nixos: add a "hardened" profile

The idea is to provide a convenient way to enable most vanilla hardening
features in one go.  The hardened profile, then, will serve as a place for
features that enhance security but cannot be enabled for all deployments
because they interfere with legitimate use cases (e.g., using ptrace to
debug problems in an already running process).

Closes https://github.com/NixOS/nixpkgs/pull/24680
This commit is contained in:
Joachim Fasting 2017-04-06 14:54:45 +02:00
parent 1b3300bf01
commit 063ac40304
No known key found for this signature in database
GPG key ID: 7544761007FE4E08

View file

@ -0,0 +1,35 @@
# A profile with most (vanilla) hardening options enabled by default,
# potentially at the cost of features and performance.
{ config, lib, pkgs, ... }:
with lib;
{
security.hideProcessInformation = mkDefault true;
security.apparmor.enable = mkDefault true;
# Restrict ptrace() usage to processes with a pre-defined relationship
# (e.g., parent/child)
boot.kernel.sysctl."kernel.yama.ptrace_scope" = mkOverride 500 1;
# Prevent replacing the running kernel image w/o reboot
boot.kernel.sysctl."kernel.kexec_load_disabled" = mkDefault true;
# Restrict access to kernel ring buffer (information leaks)
boot.kernel.sysctl."kernel.dmesg_restrict" = mkDefault true;
# Hide kptrs even for processes with CAP_SYSLOG
boot.kernel.sysctl."kernel.kptr_restrict" = mkOverride 500 2;
# Unprivileged access to bpf() has been used for privilege escalation in
# the past
boot.kernel.sysctl."kernel.unprivileged_bpf_disabled" = mkDefault true;
# Disable bpf() JIT (to eliminate spray attacks)
boot.kernel.sysctl."net.core.bpf_jit_enable" = mkDefault false;
# ... or at least apply some hardening to it
boot.kernel.sysctl."net.core.bpf_jit_harden" = mkDefault true;
}