Experience Report: Fault Triggers in Linux Operating System: from Evolution Perspective

A Metric Based On Mathematics And Information Theory For Software Reliability

“…Sullivant and Chillarege [2] examined MVS operating system failures, defined error types to classify low level programming mistakes and defined error triggers to classify events that caused latent errors in programs. Xiao et al [3] studied Linux operating system errors and defined a bug classification based on fault triggering conditions. Through the studies we have known error locations, error distributions, error types and triggering conditions across operating systems.…”

An Empirical Study of Regression Bug Chains in Linux

“…As shown in Table I, we utilized two types of data, including Linux regression bug reports and Linux Git repository commits. 2 The Linux regression bugs are obtained from [15]. In this work, among the 4035 classified bugs, there are 2020 regression bugs.…”

“…In addition, more than half of regression bugs possess high priorities. Recently, Xiao et al [15] have reported that 50.1% of classified bugs in Linux are regression bugs, and regression bugs are more likely to be bohrbugs, i.e., bugs that can be consistently reproduced under a well-defined set of conditions since their activation and/or error propagation are simple. To the best of our knowledge, our study is the first work to investigate RBCs in Linux and their characteristics based on the proposed directed bipartite network.…”

“…For example, Linux regression bug ID-51881 was introduced by commit ID-65fe1f0f, 1 whose purpose is to implement a new feature for the SATA device. Previous studies found that regression bugs account for a significant proportion (50.1%) of all classified bugs in Linux [15]. In the Google Chromium project [23], regression bugs occupy about 51.1% of all labeled bugs.…”

“…risk measurement [8]- [10], understanding [11]- [15], detecting [16]- [18], and predicting bugs [19], [20].…”