In this paper, we consider the full information and the state feedback H-2 control problems for continuous-time descriptor systems. An explicit characterization of all optimal controllers is given for these cases. As a consequence, we are able to present a parameterization of all optimal static and dynamic controllers for the LQR problem via state. feedback. We also show that, for the general state feedback H-2 optimization, the usual gain matrix defined as a linear function of a solution of a generalized Riccati equation can be non-optimal.