1/*
2 * Copyright (c) 2016 Thomas Pornin <pornin@bolet.org>
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining
5 * a copy of this software and associated documentation files (the
6 * "Software"), to deal in the Software without restriction, including
7 * without limitation the rights to use, copy, modify, merge, publish,
8 * distribute, sublicense, and/or sell copies of the Software, and to
9 * permit persons to whom the Software is furnished to do so, subject to
10 * the following conditions:
11 *
12 * The above copyright notice and this permission notice shall be
13 * included in all copies or substantial portions of the Software.
14 *
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
16 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
17 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
18 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
19 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
20 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
21 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
22 * SOFTWARE.
23 */
24
25#include "inner.h"
26
27/* see inner.h */
28void
29br_i31_mulacc(uint32_t *d, const uint32_t *a, const uint32_t *b)
30{
31	size_t alen, blen, u;
32	uint32_t dl, dh;
33
34	alen = (a[0] + 31) >> 5;
35	blen = (b[0] + 31) >> 5;
36
37	/*
38	 * We want to add the two bit lengths, but these are encoded,
39	 * which requires some extra care.
40	 */
41	dl = (a[0] & 31) + (b[0] & 31);
42	dh = (a[0] >> 5) + (b[0] >> 5);
43	d[0] = (dh << 5) + dl + (~(uint32_t)(dl - 31) >> 31);
44
45	for (u = 0; u < blen; u ++) {
46		uint32_t f;
47		size_t v;
48
49		/*
50		 * Carry always fits on 31 bits; we want to keep it in a
51		 * 32-bit register on 32-bit architectures (on a 64-bit
52		 * architecture, cast down from 64 to 32 bits means
53		 * clearing the high bits, which is not free; on a 32-bit
54		 * architecture, the same operation really means ignoring
55		 * the top register, which has negative or zero cost).
56		 */
57#if BR_64
58		uint64_t cc;
59#else
60		uint32_t cc;
61#endif
62
63		f = b[1 + u];
64		cc = 0;
65		for (v = 0; v < alen; v ++) {
66			uint64_t z;
67
68			z = (uint64_t)d[1 + u + v] + MUL31(f, a[1 + v]) + cc;
69			cc = z >> 31;
70			d[1 + u + v] = (uint32_t)z & 0x7FFFFFFF;
71		}
72		d[1 + u + alen] = (uint32_t)cc;
73	}
74}
75