The purpose of this article (...) is to derive an algorithm for solving stochastic linear quadratic control problems over infinite time horizon using a primal-dual semidefinite programming approach.