How this works

When your install is hijacked, the Arnix installer moves critical directories (like /usr and /etc) to a subdirectory in /arnix/generations. This breaks the Filesystem Hierarchy Standard (FHS) and makes everything not work anymore. Therefore we use bind mounts to fix this ASAP.

On boot, the initrd usually launches /sbin/init on the root filesystem. That file usually is your init system of choice.

However, Arnix replaces said init and uses that opportunity to set up the remaining system (remember: /usr, /etc, /var, ... are empty after the hijack).

Arnix's init mounts all the directories it hijacked and then hands off control to the real init, which is located at ... /sbin/init?

Why run /sbin/init again if that's just Arnix again?

It's not. The --bind option for mount just hides everything on the target directory. This is really convenient because we don't have to break the FHS.