Software countermeasures against power side-channel attacks are an active research area in cryptography and cryptanalysis. To date, no compiler provides dependable compilation from high-level languages since respective security properties are systematically broken at compile time. There is a strong demand for secure compilation as it is a difficult and error prone task to implement, adopt and verify handwritten assembly for each targeted platform. We provide a characterization of the probing security policy for power side-channel countermeasures, related preservation problems and sketch an approach for future work to close the gap between abstract algorithms and their secure implementation or compilation to relevant architectures.