Inter-process Communication and Process Synchronization

 

Inter-process Communication and Process Synchronization:

            Inter-process communication is a set of programming interfaces that allow a programmer to coordinate activities among different program processes that can run concurrently in an operating system. There are applications that involve each executing multiple processes concurrently. Processes work together to perform application-specific tasks. They are referred to as cooperating processes. Cooperating processes are “loosely” connected in the sense that they have independent private address spaces and run at different speeds. The relative speeds of the processes are not normally known. From time to time, they interact among themselves by exchanging information. An exchange of information among processes is called an interposes communication.

            It is stated in that one process cannot access elements from the private address space of another. Such access violations cause address exceptions, and the operating system traps them. This implies that processes need assistance from the operating system to set up communication facilities among themselves. Most operating systems implement a few different communication schemes to facilitate interposes communications. For each scheme, they support a few communication primitives (interface operations). Processes execute these primitives to exchange information among themselves.

            Ultimately, all information exchanges involve exchanging data items among the processes. The procedures of data items are called senders, and the consumers the receivers. Every data item, sent by a sender, is copied from the sender process address space to the kernel space from where a receiver process copies the data item into its own address space. Data items sent by senders remain in the kernel space until consumed by receivers.

           Figure: Inter-process communication via the kernel space

          The above-referred inter-process communication scheme is indirect as all inter-process communications are done via the kernel space. Every send or receive operation is performed by making a system call. Most operating systems also support a mechanism that allows processes to communicate among themselves directly without copying data via the kernel space. To do this a “shared memory region” is set up among the cooperating processes. The operating system maps parts of the process address spaces to the same physical memory locations so that all processes having access to the shared memory region immediately observe changes that may be made in that region. Note that shared memory regions reside in the user space, and not in the kernel space.

 

                        Figure: Inter-process communication through address space sharing

 

In indirect communication schemes, processes make system calls for each inter-process communication (send or receive of a data item). By contrast, in shared memory-based direct communication schemes, processes directly communicate by reading and modifying values in the common physical memory locations. As these memory locations are mapped to the user space, the processes do not need to make system calls to access contents of these locations. Nonetheless, they create, attach, and destroy shared memory regions through system calls.    Every inter-process communication scheme (directly or indirectly) can be modeled as schematics shown in. Ultimately, all communications involve storing and retrieving information in some shared medium. The senders store information in the medium, and the receivers retrieve the information from it. For easier manipulation, the information is structured into various data objects called shared data or shared variables. Shared variables are units of information referenced by processes. Each shared variable has a unique name / address and a type. The type defines a finite domain of values. Interface operations, and consistency semantics. The variable can store any value from the domain. Interface operations are the only means to access the shared variable. The semantics of each operation describe the permitted behavior of executions of the operation. Process communicates among themselves by manipulating values of shared variables by executing operations supported by these variables. Inter-process communication schemes vary and use shared variable of different types.

Figure: A typical inter-process communication facility

Each process executes its own operations on shared variables sequentially, as specified by its own program. Nevertheless, different processes may execute their operations on the same shared variable concurrently. That is, operation executions of different processes may overlap, and may affect one another. Note that each operation, when executed indivisibly on a shared variable, transforms the variable from one consistent value to another. However, when the operations are executed concurrently on a shared variable (without any access control), the consistency of its values may not be guaranteed. The behavior of operation executions on shared variables must be predictable for effective inter-process communications. Thus, operation executions on shared variables may need to be coordinated to ensure their consistency semantics. Coordination of accesses to shared variables (or shared space) is called synchronization. Most operating systems implement a few different synchronization schemes for process coordination purposes. Each scheme supports a set of primitives. The primitives are used when it is absolutely necessary to order executions of operations on shared variables in a particular manner.