Geodesically complete black holes

The 1965 Penrose singularity theorem demonstrates the utterly inevitable and unavoidable formation of spacetime singularities under physically reasonable assumptions, and it remains one of the main results in our understanding of black holes. It is standard lore that quantum gravitational effects will always tame these singularities in black hole interiors. However, the Penrose's theorem provides no clue as to the possible (nonsingular) geometries that may be realized in theories beyond general relativity as the result of singularity regularization. In this paper, we analyze this problem in spherically symmetric situations, being completely general otherwise, in particular regarding the dynamics of the gravitational and matter fields. Our main result is that, contrary to what one might expect, the set of regular geometries that arises is remarkably limited. We rederive geometries that have been analyzed before, but also uncover some new possibilities. Moreover, the complete catalog of possibilities that we obtain allows us to draw the novel conclusion that there is a clear tradeoff between internal and external consistency: one has to choose between models that display internal inconsistencies, or models that include significant deviations with respect to general relativity, which should therefore be amenable to observational tests via multimessenger astrophysics.