/*-------------------------------------------------------------------------
_mullong.c - routine for multiplication of 32 bit (unsigned) long
Copyright (C) 1999, Sandeep Dutta . sandeep.dutta@usa.net
Copyright (C) 1999, Jean Louis VERN jlvern@writeme.com
This library is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 2, or (at your option) any
later version.
This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this library; see the file COPYING. If not, write to the
Free Software Foundation, 51 Franklin Street, Fifth Floor, Boston,
MA 02110-1301, USA.
As a special exception, if you link this library with other files,
some of which are compiled with SDCC, to produce an executable,
this library does not by itself cause the resulting executable to
be covered by the GNU General Public License. This exception does
not however invalidate any other reasons why the executable file
might be covered by the GNU General Public License.
-------------------------------------------------------------------------*/
/* Signed and unsigned multiplication are the same - as long as the output
has the same precision as the input.
Assembler-functions are provided for:
mcs51 small
mcs51 small stack-auto
*/
#if !defined(__SDCC_USE_XSTACK) && !defined(_SDCC_NO_ASM_LIB_FUNCS)
# if defined(__SDCC_mcs51)
# if defined(__SDCC_MODEL_SMALL)
# if defined(__SDCC_STACK_AUTO) && !defined(__SDCC_PARMS_IN_BANK1)
# define _MULLONG_ASM_SMALL_AUTO
# else
# define _MULLONG_ASM_SMALL
# endif
# elif defined(__SDCC_MODEL_LARGE)
# if !defined(__SDCC_STACK_AUTO)
# define _MULLONG_ASM_LARGE
# endif
# endif
# endif
#endif
#if defined(_MULLONG_ASM_SMALL) || defined(_MULLONG_ASM_SMALL_AUTO)
void
_mullong_dummy (void) __naked
{
__asm
__mullong:
.globl __mullong
; the result c will be stored in r4...r7
#define c0 r4
#define c1 r5
#define c2 r6
#define c3 r7
#define a0 dpl
#define a1 dph
#define a2 r2
#define a3 r3
; c0 a0 * b0
; c1 a1 * b0 + a0 * b1
; c2 a2 * b0 + a1 * b1 + a0 * b2
; c3 a3 * b0 + a2 * b1 + a1 * b2 + a0 * b3
#if !defined(__SDCC_STACK_AUTO) || defined(__SDCC_PARMS_IN_BANK1)
#if defined(__SDCC_PARMS_IN_BANK1)
#define b0 (b1_0)
#define b1 (b1_1)
#define b2 (b1_2)
#define b3 (b1_3)
#else
#if defined(__SDCC_NOOVERLAY)
.area DSEG (DATA)
#else
.area OSEG (OVR,DATA)
#endif
__mullong_PARM_2:
.globl __mullong_PARM_2
.ds 4
b0 = __mullong_PARM_2
b1 = (__mullong_PARM_2+1)
b2 = (__mullong_PARM_2+2)
b3 = (__mullong_PARM_2+3)
#endif
.area CSEG (CODE)
; parameter a comes in a, b, dph, dpl
mov r2,b ; save parameter a
mov r3,a
; Byte 0
mov a,a0
mov b,b0
mul ab ; a0 * b0
mov c0,a
mov c1,b
; Byte 1
mov a,a1
mov b,b0
mul ab ; a1 * b0
add a,c1
mov c1,a
clr a
addc a,b
mov c2,a
mov a,a0
mov b,b1
mul ab ; a0 * b1
add a,c1
mov c1,a
mov a,b
addc a,c2
mov c2,a
clr a
rlc a
mov c3,a
; Byte 2
mov a,a2
mov b,b0
mul ab ; a2 * b0
add a,c2
mov c2,a
mov a,b
addc a,c3
mov c3,a
mov a,a1
mov b,b1
mul ab ; a1 * b1
add a,c2
mov c2,a
mov a,b
addc a,c3
mov c3,a
mov a,a0
mov b,b2
mul ab ; a0 * b2
add a,c2
mov c2,a
mov a,b
addc a,c3
mov c3,a
; Byte 3
mov a,a3
mov b,b0
mul ab ; a3 * b0
add a,c3
mov c3,a
mov a,a2
mov b,b1
mul ab ; a2 * b1
add a,c3
mov c3,a
mov a,a1
mov b,b2
mul ab ; a1 * b2
add a,c3
mov c3,a
mov a,a0
mov b,b3
mul ab ; a0 * b3
add a,c3
mov b,c2
mov dph,c1
mov dpl,c0
ret
#else // __SDCC_STACK_AUTO
; parameter a comes in a, b, dph, dpl
mov r2,b ; save parameter a
mov r3,a
#define a0 dpl
#define a1 dph
#define a2 r2
#define a3 r3
#define b0 r1
mov a,#-2-3 ; 1 return address 2 bytes, b 4 bytes
add a,sp ; 1
mov r0,a ; 1 r0 points to b0
; Byte 0
mov a,a0
mov b,@r0 ; b0
mov b0,b ; we need b0 several times
inc r0 ; r0 points to b1
mul ab ; a0 * b0
mov c0,a
mov c1,b
; Byte 1
mov a,a1
mov b,b0
mul ab ; a1 * b0
add a,c1
mov c1,a
clr a
addc a,b
mov c2,a
mov a,a0
mov b,@r0 ; b1
mul ab ; a0 * b1
add a,c1
mov c1,a
mov a,b
addc a,c2
mov c2,a
clr a
rlc a
mov c3,a
; Byte 2
mov a,a2
mov b,b0
mul ab ; a2 * b0
add a,c2
mov c2,a
mov a,b
addc a,c3
mov c3,a
mov a,a1
mov b,@r0 ; b1
mul ab ; a1 * b1
add a,c2
mov c2,a
mov a,b
addc a,c3
mov c3,a
mov a,a0
inc r0
mov b,@r0 ; b2
mul ab ; a0 * b2
add a,c2
mov c2,a
mov a,b
addc a,c3
mov c3,a
; Byte 3
mov a,a3
mov b,b0
mul ab ; a3 * b0
add a,c3
mov c3,a
mov a,a1
mov b,@r0 ; b2
mul ab ; a1 * b2
add a,c3
mov c3,a
mov a,a2
dec r0
mov b,@r0 ; b1
mul ab ; a2 * b1
add a,c3
mov c3,a
mov a,a0
inc r0
inc r0
mov b,@r0 ; b3
mul ab ; a0 * b3
add a,c3
mov b,c2
mov dph,c1
mov dpl,c0
ret
#endif // __SDCC_STACK_AUTO
__endasm;
}
#elif defined(_MULLONG_ASM_LARGE)
void
_mullong_dummy (void) __naked
{
__asm
__mullong:
.globl __mullong
; the result c will be stored in r4...r7
#define c0 r4
#define c1 r5
#define c2 r6
#define c3 r7
; c0 a0 * b0
; c1 a1 * b0 + a0 * b1
; c2 a2 * b0 + a1 * b1 + a0 * b2
; c3 a3 * b0 + a2 * b1 + a1 * b2 + a0 * b3
#if !defined(__SDCC_PARMS_IN_BANK1)
.area XSEG (XDATA)
__mullong_PARM_2:
.globl __mullong_PARM_2
.ds 4
#endif
.area CSEG (CODE)
; parameter a comes in a, b, dph, dpl
mov r0,dpl ; save parameter a
mov r1,dph
mov r2,b
mov r3,a
#define a0 r0
#define a1 r1
#define a2 r2
#define a3 r3
; Byte 0
mov b,a0
#if defined(__SDCC_PARMS_IN_BANK1)
mov a,b1_0 ; b0
#else
mov dptr,#__mullong_PARM_2
movx a,@dptr ; b0
#endif
mul ab ; a0 * b0
mov c0,a
mov c1,b
; Byte 1
mov b,a1
#if defined(__SDCC_PARMS_IN_BANK1)
mov a,b1_0 ; b0
#else
movx a,@dptr ; b0
#endif
mul ab ; a1 * b0
add a,c1
mov c1,a
clr a
addc a,b
mov c2,a
mov b,a0
#if defined(__SDCC_PARMS_IN_BANK1)
mov a,b1_1 ; b1
#else
inc dptr ; b1
movx a,@dptr
#endif
mul ab ; a0 * b1
add a,c1
mov c1,a
mov a,b
addc a,c2
mov c2,a
clr a
rlc a
mov c3,a
; Byte 2
mov b,a1
#if defined(__SDCC_PARMS_IN_BANK1)
mov a,b1_1 ; b1
#else
movx a,@dptr ; b1
#endif
mul ab ; a1 * b1
add a,c2
mov c2,a
mov a,b
addc a,c3
mov c3,a
mov b,a0
#if defined(__SDCC_PARMS_IN_BANK1)
mov a,b1_2 ; b2
#else
inc dptr ; b2
movx a,@dptr
#endif
mul ab ; a0 * b2
add a,c2
mov c2,a
mov a,b
addc a,c3
mov c3,a
mov b,a2
#if defined(__SDCC_PARMS_IN_BANK1)
mov a,b1_0 ; b0
#else
mov dptr,#__mullong_PARM_2
movx a,@dptr ; b0
#endif
mul ab ; a2 * b0
add a,c2
mov c2,a
mov a,b
addc a,c3
mov c3,a
; Byte 3
mov b,a3
#if defined(__SDCC_PARMS_IN_BANK1)
mov a,b1_0 ; b0
#else
movx a,@dptr ; b0
#endif
mul ab ; a3 * b0
add a,c3
mov c3,a
mov b,a2
#if defined(__SDCC_PARMS_IN_BANK1)
mov a,b1_1 ; b1
#else
inc dptr ; b1
movx a,@dptr
#endif
mul ab ; a2 * b1
add a,c3
mov c3,a
mov b,a1
#if defined(__SDCC_PARMS_IN_BANK1)
mov a,b1_2 ; b2
#else
inc dptr ; b2
movx a,@dptr
#endif
mul ab ; a1 * b2
add a,c3
mov c3,a
mov b,a0
#if defined(__SDCC_PARMS_IN_BANK1)
mov a,b1_3 ; b3
#else
inc dptr ; b3
movx a,@dptr
#endif
mul ab ; a0 * b3
add a,c3
mov b,c2
mov dph,c1
mov dpl,c0
ret
__endasm;
}
#elif defined(__SDCC_USE_XSTACK) && defined(__SDCC_STACK_AUTO)
void
_mullong_dummy (void) __naked
{
__asm
__mullong:
.globl __mullong
; the result c will be stored in r4...r7
#define c0 r4
#define c1 r5
#define c2 r6
#define c3 r7
#define a0 dpl
#define a1 dph
#define a2 r2
#define a3 r3
#define b0 r1
; c0 a0 * b0
; c1 a1 * b0 + a0 * b1
; c2 a2 * b0 + a1 * b1 + a0 * b2
; c3 a3 * b0 + a2 * b1 + a1 * b2 + a0 * b3
; parameter a comes in a, b, dph, dpl
mov r2,b ; save parameter a
mov r3,a
mov a,#-4 ; 1 b 4 bytes
add a,_spx ; 1
mov r0,a ; 1 r0 points to b0
; Byte 0
movx a,@r0 ; b0
mov b0,a ; we need b0 several times
inc r0 ; r0 points to b1
mov b,a0
mul ab ; a0 * b0
mov c0,a
mov c1,b
; Byte 1
mov a,a1
mov b,b0
mul ab ; a1 * b0
add a,c1
mov c1,a
clr a
addc a,b
mov c2,a
mov b,a0
movx a,@r0 ; b1
mul ab ; a0 * b1
add a,c1
mov c1,a
mov a,b
addc a,c2
mov c2,a
clr a
rlc a
mov c3,a
; Byte 2
mov a,a2
mov b,b0
mul ab ; a2 * b0
add a,c2
mov c2,a
mov a,b
addc a,c3
mov c3,a
mov b,a1
movx a,@r0 ; b1
mul ab ; a1 * b1
add a,c2
mov c2,a
mov a,b
addc a,c3
mov c3,a
mov b,a0
inc r0
movx a,@r0 ; b2
mul ab ; a0 * b2
add a,c2
mov c2,a
mov a,b
addc a,c3
mov c3,a
; Byte 3
mov a,a3
mov b,b0
mul ab ; a3 * b0
add a,c3
mov c3,a
mov b,a1
movx a,@r0 ; b2
mul ab ; a1 * b2
add a,c3
mov c3,a
mov b,a2
dec r0
movx a,@r0 ; b1
mul ab ; a2 * b1
add a,c3
mov c3,a
mov b,a0
inc r0
inc r0
movx a,@r0 ; b3
mul ab ; a0 * b3
add a,c3
mov b,c2
mov dph,c1
mov dpl,c0
ret
__endasm;
}
#else // _MULLONG_ASM
struct some_struct {
short a ;
char b;
long c ;};
#if defined(__SDCC_hc08) || defined(__SDCC_s08) || defined(__SDCC_stm8)
/* big endian order */
union bil {
struct {unsigned char b3,b2,b1,b0 ;} b;
struct {unsigned short hi,lo ;} i;
unsigned long l;
struct { unsigned char b3; unsigned short i12; unsigned char b0;} bi;
} ;
#else
/* little endian order */
union bil {
struct {unsigned char b0,b1,b2,b3 ;} b;
struct {unsigned short lo,hi ;} i;
unsigned long l;
struct { unsigned char b0; unsigned short i12; unsigned char b3;} bi;
} ;
#endif
#if defined(__SDCC)
#include <sdcc-lib.h>
#endif
#define bcast(x) ((union bil _AUTOMEM *)&(x))
/*
3 2 1 0
X 3 2 1 0
----------------------------
0.3 0.2 0.1 0.0
1.3 1.2 1.1 1.0
2.3 2.2 2.1 2.0
3.3 3.2 3.1 3.0
----------------------------
|3.3|1.3|0.2|0.0| A
|2.3|0.3|0.1| B
|3.2|1.2|1.0| C
|2.2|1.1| D
|3.1|2.0| E
|2.1| F
|3.0| G
|-------> only this side 32 x 32 -> 32
*/
#if defined(__SDCC_USE_XSTACK)
// currently the original code without u fails with --xstack
// it runs out of pointer registers
long
_mullong (long a, long b)
{
union bil t, u;
t.i.hi = bcast(a)->b.b0 * bcast(b)->b.b2; // A
t.i.lo = bcast(a)->b.b0 * bcast(b)->b.b0; // A
u.bi.b3 = bcast(a)->b.b0 * bcast(b)->b.b3; // B
u.bi.i12 = bcast(a)->b.b0 * bcast(b)->b.b1; // B
u.bi.b0 = 0; // B
t.l += u.l;
t.b.b3 += bcast(a)->b.b3 * bcast(b)->b.b0; // G
t.b.b3 += bcast(a)->b.b2 * bcast(b)->b.b1; // F
t.i.hi += bcast(a)->b.b2 * bcast(b)->b.b0; // E
t.i.hi += bcast(a)->b.b1 * bcast(b)->b.b1; // D
u.bi.b3 = bcast(a)->b.b1 * bcast(b)->b.b2; // C
u.bi.i12 = bcast(a)->b.b1 * bcast(b)->b.b0; // C
u.bi.b0 = 0; // C
t.l += u.l;
return t.l;
}
#elif defined(__SDCC_z80) || defined(__SDCC_gbz80) || defined(__SDCC_r2k) || defined(__SDCC_r3k)
/* 32x32->32 multiplication to be used
if 16x16->16 is faster than three 8x8->16.
2009, by M.Bodrato ( http://bodrato.it/ )
z80 and gbz80 don't have any hardware multiplication.
r2k and r3k have 16x16 hardware multiplication.
*/
long
_mullong (long a, long b)
{
unsigned short i12;
bcast(a)->i.hi *= bcast(b)->i.lo;
bcast(a)->i.hi += bcast(b)->i.hi * bcast(a)->i.lo;
/* only (a->i.lo * b->i.lo) 16x16->32 to do. asm? */
bcast(a)->i.hi += bcast(a)->b.b1 * bcast(b)->b.b1;
i12 = bcast(b)->b.b0 * bcast(a)->b.b1;
bcast(b)->bi.i12 = bcast(a)->b.b0 * bcast(b)->b.b1;
/* add up the two partial result, store carry in b3 */
bcast(b)->b.b3 = ((bcast(b)->bi.i12 += i12) < i12);
bcast(a)->i.lo = bcast(a)->b.b0 * bcast(b)->b.b0;
bcast(b)->bi.b0 = 0;
return a + b;
}
#else
long
_mullong (long a, long b)
{
union bil t;
t.i.hi = bcast(a)->b.b0 * bcast(b)->b.b2; // A
t.i.lo = bcast(a)->b.b0 * bcast(b)->b.b0; // A
t.b.b3 += bcast(a)->b.b3 * bcast(b)->b.b0; // G
t.b.b3 += bcast(a)->b.b2 * bcast(b)->b.b1; // F
t.i.hi += bcast(a)->b.b2 * bcast(b)->b.b0; // E <- b lost in .lst
// bcast(a)->i.hi is free !
t.i.hi += bcast(a)->b.b1 * bcast(b)->b.b1; // D <- b lost in .lst
bcast(a)->bi.b3 = bcast(a)->b.b1 * bcast(b)->b.b2; // C
bcast(a)->bi.i12 = bcast(a)->b.b1 * bcast(b)->b.b0; // C
bcast(b)->bi.b3 = bcast(a)->b.b0 * bcast(b)->b.b3; // B
bcast(b)->bi.i12 = bcast(a)->b.b0 * bcast(b)->b.b1; // B
bcast(b)->bi.b0 = 0; // B
bcast(a)->bi.b0 = 0; // C
t.l += a;
return t.l + b;
}
#endif
#endif // _MULLONG_ASM