_super += stripe_len; + ret = add_excluded_extent(root, cache->key.objectid, + stripe_len); + BUG_ON(ret); + } + for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) { bytenr = btrfs_sb_offset(i); ret = btrfs_rmap_block(&root->fs_info->mapping_tree, @@ -253,7 +274,7 @@ static u64 add_new_free_space(struct btrfs_block_group_cache *block_group, if (ret) break; - if (extent_start == start) { + if (extent_start <= start) { start = extent_end + 1; } else if (extent_start > start && extent_start < end) { size = extent_start - start; @@ -397,6 +418,8 @@ err: put_caching_control(caching_ctl); atomic_dec(&block_group->space_info->caching_threads); + btrfs_put_block_group(block_group); + return 0; } @@ -437,6 +460,7 @@ static int cache_block_group(struct btrfs_block_group_cache *cache) up_write(&fs_info->extent_commit_sem); atomic_inc(&cache->space_info->caching_threads); + btrfs_get_block_group(cache); tsk = kthread_run(caching_kthread, cache, "btrfs-cache-%llu\n", cache->key.objectid); @@ -476,12 +500,6 @@ struct btrfs_block_group_cache *btrfs_lookup_block_group( return cache; } -void btrfs_put_block_group(struct btrfs_block_group_cache *cache) -{ - if (atomic_dec_and_test(&cache->count)) - kfree(cache); -} - static struct btrfs_space_info *__find_space_info(struct btrfs_fs_info *info, u64 flags) { @@ -1566,23 +1584,23 @@ static int remove_extent_backref(struct btrfs_trans_handle *trans, return ret; } -#ifdef BIO_RW_DISCARD static void btrfs_issue_discard(struct block_device *bdev, u64 start, u64 len) { blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_KERNEL, DISCARD_FL_BARRIER); } -#endif static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr, u64 num_bytes) { -#ifdef BIO_RW_DISCARD int ret; u64 map_length = num_bytes; struct btrfs_multi_bio *multi = NULL; + if (!btrfs_test_opt(root, DISCARD)) + return 0; + /* Tell the block device(s) that the sectors can be discarded */ ret = btrfs_map_block(&root->fs_info->mapping_tree, READ, bytenr, &map_length, &multi, 0); @@ -1602,9 +1620,6 @@ static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr, } return ret; -#else - return 0; -#endif } int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans, @@ -2575,7 +2590,7 @@ next_block_group(struct btrfs_root *root, if (node) { cache = rb_entry(node, struct btrfs_block_group_cache, cache_node); - atomic_inc(&cache->count); + btrfs_get_block_group(cache); } else cache = NULL; spin_unlock(&root->fs_info->block_group_cache_lock); @@ -2765,67 +2780,448 @@ void btrfs_set_inode_space_info(struct btrfs_root *root, struct inode *inode) alloc_target); } +static u64 calculate_bytes_needed(struct btrfs_root *root, int num_items) +{ + u64 num_bytes; + int level; + + level = BTRFS_MAX_LEVEL - 2; + /* + * NOTE: these calculations are absolutely the worst possible case. + * This assumes that _every_ item we insert will require a new leaf, and + * that the tree has grown to its maximum level size. + */ + + /* + * for every item we insert we could insert both an extent item and a + * extent ref item. Then for ever item we insert, we will need to cow + * both the original leaf, plus the leaf to the left and right of it. + * + * Unless we are talking about the extent root, then we just want the + * number of items * 2, since we just need the extent item plus its ref. + */ + if (root == root->fs_info->extent_root) + num_bytes = num_items * 2; + else + num_bytes = (num_items + (2 * num_items)) * 3; + + /* + * num_bytes is total number of leaves we could need times the leaf + * size, and then for every leaf we could end up cow'ing 2 nodes per + * level, down to the leaf level. + */ + num_bytes = (num_bytes * root->leafsize) + + (num_bytes * (level * 2)) * root->nodesize; + + return num_bytes; +} + /* - * for now this just makes sure we have at least 5% of our metadata space free - * for use. + * Unreserve metadata space for delalloc. If we have less reserved credits than + * we have extents, this function does nothing. */ -int btrfs_check_metadata_free_space(struct btrfs_root *root) +int btrfs_unreserve_metadata_for_delalloc(struct btrfs_root *root, + struct inode *inode, int num_items) { struct btrfs_fs_info *info = root->fs_info; struct btrfs_space_info *meta_sinfo; - u64 alloc_target, thresh; - int committed = 0, ret; + u64 num_bytes; + u64 alloc_target; + bool bug = false; /* get the space info for where the metadata will live */ alloc_target = btrfs_get_alloc_profile(root, 0); meta_sinfo = __find_space_info(info, alloc_target); - if (!meta_sinfo) - goto alloc; -again: + num_bytes = calculate_bytes_needed(root->fs_info->extent_root, + num_items); + spin_lock(&meta_sinfo->lock); - if (!meta_sinfo->full) - thresh = meta_sinfo->total_bytes * 80; - else - thresh = meta_sinfo->total_bytes * 95; + spin_lock(&BTRFS_I(inode)->accounting_lock); + if (BTRFS_I(inode)->reserved_extents <= + BTRFS_I(inode)->outstanding_extents) { + spin_unlock(&BTRFS_I(inode)->accounting_lock); + spin_unlock(&meta_sinfo->lock); + return 0; + } + spin_unlock(&BTRFS_I(inode)->accounting_lock); + + BTRFS_I(inode)->reserved_extents--; + BUG_ON(BTRFS_I(inode)->reserved_extents < 0); + + if (meta_sinfo->bytes_delalloc < num_bytes) { + bug = true; + meta_sinfo->bytes_delalloc = 0; + } else { + meta_sinfo->bytes_delalloc -= num_bytes; + } + spin_unlock(&meta_sinfo->lock); + + BUG_ON(bug); + + return 0; +} +static void check_force_delalloc(struct btrfs_space_info *meta_sinfo) +{ + u64 thresh; + + thresh = meta_sinfo->bytes_used + meta_sinfo->bytes_reserved + + meta_sinfo->bytes_pinned + meta_sinfo->bytes_readonly + + meta_sinfo->bytes_super + meta_sinfo->bytes_root + + meta_sinfo->bytes_may_use; + + thresh = meta_sinfo->total_bytes - thresh; + thresh *= 80; do_div(thresh, 100); + if (thresh <= meta_sinfo->bytes_delalloc) + meta_sinfo->force_delalloc = 1; + else + meta_sinfo->force_delalloc = 0; +} - if (meta_sinfo->bytes_used + meta_sinfo->bytes_reserved + - meta_sinfo->bytes_pinned + meta_sinfo->bytes_readonly + - meta_sinfo->bytes_super > thresh) { - struct btrfs_trans_handle *trans; - if (!meta_sinfo->full) { - meta_sinfo->force_alloc = 1; +struct async_flush { + struct btrfs_root *root; + struct btrfs_space_info *info; + struct btrfs_work work; +}; + +static noinline void flush_delalloc_async(struct btrfs_work *work) +{ + struct async_flush *async; + struct btrfs_root *root; + struct btrfs_space_info *info; + + async = container_of(work, struct async_flush, work); + root = async->root; + info = async->info; + + btrfs_start_delalloc_inodes(root, 0); + wake_up(&info->flush_wait); + btrfs_wait_ordered_extents(root, 0, 0); + + spin_lock(&info->lock); + info->flushing = 0; + spin_unlock(&info->lock); + wake_up(&info->flush_wait); + + kfree(async); +} + +static void wait_on_flush(struct btrfs_space_info *info) +{ + DEFINE_WAIT(wait); + u64 used; + + while (1) { + prepare_to_wait(&info->flush_wait, &wait, + TASK_UNINTERRUPTIBLE); + spin_lock(&info->lock); + if (!info->flushing) { + spin_unlock(&info->lock); + break; + } + + used = info->bytes_used + info->bytes_reserved + + info->bytes_pinned + info->bytes_readonly + + info->bytes_super + info->bytes_root + + info->bytes_may_use + info->bytes_delalloc; + if (used < info->total_bytes) { + spin_unlock(&info->lock); + break; + } + spin_unlock(&info->lock); + schedule(); + } + finish_wait(&info->flush_wait, &wait); +} + +static void flush_delalloc(struct btrfs_root *root, + struct btrfs_space_info *info) +{ + struct async_flush *async; + bool wait = false; + + spin_lock(&info->lock); + + if (!info->flushing) { + info->flushing = 1; + init_waitqueue_head(&info->flush_wait); + } else { + wait = true; + } + + spin_unlock(&info->lock); + + if (wait) { + wait_on_flush(info); + return; + } + + async = kzalloc(sizeof(*async), GFP_NOFS); + if (!async) + goto flush; + + async->root = root; + async->info = info; + async->work.func = flush_delalloc_async; + + btrfs_queue_worker(&root->fs_info->enospc_workers, + &async->work); + wait_on_flush(info); + return; + +flush: + btrfs_start_delalloc_inodes(root, 0); + btrfs_wait_ordered_extents(root, 0, 0); + + spin_lock(&info->lock); + info->flushing = 0; + spin_unlock(&info->lock); + wake_up(&info->flush_wait); +} + +static int maybe_allocate_chunk(struct btrfs_root *root, + struct btrfs_space_info *info) +{ + struct btrfs_super_block *disk_super = &root->fs_info->super_copy; + struct btrfs_trans_handle *trans; + bool wait = false; + int ret = 0; + u64 min_metadata; + u64 free_space; + + free_space = btrfs_super_total_bytes(disk_super); + /* + * we allow the metadata to grow to a max of either 10gb or 5% of the + * space in the volume. + */ + min_metadata = min((u64)10 * 1024 * 1024 * 1024, + div64_u64(free_space * 5, 100)); + if (info->total_bytes >= min_metadata) { + spin_unlock(&info->lock); + return 0; + } + + if (info->full) { + spin_unlock(&info->lock); + return 0; + } + + if (!info->allocating_chunk) { + info->force_alloc = 1; + info->allocating_chunk = 1; + init_waitqueue_head(&info->allocate_wait); + } else { + wait = true; + } + + spin_unlock(&info->lock); + + if (wait) { + wait_event(info->allocate_wait, + !info->allocating_chunk); + return 1; + } + + trans = btrfs_start_transaction(root, 1); + if (!trans) { + ret = -ENOMEM; + goto out; + } + + ret = do_chunk_alloc(trans, root->fs_info->extent_root, + 4096 + 2 * 1024 * 1024, + info->flags, 0); + btrfs_end_transaction(trans, root); + if (ret) + goto out; +out: + spin_lock(&info->lock); + info->allocating_chunk = 0; + spin_unlock(&info->lock); + wake_up(&info->allocate_wait); + + if (ret) + return 0; + return 1; +} + +/* + * Reserve metadata space for delalloc. + */ +int btrfs_reserve_metadata_for_delalloc(struct btrfs_root *root, + struct inode *inode, int num_items) +{ + struct btrfs_fs_info *info = root->fs_info; + struct btrfs_space_info *meta_sinfo; + u64 num_bytes; + u64 used; + u64 alloc_target; + int flushed = 0; + int force_delalloc; + + /* get the space info for where the metadata will live */ + alloc_target = btrfs_get_alloc_profile(root, 0); + meta_sinfo = __find_space_info(info, alloc_target); + + num_bytes = calculate_bytes_needed(root->fs_info->extent_root, + num_items); +again: + spin_lock(&meta_sinfo->lock); + + force_delalloc = meta_sinfo->force_delalloc; + + if (unlikely(!meta_sinfo->bytes_root)) + meta_sinfo->bytes_root = calculate_bytes_needed(root, 6); + + if (!flushed) + meta_sinfo->bytes_delalloc += num_bytes; + + used = meta_sinfo->bytes_used + meta_sinfo->bytes_reserved + + meta_sinfo->bytes_pinned + meta_sinfo->bytes_readonly + + meta_sinfo->bytes_super + meta_sinfo->bytes_root + + meta_sinfo->bytes_may_use + meta_sinfo->bytes_delalloc; + + if (used > meta_sinfo->total_bytes) { + flushed++; + + if (flushed == 1) { + if (maybe_allocate_chunk(root, meta_sinfo)) + goto again; + flushed++; + } else { spin_unlock(&meta_sinfo->lock); -alloc: - trans = btrfs_start_transaction(root, 1); - if (!trans) - return -ENOMEM; + } - ret = do_chunk_alloc(trans, root->fs_info->extent_root, - 2 * 1024 * 1024, alloc_target, 0); - btrfs_end_transaction(trans, root); - if (!meta_sinfo) { - meta_sinfo = __find_space_info(info, - alloc_target); - } + if (flushed == 2) { + filemap_flush(inode->i_mapping); + goto again; + } else if (flushed == 3) { + flush_delalloc(root, meta_sinfo); goto again; } + spin_lock(&meta_sinfo->lock); + meta_sinfo->bytes_delalloc -= num_bytes; spin_unlock(&meta_sinfo->lock); + printk(KERN_ERR "enospc, has %d, reserved %d\n", + BTRFS_I(inode)->outstanding_extents, + BTRFS_I(inode)->reserved_extents); + dump_space_info(meta_sinfo, 0, 0); + return -ENOSPC; + } - if (!committed) { - committed = 1; - trans = btrfs_join_transaction(root, 1); - if (!trans) - return -ENOMEM; - ret = btrfs_commit_transaction(trans, root); - if (ret) - return ret; + BTRFS_I(inode)->reserved_extents++; + check_force_delalloc(meta_sinfo); + spin_unlock(&meta_sinfo->lock); + + if (!flushed && force_delalloc) + filemap_flush(inode->i_mapping); + + return 0; +} + +/* + * unreserve num_items number of items worth of metadata space. This needs to + * be paired with btrfs_reserve_metadata_space. + * + * NOTE: if you have the option, run this _AFTER_ you do a + * btrfs_end_transaction, since btrfs_end_transaction will run delayed ref + * oprations which will result in more used metadata, so we want to make sure we + * can do that without issue. + */ +int btrfs_unreserve_metadata_space(struct btrfs_root *root, int num_items) +{ + struct btrfs_fs_info *info = root->fs_info; + struct btrfs_space_info *meta_sinfo; + u64 num_bytes; + u64 alloc_target; + bool bug = false; + + /* get the space info for where the metadata will live */ + alloc_target = btrfs_get_alloc_profile(root, 0); + meta_sinfo = __find_space_info(info, alloc_target); + + num_bytes = calculate_bytes_needed(root, num_items); + + spin_lock(&meta_sinfo->lock); + if (meta_sinfo->bytes_may_use < num_bytes) { + bug = true; + meta_sinfo->bytes_may_use = 0; + } else { + meta_sinfo->bytes_may_use -= num_bytes; + } + spin_unlock(&meta_sinfo->lock); + + BUG_ON(bug); + + return 0; +} + +/* + * Reserve some metadata space for use. We'll calculate the worste case number + * of bytes that would be needed to modify num_items number of items. If we + * have space, fantastic, if not, you get -ENOSPC. Please call + * btrfs_unreserve_metadata_space when you are done for the _SAME_ number of + * items you reserved, since whatever metadata you needed should have already + * been allocated. + * + * This will commit the transaction to make more space if we don't have enough + * metadata space. THe only time we don't do this is if we're reserving space + * inside of a transaction, then we will just return -ENOSPC and it is the + * callers responsibility to handle it properly. + */ +int btrfs_reserve_metadata_space(struct btrfs_root *root, int num_items) +{ + struct btrfs_fs_info *info = root->fs_info; + struct btrfs_space_info *meta_sinfo; + u64 num_bytes; + u64 used; + u64 alloc_target; + int retries = 0; + + /* get the space info for where the metadata will live */ + alloc_target = btrfs_get_alloc_profile(root, 0); + meta_sinfo = __find_space_info(info, alloc_target); + + num_bytes = calculate_bytes_needed(root, num_items); +again: + spin_lock(&meta_sinfo->lock); + + if (unlikely(!meta_sinfo->bytes_root)) + meta_sinfo->bytes_root = calculate_bytes_needed(root, 6); + + if (!retries) + meta_sinfo->bytes_may_use += num_bytes; + + used = meta_sinfo->bytes_used + meta_sinfo->bytes_reserved + + meta_sinfo->bytes_pinned + meta_sinfo->bytes_readonly + + meta_sinfo->bytes_super + meta_sinfo->bytes_root + + meta_sinfo->bytes_may_use + meta_sinfo->bytes_delalloc; + + if (used > meta_sinfo->total_bytes) { + retries++; + if (retries == 1) { + if (maybe_allocate_chunk(root, meta_sinfo)) + goto again; + retries++; + } else { + spin_unlock(&meta_sinfo->lock); + } + + if (retries == 2) { + flush_delalloc(root, meta_sinfo); goto again; } + spin_lock(&meta_sinfo->lock); + meta_sinfo->bytes_may_use -= num_bytes; + spin_unlock(&meta_sinfo->lock); + + dump_space_info(meta_sinfo, 0, 0); return -ENOSPC; } + + check_force_delalloc(meta_sinfo); spin_unlock(&meta_sinfo->lock); return 0; @@ -2888,7 +3284,7 @@ alloc: spin_unlock(&data_sinfo->lock); /* commit the current transaction and try again */ - if (!committed) { + if (!committed && !root->fs_info->open_ioctl_trans) { committed = 1; trans = btrfs_join_transaction(root, 1); if (!trans) @@ -2916,7 +3312,7 @@ alloc: BTRFS_I(inode)->reserved_bytes += bytes; spin_unlock(&data_sinfo->lock); - return btrfs_check_metadata_free_space(root); + return 0; } /* @@ -3015,17 +3411,15 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans, BUG_ON(!space_info); spin_lock(&space_info->lock); - if (space_info->force_alloc) { + if (space_info->force_alloc) force = 1; - space_info->force_alloc = 0; - } if (space_info->full) { spin_unlock(&space_info->lock); goto out; } thresh = space_info->total_bytes - space_info->bytes_readonly; - thresh = div_factor(thresh, 6); + thresh = div_factor(thresh, 8); if (!force && (space_info->bytes_used + space_info->bytes_pinned + space_info->bytes_reserved + alloc_bytes) < thresh) { @@ -3039,7 +3433,7 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans, * we keep a reasonable number of metadata chunks allocated in the * FS as well. */ - if (flags & BTRFS_BLOCK_GROUP_DATA) { + if (flags & BTRFS_BLOCK_GROUP_DATA && fs_info->metadata_ratio) { fs_info->data_chunk_allocations++; if (!(fs_info->data_chunk_allocations % fs_info->metadata_ratio)) @@ -3047,8 +3441,11 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans, } ret = btrfs_alloc_chunk(trans, extent_root, flags); + spin_lock(&space_info->lock); if (ret) space_info->full = 1; + space_info->force_alloc = 0; + spin_unlock(&space_info->lock); out: mutex_unlock(&extent_root->fs_info->chunk_mutex); return ret; @@ -3073,14 +3470,6 @@ static int update_block_group(struct btrfs_trans_handle *trans, else old_val -= num_bytes; btrfs_set_super_bytes_used(&info->super_copy, old_val); - - /* block accounting for root item */ - old_val = btrfs_root_used(&root->root_item); - if (alloc) - old_val += num_bytes; - else - old_val -= num_bytes; - btrfs_set_root_used(&root->root_item, old_val); spin_unlock(&info->delalloc_lock); while (total) { @@ -3306,6 +3695,14 @@ static int pin_down_bytes(struct btrfs_trans_handle *trans, if (is_data) goto pinit; + /* + * discard is sloooow, and so triggering discards on + * individual btree blocks isn't a good plan. Just + * pin everything in discard mode. + */ + if (btrfs_test_opt(root, DISCARD)) + goto pinit; + buf = btrfs_find_tree_block(root, bytenr, num_bytes); if (!buf) goto pinit; @@ -3660,6 +4057,21 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans, return ret; } +int btrfs_free_tree_block(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + u64 bytenr, u32 blocksize, + u64 parent, u64 root_objectid, int level) +{ + u64 used; + spin_lock(&root->node_lock); + used = btrfs_root_used(&root->root_item) - blocksize; + btrfs_set_root_used(&root->root_item, used); + spin_unlock(&root->node_lock); + + return btrfs_free_extent(trans, root, bytenr, blocksize, + parent, root_objectid, level, 0); +} + static u64 stripe_align(struct btrfs_root *root, u64 val) { u64 mask = ((u64)root->stripesize - 1); @@ -3713,7 +4125,7 @@ wait_block_group_cache_done(struct btrfs_block_group_cache *cache) } enum btrfs_loop_type { - LOOP_CACHED_ONLY = 0, + LOOP_FIND_IDEAL = 0, LOOP_CACHING_NOWAIT = 1, LOOP_CACHING_WAIT = 2, LOOP_ALLOC_CHUNK = 3, @@ -3742,11 +4154,15 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans, struct btrfs_block_group_cache *block_group = NULL; int empty_cluster = 2 * 1024 * 1024; int allowed_chunk_alloc = 0; + int done_chunk_alloc = 0; struct btrfs_space_info *space_info; int last_ptr_loop = 0; int loop = 0; bool found_uncached_bg = false; bool failed_cluster_refill = false; + bool failed_alloc = false; + u64 ideal_cache_percent = 0; + u64 ideal_cache_offset = 0; WARN_ON(num_bytes < root->sectorsize); btrfs_set_key_type(ins, BTRFS_EXTENT_ITEM_KEY); @@ -3782,14 +4198,19 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans, empty_cluster = 0; if (search_start == hint_byte) { +ideal_cache: block_group = btrfs_lookup_block_group(root->fs_info, search_start); /* * we don't want to use the block group if it doesn't match our * allocation bits, or if its not cached. + * + * However if we are re-searching with an ideal block group + * picked out then we don't care that the block group is cached. */ if (block_group && block_group_bits(block_group, data) && - block_group_cache_done(block_group)) { + (block_group->cached != BTRFS_CACHE_NO || + search_start == ideal_cache_offset)) { down_read(&space_info->groups_sem); if (list_empty(&block_group->list) || block_group->ro) { @@ -3801,46 +4222,63 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans, */ btrfs_put_block_group(block_group); up_read(&space_info->groups_sem); - } else + } else { goto have_block_group; + } } else if (block_group) { btrfs_put_block_group(block_group); } } - search: down_read(&space_info->groups_sem); list_for_each_entry(block_group, &space_info->block_groups, list) { u64 offset; int cached; - atomic_inc(&block_group->count); + btrfs_get_block_group(block_group); search_start = block_group->key.objectid; have_block_group: if (unlikely(block_group->cached == BTRFS_CACHE_NO)) { + u64 free_percent; + + free_percent = btrfs_block_group_used(&block_group->item); + free_percent *= 100; + free_percent = div64_u64(free_percent, + block_group->key.offset); + free_percent = 100 - free_percent; + if (free_percent > ideal_cache_percent && + likely(!block_group->ro)) { + ideal_cache_offset = block_group->key.objectid; + ideal_cache_percent = free_percent; + } + /* - * we want to start caching kthreads, but not too many - * right off the bat so we don't overwhelm the system, - * so only start them if there are less than 2 and we're - * in the initial allocation phase. + * We only want to start kthread caching if we are at + * the point where we will wait for caching to make + * progress, or if our ideal search is over and we've + * found somebody to start caching. */ if (loop > LOOP_CACHING_NOWAIT || - atomic_read(&space_info->caching_threads) < 2) { + (loop > LOOP_FIND_IDEAL && + atomic_read(&space_info->caching_threads) < 2)) { ret = cache_block_group(block_group); BUG_ON(ret); } - } - - cached = block_group_cache_done(block_group); - if (unlikely(!cached)) { found_uncached_bg = true; - /* if we only want cached bgs, loop */ - if (loop == LOOP_CACHED_ONLY) + /* + * If loop is set for cached only, try the next block + * group. + */ + if (loop == LOOP_FIND_IDEAL) goto loop; } + cached = block_group_cache_done(block_group); + if (unlikely(!cached)) + found_uncached_bg = true; + if (unlikely(block_group->ro)) goto loop; @@ -3885,7 +4323,7 @@ have_block_group: btrfs_put_block_group(block_group); block_group = last_ptr->block_group; - atomic_inc(&block_group->count); + btrfs_get_block_group(block_group); spin_unlock(&last_ptr->lock); spin_unlock(&last_ptr->refill_lock); @@ -3951,14 +4389,23 @@ refill_cluster: offset = btrfs_find_space_for_alloc(block_group, search_start, num_bytes, empty_size); - if (!offset && (cached || (!cached && - loop == LOOP_CACHING_NOWAIT))) { - goto loop; - } else if (!offset && (!cached && - loop > LOOP_CACHING_NOWAIT)) { + /* + * If we didn't find a chunk, and we haven't failed on this + * block group before, and this block group is in the middle of + * caching and we are ok with waiting, then go ahead and wait + * for progress to be made, and set failed_alloc to true. + * + * If failed_alloc is true then we've already waited on this + * block group once and should move on to the next block group. + */ + if (!offset && !failed_alloc && !cached && + loop > LOOP_CACHING_NOWAIT) { wait_block_group_cache_progress(block_group, - num_bytes + empty_size); + num_bytes + empty_size); + failed_alloc = true; goto have_block_group; + } else if (!offset) { + goto loop; } checks: search_start = stripe_align(root, offset); @@ -4006,13 +4453,16 @@ checks: break; loop: failed_cluster_refill = false; + failed_alloc = false; btrfs_put_block_group(block_group); } up_read(&space_info->groups_sem); - /* LOOP_CACHED_ONLY, only search fully cached block groups - * LOOP_CACHING_NOWAIT, search partially cached block groups, but - * dont wait foR them to finish caching + /* LOOP_FIND_IDEAL, only search caching/cached bg's, and don't wait for + * for them to make caching progress. Also + * determine the best possible bg to cache + * LOOP_CACHING_NOWAIT, search partially cached block groups, kicking + * caching kthreads as we move along * LOOP_CACHING_WAIT, search everything, and wait if our bg is caching * LOOP_ALLOC_CHUNK, force a chunk allocation and try again * LOOP_NO_EMPTY_SIZE, set empty_size and empty_cluster to 0 and try @@ -4021,12 +4471,47 @@ loop: if (!ins->objectid && loop < LOOP_NO_EMPTY_SIZE && (found_uncached_bg || empty_size || empty_cluster || allowed_chunk_alloc)) { - if (found_uncached_bg) { + if (loop == LOOP_FIND_IDEAL && found_uncached_bg) { found_uncached_bg = false; - if (loop < LOOP_CACHING_WAIT) { - loop++; + loop++; + if (!ideal_cache_percent && + atomic_read(&space_info->caching_threads)) goto search; - } + + /* + * 1 of the following 2 things have happened so far + * + * 1) We found an ideal block group for caching that + * is mostly full and will cache quickly, so we might + * as well wait for it. + * + * 2) We searched for cached only and we didn't find + * anything, and we didn't start any caching kthreads + * either, so chances are we will loop through and + * start a couple caching kthreads, and then come back + * around and just wait for them. This will be slower + * because we will have 2 caching kthreads reading at + * the same time when we could have just started one + * and waited for it to get far enough to give us an + * allocation, so go ahead and go to the wait caching + * loop. + */ + loop = LOOP_CACHING_WAIT; + search_start = ideal_cache_offset; + ideal_cache_percent = 0; + goto ideal_cache; + } else if (loop == LOOP_FIND_IDEAL) { + /* + * Didn't find a uncached bg, wait on anything we find + * next. + */ + loop = LOOP_CACHING_WAIT; + goto search; + } + + if (loop < LOOP_CACHING_WAIT) { + loop++; + goto search; } if (loop == LOOP_ALLOC_CHUNK) { @@ -4038,7 +4523,8 @@ loop: ret = do_chunk_alloc(trans, root, num_bytes + 2 * 1024 * 1024, data, 1); allowed_chunk_alloc = 0; - } else { + done_chunk_alloc = 1; + } else if (!done_chunk_alloc) { space_info->force_alloc = 1; } @@ -4063,21 +4549,32 @@ loop: return ret; } -static void dump_space_info(struct btrfs_space_info *info, u64 bytes) +static void dump_space_info(struct btrfs_space_info *info, u64 bytes, + int dump_block_groups) { struct btrfs_block_group_cache *cache; + spin_lock(&info->lock); printk(KERN_INFO "space_info has %llu free, is %sfull\n", (unsigned long long)(info->total_bytes - info->bytes_used - - info->bytes_pinned - info->bytes_reserved), + info->bytes_pinned - info->bytes_reserved - + info->bytes_super), (info->full) ? "" : "not "); printk(KERN_INFO "space_info total=%llu, pinned=%llu, delalloc=%llu," - " may_use=%llu, used=%llu\n", + " may_use=%llu, used=%llu, root=%llu, super=%llu, reserved=%llu" + "\n", (unsigned long long)info->total_bytes, (unsigned long long)info->bytes_pinned, (unsigned long long)info->bytes_delalloc, (unsigned long long)info->bytes_may_use, - (unsigned long long)info->bytes_used); + (unsigned long long)info->bytes_used, + (unsigned long long)info->bytes_root, + (unsigned long long)info->bytes_super, + (unsigned long long)info->bytes_reserved); + spin_unlock(&info->lock); + + if (!dump_block_groups) + return; down_read(&info->groups_sem); list_for_each_entry(cache, &info->block_groups, list) { @@ -4104,7 +4601,6 @@ int btrfs_reserve_extent(struct btrfs_trans_handle *trans, { int ret; u64 search_start = 0; - struct btrfs_fs_info *info = root->fs_info; data = btrfs_get_alloc_profile(root, data); again: @@ -4112,17 +4608,9 @@ again: * the only place that sets empty_size is btrfs_realloc_node, which * is not called recursively on allocations */ - if (empty_size || root->ref_cows) { - if (!(data & BTRFS_BLOCK_GROUP_METADATA)) { - ret = do_chunk_alloc(trans, root->fs_info->extent_root, - 2 * 1024 * 1024, - BTRFS_BLOCK_GROUP_METADATA | - (info->metadata_alloc_profile & - info->avail_metadata_alloc_bits), 0); - } + if (empty_size || root->ref_cows) ret = do_chunk_alloc(trans, root->fs_info->extent_root, num_bytes + 2 * 1024 * 1024, data, 0); - } WARN_ON(num_bytes < root->sectorsize); ret = find_free_extent(trans, root, num_bytes, empty_size, @@ -4145,7 +4633,7 @@ again: printk(KERN_ERR "btrfs allocation failed flags %llu, " "wanted %llu\n", (unsigned long long)data, (unsigned long long)num_bytes); - dump_space_info(sinfo, num_bytes); + dump_space_info(sinfo, num_bytes, 1); } return ret; @@ -4423,6 +4911,14 @@ static int alloc_tree_block(struct btrfs_trans_handle *trans, extent_op); BUG_ON(ret); } + + if (root_objectid == root->root_key.objectid) { + u64 used; + spin_lock(&root->node_lock); + used = btrfs_root_used(&root->root_item) + num_bytes; + btrfs_set_root_used(&root->root_item, used); + spin_unlock(&root->node_lock); + } return ret; } @@ -4445,8 +4941,16 @@ struct extent_buffer *btrfs_init_new_buffer(struct btrfs_trans_handle *trans, btrfs_set_buffer_uptodate(buf); if (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID) { - set_extent_dirty(&root->dirty_log_pages, buf->start, - buf->start + buf->len - 1, GFP_NOFS); + /* + * we allow two log transactions at a time, use different + * EXENT bit to differentiate dirty pages. + */ + if (root->log_transid % 2 == 0) + set_extent_dirty(&root->dirty_log_pages, buf->start, + buf->start + buf->len - 1, GFP_NOFS); + else + set_extent_new(&root->dirty_log_pages, buf->start, + buf->start + buf->len - 1, GFP_NOFS); } else { set_extent_dirty(&trans->transaction->dirty_pages, buf->start, buf->start + buf->len - 1, GFP_NOFS); @@ -4506,6 +5010,7 @@ static noinline void reada_walk_down(struct btrfs_trans_handle *trans, u64 bytenr; u64 generation; u64 refs; + u64 flags; u64 last = 0; u32 nritems; u32 blocksize; @@ -4543,15 +5048,19 @@ static noinline void reada_walk_down(struct btrfs_trans_handle *trans, generation <= root->root_key.offset) continue; + /* We don't lock the tree block, it's OK to be racy here */ + ret = btrfs_lookup_extent_info(trans, root, bytenr, blocksize, + &refs, &flags); + BUG_ON(ret); + BUG_ON(refs == 0); + if (wc->stage == DROP_REFERENCE) { - ret = btrfs_lookup_extent_info(trans, root, - bytenr, blocksize, - &refs, NULL); - BUG_ON(ret); - BUG_ON(refs == 0); if (refs == 1) goto reada; + if (wc->level == 1 && + (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)) + continue; if (!wc->update_ref || generation <= root->root_key.offset) continue; @@ -4560,6 +5069,10 @@ static noinline void reada_walk_down(struct btrfs_trans_handle *trans, &wc->update_progress); if (ret < 0) continue; + } else { + if (wc->level == 1 && + (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)) + continue; } reada: ret = readahead_tree_block(root, bytenr, blocksize, @@ -4583,7 +5096,7 @@ reada: static noinline int walk_down_proc(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_path *path, - struct walk_control *wc) + struct walk_control *wc, int lookup_info) { int level = wc->level; struct extent_buffer *eb = path->nodes[level]; @@ -4598,8 +5111,9 @@ static noinline int walk_down_proc(struct btrfs_trans_handle *trans, * when reference count of tree block is 1, it won't increase * again. once full backref flag is set, we never clear it. */ - if ((wc->stage == DROP_REFERENCE && wc->refs[level] != 1) || - (wc->stage == UPDATE_BACKREF && !(wc->flags[level] & flag))) { + if (lookup_info && + ((wc->stage == DROP_REFERENCE && wc->refs[level] != 1) || + (wc->stage == UPDATE_BACKREF && !(wc->flags[level] & flag)))) { BUG_ON(!path->locks[level]); ret = btrfs_lookup_extent_info(trans, root, eb->start, eb->len, @@ -4660,7 +5174,7 @@ static noinline int walk_down_proc(struct btrfs_trans_handle *trans, static noinline int do_walk_down(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_path *path, - struct walk_control *wc) + struct walk_control *wc, int *lookup_info) { u64 bytenr; u64 generation; @@ -4680,8 +5194,10 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans, * for the subtree */ if (wc->stage == UPDATE_BACKREF && - generation <= root->root_key.offset) + generation <= root->root_key.offset) { + *lookup_info = 1; return 1; + } bytenr = btrfs_node_blockptr(path->nodes[level], path->slots[level]); blocksize = btrfs_level_size(root, level - 1); @@ -4694,14 +5210,19 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans, btrfs_tree_lock(next); btrfs_set_lock_blocking(next); - if (wc->stage == DROP_REFERENCE) { - ret = btrfs_lookup_extent_info(trans, root, bytenr, blocksize, - &wc->refs[level - 1], - &wc->flags[level - 1]); - BUG_ON(ret); - BUG_ON(wc->refs[level - 1] == 0); + ret = btrfs_lookup_extent_info(trans, root, bytenr, blocksize, + &wc->refs[level - 1], + &wc->flags[level - 1]); + BUG_ON(ret); + BUG_ON(wc->refs[level - 1] == 0); + *lookup_info = 0; + if (wc->stage == DROP_REFERENCE) { if (wc->refs[level - 1] > 1) { + if (level == 1 && + (wc->flags[0] & BTRFS_BLOCK_FLAG_FULL_BACKREF)) + goto skip; + if (!wc->update_ref || generation <= root->root_key.offset) goto skip; @@ -4715,12 +5236,17 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans, wc->stage = UPDATE_BACKREF; wc->shared_level = level - 1; } + } else { + if (level == 1 && + (wc->flags[0] & BTRFS_BLOCK_FLAG_FULL_BACKREF)) + goto skip; } if (!btrfs_buffer_uptodate(next, generation)) { btrfs_tree_unlock(next); free_extent_buffer(next); next = NULL; + *lookup_info = 1; } if (!next) { @@ -4743,21 +5269,22 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans, skip: wc->refs[level - 1] = 0; wc->flags[level - 1] = 0; + if (wc->stage == DROP_REFERENCE) { + if (wc->flags[level] & BTRFS_BLOCK_FLAG_FULL_BACKREF) { + parent = path->nodes[level]->start; + } else { + BUG_ON(root->root_key.objectid != + btrfs_header_owner(path->nodes[level])); + parent = 0; + } - if (wc->flags[level] & BTRFS_BLOCK_FLAG_FULL_BACKREF) { - parent = path->nodes[level]->start; - } else { - BUG_ON(root->root_key.objectid != - btrfs_header_owner(path->nodes[level])); - parent = 0; + ret = btrfs_free_extent(trans, root, bytenr, blocksize, parent, + root->root_key.objectid, level - 1, 0); + BUG_ON(ret); } - - ret = btrfs_free_extent(trans, root, bytenr, blocksize, parent, - root->root_key.objectid, level - 1, 0); - BUG_ON(ret); - btrfs_tree_unlock(next); free_extent_buffer(next); + *lookup_info = 1; return 1; } @@ -4871,21 +5398,22 @@ static noinline int walk_down_tree(struct btrfs_trans_handle *trans, struct walk_control *wc) { int level = wc->level; + int lookup_info = 1; int ret; while (level >= 0) { - if (path->slots[level] >= - btrfs_header_nritems(path->nodes[level])) - break; - - ret = walk_down_proc(trans, root, path, wc); + ret = walk_down_proc(trans, root, path, wc, lookup_info); if (ret > 0) break; if (level == 0) break; - ret = do_walk_down(trans, root, path, wc); + if (path->slots[level] >= + btrfs_header_nritems(path->nodes[level])) + break; + + ret = do_walk_down(trans, root, path, wc, &lookup_info); if (ret > 0) { path->slots[level]++; continue; @@ -6033,6 +6561,7 @@ static noinline int invalidate_extent_cache(struct btrfs_root *root, struct btrfs_key key; struct inode *inode = NULL; struct btrfs_file_extent_item *fi; + struct extent_state *cached_state = NULL; u64 num_bytes; u64 skip_objectid = 0; u32 nritems; @@ -6061,12 +6590,14 @@ static noinline int invalidate_extent_cache(struct btrfs_root *root, } num_bytes = btrfs_file_extent_num_bytes(leaf, fi); - lock_extent(&BTRFS_I(inode)->io_tree, key.offset, - key.offset + num_bytes - 1, GFP_NOFS); + lock_extent_bits(&BTRFS_I(inode)->io_tree, key.offset, + key.offset + num_bytes - 1, 0, &cached_state, + GFP_NOFS); btrfs_drop_extent_cache(inode, key.offset, key.offset + num_bytes - 1, 1); - unlock_extent(&BTRFS_I(inode)->io_tree, key.offset, - key.offset + num_bytes - 1, GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, key.offset, + key.offset + num_bytes - 1, &cached_state, + GFP_NOFS); cond_resched(); } iput(inode); @@ -6875,9 +7406,7 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info) wait_block_group_cache_done(block_group); btrfs_remove_free_space_cache(block_group); - - WARN_ON(atomic_read(&block_group->count) != 1); - kfree(block_group); + btrfs_put_block_group(block_group); spin_lock(&info->block_group_cache_lock); } diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 0cb88f8146ea..c99121ac5d6b 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -104,8 +104,8 @@ void extent_io_exit(void) void extent_io_tree_init(struct extent_io_tree *tree, struct address_space *mapping, gfp_t mask) { - tree->state.rb_node = NULL; - tree->buffer.rb_node = NULL; + tree->state = RB_ROOT; + tree->buffer = RB_ROOT; tree->ops = NULL; tree->dirty_bytes = 0; spin_lock_init(&tree->lock); @@ -280,6 +280,14 @@ static struct extent_buffer *buffer_search(struct extent_io_tree *tree, return NULL; } +static void merge_cb(struct extent_io_tree *tree, struct extent_state *new, + struct extent_state *other) +{ + if (tree->ops && tree->ops->merge_extent_hook) + tree->ops->merge_extent_hook(tree->mapping->host, new, + other); +} + /* * utility function to look for merge candidates inside a given range. * Any extents with matching state are merged together into a single @@ -303,6 +311,7 @@ static int merge_state(struct extent_io_tree *tree, other = rb_entry(other_node, struct extent_state, rb_node); if (other->end == state->start - 1 && other->state == state->state) { + merge_cb(tree, state, other); state->start = other->start; other->tree = NULL; rb_erase(&other->rb_node, &tree->state); @@ -314,33 +323,37 @@ static int merge_state(struct extent_io_tree *tree, other = rb_entry(other_node, struct extent_state, rb_node); if (other->start == state->end + 1 && other->state == state->state) { + merge_cb(tree, state, other); other->start = state->start; state->tree = NULL; rb_erase(&state->rb_node, &tree->state); free_extent_state(state); + state = NULL; } } + return 0; } -static void set_state_cb(struct extent_io_tree *tree, +static int set_state_cb(struct extent_io_tree *tree, struct extent_state *state, unsigned long bits) { if (tree->ops && tree->ops->set_bit_hook) { - tree->ops->set_bit_hook(tree->mapping->host, state->start, - state->end, state->state, bits); + return tree->ops->set_bit_hook(tree->mapping->host, + state->start, state->end, + state->state, bits); } + + return 0; } static void clear_state_cb(struct extent_io_tree *tree, struct extent_state *state, unsigned long bits) { - if (tree->ops && tree->ops->clear_bit_hook) { - tree->ops->clear_bit_hook(tree->mapping->host, state->start, - state->end, state->state, bits); - } + if (tree->ops && tree->ops->clear_bit_hook) + tree->ops->clear_bit_hook(tree->mapping->host, state, bits); } /* @@ -358,6 +371,7 @@ static int insert_state(struct extent_io_tree *tree, int bits) { struct rb_node *node; + int ret; if (end < start) { printk(KERN_ERR "btrfs end < start %llu %llu\n", @@ -365,11 +379,14 @@ static int insert_state(struct extent_io_tree *tree, (unsigned long long)start); WARN_ON(1); } - if (bits & EXTENT_DIRTY) - tree->dirty_bytes += end - start + 1; state->start = start; state->end = end; - set_state_cb(tree, state, bits); + ret = set_state_cb(tree, state, bits); + if (ret) + return ret; + + if (bits & EXTENT_DIRTY) + tree->dirty_bytes += end - start + 1; state->state |= bits; node = tree_insert(&tree->state, end, &state->rb_node); if (node) { @@ -387,6 +404,15 @@ static int insert_state(struct extent_io_tree *tree, return 0; } +static int split_cb(struct extent_io_tree *tree, struct extent_state *orig, + u64 split) +{ + if (tree->ops && tree->ops->split_extent_hook) + return tree->ops->split_extent_hook(tree->mapping->host, + orig, split); + return 0; +} + /* * split a given extent state struct in two, inserting the preallocated * struct 'prealloc' as the newly created second half. 'split' indicates an @@ -405,6 +431,9 @@ static int split_state(struct extent_io_tree *tree, struct extent_state *orig, struct extent_state *prealloc, u64 split) { struct rb_node *node; + + split_cb(tree, orig, split); + prealloc->start = orig->start; prealloc->end = split - 1; prealloc->state = orig->state; @@ -431,7 +460,8 @@ static int clear_state_bit(struct extent_io_tree *tree, struct extent_state *state, int bits, int wake, int delete) { - int ret = state->state & bits; + int bits_to_clear = bits & ~EXTENT_DO_ACCOUNTING; + int ret = state->state & bits_to_clear; if ((bits & EXTENT_DIRTY) && (state->state & EXTENT_DIRTY)) { u64 range = state->end - state->start + 1; @@ -439,7 +469,7 @@ static int clear_state_bit(struct extent_io_tree *tree, tree->dirty_bytes -= range; } clear_state_cb(tree, state, bits); - state->state &= ~bits; + state->state &= ~bits_to_clear; if (wake) wake_up(&state->wq); if (delete || state->state == 0) { @@ -483,7 +513,10 @@ int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, u64 last_end; int err; int set = 0; + int clear = 0; + if (bits & (EXTENT_IOBITS | EXTENT_BOUNDARY)) + clear = 1; again: if (!prealloc && (mask & __GFP_WAIT)) { prealloc = alloc_extent_state(mask); @@ -494,14 +527,20 @@ again: spin_lock(&tree->lock); if (cached_state) { cached = *cached_state; - *cached_state = NULL; - cached_state = NULL; + + if (clear) { + *cached_state = NULL; + cached_state = NULL; + } + if (cached && cached->tree && cached->start == start) { - atomic_dec(&cached->refs); + if (clear) + atomic_dec(&cached->refs); state = cached; goto hit_next; } - free_extent_state(cached); + if (clear) + free_extent_state(cached); } /* * this search will find the extents that end after @@ -542,8 +581,8 @@ hit_next: if (err) goto out; if (state->end <= end) { - set |= clear_state_bit(tree, state, bits, - wake, delete); + set |= clear_state_bit(tree, state, bits, wake, + delete); if (last_end == (u64)-1) goto out; start = last_end + 1; @@ -561,12 +600,11 @@ hit_next: prealloc = alloc_extent_state(GFP_ATOMIC); err = split_state(tree, state, prealloc, end + 1); BUG_ON(err == -EEXIST); - if (wake) wake_up(&state->wq); - set |= clear_state_bit(tree, prealloc, bits, - wake, delete); + set |= clear_state_bit(tree, prealloc, bits, wake, delete); + prealloc = NULL; goto out; } @@ -667,16 +705,23 @@ out: return 0; } -static void set_state_bits(struct extent_io_tree *tree, +static int set_state_bits(struct extent_io_tree *tree, struct extent_state *state, int bits) { + int ret; + + ret = set_state_cb(tree, state, bits); + if (ret) + return ret; + if ((bits & EXTENT_DIRTY) && !(state->state & EXTENT_DIRTY)) { u64 range = state->end - state->start + 1; tree->dirty_bytes += range; } - set_state_cb(tree, state, bits); state->state |= bits; + + return 0; } static void cache_state(struct extent_state *state, @@ -758,7 +803,10 @@ hit_next: goto out; } - set_state_bits(tree, state, bits); + err = set_state_bits(tree, state, bits); + if (err) + goto out; + cache_state(state, cached_state); merge_state(tree, state); if (last_end == (u64)-1) @@ -805,7 +853,9 @@ hit_next: if (err) goto out; if (state->end <= end) { - set_state_bits(tree, state, bits); + err = set_state_bits(tree, state, bits); + if (err) + goto out; cache_state(state, cached_state); merge_state(tree, state); if (last_end == (u64)-1) @@ -829,11 +879,13 @@ hit_next: this_end = last_start - 1; err = insert_state(tree, prealloc, start, this_end, bits); - cache_state(prealloc, cached_state); - prealloc = NULL; BUG_ON(err == -EEXIST); - if (err) + if (err) { + prealloc = NULL; goto out; + } + cache_state(prealloc, cached_state); + prealloc = NULL; start = this_end + 1; goto search_again; } @@ -852,7 +904,11 @@ hit_next: err = split_state(tree, state, prealloc, end + 1); BUG_ON(err == -EEXIST); - set_state_bits(tree, prealloc, bits); + err = set_state_bits(tree, prealloc, bits); + if (err) { + prealloc = NULL; + goto out; + } cache_state(prealloc, cached_state); merge_state(tree, prealloc); prealloc = NULL; @@ -899,18 +955,19 @@ int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end, } int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end, - gfp_t mask) + struct extent_state **cached_state, gfp_t mask) { return set_extent_bit(tree, start, end, EXTENT_DELALLOC | EXTENT_DIRTY | EXTENT_UPTODATE, - 0, NULL, NULL, mask); + 0, NULL, cached_state, mask); } int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask) { return clear_extent_bit(tree, start, end, - EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, + EXTENT_DIRTY | EXTENT_DELALLOC | + EXTENT_DO_ACCOUNTING, 0, 0, NULL, mask); } @@ -936,10 +993,11 @@ int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end, } static int clear_extent_uptodate(struct extent_io_tree *tree, u64 start, - u64 end, gfp_t mask) + u64 end, struct extent_state **cached_state, + gfp_t mask) { return clear_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, 0, - NULL, mask); + cached_state, mask); } int wait_on_extent_writeback(struct extent_io_tree *tree, u64 start, u64 end) @@ -1123,7 +1181,8 @@ out: * 1 is returned if we find something, 0 if nothing was in the tree */ static noinline u64 find_delalloc_range(struct extent_io_tree *tree, - u64 *start, u64 *end, u64 max_bytes) + u64 *start, u64 *end, u64 max_bytes, + struct extent_state **cached_state) { struct rb_node *node; struct extent_state *state; @@ -1155,8 +1214,11 @@ static noinline u64 find_delalloc_range(struct extent_io_tree *tree, *end = state->end; goto out; } - if (!found) + if (!found) { *start = state->start; + *cached_state = state; + atomic_inc(&state->refs); + } found++; *end = state->end; cur_start = state->end + 1; @@ -1288,10 +1350,11 @@ again: delalloc_start = *start; delalloc_end = 0; found = find_delalloc_range(tree, &delalloc_start, &delalloc_end, - max_bytes); + max_bytes, &cached_state); if (!found || delalloc_end <= *start) { *start = delalloc_start; *end = delalloc_end; + free_extent_state(cached_state); return found; } @@ -1355,12 +1418,7 @@ out_failed: int extent_clear_unlock_delalloc(struct inode *inode, struct extent_io_tree *tree, u64 start, u64 end, struct page *locked_page, - int unlock_pages, - int clear_unlock, - int clear_delalloc, int clear_dirty, - int set_writeback, - int end_writeback, - int set_private2) + unsigned long op) { int ret; struct page *pages[16]; @@ -1370,17 +1428,21 @@ int extent_clear_unlock_delalloc(struct inode *inode, int i; int clear_bits = 0; - if (clear_unlock) + if (op & EXTENT_CLEAR_UNLOCK) clear_bits |= EXTENT_LOCKED; - if (clear_dirty) + if (op & EXTENT_CLEAR_DIRTY) clear_bits |= EXTENT_DIRTY; - if (clear_delalloc) + if (op & EXTENT_CLEAR_DELALLOC) clear_bits |= EXTENT_DELALLOC; + if (op & EXTENT_CLEAR_ACCOUNTING) + clear_bits |= EXTENT_DO_ACCOUNTING; + clear_extent_bit(tree, start, end, clear_bits, 1, 0, NULL, GFP_NOFS); - if (!(unlock_pages || clear_dirty || set_writeback || end_writeback || - set_private2)) + if (!(op & (EXTENT_CLEAR_UNLOCK_PAGE | EXTENT_CLEAR_DIRTY | + EXTENT_SET_WRITEBACK | EXTENT_END_WRITEBACK | + EXTENT_SET_PRIVATE2))) return 0; while (nr_pages > 0) { @@ -1389,20 +1451,20 @@ int extent_clear_unlock_delalloc(struct inode *inode, nr_pages, ARRAY_SIZE(pages)), pages); for (i = 0; i < ret; i++) { - if (set_private2) + if (op & EXTENT_SET_PRIVATE2) SetPagePrivate2(pages[i]); if (pages[i] == locked_page) { page_cache_release(pages[i]); continue; } - if (clear_dirty) + if (op & EXTENT_CLEAR_DIRTY) clear_page_dirty_for_io(pages[i]); - if (set_writeback) + if (op & EXTENT_SET_WRITEBACK) set_page_writeback(pages[i]); - if (end_writeback) + if (op & EXTENT_END_WRITEBACK) end_page_writeback(pages[i]); - if (unlock_pages) + if (op & EXTENT_CLEAR_UNLOCK_PAGE) unlock_page(pages[i]); page_cache_release(pages[i]); } @@ -1675,7 +1737,7 @@ static void end_bio_extent_writepage(struct bio *bio, int err) } if (!uptodate) { - clear_extent_uptodate(tree, start, end, GFP_NOFS); + clear_extent_uptodate(tree, start, end, NULL, GFP_NOFS); ClearPageUptodate(page); SetPageError(page); } @@ -1703,7 +1765,8 @@ static void end_bio_extent_writepage(struct bio *bio, int err) static void end_bio_extent_readpage(struct bio *bio, int err) { int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); - struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1; + struct bio_vec *bvec_end = bio->bi_io_vec + bio->bi_vcnt - 1; + struct bio_vec *bvec = bio->bi_io_vec; struct extent_io_tree *tree; u64 start; u64 end; @@ -1726,7 +1789,7 @@ static void end_bio_extent_readpage(struct bio *bio, int err) else whole_page = 0; - if (--bvec >= bio->bi_io_vec) + if (++bvec <= bvec_end) prefetchw(&bvec->bv_page->flags); if (uptodate && tree->ops && tree->ops->readpage_end_io_hook) { @@ -1771,7 +1834,7 @@ static void end_bio_extent_readpage(struct bio *bio, int err) } check_page_locked(tree, page); } - } while (bvec >= bio->bi_io_vec); + } while (bvec <= bvec_end); bio_put(bio); } @@ -2657,6 +2720,7 @@ int extent_readpages(struct extent_io_tree *tree, int extent_invalidatepage(struct extent_io_tree *tree, struct page *page, unsigned long offset) { + struct extent_state *cached_state = NULL; u64 start = ((u64)page->index << PAGE_CACHE_SHIFT); u64 end = start + PAGE_CACHE_SIZE - 1; size_t blocksize = page->mapping->host->i_sb->s_blocksize; @@ -2665,11 +2729,12 @@ int extent_invalidatepage(struct extent_io_tree *tree, if (start > end) return 0; - lock_extent(tree, start, end, GFP_NOFS); + lock_extent_bits(tree, start, end, 0, &cached_state, GFP_NOFS); wait_on_page_writeback(page); clear_extent_bit(tree, start, end, - EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC, - 1, 1, NULL, GFP_NOFS); + EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC | + EXTENT_DO_ACCOUNTING, + 1, 1, &cached_state, GFP_NOFS); return 0; } @@ -2872,16 +2937,17 @@ sector_t extent_bmap(struct address_space *mapping, sector_t iblock, get_extent_t *get_extent) { struct inode *inode = mapping->host; + struct extent_state *cached_state = NULL; u64 start = iblock << inode->i_blkbits; sector_t sector = 0; size_t blksize = (1 << inode->i_blkbits); struct extent_map *em; - lock_extent(&BTRFS_I(inode)->io_tree, start, start + blksize - 1, - GFP_NOFS); + lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + blksize - 1, + 0, &cached_state, GFP_NOFS); em = get_extent(inode, NULL, 0, start, blksize, 0); - unlock_extent(&BTRFS_I(inode)->io_tree, start, start + blksize - 1, - GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, start, + start + blksize - 1, &cached_state, GFP_NOFS); if (!em || IS_ERR(em)) return 0; @@ -2903,6 +2969,7 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, u32 flags = 0; u64 disko = 0; struct extent_map *em = NULL; + struct extent_state *cached_state = NULL; int end = 0; u64 em_start = 0, em_len = 0; unsigned long emflags; @@ -2911,8 +2978,8 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, if (len == 0) return -EINVAL; - lock_extent(&BTRFS_I(inode)->io_tree, start, start + len, - GFP_NOFS); + lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + len, 0, + &cached_state, GFP_NOFS); em = get_extent(inode, NULL, 0, off, max - off, 0); if (!em) goto out; @@ -2975,8 +3042,8 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, out_free: free_extent_map(em); out: - unlock_extent(&BTRFS_I(inode)->io_tree, start, start + len, - GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, start, start + len, + &cached_state, GFP_NOFS); return ret; } @@ -3117,10 +3184,9 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree, spin_unlock(&tree->buffer_lock); goto free_eb; } - spin_unlock(&tree->buffer_lock); - /* add one reference for the tree */ atomic_inc(&eb->refs); + spin_unlock(&tree->buffer_lock); return eb; free_eb: @@ -3217,7 +3283,8 @@ int set_extent_buffer_dirty(struct extent_io_tree *tree, } int clear_extent_buffer_uptodate(struct extent_io_tree *tree, - struct extent_buffer *eb) + struct extent_buffer *eb, + struct extent_state **cached_state) { unsigned long i; struct page *page; @@ -3227,7 +3294,7 @@ int clear_extent_buffer_uptodate(struct extent_io_tree *tree, clear_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags); clear_extent_uptodate(tree, eb->start, eb->start + eb->len - 1, - GFP_NOFS); + cached_state, GFP_NOFS); for (i = 0; i < num_pages; i++) { page = extent_buffer_page(eb, i); if (page) @@ -3287,7 +3354,8 @@ int extent_range_uptodate(struct extent_io_tree *tree, } int extent_buffer_uptodate(struct extent_io_tree *tree, - struct extent_buffer *eb) + struct extent_buffer *eb, + struct extent_state *cached_state) { int ret = 0; unsigned long num_pages; @@ -3299,7 +3367,7 @@ int extent_buffer_uptodate(struct extent_io_tree *tree, return 1; ret = test_range_bit(tree, eb->start, eb->start + eb->len - 1, - EXTENT_UPTODATE, 1, NULL); + EXTENT_UPTODATE, 1, cached_state); if (ret) return ret; diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h index 14ed16fd862d..bbab4813646f 100644 --- a/fs/btrfs/extent_io.h +++ b/fs/btrfs/extent_io.h @@ -15,6 +15,7 @@ #define EXTENT_BUFFER_FILLED (1 << 8) #define EXTENT_BOUNDARY (1 << 9) #define EXTENT_NODATASUM (1 << 10) +#define EXTENT_DO_ACCOUNTING (1 << 11) #define EXTENT_IOBITS (EXTENT_LOCKED | EXTENT_WRITEBACK) /* flags for bio submission */ @@ -25,6 +26,16 @@ #define EXTENT_BUFFER_BLOCKING 1 #define EXTENT_BUFFER_DIRTY 2 +/* these are flags for extent_clear_unlock_delalloc */ +#define EXTENT_CLEAR_UNLOCK_PAGE 0x1 +#define EXTENT_CLEAR_UNLOCK 0x2 +#define EXTENT_CLEAR_DELALLOC 0x4 +#define EXTENT_CLEAR_DIRTY 0x8 +#define EXTENT_SET_WRITEBACK 0x10 +#define EXTENT_END_WRITEBACK 0x20 +#define EXTENT_SET_PRIVATE2 0x40 +#define EXTENT_CLEAR_ACCOUNTING 0x80 + /* * page->private values. Every page that is controlled by the extent * map has page->private set to one. @@ -60,8 +71,13 @@ struct extent_io_ops { struct extent_state *state, int uptodate); int (*set_bit_hook)(struct inode *inode, u64 start, u64 end, unsigned long old, unsigned long bits); - int (*clear_bit_hook)(struct inode *inode, u64 start, u64 end, - unsigned long old, unsigned long bits); + int (*clear_bit_hook)(struct inode *inode, struct extent_state *state, + unsigned long bits); + int (*merge_extent_hook)(struct inode *inode, + struct extent_state *new, + struct extent_state *other); + int (*split_extent_hook)(struct inode *inode, + struct extent_state *orig, u64 split); int (*write_cache_pages_lock_hook)(struct page *page); }; @@ -79,10 +95,14 @@ struct extent_state { u64 start; u64 end; /* inclusive */ struct rb_node rb_node; + + /* ADD NEW ELEMENTS AFTER THIS */ struct extent_io_tree *tree; wait_queue_head_t wq; atomic_t refs; unsigned long state; + u64 split_start; + u64 split_end; /* for use by the FS */ u64 private; @@ -143,6 +163,8 @@ int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask); int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end, int bits, struct extent_state **cached, gfp_t mask); int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask); +int unlock_extent_cached(struct extent_io_tree *tree, u64 start, u64 end, + struct extent_state **cached, gfp_t mask); int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask); int extent_read_full_page(struct extent_io_tree *tree, struct page *page, @@ -176,7 +198,7 @@ int clear_extent_ordered(struct extent_io_tree *tree, u64 start, u64 end, int clear_extent_ordered_metadata(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask); int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end, - gfp_t mask); + struct extent_state **cached_state, gfp_t mask); int set_extent_ordered(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask); int find_first_extent_bit(struct extent_io_tree *tree, u64 start, @@ -261,9 +283,11 @@ int test_extent_buffer_dirty(struct extent_io_tree *tree, int set_extent_buffer_uptodate(struct extent_io_tree *tree, struct extent_buffer *eb); int clear_extent_buffer_uptodate(struct extent_io_tree *tree, - struct extent_buffer *eb); + struct extent_buffer *eb, + struct extent_state **cached_state); int extent_buffer_uptodate(struct extent_io_tree *tree, - struct extent_buffer *eb); + struct extent_buffer *eb, + struct extent_state *cached_state); int map_extent_buffer(struct extent_buffer *eb, unsigned long offset, unsigned long min_len, char **token, char **map, unsigned long *map_start, @@ -279,10 +303,5 @@ int extent_range_uptodate(struct extent_io_tree *tree, int extent_clear_unlock_delalloc(struct inode *inode, struct extent_io_tree *tree, u64 start, u64 end, struct page *locked_page, - int unlock_page, - int clear_unlock, - int clear_delalloc, int clear_dirty, - int set_writeback, - int end_writeback, - int set_private2); + unsigned long op); #endif diff --git a/fs/btrfs/extent_map.c b/fs/btrfs/extent_map.c index 2c726b7b9faa..28d87ba60ce8 100644 --- a/fs/btrfs/extent_map.c +++ b/fs/btrfs/extent_map.c @@ -35,7 +35,7 @@ void extent_map_exit(void) */ void extent_map_tree_init(struct extent_map_tree *tree, gfp_t mask) { - tree->map.rb_node = NULL; + tree->map = RB_ROOT; rwlock_init(&tree->lock); } @@ -155,20 +155,6 @@ static struct rb_node *__tree_search(struct rb_root *root, u64 offset, return NULL; } -/* - * look for an offset in the tree, and if it can't be found, return - * the first offset we can find smaller than 'offset'. - */ -static inline struct rb_node *tree_search(struct rb_root *root, u64 offset) -{ - struct rb_node *prev; - struct rb_node *ret; - ret = __tree_search(root, offset, &prev, NULL); - if (!ret) - return prev; - return ret; -} - /* check to see if two extent_map structs are adjacent and safe to merge */ static int mergable_maps(struct extent_map *prev, struct extent_map *next) { @@ -208,7 +194,7 @@ int unpin_extent_cache(struct extent_map_tree *tree, u64 start, u64 len) write_lock(&tree->lock); em = lookup_extent_mapping(tree, start, len); - WARN_ON(em->start != start || !em); + WARN_ON(!em || em->start != start); if (!em) goto out; @@ -256,7 +242,7 @@ out: * Insert @em into @tree or perform a simple forward/backward merge with * existing mappings. The extent_map struct passed in will be inserted * into the tree directly, with an additional reference taken, or a - * reference dropped if the merge attempt was sucessfull. + * reference dropped if the merge attempt was successfull. */ int add_extent_mapping(struct extent_map_tree *tree, struct extent_map *em) diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index a3492a3ad96b..ee3323c7fc1c 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -123,7 +123,11 @@ static noinline int dirty_and_release_pages(struct btrfs_trans_handle *trans, root->sectorsize - 1) & ~((u64)root->sectorsize - 1); end_of_last_block = start_pos + num_bytes - 1; - btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block); + err = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block, + NULL); + if (err) + return err; + for (i = 0; i < num_pages; i++) { struct page *p = pages[i]; SetPageUptodate(p); @@ -176,18 +180,14 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end, } flags = em->flags; if (skip_pinned && test_bit(EXTENT_FLAG_PINNED, &em->flags)) { - if (em->start <= start && - (!testend || em->start + em->len >= start + len)) { + if (testend && em->start + em->len >= start + len) { free_extent_map(em); write_unlock(&em_tree->lock); break; } - if (start < em->start) { - len = em->start - start; - } else { + start = em->start + em->len; + if (testend) len = start + len - (em->start + em->len); - start = em->start + em->len; - } free_extent_map(em); write_unlock(&em_tree->lock); continue; @@ -262,324 +262,253 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end, * If an extent intersects the range but is not entirely inside the range * it is either truncated or split. Anything entirely inside the range * is deleted from the tree. - * - * inline_limit is used to tell this code which offsets in the file to keep - * if they contain inline extents. */ -noinline int btrfs_drop_extents(struct btrfs_trans_handle *trans, - struct btrfs_root *root, struct inode *inode, - u64 start, u64 end, u64 locked_end, - u64 inline_limit, u64 *hint_byte, int drop_cache) +int btrfs_drop_extents(struct btrfs_trans_handle *trans, struct inode *inode, + u64 start, u64 end, u64 *hint_byte, int drop_cache) { - u64 extent_end = 0; - u64 search_start = start; - u64 ram_bytes = 0; - u64 disk_bytenr = 0; - u64 orig_locked_end = locked_end; - u8 compression; - u8 encryption; - u16 other_encoding = 0; + struct btrfs_root *root = BTRFS_I(inode)->root; struct extent_buffer *leaf; - struct btrfs_file_extent_item *extent; + struct btrfs_file_extent_item *fi; struct btrfs_path *path; struct btrfs_key key; - struct btrfs_file_extent_item old; - int keep; - int slot; - int bookend; - int found_type = 0; - int found_extent; - int found_inline; + struct btrfs_key new_key; + u64 search_start = start; + u64 disk_bytenr = 0; + u64 num_bytes = 0; + u64 extent_offset = 0; + u64 extent_end = 0; + int del_nr = 0; + int del_slot = 0; + int extent_type; int recow; int ret; - inline_limit = 0; if (drop_cache) btrfs_drop_extent_cache(inode, start, end - 1, 0); path = btrfs_alloc_path(); if (!path) return -ENOMEM; + while (1) { recow = 0; - btrfs_release_path(root, path); ret = btrfs_lookup_file_extent(trans, root, path, inode->i_ino, search_start, -1); if (ret < 0) - goto out; - if (ret > 0) { - if (path->slots[0] == 0) { - ret = 0; - goto out; - } - path->slots[0]--; + break; + if (ret > 0 && path->slots[0] > 0 && search_start == start) { + leaf = path->nodes[0]; + btrfs_item_key_to_cpu(leaf, &key, path->slots[0] - 1); + if (key.objectid == inode->i_ino && + key.type == BTRFS_EXTENT_DATA_KEY) + path->slots[0]--; } + ret = 0; next_slot: - keep = 0; - bookend = 0; - found_extent = 0; - found_inline = 0; - compression = 0; - encryption = 0; - extent = NULL; leaf = path->nodes[0]; - slot = path->slots[0]; - ret = 0; - btrfs_item_key_to_cpu(leaf, &key, slot); - if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY && - key.offset >= end) { - goto out; - } - if (btrfs_key_type(&key) > BTRFS_EXTENT_DATA_KEY || - key.objectid != inode->i_ino) { - goto out; - } - if (recow) { - search_start = max(key.offset, start); - continue; - } - if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) { - extent = btrfs_item_ptr(leaf, slot, - struct btrfs_file_extent_item); - found_type = btrfs_file_extent_type(leaf, extent); - compression = btrfs_file_extent_compression(leaf, - extent); - encryption = btrfs_file_extent_encryption(leaf, - extent); - other_encoding = btrfs_file_extent_other_encoding(leaf, - extent); - if (found_type == BTRFS_FILE_EXTENT_REG || - found_type == BTRFS_FILE_EXTENT_PREALLOC) { - extent_end = - btrfs_file_extent_disk_bytenr(leaf, - extent); - if (extent_end) - *hint_byte = extent_end; - - extent_end = key.offset + - btrfs_file_extent_num_bytes(leaf, extent); - ram_bytes = btrfs_file_extent_ram_bytes(leaf, - extent); - found_extent = 1; - } else if (found_type == BTRFS_FILE_EXTENT_INLINE) { - found_inline = 1; - extent_end = key.offset + - btrfs_file_extent_inline_len(leaf, extent); + if (path->slots[0] >= btrfs_header_nritems(leaf)) { + BUG_ON(del_nr > 0); + ret = btrfs_next_leaf(root, path); + if (ret < 0) + break; + if (ret > 0) { + ret = 0; + break; } + leaf = path->nodes[0]; + recow = 1; + } + + btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); + if (key.objectid > inode->i_ino || + key.type > BTRFS_EXTENT_DATA_KEY || key.offset >= end) + break; + + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + extent_type = btrfs_file_extent_type(leaf, fi); + + if (extent_type == BTRFS_FILE_EXTENT_REG || + extent_type == BTRFS_FILE_EXTENT_PREALLOC) { + disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); + num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi); + extent_offset = btrfs_file_extent_offset(leaf, fi); + extent_end = key.offset + + btrfs_file_extent_num_bytes(leaf, fi); + } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) { + extent_end = key.offset + + btrfs_file_extent_inline_len(leaf, fi); } else { + WARN_ON(1); extent_end = search_start; } - /* we found nothing we can drop */ - if ((!found_extent && !found_inline) || - search_start >= extent_end) { - int nextret; - u32 nritems; - nritems = btrfs_header_nritems(leaf); - if (slot >= nritems - 1) { - nextret = btrfs_next_leaf(root, path); - if (nextret) - goto out; - recow = 1; - } else { - path->slots[0]++; - } + if (extent_end <= search_start) { + path->slots[0]++; goto next_slot; } - if (end <= extent_end && start >= key.offset && found_inline) - *hint_byte = EXTENT_MAP_INLINE; - - if (found_extent) { - read_extent_buffer(leaf, &old, (unsigned long)extent, - sizeof(old)); - } - - if (end < extent_end && end >= key.offset) { - bookend = 1; - if (found_inline && start <= key.offset) - keep = 1; + search_start = max(key.offset, start); + if (recow) { + btrfs_release_path(root, path); + continue; } - if (bookend && found_extent) { - if (locked_end < extent_end) { - ret = try_lock_extent(&BTRFS_I(inode)->io_tree, - locked_end, extent_end - 1, - GFP_NOFS); - if (!ret) { - btrfs_release_path(root, path); - lock_extent(&BTRFS_I(inode)->io_tree, - locked_end, extent_end - 1, - GFP_NOFS); - locked_end = extent_end; - continue; - } - locked_end = extent_end; + /* + * | - range to drop - | + * | -------- extent -------- | + */ + if (start > key.offset && end < extent_end) { + BUG_ON(del_nr > 0); + BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE); + + memcpy(&new_key, &key, sizeof(new_key)); + new_key.offset = start; + ret = btrfs_duplicate_item(trans, root, path, + &new_key); + if (ret == -EAGAIN) { + btrfs_release_path(root, path); + continue; } - disk_bytenr = le64_to_cpu(old.disk_bytenr); - if (disk_bytenr != 0) { + if (ret < 0) + break; + + leaf = path->nodes[0]; + fi = btrfs_item_ptr(leaf, path->slots[0] - 1, + struct btrfs_file_extent_item); + btrfs_set_file_extent_num_bytes(leaf, fi, + start - key.offset); + + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + + extent_offset += start - key.offset; + btrfs_set_file_extent_offset(leaf, fi, extent_offset); + btrfs_set_file_extent_num_bytes(leaf, fi, + extent_end - start); + btrfs_mark_buffer_dirty(leaf); + + if (disk_bytenr > 0) { ret = btrfs_inc_extent_ref(trans, root, - disk_bytenr, - le64_to_cpu(old.disk_num_bytes), 0, - root->root_key.objectid, - key.objectid, key.offset - - le64_to_cpu(old.offset)); + disk_bytenr, num_bytes, 0, + root->root_key.objectid, + new_key.objectid, + start - extent_offset); BUG_ON(ret); + *hint_byte = disk_bytenr; } + key.offset = start; } + /* + * | ---- range to drop ----- | + * | -------- extent -------- | + */ + if (start <= key.offset && end < extent_end) { + BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE); + + memcpy(&new_key, &key, sizeof(new_key)); + new_key.offset = end; + btrfs_set_item_key_safe(trans, root, path, &new_key); - if (found_inline) { - u64 mask = root->sectorsize - 1; - search_start = (extent_end + mask) & ~mask; - } else - search_start = extent_end; - - /* truncate existing extent */ - if (start > key.offset) { - u64 new_num; - u64 old_num; - keep = 1; - WARN_ON(start & (root->sectorsize - 1)); - if (found_extent) { - new_num = start - key.offset; - old_num = btrfs_file_extent_num_bytes(leaf, - extent); - *hint_byte = - btrfs_file_extent_disk_bytenr(leaf, - extent); - if (btrfs_file_extent_disk_bytenr(leaf, - extent)) { - inode_sub_bytes(inode, old_num - - new_num); - } - btrfs_set_file_extent_num_bytes(leaf, - extent, new_num); - btrfs_mark_buffer_dirty(leaf); - } else if (key.offset < inline_limit && - (end > extent_end) && - (inline_limit < extent_end)) { - u32 new_size; - new_size = btrfs_file_extent_calc_inline_size( - inline_limit - key.offset); - inode_sub_bytes(inode, extent_end - - inline_limit); - btrfs_set_file_extent_ram_bytes(leaf, extent, - new_size); - if (!compression && !encryption) { - btrfs_truncate_item(trans, root, path, - new_size, 1); - } + extent_offset += end - key.offset; + btrfs_set_file_extent_offset(leaf, fi, extent_offset); + btrfs_set_file_extent_num_bytes(leaf, fi, + extent_end - end); + btrfs_mark_buffer_dirty(leaf); + if (disk_bytenr > 0) { + inode_sub_bytes(inode, end - key.offset); + *hint_byte = disk_bytenr; } + break; } - /* delete the entire extent */ - if (!keep) { - if (found_inline) - inode_sub_bytes(inode, extent_end - - key.offset); - ret = btrfs_del_item(trans, root, path); - /* TODO update progress marker and return */ - BUG_ON(ret); - extent = NULL; - btrfs_release_path(root, path); - /* the extent will be freed later */ - } - if (bookend && found_inline && start <= key.offset) { - u32 new_size; - new_size = btrfs_file_extent_calc_inline_size( - extent_end - end); - inode_sub_bytes(inode, end - key.offset); - btrfs_set_file_extent_ram_bytes(leaf, extent, - new_size); - if (!compression && !encryption) - ret = btrfs_truncate_item(trans, root, path, - new_size, 0); - BUG_ON(ret); - } - /* create bookend, splitting the extent in two */ - if (bookend && found_extent) { - struct btrfs_key ins; - ins.objectid = inode->i_ino; - ins.offset = end; - btrfs_set_key_type(&ins, BTRFS_EXTENT_DATA_KEY); - btrfs_release_path(root, path); - path->leave_spinning = 1; - ret = btrfs_insert_empty_item(trans, root, path, &ins, - sizeof(*extent)); - BUG_ON(ret); + search_start = extent_end; + /* + * | ---- range to drop ----- | + * | -------- extent -------- | + */ + if (start > key.offset && end >= extent_end) { + BUG_ON(del_nr > 0); + BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE); - leaf = path->nodes[0]; - extent = btrfs_item_ptr(leaf, path->slots[0], - struct btrfs_file_extent_item); - write_extent_buffer(leaf, &old, - (unsigned long)extent, sizeof(old)); - - btrfs_set_file_extent_compression(leaf, extent, - compression); - btrfs_set_file_extent_encryption(leaf, extent, - encryption); - btrfs_set_file_extent_other_encoding(leaf, extent, - other_encoding); - btrfs_set_file_extent_offset(leaf, extent, - le64_to_cpu(old.offset) + end - key.offset); - WARN_ON(le64_to_cpu(old.num_bytes) < - (extent_end - end)); - btrfs_set_file_extent_num_bytes(leaf, extent, - extent_end - end); + btrfs_set_file_extent_num_bytes(leaf, fi, + start - key.offset); + btrfs_mark_buffer_dirty(leaf); + if (disk_bytenr > 0) { + inode_sub_bytes(inode, extent_end - start); + *hint_byte = disk_bytenr; + } + if (end == extent_end) + break; - /* - * set the ram bytes to the size of the full extent - * before splitting. This is a worst case flag, - * but its the best we can do because we don't know - * how splitting affects compression - */ - btrfs_set_file_extent_ram_bytes(leaf, extent, - ram_bytes); - btrfs_set_file_extent_type(leaf, extent, found_type); - - btrfs_unlock_up_safe(path, 1); - btrfs_mark_buffer_dirty(path->nodes[0]); - btrfs_set_lock_blocking(path->nodes[0]); - - path->leave_spinning = 0; - btrfs_release_path(root, path); - if (disk_bytenr != 0) - inode_add_bytes(inode, extent_end - end); + path->slots[0]++; + goto next_slot; } - if (found_extent && !keep) { - u64 old_disk_bytenr = le64_to_cpu(old.disk_bytenr); + /* + * | ---- range to drop ----- | + * | ------ extent ------ | + */ + if (start <= key.offset && end >= extent_end) { + if (del_nr == 0) { + del_slot = path->slots[0]; + del_nr = 1; + } else { + BUG_ON(del_slot + del_nr != path->slots[0]); + del_nr++; + } - if (old_disk_bytenr != 0) { + if (extent_type == BTRFS_FILE_EXTENT_INLINE) { inode_sub_bytes(inode, - le64_to_cpu(old.num_bytes)); + extent_end - key.offset); + extent_end = ALIGN(extent_end, + root->sectorsize); + } else if (disk_bytenr > 0) { ret = btrfs_free_extent(trans, root, - old_disk_bytenr, - le64_to_cpu(old.disk_num_bytes), - 0, root->root_key.objectid, + disk_bytenr, num_bytes, 0, + root->root_key.objectid, key.objectid, key.offset - - le64_to_cpu(old.offset)); + extent_offset); BUG_ON(ret); - *hint_byte = old_disk_bytenr; + inode_sub_bytes(inode, + extent_end - key.offset); + *hint_byte = disk_bytenr; } - } - if (search_start >= end) { - ret = 0; - goto out; + if (end == extent_end) + break; + + if (path->slots[0] + 1 < btrfs_header_nritems(leaf)) { + path->slots[0]++; + goto next_slot; + } + + ret = btrfs_del_items(trans, root, path, del_slot, + del_nr); + BUG_ON(ret); + + del_nr = 0; + del_slot = 0; + + btrfs_release_path(root, path); + continue; } + + BUG_ON(1); } -out: - btrfs_free_path(path); - if (locked_end > orig_locked_end) { - unlock_extent(&BTRFS_I(inode)->io_tree, orig_locked_end, - locked_end - 1, GFP_NOFS); + + if (del_nr > 0) { + ret = btrfs_del_items(trans, root, path, del_slot, del_nr); + BUG_ON(ret); } + + btrfs_free_path(path); return ret; } static int extent_mergeable(struct extent_buffer *leaf, int slot, - u64 objectid, u64 bytenr, u64 *start, u64 *end) + u64 objectid, u64 bytenr, u64 orig_offset, + u64 *start, u64 *end) { struct btrfs_file_extent_item *fi; struct btrfs_key key; @@ -595,6 +524,7 @@ static int extent_mergeable(struct extent_buffer *leaf, int slot, fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item); if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG || btrfs_file_extent_disk_bytenr(leaf, fi) != bytenr || + btrfs_file_extent_offset(leaf, fi) != key.offset - orig_offset || btrfs_file_extent_compression(leaf, fi) || btrfs_file_extent_encryption(leaf, fi) || btrfs_file_extent_other_encoding(leaf, fi)) @@ -617,23 +547,24 @@ static int extent_mergeable(struct extent_buffer *leaf, int slot, * two or three. */ int btrfs_mark_extent_written(struct btrfs_trans_handle *trans, - struct btrfs_root *root, struct inode *inode, u64 start, u64 end) { + struct btrfs_root *root = BTRFS_I(inode)->root; struct extent_buffer *leaf; struct btrfs_path *path; struct btrfs_file_extent_item *fi; struct btrfs_key key; + struct btrfs_key new_key; u64 bytenr; u64 num_bytes; u64 extent_end; u64 orig_offset; u64 other_start; u64 other_end; - u64 split = start; - u64 locked_end = end; - int extent_type; - int split_end = 1; + u64 split; + int del_nr = 0; + int del_slot = 0; + int recow; int ret; btrfs_drop_extent_cache(inode, start, end - 1, 0); @@ -641,12 +572,11 @@ int btrfs_mark_extent_written(struct btrfs_trans_handle *trans, path = btrfs_alloc_path(); BUG_ON(!path); again: + recow = 0; + split = start; key.objectid = inode->i_ino; key.type = BTRFS_EXTENT_DATA_KEY; - if (split == start) - key.offset = split; - else - key.offset = split - 1; + key.offset = split; ret = btrfs_search_slot(trans, root, &key, path, -1, 1); if (ret > 0 && path->slots[0] > 0) @@ -658,159 +588,158 @@ again: key.type != BTRFS_EXTENT_DATA_KEY); fi = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item); - extent_type = btrfs_file_extent_type(leaf, fi); - BUG_ON(extent_type != BTRFS_FILE_EXTENT_PREALLOC); + BUG_ON(btrfs_file_extent_type(leaf, fi) != + BTRFS_FILE_EXTENT_PREALLOC); extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi); BUG_ON(key.offset > start || extent_end < end); bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi); orig_offset = key.offset - btrfs_file_extent_offset(leaf, fi); + memcpy(&new_key, &key, sizeof(new_key)); - if (key.offset == start) - split = end; - - if (key.offset == start && extent_end == end) { - int del_nr = 0; - int del_slot = 0; - other_start = end; - other_end = 0; - if (extent_mergeable(leaf, path->slots[0] + 1, inode->i_ino, - bytenr, &other_start, &other_end)) { - extent_end = other_end; - del_slot = path->slots[0] + 1; - del_nr++; - ret = btrfs_free_extent(trans, root, bytenr, num_bytes, - 0, root->root_key.objectid, - inode->i_ino, orig_offset); - BUG_ON(ret); - } + if (start == key.offset && end < extent_end) { other_start = 0; other_end = start; - if (extent_mergeable(leaf, path->slots[0] - 1, inode->i_ino, - bytenr, &other_start, &other_end)) { - key.offset = other_start; - del_slot = path->slots[0]; - del_nr++; - ret = btrfs_free_extent(trans, root, bytenr, num_bytes, - 0, root->root_key.objectid, - inode->i_ino, orig_offset); - BUG_ON(ret); - } - split_end = 0; - if (del_nr == 0) { - btrfs_set_file_extent_type(leaf, fi, - BTRFS_FILE_EXTENT_REG); - goto done; - } - - fi = btrfs_item_ptr(leaf, del_slot - 1, - struct btrfs_file_extent_item); - btrfs_set_file_extent_type(leaf, fi, BTRFS_FILE_EXTENT_REG); - btrfs_set_file_extent_num_bytes(leaf, fi, - extent_end - key.offset); - btrfs_mark_buffer_dirty(leaf); - - ret = btrfs_del_items(trans, root, path, del_slot, del_nr); - BUG_ON(ret); - goto release; - } else if (split == start) { - if (locked_end < extent_end) { - ret = try_lock_extent(&BTRFS_I(inode)->io_tree, - locked_end, extent_end - 1, GFP_NOFS); - if (!ret) { - btrfs_release_path(root, path); - lock_extent(&BTRFS_I(inode)->io_tree, - locked_end, extent_end - 1, GFP_NOFS); - locked_end = extent_end; - goto again; - } - locked_end = extent_end; + if (extent_mergeable(leaf, path->slots[0] - 1, + inode->i_ino, bytenr, orig_offset, + &other_start, &other_end)) { + new_key.offset = end; + btrfs_set_item_key_safe(trans, root, path, &new_key); + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + btrfs_set_file_extent_num_bytes(leaf, fi, + extent_end - end); + btrfs_set_file_extent_offset(leaf, fi, + end - orig_offset); + fi = btrfs_item_ptr(leaf, path->slots[0] - 1, + struct btrfs_file_extent_item); + btrfs_set_file_extent_num_bytes(leaf, fi, + end - other_start); + btrfs_mark_buffer_dirty(leaf); + goto out; } - btrfs_set_file_extent_num_bytes(leaf, fi, split - key.offset); - } else { - BUG_ON(key.offset != start); - key.offset = split; - btrfs_set_file_extent_offset(leaf, fi, key.offset - - orig_offset); - btrfs_set_file_extent_num_bytes(leaf, fi, extent_end - split); - btrfs_set_item_key_safe(trans, root, path, &key); - extent_end = split; } - if (extent_end == end) { - split_end = 0; - extent_type = BTRFS_FILE_EXTENT_REG; - } - if (extent_end == end && split == start) { + if (start > key.offset && end == extent_end) { other_start = end; other_end = 0; - if (extent_mergeable(leaf, path->slots[0] + 1, inode->i_ino, - bytenr, &other_start, &other_end)) { - path->slots[0]++; + if (extent_mergeable(leaf, path->slots[0] + 1, + inode->i_ino, bytenr, orig_offset, + &other_start, &other_end)) { fi = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item); - key.offset = split; - btrfs_set_item_key_safe(trans, root, path, &key); - btrfs_set_file_extent_offset(leaf, fi, key.offset - - orig_offset); btrfs_set_file_extent_num_bytes(leaf, fi, - other_end - split); - goto done; - } - } - if (extent_end == end && split == end) { - other_start = 0; - other_end = start; - if (extent_mergeable(leaf, path->slots[0] - 1 , inode->i_ino, - bytenr, &other_start, &other_end)) { - path->slots[0]--; + start - key.offset); + path->slots[0]++; + new_key.offset = start; + btrfs_set_item_key_safe(trans, root, path, &new_key); + fi = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item); - btrfs_set_file_extent_num_bytes(leaf, fi, extent_end - - other_start); - goto done; + btrfs_set_file_extent_num_bytes(leaf, fi, + other_end - start); + btrfs_set_file_extent_offset(leaf, fi, + start - orig_offset); + btrfs_mark_buffer_dirty(leaf); + goto out; } } - btrfs_mark_buffer_dirty(leaf); + while (start > key.offset || end < extent_end) { + if (key.offset == start) + split = end; - ret = btrfs_inc_extent_ref(trans, root, bytenr, num_bytes, 0, - root->root_key.objectid, - inode->i_ino, orig_offset); - BUG_ON(ret); - btrfs_release_path(root, path); + new_key.offset = split; + ret = btrfs_duplicate_item(trans, root, path, &new_key); + if (ret == -EAGAIN) { + btrfs_release_path(root, path); + goto again; + } + BUG_ON(ret < 0); - key.offset = start; - ret = btrfs_insert_empty_item(trans, root, path, &key, sizeof(*fi)); - BUG_ON(ret); + leaf = path->nodes[0]; + fi = btrfs_item_ptr(leaf, path->slots[0] - 1, + struct btrfs_file_extent_item); + btrfs_set_file_extent_num_bytes(leaf, fi, + split - key.offset); - leaf = path->nodes[0]; - fi = btrfs_item_ptr(leaf, path->slots[0], - struct btrfs_file_extent_item); - btrfs_set_file_extent_generation(leaf, fi, trans->transid); - btrfs_set_file_extent_type(leaf, fi, extent_type); - btrfs_set_file_extent_disk_bytenr(leaf, fi, bytenr); - btrfs_set_file_extent_disk_num_bytes(leaf, fi, num_bytes); - btrfs_set_file_extent_offset(leaf, fi, key.offset - orig_offset); - btrfs_set_file_extent_num_bytes(leaf, fi, extent_end - key.offset); - btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes); - btrfs_set_file_extent_compression(leaf, fi, 0); - btrfs_set_file_extent_encryption(leaf, fi, 0); - btrfs_set_file_extent_other_encoding(leaf, fi, 0); -done: - btrfs_mark_buffer_dirty(leaf); - -release: - btrfs_release_path(root, path); - if (split_end && split == start) { - split = end; - goto again; + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + + btrfs_set_file_extent_offset(leaf, fi, split - orig_offset); + btrfs_set_file_extent_num_bytes(leaf, fi, + extent_end - split); + btrfs_mark_buffer_dirty(leaf); + + ret = btrfs_inc_extent_ref(trans, root, bytenr, num_bytes, 0, + root->root_key.objectid, + inode->i_ino, orig_offset); + BUG_ON(ret); + + if (split == start) { + key.offset = start; + } else { + BUG_ON(start != key.offset); + path->slots[0]--; + extent_end = end; + } + recow = 1; + } + + other_start = end; + other_end = 0; + if (extent_mergeable(leaf, path->slots[0] + 1, + inode->i_ino, bytenr, orig_offset, + &other_start, &other_end)) { + if (recow) { + btrfs_release_path(root, path); + goto again; + } + extent_end = other_end; + del_slot = path->slots[0] + 1; + del_nr++; + ret = btrfs_free_extent(trans, root, bytenr, num_bytes, + 0, root->root_key.objectid, + inode->i_ino, orig_offset); + BUG_ON(ret); } - if (locked_end > end) { - unlock_extent(&BTRFS_I(inode)->io_tree, end, locked_end - 1, - GFP_NOFS); + other_start = 0; + other_end = start; + if (extent_mergeable(leaf, path->slots[0] - 1, + inode->i_ino, bytenr, orig_offset, + &other_start, &other_end)) { + if (recow) { + btrfs_release_path(root, path); + goto again; + } + key.offset = other_start; + del_slot = path->slots[0]; + del_nr++; + ret = btrfs_free_extent(trans, root, bytenr, num_bytes, + 0, root->root_key.objectid, + inode->i_ino, orig_offset); + BUG_ON(ret); } + if (del_nr == 0) { + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + btrfs_set_file_extent_type(leaf, fi, + BTRFS_FILE_EXTENT_REG); + btrfs_mark_buffer_dirty(leaf); + } else { + fi = btrfs_item_ptr(leaf, del_slot - 1, + struct btrfs_file_extent_item); + btrfs_set_file_extent_type(leaf, fi, + BTRFS_FILE_EXTENT_REG); + btrfs_set_file_extent_num_bytes(leaf, fi, + extent_end - key.offset); + btrfs_mark_buffer_dirty(leaf); + + ret = btrfs_del_items(trans, root, path, del_slot, del_nr); + BUG_ON(ret); + } +out: btrfs_free_path(path); return 0; } @@ -825,6 +754,7 @@ static noinline int prepare_pages(struct btrfs_root *root, struct file *file, loff_t pos, unsigned long first_index, unsigned long last_index, size_t write_bytes) { + struct extent_state *cached_state = NULL; int i; unsigned long index = pos >> PAGE_CACHE_SHIFT; struct inode *inode = fdentry(file)->d_inode; @@ -853,16 +783,18 @@ again: } if (start_pos < inode->i_size) { struct btrfs_ordered_extent *ordered; - lock_extent(&BTRFS_I(inode)->io_tree, - start_pos, last_pos - 1, GFP_NOFS); + lock_extent_bits(&BTRFS_I(inode)->io_tree, + start_pos, last_pos - 1, 0, &cached_state, + GFP_NOFS); ordered = btrfs_lookup_first_ordered_extent(inode, last_pos - 1); if (ordered && ordered->file_offset + ordered->len > start_pos && ordered->file_offset < last_pos) { btrfs_put_ordered_extent(ordered); - unlock_extent(&BTRFS_I(inode)->io_tree, - start_pos, last_pos - 1, GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, + start_pos, last_pos - 1, + &cached_state, GFP_NOFS); for (i = 0; i < num_pages; i++) { unlock_page(pages[i]); page_cache_release(pages[i]); @@ -874,11 +806,13 @@ again: if (ordered) btrfs_put_ordered_extent(ordered); - clear_extent_bits(&BTRFS_I(inode)->io_tree, start_pos, - last_pos - 1, EXTENT_DIRTY | EXTENT_DELALLOC, + clear_extent_bit(&BTRFS_I(inode)->io_tree, start_pos, + last_pos - 1, EXTENT_DIRTY | EXTENT_DELALLOC | + EXTENT_DO_ACCOUNTING, 0, 0, &cached_state, GFP_NOFS); - unlock_extent(&BTRFS_I(inode)->io_tree, - start_pos, last_pos - 1, GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, + start_pos, last_pos - 1, &cached_state, + GFP_NOFS); } for (i = 0; i < num_pages; i++) { clear_page_dirty_for_io(pages[i]); @@ -905,7 +839,7 @@ static ssize_t btrfs_file_write(struct file *file, const char __user *buf, unsigned long last_index; int will_write; - will_write = ((file->f_flags & O_SYNC) || IS_SYNC(inode) || + will_write = ((file->f_flags & O_DSYNC) || IS_SYNC(inode) || (file->f_flags & O_DIRECT)); nrptrs = min((count + PAGE_CACHE_SIZE - 1) / PAGE_CACHE_SIZE, @@ -917,21 +851,35 @@ static ssize_t btrfs_file_write(struct file *file, const char __user *buf, start_pos = pos; vfs_check_frozen(inode->i_sb, SB_FREEZE_WRITE); + + /* do the reserve before the mutex lock in case we have to do some + * flushing. We wouldn't deadlock, but this is more polite. + */ + err = btrfs_reserve_metadata_for_delalloc(root, inode, 1); + if (err) + goto out_nolock; + + mutex_lock(&inode->i_mutex); + current->backing_dev_info = inode->i_mapping->backing_dev_info; err = generic_write_checks(file, &pos, &count, S_ISBLK(inode->i_mode)); if (err) - goto out_nolock; + goto out; + if (count == 0) - goto out_nolock; + goto out; err = file_remove_suid(file); if (err) - goto out_nolock; + goto out; + file_update_time(file); pages = kmalloc(nrptrs * sizeof(struct page *), GFP_KERNEL); - mutex_lock(&inode->i_mutex); + /* generic_write_checks can change our pos */ + start_pos = pos; + BTRFS_I(inode)->sequence++; first_index = pos >> PAGE_CACHE_SHIFT; last_index = (pos + count) >> PAGE_CACHE_SHIFT; @@ -1005,9 +953,8 @@ static ssize_t btrfs_file_write(struct file *file, const char __user *buf, } if (will_write) { - btrfs_fdatawrite_range(inode->i_mapping, pos, - pos + write_bytes - 1, - WB_SYNC_ALL); + filemap_fdatawrite_range(inode->i_mapping, pos, + pos + write_bytes - 1); } else { balance_dirty_pages_ratelimited_nr(inode->i_mapping, num_pages); @@ -1028,6 +975,7 @@ out: mutex_unlock(&inode->i_mutex); if (ret) err = ret; + btrfs_unreserve_metadata_for_delalloc(root, inode, 1); out_nolock: kfree(pages); @@ -1058,7 +1006,7 @@ out_nolock: if (err) num_written = err; - if ((file->f_flags & O_SYNC) || IS_SYNC(inode)) { + if ((file->f_flags & O_DSYNC) || IS_SYNC(inode)) { trans = btrfs_start_transaction(root, 1); ret = btrfs_log_dentry_safe(trans, root, file->f_dentry); @@ -1068,8 +1016,10 @@ out_nolock: btrfs_end_transaction(trans, root); else btrfs_commit_transaction(trans, root); - } else { + } else if (ret != BTRFS_NO_LOG_SYNC) { btrfs_commit_transaction(trans, root); + } else { + btrfs_end_transaction(trans, root); } } if (file->f_flags & O_DIRECT) { @@ -1119,6 +1069,13 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync) int ret = 0; struct btrfs_trans_handle *trans; + + /* we wait first, since the writeback may change the inode */ + root->log_batch++; + /* the VFS called filemap_fdatawrite for us */ + btrfs_wait_ordered_range(inode, 0, (u64)-1); + root->log_batch++; + /* * check the transaction that last modified this inode * and see if its already been committed @@ -1126,6 +1083,11 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync) if (!BTRFS_I(inode)->last_trans) goto out; + /* + * if the last transaction that changed this file was before + * the current transaction, we can bail out now without any + * syncing + */ mutex_lock(&root->fs_info->trans_mutex); if (BTRFS_I(inode)->last_trans <= root->fs_info->last_trans_committed) { @@ -1135,13 +1097,6 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync) } mutex_unlock(&root->fs_info->trans_mutex); - root->log_batch++; - filemap_fdatawrite(inode->i_mapping); - btrfs_wait_ordered_range(inode, 0, (u64)-1); - root->log_batch++; - - if (datasync && !(inode->i_state & I_DIRTY_PAGES)) - goto out; /* * ok we haven't committed the transaction yet, lets do a commit */ @@ -1170,18 +1125,22 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync) */ mutex_unlock(&dentry->d_inode->i_mutex); - if (ret > 0) { - ret = btrfs_commit_transaction(trans, root); - } else { - ret = btrfs_sync_log(trans, root); - if (ret == 0) - ret = btrfs_end_transaction(trans, root); - else + if (ret != BTRFS_NO_LOG_SYNC) { + if (ret > 0) { ret = btrfs_commit_transaction(trans, root); + } else { + ret = btrfs_sync_log(trans, root); + if (ret == 0) + ret = btrfs_end_transaction(trans, root); + else + ret = btrfs_commit_transaction(trans, root); + } + } else { + ret = btrfs_end_transaction(trans, root); } mutex_lock(&dentry->d_inode->i_mutex); out: - return ret > 0 ? EIO : ret; + return ret > 0 ? -EIO : ret; } static const struct vm_operations_struct btrfs_file_vm_ops = { @@ -1196,7 +1155,7 @@ static int btrfs_file_mmap(struct file *filp, struct vm_area_struct *vma) return 0; } -struct file_operations btrfs_file_operations = { +const struct file_operations btrfs_file_operations = { .llseek = generic_file_llseek, .read = do_sync_read, .aio_read = generic_file_aio_read, diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c index 5c2caad76212..dd831ed31eea 100644 --- a/fs/btrfs/free-space-cache.c +++ b/fs/btrfs/free-space-cache.c @@ -870,7 +870,7 @@ __btrfs_return_cluster_to_free_space( tree_insert_offset(&block_group->free_space_offset, entry->offset, &entry->offset_index, 0); } - cluster->root.rb_node = NULL; + cluster->root = RB_ROOT; out: spin_unlock(&cluster->lock); @@ -1296,7 +1296,7 @@ again: window_start = entry->offset; window_free = entry->bytes; last = entry; - max_extent = 0; + max_extent = entry->bytes; } else { last = next; window_free += next->bytes; @@ -1355,7 +1355,7 @@ void btrfs_init_free_cluster(struct btrfs_free_cluster *cluster) { spin_lock_init(&cluster->lock); spin_lock_init(&cluster->refill_lock); - cluster->root.rb_node = NULL; + cluster->root = RB_ROOT; cluster->max_size = 0; cluster->points_to_bitmap = false; INIT_LIST_HEAD(&cluster->block_group_list); diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index e9b76bcd1c12..02bb099845fd 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -62,7 +62,7 @@ static const struct inode_operations btrfs_special_inode_operations; static const struct inode_operations btrfs_file_inode_operations; static const struct address_space_operations btrfs_aops; static const struct address_space_operations btrfs_symlink_aops; -static struct file_operations btrfs_dir_file_operations; +static const struct file_operations btrfs_dir_file_operations; static struct extent_io_ops btrfs_extent_io_ops; static struct kmem_cache *btrfs_inode_cachep; @@ -88,13 +88,14 @@ static noinline int cow_file_range(struct inode *inode, u64 start, u64 end, int *page_started, unsigned long *nr_written, int unlock); -static int btrfs_init_inode_security(struct inode *inode, struct inode *dir) +static int btrfs_init_inode_security(struct btrfs_trans_handle *trans, + struct inode *inode, struct inode *dir) { int err; - err = btrfs_init_acl(inode, dir); + err = btrfs_init_acl(trans, inode, dir); if (!err) - err = btrfs_xattr_security_init(inode, dir); + err = btrfs_xattr_security_init(trans, inode, dir); return err; } @@ -188,8 +189,18 @@ static noinline int insert_inline_extent(struct btrfs_trans_handle *trans, btrfs_mark_buffer_dirty(leaf); btrfs_free_path(path); + /* + * we're an inline extent, so nobody can + * extend the file past i_size without locking + * a page we already have locked. + * + * We must do any isize and inode updates + * before we unlock the pages. Otherwise we + * could end up racing with unlink. + */ BTRFS_I(inode)->disk_i_size = inode->i_size; btrfs_update_inode(trans, root, inode); + return 0; fail: btrfs_free_path(path); @@ -230,8 +241,7 @@ static noinline int cow_file_range_inline(struct btrfs_trans_handle *trans, return 1; } - ret = btrfs_drop_extents(trans, root, inode, start, - aligned_end, aligned_end, start, + ret = btrfs_drop_extents(trans, inode, start, aligned_end, &hint_byte, 1); BUG_ON(ret); @@ -369,7 +379,8 @@ again: * change at any time if we discover bad compression ratios. */ if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NOCOMPRESS) && - btrfs_test_opt(root, COMPRESS)) { + (btrfs_test_opt(root, COMPRESS) || + (BTRFS_I(inode)->force_compress))) { WARN_ON(pages); pages = kzalloc(sizeof(struct page *) * nr_pages, GFP_NOFS); @@ -416,7 +427,6 @@ again: start, end, total_compressed, pages); } - btrfs_end_transaction(trans, root); if (ret == 0) { /* * inline extent creation worked, we don't need @@ -424,12 +434,17 @@ again: * and free up our temp pages. */ extent_clear_unlock_delalloc(inode, - &BTRFS_I(inode)->io_tree, - start, end, NULL, 1, 0, - 0, 1, 1, 1, 0); - ret = 0; + &BTRFS_I(inode)->io_tree, + start, end, NULL, + EXTENT_CLEAR_UNLOCK_PAGE | EXTENT_CLEAR_DIRTY | + EXTENT_CLEAR_DELALLOC | + EXTENT_CLEAR_ACCOUNTING | + EXTENT_SET_WRITEBACK | EXTENT_END_WRITEBACK); + + btrfs_end_transaction(trans, root); goto free_pages_out; } + btrfs_end_transaction(trans, root); } if (will_compress) { @@ -469,7 +484,10 @@ again: nr_pages_ret = 0; /* flag the file so we don't compress in the future */ - BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS; + if (!btrfs_test_opt(root, FORCE_COMPRESS) && + !(BTRFS_I(inode)->force_compress)) { + BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS; + } } if (will_compress) { *num_added += 1; @@ -535,12 +553,11 @@ static noinline int submit_compressed_extents(struct inode *inode, struct btrfs_root *root = BTRFS_I(inode)->root; struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; struct extent_io_tree *io_tree; - int ret; + int ret = 0; if (list_empty(&async_cow->extents)) return 0; - trans = btrfs_join_transaction(root, 1); while (!list_empty(&async_cow->extents)) { async_extent = list_entry(async_cow->extents.next, @@ -549,21 +566,22 @@ static noinline int submit_compressed_extents(struct inode *inode, io_tree = &BTRFS_I(inode)->io_tree; +retry: /* did the compression code fall back to uncompressed IO? */ if (!async_extent->pages) { int page_started = 0; unsigned long nr_written = 0; lock_extent(io_tree, async_extent->start, - async_extent->start + - async_extent->ram_size - 1, GFP_NOFS); + async_extent->start + + async_extent->ram_size - 1, GFP_NOFS); /* allocate blocks */ - cow_file_range(inode, async_cow->locked_page, - async_extent->start, - async_extent->start + - async_extent->ram_size - 1, - &page_started, &nr_written, 0); + ret = cow_file_range(inode, async_cow->locked_page, + async_extent->start, + async_extent->start + + async_extent->ram_size - 1, + &page_started, &nr_written, 0); /* * if page_started, cow_file_range inserted an @@ -571,7 +589,7 @@ static noinline int submit_compressed_extents(struct inode *inode, * and IO for us. Otherwise, we need to submit * all those pages down to the drive. */ - if (!page_started) + if (!page_started && !ret) extent_write_locked_range(io_tree, inode, async_extent->start, async_extent->start + @@ -586,6 +604,30 @@ static noinline int submit_compressed_extents(struct inode *inode, lock_extent(io_tree, async_extent->start, async_extent->start + async_extent->ram_size - 1, GFP_NOFS); + + trans = btrfs_join_transaction(root, 1); + ret = btrfs_reserve_extent(trans, root, + async_extent->compressed_size, + async_extent->compressed_size, + 0, alloc_hint, + (u64)-1, &ins, 1); + btrfs_end_transaction(trans, root); + + if (ret) { + int i; + for (i = 0; i < async_extent->nr_pages; i++) { + WARN_ON(async_extent->pages[i]->mapping); + page_cache_release(async_extent->pages[i]); + } + kfree(async_extent->pages); + async_extent->nr_pages = 0; + async_extent->pages = NULL; + unlock_extent(io_tree, async_extent->start, + async_extent->start + + async_extent->ram_size - 1, GFP_NOFS); + goto retry; + } + /* * here we're doing allocation and writeback of the * compressed pages @@ -594,12 +636,6 @@ static noinline int submit_compressed_extents(struct inode *inode, async_extent->start + async_extent->ram_size - 1, 0); - ret = btrfs_reserve_extent(trans, root, - async_extent->compressed_size, - async_extent->compressed_size, - 0, alloc_hint, - (u64)-1, &ins, 1); - BUG_ON(ret); em = alloc_extent_map(GFP_NOFS); em->start = async_extent->start; em->len = async_extent->ram_size; @@ -631,17 +667,18 @@ static noinline int submit_compressed_extents(struct inode *inode, BTRFS_ORDERED_COMPRESSED); BUG_ON(ret); - btrfs_end_transaction(trans, root); - /* * clear dirty, set writeback and unlock the pages. */ extent_clear_unlock_delalloc(inode, - &BTRFS_I(inode)->io_tree, - async_extent->start, - async_extent->start + - async_extent->ram_size - 1, - NULL, 1, 1, 0, 1, 1, 0, 0); + &BTRFS_I(inode)->io_tree, + async_extent->start, + async_extent->start + + async_extent->ram_size - 1, + NULL, EXTENT_CLEAR_UNLOCK_PAGE | + EXTENT_CLEAR_UNLOCK | + EXTENT_CLEAR_DELALLOC | + EXTENT_CLEAR_DIRTY | EXTENT_SET_WRITEBACK); ret = btrfs_submit_compressed_write(inode, async_extent->start, @@ -651,13 +688,11 @@ static noinline int submit_compressed_extents(struct inode *inode, async_extent->nr_pages); BUG_ON(ret); - trans = btrfs_join_transaction(root, 1); alloc_hint = ins.objectid + ins.offset; kfree(async_extent); cond_resched(); } - btrfs_end_transaction(trans, root); return 0; } @@ -712,9 +747,16 @@ static noinline int cow_file_range(struct inode *inode, start, end, 0, NULL); if (ret == 0) { extent_clear_unlock_delalloc(inode, - &BTRFS_I(inode)->io_tree, - start, end, NULL, 1, 1, - 1, 1, 1, 1, 0); + &BTRFS_I(inode)->io_tree, + start, end, NULL, + EXTENT_CLEAR_UNLOCK_PAGE | + EXTENT_CLEAR_UNLOCK | + EXTENT_CLEAR_DELALLOC | + EXTENT_CLEAR_ACCOUNTING | + EXTENT_CLEAR_DIRTY | + EXTENT_SET_WRITEBACK | + EXTENT_END_WRITEBACK); + *nr_written = *nr_written + (end - start + PAGE_CACHE_SIZE) / PAGE_CACHE_SIZE; *page_started = 1; @@ -731,13 +773,29 @@ static noinline int cow_file_range(struct inode *inode, em = search_extent_mapping(&BTRFS_I(inode)->extent_tree, start, num_bytes); if (em) { - alloc_hint = em->block_start; - free_extent_map(em); + /* + * if block start isn't an actual block number then find the + * first block in this inode and use that as a hint. If that + * block is also bogus then just don't worry about it. + */ + if (em->block_start >= EXTENT_MAP_LAST_BYTE) { + free_extent_map(em); + em = search_extent_mapping(em_tree, 0, 0); + if (em && em->block_start < EXTENT_MAP_LAST_BYTE) + alloc_hint = em->block_start; + if (em) + free_extent_map(em); + } else { + alloc_hint = em->block_start; + free_extent_map(em); + } } read_unlock(&BTRFS_I(inode)->extent_tree.lock); btrfs_drop_extent_cache(inode, start, start + num_bytes - 1, 0); while (disk_num_bytes > 0) { + unsigned long op; + cur_alloc_size = min(disk_num_bytes, root->fs_info->max_extent); ret = btrfs_reserve_extent(trans, root, cur_alloc_size, root->sectorsize, 0, alloc_hint, @@ -789,10 +847,13 @@ static noinline int cow_file_range(struct inode *inode, * Do set the Private2 bit so we know this page was properly * setup for writepage */ + op = unlock ? EXTENT_CLEAR_UNLOCK_PAGE : 0; + op |= EXTENT_CLEAR_UNLOCK | EXTENT_CLEAR_DELALLOC | + EXTENT_SET_PRIVATE2; + extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree, start, start + ram_size - 1, - locked_page, unlock, 1, - 1, 0, 0, 0, 1); + locked_page, op); disk_num_bytes -= cur_alloc_size; num_bytes -= cur_alloc_size; alloc_hint = ins.objectid + ins.offset; @@ -864,8 +925,8 @@ static int cow_file_range_async(struct inode *inode, struct page *locked_page, u64 cur_end; int limit = 10 * 1024 * 1042; - clear_extent_bit(&BTRFS_I(inode)->io_tree, start, end, EXTENT_LOCKED | - EXTENT_DELALLOC, 1, 0, NULL, GFP_NOFS); + clear_extent_bit(&BTRFS_I(inode)->io_tree, start, end, EXTENT_LOCKED, + 1, 0, NULL, GFP_NOFS); while (start < end) { async_cow = kmalloc(sizeof(*async_cow), GFP_NOFS); async_cow->inode = inode; @@ -1006,6 +1067,7 @@ next_slot: if (found_key.offset > cur_offset) { extent_end = found_key.offset; + extent_type = 0; goto out_check; } @@ -1112,8 +1174,10 @@ out_check: BUG_ON(ret); extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree, - cur_offset, cur_offset + num_bytes - 1, - locked_page, 1, 1, 1, 0, 0, 0, 1); + cur_offset, cur_offset + num_bytes - 1, + locked_page, EXTENT_CLEAR_UNLOCK_PAGE | + EXTENT_CLEAR_UNLOCK | EXTENT_CLEAR_DELALLOC | + EXTENT_SET_PRIVATE2); cur_offset = extent_end; if (cur_offset > end) break; @@ -1150,7 +1214,8 @@ static int run_delalloc_range(struct inode *inode, struct page *locked_page, else if (BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC) ret = run_delalloc_nocow(inode, locked_page, start, end, page_started, 0, nr_written); - else if (!btrfs_test_opt(root, COMPRESS)) + else if (!btrfs_test_opt(root, COMPRESS) && + !(BTRFS_I(inode)->force_compress)) ret = cow_file_range(inode, locked_page, start, end, page_started, nr_written, 1); else @@ -1159,6 +1224,89 @@ static int run_delalloc_range(struct inode *inode, struct page *locked_page, return ret; } +static int btrfs_split_extent_hook(struct inode *inode, + struct extent_state *orig, u64 split) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + u64 size; + + if (!(orig->state & EXTENT_DELALLOC)) + return 0; + + size = orig->end - orig->start + 1; + if (size > root->fs_info->max_extent) { + u64 num_extents; + u64 new_size; + + new_size = orig->end - split + 1; + num_extents = div64_u64(size + root->fs_info->max_extent - 1, + root->fs_info->max_extent); + + /* + * if we break a large extent up then leave oustanding_extents + * be, since we've already accounted for the large extent. + */ + if (div64_u64(new_size + root->fs_info->max_extent - 1, + root->fs_info->max_extent) < num_extents) + return 0; + } + + spin_lock(&BTRFS_I(inode)->accounting_lock); + BTRFS_I(inode)->outstanding_extents++; + spin_unlock(&BTRFS_I(inode)->accounting_lock); + + return 0; +} + +/* + * extent_io.c merge_extent_hook, used to track merged delayed allocation + * extents so we can keep track of new extents that are just merged onto old + * extents, such as when we are doing sequential writes, so we can properly + * account for the metadata space we'll need. + */ +static int btrfs_merge_extent_hook(struct inode *inode, + struct extent_state *new, + struct extent_state *other) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + u64 new_size, old_size; + u64 num_extents; + + /* not delalloc, ignore it */ + if (!(other->state & EXTENT_DELALLOC)) + return 0; + + old_size = other->end - other->start + 1; + if (new->start < other->start) + new_size = other->end - new->start + 1; + else + new_size = new->end - other->start + 1; + + /* we're not bigger than the max, unreserve the space and go */ + if (new_size <= root->fs_info->max_extent) { + spin_lock(&BTRFS_I(inode)->accounting_lock); + BTRFS_I(inode)->outstanding_extents--; + spin_unlock(&BTRFS_I(inode)->accounting_lock); + return 0; + } + + /* + * If we grew by another max_extent, just return, we want to keep that + * reserved amount. + */ + num_extents = div64_u64(old_size + root->fs_info->max_extent - 1, + root->fs_info->max_extent); + if (div64_u64(new_size + root->fs_info->max_extent - 1, + root->fs_info->max_extent) > num_extents) + return 0; + + spin_lock(&BTRFS_I(inode)->accounting_lock); + BTRFS_I(inode)->outstanding_extents--; + spin_unlock(&BTRFS_I(inode)->accounting_lock); + + return 0; +} + /* * extent_io.c set_bit_hook, used to track delayed allocation * bytes in this file, and to maintain the list of inodes that @@ -1167,6 +1315,7 @@ static int run_delalloc_range(struct inode *inode, struct page *locked_page, static int btrfs_set_bit_hook(struct inode *inode, u64 start, u64 end, unsigned long old, unsigned long bits) { + /* * set_bit and clear bit hooks normally require _irqsave/restore * but in this case, we are only testeing for the DELALLOC @@ -1174,6 +1323,10 @@ static int btrfs_set_bit_hook(struct inode *inode, u64 start, u64 end, */ if (!(old & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) { struct btrfs_root *root = BTRFS_I(inode)->root; + + spin_lock(&BTRFS_I(inode)->accounting_lock); + BTRFS_I(inode)->outstanding_extents++; + spin_unlock(&BTRFS_I(inode)->accounting_lock); btrfs_delalloc_reserve_space(root, inode, end - start + 1); spin_lock(&root->fs_info->delalloc_lock); BTRFS_I(inode)->delalloc_bytes += end - start + 1; @@ -1190,22 +1343,31 @@ static int btrfs_set_bit_hook(struct inode *inode, u64 start, u64 end, /* * extent_io.c clear_bit_hook, see set_bit_hook for why */ -static int btrfs_clear_bit_hook(struct inode *inode, u64 start, u64 end, - unsigned long old, unsigned long bits) +static int btrfs_clear_bit_hook(struct inode *inode, + struct extent_state *state, unsigned long bits) { /* * set_bit and clear bit hooks normally require _irqsave/restore * but in this case, we are only testeing for the DELALLOC * bit, which is only set or cleared with irqs on */ - if ((old & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) { + if ((state->state & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) { struct btrfs_root *root = BTRFS_I(inode)->root; + if (bits & EXTENT_DO_ACCOUNTING) { + spin_lock(&BTRFS_I(inode)->accounting_lock); + BTRFS_I(inode)->outstanding_extents--; + spin_unlock(&BTRFS_I(inode)->accounting_lock); + btrfs_unreserve_metadata_for_delalloc(root, inode, 1); + } + spin_lock(&root->fs_info->delalloc_lock); - if (end - start + 1 > root->fs_info->delalloc_bytes) { + if (state->end - state->start + 1 > + root->fs_info->delalloc_bytes) { printk(KERN_INFO "btrfs warning: delalloc account " "%llu %llu\n", - (unsigned long long)end - start + 1, + (unsigned long long) + state->end - state->start + 1, (unsigned long long) root->fs_info->delalloc_bytes); btrfs_delalloc_free_space(root, inode, (u64)-1); @@ -1213,9 +1375,12 @@ static int btrfs_clear_bit_hook(struct inode *inode, u64 start, u64 end, BTRFS_I(inode)->delalloc_bytes = 0; } else { btrfs_delalloc_free_space(root, inode, - end - start + 1); - root->fs_info->delalloc_bytes -= end - start + 1; - BTRFS_I(inode)->delalloc_bytes -= end - start + 1; + state->end - + state->start + 1); + root->fs_info->delalloc_bytes -= state->end - + state->start + 1; + BTRFS_I(inode)->delalloc_bytes -= state->end - + state->start + 1; } if (BTRFS_I(inode)->delalloc_bytes == 0 && !list_empty(&BTRFS_I(inode)->delalloc_inodes)) { @@ -1347,12 +1512,13 @@ static noinline int add_pending_csums(struct btrfs_trans_handle *trans, return 0; } -int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end) +int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end, + struct extent_state **cached_state) { if ((end & (PAGE_CACHE_SIZE - 1)) == 0) WARN_ON(1); return set_extent_delalloc(&BTRFS_I(inode)->io_tree, start, end, - GFP_NOFS); + cached_state, GFP_NOFS); } /* see btrfs_writepage_start_hook for details on why this is required */ @@ -1365,6 +1531,7 @@ static void btrfs_writepage_fixup_worker(struct btrfs_work *work) { struct btrfs_writepage_fixup *fixup; struct btrfs_ordered_extent *ordered; + struct extent_state *cached_state = NULL; struct page *page; struct inode *inode; u64 page_start; @@ -1383,7 +1550,8 @@ again: page_start = page_offset(page); page_end = page_offset(page) + PAGE_CACHE_SIZE - 1; - lock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end, GFP_NOFS); + lock_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end, 0, + &cached_state, GFP_NOFS); /* already ordered? We're done */ if (PagePrivate2(page)) @@ -1391,17 +1559,18 @@ again: ordered = btrfs_lookup_ordered_extent(inode, page_start); if (ordered) { - unlock_extent(&BTRFS_I(inode)->io_tree, page_start, - page_end, GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, page_start, + page_end, &cached_state, GFP_NOFS); unlock_page(page); btrfs_start_ordered_extent(inode, ordered, 1); goto again; } - btrfs_set_extent_delalloc(inode, page_start, page_end); + btrfs_set_extent_delalloc(inode, page_start, page_end, &cached_state); ClearPageChecked(page); out: - unlock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end, GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, page_start, page_end, + &cached_state, GFP_NOFS); out_page: unlock_page(page); page_cache_release(page); @@ -1447,7 +1616,6 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans, struct inode *inode, u64 file_pos, u64 disk_bytenr, u64 disk_num_bytes, u64 num_bytes, u64 ram_bytes, - u64 locked_end, u8 compression, u8 encryption, u16 other_encoding, int extent_type) { @@ -1473,9 +1641,8 @@ 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, root, inode, file_pos, - file_pos + num_bytes, locked_end, - file_pos, &hint, 0); + ret = btrfs_drop_extents(trans, inode, file_pos, file_pos + num_bytes, + &hint, 0); BUG_ON(ret); ins.objectid = inode->i_ino; @@ -1522,24 +1689,6 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans, * before we start the transaction. It limits the amount of btree * reads required while inside the transaction. */ -static noinline void reada_csum(struct btrfs_root *root, - struct btrfs_path *path, - struct btrfs_ordered_extent *ordered_extent) -{ - struct btrfs_ordered_sum *sum; - u64 bytenr; - - sum = list_entry(ordered_extent->list.next, struct btrfs_ordered_sum, - list); - bytenr = sum->sums[0].bytenr; - - /* - * we don't care about the results, the point of this search is - * just to get the btree leaves into ram - */ - btrfs_lookup_csum(NULL, root->fs_info->csum_root, path, bytenr, 0); -} - /* as ordered data IO finishes, this gets called so we can finish * an ordered extent if the range of bytes in the file it covers are * fully written. @@ -1550,54 +1699,39 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) struct btrfs_trans_handle *trans; struct btrfs_ordered_extent *ordered_extent = NULL; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; - struct btrfs_path *path; + struct extent_state *cached_state = NULL; int compressed = 0; int ret; - ret = btrfs_dec_test_ordered_pending(inode, start, end - start + 1); + ret = btrfs_dec_test_ordered_pending(inode, &ordered_extent, start, + end - start + 1); if (!ret) return 0; + BUG_ON(!ordered_extent); - /* - * before we join the transaction, try to do some of our IO. - * This will limit the amount of IO that we have to do with - * the transaction running. We're unlikely to need to do any - * IO if the file extents are new, the disk_i_size checks - * covers the most common case. - */ - if (start < BTRFS_I(inode)->disk_i_size) { - path = btrfs_alloc_path(); - if (path) { - ret = btrfs_lookup_file_extent(NULL, root, path, - inode->i_ino, - start, 0); - ordered_extent = btrfs_lookup_ordered_extent(inode, - start); - if (!list_empty(&ordered_extent->list)) { - btrfs_release_path(root, path); - reada_csum(root, path, ordered_extent); - } - btrfs_free_path(path); + if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) { + BUG_ON(!list_empty(&ordered_extent->list)); + ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent); + if (!ret) { + trans = btrfs_join_transaction(root, 1); + ret = btrfs_update_inode(trans, root, inode); + BUG_ON(ret); + btrfs_end_transaction(trans, root); } + goto out; } - trans = btrfs_join_transaction(root, 1); - - if (!ordered_extent) - ordered_extent = btrfs_lookup_ordered_extent(inode, start); - BUG_ON(!ordered_extent); - if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) - goto nocow; + lock_extent_bits(io_tree, ordered_extent->file_offset, + ordered_extent->file_offset + ordered_extent->len - 1, + 0, &cached_state, GFP_NOFS); - lock_extent(io_tree, ordered_extent->file_offset, - ordered_extent->file_offset + ordered_extent->len - 1, - GFP_NOFS); + trans = btrfs_join_transaction(root, 1); if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags)) compressed = 1; if (test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) { BUG_ON(compressed); - ret = btrfs_mark_extent_written(trans, root, inode, + ret = btrfs_mark_extent_written(trans, inode, ordered_extent->file_offset, ordered_extent->file_offset + ordered_extent->len); @@ -1609,8 +1743,6 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) ordered_extent->disk_len, ordered_extent->len, ordered_extent->len, - ordered_extent->file_offset + - ordered_extent->len, compressed, 0, 0, BTRFS_FILE_EXTENT_REG); unpin_extent_cache(&BTRFS_I(inode)->extent_tree, @@ -1618,25 +1750,24 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) ordered_extent->len); BUG_ON(ret); } - unlock_extent(io_tree, ordered_extent->file_offset, - ordered_extent->file_offset + ordered_extent->len - 1, - GFP_NOFS); -nocow: + unlock_extent_cached(io_tree, ordered_extent->file_offset, + ordered_extent->file_offset + + ordered_extent->len - 1, &cached_state, GFP_NOFS); + add_pending_csums(trans, inode, ordered_extent->file_offset, &ordered_extent->list); - mutex_lock(&BTRFS_I(inode)->extent_mutex); - btrfs_ordered_update_i_size(inode, ordered_extent); - btrfs_update_inode(trans, root, inode); - btrfs_remove_ordered_extent(inode, ordered_extent); - mutex_unlock(&BTRFS_I(inode)->extent_mutex); - + /* this also removes the ordered extent from the tree */ + btrfs_ordered_update_i_size(inode, 0, ordered_extent); + ret = btrfs_update_inode(trans, root, inode); + BUG_ON(ret); + btrfs_end_transaction(trans, root); +out: /* once for us */ btrfs_put_ordered_extent(ordered_extent); /* once for the tree */ btrfs_put_ordered_extent(ordered_extent); - btrfs_end_transaction(trans, root); return 0; } @@ -1859,6 +1990,54 @@ zeroit: return -EIO; } +struct delayed_iput { + struct list_head list; + struct inode *inode; +}; + +void btrfs_add_delayed_iput(struct inode *inode) +{ + struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info; + struct delayed_iput *delayed; + + if (atomic_add_unless(&inode->i_count, -1, 1)) + return; + + delayed = kmalloc(sizeof(*delayed), GFP_NOFS | __GFP_NOFAIL); + delayed->inode = inode; + + spin_lock(&fs_info->delayed_iput_lock); + list_add_tail(&delayed->list, &fs_info->delayed_iputs); + spin_unlock(&fs_info->delayed_iput_lock); +} + +void btrfs_run_delayed_iputs(struct btrfs_root *root) +{ + LIST_HEAD(list); + struct btrfs_fs_info *fs_info = root->fs_info; + struct delayed_iput *delayed; + int empty; + + spin_lock(&fs_info->delayed_iput_lock); + empty = list_empty(&fs_info->delayed_iputs); + spin_unlock(&fs_info->delayed_iput_lock); + 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); + + while (!list_empty(&list)) { + delayed = list_entry(list.next, struct delayed_iput, list); + list_del(&delayed->list); + iput(delayed->inode); + kfree(delayed); + } + up_read(&root->fs_info->cleanup_work_sem); +} + /* * This creates an orphan entry for the given inode in case something goes * wrong in the middle of an unlink/truncate. @@ -1931,16 +2110,17 @@ void btrfs_orphan_cleanup(struct btrfs_root *root) struct inode *inode; int ret = 0, nr_unlink = 0, nr_truncate = 0; - path = btrfs_alloc_path(); - if (!path) + if (!xchg(&root->clean_orphans, 0)) return; + + path = btrfs_alloc_path(); + BUG_ON(!path); path->reada = -1; key.objectid = BTRFS_ORPHAN_OBJECTID; btrfs_set_key_type(&key, BTRFS_ORPHAN_ITEM_KEY); key.offset = (u64)-1; - while (1) { ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); if (ret < 0) { @@ -1982,7 +2162,7 @@ void btrfs_orphan_cleanup(struct btrfs_root *root) found_key.objectid = found_key.offset; found_key.type = BTRFS_INODE_ITEM_KEY; found_key.offset = 0; - inode = btrfs_iget(root->fs_info->sb, &found_key, root); + inode = btrfs_iget(root->fs_info->sb, &found_key, root, NULL); if (IS_ERR(inode)) break; @@ -2354,7 +2534,19 @@ static int btrfs_unlink(struct inode *dir, struct dentry *dentry) root = BTRFS_I(dir)->root; + /* + * 5 items for unlink inode + * 1 for orphan + */ + ret = btrfs_reserve_metadata_space(root, 6); + if (ret) + return ret; + trans = btrfs_start_transaction(root, 1); + if (IS_ERR(trans)) { + btrfs_unreserve_metadata_space(root, 6); + return PTR_ERR(trans); + } btrfs_set_trans_block_group(trans, dir); @@ -2369,6 +2561,7 @@ static int btrfs_unlink(struct inode *dir, struct dentry *dentry) nr = trans->blocks_used; btrfs_end_transaction_throttle(trans, root); + btrfs_unreserve_metadata_space(root, 6); btrfs_btree_balance_dirty(root, nr); return ret; } @@ -2449,7 +2642,16 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry) inode->i_ino == BTRFS_FIRST_FREE_OBJECTID) return -ENOTEMPTY; + ret = btrfs_reserve_metadata_space(root, 5); + if (ret) + return ret; + trans = btrfs_start_transaction(root, 1); + if (IS_ERR(trans)) { + btrfs_unreserve_metadata_space(root, 5); + return PTR_ERR(trans); + } + btrfs_set_trans_block_group(trans, dir); if (unlikely(inode->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)) { @@ -2472,6 +2674,7 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry) out: nr = trans->blocks_used; ret = btrfs_end_transaction_throttle(trans, root); + btrfs_unreserve_metadata_space(root, 5); btrfs_btree_balance_dirty(root, nr); if (ret && !err) @@ -2662,37 +2865,40 @@ out: * min_type is the minimum key type to truncate down to. If set to 0, this * will kill all the items on this inode, including the INODE_ITEM_KEY. */ -noinline int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct inode *inode, - u64 new_size, u32 min_type) +int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct inode *inode, + u64 new_size, u32 min_type) { - int ret; struct btrfs_path *path; - struct btrfs_key key; - struct btrfs_key found_key; - u32 found_type = (u8)-1; struct extent_buffer *leaf; struct btrfs_file_extent_item *fi; + struct btrfs_key key; + struct btrfs_key found_key; u64 extent_start = 0; u64 extent_num_bytes = 0; u64 extent_offset = 0; u64 item_end = 0; + u64 mask = root->sectorsize - 1; + u32 found_type = (u8)-1; int found_extent; int del_item; int pending_del_nr = 0; int pending_del_slot = 0; int extent_type = -1; int encoding; - u64 mask = root->sectorsize - 1; + int ret; + int err = 0; + + BUG_ON(new_size > 0 && min_type != BTRFS_EXTENT_DATA_KEY); if (root->ref_cows) btrfs_drop_extent_cache(inode, new_size & (~mask), (u64)-1, 0); + path = btrfs_alloc_path(); BUG_ON(!path); path->reada = -1; - /* FIXME, add redo link to tree so we don't leak on crash */ key.objectid = inode->i_ino; key.offset = (u64)-1; key.type = (u8)-1; @@ -2700,17 +2906,17 @@ noinline int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans, search_again: path->leave_spinning = 1; ret = btrfs_search_slot(trans, root, &key, path, -1, 1); - if (ret < 0) - goto error; + if (ret < 0) { + err = ret; + goto out; + } if (ret > 0) { /* there are no items in the tree for us to truncate, we're * done */ - if (path->slots[0] == 0) { - ret = 0; - goto error; - } + if (path->slots[0] == 0) + goto out; path->slots[0]--; } @@ -2745,28 +2951,17 @@ search_again: } item_end--; } - if (item_end < new_size) { - if (found_type == BTRFS_DIR_ITEM_KEY) - found_type = BTRFS_INODE_ITEM_KEY; - else if (found_type == BTRFS_EXTENT_ITEM_KEY) - found_type = BTRFS_EXTENT_DATA_KEY; - else if (found_type == BTRFS_EXTENT_DATA_KEY) - found_type = BTRFS_XATTR_ITEM_KEY; - else if (found_type == BTRFS_XATTR_ITEM_KEY) - found_type = BTRFS_INODE_REF_KEY; - else if (found_type) - found_type--; - else + if (found_type > min_type) { + del_item = 1; + } else { + if (item_end < new_size) break; - btrfs_set_key_type(&key, found_type); - goto next; + if (found_key.offset >= new_size) + del_item = 1; + else + del_item = 0; } - if (found_key.offset >= new_size) - del_item = 1; - else - del_item = 0; found_extent = 0; - /* FIXME, shrink the extent if the ref count is only 1 */ if (found_type != BTRFS_EXTENT_DATA_KEY) goto delete; @@ -2853,42 +3048,36 @@ delete: inode->i_ino, extent_offset); BUG_ON(ret); } -next: - if (path->slots[0] == 0) { - if (pending_del_nr) - goto del_pending; - btrfs_release_path(root, path); - if (found_type == BTRFS_INODE_ITEM_KEY) - break; - goto search_again; - } - path->slots[0]--; - if (pending_del_nr && - path->slots[0] + 1 != pending_del_slot) { - struct btrfs_key debug; -del_pending: - btrfs_item_key_to_cpu(path->nodes[0], &debug, - pending_del_slot); - ret = btrfs_del_items(trans, root, path, - pending_del_slot, - pending_del_nr); - BUG_ON(ret); - pending_del_nr = 0; + if (found_type == BTRFS_INODE_ITEM_KEY) + break; + + if (path->slots[0] == 0 || + path->slots[0] != pending_del_slot) { + if (root->ref_cows) { + err = -EAGAIN; + goto out; + } + if (pending_del_nr) { + ret = btrfs_del_items(trans, root, path, + pending_del_slot, + pending_del_nr); + BUG_ON(ret); + pending_del_nr = 0; + } btrfs_release_path(root, path); - if (found_type == BTRFS_INODE_ITEM_KEY) - break; goto search_again; + } else { + path->slots[0]--; } } - ret = 0; -error: +out: if (pending_del_nr) { ret = btrfs_del_items(trans, root, path, pending_del_slot, pending_del_nr); } btrfs_free_path(path); - return ret; + return err; } /* @@ -2901,6 +3090,7 @@ static int btrfs_truncate_page(struct address_space *mapping, loff_t from) struct btrfs_root *root = BTRFS_I(inode)->root; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct btrfs_ordered_extent *ordered; + struct extent_state *cached_state = NULL; char *kaddr; u32 blocksize = root->sectorsize; pgoff_t index = from >> PAGE_CACHE_SHIFT; @@ -2912,12 +3102,22 @@ static int btrfs_truncate_page(struct address_space *mapping, loff_t from) if ((offset & (blocksize - 1)) == 0) goto out; + ret = btrfs_check_data_free_space(root, inode, PAGE_CACHE_SIZE); + if (ret) + goto out; + + ret = btrfs_reserve_metadata_for_delalloc(root, inode, 1); + if (ret) + goto out; ret = -ENOMEM; again: page = grab_cache_page(mapping, index); - if (!page) + if (!page) { + btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE); + btrfs_unreserve_metadata_for_delalloc(root, inode, 1); goto out; + } page_start = page_offset(page); page_end = page_start + PAGE_CACHE_SIZE - 1; @@ -2937,12 +3137,14 @@ again: } wait_on_page_writeback(page); - lock_extent(io_tree, page_start, page_end, GFP_NOFS); + lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state, + GFP_NOFS); set_page_extent_mapped(page); ordered = btrfs_lookup_ordered_extent(inode, page_start); if (ordered) { - unlock_extent(io_tree, page_start, page_end, GFP_NOFS); + unlock_extent_cached(io_tree, page_start, page_end, + &cached_state, GFP_NOFS); unlock_page(page); page_cache_release(page); btrfs_start_ordered_extent(inode, ordered, 1); @@ -2950,7 +3152,18 @@ again: goto again; } - btrfs_set_extent_delalloc(inode, page_start, page_end); + clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, page_end, + EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING, + 0, 0, &cached_state, GFP_NOFS); + + ret = btrfs_set_extent_delalloc(inode, page_start, page_end, + &cached_state); + if (ret) { + unlock_extent_cached(io_tree, page_start, page_end, + &cached_state, GFP_NOFS); + goto out_unlock; + } + ret = 0; if (offset != PAGE_CACHE_SIZE) { kaddr = kmap(page); @@ -2960,9 +3173,13 @@ again: } ClearPageChecked(page); set_page_dirty(page); - unlock_extent(io_tree, page_start, page_end, GFP_NOFS); + unlock_extent_cached(io_tree, page_start, page_end, &cached_state, + GFP_NOFS); out_unlock: + if (ret) + btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE); + btrfs_unreserve_metadata_for_delalloc(root, inode, 1); unlock_page(page); page_cache_release(page); out: @@ -2975,38 +3192,32 @@ int btrfs_cont_expand(struct inode *inode, loff_t size) struct btrfs_root *root = BTRFS_I(inode)->root; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct extent_map *em; + struct extent_state *cached_state = NULL; u64 mask = root->sectorsize - 1; u64 hole_start = (inode->i_size + mask) & ~mask; u64 block_end = (size + mask) & ~mask; u64 last_byte; u64 cur_offset; u64 hole_size; - int err; + int err = 0; if (size <= hole_start) return 0; - err = btrfs_check_metadata_free_space(root); - if (err) - return err; - - btrfs_truncate_page(inode->i_mapping, inode->i_size); - while (1) { struct btrfs_ordered_extent *ordered; btrfs_wait_ordered_range(inode, hole_start, block_end - hole_start); - lock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS); + lock_extent_bits(io_tree, hole_start, block_end - 1, 0, + &cached_state, GFP_NOFS); ordered = btrfs_lookup_ordered_extent(inode, hole_start); if (!ordered) break; - unlock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS); + unlock_extent_cached(io_tree, hole_start, block_end - 1, + &cached_state, GFP_NOFS); btrfs_put_ordered_extent(ordered); } - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); - cur_offset = hole_start; while (1) { em = btrfs_get_extent(inode, NULL, 0, cur_offset, @@ -3014,34 +3225,121 @@ int btrfs_cont_expand(struct inode *inode, loff_t size) BUG_ON(IS_ERR(em) || !em); last_byte = min(extent_map_end(em), block_end); last_byte = (last_byte + mask) & ~mask; - if (test_bit(EXTENT_FLAG_VACANCY, &em->flags)) { + if (!test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) { u64 hint_byte = 0; hole_size = last_byte - cur_offset; - err = btrfs_drop_extents(trans, root, inode, - cur_offset, - cur_offset + hole_size, - block_end, - cur_offset, &hint_byte, 1); + + err = btrfs_reserve_metadata_space(root, 2); if (err) break; + + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, inode); + + err = btrfs_drop_extents(trans, inode, cur_offset, + cur_offset + hole_size, + &hint_byte, 1); + BUG_ON(err); + err = btrfs_insert_file_extent(trans, root, inode->i_ino, cur_offset, 0, 0, hole_size, 0, hole_size, 0, 0, 0); + BUG_ON(err); + btrfs_drop_extent_cache(inode, hole_start, last_byte - 1, 0); + + btrfs_end_transaction(trans, root); + btrfs_unreserve_metadata_space(root, 2); } free_extent_map(em); cur_offset = last_byte; - if (err || cur_offset >= block_end) + if (cur_offset >= block_end) break; } - btrfs_end_transaction(trans, root); - unlock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS); + unlock_extent_cached(io_tree, hole_start, block_end - 1, &cached_state, + GFP_NOFS); return err; } +static int btrfs_setattr_size(struct inode *inode, struct iattr *attr) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_trans_handle *trans; + unsigned long nr; + int ret; + + if (attr->ia_size == inode->i_size) + return 0; + + if (attr->ia_size > inode->i_size) { + unsigned long limit; + limit = current->signal->rlim[RLIMIT_FSIZE].rlim_cur; + if (attr->ia_size > inode->i_sb->s_maxbytes) + return -EFBIG; + if (limit != RLIM_INFINITY && attr->ia_size > limit) { + send_sig(SIGXFSZ, current, 0); + return -EFBIG; + } + } + + ret = btrfs_reserve_metadata_space(root, 1); + if (ret) + return ret; + + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, inode); + + ret = btrfs_orphan_add(trans, inode); + BUG_ON(ret); + + nr = trans->blocks_used; + btrfs_end_transaction(trans, root); + btrfs_unreserve_metadata_space(root, 1); + btrfs_btree_balance_dirty(root, nr); + + if (attr->ia_size > inode->i_size) { + ret = btrfs_cont_expand(inode, attr->ia_size); + if (ret) { + btrfs_truncate(inode); + return ret; + } + + i_size_write(inode, attr->ia_size); + btrfs_ordered_update_i_size(inode, inode->i_size, NULL); + + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, inode); + + ret = btrfs_update_inode(trans, root, inode); + BUG_ON(ret); + if (inode->i_nlink > 0) { + ret = btrfs_orphan_del(trans, inode); + BUG_ON(ret); + } + nr = trans->blocks_used; + btrfs_end_transaction(trans, root); + btrfs_btree_balance_dirty(root, nr); + return 0; + } + + /* + * We're truncating a file that used to have good data down to + * zero. Make sure it gets into the ordered flush list so that + * any new writes get down to disk quickly. + */ + if (attr->ia_size == 0) + BTRFS_I(inode)->ordered_data_close = 1; + + /* we don't support swapfiles, so vmtruncate shouldn't fail */ + ret = vmtruncate(inode, attr->ia_size); + BUG_ON(ret); + + return 0; +} + static int btrfs_setattr(struct dentry *dentry, struct iattr *attr) { struct inode *inode = dentry->d_inode; @@ -3052,23 +3350,14 @@ static int btrfs_setattr(struct dentry *dentry, struct iattr *attr) return err; if (S_ISREG(inode->i_mode) && (attr->ia_valid & ATTR_SIZE)) { - if (attr->ia_size > inode->i_size) { - err = btrfs_cont_expand(inode, attr->ia_size); - if (err) - return err; - } else if (inode->i_size > 0 && - attr->ia_size == 0) { - - /* we're truncating a file that used to have good - * data down to zero. Make sure it gets into - * the ordered flush list so that any new writes - * get down to disk quickly. - */ - BTRFS_I(inode)->ordered_data_close = 1; - } + err = btrfs_setattr_size(inode, attr); + if (err) + return err; } + attr->ia_valid &= ~ATTR_SIZE; - err = inode_setattr(inode, attr); + if (attr->ia_valid) + err = inode_setattr(inode, attr); if (!err && ((attr->ia_valid & ATTR_MODE))) err = btrfs_acl_chmod(inode); @@ -3089,36 +3378,43 @@ void btrfs_delete_inode(struct inode *inode) } btrfs_wait_ordered_range(inode, 0, (u64)-1); + if (root->fs_info->log_root_recovering) { + BUG_ON(!list_empty(&BTRFS_I(inode)->i_orphan)); + goto no_delete; + } + if (inode->i_nlink > 0) { BUG_ON(btrfs_root_refs(&root->root_item) != 0); goto no_delete; } btrfs_i_size_write(inode, 0); - trans = btrfs_join_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); - ret = btrfs_truncate_inode_items(trans, root, inode, inode->i_size, 0); - if (ret) { - btrfs_orphan_del(NULL, inode); - goto no_delete_lock; - } + while (1) { + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, inode); + ret = btrfs_truncate_inode_items(trans, root, inode, 0, 0); - btrfs_orphan_del(trans, inode); + if (ret != -EAGAIN) + break; - nr = trans->blocks_used; - clear_inode(inode); + nr = trans->blocks_used; + btrfs_end_transaction(trans, root); + trans = NULL; + btrfs_btree_balance_dirty(root, nr); + } - btrfs_end_transaction(trans, root); - btrfs_btree_balance_dirty(root, nr); - return; + if (ret == 0) { + ret = btrfs_orphan_del(trans, inode); + BUG_ON(ret); + } -no_delete_lock: nr = trans->blocks_used; btrfs_end_transaction(trans, root); btrfs_btree_balance_dirty(root, nr); no_delete: clear_inode(inode); + return; } /* @@ -3353,6 +3649,7 @@ static noinline void init_btrfs_i(struct inode *inode) bi->generation = 0; bi->sequence = 0; bi->last_trans = 0; + bi->last_sub_trans = 0; bi->logged_trans = 0; bi->delalloc_bytes = 0; bi->reserved_bytes = 0; @@ -3361,6 +3658,7 @@ static noinline void init_btrfs_i(struct inode *inode) bi->index_cnt = (u64)-1; bi->last_unlink_trans = 0; bi->ordered_data_close = 0; + bi->force_compress = 0; extent_map_tree_init(&BTRFS_I(inode)->extent_tree, GFP_NOFS); extent_io_tree_init(&BTRFS_I(inode)->io_tree, inode->i_mapping, GFP_NOFS); @@ -3370,7 +3668,6 @@ static noinline void init_btrfs_i(struct inode *inode) INIT_LIST_HEAD(&BTRFS_I(inode)->ordered_operations); RB_CLEAR_NODE(&BTRFS_I(inode)->rb_node); btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree); - mutex_init(&BTRFS_I(inode)->extent_mutex); mutex_init(&BTRFS_I(inode)->log_mutex); } @@ -3410,7 +3707,7 @@ static struct inode *btrfs_iget_locked(struct super_block *s, * Returns in *is_new if the inode was read from disk */ struct inode *btrfs_iget(struct super_block *s, struct btrfs_key *location, - struct btrfs_root *root) + struct btrfs_root *root, int *new) { struct inode *inode; @@ -3425,6 +3722,8 @@ struct inode *btrfs_iget(struct super_block *s, struct btrfs_key *location, inode_tree_add(inode); unlock_new_inode(inode); + if (new) + *new = 1; } return inode; @@ -3477,7 +3776,7 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry) return NULL; if (location.type == BTRFS_INODE_ITEM_KEY) { - inode = btrfs_iget(dir->i_sb, &location, root); + inode = btrfs_iget(dir->i_sb, &location, root, NULL); return inode; } @@ -3492,10 +3791,17 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry) else inode = new_simple_dir(dir->i_sb, &location, sub_root); } else { - inode = btrfs_iget(dir->i_sb, &location, sub_root); + inode = btrfs_iget(dir->i_sb, &location, sub_root, NULL); } srcu_read_unlock(&root->fs_info->subvol_srcu, index); + if (root != sub_root) { + down_read(&root->fs_info->cleanup_work_sem); + if (!(inode->i_sb->s_flags & MS_RDONLY)) + btrfs_orphan_cleanup(sub_root); + up_read(&root->fs_info->cleanup_work_sem); + } + return inode; } @@ -3503,12 +3809,14 @@ static int btrfs_dentry_delete(struct dentry *dentry) { struct btrfs_root *root; - if (!dentry->d_inode) - return 0; + if (!dentry->d_inode && !IS_ROOT(dentry)) + dentry = dentry->d_parent; - root = BTRFS_I(dentry->d_inode)->root; - if (btrfs_root_refs(&root->root_item) == 0) - return 1; + if (dentry->d_inode) { + root = BTRFS_I(dentry->d_inode)->root; + if (btrfs_root_refs(&root->root_item) == 0) + return 1; + } return 0; } @@ -3668,7 +3976,11 @@ skip: /* Reached end of directory/root. Bump pos past the last item. */ if (key_type == BTRFS_DIR_INDEX_KEY) - filp->f_pos = INT_LIMIT(off_t); + /* + * 32-bit glibc will use getdents64, but then strtol - + * so the last number we can serve is this. + */ + filp->f_pos = 0x7fffffff; else filp->f_pos++; nopos: @@ -3678,7 +3990,7 @@ err: return ret; } -int btrfs_write_inode(struct inode *inode, int wait) +int btrfs_write_inode(struct inode *inode, struct writeback_control *wbc) { struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_trans_handle *trans; @@ -3687,7 +3999,7 @@ int btrfs_write_inode(struct inode *inode, int wait) if (root->fs_info->btree_inode == inode) return 0; - if (wait) { + if (wbc->sync_mode == WB_SYNC_ALL) { trans = btrfs_join_transaction(root, 1); btrfs_set_trans_block_group(trans, inode); ret = btrfs_commit_transaction(trans, root); @@ -3990,11 +4302,18 @@ static int btrfs_mknod(struct inode *dir, struct dentry *dentry, if (!new_valid_dev(rdev)) return -EINVAL; - err = btrfs_check_metadata_free_space(root); + /* + * 2 for inode item and ref + * 2 for dir items + * 1 for xattr if selinux is on + */ + err = btrfs_reserve_metadata_space(root, 5); if (err) - goto fail; + return err; trans = btrfs_start_transaction(root, 1); + if (!trans) + goto fail; btrfs_set_trans_block_group(trans, dir); err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid); @@ -4011,7 +4330,7 @@ static int btrfs_mknod(struct inode *dir, struct dentry *dentry, if (IS_ERR(inode)) goto out_unlock; - err = btrfs_init_inode_security(inode, dir); + err = btrfs_init_inode_security(trans, inode, dir); if (err) { drop_inode = 1; goto out_unlock; @@ -4032,6 +4351,7 @@ out_unlock: nr = trans->blocks_used; btrfs_end_transaction_throttle(trans, root); fail: + btrfs_unreserve_metadata_space(root, 5); if (drop_inode) { inode_dec_link_count(inode); iput(inode); @@ -4052,10 +4372,18 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry, u64 objectid; u64 index = 0; - err = btrfs_check_metadata_free_space(root); + /* + * 2 for inode item and ref + * 2 for dir items + * 1 for xattr if selinux is on + */ + err = btrfs_reserve_metadata_space(root, 5); if (err) - goto fail; + return err; + trans = btrfs_start_transaction(root, 1); + if (!trans) + goto fail; btrfs_set_trans_block_group(trans, dir); err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid); @@ -4073,7 +4401,7 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry, if (IS_ERR(inode)) goto out_unlock; - err = btrfs_init_inode_security(inode, dir); + err = btrfs_init_inode_security(trans, inode, dir); if (err) { drop_inode = 1; goto out_unlock; @@ -4096,6 +4424,7 @@ out_unlock: nr = trans->blocks_used; btrfs_end_transaction_throttle(trans, root); fail: + btrfs_unreserve_metadata_space(root, 5); if (drop_inode) { inode_dec_link_count(inode); iput(inode); @@ -4118,10 +4447,20 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir, if (inode->i_nlink == 0) return -ENOENT; - btrfs_inc_nlink(inode); - err = btrfs_check_metadata_free_space(root); + /* do not allow sys_link's with other subvols of the same device */ + if (root->objectid != BTRFS_I(inode)->root->objectid) + return -EPERM; + + /* + * 1 item for inode ref + * 2 items for dir items + */ + err = btrfs_reserve_metadata_space(root, 3); if (err) - goto fail; + return err; + + btrfs_inc_nlink(inode); + err = btrfs_set_inode_index(dir, &index); if (err) goto fail; @@ -4145,6 +4484,7 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir, nr = trans->blocks_used; btrfs_end_transaction_throttle(trans, root); fail: + btrfs_unreserve_metadata_space(root, 3); if (drop_inode) { inode_dec_link_count(inode); iput(inode); @@ -4164,22 +4504,26 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode) u64 index = 0; unsigned long nr = 1; - err = btrfs_check_metadata_free_space(root); + /* + * 2 items for inode and ref + * 2 items for dir items + * 1 for xattr if selinux is on + */ + err = btrfs_reserve_metadata_space(root, 5); if (err) - goto out_unlock; + return err; trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, dir); - - if (IS_ERR(trans)) { - err = PTR_ERR(trans); + if (!trans) { + err = -ENOMEM; goto out_unlock; } + btrfs_set_trans_block_group(trans, dir); err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid); if (err) { err = -ENOSPC; - goto out_unlock; + goto out_fail; } inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name, @@ -4194,7 +4538,7 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode) drop_on_err = 1; - err = btrfs_init_inode_security(inode, dir); + err = btrfs_init_inode_security(trans, inode, dir); if (err) goto out_fail; @@ -4223,6 +4567,7 @@ out_fail: btrfs_end_transaction_throttle(trans, root); out_unlock: + btrfs_unreserve_metadata_space(root, 5); if (drop_on_err) iput(inode); btrfs_btree_balance_dirty(root, nr); @@ -4656,6 +5001,7 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) { struct extent_io_tree *tree; struct btrfs_ordered_extent *ordered; + struct extent_state *cached_state = NULL; u64 page_start = page_offset(page); u64 page_end = page_start + PAGE_CACHE_SIZE - 1; @@ -4674,7 +5020,8 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) btrfs_releasepage(page, GFP_NOFS); return; } - lock_extent(tree, page_start, page_end, GFP_NOFS); + lock_extent_bits(tree, page_start, page_end, 0, &cached_state, + GFP_NOFS); ordered = btrfs_lookup_ordered_extent(page->mapping->host, page_offset(page)); if (ordered) { @@ -4684,7 +5031,8 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) */ clear_extent_bit(tree, page_start, page_end, EXTENT_DIRTY | EXTENT_DELALLOC | - EXTENT_LOCKED, 1, 0, NULL, GFP_NOFS); + EXTENT_LOCKED | EXTENT_DO_ACCOUNTING, 1, 0, + &cached_state, GFP_NOFS); /* * whoever cleared the private bit is responsible * for the finish_ordered_io @@ -4694,11 +5042,13 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) page_start, page_end); } btrfs_put_ordered_extent(ordered); - lock_extent(tree, page_start, page_end, GFP_NOFS); + cached_state = NULL; + lock_extent_bits(tree, page_start, page_end, 0, &cached_state, + GFP_NOFS); } clear_extent_bit(tree, page_start, page_end, - EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC, - 1, 1, NULL, GFP_NOFS); + EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC | + EXTENT_DO_ACCOUNTING, 1, 1, &cached_state, GFP_NOFS); __btrfs_releasepage(page, GFP_NOFS); ClearPageChecked(page); @@ -4731,6 +5081,7 @@ int btrfs_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf) struct btrfs_root *root = BTRFS_I(inode)->root; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct btrfs_ordered_extent *ordered; + struct extent_state *cached_state = NULL; char *kaddr; unsigned long zero_start; loff_t size; @@ -4747,6 +5098,13 @@ int btrfs_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf) goto out; } + ret = btrfs_reserve_metadata_for_delalloc(root, inode, 1); + if (ret) { + btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE); + ret = VM_FAULT_SIGBUS; + goto out; + } + ret = VM_FAULT_NOPAGE; /* make the VM retry the fault */ again: lock_page(page); @@ -4762,7 +5120,8 @@ again: } wait_on_page_writeback(page); - lock_extent(io_tree, page_start, page_end, GFP_NOFS); + lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state, + GFP_NOFS); set_page_extent_mapped(page); /* @@ -4771,14 +5130,34 @@ again: */ ordered = btrfs_lookup_ordered_extent(inode, page_start); if (ordered) { - unlock_extent(io_tree, page_start, page_end, GFP_NOFS); + unlock_extent_cached(io_tree, page_start, page_end, + &cached_state, GFP_NOFS); unlock_page(page); btrfs_start_ordered_extent(inode, ordered, 1); btrfs_put_ordered_extent(ordered); goto again; } - btrfs_set_extent_delalloc(inode, page_start, page_end); + /* + * XXX - page_mkwrite gets called every time the page is dirtied, even + * if it was already dirty, so for space accounting reasons we need to + * clear any delalloc bits for the range we are fixing to save. There + * is probably a better way to do this, but for now keep consistent with + * 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, + 0, 0, &cached_state, GFP_NOFS); + + ret = btrfs_set_extent_delalloc(inode, page_start, page_end, + &cached_state); + if (ret) { + unlock_extent_cached(io_tree, page_start, page_end, + &cached_state, GFP_NOFS); + ret = VM_FAULT_SIGBUS; + btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE); + goto out_unlock; + } ret = 0; /* page is wholly or partially inside EOF */ @@ -4797,10 +5176,13 @@ again: set_page_dirty(page); SetPageUptodate(page); - BTRFS_I(inode)->last_trans = root->fs_info->generation + 1; - unlock_extent(io_tree, page_start, page_end, GFP_NOFS); + BTRFS_I(inode)->last_trans = root->fs_info->generation; + BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid; + + unlock_extent_cached(io_tree, page_start, page_end, &cached_state, GFP_NOFS); out_unlock: + btrfs_unreserve_metadata_for_delalloc(root, inode, 1); if (!ret) return VM_FAULT_LOCKED; unlock_page(page); @@ -4816,15 +5198,20 @@ static void btrfs_truncate(struct inode *inode) unsigned long nr; u64 mask = root->sectorsize - 1; - if (!S_ISREG(inode->i_mode)) + if (!S_ISREG(inode->i_mode)) { + WARN_ON(1); return; - if (IS_APPEND(inode) || IS_IMMUTABLE(inode)) + } + + ret = btrfs_truncate_page(inode->i_mapping, inode->i_size); + if (ret) return; - btrfs_truncate_page(inode->i_mapping, inode->i_size); btrfs_wait_ordered_range(inode, inode->i_size & (~mask), (u64)-1); + btrfs_ordered_update_i_size(inode, inode->i_size, NULL); trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, inode); /* * setattr is responsible for setting the ordered_data_close flag, @@ -4846,21 +5233,32 @@ static void btrfs_truncate(struct inode *inode) if (inode->i_size == 0 && BTRFS_I(inode)->ordered_data_close) btrfs_add_ordered_operation(trans, root, inode); - btrfs_set_trans_block_group(trans, inode); - btrfs_i_size_write(inode, inode->i_size); + while (1) { + ret = btrfs_truncate_inode_items(trans, root, inode, + inode->i_size, + BTRFS_EXTENT_DATA_KEY); + if (ret != -EAGAIN) + break; - ret = btrfs_orphan_add(trans, inode); - if (ret) - goto out; - /* FIXME, add redo link to tree so we don't leak on crash */ - ret = btrfs_truncate_inode_items(trans, root, inode, inode->i_size, - BTRFS_EXTENT_DATA_KEY); - btrfs_update_inode(trans, root, inode); + ret = btrfs_update_inode(trans, root, inode); + BUG_ON(ret); + + nr = trans->blocks_used; + btrfs_end_transaction(trans, root); + btrfs_btree_balance_dirty(root, nr); + + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, inode); + } + + if (ret == 0 && inode->i_nlink > 0) { + ret = btrfs_orphan_del(trans, inode); + BUG_ON(ret); + } - ret = btrfs_orphan_del(trans, inode); + ret = btrfs_update_inode(trans, root, inode); BUG_ON(ret); -out: nr = trans->blocks_used; ret = btrfs_end_transaction_throttle(trans, root); BUG_ON(ret); @@ -4916,7 +5314,12 @@ struct inode *btrfs_alloc_inode(struct super_block *sb) if (!ei) return NULL; ei->last_trans = 0; + ei->last_sub_trans = 0; ei->logged_trans = 0; + ei->outstanding_extents = 0; + ei->reserved_extents = 0; + ei->root = NULL; + spin_lock_init(&ei->accounting_lock); btrfs_ordered_inode_tree_init(&ei->ordered_tree); INIT_LIST_HEAD(&ei->i_orphan); INIT_LIST_HEAD(&ei->ordered_operations); @@ -4932,6 +5335,14 @@ void btrfs_destroy_inode(struct inode *inode) WARN_ON(inode->i_data.nrpages); /* + * This can happen where we create an inode, but somebody else also + * created the same inode and we need to destroy the one we already + * created. + */ + if (!root) + goto free; + + /* * Make sure we're properly removed from the ordered operation * lists. */ @@ -4944,9 +5355,9 @@ void btrfs_destroy_inode(struct inode *inode) spin_lock(&root->list_lock); if (!list_empty(&BTRFS_I(inode)->i_orphan)) { - printk(KERN_ERR "BTRFS: inode %lu: inode still on the orphan" - " list\n", inode->i_ino); - dump_stack(); + printk(KERN_INFO "BTRFS: inode %lu still on the orphan list\n", + inode->i_ino); + list_del_init(&BTRFS_I(inode)->i_orphan); } spin_unlock(&root->list_lock); @@ -4966,6 +5377,7 @@ void btrfs_destroy_inode(struct inode *inode) } inode_tree_del(inode); btrfs_drop_extent_cache(inode, 0, (u64)-1, 0); +free: kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode)); } @@ -5070,7 +5482,15 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry, new_inode->i_size > BTRFS_EMPTY_DIR_SIZE) return -ENOTEMPTY; - ret = btrfs_check_metadata_free_space(root); + /* + * We want to reserve the absolute worst case amount of items. So if + * both inodes are subvols and we need to unlink them then that would + * require 4 item modifications, but if they are both normal inodes it + * would require 5 item modifications, so we'll assume their normal + * inodes. So 5 * 2 is 10, plus 1 for the new link, so 11 total items + * should cover the worst case number of items we'll modify. + */ + ret = btrfs_reserve_metadata_space(root, 11); if (ret) return ret; @@ -5185,6 +5605,8 @@ out_fail: if (old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID) up_read(&root->fs_info->subvol_sem); + + btrfs_unreserve_metadata_space(root, 11); return ret; } @@ -5192,7 +5614,7 @@ out_fail: * some fairly slow code that needs optimization. This walks the list * of all the inodes with pending delalloc and forces them to disk. */ -int btrfs_start_delalloc_inodes(struct btrfs_root *root) +int btrfs_start_delalloc_inodes(struct btrfs_root *root, int delay_iput) { struct list_head *head = &root->fs_info->delalloc_inodes; struct btrfs_inode *binode; @@ -5211,7 +5633,10 @@ int btrfs_start_delalloc_inodes(struct btrfs_root *root) spin_unlock(&root->fs_info->delalloc_lock); if (inode) { filemap_flush(inode->i_mapping); - iput(inode); + if (delay_iput) + btrfs_add_delayed_iput(inode); + else + iput(inode); } cond_resched(); spin_lock(&root->fs_info->delalloc_lock); @@ -5256,11 +5681,18 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry, if (name_len > BTRFS_MAX_INLINE_DATA_SIZE(root)) return -ENAMETOOLONG; - err = btrfs_check_metadata_free_space(root); + /* + * 2 items for inode item and ref + * 2 items for dir items + * 1 item for xattr if selinux is on + */ + err = btrfs_reserve_metadata_space(root, 5); if (err) - goto out_fail; + return err; trans = btrfs_start_transaction(root, 1); + if (!trans) + goto out_fail; btrfs_set_trans_block_group(trans, dir); err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid); @@ -5278,7 +5710,7 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry, if (IS_ERR(inode)) goto out_unlock; - err = btrfs_init_inode_security(inode, dir); + err = btrfs_init_inode_security(trans, inode, dir); if (err) { drop_inode = 1; goto out_unlock; @@ -5341,6 +5773,7 @@ out_unlock: nr = trans->blocks_used; btrfs_end_transaction_throttle(trans, root); out_fail: + btrfs_unreserve_metadata_space(root, 5); if (drop_inode) { inode_dec_link_count(inode); iput(inode); @@ -5349,56 +5782,83 @@ out_fail: return err; } -static int prealloc_file_range(struct btrfs_trans_handle *trans, - struct inode *inode, u64 start, u64 end, - u64 locked_end, u64 alloc_hint, int mode) +static int prealloc_file_range(struct inode *inode, u64 start, u64 end, + u64 alloc_hint, int mode, loff_t actual_len) { + struct btrfs_trans_handle *trans; struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_key ins; u64 alloc_size; u64 cur_offset = start; u64 num_bytes = end - start; int ret = 0; + u64 i_size; while (num_bytes > 0) { alloc_size = min(num_bytes, root->fs_info->max_extent); + + trans = btrfs_start_transaction(root, 1); + ret = btrfs_reserve_extent(trans, root, alloc_size, root->sectorsize, 0, alloc_hint, (u64)-1, &ins, 1); if (ret) { WARN_ON(1); - goto out; + goto stop_trans; } + + ret = btrfs_reserve_metadata_space(root, 3); + if (ret) { + btrfs_free_reserved_extent(root, ins.objectid, + ins.offset); + goto stop_trans; + } + ret = insert_reserved_file_extent(trans, inode, cur_offset, ins.objectid, ins.offset, ins.offset, - ins.offset, locked_end, - 0, 0, 0, + ins.offset, 0, 0, 0, BTRFS_FILE_EXTENT_PREALLOC); BUG_ON(ret); btrfs_drop_extent_cache(inode, cur_offset, cur_offset + ins.offset -1, 0); + num_bytes -= ins.offset; cur_offset += ins.offset; alloc_hint = ins.objectid + ins.offset; - } -out: - if (cur_offset > start) { + inode->i_ctime = CURRENT_TIME; BTRFS_I(inode)->flags |= BTRFS_INODE_PREALLOC; if (!(mode & FALLOC_FL_KEEP_SIZE) && - cur_offset > i_size_read(inode)) - btrfs_i_size_write(inode, cur_offset); + (actual_len > inode->i_size) && + (cur_offset > inode->i_size)) { + + if (cur_offset > actual_len) + i_size = actual_len; + else + i_size = cur_offset; + i_size_write(inode, i_size); + btrfs_ordered_update_i_size(inode, i_size, NULL); + } + ret = btrfs_update_inode(trans, root, inode); BUG_ON(ret); + + btrfs_end_transaction(trans, root); + btrfs_unreserve_metadata_space(root, 3); } + return ret; +stop_trans: + btrfs_end_transaction(trans, root); return ret; + } static long btrfs_fallocate(struct inode *inode, int mode, loff_t offset, loff_t len) { + struct extent_state *cached_state = NULL; u64 cur_offset; u64 last_byte; u64 alloc_start; @@ -5407,8 +5867,6 @@ static long btrfs_fallocate(struct inode *inode, int mode, u64 locked_end; u64 mask = BTRFS_I(inode)->root->sectorsize - 1; struct extent_map *em; - struct btrfs_trans_handle *trans; - struct btrfs_root *root; int ret; alloc_start = offset & ~mask; @@ -5427,9 +5885,7 @@ static long btrfs_fallocate(struct inode *inode, int mode, goto out; } - root = BTRFS_I(inode)->root; - - ret = btrfs_check_data_free_space(root, inode, + ret = btrfs_check_data_free_space(BTRFS_I(inode)->root, inode, alloc_end - alloc_start); if (ret) goto out; @@ -5438,27 +5894,20 @@ static long btrfs_fallocate(struct inode *inode, int mode, while (1) { struct btrfs_ordered_extent *ordered; - trans = btrfs_start_transaction(BTRFS_I(inode)->root, 1); - if (!trans) { - ret = -EIO; - goto out_free; - } - /* the extent lock is ordered inside the running * transaction */ - lock_extent(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, - GFP_NOFS); + lock_extent_bits(&BTRFS_I(inode)->io_tree, alloc_start, + locked_end, 0, &cached_state, GFP_NOFS); ordered = btrfs_lookup_first_ordered_extent(inode, alloc_end - 1); if (ordered && ordered->file_offset + ordered->len > alloc_start && ordered->file_offset < alloc_end) { btrfs_put_ordered_extent(ordered); - unlock_extent(&BTRFS_I(inode)->io_tree, - alloc_start, locked_end, GFP_NOFS); - btrfs_end_transaction(trans, BTRFS_I(inode)->root); - + unlock_extent_cached(&BTRFS_I(inode)->io_tree, + alloc_start, locked_end, + &cached_state, GFP_NOFS); /* * we can't wait on the range with the transaction * running or with the extent lock held @@ -5479,10 +5928,12 @@ static long btrfs_fallocate(struct inode *inode, int mode, BUG_ON(IS_ERR(em) || !em); last_byte = min(extent_map_end(em), alloc_end); last_byte = (last_byte + mask) & ~mask; - if (em->block_start == EXTENT_MAP_HOLE) { - ret = prealloc_file_range(trans, inode, cur_offset, - last_byte, locked_end + 1, - alloc_hint, mode); + if (em->block_start == EXTENT_MAP_HOLE || + (cur_offset >= inode->i_size && + !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) { + ret = prealloc_file_range(inode, + cur_offset, last_byte, + alloc_hint, mode, offset+len); if (ret < 0) { free_extent_map(em); break; @@ -5498,12 +5949,11 @@ static long btrfs_fallocate(struct inode *inode, int mode, break; } } - unlock_extent(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, - GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, + &cached_state, GFP_NOFS); - btrfs_end_transaction(trans, BTRFS_I(inode)->root); -out_free: - btrfs_free_reserved_data_space(root, inode, alloc_end - alloc_start); + btrfs_free_reserved_data_space(BTRFS_I(inode)->root, inode, + alloc_end - alloc_start); out: mutex_unlock(&inode->i_mutex); return ret; @@ -5544,7 +5994,7 @@ static const struct inode_operations btrfs_dir_ro_inode_operations = { .permission = btrfs_permission, }; -static struct file_operations btrfs_dir_file_operations = { +static const struct file_operations btrfs_dir_file_operations = { .llseek = generic_file_llseek, .read = generic_read_dir, .readdir = btrfs_real_readdir, @@ -5566,6 +6016,8 @@ static struct extent_io_ops btrfs_extent_io_ops = { .readpage_io_failed_hook = btrfs_io_failed_hook, .set_bit_hook = btrfs_set_bit_hook, .clear_bit_hook = btrfs_clear_bit_hook, + .merge_extent_hook = btrfs_merge_extent_hook, + .split_extent_hook = btrfs_split_extent_hook, }; /* @@ -5632,6 +6084,6 @@ static const struct inode_operations btrfs_symlink_inode_operations = { .removexattr = btrfs_removexattr, }; -struct dentry_operations btrfs_dentry_operations = { +const struct dentry_operations btrfs_dentry_operations = { .d_delete = btrfs_dentry_delete, }; diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index a8577a7f26ab..2845c6ceecd2 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -48,6 +48,7 @@ #include "print-tree.h" #include "volumes.h" #include "locking.h" +#include "ctree.h" /* Mask out flags that are inappropriate for the given type of inode. */ static inline __u32 btrfs_mask_flags(umode_t mode, __u32 flags) @@ -237,9 +238,14 @@ static noinline int create_subvol(struct btrfs_root *root, u64 objectid; u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID; u64 index = 0; - unsigned long nr = 1; - ret = btrfs_check_metadata_free_space(root); + /* + * 1 - inode item + * 2 - refs + * 1 - root item + * 2 - dir items + */ + ret = btrfs_reserve_metadata_space(root, 6); if (ret) return ret; @@ -284,7 +290,7 @@ static noinline int create_subvol(struct btrfs_root *root, btrfs_set_root_generation(&root_item, trans->transid); btrfs_set_root_level(&root_item, 0); btrfs_set_root_refs(&root_item, 1); - btrfs_set_root_used(&root_item, 0); + btrfs_set_root_used(&root_item, leaf->len); btrfs_set_root_last_snapshot(&root_item, 0); memset(&root_item.drop_progress, 0, sizeof(root_item.drop_progress)); @@ -336,39 +342,47 @@ static noinline int create_subvol(struct btrfs_root *root, d_instantiate(dentry, btrfs_lookup_dentry(dir, dentry)); fail: - nr = trans->blocks_used; err = btrfs_commit_transaction(trans, root); if (err && !ret) ret = err; + + btrfs_unreserve_metadata_space(root, 6); return ret; } static int create_snapshot(struct btrfs_root *root, struct dentry *dentry, char *name, int namelen) { + struct inode *inode; struct btrfs_pending_snapshot *pending_snapshot; struct btrfs_trans_handle *trans; - int ret = 0; - int err; - unsigned long nr = 0; + int ret; if (!root->ref_cows) return -EINVAL; - ret = btrfs_check_metadata_free_space(root); + /* + * 1 - inode item + * 2 - refs + * 1 - root item + * 2 - dir items + */ + ret = btrfs_reserve_metadata_space(root, 6); if (ret) - goto fail_unlock; + goto fail; pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS); if (!pending_snapshot) { ret = -ENOMEM; - goto fail_unlock; + btrfs_unreserve_metadata_space(root, 6); + goto fail; } pending_snapshot->name = kmalloc(namelen + 1, GFP_NOFS); if (!pending_snapshot->name) { ret = -ENOMEM; kfree(pending_snapshot); - goto fail_unlock; + btrfs_unreserve_metadata_space(root, 6); + goto fail; } memcpy(pending_snapshot->name, name, namelen); pending_snapshot->name[namelen] = '\0'; @@ -378,10 +392,19 @@ static int create_snapshot(struct btrfs_root *root, struct dentry *dentry, pending_snapshot->root = root; list_add(&pending_snapshot->list, &trans->transaction->pending_snapshots); - err = btrfs_commit_transaction(trans, root); + ret = btrfs_commit_transaction(trans, root); + BUG_ON(ret); + btrfs_unreserve_metadata_space(root, 6); -fail_unlock: - btrfs_btree_balance_dirty(root, nr); + inode = btrfs_lookup_dentry(dentry->d_parent->d_inode, dentry); + if (IS_ERR(inode)) { + ret = PTR_ERR(inode); + goto fail; + } + BUG_ON(!inode); + d_instantiate(dentry, inode); + ret = 0; +fail: return ret; } @@ -452,7 +475,79 @@ out_unlock: return error; } -static int btrfs_defrag_file(struct file *file) +static int should_defrag_range(struct inode *inode, u64 start, u64 len, + int thresh, u64 *last_len, u64 *skip, + u64 *defrag_end) +{ + struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; + struct extent_map *em = NULL; + struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + int ret = 1; + + + if (thresh == 0) + thresh = 256 * 1024; + + /* + * make sure that once we start defragging and extent, we keep on + * defragging it + */ + if (start < *defrag_end) + return 1; + + *skip = 0; + + /* + * hopefully we have this extent in the tree already, try without + * the full extent lock + */ + read_lock(&em_tree->lock); + em = lookup_extent_mapping(em_tree, start, len); + read_unlock(&em_tree->lock); + + if (!em) { + /* get the big lock and read metadata off disk */ + lock_extent(io_tree, start, start + len - 1, GFP_NOFS); + em = btrfs_get_extent(inode, NULL, 0, start, len, 0); + unlock_extent(io_tree, start, start + len - 1, GFP_NOFS); + + if (!em) + return 0; + } + + /* this will cover holes, and inline extents */ + if (em->block_start >= EXTENT_MAP_LAST_BYTE) + ret = 0; + + /* + * we hit a real extent, if it is big don't bother defragging it again + */ + if ((*last_len == 0 || *last_len >= thresh) && em->len >= thresh) + ret = 0; + + /* + * last_len ends up being a counter of how many bytes we've defragged. + * every time we choose not to defrag an extent, we reset *last_len + * so that the next tiny extent will force a defrag. + * + * The end result of this is that tiny extents before a single big + * extent will force at least part of that big extent to be defragged. + */ + if (ret) { + *last_len += len; + *defrag_end = extent_map_end(em); + } else { + *last_len = 0; + *skip = extent_map_end(em); + *defrag_end = 0; + } + + free_extent_map(em); + return ret; +} + +static int btrfs_defrag_file(struct file *file, + struct btrfs_ioctl_defrag_range_args *range) { struct inode *inode = fdentry(file)->d_inode; struct btrfs_root *root = BTRFS_I(inode)->root; @@ -464,37 +559,96 @@ static int btrfs_defrag_file(struct file *file) unsigned long total_read = 0; u64 page_start; u64 page_end; + u64 last_len = 0; + u64 skip = 0; + u64 defrag_end = 0; unsigned long i; int ret; - ret = btrfs_check_data_free_space(root, inode, inode->i_size); - if (ret) - return -ENOSPC; + if (inode->i_size == 0) + return 0; + + if (range->start + range->len > range->start) { + last_index = min_t(u64, inode->i_size - 1, + range->start + range->len - 1) >> PAGE_CACHE_SHIFT; + } else { + last_index = (inode->i_size - 1) >> PAGE_CACHE_SHIFT; + } + + i = range->start >> PAGE_CACHE_SHIFT; + while (i <= last_index) { + if (!should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT, + PAGE_CACHE_SIZE, + range->extent_thresh, + &last_len, &skip, + &defrag_end)) { + unsigned long next; + /* + * the should_defrag function tells us how much to skip + * bump our counter by the suggested amount + */ + next = (skip + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; + i = max(i + 1, next); + continue; + } - mutex_lock(&inode->i_mutex); - last_index = inode->i_size >> PAGE_CACHE_SHIFT; - for (i = 0; i <= last_index; i++) { if (total_read % ra_pages == 0) { btrfs_force_ra(inode->i_mapping, &file->f_ra, file, i, min(last_index, i + ra_pages - 1)); } total_read++; + mutex_lock(&inode->i_mutex); + if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS) + BTRFS_I(inode)->force_compress = 1; + + ret = btrfs_check_data_free_space(root, inode, PAGE_CACHE_SIZE); + if (ret) { + ret = -ENOSPC; + break; + } + + ret = btrfs_reserve_metadata_for_delalloc(root, inode, 1); + if (ret) { + btrfs_free_reserved_data_space(root, inode, + PAGE_CACHE_SIZE); + ret = -ENOSPC; + break; + } again: + if (inode->i_size == 0 || + i > ((inode->i_size - 1) >> PAGE_CACHE_SHIFT)) { + ret = 0; + goto err_reservations; + } + page = grab_cache_page(inode->i_mapping, i); if (!page) - goto out_unlock; + goto err_reservations; + if (!PageUptodate(page)) { btrfs_readpage(NULL, page); lock_page(page); if (!PageUptodate(page)) { unlock_page(page); page_cache_release(page); - goto out_unlock; + goto err_reservations; } } + if (page->mapping != inode->i_mapping) { + unlock_page(page); + page_cache_release(page); + goto again; + } + wait_on_page_writeback(page); + if (PageDirty(page)) { + btrfs_free_reserved_data_space(root, inode, + PAGE_CACHE_SIZE); + goto loop_unlock; + } + page_start = (u64)page->index << PAGE_CACHE_SHIFT; page_end = page_start + PAGE_CACHE_SIZE - 1; lock_extent(io_tree, page_start, page_end, GFP_NOFS); @@ -515,18 +669,54 @@ again: * page if it is dirtied again later */ clear_page_dirty_for_io(page); + clear_extent_bits(&BTRFS_I(inode)->io_tree, page_start, + page_end, EXTENT_DIRTY | EXTENT_DELALLOC | + EXTENT_DO_ACCOUNTING, GFP_NOFS); - btrfs_set_extent_delalloc(inode, page_start, page_end); + btrfs_set_extent_delalloc(inode, page_start, page_end, NULL); + ClearPageChecked(page); set_page_dirty(page); unlock_extent(io_tree, page_start, page_end, GFP_NOFS); + +loop_unlock: unlock_page(page); page_cache_release(page); + mutex_unlock(&inode->i_mutex); + + btrfs_unreserve_metadata_for_delalloc(root, inode, 1); balance_dirty_pages_ratelimited_nr(inode->i_mapping, 1); + i++; + } + + if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO)) + filemap_flush(inode->i_mapping); + + if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { + /* the filemap_flush will queue IO into the worker threads, but + * we have to make sure the IO is actually started and that + * ordered extents get created before we return + */ + atomic_inc(&root->fs_info->async_submit_draining); + while (atomic_read(&root->fs_info->nr_async_submits) || + atomic_read(&root->fs_info->async_delalloc_pages)) { + wait_event(root->fs_info->async_submit_wait, + (atomic_read(&root->fs_info->nr_async_submits) == 0 && + atomic_read(&root->fs_info->async_delalloc_pages) == 0)); + } + atomic_dec(&root->fs_info->async_submit_draining); + + mutex_lock(&inode->i_mutex); + BTRFS_I(inode)->force_compress = 0; + mutex_unlock(&inode->i_mutex); } -out_unlock: - mutex_unlock(&inode->i_mutex); return 0; + +err_reservations: + mutex_unlock(&inode->i_mutex); + btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE); + btrfs_unreserve_metadata_for_delalloc(root, inode, 1); + return ret; } static noinline int btrfs_ioctl_resize(struct btrfs_root *root, @@ -586,7 +776,7 @@ static noinline int btrfs_ioctl_resize(struct btrfs_root *root, mod = 1; sizestr++; } - new_size = btrfs_parse_size(sizestr); + new_size = memparse(sizestr, NULL); if (new_size == 0) { ret = -EINVAL; goto out_unlock; @@ -721,6 +911,327 @@ out: return ret; } +static noinline int key_in_sk(struct btrfs_key *key, + struct btrfs_ioctl_search_key *sk) +{ + struct btrfs_key test; + int ret; + + test.objectid = sk->min_objectid; + test.type = sk->min_type; + test.offset = sk->min_offset; + + ret = btrfs_comp_cpu_keys(key, &test); + if (ret < 0) + return 0; + + test.objectid = sk->max_objectid; + test.type = sk->max_type; + test.offset = sk->max_offset; + + ret = btrfs_comp_cpu_keys(key, &test); + if (ret > 0) + return 0; + return 1; +} + +static noinline int copy_to_sk(struct btrfs_root *root, + struct btrfs_path *path, + struct btrfs_key *key, + struct btrfs_ioctl_search_key *sk, + char *buf, + unsigned long *sk_offset, + int *num_found) +{ + u64 found_transid; + struct extent_buffer *leaf; + struct btrfs_ioctl_search_header sh; + unsigned long item_off; + unsigned long item_len; + int nritems; + int i; + int slot; + int found = 0; + int ret = 0; + + leaf = path->nodes[0]; + slot = path->slots[0]; + nritems = btrfs_header_nritems(leaf); + + if (btrfs_header_generation(leaf) > sk->max_transid) { + i = nritems; + goto advance_key; + } + found_transid = btrfs_header_generation(leaf); + + for (i = slot; i < nritems; i++) { + item_off = btrfs_item_ptr_offset(leaf, i); + item_len = btrfs_item_size_nr(leaf, i); + + if (item_len > BTRFS_SEARCH_ARGS_BUFSIZE) + item_len = 0; + + if (sizeof(sh) + item_len + *sk_offset > + BTRFS_SEARCH_ARGS_BUFSIZE) { + ret = 1; + goto overflow; + } + + btrfs_item_key_to_cpu(leaf, key, i); + if (!key_in_sk(key, sk)) + continue; + + sh.objectid = key->objectid; + sh.offset = key->offset; + sh.type = key->type; + sh.len = item_len; + sh.transid = found_transid; + + /* copy search result header */ + memcpy(buf + *sk_offset, &sh, sizeof(sh)); + *sk_offset += sizeof(sh); + + if (item_len) { + char *p = buf + *sk_offset; + /* copy the item */ + read_extent_buffer(leaf, p, + item_off, item_len); + *sk_offset += item_len; + } + found++; + + if (*num_found >= sk->nr_items) + break; + } +advance_key: + ret = 0; + if (key->offset < (u64)-1 && key->offset < sk->max_offset) + key->offset++; + else if (key->type < (u8)-1 && key->type < sk->max_type) { + key->offset = 0; + key->type++; + } else if (key->objectid < (u64)-1 && key->objectid < sk->max_objectid) { + key->offset = 0; + key->type = 0; + key->objectid++; + } else + ret = 1; +overflow: + *num_found += found; + return ret; +} + +static noinline int search_ioctl(struct inode *inode, + struct btrfs_ioctl_search_args *args) +{ + struct btrfs_root *root; + struct btrfs_key key; + struct btrfs_key max_key; + struct btrfs_path *path; + struct btrfs_ioctl_search_key *sk = &args->key; + struct btrfs_fs_info *info = BTRFS_I(inode)->root->fs_info; + int ret; + int num_found = 0; + unsigned long sk_offset = 0; + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + + if (sk->tree_id == 0) { + /* search the root of the inode that was passed */ + root = BTRFS_I(inode)->root; + } else { + key.objectid = sk->tree_id; + key.type = BTRFS_ROOT_ITEM_KEY; + key.offset = (u64)-1; + root = btrfs_read_fs_root_no_name(info, &key); + if (IS_ERR(root)) { + printk(KERN_ERR "could not find root %llu\n", + sk->tree_id); + btrfs_free_path(path); + return -ENOENT; + } + } + + key.objectid = sk->min_objectid; + key.type = sk->min_type; + key.offset = sk->min_offset; + + max_key.objectid = sk->max_objectid; + max_key.type = sk->max_type; + max_key.offset = sk->max_offset; + + path->keep_locks = 1; + + while(1) { + ret = btrfs_search_forward(root, &key, &max_key, path, 0, + sk->min_transid); + if (ret != 0) { + if (ret > 0) + ret = 0; + goto err; + } + ret = copy_to_sk(root, path, &key, sk, args->buf, + &sk_offset, &num_found); + btrfs_release_path(root, path); + if (ret || num_found >= sk->nr_items) + break; + + } + ret = 0; +err: + sk->nr_items = num_found; + btrfs_free_path(path); + return ret; +} + +static noinline int btrfs_ioctl_tree_search(struct file *file, + void __user *argp) +{ + struct btrfs_ioctl_search_args *args; + struct inode *inode; + int ret; + + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + + args = kmalloc(sizeof(*args), GFP_KERNEL); + if (!args) + return -ENOMEM; + + if (copy_from_user(args, argp, sizeof(*args))) { + kfree(args); + return -EFAULT; + } + inode = fdentry(file)->d_inode; + ret = search_ioctl(inode, args); + if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) + ret = -EFAULT; + kfree(args); + return ret; +} + +/* + * Search INODE_REFs to identify path name of 'dirid' directory + * in a 'tree_id' tree. and sets path name to 'name'. + */ +static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, + u64 tree_id, u64 dirid, char *name) +{ + struct btrfs_root *root; + struct btrfs_key key; + char *ptr; + int ret = -1; + int slot; + int len; + int total_len = 0; + struct btrfs_inode_ref *iref; + struct extent_buffer *l; + struct btrfs_path *path; + + if (dirid == BTRFS_FIRST_FREE_OBJECTID) { + name[0]='\0'; + return 0; + } + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + + ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX]; + + key.objectid = tree_id; + key.type = BTRFS_ROOT_ITEM_KEY; + key.offset = (u64)-1; + root = btrfs_read_fs_root_no_name(info, &key); + if (IS_ERR(root)) { + printk(KERN_ERR "could not find root %llu\n", tree_id); + ret = -ENOENT; + goto out; + } + + key.objectid = dirid; + key.type = BTRFS_INODE_REF_KEY; + key.offset = (u64)-1; + + while(1) { + ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); + if (ret < 0) + goto out; + + l = path->nodes[0]; + slot = path->slots[0]; + if (ret > 0 && slot > 0) + slot--; + btrfs_item_key_to_cpu(l, &key, slot); + + if (ret > 0 && (key.objectid != dirid || + key.type != BTRFS_INODE_REF_KEY)) { + ret = -ENOENT; + goto out; + } + + iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref); + len = btrfs_inode_ref_name_len(l, iref); + ptr -= len + 1; + total_len += len + 1; + if (ptr < name) + goto out; + + *(ptr + len) = '/'; + read_extent_buffer(l, ptr,(unsigned long)(iref + 1), len); + + if (key.offset == BTRFS_FIRST_FREE_OBJECTID) + break; + + btrfs_release_path(root, path); + key.objectid = key.offset; + key.offset = (u64)-1; + dirid = key.objectid; + + } + if (ptr < name) + goto out; + memcpy(name, ptr, total_len); + name[total_len]='\0'; + ret = 0; +out: + btrfs_free_path(path); + return ret; +} + +static noinline int btrfs_ioctl_ino_lookup(struct file *file, + void __user *argp) +{ + struct btrfs_ioctl_ino_lookup_args *args; + struct inode *inode; + int ret; + + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + + args = kmalloc(sizeof(*args), GFP_KERNEL); + if (copy_from_user(args, argp, sizeof(*args))) { + kfree(args); + return -EFAULT; + } + inode = fdentry(file)->d_inode; + + if (args->treeid == 0) + args->treeid = BTRFS_I(inode)->root->root_key.objectid; + + ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info, + args->treeid, args->objectid, + args->name); + + if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) + ret = -EFAULT; + + kfree(args); + return ret; +} + static noinline int btrfs_ioctl_snap_destroy(struct file *file, void __user *arg) { @@ -813,6 +1324,7 @@ out_up_write: out_unlock: mutex_unlock(&inode->i_mutex); if (!err) { + shrink_dcache_sb(root->fs_info->sb); btrfs_invalidate_inodes(dest); d_delete(dentry); } @@ -826,10 +1338,11 @@ out: return err; } -static int btrfs_ioctl_defrag(struct file *file) +static int btrfs_ioctl_defrag(struct file *file, void __user *argp) { struct inode *inode = fdentry(file)->d_inode; struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_ioctl_defrag_range_args *range; int ret; ret = mnt_want_write(file->f_path.mnt); @@ -850,7 +1363,30 @@ static int btrfs_ioctl_defrag(struct file *file) ret = -EINVAL; goto out; } - btrfs_defrag_file(file); + + range = kzalloc(sizeof(*range), GFP_KERNEL); + if (!range) { + ret = -ENOMEM; + goto out; + } + + if (argp) { + if (copy_from_user(range, argp, + sizeof(*range))) { + ret = -EFAULT; + kfree(range); + } + /* compression requires us to start the IO */ + if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { + range->flags |= BTRFS_DEFRAG_RANGE_START_IO; + range->extent_thresh = (u32)-1; + } + } else { + /* the rest are all set to zero by kzalloc */ + range->len = (u64)-1; + } + btrfs_defrag_file(file, range); + kfree(range); break; } out: @@ -1009,8 +1545,7 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd, BUG_ON(!trans); /* punch hole in destination first */ - btrfs_drop_extents(trans, root, inode, off, off + len, - off + len, 0, &hint_byte, 1); + btrfs_drop_extents(trans, inode, off, off + len, &hint_byte, 1); /* clone data */ key.objectid = src->i_ino; @@ -1105,8 +1640,10 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd, datao += off - key.offset; datal -= off - key.offset; } - if (key.offset + datao + datal > off + len) - datal = off + len - key.offset - datao; + + if (key.offset + datal > off + len) + datal = off + len - key.offset; + /* disko == 0 means it's a hole */ if (!disko) datao = 0; @@ -1215,15 +1752,15 @@ static long btrfs_ioctl_trans_start(struct file *file) struct inode *inode = fdentry(file)->d_inode; struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_trans_handle *trans; - int ret = 0; + int ret; + ret = -EPERM; if (!capable(CAP_SYS_ADMIN)) - return -EPERM; + goto out; - if (file->private_data) { - ret = -EINPROGRESS; + ret = -EINPROGRESS; + if (file->private_data) goto out; - } ret = mnt_want_write(file->f_path.mnt); if (ret) @@ -1233,13 +1770,171 @@ static long btrfs_ioctl_trans_start(struct file *file) root->fs_info->open_ioctl_trans++; mutex_unlock(&root->fs_info->trans_mutex); + ret = -ENOMEM; trans = btrfs_start_ioctl_transaction(root, 0); - if (trans) - file->private_data = trans; - else - ret = -ENOMEM; - /*printk(KERN_INFO "btrfs_ioctl_trans_start on %p\n", file);*/ + if (!trans) + goto out_drop; + + file->private_data = trans; + return 0; + +out_drop: + mutex_lock(&root->fs_info->trans_mutex); + root->fs_info->open_ioctl_trans--; + mutex_unlock(&root->fs_info->trans_mutex); + mnt_drop_write(file->f_path.mnt); +out: + return ret; +} + +static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp) +{ + struct inode *inode = fdentry(file)->d_inode; + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_root *new_root; + struct btrfs_dir_item *di; + struct btrfs_trans_handle *trans; + struct btrfs_path *path; + struct btrfs_key location; + struct btrfs_disk_key disk_key; + struct btrfs_super_block *disk_super; + u64 features; + u64 objectid = 0; + u64 dir_id; + + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + + if (copy_from_user(&objectid, argp, sizeof(objectid))) + return -EFAULT; + + if (!objectid) + objectid = root->root_key.objectid; + + location.objectid = objectid; + location.type = BTRFS_ROOT_ITEM_KEY; + location.offset = (u64)-1; + + new_root = btrfs_read_fs_root_no_name(root->fs_info, &location); + if (IS_ERR(new_root)) + return PTR_ERR(new_root); + + if (btrfs_root_refs(&new_root->root_item) == 0) + return -ENOENT; + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + path->leave_spinning = 1; + + trans = btrfs_start_transaction(root, 1); + if (!trans) { + btrfs_free_path(path); + return -ENOMEM; + } + + dir_id = btrfs_super_root_dir(&root->fs_info->super_copy); + di = btrfs_lookup_dir_item(trans, root->fs_info->tree_root, path, + dir_id, "default", 7, 1); + if (!di) { + btrfs_free_path(path); + btrfs_end_transaction(trans, root); + printk(KERN_ERR "Umm, you don't have the default dir item, " + "this isn't going to work\n"); + return -ENOENT; + } + + btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key); + btrfs_set_dir_item_key(path->nodes[0], di, &disk_key); + btrfs_mark_buffer_dirty(path->nodes[0]); + btrfs_free_path(path); + + disk_super = &root->fs_info->super_copy; + features = btrfs_super_incompat_flags(disk_super); + if (!(features & BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL)) { + features |= BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL; + btrfs_set_super_incompat_flags(disk_super, features); + } + btrfs_end_transaction(trans, root); + + return 0; +} + +long btrfs_ioctl_space_info(struct btrfs_root *root, void __user *arg) +{ + struct btrfs_ioctl_space_args space_args; + struct btrfs_ioctl_space_info space; + struct btrfs_ioctl_space_info *dest; + struct btrfs_ioctl_space_info *dest_orig; + struct btrfs_ioctl_space_info *user_dest; + struct btrfs_space_info *info; + int alloc_size; + int ret = 0; + int slot_count = 0; + + if (copy_from_user(&space_args, + (struct btrfs_ioctl_space_args __user *)arg, + sizeof(space_args))) + return -EFAULT; + + /* first we count slots */ + rcu_read_lock(); + list_for_each_entry_rcu(info, &root->fs_info->space_info, list) + slot_count++; + rcu_read_unlock(); + + /* space_slots == 0 means they are asking for a count */ + if (space_args.space_slots == 0) { + space_args.total_spaces = slot_count; + goto out; + } + alloc_size = sizeof(*dest) * slot_count; + /* we generally have at most 6 or so space infos, one for each raid + * level. So, a whole page should be more than enough for everyone + */ + if (alloc_size > PAGE_CACHE_SIZE) + return -ENOMEM; + + space_args.total_spaces = 0; + dest = kmalloc(alloc_size, GFP_NOFS); + if (!dest) + return -ENOMEM; + dest_orig = dest; + + /* now we have a buffer to copy into */ + rcu_read_lock(); + list_for_each_entry_rcu(info, &root->fs_info->space_info, list) { + /* make sure we don't copy more than we allocated + * in our buffer + */ + if (slot_count == 0) + break; + slot_count--; + + /* make sure userland has enough room in their buffer */ + if (space_args.total_spaces >= space_args.space_slots) + break; + + space.flags = info->flags; + space.total_bytes = info->total_bytes; + space.used_bytes = info->bytes_used; + memcpy(dest, &space, sizeof(space)); + dest++; + space_args.total_spaces++; + } + rcu_read_unlock(); + + user_dest = (struct btrfs_ioctl_space_info *) + (arg + sizeof(struct btrfs_ioctl_space_args)); + + if (copy_to_user(user_dest, dest_orig, alloc_size)) + ret = -EFAULT; + + kfree(dest_orig); out: + if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args))) + ret = -EFAULT; + return ret; } @@ -1254,24 +1949,20 @@ long btrfs_ioctl_trans_end(struct file *file) struct inode *inode = fdentry(file)->d_inode; struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_trans_handle *trans; - int ret = 0; trans = file->private_data; - if (!trans) { - ret = -EINVAL; - goto out; - } - btrfs_end_transaction(trans, root); + if (!trans) + return -EINVAL; file->private_data = NULL; + btrfs_end_transaction(trans, root); + mutex_lock(&root->fs_info->trans_mutex); root->fs_info->open_ioctl_trans--; mutex_unlock(&root->fs_info->trans_mutex); mnt_drop_write(file->f_path.mnt); - -out: - return ret; + return 0; } long btrfs_ioctl(struct file *file, unsigned int @@ -1293,8 +1984,12 @@ long btrfs_ioctl(struct file *file, unsigned int return btrfs_ioctl_snap_create(file, argp, 1); case BTRFS_IOC_SNAP_DESTROY: return btrfs_ioctl_snap_destroy(file, argp); + case BTRFS_IOC_DEFAULT_SUBVOL: + return btrfs_ioctl_default_subvol(file, argp); case BTRFS_IOC_DEFRAG: - return btrfs_ioctl_defrag(file); + return btrfs_ioctl_defrag(file, NULL); + case BTRFS_IOC_DEFRAG_RANGE: + return btrfs_ioctl_defrag(file, argp); case BTRFS_IOC_RESIZE: return btrfs_ioctl_resize(root, argp); case BTRFS_IOC_ADD_DEV: @@ -1311,6 +2006,12 @@ long btrfs_ioctl(struct file *file, unsigned int return btrfs_ioctl_trans_start(file); case BTRFS_IOC_TRANS_END: return btrfs_ioctl_trans_end(file); + case BTRFS_IOC_TREE_SEARCH: + return btrfs_ioctl_tree_search(file, argp); + case BTRFS_IOC_INO_LOOKUP: + return btrfs_ioctl_ino_lookup(file, argp); + case BTRFS_IOC_SPACE_INFO: + return btrfs_ioctl_space_info(root, argp); case BTRFS_IOC_SYNC: btrfs_sync_fs(file->f_dentry->d_sb, 1); return 0; diff --git a/fs/btrfs/ioctl.h b/fs/btrfs/ioctl.h index bc49914475eb..424694aa517f 100644 --- a/fs/btrfs/ioctl.h +++ b/fs/btrfs/ioctl.h @@ -30,12 +30,114 @@ struct btrfs_ioctl_vol_args { char name[BTRFS_PATH_NAME_MAX + 1]; }; +#define BTRFS_INO_LOOKUP_PATH_MAX 4080 +struct btrfs_ioctl_ino_lookup_args { + __u64 treeid; + __u64 objectid; + char name[BTRFS_INO_LOOKUP_PATH_MAX]; +}; + +struct btrfs_ioctl_search_key { + /* which root are we searching. 0 is the tree of tree roots */ + __u64 tree_id; + + /* keys returned will be >= min and <= max */ + __u64 min_objectid; + __u64 max_objectid; + + /* keys returned will be >= min and <= max */ + __u64 min_offset; + __u64 max_offset; + + /* max and min transids to search for */ + __u64 min_transid; + __u64 max_transid; + + /* keys returned will be >= min and <= max */ + __u32 min_type; + __u32 max_type; + + /* + * how many items did userland ask for, and how many are we + * returning + */ + __u32 nr_items; + + /* align to 64 bits */ + __u32 unused; + + /* some extra for later */ + __u64 unused1; + __u64 unused2; + __u64 unused3; + __u64 unused4; +}; + +struct btrfs_ioctl_search_header { + __u64 transid; + __u64 objectid; + __u64 offset; + __u32 type; + __u32 len; +}; + +#define BTRFS_SEARCH_ARGS_BUFSIZE (4096 - sizeof(struct btrfs_ioctl_search_key)) +/* + * the buf is an array of search headers where + * each header is followed by the actual item + * the type field is expanded to 32 bits for alignment + */ +struct btrfs_ioctl_search_args { + struct btrfs_ioctl_search_key key; + char buf[BTRFS_SEARCH_ARGS_BUFSIZE]; +}; + struct btrfs_ioctl_clone_range_args { __s64 src_fd; __u64 src_offset, src_length; __u64 dest_offset; }; +/* flags for the defrag range ioctl */ +#define BTRFS_DEFRAG_RANGE_COMPRESS 1 +#define BTRFS_DEFRAG_RANGE_START_IO 2 + +struct btrfs_ioctl_defrag_range_args { + /* start of the defrag operation */ + __u64 start; + + /* number of bytes to defrag, use (u64)-1 to say all */ + __u64 len; + + /* + * flags for the operation, which can include turning + * on compression for this one defrag + */ + __u64 flags; + + /* + * any extent bigger than this will be considered + * already defragged. Use 0 to take the kernel default + * Use 1 to say every single extent must be rewritten + */ + __u32 extent_thresh; + + /* spare for later */ + __u32 unused[5]; +}; + +struct btrfs_ioctl_space_info { + __u64 flags; + __u64 total_bytes; + __u64 used_bytes; +}; + +struct btrfs_ioctl_space_args { + __u64 space_slots; + __u64 total_spaces; + struct btrfs_ioctl_space_info spaces[0]; +}; + #define BTRFS_IOC_SNAP_CREATE _IOW(BTRFS_IOCTL_MAGIC, 1, \ struct btrfs_ioctl_vol_args) #define BTRFS_IOC_DEFRAG _IOW(BTRFS_IOCTL_MAGIC, 2, \ @@ -67,4 +169,13 @@ struct btrfs_ioctl_clone_range_args { struct btrfs_ioctl_vol_args) #define BTRFS_IOC_SNAP_DESTROY _IOW(BTRFS_IOCTL_MAGIC, 15, \ struct btrfs_ioctl_vol_args) +#define BTRFS_IOC_DEFRAG_RANGE _IOW(BTRFS_IOCTL_MAGIC, 16, \ + struct btrfs_ioctl_defrag_range_args) +#define BTRFS_IOC_TREE_SEARCH _IOWR(BTRFS_IOCTL_MAGIC, 17, \ + struct btrfs_ioctl_search_args) +#define BTRFS_IOC_INO_LOOKUP _IOWR(BTRFS_IOCTL_MAGIC, 18, \ + struct btrfs_ioctl_ino_lookup_args) +#define BTRFS_IOC_DEFAULT_SUBVOL _IOW(BTRFS_IOCTL_MAGIC, 19, u64) +#define BTRFS_IOC_SPACE_INFO _IOWR(BTRFS_IOCTL_MAGIC, 20, \ + struct btrfs_ioctl_space_args) #endif diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c index b5d6d24726b0..a8ffecd0b491 100644 --- a/fs/btrfs/ordered-data.c +++ b/fs/btrfs/ordered-data.c @@ -174,7 +174,6 @@ int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset, if (!entry) return -ENOMEM; - mutex_lock(&tree->mutex); entry->file_offset = file_offset; entry->start = start; entry->len = len; @@ -190,16 +189,17 @@ int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset, INIT_LIST_HEAD(&entry->list); INIT_LIST_HEAD(&entry->root_extent_list); + spin_lock(&tree->lock); node = tree_insert(&tree->tree, file_offset, &entry->rb_node); BUG_ON(node); + spin_unlock(&tree->lock); spin_lock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock); list_add_tail(&entry->root_extent_list, &BTRFS_I(inode)->root->fs_info->ordered_extents); spin_unlock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock); - mutex_unlock(&tree->mutex); BUG_ON(node); return 0; } @@ -216,9 +216,9 @@ int btrfs_add_ordered_sum(struct inode *inode, struct btrfs_ordered_inode_tree *tree; tree = &BTRFS_I(inode)->ordered_tree; - mutex_lock(&tree->mutex); + spin_lock(&tree->lock); list_add_tail(&sum->list, &entry->list); - mutex_unlock(&tree->mutex); + spin_unlock(&tree->lock); return 0; } @@ -232,15 +232,16 @@ int btrfs_add_ordered_sum(struct inode *inode, * to make sure this function only returns 1 once for a given ordered extent. */ int btrfs_dec_test_ordered_pending(struct inode *inode, + struct btrfs_ordered_extent **cached, u64 file_offset, u64 io_size) { struct btrfs_ordered_inode_tree *tree; struct rb_node *node; - struct btrfs_ordered_extent *entry; + struct btrfs_ordered_extent *entry = NULL; int ret; tree = &BTRFS_I(inode)->ordered_tree; - mutex_lock(&tree->mutex); + spin_lock(&tree->lock); node = tree_search(tree, file_offset); if (!node) { ret = 1; @@ -264,7 +265,11 @@ int btrfs_dec_test_ordered_pending(struct inode *inode, else ret = 1; out: - mutex_unlock(&tree->mutex); + if (!ret && cached && entry) { + *cached = entry; + atomic_inc(&entry->refs); + } + spin_unlock(&tree->lock); return ret == 0; } @@ -291,21 +296,27 @@ int btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry) /* * remove an ordered extent from the tree. No references are dropped - * but, anyone waiting on this extent is woken up. + * and you must wake_up entry->wait. You must hold the tree lock + * while you call this function. */ -int btrfs_remove_ordered_extent(struct inode *inode, +static int __btrfs_remove_ordered_extent(struct inode *inode, struct btrfs_ordered_extent *entry) { struct btrfs_ordered_inode_tree *tree; struct rb_node *node; tree = &BTRFS_I(inode)->ordered_tree; - mutex_lock(&tree->mutex); node = &entry->rb_node; rb_erase(node, &tree->tree); tree->last = NULL; set_bit(BTRFS_ORDERED_COMPLETE, &entry->flags); + spin_lock(&BTRFS_I(inode)->accounting_lock); + BTRFS_I(inode)->outstanding_extents--; + spin_unlock(&BTRFS_I(inode)->accounting_lock); + btrfs_unreserve_metadata_for_delalloc(BTRFS_I(inode)->root, + inode, 1); + spin_lock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock); list_del_init(&entry->root_extent_list); @@ -320,16 +331,34 @@ int btrfs_remove_ordered_extent(struct inode *inode, } spin_unlock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock); - mutex_unlock(&tree->mutex); - wake_up(&entry->wait); return 0; } /* + * remove an ordered extent from the tree. No references are dropped + * but any waiters are woken. + */ +int btrfs_remove_ordered_extent(struct inode *inode, + struct btrfs_ordered_extent *entry) +{ + struct btrfs_ordered_inode_tree *tree; + int ret; + + tree = &BTRFS_I(inode)->ordered_tree; + spin_lock(&tree->lock); + ret = __btrfs_remove_ordered_extent(inode, entry); + spin_unlock(&tree->lock); + wake_up(&entry->wait); + + return ret; +} + +/* * wait for all the ordered extents in a root. This is done when balancing * space between drives. */ -int btrfs_wait_ordered_extents(struct btrfs_root *root, int nocow_only) +int btrfs_wait_ordered_extents(struct btrfs_root *root, + int nocow_only, int delay_iput) { struct list_head splice; struct list_head *cur; @@ -366,7 +395,10 @@ int btrfs_wait_ordered_extents(struct btrfs_root *root, int nocow_only) if (inode) { btrfs_start_ordered_extent(inode, ordered, 1); btrfs_put_ordered_extent(ordered); - iput(inode); + if (delay_iput) + btrfs_add_delayed_iput(inode); + else + iput(inode); } else { btrfs_put_ordered_extent(ordered); } @@ -424,7 +456,7 @@ again: btrfs_wait_ordered_range(inode, 0, (u64)-1); else filemap_flush(inode->i_mapping); - iput(inode); + btrfs_add_delayed_iput(inode); } cond_resched(); @@ -458,7 +490,7 @@ void btrfs_start_ordered_extent(struct inode *inode, * start IO on any dirty ones so the wait doesn't stall waiting * for pdflush to find them */ - btrfs_fdatawrite_range(inode->i_mapping, start, end, WB_SYNC_ALL); + filemap_fdatawrite_range(inode->i_mapping, start, end); if (wait) { wait_event(entry->wait, test_bit(BTRFS_ORDERED_COMPLETE, &entry->flags)); @@ -488,17 +520,15 @@ again: /* start IO across the range first to instantiate any delalloc * extents */ - btrfs_fdatawrite_range(inode->i_mapping, start, orig_end, WB_SYNC_ALL); + filemap_fdatawrite_range(inode->i_mapping, start, orig_end); /* The compression code will leave pages locked but return from * writepage without setting the page writeback. Starting again * with WB_SYNC_ALL will end up waiting for the IO to actually start. */ - btrfs_fdatawrite_range(inode->i_mapping, start, orig_end, WB_SYNC_ALL); + filemap_fdatawrite_range(inode->i_mapping, start, orig_end); - btrfs_wait_on_page_writeback_range(inode->i_mapping, - start >> PAGE_CACHE_SHIFT, - orig_end >> PAGE_CACHE_SHIFT); + filemap_fdatawait_range(inode->i_mapping, start, orig_end); end = orig_end; found = 0; @@ -542,7 +572,7 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode, struct btrfs_ordered_extent *entry = NULL; tree = &BTRFS_I(inode)->ordered_tree; - mutex_lock(&tree->mutex); + spin_lock(&tree->lock); node = tree_search(tree, file_offset); if (!node) goto out; @@ -553,7 +583,7 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode, if (entry) atomic_inc(&entry->refs); out: - mutex_unlock(&tree->mutex); + spin_unlock(&tree->lock); return entry; } @@ -569,7 +599,7 @@ btrfs_lookup_first_ordered_extent(struct inode *inode, u64 file_offset) struct btrfs_ordered_extent *entry = NULL; tree = &BTRFS_I(inode)->ordered_tree; - mutex_lock(&tree->mutex); + spin_lock(&tree->lock); node = tree_search(tree, file_offset); if (!node) goto out; @@ -577,7 +607,7 @@ btrfs_lookup_first_ordered_extent(struct inode *inode, u64 file_offset) entry = rb_entry(node, struct btrfs_ordered_extent, rb_node); atomic_inc(&entry->refs); out: - mutex_unlock(&tree->mutex); + spin_unlock(&tree->lock); return entry; } @@ -585,7 +615,7 @@ out: * After an extent is done, call this to conditionally update the on disk * i_size. i_size is updated to cover any fully written part of the file. */ -int btrfs_ordered_update_i_size(struct inode *inode, +int btrfs_ordered_update_i_size(struct inode *inode, u64 offset, struct btrfs_ordered_extent *ordered) { struct btrfs_ordered_inode_tree *tree = &BTRFS_I(inode)->ordered_tree; @@ -593,18 +623,32 @@ int btrfs_ordered_update_i_size(struct inode *inode, u64 disk_i_size; u64 new_i_size; u64 i_size_test; + u64 i_size = i_size_read(inode); struct rb_node *node; + struct rb_node *prev = NULL; struct btrfs_ordered_extent *test; + int ret = 1; + + if (ordered) + offset = entry_end(ordered); + else + offset = ALIGN(offset, BTRFS_I(inode)->root->sectorsize); - mutex_lock(&tree->mutex); + spin_lock(&tree->lock); disk_i_size = BTRFS_I(inode)->disk_i_size; + /* truncate file */ + if (disk_i_size > i_size) { + BTRFS_I(inode)->disk_i_size = i_size; + ret = 0; + goto out; + } + /* * if the disk i_size is already at the inode->i_size, or * this ordered extent is inside the disk i_size, we're done */ - if (disk_i_size >= inode->i_size || - ordered->file_offset + ordered->len <= disk_i_size) { + if (disk_i_size == i_size || offset <= disk_i_size) { goto out; } @@ -612,8 +656,7 @@ int btrfs_ordered_update_i_size(struct inode *inode, * we can't update the disk_isize if there are delalloc bytes * between disk_i_size and this ordered extent */ - if (test_range_bit(io_tree, disk_i_size, - ordered->file_offset + ordered->len - 1, + if (test_range_bit(io_tree, disk_i_size, offset - 1, EXTENT_DELALLOC, 0, NULL)) { goto out; } @@ -622,20 +665,32 @@ int btrfs_ordered_update_i_size(struct inode *inode, * if we find an ordered extent then we can't update disk i_size * yet */ - node = &ordered->rb_node; - while (1) { - node = rb_prev(node); - if (!node) - break; + if (ordered) { + node = rb_prev(&ordered->rb_node); + } else { + prev = tree_search(tree, offset); + /* + * we insert file extents without involving ordered struct, + * so there should be no ordered struct cover this offset + */ + if (prev) { + test = rb_entry(prev, struct btrfs_ordered_extent, + rb_node); + BUG_ON(offset_in_entry(test, offset)); + } + node = prev; + } + while (node) { test = rb_entry(node, struct btrfs_ordered_extent, rb_node); if (test->file_offset + test->len <= disk_i_size) break; - if (test->file_offset >= inode->i_size) + if (test->file_offset >= i_size) break; if (test->file_offset >= disk_i_size) goto out; + node = rb_prev(node); } - new_i_size = min_t(u64, entry_end(ordered), i_size_read(inode)); + new_i_size = min_t(u64, offset, i_size); /* * at this point, we know we can safely update i_size to at least @@ -643,7 +698,14 @@ int btrfs_ordered_update_i_size(struct inode *inode, * walk forward and see if ios from higher up in the file have * finished. */ - node = rb_next(&ordered->rb_node); + if (ordered) { + node = rb_next(&ordered->rb_node); + } else { + if (prev) + node = rb_next(prev); + else + node = rb_first(&tree->tree); + } i_size_test = 0; if (node) { /* @@ -651,10 +713,10 @@ int btrfs_ordered_update_i_size(struct inode *inode, * between our ordered extent and the next one. */ test = rb_entry(node, struct btrfs_ordered_extent, rb_node); - if (test->file_offset > entry_end(ordered)) + if (test->file_offset > offset) i_size_test = test->file_offset; } else { - i_size_test = i_size_read(inode); + i_size_test = i_size; } /* @@ -663,15 +725,25 @@ int btrfs_ordered_update_i_size(struct inode *inode, * are no delalloc bytes in this area, it is safe to update * disk_i_size to the end of the region. */ - if (i_size_test > entry_end(ordered) && - !test_range_bit(io_tree, entry_end(ordered), i_size_test - 1, - EXTENT_DELALLOC, 0, NULL)) { - new_i_size = min_t(u64, i_size_test, i_size_read(inode)); + if (i_size_test > offset && + !test_range_bit(io_tree, offset, i_size_test - 1, + EXTENT_DELALLOC, 0, NULL)) { + new_i_size = min_t(u64, i_size_test, i_size); } BTRFS_I(inode)->disk_i_size = new_i_size; + ret = 0; out: - mutex_unlock(&tree->mutex); - return 0; + /* + * we need to remove the ordered extent with the tree lock held + * so that other people calling this function don't find our fully + * processed ordered entry and skip updating the i_size + */ + if (ordered) + __btrfs_remove_ordered_extent(inode, ordered); + spin_unlock(&tree->lock); + if (ordered) + wake_up(&ordered->wait); + return ret; } /* @@ -695,7 +767,7 @@ int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr, if (!ordered) return 1; - mutex_lock(&tree->mutex); + spin_lock(&tree->lock); list_for_each_entry_reverse(ordered_sum, &ordered->list, list) { if (disk_bytenr >= ordered_sum->bytenr) { num_sectors = ordered_sum->len / sectorsize; @@ -710,95 +782,12 @@ int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr, } } out: - mutex_unlock(&tree->mutex); + spin_unlock(&tree->lock); btrfs_put_ordered_extent(ordered); return ret; } -/** - * taken from mm/filemap.c because it isn't exported - * - * __filemap_fdatawrite_range - start writeback on mapping dirty pages in range - * @mapping: address space structure to write - * @start: offset in bytes where the range starts - * @end: offset in bytes where the range ends (inclusive) - * @sync_mode: enable synchronous operation - * - * Start writeback against all of a mapping's dirty pages that lie - * within the byte offsets <start, end> inclusive. - * - * If sync_mode is WB_SYNC_ALL then this is a "data integrity" operation, as - * opposed to a regular memory cleansing writeback. The difference between - * these two operations is that if a dirty page/buffer is encountered, it must - * be waited upon, and not just skipped over. - */ -int btrfs_fdatawrite_range(struct address_space *mapping, loff_t start, - loff_t end, int sync_mode) -{ - struct writeback_control wbc = { - .sync_mode = sync_mode, - .nr_to_write = mapping->nrpages * 2, - .range_start = start, - .range_end = end, - }; - return btrfs_writepages(mapping, &wbc); -} - -/** - * taken from mm/filemap.c because it isn't exported - * - * wait_on_page_writeback_range - wait for writeback to complete - * @mapping: target address_space - * @start: beginning page index - * @end: ending page index - * - * Wait for writeback to complete against pages indexed by start->end - * inclusive - */ -int btrfs_wait_on_page_writeback_range(struct address_space *mapping, - pgoff_t start, pgoff_t end) -{ - struct pagevec pvec; - int nr_pages; - int ret = 0; - pgoff_t index; - - if (end < start) - return 0; - - pagevec_init(&pvec, 0); - index = start; - while ((index <= end) && - (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, - PAGECACHE_TAG_WRITEBACK, - min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1)) != 0) { - unsigned i; - - for (i = 0; i < nr_pages; i++) { - struct page *page = pvec.pages[i]; - - /* until radix tree lookup accepts end_index */ - if (page->index > end) - continue; - - wait_on_page_writeback(page); - if (PageError(page)) - ret = -EIO; - } - pagevec_release(&pvec); - cond_resched(); - } - - /* Check for outstanding write errors */ - if (test_and_clear_bit(AS_ENOSPC, &mapping->flags)) - ret = -ENOSPC; - if (test_and_clear_bit(AS_EIO, &mapping->flags)) - ret = -EIO; - - return ret; -} - /* * add a given inode to the list of inodes that must be fully on * disk before a transaction commit finishes. diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h index 993a7ea45c70..c82f76a9f040 100644 --- a/fs/btrfs/ordered-data.h +++ b/fs/btrfs/ordered-data.h @@ -21,7 +21,7 @@ /* one of these per inode */ struct btrfs_ordered_inode_tree { - struct mutex mutex; + spinlock_t lock; struct rb_root tree; struct rb_node *last; }; @@ -128,8 +128,8 @@ static inline int btrfs_ordered_sum_size(struct btrfs_root *root, static inline void btrfs_ordered_inode_tree_init(struct btrfs_ordered_inode_tree *t) { - mutex_init(&t->mutex); - t->tree.rb_node = NULL; + spin_lock_init(&t->lock); + t->tree = RB_ROOT; t->last = NULL; } @@ -137,7 +137,8 @@ int btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry); int btrfs_remove_ordered_extent(struct inode *inode, struct btrfs_ordered_extent *entry); int btrfs_dec_test_ordered_pending(struct inode *inode, - u64 file_offset, u64 io_size); + struct btrfs_ordered_extent **cached, + u64 file_offset, u64 io_size); int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset, u64 start, u64 len, u64 disk_len, int tyep); int btrfs_add_ordered_sum(struct inode *inode, @@ -150,16 +151,13 @@ void btrfs_start_ordered_extent(struct inode *inode, int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len); struct btrfs_ordered_extent * btrfs_lookup_first_ordered_extent(struct inode * inode, u64 file_offset); -int btrfs_ordered_update_i_size(struct inode *inode, +int btrfs_ordered_update_i_size(struct inode *inode, u64 offset, struct btrfs_ordered_extent *ordered); int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr, u32 *sum); -int btrfs_wait_on_page_writeback_range(struct address_space *mapping, - pgoff_t start, pgoff_t end); -int btrfs_fdatawrite_range(struct address_space *mapping, loff_t start, - loff_t end, int sync_mode); -int btrfs_wait_ordered_extents(struct btrfs_root *root, int nocow_only); int btrfs_run_ordered_operations(struct btrfs_root *root, int wait); int btrfs_add_ordered_operation(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct inode *inode); +int btrfs_wait_ordered_extents(struct btrfs_root *root, + int nocow_only, int delay_iput); #endif diff --git a/fs/btrfs/ref-cache.h b/fs/btrfs/ref-cache.h index bc283ad2db73..e2a55cb2072b 100644 --- a/fs/btrfs/ref-cache.h +++ b/fs/btrfs/ref-cache.h @@ -52,7 +52,7 @@ static inline size_t btrfs_leaf_ref_size(int nr_extents) static inline void btrfs_leaf_ref_tree_init(struct btrfs_leaf_ref_tree *tree) { - tree->root.rb_node = NULL; + tree->root = RB_ROOT; INIT_LIST_HEAD(&tree->list); spin_lock_init(&tree->lock); } diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c index 361ad323faac..0b23942cbc0d 100644 --- a/fs/btrfs/relocation.c +++ b/fs/btrfs/relocation.c @@ -170,14 +170,14 @@ struct async_merge { static void mapping_tree_init(struct mapping_tree *tree) { - tree->rb_root.rb_node = NULL; + tree->rb_root = RB_ROOT; spin_lock_init(&tree->lock); } static void backref_cache_init(struct backref_cache *cache) { int i; - cache->rb_root.rb_node = NULL; + cache->rb_root = RB_ROOT; for (i = 0; i < BTRFS_MAX_LEVEL; i++) INIT_LIST_HEAD(&cache->pending[i]); spin_lock_init(&cache->lock); @@ -1561,6 +1561,20 @@ static int invalidate_extent_cache(struct btrfs_root *root, return 0; } +static void put_inodes(struct list_head *list) +{ + struct inodevec *ivec; + while (!list_empty(list)) { + ivec = list_entry(list->next, struct inodevec, list); + list_del(&ivec->list); + while (ivec->nr > 0) { + ivec->nr--; + iput(ivec->inode[ivec->nr]); + } + kfree(ivec); + } +} + static int find_next_key(struct btrfs_path *path, int level, struct btrfs_key *key) @@ -1723,6 +1737,11 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc, btrfs_btree_balance_dirty(root, nr); + /* + * put inodes outside transaction, otherwise we may deadlock. + */ + put_inodes(&inode_list); + if (replaced && rc->stage == UPDATE_DATA_PTRS) invalidate_extent_cache(root, &key, &next_key); } @@ -1752,19 +1771,7 @@ out: btrfs_btree_balance_dirty(root, nr); - /* - * put inodes while we aren't holding the tree locks - */ - while (!list_empty(&inode_list)) { - struct inodevec *ivec; - ivec = list_entry(inode_list.next, struct inodevec, list); - list_del(&ivec->list); - while (ivec->nr > 0) { - ivec->nr--; - iput(ivec->inode[ivec->nr]); - } - kfree(ivec); - } + put_inodes(&inode_list); if (replaced && rc->stage == UPDATE_DATA_PTRS) invalidate_extent_cache(root, &key, &next_key); @@ -2652,7 +2659,7 @@ static int relocate_file_extent_cluster(struct inode *inode, EXTENT_BOUNDARY, GFP_NOFS); nr++; } - btrfs_set_extent_delalloc(inode, page_start, page_end); + btrfs_set_extent_delalloc(inode, page_start, page_end, NULL); set_page_dirty(page); dirty_page++; @@ -3274,8 +3281,10 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc) return -ENOMEM; path = btrfs_alloc_path(); - if (!path) + if (!path) { + kfree(cluster); return -ENOMEM; + } rc->extents_found = 0; rc->extents_skipped = 0; @@ -3478,7 +3487,7 @@ static struct inode *create_reloc_inode(struct btrfs_fs_info *fs_info, key.objectid = objectid; key.type = BTRFS_INODE_ITEM_KEY; key.offset = 0; - inode = btrfs_iget(root->fs_info->sb, &key, root); + inode = btrfs_iget(root->fs_info->sb, &key, root, NULL); BUG_ON(IS_ERR(inode) || is_bad_inode(inode)); BTRFS_I(inode)->index_cnt = group->key.objectid; @@ -3518,7 +3527,7 @@ int btrfs_relocate_block_group(struct btrfs_root *extent_root, u64 group_start) BUG_ON(!rc->block_group); btrfs_init_workers(&rc->workers, "relocate", - fs_info->thread_pool_size); + fs_info->thread_pool_size, NULL); rc->extent_root = extent_root; btrfs_prepare_block_group_relocation(extent_root, rc->block_group); @@ -3534,8 +3543,8 @@ int btrfs_relocate_block_group(struct btrfs_root *extent_root, u64 group_start) (unsigned long long)rc->block_group->key.objectid, (unsigned long long)rc->block_group->flags); - btrfs_start_delalloc_inodes(fs_info->tree_root); - btrfs_wait_ordered_extents(fs_info->tree_root, 0); + btrfs_start_delalloc_inodes(fs_info->tree_root, 0); + btrfs_wait_ordered_extents(fs_info->tree_root, 0, 0); while (1) { rc->extents_found = 0; @@ -3701,7 +3710,7 @@ int btrfs_recover_relocation(struct btrfs_root *root) mapping_tree_init(&rc->reloc_root_tree); INIT_LIST_HEAD(&rc->reloc_roots); btrfs_init_workers(&rc->workers, "relocate", - root->fs_info->thread_pool_size); + root->fs_info->thread_pool_size, NULL); rc->extent_root = root->fs_info->extent_root; set_reloc_control(rc); @@ -3755,6 +3764,8 @@ out: BTRFS_DATA_RELOC_TREE_OBJECTID); if (IS_ERR(fs_root)) err = PTR_ERR(fs_root); + else + btrfs_orphan_cleanup(fs_root); } return err; } diff --git a/fs/btrfs/root-tree.c b/fs/btrfs/root-tree.c index 9351428f30e2..67fa2d29d663 100644 --- a/fs/btrfs/root-tree.c +++ b/fs/btrfs/root-tree.c @@ -159,7 +159,6 @@ int btrfs_update_root(struct btrfs_trans_handle *trans, struct btrfs_root write_extent_buffer(l, item, ptr, sizeof(*item)); btrfs_mark_buffer_dirty(path->nodes[0]); out: - btrfs_release_path(root, path); btrfs_free_path(path); return ret; } @@ -332,7 +331,6 @@ int btrfs_del_root(struct btrfs_trans_handle *trans, struct btrfs_root *root, BUG_ON(refs != 0); ret = btrfs_del_item(trans, root, path); out: - btrfs_release_path(root, path); btrfs_free_path(path); return ret; } diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 67035385444c..9ac612e6ca60 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -63,15 +63,18 @@ static void btrfs_put_super(struct super_block *sb) } enum { - Opt_degraded, Opt_subvol, Opt_device, Opt_nodatasum, Opt_nodatacow, - Opt_max_extent, Opt_max_inline, Opt_alloc_start, Opt_nobarrier, - Opt_ssd, Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl, - Opt_compress, Opt_notreelog, Opt_ratio, Opt_flushoncommit, Opt_err, + Opt_degraded, Opt_subvol, Opt_subvolid, Opt_device, Opt_nodatasum, + Opt_nodatacow, Opt_max_extent, Opt_max_inline, Opt_alloc_start, + Opt_nobarrier, Opt_ssd, Opt_nossd, Opt_ssd_spread, Opt_thread_pool, + Opt_noacl, Opt_compress, Opt_compress_force, Opt_notreelog, Opt_ratio, + Opt_flushoncommit, + Opt_discard, Opt_err, }; static match_table_t tokens = { {Opt_degraded, "degraded"}, {Opt_subvol, "subvol=%s"}, + {Opt_subvolid, "subvolid=%d"}, {Opt_device, "device=%s"}, {Opt_nodatasum, "nodatasum"}, {Opt_nodatacow, "nodatacow"}, @@ -81,6 +84,7 @@ static match_table_t tokens = { {Opt_alloc_start, "alloc_start=%s"}, {Opt_thread_pool, "thread_pool=%d"}, {Opt_compress, "compress"}, + {Opt_compress_force, "compress-force"}, {Opt_ssd, "ssd"}, {Opt_ssd_spread, "ssd_spread"}, {Opt_nossd, "nossd"}, @@ -88,34 +92,10 @@ static match_table_t tokens = { {Opt_notreelog, "notreelog"}, {Opt_flushoncommit, "flushoncommit"}, {Opt_ratio, "metadata_ratio=%d"}, + {Opt_discard, "discard"}, {Opt_err, NULL}, }; -u64 btrfs_parse_size(char *str) -{ - u64 res; - int mult = 1; - char *end; - char last; - - res = simple_strtoul(str, &end, 10); - - last = end[0]; - if (isalpha(last)) { - last = tolower(last); - switch (last) { - case 'g': - mult *= 1024; - case 'm': - mult *= 1024; - case 'k': - mult *= 1024; - } - res = res * mult; - } - return res; -} - /* * Regular mount options parser. Everything that is needed only when * reading in a new superblock is parsed here. @@ -124,8 +104,9 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) { struct btrfs_fs_info *info = root->fs_info; substring_t args[MAX_OPT_ARGS]; - char *p, *num; + char *p, *num, *orig; int intarg; + int ret = 0; if (!options) return 0; @@ -138,6 +119,7 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) if (!options) return -ENOMEM; + orig = options; while ((p = strsep(&options, ",")) != NULL) { int token; @@ -151,6 +133,7 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) btrfs_set_opt(info->mount_opt, DEGRADED); break; case Opt_subvol: + case Opt_subvolid: case Opt_device: /* * These are parsed by btrfs_parse_early_options @@ -170,6 +153,11 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) printk(KERN_INFO "btrfs: use compression\n"); btrfs_set_opt(info->mount_opt, COMPRESS); break; + case Opt_compress_force: + printk(KERN_INFO "btrfs: forcing compression\n"); + btrfs_set_opt(info->mount_opt, FORCE_COMPRESS); + btrfs_set_opt(info->mount_opt, COMPRESS); + break; case Opt_ssd: printk(KERN_INFO "btrfs: use ssd allocation scheme\n"); btrfs_set_opt(info->mount_opt, SSD); @@ -203,7 +191,7 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) case Opt_max_extent: num = match_strdup(&args[0]); if (num) { - info->max_extent = btrfs_parse_size(num); + info->max_extent = memparse(num, NULL); kfree(num); info->max_extent = max_t(u64, @@ -215,7 +203,7 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) case Opt_max_inline: num = match_strdup(&args[0]); if (num) { - info->max_inline = btrfs_parse_size(num); + info->max_inline = memparse(num, NULL); kfree(num); if (info->max_inline) { @@ -230,7 +218,7 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) case Opt_alloc_start: num = match_strdup(&args[0]); if (num) { - info->alloc_start = btrfs_parse_size(num); + info->alloc_start = memparse(num, NULL); kfree(num); printk(KERN_INFO "btrfs: allocations start at %llu\n", @@ -257,12 +245,21 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) info->metadata_ratio); } break; + case Opt_discard: + btrfs_set_opt(info->mount_opt, DISCARD); + break; + case Opt_err: + printk(KERN_INFO "btrfs: unrecognized mount option " + "'%s'\n", p); + ret = -EINVAL; + goto out; default: break; } } - kfree(options); - return 0; +out: + kfree(orig); + return ret; } /* @@ -272,12 +269,13 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) * only when we need to allocate a new super block. */ static int btrfs_parse_early_options(const char *options, fmode_t flags, - void *holder, char **subvol_name, + void *holder, char **subvol_name, u64 *subvol_objectid, struct btrfs_fs_devices **fs_devices) { substring_t args[MAX_OPT_ARGS]; char *opts, *p; int error = 0; + int intarg; if (!options) goto out; @@ -300,6 +298,18 @@ static int btrfs_parse_early_options(const char *options, fmode_t flags, case Opt_subvol: *subvol_name = match_strdup(&args[0]); break; + case Opt_subvolid: + intarg = 0; + error = match_int(&args[0], &intarg); + if (!error) { + /* we want the original fs_tree */ + if (!intarg) + *subvol_objectid = + BTRFS_FS_TREE_OBJECTID; + else + *subvol_objectid = intarg; + } + break; case Opt_device: error = btrfs_scan_one_device(match_strdup(&args[0]), flags, holder, fs_devices); @@ -327,6 +337,110 @@ static int btrfs_parse_early_options(const char *options, fmode_t flags, return error; } +static struct dentry *get_default_root(struct super_block *sb, + u64 subvol_objectid) +{ + struct btrfs_root *root = sb->s_fs_info; + struct btrfs_root *new_root; + struct btrfs_dir_item *di; + struct btrfs_path *path; + struct btrfs_key location; + struct inode *inode; + struct dentry *dentry; + u64 dir_id; + int new = 0; + + /* + * We have a specific subvol we want to mount, just setup location and + * go look up the root. + */ + if (subvol_objectid) { + location.objectid = subvol_objectid; + location.type = BTRFS_ROOT_ITEM_KEY; + location.offset = (u64)-1; + goto find_root; + } + + path = btrfs_alloc_path(); + if (!path) + return ERR_PTR(-ENOMEM); + path->leave_spinning = 1; + + /* + * Find the "default" dir item which points to the root item that we + * will mount by default if we haven't been given a specific subvolume + * to mount. + */ + dir_id = btrfs_super_root_dir(&root->fs_info->super_copy); + di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0); + if (!di) { + /* + * Ok the default dir item isn't there. This is weird since + * it's always been there, but don't freak out, just try and + * mount to root most subvolume. + */ + btrfs_free_path(path); + dir_id = BTRFS_FIRST_FREE_OBJECTID; + new_root = root->fs_info->fs_root; + goto setup_root; + } + + btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location); + btrfs_free_path(path); + +find_root: + new_root = btrfs_read_fs_root_no_name(root->fs_info, &location); + if (IS_ERR(new_root)) + return ERR_PTR(PTR_ERR(new_root)); + + if (btrfs_root_refs(&new_root->root_item) == 0) + return ERR_PTR(-ENOENT); + + dir_id = btrfs_root_dirid(&new_root->root_item); +setup_root: + location.objectid = dir_id; + location.type = BTRFS_INODE_ITEM_KEY; + location.offset = 0; + + inode = btrfs_iget(sb, &location, new_root, &new); + if (!inode) + return ERR_PTR(-ENOMEM); + + /* + * If we're just mounting the root most subvol put the inode and return + * a reference to the dentry. We will have already gotten a reference + * to the inode in btrfs_fill_super so we're good to go. + */ + if (!new && sb->s_root->d_inode == inode) { + iput(inode); + return dget(sb->s_root); + } + + if (new) { + const struct qstr name = { .name = "/", .len = 1 }; + + /* + * New inode, we need to make the dentry a sibling of s_root so + * everything gets cleaned up properly on unmount. + */ + dentry = d_alloc(sb->s_root, &name); + if (!dentry) { + iput(inode); + return ERR_PTR(-ENOMEM); + } + d_splice_alias(inode, dentry); + } else { + /* + * We found the inode in cache, just find a dentry for it and + * put the reference to the inode we just got. + */ + dentry = d_find_alias(inode); + iput(inode); + } + + return dentry; +} + static int btrfs_fill_super(struct super_block *sb, struct btrfs_fs_devices *fs_devices, void *data, int silent) @@ -344,7 +458,9 @@ static int btrfs_fill_super(struct super_block *sb, sb->s_export_op = &btrfs_export_ops; sb->s_xattr = btrfs_xattr_handlers; sb->s_time_gran = 1; +#ifdef CONFIG_BTRFS_FS_POSIX_ACL sb->s_flags |= MS_POSIXACL; +#endif tree_root = open_ctree(sb, fs_devices, (char *)data); @@ -358,7 +474,7 @@ static int btrfs_fill_super(struct super_block *sb, key.objectid = BTRFS_FIRST_FREE_OBJECTID; key.type = BTRFS_INODE_ITEM_KEY; key.offset = 0; - inode = btrfs_iget(sb, &key, tree_root->fs_info->fs_root); + inode = btrfs_iget(sb, &key, tree_root->fs_info->fs_root, NULL); if (IS_ERR(inode)) { err = PTR_ERR(inode); goto fail_close; @@ -370,12 +486,6 @@ static int btrfs_fill_super(struct super_block *sb, err = -ENOMEM; goto fail_close; } -#if 0 - /* this does the super kobj at the same time */ - err = btrfs_sysfs_add_super(tree_root->fs_info); - if (err) - goto fail_close; -#endif sb->s_root = root_dentry; @@ -398,8 +508,8 @@ int btrfs_sync_fs(struct super_block *sb, int wait) return 0; } - btrfs_start_delalloc_inodes(root); - btrfs_wait_ordered_extents(root, 0); + btrfs_start_delalloc_inodes(root, 0); + btrfs_wait_ordered_extents(root, 0, 0); trans = btrfs_start_transaction(root, 1); ret = btrfs_commit_transaction(trans, root); @@ -443,6 +553,8 @@ static int btrfs_show_options(struct seq_file *seq, struct vfsmount *vfs) seq_puts(seq, ",notreelog"); if (btrfs_test_opt(root, FLUSHONCOMMIT)) seq_puts(seq, ",flushoncommit"); + if (btrfs_test_opt(root, DISCARD)) + seq_puts(seq, ",discard"); if (!(root->fs_info->sb->s_flags & MS_POSIXACL)) seq_puts(seq, ",noacl"); return 0; @@ -465,19 +577,22 @@ static int btrfs_test_super(struct super_block *s, void *data) static int btrfs_get_sb(struct file_system_type *fs_type, int flags, const char *dev_name, void *data, struct vfsmount *mnt) { - char *subvol_name = NULL; struct block_device *bdev = NULL; struct super_block *s; struct dentry *root; struct btrfs_fs_devices *fs_devices = NULL; fmode_t mode = FMODE_READ; + char *subvol_name = NULL; + u64 subvol_objectid = 0; int error = 0; + int found = 0; if (!(flags & MS_RDONLY)) mode |= FMODE_WRITE; error = btrfs_parse_early_options(data, mode, fs_type, - &subvol_name, &fs_devices); + &subvol_name, &subvol_objectid, + &fs_devices); if (error) return error; @@ -506,6 +621,7 @@ static int btrfs_get_sb(struct file_system_type *fs_type, int flags, goto error_close_devices; } + found = 1; btrfs_close_devices(fs_devices); } else { char b[BDEVNAME_SIZE]; @@ -523,25 +639,35 @@ static int btrfs_get_sb(struct file_system_type *fs_type, int flags, s->s_flags |= MS_ACTIVE; } - if (!strcmp(subvol_name, ".")) - root = dget(s->s_root); - else { - mutex_lock(&s->s_root->d_inode->i_mutex); - root = lookup_one_len(subvol_name, s->s_root, + root = get_default_root(s, subvol_objectid); + if (IS_ERR(root)) { + error = PTR_ERR(root); + deactivate_locked_super(s); + goto error; + } + /* if they gave us a subvolume name bind mount into that */ + if (strcmp(subvol_name, ".")) { + struct dentry *new_root; + mutex_lock(&root->d_inode->i_mutex); + new_root = lookup_one_len(subvol_name, root, strlen(subvol_name)); - mutex_unlock(&s->s_root->d_inode->i_mutex); + mutex_unlock(&root->d_inode->i_mutex); - if (IS_ERR(root)) { + if (IS_ERR(new_root)) { deactivate_locked_super(s); - error = PTR_ERR(root); - goto error_free_subvol_name; + error = PTR_ERR(new_root); + dput(root); + goto error_close_devices; } - if (!root->d_inode) { + if (!new_root->d_inode) { dput(root); + dput(new_root); deactivate_locked_super(s); error = -ENXIO; - goto error_free_subvol_name; + goto error_close_devices; } + dput(root); + root = new_root; } mnt->mnt_sb = s; @@ -556,6 +682,7 @@ error_close_devices: btrfs_close_devices(fs_devices); error_free_subvol_name: kfree(subvol_name); +error: return error; } @@ -600,14 +727,37 @@ static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf) { struct btrfs_root *root = btrfs_sb(dentry->d_sb); struct btrfs_super_block *disk_super = &root->fs_info->super_copy; + struct list_head *head = &root->fs_info->space_info; + struct btrfs_space_info *found; + u64 total_used = 0; + u64 data_used = 0; int bits = dentry->d_sb->s_blocksize_bits; __be32 *fsid = (__be32 *)root->fs_info->fsid; + rcu_read_lock(); + list_for_each_entry_rcu(found, head, list) { + if (found->flags & (BTRFS_BLOCK_GROUP_DUP| + BTRFS_BLOCK_GROUP_RAID10| + BTRFS_BLOCK_GROUP_RAID1)) { + total_used += found->bytes_used; + if (found->flags & BTRFS_BLOCK_GROUP_DATA) + data_used += found->bytes_used; + else + data_used += found->total_bytes; + } + + total_used += found->bytes_used; + if (found->flags & BTRFS_BLOCK_GROUP_DATA) + data_used += found->bytes_used; + else + data_used += found->total_bytes; + } + rcu_read_unlock(); + buf->f_namelen = BTRFS_NAME_LEN; buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits; - buf->f_bfree = buf->f_blocks - - (btrfs_super_bytes_used(disk_super) >> bits); - buf->f_bavail = buf->f_bfree; + buf->f_bfree = buf->f_blocks - (total_used >> bits); + buf->f_bavail = buf->f_blocks - (data_used >> bits); buf->f_bsize = dentry->d_sb->s_blocksize; buf->f_type = BTRFS_SUPER_MAGIC; diff --git a/fs/btrfs/sysfs.c b/fs/btrfs/sysfs.c index a240b6fa81df..4ce16ef702a3 100644 --- a/fs/btrfs/sysfs.c +++ b/fs/btrfs/sysfs.c @@ -164,12 +164,12 @@ static void btrfs_root_release(struct kobject *kobj) complete(&root->kobj_unregister); } -static struct sysfs_ops btrfs_super_attr_ops = { +static const struct sysfs_ops btrfs_super_attr_ops = { .show = btrfs_super_attr_show, .store = btrfs_super_attr_store, }; -static struct sysfs_ops btrfs_root_attr_ops = { +static const struct sysfs_ops btrfs_root_attr_ops = { .show = btrfs_root_attr_show, .store = btrfs_root_attr_store, }; diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 88f866f85e7a..2d654c1c794d 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -69,7 +69,7 @@ static noinline int join_transaction(struct btrfs_root *root) cur_trans->commit_done = 0; cur_trans->start_time = get_seconds(); - cur_trans->delayed_refs.root.rb_node = NULL; + cur_trans->delayed_refs.root = RB_ROOT; cur_trans->delayed_refs.num_entries = 0; cur_trans->delayed_refs.num_heads_ready = 0; cur_trans->delayed_refs.num_heads = 0; @@ -163,8 +163,14 @@ static void wait_current_trans(struct btrfs_root *root) } } +enum btrfs_trans_type { + TRANS_START, + TRANS_JOIN, + TRANS_USERSPACE, +}; + static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root, - int num_blocks, int wait) + int num_blocks, int type) { struct btrfs_trans_handle *h = kmem_cache_alloc(btrfs_trans_handle_cachep, GFP_NOFS); @@ -172,7 +178,8 @@ static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root, mutex_lock(&root->fs_info->trans_mutex); if (!root->fs_info->log_root_recovering && - ((wait == 1 && !root->fs_info->open_ioctl_trans) || wait == 2)) + ((type == TRANS_START && !root->fs_info->open_ioctl_trans) || + type == TRANS_USERSPACE)) wait_current_trans(root); ret = join_transaction(root); BUG_ON(ret); @@ -186,6 +193,9 @@ static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root, h->alloc_exclude_start = 0; h->delayed_ref_updates = 0; + if (!current->journal_info && type != TRANS_USERSPACE) + current->journal_info = h; + root->fs_info->running_transaction->use_count++; record_root_in_trans(h, root); mutex_unlock(&root->fs_info->trans_mutex); @@ -195,18 +205,18 @@ static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root, struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root, int num_blocks) { - return start_transaction(root, num_blocks, 1); + return start_transaction(root, num_blocks, TRANS_START); } struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root, int num_blocks) { - return start_transaction(root, num_blocks, 0); + return start_transaction(root, num_blocks, TRANS_JOIN); } struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *r, int num_blocks) { - return start_transaction(r, num_blocks, 2); + return start_transaction(r, num_blocks, TRANS_USERSPACE); } /* wait for a transaction commit to be fully complete */ @@ -317,9 +327,15 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans, wake_up(&cur_trans->writer_wait); put_transaction(cur_trans); mutex_unlock(&info->trans_mutex); + + if (current->journal_info == trans) + current->journal_info = NULL; memset(trans, 0, sizeof(*trans)); kmem_cache_free(btrfs_trans_handle_cachep, trans); + if (throttle) + btrfs_run_delayed_iputs(root); + return 0; } @@ -338,10 +354,10 @@ int btrfs_end_transaction_throttle(struct btrfs_trans_handle *trans, /* * when btree blocks are allocated, they have some corresponding bits set for * them in one of two extent_io trees. This is used to make sure all of - * those extents are on disk for transaction or log commit + * those extents are sent to disk but does not wait on them */ -int btrfs_write_and_wait_marked_extents(struct btrfs_root *root, - struct extent_io_tree *dirty_pages) +int btrfs_write_marked_extents(struct btrfs_root *root, + struct extent_io_tree *dirty_pages, int mark) { int ret; int err = 0; @@ -354,7 +370,7 @@ int btrfs_write_and_wait_marked_extents(struct btrfs_root *root, while (1) { ret = find_first_extent_bit(dirty_pages, start, &start, &end, - EXTENT_DIRTY); + mark); if (ret) break; while (start <= end) { @@ -388,13 +404,36 @@ int btrfs_write_and_wait_marked_extents(struct btrfs_root *root, page_cache_release(page); } } + if (err) + werr = err; + return werr; +} + +/* + * when btree blocks are allocated, they have some corresponding bits set for + * them in one of two extent_io trees. This is used to make sure all of + * those extents are on disk for transaction or log commit. We wait + * on all the pages and clear them from the dirty pages state tree + */ +int btrfs_wait_marked_extents(struct btrfs_root *root, + struct extent_io_tree *dirty_pages, int mark) +{ + int ret; + int err = 0; + int werr = 0; + struct page *page; + struct inode *btree_inode = root->fs_info->btree_inode; + u64 start = 0; + u64 end; + unsigned long index; + while (1) { - ret = find_first_extent_bit(dirty_pages, 0, &start, &end, - EXTENT_DIRTY); + ret = find_first_extent_bit(dirty_pages, start, &start, &end, + mark); if (ret) break; - clear_extent_dirty(dirty_pages, start, end, GFP_NOFS); + clear_extent_bits(dirty_pages, start, end, mark, GFP_NOFS); while (start <= end) { index = start >> PAGE_CACHE_SHIFT; start = (u64)(index + 1) << PAGE_CACHE_SHIFT; @@ -418,6 +457,22 @@ int btrfs_write_and_wait_marked_extents(struct btrfs_root *root, return werr; } +/* + * when btree blocks are allocated, they have some corresponding bits set for + * them in one of two extent_io trees. This is used to make sure all of + * those extents are on disk for transaction or log commit + */ +int btrfs_write_and_wait_marked_extents(struct btrfs_root *root, + struct extent_io_tree *dirty_pages, int mark) +{ + int ret; + int ret2; + + ret = btrfs_write_marked_extents(root, dirty_pages, mark); + ret2 = btrfs_wait_marked_extents(root, dirty_pages, mark); + return ret || ret2; +} + int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans, struct btrfs_root *root) { @@ -427,7 +482,8 @@ int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans, return filemap_write_and_wait(btree_inode->i_mapping); } return btrfs_write_and_wait_marked_extents(root, - &trans->transaction->dirty_pages); + &trans->transaction->dirty_pages, + EXTENT_DIRTY); } /* @@ -445,13 +501,16 @@ static int update_cowonly_root(struct btrfs_trans_handle *trans, { int ret; u64 old_root_bytenr; + u64 old_root_used; struct btrfs_root *tree_root = root->fs_info->tree_root; + old_root_used = btrfs_root_used(&root->root_item); btrfs_write_dirty_block_groups(trans, root); while (1) { old_root_bytenr = btrfs_root_bytenr(&root->root_item); - if (old_root_bytenr == root->node->start) + if (old_root_bytenr == root->node->start && + old_root_used == btrfs_root_used(&root->root_item)) break; btrfs_set_root_node(&root->root_item, root->node); @@ -460,6 +519,7 @@ static int update_cowonly_root(struct btrfs_trans_handle *trans, &root->root_item); BUG_ON(ret); + old_root_used = btrfs_root_used(&root->root_item); ret = btrfs_write_dirty_block_groups(trans, root); BUG_ON(ret); } @@ -754,7 +814,6 @@ static noinline int finish_pending_snapshot(struct btrfs_fs_info *fs_info, u64 index = 0; struct btrfs_trans_handle *trans; struct inode *parent_inode; - struct inode *inode; struct btrfs_root *parent_root; parent_inode = pending->dentry->d_parent->d_inode; @@ -786,8 +845,6 @@ static noinline int finish_pending_snapshot(struct btrfs_fs_info *fs_info, BUG_ON(ret); - inode = btrfs_lookup_dentry(parent_inode, pending->dentry); - d_instantiate(pending->dentry, inode); fail: btrfs_end_transaction(trans, fs_info->fs_root); return ret; @@ -940,12 +997,9 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans, mutex_unlock(&root->fs_info->trans_mutex); - if (flush_on_commit) { - btrfs_start_delalloc_inodes(root); - ret = btrfs_wait_ordered_extents(root, 0); - BUG_ON(ret); - } else if (snap_pending) { - ret = btrfs_wait_ordered_extents(root, 1); + if (flush_on_commit || snap_pending) { + btrfs_start_delalloc_inodes(root, 1); + ret = btrfs_wait_ordered_extents(root, 0, 1); BUG_ON(ret); } @@ -1059,7 +1113,14 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans, mutex_unlock(&root->fs_info->trans_mutex); + if (current->journal_info == trans) + current->journal_info = NULL; + kmem_cache_free(btrfs_trans_handle_cachep, trans); + + if (current != root->fs_info->transaction_kthread) + btrfs_run_delayed_iputs(root); + return ret; } diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h index 663c67404918..93c7ccb33118 100644 --- a/fs/btrfs/transaction.h +++ b/fs/btrfs/transaction.h @@ -79,6 +79,7 @@ static inline void btrfs_set_inode_last_trans(struct btrfs_trans_handle *trans, struct inode *inode) { BTRFS_I(inode)->last_trans = trans->transaction->transid; + BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid; } int btrfs_end_transaction(struct btrfs_trans_handle *trans, @@ -106,6 +107,10 @@ void btrfs_throttle(struct btrfs_root *root); int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans, struct btrfs_root *root); int btrfs_write_and_wait_marked_extents(struct btrfs_root *root, - struct extent_io_tree *dirty_pages); + struct extent_io_tree *dirty_pages, int mark); +int btrfs_write_marked_extents(struct btrfs_root *root, + struct extent_io_tree *dirty_pages, int mark); +int btrfs_wait_marked_extents(struct btrfs_root *root, + struct extent_io_tree *dirty_pages, int mark); int btrfs_transaction_in_commit(struct btrfs_fs_info *info); #endif diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c index 7827841b55cb..1255fcc8ade5 100644 --- a/fs/btrfs/tree-log.c +++ b/fs/btrfs/tree-log.c @@ -137,11 +137,20 @@ static int start_log_trans(struct btrfs_trans_handle *trans, mutex_lock(&root->log_mutex); if (root->log_root) { + if (!root->log_start_pid) { + root->log_start_pid = current->pid; + root->log_multiple_pids = false; + } else if (root->log_start_pid != current->pid) { + root->log_multiple_pids = true; + } + root->log_batch++; atomic_inc(&root->log_writers); mutex_unlock(&root->log_mutex); return 0; } + root->log_multiple_pids = false; + root->log_start_pid = current->pid; mutex_lock(&root->fs_info->tree_log_mutex); if (!root->fs_info->log_root_tree) { ret = btrfs_init_log_root_tree(trans, root->fs_info); @@ -436,7 +445,7 @@ static noinline struct inode *read_one_inode(struct btrfs_root *root, key.objectid = objectid; key.type = BTRFS_INODE_ITEM_KEY; key.offset = 0; - inode = btrfs_iget(root->fs_info->sb, &key, root); + inode = btrfs_iget(root->fs_info->sb, &key, root, NULL); if (IS_ERR(inode)) { inode = NULL; } else if (is_bad_inode(inode)) { @@ -533,8 +542,8 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans, saved_nbytes = inode_get_bytes(inode); /* drop any overlapping extents */ - ret = btrfs_drop_extents(trans, root, inode, - start, extent_end, extent_end, start, &alloc_hint, 1); + ret = btrfs_drop_extents(trans, inode, start, extent_end, + &alloc_hint, 1); BUG_ON(ret); if (found_type == BTRFS_FILE_EXTENT_REG || @@ -921,6 +930,17 @@ out_nowrite: return 0; } +static int insert_orphan_item(struct btrfs_trans_handle *trans, + struct btrfs_root *root, u64 offset) +{ + int ret; + ret = btrfs_find_orphan_item(root, offset); + if (ret > 0) + ret = btrfs_insert_orphan_item(trans, root, offset); + return ret; +} + + /* * There are a few corners where the link count of the file can't * be properly maintained during replay. So, instead of adding @@ -988,9 +1008,13 @@ static noinline int fixup_inode_link_count(struct btrfs_trans_handle *trans, } BTRFS_I(inode)->index_cnt = (u64)-1; - if (inode->i_nlink == 0 && S_ISDIR(inode->i_mode)) { - ret = replay_dir_deletes(trans, root, NULL, path, - inode->i_ino, 1); + if (inode->i_nlink == 0) { + if (S_ISDIR(inode->i_mode)) { + ret = replay_dir_deletes(trans, root, NULL, path, + inode->i_ino, 1); + BUG_ON(ret); + } + ret = insert_orphan_item(trans, root, inode->i_ino); BUG_ON(ret); } btrfs_free_path(path); @@ -1578,7 +1602,6 @@ static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb, /* inode keys are done during the first stage */ if (key.type == BTRFS_INODE_ITEM_KEY && wc->stage == LOG_WALK_REPLAY_INODES) { - struct inode *inode; struct btrfs_inode_item *inode_item; u32 mode; @@ -1594,31 +1617,16 @@ static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb, eb, i, &key); BUG_ON(ret); - /* for regular files, truncate away - * extents past the new EOF + /* for regular files, make sure corresponding + * orhpan item exist. extents past the new EOF + * will be truncated later by orphan cleanup. */ if (S_ISREG(mode)) { - inode = read_one_inode(root, - key.objectid); - BUG_ON(!inode); - - ret = btrfs_truncate_inode_items(wc->trans, - root, inode, inode->i_size, - BTRFS_EXTENT_DATA_KEY); + ret = insert_orphan_item(wc->trans, root, + key.objectid); BUG_ON(ret); - - /* if the nlink count is zero here, the iput - * will free the inode. We bump it to make - * sure it doesn't get freed until the link - * count fixup is done - */ - if (inode->i_nlink == 0) { - btrfs_inc_nlink(inode); - btrfs_update_inode(wc->trans, - root, inode); - } - iput(inode); } + ret = link_to_fixup_dir(wc->trans, root, path, key.objectid); BUG_ON(ret); @@ -1968,9 +1976,11 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, { int index1; int index2; + int mark; int ret; struct btrfs_root *log = root->log_root; struct btrfs_root *log_root_tree = root->fs_info->log_root_tree; + unsigned long log_transid = 0; mutex_lock(&root->log_mutex); index1 = root->log_transid % 2; @@ -1987,10 +1997,11 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, while (1) { unsigned long batch = root->log_batch; - mutex_unlock(&root->log_mutex); - schedule_timeout_uninterruptible(1); - mutex_lock(&root->log_mutex); - + if (root->log_multiple_pids) { + mutex_unlock(&root->log_mutex); + schedule_timeout_uninterruptible(1); + mutex_lock(&root->log_mutex); + } wait_for_writer(trans, root); if (batch == root->log_batch) break; @@ -2003,7 +2014,16 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, goto out; } - ret = btrfs_write_and_wait_marked_extents(log, &log->dirty_log_pages); + log_transid = root->log_transid; + if (log_transid % 2 == 0) + mark = EXTENT_DIRTY; + else + mark = EXTENT_NEW; + + /* we start IO on all the marked extents here, but we don't actually + * wait for them until later. + */ + ret = btrfs_write_marked_extents(log, &log->dirty_log_pages, mark); BUG_ON(ret); btrfs_set_root_node(&log->root_item, log->node); @@ -2011,11 +2031,12 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, root->log_batch = 0; root->log_transid++; log->log_transid = root->log_transid; + root->log_start_pid = 0; smp_mb(); /* - * log tree has been flushed to disk, new modifications of - * the log will be written to new positions. so it's safe to - * allow log writers to go in. + * IO has been started, blocks of the log tree have WRITTEN flag set + * in their headers. new modifications of the log will be written to + * new positions. so it's safe to allow log writers to go in. */ mutex_unlock(&root->log_mutex); @@ -2036,6 +2057,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, index2 = log_root_tree->log_transid % 2; if (atomic_read(&log_root_tree->log_commit[index2])) { + btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark); wait_log_commit(trans, log_root_tree, log_root_tree->log_transid); mutex_unlock(&log_root_tree->log_mutex); @@ -2055,14 +2077,17 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, * check the full commit flag again */ if (root->fs_info->last_trans_log_full_commit == trans->transid) { + btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark); mutex_unlock(&log_root_tree->log_mutex); ret = -EAGAIN; goto out_wake_log_root; } ret = btrfs_write_and_wait_marked_extents(log_root_tree, - &log_root_tree->dirty_log_pages); + &log_root_tree->dirty_log_pages, + EXTENT_DIRTY | EXTENT_NEW); BUG_ON(ret); + btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark); btrfs_set_super_log_root(&root->fs_info->super_for_commit, log_root_tree->node->start); @@ -2082,9 +2107,14 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, * the running transaction open, so a full commit can't hop * in and cause problems either. */ - write_ctree_super(trans, root->fs_info->tree_root, 2); + write_ctree_super(trans, root->fs_info->tree_root, 1); ret = 0; + mutex_lock(&root->log_mutex); + if (root->last_log_commit < log_transid) + root->last_log_commit = log_transid; + mutex_unlock(&root->log_mutex); + out_wake_log_root: atomic_set(&log_root_tree->log_commit[index2], 0); smp_mb(); @@ -2123,12 +2153,12 @@ int btrfs_free_log(struct btrfs_trans_handle *trans, struct btrfs_root *root) while (1) { ret = find_first_extent_bit(&log->dirty_log_pages, - 0, &start, &end, EXTENT_DIRTY); + 0, &start, &end, EXTENT_DIRTY | EXTENT_NEW); if (ret) break; - clear_extent_dirty(&log->dirty_log_pages, - start, end, GFP_NOFS); + clear_extent_bits(&log->dirty_log_pages, start, end, + EXTENT_DIRTY | EXTENT_NEW, GFP_NOFS); } if (log->log_transid > 0) { @@ -2852,6 +2882,21 @@ out: return ret; } +static int inode_in_log(struct btrfs_trans_handle *trans, + struct inode *inode) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + int ret = 0; + + mutex_lock(&root->log_mutex); + if (BTRFS_I(inode)->logged_trans == trans->transid && + BTRFS_I(inode)->last_sub_trans <= root->last_log_commit) + ret = 1; + mutex_unlock(&root->log_mutex); + return ret; +} + + /* * helper function around btrfs_log_inode to make sure newly created * parent directories also end up in the log. A minimal inode and backref @@ -2891,6 +2936,11 @@ int btrfs_log_inode_parent(struct btrfs_trans_handle *trans, if (ret) goto end_no_trans; + if (inode_in_log(trans, inode)) { + ret = BTRFS_NO_LOG_SYNC; + goto end_no_trans; + } + start_log_trans(trans, root); ret = btrfs_log_inode(trans, root, inode, inode_only); diff --git a/fs/btrfs/tree-log.h b/fs/btrfs/tree-log.h index d09c7609e16b..0776eacb5083 100644 --- a/fs/btrfs/tree-log.h +++ b/fs/btrfs/tree-log.h @@ -19,6 +19,9 @@ #ifndef __TREE_LOG_ #define __TREE_LOG_ +/* return value for btrfs_log_dentry_safe that means we don't need to log it at all */ +#define BTRFS_NO_LOG_SYNC 256 + int btrfs_sync_log(struct btrfs_trans_handle *trans, struct btrfs_root *root); int btrfs_free_log(struct btrfs_trans_handle *trans, struct btrfs_root *root); diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 23e7d36ff325..9df8e3f1ccab 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -256,13 +256,13 @@ loop_lock: wake_up(&fs_info->async_submit_wait); BUG_ON(atomic_read(&cur->bi_cnt) == 0); - submit_bio(cur->bi_rw, cur); - num_run++; - batch_run++; if (bio_rw_flagged(cur, BIO_RW_SYNCIO)) num_sync_run++; + submit_bio(cur->bi_rw, cur); + num_run++; + batch_run++; if (need_resched()) { if (num_sync_run) { blk_run_backing_dev(bdi, NULL); @@ -325,16 +325,6 @@ loop_lock: num_sync_run = 0; blk_run_backing_dev(bdi, NULL); } - - cond_resched(); - if (again) - goto loop; - - spin_lock(&device->io_lock); - if (device->pending_bios.head || device->pending_sync_bios.head) - goto loop_lock; - spin_unlock(&device->io_lock); - /* * IO has already been through a long path to get here. Checksumming, * async helper threads, perhaps compression. We've done a pretty @@ -346,6 +336,16 @@ loop_lock: * cared about found its way down here. */ blk_run_backing_dev(bdi, NULL); + + cond_resched(); + if (again) + goto loop; + + spin_lock(&device->io_lock); + if (device->pending_bios.head || device->pending_sync_bios.head) + goto loop_lock; + spin_unlock(&device->io_lock); + done: return 0; } @@ -365,6 +365,7 @@ static noinline int device_list_add(const char *path, struct btrfs_device *device; struct btrfs_fs_devices *fs_devices; u64 found_transid = btrfs_super_generation(disk_super); + char *name; fs_devices = find_fsid(disk_super->fsid); if (!fs_devices) { @@ -411,6 +412,12 @@ static noinline int device_list_add(const char *path, device->fs_devices = fs_devices; fs_devices->num_devices++; + } else if (strcmp(device->name, path)) { + name = kstrdup(path, GFP_NOFS); + if (!name) + return -ENOMEM; + kfree(device->name); + device->name = name; } if (found_transid > fs_devices->latest_trans) { @@ -446,8 +453,10 @@ static struct btrfs_fs_devices *clone_fs_devices(struct btrfs_fs_devices *orig) goto error; device->name = kstrdup(orig_dev->name, GFP_NOFS); - if (!device->name) + if (!device->name) { + kfree(device); goto error; + } device->devid = orig_dev->devid; device->work.func = pending_bios_fn; @@ -590,7 +599,7 @@ static int __btrfs_open_devices(struct btrfs_fs_devices *fs_devices, goto error_close; disk_super = (struct btrfs_super_block *)bh->b_data; - devid = le64_to_cpu(disk_super->dev_item.devid); + devid = btrfs_stack_device_id(&disk_super->dev_item); if (devid != device->devid) goto error_brelse; @@ -692,7 +701,7 @@ int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder, goto error_close; } disk_super = (struct btrfs_super_block *)bh->b_data; - devid = le64_to_cpu(disk_super->dev_item.devid); + devid = btrfs_stack_device_id(&disk_super->dev_item); transid = btrfs_super_generation(disk_super); if (disk_super->label[0]) printk(KERN_INFO "device label %s ", disk_super->label); @@ -1133,7 +1142,7 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path) root->fs_info->avail_metadata_alloc_bits; if ((all_avail & BTRFS_BLOCK_GROUP_RAID10) && - root->fs_info->fs_devices->rw_devices <= 4) { + root->fs_info->fs_devices->num_devices <= 4) { printk(KERN_ERR "btrfs: unable to go below four devices " "on raid10\n"); ret = -EINVAL; @@ -1141,7 +1150,7 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path) } if ((all_avail & BTRFS_BLOCK_GROUP_RAID1) && - root->fs_info->fs_devices->rw_devices <= 2) { + root->fs_info->fs_devices->num_devices <= 2) { printk(KERN_ERR "btrfs: unable to go below two " "devices on raid1\n"); ret = -EINVAL; @@ -1185,7 +1194,7 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path) goto error_close; } disk_super = (struct btrfs_super_block *)bh->b_data; - devid = le64_to_cpu(disk_super->dev_item.devid); + devid = btrfs_stack_device_id(&disk_super->dev_item); dev_uuid = disk_super->dev_item.uuid; device = btrfs_find_device(root, devid, dev_uuid, disk_super->fsid); @@ -1432,8 +1441,8 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path) return -EINVAL; bdev = open_bdev_exclusive(device_path, 0, root->fs_info->bdev_holder); - if (!bdev) - return -EIO; + if (IS_ERR(bdev)) + return PTR_ERR(bdev); if (root->fs_info->fs_devices->seeding) { seeding_dev = 1; @@ -2207,7 +2216,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans, max_chunk_size = 10 * calc_size; min_stripe_size = 64 * 1024 * 1024; } else if (type & BTRFS_BLOCK_GROUP_METADATA) { - max_chunk_size = 4 * calc_size; + max_chunk_size = 256 * 1024 * 1024; min_stripe_size = 32 * 1024 * 1024; } else if (type & BTRFS_BLOCK_GROUP_SYSTEM) { calc_size = 8 * 1024 * 1024; @@ -2536,6 +2545,11 @@ int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset) if (!em) return 1; + if (btrfs_test_opt(root, DEGRADED)) { + free_extent_map(em); + return 0; + } + map = (struct map_lookup *)em->bdev; for (i = 0; i < map->num_stripes; i++) { if (!map->stripes[i].dev->writeable) { @@ -2647,8 +2661,10 @@ again: em = lookup_extent_mapping(em_tree, logical, *length); read_unlock(&em_tree->lock); - if (!em && unplug_page) + if (!em && unplug_page) { + kfree(multi); return 0; + } if (!em) { printk(KERN_CRIT "unable to find logical %llu len %llu\n", diff --git a/fs/btrfs/xattr.c b/fs/btrfs/xattr.c index a9d3bf4d2689..193b58f7d3f3 100644 --- a/fs/btrfs/xattr.c +++ b/fs/btrfs/xattr.c @@ -85,22 +85,23 @@ out: return ret; } -int __btrfs_setxattr(struct inode *inode, const char *name, - const void *value, size_t size, int flags) +static int do_setxattr(struct btrfs_trans_handle *trans, + struct inode *inode, const char *name, + const void *value, size_t size, int flags) { struct btrfs_dir_item *di; struct btrfs_root *root = BTRFS_I(inode)->root; - struct btrfs_trans_handle *trans; struct btrfs_path *path; - int ret = 0, mod = 0; + size_t name_len = strlen(name); + int ret = 0; + + if (name_len + size > BTRFS_MAX_XATTR_SIZE(root)) + return -ENOSPC; path = btrfs_alloc_path(); if (!path) return -ENOMEM; - trans = btrfs_join_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); - /* first lets see if we already have this xattr */ di = btrfs_lookup_xattr(trans, root, path, inode->i_ino, name, strlen(name), -1); @@ -118,15 +119,12 @@ int __btrfs_setxattr(struct inode *inode, const char *name, } ret = btrfs_delete_one_dir_name(trans, root, path, di); - if (ret) - goto out; + BUG_ON(ret); btrfs_release_path(root, path); /* if we don't have a value then we are removing the xattr */ - if (!value) { - mod = 1; + if (!value) goto out; - } } else { btrfs_release_path(root, path); @@ -138,20 +136,45 @@ int __btrfs_setxattr(struct inode *inode, const char *name, } /* ok we have to create a completely new xattr */ - ret = btrfs_insert_xattr_item(trans, root, name, strlen(name), - value, size, inode->i_ino); + ret = btrfs_insert_xattr_item(trans, root, path, inode->i_ino, + name, name_len, value, size); + BUG_ON(ret); +out: + btrfs_free_path(path); + return ret; +} + +int __btrfs_setxattr(struct btrfs_trans_handle *trans, + struct inode *inode, const char *name, + const void *value, size_t size, int flags) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + int ret; + + if (trans) + return do_setxattr(trans, inode, name, value, size, flags); + + ret = btrfs_reserve_metadata_space(root, 2); if (ret) - goto out; - mod = 1; + return ret; -out: - if (mod) { - inode->i_ctime = CURRENT_TIME; - ret = btrfs_update_inode(trans, root, inode); + trans = btrfs_start_transaction(root, 1); + if (!trans) { + ret = -ENOMEM; + goto out; } + btrfs_set_trans_block_group(trans, inode); - btrfs_end_transaction(trans, root); - btrfs_free_path(path); + ret = do_setxattr(trans, inode, name, value, size, flags); + if (ret) + goto out; + + inode->i_ctime = CURRENT_TIME; + ret = btrfs_update_inode(trans, root, inode); + BUG_ON(ret); +out: + btrfs_end_transaction_throttle(trans, root); + btrfs_unreserve_metadata_space(root, 2); return ret; } @@ -260,7 +283,7 @@ err: * attributes are handled directly. */ struct xattr_handler *btrfs_xattr_handlers[] = { -#ifdef CONFIG_FS_POSIX_ACL +#ifdef CONFIG_BTRFS_FS_POSIX_ACL &btrfs_xattr_acl_access_handler, &btrfs_xattr_acl_default_handler, #endif @@ -314,7 +337,9 @@ int btrfs_setxattr(struct dentry *dentry, const char *name, const void *value, if (size == 0) value = ""; /* empty EA, do not remove */ - return __btrfs_setxattr(dentry->d_inode, name, value, size, flags); + + return __btrfs_setxattr(NULL, dentry->d_inode, name, value, size, + flags); } int btrfs_removexattr(struct dentry *dentry, const char *name) @@ -329,10 +354,13 @@ int btrfs_removexattr(struct dentry *dentry, const char *name) if (!btrfs_is_valid_xattr(name)) return -EOPNOTSUPP; - return __btrfs_setxattr(dentry->d_inode, name, NULL, 0, XATTR_REPLACE); + + return __btrfs_setxattr(NULL, dentry->d_inode, name, NULL, 0, + XATTR_REPLACE); } -int btrfs_xattr_security_init(struct inode *inode, struct inode *dir) +int btrfs_xattr_security_init(struct btrfs_trans_handle *trans, + struct inode *inode, struct inode *dir) { int err; size_t len; @@ -354,7 +382,7 @@ int btrfs_xattr_security_init(struct inode *inode, struct inode *dir) } else { strcpy(name, XATTR_SECURITY_PREFIX); strcpy(name + XATTR_SECURITY_PREFIX_LEN, suffix); - err = __btrfs_setxattr(inode, name, value, len, 0); + err = __btrfs_setxattr(trans, inode, name, value, len, 0); kfree(name); } diff --git a/fs/btrfs/xattr.h b/fs/btrfs/xattr.h index c71e9c3cf3f7..721efa0346e0 100644 --- a/fs/btrfs/xattr.h +++ b/fs/btrfs/xattr.h @@ -27,15 +27,16 @@ extern struct xattr_handler *btrfs_xattr_handlers[]; extern ssize_t __btrfs_getxattr(struct inode *inode, const char *name, void *buffer, size_t size); -extern int __btrfs_setxattr(struct inode *inode, const char *name, - const void *value, size_t size, int flags); - +extern int __btrfs_setxattr(struct btrfs_trans_handle *trans, + struct inode *inode, const char *name, + const void *value, size_t size, int flags); extern ssize_t btrfs_getxattr(struct dentry *dentry, const char *name, void *buffer, size_t size); extern int btrfs_setxattr(struct dentry *dentry, const char *name, const void *value, size_t size, int flags); extern int btrfs_removexattr(struct dentry *dentry, const char *name); -extern int btrfs_xattr_security_init(struct inode *inode, struct inode *dir); +extern int btrfs_xattr_security_init(struct btrfs_trans_handle *trans, + struct inode *inode, struct inode *dir); #endif /* __XATTR__ */ |