Scheduling in theCo - Synthesis of Reactive Real - Time Systems

{ Existing software scheduling techniques limit the functions that can be implemented in software to those with a restricted class of timing constraints, in particular those with a coarse-grained, uniform, periodic behavior. In practice, however , many systems change their I/O behavior in response to the inputs from the environment. This paper considers one such class of systems, called reactive real-time systems, where timing requirements can include sequencing, rate, and response time constraints. We present a static, non-preemptive, ne-grained software scheduling algorithm to meet these constraints. This algorithm is suitable for control-dominated embedded systems with hard real-time constraints, and is part of the core of a hardware/software co-synthesis system. An embedded system is a special purpose computer consisting of one or more controllers and peripheral devices. A reactive system is an embedded system that changes its I/O behavior in response to inputs from the environment. This is in contrast to those systems with a uniform periodic behavior that is independent of their input. Many reactive systems must also meet hard timing constraints of various types imposed by the devices' protocols or by the required system behavior. These systems are referred to as reactive real-time systems. Esterel 1] and StateCharts 4] have been used for specifying reactive systems. Reactive behavior can be succinctly and conveniently captured with parallelism and watchdogs. A watchdog is a wait-on-signal statement that encloses a statement block, and breaks control ow out of the block upon receiving the signal. Both Es-terel and StateCharts assume an idealized timing model, where simple computations are assumed to take zero time to perform. However, lengthy computations that violate this assumption are extracted and treated as external signals , and timing constraints cannot be speciied on them for scheduling. While this assumption simpliies semantics , it also restricts the class of applications that can be speciied. The watchdog-concurrency reactive programming model has been augmented with timing constraints in 2]. The system behavior is divided into a number of modes. A mode speciies a scope within which a set of timing constraints must be met, until one of the watchdogs detects an event and disables, or causes a transition out of, the mode. When such a transition is initiated, each concurrent branch to be disabled is scheduled to run until a safe exit point is reached. This enables interleaving while maintaining the integrity of I/O protocols and program state. A timing constraint speciies …