Operating systems
Second laboratory exercise

During this exercise, you will need to use complex system calls whose behaviour needs to be analyzed beforehand to ensure that your shell functions properly. The main source of information regarding system calls and library functions are the so called man pages, accessed using the man command. The man command takes a program or function name as an argument and displays the corresponding documentation page. It is important to note that the man pages are divided into discrete sections, grouped by the nature of the content they describe. For instance, the third section documents library functions, the second section documents system calls, and the first section documents builtin shell commands and programs. If invoked with only a name argument, the man function will sequentially scan the sections until it finds the first matching documentation page. Unfortunately, this can lead to confusing situations as some programs and functions share names, leading to man displaying the wrong documentation page, as depicted in the following listing.

$ man sleep
SLEEP(1)   User Commands   SLEEP(1)

NAME
       sleep - delay for a specified amount of time
SYNOPSIS
       sleep NUMBER[SUFFIX]...
       sleep OPTION
...

$ man 3 sleep
sleep(3)   Library Functions Manual   sleep(3)

NAME
       sleep - sleep for a specified number of seconds

LIBRARY
       Standard C library (libc, -lc)

SYNOPSIS
       #include <unistd.h>
       unsigned int sleep(unsigned int seconds);
...

Fortunately, the man command takes an optional section number argument, searching the documentation page only in the specified section. Detailed information for any system call can be obtained using the man 2 <syscall_name> command. Detailed information for any standard C library function can be obtained using the man 3 <function_name> command.

Pseudocode 1 roughly covers the basic shell execution flow. Before accepting user input, your shell must print "fsh> " to signal that it is ready to process the user's input. Each command consists of a command name and, optionally, multiple arguments that are separated using one or more whitespace characters. You don't need to implement deleting characters when typing commands.

while(1){
 print "fsh> ";

 wait for user input;
 read user input and parse;

 command = find_command();
 if(command not builtin){
   create a new process;
   execute the command in the child process;
   wait until the child process has exited;
 }
}

fsh> /usr/bin/pwd
/home/student
fsh> /usr/bin/echo "hello"
hello
fsh> /usr/bin/ls -lh
# directory contents
fsh> /bin/asdf
fsh: Unknown command: /bin/asdf

The basic command execution feature required the user to provide the full path for a command. Since typing full program paths is a cumbersome process, all modern shells provide a more refined mechanism that executes programs using only their names. After the user inputs the command name, the shell then tries to find the corresponding program in a set of predefined directories listed in the PATH environment variable ². The value of the PATH environment variable consists of a series of directory paths separated using the ":" character. Listing 4 shows an example of the PATH environment variable contents.

> echo $PATH
/usr/local/bin:/usr/bin:/bin:/usr/local/sbin

Your shell must fetch the contents of the PATH environment variable using the getenv function, process it, and use it to search the corresponding directories before executing a command.

process_path() {
    ...
    char *contents = getenv("PATH");
    ...
}