Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  btrfs: rename the option to nospace_cache
  Btrfs: handle bio_add_page failure gracefully in scrub
  Btrfs: fix deadlock caused by the race between relocation
  Btrfs: only map pages if we know we need them when reading the space cache
  Btrfs: fix orphan backref nodes
  Btrfs: Abstract similar code for btrfs_block_rsv_add{, _noflush}
  Btrfs: fix unreleased path in btrfs_orphan_cleanup()
  Btrfs: fix no reserved space for writing out inode cache
  Btrfs: fix nocow when deleting the item
  Btrfs: tweak the delayed inode reservations again
  Btrfs: rework error handling in btrfs_mount()
  Btrfs: close devices on all error paths in open_ctree()
  Btrfs: avoid null dereference and leaks when bailing from open_ctree()
  Btrfs: fix subvol_name leak on error in btrfs_mount()
  Btrfs: fix memory leak in btrfs_parse_early_options()
  Btrfs: fix our reservations for updating an inode when completing io
  Btrfs: fix oops on NULL trans handle in btrfs_truncate
  btrfs: fix double-free 'tree_root' in 'btrfs_mount()'

fs/btrfs/btrfs_inode.h

@@ -147,14 +147,12 @@ struct btrfs_inode {
 	 * the btrfs file release call will add this inode to the
 	 * ordered operations list so that we make sure to flush out any
 	 * new data the application may have written before commit.
-	 *
-	 * yes, its silly to have a single bitflag, but we might grow more
-	 * of these.
 	 */
 	unsigned ordered_data_close:1;
 	unsigned orphan_meta_reserved:1;
 	unsigned dummy_inode:1;
 	unsigned in_defrag:1;
+	unsigned delalloc_meta_reserved:1;
 
 	/*
 	 * always compress this one file

fs/btrfs/delayed-inode.c

@@ -617,12 +617,14 @@ static void btrfs_delayed_item_release_metadata(struct btrfs_root *root,
 static int btrfs_delayed_inode_reserve_metadata(
 					struct btrfs_trans_handle *trans,
 					struct btrfs_root *root,
+					struct inode *inode,
 					struct btrfs_delayed_node *node)
 {
 	struct btrfs_block_rsv *src_rsv;
 	struct btrfs_block_rsv *dst_rsv;
 	u64 num_bytes;
 	int ret;
+	int release = false;
 
 	src_rsv = trans->block_rsv;
 	dst_rsv = &root->fs_info->delayed_block_rsv;
@@ -652,12 +654,65 @@ static int btrfs_delayed_inode_reserve_metadata(
 		if (!ret)
 			node->bytes_reserved = num_bytes;
 		return ret;
+	} else if (src_rsv == &root->fs_info->delalloc_block_rsv) {
+		spin_lock(&BTRFS_I(inode)->lock);
+		if (BTRFS_I(inode)->delalloc_meta_reserved) {
+			BTRFS_I(inode)->delalloc_meta_reserved = 0;
+			spin_unlock(&BTRFS_I(inode)->lock);
+			release = true;
+			goto migrate;
 		}
+		spin_unlock(&BTRFS_I(inode)->lock);
+
+		/* Ok we didn't have space pre-reserved.  This shouldn't happen
+		 * too often but it can happen if we do delalloc to an existing
+		 * inode which gets dirtied because of the time update, and then
+		 * isn't touched again until after the transaction commits and
+		 * then we try to write out the data.  First try to be nice and
+		 * reserve something strictly for us.  If not be a pain and try
+		 * to steal from the delalloc block rsv.
+		 */
+		ret = btrfs_block_rsv_add_noflush(root, dst_rsv, num_bytes);
+		if (!ret)
+			goto out;
 
 		ret = btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes);
+		if (!ret)
+			goto out;
+
+		/*
+		 * Ok this is a problem, let's just steal from the global rsv
+		 * since this really shouldn't happen that often.
+		 */
+		WARN_ON(1);
+		ret = btrfs_block_rsv_migrate(&root->fs_info->global_block_rsv,
+					      dst_rsv, num_bytes);
+		goto out;
+	}
+
+migrate:
+	ret = btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes);
+
+out:
+	/*
+	 * Migrate only takes a reservation, it doesn't touch the size of the
+	 * block_rsv.  This is to simplify people who don't normally have things
+	 * migrated from their block rsv.  If they go to release their
+	 * reservation, that will decrease the size as well, so if migrate
+	 * reduced size we'd end up with a negative size.  But for the
+	 * delalloc_meta_reserved stuff we will only know to drop 1 reservation,
+	 * but we could in fact do this reserve/migrate dance several times
+	 * between the time we did the original reservation and we'd clean it
+	 * up.  So to take care of this, release the space for the meta
+	 * reservation here.  I think it may be time for a documentation page on
+	 * how block rsvs. work.
+	 */
 	if (!ret)
 		node->bytes_reserved = num_bytes;
 
+	if (release)
+		btrfs_block_rsv_release(root, src_rsv, num_bytes);
+
 	return ret;
 }
 
@@ -1708,7 +1763,8 @@ int btrfs_delayed_update_inode(struct btrfs_trans_handle *trans,
 		goto release_node;
 	}
 
-	ret = btrfs_delayed_inode_reserve_metadata(trans, root, delayed_node);
+	ret = btrfs_delayed_inode_reserve_metadata(trans, root, inode,
+						   delayed_node);
 	if (ret)
 		goto release_node;
 

fs/btrfs/disk-io.c

@@ -1890,31 +1890,32 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	u64 features;
 	struct btrfs_key location;
 	struct buffer_head *bh;
-	struct btrfs_root *extent_root = kzalloc(sizeof(struct btrfs_root),
-						 GFP_NOFS);
-	struct btrfs_root *csum_root = kzalloc(sizeof(struct btrfs_root),
-						 GFP_NOFS);
+	struct btrfs_super_block *disk_super;
 	struct btrfs_root *tree_root = btrfs_sb(sb);
-	struct btrfs_fs_info *fs_info = NULL;
-	struct btrfs_root *chunk_root = kzalloc(sizeof(struct btrfs_root),
-						GFP_NOFS);
-	struct btrfs_root *dev_root = kzalloc(sizeof(struct btrfs_root),
-					      GFP_NOFS);
+	struct btrfs_fs_info *fs_info = tree_root->fs_info;
+	struct btrfs_root *extent_root;
+	struct btrfs_root *csum_root;
+	struct btrfs_root *chunk_root;
+	struct btrfs_root *dev_root;
 	struct btrfs_root *log_tree_root;
-
 	int ret;
 	int err = -EINVAL;
 	int num_backups_tried = 0;
 	int backup_index = 0;
 
-	struct btrfs_super_block *disk_super;
+	extent_root = fs_info->extent_root =
+		kzalloc(sizeof(struct btrfs_root), GFP_NOFS);
+	csum_root = fs_info->csum_root =
+		kzalloc(sizeof(struct btrfs_root), GFP_NOFS);
+	chunk_root = fs_info->chunk_root =
+		kzalloc(sizeof(struct btrfs_root), GFP_NOFS);
+	dev_root = fs_info->dev_root =
+		kzalloc(sizeof(struct btrfs_root), GFP_NOFS);
 
-	if (!extent_root || !tree_root || !tree_root->fs_info ||
-	    !chunk_root || !dev_root || !csum_root) {
+	if (!extent_root || !csum_root || !chunk_root || !dev_root) {
 		err = -ENOMEM;
 		goto fail;
 	}
-	fs_info = tree_root->fs_info;
 
 	ret = init_srcu_struct(&fs_info->subvol_srcu);
 	if (ret) {
@@ -1954,12 +1955,6 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	mutex_init(&fs_info->reloc_mutex);
 
 	init_completion(&fs_info->kobj_unregister);
-	fs_info->tree_root = tree_root;
-	fs_info->extent_root = extent_root;
-	fs_info->csum_root = csum_root;
-	fs_info->chunk_root = chunk_root;
-	fs_info->dev_root = dev_root;
-	fs_info->fs_devices = fs_devices;
 	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
 	INIT_LIST_HEAD(&fs_info->space_info);
 	btrfs_mapping_init(&fs_info->mapping_tree);
@@ -2465,21 +2460,20 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	btrfs_stop_workers(&fs_info->caching_workers);
 fail_alloc:
 fail_iput:
+	btrfs_mapping_tree_free(&fs_info->mapping_tree);
+
 	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
 	iput(fs_info->btree_inode);
-
-	btrfs_close_devices(fs_info->fs_devices);
-	btrfs_mapping_tree_free(&fs_info->mapping_tree);
 fail_bdi:
 	bdi_destroy(&fs_info->bdi);
 fail_srcu:
 	cleanup_srcu_struct(&fs_info->subvol_srcu);
 fail:
+	btrfs_close_devices(fs_info->fs_devices);
 	free_fs_info(fs_info);
 	return ERR_PTR(err);
 
 recovery_tree_root:
-
 	if (!btrfs_test_opt(tree_root, RECOVERY))
 		goto fail_tree_roots;
 

fs/btrfs/extent-tree.c

@@ -3797,16 +3797,16 @@ void btrfs_free_block_rsv(struct btrfs_root *root,
 	kfree(rsv);
 }
 
-int btrfs_block_rsv_add(struct btrfs_root *root,
+static inline int __block_rsv_add(struct btrfs_root *root,
 				  struct btrfs_block_rsv *block_rsv,
-			u64 num_bytes)
+				  u64 num_bytes, int flush)
 {
 	int ret;
 
 	if (num_bytes == 0)
 		return 0;
 
-	ret = reserve_metadata_bytes(root, block_rsv, num_bytes, 1);
+	ret = reserve_metadata_bytes(root, block_rsv, num_bytes, flush);
 	if (!ret) {
 		block_rsv_add_bytes(block_rsv, num_bytes, 1);
 		return 0;
@@ -3815,22 +3815,18 @@ int btrfs_block_rsv_add(struct btrfs_root *root,
 	return ret;
 }
 
-int btrfs_block_rsv_add_noflush(struct btrfs_root *root,
+int btrfs_block_rsv_add(struct btrfs_root *root,
 			struct btrfs_block_rsv *block_rsv,
 			u64 num_bytes)
 {
-	int ret;
-
-	if (num_bytes == 0)
-		return 0;
-
-	ret = reserve_metadata_bytes(root, block_rsv, num_bytes, 0);
-	if (!ret) {
-		block_rsv_add_bytes(block_rsv, num_bytes, 1);
-		return 0;
-	}
+	return __block_rsv_add(root, block_rsv, num_bytes, 1);
+}
 
-	return ret;
+int btrfs_block_rsv_add_noflush(struct btrfs_root *root,
+				struct btrfs_block_rsv *block_rsv,
+				u64 num_bytes)
+{
+	return __block_rsv_add(root, block_rsv, num_bytes, 0);
 }
 
 int btrfs_block_rsv_check(struct btrfs_root *root,
@@ -4064,23 +4060,30 @@ int btrfs_snap_reserve_metadata(struct btrfs_trans_handle *trans,
  */
 static unsigned drop_outstanding_extent(struct inode *inode)
 {
+	unsigned drop_inode_space = 0;
 	unsigned dropped_extents = 0;
 
 	BUG_ON(!BTRFS_I(inode)->outstanding_extents);
 	BTRFS_I(inode)->outstanding_extents--;
 
+	if (BTRFS_I(inode)->outstanding_extents == 0 &&
+	    BTRFS_I(inode)->delalloc_meta_reserved) {
+		drop_inode_space = 1;
+		BTRFS_I(inode)->delalloc_meta_reserved = 0;
+	}
+
 	/*
 	 * If we have more or the same amount of outsanding extents than we have
 	 * reserved then we need to leave the reserved extents count alone.
 	 */
 	if (BTRFS_I(inode)->outstanding_extents >=
 	    BTRFS_I(inode)->reserved_extents)
-		return 0;
+		return drop_inode_space;
 
 	dropped_extents = BTRFS_I(inode)->reserved_extents -
 		BTRFS_I(inode)->outstanding_extents;
 	BTRFS_I(inode)->reserved_extents -= dropped_extents;
-	return dropped_extents;
+	return dropped_extents + drop_inode_space;
 }
 
 /**
@@ -4166,9 +4169,18 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes)
 		nr_extents = BTRFS_I(inode)->outstanding_extents -
 			BTRFS_I(inode)->reserved_extents;
 		BTRFS_I(inode)->reserved_extents += nr_extents;
+	}
 
-		to_reserve = btrfs_calc_trans_metadata_size(root, nr_extents);
+	/*
+	 * Add an item to reserve for updating the inode when we complete the
+	 * delalloc io.
+	 */
+	if (!BTRFS_I(inode)->delalloc_meta_reserved) {
+		nr_extents++;
+		BTRFS_I(inode)->delalloc_meta_reserved = 1;
 	}
+
+	to_reserve = btrfs_calc_trans_metadata_size(root, nr_extents);
 	to_reserve += calc_csum_metadata_size(inode, num_bytes, 1);
 	spin_unlock(&BTRFS_I(inode)->lock);
 

fs/btrfs/free-space-cache.c

@@ -537,6 +537,13 @@ static int io_ctl_read_entry(struct io_ctl *io_ctl,
 			    struct btrfs_free_space *entry, u8 *type)
 {
 	struct btrfs_free_space_entry *e;
+	int ret;
+
+	if (!io_ctl->cur) {
+		ret = io_ctl_check_crc(io_ctl, io_ctl->index);
+		if (ret)
+			return ret;
+	}
 
 	e = io_ctl->cur;
 	entry->offset = le64_to_cpu(e->offset);
@@ -550,10 +557,7 @@ static int io_ctl_read_entry(struct io_ctl *io_ctl,
 
 	io_ctl_unmap_page(io_ctl);
 
-	if (io_ctl->index >= io_ctl->num_pages)
 	return 0;
-
-	return io_ctl_check_crc(io_ctl, io_ctl->index);
 }
 
 static int io_ctl_read_bitmap(struct io_ctl *io_ctl,
@@ -561,9 +565,6 @@ static int io_ctl_read_bitmap(struct io_ctl *io_ctl,
 {
 	int ret;
 
-	if (io_ctl->cur && io_ctl->cur != io_ctl->orig)
-		io_ctl_unmap_page(io_ctl);
-
 	ret = io_ctl_check_crc(io_ctl, io_ctl->index);
 	if (ret)
 		return ret;
@@ -699,6 +700,8 @@ int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
 		num_entries--;
 	}
 
+	io_ctl_unmap_page(&io_ctl);
+
 	/*
 	 * We add the bitmaps at the end of the entries in order that
 	 * the bitmap entries are added to the cache.

fs/btrfs/inode-map.c

@@ -398,6 +398,8 @@ int btrfs_save_ino_cache(struct btrfs_root *root,
 	struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
 	struct btrfs_path *path;
 	struct inode *inode;
+	struct btrfs_block_rsv *rsv;
+	u64 num_bytes;
 	u64 alloc_hint = 0;
 	int ret;
 	int prealloc;
@@ -421,11 +423,26 @@ int btrfs_save_ino_cache(struct btrfs_root *root,
 	if (!path)
 		return -ENOMEM;
 
+	rsv = trans->block_rsv;
+	trans->block_rsv = &root->fs_info->trans_block_rsv;
+
+	num_bytes = trans->bytes_reserved;
+	/*
+	 * 1 item for inode item insertion if need
+	 * 3 items for inode item update (in the worst case)
+	 * 1 item for free space object
+	 * 3 items for pre-allocation
+	 */
+	trans->bytes_reserved = btrfs_calc_trans_metadata_size(root, 8);
+	ret = btrfs_block_rsv_add_noflush(root, trans->block_rsv,
+					  trans->bytes_reserved);
+	if (ret)
+		goto out;
 again:
 	inode = lookup_free_ino_inode(root, path);
 	if (IS_ERR(inode) && PTR_ERR(inode) != -ENOENT) {
 		ret = PTR_ERR(inode);
-		goto out;
+		goto out_release;
 	}
 
 	if (IS_ERR(inode)) {
@@ -434,7 +451,7 @@ int btrfs_save_ino_cache(struct btrfs_root *root,
 
 		ret = create_free_ino_inode(root, trans, path);
 		if (ret)
-			goto out;
+			goto out_release;
 		goto again;
 	}
 
@@ -477,11 +494,14 @@ int btrfs_save_ino_cache(struct btrfs_root *root,
 	}
 	btrfs_free_reserved_data_space(inode, prealloc);
 
+	ret = btrfs_write_out_ino_cache(root, trans, path);
 out_put:
 	iput(inode);
+out_release:
+	btrfs_block_rsv_release(root, trans->block_rsv, trans->bytes_reserved);
 out:
-	if (ret == 0)
-		ret = btrfs_write_out_ino_cache(root, trans, path);
+	trans->block_rsv = rsv;
+	trans->bytes_reserved = num_bytes;
 
 	btrfs_free_path(path);
 	return ret;

fs/btrfs/inode.c

@@ -93,6 +93,8 @@ static noinline int cow_file_range(struct inode *inode,
 				   struct page *locked_page,
 				   u64 start, u64 end, int *page_started,
 				   unsigned long *nr_written, int unlock);
+static noinline int btrfs_update_inode_fallback(struct btrfs_trans_handle *trans,
+				struct btrfs_root *root, struct inode *inode);
 
 static int btrfs_init_inode_security(struct btrfs_trans_handle *trans,
 				     struct inode *inode,  struct inode *dir,
@@ -1741,7 +1743,7 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 				trans = btrfs_join_transaction(root);
 			BUG_ON(IS_ERR(trans));
 			trans->block_rsv = &root->fs_info->delalloc_block_rsv;
-			ret = btrfs_update_inode(trans, root, inode);
+			ret = btrfs_update_inode_fallback(trans, root, inode);
 			BUG_ON(ret);
 		}
 		goto out;
@@ -1791,7 +1793,7 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 
 	ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent);
 	if (!ret || !test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) {
-		ret = btrfs_update_inode(trans, root, inode);
+		ret = btrfs_update_inode_fallback(trans, root, inode);
 		BUG_ON(ret);
 	}
 	ret = 0;
@@ -2199,6 +2201,9 @@ int btrfs_orphan_cleanup(struct btrfs_root *root)
 		if (ret)
 			goto out;
 	}
+	/* release the path since we're done with it */
+	btrfs_release_path(path);
+
 	root->orphan_cleanup_state = ORPHAN_CLEANUP_DONE;
 
 	if (root->orphan_block_rsv)
@@ -2426,7 +2431,7 @@ static void fill_inode_item(struct btrfs_trans_handle *trans,
 /*
  * copy everything in the in-memory inode into the btree.
  */
-noinline int btrfs_update_inode(struct btrfs_trans_handle *trans,
+static noinline int btrfs_update_inode_item(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root, struct inode *inode)
 {
 	struct btrfs_inode_item *inode_item;
@@ -2434,21 +2439,6 @@ noinline int btrfs_update_inode(struct btrfs_trans_handle *trans,
 	struct extent_buffer *leaf;
 	int ret;
 
-	/*
-	 * If the inode is a free space inode, we can deadlock during commit
-	 * if we put it into the delayed code.
-	 *
-	 * The data relocation inode should also be directly updated
-	 * without delay
-	 */
-	if (!btrfs_is_free_space_inode(root, inode)
-	    && root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID) {
-		ret = btrfs_delayed_update_inode(trans, root, inode);
-		if (!ret)
-			btrfs_set_inode_last_trans(trans, inode);
-		return ret;
-	}
-
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
@@ -2476,6 +2466,43 @@ noinline int btrfs_update_inode(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
+/*
+ * copy everything in the in-memory inode into the btree.
+ */
+noinline int btrfs_update_inode(struct btrfs_trans_handle *trans,
+				struct btrfs_root *root, struct inode *inode)
+{
+	int ret;
+
+	/*
+	 * If the inode is a free space inode, we can deadlock during commit
+	 * if we put it into the delayed code.
+	 *
+	 * The data relocation inode should also be directly updated
+	 * without delay
+	 */
+	if (!btrfs_is_free_space_inode(root, inode)
+	    && root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID) {
+		ret = btrfs_delayed_update_inode(trans, root, inode);
+		if (!ret)
+			btrfs_set_inode_last_trans(trans, inode);
+		return ret;
+	}
+
+	return btrfs_update_inode_item(trans, root, inode);
+}
+
+static noinline int btrfs_update_inode_fallback(struct btrfs_trans_handle *trans,
+				struct btrfs_root *root, struct inode *inode)
+{
+	int ret;
+
+	ret = btrfs_update_inode(trans, root, inode);
+	if (ret == -ENOSPC)
+		return btrfs_update_inode_item(trans, root, inode);
+	return ret;
+}
+
 /*
  * unlink helper that gets used here in inode.c and in the tree logging
  * recovery code.  It remove a link in a directory with a given name, and
@@ -5632,7 +5659,7 @@ static void btrfs_endio_direct_write(struct bio *bio, int err)
 	if (test_bit(BTRFS_ORDERED_NOCOW, &ordered->flags)) {
 		ret = btrfs_ordered_update_i_size(inode, 0, ordered);
 		if (!ret)
-			err = btrfs_update_inode(trans, root, inode);
+			err = btrfs_update_inode_fallback(trans, root, inode);
 		goto out;
 	}
 
@@ -5670,7 +5697,7 @@ static void btrfs_endio_direct_write(struct bio *bio, int err)
 	add_pending_csums(trans, inode, ordered->file_offset, &ordered->list);
 	ret = btrfs_ordered_update_i_size(inode, 0, ordered);
 	if (!ret || !test_bit(BTRFS_ORDERED_PREALLOC, &ordered->flags))
-		btrfs_update_inode(trans, root, inode);
+		btrfs_update_inode_fallback(trans, root, inode);
 	ret = 0;
 out_unlock:
 	unlock_extent_cached(&BTRFS_I(inode)->io_tree, ordered->file_offset,
@@ -6529,6 +6556,7 @@ static int btrfs_truncate(struct inode *inode)
 		ret = btrfs_orphan_del(NULL, inode);
 	}
 
+	if (trans) {
 		trans->block_rsv = &root->fs_info->trans_block_rsv;
 		ret = btrfs_update_inode(trans, root, inode);
 		if (ret && !err)
@@ -6537,6 +6565,7 @@ static int btrfs_truncate(struct inode *inode)
 		nr = trans->blocks_used;
 		ret = btrfs_end_transaction_throttle(trans, root);
 		btrfs_btree_balance_dirty(root, nr);
+	}
 
 out:
 	btrfs_free_block_rsv(root, rsv);
@@ -6605,6 +6634,7 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
 	ei->orphan_meta_reserved = 0;
 	ei->dummy_inode = 0;
 	ei->in_defrag = 0;
+	ei->delalloc_meta_reserved = 0;
 	ei->force_compress = BTRFS_COMPRESS_NONE;
 
 	ei->delayed_node = NULL;

fs/btrfs/relocation.c

@@ -1174,6 +1174,8 @@ static int clone_backref_node(struct btrfs_trans_handle *trans,
 			list_add_tail(&new_edge->list[UPPER],
 				      &new_node->lower);
 		}
+	} else {
+		list_add_tail(&new_node->lower, &cache->leaves);
 	}
 
 	rb_node = tree_insert(&cache->rb_root, new_node->bytenr,

fs/btrfs/scrub.c

@@ -944,50 +944,18 @@ static int scrub_checksum_super(struct scrub_bio *sbio, void *buffer)
 static int scrub_submit(struct scrub_dev *sdev)
 {
 	struct scrub_bio *sbio;
-	struct bio *bio;
-	int i;
 
 	if (sdev->curr == -1)
 		return 0;
 
 	sbio = sdev->bios[sdev->curr];
-
-	bio = bio_alloc(GFP_NOFS, sbio->count);
-	if (!bio)
-		goto nomem;
-
-	bio->bi_private = sbio;
-	bio->bi_end_io = scrub_bio_end_io;
-	bio->bi_bdev = sdev->dev->bdev;
-	bio->bi_sector = sbio->physical >> 9;
-
-	for (i = 0; i < sbio->count; ++i) {
-		struct page *page;
-		int ret;
-
-		page = alloc_page(GFP_NOFS);
-		if (!page)
-			goto nomem;
-
-		ret = bio_add_page(bio, page, PAGE_SIZE, 0);
-		if (!ret) {
-			__free_page(page);
-			goto nomem;
-		}
-	}
-
 	sbio->err = 0;
 	sdev->curr = -1;
 	atomic_inc(&sdev->in_flight);
 
-	submit_bio(READ, bio);
+	submit_bio(READ, sbio->bio);
 
 	return 0;
-
-nomem:
-	scrub_free_bio(bio);
-
-	return -ENOMEM;
 }
 
 static int scrub_page(struct scrub_dev *sdev, u64 logical, u64 len,
@@ -995,6 +963,8 @@ static int scrub_page(struct scrub_dev *sdev, u64 logical, u64 len,
 		      u8 *csum, int force)
 {
 	struct scrub_bio *sbio;
+	struct page *page;
+	int ret;
 
 again:
 	/*
@@ -1015,12 +985,22 @@ static int scrub_page(struct scrub_dev *sdev, u64 logical, u64 len,
 	}
 	sbio = sdev->bios[sdev->curr];
 	if (sbio->count == 0) {
+		struct bio *bio;
+
 		sbio->physical = physical;
 		sbio->logical = logical;
+		bio = bio_alloc(GFP_NOFS, SCRUB_PAGES_PER_BIO);
+		if (!bio)
+			return -ENOMEM;
+
+		bio->bi_private = sbio;
+		bio->bi_end_io = scrub_bio_end_io;
+		bio->bi_bdev = sdev->dev->bdev;
+		bio->bi_sector = sbio->physical >> 9;
+		sbio->err = 0;
+		sbio->bio = bio;
 	} else if (sbio->physical + sbio->count * PAGE_SIZE != physical ||
 		   sbio->logical + sbio->count * PAGE_SIZE != logical) {
-		int ret;
-
 		ret = scrub_submit(sdev);
 		if (ret)
 			return ret;
@@ -1030,6 +1010,20 @@ static int scrub_page(struct scrub_dev *sdev, u64 logical, u64 len,
 	sbio->spag[sbio->count].generation = gen;
 	sbio->spag[sbio->count].have_csum = 0;
 	sbio->spag[sbio->count].mirror_num = mirror_num;
+
+	page = alloc_page(GFP_NOFS);
+	if (!page)
+		return -ENOMEM;
+
+	ret = bio_add_page(sbio->bio, page, PAGE_SIZE, 0);
+	if (!ret) {
+		__free_page(page);
+		ret = scrub_submit(sdev);
+		if (ret)
+			return ret;
+		goto again;
+	}
+
 	if (csum) {
 		sbio->spag[sbio->count].have_csum = 1;
 		memcpy(sbio->spag[sbio->count].csum, csum, sdev->csum_size);

fs/btrfs/super.c

@@ -197,7 +197,7 @@ static match_table_t tokens = {
 	{Opt_subvolrootid, "subvolrootid=%d"},
 	{Opt_defrag, "autodefrag"},
 	{Opt_inode_cache, "inode_cache"},
-	{Opt_no_space_cache, "no_space_cache"},
+	{Opt_no_space_cache, "nospace_cache"},
 	{Opt_recovery, "recovery"},
 	{Opt_err, NULL},
 };
@@ -448,6 +448,7 @@ static int btrfs_parse_early_options(const char *options, fmode_t flags,
 		token = match_token(p, tokens, args);
 		switch (token) {
 		case Opt_subvol:
+			kfree(*subvol_name);
 			*subvol_name = match_strdup(&args[0]);
 			break;
 		case Opt_subvolid:
@@ -710,7 +711,7 @@ static int btrfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
 	if (btrfs_test_opt(root, SPACE_CACHE))
 		seq_puts(seq, ",space_cache");
 	else
-		seq_puts(seq, ",no_space_cache");
+		seq_puts(seq, ",nospace_cache");
 	if (btrfs_test_opt(root, CLEAR_CACHE))
 		seq_puts(seq, ",clear_cache");
 	if (btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED))
@@ -890,7 +891,6 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
 	struct super_block *s;
 	struct dentry *root;
 	struct btrfs_fs_devices *fs_devices = NULL;
-	struct btrfs_root *tree_root = NULL;
 	struct btrfs_fs_info *fs_info = NULL;
 	fmode_t mode = FMODE_READ;
 	char *subvol_name = NULL;
@@ -904,8 +904,10 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
 	error = btrfs_parse_early_options(data, mode, fs_type,
 					  &subvol_name, &subvol_objectid,
 					  &subvol_rootid, &fs_devices);
-	if (error)
+	if (error) {
+		kfree(subvol_name);
 		return ERR_PTR(error);
+	}
 
 	if (subvol_name) {
 		root = mount_subvol(subvol_name, flags, device_name, data);
@@ -917,15 +919,6 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
 	if (error)
 		return ERR_PTR(error);
 
-	error = btrfs_open_devices(fs_devices, mode, fs_type);
-	if (error)
-		return ERR_PTR(error);
-
-	if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
-		error = -EACCES;
-		goto error_close_devices;
-	}
-
 	/*
 	 * Setup a dummy root and fs_info for test/set super.  This is because
 	 * we don't actually fill this stuff out until open_ctree, but we need
@@ -933,24 +926,36 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
 	 * then open_ctree will properly initialize everything later.
 	 */
 	fs_info = kzalloc(sizeof(struct btrfs_fs_info), GFP_NOFS);
-	tree_root = kzalloc(sizeof(struct btrfs_root), GFP_NOFS);
-	if (!fs_info || !tree_root) {
+	if (!fs_info)
+		return ERR_PTR(-ENOMEM);
+
+	fs_info->tree_root = kzalloc(sizeof(struct btrfs_root), GFP_NOFS);
+	if (!fs_info->tree_root) {
 		error = -ENOMEM;
-		goto error_close_devices;
+		goto error_fs_info;
 	}
-	fs_info->tree_root = tree_root;
+	fs_info->tree_root->fs_info = fs_info;
 	fs_info->fs_devices = fs_devices;
-	tree_root->fs_info = fs_info;
 
 	fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
 	fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
 	if (!fs_info->super_copy || !fs_info->super_for_commit) {
 		error = -ENOMEM;
+		goto error_fs_info;
+	}
+
+	error = btrfs_open_devices(fs_devices, mode, fs_type);
+	if (error)
+		goto error_fs_info;
+
+	if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
+		error = -EACCES;
 		goto error_close_devices;
 	}
 
 	bdev = fs_devices->latest_bdev;
-	s = sget(fs_type, btrfs_test_super, btrfs_set_super, tree_root);
+	s = sget(fs_type, btrfs_test_super, btrfs_set_super,
+		 fs_info->tree_root);
 	if (IS_ERR(s)) {
 		error = PTR_ERR(s);
 		goto error_close_devices;
@@ -959,12 +964,12 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
 	if (s->s_root) {
 		if ((flags ^ s->s_flags) & MS_RDONLY) {
 			deactivate_locked_super(s);
-			return ERR_PTR(-EBUSY);
+			error = -EBUSY;
+			goto error_close_devices;
 		}
 
 		btrfs_close_devices(fs_devices);
 		free_fs_info(fs_info);
-		kfree(tree_root);
 	} else {
 		char b[BDEVNAME_SIZE];
 
@@ -991,8 +996,8 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
 
 error_close_devices:
 	btrfs_close_devices(fs_devices);
+error_fs_info:
 	free_fs_info(fs_info);
-	kfree(tree_root);
 	return ERR_PTR(error);
 }
 

fs/btrfs/transaction.c

@@ -882,7 +882,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	btrfs_reloc_pre_snapshot(trans, pending, &to_reserve);
 
 	if (to_reserve > 0) {
-		ret = btrfs_block_rsv_add(root, &pending->block_rsv,
+		ret = btrfs_block_rsv_add_noflush(root, &pending->block_rsv,
 						  to_reserve);
 		if (ret) {
 			pending->error = ret;

fs/btrfs/volumes.c

@@ -999,7 +999,7 @@ static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans,
 	key.objectid = device->devid;
 	key.offset = start;
 	key.type = BTRFS_DEV_EXTENT_KEY;
-
+again:
 	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
 	if (ret > 0) {
 		ret = btrfs_previous_item(root, path, key.objectid,
@@ -1012,6 +1012,9 @@ static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans,
 					struct btrfs_dev_extent);
 		BUG_ON(found_key.offset > start || found_key.offset +
 		       btrfs_dev_extent_length(leaf, extent) < start);
+		key = found_key;
+		btrfs_release_path(path);
+		goto again;
 	} else if (ret == 0) {
 		leaf = path->nodes[0];
 		extent = btrfs_item_ptr(leaf, path->slots[0],