Linux FAT Filesystem Journaling Implementation using LKL

---

Overview

This repository documents a project for the Operating Systems course. The primary objective was to enhance the reliability of the legacy FAT filesystem by implementing a rudimentary journaling/logging mechanism.

The modifications were carried out within the Linux kernel filesystem layer, utilizing the Linux Kernel Library (LKL). LKL allowed the modified kernel filesystem code to run in user-space, enabling easier development, testing, and debugging of the changes.

---

Project Goal

The core goal was to safeguard the integrity of the FAT filesystem against unexpected crashes (e.g., power loss) during file operations. This was achieved by:

1. Implementing a Transaction Log: Creating a designated journal file (e.g., journal.txt) to log critical metadata changes before they are committed to the disk.
2. Ensuring Atomicity: Guaranteeing that key metadata operations (like directory entry creation, deletion, or renaming) are performed atomically or can be safely recovered upon reboot.
3. FAT Driver Modification: Integrating the logging calls directly into the relevant functions of the kernel's FAT driver.

---

Technical Implementation

The work involved modifying the FAT filesystem code within the context of the Linux Kernel's Virtual Filesystem (VFS) layer. Key aspects of the implementation included:

• Linux Kernel Library (LKL): Used as the primary environment to compile and run the modified FAT driver in user-space.
• FAT Structure Analysis: Deep dive into the structure of FAT, including the File Allocation Table (FAT), Superblocks, and Inodes, to identify the exact points where metadata updates occur.
• System Call Interface: Exploration of how user applications interact with the kernel through the System Call Interface, specifically focusing on functions like sys_open and sys_write to understand file manipulation flow.
• Journaling Logic: Implementing logic to write transaction records to the journal file for critical operations related to directory entries and file allocation chains.

---

Known Challenges & Development Notes

A significant challenge encountered during development was the handling of the journal file from within the kernel context:

• Journal Write/Read Issues: Attempts to write certain metadata structures (like directory entry slot information) to the journal.txt file from the modified FAT functions were successful, but reading this data back via the standard VFS/inode.c mechanisms proved unreliable.
• Data Integrity Risk: Directly calling read from the logging function led to undesirable side effects, including data corruption/loss in the inode.c context.