Mini Linux shell using C

About me

I am a Teacher and Developer. I have completed my Masters in Computer Engineering. I have been teaching since 2011 and coding since my student days. I have worked for multiple Software Development companies to help build their products.

I have also done many fun projects. I am sharing one of them here.

The problem I wanted to solve

I came across one such fun project when I was working as an online tutor. The project was about writing your own Linux Shell. I wanted to work on it as it was the first time I would be developing something of this kind.

What is a mini Linux shell?

A mini Linux shell is a program that behaves exactly lie the Linux shell, albeit with limited functionality.

• It supports built-in shell commands like "cd" and "exit".
• It supports input and output redirection.
• It supports background process.

Tech stack

Since it's a Linux shell, the only feasible language to use was C; as it is native to Linux. Also, the POSIX library was required for the implementation of the shell.

The process of building the shell

It was difficult to think of where to start. I was dealing with the fork() function for the first time. fork() creates a child process that can run simultaneously with the parent process in the system. The child process can be given a different task to perform. In this case, the child process would execute the command typed by the user in the shell. So I started reading about fork(). I, of course, started with the man page for fork(). But that was not enough, so I read some articles that explained the usage of fork() with examples.

Now armed with that knowledge, I went ahead to find how the child process would execute the command entered by the user. Do I have to code for all the possible commands? Well, the answer was no. There's a really useful family of functions in Linux, viz. exec(). These functions help one to execute a command in Linux. So I read the man page for that and found out that the execvp() function is the best match for my code.

With that sorted out, the next was just logically coding the entire shell program, which turned out to be quite simple. You may find the entire code on Github here.

Challenges I faced

The hardest parts were understanding the usage of fork() and installing the signal handler.

The signal handler is required as this shell supports background processes. The child process that is sent to the background must signal the parent process once it's done with its execution. This can be done using the waitpid() function with WNOHANG option. I had to read multiple articles on this topic to understand it properly.

Key learnings

The new things I learned

• fork()
• execvp()
• Signal handling

Tips and advice

The best way to tackle a huge problem is to break it down into smaller parts and then attack one part at a time. I started with creating child process and then moved on to executing the user command. After that was done, I implemented the support for background process part.

Final thoughts and next steps

Projects like these might seem really difficult and one might think that the implementation might go completely over their heads (especially for newbies). But, that's no reason to give up. If one keeps trying to solve a problem like that, eventually they will get success and learn many new things on the way.

This project taught me a lot, but now it offers nothing more. So I won't be lingering on this any longer. Many more new fun projects await!