1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
|
! SPARC __mpn_add_n -- Add two limb vectors of the same length > 0 and store
! sum in a third limb vector.
! Copyright (C) 1995, 1996 Free Software Foundation, Inc.
! This file is part of the GNU MP Library.
! The GNU MP Library is free software; you can redistribute it and/or modify
! it under the terms of the GNU Library General Public License as published by
! the Free Software Foundation; either version 2 of the License, or (at your
! option) any later version.
! The GNU MP 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 Library General Public
! License for more details.
! You should have received a copy of the GNU Library General Public License
! along with the GNU MP Library; see the file COPYING.LIB. If not, write to
! the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
! MA 02111-1307, USA.
! INPUT PARAMETERS
#define res_ptr %o0
#define s1_ptr %o1
#define s2_ptr %o2
#define size %o3
#include "sysdep.h"
.text
.align 4
.global C_SYMBOL_NAME(__mpn_add_n)
C_SYMBOL_NAME(__mpn_add_n):
xor s2_ptr,res_ptr,%g1
andcc %g1,4,%g0
bne L1 ! branch if alignment differs
nop
! ** V1a **
L0: andcc res_ptr,4,%g0 ! res_ptr unaligned? Side effect: cy=0
be L_v1 ! if no, branch
nop
/* Add least significant limb separately to align res_ptr and s2_ptr */
ld [s1_ptr],%g4
add s1_ptr,4,s1_ptr
ld [s2_ptr],%g2
add s2_ptr,4,s2_ptr
add size,-1,size
addcc %g4,%g2,%o4
st %o4,[res_ptr]
add res_ptr,4,res_ptr
L_v1: addx %g0,%g0,%o4 ! save cy in register
cmp size,2 ! if size < 2 ...
bl Lend2 ! ... branch to tail code
subcc %g0,%o4,%g0 ! restore cy
ld [s1_ptr+0],%g4
addcc size,-10,size
ld [s1_ptr+4],%g1
ldd [s2_ptr+0],%g2
blt Lfin1
subcc %g0,%o4,%g0 ! restore cy
/* Add blocks of 8 limbs until less than 8 limbs remain */
Loop1: addxcc %g4,%g2,%o4
ld [s1_ptr+8],%g4
addxcc %g1,%g3,%o5
ld [s1_ptr+12],%g1
ldd [s2_ptr+8],%g2
std %o4,[res_ptr+0]
addxcc %g4,%g2,%o4
ld [s1_ptr+16],%g4
addxcc %g1,%g3,%o5
ld [s1_ptr+20],%g1
ldd [s2_ptr+16],%g2
std %o4,[res_ptr+8]
addxcc %g4,%g2,%o4
ld [s1_ptr+24],%g4
addxcc %g1,%g3,%o5
ld [s1_ptr+28],%g1
ldd [s2_ptr+24],%g2
std %o4,[res_ptr+16]
addxcc %g4,%g2,%o4
ld [s1_ptr+32],%g4
addxcc %g1,%g3,%o5
ld [s1_ptr+36],%g1
ldd [s2_ptr+32],%g2
std %o4,[res_ptr+24]
addx %g0,%g0,%o4 ! save cy in register
addcc size,-8,size
add s1_ptr,32,s1_ptr
add s2_ptr,32,s2_ptr
add res_ptr,32,res_ptr
bge Loop1
subcc %g0,%o4,%g0 ! restore cy
Lfin1: addcc size,8-2,size
blt Lend1
subcc %g0,%o4,%g0 ! restore cy
/* Add blocks of 2 limbs until less than 2 limbs remain */
Loope1: addxcc %g4,%g2,%o4
ld [s1_ptr+8],%g4
addxcc %g1,%g3,%o5
ld [s1_ptr+12],%g1
ldd [s2_ptr+8],%g2
std %o4,[res_ptr+0]
addx %g0,%g0,%o4 ! save cy in register
addcc size,-2,size
add s1_ptr,8,s1_ptr
add s2_ptr,8,s2_ptr
add res_ptr,8,res_ptr
bge Loope1
subcc %g0,%o4,%g0 ! restore cy
Lend1: addxcc %g4,%g2,%o4
addxcc %g1,%g3,%o5
std %o4,[res_ptr+0]
addx %g0,%g0,%o4 ! save cy in register
andcc size,1,%g0
be Lret1
subcc %g0,%o4,%g0 ! restore cy
/* Add last limb */
ld [s1_ptr+8],%g4
ld [s2_ptr+8],%g2
addxcc %g4,%g2,%o4
st %o4,[res_ptr+8]
Lret1: retl
addx %g0,%g0,%o0 ! return carry-out from most sign. limb
L1: xor s1_ptr,res_ptr,%g1
andcc %g1,4,%g0
bne L2
nop
! ** V1b **
mov s2_ptr,%g1
mov s1_ptr,s2_ptr
b L0
mov %g1,s1_ptr
! ** V2 **
/* If we come here, the alignment of s1_ptr and res_ptr as well as the
alignment of s2_ptr and res_ptr differ. Since there are only two ways
things can be aligned (that we care about) we now know that the alignment
of s1_ptr and s2_ptr are the same. */
L2: cmp size,1
be Ljone
nop
andcc s1_ptr,4,%g0 ! s1_ptr unaligned? Side effect: cy=0
be L_v2 ! if no, branch
nop
/* Add least significant limb separately to align s1_ptr and s2_ptr */
ld [s1_ptr],%g4
add s1_ptr,4,s1_ptr
ld [s2_ptr],%g2
add s2_ptr,4,s2_ptr
add size,-1,size
addcc %g4,%g2,%o4
st %o4,[res_ptr]
add res_ptr,4,res_ptr
L_v2: addx %g0,%g0,%o4 ! save cy in register
addcc size,-8,size
blt Lfin2
subcc %g0,%o4,%g0 ! restore cy
/* Add blocks of 8 limbs until less than 8 limbs remain */
Loop2: ldd [s1_ptr+0],%g2
ldd [s2_ptr+0],%o4
addxcc %g2,%o4,%g2
st %g2,[res_ptr+0]
addxcc %g3,%o5,%g3
st %g3,[res_ptr+4]
ldd [s1_ptr+8],%g2
ldd [s2_ptr+8],%o4
addxcc %g2,%o4,%g2
st %g2,[res_ptr+8]
addxcc %g3,%o5,%g3
st %g3,[res_ptr+12]
ldd [s1_ptr+16],%g2
ldd [s2_ptr+16],%o4
addxcc %g2,%o4,%g2
st %g2,[res_ptr+16]
addxcc %g3,%o5,%g3
st %g3,[res_ptr+20]
ldd [s1_ptr+24],%g2
ldd [s2_ptr+24],%o4
addxcc %g2,%o4,%g2
st %g2,[res_ptr+24]
addxcc %g3,%o5,%g3
st %g3,[res_ptr+28]
addx %g0,%g0,%o4 ! save cy in register
addcc size,-8,size
add s1_ptr,32,s1_ptr
add s2_ptr,32,s2_ptr
add res_ptr,32,res_ptr
bge Loop2
subcc %g0,%o4,%g0 ! restore cy
Lfin2: addcc size,8-2,size
blt Lend2
subcc %g0,%o4,%g0 ! restore cy
Loope2: ldd [s1_ptr+0],%g2
ldd [s2_ptr+0],%o4
addxcc %g2,%o4,%g2
st %g2,[res_ptr+0]
addxcc %g3,%o5,%g3
st %g3,[res_ptr+4]
addx %g0,%g0,%o4 ! save cy in register
addcc size,-2,size
add s1_ptr,8,s1_ptr
add s2_ptr,8,s2_ptr
add res_ptr,8,res_ptr
bge Loope2
subcc %g0,%o4,%g0 ! restore cy
Lend2: andcc size,1,%g0
be Lret2
subcc %g0,%o4,%g0 ! restore cy
/* Add last limb */
Ljone: ld [s1_ptr],%g4
ld [s2_ptr],%g2
addxcc %g4,%g2,%o4
st %o4,[res_ptr]
Lret2: retl
addx %g0,%g0,%o0 ! return carry-out from most sign. limb
|
