The Linux kernel: a case study of build system variability. Issue 8 (18th April 2013)

• Record Type: Journal Article
• Authors: Nadi, Sarah; Holt, Ric; Mens, T.; Cleve, A.
• Is Part Of: Journal of software. Volume 26:Issue 8(2014:Aug.)
• Page Start: 730
• Page End: 746
• Publication Date: 2013-04-18
• Subjects: Software engineering -- Periodicals; Computer software -- Development -- Periodicals; Software maintenance -- Periodicals; 005.1
• DOI: 10.1002/smr.1595
• Languages: English
• ISSNs: 2047-7473

Abstract

<abstract abstract-type="main"> <title>SUMMARY</title> <p>Although build systems control what code gets compiled into the final built product, they are often overlooked when studying software variability. The Linux kernel is one of the biggest open source software systems supporting variability and contains over 10, 000 configurable features described in its <sc>Kconfig</sc> files. To understand the role of the build system in variability implementation, we use Linux as a case study. We study its build system, <sc>Kbuild</sc>, and extract the variability constraints in its Makefiles. We first provide a quantitative analysis of the variability in <sc>Kbuild</sc>. We then study how the variability constraints in the build system affect variability anomalies detected in Linux. We concentrate on dead and undead artifacts, and by extending previous work, we show that considering build system variability constraints allows more anomalies to be detected. We provide examples of such anomalies on both the code block and source file level. Our work shows that <sc>Kbuild</sc> contains a large percentage of the variability information in Linux, so it should not be ignored during variability analysis. Nonetheless, the anomalies we find suggest that variability on the file level in <sc>Kbuild</sc> is consistent with <sc>Kconfig</sc>, whereas the constraints on the code level are harder to keep consistent with both <sc>Kbuild</sc> and <sc>Kconfig</sc>. Copyright © 2013 John Wiley & Sons, Ltd.</p> </abstract>