comparison libtommath/bn_s_mp_sqr.c @ 284:eed26cff980b

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_S_MP_SQR_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 /* low level squaring, b = a*a, HAC pp.596-597, Algorithm 14.16 */
19 int s_mp_sqr (mp_int * a, mp_int * b)
20 {
21 mp_int t;
22 int res, ix, iy, pa;
23 mp_word r;
24 mp_digit u, tmpx, *tmpt;
25
26 pa = a->used;
27 if ((res = mp_init_size (&t, 2*pa + 1)) != MP_OKAY) {
28 return res;
29 }
30
31 /* default used is maximum possible size */
32 t.used = 2*pa + 1;
33
34 for (ix = 0; ix < pa; ix++) {
35 /* first calculate the digit at 2*ix */
36 /* calculate double precision result */
37 r = ((mp_word) t.dp[2*ix]) +
38 ((mp_word)a->dp[ix])*((mp_word)a->dp[ix]);
39
40 /* store lower part in result */
41 t.dp[ix+ix] = (mp_digit) (r & ((mp_word) MP_MASK));
42
43 /* get the carry */
44 u = (mp_digit)(r >> ((mp_word) DIGIT_BIT));
45
46 /* left hand side of A[ix] * A[iy] */
47 tmpx = a->dp[ix];
48
49 /* alias for where to store the results */
50 tmpt = t.dp + (2*ix + 1);
51
52 for (iy = ix + 1; iy < pa; iy++) {
53 /* first calculate the product */
54 r = ((mp_word)tmpx) * ((mp_word)a->dp[iy]);
55
56 /* now calculate the double precision result, note we use
57 * addition instead of *2 since it's easier to optimize
58 */
59 r = ((mp_word) *tmpt) + r + r + ((mp_word) u);
60
61 /* store lower part */
62 *tmpt++ = (mp_digit) (r & ((mp_word) MP_MASK));
63
64 /* get carry */
65 u = (mp_digit)(r >> ((mp_word) DIGIT_BIT));
66 }
67 /* propagate upwards */
68 while (u != ((mp_digit) 0)) {
69 r = ((mp_word) *tmpt) + ((mp_word) u);
70 *tmpt++ = (mp_digit) (r & ((mp_word) MP_MASK));
71 u = (mp_digit)(r >> ((mp_word) DIGIT_BIT));
72 }
73 }
74
75 mp_clamp (&t);
76 mp_exch (&t, b);
77 mp_clear (&t);
78 return MP_OKAY;
79 }
80 #endif