1/* IBM POWER addmul_1 -- Multiply a limb vector with a limb and add
2 *			 the result to a second limb vector.
3 *
4 *      Copyright (C) 1992, 1994, 1999, 2002 Free Software Foundation, Inc.
5 *
6 * This file is part of Libgcrypt.
7 *
8 * Libgcrypt is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU Lesser General Public License as
10 * published by the Free Software Foundation; either version 2.1 of
11 * the License, or (at your option) any later version.
12 *
13 * Libgcrypt is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
16 * GNU Lesser General Public License for more details.
17 *
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with this program; if not, write to the Free Software
20 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA
21 */
22
23#include "sysdep.h"
24#include "asm-syntax.h"
25
26
27
28/*
29# INPUT PARAMETERS
30# res_ptr	r3
31# s1_ptr	r4
32# size		r5
33# s2_limb	r6
34
35# The RS/6000 has no unsigned 32x32->64 bit multiplication instruction.  To
36# obtain that operation, we have to use the 32x32->64 signed multiplication
37# instruction, and add the appropriate compensation to the high limb of the
38# result.  We add the multiplicand if the multiplier has its most significant
39# bit set, and we add the multiplier if the multiplicand has its most
40# significant bit set.	We need to preserve the carry flag between each
41# iteration, so we have to compute the compensation carefully (the natural,
42# srai+and doesn't work).  Since the POWER architecture has a branch unit
43# we can branch in zero cycles, so that's how we perform the additions.
44 */
45
46	.toc
47	.csect ._gcry_mpih_addmul_1[PR]
48	.align 2
49	.globl _gcry_mpih_addmul_1
50	.globl ._gcry_mpih_addmul_1
51	.csect _gcry_mpih_addmul_1[DS]
52_gcry_mpih_addmul_1:
53	.long ._gcry_mpih_addmul_1[PR], TOC[tc0], 0
54	.csect ._gcry_mpih_addmul_1[PR]
55._gcry_mpih_addmul_1:
56
57	cal	3,-4(3)
58	l	0,0(4)
59	cmpi	0,6,0
60	mtctr	5
61	mul	9,0,6
62	srai	7,0,31
63	and	7,7,6
64	mfmq	8
65	cax	9,9,7
66	l	7,4(3)
67	a	8,8,7		# add res_limb
68	blt	Lneg
69Lpos:	bdz	Lend
70
71Lploop: lu	0,4(4)
72	stu	8,4(3)
73	cmpi	0,0,0
74	mul	10,0,6
75	mfmq	0
76	ae	8,0,9		# low limb + old_cy_limb + old cy
77	l	7,4(3)
78	aze	10,10		# propagate cy to new cy_limb
79	a	8,8,7		# add res_limb
80	bge	Lp0
81	cax	10,10,6 	# adjust high limb for negative limb from s1
82Lp0:	bdz	Lend0
83	lu	0,4(4)
84	stu	8,4(3)
85	cmpi	0,0,0
86	mul	9,0,6
87	mfmq	0
88	ae	8,0,10
89	l	7,4(3)
90	aze	9,9
91	a	8,8,7
92	bge	Lp1
93	cax	9,9,6		# adjust high limb for negative limb from s1
94Lp1:	bdn	Lploop
95
96	b	Lend
97
98Lneg:	cax	9,9,0
99	bdz	Lend
100Lnloop: lu	0,4(4)
101	stu	8,4(3)
102	cmpi	0,0,0
103	mul	10,0,6
104	mfmq	7
105	ae	8,7,9
106	l	7,4(3)
107	ae	10,10,0 	# propagate cy to new cy_limb
108	a	8,8,7		# add res_limb
109	bge	Ln0
110	cax	10,10,6 	# adjust high limb for negative limb from s1
111Ln0:	bdz	Lend0
112	lu	0,4(4)
113	stu	8,4(3)
114	cmpi	0,0,0
115	mul	9,0,6
116	mfmq	7
117	ae	8,7,10
118	l	7,4(3)
119	ae	9,9,0		# propagate cy to new cy_limb
120	a	8,8,7		# add res_limb
121	bge	Ln1
122	cax	9,9,6		# adjust high limb for negative limb from s1
123Ln1:	bdn	Lnloop
124	b	Lend
125
126Lend0:	cal	9,0(10)
127Lend:	st	8,4(3)
128	aze	3,9
129	br
130
131