A Study of the Impact of Callbacks in Staged Static+Dynamic Partial Analysis

A Study of the Impact of Callbacks in Staged Static+Dynamic Partial Analysis

연구 분야: Verification

학회: SPLASH Companion 2022: Companion Proceedings of the 2022 ACM SIGPLAN International Conference on Systems, Programming, Languages, and Applications: Software for Humanity

Partial analysis is a program analysis technique used in compilation systems when the whole program is not available. Many recent promising approaches perform partial analysis statically that involves identifying the interprocedural dependencies across program elements. These generated dependencies further get evaluated during runtime while generating the final analysis result. However, as the application and library methods are analyzed independently during static analysis, these approaches do not account for the effect of dynamic features such as callbacks. Consequently, in such scenarios, the runtime (say the Java Virtual Machine) needs to discard the static-analysis results and use the existing imprecise builtin analyses. The primary goal of this work is to find out the percentage of objects and methods that may get affected by callbacks, and to propose possible techniques to enhance the generation of dependencies in their presence.