The various modes of spallative LiBeB production are summarized, and classified according to their dependence or independence on the abundance of medium heavy elements (CNO) illustrated by that of oxygen in the interstellar medium. The predictions of the models are confronted to the available observational correlations (Be, B vs O). Clearly, a primary mechanism should lead to a slope one in the lg(Be/H) vs [O/H] plot and a secondary mechanism to a slope two. Due to the ambiguity of the O data, another criterion, based on energetics, can help us to select an adequate model. A purely secondary origin in the very early Galaxy is much more energy demanding than a primary one. Indeed, magnesium seems to be a possible surrogate of oxygen and iron since i) it is spectroscopically more easy to cope with and ii) its nucleosynthetic yield is independent of the mass cut and does not depend on metallicity.