Home → Magazine Archive → May 2009 (Vol. 52, No. 5) → Lest We Remember: Cold-Boot Attacks on Encryption... → Abstract

Lest We Remember: Cold-Boot Attacks on Encryption Keys

By J. Alex Halderman, Seth D. Schoen, Nadia Heninger, William Clarkson, William Paul, Joseph A. Calandrino, Ariel J. Feldman, Jacob Appelbaum, Edward W. Felten

Communications of the ACM, Vol. 52 No. 5, Pages 91-98

[article image]

DRAM retains its contents for several seconds after power is lost. Although DRAM becomes less reliable when it is not refreshed, it is not immediately erased, and its contents persist sufficiently for malicious (or forensic) acquisition of usable full-system memory images.

The full text of this article is premium content


Lark Allen

Self-Encrypting Drives: The Solution to Cold-Boot Attacks

The ACM May, 2009, article "Lest We Remember: Cold-Boot Attacks on
Encryption Keys" is incorrect in its conclusion, stating "Though we
discuss several strategies for mitigating these risks, we know of no
simple remedy that would eliminate them." Self-encrypting drives (SED)
are commercially available from the leading drive vendors today and solve the security issues raised by OS and memory attacks by moving all encryption keys and authentication secrets out of the operating system and main memory and into the hardware-protected environment of the drive. Since all encryption/decryption of data is done in the disk controller and authenticated access to the drive is also performed inside the secure drive enclosure, none of these secrets are ever exposed to the kinds of attacks which can occur with software full disk encryption solutions running as application software or even system-level software.

While the authors briefly mention encryption in the drive controller in
Section 7.5 of the article as an alternative approach, it is clear that
they do not understand the security model and data protection solution provided by self-encrypting drives. SEDs solve the fundamental data at rest problem of making sure that only authorized users can access the encrypted data after the drive has been locked by any power-off event such as removing the drive, shutting the lid, etc. The authors describe a scenario where SED drives from Seagate and Hitachi were unlocked by providing the correct authentication passwords, this is exactly what they were designed to do for 'authorized' users. But then the authors talk about being able to 'transplant' the unlocked drives from one system to another by keeping the SATA power cables connected. If the authors have access to an unlocked drive, why not read the data then, instead of needlessly moving the drive? The data at rest security model for SED drives is very straight forward: the drive locks whenever the power is removed from the drive and can only be unlocked by authorized users at the pre-boot time, before any OS or applications are available. Data access is provided to all authenticated users and 'transplantation' of a powered up, unlocked drive is a technical feature, not a security exposure.

Self-encrypting drives provide the industry's best solution to the
described problems of memory attacks against encryption keys and
authentication credentials held in system memory. In addition,
SEDs are faster, easier to use and manage, and are more secure than any of today's software full disk encryption alternatives.

Displaying 1 comment