I found the answer in one of the problems in Peskin.
Because in the corresponding quantum theory, this spinor field will be anti-commuting variables, therefore it makes sense to start with classical anti-commuting Grassmann numbers for the classical field. In this sense, the mass term will not vanish.