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
|
/* mpn_brootinv, compute r such that r^k * y = 1 (mod 2^b).
Contributed to the GNU project by Martin Boij (as part of perfpow.c).
Copyright 2009, 2010, 2012, 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"
/* Computes a^e (mod B). Uses right-to-left binary algorithm, since
typical use will have e small. */
static mp_limb_t
powlimb (mp_limb_t a, mp_limb_t e)
{
mp_limb_t r;
for (r = 1; e > 0; e >>= 1, a *= a)
if (e & 1)
r *= a;
return r;
}
/* Compute r such that r^k * y = 1 (mod B^n).
Iterates
r' <-- k^{-1} ((k+1) r - r^{k+1} y) (mod 2^b)
using Hensel lifting, each time doubling the number of known bits in r.
Works just for odd k. Else the Hensel lifting degenerates.
FIXME:
(1) Make it work for k == GMP_LIMB_MAX (k+1 below overflows).
(2) Rewrite iteration as
r' <-- r - k^{-1} r (r^k y - 1)
and take advantage of the zero low part of r^k y - 1.
(3) Use wrap-around trick.
(4) Use a small table to get starting value.
Scratch need: 5*bn, where bn = ceil (bnb / GMP_NUMB_BITS).
*/
void
mpn_brootinv (mp_ptr rp, mp_srcptr yp, mp_size_t bn, mp_limb_t k, mp_ptr tp)
{
mp_ptr tp2, tp3;
mp_limb_t kinv, k2, r0, y0;
mp_size_t order[GMP_LIMB_BITS + 1];
int i, d;
ASSERT (bn > 0);
ASSERT ((k & 1) != 0);
tp2 = tp + bn;
tp3 = tp + 2 * bn;
k2 = k + 1;
binvert_limb (kinv, k);
/* 4-bit initial approximation:
y%16 | 1 3 5 7 9 11 13 15,
k%4 +-------------------------+k2%4
1 | 1 11 13 7 9 3 5 15 | 2
3 | 1 3 5 7 9 11 13 15 | 0
*/
y0 = yp[0];
r0 = y0 ^ (((y0 << 1) ^ (y0 << 2)) & (k2 << 2) & 8); /* 4 bits */
r0 = kinv * (k2 * r0 - y0 * powlimb(r0, k2 & 0x7f)); /* 8 bits */
r0 = kinv * (k2 * r0 - y0 * powlimb(r0, k2 & 0x7fff)); /* 16 bits */
#if GMP_NUMB_BITS > 16
{
unsigned prec = 16;
do
{
r0 = kinv * (k2 * r0 - y0 * powlimb(r0, k2));
prec *= 2;
}
while (prec < GMP_NUMB_BITS);
}
#endif
rp[0] = r0;
if (bn == 1)
return;
/* This initialization doesn't matter for the result (any garbage is
cancelled in the iteration), but proper initialization makes
valgrind happier. */
MPN_ZERO (rp+1, bn-1);
d = 0;
for (; bn > 1; bn = (bn + 1) >> 1)
order[d++] = bn;
for (i = d - 1; i >= 0; i--)
{
bn = order[i];
mpn_mul_1 (tp, rp, bn, k2);
mpn_powlo (tp2, rp, &k2, 1, bn, tp3);
mpn_mullo_n (rp, yp, tp2, bn);
mpn_sub_n (tp2, tp, rp, bn);
mpn_pi1_bdiv_q_1 (rp, tp2, bn, k, kinv, 0);
}
}
|
