I have a file-server which has already received numerous backups from other machines. I want to deduplicate existing files by the virtue of link-by-hash, however I'd prefer not copying the data around again - instead, I want the specified hash-directory created and used to link existing identical files into the same inode.

Also of use would be a "relaxed" vs. "strict" mode to allow or forbid creation of such hardlinks for files with same content and differing metadata (owner, date, permissions, etc.)