FPCL and FPZL Schedulability Analysis

This paper presents the Fixed Priority until Critical Laxity (FPCL) and Fixed Priority until Zero Laxity (FPZL) and scheduling algorithms for multiprocessor real-time systems. FPZL is similar to global fixed priority pre-emptive scheduling; however, whenever a task reaches a state of zero laxity it is given the highest priority. FPCL is a variant of FPZL that introduces no additional scheduling points beyond those present with fixed priority scheduling. FPZL and FPCL are minimally dynamic algorithms, in that the priority of a job can change at most once during its execution, bounding the number of pre-emptions. Polynomial time and pseudo-polynomial time sufficient schedulability tests are derived which are applicable to both FPCL and FPZL. These tests are then improved by computing upper bounds on the amount of execution that each task can perform in the zero-laxity / critical-laxity state. An empirical evaluation shows that FPCL and FPZL are highly effective, with a significantly larger number of tasksets deemed schedulable by the tests derived in this paper, than by state-of-the-art schedulability tests for EDZL scheduling. Extended version This paper builds on the paper “FPZL Schedulability Analysis ” [34] published in the proceedings of RTAS 2011. This paper extends the analysis given in [34] to cover the FPCL scheduling algorithm as well as FPZL. The analysis for FPCL is a superset of that provided for FPZL in [34] and reduces to it if the critical laxity parameter of each task is set to zero. In this paper, the experimental evaluation given in [34] has been extended to include schedulability tests and simulation of FPCL as well as FPZL. We also report on a prototype implementation of FPCL and FPZL in the Linux kernel, running on an Intel Core 2 Quad processor (Q9650)