This project is based on Linux Driver to manage system memory, deadlocks and crash.

• Technique to feed random inputs to system calls present in running kernel
• Achieved through automated programs.
• Continuous Process. Keep on thrashing the system calls.
• Based on Coverage guided fuzzing
• Starts with basic random input test case
• Mutates into new test cases
• Report valid test cases back to corpus
• Remove redundant ones.

Blockers Encountered –

Custom Kernel

• No official standard documentation for building custom kernel image from RedHat or Ununtu.
• Documentation works for one version and fails for another.
• Not all the kernel source code provided was compiled successfully with the config flags required for fuzzing.
• Builds with base kernel version and do not include patch/release in final kernel image.
• Don’t share which config file to pick for building custom kernel
• Issues faced while building a final package of custom kernel image.

DCS Drivers

• Unable to load DCS drivers on custom kernel image.
• No build framework available for building DCS drivers for a particular custom kernel version.
• Can’t use the existing kernel headers for the same kernel
• No intelligence in build scripts to separate out Fuzzing kmod and regular kmod packages

Implementation of Linux Fuzzing

Setting up the Environment

• Install a Linux distribution (e.g., Ubuntu, Debian) on a virtual machine or a dedicated testing environment.
• Install essential tools like gcc (for compiling), gdb (for debugging), and other dependencies.

Fuzzing Tool Used

• Skyzkaller: Specifically designed for fuzzing the Linux kernel.

Linux fuzzing is a critical part of securing Linux systems and applications, helping developers identify and fix vulnerabilities before they can be exploited. Proper implementation involves setting up a suitable environment, selecting the right tools, configuring them, and thoroughly analyzing the results to improve the security and stability of the software.