摘要
Java语言的动态特性使程序的最坏执行时间分析较悲观和难以预测,提出一种精确最坏执行时间分析方法,在高层分析中,引入一种标记方法,对带有标记的Java类文件进行反编译提取控制流程,得到每一个基本块中的Java字节码指令的最坏情况下的执行次数,在底层分析中,建立结合流水线和高级缓存影响的时间模型,得到每条指令所对应的执行时间,最后结合高层分析和底层分析的结果得到程序的最坏情况下的执行时间。实验表明,该方法可以使对实时Java程序的最坏情况执行时间预测更加安全和精确。
The dynamic character of Java language makes it pessimistic and hard to predict for program worst-case execution time analysis,a precise program worst-case execution time analysis method is proposed.In high level analysis,an annotation method is introduced,through decompiling Java class file with the introduced annotation,the approach extract control flow and get execution times of the Java bytecode instruction of every basic block in the worst case,in low level analysis,time model combining pipeline and Cache effect is built and execution time of corresponding instruction is gotten.By combining the result of high level and low level,the approach get the program execution time in the worst case.Experiments show that the method makes it more safe and precise for real-time Java program worst-case execution time prediction.
出处
《计算机工程与应用》
CSCD
北大核心
2010年第18期60-64,共5页
Computer Engineering and Applications
基金
国家自然科学基金No.60703101~~