Btrfs: fix EIO on reading file after ioctl clone works on it

For inline data extent, we need to make its length aligned, otherwise,
we can get a phantom extent map which confuses readpages() to return -EIO.

This can be detected by xfstests/btrfs/035.

Reported-by: David Disseldorp <ddiss@suse.de>
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Chris Mason <clm@fb.com>

fs/btrfs/ioctl.c

@@ -3142,6 +3142,8 @@ static int btrfs_clone(struct inode *src, struct inode *inode,
 			} else if (type == BTRFS_FILE_EXTENT_INLINE) {
 				u64 skip = 0;
 				u64 trim = 0;
+				u64 aligned_end = 0;
+
 				if (off > key.offset) {
 					skip = off - key.offset;
 					new_key.offset += skip;
@@ -3158,9 +3160,11 @@ static int btrfs_clone(struct inode *src, struct inode *inode,
 				size -= skip + trim;
 				datal -= skip + trim;
 
+				aligned_end = ALIGN(new_key.offset + datal,
+						    root->sectorsize);
 				ret = btrfs_drop_extents(trans, root, inode,
 							 new_key.offset,
-							 new_key.offset + datal,
+							 aligned_end,
 							 1);
 				if (ret) {
 					if (ret != -EOPNOTSUPP)