1 /* md5.c - 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, 1996, 2001 Free Software Foundation, Inc.
4    NOTE: The canonical source of this file is maintained with the GNU C
5    Library.  Bugs can be reported to bug-glibc@prep.ai.mit.edu.
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 2, 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, write to the Free Software Foundation,
19    Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA.
20 */
21 
22 /* Written by Ulrich Drepper <drepper@gnu.ai.mit.edu>, 1995.  */
23 
24 #include <Rconfig.h>
25 #include <sys/types.h>
26 
27 # include <stdlib.h>
28 # include <string.h>
29 
30 #include "md5.h"
31 
32 #ifdef WORDS_BIGENDIAN
33 # define SWAP(n)							\
34     (((n) << 24) | (((n) & 0xff00) << 8) | (((n) >> 8) & 0xff00) | ((n) >> 24))
35 #else
36 # define SWAP(n) (n)
37 #endif
38 
39 /* moved from md5.h */
40 static void md5_process_block __P ((const void *buffer, size_t len,
41 				    struct md5_ctx *ctx));
42 static void md5_process_bytes __P ((const void *buffer, size_t len,
43 				    struct md5_ctx *ctx));
44 
45 
46 /* This array contains the bytes used to pad the buffer to the next
47    64-byte boundary.  (RFC 1321, 3.1: Step 1)  */
48 static const unsigned char fillbuf[64] = { 0x80, 0 /* , 0, 0, ...  */ };
49 
50 
51 /* Initialize structure containing state of computation.
52    (RFC 1321, 3.3: Step 3)  */
53 static void
md5_init_ctx(struct md5_ctx * ctx)54 md5_init_ctx (struct md5_ctx *ctx)
55 {
56   ctx->A = 0x67452301;
57   ctx->B = 0xefcdab89;
58   ctx->C = 0x98badcfe;
59   ctx->D = 0x10325476;
60 
61   ctx->total[0] = ctx->total[1] = 0;
62   ctx->buflen = 0;
63 }
64 
65 /* Put result from CTX in first 16 bytes following RESBUF.  The result
66    must be in little endian byte order.
67 
68    IMPORTANT: On some systems it is required that RESBUF is correctly
69    aligned for a 32 bits value.  */
70 static void *
md5_read_ctx(const struct md5_ctx * ctx,void * resbuf)71 md5_read_ctx (const struct md5_ctx *ctx, void *resbuf)
72 {
73   ((md5_uint32 *) resbuf)[0] = SWAP (ctx->A);
74   ((md5_uint32 *) resbuf)[1] = SWAP (ctx->B);
75   ((md5_uint32 *) resbuf)[2] = SWAP (ctx->C);
76   ((md5_uint32 *) resbuf)[3] = SWAP (ctx->D);
77 
78   return resbuf;
79 }
80 
81 /* Process the remaining bytes in the internal buffer and the usual
82    prolog according to the standard and write the result to RESBUF.
83 
84    IMPORTANT: On some systems it is required that RESBUF is correctly
85    aligned for a 32 bits value.  */
86 static void *
md5_finish_ctx(struct md5_ctx * ctx,void * resbuf)87 md5_finish_ctx (struct md5_ctx *ctx, void *resbuf)
88 {
89   /* Take yet unprocessed bytes into account.  */
90   md5_uint32 bytes = ctx->buflen;
91   size_t pad;
92 
93   /* Now count remaining bytes.  */
94   ctx->total[0] += bytes;
95   if (ctx->total[0] < bytes)
96     ++ctx->total[1];
97 
98   pad = bytes >= 56 ? 64 + 56 - bytes : 56 - bytes;
99   memcpy (&ctx->buffer[bytes], fillbuf, pad);
100 
101   /* Put the 64-bit file length in *bits* at the end of the buffer.  */
102   *(md5_uint32 *) &ctx->buffer[bytes + pad] = SWAP (ctx->total[0] << 3);
103   *(md5_uint32 *) &ctx->buffer[bytes + pad + 4] = SWAP ((ctx->total[1] << 3) |
104 							(ctx->total[0] >> 29));
105 
106   /* Process last bytes.  */
107   md5_process_block (ctx->buffer, bytes + pad + 8, ctx);
108 
109   return md5_read_ctx (ctx, resbuf);
110 }
111 
112 /* Compute MD5 message digest for bytes read from STREAM.  The
113    resulting message digest number will be written into the 16 bytes
114    beginning at RESBLOCK.  */
115 int
md5_stream(FILE * stream,void * resblock)116 md5_stream (FILE *stream, void *resblock)
117 {
118   /* Important: BLOCKSIZE must be a multiple of 64.  */
119 #define BLOCKSIZE 4096
120   struct md5_ctx ctx;
121   char buffer[BLOCKSIZE + 72];
122   size_t sum;
123 
124   /* Initialize the computation context.  */
125   md5_init_ctx (&ctx);
126 
127   /* Iterate over full file contents.  */
128   while (1)
129     {
130       /* We read the file in blocks of BLOCKSIZE bytes.  One call of the
131 	 computation function processes the whole buffer so that with the
132 	 next round of the loop another block can be read.  */
133       size_t n;
134       sum = 0;
135 
136       /* Read block.  Take care for partial reads.  */
137       do
138 	{
139 	  n = fread (buffer + sum, 1, BLOCKSIZE - sum, stream);
140 
141 	  sum += n;
142 	}
143       while (sum < BLOCKSIZE && n != 0);
144       if (n == 0 && ferror (stream))
145         return 1;
146 
147       /* If end of file is reached, end the loop.  */
148       if (n == 0)
149 	break;
150 
151       /* Process buffer with BLOCKSIZE bytes.  Note that
152 			BLOCKSIZE % 64 == 0
153        */
154       md5_process_block (buffer, BLOCKSIZE, &ctx);
155     }
156 
157   /* Add the last bytes if necessary.  */
158   if (sum > 0)
159     md5_process_bytes (buffer, sum, &ctx);
160 
161   /* Construct result in desired memory.  */
162   md5_finish_ctx (&ctx, resblock);
163   return 0;
164 }
165 
166 #ifdef UNUSED
167 /* Compute MD5 message digest for LEN bytes beginning at BUFFER.  The
168    result is always in little endian byte order, so that a byte-wise
169    output yields to the wanted ASCII representation of the message
170    digest.  */
171 static void *
md5_buffer(const char * buffer,size_t len,void * resblock)172 md5_buffer (const char *buffer, size_t len, void *resblock)
173 {
174   struct md5_ctx ctx;
175 
176   /* Initialize the computation context.  */
177   md5_init_ctx (&ctx);
178 
179   /* Process whole buffer but last len % 64 bytes.  */
180   md5_process_bytes (buffer, len, &ctx);
181 
182   /* Put result in desired memory area.  */
183   return md5_finish_ctx (&ctx, resblock);
184 }
185 #endif
186 
187 static void
md5_process_bytes(const void * buffer,size_t len,struct md5_ctx * ctx)188  md5_process_bytes (const void *buffer, size_t len, struct md5_ctx *ctx)
189 {
190   /* When we already have some bits in our internal buffer concatenate
191      both inputs first.  */
192   if (ctx->buflen != 0)
193     {
194       size_t left_over = ctx->buflen;
195       size_t add = 128 - left_over > len ? len : 128 - left_over;
196 
197       memcpy (&ctx->buffer[left_over], buffer, add);
198       ctx->buflen += add;
199 
200       if (left_over + add > 64)
201 	{
202 	  md5_process_block (ctx->buffer, (left_over + add) & ~63, ctx);
203 	  /* The regions in the following copy operation cannot overlap.  */
204 	  memcpy (ctx->buffer, &ctx->buffer[(left_over + add) & ~63],
205 		  (left_over + add) & 63);
206 	  ctx->buflen = (left_over + add) & 63;
207 	}
208 
209       buffer = (const char *) buffer + add;
210       len -= add;
211     }
212 
213   /* Process available complete blocks.  */
214   if (len > 64)
215     {
216       md5_process_block (buffer, len & ~63, ctx);
217       buffer = (const char *) buffer + (len & ~63);
218       len &= 63;
219     }
220 
221   /* Move remaining bytes in internal buffer.  */
222   if (len > 0)
223     {
224       memcpy (ctx->buffer, buffer, len);
225       ctx->buflen = (md5_uint32) len;
226     }
227 }
228 
229 
230 /* These are the four functions used in the four steps of the MD5 algorithm
231    and defined in the RFC 1321.  The first function is a little bit optimized
232    (as found in Colin Plumbs public domain implementation).  */
233 /* #define FF(b, c, d) ((b & c) | (~b & d)) */
234 #define FF(b, c, d) (d ^ (b & (c ^ d)))
235 #define FG(b, c, d) FF (d, b, c)
236 #define FH(b, c, d) (b ^ c ^ d)
237 #define FI(b, c, d) (c ^ (b | ~d))
238 
239 /* Process LEN bytes of BUFFER, accumulating context into CTX.
240    It is assumed that LEN % 64 == 0.  */
241 
242 static void
md5_process_block(const void * buffer,size_t len,struct md5_ctx * ctx)243 md5_process_block (const void *buffer, size_t len, struct md5_ctx *ctx)
244 {
245   md5_uint32 correct_words[16];
246   const md5_uint32 *words = buffer;
247   size_t nwords = len / sizeof (md5_uint32);
248   const md5_uint32 *endp = words + nwords;
249   md5_uint32 A = ctx->A;
250   md5_uint32 B = ctx->B;
251   md5_uint32 C = ctx->C;
252   md5_uint32 D = ctx->D;
253 
254   /* First increment the byte count.  RFC 1321 specifies the possible
255      length of the file up to 2^64 bits.  Here we only compute the
256      number of bytes.  Do a double word increment.  */
257   ctx->total[0] += len;
258   if (ctx->total[0] < len)
259     ++ctx->total[1];
260 
261   /* Process all bytes in the buffer with 64 bytes in each round of
262      the loop.  */
263   while (words < endp)
264     {
265       md5_uint32 *cwp = correct_words;
266       md5_uint32 A_save = A;
267       md5_uint32 B_save = B;
268       md5_uint32 C_save = C;
269       md5_uint32 D_save = D;
270 
271       /* First round: using the given function, the context and a constant
272 	 the next context is computed.  Because the algorithms processing
273 	 unit is a 32-bit word and it is determined to work on words in
274 	 little endian byte order we perhaps have to change the byte order
275 	 before the computation.  To reduce the work for the next steps
276 	 we store the swapped words in the array CORRECT_WORDS.  */
277 
278 #define OP(a, b, c, d, s, T)						\
279       do								\
280         {								\
281 	  a += FF (b, c, d) + (*cwp++ = SWAP (*words)) + T;		\
282 	  ++words;							\
283 	  a = rol (a, s);						\
284 	  a += b;							\
285         }								\
286       while (0)
287 
288       /* Before we start, one word to the strange constants.
289 	 They are defined in RFC 1321 as
290 
291 	 T[i] = (int) (4294967296.0 * fabs (sin (i))), i=1..64, or
292 	 perl -e 'foreach(1..64){printf "0x%08x\n", int (4294967296 * abs (sin $_))}'
293        */
294 
295       /* Round 1.  */
296       OP (A, B, C, D,  7, 0xd76aa478);
297       OP (D, A, B, C, 12, 0xe8c7b756);
298       OP (C, D, A, B, 17, 0x242070db);
299       OP (B, C, D, A, 22, 0xc1bdceee);
300       OP (A, B, C, D,  7, 0xf57c0faf);
301       OP (D, A, B, C, 12, 0x4787c62a);
302       OP (C, D, A, B, 17, 0xa8304613);
303       OP (B, C, D, A, 22, 0xfd469501);
304       OP (A, B, C, D,  7, 0x698098d8);
305       OP (D, A, B, C, 12, 0x8b44f7af);
306       OP (C, D, A, B, 17, 0xffff5bb1);
307       OP (B, C, D, A, 22, 0x895cd7be);
308       OP (A, B, C, D,  7, 0x6b901122);
309       OP (D, A, B, C, 12, 0xfd987193);
310       OP (C, D, A, B, 17, 0xa679438e);
311       OP (B, C, D, A, 22, 0x49b40821);
312 
313       /* For the second to fourth round we have the possibly swapped words
314 	 in CORRECT_WORDS.  Redefine the macro to take an additional first
315 	 argument specifying the function to use.  */
316 #undef OP
317 #define OP(f, a, b, c, d, k, s, T)					\
318       do 								\
319 	{								\
320 	  a += f (b, c, d) + correct_words[k] + T;			\
321 	  a = rol (a, s);						\
322 	  a += b;							\
323 	}								\
324       while (0)
325 
326       /* Round 2.  */
327       OP (FG, A, B, C, D,  1,  5, 0xf61e2562);
328       OP (FG, D, A, B, C,  6,  9, 0xc040b340);
329       OP (FG, C, D, A, B, 11, 14, 0x265e5a51);
330       OP (FG, B, C, D, A,  0, 20, 0xe9b6c7aa);
331       OP (FG, A, B, C, D,  5,  5, 0xd62f105d);
332       OP (FG, D, A, B, C, 10,  9, 0x02441453);
333       OP (FG, C, D, A, B, 15, 14, 0xd8a1e681);
334       OP (FG, B, C, D, A,  4, 20, 0xe7d3fbc8);
335       OP (FG, A, B, C, D,  9,  5, 0x21e1cde6);
336       OP (FG, D, A, B, C, 14,  9, 0xc33707d6);
337       OP (FG, C, D, A, B,  3, 14, 0xf4d50d87);
338       OP (FG, B, C, D, A,  8, 20, 0x455a14ed);
339       OP (FG, A, B, C, D, 13,  5, 0xa9e3e905);
340       OP (FG, D, A, B, C,  2,  9, 0xfcefa3f8);
341       OP (FG, C, D, A, B,  7, 14, 0x676f02d9);
342       OP (FG, B, C, D, A, 12, 20, 0x8d2a4c8a);
343 
344       /* Round 3.  */
345       OP (FH, A, B, C, D,  5,  4, 0xfffa3942);
346       OP (FH, D, A, B, C,  8, 11, 0x8771f681);
347       OP (FH, C, D, A, B, 11, 16, 0x6d9d6122);
348       OP (FH, B, C, D, A, 14, 23, 0xfde5380c);
349       OP (FH, A, B, C, D,  1,  4, 0xa4beea44);
350       OP (FH, D, A, B, C,  4, 11, 0x4bdecfa9);
351       OP (FH, C, D, A, B,  7, 16, 0xf6bb4b60);
352       OP (FH, B, C, D, A, 10, 23, 0xbebfbc70);
353       OP (FH, A, B, C, D, 13,  4, 0x289b7ec6);
354       OP (FH, D, A, B, C,  0, 11, 0xeaa127fa);
355       OP (FH, C, D, A, B,  3, 16, 0xd4ef3085);
356       OP (FH, B, C, D, A,  6, 23, 0x04881d05);
357       OP (FH, A, B, C, D,  9,  4, 0xd9d4d039);
358       OP (FH, D, A, B, C, 12, 11, 0xe6db99e5);
359       OP (FH, C, D, A, B, 15, 16, 0x1fa27cf8);
360       OP (FH, B, C, D, A,  2, 23, 0xc4ac5665);
361 
362       /* Round 4.  */
363       OP (FI, A, B, C, D,  0,  6, 0xf4292244);
364       OP (FI, D, A, B, C,  7, 10, 0x432aff97);
365       OP (FI, C, D, A, B, 14, 15, 0xab9423a7);
366       OP (FI, B, C, D, A,  5, 21, 0xfc93a039);
367       OP (FI, A, B, C, D, 12,  6, 0x655b59c3);
368       OP (FI, D, A, B, C,  3, 10, 0x8f0ccc92);
369       OP (FI, C, D, A, B, 10, 15, 0xffeff47d);
370       OP (FI, B, C, D, A,  1, 21, 0x85845dd1);
371       OP (FI, A, B, C, D,  8,  6, 0x6fa87e4f);
372       OP (FI, D, A, B, C, 15, 10, 0xfe2ce6e0);
373       OP (FI, C, D, A, B,  6, 15, 0xa3014314);
374       OP (FI, B, C, D, A, 13, 21, 0x4e0811a1);
375       OP (FI, A, B, C, D,  4,  6, 0xf7537e82);
376       OP (FI, D, A, B, C, 11, 10, 0xbd3af235);
377       OP (FI, C, D, A, B,  2, 15, 0x2ad7d2bb);
378       OP (FI, B, C, D, A,  9, 21, 0xeb86d391);
379 
380       /* Add the starting values of the context.  */
381       A += A_save;
382       B += B_save;
383       C += C_save;
384       D += D_save;
385     }
386 
387   /* Put checksum in context given as argument.  */
388   ctx->A = A;
389   ctx->B = B;
390   ctx->C = C;
391   ctx->D = D;
392 }
393