Mercurial > dropbear

comparison libtommath/bn_fast_s_mp_mul_digs.c @ 284:eed26cff980b

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
propagate from branch 'au.asn.ucc.matt.ltm.dropbear' (head 6c790cad5a7fa866ad062cb3a0c279f7ba788583) to branch 'au.asn.ucc.matt.dropbear' (head fff0894a0399405a9410ea1c6d118f342cf2aa64)
author Matt Johnston <matt@ucc.asn.au>
date Wed, 08 Mar 2006 13:23:49 +0000
parents
children 5ff8218bcee9
comparison
equal deleted inserted replaced
283:bd240aa12ba7 284:eed26cff980b
1 #include <tommath.h>
2 #ifdef BN_FAST_S_MP_MUL_DIGS_C
3 /* LibTomMath, multiple-precision integer library -- Tom St Denis
4 *
5 * LibTomMath is a library that provides multiple-precision
6 * integer arithmetic as well as number theoretic functionality.
7 *
8 * The library was designed directly after the MPI library by
9 * Michael Fromberger but has been written from scratch with
10 * additional optimizations in place.
11 *
12 * The library is free for all purposes without any express
13 * guarantee it works.
14 *
15 * Tom St Denis, [email protected], http://math.libtomcrypt.org
16 */
17
18 /* Fast (comba) multiplier
19 *
20 * This is the fast column-array [comba] multiplier. It is
21 * designed to compute the columns of the product first
22 * then handle the carries afterwards. This has the effect
23 * of making the nested loops that compute the columns very
24 * simple and schedulable on super-scalar processors.
25 *
26 * This has been modified to produce a variable number of
27 * digits of output so if say only a half-product is required
28 * you don't have to compute the upper half (a feature
29 * required for fast Barrett reduction).
30 *
31 * Based on Algorithm 14.12 on pp.595 of HAC.
32 *
33 */
34 int fast_s_mp_mul_digs (mp_int * a, mp_int * b, mp_int * c, int digs)
35 {
36 int olduse, res, pa, ix, iz;
37 mp_digit W[MP_WARRAY];
38 register mp_word _W;
39
40 /* grow the destination as required */
41 if (c->alloc < digs) {
42 if ((res = mp_grow (c, digs)) != MP_OKAY) {
43 return res;
44 }
45 }
46
47 /* number of output digits to produce */
48 pa = MIN(digs, a->used + b->used);
49
50 /* clear the carry */
51 _W = 0;
52 for (ix = 0; ix < pa; ix++) {
53 int tx, ty;
54 int iy;
55 mp_digit *tmpx, *tmpy;
56
57 /* get offsets into the two bignums */
58 ty = MIN(b->used-1, ix);
59 tx = ix - ty;
60
61 /* setup temp aliases */
62 tmpx = a->dp + tx;
63 tmpy = b->dp + ty;
64
65 /* this is the number of times the loop will iterrate, essentially
66 while (tx++ < a->used && ty-- >= 0) { ... }
67 */
68 iy = MIN(a->used-tx, ty+1);
69
70 /* execute loop */
71 for (iz = 0; iz < iy; ++iz) {
72 _W += ((mp_word)*tmpx++)*((mp_word)*tmpy--);
73 }
74
75 /* store term */
76 W[ix] = ((mp_digit)_W) & MP_MASK;
77
78 /* make next carry */
79 _W = _W >> ((mp_word)DIGIT_BIT);
80 }
81
82 /* store final carry */
83 W[ix] = (mp_digit)(_W & MP_MASK);
84
85 /* setup dest */
86 olduse = c->used;
87 c->used = pa;
88
89 {
90 register mp_digit *tmpc;
91 tmpc = c->dp;
92 for (ix = 0; ix < pa+1; ix++) {
93 /* now extract the previous digit [below the carry] */
94 *tmpc++ = W[ix];
95 }
96
97 /* clear unused digits [that existed in the old copy of c] */
98 for (; ix < olduse; ix++) {
99 *tmpc++ = 0;
100 }
101 }
102 mp_clamp (c);
103 return MP_OKAY;
104 }
105 #endif