MOV EDI,EDI指令的解释（整理）

源文：http://blog.csdn.net/qq276592716/article/details/7007396

XP系统程序中开头的MOV  EDI,EDI指令的解释：

在VS .NET 2003的VC7\INCLUDE目录中的listing.inc文件中定义了１到７个字节的无破坏性NOP操作的宏

MOV  EDI,EDI　就是两个字节的NOP
在程序中与NOP指令的意义相同。

为什么要用MOV  EDI,EDI 而不用两个NOP?
我的理解是：
用两个NOP指令耗费的CPU时钟周期要比用MOV  EDI,EDI指令要长，为了提高效率，就采用了MOV  EDI,EDI


listing.inc文件的内容

;; LISTING.INC
;;
;; This file contains assembler macros and is included by the files created
;; with the -FA compiler switch to be assembled by MASM (Microsoft Macro
;; Assembler).
;;
;; Copyright (c) 1993, Microsoft Corporation. All rights reserved.

;; non destructive nops
npad macro size
if size eq 1
  nop
else
 if size eq 2
   mov edi, edi
 else
  if size eq 3
    ; lea ecx, [ecx+00]
    DB 8DH, 49H, 00H
  else
   if size eq 4
     ; lea esp, [esp+00]
     DB 8DH, 64H, 24H, 00H
   else
    if size eq 5
      add eax, DWORD PTR 0
    else
     if size eq 6
       ; lea ebx, [ebx+00000000]
       DB 8DH, 9BH, 00H, 00H, 00H, 00H
     else
      if size eq 7
  ; lea esp, [esp+00000000]
  DB 8DH, 0A4H, 24H, 00H, 00H, 00H, 00H 
      else
  %out error: unsupported npad size
  .err
      endif
     endif
    endif
   endif
  endif
 endif
endif
endm

;; destructive nops
dpad macro size, reg
if size eq 1
  inc reg
else
  %out error: unsupported dpad size
  .err
endif
endm

 

Ishai的解释~

The PUSH  EBP and MOV EBP, ESP instructions are standard frame establishment, but what is the purpose of the MOV EDI,EDI instruction?  Seems like a 2-byte NOP instruction.

 

MOV EDI,EDI is indeed a 2-byte no-op that is there to enable hot-patching.   It enables the application of a hot-fix to a function without a need for a reboot, or even a restart of a running application.   Instead, at runtime, the 2-byte NOP is replaced by a short jump to a long jump instruction that jumps to the hot-fix function.   A 2-byte instruction is required so that when patching the instruction pointer will not point in a middle of an instruction.

Detours change a binary file offline. Hot patching is done on a running executable and they want to guarantee that the instruction pointer does not point in the middle of the patched area. 

Using the Detours method on a live process would require suspending threads and making sure no thread instruction pointer is pointing at the second, third, forth, or fifth byte of a function that is being Detoured and handling the case that it does. 

A Detour will also put limitation on the code generation (i.e. never jump to instructions in bytes 2-5). 

Seems to be possible but more complicated than placing a gap between functions and ensuring a 2-byte first instruction. 

个人认同Ishai的说法~方便热补丁~~