1 files changed, 82 insertions, 60 deletions

diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index c6a872a8a468..7579f6d0b854 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -62,7 +62,7 @@ void btrfs_put_transaction(struct btrfs_transaction *transaction)
 	WARN_ON(atomic_read(&transaction->use_count) == 0);
 	if (atomic_dec_and_test(&transaction->use_count)) {
 		BUG_ON(!list_empty(&transaction->list));
-		WARN_ON(transaction->delayed_refs.root.rb_node);
+		WARN_ON(!RB_EMPTY_ROOT(&transaction->delayed_refs.href_root));
 		while (!list_empty(&transaction->pending_chunks)) {
 			struct extent_map *em;
 
@@ -75,10 +75,21 @@ void btrfs_put_transaction(struct btrfs_transaction *transaction)
 	}
 }
 
-static noinline void switch_commit_root(struct btrfs_root *root)
+static noinline void switch_commit_roots(struct btrfs_transaction *trans,
+					 struct btrfs_fs_info *fs_info)
 {
-	free_extent_buffer(root->commit_root);
-	root->commit_root = btrfs_root_node(root);
+	struct btrfs_root *root, *tmp;
+
+	down_write(&fs_info->commit_root_sem);
+	list_for_each_entry_safe(root, tmp, &trans->switch_commits,
+				 dirty_list) {
+		list_del_init(&root->dirty_list);
+		free_extent_buffer(root->commit_root);
+		root->commit_root = btrfs_root_node(root);
+		if (is_fstree(root->objectid))
+			btrfs_unpin_free_ino(root);
+	}
+	up_write(&fs_info->commit_root_sem);
 }
 
 static inline void extwriter_counter_inc(struct btrfs_transaction *trans,
@@ -183,8 +194,8 @@ loop:
 	atomic_set(&cur_trans->use_count, 2);
 	cur_trans->start_time = get_seconds();
 
-	cur_trans->delayed_refs.root = RB_ROOT;
-	cur_trans->delayed_refs.num_entries = 0;
+	cur_trans->delayed_refs.href_root = RB_ROOT;
+	atomic_set(&cur_trans->delayed_refs.num_entries, 0);
 	cur_trans->delayed_refs.num_heads_ready = 0;
 	cur_trans->delayed_refs.num_heads = 0;
 	cur_trans->delayed_refs.flushing = 0;
@@ -196,21 +207,19 @@ loop:
 	 */
 	smp_mb();
 	if (!list_empty(&fs_info->tree_mod_seq_list))
-		WARN(1, KERN_ERR "btrfs: tree_mod_seq_list not empty when "
+		WARN(1, KERN_ERR "BTRFS: tree_mod_seq_list not empty when "
 			"creating a fresh transaction\n");
 	if (!RB_EMPTY_ROOT(&fs_info->tree_mod_log))
-		WARN(1, KERN_ERR "btrfs: tree_mod_log rb tree not empty when "
+		WARN(1, KERN_ERR "BTRFS: tree_mod_log rb tree not empty when "
 			"creating a fresh transaction\n");
 	atomic64_set(&fs_info->tree_mod_seq, 0);
 
 	spin_lock_init(&cur_trans->delayed_refs.lock);
-	atomic_set(&cur_trans->delayed_refs.procs_running_refs, 0);
-	atomic_set(&cur_trans->delayed_refs.ref_seq, 0);
-	init_waitqueue_head(&cur_trans->delayed_refs.wait);
 
 	INIT_LIST_HEAD(&cur_trans->pending_snapshots);
 	INIT_LIST_HEAD(&cur_trans->ordered_operations);
 	INIT_LIST_HEAD(&cur_trans->pending_chunks);
+	INIT_LIST_HEAD(&cur_trans->switch_commits);
 	list_add_tail(&cur_trans->list, &fs_info->trans_list);
 	extent_io_tree_init(&cur_trans->dirty_pages,
 			     fs_info->btree_inode->i_mapping);
@@ -378,7 +387,8 @@ start_transaction(struct btrfs_root *root, u64 num_items, unsigned int type,
 	if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state))
 		return ERR_PTR(-EROFS);
 
-	if (current->journal_info) {
+	if (current->journal_info &&
+	    current->journal_info != (void *)BTRFS_SEND_TRANS_STUB) {
 		WARN_ON(type & TRANS_EXTWRITERS);
 		h = current->journal_info;
 		h->use_count++;
@@ -472,6 +482,7 @@ again:
 	h->type = type;
 	h->allocating_chunk = false;
 	h->reloc_reserved = false;
+	h->sync = false;
 	INIT_LIST_HEAD(&h->qgroup_ref_list);
 	INIT_LIST_HEAD(&h->new_bgs);
 
@@ -647,7 +658,7 @@ static int should_end_transaction(struct btrfs_trans_handle *trans,
 				  struct btrfs_root *root)
 {
 	if (root->fs_info->global_block_rsv.space_info->full &&
-	    btrfs_should_throttle_delayed_refs(trans, root))
+	    btrfs_check_space_for_delayed_refs(trans, root))
 		return 1;
 
 	return !!btrfs_block_rsv_check(root, &root->fs_info->global_block_rsv, 5);
@@ -685,7 +696,8 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 	int lock = (trans->type != TRANS_JOIN_NOLOCK);
 	int err = 0;
 
-	if (--trans->use_count) {
+	if (trans->use_count > 1) {
+		trans->use_count--;
 		trans->block_rsv = trans->orig_rsv;
 		return 0;
 	}
@@ -711,8 +723,8 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 		btrfs_create_pending_block_groups(trans, root);
 
 	trans->delayed_ref_updates = 0;
-	if (btrfs_should_throttle_delayed_refs(trans, root)) {
-		cur = max_t(unsigned long, cur, 1);
+	if (!trans->sync && btrfs_should_throttle_delayed_refs(trans, root)) {
+		cur = max_t(unsigned long, cur, 32);
 		trans->delayed_ref_updates = 0;
 		btrfs_run_delayed_refs(trans, root, cur);
 	}
@@ -733,17 +745,10 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 	}
 
 	if (lock && ACCESS_ONCE(cur_trans->state) == TRANS_STATE_BLOCKED) {
-		if (throttle) {
-			/*
-			 * We may race with somebody else here so end up having
-			 * to call end_transaction on ourselves again, so inc
-			 * our use_count.
-			 */
-			trans->use_count++;
+		if (throttle)
 			return btrfs_commit_transaction(trans, root);
-		} else {
+		else
 			wake_up_process(info->transaction_kthread);
-		}
 	}
 
 	if (trans->type & __TRANS_FREEZABLE)
@@ -788,12 +793,6 @@ int btrfs_end_transaction_throttle(struct btrfs_trans_handle *trans,
 	return __btrfs_end_transaction(trans, root, 1);
 }
 
-int btrfs_end_transaction_dmeta(struct btrfs_trans_handle *trans,
-				struct btrfs_root *root)
-{
-	return __btrfs_end_transaction(trans, root, 1);
-}
-
 /*
  * 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
@@ -933,9 +932,6 @@ static int update_cowonly_root(struct btrfs_trans_handle *trans,
 			return ret;
 	}
 
-	if (root != root->fs_info->extent_root)
-		switch_commit_root(root);
-
 	return 0;
 }
 
@@ -991,15 +987,16 @@ static noinline int commit_cowonly_roots(struct btrfs_trans_handle *trans,
 		list_del_init(next);
 		root = list_entry(next, struct btrfs_root, dirty_list);
 
+		if (root != fs_info->extent_root)
+			list_add_tail(&root->dirty_list,
+				      &trans->transaction->switch_commits);
 		ret = update_cowonly_root(trans, root);
 		if (ret)
 			return ret;
 	}
 
-	down_write(&fs_info->extent_commit_sem);
-	switch_commit_root(fs_info->extent_root);
-	up_write(&fs_info->extent_commit_sem);
-
+	list_add_tail(&fs_info->extent_root->dirty_list,
+		      &trans->transaction->switch_commits);
 	btrfs_after_dev_replace_commit(fs_info);
 
 	return 0;
@@ -1056,11 +1053,8 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans,
 			smp_wmb();
 
 			if (root->commit_root != root->node) {
-				mutex_lock(&root->fs_commit_mutex);
-				switch_commit_root(root);
-				btrfs_unpin_free_ino(root);
-				mutex_unlock(&root->fs_commit_mutex);
-
+				list_add_tail(&root->dirty_list,
+					&trans->transaction->switch_commits);
 				btrfs_set_root_node(&root->root_item,
 						    root->node);
 			}
@@ -1105,7 +1099,7 @@ int btrfs_defrag_root(struct btrfs_root *root)
 			break;
 
 		if (btrfs_defrag_cancelled(root->fs_info)) {
-			printk(KERN_DEBUG "btrfs: defrag_root cancelled\n");
+			pr_debug("BTRFS: defrag_root cancelled\n");
 			ret = -EAGAIN;
 			break;
 		}
@@ -1586,10 +1580,9 @@ static void cleanup_transaction(struct btrfs_trans_handle *trans,
 
 	trace_btrfs_transaction_commit(root);
 
-	btrfs_scrub_continue(root);
-
 	if (current->journal_info == trans)
 		current->journal_info = NULL;
+	btrfs_scrub_cancel(root->fs_info);
 
 	kmem_cache_free(btrfs_trans_handle_cachep, trans);
 }
@@ -1629,7 +1622,7 @@ static int btrfs_flush_all_pending_stuffs(struct btrfs_trans_handle *trans,
 static inline int btrfs_start_delalloc_flush(struct btrfs_fs_info *fs_info)
 {
 	if (btrfs_test_opt(fs_info->tree_root, FLUSHONCOMMIT))
-		return btrfs_start_delalloc_roots(fs_info, 1);
+		return btrfs_start_delalloc_roots(fs_info, 1, -1);
 	return 0;
 }
 
@@ -1746,6 +1739,8 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 		goto cleanup_transaction;
 
 	btrfs_wait_delalloc_flush(root->fs_info);
+
+	btrfs_scrub_pause(root);
 	/*
 	 * Ok now we need to make sure to block out any other joins while we
 	 * commit the transaction.  We could have started a join before setting
@@ -1760,7 +1755,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	/* ->aborted might be set after the previous check, so check it */
 	if (unlikely(ACCESS_ONCE(cur_trans->aborted))) {
 		ret = cur_trans->aborted;
-		goto cleanup_transaction;
+		goto scrub_continue;
 	}
 	/*
 	 * the reloc mutex makes sure that we stop
@@ -1777,7 +1772,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	ret = create_pending_snapshots(trans, root->fs_info);
 	if (ret) {
 		mutex_unlock(&root->fs_info->reloc_mutex);
-		goto cleanup_transaction;
+		goto scrub_continue;
 	}
 
 	/*
@@ -1793,13 +1788,13 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	ret = btrfs_run_delayed_items(trans, root);
 	if (ret) {
 		mutex_unlock(&root->fs_info->reloc_mutex);
-		goto cleanup_transaction;
+		goto scrub_continue;
 	}
 
 	ret = btrfs_run_delayed_refs(trans, root, (unsigned long)-1);
 	if (ret) {
 		mutex_unlock(&root->fs_info->reloc_mutex);
-		goto cleanup_transaction;
+		goto scrub_continue;
 	}
 
 	/*
@@ -1810,7 +1805,6 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 
 	WARN_ON(cur_trans != trans->transaction);
 
-	btrfs_scrub_pause(root);
 	/* btrfs_commit_tree_roots is responsible for getting the
 	 * various roots consistent with each other.  Every pointer
 	 * in the tree of tree roots has to point to the most up to date
@@ -1830,9 +1824,18 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	if (ret) {
 		mutex_unlock(&root->fs_info->tree_log_mutex);
 		mutex_unlock(&root->fs_info->reloc_mutex);
-		goto cleanup_transaction;
+		goto scrub_continue;
 	}
 
+	/*
+	 * Since the transaction is done, we should set the inode map cache flag
+	 * before any other comming transaction.
+	 */
+	if (btrfs_test_opt(root, CHANGE_INODE_CACHE))
+		btrfs_set_opt(root->fs_info->mount_opt, INODE_MAP_CACHE);
+	else
+		btrfs_clear_opt(root->fs_info->mount_opt, INODE_MAP_CACHE);
+
 	/* commit_fs_roots gets rid of all the tree log roots, it is now
 	 * safe to free the root of tree log roots
 	 */
@@ -1842,7 +1845,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	if (ret) {
 		mutex_unlock(&root->fs_info->tree_log_mutex);
 		mutex_unlock(&root->fs_info->reloc_mutex);
-		goto cleanup_transaction;
+		goto scrub_continue;
 	}
 
 	/*
@@ -1853,7 +1856,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 		ret = cur_trans->aborted;
 		mutex_unlock(&root->fs_info->tree_log_mutex);
 		mutex_unlock(&root->fs_info->reloc_mutex);
-		goto cleanup_transaction;
+		goto scrub_continue;
 	}
 
 	btrfs_prepare_extent_commit(trans, root);
@@ -1862,11 +1865,15 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 
 	btrfs_set_root_node(&root->fs_info->tree_root->root_item,
 			    root->fs_info->tree_root->node);
-	switch_commit_root(root->fs_info->tree_root);
+	list_add_tail(&root->fs_info->tree_root->dirty_list,
+		      &cur_trans->switch_commits);
 
 	btrfs_set_root_node(&root->fs_info->chunk_root->root_item,
 			    root->fs_info->chunk_root->node);
-	switch_commit_root(root->fs_info->chunk_root);
+	list_add_tail(&root->fs_info->chunk_root->dirty_list,
+		      &cur_trans->switch_commits);
+
+	switch_commit_roots(cur_trans, root->fs_info);
 
 	assert_qgroups_uptodate(trans);
 	update_super_roots(root);
@@ -1889,13 +1896,13 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 		btrfs_error(root->fs_info, ret,
 			    "Error while writing out transaction");
 		mutex_unlock(&root->fs_info->tree_log_mutex);
-		goto cleanup_transaction;
+		goto scrub_continue;
 	}
 
 	ret = write_ctree_super(trans, root, 0);
 	if (ret) {
 		mutex_unlock(&root->fs_info->tree_log_mutex);
-		goto cleanup_transaction;
+		goto scrub_continue;
 	}
 
 	/*
@@ -1938,6 +1945,8 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 
 	return ret;
 
+scrub_continue:
+	btrfs_scrub_continue(root);
 cleanup_transaction:
 	btrfs_trans_release_metadata(trans, root);
 	trans->block_rsv = NULL;
@@ -1975,10 +1984,23 @@ int btrfs_clean_one_deleted_snapshot(struct btrfs_root *root)
 	}
 	root = list_first_entry(&fs_info->dead_roots,
 			struct btrfs_root, root_list);
+	/*
+	 * Make sure root is not involved in send,
+	 * if we fail with first root, we return
+	 * directly rather than continue.
+	 */
+	spin_lock(&root->root_item_lock);
+	if (root->send_in_progress) {
+		spin_unlock(&fs_info->trans_lock);
+		spin_unlock(&root->root_item_lock);
+		return 0;
+	}
+	spin_unlock(&root->root_item_lock);
+
 	list_del_init(&root->root_list);
 	spin_unlock(&fs_info->trans_lock);
 
-	pr_debug("btrfs: cleaner removing %llu\n", root->objectid);
+	pr_debug("BTRFS: cleaner removing %llu\n", root->objectid);
 
 	btrfs_kill_all_delayed_nodes(root);