NASM部件号到基本的转换

Question

I have what seems to me to be a fairly straightforward project. Ask user for a number between 0-255 and base between 2-9. And output in that base. I plan on simply doing a usual division algorithm and getting the remainders, pushing onto the stack, and poping back off to get the reverse order to output back to the user. However after inserting a few print debugging statements I am getting very strange output for the remainder

; Run using nasm -f elf -g -F stabs proj4.asm ; gcc -o proj4 proj4.o -m32 ; To execute type proj4 ; Run using nasm -f elf -g -F stabs proj4.asm ; gcc -o proj4 proj4.o -m32 ; To execute type proj4 %macro SAVE_REGS 0 push eax push ecx push edx %endmacro %macro RESTORE_REGS 0 pop edx pop ecx pop eax %endmacro %macro CALL_PUTS 1 push %1 call puts add esp, 4 %endmacro %macro CALL_SCANF 2 push %1 push %2 call scanf add esp, 8 %endmacro %macro CALL_PRINTF1 1 push %1 ;The address of the string to print call printf add esp, 4 %endmacro %macro CALL_PRINTF2 2 push %1 ;The formatted string with a %char place holder push %2 ;The item to place into the place holder call printf add esp, 8 %endmacro SECTION .data prmptNumMsg: db "Enter a number between 0 and 255: ", 0 prmptBaseMsg: db "Enter a base between 2 and 9: ", 0 remShow: db 'The the remainder is: %d', 10, 0 numShow: db 'The number is %d', 10, 0 baseShow: db 'The base is %d', 10, 0 printed: db 'Looped in division method', 10, 0 poped: db 'Looped in pop method',10, 0 ansShow: db '8d', 10, 0
numFormat db '%d', 0 stringFormat db '%s', 0 prmptNumMsg: db "Enter a number between 0 and 255: ", 0 prmptBaseMsg: db "Enter a base between 2 and 9: ", 0 remShow: db 'The the remainder is: %d', 10, 0 numShow: db 'The number is %d', 10, 0 baseShow: db 'The base is %d', 10, 0 printed: db 'Looped in division method', 10, 0 poped: db 'Looped in pop method',10, 0 ansShow: db '8d', 10, 0
numFormat db '%d', 0 stringFormat db '%s', 0 prmptNumMsg: db "Enter a number between 0 and 255: ", 0 prmptBaseMsg: db "Enter a base between 2 and 9: ", 0 remShow: db 'The the remainder is: %d', 10, 0 numShow: db 'The number is %d', 10, 0 baseShow: db 'The base is %d', 10, 0 printed: db 'Looped in division method', 10, 0 poped: db 'Looped in pop method',10, 0 ansShow: db '8d', 10, 0
numFormat db '%d', 0 stringFormat db '%s', 0 SECTION .bss numVal resd 1 baseVal resd 1 ans resd 9 i resd 1 n resd 1 j resd 1

SECTION .text global main extern puts extern scanf extern printf main: push ebp ; Set up stack frame for debugger mov ebp, esp push ebx push esi push edi ;Everything before this is boilerplate push ebp ; Set up stack frame for debugger mov ebp, esp push ebx push esi push edi ;Everything before this is boilerplate
getInt: CALL_PRINTF1 prmptNumMsg ;push numVal ;push numFormat ;call scanf ;add esp, 8 CALL_SCANF numVal, numFormat
mov eax, dword[numVal] mov ebx, dword 0 ;check below 0 cmp eax, ebx jb getInt mov eax, dword[numVal] mov ebx, dword 255 ;check above 255 cmp eax , ebx ja getInt ;VALID INTEGER PAST THIS POINT
getBase: CALL_PRINTF1 prmptBaseMsg
CALL_SCANF baseVal, numFormat getBase: CALL_PRINTF1 prmptBaseMsg
CALL_SCANF baseVal, numFormat mov eax, dword[baseVal] mov ebx, dword 0 cmp eax, ebx jb getBase mov eax, dword[baseVal] mov ebx, dword 9 cmp eax, ebx ja getBase ;END GETBASE ;VALID BASE NUMBER PAST THIS POINT mov eax, dword[numVal] mov [n], eax ;set n to the current number value CALL_PRINTF2 eax, numShow mov eax, dword 0 mov [i], eax mov eax, dword[baseVal] CALL_PRINTF2 eax, baseShow doDivision: ;CALL_PRINTF1 printed xor edx, edx mov eax, dword[n] mov ebx, dword[baseVal] div ebx ;edx = remainder eax = quotient mov [n], eax ;save quotient
CALL_PRINTF2 eax, numShow CALL_PRINTF2 edx, remShow ;END GETBASE ;VALID BASE NUMBER PAST THIS POINT mov eax, dword[numVal] mov [n], eax ;set n to the current number value CALL_PRINTF2 eax, numShow mov eax, dword 0 mov [i], eax mov eax, dword[baseVal] CALL_PRINTF2 eax, baseShow doDivision: ;CALL_PRINTF1 printed xor edx, edx mov eax, dword[n] mov ebx, dword[baseVal] div ebx ;edx = remainder eax = quotient mov [n], eax ;save quotient
CALL_PRINTF2 eax, numShow CALL_PRINTF2 edx, remShow
;push edx ;save remainder on stack to pop in reverse order later ;mov [n], eax ;move quotient to eax mov ebx, dword[i] inc ebx mov [i], ebx ;i++ ;mov eax, [i] mov ecx, dword 8 cmp ebx, ecx jb doDivision ;END DO DIVISION

end: ;Everything after this is boilerplate pop edi pop esi pop ebx mov esp, ebp pop ebp ret

When the program is run my output is as follows Enter a number between 0 and 255: 105 Enter a base between 2 and 9: 4 The number is 105 The base is 4 The number is 26 The the remainder is: 13144896 The number is 6 The the remainder is: 13144896 The number is 1 The the remainder is: 13144896 The number is 0 The the remainder is: 13144896 The number is 0 The the remainder is: 13144896 The number is 0 The the remainder is: 13144896 The number is 0 The the remainder is: 13144896 The number is 0 The the remainder is: 13144896 I want it to loop 8 times which is correct, the quotient of each div operation is correct, however I'm getting crazy numbers for the remainder which would be held in edx. I don't understand whats causing this.

Answer 1

When you call a routine (for example: printf), there are rules that govern how this works. For example, do you put the parameters on the stack? Or pass them in registers? Do you push parameters from left to right? Or right to left? Does the caller pop the parameters off the stack? Or does the callee? Where will return values be located?

Most importantly for your question, does the callee need to make sure that all registers will have the same value when it returns? Or can it overwrite some of them?

The answer to these questions are called the "calling conventions" (or sometimes ABI). And there isn't just one answer. For example cdecl, pascal, and fastcall are all common calling conventions on the x86, and they all answer these questions in slightly different ways.

I believe you will find that printf is cdecl. With that in mind, you might check out http://en.wikipedia.org/wiki/X86_calling_conventions#cdecl . That should help you understand what is happening to edx in this example.