1/*
 2 * ChaCha20 256-bit cipher algorithm, RFC7539
 3 *
 4 * Copyright (C) 2015 Martin Willi
 5 *
 6 * This program is free software; you can redistribute it and/or modify
 7 * it under the terms of the GNU General Public License as published by
 8 * the Free Software Foundation; either version 2 of the License, or
 9 * (at your option) any later version.
10 */
11
12#include <linux/kernel.h>
13#include <linux/export.h>
14#include <linux/bitops.h>
15#include <linux/cryptohash.h>
16#include <asm/unaligned.h>
17#include <crypto/chacha20.h>
18
19static inline u32 rotl32(u32 v, u8 n)
20{
21	return (v << n) | (v >> (sizeof(v) * 8 - n));
22}
23
24extern void chacha20_block(u32 *state, void *stream)
25{
26	u32 x[16], *out = stream;
27	int i;
28
29	for (i = 0; i < ARRAY_SIZE(x); i++)
30		x[i] = state[i];
31
32	for (i = 0; i < 20; i += 2) {
33		x[0]  += x[4];    x[12] = rotl32(x[12] ^ x[0],  16);
34		x[1]  += x[5];    x[13] = rotl32(x[13] ^ x[1],  16);
35		x[2]  += x[6];    x[14] = rotl32(x[14] ^ x[2],  16);
36		x[3]  += x[7];    x[15] = rotl32(x[15] ^ x[3],  16);
37
38		x[8]  += x[12];   x[4]  = rotl32(x[4]  ^ x[8],  12);
39		x[9]  += x[13];   x[5]  = rotl32(x[5]  ^ x[9],  12);
40		x[10] += x[14];   x[6]  = rotl32(x[6]  ^ x[10], 12);
41		x[11] += x[15];   x[7]  = rotl32(x[7]  ^ x[11], 12);
42
43		x[0]  += x[4];    x[12] = rotl32(x[12] ^ x[0],   8);
44		x[1]  += x[5];    x[13] = rotl32(x[13] ^ x[1],   8);
45		x[2]  += x[6];    x[14] = rotl32(x[14] ^ x[2],   8);
46		x[3]  += x[7];    x[15] = rotl32(x[15] ^ x[3],   8);
47
48		x[8]  += x[12];   x[4]  = rotl32(x[4]  ^ x[8],   7);
49		x[9]  += x[13];   x[5]  = rotl32(x[5]  ^ x[9],   7);
50		x[10] += x[14];   x[6]  = rotl32(x[6]  ^ x[10],  7);
51		x[11] += x[15];   x[7]  = rotl32(x[7]  ^ x[11],  7);
52
53		x[0]  += x[5];    x[15] = rotl32(x[15] ^ x[0],  16);
54		x[1]  += x[6];    x[12] = rotl32(x[12] ^ x[1],  16);
55		x[2]  += x[7];    x[13] = rotl32(x[13] ^ x[2],  16);
56		x[3]  += x[4];    x[14] = rotl32(x[14] ^ x[3],  16);
57
58		x[10] += x[15];   x[5]  = rotl32(x[5]  ^ x[10], 12);
59		x[11] += x[12];   x[6]  = rotl32(x[6]  ^ x[11], 12);
60		x[8]  += x[13];   x[7]  = rotl32(x[7]  ^ x[8],  12);
61		x[9]  += x[14];   x[4]  = rotl32(x[4]  ^ x[9],  12);
62
63		x[0]  += x[5];    x[15] = rotl32(x[15] ^ x[0],   8);
64		x[1]  += x[6];    x[12] = rotl32(x[12] ^ x[1],   8);
65		x[2]  += x[7];    x[13] = rotl32(x[13] ^ x[2],   8);
66		x[3]  += x[4];    x[14] = rotl32(x[14] ^ x[3],   8);
67
68		x[10] += x[15];   x[5]  = rotl32(x[5]  ^ x[10],  7);
69		x[11] += x[12];   x[6]  = rotl32(x[6]  ^ x[11],  7);
70		x[8]  += x[13];   x[7]  = rotl32(x[7]  ^ x[8],   7);
71		x[9]  += x[14];   x[4]  = rotl32(x[4]  ^ x[9],   7);
72	}
73
74	for (i = 0; i < ARRAY_SIZE(x); i++)
75		out[i] = cpu_to_le32(x[i] + state[i]);
76
77	state[12]++;
78}
79EXPORT_SYMBOL(chacha20_block);