Why are inner horizons Cauchy horizons? 
I know that RN black hole has two horizons, one outer one and one inner one. The outer one is the event horizon.
As far as I know, a Cauchy horizon is the boundary of the domain of dependence of a spatial hypersurface(the Cauchy surface). According to the Penrose diagram, the inner horizons are just part of the boundary. So why are the inner horizons the Cauchy horizons?
 A: Any event beyond the inner horizon, will have part of the timelike singularity in its causal past. Consequently, starting from the inner horizon the spacetime cannot be obtained as a Cauchy evolution of the initial data on the hypersurface $\Sigma$; it needs additional input in the form of a boundary condition at the singularity.
EDIT:
So, it appears the actual question was: why are the other parts of the boundary of the domain of dependence in the Penrose diagram not part of the Cauchy horizon? The other parts of the boundary are past and future null, timelike and spacelike infinity (in different asymptotic regions). The reason that these are not part of the Cauchy horizon is that they are not part of the Reissner-Nordström manifold in the first place. As the "infinity" in their names implies, these parts of the Penrose diagram cannot actually be reached by a geodesic. The actual manifold consists only of the interior of the Penrose diagram. Consequently the boundary of the domain of dependence of the hypersurface $\Sigma$ as a subset of the maximally extended Reissner-Nordström manifold, consists only of the past and future innerhorizons.
