IMPLEMENTING MODE CHANGES AND FAULT RECOVERY FOR HARD REAL-TIME SYSTEMS IN ADA

Abstract

The Ada language was designed with the purpose of developing real-time embedded applications. Although the language shows some difficulties when it is used in such application domain, its tasking model is well-suited to scheduling theories based on statically assigning priorities to processes. In particular, the rate monotonic and deadline monotonic methods have been the subject of research to extend its range of applicability and to implement its concepts in Ada. In this paper we present an implementation of a mode change protocol and an approach to attain a certain degree of fault recovery in hard real-time systems in Ada. The mode change implementation uses standard Ada, with the only exception that dynamic priority setting is required. This approach assures data consistency and deadline meeting, during the execution of the mode change protocol. The fault recovery approach is based on recovery groups, which are composed of related processes with a common fault recovery behaviour. When any process in the group fails, the rest of the processes in the group end their execution and a recovery process, in charge of executing the recovery actions, starts. Those features have been successfully implemented and tested as part of the development of a reusable tasking executive intended to develop hard real-time systems in Ada.