Resource Usage Tracker & Limiter for Linux Kernel 6.1.6 and 6.13.4

Developed as part of a COL331 course project.

---

Table of Contents

• Overview
• Features
• Supported Kernel Versions
• Directory Structure
• Implementation Details
  • System Calls
  • Data Structures
  • Resource Usage Tracking & Limiting
  • Helper Modules
• Applying the Patch
• Building & Installing the Kernel
• Usage Instructions
• Extra Features
• References

---

Overview

This repository provides a Linux kernel patch (res_usage.patch) and supporting modules to enable tracking and limiting of heap memory usage and open file descriptors for specified processes. The solution is compatible with Linux kernel versions 6.1.6 and 6.13.4. It introduces new system calls, helper modules, and /proc integration for real-time resource monitoring and enforcement.

---

Features

• Custom System Calls for process registration, resource fetching, deregistration, quota setting, and quota resetting.
• Automatic Quota Enforcement: Processes exceeding set quotas are killed with SIGKILL.
• Procfs Interface: /proc/tracker_status displays all monitored processes and their current resource usage.
• Automatic Cleanup: Monitored entries are removed when processes exit, using a kprobe on do_exit.
• Cross-architecture Support: Works on both x86_64 and ARM64.
• Robust Edge Case Handling: Handles process death during registration, PID reuse, and concurrent access.

---

Supported Kernel Versions

• Linux 6.1.6
• Linux 6.13.4

The patch (res_usage.patch) is provided for both versions and has been tested on x86_64 and ARM64.

---

Directory Structure for modified files

---

Implementation Details

System Calls

System Call	Purpose
sys_register	Register a process for monitoring
sys_fetch	Fetch resource usage for a monitored process
sys_deregister	Remove a process from monitoring
sys_resource_cap	Set heap/file quotas for a monitored process
sys_resource_reset	Reset quotas to unlimited for a monitored process

Syscall numbers are assigned in the appropriate syscall tables for each architecture and kernel version.

---

Data Structures

Defined in include/linux/resource_tracker.h:

struct per_proc_resource {
pid_t pid;
unsigned long heapsize;
unsigned long openfile_count;
};

struct pid_node {
struct per_proc_resource *proc_resource;
struct list_head next_prev_list;
};

Each monitored process is represented as a node in a doubly-linked list, protected by a spinlock.

---

Resource Usage Tracking & Limiting

• Registration: Validates PID, ensures process is alive, allocates and initializes a tracking node, and adds it to the monitored list.
• Fetching: Copies current resource usage from kernel space to user space.
• Deregistration: Removes the process from the monitored list and frees memory.
• Quota Setting: Adds heap_quota and file_quota fields to task_struct. Exceeding quotas results in SIGKILL.
• Quota Reset: Resets quotas to unlimited (-1).
• Syscall Hooking: Modifies mmap, brk, open, openat, openat2, and close syscalls to update tracked usage.

---

Helper Modules

• cleanup_kprobe: Uses a kprobe on do_exit to ensure monitored entries are cleaned up when a process exits.
• tracker: Creates /proc/tracker_status for real-time monitoring of all tracked processes.

---

Applying the Patch

1. Download the Kernel Source

2. Apply the Patch:

git apply < /path/to/res_usage.patch

---

Building & Installing the Kernel

1. Configure the Kernel:

cp -v /boot/config-$(uname -r) .config
make menuconfig # Or 'make oldconfig'

2. Build the Kernel:

make -j$(nproc)
sudo make modules_install install

3. Update Bootloader and Reboot:

sudo update-grub
sudo reboot

4. Verify Kernel Version:

uname -r

---

Usage Instructions

• Register a Process:

syscall(SYS_register, pid);

• Fetch Resource Usage:

struct per_proc_resource stats;
syscall(SYS_fetch, &stats, pid);

• Deregister a Process:

syscall(SYS_deregister, pid);

• Set Quotas:

syscall(SYS_resource_cap, pid, heap_quota_MB, file_quota);

• Reset Quotas:

syscall(SYS_resource_reset, pid);

• View Monitored Processes:

cat /proc/tracker_status

---

Extra Features

• Works on both x86_64 and ARM64.
• Automatic cleanup of monitored entries on process exit.
• Procfs visualization for real-time monitoring.
• Detailed documentation and robust error handling.
• Efficient memory management and concurrency protection.

---

References

• Assignment instructions and requirements.
• Implementation report (report.pdf).
• Linux kernel development documentation.

---