Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
1// SPDX-License-Identifier: GPL-2.0 OR MIT
2/*
3 * Copyright (C) 2015-2019 Jason A. Donenfeld <Jason@zx2c4.com>. All Rights Reserved.
4 * Copyright 2025 Google LLC
5 *
6 * This is an implementation of the BLAKE2b hash and PRF functions.
7 *
8 * Information: https://blake2.net/
9 */
10
11#include <crypto/blake2b.h>
12#include <linux/bug.h>
13#include <linux/export.h>
14#include <linux/kernel.h>
15#include <linux/module.h>
16#include <linux/string.h>
17#include <linux/unroll.h>
18#include <linux/types.h>
19
20static const u8 blake2b_sigma[12][16] = {
21 { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 },
22 { 14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 },
23 { 11, 8, 12, 0, 5, 2, 15, 13, 10, 14, 3, 6, 7, 1, 9, 4 },
24 { 7, 9, 3, 1, 13, 12, 11, 14, 2, 6, 5, 10, 4, 0, 15, 8 },
25 { 9, 0, 5, 7, 2, 4, 10, 15, 14, 1, 11, 12, 6, 8, 3, 13 },
26 { 2, 12, 6, 10, 0, 11, 8, 3, 4, 13, 7, 5, 15, 14, 1, 9 },
27 { 12, 5, 1, 15, 14, 13, 4, 10, 0, 7, 6, 3, 9, 2, 8, 11 },
28 { 13, 11, 7, 14, 12, 1, 3, 9, 5, 0, 15, 4, 8, 6, 2, 10 },
29 { 6, 15, 14, 9, 11, 3, 0, 8, 12, 2, 13, 7, 1, 4, 10, 5 },
30 { 10, 2, 8, 4, 7, 6, 1, 5, 15, 11, 9, 14, 3, 12, 13, 0 },
31 { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 },
32 { 14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 }
33};
34
35static inline void blake2b_increment_counter(struct blake2b_ctx *ctx, u32 inc)
36{
37 ctx->t[0] += inc;
38 ctx->t[1] += (ctx->t[0] < inc);
39}
40
41static void __maybe_unused
42blake2b_compress_generic(struct blake2b_ctx *ctx,
43 const u8 *data, size_t nblocks, u32 inc)
44{
45 u64 m[16];
46 u64 v[16];
47 int i;
48
49 WARN_ON(IS_ENABLED(DEBUG) &&
50 (nblocks > 1 && inc != BLAKE2B_BLOCK_SIZE));
51
52 while (nblocks > 0) {
53 blake2b_increment_counter(ctx, inc);
54 memcpy(m, data, BLAKE2B_BLOCK_SIZE);
55 le64_to_cpu_array(m, ARRAY_SIZE(m));
56 memcpy(v, ctx->h, 64);
57 v[ 8] = BLAKE2B_IV0;
58 v[ 9] = BLAKE2B_IV1;
59 v[10] = BLAKE2B_IV2;
60 v[11] = BLAKE2B_IV3;
61 v[12] = BLAKE2B_IV4 ^ ctx->t[0];
62 v[13] = BLAKE2B_IV5 ^ ctx->t[1];
63 v[14] = BLAKE2B_IV6 ^ ctx->f[0];
64 v[15] = BLAKE2B_IV7 ^ ctx->f[1];
65
66#define G(r, i, a, b, c, d) do { \
67 a += b + m[blake2b_sigma[r][2 * i + 0]]; \
68 d = ror64(d ^ a, 32); \
69 c += d; \
70 b = ror64(b ^ c, 24); \
71 a += b + m[blake2b_sigma[r][2 * i + 1]]; \
72 d = ror64(d ^ a, 16); \
73 c += d; \
74 b = ror64(b ^ c, 63); \
75} while (0)
76
77#ifdef CONFIG_64BIT
78 /*
79 * Unroll the rounds loop to enable constant-folding of the
80 * blake2b_sigma values. Seems worthwhile on 64-bit kernels.
81 * Not worthwhile on 32-bit kernels because the code size is
82 * already so large there due to BLAKE2b using 64-bit words.
83 */
84 unrolled_full
85#endif
86 for (int r = 0; r < 12; r++) {
87 G(r, 0, v[0], v[4], v[8], v[12]);
88 G(r, 1, v[1], v[5], v[9], v[13]);
89 G(r, 2, v[2], v[6], v[10], v[14]);
90 G(r, 3, v[3], v[7], v[11], v[15]);
91 G(r, 4, v[0], v[5], v[10], v[15]);
92 G(r, 5, v[1], v[6], v[11], v[12]);
93 G(r, 6, v[2], v[7], v[8], v[13]);
94 G(r, 7, v[3], v[4], v[9], v[14]);
95 }
96#undef G
97
98 for (i = 0; i < 8; ++i)
99 ctx->h[i] ^= v[i] ^ v[i + 8];
100
101 data += BLAKE2B_BLOCK_SIZE;
102 --nblocks;
103 }
104}
105
106#ifdef CONFIG_CRYPTO_LIB_BLAKE2B_ARCH
107#include "blake2b.h" /* $(SRCARCH)/blake2b.h */
108#else
109#define blake2b_compress blake2b_compress_generic
110#endif
111
112static inline void blake2b_set_lastblock(struct blake2b_ctx *ctx)
113{
114 ctx->f[0] = -1;
115}
116
117void blake2b_update(struct blake2b_ctx *ctx, const u8 *in, size_t inlen)
118{
119 const size_t fill = BLAKE2B_BLOCK_SIZE - ctx->buflen;
120
121 if (unlikely(!inlen))
122 return;
123 if (inlen > fill) {
124 memcpy(ctx->buf + ctx->buflen, in, fill);
125 blake2b_compress(ctx, ctx->buf, 1, BLAKE2B_BLOCK_SIZE);
126 ctx->buflen = 0;
127 in += fill;
128 inlen -= fill;
129 }
130 if (inlen > BLAKE2B_BLOCK_SIZE) {
131 const size_t nblocks = DIV_ROUND_UP(inlen, BLAKE2B_BLOCK_SIZE);
132
133 blake2b_compress(ctx, in, nblocks - 1, BLAKE2B_BLOCK_SIZE);
134 in += BLAKE2B_BLOCK_SIZE * (nblocks - 1);
135 inlen -= BLAKE2B_BLOCK_SIZE * (nblocks - 1);
136 }
137 memcpy(ctx->buf + ctx->buflen, in, inlen);
138 ctx->buflen += inlen;
139}
140EXPORT_SYMBOL(blake2b_update);
141
142void blake2b_final(struct blake2b_ctx *ctx, u8 *out)
143{
144 WARN_ON(IS_ENABLED(DEBUG) && !out);
145 blake2b_set_lastblock(ctx);
146 memset(ctx->buf + ctx->buflen, 0,
147 BLAKE2B_BLOCK_SIZE - ctx->buflen); /* Padding */
148 blake2b_compress(ctx, ctx->buf, 1, ctx->buflen);
149 cpu_to_le64_array(ctx->h, ARRAY_SIZE(ctx->h));
150 memcpy(out, ctx->h, ctx->outlen);
151 memzero_explicit(ctx, sizeof(*ctx));
152}
153EXPORT_SYMBOL(blake2b_final);
154
155#ifdef blake2b_mod_init_arch
156static int __init blake2b_mod_init(void)
157{
158 blake2b_mod_init_arch();
159 return 0;
160}
161subsys_initcall(blake2b_mod_init);
162
163static void __exit blake2b_mod_exit(void)
164{
165}
166module_exit(blake2b_mod_exit);
167#endif
168
169MODULE_DESCRIPTION("BLAKE2b hash function");
170MODULE_LICENSE("GPL");