When the prologue asks the guard for epilogue propaga-tion (relay), a test on an active guarded section is performed. The epilogue is made pending, and the interrupted process is resumed, if the request for propagation overlaps with an active guarded section. If overlapping does not take place, a number of follow-up steps take place to make sure that (a) propagated epilogues run interrupts enabled, (b) stack ex-pansion is constrained by a variable but maximum number of nested interrupts, and (c) no pending epilogue has been left after propagation finished. The order of the individual steps is of importance yet: (1) the guard is to be locked,
(2) the actual interrupt priority level is to be remembered and then are interrupts to be enabled, (3) all pending epi-logues are to be processed, (4) the interrupt priority level is to be restored, and (5) the guard is to be unlocked. Finally, a check on further epilogues that may have been made pend-ing in the meantime is to be performed and, if necessary, the entire sequence is to be repeated.