EventHealer: Bypassing data races in event-driven programs
- Tchamgoue, Guy Martin; Kim, Kyong Hoon; Jun, Yong-Kee
- Issue Date
- ELSEVIER SCIENCE INC
- Data races; Healing techniques; Event-driven programs
- JOURNAL OF SYSTEMS AND SOFTWARE, v.118, pp.208 - 220
- Journal Title
- JOURNAL OF SYSTEMS AND SOFTWARE
- Start Page
- End Page
- Data races represent a serious threat to the reliability of shared-memory concurrent programs including event-driven programs which handle asynchronous events. Despite the important number of existing testing and detection tools, data races often remain undetectable until the exploitation phase leading the application into unpredictable executions sometimes with disastrous consequences. To heal data races, current approaches which focus only on multithreaded programs are not directly applicable to event driven programs since they are still incomplete or incur a high runtime overhead which makes them also inappropriate for the exploitation phase. Thus, this paper proposes a hybrid technique that statically disassembles a program binary to collect information about critical sections and event handlers accessing each shared variable, to dynamically prevent data races from occurring by injecting a disabling and an enabling instruction respectively before and after every critical section, in order to enforce the synchronization property of the potentially harmful events. We implemented a prototype of this technique for sequential programs with signal handlers on top of the Pin instrumentation framework. An evaluation of this prototype proved its effectiveness and showed that our technique incurs only a negligible overhead to the monitored program. (C) 2016 Elsevier Inc. All rights reserved.
- Files in This Item
- There are no files associated with this item.
- Appears in
- 공과대학 > Department of Aerospace and Software Engineering > Journal Articles
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.