Rights to individual papers remain with the author or the author's employer. Permission is granted for noncommercial reproduction of the work for educational or research purposes. This copyright notice must be included in the reproduced paper. USENIX acknowledges all trademarks herein. For more information about the USENIX Association:

This article describes concrete results and practically approved countermeasures concerning differential fault attacks on RSA using the CRT. It especially investigates smartcards with a RSA coprocessor where any hardware countermeasure to defeat such fault attacks have been switched off. This scenario has been chosen in order to completely analyze the resulting effects and errors occurring inside the hardware. Using the results of this kind of physical stress attack enables the development of completely reliable software countermeasures. Although

Elliptic curve cryptosystems in the presence of faults were studied by Biehl, Meyer and Müller (2000). The rst fault model they consider requires that the input point P in the computation of dP is chosen by the adversary. Their second and third fault models only require the knowledge of P . But these two latter models are less `practical' in the sense that they assume that only a few bits of error are inserted (typically exactly one bit is supposed to be disturbed) either into P just prior to the point multiplication or during the course of the computation in a chosen location. This paper

In this paper we show that, paradoxically, what seems like a "universal improvement" or a "straight-forward improvement" which enables better security and better reliability on a theoretical level, may in fact, within certain operational contexts, introduce new exposures and attacks, resulting in a weaker operational cryptosystem. We demonstrate a number of such dangerous "improvements". This implies that careful considerations should be given to the fact that an implemented cryptosystem exists within certain operational environments (which may enable certain types of tampering and other observed information channels via faults, side-channel attacks or behavior of system operators).

This paper experimentally evaluates and models the error-caused security vulnerabilities and the resulting security violations on two Linux kernel firewalls: IPChains and Netfilter. There are two major aspects to this work: to conduct extensive error injection experiments on the Linux kernel and to quantify the possibility of error-caused security violations using a Stochastic Activity Network (SAN) model. The error injection experiments show that about 2 % of errors injected into the firewall code segment cause security vulnerabilities. Two types of error-caused security vulnerabilities are distinguished: temporary, which disappear when the error disappears, and permanent, which persist even after the error is removed, as long as the system is not rebooted. Results from simulating the SAN model indicate that under an error rate of 0.1 error per day during a 1-year period in a networked system protected by 20 firewalls, two machines (on the average) will experience security violations. This indicates that error-caused security vulnerabilities can be a non-negligible source of a security threat to a highly secure system.

This paper presents a simple and e#cient method of protection against fault analysis when the underpinning cryptosystem uses modular arithmetic. The proposed method applies whatever the modular function to be evaluated and the used algorithms. Moreover, it only requires a very little overhead of extra computations, especially when the modulus is represented in diminished-radix form or when at least one factor of the modulus is known.