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
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
|
/* mpn_mod_1_1p (ap, n, b, cps)
Divide (ap,,n) by b. Return the single-limb remainder.
Contributed to the GNU project by Torbjorn Granlund and Niels Möller.
Based on a suggestion by Peter L. Montgomery.
THE FUNCTIONS IN THIS FILE ARE INTERNAL WITH MUTABLE INTERFACES. IT IS ONLY
SAFE TO REACH THEM THROUGH DOCUMENTED INTERFACES. IN FACT, IT IS ALMOST
GUARANTEED THAT THEY WILL CHANGE OR DISAPPEAR IN A FUTURE GNU MP RELEASE.
Copyright 2008-2011, 2013 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 either:
* the GNU Lesser General Public License as published by the Free
Software Foundation; either version 3 of the License, or (at your
option) any later version.
or
* the GNU General Public License as published by the Free Software
Foundation; either version 2 of the License, or (at your option) any
later version.
or both in parallel, as here.
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 General Public License
for more details.
You should have received copies of the GNU General Public License and the
GNU Lesser General Public License along with the GNU MP Library. If not,
see https://www.gnu.org/licenses/. */
#include "gmp-impl.h"
#include "longlong.h"
#ifndef MOD_1_1P_METHOD
# define MOD_1_1P_METHOD 1 /* need to make sure this is 2 for asm testing */
#endif
/* Define some longlong.h-style macros, but for wider operations.
* add_mssaaaa is like longlong.h's add_ssaaaa, but also generates
* carry out, in the form of a mask. */
#if defined (__GNUC__) && ! defined (NO_ASM)
#if HAVE_HOST_CPU_FAMILY_x86 && W_TYPE_SIZE == 32
#define add_mssaaaa(m, s1, s0, a1, a0, b1, b0) \
__asm__ ( "add %6, %k2\n\t" \
"adc %4, %k1\n\t" \
"sbb %k0, %k0" \
: "=r" (m), "=r" (s1), "=&r" (s0) \
: "1" ((USItype)(a1)), "g" ((USItype)(b1)), \
"%2" ((USItype)(a0)), "g" ((USItype)(b0)))
#endif
#if HAVE_HOST_CPU_FAMILY_x86_64 && W_TYPE_SIZE == 64
#define add_mssaaaa(m, s1, s0, a1, a0, b1, b0) \
__asm__ ( "add %6, %q2\n\t" \
"adc %4, %q1\n\t" \
"sbb %q0, %q0" \
: "=r" (m), "=r" (s1), "=&r" (s0) \
: "1" ((UDItype)(a1)), "rme" ((UDItype)(b1)), \
"%2" ((UDItype)(a0)), "rme" ((UDItype)(b0)))
#endif
#if defined (__sparc__) && W_TYPE_SIZE == 32
#define add_mssaaaa(m, sh, sl, ah, al, bh, bl) \
__asm__ ( "addcc %r5, %6, %2\n\t" \
"addxcc %r3, %4, %1\n\t" \
"subx %%g0, %%g0, %0" \
: "=r" (m), "=r" (sh), "=&r" (sl) \
: "rJ" (ah), "rI" (bh), "%rJ" (al), "rI" (bl) \
__CLOBBER_CC)
#endif
#if defined (__sparc__) && W_TYPE_SIZE == 64
#define add_mssaaaa(m, sh, sl, ah, al, bh, bl) \
__asm__ ( "addcc %r5, %6, %2\n\t" \
"addccc %r7, %8, %%g0\n\t" \
"addccc %r3, %4, %1\n\t" \
"clr %0\n\t" \
"movcs %%xcc, -1, %0" \
: "=r" (m), "=r" (sh), "=&r" (sl) \
: "rJ" (ah), "rI" (bh), "%rJ" (al), "rI" (bl), \
"rJ" ((al) >> 32), "rI" ((bl) >> 32) \
__CLOBBER_CC)
#if __VIS__ >= 0x300
#undef add_mssaaaa
#define add_mssaaaa(m, sh, sl, ah, al, bh, bl) \
__asm__ ( "addcc %r5, %6, %2\n\t" \
"addxccc %r3, %4, %1\n\t" \
"clr %0\n\t" \
"movcs %%xcc, -1, %0" \
: "=r" (m), "=r" (sh), "=&r" (sl) \
: "rJ" (ah), "rI" (bh), "%rJ" (al), "rI" (bl) \
__CLOBBER_CC)
#endif
#endif
#if HAVE_HOST_CPU_FAMILY_powerpc && !defined (_LONG_LONG_LIMB)
/* This works fine for 32-bit and 64-bit limbs, except for 64-bit limbs with a
processor running in 32-bit mode, since the carry flag then gets the 32-bit
carry. */
#define add_mssaaaa(m, s1, s0, a1, a0, b1, b0) \
__asm__ ( "add%I6c %2, %5, %6\n\t" \
"adde %1, %3, %4\n\t" \
"subfe %0, %0, %0\n\t" \
"nor %0, %0, %0" \
: "=r" (m), "=r" (s1), "=&r" (s0) \
: "r" (a1), "r" (b1), "%r" (a0), "rI" (b0) \
__CLOBBER_CC)
#endif
#if defined (__s390x__) && W_TYPE_SIZE == 64
#define add_mssaaaa(m, s1, s0, a1, a0, b1, b0) \
__asm__ ( "algr %2, %6\n\t" \
"alcgr %1, %4\n\t" \
"lghi %0, 0\n\t" \
"alcgr %0, %0\n\t" \
"lcgr %0, %0" \
: "=r" (m), "=r" (s1), "=&r" (s0) \
: "1" ((UDItype)(a1)), "r" ((UDItype)(b1)), \
"%2" ((UDItype)(a0)), "r" ((UDItype)(b0)) __CLOBBER_CC)
#endif
#if defined (__arm__) && !defined (__thumb__) && W_TYPE_SIZE == 32
#define add_mssaaaa(m, sh, sl, ah, al, bh, bl) \
__asm__ ( "adds %2, %5, %6\n\t" \
"adcs %1, %3, %4\n\t" \
"movcc %0, #0\n\t" \
"movcs %0, #-1" \
: "=r" (m), "=r" (sh), "=&r" (sl) \
: "r" (ah), "rI" (bh), "%r" (al), "rI" (bl) __CLOBBER_CC)
#endif
#if defined (__aarch64__) && W_TYPE_SIZE == 64
#define add_mssaaaa(m, sh, sl, ah, al, bh, bl) \
__asm__ ( "adds %2, %x5, %6\n\t" \
"adcs %1, %x3, %x4\n\t" \
"csinv %0, xzr, xzr, cc\n\t" \
: "=r" (m), "=r" (sh), "=&r" (sl) \
: "rZ" (ah), "rZ" (bh), "%rZ" (al), "rI" (bl) __CLOBBER_CC)
#endif
#endif /* defined (__GNUC__) */
#ifndef add_mssaaaa
#define add_mssaaaa(m, s1, s0, a1, a0, b1, b0) \
do { \
UWtype __s0, __s1, __c0, __c1; \
__s0 = (a0) + (b0); \
__s1 = (a1) + (b1); \
__c0 = __s0 < (a0); \
__c1 = __s1 < (a1); \
(s0) = __s0; \
__s1 = __s1 + __c0; \
(s1) = __s1; \
(m) = - (__c1 + (__s1 < __c0)); \
} while (0)
#endif
#if MOD_1_1P_METHOD == 1
void
mpn_mod_1_1p_cps (mp_limb_t cps[4], mp_limb_t b)
{
mp_limb_t bi;
mp_limb_t B1modb, B2modb;
int cnt;
count_leading_zeros (cnt, b);
b <<= cnt;
invert_limb (bi, b);
cps[0] = bi;
cps[1] = cnt;
B1modb = -b;
if (LIKELY (cnt != 0))
B1modb *= ((bi >> (GMP_LIMB_BITS-cnt)) | (CNST_LIMB(1) << cnt));
ASSERT (B1modb <= b); /* NB: not fully reduced mod b */
cps[2] = B1modb >> cnt;
/* In the normalized case, this can be simplified to
*
* B2modb = - b * bi;
* ASSERT (B2modb <= b); // NB: equality iff b = B/2
*/
udiv_rnnd_preinv (B2modb, B1modb, CNST_LIMB(0), b, bi);
cps[3] = B2modb >> cnt;
}
mp_limb_t
mpn_mod_1_1p (mp_srcptr ap, mp_size_t n, mp_limb_t b, const mp_limb_t bmodb[4])
{
mp_limb_t rh, rl, bi, ph, pl, r;
mp_limb_t B1modb, B2modb;
mp_size_t i;
int cnt;
mp_limb_t mask;
ASSERT (n >= 2); /* fix tuneup.c if this is changed */
B1modb = bmodb[2];
B2modb = bmodb[3];
rl = ap[n - 1];
umul_ppmm (ph, pl, rl, B1modb);
add_ssaaaa (rh, rl, ph, pl, CNST_LIMB(0), ap[n - 2]);
for (i = n - 3; i >= 0; i -= 1)
{
/* rr = ap[i] < B
+ LO(rr) * (B mod b) <= (B-1)(b-1)
+ HI(rr) * (B^2 mod b) <= (B-1)(b-1)
*/
umul_ppmm (ph, pl, rl, B1modb);
add_ssaaaa (ph, pl, ph, pl, CNST_LIMB(0), ap[i]);
umul_ppmm (rh, rl, rh, B2modb);
add_ssaaaa (rh, rl, rh, rl, ph, pl);
}
cnt = bmodb[1];
bi = bmodb[0];
if (LIKELY (cnt != 0))
rh = (rh << cnt) | (rl >> (GMP_LIMB_BITS - cnt));
mask = -(mp_limb_t) (rh >= b);
rh -= mask & b;
udiv_rnnd_preinv (r, rh, rl << cnt, b, bi);
return r >> cnt;
}
#endif /* MOD_1_1P_METHOD == 1 */
#if MOD_1_1P_METHOD == 2
void
mpn_mod_1_1p_cps (mp_limb_t cps[4], mp_limb_t b)
{
mp_limb_t bi;
mp_limb_t B2modb;
int cnt;
count_leading_zeros (cnt, b);
b <<= cnt;
invert_limb (bi, b);
cps[0] = bi;
cps[1] = cnt;
if (LIKELY (cnt != 0))
{
mp_limb_t B1modb = -b;
B1modb *= ((bi >> (GMP_LIMB_BITS-cnt)) | (CNST_LIMB(1) << cnt));
ASSERT (B1modb <= b); /* NB: not fully reduced mod b */
cps[2] = B1modb >> cnt;
}
B2modb = - b * bi;
ASSERT (B2modb <= b); // NB: equality iff b = B/2
cps[3] = B2modb;
}
mp_limb_t
mpn_mod_1_1p (mp_srcptr ap, mp_size_t n, mp_limb_t b, const mp_limb_t bmodb[4])
{
int cnt;
mp_limb_t bi, B1modb;
mp_limb_t r0, r1;
mp_limb_t r;
ASSERT (n >= 2); /* fix tuneup.c if this is changed */
r0 = ap[n-2];
r1 = ap[n-1];
if (n > 2)
{
mp_limb_t B2modb, B2mb;
mp_limb_t p0, p1;
mp_limb_t r2;
mp_size_t j;
B2modb = bmodb[3];
B2mb = B2modb - b;
umul_ppmm (p1, p0, r1, B2modb);
add_mssaaaa (r2, r1, r0, r0, ap[n-3], p1, p0);
for (j = n-4; j >= 0; j--)
{
mp_limb_t cy;
/* mp_limb_t t = r0 + B2mb; */
umul_ppmm (p1, p0, r1, B2modb);
ADDC_LIMB (cy, r0, r0, r2 & B2modb);
/* Alternative, for cmov: if (cy) r0 = t; */
r0 -= (-cy) & b;
add_mssaaaa (r2, r1, r0, r0, ap[j], p1, p0);
}
r1 -= (r2 & b);
}
cnt = bmodb[1];
if (LIKELY (cnt != 0))
{
mp_limb_t t;
mp_limb_t B1modb = bmodb[2];
umul_ppmm (r1, t, r1, B1modb);
r0 += t;
r1 += (r0 < t);
/* Normalize */
r1 = (r1 << cnt) | (r0 >> (GMP_LIMB_BITS - cnt));
r0 <<= cnt;
/* NOTE: Might get r1 == b here, but udiv_rnnd_preinv allows that. */
}
else
{
mp_limb_t mask = -(mp_limb_t) (r1 >= b);
r1 -= mask & b;
}
bi = bmodb[0];
udiv_rnnd_preinv (r, r1, r0, b, bi);
return r >> cnt;
}
#endif /* MOD_1_1P_METHOD == 2 */
|
