Possible New Threat: Malware That Targets Hardware
Researchers demonstrate proof-of-concept for developing malware that attacks specific hardware processors with 'surgical' precision
French researchers say it's possible to write malware that attacks specific hardware processors rather than operating systems or applications.
Anthony Desnos, Robert Erra, and Eric Filiol, of Ecole Sup�rieure d'Informatique Electronique Automatique (ESIEA) in Paris, have developed a proof-of-concept for hardware-specific malware, which they consider a step up from Stuxnet and a potentially key weapon in cyberwarfare. The malware can easily identify and target specific hardware systems based on the on-board processor chip, the researchers say.
More Security Insights
- Forrester Study: The Total Economic Impact of VMware View
- Securing Executives and Highly Sensitive Documents of Corporations Globally
- Innovations in Integration: Achieving Holistic Rapid Detection and Response
- Optimize Your SQL Environment for Performance & Flexibility
They used the so-called floating point arithmetic (FPA) to help identify processors, including AMD, Intel Dual-Core and Atom, SPARC, Digital Alpha, Cell, and Atom. Hardware malware doesn't exploit vulnerabilities in hardware -- it preys on actual features: "We just exploit differences in processor features. There will be always such differences," Filiol says.
In order to pinpoint the type of processor, the malware would see how a processor handles certain mathematical calculations. This breed of malware is not any more difficult to create than malware that targets software vulnerabilities, Filiol says. "The malware algorithm is the same. You just have to know which processor-specific information to use to trigger the attack," he says. The tricky part is that information is often a closely held secret, he says.
The researchers maintain that targeted attacks like Stuxnet are a major threat, but it's not always so simple for the attacker to be sure what software is running on a targeted machine. "While it can be very difficult to forecast and envisage which kind of applications is likely to be present on the target system (it can be a secret information), the variety in terms of hardware -- and especially as far as processors are concerned -- is far more reduced due to the very limited number of hardware manufacturers," the researchers wrote in their paper on the malware research.
Hardware malware gives cyberwarfare another weapon. "You can arrange things in such a way that effectively Iran buys a set of computers with Intel processor of a given type and family. Then you can strike them selectively -- and only these computers -- whatever Iran has installed on those computers, [whether it's] Linux, Windows, or any application," Filiol says.
Marc Maiffret, chief technology officer at eEye Digital Security, says he doesn't see hardware malware posing a major threat anytime soon. "While it is interesting to perform this sort of processor fingerprinting, malware will still need to look at other factors to make sure it is hitting the right target, as there is plenty of overlap in systems and what processors they use," Maiffret says. "To put it another way, I think we will continue to see targeting happening more in the way that Stuxnet did it than via processor fingerprinting."
Filiol, meanwhile, says he and his colleagues decided to publish part of their research to raise awareness about this threat. "Even rogue countries and bad guys are doing research. So attacks using those techniques can strike our own countries. That is why we have decided to publish part of our research: to make people aware of the threat," he says.
The malware could be used to wage Distributed Denial-of-Service (DDoS) and any other attack software malware can execute. The idea is for "far more precise and targeted attacks, at a finer level (surgical strikes) in a large network of heterogeneous machines but with generic malware," the research paper says.
There's no way for a processor manufacturer to mitigate such a targeted attack by "patching," either, "unless manufacturers would accept to use the same computation techniques and the same processor designs," he says. But that's obviously not a realistic option, he says.
A full copy of the research is available here (PDF) for download.
Have a comment on this story? Please click "Discuss" below. If you'd like to contact Dark Reading's editors directly, send us a message.