3. Process Management in Unix/Linux

This chapter covers process management in Unix/Linux. It explains the principle of multitasking and introduces the process concept. It uses a programming example to illustrate the principles and techniques of multitasking, context switching and processes. The multitasking system supports dynamic process creation, process termination, process synchronization by sleep and wakeup, process relations and implementation of process family tree as a binary tree, allowing parent process to wait for child process termination. It provides a concrete example of how process management functions work in an operating system kernel. Then it explains the origin of processes in Unix/Linux, from the initial process during booting to INIT process, daemon processes, login processes and sh process for user command execution. Next, it explains the execution modes of processes, transitions from User mode to Kernel mode by interrupts, exceptions and system calls. Then it describes Unix/Linux system calls for process management, which include fork, wait, exec and exit. It explains the relationship between parent and child processes, including a detailed description between process termination and the wait operation by parent process. It explains how to handle orphan processes by the INIT process, including subreaper processes in current Linux, and it demonstrates subreaper process by example. Then it explains changing process execution image by exec in detail, which includes the execve system call, command-line parameters and environment variables. It explains the principles and techniques of I/O redirections, pipes and shows pipe programming by examples. The programming project of this chapter is for the reader to integrate the concepts and techniques of process management to implement a sh simulator for command execution. The sh simulator works exactly the same as the standard sh. It supports executions of simple commands, commands with I/O redirections and multiple commands connected by pipes.