1 /* Functions to compute MD5 message digest of files or memory blocks.
2 according to the definition of MD5 in RFC 1321 from April 1992.
3 Copyright (C) 1995-1997, 1999-2001, 2005-2006, 2008-2020 Free Software
4 Foundation, Inc.
5 This file is part of the GNU C Library.
6
7 This program is free software; you can redistribute it and/or modify it
8 under the terms of the GNU General Public License as published by the
9 Free Software Foundation; either version 3, or (at your option) any
10 later version.
11
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
16
17 You should have received a copy of the GNU General Public License
18 along with this program; if not, see <https://www.gnu.org/licenses/>. */
19
20 /* Written by Ulrich Drepper <drepper@gnu.ai.mit.edu>, 1995. */
21
22 #include <config.h>
23
24 #if HAVE_OPENSSL_MD5
25 # define GL_OPENSSL_INLINE _GL_EXTERN_INLINE
26 #endif
27 #include "md5.h"
28
29 #include <stdalign.h>
30 #include <stdint.h>
31 #include <stdlib.h>
32 #include <string.h>
33 #include <sys/types.h>
34
35 #if USE_UNLOCKED_IO
36 # include "unlocked-io.h"
37 #endif
38
39 #ifdef _LIBC
40 # include <endian.h>
41 # if __BYTE_ORDER == __BIG_ENDIAN
42 # define WORDS_BIGENDIAN 1
43 # endif
44 /* We need to keep the namespace clean so define the MD5 function
45 protected using leading __ . */
46 # define md5_init_ctx __md5_init_ctx
47 # define md5_process_block __md5_process_block
48 # define md5_process_bytes __md5_process_bytes
49 # define md5_finish_ctx __md5_finish_ctx
50 # define md5_read_ctx __md5_read_ctx
51 # define md5_stream __md5_stream
52 # define md5_buffer __md5_buffer
53 #endif
54
55 #include <byteswap.h>
56 #ifdef WORDS_BIGENDIAN
57 # define SWAP(n) bswap_32 (n)
58 #else
59 # define SWAP(n) (n)
60 #endif
61
62 #define BLOCKSIZE 32768
63 #if BLOCKSIZE % 64 != 0
64 # error "invalid BLOCKSIZE"
65 #endif
66
67 #if ! HAVE_OPENSSL_MD5
68 /* This array contains the bytes used to pad the buffer to the next
69 64-byte boundary. (RFC 1321, 3.1: Step 1) */
70 static const unsigned char fillbuf[64] = { 0x80, 0 /* , 0, 0, ... */ };
71
72
73 /* Initialize structure containing state of computation.
74 (RFC 1321, 3.3: Step 3) */
75 void
md5_init_ctx(struct md5_ctx * ctx)76 md5_init_ctx (struct md5_ctx *ctx)
77 {
78 ctx->A = 0x67452301;
79 ctx->B = 0xefcdab89;
80 ctx->C = 0x98badcfe;
81 ctx->D = 0x10325476;
82
83 ctx->total[0] = ctx->total[1] = 0;
84 ctx->buflen = 0;
85 }
86
87 /* Copy the 4 byte value from v into the memory location pointed to by *cp,
88 If your architecture allows unaligned access this is equivalent to
89 * (uint32_t *) cp = v */
90 static void
set_uint32(char * cp,uint32_t v)91 set_uint32 (char *cp, uint32_t v)
92 {
93 memcpy (cp, &v, sizeof v);
94 }
95
96 /* Put result from CTX in first 16 bytes following RESBUF. The result
97 must be in little endian byte order. */
98 void *
md5_read_ctx(const struct md5_ctx * ctx,void * resbuf)99 md5_read_ctx (const struct md5_ctx *ctx, void *resbuf)
100 {
101 char *r = resbuf;
102 set_uint32 (r + 0 * sizeof ctx->A, SWAP (ctx->A));
103 set_uint32 (r + 1 * sizeof ctx->B, SWAP (ctx->B));
104 set_uint32 (r + 2 * sizeof ctx->C, SWAP (ctx->C));
105 set_uint32 (r + 3 * sizeof ctx->D, SWAP (ctx->D));
106
107 return resbuf;
108 }
109
110 /* Process the remaining bytes in the internal buffer and the usual
111 prolog according to the standard and write the result to RESBUF. */
112 void *
md5_finish_ctx(struct md5_ctx * ctx,void * resbuf)113 md5_finish_ctx (struct md5_ctx *ctx, void *resbuf)
114 {
115 /* Take yet unprocessed bytes into account. */
116 uint32_t bytes = ctx->buflen;
117 size_t size = (bytes < 56) ? 64 / 4 : 64 * 2 / 4;
118
119 /* Now count remaining bytes. */
120 ctx->total[0] += bytes;
121 if (ctx->total[0] < bytes)
122 ++ctx->total[1];
123
124 /* Put the 64-bit file length in *bits* at the end of the buffer. */
125 ctx->buffer[size - 2] = SWAP (ctx->total[0] << 3);
126 ctx->buffer[size - 1] = SWAP ((ctx->total[1] << 3) | (ctx->total[0] >> 29));
127
128 memcpy (&((char *) ctx->buffer)[bytes], fillbuf, (size - 2) * 4 - bytes);
129
130 /* Process last bytes. */
131 md5_process_block (ctx->buffer, size * 4, ctx);
132
133 return md5_read_ctx (ctx, resbuf);
134 }
135 #endif
136
137 #if defined _LIBC || defined GL_COMPILE_CRYPTO_STREAM
138
139 #include "af_alg.h"
140
141 /* Compute MD5 message digest for bytes read from STREAM. The
142 resulting message digest number will be written into the 16 bytes
143 beginning at RESBLOCK. */
144 int
md5_stream(FILE * stream,void * resblock)145 md5_stream (FILE *stream, void *resblock)
146 {
147 switch (afalg_stream (stream, "md5", resblock, MD5_DIGEST_SIZE))
148 {
149 case 0: return 0;
150 case -EIO: return 1;
151 }
152
153 char *buffer = malloc (BLOCKSIZE + 72);
154 if (!buffer)
155 return 1;
156
157 struct md5_ctx ctx;
158 md5_init_ctx (&ctx);
159 size_t sum;
160
161 /* Iterate over full file contents. */
162 while (1)
163 {
164 /* We read the file in blocks of BLOCKSIZE bytes. One call of the
165 computation function processes the whole buffer so that with the
166 next round of the loop another block can be read. */
167 size_t n;
168 sum = 0;
169
170 /* Read block. Take care for partial reads. */
171 while (1)
172 {
173 /* Either process a partial fread() from this loop,
174 or the fread() in afalg_stream may have gotten EOF.
175 We need to avoid a subsequent fread() as EOF may
176 not be sticky. For details of such systems, see:
177 https://sourceware.org/bugzilla/show_bug.cgi?id=1190 */
178 if (feof (stream))
179 goto process_partial_block;
180
181 n = fread (buffer + sum, 1, BLOCKSIZE - sum, stream);
182
183 sum += n;
184
185 if (sum == BLOCKSIZE)
186 break;
187
188 if (n == 0)
189 {
190 /* Check for the error flag IFF N == 0, so that we don't
191 exit the loop after a partial read due to e.g., EAGAIN
192 or EWOULDBLOCK. */
193 if (ferror (stream))
194 {
195 free (buffer);
196 return 1;
197 }
198 goto process_partial_block;
199 }
200 }
201
202 /* Process buffer with BLOCKSIZE bytes. Note that
203 BLOCKSIZE % 64 == 0
204 */
205 md5_process_block (buffer, BLOCKSIZE, &ctx);
206 }
207
208 process_partial_block:
209
210 /* Process any remaining bytes. */
211 if (sum > 0)
212 md5_process_bytes (buffer, sum, &ctx);
213
214 /* Construct result in desired memory. */
215 md5_finish_ctx (&ctx, resblock);
216 free (buffer);
217 return 0;
218 }
219 #endif
220
221 #if ! HAVE_OPENSSL_MD5
222 /* Compute MD5 message digest for LEN bytes beginning at BUFFER. The
223 result is always in little endian byte order, so that a byte-wise
224 output yields to the wanted ASCII representation of the message
225 digest. */
226 void *
md5_buffer(const char * buffer,size_t len,void * resblock)227 md5_buffer (const char *buffer, size_t len, void *resblock)
228 {
229 struct md5_ctx ctx;
230
231 /* Initialize the computation context. */
232 md5_init_ctx (&ctx);
233
234 /* Process whole buffer but last len % 64 bytes. */
235 md5_process_bytes (buffer, len, &ctx);
236
237 /* Put result in desired memory area. */
238 return md5_finish_ctx (&ctx, resblock);
239 }
240
241
242 void
md5_process_bytes(const void * buffer,size_t len,struct md5_ctx * ctx)243 md5_process_bytes (const void *buffer, size_t len, struct md5_ctx *ctx)
244 {
245 /* When we already have some bits in our internal buffer concatenate
246 both inputs first. */
247 if (ctx->buflen != 0)
248 {
249 size_t left_over = ctx->buflen;
250 size_t add = 128 - left_over > len ? len : 128 - left_over;
251
252 memcpy (&((char *) ctx->buffer)[left_over], buffer, add);
253 ctx->buflen += add;
254
255 if (ctx->buflen > 64)
256 {
257 md5_process_block (ctx->buffer, ctx->buflen & ~63, ctx);
258
259 ctx->buflen &= 63;
260 /* The regions in the following copy operation cannot overlap,
261 because ctx->buflen < 64 ≤ (left_over + add) & ~63. */
262 memcpy (ctx->buffer,
263 &((char *) ctx->buffer)[(left_over + add) & ~63],
264 ctx->buflen);
265 }
266
267 buffer = (const char *) buffer + add;
268 len -= add;
269 }
270
271 /* Process available complete blocks. */
272 if (len >= 64)
273 {
274 #if !(_STRING_ARCH_unaligned || _STRING_INLINE_unaligned)
275 # define UNALIGNED_P(p) ((uintptr_t) (p) % alignof (uint32_t) != 0)
276 if (UNALIGNED_P (buffer))
277 while (len > 64)
278 {
279 md5_process_block (memcpy (ctx->buffer, buffer, 64), 64, ctx);
280 buffer = (const char *) buffer + 64;
281 len -= 64;
282 }
283 else
284 #endif
285 {
286 md5_process_block (buffer, len & ~63, ctx);
287 buffer = (const char *) buffer + (len & ~63);
288 len &= 63;
289 }
290 }
291
292 /* Move remaining bytes in internal buffer. */
293 if (len > 0)
294 {
295 size_t left_over = ctx->buflen;
296
297 memcpy (&((char *) ctx->buffer)[left_over], buffer, len);
298 left_over += len;
299 if (left_over >= 64)
300 {
301 md5_process_block (ctx->buffer, 64, ctx);
302 left_over -= 64;
303 /* The regions in the following copy operation cannot overlap,
304 because left_over ≤ 64. */
305 memcpy (ctx->buffer, &ctx->buffer[16], left_over);
306 }
307 ctx->buflen = left_over;
308 }
309 }
310
311
312 /* These are the four functions used in the four steps of the MD5 algorithm
313 and defined in the RFC 1321. The first function is a little bit optimized
314 (as found in Colin Plumbs public domain implementation). */
315 /* #define FF(b, c, d) ((b & c) | (~b & d)) */
316 #define FF(b, c, d) (d ^ (b & (c ^ d)))
317 #define FG(b, c, d) FF (d, b, c)
318 #define FH(b, c, d) (b ^ c ^ d)
319 #define FI(b, c, d) (c ^ (b | ~d))
320
321 /* Process LEN bytes of BUFFER, accumulating context into CTX.
322 It is assumed that LEN % 64 == 0. */
323
324 void
md5_process_block(const void * buffer,size_t len,struct md5_ctx * ctx)325 md5_process_block (const void *buffer, size_t len, struct md5_ctx *ctx)
326 {
327 uint32_t correct_words[16];
328 const uint32_t *words = buffer;
329 size_t nwords = len / sizeof (uint32_t);
330 const uint32_t *endp = words + nwords;
331 uint32_t A = ctx->A;
332 uint32_t B = ctx->B;
333 uint32_t C = ctx->C;
334 uint32_t D = ctx->D;
335 uint32_t lolen = len;
336
337 /* First increment the byte count. RFC 1321 specifies the possible
338 length of the file up to 2^64 bits. Here we only compute the
339 number of bytes. Do a double word increment. */
340 ctx->total[0] += lolen;
341 ctx->total[1] += (len >> 31 >> 1) + (ctx->total[0] < lolen);
342
343 /* Process all bytes in the buffer with 64 bytes in each round of
344 the loop. */
345 while (words < endp)
346 {
347 uint32_t *cwp = correct_words;
348 uint32_t A_save = A;
349 uint32_t B_save = B;
350 uint32_t C_save = C;
351 uint32_t D_save = D;
352
353 /* First round: using the given function, the context and a constant
354 the next context is computed. Because the algorithms processing
355 unit is a 32-bit word and it is determined to work on words in
356 little endian byte order we perhaps have to change the byte order
357 before the computation. To reduce the work for the next steps
358 we store the swapped words in the array CORRECT_WORDS. */
359
360 #define OP(a, b, c, d, s, T) \
361 do \
362 { \
363 a += FF (b, c, d) + (*cwp++ = SWAP (*words)) + T; \
364 ++words; \
365 CYCLIC (a, s); \
366 a += b; \
367 } \
368 while (0)
369
370 /* It is unfortunate that C does not provide an operator for
371 cyclic rotation. Hope the C compiler is smart enough. */
372 #define CYCLIC(w, s) (w = (w << s) | (w >> (32 - s)))
373
374 /* Before we start, one word to the strange constants.
375 They are defined in RFC 1321 as
376
377 T[i] = (int) (4294967296.0 * fabs (sin (i))), i=1..64
378
379 Here is an equivalent invocation using Perl:
380
381 perl -e 'foreach(1..64){printf "0x%08x\n", int (4294967296 * abs (sin $_))}'
382 */
383
384 /* Round 1. */
385 OP (A, B, C, D, 7, 0xd76aa478);
386 OP (D, A, B, C, 12, 0xe8c7b756);
387 OP (C, D, A, B, 17, 0x242070db);
388 OP (B, C, D, A, 22, 0xc1bdceee);
389 OP (A, B, C, D, 7, 0xf57c0faf);
390 OP (D, A, B, C, 12, 0x4787c62a);
391 OP (C, D, A, B, 17, 0xa8304613);
392 OP (B, C, D, A, 22, 0xfd469501);
393 OP (A, B, C, D, 7, 0x698098d8);
394 OP (D, A, B, C, 12, 0x8b44f7af);
395 OP (C, D, A, B, 17, 0xffff5bb1);
396 OP (B, C, D, A, 22, 0x895cd7be);
397 OP (A, B, C, D, 7, 0x6b901122);
398 OP (D, A, B, C, 12, 0xfd987193);
399 OP (C, D, A, B, 17, 0xa679438e);
400 OP (B, C, D, A, 22, 0x49b40821);
401
402 /* For the second to fourth round we have the possibly swapped words
403 in CORRECT_WORDS. Redefine the macro to take an additional first
404 argument specifying the function to use. */
405 #undef OP
406 #define OP(f, a, b, c, d, k, s, T) \
407 do \
408 { \
409 a += f (b, c, d) + correct_words[k] + T; \
410 CYCLIC (a, s); \
411 a += b; \
412 } \
413 while (0)
414
415 /* Round 2. */
416 OP (FG, A, B, C, D, 1, 5, 0xf61e2562);
417 OP (FG, D, A, B, C, 6, 9, 0xc040b340);
418 OP (FG, C, D, A, B, 11, 14, 0x265e5a51);
419 OP (FG, B, C, D, A, 0, 20, 0xe9b6c7aa);
420 OP (FG, A, B, C, D, 5, 5, 0xd62f105d);
421 OP (FG, D, A, B, C, 10, 9, 0x02441453);
422 OP (FG, C, D, A, B, 15, 14, 0xd8a1e681);
423 OP (FG, B, C, D, A, 4, 20, 0xe7d3fbc8);
424 OP (FG, A, B, C, D, 9, 5, 0x21e1cde6);
425 OP (FG, D, A, B, C, 14, 9, 0xc33707d6);
426 OP (FG, C, D, A, B, 3, 14, 0xf4d50d87);
427 OP (FG, B, C, D, A, 8, 20, 0x455a14ed);
428 OP (FG, A, B, C, D, 13, 5, 0xa9e3e905);
429 OP (FG, D, A, B, C, 2, 9, 0xfcefa3f8);
430 OP (FG, C, D, A, B, 7, 14, 0x676f02d9);
431 OP (FG, B, C, D, A, 12, 20, 0x8d2a4c8a);
432
433 /* Round 3. */
434 OP (FH, A, B, C, D, 5, 4, 0xfffa3942);
435 OP (FH, D, A, B, C, 8, 11, 0x8771f681);
436 OP (FH, C, D, A, B, 11, 16, 0x6d9d6122);
437 OP (FH, B, C, D, A, 14, 23, 0xfde5380c);
438 OP (FH, A, B, C, D, 1, 4, 0xa4beea44);
439 OP (FH, D, A, B, C, 4, 11, 0x4bdecfa9);
440 OP (FH, C, D, A, B, 7, 16, 0xf6bb4b60);
441 OP (FH, B, C, D, A, 10, 23, 0xbebfbc70);
442 OP (FH, A, B, C, D, 13, 4, 0x289b7ec6);
443 OP (FH, D, A, B, C, 0, 11, 0xeaa127fa);
444 OP (FH, C, D, A, B, 3, 16, 0xd4ef3085);
445 OP (FH, B, C, D, A, 6, 23, 0x04881d05);
446 OP (FH, A, B, C, D, 9, 4, 0xd9d4d039);
447 OP (FH, D, A, B, C, 12, 11, 0xe6db99e5);
448 OP (FH, C, D, A, B, 15, 16, 0x1fa27cf8);
449 OP (FH, B, C, D, A, 2, 23, 0xc4ac5665);
450
451 /* Round 4. */
452 OP (FI, A, B, C, D, 0, 6, 0xf4292244);
453 OP (FI, D, A, B, C, 7, 10, 0x432aff97);
454 OP (FI, C, D, A, B, 14, 15, 0xab9423a7);
455 OP (FI, B, C, D, A, 5, 21, 0xfc93a039);
456 OP (FI, A, B, C, D, 12, 6, 0x655b59c3);
457 OP (FI, D, A, B, C, 3, 10, 0x8f0ccc92);
458 OP (FI, C, D, A, B, 10, 15, 0xffeff47d);
459 OP (FI, B, C, D, A, 1, 21, 0x85845dd1);
460 OP (FI, A, B, C, D, 8, 6, 0x6fa87e4f);
461 OP (FI, D, A, B, C, 15, 10, 0xfe2ce6e0);
462 OP (FI, C, D, A, B, 6, 15, 0xa3014314);
463 OP (FI, B, C, D, A, 13, 21, 0x4e0811a1);
464 OP (FI, A, B, C, D, 4, 6, 0xf7537e82);
465 OP (FI, D, A, B, C, 11, 10, 0xbd3af235);
466 OP (FI, C, D, A, B, 2, 15, 0x2ad7d2bb);
467 OP (FI, B, C, D, A, 9, 21, 0xeb86d391);
468
469 /* Add the starting values of the context. */
470 A += A_save;
471 B += B_save;
472 C += C_save;
473 D += D_save;
474 }
475
476 /* Put checksum in context given as argument. */
477 ctx->A = A;
478 ctx->B = B;
479 ctx->C = C;
480 ctx->D = D;
481 }
482 #endif
483
484 /*
485 * Hey Emacs!
486 * Local Variables:
487 * coding: utf-8
488 * End:
489 */
490