1 files changed, 435 insertions, 82 deletions

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 6b7fe291a174..d2be95cfb6d1 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -455,7 +455,7 @@ again:
 
 	/*
 	 * skip compression for a small file range(<=blocksize) that
-	 * isn't an inline extent, since it dosen't save disk space at all.
+	 * isn't an inline extent, since it doesn't save disk space at all.
 	 */
 	if (total_compressed <= blocksize &&
 	   (start > 0 || end + 1 < BTRFS_I(inode)->disk_i_size))
@@ -824,6 +824,7 @@ retry:
 						async_extent->ram_size - 1, 0);
 			goto out_free_reserve;
 		}
+		btrfs_dec_block_group_reservations(root->fs_info, ins.objectid);
 
 		/*
 		 * clear dirty, set writeback and unlock the pages.
@@ -861,6 +862,7 @@ retry:
 	}
 	return;
 out_free_reserve:
+	btrfs_dec_block_group_reservations(root->fs_info, ins.objectid);
 	btrfs_free_reserved_extent(root, ins.objectid, ins.offset, 1);
 out_free:
 	extent_clear_unlock_delalloc(inode, async_extent->start,
@@ -1038,6 +1040,8 @@ static noinline int cow_file_range(struct inode *inode,
 				goto out_drop_extent_cache;
 		}
 
+		btrfs_dec_block_group_reservations(root->fs_info, ins.objectid);
+
 		if (disk_num_bytes < cur_alloc_size)
 			break;
 
@@ -1066,6 +1070,7 @@ out:
 out_drop_extent_cache:
 	btrfs_drop_extent_cache(inode, start, start + ram_size - 1, 0);
 out_reserve:
+	btrfs_dec_block_group_reservations(root->fs_info, ins.objectid);
 	btrfs_free_reserved_extent(root, ins.objectid, ins.offset, 1);
 out_unlock:
 	extent_clear_unlock_delalloc(inode, start, end, locked_page,
@@ -1377,6 +1382,9 @@ next_slot:
 			 */
 			if (csum_exist_in_range(root, disk_bytenr, num_bytes))
 				goto out_check;
+			if (!btrfs_inc_nocow_writers(root->fs_info,
+						     disk_bytenr))
+				goto out_check;
 			nocow = 1;
 		} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
 			extent_end = found_key.offset +
@@ -1391,6 +1399,9 @@ out_check:
 			path->slots[0]++;
 			if (!nolock && nocow)
 				btrfs_end_write_no_snapshoting(root);
+			if (nocow)
+				btrfs_dec_nocow_writers(root->fs_info,
+							disk_bytenr);
 			goto next_slot;
 		}
 		if (!nocow) {
@@ -1411,6 +1422,9 @@ out_check:
 			if (ret) {
 				if (!nolock && nocow)
 					btrfs_end_write_no_snapshoting(root);
+				if (nocow)
+					btrfs_dec_nocow_writers(root->fs_info,
+								disk_bytenr);
 				goto error;
 			}
 			cow_start = (u64)-1;
@@ -1453,6 +1467,8 @@ out_check:
 
 		ret = btrfs_add_ordered_extent(inode, cur_offset, disk_bytenr,
 					       num_bytes, num_bytes, type);
+		if (nocow)
+			btrfs_dec_nocow_writers(root->fs_info, disk_bytenr);
 		BUG_ON(ret); /* -ENOMEM */
 
 		if (root->root_key.objectid ==
@@ -1962,7 +1978,7 @@ int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end,
 {
 	WARN_ON((end & (PAGE_SIZE - 1)) == 0);
 	return set_extent_delalloc(&BTRFS_I(inode)->io_tree, start, end,
-				   cached_state, GFP_NOFS);
+				   cached_state);
 }
 
 /* see btrfs_writepage_start_hook for details on why this is required */
@@ -3103,8 +3119,7 @@ static int btrfs_readpage_end_io_hook(struct btrfs_io_bio *io_bio,
 
 	if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID &&
 	    test_range_bit(io_tree, start, end, EXTENT_NODATASUM, 1, NULL)) {
-		clear_extent_bits(io_tree, start, end, EXTENT_NODATASUM,
-				  GFP_NOFS);
+		clear_extent_bits(io_tree, start, end, EXTENT_NODATASUM);
 		return 0;
 	}
 
@@ -3256,7 +3271,16 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode)
 	/* grab metadata reservation from transaction handle */
 	if (reserve) {
 		ret = btrfs_orphan_reserve_metadata(trans, inode);
-		BUG_ON(ret); /* -ENOSPC in reservation; Logic error? JDM */
+		ASSERT(!ret);
+		if (ret) {
+			atomic_dec(&root->orphan_inodes);
+			clear_bit(BTRFS_INODE_ORPHAN_META_RESERVED,
+				  &BTRFS_I(inode)->runtime_flags);
+			if (insert)
+				clear_bit(BTRFS_INODE_HAS_ORPHAN_ITEM,
+					  &BTRFS_I(inode)->runtime_flags);
+			return ret;
+		}
 	}
 
 	/* insert an orphan item to track this unlinked/truncated file */
@@ -3706,7 +3730,7 @@ cache_index:
 	 * and doesn't have an inode ref with the name "bar" anymore.
 	 *
 	 * Setting last_unlink_trans to last_trans is a pessimistic approach,
-	 * but it guarantees correctness at the expense of ocassional full
+	 * but it guarantees correctness at the expense of occasional full
 	 * transaction commits on fsync if our inode is a directory, or if our
 	 * inode is not a directory, logging its parent unnecessarily.
 	 */
@@ -4962,7 +4986,7 @@ static int btrfs_setsize(struct inode *inode, struct iattr *attr)
 		 * be instantly completed which will give us extents that need
 		 * to be truncated.  If we fail to get an orphan inode down we
 		 * could have left over extents that were never meant to live,
-		 * so we need to garuntee from this point on that everything
+		 * so we need to guarantee from this point on that everything
 		 * will be consistent.
 		 */
 		ret = btrfs_orphan_add(trans, inode);
@@ -5232,7 +5256,7 @@ void btrfs_evict_inode(struct inode *inode)
 		}
 
 		/*
-		 * We can't just steal from the global reserve, we need tomake
+		 * We can't just steal from the global reserve, we need to make
 		 * sure there is room to do it, if not we need to commit and try
 		 * again.
 		 */
@@ -6964,7 +6988,18 @@ insert:
 		 * existing will always be non-NULL, since there must be
 		 * extent causing the -EEXIST.
 		 */
-		if (start >= extent_map_end(existing) ||
+		if (existing->start == em->start &&
+		    extent_map_end(existing) == extent_map_end(em) &&
+		    em->block_start == existing->block_start) {
+			/*
+			 * these two extents are the same, it happens
+			 * with inlines especially
+			 */
+			free_extent_map(em);
+			em = existing;
+			err = 0;
+
+		} else if (start >= extent_map_end(existing) ||
 		    start <= existing->start) {
 			/*
 			 * The existing extent map is the one nearest to
@@ -7129,6 +7164,43 @@ out:
 	return em;
 }
 
+static struct extent_map *btrfs_create_dio_extent(struct inode *inode,
+						  const u64 start,
+						  const u64 len,
+						  const u64 orig_start,
+						  const u64 block_start,
+						  const u64 block_len,
+						  const u64 orig_block_len,
+						  const u64 ram_bytes,
+						  const int type)
+{
+	struct extent_map *em = NULL;
+	int ret;
+
+	down_read(&BTRFS_I(inode)->dio_sem);
+	if (type != BTRFS_ORDERED_NOCOW) {
+		em = create_pinned_em(inode, start, len, orig_start,
+				      block_start, block_len, orig_block_len,
+				      ram_bytes, type);
+		if (IS_ERR(em))
+			goto out;
+	}
+	ret = btrfs_add_ordered_extent_dio(inode, start, block_start,
+					   len, block_len, type);
+	if (ret) {
+		if (em) {
+			free_extent_map(em);
+			btrfs_drop_extent_cache(inode, start,
+						start + len - 1, 0);
+		}
+		em = ERR_PTR(ret);
+	}
+ out:
+	up_read(&BTRFS_I(inode)->dio_sem);
+
+	return em;
+}
+
 static struct extent_map *btrfs_new_extent_direct(struct inode *inode,
 						  u64 start, u64 len)
 {
@@ -7144,41 +7216,13 @@ static struct extent_map *btrfs_new_extent_direct(struct inode *inode,
 	if (ret)
 		return ERR_PTR(ret);
 
-	/*
-	 * Create the ordered extent before the extent map. This is to avoid
-	 * races with the fast fsync path that would lead to it logging file
-	 * extent items that point to disk extents that were not yet written to.
-	 * The fast fsync path collects ordered extents into a local list and
-	 * then collects all the new extent maps, so we must create the ordered
-	 * extent first and make sure the fast fsync path collects any new
-	 * ordered extents after collecting new extent maps as well.
-	 * The fsync path simply can not rely on inode_dio_wait() because it
-	 * causes deadlock with AIO.
-	 */
-	ret = btrfs_add_ordered_extent_dio(inode, start, ins.objectid,
-					   ins.offset, ins.offset, 0);
-	if (ret) {
+	em = btrfs_create_dio_extent(inode, start, ins.offset, start,
+				     ins.objectid, ins.offset, ins.offset,
+				     ins.offset, 0);
+	btrfs_dec_block_group_reservations(root->fs_info, ins.objectid);
+	if (IS_ERR(em))
 		btrfs_free_reserved_extent(root, ins.objectid, ins.offset, 1);
-		return ERR_PTR(ret);
-	}
-
-	em = create_pinned_em(inode, start, ins.offset, start, ins.objectid,
-			      ins.offset, ins.offset, ins.offset, 0);
-	if (IS_ERR(em)) {
-		struct btrfs_ordered_extent *oe;
 
-		btrfs_free_reserved_extent(root, ins.objectid, ins.offset, 1);
-		oe = btrfs_lookup_ordered_extent(inode, start);
-		ASSERT(oe);
-		if (WARN_ON(!oe))
-			return em;
-		set_bit(BTRFS_ORDERED_IOERR, &oe->flags);
-		set_bit(BTRFS_ORDERED_IO_DONE, &oe->flags);
-		btrfs_remove_ordered_extent(inode, oe);
-		/* Once for our lookup and once for the ordered extents tree. */
-		btrfs_put_ordered_extent(oe);
-		btrfs_put_ordered_extent(oe);
-	}
 	return em;
 }
 
@@ -7408,7 +7452,7 @@ static int lock_extent_direct(struct inode *inode, u64 lockstart, u64 lockend,
 				 cached_state);
 		/*
 		 * We're concerned with the entire range that we're going to be
-		 * doing DIO to, so we need to make sure theres no ordered
+		 * doing DIO to, so we need to make sure there's no ordered
 		 * extents in this range.
 		 */
 		ordered = btrfs_lookup_ordered_range(inode, lockstart,
@@ -7570,7 +7614,7 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
 	if (current->journal_info) {
 		/*
 		 * Need to pull our outstanding extents and set journal_info to NULL so
-		 * that anything that needs to check if there's a transction doesn't get
+		 * that anything that needs to check if there's a transaction doesn't get
 		 * confused.
 		 */
 		dio_data = current->journal_info;
@@ -7603,7 +7647,7 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
 	 * decompress it, so there will be buffering required no matter what we
 	 * do, so go ahead and fallback to buffered.
 	 *
-	 * We return -ENOTBLK because thats what makes DIO go ahead and go back
+	 * We return -ENOTBLK because that's what makes DIO go ahead and go back
 	 * to buffered IO.  Don't blame me, this is the price we pay for using
 	 * the generic code.
 	 */
@@ -7650,24 +7694,21 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
 		block_start = em->block_start + (start - em->start);
 
 		if (can_nocow_extent(inode, start, &len, &orig_start,
-				     &orig_block_len, &ram_bytes) == 1) {
+				     &orig_block_len, &ram_bytes) == 1 &&
+		    btrfs_inc_nocow_writers(root->fs_info, block_start)) {
+			struct extent_map *em2;
+
+			em2 = btrfs_create_dio_extent(inode, start, len,
+						      orig_start, block_start,
+						      len, orig_block_len,
+						      ram_bytes, type);
+			btrfs_dec_nocow_writers(root->fs_info, block_start);
 			if (type == BTRFS_ORDERED_PREALLOC) {
 				free_extent_map(em);
-				em = create_pinned_em(inode, start, len,
-						       orig_start,
-						       block_start, len,
-						       orig_block_len,
-						       ram_bytes, type);
-				if (IS_ERR(em)) {
-					ret = PTR_ERR(em);
-					goto unlock_err;
-				}
+				em = em2;
 			}
-
-			ret = btrfs_add_ordered_extent_dio(inode, start,
-					   block_start, len, len, type);
-			if (ret) {
-				free_extent_map(em);
+			if (em2 && IS_ERR(em2)) {
+				ret = PTR_ERR(em2);
 				goto unlock_err;
 			}
 			goto unlock;
@@ -9019,7 +9060,7 @@ static int btrfs_truncate(struct inode *inode)
 		return ret;
 
 	/*
-	 * Yes ladies and gentelment, this is indeed ugly.  The fact is we have
+	 * Yes ladies and gentlemen, this is indeed ugly.  The fact is we have
 	 * 3 things going on here
 	 *
 	 * 1) We need to reserve space for our orphan item and the space to
@@ -9033,15 +9074,15 @@ static int btrfs_truncate(struct inode *inode)
 	 * space reserved in case it uses space during the truncate (thank you
 	 * very much snapshotting).
 	 *
-	 * And we need these to all be seperate.  The fact is we can use alot of
+	 * And we need these to all be separate.  The fact is we can use a lot of
 	 * space doing the truncate, and we have no earthly idea how much space
-	 * we will use, so we need the truncate reservation to be seperate so it
+	 * we will use, so we need the truncate reservation to be separate so it
 	 * doesn't end up using space reserved for updating the inode or
 	 * removing the orphan item.  We also need to be able to stop the
 	 * transaction and start a new one, which means we need to be able to
 	 * update the inode several times, and we have no idea of knowing how
 	 * many times that will be, so we can't just reserve 1 item for the
-	 * entirety of the opration, so that has to be done seperately as well.
+	 * entirety of the operation, so that has to be done separately as well.
 	 * Then there is the orphan item, which does indeed need to be held on
 	 * to for the whole operation, and we need nobody to touch this reserved
 	 * space except the orphan code.
@@ -9230,6 +9271,7 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
 	INIT_LIST_HEAD(&ei->delalloc_inodes);
 	INIT_LIST_HEAD(&ei->delayed_iput);
 	RB_CLEAR_NODE(&ei->rb_node);
+	init_rwsem(&ei->dio_sem);
 
 	return inode;
 }
@@ -9387,10 +9429,281 @@ static int btrfs_getattr(struct vfsmount *mnt,
 	return 0;
 }
 
+static int btrfs_rename_exchange(struct inode *old_dir,
+			      struct dentry *old_dentry,
+			      struct inode *new_dir,
+			      struct dentry *new_dentry)
+{
+	struct btrfs_trans_handle *trans;
+	struct btrfs_root *root = BTRFS_I(old_dir)->root;
+	struct btrfs_root *dest = BTRFS_I(new_dir)->root;
+	struct inode *new_inode = new_dentry->d_inode;
+	struct inode *old_inode = old_dentry->d_inode;
+	struct timespec ctime = CURRENT_TIME;
+	struct dentry *parent;
+	u64 old_ino = btrfs_ino(old_inode);
+	u64 new_ino = btrfs_ino(new_inode);
+	u64 old_idx = 0;
+	u64 new_idx = 0;
+	u64 root_objectid;
+	int ret;
+	bool root_log_pinned = false;
+	bool dest_log_pinned = false;
+
+	/* we only allow rename subvolume link between subvolumes */
+	if (old_ino != BTRFS_FIRST_FREE_OBJECTID && root != dest)
+		return -EXDEV;
+
+	/* close the race window with snapshot create/destroy ioctl */
+	if (old_ino == BTRFS_FIRST_FREE_OBJECTID)
+		down_read(&root->fs_info->subvol_sem);
+	if (new_ino == BTRFS_FIRST_FREE_OBJECTID)
+		down_read(&dest->fs_info->subvol_sem);
+
+	/*
+	 * 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 2 for the new links, so 12 total items
+	 * should cover the worst case number of items we'll modify.
+	 */
+	trans = btrfs_start_transaction(root, 12);
+	if (IS_ERR(trans)) {
+		ret = PTR_ERR(trans);
+		goto out_notrans;
+	}
+
+	/*
+	 * We need to find a free sequence number both in the source and
+	 * in the destination directory for the exchange.
+	 */
+	ret = btrfs_set_inode_index(new_dir, &old_idx);
+	if (ret)
+		goto out_fail;
+	ret = btrfs_set_inode_index(old_dir, &new_idx);
+	if (ret)
+		goto out_fail;
+
+	BTRFS_I(old_inode)->dir_index = 0ULL;
+	BTRFS_I(new_inode)->dir_index = 0ULL;
+
+	/* Reference for the source. */
+	if (old_ino == BTRFS_FIRST_FREE_OBJECTID) {
+		/* force full log commit if subvolume involved. */
+		btrfs_set_log_full_commit(root->fs_info, trans);
+	} else {
+		btrfs_pin_log_trans(root);
+		root_log_pinned = true;
+		ret = btrfs_insert_inode_ref(trans, dest,
+					     new_dentry->d_name.name,
+					     new_dentry->d_name.len,
+					     old_ino,
+					     btrfs_ino(new_dir), old_idx);
+		if (ret)
+			goto out_fail;
+	}
+
+	/* And now for the dest. */
+	if (new_ino == BTRFS_FIRST_FREE_OBJECTID) {
+		/* force full log commit if subvolume involved. */
+		btrfs_set_log_full_commit(dest->fs_info, trans);
+	} else {
+		btrfs_pin_log_trans(dest);
+		dest_log_pinned = true;
+		ret = btrfs_insert_inode_ref(trans, root,
+					     old_dentry->d_name.name,
+					     old_dentry->d_name.len,
+					     new_ino,
+					     btrfs_ino(old_dir), new_idx);
+		if (ret)
+			goto out_fail;
+	}
+
+	/* Update inode version and ctime/mtime. */
+	inode_inc_iversion(old_dir);
+	inode_inc_iversion(new_dir);
+	inode_inc_iversion(old_inode);
+	inode_inc_iversion(new_inode);
+	old_dir->i_ctime = old_dir->i_mtime = ctime;
+	new_dir->i_ctime = new_dir->i_mtime = ctime;
+	old_inode->i_ctime = ctime;
+	new_inode->i_ctime = ctime;
+
+	if (old_dentry->d_parent != new_dentry->d_parent) {
+		btrfs_record_unlink_dir(trans, old_dir, old_inode, 1);
+		btrfs_record_unlink_dir(trans, new_dir, new_inode, 1);
+	}
+
+	/* src is a subvolume */
+	if (old_ino == BTRFS_FIRST_FREE_OBJECTID) {
+		root_objectid = BTRFS_I(old_inode)->root->root_key.objectid;
+		ret = btrfs_unlink_subvol(trans, root, old_dir,
+					  root_objectid,
+					  old_dentry->d_name.name,
+					  old_dentry->d_name.len);
+	} else { /* src is an inode */
+		ret = __btrfs_unlink_inode(trans, root, old_dir,
+					   old_dentry->d_inode,
+					   old_dentry->d_name.name,
+					   old_dentry->d_name.len);
+		if (!ret)
+			ret = btrfs_update_inode(trans, root, old_inode);
+	}
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out_fail;
+	}
+
+	/* dest is a subvolume */
+	if (new_ino == BTRFS_FIRST_FREE_OBJECTID) {
+		root_objectid = BTRFS_I(new_inode)->root->root_key.objectid;
+		ret = btrfs_unlink_subvol(trans, dest, new_dir,
+					  root_objectid,
+					  new_dentry->d_name.name,
+					  new_dentry->d_name.len);
+	} else { /* dest is an inode */
+		ret = __btrfs_unlink_inode(trans, dest, new_dir,
+					   new_dentry->d_inode,
+					   new_dentry->d_name.name,
+					   new_dentry->d_name.len);
+		if (!ret)
+			ret = btrfs_update_inode(trans, dest, new_inode);
+	}
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out_fail;
+	}
+
+	ret = btrfs_add_link(trans, new_dir, old_inode,
+			     new_dentry->d_name.name,
+			     new_dentry->d_name.len, 0, old_idx);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out_fail;
+	}
+
+	ret = btrfs_add_link(trans, old_dir, new_inode,
+			     old_dentry->d_name.name,
+			     old_dentry->d_name.len, 0, new_idx);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out_fail;
+	}
+
+	if (old_inode->i_nlink == 1)
+		BTRFS_I(old_inode)->dir_index = old_idx;
+	if (new_inode->i_nlink == 1)
+		BTRFS_I(new_inode)->dir_index = new_idx;
+
+	if (root_log_pinned) {
+		parent = new_dentry->d_parent;
+		btrfs_log_new_name(trans, old_inode, old_dir, parent);
+		btrfs_end_log_trans(root);
+		root_log_pinned = false;
+	}
+	if (dest_log_pinned) {
+		parent = old_dentry->d_parent;
+		btrfs_log_new_name(trans, new_inode, new_dir, parent);
+		btrfs_end_log_trans(dest);
+		dest_log_pinned = false;
+	}
+out_fail:
+	/*
+	 * If we have pinned a log and an error happened, we unpin tasks
+	 * trying to sync the log and force them to fallback to a transaction
+	 * commit if the log currently contains any of the inodes involved in
+	 * this rename operation (to ensure we do not persist a log with an
+	 * inconsistent state for any of these inodes or leading to any
+	 * inconsistencies when replayed). If the transaction was aborted, the
+	 * abortion reason is propagated to userspace when attempting to commit
+	 * the transaction. If the log does not contain any of these inodes, we
+	 * allow the tasks to sync it.
+	 */
+	if (ret && (root_log_pinned || dest_log_pinned)) {
+		if (btrfs_inode_in_log(old_dir, root->fs_info->generation) ||
+		    btrfs_inode_in_log(new_dir, root->fs_info->generation) ||
+		    btrfs_inode_in_log(old_inode, root->fs_info->generation) ||
+		    (new_inode &&
+		     btrfs_inode_in_log(new_inode, root->fs_info->generation)))
+		    btrfs_set_log_full_commit(root->fs_info, trans);
+
+		if (root_log_pinned) {
+			btrfs_end_log_trans(root);
+			root_log_pinned = false;
+		}
+		if (dest_log_pinned) {
+			btrfs_end_log_trans(dest);
+			dest_log_pinned = false;
+		}
+	}
+	ret = btrfs_end_transaction(trans, root);
+out_notrans:
+	if (new_ino == BTRFS_FIRST_FREE_OBJECTID)
+		up_read(&dest->fs_info->subvol_sem);
+	if (old_ino == BTRFS_FIRST_FREE_OBJECTID)
+		up_read(&root->fs_info->subvol_sem);
+
+	return ret;
+}
+
+static int btrfs_whiteout_for_rename(struct btrfs_trans_handle *trans,
+				     struct btrfs_root *root,
+				     struct inode *dir,
+				     struct dentry *dentry)
+{
+	int ret;
+	struct inode *inode;
+	u64 objectid;
+	u64 index;
+
+	ret = btrfs_find_free_ino(root, &objectid);
+	if (ret)
+		return ret;
+
+	inode = btrfs_new_inode(trans, root, dir,
+				dentry->d_name.name,
+				dentry->d_name.len,
+				btrfs_ino(dir),
+				objectid,
+				S_IFCHR | WHITEOUT_MODE,
+				&index);
+
+	if (IS_ERR(inode)) {
+		ret = PTR_ERR(inode);
+		return ret;
+	}
+
+	inode->i_op = &btrfs_special_inode_operations;
+	init_special_inode(inode, inode->i_mode,
+		WHITEOUT_DEV);
+
+	ret = btrfs_init_inode_security(trans, inode, dir,
+				&dentry->d_name);
+	if (ret)
+		goto out;
+
+	ret = btrfs_add_nondir(trans, dir, dentry,
+				inode, 0, index);
+	if (ret)
+		goto out;
+
+	ret = btrfs_update_inode(trans, root, inode);
+out:
+	unlock_new_inode(inode);
+	if (ret)
+		inode_dec_link_count(inode);
+	iput(inode);
+
+	return ret;
+}
+
 static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
-			   struct inode *new_dir, struct dentry *new_dentry)
+			   struct inode *new_dir, struct dentry *new_dentry,
+			   unsigned int flags)
 {
 	struct btrfs_trans_handle *trans;
+	unsigned int trans_num_items;
 	struct btrfs_root *root = BTRFS_I(old_dir)->root;
 	struct btrfs_root *dest = BTRFS_I(new_dir)->root;
 	struct inode *new_inode = d_inode(new_dentry);
@@ -9399,6 +9712,7 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	u64 root_objectid;
 	int ret;
 	u64 old_ino = btrfs_ino(old_inode);
+	bool log_pinned = false;
 
 	if (btrfs_ino(new_dir) == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)
 		return -EPERM;
@@ -9449,15 +9763,21 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	 * 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
+	 * would require 5 item modifications, so we'll assume they are 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.
+	 * If our rename has the whiteout flag, we need more 5 units for the
+	 * new inode (1 inode item, 1 inode ref, 2 dir items and 1 xattr item
+	 * when selinux is enabled).
 	 */
-	trans = btrfs_start_transaction(root, 11);
+	trans_num_items = 11;
+	if (flags & RENAME_WHITEOUT)
+		trans_num_items += 5;
+	trans = btrfs_start_transaction(root, trans_num_items);
 	if (IS_ERR(trans)) {
-                ret = PTR_ERR(trans);
-                goto out_notrans;
-        }
+		ret = PTR_ERR(trans);
+		goto out_notrans;
+	}
 
 	if (dest != root)
 		btrfs_record_root_in_trans(trans, dest);
@@ -9471,6 +9791,8 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 		/* force full log commit if subvolume involved. */
 		btrfs_set_log_full_commit(root->fs_info, trans);
 	} else {
+		btrfs_pin_log_trans(root);
+		log_pinned = true;
 		ret = btrfs_insert_inode_ref(trans, dest,
 					     new_dentry->d_name.name,
 					     new_dentry->d_name.len,
@@ -9478,14 +9800,6 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 					     btrfs_ino(new_dir), index);
 		if (ret)
 			goto out_fail;
-		/*
-		 * this is an ugly little race, but the rename is required
-		 * to make sure that if we crash, the inode is either at the
-		 * old name or the new one.  pinning the log transaction lets
-		 * us make sure we don't allow a log commit to come in after
-		 * we unlink the name but before we add the new name back in.
-		 */
-		btrfs_pin_log_trans(root);
 	}
 
 	inode_inc_iversion(old_dir);
@@ -9552,12 +9866,46 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	if (old_inode->i_nlink == 1)
 		BTRFS_I(old_inode)->dir_index = index;
 
-	if (old_ino != BTRFS_FIRST_FREE_OBJECTID) {
+	if (log_pinned) {
 		struct dentry *parent = new_dentry->d_parent;
+
 		btrfs_log_new_name(trans, old_inode, old_dir, parent);
 		btrfs_end_log_trans(root);
+		log_pinned = false;
+	}
+
+	if (flags & RENAME_WHITEOUT) {
+		ret = btrfs_whiteout_for_rename(trans, root, old_dir,
+						old_dentry);
+
+		if (ret) {
+			btrfs_abort_transaction(trans, root, ret);
+			goto out_fail;
+		}
 	}
 out_fail:
+	/*
+	 * If we have pinned the log and an error happened, we unpin tasks
+	 * trying to sync the log and force them to fallback to a transaction
+	 * commit if the log currently contains any of the inodes involved in
+	 * this rename operation (to ensure we do not persist a log with an
+	 * inconsistent state for any of these inodes or leading to any
+	 * inconsistencies when replayed). If the transaction was aborted, the
+	 * abortion reason is propagated to userspace when attempting to commit
+	 * the transaction. If the log does not contain any of these inodes, we
+	 * allow the tasks to sync it.
+	 */
+	if (ret && log_pinned) {
+		if (btrfs_inode_in_log(old_dir, root->fs_info->generation) ||
+		    btrfs_inode_in_log(new_dir, root->fs_info->generation) ||
+		    btrfs_inode_in_log(old_inode, root->fs_info->generation) ||
+		    (new_inode &&
+		     btrfs_inode_in_log(new_inode, root->fs_info->generation)))
+		    btrfs_set_log_full_commit(root->fs_info, trans);
+
+		btrfs_end_log_trans(root);
+		log_pinned = false;
+	}
 	btrfs_end_transaction(trans, root);
 out_notrans:
 	if (old_ino == BTRFS_FIRST_FREE_OBJECTID)
@@ -9570,10 +9918,14 @@ static int btrfs_rename2(struct inode *old_dir, struct dentry *old_dentry,
 			 struct inode *new_dir, struct dentry *new_dentry,
 			 unsigned int flags)
 {
-	if (flags & ~RENAME_NOREPLACE)
+	if (flags & ~(RENAME_NOREPLACE | RENAME_EXCHANGE | RENAME_WHITEOUT))
 		return -EINVAL;
 
-	return btrfs_rename(old_dir, old_dentry, new_dir, new_dentry);
+	if (flags & RENAME_EXCHANGE)
+		return btrfs_rename_exchange(old_dir, old_dentry, new_dir,
+					  new_dentry);
+
+	return btrfs_rename(old_dir, old_dentry, new_dir, new_dentry, flags);
 }
 
 static void btrfs_run_delalloc_work(struct btrfs_work *work)
@@ -9942,6 +10294,7 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 				btrfs_end_transaction(trans, root);
 			break;
 		}
+		btrfs_dec_block_group_reservations(root->fs_info, ins.objectid);
 
 		last_alloc = ins.offset;
 		ret = insert_reserved_file_extent(trans, inode,
@@ -10184,7 +10537,7 @@ static const struct file_operations btrfs_dir_file_operations = {
 	.iterate	= btrfs_real_readdir,
 	.unlocked_ioctl	= btrfs_ioctl,
 #ifdef CONFIG_COMPAT
-	.compat_ioctl	= btrfs_ioctl,
+	.compat_ioctl	= btrfs_compat_ioctl,
 #endif
 	.release        = btrfs_release_file,
 	.fsync		= btrfs_sync_file,