Do not store the return value of called functions in the same variable
as the (future) return value of the current function.

This makes tracking the origin of the value easier and reduces
the chance of introducing a new point of exit without resetting
the return value back to -1.