Paul Regnier

RUN: Optimal Multiprocessor Real-Time Scheduling via Reduction to Uniprocessor

Optimal multiprocessor real-time schedulers incur significant overhead for preemptions and migrations. We present RUN, an efficient scheduler that reduces the multiprocessor problem to a series of uniprocessor problems. RUN significantly outperforms existing optimal algorithms with an upper bound of O(\log m) average preemptions per job on m processors (less than 3 per job in all of our simulated task sets) and reduces to Partitioned EDF whenever a proper partitioning is found.

Evaluation of interrupt handling timeliness in real-time Linux operating systems

Several real-time Linux extensions are available nowadays. Two of those extensions that have received special attention recently are Preempt-RT and Xenomai. This paper evaluates to what extent they provide deterministic guarantees when reacting to external events, an essential characteristic when it comes to real-time systems. For this, we define two simple experimental approaches. Our results indicate that Preempt-RT is more prone to temporal variations than Xenomai when the system is subject to overload scenarios.

A TLA+ Formal Specification and Verification of a New Real-Time Communication Protocol

We describe the formal specification and verification of a new fault-tolerant real-time communication protocol, called DoRiS, which is designed for supporting distributed real-time systems that use a shared high-bandwidth medium. Since such a kind of protocol is reasonably complex and requires high levels of confidence on both timing and safety properties, formal methods are useful. Indeed, the design of DoRiS was strongly based on formal methods, where the TLA+ language and its associated model-checker TLC were the supporting design tool. The protocol conception was improved by using information provided by its formal specification and verification. In the end, a precise and highly reliable protocol description is provided.

Quasi-partitioned scheduling: optimality and adaptation in multiprocessor real-time systems

We describe a new algorithm, called quasi-partitioned scheduling (QPS), capable of scheduling any feasible system composed of independent implicit-deadline sporadic tasks on identical processors. QPS partitions the system tasks into subsets, each of which is either scheduled by EDF on a single processor or by a set of servers on two or more processors. More precisely, QPS uses an efficient scheme to switch between partitioned EDF and global-like scheduling rules in response to system load variation, providing dynamic adaptation in the system. Extensive simulation compares QPS favorably against related work, showing that it has very low preemption and migration overheads.

OUTSTANDING PAPER: Optimal and Adaptive Multiprocessor Real-Time Scheduling: The Quasi-Partitioning Approach

We describe a new algorithm, called Quasi-Partitioned Scheduling (QPS), capable of scheduling any feasible system composed of independent implicit-deadline sporadic tasks on identical processors. QPS partitions the system tasks into subsets, each of which is either scheduled by EDF on a single processor or by a set of servers on two or more processors. More precisely, QPS uses an efficient scheme to switch between partitioned EDF and global-like scheduling rules in response to system load variation, providing dynamic adaptation in the system. Extensive simulation compares QPS favorably against related work, showing that it has very low preemption and migration overheads.

Revealing the Secrets of RUN and QPS: New Trends for Optimal Real-Time Multiprocessor Scheduling

Until recently there has been a common belief that optimal multiprocessor real-time scheduling algorithms necessarily incur a high number of task preemptions and migrations. New scheduling algorithms have shown that this is not the case. In this paper we explain why two of these algorithms, RUN and QPS, achieve optimality with only a few preemptions and migrations. We also compare these two algorithms, exhibiting their similarities and differences. By putting RUN and QPS side-by-side, we bring about their fundamental properties and help in the understanding of the multiprocessor real-time scheduling problem.

From RUN to QPS: new trends for optimal real-time multiprocessor scheduling

Until recently, there has been a common belief that optimal multiprocessor real-time scheduling algorithms necessarily incur a high number of task preemptions and migrations. New scheduling algorithms have shown that this is not the case. In this paper, we explain why two of these algorithms, reduction to uniprocessor (RUN) and quasi-partition scheduling (QPS), achieve optimality with only a few preemptions and migrations, exhibiting their similarities and differences. We also compare these two algorithms via simulations, highlighting the consequences of their differences in their performance. By putting RUN and QPS side-by-side, we bring about their fundamental properties and help in the understanding of the multiprocessor real-time scheduling problem.