Added new div() function written by Greg King

git-svn-id: svn://svn.cc65.org/cc65/trunk@1469 b7a2c559-68d2-44c3-8de9-860c34a00d81

libsrc/runtime/Makefile

@@ -54,6 +54,7 @@ OBJS = 	add.o		\
 	decsp7.o	\
 	decsp8.o	\
        	div.o		\
+        divt.o          \
        	enter.o		\
        	eq.o		\
        	ge.o		\

libsrc/runtime/divt.s

@@ -0,0 +1,43 @@
+; divt.s
+;
+; 2002-10-22, Greg King
+;
+; This signed-division function returns both the quotient and the remainder,
+; in this structure:
+;
+; typedef struct {
+;     int rem, quot;
+; } div_t;
+;
+; div_t __fastcall__ div (int numer, int denom);
+;
+; Both sides of an assignment-expression must be cast to (long)
+; because cc65 functions can't return structures -- yet.  Example:
+;
+; #include <stdlib.h>
+; div_t answer;
+;
+; (long)answer = (long)div(-41, +3);
+; printf("The quotient is %d, and the remainder is %d.\n",
+;	answer.quot, answer.rem);
+
+	.export		_div
+
+	.import		tosdivax, negax
+	.importzp	sreg, ptr1, tmp1
+
+_div:	jsr	tosdivax	; Division-operator does most of the work
+        sta     sreg		; Quotient is in sreg
+        stx     sreg+1
+        lda     ptr1  		; Unsigned remainder is in ptr1
+        ldx     ptr1+1
+
+; Adjust the sign of the remainder.
+; It must be the same as the sign of the dividend.
+;
+	bit	tmp1		; Load high-byte of left argument
+	bpl	Pos		; Jump if sign-of-result is positive
+	jmp	negax		; Result is negative, adjust the sign
+
+Pos:	rts
+