The native app development is increased in Android systems to implement CPU-intensive applications such as game engines, signal processing, and physics simulation. However, native code analysis is very difficult and requires a lot of time which explains the limited number of systems that track information flow in native libraries. But, none of them detects the sensitive information leakage through control flows at native level. In this paper, we combine dynamic and static taint analysis to propagate taint along control dependencies. Our approach has proven to be effective in analyzing several malicious Android applications that invoke native librairies with reasonable performance overheads.