- CPU 包括: 运算单元, 数据单元, 控制单元
    - 运算单元 不知道算哪些数据, 结果放哪
    - 数据单元 包括 CPU 内部缓存和寄存器, 暂时存放数据和结果
    - 控制单元 获取下一条指令, 指导运算单元取数据, 计算, 存放结果
- 进程包含代码段, 数据段等, 以下为 CPU 执行过程:
    - 控制单元 通过指令指针寄存器(IP), 取下一条指令, 放入指令寄存器中
        - 指令包括操作和目标数据
    - 数据单元 根据控制单元的指令, 从数据段读数据到数据寄存器中
    - 运算单元 开始计算, 结果暂时存放到数据寄存器
- 两个寄存器, 存当前进程代码段和数据段起始地址, 在进程间切换
- 总线包含两类数据: 地址总线和数据总线
---
- x86 开放, 统一, 兼容
- 数据单元 包含 8个 16位通用寄存器, 可分为 2个 8位使用
- 控制单元 包含 IP(指令指针寄存器) 以及 4个段寄存器 CS DS SS ES
    - IP 存放指令偏移量
    - 数据偏移量存放在通用寄存器中
    - `段地址<<4 + 偏移量` 得到地址
---
- 32 位处理器
- 通用寄存器 从 8个 16位拓展为 8个 32位, 保留 16位和 8位使用方式
- IP 从 16位扩展为 32位, 保持兼容
- 段寄存器仍为 16位, 由段描述符(表格, 缓存到 CPU 中)存储段的起始地址, 由段寄存器选择其中一项
    - 保证段地址灵活性与兼容性
---
- 16位为实模式, 32位为保护模式
- 刚开机为实模式, 需要更多内存切换到保护模式

老师讲地太精彩了 点赞。由于基础弱，还有几个问题希望老师万忙中答疑一下，谢谢：

问题1：程序编译成二进制代码的时候，包含有指令起始地址吗？若包含那么后续每一行指令的涉及到的指令地址是计算出来的？ 或者说加载进程的程序的时候才会确定起始地址？很好奇这个指令的指针寄存器里的值是什么时候、怎么放进去的？

问题2： CPU两个寄存器处理保存当前进程代码段的起始地址，已经数据起始地址。切换进程时会将这两个寄存器里的值一并切换，那么同一个进程出现多线程的时候了？

问题3：数据总线拿数据的时候没有限制大小吗，若数据很大，数据单元里的数据寄存器放不下怎么办的？

已经有童鞋回复，再补充点
MOV: load value. this instruction name is misnomer, resulting in some confusion (data is not movedbut copied), in other architectures the same instructions is usually named “LOAD” and/or “STORE”or something like that.One important thing: if you set the low 16-bit part of a 32-bit register in 32-bit mode, the high 16bits remains as they were. But if you modify the low 32-bit part of the register in 64-bit mode, thehigh 32 bits of the register will be cleared.Supposedly, it was done to simplify porting code to x86-64.

CALL:call another function:PUSH address_after_CALL_instruction; JMP label.
JMP: jump to another address. The opcode has ajump offset.

INT(M):INT x is analogous to PUSHF; CALL dword ptr [x*4]in 16-bit environment. It was widely used in MS-DOS, functioning as a syscall vector. The registers AX/BX/CX/DX/SI/DI were filled with the arguments and then the flow jumped to the address in the Interrupt Vector Table (located at thebeginning of the address space). It was popular because INT has a short opcode (2 bytes) and the program which needs some MS-DOS services is not bother to determine the address of the service’sentry point. The interrupt handler returns the control flow to caller using the IRET instruction.The most busy MS-DOS interrupt number was 0x21, serving a huge part of itsAPI. In the post-MS-DOS era, this instruction was still used as syscall both in Linux and Windows (6.3 onpage 750), but was later replaced by the SYSENTER or SYSCALL instructions.

RET: return from subroutine:POP tmp; JMP tmp.
    In fact, RET is an assembly language macro, in Windows and *NIX environment it is translated into RETN (“return near”) or, in MS-DOS times, where the memory was addressed differently, into RETF (“return far”).
    RET can have an operand. Then it works like this:
    POP tmp; ADD ESP op1; JMP tmp.RETwith an operand usually ends functions in the stdcall calling convention.

补充2
ADD: add two values.    
OR: logical “or”.
XOR op1, op2:is in fact just “eXclusive OR”,but the compilers often use it instead of MOV EAX, 0—again becauseit is a slightly shorter opcode (2 bytes for XOR against 5 for MOV).

SHL/SHR: shift value left/right.
   Bit shifts in C/C++ are implemented using≪and≫operators. The x86 ISA has the SHL (SHift Left) and SHR (SHift Right) instructions for this. Shift instructions are often used in division and multiplications bypowers of two:2n(e.g., 1, 2, 4, 8, etc.): Shifting operations are also so important because they are often used for specific bit isolation or forconstructing a value of several scattered bits.

PUSH: push a value into the stack:ESP=ESP-4 (or 8); SS:[ESP]=value.
POP: get a value from the stack:value=SS:[ESP]; ESP=ESP+4 (or 8).
    The most frequently used stack access instructions are PUSH and POP(in both x86 and ARM Thumb-mode).PUSH subtracts from ESP/RSP/SP4 in 32-bit mode (or 8 in 64-bit mode) and then writes the contents of its sole operand to the memory address pointed by ESP/RSP/SP.
    POP is the reverse operation: retrieve the data from the memory location that SP points to, load it into the instruction operand (often a register) and then add 4 (or 8) to the stack pointer.

INC: increment. Unlike other arithmetic instructions,INC doesn’t modify CF flag.
DEC: decrement. Unlike other arithmetic instructions,DEC doesn’t modify CF flag.
SUB: subtract values. A frequently occurring pattern is SUB reg.
CMP: compare values and set flags, the same as SUB but without writing the result.