Define the Program
把文件中的所有字符都变成用十六进制书写的格式。
Starting with Pseudo-Code
1、最初的伪代码设计如下：
| Read a char from the input file convert the character to hex string Write the hex string to the output file Repeat until done. |
2、第一次改进，增添一些细节：
| Read a character from standard input (stdin) Apart the 8-bit character into two nybbles, change all of them to hex character, and then put them together Witer the hex string to standard output (stdout) Repeat until done. Exit the program by calling sys_exit. |
3、EOF
| Read a character from standard input (stdin) Test if we have reached End Of File (EOF) If we have reached EOF, we’re done, so jump to exit Apart the 8-bit character into two nybbles, change all of them to hex character, and then put them together Witer the hex string to standard output (stdout) Go back and read another character. Exit the program by calling sys_exit. |
4、加上标签
| Read: Set up registers for the sys_read kernel call. Call sys_read to read from stdin. Test for EOF. If we are at EOF, jump to Exit Apart the 8-bit character into two nybbles Change all of them into Hex character Put the two Hex char together Write: Set up registers for the sys_write kernel call. Call sys_write to write stdout. Jump back to Read and get another character Exit: Set up registers for termianting the program via sys_exit. Call sys_exit. |
5、进一步细化如何把nybble转换成Hex character
| Set a looking up table for mapping Decimal to Hex: Digits “0123456789ABCDEF” Set a output table for putting two Hex char together: HexStr “ 00” Read: Set up registers for the sys_read kernel call. Call sys_read to read from stdin. Test for EOF. If we are at EOF, jump to Exit Apart the 8-bit character into two nybbles For each nybble, the decimal value of this nybble plus the address of Digits, this byte in the Digits is the Hex character Put the Least Significant Hex in [HexStr+2]. Put the Most Significant Hex in [HexStr+1]. Write: Set up registers for the sys_write kernel call. Call sys_write to write stdout. Jump back to Read and get another character Exit: Set up registers for termianting the program via sys_exit. Call sys_exit. |
这一步细化之后的代码，完全按照这一步的细化方案来写的：
; Executable name : myhexdump1 ; Version : 1.0 ; Created data : 3/22/2011 ; Last update : 3/22/2001 ; Author : Eric Wang ; Description : A simple program in assembly for Linux, using NASM 2.05, ; Demonstrating the conversion of binary values to hexadecial strings. ; It acts as a very simple hex dump utility for files. ; ; Run it this way: ; myhexdump1 < (input file) ; ; Build using these commands: ; nasm -f elf -g -F dwarf myhexdump1.asm ; ld -o myhexdump1 myhexdump1.o ; section .bss ; Section containing uninitialized data Buff: resb 1 ; Text buffer itself section .data ; Section containing initialized data Digits: db "0123456789ABCDEF" HexStr: db " 00" HEXLEN: equ $-HexStr section .text ; Section containing code global _start ; Linker needs this to find the entry point! _start: nop ; This no-op keeps gdb happy... ; Read from stdin Read: mov eax,3 ; Specify sys_read call mov ebx,0 ; Specify File Descriptor 0: Standard Input mov ecx,Buff ; Pass address of the buffer to read to mov edx,1 ; Tell sys_read to read one char from stdin int 80h ; Call sys_read cmp eax,0 ; If eax==0, sys_read reached EOF on stdin je Done ; Jump If Equal (to 0, from compare) ; Apart the 8-bit char into two nybbles xor eax,eax ; Clear eax to 0 mov al,byte [Buff] ; Put the char into eax register, for low nybble mov ebx,eax ; For high nybble and al,0fh ; mask the high 4-bits in this character mov al,byte [Digits+eax] ; Look up table for finding the Hex char mov byte [HexStr+2],al ; Put the Hex char into HexStr at least significant place shr bl,4 ; Shift the high 4-bits to the low 4-bits mov bl,byte [Digits+ebx] ; Loop up table for finding the Hex char mov byte [HexStr+1],bl ; Put the Hex char into HexStr at most significant place Write: mov eax,4 ; Specify sys_write call mov ebx,1 ; Specify File Descriptor 1: Standard Output mov ecx,HexStr ; Pass address of the character to write mov edx,HEXLEN ; Pass number of characters to write int 80h ; Call sys_write jmp Read ; Jump back to Read and get another character ; All done! Done: mov eax,1 ; Specify sys_exit call mov ebx,0 ; Return code of zero to Linux int 80h ; Call sys_exit
6、Buffered file scan
| Set a 16 bytes buffer. Set a looking up table for mapping Decimal to Hex: Digits “0123456789ABCDEF” Set a output table for putting two Hex char together: HexStr “ 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00”,10 Read: Set up registers for the sys_read kernel call. Call sys_read to read a buffer full of characters from stdin. Test for EOF. If we are at EOF, jump to Exit Set up registers as a pointer to scan the buffer Scan: Apart the 8-bit character at buffer pointer into two nybbles For each nybble, the decimal value of this nybble plus the address of Digits, this byte in the Digits is the Hex character Put the Least Significant Hex in [HexStr+bufferPointer*3+2]. Put the Most Significant Hex in [HexStr+bufferPointer*3+1]. Increment buffer pointer. If we still have characters in the buffer, jump to Scan. Write: Set up registers for the sys_write kernel call. Call sys_write to write the processed HexStr to stdout. Jump back to Read and get another buffer full of characters. Exit: Set up registers for termianting the program via sys_exit. Call sys_exit. |
7、Start talking specifics: which register do what:
| Set a 16 bytes buffer. Set a looking up table for mapping Decimal to Hex: Digits: db “0123456789ABCDEF” Set a output table for putting two Hex char together: HexStr: db “ 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00”,10 Read: Set up registers for the sys_read kernel call. Call sys_read to read a buffer full of characters from stdin. Store the number of characters read in esi Test for EOF (eax==0). If we are at EOF, jump to Exit. Set ebp to 0 to point to the first character address in buffer Scan: Move the byte at [Buff+ebp] to al Copy the value of eax to ebx (Then the low 4-bits in al and the high 4-bits in bl are two nybbles) Mask the al with 0Fh, plus this value with the address of Digits, this byte in the Digits is the Least Significant Hex character Shift right for 4 bits of the bl, plus this value with the Digits’ address, this byte in Digits is the Most Significant Hex character Put the Least Significant Hex in [HexStr+bufferPointer*3+2]. Put the Most Significant Hex in [HexStr+bufferPointer*3+1]. Increment buffer pointer. If ebp < esi, jump to Scan. Write: Set up registers for the sys_write kernel call. Call sys_write to write the processed HexStr to stdout. Jump back to Read and get another buffer full of characters. Exit: Set up registers for termianting the program via sys_exit. Call sys_exit. |
代码，粗体部分是修改的：
; Executable name : myhexdump2 ; Version : 2.0 ; Created data : 3/22/2011 ; Last update : 3/22/2001 ; Author : Eric Wang ; Description : A simple program in assembly for Linux, using NASM 2.05, ; Demonstrating the conversion of binary values to hexadecial strings. ; It acts as a very simple hex dump utility for files with Buffer Scanning. ; ; Run it this way: ; myhexdump2 < (input file) ; ; Build using these commands: ; nasm -f elf -g -F dwarf myhexdump2.asm ; ld -o myhexdump2 myhexdump2.o ; section .bss ; Section containing uninitialized data BUFFLEN: equ 16 ; Our buffer lenth is 16 Buff: resb BUFFLEN ; Text buffer itself, resb may means "reset to byte" section .data ; Section containing initialized data Digits: db "0123456789ABCDEF" HexStr: db " 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00",10 HEXLEN: equ $-HexStr section .text ; Section containing code global _start ; Linker needs this to find the entry point! _start: nop ; This no-op keeps gdb happy... ; Read from stdin Read: mov eax,3 ; Specify sys_read call mov ebx,0 ; Specify File Descriptor 0: Standard Input mov ecx,Buff ; Pass address of the buffer to read to mov edx,BUFFLEN ; Tell sys_read to read BUFFLEN characters from stdin int 80h ; Call sys_read mov esi,eax ; Store the number of characters read in esi cmp eax,0 ; If eax==0, sys_read reached EOF on stdin je Done ; Jump If Equal (to 0, from compare) xor ebp,ebp ; set ebp to 0 to point to the first char in buffer ; Apart the 8-bit char into two nybbles Scan: xor eax,eax ; Clear eax to 0 mov al,byte [Buff+ebp] ; Put the char into eax register, for low nybble mov ebx,eax ; For high nybble and al,0fh ; mask the high 4-bits in this character mov al,byte [Digits+eax] ; Look up table for finding the Hex char mov ecx,ebp ; The first step for X3. ecx = ebp shl ecx,1 ; ecx = ecx * 2, so ecx = ebp * 2 add ecx,ebp ; ecx = ebp + ecx, so ecx = ebp * 3 mov byte [HexStr+ecx+2],al ; Put the Hex char into HexStr at least significant place shr bl,4 ; Shift the high 4-bits to the low 4-bits mov bl,byte [Digits+ebx] ; Loop up table for finding the Hex char mov ecx,ebp ; The first step for X3. ecx = ebp shl ecx,1 ; ecx = ecx * 2, so ecx = ebp * 2 add ecx,ebp ; ecx = ebp + ecx, so ecx = ebp * 3 mov byte [HexStr+ecx+1],bl ; Put the Hex char into HexStr at most significant place inc ebp ; increse buffer pointer cmp ebp,esi ; Compare ebp and esi jb Scan ; Jump if Below (if ebp < esi, jump to Scan) Write: mov eax,4 ; Specify sys_write call mov ebx,1 ; Specify File Descriptor 1: Standard Output mov ecx,HexStr ; Pass address of the character to write mov edx,HEXLEN ; Pass number of characters to write int 80h ; Call sys_write jmp Read ; Jump back to Read and get another character ; All done! Done: mov eax,1 ; Specify sys_exit call mov ebx,0 ; Return code of zero to Linux int 80h ; Call sys_exit
最后，这个程序还有一个问题：如果最后一次从文件读入的数据不足16个byte，那么HexStr的最后结果就会和上一次的尾部相同。更改的部分如下：
Write: mov eax,4 ; Specify sys_write call mov ebx,1 ; Specify File Descriptor 1: Standard Output mov ecx,HexStr ; Pass address of the character to write cmp esi,BUFFLEN ; For the last read. je Normal ; If esi = HEXLEN, this is not the last read, and the buffer is full. mov edx,esi ; The real characters in the buffer shl edx,1 ; edx = esi * 2 add edx,esi ; edx = esi * 3 mov byte [HexStr+edx],10 ; New line inc edx ; edx plus one jmp Call ; Do not mov edx,HEXLEN Normal: mov edx,HEXLEN ; Pass number of characters to write Call: int 80h ; Call sys_write jmp Read ; Jump back to Read and get another character