Merge tag 'for-5.8-rc2-tag' of git://git.kernel.org/pub/scm/linux/ker…

…nel/git/kdave/linux

Pull btrfs fixes from David Sterba:
 "A number of fixes, located in two areas, one performance fix and one
  fixup for better integration with another patchset.

   - bug fixes in nowait aio:
       - fix snapshot creation hang after nowait-aio was used
       - fix failure to write to prealloc extent past EOF
       - don't block when extent range is locked

   - block group fixes:
       - relocation failure when scrub runs in parallel
       - refcount fix when removing fails
       - fix race between removal and creation
       - space accounting fixes

   - reinstante fast path check for log tree at unlink time, fixes
     performance drop up to 30% in REAIM

   - kzfree/kfree fixup to ease treewide patchset renaming kzfree"

* tag 'for-5.8-rc2-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
  btrfs: use kfree() in btrfs_ioctl_get_subvol_info()
  btrfs: fix RWF_NOWAIT writes blocking on extent locks and waiting for IO
  btrfs: fix RWF_NOWAIT write not failling when we need to cow
  btrfs: fix failure of RWF_NOWAIT write into prealloc extent beyond eof
  btrfs: fix hang on snapshot creation after RWF_NOWAIT write
  btrfs: check if a log root exists before locking the log_mutex on unlink
  btrfs: fix bytes_may_use underflow when running balance and scrub in parallel
  btrfs: fix data block group relocation failure due to concurrent scrub
  btrfs: fix race between block group removal and block group creation
  btrfs: fix a block group ref counter leak after failure to remove block group

fs/btrfs/block-group.c

@@ -940,7 +940,7 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 	path = btrfs_alloc_path();
 	if (!path) {
 		ret = -ENOMEM;
-		goto out_put_group;
+		goto out;
 	}
 
 	/*
@@ -978,7 +978,7 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 		ret = btrfs_orphan_add(trans, BTRFS_I(inode));
 		if (ret) {
 			btrfs_add_delayed_iput(inode);
-			goto out_put_group;
+			goto out;
 		}
 		clear_nlink(inode);
 		/* One for the block groups ref */
@@ -1001,13 +1001,13 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 
 	ret = btrfs_search_slot(trans, tree_root, &key, path, -1, 1);
 	if (ret < 0)
-		goto out_put_group;
+		goto out;
 	if (ret > 0)
 		btrfs_release_path(path);
 	if (ret == 0) {
 		ret = btrfs_del_item(trans, tree_root, path);
 		if (ret)
-			goto out_put_group;
+			goto out;
 		btrfs_release_path(path);
 	}
 
@@ -1016,6 +1016,9 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 		 &fs_info->block_group_cache_tree);
 	RB_CLEAR_NODE(&block_group->cache_node);
 
+	/* Once for the block groups rbtree */
+	btrfs_put_block_group(block_group);
+
 	if (fs_info->first_logical_byte == block_group->start)
 		fs_info->first_logical_byte = (u64)-1;
 	spin_unlock(&fs_info->block_group_cache_lock);
@@ -1089,6 +1092,25 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 
 	spin_unlock(&block_group->space_info->lock);
 
+	/*
+	 * Remove the free space for the block group from the free space tree
+	 * and the block group's item from the extent tree before marking the
+	 * block group as removed. This is to prevent races with tasks that
+	 * freeze and unfreeze a block group, this task and another task
+	 * allocating a new block group - the unfreeze task ends up removing
+	 * the block group's extent map before the task calling this function
+	 * deletes the block group item from the extent tree, allowing for
+	 * another task to attempt to create another block group with the same
+	 * item key (and failing with -EEXIST and a transaction abort).
+	 */
+	ret = remove_block_group_free_space(trans, block_group);
+	if (ret)
+		goto out;
+
+	ret = remove_block_group_item(trans, path, block_group);
+	if (ret < 0)
+		goto out;
+
 	mutex_lock(&fs_info->chunk_mutex);
 	spin_lock(&block_group->lock);
 	block_group->removed = 1;
@@ -1123,17 +1145,6 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 
 	mutex_unlock(&fs_info->chunk_mutex);
 
-	ret = remove_block_group_free_space(trans, block_group);
-	if (ret)
-		goto out_put_group;
-
-	/* Once for the block groups rbtree */
-	btrfs_put_block_group(block_group);
-
-	ret = remove_block_group_item(trans, path, block_group);
-	if (ret < 0)
-		goto out;
-
 	if (remove_em) {
 		struct extent_map_tree *em_tree;
 
@@ -1145,10 +1156,9 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 		free_extent_map(em);
 	}
 
-out_put_group:
+out:
 	/* Once for the lookup reference */
 	btrfs_put_block_group(block_group);
-out:
 	if (remove_rsv)
 		btrfs_delayed_refs_rsv_release(fs_info, 1);
 	btrfs_free_path(path);

fs/btrfs/ctree.h

@@ -1009,6 +1009,8 @@ enum {
 	BTRFS_ROOT_DEAD_RELOC_TREE,
 	/* Mark dead root stored on device whose cleanup needs to be resumed */
 	BTRFS_ROOT_DEAD_TREE,
+	/* The root has a log tree. Used only for subvolume roots. */
+	BTRFS_ROOT_HAS_LOG_TREE,
 };
 
 /*

fs/btrfs/file.c

@@ -1533,35 +1533,51 @@ lock_and_cleanup_extent_if_need(struct btrfs_inode *inode, struct page **pages,
 }
 
 static noinline int check_can_nocow(struct btrfs_inode *inode, loff_t pos,
-				    size_t *write_bytes)
+				    size_t *write_bytes, bool nowait)
 {
 	struct btrfs_fs_info *fs_info = inode->root->fs_info;
 	struct btrfs_root *root = inode->root;
 	u64 lockstart, lockend;
 	u64 num_bytes;
 	int ret;
 
-	if (!btrfs_drew_try_write_lock(&root->snapshot_lock))
+	if (!nowait && !btrfs_drew_try_write_lock(&root->snapshot_lock))
 		return -EAGAIN;
 
 	lockstart = round_down(pos, fs_info->sectorsize);
 	lockend = round_up(pos + *write_bytes,
 			   fs_info->sectorsize) - 1;
+	num_bytes = lockend - lockstart + 1;
 
-	btrfs_lock_and_flush_ordered_range(inode, lockstart,
-					   lockend, NULL);
+	if (nowait) {
+		struct btrfs_ordered_extent *ordered;
+
+		if (!try_lock_extent(&inode->io_tree, lockstart, lockend))
+			return -EAGAIN;
+
+		ordered = btrfs_lookup_ordered_range(inode, lockstart,
+						     num_bytes);
+		if (ordered) {
+			btrfs_put_ordered_extent(ordered);
+			ret = -EAGAIN;
+			goto out_unlock;
+		}
+	} else {
+		btrfs_lock_and_flush_ordered_range(inode, lockstart,
+						   lockend, NULL);
+	}
 
-	num_bytes = lockend - lockstart + 1;
 	ret = can_nocow_extent(&inode->vfs_inode, lockstart, &num_bytes,
 			NULL, NULL, NULL);
 	if (ret <= 0) {
 		ret = 0;
-		btrfs_drew_write_unlock(&root->snapshot_lock);
+		if (!nowait)
+			btrfs_drew_write_unlock(&root->snapshot_lock);
 	} else {
 		*write_bytes = min_t(size_t, *write_bytes ,
 				     num_bytes - pos + lockstart);
 	}
-
+out_unlock:
 	unlock_extent(&inode->io_tree, lockstart, lockend);
 
 	return ret;
@@ -1633,7 +1649,7 @@ static noinline ssize_t btrfs_buffered_write(struct kiocb *iocb,
 			if ((BTRFS_I(inode)->flags & (BTRFS_INODE_NODATACOW |
 						      BTRFS_INODE_PREALLOC)) &&
 			    check_can_nocow(BTRFS_I(inode), pos,
-					&write_bytes) > 0) {
+					    &write_bytes, false) > 0) {
 				/*
 				 * For nodata cow case, no need to reserve
 				 * data space.
@@ -1904,13 +1920,25 @@ static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
 	pos = iocb->ki_pos;
 	count = iov_iter_count(from);
 	if (iocb->ki_flags & IOCB_NOWAIT) {
+		size_t nocow_bytes = count;
+
 		/*
 		 * We will allocate space in case nodatacow is not set,
 		 * so bail
 		 */
 		if (!(BTRFS_I(inode)->flags & (BTRFS_INODE_NODATACOW |
 					      BTRFS_INODE_PREALLOC)) ||
-		    check_can_nocow(BTRFS_I(inode), pos, &count) <= 0) {
+		    check_can_nocow(BTRFS_I(inode), pos, &nocow_bytes,
+				    true) <= 0) {
+			inode_unlock(inode);
+			return -EAGAIN;
+		}
+		/*
+		 * There are holes in the range or parts of the range that must
+		 * be COWed (shared extents, RO block groups, etc), so just bail
+		 * out.
+		 */
+		if (nocow_bytes < count) {
 			inode_unlock(inode);
 			return -EAGAIN;
 		}

fs/btrfs/inode.c

@@ -985,6 +985,7 @@ static noinline int cow_file_range(struct inode *inode,
 	u64 num_bytes;
 	unsigned long ram_size;
 	u64 cur_alloc_size = 0;
+	u64 min_alloc_size;
 	u64 blocksize = fs_info->sectorsize;
 	struct btrfs_key ins;
 	struct extent_map *em;
@@ -1035,10 +1036,26 @@ static noinline int cow_file_range(struct inode *inode,
 	btrfs_drop_extent_cache(BTRFS_I(inode), start,
 			start + num_bytes - 1, 0);
 
+	/*
+	 * Relocation relies on the relocated extents to have exactly the same
+	 * size as the original extents. Normally writeback for relocation data
+	 * extents follows a NOCOW path because relocation preallocates the
+	 * extents. However, due to an operation such as scrub turning a block
+	 * group to RO mode, it may fallback to COW mode, so we must make sure
+	 * an extent allocated during COW has exactly the requested size and can
+	 * not be split into smaller extents, otherwise relocation breaks and
+	 * fails during the stage where it updates the bytenr of file extent
+	 * items.
+	 */
+	if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID)
+		min_alloc_size = num_bytes;
+	else
+		min_alloc_size = fs_info->sectorsize;
+
 	while (num_bytes > 0) {
 		cur_alloc_size = num_bytes;
 		ret = btrfs_reserve_extent(root, cur_alloc_size, cur_alloc_size,
-					   fs_info->sectorsize, 0, alloc_hint,
+					   min_alloc_size, 0, alloc_hint,
 					   &ins, 1, 1);
 		if (ret < 0)
 			goto out_unlock;
@@ -1361,6 +1378,8 @@ static int fallback_to_cow(struct inode *inode, struct page *locked_page,
 			   int *page_started, unsigned long *nr_written)
 {
 	const bool is_space_ino = btrfs_is_free_space_inode(BTRFS_I(inode));
+	const bool is_reloc_ino = (BTRFS_I(inode)->root->root_key.objectid ==
+				   BTRFS_DATA_RELOC_TREE_OBJECTID);
 	const u64 range_bytes = end + 1 - start;
 	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
 	u64 range_start = start;
@@ -1391,18 +1410,23 @@ static int fallback_to_cow(struct inode *inode, struct page *locked_page,
 	 *    data space info, which we incremented in the step above.
 	 *
 	 * If we need to fallback to cow and the inode corresponds to a free
-	 * space cache inode, we must also increment bytes_may_use of the data
-	 * space_info for the same reason. Space caches always get a prealloc
+	 * space cache inode or an inode of the data relocation tree, we must
+	 * also increment bytes_may_use of the data space_info for the same
+	 * reason. Space caches and relocated data extents always get a prealloc
 	 * extent for them, however scrub or balance may have set the block
-	 * group that contains that extent to RO mode.
+	 * group that contains that extent to RO mode and therefore force COW
+	 * when starting writeback.
 	 */
 	count = count_range_bits(io_tree, &range_start, end, range_bytes,
 				 EXTENT_NORESERVE, 0);
-	if (count > 0 || is_space_ino) {
-		const u64 bytes = is_space_ino ? range_bytes : count;
+	if (count > 0 || is_space_ino || is_reloc_ino) {
+		u64 bytes = count;
 		struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
 		struct btrfs_space_info *sinfo = fs_info->data_sinfo;
 
+		if (is_space_ino || is_reloc_ino)
+			bytes = range_bytes;
+
 		spin_lock(&sinfo->lock);
 		btrfs_space_info_update_bytes_may_use(fs_info, sinfo, bytes);
 		spin_unlock(&sinfo->lock);
@@ -7865,9 +7889,6 @@ static ssize_t btrfs_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
 			dio_data.overwrite = 1;
 			inode_unlock(inode);
 			relock = true;
-		} else if (iocb->ki_flags & IOCB_NOWAIT) {
-			ret = -EAGAIN;
-			goto out;
 		}
 		ret = btrfs_delalloc_reserve_space(inode, &data_reserved,
 						   offset, count);

fs/btrfs/ioctl.c

@@ -2692,7 +2692,7 @@ static int btrfs_ioctl_get_subvol_info(struct file *file, void __user *argp)
 	btrfs_put_root(root);
 out_free:
 	btrfs_free_path(path);
-	kzfree(subvol_info);
+	kfree(subvol_info);
 	return ret;
 }
 

fs/btrfs/tree-log.c

@@ -169,6 +169,7 @@ static int start_log_trans(struct btrfs_trans_handle *trans,
 		if (ret)
 			goto out;
 
+		set_bit(BTRFS_ROOT_HAS_LOG_TREE, &root->state);
 		clear_bit(BTRFS_ROOT_MULTI_LOG_TASKS, &root->state);
 		root->log_start_pid = current->pid;
 	}
@@ -195,6 +196,9 @@ static int join_running_log_trans(struct btrfs_root *root)
 {
 	int ret = -ENOENT;
 
+	if (!test_bit(BTRFS_ROOT_HAS_LOG_TREE, &root->state))
+		return ret;
+
 	mutex_lock(&root->log_mutex);
 	if (root->log_root) {
 		ret = 0;
@@ -3303,6 +3307,7 @@ int btrfs_free_log(struct btrfs_trans_handle *trans, struct btrfs_root *root)
 	if (root->log_root) {
 		free_log_tree(trans, root->log_root);
 		root->log_root = NULL;
+		clear_bit(BTRFS_ROOT_HAS_LOG_TREE, &root->state);
 	}
 	return 0;
 }