Computer computations are generally irreversible while the laws of physics are reversible. This mismatch is penalized by among other things generating excess thermic entropy in the computation. Computing performance has improved to the extent that efficiency degrades unless all algorithms are executed reversibly, for example by a universal reversible simulation of irreversible computations. All known reversible simulations are either space hungry or time hungry. The leanest method was proposed by Bennett and can be analyzed using a simple 'reversible' pebble game. The reachable reversible simulation instantaneous descriptions (pebble configurations) of such pebble games are characterized completely. As a corollary we obtain the reversible simulation by Bennett and, moreover, show that it is a space-optimal pebble game. We also introduce irreversible steps and give a theorem on the tradeoff between the number of allowed irreversible steps and the memory gain in the pebble game. In this resource-bounded setting the limited erasing needs to be performed at precise instants during the simulation. The reversible simulation can be modified so that it is applicable also when the simulated computation time is unknown.