摘要： 通过基于包含的指针分析在线优化技术中的纵向传播方法，改进基于调用图上下文敏感指针分析的环消除技术，提出一种上下文敏感的纵向传播算法。初始化约束图，在约束图中进行环探测及其合并，执行差异传播并循环处理复杂约束，直到一次调用纵向传播例程后图中所有结点指向集不变为止，从而得到图中各结点到其指向集的映射。应用CIL工具的实验结果表明，该算法能有效地对源程序进行上下文敏感的指针分析，与环消除技术相比，在分析大规模程序时具有更高的时间效率。

关键词: 在线优化, 纵向传播, 调用图, 上下文敏感指针分析, 环消除, 约束图

Abstract: Through using Deep Propagation(DP) algorithm, which is the techniques of inclusion based pointer analysis on-line optimization technology, to optimize the technology of cycle elimination based on invocation graph context-sensitive pointer analysis, and then puts forward a new context-sensitive DP algorithm. It initializes constraint graph, explores and merges ring, executes difference propagation and circulates processing complex constraints in the diagram, until all the point-to set of all nodes no longer change after executive vertical transmission routines for once, then gets mapping of each node in the graph to the point-to set. Using the CIL tool for experiment, results show that the proposed algorithm has higher time efficiency when analyzing the same large scale program compared with cycle elimination technology.

Key words: on-line optimization, Deep Propagation(DP), invocation graph, context-sensitive pointer analysis, cycle elimination, constraint graph

中图分类号: 

• TP311