Recent experiments [N\'eel et al., PRL, vol. 91, p. 226103, 2003] have established that one- and two-dimensional instabilities, bunching and meandering respectively, occur simultaneously during step-flow growth on copper vicinal surfaces, in contrast to the predictions of existing models that trace back to the 1951 article by Burton, Cabrera, and Frank. Indeed, in the BCF framework, meandering is contingent upon the existence of an Ehrlich--Schwoebel barrier, one that favors adatom attachment to ascending steps, whereas bunching requires an inverse ES effect (whereby the barrier to adatom incorporation to a descending step is lower than that which characterizes ordinary diffusion). Bunching and meandering appear therefore to be a priori mutually exclusive. In this talk, a thermodynamically consistent theory is presented that resolves this apparent paradox, in the sense that it yields bunching under the assumption of an ES barrier. A main ingredient of the theory is the step chemical potential for which a generalized Gibbs--Thomson relation is derived via a direct energy-rate calculation, resulting in boundary conditions along step edges that couple adjacent terraces. Specialization to the case of a periodic train of initially equidistant rectilinear steps reveals a competition between the stabilizing ES effect and a destabilizing energetic correction that, for sufficiently high adatom equilibrium coverage, leads to step collisions. The underlying physics can be understood in terms of the tendency of the crystalline surface to minimize its total grand canonical potential. An extension of the model to account for the growth of binary compounds will also be discussed. [Joint work with P. Cermelli.]