SuperAPI AI Directory

kernel-security

Security & auditing

View repository

Updated 10-15 15:45

About this skill

Kernel Security

Purpose

Guide agents through Linux kernel security: LSM frameworks (SELinux, AppArmor), seccomp-bpf with libseccomp, KASLR and bypass mitigations, Intel CET (Shadow Stack + IBT), ARM PAC and BTI, kernel sanitizers (KASAN, KMSAN), and CVE triage for kernel vulnerabilities.

When to Use

  • Writing SELinux or AppArmor policies for confined services
  • Sandboxing processes with seccomp-bpf filters
  • Hardening binaries with CET, PAC, or BTI
  • Enabling KASAN on kernel builds for vulnerability research
  • Triaging kernel CVE impact on your distro/kernel version
  • Designing container or microservice security boundaries

Workflow

1. LSM framework overview

Application syscall
    → DAC (uid/gid, file mode)
    → LSM hook (SELinux/AppArmor/Yama/...)
    → Capability check
    → seccomp filter
    → Kernel
# Active LSM
cat /sys/kernel/security/lsm
# common: lockdown,capability,yama,apparmor,safesetid

2. SELinux policy

# Check SELinux status
getenforce
sestatus

# Context of file/process
ls -Z /usr/sbin/nginx
ps -eZ | grep nginx

# Audit denials
ausearch -m avc -ts recent
sealert -a /var/log/audit/audit.log

Policy module example:

# myapp.te
policy_module(myapp, 1.0.0)

type myapp_t;
type myapp_exec_t;
type myapp_log_t;

init_daemon_domain(myapp_t, myapp_exec_t)

allow myapp_t myapp_log_t:file { create write append open };
allow myapp_t self:tcp_socket { create bind listen accept };
checkmodule -M -m -o myapp.mod myapp.te
semodule_package -o myapp.pp -m myapp.mod
semodule -i myapp.pp

3. AppArmor profiles

# Generate complain-mode profile
aa-genprof /usr/bin/myapp

# Enforce
aa-enforce /etc/apparmor.d/usr.bin.myapp

# Check status
aa-status
# /etc/apparmor.d/usr.bin.myapp
#include <tunables/global>

/usr/bin/myapp {
  #include <abstractions/base>

  /usr/bin/myapp mr,
  /var/log/myapp.log w,
  /etc/myapp/config r,
  network bind tcp,
  deny /etc/shadow r,
}

4. seccomp-bpf with libseccomp

#include <seccomp.h>

int sandbox(void) {
    scmp_filter_ctx ctx = seccomp_init(SCMP_ACT_KILL_PROCESS);

    // Allow read/write/exit/mmap
    seccomp_rule_add(ctx, SCMP_ACT_ALLOW, SCMP_SYS(read), 0);
    seccomp_rule_add(ctx, SCMP_ACT_ALLOW, SCMP_SYS(write), 0);
    seccomp_rule_add(ctx, SCMP_ACT_ALLOW, SCMP_SYS(exit_group), 0);
    seccomp_rule_add(ctx, SCMP_ACT_ALLOW, SCMP_SYS(mmap), 0);
    seccomp_rule_add(ctx, SCMP_ACT_ALLOW, SCMP_SYS(brk), 0);

    // Return EPERM instead of kill for open
    seccomp_rule_add(ctx, SCMP_ACT_ERRNO(EPERM), SCMP_SYS(open), 0);
    seccomp_rule_add(ctx, SCMP_ACT_ERRNO(EPERM), SCMP_SYS(openat), 0);

    return seccomp_load(ctx);
}
# Export BPF filter for audit
scmp_filter_ctx ctx = ...;
seccomp_export_bpf(ctx, fd);

# strace to find needed syscalls before tightening
strace -c ./myapp

5. KASLR

# Check KASLR enabled
cat /proc/sys/kernel/randomize_va_space   # 2 = full
dmesg | grep KASLR

# Kernel cmdline
grep kaslr /proc/cmdline

Mitigations against KASLR leaks:

  • No /proc/<pid>/maps to untrusted
  • Pointer hashing in %pK printk
  • eBPF restricted on unprivileged

6. Intel CET

# Compile with CET
gcc -fcf-protection=full -o app app.c

# Verify shadow stack and IBT
readelf -n app | grep -E 'SHSTK|IBT'
readelf --notes app | grep -A2 GNU_PROPERTY
Feature Protects against
SHSTK (Shadow Stack) ROP return address overwrites
IBT (Indirect Branch Tracking) CALL/JMP to non-ENDBR targets

Requires CPU with CET (Intel Tiger Lake+; AMD Zen 3+ on CPUs with shadow-stack support) and kernel CET support.

7. ARM PAC and BTI

# GCC/Clang branch protection
gcc -mbranch-protection=standard -o app app.c
# PAC (pointer authentication) + BTI (branch target identification)

# Verify
readelf -n app | grep -E 'GNU_PROPERTY_AARCH64_FEATURE_1'
llvm-objdump -d app | grep bti

PAC signs return addresses and pointers with cryptographic keys (ARMv8.3+). BTI marks valid branch targets — invalid jumps fault.

8. Kernel memory tagging

# KASAN kernel build
# CONFIG_KASAN=y in kernel .config
make menuconfig  # Kernel hacking → KASAN

# Boot with KASAN kernel
# Reports use-after-free, OOB with stack trace

# KMSAN (uninitialized memory)
# CONFIG_KMSAN=y — kernel equivalent of MSan
# KASAN report example fields
# BUG: KASAN: slab-out-of-bounds in ...
# Call trace: ...

9. CVE triage workflow

# Check kernel version
uname -r

# Distro security tracker
# Ubuntu: ubuntu-security-notices
# RHEL: errata

# NVD lookup
# https://nvd.nist.gov/vuln/detail/CVE-XXXX-XXXXX

# Is patch backported?
zgrep -l CVE-2024-XXXX /usr/share/doc/linux-*/changelog.Debian.gz

Triage checklist:

  1. Affected subsystem (net, fs, drivers)?
  2. Local or remote exploit?
  3. Fixed in your kernel version?
  4. Mitigation without patch (disable module, sysctl)?

Common Problems

Symptom Cause Fix
SELinux denials Missing allow rule audit2allow; refine policy
AppArmor profile break Path mismatch Update profile paths; use globs
seccomp kills app Missing syscall strace; add allow rule
CET not active Old CPU/kernel Check /proc/cpuinfo flags
KASAN kernel slow 2-5x overhead Use only in test VMs
False sense of security LSM bypass via kernel bug Defense in depth; keep kernel updated

Related Skills

  • skills/virtualization/containers-internals — container seccomp and caps
  • skills/runtimes/binary-hardening — userspace CET, RELRO, PIE
  • skills/runtimes/sanitizers — ASan/HWASan userspace counterparts
  • skills/observability/ebpf — LSM BPF programs
  • skills/kernel/kernel-debugging — analyze KASAN reports
  • skills/security/reverse-engineering — exploit analysis