1 files changed, 174 insertions, 127 deletions

diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 3388c836b9a5..05ee4391c83a 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -32,6 +32,8 @@
 #include "ioctl.h"
 #include "relocation.h"
 #include "scrub.h"
+#include "ordered-data.h"
+#include "delayed-inode.h"
 
 static struct kmem_cache *btrfs_trans_handle_cachep;
 
@@ -103,7 +105,7 @@ static struct kmem_cache *btrfs_trans_handle_cachep;
  * | attached to transid N+1.			    |
  * |						    |
  * | To next stage:				    |
- * |  Until all tree blocks are super blocks are    |
+ * |  Until all tree blocks and super blocks are    |
  * |  written to block devices			    |
  * V						    |
  * Transaction N [[TRANS_STATE_COMPLETED]]	    V
@@ -138,13 +140,10 @@ static const unsigned int btrfs_blocked_trans_types[TRANS_STATE_MAX] = {
 
 void btrfs_put_transaction(struct btrfs_transaction *transaction)
 {
-	WARN_ON(refcount_read(&transaction->use_count) == 0);
 	if (refcount_dec_and_test(&transaction->use_count)) {
 		BUG_ON(!list_empty(&transaction->list));
-		WARN_ON(!RB_EMPTY_ROOT(
-				&transaction->delayed_refs.href_root.rb_root));
-		WARN_ON(!RB_EMPTY_ROOT(
-				&transaction->delayed_refs.dirty_extent_root));
+		WARN_ON(!xa_empty(&transaction->delayed_refs.head_refs));
+		WARN_ON(!xa_empty(&transaction->delayed_refs.dirty_extents));
 		if (transaction->delayed_refs.pending_csums)
 			btrfs_err(transaction->fs_info,
 				  "pending csums is %llu",
@@ -162,7 +161,13 @@ void btrfs_put_transaction(struct btrfs_transaction *transaction)
 			cache = list_first_entry(&transaction->deleted_bgs,
 						 struct btrfs_block_group,
 						 bg_list);
+			/*
+			 * Not strictly necessary to lock, as no other task will be using a
+			 * block_group on the deleted_bgs list during a transaction abort.
+			 */
+			spin_lock(&transaction->fs_info->unused_bgs_lock);
 			list_del_init(&cache->bg_list);
+			spin_unlock(&transaction->fs_info->unused_bgs_lock);
 			btrfs_unfreeze_block_group(cache);
 			btrfs_put_block_group(cache);
 		}
@@ -181,7 +186,8 @@ static noinline void switch_commit_roots(struct btrfs_trans_handle *trans)
 	 * At this point no one can be using this transaction to modify any tree
 	 * and no one can start another transaction to modify any tree either.
 	 */
-	ASSERT(cur_trans->state == TRANS_STATE_COMMIT_DOING);
+	ASSERT(cur_trans->state == TRANS_STATE_COMMIT_DOING,
+	       "cur_trans->state=%d", cur_trans->state);
 
 	down_write(&fs_info->commit_root_sem);
 
@@ -193,7 +199,7 @@ static noinline void switch_commit_roots(struct btrfs_trans_handle *trans)
 		list_del_init(&root->dirty_list);
 		free_extent_buffer(root->commit_root);
 		root->commit_root = btrfs_root_node(root);
-		extent_io_tree_release(&root->dirty_log_pages);
+		btrfs_extent_io_tree_release(&root->dirty_log_pages);
 		btrfs_qgroup_clean_swapped_blocks(root);
 	}
 
@@ -276,8 +282,10 @@ loop:
 	cur_trans = fs_info->running_transaction;
 	if (cur_trans) {
 		if (TRANS_ABORTED(cur_trans)) {
+			const int abort_error = cur_trans->aborted;
+
 			spin_unlock(&fs_info->trans_lock);
-			return cur_trans->aborted;
+			return abort_error;
 		}
 		if (btrfs_blocked_trans_types[cur_trans->state] & type) {
 			spin_unlock(&fs_info->trans_lock);
@@ -350,9 +358,8 @@ loop:
 
 	memset(&cur_trans->delayed_refs, 0, sizeof(cur_trans->delayed_refs));
 
-	cur_trans->delayed_refs.href_root = RB_ROOT_CACHED;
-	cur_trans->delayed_refs.dirty_extent_root = RB_ROOT;
-	atomic_set(&cur_trans->delayed_refs.num_entries, 0);
+	xa_init(&cur_trans->delayed_refs.head_refs);
+	xa_init(&cur_trans->delayed_refs.dirty_extents);
 
 	/*
 	 * although the tree mod log is per file system and not per transaction,
@@ -378,10 +385,10 @@ loop:
 	INIT_LIST_HEAD(&cur_trans->deleted_bgs);
 	spin_lock_init(&cur_trans->dropped_roots_lock);
 	list_add_tail(&cur_trans->list, &fs_info->trans_list);
-	extent_io_tree_init(fs_info, &cur_trans->dirty_pages,
-			IO_TREE_TRANS_DIRTY_PAGES);
-	extent_io_tree_init(fs_info, &cur_trans->pinned_extents,
-			IO_TREE_FS_PINNED_EXTENTS);
+	btrfs_extent_io_tree_init(fs_info, &cur_trans->dirty_pages,
+				  IO_TREE_TRANS_DIRTY_PAGES);
+	btrfs_extent_io_tree_init(fs_info, &cur_trans->pinned_extents,
+				  IO_TREE_FS_PINNED_EXTENTS);
 	btrfs_set_fs_generation(fs_info, fs_info->generation + 1);
 	cur_trans->transid = fs_info->generation;
 	fs_info->running_transaction = cur_trans;
@@ -399,13 +406,13 @@ loop:
  */
 static int record_root_in_trans(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root,
-			       int force)
+			       bool force)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	int ret = 0;
 
 	if ((test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
-	    root->last_trans < trans->transid) || force) {
+	    btrfs_get_root_last_trans(root) < trans->transid) || force) {
 		WARN_ON(!force && root->commit_root != root->node);
 
 		/*
@@ -421,7 +428,7 @@ static int record_root_in_trans(struct btrfs_trans_handle *trans,
 		smp_wmb();
 
 		spin_lock(&fs_info->fs_roots_radix_lock);
-		if (root->last_trans == trans->transid && !force) {
+		if (btrfs_get_root_last_trans(root) == trans->transid && !force) {
 			spin_unlock(&fs_info->fs_roots_radix_lock);
 			return 0;
 		}
@@ -429,7 +436,7 @@ static int record_root_in_trans(struct btrfs_trans_handle *trans,
 				   (unsigned long)btrfs_root_id(root),
 				   BTRFS_ROOT_TRANS_TAG);
 		spin_unlock(&fs_info->fs_roots_radix_lock);
-		root->last_trans = trans->transid;
+		btrfs_set_root_last_trans(root, trans->transid);
 
 		/* this is pretty tricky.  We don't want to
 		 * take the relocation lock in btrfs_record_root_in_trans
@@ -491,7 +498,7 @@ int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
 	 * and barriers
 	 */
 	smp_rmb();
-	if (root->last_trans == trans->transid &&
+	if (btrfs_get_root_last_trans(root) == trans->transid &&
 	    !test_bit(BTRFS_ROOT_IN_TRANS_SETUP, &root->state))
 		return 0;
 
@@ -533,15 +540,15 @@ static void wait_current_trans(struct btrfs_fs_info *fs_info)
 	}
 }
 
-static int may_wait_transaction(struct btrfs_fs_info *fs_info, int type)
+static bool may_wait_transaction(struct btrfs_fs_info *fs_info, int type)
 {
 	if (test_bit(BTRFS_FS_LOG_RECOVERING, &fs_info->flags))
-		return 0;
+		return false;
 
 	if (type == TRANS_START)
-		return 1;
+		return true;
 
-	return 0;
+	return false;
 }
 
 static inline bool need_reserve_reloc_root(struct btrfs_root *root)
@@ -570,7 +577,7 @@ static int btrfs_reserve_trans_metadata(struct btrfs_fs_info *fs_info,
 	 * We want to reserve all the bytes we may need all at once, so we only
 	 * do 1 enospc flushing cycle per transaction start.
 	 */
-	ret = btrfs_reserve_metadata_bytes(fs_info, si, bytes, flush);
+	ret = btrfs_reserve_metadata_bytes(si, bytes, flush);
 
 	/*
 	 * If we are an emergency flush, which can steal from the global block
@@ -580,7 +587,7 @@ static int btrfs_reserve_trans_metadata(struct btrfs_fs_info *fs_info,
 	if (ret && flush == BTRFS_RESERVE_FLUSH_ALL_STEAL) {
 		bytes -= *delayed_refs_bytes;
 		*delayed_refs_bytes = 0;
-		ret = btrfs_reserve_metadata_bytes(fs_info, si, bytes, flush);
+		ret = btrfs_reserve_metadata_bytes(si, bytes, flush);
 	}
 
 	return ret;
@@ -756,9 +763,10 @@ got_it:
 	 * value here.
 	 */
 	if (do_chunk_alloc && num_bytes) {
-		u64 flags = h->block_rsv->space_info->flags;
+		struct btrfs_space_info *space_info = h->block_rsv->space_info;
+		u64 flags = space_info->flags;
 
-		btrfs_chunk_alloc(h, btrfs_get_alloc_profile(fs_info, flags),
+		btrfs_chunk_alloc(h, space_info, btrfs_get_alloc_profile(fs_info, flags),
 				  CHUNK_ALLOC_NO_FORCE);
 	}
 
@@ -798,8 +806,7 @@ alloc_fail:
 	if (num_bytes)
 		btrfs_block_rsv_release(fs_info, trans_rsv, num_bytes, NULL);
 	if (delayed_refs_bytes)
-		btrfs_space_info_free_bytes_may_use(fs_info, trans_rsv->space_info,
-						    delayed_refs_bytes);
+		btrfs_space_info_free_bytes_may_use(trans_rsv->space_info, delayed_refs_bytes);
 reserve_fail:
 	btrfs_qgroup_free_meta_prealloc(root, qgroup_reserved);
 	return ERR_PTR(ret);
@@ -1019,13 +1026,18 @@ static void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans)
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 
 	if (!trans->block_rsv) {
-		ASSERT(!trans->bytes_reserved);
-		ASSERT(!trans->delayed_refs_bytes_reserved);
+		ASSERT(trans->bytes_reserved == 0,
+		       "trans->bytes_reserved=%llu", trans->bytes_reserved);
+		ASSERT(trans->delayed_refs_bytes_reserved == 0,
+		       "trans->delayed_refs_bytes_reserved=%llu",
+		       trans->delayed_refs_bytes_reserved);
 		return;
 	}
 
 	if (!trans->bytes_reserved) {
-		ASSERT(!trans->delayed_refs_bytes_reserved);
+		ASSERT(trans->delayed_refs_bytes_reserved == 0,
+		       "trans->delayed_refs_bytes_reserved=%llu",
+		       trans->delayed_refs_bytes_reserved);
 		return;
 	}
 
@@ -1124,13 +1136,13 @@ int btrfs_write_marked_extents(struct btrfs_fs_info *fs_info,
 	u64 start = 0;
 	u64 end;
 
-	while (find_first_extent_bit(dirty_pages, start, &start, &end,
-				     mark, &cached_state)) {
+	while (btrfs_find_first_extent_bit(dirty_pages, start, &start, &end,
+					   mark, &cached_state)) {
 		bool wait_writeback = false;
 
-		ret = convert_extent_bit(dirty_pages, start, end,
-					 EXTENT_NEED_WAIT,
-					 mark, &cached_state);
+		ret = btrfs_convert_extent_bit(dirty_pages, start, end,
+					       EXTENT_NEED_WAIT,
+					       mark, &cached_state);
 		/*
 		 * convert_extent_bit can return -ENOMEM, which is most of the
 		 * time a temporary error. So when it happens, ignore the error
@@ -1151,8 +1163,8 @@ int btrfs_write_marked_extents(struct btrfs_fs_info *fs_info,
 		if (!ret)
 			ret = filemap_fdatawrite_range(mapping, start, end);
 		if (!ret && wait_writeback)
-			ret = filemap_fdatawait_range(mapping, start, end);
-		free_extent_state(cached_state);
+			btrfs_btree_wait_writeback_range(fs_info, start, end);
+		btrfs_free_extent_state(cached_state);
 		if (ret)
 			break;
 		cached_state = NULL;
@@ -1171,14 +1183,13 @@ int btrfs_write_marked_extents(struct btrfs_fs_info *fs_info,
 static int __btrfs_wait_marked_extents(struct btrfs_fs_info *fs_info,
 				       struct extent_io_tree *dirty_pages)
 {
-	struct address_space *mapping = fs_info->btree_inode->i_mapping;
 	struct extent_state *cached_state = NULL;
 	u64 start = 0;
 	u64 end;
 	int ret = 0;
 
-	while (find_first_extent_bit(dirty_pages, start, &start, &end,
-				     EXTENT_NEED_WAIT, &cached_state)) {
+	while (btrfs_find_first_extent_bit(dirty_pages, start, &start, &end,
+					   EXTENT_NEED_WAIT, &cached_state)) {
 		/*
 		 * Ignore -ENOMEM errors returned by clear_extent_bit().
 		 * When committing the transaction, we'll remove any entries
@@ -1187,13 +1198,13 @@ static int __btrfs_wait_marked_extents(struct btrfs_fs_info *fs_info,
 		 * concurrently - we do it only at transaction commit time when
 		 * it's safe to do it (through extent_io_tree_release()).
 		 */
-		ret = clear_extent_bit(dirty_pages, start, end,
-				       EXTENT_NEED_WAIT, &cached_state);
+		ret = btrfs_clear_extent_bit(dirty_pages, start, end,
+					     EXTENT_NEED_WAIT, &cached_state);
 		if (ret == -ENOMEM)
 			ret = 0;
 		if (!ret)
-			ret = filemap_fdatawait_range(mapping, start, end);
-		free_extent_state(cached_state);
+			btrfs_btree_wait_writeback_range(fs_info, start, end);
+		btrfs_free_extent_state(cached_state);
 		if (ret)
 			break;
 		cached_state = NULL;
@@ -1207,15 +1218,15 @@ static int btrfs_wait_extents(struct btrfs_fs_info *fs_info,
 		       struct extent_io_tree *dirty_pages)
 {
 	bool errors = false;
-	int err;
+	int ret;
 
-	err = __btrfs_wait_marked_extents(fs_info, dirty_pages);
+	ret = __btrfs_wait_marked_extents(fs_info, dirty_pages);
 	if (test_and_clear_bit(BTRFS_FS_BTREE_ERR, &fs_info->flags))
 		errors = true;
 
-	if (errors && !err)
-		err = -EIO;
-	return err;
+	if (errors && !ret)
+		ret = -EIO;
+	return ret;
 }
 
 int btrfs_wait_tree_log_extents(struct btrfs_root *log_root, int mark)
@@ -1223,22 +1234,23 @@ int btrfs_wait_tree_log_extents(struct btrfs_root *log_root, int mark)
 	struct btrfs_fs_info *fs_info = log_root->fs_info;
 	struct extent_io_tree *dirty_pages = &log_root->dirty_log_pages;
 	bool errors = false;
-	int err;
+	int ret;
 
-	ASSERT(btrfs_root_id(log_root) == BTRFS_TREE_LOG_OBJECTID);
+	ASSERT(btrfs_root_id(log_root) == BTRFS_TREE_LOG_OBJECTID,
+	       "root_id(log_root)=%llu", btrfs_root_id(log_root));
 
-	err = __btrfs_wait_marked_extents(fs_info, dirty_pages);
-	if ((mark & EXTENT_DIRTY) &&
+	ret = __btrfs_wait_marked_extents(fs_info, dirty_pages);
+	if ((mark & EXTENT_DIRTY_LOG1) &&
 	    test_and_clear_bit(BTRFS_FS_LOG1_ERR, &fs_info->flags))
 		errors = true;
 
-	if ((mark & EXTENT_NEW) &&
+	if ((mark & EXTENT_DIRTY_LOG2) &&
 	    test_and_clear_bit(BTRFS_FS_LOG2_ERR, &fs_info->flags))
 		errors = true;
 
-	if (errors && !err)
-		err = -EIO;
-	return err;
+	if (errors && !ret)
+		ret = -EIO;
+	return ret;
 }
 
 /*
@@ -1261,7 +1273,7 @@ static int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans)
 	blk_finish_plug(&plug);
 	ret2 = btrfs_wait_extents(fs_info, dirty_pages);
 
-	extent_io_tree_release(&trans->transaction->dirty_pages);
+	btrfs_extent_io_tree_release(&trans->transaction->dirty_pages);
 
 	if (ret)
 		return ret;
@@ -1323,7 +1335,6 @@ static noinline int commit_cowonly_roots(struct btrfs_trans_handle *trans)
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 	struct list_head *dirty_bgs = &trans->transaction->dirty_bgs;
 	struct list_head *io_bgs = &trans->transaction->io_bgs;
-	struct list_head *next;
 	struct extent_buffer *eb;
 	int ret;
 
@@ -1331,7 +1342,8 @@ static noinline int commit_cowonly_roots(struct btrfs_trans_handle *trans)
 	 * At this point no one can be using this transaction to modify any tree
 	 * and no one can start another transaction to modify any tree either.
 	 */
-	ASSERT(trans->transaction->state == TRANS_STATE_COMMIT_DOING);
+	ASSERT(trans->transaction->state == TRANS_STATE_COMMIT_DOING,
+	       "trans->transaction->state=%d", trans->transaction->state);
 
 	eb = btrfs_lock_root_node(fs_info->tree_root);
 	ret = btrfs_cow_block(trans, fs_info->tree_root, eb, NULL,
@@ -1359,13 +1371,13 @@ static noinline int commit_cowonly_roots(struct btrfs_trans_handle *trans)
 again:
 	while (!list_empty(&fs_info->dirty_cowonly_roots)) {
 		struct btrfs_root *root;
-		next = fs_info->dirty_cowonly_roots.next;
-		list_del_init(next);
-		root = list_entry(next, struct btrfs_root, dirty_list);
+
+		root = list_first_entry(&fs_info->dirty_cowonly_roots,
+					struct btrfs_root, dirty_list);
 		clear_bit(BTRFS_ROOT_DIRTY, &root->state);
+		list_move_tail(&root->dirty_list,
+			       &trans->transaction->switch_commits);
 
-		list_add_tail(&root->dirty_list,
-			      &trans->transaction->switch_commits);
 		ret = update_cowonly_root(trans, root);
 		if (ret)
 			return ret;
@@ -1465,7 +1477,8 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans)
 	 * At this point no one can be using this transaction to modify any tree
 	 * and no one can start another transaction to modify any tree either.
 	 */
-	ASSERT(trans->transaction->state == TRANS_STATE_COMMIT_DOING);
+	ASSERT(trans->transaction->state == TRANS_STATE_COMMIT_DOING,
+	       "trans->transaction->state=%d", trans->transaction->state);
 
 	spin_lock(&fs_info->fs_roots_radix_lock);
 	while (1) {
@@ -1483,9 +1496,15 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans)
 			 * At this point we can neither have tasks logging inodes
 			 * from a root nor trying to commit a log tree.
 			 */
-			ASSERT(atomic_read(&root->log_writers) == 0);
-			ASSERT(atomic_read(&root->log_commit[0]) == 0);
-			ASSERT(atomic_read(&root->log_commit[1]) == 0);
+			ASSERT(atomic_read(&root->log_writers) == 0,
+			       "atomic_read(&root->log_writers)=%d",
+			       atomic_read(&root->log_writers));
+			ASSERT(atomic_read(&root->log_commit[0]) == 0,
+			       "atomic_read(&root->log_commit[0])=%d",
+			       atomic_read(&root->log_commit[0]));
+			ASSERT(atomic_read(&root->log_commit[1]) == 0,
+			       "atomic_read(&root->log_commit[1])=%d",
+			       atomic_read(&root->log_commit[1]));
 
 			radix_tree_tag_clear(&fs_info->fs_roots_radix,
 					(unsigned long)btrfs_root_id(root),
@@ -1566,7 +1585,7 @@ static int qgroup_account_snapshot(struct btrfs_trans_handle *trans,
 	 * qgroup counters could end up wrong.
 	 */
 	ret = btrfs_run_delayed_refs(trans, U64_MAX);
-	if (ret) {
+	if (unlikely(ret)) {
 		btrfs_abort_transaction(trans, ret);
 		return ret;
 	}
@@ -1637,8 +1656,8 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	struct btrfs_root *root = pending->root;
 	struct btrfs_root *parent_root;
 	struct btrfs_block_rsv *rsv;
-	struct inode *parent_inode = pending->dir;
-	struct btrfs_path *path;
+	struct btrfs_inode *parent_inode = pending->dir;
+	BTRFS_PATH_AUTO_FREE(path);
 	struct btrfs_dir_item *dir_item;
 	struct extent_buffer *tmp;
 	struct extent_buffer *old;
@@ -1663,7 +1682,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	 * filesystem.
 	 */
 	nofs_flags = memalloc_nofs_save();
-	pending->error = fscrypt_setup_filename(parent_inode,
+	pending->error = fscrypt_setup_filename(&parent_inode->vfs_inode,
 						&pending->dentry->d_name, 0,
 						&fname);
 	memalloc_nofs_restore(nofs_flags);
@@ -1691,34 +1710,30 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 			goto clear_skip_qgroup;
 	}
 
-	key.objectid = objectid;
-	key.offset = (u64)-1;
-	key.type = BTRFS_ROOT_ITEM_KEY;
-
 	rsv = trans->block_rsv;
 	trans->block_rsv = &pending->block_rsv;
 	trans->bytes_reserved = trans->block_rsv->reserved;
 	trace_btrfs_space_reservation(fs_info, "transaction",
 				      trans->transid,
 				      trans->bytes_reserved, 1);
-	parent_root = BTRFS_I(parent_inode)->root;
+	parent_root = parent_inode->root;
 	ret = record_root_in_trans(trans, parent_root, 0);
 	if (ret)
 		goto fail;
-	cur_time = current_time(parent_inode);
+	cur_time = current_time(&parent_inode->vfs_inode);
 
 	/*
 	 * insert the directory item
 	 */
-	ret = btrfs_set_inode_index(BTRFS_I(parent_inode), &index);
-	if (ret) {
+	ret = btrfs_set_inode_index(parent_inode, &index);
+	if (unlikely(ret)) {
 		btrfs_abort_transaction(trans, ret);
 		goto fail;
 	}
 
 	/* check if there is a file/dir which has the same name. */
 	dir_item = btrfs_lookup_dir_item(NULL, parent_root, path,
-					 btrfs_ino(BTRFS_I(parent_inode)),
+					 btrfs_ino(parent_inode),
 					 &fname.disk_name, 0);
 	if (dir_item != NULL && !IS_ERR(dir_item)) {
 		pending->error = -EEXIST;
@@ -1732,8 +1747,10 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 
 	ret = btrfs_create_qgroup(trans, objectid);
 	if (ret && ret != -EEXIST) {
-		btrfs_abort_transaction(trans, ret);
-		goto fail;
+		if (unlikely(ret != -ENOTCONN || btrfs_qgroup_enabled(fs_info))) {
+			btrfs_abort_transaction(trans, ret);
+			goto fail;
+		}
 	}
 
 	/*
@@ -1743,13 +1760,13 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	 * snapshot
 	 */
 	ret = btrfs_run_delayed_items(trans);
-	if (ret) {	/* Transaction aborted */
+	if (unlikely(ret)) {
 		btrfs_abort_transaction(trans, ret);
 		goto fail;
 	}
 
 	ret = record_root_in_trans(trans, root, 0);
-	if (ret) {
+	if (unlikely(ret)) {
 		btrfs_abort_transaction(trans, ret);
 		goto fail;
 	}
@@ -1784,7 +1801,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	old = btrfs_lock_root_node(root);
 	ret = btrfs_cow_block(trans, root, old, NULL, 0, &old,
 			      BTRFS_NESTING_COW);
-	if (ret) {
+	if (unlikely(ret)) {
 		btrfs_tree_unlock(old);
 		free_extent_buffer(old);
 		btrfs_abort_transaction(trans, ret);
@@ -1795,21 +1812,23 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	/* clean up in any case */
 	btrfs_tree_unlock(old);
 	free_extent_buffer(old);
-	if (ret) {
+	if (unlikely(ret)) {
 		btrfs_abort_transaction(trans, ret);
 		goto fail;
 	}
 	/* see comments in should_cow_block() */
 	set_bit(BTRFS_ROOT_FORCE_COW, &root->state);
-	smp_wmb();
+	smp_mb__after_atomic();
 
 	btrfs_set_root_node(new_root_item, tmp);
 	/* record when the snapshot was created in key.offset */
+	key.objectid = objectid;
+	key.type = BTRFS_ROOT_ITEM_KEY;
 	key.offset = trans->transid;
 	ret = btrfs_insert_root(trans, tree_root, &key, new_root_item);
 	btrfs_tree_unlock(tmp);
 	free_extent_buffer(tmp);
-	if (ret) {
+	if (unlikely(ret)) {
 		btrfs_abort_transaction(trans, ret);
 		goto fail;
 	}
@@ -1819,9 +1838,9 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	 */
 	ret = btrfs_add_root_ref(trans, objectid,
 				 btrfs_root_id(parent_root),
-				 btrfs_ino(BTRFS_I(parent_inode)), index,
+				 btrfs_ino(parent_inode), index,
 				 &fname.disk_name);
-	if (ret) {
+	if (unlikely(ret)) {
 		btrfs_abort_transaction(trans, ret);
 		goto fail;
 	}
@@ -1836,7 +1855,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	}
 
 	ret = btrfs_reloc_post_snapshot(trans, pending);
-	if (ret) {
+	if (unlikely(ret)) {
 		btrfs_abort_transaction(trans, ret);
 		goto fail;
 	}
@@ -1857,26 +1876,26 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 		goto fail;
 
 	ret = btrfs_insert_dir_item(trans, &fname.disk_name,
-				    BTRFS_I(parent_inode), &key, BTRFS_FT_DIR,
+				    parent_inode, &key, BTRFS_FT_DIR,
 				    index);
-	if (ret) {
+	if (unlikely(ret)) {
 		btrfs_abort_transaction(trans, ret);
 		goto fail;
 	}
 
-	btrfs_i_size_write(BTRFS_I(parent_inode), parent_inode->i_size +
+	btrfs_i_size_write(parent_inode, parent_inode->vfs_inode.i_size +
 						  fname.disk_name.len * 2);
-	inode_set_mtime_to_ts(parent_inode,
-			      inode_set_ctime_current(parent_inode));
-	ret = btrfs_update_inode_fallback(trans, BTRFS_I(parent_inode));
-	if (ret) {
+	inode_set_mtime_to_ts(&parent_inode->vfs_inode,
+			      inode_set_ctime_current(&parent_inode->vfs_inode));
+	ret = btrfs_update_inode_fallback(trans, parent_inode);
+	if (unlikely(ret)) {
 		btrfs_abort_transaction(trans, ret);
 		goto fail;
 	}
 	ret = btrfs_uuid_tree_add(trans, new_root_item->uuid,
 				  BTRFS_UUID_KEY_SUBVOL,
 				  objectid);
-	if (ret) {
+	if (unlikely(ret)) {
 		btrfs_abort_transaction(trans, ret);
 		goto fail;
 	}
@@ -1884,7 +1903,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 		ret = btrfs_uuid_tree_add(trans, new_root_item->received_uuid,
 					  BTRFS_UUID_KEY_RECEIVED_SUBVOL,
 					  objectid);
-		if (ret && ret != -EEXIST) {
+		if (unlikely(ret && ret != -EEXIST)) {
 			btrfs_abort_transaction(trans, ret);
 			goto fail;
 		}
@@ -1902,7 +1921,6 @@ free_fname:
 free_pending:
 	kfree(new_root_item);
 	pending->root_item = NULL;
-	btrfs_free_path(path);
 	pending->path = NULL;
 
 	return ret;
@@ -1989,6 +2007,25 @@ void btrfs_commit_transaction_async(struct btrfs_trans_handle *trans)
 	btrfs_put_transaction(cur_trans);
 }
 
+/*
+ * If there is a running transaction commit it or if it's already committing,
+ * wait for its commit to complete. Does not start and commit a new transaction
+ * if there isn't any running.
+ */
+int btrfs_commit_current_transaction(struct btrfs_root *root)
+{
+	struct btrfs_trans_handle *trans;
+
+	trans = btrfs_attach_transaction_barrier(root);
+	if (IS_ERR(trans)) {
+		int ret = PTR_ERR(trans);
+
+		return (ret == -ENOENT) ? 0 : ret;
+	}
+
+	return btrfs_commit_transaction(trans);
+}
+
 static void cleanup_transaction(struct btrfs_trans_handle *trans, int err)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
@@ -2034,7 +2071,7 @@ static void cleanup_transaction(struct btrfs_trans_handle *trans, int err)
 
 	spin_unlock(&fs_info->trans_lock);
 
-	btrfs_cleanup_one_transaction(trans->transaction, fs_info);
+	btrfs_cleanup_one_transaction(trans->transaction);
 
 	spin_lock(&fs_info->trans_lock);
 	if (cur_trans == fs_info->running_transaction)
@@ -2079,7 +2116,14 @@ static void btrfs_cleanup_pending_block_groups(struct btrfs_trans_handle *trans)
 
        list_for_each_entry_safe(block_group, tmp, &trans->new_bgs, bg_list) {
                btrfs_dec_delayed_refs_rsv_bg_inserts(fs_info);
+		/*
+		* Not strictly necessary to lock, as no other task will be using a
+		* block_group on the new_bgs list during a transaction abort.
+		*/
+	       spin_lock(&fs_info->unused_bgs_lock);
                list_del_init(&block_group->bg_list);
+	       btrfs_put_block_group(block_group);
+	       spin_unlock(&fs_info->unused_bgs_lock);
        }
 }
 
@@ -2110,7 +2154,7 @@ static inline int btrfs_start_delalloc_flush(struct btrfs_fs_info *fs_info)
 static inline void btrfs_wait_delalloc_flush(struct btrfs_fs_info *fs_info)
 {
 	if (btrfs_test_opt(fs_info, FLUSHONCOMMIT))
-		btrfs_wait_ordered_roots(fs_info, U64_MAX, 0, (u64)-1);
+		btrfs_wait_ordered_roots(fs_info, U64_MAX, NULL);
 }
 
 /*
@@ -2129,18 +2173,25 @@ static void add_pending_snapshot(struct btrfs_trans_handle *trans)
 		return;
 
 	lockdep_assert_held(&trans->fs_info->trans_lock);
-	ASSERT(cur_trans->state >= TRANS_STATE_COMMIT_PREP);
+	ASSERT(cur_trans->state >= TRANS_STATE_COMMIT_PREP,
+	       "cur_trans->state=%d", cur_trans->state);
 
 	list_add(&trans->pending_snapshot->list, &cur_trans->pending_snapshots);
 }
 
-static void update_commit_stats(struct btrfs_fs_info *fs_info, ktime_t interval)
+static void update_commit_stats(struct btrfs_fs_info *fs_info)
 {
+	ktime_t now = ktime_get_ns();
+	ktime_t interval = now - fs_info->commit_stats.critical_section_start_time;
+
+	ASSERT(fs_info->commit_stats.critical_section_start_time);
+
 	fs_info->commit_stats.commit_count++;
 	fs_info->commit_stats.last_commit_dur = interval;
 	fs_info->commit_stats.max_commit_dur =
 			max_t(u64, fs_info->commit_stats.max_commit_dur, interval);
 	fs_info->commit_stats.total_commit_dur += interval;
+	fs_info->commit_stats.critical_section_start_time = 0;
 }
 
 int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
@@ -2149,10 +2200,9 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	struct btrfs_transaction *cur_trans = trans->transaction;
 	struct btrfs_transaction *prev_trans = NULL;
 	int ret;
-	ktime_t start_time;
-	ktime_t interval;
 
-	ASSERT(refcount_read(&trans->use_count) == 1);
+	ASSERT(refcount_read(&trans->use_count) == 1,
+	       "refcount_read(&trans->use_count)=%d", refcount_read(&trans->use_count));
 	btrfs_trans_state_lockdep_acquire(fs_info, BTRFS_LOCKDEP_TRANS_COMMIT_PREP);
 
 	clear_bit(BTRFS_FS_NEED_TRANS_COMMIT, &fs_info->flags);
@@ -2241,14 +2291,13 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	wake_up(&fs_info->transaction_blocked_wait);
 	btrfs_trans_state_lockdep_release(fs_info, BTRFS_LOCKDEP_TRANS_COMMIT_PREP);
 
-	if (cur_trans->list.prev != &fs_info->trans_list) {
+	if (!list_is_first(&cur_trans->list, &fs_info->trans_list)) {
 		enum btrfs_trans_state want_state = TRANS_STATE_COMPLETED;
 
 		if (trans->in_fsync)
 			want_state = TRANS_STATE_SUPER_COMMITTED;
 
-		prev_trans = list_entry(cur_trans->list.prev,
-					struct btrfs_transaction, list);
+		prev_trans = list_prev_entry(cur_trans, list);
 		if (prev_trans->state < want_state) {
 			refcount_inc(&prev_trans->use_count);
 			spin_unlock(&fs_info->trans_lock);
@@ -2284,8 +2333,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	 * Get the time spent on the work done by the commit thread and not
 	 * the time spent waiting on a previous commit
 	 */
-	start_time = ktime_get_ns();
-
+	fs_info->commit_stats.critical_section_start_time = ktime_get_ns();
 	extwriter_counter_dec(cur_trans, trans->type);
 
 	ret = btrfs_start_delalloc_flush(fs_info);
@@ -2390,7 +2438,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	 * them.
 	 *
 	 * We needn't worry that this operation will corrupt the snapshots,
-	 * because all the tree which are snapshoted will be forced to COW
+	 * because all the tree which are snapshotted will be forced to COW
 	 * the nodes and leaves.
 	 */
 	ret = btrfs_run_delayed_items(trans);
@@ -2517,6 +2565,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	if (ret)
 		goto scrub_continue;
 
+	update_commit_stats(fs_info);
 	/*
 	 * We needn't acquire the lock here because there is no other task
 	 * which can change it.
@@ -2525,7 +2574,9 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	wake_up(&cur_trans->commit_wait);
 	btrfs_trans_state_lockdep_release(fs_info, BTRFS_LOCKDEP_TRANS_SUPER_COMMITTED);
 
-	btrfs_finish_extent_commit(trans);
+	ret = btrfs_finish_extent_commit(trans);
+	if (ret)
+		goto scrub_continue;
 
 	if (test_bit(BTRFS_TRANS_HAVE_FREE_BGS, &cur_trans->flags))
 		btrfs_clear_space_info_full(fs_info);
@@ -2551,8 +2602,6 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 
 	trace_btrfs_transaction_commit(fs_info);
 
-	interval = ktime_get_ns() - start_time;
-
 	btrfs_scrub_continue(fs_info);
 
 	if (current->journal_info == trans)
@@ -2560,8 +2609,6 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 
 	kmem_cache_free(btrfs_trans_handle_cachep, trans);
 
-	update_commit_stats(fs_info, interval);
-
 	return ret;
 
 unlock_reloc:
@@ -2625,9 +2672,9 @@ int btrfs_clean_one_deleted_snapshot(struct btrfs_fs_info *fs_info)
 
 	if (btrfs_header_backref_rev(root->node) <
 			BTRFS_MIXED_BACKREF_REV)
-		ret = btrfs_drop_snapshot(root, 0, 0);
+		ret = btrfs_drop_snapshot(root, false, false);
 	else
-		ret = btrfs_drop_snapshot(root, 1, 0);
+		ret = btrfs_drop_snapshot(root, true, false);
 
 	btrfs_put_root(root);
 	return (ret < 0) ? 0 : 1;