Author Topic: mixing nasm and pascal  (Read 8234 times)

nobody

  • Guest
mixing nasm and pascal
« on: September 02, 2009, 07:12:08 AM »
Hello

I have reading nasm-doc looking for information about mixing pascal and nasm assembler. Well what i found did not satisfied me. And that's the question how can i write an nasm procedure to be externally called by a pascal routine?. Let's say:

; Nasm-routine (PasAsN1.asm)
[section .text]
  global WriteChar
  WriteChar:
    MOV    AH, 02h
    MOV    DL, BYTE [BP+6]
    INT    21h
  retf   2
; nasmw -f obj PasAsN1.asm -o PasAsN1.obj


Program MixPascalNasm; (* PasAsm.pas *)
uses crt;
PROCEDURE WriteChar(Car: Char); External;
{$L PasAsN1.obj}
begin
  WriteChar('A');
  ReadKey;
end.


Thank you very much

nobody

  • Guest
Re: mixing nasm and pascal
« Reply #1 on: September 02, 2009, 07:37:19 AM »
Well... I have no experience at all with Pascal, but I think I understand how it would work... I think you're quite close (assuming Pascal doesn't complain about this), but you've missed setting up your "stack frame":


 WriteChar:

push bp
mov bp, sp
; or "enter 0, 0"

MOV    AH, 02h
    MOV    DL, BYTE [BP+6]
    INT    21h

mov sp, bp
pop bp
; or "leave"

retf   2

See if that helps...

Best,
Frank

nobody

  • Guest
Re: mixing nasm and pascal
« Reply #2 on: September 03, 2009, 07:11:56 AM »
Thank you for your replay

I don't believe using SP it will work. BP-SP are used to define internal variables if they are needed. In this case i don't need them.

The problem is that with these codes, nasm gives an obj file, but turbo pascal says "Invalid PUBLIC definition (PASASN1.OBJ)."

Offline Frank Kotler

  • NASM Developer
  • Hero Member
  • *****
  • Posts: 2667
  • Country: us
Re: mixing nasm and pascal
« Reply #3 on: September 03, 2009, 10:31:02 AM »
You expect "mov dl, [bp + 6]" to work by magic? Well, once you get an executable, you'll be able to find out easily. I don't know what to do about the error message. "PUBLIC" is what Nasm calls "global", so apparently it's "global WriteChar" that it's complaining about - but that looks okay to me. The only thing I can think of to try is to replace "[section .text]" with "section code class=code". Nasm's "-f obj" (OMF) output format doesn't know any "special names" like ".text" (most other output formats know what to do with ".text"), so specifying "class=code" might help. Since it's "far", the name shouldn't matter (?). Hmmm, I guess that's not right. Looking at the Friendly Nasm Manual:

# Procedures and functions must be in a segment whose name is either CODE, CSEG, or something ending in _TEXT.

So maybe make it "section CODE class=code"... maybe if you use one of those names, you wouldn't need "class=code"... Fool around in that area and see if it changes your results any.

Best,
Frank