Static Race Detection for RTOS Applications
Abstract
We present a static analysis technique for detecting data races in Real-Time Operating System (RTOS) applications. These applications are often employed in safety-critical tasks and the presence of races may lead to erroneous behaviour with serious consequences. Analyzing these applications is challenging due to the variety of non-standard synchronization mechanisms they use. We propose a technique based on the notion of an "occurs-in-between" relation between statements. This notion enables us to capture the interplay of various synchronization mechanisms. We use a pre-analysis and a small set of not-occurs-in-between patterns to detect whether two statements may race with each other. Our experimental evaluation shows that the technique is efficient and effective in identifying races with high precision.
Cite
@article{arxiv.2010.02642,
title = {Static Race Detection for RTOS Applications},
author = {Rishi Tulsyan and Rekha Pai and Deepak D'Souza},
journal= {arXiv preprint arXiv:2010.02642},
year = {2020}
}
Comments
18 pages Accepted in FSTTCS 2020 This version contains detailed semantics