According to https://xfs.org/index.php/Shrinking_Support :
"Currently XFS Filesystems can't be shrunk."
You need backup the data and recreate the file system to shrink it.
But if you have a big xfs filesystem with not too much space used, For example, a 2T xfs volume with 100G data.
And you can take some risk to lose data.
Here is a way to shrink it without using new space.
The idea is:
1.temporary reduce the logic volume size, you can still read from xfs filesystem even after the partition or volume is shrunk if the real data is not wrote at the shrunk part
2.create a new volume
3.copy data
4.drop old volume then rename volume name back.
Notice: You may lost your data. Don't try it on important data.
You will break your file system if there's not enough free space in the filesytem.
./freecount.pl /dev/mapper/cl-data
Stop all processes using the filesystem, then reduce the volume, free enough space to copy the files.
lvreduce -L -200G /dev/mapper/cl-data
use ext4 if you may need reduce it later
lvcreate -L 200G -n newdata cl
mkfs.ext4 /dev/mapper/cl-newdata
mount /dev/mapper/cl-newdata /mnt
tar --ignore-failed-read -C /data -cf - . | tar --ignore-failed-read -v -C /mnt -xf -
umount /data
umount /mnt
lvremove /dev/mapper/cl-data
lvrename cl newdata data
lvextend -L +200G /dev/mapper/cl-data
e2fsck -f /dev/mapper/cl-data
resize2fs /dev/mapper/cl-data
mount /dev/mapper/cl-data /data
lvextend -L +200G /dev/mapper/cl-root
xfs_growfs /dev/mapper/cl-root