On the Implementation of Data-Breakpoints Based Race Detection for Linux Kernel Modules.

On the Implementation of Data-Breakpoints Based Race Detection for Linux Kernel Modules.

Authors

Nikita Komarov.

Abstract

An important class of problems in software are race conditions. Errors of this class are becoming more common and more dangerous with the development of multi-processor and multi-core systems, especially in such a fundamentally parallel environment as an operating system kernel. The paper overviews some of existing approaches to detect race conditions including DataCollider system based on concurrent memory access tracking. RaceHound, a race condition detection system for Linux drivers based on similar principles as DataCollider is presented.

Full text of the paper in pdf

Keywords

verification; driver verification; race condition; linux kernel; dynamic verification; operating system

Edition

Proceedings of the Spring/Summer Young Researchers' Colloquium on Software Engineering, 2013 (7)

DOI: 10.15514/SYRCOSE-2013-7-14

978-5-91474-020-4

Research Group

Software Engineering

All publications during 2013 All publications