L’interface est localisée ; les présentations et documents des Skills sont traduits automatiquement par IA dans votre langue. Le code, les commandes, chemins et termes techniques restent en version originale, et le document de l’auteur fait foi.
Explorer les Skills
Données et recherche · SKILL.md

analyzing-linux-kernel-rootkits

analyzing-linux-kernel-rootkits

Detect kernel-level rootkits in Linux memory dumps using Volatility3 linux plugins (check_syscall, lsmod, hidden_modules), rkhunter system scanning, and /proc vs /sys discrepancy analysis to identify hooked syscalls, hidden kernel modules, and tampered system structures.

Independent GitHub snapshot · Automated classification · Original documentation governs

Documentation de la Skill

GitHub

name: analyzing-linux-kernel-rootkits description: Detect kernel-level rootkits in Linux memory dumps using Volatility3 linux plugins (check_syscall, lsmod, hidden_modules), rkhunter system scanning, and /proc vs /sys discrepancy analysis to identify hooked syscalls, hidden kernel modules, and tampered system structures. domain: cybersecurity subdomain: digital-forensics tags: [rootkit, linux, kernel, volatility3, memory-forensics, malware-analysis, rkhunter, forensics] version: "1.0" author: mahipal license: Apache-2.0

Analyzing Linux Kernel Rootkits

Overview

Linux kernel rootkits operate at ring 0, modifying kernel data structures to hide processes, files, network connections, and kernel modules from userspace tools. Detection requires either memory forensics (analyzing physical memory dumps with Volatility3) or cross-view analysis (comparing /proc, /sys, and kernel data structures for inconsistencies). This skill covers using Volatility3 Linux plugins to detect syscall table hooks, hidden kernel modules, and modified function pointers, supplemented by live system scanning with rkhunter and chkrootkit.

Prerequisites

  • Volatility3 installed (pip install volatility3)
  • Linux memory dump (acquired via LiME, AVML, or /proc/kcore)
  • Volatility3 Linux symbol table (ISF) matching the target kernel version
  • rkhunter and chkrootkit for live system scanning
  • Reference known-good kernel image for comparison

Steps

Step 1: Acquire Memory Dump

Capture Linux physical memory using LiME kernel module or AVML for cloud instances.

Step 2: Analyze with Volatility3

Run linux.check_syscall, linux.lsmod, linux.hidden_modules, and linux.check_idt plugins to detect rootkit artifacts.

Step 3: Cross-View Analysis

Compare module lists from /proc/modules, lsmod, and /sys/module to identify modules hidden from one view but present in another.

Step 4: Live System Scanning

Run rkhunter and chkrootkit to detect known rootkit signatures, suspicious files, and modified system binaries.

Expected Output

JSON report containing detected syscall hooks, hidden kernel modules, modified IDT entries, suspicious /proc discrepancies, and rkhunter findings.

Skills enregistrées