动态二进制翻译中的标志位优化算法  

Condition code optimization in dynamic binary translation

作　　者：王荣华[1] 孟建熠[1] 陈志坚[1] 严晓浪[1] 

机构地区：[1]浙江大学超大规模集成电路设计研究所,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2014年第1期124-129,共6页Journal of Zhejiang University：Engineering Science

基　　金：中央高校基本科研业务费资助项目(2012QNA5004)

摘　　要：为了提高动态翻译器对标志位的模拟与处理效率，针对程序中比例较高的“比较一条件转移”指令对，提出标志位快速映射方法．该方法通过动态识别与提取源程序翻译块内的“比较一条件转移”指令对，利用目标架构的条件依赖关系特征实现“比较一条件转移”指令的高效映射，避免了对这类特殊的标志位定值与引用实施统一而复杂的处理，从而提高动态翻译与执行的速度．基于QEMU的模拟器运行基准程序显示，基于该方法翻译生成的目标标志位处理指令总数比采用其他主流方法减少约20％～90％．An efficient mapping method named compare and condition branch fast mapping algorithm was proposed in order to improve emulating and processing speed of condition flags. The algorithm mainly focuses on ＇compare and condition branch＇ instruction pairs which occupy a large proportion of condition code defining and using instruction pair. The method dynamically identifies and extracts the ＂compare and condition branch＂ instruction pair in the source block and completes instruction mapping by using the in- herent conditional dependencies of the target machine. By avoiding the complex and uniform traditional processes for these special instruction pairs, dynamic binary translator has achieved great performance im- provement. Results of benchmark on QEMU emulator showed that the generated instruction number for translating condition code was reduced by 20% to 90% than that of traditional methods.

关 键 词：动态二进制翻译 标志位定值与引用 比较一条件转移指令对 

分 类 号：TP314[自动化与计算机技术—计算机软件与理论] TN332[自动化与计算机技术—计算机科学与技术]