mips 指令简介

Machine language vs. assembly language

• Real machine language level programming means to handle the bit encodings of machine instructions 

Example (MIPS CPU: addition $t0 ← $t0 + $t1): 1000010010100000000100000

 • Assembly language introduces symbolic names (mnemonics) for machine instructions and makes programming less error-prone: 

Example (MIPS CPU: addition $t0 ← $t0 + $t1): add $t0, $t0, $t1 

• An assembler translates mnemonics into machine instructions – Normally: mnemonic 1:1 ←→ machine instruction 

– Also: the assembler supports pseudo instructions which are translated into series of machine instructions (mnemonic 1:n ←→ machine instruction)

mips是RISC的鼻祖，每条指令固定的占用4个字节，指令格式分为以下几类：

• R-type instructions

Converting an R mnemonic into the equivalent binary machine code is performed in the following way:

opcode	rs	rt	rd	shift (shamt)	funct
6 bits	5 bits	5 bits	5 bits	5 bits	6 bits

All R-type instructions use opcode 000000.

• I-type instructions

opcode	rs	rt	IMM
6 bits	5 bits	5 bits	16 bits

All opcodes except 000000, 00001x, and 0100xx are used for I-type instructions.

• J-type instructions

Opcode

Pseudo-Address

J-type instructions use opcodes 00001x.

• coprocessor Instructions

MIPS processors all have two standard coprocessors, CP0 and CP1. CP0 processes various kinds of program exceptions. CP1 is a floating point processor. The MIPS architecture makes allowance for future inclusion of two additional coprocessors, CP2 and CP3.

All coprocessor instructions instructions use opcodes 0100xx.

mips指令的作用主要分为以下几大类：

1. load ，store， and Data Movement

Typical for the RISC design, MIPS is a load-store architecture:

 – Memory is accessed only by explicit load and store instructions 

– Computation (e.g., arithmetics) reads operands from registers and writes results back into registers

• MIPS: load word/halfword/byte at address a into target register r (r ← (a)): 

• MIPS: store word/halfword/byte in register r at address a (a ← r):

MIPS can move data between registers directly (no memory access involved)

除了显示的load-store可访问内存外，其他的操作都是在寄存器之间进行的，注意这里的内存仅仅指：32个通用寄存器，一个pc寄存器，一个HI和LO寄存器，另外还有协处理器自己的寄存器，除此之外所有的都看做外部内存，包括设备寄存器，即通过总线接口单元BIU单元访问的都是外部内存，这时访问需要耗费较多cpu时间。

 2.arithmetic instructions

3.shift/rotate instructions

4.logical instructions

• MIPS CPUs provide instructions to compute common boolean functions

5.comparison instructions

• Compare the values of two registers (or one register and a constant) 

– Comparison successful: r ← 1 fails: r ← 0

 6.branch and jump instructions

7.special  instructions

nop

ssnop

sync

cache

另外在汇编程序中，常见一些伪指令，伪指令的目的是方便程序编写

The assembler translates pseudo instructions into real MIPS instructions

感悟：现在终于知道为啥计算机叫计算机了，本质上就是一系列的计算，操作码+操作数，load+store可将外部内存看做计算数据的来源和计算结果的仓库，另外数据还可以改变计算执行的序列，仅此而已。

引申：java源码首先被编译为class字节码，其实就是一些抽象机器指令序列，本质上还是操作码+操作数，只是这些抽象指令不是在具体机器上去执行的，而是在虚拟机上执行的，虚拟机取出一条抽象指令作为数据，拆分出操作码，操作码作为数据影响虚拟机的执行，根据操作码跳转到虚拟机对应的代码序列中，然后将操作数作为数据参与计算，这样就完成执行了一条抽象指令，然后循环取出下一条抽象指令执行，这就是java虚拟机的实现原理。

虚拟机程序：

loop（抽象指令=read（class文件））

{

    操作码=getop（抽象指令）；

    操作数=getdata（抽象指令）；

    switch（操作码）

     {

           case 加法：

            add 操作数; 

            break；

            case 减法：

            sub 操作数;

            break；

            ....

      }

}

参考文档：

《MIPS32® Architecture For Programmers Volume II: The MIPS32® Instruction Set》

《MIPS32-指令集-简表.pdf》

http://www.eecs.harvard.edu/~ellard/Courses/cs50-asm.pdf

http://www.cs.unibo.it/~solmi/te ... LanguageProgDoc.pdf

http://www.inf.uni-konstanz.de/d ... /download/rs-05.pdf

https://en.wikibooks.org/wiki/MIPS_Assembly/Instruction_Formats