Your program should print out a prompt like myshell> when it is ready to accept input. It must read a line of input, accepting several possible commands. The start command will start another program with command line arguments, print out the process number of the running program, and then accept another line of input. For example:
myshell> start cp data.txt copy.txt myshell: process 346 started myshell>
The wait command causes the shell to wait for any child process to exit. When this happens, indicate whether the exit was normal or abnormal, along with the exit code or signal number and name, respectively. Display any errors encountered. For example:
myshell> wait myshell: process 502 exited normally with status 5 myshell> wait myshell: process 347 exited abnormally with signal 11: Segmentation fault. myshell> wait myshell: No children.
The waitfor command does the same thing, but waits for a specific child process to exit:
myshell> waitfor 346 myshell: process 346 exited normally with status 0 myshell> waitfor 346 myshell: No such process.The run command should combine the behavior of start and waitfor. run should start a program, possibly with command line arguments, wait for that particular process to finish, and print the exit status. For example:
myshell> run date Mon Jan 19 11:51:57 EST 2009 myshell: process 348 exited normally with status 0The watchdog takes a timeout (in seconds) and a command to run, and then executes the command in the same way as run. However, if the command takes too long, then the shell should send the child process a SIGKILL signal, wait for it to exit, and then display the final status. For example:
myshell> watchdog 3 sleep 10 . . . myshell: process 61 exceeded the time limit, killing it... myshell: process 61 exited abnormally with signal 9: Killed.The chdir command should cause the shell to change its working directory to the named directory:
myshell> chdir /tmpThe pwd command should cause the shell to print the current working directory:
myshell> pwd /escnfs/home/dthainAfter each command completes, your program must continue to print a prompt and accept another line of input. The shell should exit with status zero if the command is quit or exit or the input reaches end-of-file. If the user types a blank line, simply print another prompt and accept a new line of input. If the user types any other command, the shell should print a reasonable error message:
myshell> bargle ls -la myshell: unknown command: bargle
To keep things simple, your shell doesn't need to deal with arbitrarily long commands. It must accept input lines of up to 1000 characters, and must handle up to 100 distinct words on a line.
fork, execvp, wait, waitpid, kill, exit, signal, printf, fgets, strtok, strcmp, strsignal, atoi, chdir, getwdUse fgets to read one line of text after printing the prompt. Note that if you printf a prompt that has no newline on the end, it will not immediately display. Call fflush(stdout) to force the output.
Breaking the input line into separate word is a little tricky, but is only a few lines of code once you get it right. Call strtok(line," \t\n") once to obtain the first word, and then strtok(0," \t\n") repeatedly to get the rest, until it returns null. Declare an array of pointers char *words, then, for each word found by strtok, save a pointer to the word in words[i]. Keep track of the number of words as nwords, then set words[nwords] = 0; when you have found the last one.
Once you have broken the input line into words, you can check words for the command name, use strcmp to check for string equality and atoi to convert a string to an integer.
watchdog is a little tricky, because you want the shell to stop and wait for either the time to expire, or the child process to exit. The trick here is to catch the SIGCHLD signal, which is delivered whenever a child process exits. Use signal to set up a function as a signal handler that gets called with SIGCHLD is delivered. Start the child process and go to sleep. If the signal arrives while you are sleeping, the sleep system call will be interrupted and return an error indicating that.
Make sure to stop if fgets returns null, indicating end-of-file. This allows you to run myshell and read commands from a file. For example, if you create myscript with the following contents:
start ls wait start date waitThen, you can run the shell on that input like this:
./myshell < myscript
Make sure to carefully handle all possible error conditions. Every system call can fail in a number of ways. You must cleanly handle all possible errors with a reasonable error message, as discussed in Project I. It is up to you to read the man pages carefully and learn what errors are possible.
Turn in the file myshell.c and a Makefile that builds the executable from source. Your dropbox is mounted on the student machines at this location:
/escnfs/courses/sp20-cse-30341.01/dropbox/YOURNETIDTo submit your files, make a directory called project2 in your dropbox, and copy your files there:
mkdir /escnfs/courses/sp20-cse-30341.01/dropbox/YOURNETID/project2 cp file1 file2 file3 ... /escnfs/courses/sp20-cse-30341.01/dropbox/YOURNETID/project2And double check that the right items are present:
ls -la /escnfs/courses/sp20-cse-30341.01/dropbox/YOURNETID/project2