 File deletion

File deletion removes a file from the live naming and allocation structures of a file system. It must be distinguished from data sanitization, which concerns whether former information remains recoverable. The exact lifetime rules depend on the operating system and file system; the Unix example here uses the documented Linux unlink interface.

A successful unlink removes a name, represented by a directory entry. If another hard link names the same file, that other name still works. If the final name disappears while the file remains open, existing file descriptors can continue to access it; storage becomes reusable after the final open reference is closed.^[ref]

For an original lifetime trace, let names a and b refer to inode 42, and let a process hold an open file descriptor for it. Removing a reduces its hard link count from two to one. Removing b reduces the count to zero, but the open reference still reaches inode 42. Closing that final reference permits reclamation. The counts describe naming and lifetime, not how many times the file data blocks have been overwritten.

Deletion also follows the named object's type. If the final path component is a symbolic link, unlink removes that link rather than its target.^[ref] A deleted name, a reclaimed allocation and physically erased content are separate claims; observing one does not establish all three.