Quality Assurance for Non-trivial Systems: Use Case GCC Plugins

There are software systems that function without human interven- tion. Such systems work according to preset rules, algorithms, or data. In this doctoral research, we aim to explore the testing-related challenges associated with such systems with GCC plugins as our use case. The GCC compiler family is widely known especially as a compiler for C and C++. GCC allows extensions to be added through the GCC Plugin API. GCC plugins execute at the com- piler level and influence the compiler at different compiling stages. Plugins depend on the GCC internal data structures such as AST, GIMPLE, Basic Blocks and RTL. Therefore, it is difficult to test the correctness of a GCC plugin. Testing is essential for ensuring the quality and the correctness of the plugins. The attempts made in the past are insufficient. They depend on testing the plugins from a black-box perspective. Internal issues may remain hidden when only testing from a black-box perspective is considered. Testing the plugins from both white-box and black-box perspectives is essential to guarantee their quality. We intend to shed light on the complex- ity and challenges of GCC plugin testing. We propose a four-tiered approach for GCC plugin testing. The approach consists of static and dynamic testing techniques. The main challenge is white box testing. This research aims to address the challenge and provide a solution by utilising logs focusing on higher test coverage. We intend to utilise the findings of the study with other comparable systems