Passing an Integer from C to MASM, Returning a Double (39520)

MORE INFORMATION

Below is the C program that calls the MASM routine. It should be compiled for large-memory model and either 8087 or emulator-math library.

Also shown is the MASM routine. The main piece to this program is the FILD instruction that transforms the integer to a floating-point real, then pushes the value on the co-processor stack. The other point is to obtain the segment and offset of the __fac variable. A value requiring more than 4 bytes cannot be stored in the DX and AX registers for return, so a double is returned by storing its address in DX:AX.

Sample Code

/* Compile options needed: /AL /FPi87 (or /FPi)
*/ 

#include <stdio.h>

extern double abc( int x );

void main()
{
   int x;
   double y;

   x = 7;
   y = abc( x );
   printf( "%The value of y is %lf.\n", y );
}

----------------------------------------------------------------------

; Assemble options needed: /Mx

        .MODEL LARGE,C    ; so Masm will use C naming conventions
        .DATA
        .CODE
EXTRN _fac:QWORD          ; __fac is the global variable used by C
                          ; for storing floating point accumulations
PUBLIC  abc
abc     PROC    FAR
        push    bp
        mov     bp,sp
        fild    WORD PTR [bp+6]
        mov     dx,SEG _fac
        mov     es,dx
        fstp    QWORD PTR es:_fac  ; this pops the value off of
        mov     ax,OFFSET _fac     ; the co-processor and puts it
        fwait                      ; in __fac
        mov     sp,bp
        pop     bp
        ret

abc     ENDP
        END
				

Program Output:

The value of y is 7.000000.