Home | History | Annotate | Line # | Download | only in gen
      1  1.1  christos /*	$NetBSD: umul.S,v 1.1 2005/12/20 19:28:50 christos Exp $	*/
      2  1.1  christos 
      3  1.1  christos /*
      4  1.1  christos  * Copyright (c) 1992, 1993
      5  1.1  christos  *	The Regents of the University of California.  All rights reserved.
      6  1.1  christos  *
      7  1.1  christos  * This software was developed by the Computer Systems Engineering group
      8  1.1  christos  * at Lawrence Berkeley Laboratory under DARPA contract BG 91-66 and
      9  1.1  christos  * contributed to Berkeley.
     10  1.1  christos  *
     11  1.1  christos  * Redistribution and use in source and binary forms, with or without
     12  1.1  christos  * modification, are permitted provided that the following conditions
     13  1.1  christos  * are met:
     14  1.1  christos  * 1. Redistributions of source code must retain the above copyright
     15  1.1  christos  *    notice, this list of conditions and the following disclaimer.
     16  1.1  christos  * 2. Redistributions in binary form must reproduce the above copyright
     17  1.1  christos  *    notice, this list of conditions and the following disclaimer in the
     18  1.1  christos  *    documentation and/or other materials provided with the distribution.
     19  1.1  christos  * 3. Neither the name of the University nor the names of its contributors
     20  1.1  christos  *    may be used to endorse or promote products derived from this software
     21  1.1  christos  *    without specific prior written permission.
     22  1.1  christos  *
     23  1.1  christos  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     24  1.1  christos  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     25  1.1  christos  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     26  1.1  christos  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     27  1.1  christos  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     28  1.1  christos  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     29  1.1  christos  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     30  1.1  christos  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     31  1.1  christos  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     32  1.1  christos  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     33  1.1  christos  * SUCH DAMAGE.
     34  1.1  christos  *
     35  1.1  christos  * from: Header: umul.s,v 1.4 92/06/25 13:24:05 torek Exp
     36  1.1  christos  */
     37  1.1  christos 
     38  1.1  christos #include <machine/asm.h>
     39  1.1  christos #if defined(LIBC_SCCS) && !defined(lint)
     40  1.1  christos #if 0
     41  1.1  christos 	.asciz "@(#)umul.s	8.1 (Berkeley) 6/4/93"
     42  1.1  christos #else
     43  1.1  christos 	RCSID("$NetBSD: umul.S,v 1.1 2005/12/20 19:28:50 christos Exp $")
     44  1.1  christos #endif
     45  1.1  christos #endif /* LIBC_SCCS and not lint */
     46  1.1  christos 
     47  1.1  christos /*
     48  1.1  christos  * Unsigned multiply.  Returns %o0 * %o1 in %o1%o0 (i.e., %o1 holds the
     49  1.1  christos  * upper 32 bits of the 64-bit product).
     50  1.1  christos  *
     51  1.1  christos  * This code optimizes short (less than 13-bit) multiplies.  Short
     52  1.1  christos  * multiplies require 25 instruction cycles, and long ones require
     53  1.1  christos  * 45 instruction cycles.
     54  1.1  christos  *
     55  1.1  christos  * On return, overflow has occurred (%o1 is not zero) if and only if
     56  1.1  christos  * the Z condition code is clear, allowing, e.g., the following:
     57  1.1  christos  *
     58  1.1  christos  *	call	.umul
     59  1.1  christos  *	nop
     60  1.1  christos  *	bnz	overflow	(or tnz)
     61  1.1  christos  */
     62  1.1  christos 
     63  1.1  christos FUNC(.umul)
     64  1.1  christos 	or	%o0, %o1, %o4
     65  1.1  christos 	mov	%o0, %y		! multiplier -> Y
     66  1.1  christos 	andncc	%o4, 0xfff, %g0	! test bits 12..31 of *both* args
     67  1.1  christos 	be	Lmul_shortway	! if zero, can do it the short way
     68  1.1  christos 	andcc	%g0, %g0, %o4	! zero the partial product and clear N and V
     69  1.1  christos 
     70  1.1  christos 	/*
     71  1.1  christos 	 * Long multiply.  32 steps, followed by a final shift step.
     72  1.1  christos 	 */
     73  1.1  christos 	mulscc	%o4, %o1, %o4	! 1
     74  1.1  christos 	mulscc	%o4, %o1, %o4	! 2
     75  1.1  christos 	mulscc	%o4, %o1, %o4	! 3
     76  1.1  christos 	mulscc	%o4, %o1, %o4	! 4
     77  1.1  christos 	mulscc	%o4, %o1, %o4	! 5
     78  1.1  christos 	mulscc	%o4, %o1, %o4	! 6
     79  1.1  christos 	mulscc	%o4, %o1, %o4	! 7
     80  1.1  christos 	mulscc	%o4, %o1, %o4	! 8
     81  1.1  christos 	mulscc	%o4, %o1, %o4	! 9
     82  1.1  christos 	mulscc	%o4, %o1, %o4	! 10
     83  1.1  christos 	mulscc	%o4, %o1, %o4	! 11
     84  1.1  christos 	mulscc	%o4, %o1, %o4	! 12
     85  1.1  christos 	mulscc	%o4, %o1, %o4	! 13
     86  1.1  christos 	mulscc	%o4, %o1, %o4	! 14
     87  1.1  christos 	mulscc	%o4, %o1, %o4	! 15
     88  1.1  christos 	mulscc	%o4, %o1, %o4	! 16
     89  1.1  christos 	mulscc	%o4, %o1, %o4	! 17
     90  1.1  christos 	mulscc	%o4, %o1, %o4	! 18
     91  1.1  christos 	mulscc	%o4, %o1, %o4	! 19
     92  1.1  christos 	mulscc	%o4, %o1, %o4	! 20
     93  1.1  christos 	mulscc	%o4, %o1, %o4	! 21
     94  1.1  christos 	mulscc	%o4, %o1, %o4	! 22
     95  1.1  christos 	mulscc	%o4, %o1, %o4	! 23
     96  1.1  christos 	mulscc	%o4, %o1, %o4	! 24
     97  1.1  christos 	mulscc	%o4, %o1, %o4	! 25
     98  1.1  christos 	mulscc	%o4, %o1, %o4	! 26
     99  1.1  christos 	mulscc	%o4, %o1, %o4	! 27
    100  1.1  christos 	mulscc	%o4, %o1, %o4	! 28
    101  1.1  christos 	mulscc	%o4, %o1, %o4	! 29
    102  1.1  christos 	mulscc	%o4, %o1, %o4	! 30
    103  1.1  christos 	mulscc	%o4, %o1, %o4	! 31
    104  1.1  christos 	mulscc	%o4, %o1, %o4	! 32
    105  1.1  christos 	mulscc	%o4, %g0, %o4	! final shift
    106  1.1  christos 
    107  1.1  christos 
    108  1.1  christos 	/*
    109  1.1  christos 	 * Normally, with the shift-and-add approach, if both numbers are
    110  1.1  christos 	 * positive you get the correct result.  WIth 32-bit two's-complement
    111  1.1  christos 	 * numbers, -x is represented as
    112  1.1  christos 	 *
    113  1.1  christos 	 *		  x		    32
    114  1.1  christos 	 *	( 2  -  ------ ) mod 2  *  2
    115  1.1  christos 	 *		   32
    116  1.1  christos 	 *		  2
    117  1.1  christos 	 *
    118  1.1  christos 	 * (the `mod 2' subtracts 1 from 1.bbbb).  To avoid lots of 2^32s,
    119  1.1  christos 	 * we can treat this as if the radix point were just to the left
    120  1.1  christos 	 * of the sign bit (multiply by 2^32), and get
    121  1.1  christos 	 *
    122  1.1  christos 	 *	-x  =  (2 - x) mod 2
    123  1.1  christos 	 *
    124  1.1  christos 	 * Then, ignoring the `mod 2's for convenience:
    125  1.1  christos 	 *
    126  1.1  christos 	 *   x *  y	= xy
    127  1.1  christos 	 *  -x *  y	= 2y - xy
    128  1.1  christos 	 *   x * -y	= 2x - xy
    129  1.1  christos 	 *  -x * -y	= 4 - 2x - 2y + xy
    130  1.1  christos 	 *
    131  1.1  christos 	 * For signed multiplies, we subtract (x << 32) from the partial
    132  1.1  christos 	 * product to fix this problem for negative multipliers (see mul.s).
    133  1.1  christos 	 * Because of the way the shift into the partial product is calculated
    134  1.1  christos 	 * (N xor V), this term is automatically removed for the multiplicand,
    135  1.1  christos 	 * so we don't have to adjust.
    136  1.1  christos 	 *
    137  1.1  christos 	 * But for unsigned multiplies, the high order bit wasn't a sign bit,
    138  1.1  christos 	 * and the correction is wrong.  So for unsigned multiplies where the
    139  1.1  christos 	 * high order bit is one, we end up with xy - (y << 32).  To fix it
    140  1.1  christos 	 * we add y << 32.
    141  1.1  christos 	 */
    142  1.1  christos 	tst	%o1
    143  1.1  christos 	bl,a	1f		! if %o1 < 0 (high order bit = 1),
    144  1.1  christos 	add	%o4, %o0, %o4	! %o4 += %o0 (add y to upper half)
    145  1.1  christos 1:	rd	%y, %o0		! get lower half of product
    146  1.1  christos 	retl
    147  1.1  christos 	addcc	%o4, %g0, %o1	! put upper half in place and set Z for %o1==0
    148  1.1  christos 
    149  1.1  christos Lmul_shortway:
    150  1.1  christos 	/*
    151  1.1  christos 	 * Short multiply.  12 steps, followed by a final shift step.
    152  1.1  christos 	 * The resulting bits are off by 12 and (32-12) = 20 bit positions,
    153  1.1  christos 	 * but there is no problem with %o0 being negative (unlike above),
    154  1.1  christos 	 * and overflow is impossible (the answer is at most 24 bits long).
    155  1.1  christos 	 */
    156  1.1  christos 	mulscc	%o4, %o1, %o4	! 1
    157  1.1  christos 	mulscc	%o4, %o1, %o4	! 2
    158  1.1  christos 	mulscc	%o4, %o1, %o4	! 3
    159  1.1  christos 	mulscc	%o4, %o1, %o4	! 4
    160  1.1  christos 	mulscc	%o4, %o1, %o4	! 5
    161  1.1  christos 	mulscc	%o4, %o1, %o4	! 6
    162  1.1  christos 	mulscc	%o4, %o1, %o4	! 7
    163  1.1  christos 	mulscc	%o4, %o1, %o4	! 8
    164  1.1  christos 	mulscc	%o4, %o1, %o4	! 9
    165  1.1  christos 	mulscc	%o4, %o1, %o4	! 10
    166  1.1  christos 	mulscc	%o4, %o1, %o4	! 11
    167  1.1  christos 	mulscc	%o4, %o1, %o4	! 12
    168  1.1  christos 	mulscc	%o4, %g0, %o4	! final shift
    169  1.1  christos 
    170  1.1  christos 	/*
    171  1.1  christos 	 * %o4 has 20 of the bits that should be in the result; %y has
    172  1.1  christos 	 * the bottom 12 (as %y's top 12).  That is:
    173  1.1  christos 	 *
    174  1.1  christos 	 *	  %o4		    %y
    175  1.1  christos 	 * +----------------+----------------+
    176  1.1  christos 	 * | -12- |   -20-  | -12- |   -20-  |
    177  1.1  christos 	 * +------(---------+------)---------+
    178  1.1  christos 	 *	   -----result-----
    179  1.1  christos 	 *
    180  1.1  christos 	 * The 12 bits of %o4 left of the `result' area are all zero;
    181  1.1  christos 	 * in fact, all top 20 bits of %o4 are zero.
    182  1.1  christos 	 */
    183  1.1  christos 
    184  1.1  christos 	rd	%y, %o5
    185  1.1  christos 	sll	%o4, 12, %o0	! shift middle bits left 12
    186  1.1  christos 	srl	%o5, 20, %o5	! shift low bits right 20
    187  1.1  christos 	or	%o5, %o0, %o0
    188  1.1  christos 	retl
    189  1.1  christos 	addcc	%g0, %g0, %o1	! %o1 = zero, and set Z
    190