The classic chroot(2) system call just changes the directory entry that's used for the root in that process. It only affects pathname resolution (including reso

The classic chroot(2) system call just changes the directory entry that's used for the root in that process. It only affects pathname resolution (including resolution of the pathnames in symbolic links). If an unconstrained uid zero process is inside the chroot but can access or create device nodes, it can mount devices within the chroot just fine.