Proceedings of ISP RAS


Method for exploitability estimation of program bugs

A.N. Fedotov (ISP RAS, Moscow, Russia)

Abstract

The method for exploitability estimation of program bugs is presented. Using this technique allows to prioritize software bugs that were found. Thus, it gives an opportunity for a developer to fix bugs, which are most security critical at first. The method is based on combining preliminary classification of program bugs and automatic exploit generation. Preliminary classification is used to filter non-exploitable software defects. For potentially exploitable bugs corresponding exploit generation algorithm is chosen. In case of a successful exploit generation the operability of exploit is checked in program emulator. There are various ways that used for finding software bugs. Fuzzing and dynamic symbolic execution are often used for this purpose. The main requirement for the use of the proposed method is an opportunity to get input data, which cause program to crash. The technique could be applied to program binaries and does not require debug information. Implementation of the method is a set of software tools, which are interconnected with control scripts. The preliminary classification method and automatic exploit generation method are implemented as stand-alone tools, and could be used separately. The technique was used to analyze 274 program crashes, which were obtained by fuzzing. The analysis managed to detect 13 exploitable bugs, for which successfully workable exploits were generated.

Keywords

vulnerability; buffer overflow; symbolic execution; exploit; binary code

Edition

Proceedings of the Institute for System Programming, vol. 28, issue 4, 2016, pp. 137-148.

ISSN 2220-6426 (Online), ISSN 2079-8156 (Print).

DOI: 10.15514/ISPRAS-2016-28(4)-8

Full text of the paper in pdf (in Russian) Back to the contents of the volume