Very interesting site. It does add to the how of the NSA exploit. It also shows how this could also be used for another nefarious purpose namely DRM (preventing disk cloning). Yes this could be used to re install malware after the server was cleaned up, and this attack could work on a GNU/Linux system. The practical reality is that the attacker would need either root on the server / and or physical access to the hard drive. So it come down to the question how does the attacker get root in the first place? It is at this point where propriety operating systems provide a huge advantage to the attacker. Microsoft regularly provides access to the source code and advance knowledge of vulnerabilities to agencies such as the NSA, PLA and FSB. This creates a hugely uneven playing field since the attackers have access to the source code and vulnerabilities while the defenders do not. Stuxnet is a prime example of what can happen. In addition operating systems that support DRM must have hidden and obtuse parts to support the DRM. We must not forget that DRM and malware are in reality accomplishing the same thing. It is easy to spy on a user of say Windows, OS X or IOS  when those operating systems by design treat that same user as an adversary not to be trusted.  With GNU/Linux everyone has access to the source code creating a level playing field, and the operating system does not treat the user as an adversary not to be trusted. Big difference.