michael@0: /* This Source Code Form is subject to the terms of the Mozilla Public
michael@0:  * License, v. 2.0. If a copy of the MPL was not distributed with this
michael@0:  * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
michael@0: 
michael@0: /* This file contains routines that perform vector multiplication.  */
michael@0: 
michael@0: #include "mpi-priv.h"
michael@0: #include <unistd.h>
michael@0: 
michael@0: #include <stddef.h>
michael@0: /* #include <sys/systeminfo.h> */
michael@0: #include <strings.h>
michael@0: 
michael@0: extern void multacc512( 
michael@0:    int             length,        /* doublewords in multiplicand vector. */
michael@0:    const mp_digit *scalaraddr,    /* Address of scalar. */
michael@0:    const mp_digit *multiplicand,  /* The multiplicand vector. */
michael@0:    mp_digit *      result);       /* Where to accumulate the result. */
michael@0: 
michael@0: extern void maxpy_little(
michael@0:    int             length,        /* doublewords in multiplicand vector. */
michael@0:    const mp_digit *scalaraddr,    /* Address of scalar. */
michael@0:    const mp_digit *multiplicand,  /* The multiplicand vector. */
michael@0:    mp_digit *      result);       /* Where to accumulate the result. */
michael@0: 
michael@0: extern void add_diag_little(
michael@0:    int            length,       /* doublewords in input vector. */
michael@0:    const mp_digit *root,         /* The vector to square. */
michael@0:    mp_digit *      result);      /* Where to accumulate the result. */
michael@0: 
michael@0: void 
michael@0: s_mpv_sqr_add_prop(const mp_digit *pa, mp_size a_len, mp_digit *ps)
michael@0: {
michael@0:     add_diag_little(a_len, pa, ps);
michael@0: }
michael@0: 
michael@0: #define MAX_STACK_DIGITS 258
michael@0: #define MULTACC512_LEN   (512 / MP_DIGIT_BIT)
michael@0: #define HP_MPY_ADD_FN    (a_len == MULTACC512_LEN ? multacc512 : maxpy_little)
michael@0: 
michael@0: /* c = a * b */
michael@0: void 
michael@0: s_mpv_mul_d(const mp_digit *a, mp_size a_len, mp_digit b, mp_digit *c)
michael@0: {
michael@0:     mp_digit x[MAX_STACK_DIGITS];
michael@0:     mp_digit *px = x;
michael@0:     size_t   xSize = 0;
michael@0: 
michael@0:     if (a == c) {
michael@0: 	if (a_len > MAX_STACK_DIGITS) {
michael@0: 	    xSize = sizeof(mp_digit) * (a_len + 2);
michael@0: 	    px = malloc(xSize);
michael@0: 	    if (!px)
michael@0: 		return;
michael@0: 	}
michael@0: 	memcpy(px, a, a_len * sizeof(*a));
michael@0: 	a = px;
michael@0:     }
michael@0:     s_mp_setz(c, a_len + 1);
michael@0:     HP_MPY_ADD_FN(a_len, &b, a, c);
michael@0:     if (px != x && px) {
michael@0: 	memset(px, 0, xSize);
michael@0: 	free(px);
michael@0:     }
michael@0: }
michael@0: 
michael@0: /* c += a * b, where a is a_len words long. */
michael@0: void     
michael@0: s_mpv_mul_d_add(const mp_digit *a, mp_size a_len, mp_digit b, mp_digit *c)
michael@0: {
michael@0:     c[a_len] = 0;	/* so carry propagation stops here. */
michael@0:     HP_MPY_ADD_FN(a_len, &b, a, c);
michael@0: }
michael@0: 
michael@0: /* c += a * b, where a is y words long. */
michael@0: void     
michael@0: s_mpv_mul_d_add_prop(const mp_digit *a, mp_size a_len, mp_digit b, 
michael@0: 			 mp_digit *c)
michael@0: {
michael@0:     HP_MPY_ADD_FN(a_len, &b, a, c);
michael@0: }
michael@0: