189 lines
5.7 KiB
ArmAsm
189 lines
5.7 KiB
ArmAsm
/* $NetBSD: mul.S,v 1.4 1998/11/26 19:08:21 eeh Exp $ */
|
|
|
|
/*
|
|
* Copyright (c) 1992, 1993
|
|
* The Regents of the University of California. All rights reserved.
|
|
*
|
|
* This software was developed by the Computer Systems Engineering group
|
|
* at Lawrence Berkeley Laboratory under DARPA contract BG 91-66 and
|
|
* contributed to Berkeley.
|
|
*
|
|
* Redistribution and use in source and binary forms, with or without
|
|
* modification, are permitted provided that the following conditions
|
|
* are met:
|
|
* 1. Redistributions of source code must retain the above copyright
|
|
* notice, this list of conditions and the following disclaimer.
|
|
* 2. Redistributions in binary form must reproduce the above copyright
|
|
* notice, this list of conditions and the following disclaimer in the
|
|
* documentation and/or other materials provided with the distribution.
|
|
* 3. All advertising materials mentioning features or use of this software
|
|
* must display the following acknowledgement:
|
|
* This product includes software developed by the University of
|
|
* California, Berkeley and its contributors.
|
|
* 4. Neither the name of the University nor the names of its contributors
|
|
* may be used to endorse or promote products derived from this software
|
|
* without specific prior written permission.
|
|
*
|
|
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
|
|
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
|
|
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
|
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
|
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
|
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
|
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
|
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
|
* SUCH DAMAGE.
|
|
*
|
|
* Header: mul.s,v 1.5 92/06/25 13:24:03 torek Exp
|
|
*/
|
|
|
|
#if defined(LIBC_SCCS) && !defined(lint)
|
|
#ifdef notdef
|
|
.asciz "@(#)mul.s 8.1 (Berkeley) 6/4/93"
|
|
#endif
|
|
.asciz "$NetBSD: mul.S,v 1.4 1998/11/26 19:08:21 eeh Exp $"
|
|
#endif /* LIBC_SCCS and not lint */
|
|
|
|
/*
|
|
* This doesn't really belong here and should be handled by toolchain
|
|
* magic.
|
|
*/
|
|
|
|
.globl __mul64, __div64, __udiv64, __rem64, __urem64
|
|
|
|
__mul64:
|
|
ba,a,pt %icc,__muldi3
|
|
__div64:
|
|
ba,a,pt %icc,__divdi3
|
|
__udiv64:
|
|
ba,a,pt %icc,__udivdi3
|
|
__rem64:
|
|
ba,a,pt %icc,__moddi3
|
|
__urem64:
|
|
ba,a,pt %icc,__umoddi3
|
|
|
|
/*
|
|
* Signed multiply, from Appendix E of the Sparc Version 8
|
|
* Architecture Manual.
|
|
*
|
|
* Returns %o0 * %o1 in %o1%o0 (i.e., %o1 holds the upper 32 bits of
|
|
* the 64-bit product).
|
|
*
|
|
* This code optimizes short (less than 13-bit) multiplies.
|
|
*/
|
|
|
|
#include "DEFS.h"
|
|
FUNC(.mul)
|
|
#if 1
|
|
cmp %o0, %o1 ! Most efficient if smaller is in rs1
|
|
blt,a %icc, 1f
|
|
mulx %o0, %o1, %o0
|
|
mulx %o1, %o0, %o0
|
|
1:
|
|
srax %o0, 32, %o1
|
|
retl
|
|
sll %o0, 0, %o0 ! Sign extend
|
|
#else
|
|
mov %o0, %y ! multiplier -> Y
|
|
andncc %o0, 0xfff, %g0 ! test bits 12..31
|
|
be Lmul_shortway ! if zero, can do it the short way
|
|
andcc %g0, %g0, %o4 ! zero the partial product and clear N and V
|
|
|
|
/*
|
|
* Long multiply. 32 steps, followed by a final shift step.
|
|
*/
|
|
mulscc %o4, %o1, %o4 ! 1
|
|
mulscc %o4, %o1, %o4 ! 2
|
|
mulscc %o4, %o1, %o4 ! 3
|
|
mulscc %o4, %o1, %o4 ! 4
|
|
mulscc %o4, %o1, %o4 ! 5
|
|
mulscc %o4, %o1, %o4 ! 6
|
|
mulscc %o4, %o1, %o4 ! 7
|
|
mulscc %o4, %o1, %o4 ! 8
|
|
mulscc %o4, %o1, %o4 ! 9
|
|
mulscc %o4, %o1, %o4 ! 10
|
|
mulscc %o4, %o1, %o4 ! 11
|
|
mulscc %o4, %o1, %o4 ! 12
|
|
mulscc %o4, %o1, %o4 ! 13
|
|
mulscc %o4, %o1, %o4 ! 14
|
|
mulscc %o4, %o1, %o4 ! 15
|
|
mulscc %o4, %o1, %o4 ! 16
|
|
mulscc %o4, %o1, %o4 ! 17
|
|
mulscc %o4, %o1, %o4 ! 18
|
|
mulscc %o4, %o1, %o4 ! 19
|
|
mulscc %o4, %o1, %o4 ! 20
|
|
mulscc %o4, %o1, %o4 ! 21
|
|
mulscc %o4, %o1, %o4 ! 22
|
|
mulscc %o4, %o1, %o4 ! 23
|
|
mulscc %o4, %o1, %o4 ! 24
|
|
mulscc %o4, %o1, %o4 ! 25
|
|
mulscc %o4, %o1, %o4 ! 26
|
|
mulscc %o4, %o1, %o4 ! 27
|
|
mulscc %o4, %o1, %o4 ! 28
|
|
mulscc %o4, %o1, %o4 ! 29
|
|
mulscc %o4, %o1, %o4 ! 30
|
|
mulscc %o4, %o1, %o4 ! 31
|
|
mulscc %o4, %o1, %o4 ! 32
|
|
mulscc %o4, %g0, %o4 ! final shift
|
|
|
|
! If %o0 was negative, the result is
|
|
! (%o0 * %o1) + (%o1 << 32))
|
|
! We fix that here.
|
|
|
|
tst %o0
|
|
bge 1f
|
|
rd %y, %o0
|
|
|
|
! %o0 was indeed negative; fix upper 32 bits of result by subtracting
|
|
! %o1 (i.e., return %o4 - %o1 in %o1).
|
|
retl
|
|
sub %o4, %o1, %o1
|
|
|
|
1:
|
|
retl
|
|
mov %o4, %o1
|
|
|
|
Lmul_shortway:
|
|
/*
|
|
* Short multiply. 12 steps, followed by a final shift step.
|
|
* The resulting bits are off by 12 and (32-12) = 20 bit positions,
|
|
* but there is no problem with %o0 being negative (unlike above).
|
|
*/
|
|
mulscc %o4, %o1, %o4 ! 1
|
|
mulscc %o4, %o1, %o4 ! 2
|
|
mulscc %o4, %o1, %o4 ! 3
|
|
mulscc %o4, %o1, %o4 ! 4
|
|
mulscc %o4, %o1, %o4 ! 5
|
|
mulscc %o4, %o1, %o4 ! 6
|
|
mulscc %o4, %o1, %o4 ! 7
|
|
mulscc %o4, %o1, %o4 ! 8
|
|
mulscc %o4, %o1, %o4 ! 9
|
|
mulscc %o4, %o1, %o4 ! 10
|
|
mulscc %o4, %o1, %o4 ! 11
|
|
mulscc %o4, %o1, %o4 ! 12
|
|
mulscc %o4, %g0, %o4 ! final shift
|
|
|
|
/*
|
|
* %o4 has 20 of the bits that should be in the low part of the
|
|
* result; %y has the bottom 12 (as %y's top 12). That is:
|
|
*
|
|
* %o4 %y
|
|
* +----------------+----------------+
|
|
* | -12- | -20- | -12- | -20- |
|
|
* +------(---------+------)---------+
|
|
* --hi-- ----low-part----
|
|
*
|
|
* The upper 12 bits of %o4 should be sign-extended to form the
|
|
* high part of the product (i.e., highpart = %o4 >> 20).
|
|
*/
|
|
|
|
rd %y, %o5
|
|
sll %o4, 12, %o0 ! shift middle bits left 12
|
|
srl %o5, 20, %o5 ! shift low bits right 20, zero fill at left
|
|
or %o5, %o0, %o0 ! construct low part of result
|
|
retl
|
|
sra %o4, 20, %o1 ! ... and extract high part of result
|
|
#endif
|