Getting address of data variable in x86 AT&T Assembly

问题

Possible duplicate exist, but I couldnt figure out how to apply this or othere solutions to similar problems so here I am.

I am creating a function that returns and integer as a string in x86 AT&T Assembly.

I have this code to declare the variable resdes.

        .data
    .align 4
resdes: .long 12

resdes now points to a memory location followed by 11 other bytes free for me to use (I have understood this correctly?).

I want to load one digit at a time from the integer into the bytes one by one. this is my code:

ifd:
    movl        (%esp, %ecx), %eax  //This loads %eax with my int
    movl        resdes, %ecx     //This is incorrect and causes errors later
    inc         %ecx
    movl        $10, %ebx        //Division by 10 to basically do a modulo operation
    cdq

divloop:
    div         %ebx

    movb        %dl, (%ecx)      //This is where I move the digit into the memory
                                 //And here I get the ERROR because (%ecx) does 
                                 //not contain the proper address

    inc         %ecx             //And set the pointer to point to the next byte

    cmp         $0, %eax         //If there are noe more digits left we are finished
    je          divfinish1

    jmp         divloop          //I leave out alot of the code that I know 
                                 //work because it's not relevant

My problem is getting this actual address of resdes into the %ecxregister, the first line in the above code. As far as I know the line moves the contents of the resdes-address into %ecx, and this is not what I want.

回答1:

You want to load ECX with a constant which will be computed at "assemble time". So use '$' as prefix: movl $resdes, %ecx to get the offset of resdes as constant.

回答2:

OK, so I "solved" this on my own. I basically removed the memory-storage all together, and instead use the stack.

This does not answer my question about how to correctly get addresses, but it made my code work so I put it up here.

ifd:

    movl        (%esp, %ecx), %eax
    add         $4, %ecx
    movl        $10, %ebx
    pushl       $0

divloop:
    cdq
    div         %ebx

    add         $48, %edx
    pushl       %edx

    cmp         $0, %eax
    je          divfinish1

    jmp         divloop

divfinish1:
    movl        tempb, %ebx

divfinish2:
    popl        %edx

    cmp         $0, %edx
    je          divfinish3    //this is the end

    movb        %dl, (%ebx)
    inc         %ebx

    jmp         divfinish2

回答3:

The problem seems to be that you don't save the result of the division in the %ebx register after performing a division, and thus the division keeps on going forever, always returning the same modulo and division result -- you keep on dividing the same number. This, in turn, leads to a SIGSEGV because the result eventually overflows the 12 bytes reserved for the result. Here's a possible solution:

ifd:
    movl        resdes, %ecx   
    inc         %ecx
    movl        $10, %ebx       
    cdq

divloop:
    div         %ebx

    movb        %dl, (%ecx)   
    inc         %ecx    

    movl        %eax,%ebx         //Stores the new value in %ebx

    cmp         $0, %eax        
    je          divfinish1

    jmp         divloop          

resdes now points to a memory location followed by 11 other bytes free for me to use (I have understood this correctly?).

Yes.

来源：https://stackoverflow.com/questions/23383393/getting-address-of-data-variable-in-x86-att-assembly