ABSTRACT:
Based on lessons learned in developing and maintaining LITMUS^RT,
arguments in favor of an intensified collaboration between the
academic and the open-source real-time communities are presented, and
several ways in which ongoing efforts in these two communities may
benefit each other are outlined.
Some (unfortunately) commonly-encountered sources of friction and mutual misconceptions, which result from differing backgrounds and objectives in these two communities, are exposed, with the goal of finding common ground.
Further, a wish list is presented of possible additions and changes to PREEMPT-RT that would enhance Linuxs viability as the platform of choice for real-time-systems research. These improvements are substantiated by examining the implementation requirements of EDF-HSB, a candidate algorithm for earliest-deadline-first (EDF) scheduling support in Linux that integrates hard and soft real-time guarantees as well as best-effort scheduling with bandwidth reservations.