Leveraging EM Side-Channel Information to Detect Rowhammer Attacks

摘要: The rowhammer bug belongs to software-induced hardware faults, and has been exploited form a wide range of powerful attacks. Yet, how effectively detect such attacks remains challenging problem. In this paper, we propose novel approach named RADAR (Rowhammer Attack Detection via A Radio) that leverages certain electromagnetic (EM) signals particular, have found there are recognizable hammering-correlated sideband patterns in the spectrum DRAM clock signal. As inevitable physical side effects hammering DRAM, they can "expose" any potential including extremely elusive ones hidden inside encrypted isolated environments like Intel SGX enclaves. However, interest may become unapparent due common use spread-spectrum clocking (SSC) computer systems. We de-spreading method reassemble scattered by SSC. Using classification technique, achieve both effective robust detection-based defense against attacks, as evaluated on prototype under various scenarios. addition, our does not impose performance overhead protected system. There little prior work uses side-channel information perform defenses, best knowledge, is first investigation leveraging EM for purpose.