CODEWITHSUNDEEP
assemblystack
Akrilla
Akrilla

Reputation: 39

Issues with pushing variables directly on to stack

My current code works, and is as follows:

movzx  ecx, var1
lea    eax, var2

push eax             
push ecx             

call func1

//...

and func1 is:

push ebp           
mov ebp, esp       

mov ecx, [ebp + 8] 
mov eax, [ebp + 12]

push edi           
push ecx           

not byte ptr[eax] 

//...

However, I wish to push the address of var2 directly, along with simply pushing the value of var1 like so, but I'm having trouble figuring out how:

push var1            
push var2 //address of)

call func1

Any and all help would be greatly appreciated.

Upvotes: 0

Views: 102

Answers (3)

Jose Manuel Abarca Rodríguez
Jose Manuel Abarca Rodríguez

Reputation: 10391

With compiler EMU8086 you can do it. Next is your code with some changes (I'm using 16 bit registers, that's why BP+4 and BP+6) :

.stack 100h
.data  
var1 dw 25              ;VAR1 = 25.
var2 dw 800             ;VAR2 = 800.
.code
  mov  ax, @data
  mov  ds, ax

  push var1             ;PUSH 25.
  push offset var2      ;PUSH ADDRESS OF 800.
  call func1

  mov  ax, 4c00h 
  int  21h              ;FINISH.

func1 proc
  push bp
  mov  bp, sp
  mov  si, [ bp + 4 ]   ;POP ADDRESS OF 800.
  mov  ax, [ bp + 6 ]   ;POP 25.  

  not  [ byte ptr si ]  ;CX CANNOT BE USED AS POINTER.

  pop  bp
  ret
func1 endp

The address of 800 is extracted into SI instead of CX because CX cannot be used to point to the address.

Remember that things in stack are stored upside down : 25 is pushed first then the address of 800, that's why the address is in position BP+4 and 25 in BP+6.

Upvotes: 0

zetavolt
zetavolt

Reputation: 3227

There is no way to get the address of a register value.

However, there are two things you can do:

If the value is in the psuedo-register var2 and is created through the operation of your program (i.e, isn't a syscall return register, etc.), you could simply do the following instead of mov var2, [avar2]

mov var2, avar2

If that doesn't suit you, you can push var2 to the stack and get the address from esp, e.g 

push var2
lea avar2, [esp-4]

Upvotes: 1

zx485
zx485

Reputation: 29052

Well, if the byte variable var1 has a DWORD space(4-bytes) in memory and is zero extended, you could just PUSH it. The LEA of var2' could be replaced if the address of var2 is at a constant address in the .data segment. Like this:

push var1
push offset var2            
call func1

Under these conditions your requirements would be easily satisfiable.

Upvotes: 2

Related Questions