Frappé: Querying the Linux Kernel Dependency Graph

Abstract

Frappé is a developer tool for querying and visualizing the dependencies of large C/C++ software systems to the order of 10s of millions of lines of code in size. It supports developers with a range of code comprehension queries such as Does function X or something it calls write to global variable Y? and How much code could be affected if I change this macro? Results are overlaid on a visualization of the dependency graph data based on a cartographic map metaphor. In this paper, we give a brief overview of Frappé and describe our experiences implementing it on top of the Neo4j graph database. We detail the graph model used by Frappé and outline its key use cases using representative queries and their runtimes with the dependency graph data of the Unbreakable Enterprise Kernel. Finally, we discuss some of the open challenges in supporting source code queries across single and multiple versions of an evolving codebase with current property graph database technologies: performance, efficient storage, and the expressivity of the graph querying language given a graph model.