Joakim Tjernlund writes: > Is it too much trouble to define r7 to always have > a zero > value at process startup? That's not much trouble, but what will you do for existing kernels? Won't you have to do something like testing *r1 anyway? Paul.