PTSAnalysis - A Static Program Analysis Tool to Capture Inter-Function Data Dependencies

Data dependencies are a crucial aspect when software needs to be run on multiprocessor systems-on-chip (MPSoC). These dependencies between different code sections can lead to data exchange between the different cores. With the increasing popularity of MPSoCs, developers are aiming to restructure legacy code to run on multiple cores and best utilize the available compute power. A key challenge is how these data dependencies are captured and made available in an effective manner. To this end we develop PTSAnalysis, a static program analysis tool that captures inter-function data dependencies. The tool is based on an open-source library and uses Pointer Analysis to determine how (via read or write) and where (in which functions) memory objects are being accessed. To make the resulting data dependencies available to developers, we implement a VSCode extension and its dedicated language server. Finally, we apply our tool to two examples and highlight how it can be used during software development as well as for further explorations of parallelization opportunities.