filededup does not check when it crosses filesystem boundaries.

[HankB]

If the root directory for the search includes multiple filesystems, they will all be searched for duplicate files. If duplicates are found within different filesystems, the program will attempt to and fail to link them. This has two potential consequences.

1. A lot of time could be spent searching child filesystems when duplicates cannot be linked.
2. It is unknown (e.g. no test cases) if duplicates within the child filesystems will be correctly handled.

The second issue opens the question of what should be correct behavior. Either stick to one filesystem or properly link files in child filesystems. It might make sense to add an additional flag to force a single filesystem on demand and make sure that duplicates in nested filesystems are handled correctly.

This is marked as a bug because a quick test indicates that duplicate files in a nested filesystem are not hard linked.

Severity is low as no data is lost, it just does not do what is desired.