In reply to svishnu:
The sign of the result of (m-n) is unsigned - there is no sign. You cannot find out which has the large magnitude without extra logic. You need to explain why you cannot use the relational operators < or >, which is the easiest.
This is all explained as unsigned overflow.