Mercurial > dropbear
comparison libtommath/bn_s_mp_mul_digs_fast.c @ 1692:1051e4eea25a
Update LibTomMath to 1.2.0 (#84)
* update C files
* update other files
* update headers
* update makefiles
* remove mp_set/get_double()
* use ltm 1.2.0 API
* update ltm_desc
* use bundled tommath if system-tommath is too old
* XMALLOC etc. were changed to MP_MALLOC etc.
author | Steffen Jaeckel <s@jaeckel.eu> |
---|---|
date | Tue, 26 May 2020 17:36:47 +0200 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
1691:2d3745d58843 | 1692:1051e4eea25a |
---|---|
1 #include "tommath_private.h" | |
2 #ifdef BN_S_MP_MUL_DIGS_FAST_C | |
3 /* LibTomMath, multiple-precision integer library -- Tom St Denis */ | |
4 /* SPDX-License-Identifier: Unlicense */ | |
5 | |
6 /* Fast (comba) multiplier | |
7 * | |
8 * This is the fast column-array [comba] multiplier. It is | |
9 * designed to compute the columns of the product first | |
10 * then handle the carries afterwards. This has the effect | |
11 * of making the nested loops that compute the columns very | |
12 * simple and schedulable on super-scalar processors. | |
13 * | |
14 * This has been modified to produce a variable number of | |
15 * digits of output so if say only a half-product is required | |
16 * you don't have to compute the upper half (a feature | |
17 * required for fast Barrett reduction). | |
18 * | |
19 * Based on Algorithm 14.12 on pp.595 of HAC. | |
20 * | |
21 */ | |
22 mp_err s_mp_mul_digs_fast(const mp_int *a, const mp_int *b, mp_int *c, int digs) | |
23 { | |
24 int olduse, pa, ix, iz; | |
25 mp_err err; | |
26 mp_digit W[MP_WARRAY]; | |
27 mp_word _W; | |
28 | |
29 /* grow the destination as required */ | |
30 if (c->alloc < digs) { | |
31 if ((err = mp_grow(c, digs)) != MP_OKAY) { | |
32 return err; | |
33 } | |
34 } | |
35 | |
36 /* number of output digits to produce */ | |
37 pa = MP_MIN(digs, a->used + b->used); | |
38 | |
39 /* clear the carry */ | |
40 _W = 0; | |
41 for (ix = 0; ix < pa; ix++) { | |
42 int tx, ty; | |
43 int iy; | |
44 mp_digit *tmpx, *tmpy; | |
45 | |
46 /* get offsets into the two bignums */ | |
47 ty = MP_MIN(b->used-1, ix); | |
48 tx = ix - ty; | |
49 | |
50 /* setup temp aliases */ | |
51 tmpx = a->dp + tx; | |
52 tmpy = b->dp + ty; | |
53 | |
54 /* this is the number of times the loop will iterrate, essentially | |
55 while (tx++ < a->used && ty-- >= 0) { ... } | |
56 */ | |
57 iy = MP_MIN(a->used-tx, ty+1); | |
58 | |
59 /* execute loop */ | |
60 for (iz = 0; iz < iy; ++iz) { | |
61 _W += (mp_word)*tmpx++ * (mp_word)*tmpy--; | |
62 | |
63 } | |
64 | |
65 /* store term */ | |
66 W[ix] = (mp_digit)_W & MP_MASK; | |
67 | |
68 /* make next carry */ | |
69 _W = _W >> (mp_word)MP_DIGIT_BIT; | |
70 } | |
71 | |
72 /* setup dest */ | |
73 olduse = c->used; | |
74 c->used = pa; | |
75 | |
76 { | |
77 mp_digit *tmpc; | |
78 tmpc = c->dp; | |
79 for (ix = 0; ix < pa; ix++) { | |
80 /* now extract the previous digit [below the carry] */ | |
81 *tmpc++ = W[ix]; | |
82 } | |
83 | |
84 /* clear unused digits [that existed in the old copy of c] */ | |
85 MP_ZERO_DIGITS(tmpc, olduse - ix); | |
86 } | |
87 mp_clamp(c); | |
88 return MP_OKAY; | |
89 } | |
90 #endif |