A Safer Wrapper Is Not a Modernisation Strategy

Mon, 05 Oct 2026 08:00:00 +0200

Moving a legacy product to a modern operating system is useful engineering. It can deliver a maintained kernel, current libraries, better compiler defaults, ASLR, mandatory access control, namespaces, cgroups and service supervision. All of those can make exploitation harder or contain a failure.

It can also leave the product’s old trust model almost entirely intact.

The new kernel may still launch the same privileged packet engine, the same configuration daemon, the same PHP application and the same Perl and shell helpers. A process running as root retains broad authority unless capabilities, namespaces or mandatory policy materially constrain it. A shell command built from attacker-controlled text remains a shell command. A parser written in C remains memory unsafe even when systemd starts it.

The wrapper buys time and lowers some risks. It does not retire the engineering debt inside it.

Security Engineering on Martin's Blog

A Safer Wrapper Is Not a Modernisation Strategy