CSCI 3500: Studio 5

Fork and Exec

Executing processes is fundamentally the reason why operating systems exist. Beyond that, it's an extremely common activity in modern systems- every time you look at a file, compile a program, or execute a terminal command you're probably creating at least one process, if not more. Understanding the fork() and exec() mechanisms provided by the operating system will give you insight to the process lifecycle in Linux.

In this studio, you will:

  1. Get process IDs with getpid()
  2. Create new processes with fork()
  3. Wait for child processes with wait()
  4. Execute programs with exec()

Please complete the required exercises below, as well as any optional enrichment exercises that you wish to complete.

As you work through these exercises, please record your answers in a text file. When finished, submit your work via the Git repository.

Make sure that the name of each person who worked on these exercises is listed in the first answer, and make sure you number each of your responses so it is easy to match your responses with each exercise.

Required Exercises

  1. As the answer to the first exercise, list the names of the people who worked together on this studio.

  2. Every process in Linux (and other OSes) has a unique identifying number called a Process Identifier (or PID). This number is used whenever you need to programmatically reference or interact with a specific process in the system.

    Create a new program called fork.c. Use printf() and getpid() to print out the process' PID. Run the program three times and copy-paste the output as the answer to this exercise.

  3. Now we will create a new process. Recall from lecture that this involves two different operations: fork()-ing and exec()-ing. The fork() step creates a new process (the "container" for executing a program), and the exec() step loads and executes a desired program within that new process.

    When fork() is called, the system creates a near-identical replica of the calling process. After the fork you will have two processes that both execute the same program. Insert the call "fork();" immediately after the start of main() in your program fork.c, and before the lines where you get the program PID and the print it out.

    What happens when you run your program now? Copy and paste the output.

  4. Look up the man page for fork(). This is a unique function in that you get two return values for each valid call to fork() (one in the parent, and another in the child). What are the return values listed?

  5. Use fork()'s return values to modify your program so that the parent and child processes identify themselves as such when they print out their PIDs. Copy and paste your program output.

  6. Sometimes it is useful for a parent process to wait for one or more child processes to finish executing. Otherwise, you might get undesirable behavior, or behavior that doesn't make sense. The sleep() function causes a program to suspend for a specified length of time before continuting, and we can use it to pretend that the child process takes a long time to execute. Modify your program so the child process executes the following call before it prints out its PID:


    This will cause your child process to go to sleep for about three seconds before printing its output. Now run your program... what happens? Does it behave the way you would expect?

  7. Because the child process takes a while to execute, we're in a situation where it might make sense for the parent process to explicitly wait for the child to finish. The waitpid() function will halt the parent until a desired child process returns. Modify your program so the parent process executes the following call before it prints out its PID:

    waitpid( child_PID, NULL, 0 );

    (The second and third parameters allow us some extra control over how this function operates, but we don't need them for now.)

    Execute your program again. How is the behavior different?

  8. Lastly, we can use the exec family of functions so that the child process can load and execute a new program on behalf of the parent. In particular we will use the execvp() function. (There are many forms of exec(), but this one is arguably the easiest to get started with, see the rest at man 3 exec.)

    Before we use execvp(), you should execute the command we want to run so that you know what to expect. We're going to have our child process execute the ls program in list-mode. Execute the command "ls -l" and copy-paste the output as the answer to this question.

  9. Now we will create a new, static, argv variable for the child. This is only useful when you know exactly what you're going to execute when you're writing your program, but we will explore more flexible methods in future studios. Create the following variables:

    char* child_argv[] = {"ls", "-l", NULL};
    char* child_prog = child_argv[0];

    Notice that child_argv has the exact same layout as if you were to try and access argv in a regular program. Now, in the child process, insert the call:

    execvp( child_prog, child_argv );

    If everything is correct, you should see your child process execute the "ls -l" command, and then your parent process should wait and print out its PID. Copy and paste your program output as the answer to this exercise.

Optional Enrichment Exercises

  1. Rearrange your call to the execvp() function so that you can call a different program with different arguments.

  2. Rearrange your execvp() call so that you can execute an arbitrary program, for example as a result of user input. This is a vital part of Lab 2, and we will explore it further in a future studio.