1 // SPDX-License-Identifier: BSD-2-Clause
2 /*
3 * Copyright Copyright (c) 2019 Huawei Technologies Co., Ltd
4 */
5 /*
6 * SM4 Encryption algorithm (SMS4 algorithm)
7 * GM/T 0002-2012 Chinese National Standard ref:http://www.oscca.gov.cn/
8 * thanks to Xyssl
9 * thnaks and refers to http://hi.baidu.com/numax/blog/item/80addfefddfb93e4cf1b3e61.html
10 * author:goldboar
11 * email:goldboar@163.com
12 * 2012-4-20
13 */
14
15 #include "sm4.h"
16 #include <assert.h>
17 #include <string.h>
18
19 #define GET_UINT32_BE(n, b, i) \
20 do { \
21 (n) = ((uint32_t)(b)[(i)] << 24) | \
22 ((uint32_t)(b)[(i) + 1] << 16) | \
23 ((uint32_t)(b)[(i) + 2] << 8) | \
24 ((uint32_t)(b)[(i) + 3]); \
25 } while (0)
26
27 #define PUT_UINT32_BE(n, b, i) \
28 do { \
29 (b)[(i)] = (uint8_t)((n) >> 24); \
30 (b)[(i) + 1] = (uint8_t)((n) >> 16); \
31 (b)[(i) + 2] = (uint8_t)((n) >> 8); \
32 (b)[(i) + 3] = (uint8_t)((n)); \
33 } while (0)
34
35 #define SHL(x, n) (((x) & 0xFFFFFFFF) << (n))
36 #define ROTL(x, n) (SHL((x), (n)) | ((x) >> (32 - (n))))
37
38 #define SWAP(a, b) { uint32_t t = a; a = b; b = t; t = 0; }
39
40 /*
41 * Expanded SM4 S-boxes
42 */
43 static const uint8_t SboxTable[16][16] = {
44 {0xd6, 0x90, 0xe9, 0xfe, 0xcc, 0xe1, 0x3d, 0xb7,
45 0x16, 0xb6, 0x14, 0xc2, 0x28, 0xfb, 0x2c, 0x05},
46 {0x2b, 0x67, 0x9a, 0x76, 0x2a, 0xbe, 0x04, 0xc3,
47 0xaa, 0x44, 0x13, 0x26, 0x49, 0x86, 0x06, 0x99},
48 {0x9c, 0x42, 0x50, 0xf4, 0x91, 0xef, 0x98, 0x7a,
49 0x33, 0x54, 0x0b, 0x43, 0xed, 0xcf, 0xac, 0x62},
50 {0xe4, 0xb3, 0x1c, 0xa9, 0xc9, 0x08, 0xe8, 0x95,
51 0x80, 0xdf, 0x94, 0xfa, 0x75, 0x8f, 0x3f, 0xa6},
52 {0x47, 0x07, 0xa7, 0xfc, 0xf3, 0x73, 0x17, 0xba,
53 0x83, 0x59, 0x3c, 0x19, 0xe6, 0x85, 0x4f, 0xa8},
54 {0x68, 0x6b, 0x81, 0xb2, 0x71, 0x64, 0xda, 0x8b,
55 0xf8, 0xeb, 0x0f, 0x4b, 0x70, 0x56, 0x9d, 0x35},
56 {0x1e, 0x24, 0x0e, 0x5e, 0x63, 0x58, 0xd1, 0xa2,
57 0x25, 0x22, 0x7c, 0x3b, 0x01, 0x21, 0x78, 0x87},
58 {0xd4, 0x00, 0x46, 0x57, 0x9f, 0xd3, 0x27, 0x52,
59 0x4c, 0x36, 0x02, 0xe7, 0xa0, 0xc4, 0xc8, 0x9e},
60 {0xea, 0xbf, 0x8a, 0xd2, 0x40, 0xc7, 0x38, 0xb5,
61 0xa3, 0xf7, 0xf2, 0xce, 0xf9, 0x61, 0x15, 0xa1},
62 {0xe0, 0xae, 0x5d, 0xa4, 0x9b, 0x34, 0x1a, 0x55,
63 0xad, 0x93, 0x32, 0x30, 0xf5, 0x8c, 0xb1, 0xe3},
64 {0x1d, 0xf6, 0xe2, 0x2e, 0x82, 0x66, 0xca, 0x60,
65 0xc0, 0x29, 0x23, 0xab, 0x0d, 0x53, 0x4e, 0x6f},
66 {0xd5, 0xdb, 0x37, 0x45, 0xde, 0xfd, 0x8e, 0x2f,
67 0x03, 0xff, 0x6a, 0x72, 0x6d, 0x6c, 0x5b, 0x51},
68 {0x8d, 0x1b, 0xaf, 0x92, 0xbb, 0xdd, 0xbc, 0x7f,
69 0x11, 0xd9, 0x5c, 0x41, 0x1f, 0x10, 0x5a, 0xd8},
70 {0x0a, 0xc1, 0x31, 0x88, 0xa5, 0xcd, 0x7b, 0xbd,
71 0x2d, 0x74, 0xd0, 0x12, 0xb8, 0xe5, 0xb4, 0xb0},
72 {0x89, 0x69, 0x97, 0x4a, 0x0c, 0x96, 0x77, 0x7e,
73 0x65, 0xb9, 0xf1, 0x09, 0xc5, 0x6e, 0xc6, 0x84},
74 {0x18, 0xf0, 0x7d, 0xec, 0x3a, 0xdc, 0x4d, 0x20,
75 0x79, 0xee, 0x5f, 0x3e, 0xd7, 0xcb, 0x39, 0x48}
76 };
77
78 /* System parameter */
79 static const uint32_t FK[4] = {
80 0xa3b1bac6, 0x56aa3350, 0x677d9197, 0xb27022dc
81 };
82
83 /* Fixed parameter */
84 static const uint32_t CK[32] = {
85 0x00070e15, 0x1c232a31, 0x383f464d, 0x545b6269,
86 0x70777e85, 0x8c939aa1, 0xa8afb6bd, 0xc4cbd2d9,
87 0xe0e7eef5, 0xfc030a11, 0x181f262d, 0x343b4249,
88 0x50575e65, 0x6c737a81, 0x888f969d, 0xa4abb2b9,
89 0xc0c7ced5, 0xdce3eaf1, 0xf8ff060d, 0x141b2229,
90 0x30373e45, 0x4c535a61, 0x686f767d, 0x848b9299,
91 0xa0a7aeb5, 0xbcc3cad1, 0xd8dfe6ed, 0xf4fb0209,
92 0x10171e25, 0x2c333a41, 0x484f565d, 0x646b7279
93 };
94
sm4Sbox(uint8_t inch)95 static uint8_t sm4Sbox(uint8_t inch)
96 {
97 uint8_t *tab = (uint8_t *)SboxTable;
98
99 return tab[inch];
100 }
101
sm4Lt(uint32_t ka)102 static uint32_t sm4Lt(uint32_t ka)
103 {
104 uint32_t bb = 0;
105 uint8_t a[4];
106 uint8_t b[4];
107
108 PUT_UINT32_BE(ka, a, 0);
109 b[0] = sm4Sbox(a[0]);
110 b[1] = sm4Sbox(a[1]);
111 b[2] = sm4Sbox(a[2]);
112 b[3] = sm4Sbox(a[3]);
113 GET_UINT32_BE(bb, b, 0);
114
115 return bb ^ ROTL(bb, 2) ^ ROTL(bb, 10) ^ ROTL(bb, 18) ^ ROTL(bb, 24);
116 }
117
sm4F(uint32_t x0,uint32_t x1,uint32_t x2,uint32_t x3,uint32_t rk)118 static uint32_t sm4F(uint32_t x0, uint32_t x1, uint32_t x2, uint32_t x3,
119 uint32_t rk)
120 {
121 return x0 ^ sm4Lt(x1 ^ x2 ^ x3 ^ rk);
122 }
123
sm4CalciRK(uint32_t ka)124 static uint32_t sm4CalciRK(uint32_t ka)
125 {
126 uint32_t bb = 0;
127 uint8_t a[4];
128 uint8_t b[4];
129
130 PUT_UINT32_BE(ka, a, 0);
131 b[0] = sm4Sbox(a[0]);
132 b[1] = sm4Sbox(a[1]);
133 b[2] = sm4Sbox(a[2]);
134 b[3] = sm4Sbox(a[3]);
135 GET_UINT32_BE(bb, b, 0);
136
137 return bb ^ ROTL(bb, 13) ^ ROTL(bb, 23);
138 }
139
sm4_setkey(uint32_t SK[32],const uint8_t key[16])140 static void sm4_setkey(uint32_t SK[32], const uint8_t key[16])
141 {
142 uint32_t MK[4];
143 uint32_t k[36];
144 uint32_t i = 0;
145
146 GET_UINT32_BE(MK[0], key, 0);
147 GET_UINT32_BE(MK[1], key, 4);
148 GET_UINT32_BE(MK[2], key, 8);
149 GET_UINT32_BE(MK[3], key, 12);
150
151 k[0] = MK[0] ^ FK[0];
152 k[1] = MK[1] ^ FK[1];
153 k[2] = MK[2] ^ FK[2];
154 k[3] = MK[3] ^ FK[3];
155
156 for (i = 0; i < 32; i++) {
157 k[i + 4] = k[i] ^ sm4CalciRK(k[i + 1] ^ k[i + 2] ^ k[i + 3] ^
158 CK[i]);
159 SK[i] = k[i + 4];
160 }
161 }
162
sm4_one_round(uint32_t sk[32],const uint8_t input[16],uint8_t output[16])163 static void sm4_one_round(uint32_t sk[32], const uint8_t input[16],
164 uint8_t output[16])
165 {
166 uint32_t i = 0;
167 uint32_t ulbuf[36];
168
169 memset(ulbuf, 0, sizeof(ulbuf));
170
171 GET_UINT32_BE(ulbuf[0], input, 0);
172 GET_UINT32_BE(ulbuf[1], input, 4);
173 GET_UINT32_BE(ulbuf[2], input, 8);
174 GET_UINT32_BE(ulbuf[3], input, 12);
175
176 for (i = 0; i < 32; i++)
177 ulbuf[i + 4] = sm4F(ulbuf[i], ulbuf[i + 1], ulbuf[i + 2],
178 ulbuf[i + 3], sk[i]);
179
180 PUT_UINT32_BE(ulbuf[35], output, 0);
181 PUT_UINT32_BE(ulbuf[34], output, 4);
182 PUT_UINT32_BE(ulbuf[33], output, 8);
183 PUT_UINT32_BE(ulbuf[32], output, 12);
184 }
185
sm4_setkey_enc(struct sm4_context * ctx,const uint8_t key[16])186 void sm4_setkey_enc(struct sm4_context *ctx, const uint8_t key[16])
187 {
188 ctx->mode = SM4_ENCRYPT;
189 sm4_setkey(ctx->sk, key);
190 }
191
sm4_setkey_dec(struct sm4_context * ctx,const uint8_t key[16])192 void sm4_setkey_dec(struct sm4_context *ctx, const uint8_t key[16])
193 {
194 int i;
195
196 ctx->mode = SM4_DECRYPT;
197 sm4_setkey(ctx->sk, key);
198
199 for (i = 0; i < 16; i++)
200 SWAP(ctx->sk[i], ctx->sk[31 - i]);
201 }
202
sm4_crypt_ecb(struct sm4_context * ctx,size_t length,const uint8_t * input,uint8_t * output)203 void sm4_crypt_ecb(struct sm4_context *ctx, size_t length, const uint8_t *input,
204 uint8_t *output)
205 {
206 assert(!(length % 16));
207
208 while (length > 0) {
209 sm4_one_round(ctx->sk, input, output);
210 input += 16;
211 output += 16;
212 length -= 16;
213 }
214 }
215
sm4_crypt_cbc(struct sm4_context * ctx,size_t length,uint8_t iv[16],const uint8_t * input,uint8_t * output)216 void sm4_crypt_cbc(struct sm4_context *ctx, size_t length, uint8_t iv[16],
217 const uint8_t *input, uint8_t *output)
218 {
219 int i;
220 uint8_t temp[16];
221
222 assert(!(length % 16));
223
224 if (ctx->mode == SM4_ENCRYPT) {
225 while (length > 0) {
226 for (i = 0; i < 16; i++)
227 output[i] = (uint8_t)(input[i] ^ iv[i]);
228 sm4_one_round(ctx->sk, output, output);
229 memcpy(iv, output, 16);
230 input += 16;
231 output += 16;
232 length -= 16;
233 }
234 } else {
235 /* SM4_DECRYPT */
236 while (length > 0) {
237 memcpy(temp, input, 16);
238 sm4_one_round(ctx->sk, input, output);
239 for (i = 0; i < 16; i++)
240 output[i] = (uint8_t)(output[i] ^ iv[i]);
241 memcpy(iv, temp, 16);
242 input += 16;
243 output += 16;
244 length -= 16;
245 }
246 }
247 }
248
sm4_crypt_ctr(struct sm4_context * ctx,size_t length,uint8_t ctr[16],const uint8_t * input,uint8_t * output)249 void sm4_crypt_ctr(struct sm4_context *ctx, size_t length, uint8_t ctr[16],
250 const uint8_t *input, uint8_t *output)
251 {
252 int i;
253 uint8_t temp[16];
254
255 assert(!(length % 16));
256
257 while (length > 0) {
258 memcpy(temp, ctr, 16);
259 sm4_one_round(ctx->sk, ctr, ctr);
260 for (i = 0; i < 16; i++)
261 output[i] = (uint8_t)(input[i] ^ ctr[i]);
262 memcpy(ctr, temp, 16);
263 for (i = 16; i > 0; i--)
264 if (++ctr[i - 1])
265 break;
266 input += 16;
267 output += 16;
268 length -= 16;
269 }
270 }
271
xts_multi(unsigned char * in,unsigned char * out)272 static void xts_multi(unsigned char *in, unsigned char *out)
273 {
274 uint8_t tt = 0;
275 uint8_t t = 0;
276 int i = 0;
277
278 for (i = 0; i < 16; i++) {
279 tt = in[i] >> 7;
280 out[i] = ((in[i] << 1) | t) & 0xFF;
281 t = tt;
282 }
283
284 out[0] ^= (0x87 & (0 - tt));
285 }
286
xor_128(const uint8_t a[16],const uint8_t b[16],uint8_t c[16])287 static void xor_128(const uint8_t a[16], const uint8_t b[16], uint8_t c[16])
288 {
289 int i = 0;
290
291 for (i = 0; i < 16; i++)
292 c[i] = a[i] ^ b[i];
293 }
294
sm4_crypt_xts(struct sm4_context * ctx,struct sm4_context * ctx_ek,struct sm4_context * ctx_dk,size_t len,uint8_t * iv,const uint8_t * input,uint8_t * output)295 void sm4_crypt_xts(struct sm4_context *ctx, struct sm4_context *ctx_ek,
296 struct sm4_context *ctx_dk, size_t len, uint8_t *iv,
297 const uint8_t *input, uint8_t *output)
298 {
299 uint8_t tweak[16] = { };
300 uint8_t tweak1[16] = { };
301 uint8_t ct[16] = { };
302 size_t i = 0;
303
304 assert(len >= 16);
305
306 sm4_one_round(ctx_ek->sk, iv, tweak);
307
308 if (ctx->mode == SM4_DECRYPT && (len % 16))
309 len -= 16;
310
311 while (len >= 16) {
312 xor_128(input, tweak, ct);
313 sm4_one_round(ctx->sk, ct, ct);
314 xor_128(ct, tweak, output);
315
316 xts_multi(tweak, tweak);
317 len -= 16;
318 if (len == 0) {
319 sm4_one_round(ctx_dk->sk, tweak, iv);
320 return;
321 }
322 input += 16;
323 output += 16;
324 }
325
326 if (ctx->mode == SM4_ENCRYPT) {
327 memcpy(ct, output - 16, 16);
328 for (i = 0; i < len; i++) {
329 output[i] = ct[i];
330 ct[i] = input[i];
331 }
332
333 xor_128(ct, tweak, ct);
334 sm4_one_round(ctx->sk, ct, ct);
335 xor_128(ct, tweak, ct);
336 memcpy(output - 16, ct, 16);
337 } else {
338 xts_multi(tweak, tweak1);
339 xor_128(input, tweak1, ct);
340 sm4_one_round(ctx->sk, ct, ct);
341 xor_128(ct, tweak1, ct);
342
343 for (i = 0; i < len; ++i) {
344 output[16 + i] = ct[i];
345 ct[i] = input[16 + i];
346 }
347 xor_128(ct, tweak, ct);
348 sm4_one_round(ctx->sk, ct, ct);
349 xor_128(ct, tweak, output);
350 }
351 }
352