xref: /reactos/drivers/filesystems/btrfs/sha256.c (revision 3ba42399)
1 #include <stdint.h>
2 #include <string.h>
3 
4 // Public domain code from https://github.com/amosnier/sha-2
5 
6 // FIXME - x86 SHA extensions
7 
8 #define CHUNK_SIZE 64
9 #define TOTAL_LEN_LEN 8
10 
11 /*
12  * ABOUT bool: this file does not use bool in order to be as pre-C99 compatible as possible.
13  */
14 
15 /*
16  * Comments from pseudo-code at https://en.wikipedia.org/wiki/SHA-2 are reproduced here.
17  * When useful for clarification, portions of the pseudo-code are reproduced here too.
18  */
19 
20 /*
21  * Initialize array of round constants:
22  * (first 32 bits of the fractional parts of the cube roots of the first 64 primes 2..311):
23  */
24 static const uint32_t k[] = {
25 	0x428a2f98, 0x71374491, 0xb5c0fbcf, 0xe9b5dba5, 0x3956c25b, 0x59f111f1, 0x923f82a4, 0xab1c5ed5,
26 	0xd807aa98, 0x12835b01, 0x243185be, 0x550c7dc3, 0x72be5d74, 0x80deb1fe, 0x9bdc06a7, 0xc19bf174,
27 	0xe49b69c1, 0xefbe4786, 0x0fc19dc6, 0x240ca1cc, 0x2de92c6f, 0x4a7484aa, 0x5cb0a9dc, 0x76f988da,
28 	0x983e5152, 0xa831c66d, 0xb00327c8, 0xbf597fc7, 0xc6e00bf3, 0xd5a79147, 0x06ca6351, 0x14292967,
29 	0x27b70a85, 0x2e1b2138, 0x4d2c6dfc, 0x53380d13, 0x650a7354, 0x766a0abb, 0x81c2c92e, 0x92722c85,
30 	0xa2bfe8a1, 0xa81a664b, 0xc24b8b70, 0xc76c51a3, 0xd192e819, 0xd6990624, 0xf40e3585, 0x106aa070,
31 	0x19a4c116, 0x1e376c08, 0x2748774c, 0x34b0bcb5, 0x391c0cb3, 0x4ed8aa4a, 0x5b9cca4f, 0x682e6ff3,
32 	0x748f82ee, 0x78a5636f, 0x84c87814, 0x8cc70208, 0x90befffa, 0xa4506ceb, 0xbef9a3f7, 0xc67178f2
33 };
34 
35 struct buffer_state {
36 	const uint8_t * p;
37 	size_t len;
38 	size_t total_len;
39 	int single_one_delivered; /* bool */
40 	int total_len_delivered; /* bool */
41 };
42 
right_rot(uint32_t value,unsigned int count)43 static inline uint32_t right_rot(uint32_t value, unsigned int count)
44 {
45 	/*
46 	 * Defined behaviour in standard C for all count where 0 < count < 32,
47 	 * which is what we need here.
48 	 */
49 	return value >> count | value << (32 - count);
50 }
51 
init_buf_state(struct buffer_state * state,const void * input,size_t len)52 static void init_buf_state(struct buffer_state * state, const void * input, size_t len)
53 {
54 	state->p = input;
55 	state->len = len;
56 	state->total_len = len;
57 	state->single_one_delivered = 0;
58 	state->total_len_delivered = 0;
59 }
60 
61 /* Return value: bool */
calc_chunk(uint8_t chunk[CHUNK_SIZE],struct buffer_state * state)62 static int calc_chunk(uint8_t chunk[CHUNK_SIZE], struct buffer_state * state)
63 {
64 	size_t space_in_chunk;
65 
66 	if (state->total_len_delivered) {
67 		return 0;
68 	}
69 
70 	if (state->len >= CHUNK_SIZE) {
71 		memcpy(chunk, state->p, CHUNK_SIZE);
72 		state->p += CHUNK_SIZE;
73 		state->len -= CHUNK_SIZE;
74 		return 1;
75 	}
76 
77 	memcpy(chunk, state->p, state->len);
78 	chunk += state->len;
79 	space_in_chunk = CHUNK_SIZE - state->len;
80 	state->p += state->len;
81 	state->len = 0;
82 
83 	/* If we are here, space_in_chunk is one at minimum. */
84 	if (!state->single_one_delivered) {
85 		*chunk++ = 0x80;
86 		space_in_chunk -= 1;
87 		state->single_one_delivered = 1;
88 	}
89 
90 	/*
91 	 * Now:
92 	 * - either there is enough space left for the total length, and we can conclude,
93 	 * - or there is too little space left, and we have to pad the rest of this chunk with zeroes.
94 	 * In the latter case, we will conclude at the next invokation of this function.
95 	 */
96 	if (space_in_chunk >= TOTAL_LEN_LEN) {
97 		const size_t left = space_in_chunk - TOTAL_LEN_LEN;
98 		size_t len = state->total_len;
99 		int i;
100 		memset(chunk, 0x00, left);
101 		chunk += left;
102 
103 		/* Storing of len * 8 as a big endian 64-bit without overflow. */
104 		chunk[7] = (uint8_t) (len << 3);
105 		len >>= 5;
106 		for (i = 6; i >= 0; i--) {
107 			chunk[i] = (uint8_t) len;
108 			len >>= 8;
109 		}
110 		state->total_len_delivered = 1;
111 	} else {
112 		memset(chunk, 0x00, space_in_chunk);
113 	}
114 
115 	return 1;
116 }
117 
118 /*
119  * Limitations:
120  * - Since input is a pointer in RAM, the data to hash should be in RAM, which could be a problem
121  *   for large data sizes.
122  * - SHA algorithms theoretically operate on bit strings. However, this implementation has no support
123  *   for bit string lengths that are not multiples of eight, and it really operates on arrays of bytes.
124  *   In particular, the len parameter is a number of bytes.
125  */
calc_sha256(uint8_t * hash,const void * input,size_t len)126 void calc_sha256(uint8_t* hash, const void* input, size_t len)
127 {
128 	/*
129 	 * Note 1: All integers (expect indexes) are 32-bit unsigned integers and addition is calculated modulo 2^32.
130 	 * Note 2: For each round, there is one round constant k[i] and one entry in the message schedule array w[i], 0 = i = 63
131 	 * Note 3: The compression function uses 8 working variables, a through h
132 	 * Note 4: Big-endian convention is used when expressing the constants in this pseudocode,
133 	 *     and when parsing message block data from bytes to words, for example,
134 	 *     the first word of the input message "abc" after padding is 0x61626380
135 	 */
136 
137 	/*
138 	 * Initialize hash values:
139 	 * (first 32 bits of the fractional parts of the square roots of the first 8 primes 2..19):
140 	 */
141 	uint32_t h[] = { 0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a, 0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19 };
142 	unsigned i, j;
143 
144 	/* 512-bit chunks is what we will operate on. */
145 	uint8_t chunk[64];
146 
147 	struct buffer_state state;
148 
149 	init_buf_state(&state, input, len);
150 
151 	while (calc_chunk(chunk, &state)) {
152 		uint32_t ah[8];
153 
154 		const uint8_t *p = chunk;
155 
156 		/* Initialize working variables to current hash value: */
157 		for (i = 0; i < 8; i++)
158 			ah[i] = h[i];
159 
160 		/* Compression function main loop: */
161 		for (i = 0; i < 4; i++) {
162 			/*
163 			 * The w-array is really w[64], but since we only need
164 			 * 16 of them at a time, we save stack by calculating
165 			 * 16 at a time.
166 			 *
167 			 * This optimization was not there initially and the
168 			 * rest of the comments about w[64] are kept in their
169 			 * initial state.
170 			 */
171 
172 			/*
173 			 * create a 64-entry message schedule array w[0..63] of 32-bit words
174 			 * (The initial values in w[0..63] don't matter, so many implementations zero them here)
175 			 * copy chunk into first 16 words w[0..15] of the message schedule array
176 			 */
177 			uint32_t w[16];
178 
179 			for (j = 0; j < 16; j++) {
180 				if (i == 0) {
181 					w[j] = (uint32_t) p[0] << 24 | (uint32_t) p[1] << 16 |
182 						(uint32_t) p[2] << 8 | (uint32_t) p[3];
183 					p += 4;
184 				} else {
185 					/* Extend the first 16 words into the remaining 48 words w[16..63] of the message schedule array: */
186 					const uint32_t s0 = right_rot(w[(j + 1) & 0xf], 7) ^ right_rot(w[(j + 1) & 0xf], 18) ^ (w[(j + 1) & 0xf] >> 3);
187 					const uint32_t s1 = right_rot(w[(j + 14) & 0xf], 17) ^ right_rot(w[(j + 14) & 0xf], 19) ^ (w[(j + 14) & 0xf] >> 10);
188 					w[j] = w[j] + s0 + w[(j + 9) & 0xf] + s1;
189 				}
190 				{
191 					const uint32_t s1 = right_rot(ah[4], 6) ^ right_rot(ah[4], 11) ^ right_rot(ah[4], 25);
192 					const uint32_t ch = (ah[4] & ah[5]) ^ (~ah[4] & ah[6]);
193 					const uint32_t temp1 = ah[7] + s1 + ch + k[i << 4 | j] + w[j];
194 					const uint32_t s0 = right_rot(ah[0], 2) ^ right_rot(ah[0], 13) ^ right_rot(ah[0], 22);
195 					const uint32_t maj = (ah[0] & ah[1]) ^ (ah[0] & ah[2]) ^ (ah[1] & ah[2]);
196 					const uint32_t temp2 = s0 + maj;
197 
198 					ah[7] = ah[6];
199 					ah[6] = ah[5];
200 					ah[5] = ah[4];
201 					ah[4] = ah[3] + temp1;
202 					ah[3] = ah[2];
203 					ah[2] = ah[1];
204 					ah[1] = ah[0];
205 					ah[0] = temp1 + temp2;
206 				}
207 			}
208 		}
209 
210 		/* Add the compressed chunk to the current hash value: */
211 		for (i = 0; i < 8; i++)
212 			h[i] += ah[i];
213 	}
214 
215 	/* Produce the final hash value (big-endian): */
216 	for (i = 0, j = 0; i < 8; i++)
217 	{
218 		hash[j++] = (uint8_t) (h[i] >> 24);
219 		hash[j++] = (uint8_t) (h[i] >> 16);
220 		hash[j++] = (uint8_t) (h[i] >> 8);
221 		hash[j++] = (uint8_t) h[i];
222 	}
223 }
224