xref: /netbsd-src/crypto/external/bsd/heimdal/dist/lib/hcrypto/libtommath/bn_s_mp_sqr.c (revision d3273b5b76f5afaafe308cead5511dbb8df8c5e9)
1 /*	$NetBSD: bn_s_mp_sqr.c,v 1.2 2017/01/28 21:31:47 christos Exp $	*/
2 
3 #include <tommath.h>
4 #ifdef BN_S_MP_SQR_C
5 /* LibTomMath, multiple-precision integer library -- Tom St Denis
6  *
7  * LibTomMath is a library that provides multiple-precision
8  * integer arithmetic as well as number theoretic functionality.
9  *
10  * The library was designed directly after the MPI library by
11  * Michael Fromberger but has been written from scratch with
12  * additional optimizations in place.
13  *
14  * The library is free for all purposes without any express
15  * guarantee it works.
16  *
17  * Tom St Denis, tomstdenis@gmail.com, http://libtom.org
18  */
19 
20 /* low level squaring, b = a*a, HAC pp.596-597, Algorithm 14.16 */
s_mp_sqr(mp_int * a,mp_int * b)21 int s_mp_sqr (mp_int * a, mp_int * b)
22 {
23   mp_int  t;
24   int     res, ix, iy, pa;
25   mp_word r;
26   mp_digit u, tmpx, *tmpt;
27 
28   pa = a->used;
29   if ((res = mp_init_size (&t, 2*pa + 1)) != MP_OKAY) {
30     return res;
31   }
32 
33   /* default used is maximum possible size */
34   t.used = 2*pa + 1;
35 
36   for (ix = 0; ix < pa; ix++) {
37     /* first calculate the digit at 2*ix */
38     /* calculate double precision result */
39     r = ((mp_word) t.dp[2*ix]) +
40         ((mp_word)a->dp[ix])*((mp_word)a->dp[ix]);
41 
42     /* store lower part in result */
43     t.dp[ix+ix] = (mp_digit) (r & ((mp_word) MP_MASK));
44 
45     /* get the carry */
46     u           = (mp_digit)(r >> ((mp_word) DIGIT_BIT));
47 
48     /* left hand side of A[ix] * A[iy] */
49     tmpx        = a->dp[ix];
50 
51     /* alias for where to store the results */
52     tmpt        = t.dp + (2*ix + 1);
53 
54     for (iy = ix + 1; iy < pa; iy++) {
55       /* first calculate the product */
56       r       = ((mp_word)tmpx) * ((mp_word)a->dp[iy]);
57 
58       /* now calculate the double precision result, note we use
59        * addition instead of *2 since it's easier to optimize
60        */
61       r       = ((mp_word) *tmpt) + r + r + ((mp_word) u);
62 
63       /* store lower part */
64       *tmpt++ = (mp_digit) (r & ((mp_word) MP_MASK));
65 
66       /* get carry */
67       u       = (mp_digit)(r >> ((mp_word) DIGIT_BIT));
68     }
69     /* propagate upwards */
70     while (u != ((mp_digit) 0)) {
71       r       = ((mp_word) *tmpt) + ((mp_word) u);
72       *tmpt++ = (mp_digit) (r & ((mp_word) MP_MASK));
73       u       = (mp_digit)(r >> ((mp_word) DIGIT_BIT));
74     }
75   }
76 
77   mp_clamp (&t);
78   mp_exch (&t, b);
79   mp_clear (&t);
80   return MP_OKAY;
81 }
82 #endif
83 
84 /* Source: /cvs/libtom/libtommath/bn_s_mp_sqr.c,v  */
85 /* Revision: 1.4  */
86 /* Date: 2006/12/28 01:25:13  */
87