The same idea was used in a real-world attack. It used to be possible on the Tenex system in the early seventies to recover a password by laying out an attempted password across a page boundary, and observing whether the checker incurred a page fault.
The bug was that you could align the given password string so that it was at the end of a page boundary, and have the next page in the address space mapped to a non-existant page of a write-protected file. Normally, Tenex would create a page when you tried to access a non-existant page, but in this case it couldn't (since the file was write-protected).
So, you did a password-checking system call (e.g. the system call which tried to obtain owner access to another directory) set up in this way and you would get one of two failures: Incorrect Password meant that your prefix was wrong, Page Fault meant that your prefix was right but that you need more characters.