Blind Format String Attacks

Abstract

AbstractAlthough Format String Attacks (FSAs) are known for many years there is still a number of applications that have been found to be vulnerable to such attacks in the recent years. According to the CVE database, the number of FSA vulnerabilities is stable over the last 5 years, even as FSA vulnerabilities are assumingly easy to detect. Thus we can assume, that this type of bugs will still be present in future. Current compiler-based or system-based protection mechanisms are helping to restrict the exploitation this kind of vulnerabilities, but are insufficient to circumvent an attack in all cases.Currently FSAs are mainly used to leak information such as pointer addresses to circumvent protection mechanisms like Address Space Layout Randomization (ASLR). So current attacks are also interested in the output of the format string. In this paper we present a novel method for attacking format string vulnerabilities in a blind manner. Our method does not require any memory leakage or output to the attacker. In addition, we show a way to exploit format string vulnerabilities on the heap, where we can not benefit from direct destination control, i.e. we can not place arbitrary addresses onto the stack, as is possible in stack-based format string.KeywordsSecurityFormat string attacks