An Empirical Study on Concurrency Bugs in Interrupt-Driven Embedded Software

Abstract

Interrupt-driven embedded software is widely used in aerospace, automotive electronics, medical equipment, IoT, and other industrial fields. This type of software is usually programmed with interrupts to interact with hardware and respond to external stimuli on time. However, uncertain interleaving execution of interrupts may cause concurrency bugs, resulting in task failure or serious safety issues. A deep understanding of real-world concurrency bugs in embedded software will significantly improve the ability of techniques in combating concurrency bugs, such as bug detection, testing and fixing.This paper performs the first comprehensive and large-scale empirical study on concurrency bugs in industrial interrupt-driven embedded software. A total number of 132 real-world concurrency bugs in 102 industrial embedded software have been rigorously analyzed. Not only have the root causes, impacts and fix strategies of bugs been studied, but also the manifestation, including triggering scopes, racing variables, access interleaving patterns, and variables correlations. This study reveals several significant findings, which can guide future research in developing techniques and tools to combat concurrency bugs for interrupt-driven embedded software.