1// SPDX-License-Identifier: GPL-2.0-or-later
  2/*
  3 * The "hash function" used as the core of the ChaCha stream cipher (RFC7539)
  4 *
  5 * Copyright (C) 2015 Martin Willi
  6 */
  7
  8#include <linux/kernel.h>
  9#include <linux/export.h>
 10#include <linux/bitops.h>
 11#include <linux/cryptohash.h>
 12#include <asm/unaligned.h>
 13#include <crypto/chacha.h>
 14
 15static void chacha_permute(u32 *x, int nrounds)
 16{
 17	int i;
 18
 19	/* whitelist the allowed round counts */
 20	WARN_ON_ONCE(nrounds != 20 && nrounds != 12);
 21
 22	for (i = 0; i < nrounds; i += 2) {
 23		x[0]  += x[4];    x[12] = rol32(x[12] ^ x[0],  16);
 24		x[1]  += x[5];    x[13] = rol32(x[13] ^ x[1],  16);
 25		x[2]  += x[6];    x[14] = rol32(x[14] ^ x[2],  16);
 26		x[3]  += x[7];    x[15] = rol32(x[15] ^ x[3],  16);
 27
 28		x[8]  += x[12];   x[4]  = rol32(x[4]  ^ x[8],  12);
 29		x[9]  += x[13];   x[5]  = rol32(x[5]  ^ x[9],  12);
 30		x[10] += x[14];   x[6]  = rol32(x[6]  ^ x[10], 12);
 31		x[11] += x[15];   x[7]  = rol32(x[7]  ^ x[11], 12);
 32
 33		x[0]  += x[4];    x[12] = rol32(x[12] ^ x[0],   8);
 34		x[1]  += x[5];    x[13] = rol32(x[13] ^ x[1],   8);
 35		x[2]  += x[6];    x[14] = rol32(x[14] ^ x[2],   8);
 36		x[3]  += x[7];    x[15] = rol32(x[15] ^ x[3],   8);
 37
 38		x[8]  += x[12];   x[4]  = rol32(x[4]  ^ x[8],   7);
 39		x[9]  += x[13];   x[5]  = rol32(x[5]  ^ x[9],   7);
 40		x[10] += x[14];   x[6]  = rol32(x[6]  ^ x[10],  7);
 41		x[11] += x[15];   x[7]  = rol32(x[7]  ^ x[11],  7);
 42
 43		x[0]  += x[5];    x[15] = rol32(x[15] ^ x[0],  16);
 44		x[1]  += x[6];    x[12] = rol32(x[12] ^ x[1],  16);
 45		x[2]  += x[7];    x[13] = rol32(x[13] ^ x[2],  16);
 46		x[3]  += x[4];    x[14] = rol32(x[14] ^ x[3],  16);
 47
 48		x[10] += x[15];   x[5]  = rol32(x[5]  ^ x[10], 12);
 49		x[11] += x[12];   x[6]  = rol32(x[6]  ^ x[11], 12);
 50		x[8]  += x[13];   x[7]  = rol32(x[7]  ^ x[8],  12);
 51		x[9]  += x[14];   x[4]  = rol32(x[4]  ^ x[9],  12);
 52
 53		x[0]  += x[5];    x[15] = rol32(x[15] ^ x[0],   8);
 54		x[1]  += x[6];    x[12] = rol32(x[12] ^ x[1],   8);
 55		x[2]  += x[7];    x[13] = rol32(x[13] ^ x[2],   8);
 56		x[3]  += x[4];    x[14] = rol32(x[14] ^ x[3],   8);
 57
 58		x[10] += x[15];   x[5]  = rol32(x[5]  ^ x[10],  7);
 59		x[11] += x[12];   x[6]  = rol32(x[6]  ^ x[11],  7);
 60		x[8]  += x[13];   x[7]  = rol32(x[7]  ^ x[8],   7);
 61		x[9]  += x[14];   x[4]  = rol32(x[4]  ^ x[9],   7);
 62	}
 63}
 64
 65/**
 66 * chacha_block - generate one keystream block and increment block counter
 67 * @state: input state matrix (16 32-bit words)
 68 * @stream: output keystream block (64 bytes)
 69 * @nrounds: number of rounds (20 or 12; 20 is recommended)
 70 *
 71 * This is the ChaCha core, a function from 64-byte strings to 64-byte strings.
 72 * The caller has already converted the endianness of the input.  This function
 73 * also handles incrementing the block counter in the input matrix.
 74 */
 75void chacha_block(u32 *state, u8 *stream, int nrounds)
 76{
 77	u32 x[16];
 78	int i;
 79
 80	memcpy(x, state, 64);
 81
 82	chacha_permute(x, nrounds);
 83
 84	for (i = 0; i < ARRAY_SIZE(x); i++)
 85		put_unaligned_le32(x[i] + state[i], &stream[i * sizeof(u32)]);
 86
 87	state[12]++;
 88}
 89EXPORT_SYMBOL(chacha_block);
 90
 91/**
 92 * hchacha_block - abbreviated ChaCha core, for XChaCha
 93 * @in: input state matrix (16 32-bit words)
 94 * @out: output (8 32-bit words)
 95 * @nrounds: number of rounds (20 or 12; 20 is recommended)
 96 *
 97 * HChaCha is the ChaCha equivalent of HSalsa and is an intermediate step
 98 * towards XChaCha (see https://cr.yp.to/snuffle/xsalsa-20081128.pdf).  HChaCha
 99 * skips the final addition of the initial state, and outputs only certain words
100 * of the state.  It should not be used for streaming directly.
101 */
102void hchacha_block(const u32 *in, u32 *out, int nrounds)
103{
104	u32 x[16];
105
106	memcpy(x, in, 64);
107
108	chacha_permute(x, nrounds);
109
110	memcpy(&out[0], &x[0], 16);
111	memcpy(&out[4], &x[12], 16);
112}
113EXPORT_SYMBOL(hchacha_block);