一种粗细粒度结合的动态污点分析方法

针对当前污点分析工具不能兼顾速度和精确度的缺陷，研究并实现一种粗细粒度结合的二进制代码动态污点分析方法。对比粗粒度污点分析和细粒度污点分析的实现过程，提出两者结合的新型分析框架。预先在线执行粗粒度污点分析以筛选有效指令，之后离线执行细粒度污点分析以计算污点信息。根据粒度的差异分别建立粗细粒度污点数据的引入标记方法，制定粗细粒度条件下的数据流和控制流传播策略，设计离线轨迹记录结构作为粗细粒度污点分析的传递文件。在原型系统上的测试结果表明，该方法通过在线粗粒度模式保证了污点分析信息采集的快速性，同时采用离线细粒度模式以合理的时间消耗提升了污点分析的精确度。

A Dynamic Taint Analysis Method Combined with Coarse-grained and Fine-grained

Combination of coarse-grained and fine-grained Dynamic Taint Analysis（DTA） is developed to take speed and accuracy both into account. By comparing the realization process of coarse-grained DTA and fine-grained DTA, this paper proposes a new analysis framework. It executes online coarse-grained DTA to filter useful instruction, uses offiine fine-grained DTA to calculate taint information. Coarse-grained and fine-grained taint mark methods are established respectively by comparing the difference of taint analysis. Data-flow property strategy and control-flow property strategy are developed under the condition of coarse-grained DTA and fine-grained DTA. As a transfer file, offiine track record structure is designed to provide necessary information for fine-grained analysis. A prototype system is implemented and the experimental result proves that this method can ensure the rapid collection of taint information through online coarse-grained mode, and use offiine fine-grained mode to improve the accuracy with accepted time consumption.