CSCI 380 -- Lab 1b

Writing Your Own Unix Shell -- Job Control

The purpose of this assignment is to become more familiar with the concepts of process control and signaling. You’ll do this by expanding your previously created simple Unix shell program

I have a new handout which you should then integrate your last shell into: lab1b-handout.tar

Tasks to Perform

You are tasked with implementing a unix shell! This sounds intimidating but it's really not so bad.

Your shell must be able to:

• do everything the last shell needed to do
• able to run additional built-in commands:
  • bg <%jid|pid> -- change a stopped background job to a running background job
  • kill <%jid|pid> -- send SIGINT to the specified background job
  • jobs -- list the running and stopped background jobs
  • fg should also now require a job id or process id as an argument

REMINDER: ONLY ONE JOB CAN RUN IN THE FOREGROUND AT A TIME

I'm providing a bit of additional code to help with job control by means of a jobctrl library. The header file (public API) is included in this new tsh.c skeleton code. DO NOT CHANGE jobctrl.c IN ANY WAY

Example

tsh> /bin/sleep 100 &
[1] (823) /bin/sleep 100 &
tsh> /bin/sleep 90 &
[2] (824) /bin/sleep 90 &
tsh> /bin/sleep 80 &
[3] (825) /bin/sleep 80 &
tsh> /bin/sleep 70 &
[4] (826) /bin/sleep 70 &
tsh> jobs
[1] (823) Running /bin/sleep 100 &
[2] (824) Running /bin/sleep 90 &
[3] (825) Running /bin/sleep 80 &
[4] (826) Running /bin/sleep 70 &
tsh> fg %1
tsh> 
Job [1] (823) stopped by signal 20 <<<< Ctrl+Z pressed here
tsh> jobs
[1] (823) Stopped /bin/sleep 100 &
[2] (824) Running /bin/sleep 90 &
[3] (825) Running /bin/sleep 80 &
[4] (826) Running /bin/sleep 70 &
tsh> bg 999
999: No such pid
tsh> bg %5
%5: No such job
tsh> bg %1
tsh> jobs
[1] (823) Running /bin/sleep 100 &
[2] (824) Running /bin/sleep 90 &
[3] (825) Running /bin/sleep 80 &
[4] (826) Running /bin/sleep 70 &
tsh> kill %3
Job [3] (825) terminated by signal 2
tsh> jobs
[1] (823) Running /bin/sleep 100 &
[2] (824) Running /bin/sleep 90 &
[4] (826) Running /bin/sleep 70 &
tsh> fg 823
tsh> 
Job [1] (823) terminated by signal 2 <<<< Ctrl+C pressed here
tsh> kill 826
Job [4] (826) terminated by signal 2
tsh> jobs
[2] (824) Running    /bin/sleep 90 &

Shell Specification

• Prompt should be the string tsh>
• Typing ctrl-c should cause a SIGINT signal to be sent to the current foreground job as well as any descendents of the job. If there is no foreground job, then the signal should have no effect
• Typing ctrl-z should cause a SIGTSTP signal to be sent to the current foreground job as well as any descendents of the job. If there is no foreground job, then the signal should have no effect
• If the command line ends with an ampersand &, then the new job should be run in the background. Otherwise, it should run the job in the foreground
• Support the following built-in commands:
  • quit -- terminates the shell
  • jobs -- lists all jobs. HINT: jobctrl has a function that might do exactly that
  • bg <job> -- restarts <job> by sending SIGCONT and runs it in the background. The <job> argument can be either a PID (pid) or JID (%jid)
  • fg <job> -- restarts <job> by sending SIGCONT and runs it in the forground. The <job> argument can be either a PID (pid) or JID (%jid)
  • kill <job> -- interrupt <job> by sending SIGINT to the process group. The <job> argument can be either a PID (pid) or JID (%jid)
• Reap all of its zombie children. If any job terminates because it recevies a signal that it didn't catch, then tsh should recognize this event and print a message with the job's PID and a description of the offending signal

Structure

You are welcome to organize this in any way, but you may find it helpful to structure your code into the following methods:

• eval -- main routine that parses and interprets the command line
• builtin_cmd -- recognize and interprets the built-in commands
• waitfg -- waits for a foreground job to complete
• do_bgfgkill -- process the fg, bg, and kill built-ins
• sigchld_handler -- catches SIGCHILD signals
• sigint_handler -- catches SIGINT (ctrl-c) signals
• sigstp_handler -- catches SIGTSTP (ctrl-z) signals

Checking Your Work

I have made a reference solution available which will run on the linux lab machines tshref. You are welcome to compare functionality and output between your solution and mine

I also have a series of autograded tests when you submit on autolab

Hints

• All hints from the prior lab still apply
• Read every single word of Chapter 8 in the textbook
• Each job can be identified by either a process ID (pid) or a job ID (jid). JIDs should be denoted on the command line by the prefix %. For example, %5 denotes JID 5 while 5 denotes PID 5.
• When you run a new job, add it to the jobctrl list. When you reap a child, remove it from the list. Please ensure that you block signals when you need to -- avoid race conditions with updating the job list
• waitfg can be drastically simplified -- "while my foreground process ID is the same, suspend"

TAKE ADVANTAGE OF THE CODE I ADDITIONALLY PROVIDE

Most importantly, start early.

Extra Credit (+5pts):

Expand the kill built-in command to accept an optional argument which specifies the signal to send (the default should be SIGINT -- or signal 2)

tsh> /bin/sleep 10 &
[1] (823) /bin/sleep 10 &
tsh> kill -9 %1
Job [1] (823) terminated by signal 9

You can view a list of signal numbers by looking at the man page for signal: man 3 signal

Submission

Upload/submit tsh.c to autolab.

Grading Criteria

This lab will be graded out of 100 points.

• 85pt -- autograded
• 15pt -- style, design, naming, formatting