| This section expands on the process chain of Program 3.1. The chain is a vehicle for experimenting with wait and with sharing of devices. All of the processes in the chain created by Program 3.1 share standard input, standard output and standard error. The fprintf to standard error is a critical section of the program. This exercise explores some implications of critical sections. Later chapters extend this exercise to critical sections involving other devices (Chapter 6) and a token-ring simulation (Chapter 7). Program 3.1 creates a chain of processes. It takes a single command-line argument that specifies the number of processes to create. Before exiting, each process outputs its i value, its process ID, its parent process ID and the process ID of its child. The parent does not execute wait . If the parent exits before the child, the child becomes an orphan. In this case, the child process is adopted by a special system process (which traditionally is a process, init , with process ID of 1). As a result, some of the processes may indicate a parent process ID of 1. Do not attempt this exercise on a machine with other users because it strains the resources of the machine. -
Run Program 3.1 and observe the results for different numbers of processes. -
Fill in the actual process IDs of the processes in the diagram of Figure 3.2 for a run with command-line argument value of 4. -
Experiment with different values for the command-line argument to find out the largest number of processes that the program can generate. Observe the fraction that are adopted by init . -
Place sleep(10); directly before the final fprintf statement in Program 3.1. What is the maximum number of processes generated in this case? -
Put a loop around the final fprintf in Program 3.1. Have the loop execute k times. Put sleep(m); inside this loop after the fprintf . Pass k and m on the command line. Run the program for several values of n , k and m . Observe the results. -
Modify Program 3.1 by putting a wait function call before the final fprintf statement. How does this affect the output of the program? -
Modify Program 3.1 by replacing the final fprintf statement with four fprintf statements, one each for the four integers displayed. Only the last one should output a newline. What happens when you run this program? Can you tell which process generated each part of the output? Run the program several times and see if there is a difference in the output. -
Modify Program 3.1 by replacing the final fprintf statement with a loop that reads nchars characters from standard input, one character at a time, and puts them in an array called mybuf . The values of n and nchars should be passed as command-line arguments. After the loop, put a '\0' character in entry nchars of the array so that it contains a string. Output to standard error in a single fprintf the process ID followed by a colon followed by the string in mybuf . Run the program for several values of n and nchars . Observe the results. Press the Return key often and continue typing at the keyboard until all of the processes have exited. |
