PatchNet: Hierarchical Deep Learning-Based Stable Patch Identification for the Linux Kernel

Abstract

Linux kernel stable versions serve the needs of users who value stability of the kernel over new features. The quality of such stable versions depends on the initiative of kernel developers and maintainers to propagate bug fixing patches to the stable versions. Thus, it is desirable to consider to what extent this process can be automated. A previous approach relies on words from commit messages and a small set of manually constructed code features. This approach, however, shows only moderate accuracy. In this paper, we investigate whether deep learning can provide a more accurate solution. We propose PatchNet, a hierarchical deep learning-based approach capable of automatically extracting features from commit messages and commit code and using them to identify stable patches. PatchNet contains a deep hierarchical structure that mirrors the hierarchical and sequential structure of commit code, making it distinctive from the existing deep learning models on source code. Experiments on 82,403 recent Linux patches confirm the superiority of PatchNet against various state-of-the-art baselines, including the one recently-adopted by Linux kernel maintainers.

CC2Vec

“…Two techniques using deep neural networks, PatchNet [22] and DeepJIT [21], are most similar to our work. However, as discussed earlier, our work differs from theirs in various ways.…”

“…State-of-the-art Approach. The state-of-the-art approach is PatchNet [22], which represents the removed (added) code as a three dimensional matrix. The dimensions of the matrix are the number of hunks, the number of lines in each hunk, and the number of words in each line.…”

“…The code change indicates the lines of code to remove or add across one or multiple files. Research has shown that the study of historical patches can be employed to solve software engineering problems, such as just-in-time defect prediction [21,28], identification of bug fixing patches [22,57], tangled change prediction [34], recommendation of a code reviewer for a patch [50], and many more.…”

“…To evaluate the effectiveness of CC2Vec, we employ the representation learned by CC2Vec in three software engineering tasks: 1) log message generation [39] 2) bug fixing patch identification [22] and 3) just-in-time defect prediction [21]. In the first task of log message generation, we generate the first line of a log message given a code change.…”

Mining Software Repositories with a Collaborative Heuristic Repository

“…We plan to add additional heuristics that look at other data sources, such as actual change patterns, to create an improved labelling model. We also plan to define models that look at data beyond the commit message, such as change metrics or actual source code changes [24]. A further way to improve performance will also be to scale up the end model, by training a larger capacity model on a larger dataset.…”