Adding Processes To The Workbench

This section describes methods for representing processes and inter-process communication mechanisms in ISPS and the Workbench, and two methods that could be used for implementing those processes and their communication in hardware. A transformation for creating these processes is described in the next section, along with its effect on the VT and on the final design. 

In the field of computer science, processes, concurrency, and interprocess communication are well understood [Dijkstra68, Hoare78, Andrews83]. The execution of a set of sequential statements is called a process, and the parallel execution of two or more processes is referred to as concurrent execution. The transfer of data between two processes is called inter-process communication, and insuring that inter-process communication takes place at a mutually-agreed-upon time is called synchronization. 

Several mechanisms are used to specify that two or more processes are to execute concurrently. One of the earliest, the coroutine, allows two or more processes to transfer control between themselves by means of the RESUME statement. Another early mechanism, the FORK statement, explicitly invoked a concurrent process, and then re synchronized on its completion by means of the JOIN statement. Finally, many modern languages simply label blocks of code with an explicit process declaration, with the understanding that all processes are to be executed concurrently. 

In message passing, SEND and RECEIVE operations are used to synchronize, and to pass data via a specified communication channel. At one end of the message-passing spectrum, the SEND can refer to the specific process to which the information is sent, and the RECEIVE can refer to the specific process from which the information is received this option is referred to as direct naming. At the other end of the spectrum, the SEND and RECEIVE can refer to global mailboxes. Since the mailbox is not directly tied to a particular process, this method is more general, and provides better support for client / server interaction, where there are multiple clients or servers. 

Synchronization of ISPS processes occurs though process activation, and through the WAIT operation. A WAIT operation continuously evaluates an expression, and execution proceeds with the next statement when the value of the expression is not equal to 0. The last non-zero value of the expression is also returned by the WAJT. 

---