摘要
为有效降低X86源指令集体系结构标志位的模拟开销,针对软件X86系统仿真中的动态二进制翻译过程,提出了以寄存器直接映射技术为基础的、由自定义指令控制模拟标志寄存器读写的软硬协同标志位处理优化方法。该方法能够有效降低翻译后目标代码膨胀率,提高翻译后目标代码执行效率。测试结果表明,该协同设计能够正确完成带标志位操作的X86指令翻译、模拟标志寄存器的读取与修改,在Linux-0.2操作系统启动过程测试集下,翻译后目标代码膨胀率相对优化前降低了29.6%。
To reduce the emulation cost of the X8 6 instruction set architecture’s condition code during the dynamic binary translation of software X8 6 system emulation,a co-designed condition code optimization approach based on direct register mapping and the custom in-struction controlled EEFLAG (emulated Eflags)was presented.The approach could shrink the expand rate of translated code efficiently, and improve the execution efficiency of translated code.Verification and evaluation showed that the co-designed condition code optimiza-tion approach could correctly complete the translation of X86 instructions,read or modify the EEFLAG.In the process of booting Linux-0. 2 operating system,the expand rate of translated code after optimization came down by 2 9. 6%.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第7期2476-2481,共6页
Computer Engineering and Design
基金
国家863高技术研究发展计划基金项目(2009AA012200)
关键词
标志位优化
软硬协同
系统仿真
自定义指令
寄存器映射
condition code optimization
hardware software co-design
system emulation
custom instruction
register mapping