基于MIPS架构的内存虚拟化研究  

Memory Virtulization on MIPS Architecture

作　　者：蔡万伟[1,2,3,4] 台运方[1,2,3] 刘奇[1,3] 张戈[3,4] 

机构地区：[1]中国科学院计算技术研究所,北京100190 [2]中国科学院大学,北京100049 [3]龙芯中科技术有限公司,北京100190 [4]中国科学院重庆绿色智能技术研究院,重庆401122

出　　处：《计算机研究与发展》2013年第10期2247-2252,共6页Journal of Computer Research and Development

基　　金："核高基"国家科技重大专项基金项目(2009ZX01028-002-003;2009ZX01029-001-003;2010ZX01036-001-002);国家自然科学基金项目(60921002;61003064;61050002;61070025;61100163;61133004;61173001;61232009);国家"八六三"高技术研究发展计划基金项目(2012AA010901;2012AA011002;2012AA012202;2013AA014301);重庆市科技攻关资助项目(cstc2011ggC40010)

摘　　要：内存虚拟化是系统虚拟化中如何有效抽象、利用、隔离计算机物理内存的重要方法,决定着系统虚拟化的整体性能.传统的纯软件内存虚拟化方法会产生较大的资源开销并且兼容性差,而硬件辅助的内存虚拟化方法需要重新设计处理器硬件架构.基于MIPS架构处理器提出一种软硬件协同的内存虚拟化方法,在不增加硬件支持的情况下提高内存虚拟化性能.提出的多层虚拟地址空间模型不仅可以解决MIPS架构处理器存在的虚拟化缺陷,而且可以在已有的内存虚拟化方法上提高性能.在多层虚拟地址空间模型的基础上,提出基于地址空间标识码(address space identity,ASID)、动态划分的旁路转换缓冲(translation lookaside buffer,TLB)共享方法,降低了虚拟机切换的开销.最终,在MIPS架构的龙芯3号处理器上实现了系统虚拟机VIRT-LOONGSON.性能测试表明,提出的方法可以提高大多数测试程序的性能,达到二进制翻译执行性能的3~5倍,并在TLB模拟方法的基础上提高了5%~16%的性能.Memory virtualization is one of the most important methods to effectively abstract, utilize and separate computer＇s physic memory, and it decides overall performance of system virtualization. However, the traditional software-based methods often suffer from the inefficiency and complexity. The traditional hardware-assisted methods require the unavoidable re-design of the processor architecture. This paper presents a novel hardware-software co-designed method to accelerate the memory virtualization method on MIPS architecture processor. It improves the system performance without increasing any other hardware. This paper introduces MLASM （multiple layer address space model）, which not only fills memory virtualization hole in MIPS architecture processor, but also enhances performance on the basis of the existing memory virtualization method. Meanwhile, this paper introduces （translation lookaside buffer, TLB） share method based on multiple layer address space, reduces the expenses when virtual machines are switching. Finally, a system virtual machine called VIRT-LOONGSON on MIPS architecture processor LOONGSON-3 is implemented. Performance evaluation shows that the proposed method can speedup most benchmark programs by nearly 3 to 5 times compared with binary translation method, and improves 5% to 16% performance compared with TI.B simulation method.

关 键 词：系统虚拟化 内存虚拟化 旁路转换缓冲共享 MIPS 龙芯3号处理器 

分 类 号：TP316[自动化与计算机技术—计算机软件与理论]