Implementation and implications of a stealth hard-drive backdoor

Modern workstations and servers implicitly trust hard disks to act as well-behaved block devices. This paper analyzes the catastrophic loss of security that occurs when hard disks

are not trustworthy. First, we show that it is possible to compromise the rmware of a commercial o-the-shelf hard drive, by resorting only to public information and reverse engineering. Using such a compromised rmware, we present a stealth rootkit that replaces arbitrary blocks from the disk while they are written, providing a data replacement back-door. The measured performance overhead of the compromised disk drive is less than 1% compared with a normal, non-malicious disk drive. We then demonstrate that a remote attacker can even establish a communication channel with a compromised disk to inltrate commands and to ex-filtrate data. In our example, this channel is established over the Internet to an unmodied web server that relies on the compromised drive for its storage, passing through the original webserver, database server, database storage engine, lesystem driver, and block device driver. Additional experiments, performed in an emulated disk-drive environment, could automatically extract sensitive data such as /etc/shadow (or a secret key le) in less than a minute. This paper claims that the diculty of implementing such an attack is not limited to the area of government cyber-warfare; rather, it is well within the reach of moderately funded criminals, botnet herders and academic researchers.