Objective

You will learn how to develop and use kernel modules and character devices. You will also learn how the Linux OS maintains list of processes and their execution states.

Part A: Accessing your Virtual Machines

This assignment will be carried out on a Linux virtual machine. Each student has been assigned a VM with a unique IP address. The login credentials have been emailed to you. You can either use the VM assigned to you or you can use your own Linux VM on your personal computer.

Instructions to access your virtual machine remotely

Part B: Learn how to write and execute a Linux kernel module.

Kernel modules allow the code to be dynamically added to the kernel, dynamically. Try compiling and inserting the kernel modules hello.c and hellon.c from kernel module slides.

1. Save the Makefile, hello.c, and hellon.c in your home directory. The files hello.c has initialization and cleanup functions that are invoked upon loading and unloading the kernel module. The second file hellon.c is similar, except that it takes command-line arguments with insmod command to print a custom message.
2. Compile the kernel module: make
3. Load the kernel module: sudo insmod hello.ko

If the module was successfully inserted, you will see mymodule: Hello World! message in kernel log. Use commands dmesg or cat /var/log/kern.log to see the kernel log messages.

4. Unload the kernel module: sudo rmmod hello

Upon removing the kernel module, use the dmesg command to check if kernel log prints the message mymodule: Goodbye, cruel world!!

5. For hellon.c, module insertion command is sudo insmod hellon.ko whom=class howmany=10 (or any other argument values you want to give). Removal is the same as for hello.c.

Part C: Learn how to write a miscellaneous character device driver

Follow this example to learn how to create a miscellaneous character device in Linux kernel.

Modify the example to implement the .read file operation to return Hello World! message to a user space program that invokes the read() system call on your miscellaneous device.

Part D: How to browse Linux source code

There are a couple of ways you can browse linux source code.

Use the Linux cross reference website

Download the entire linux source code to your VM (or any other Linux computer) and use the cscope tool to browse. (slightly faster as you get familiar with cscope).

wget https://cdn.kernel.org/pub/linux/kernel/v4.x/linux-4.19.80.tar.xz (replace with whatever kernel version you are developing your module against)

sudo apt install cscope Cscope tutorial

Part E: Write a kernel module to return list of processes to user space via a character device

Implement a kernel module that creates a /dev/process_list character device. The character device should support the read() operation. When the read() system call is invoked on your character device from a user space process, your kernel module should return to the following information about all currently running processes:

process ID parent process ID the CPU on which the process is running its current state.

Note that the state field in task_struct can be -1, 0 or greater than 0. A state value of 1 or more indicates a combination (bitwise OR) values listed here.

So you will have to parse the task>state value to figure out which of the states apply to a process. You can pass the raw state value (as a long value) from kernel to user space and decode the state values in your user space program.

Beware that bugs in kernel code may either crash your kernel immediately or may have no immediate visible effect, but may have a delayed effect. Therefore, you cannot assume that the thing you did most recently is necessarily the cause of a crash.

Part F: Write a user space program to retrieve the list of processes from the kernel module

Implement a user-space C program that opens your character device and outputs the list of processes retrieved from your character device.

One such application could be written as follows (please fill in the missing code):

char *buffer;

/* allocate memory for character buffers HERE before you use them */

fd = open(/dev/process_list, O_RDONLY);

/* check for errors HERE */

while(!some termination condition)

{

bytes_read = read(fd, buffer, buffer_length);

/* check for errors HERE. Exit loop if all processes have been retrieved. */

/* print the output you have read so far. */

} close(fd);

Your programs output could be as follows:

PID=1 PPID=0 CPU=4 STATE=TASK_RUNNING

PID=2 PPID=0 CPU=2 STATE=TASK_INTERRUPTIBLE

PID=10 PPID=2 CPU=0 STATE=TASK_UNINTERRUPTIBLE

PID=16434 PPID=16424 CPU=10 STATE=TASK_DEAD

PID=14820 PPID=16424 CPU=8 STATE=TASK_WAKEKILL,TASK_UNINTERRUPTIBLE and so forth Grading Guidelines

References

Example kernel modules hello.c and hellon.c Kernel Cross Reference for_each_process() macro can help you iterate over all processes in the system.

struct task_struct definition can be found here

Process states are defined here

The function task_cpu() can help you extract the currently assigned CPU for a process.

Introductory material on Linux Kernel

Chapters 1 and 2 of the following online book provide a good introduction to the kernel, though with a bias towards device-driver development.

http://lwn.net/Kernel/LDD3/

For more kernel programming help, just google Linux Kernel and youll get lots more.