sha1_file.c

   1 /*
   2  * GIT - The information manager from hell
   3  *
   4  * Copyright (C) Linus Torvalds, 2005
   5  *
   6  * This handles basic git sha1 object files - packing, unpacking,
   7  * creation etc.
   8  */
   9 #include <stdarg.h>
  10 #include "cache.h"
  11
  12 const char *sha1_file_directory = NULL;
  13
  14 #ifndef O_NOATIME
  15 #if defined(__linux__) && (defined(__i386__) || defined(__PPC__))
  16 #define O_NOATIME 01000000
  17 #else
  18 #define O_NOATIME 0
  19 #endif
  20 #endif
  21
  22 static unsigned int sha1_file_open_flag = O_NOATIME;
  23
  24 static unsigned hexval(char c)
  25 {
  26         if (c >= '0' && c <= '9')
  27                 return c - '0';
  28         if (c >= 'a' && c <= 'f')
  29                 return c - 'a' + 10;
  30         if (c >= 'A' && c <= 'F')
  31                 return c - 'A' + 10;
  32         return ~0;
  33 }
  34
  35 int get_sha1_hex(const char *hex, unsigned char *sha1)
  36 {
  37         int i;
  38         for (i = 0; i < 20; i++) {
  39                 unsigned int val = (hexval(hex[0]) << 4) | hexval(hex[1]);
  40                 if (val & ~0xff)
  41                         return -1;
  42                 *sha1++ = val;
  43                 hex += 2;
  44         }
  45         return 0;
  46 }
  47
  48 int get_sha1_file(const char *path, unsigned char *result)
  49 {
  50         char buffer[60];
  51         int fd = open(path, O_RDONLY);
  52         int len;
  53
  54         if (fd < 0)
  55                 return -1;
  56         len = read(fd, buffer, sizeof(buffer));
  57         close(fd);
  58         if (len < 40)
  59                 return -1;
  60         return get_sha1_hex(buffer, result);
  61 }
  62
  63 int get_sha1(const char *str, unsigned char *sha1)
  64 {
  65         static char pathname[PATH_MAX];
  66         static const char *prefix[] = {
  67                 "",
  68                 "refs",
  69                 "refs/tags",
  70                 "refs/heads",
  71                 "refs/snap",
  72                 NULL
  73         };
  74         const char *gitdir;
  75         const char **p;
  76
  77         if (!get_sha1_hex(str, sha1))
  78                 return 0;
  79
  80         gitdir = ".git";
  81         for (p = prefix; *p; p++) {
  82                 snprintf(pathname, sizeof(pathname), "%s/%s/%s", gitdir, *p, str);
  83                 if (!get_sha1_file(pathname, sha1))
  84                         return 0;
  85         }
  86
  87         return -1;
  88 }
  89
  90 char * sha1_to_hex(const unsigned char *sha1)
  91 {
  92         static char buffer[50];
  93         static const char hex[] = "0123456789abcdef";
  94         char *buf = buffer;
  95         int i;
  96
  97         for (i = 0; i < 20; i++) {
  98                 unsigned int val = *sha1++;
  99                 *buf++ = hex[val >> 4];
 100                 *buf++ = hex[val & 0xf];
 101         }
 102         return buffer;
 103 }
 104
 105 /*
 106  * NOTE! This returns a statically allocated buffer, so you have to be
 107  * careful about using it. Do a "strdup()" if you need to save the
 108  * filename.
 109  */
 110 char *sha1_file_name(const unsigned char *sha1)
 111 {
 112         int i;
 113         static char *name, *base;
 114
 115         if (!base) {
 116                 char *sha1_file_directory = getenv(DB_ENVIRONMENT) ? : DEFAULT_DB_ENVIRONMENT;
 117                 int len = strlen(sha1_file_directory);
 118                 base = xmalloc(len + 60);
 119                 memcpy(base, sha1_file_directory, len);
 120                 memset(base+len, 0, 60);
 121                 base[len] = '/';
 122                 base[len+3] = '/';
 123                 name = base + len + 1;
 124         }
 125         for (i = 0; i < 20; i++) {
 126                 static char hex[] = "0123456789abcdef";
 127                 unsigned int val = sha1[i];
 128                 char *pos = name + i*2 + (i > 0);
 129                 *pos++ = hex[val >> 4];
 130                 *pos = hex[val & 0xf];
 131         }
 132         return base;
 133 }
 134
 135 int check_sha1_signature(unsigned char *sha1, void *map, unsigned long size, const char *type)
 136 {
 137         char header[100];
 138         unsigned char real_sha1[20];
 139         SHA_CTX c;
 140
 141         SHA1_Init(&c);
 142         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 143         SHA1_Update(&c, map, size);
 144         SHA1_Final(real_sha1, &c);
 145         return memcmp(sha1, real_sha1, 20) ? -1 : 0;
 146 }
 147
 148 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 149 {
 150         char *filename = sha1_file_name(sha1);
 151         struct stat st;
 152         void *map;
 153         int fd;
 154
 155         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 156         if (fd < 0) {
 157                 /* See if it works without O_NOATIME */
 158                 switch (sha1_file_open_flag) {
 159                 default:
 160                         fd = open(filename, O_RDONLY);
 161                         if (fd >= 0)
 162                                 break;
 163                 /* Fallthrough */
 164                 case 0:
 165                         perror(filename);
 166                         return NULL;
 167                 }
 168
 169                 /* If it failed once, it will probably fail again. Stop using O_NOATIME */
 170                 sha1_file_open_flag = 0;
 171         }
 172         if (fstat(fd, &st) < 0) {
 173                 close(fd);
 174                 return NULL;
 175         }
 176         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 177         close(fd);
 178         if (-1 == (int)(long)map)
 179                 return NULL;
 180         *size = st.st_size;
 181         return map;
 182 }
 183
 184 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 185 {
 186         int ret, bytes;
 187         z_stream stream;
 188         char buffer[8192];
 189         char *buf;
 190
 191         /* Get the data stream */
 192         memset(&stream, 0, sizeof(stream));
 193         stream.next_in = map;
 194         stream.avail_in = mapsize;
 195         stream.next_out = buffer;
 196         stream.avail_out = sizeof(buffer);
 197
 198         inflateInit(&stream);
 199         ret = inflate(&stream, 0);
 200         if (ret < Z_OK)
 201                 return NULL;
 202         if (sscanf(buffer, "%10s %lu", type, size) != 2)
 203                 return NULL;
 204
 205         bytes = strlen(buffer) + 1;
 206         buf = xmalloc(*size);
 207
 208         memcpy(buf, buffer + bytes, stream.total_out - bytes);
 209         bytes = stream.total_out - bytes;
 210         if (bytes < *size && ret == Z_OK) {
 211                 stream.next_out = buf + bytes;
 212                 stream.avail_out = *size - bytes;
 213                 while (inflate(&stream, Z_FINISH) == Z_OK)
 214                         /* nothing */;
 215         }
 216         inflateEnd(&stream);
 217         return buf;
 218 }
 219
 220 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
 221 {
 222         unsigned long mapsize;
 223         void *map, *buf;
 224
 225         map = map_sha1_file(sha1, &mapsize);
 226         if (map) {
 227                 buf = unpack_sha1_file(map, mapsize, type, size);
 228                 munmap(map, mapsize);
 229                 return buf;
 230         }
 231         return NULL;
 232 }
 233
 234 void *read_object_with_reference(const unsigned char *sha1,
 235                                  const unsigned char *required_type,
 236                                  unsigned long *size,
 237                                  unsigned char *actual_sha1_return)
 238 {
 239         char type[20];
 240         void *buffer;
 241         unsigned long isize;
 242         unsigned char actual_sha1[20];
 243
 244         memcpy(actual_sha1, sha1, 20);
 245         while (1) {
 246                 int ref_length = -1;
 247                 const char *ref_type = NULL;
 248
 249                 buffer = read_sha1_file(actual_sha1, type, &isize);
 250                 if (!buffer)
 251                         return NULL;
 252                 if (!strcmp(type, required_type)) {
 253                         *size = isize;
 254                         if (actual_sha1_return)
 255                                 memcpy(actual_sha1_return, actual_sha1, 20);
 256                         return buffer;
 257                 }
 258                 /* Handle references */
 259                 else if (!strcmp(type, "commit"))
 260                         ref_type = "tree ";
 261                 else if (!strcmp(type, "tag"))
 262                         ref_type = "object ";
 263                 else {
 264                         free(buffer);
 265                         return NULL;
 266                 }
 267                 ref_length = strlen(ref_type);
 268
 269                 if (memcmp(buffer, ref_type, ref_length) ||
 270                     get_sha1_hex(buffer + ref_length, actual_sha1)) {
 271                         free(buffer);
 272                         return NULL;
 273                 }
 274                 /* Now we have the ID of the referred-to object in
 275                  * actual_sha1.  Check again. */
 276         }
 277 }
 278
 279 int write_sha1_file(char *buf, unsigned long len, const char *type, unsigned char *returnsha1)
 280 {
 281         int size;
 282         char *compressed;
 283         z_stream stream;
 284         unsigned char sha1[20];
 285         SHA_CTX c;
 286         char *filename;
 287         char hdr[50];
 288         int fd, hdrlen;
 289
 290         /* Generate the header */
 291         hdrlen = sprintf(hdr, "%s %lu", type, len)+1;
 292
 293         /* Sha1.. */
 294         SHA1_Init(&c);
 295         SHA1_Update(&c, hdr, hdrlen);
 296         SHA1_Update(&c, buf, len);
 297         SHA1_Final(sha1, &c);
 298
 299         if (returnsha1)
 300                 memcpy(returnsha1, sha1, 20);
 301
 302         filename = sha1_file_name(sha1);
 303         fd = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
 304         if (fd < 0) {
 305                 if (errno != EEXIST)
 306                         return -1;
 307
 308                 /*
 309                  * We might do collision checking here, but we'd need to
 310                  * uncompress the old file and check it. Later.
 311                  */
 312                 return 0;
 313         }
 314
 315         /* Set it up */
 316         memset(&stream, 0, sizeof(stream));
 317         deflateInit(&stream, Z_BEST_COMPRESSION);
 318         size = deflateBound(&stream, len+hdrlen);
 319         compressed = xmalloc(size);
 320
 321         /* Compress it */
 322         stream.next_out = compressed;
 323         stream.avail_out = size;
 324
 325         /* First header.. */
 326         stream.next_in = hdr;
 327         stream.avail_in = hdrlen;
 328         while (deflate(&stream, 0) == Z_OK)
 329                 /* nothing */
 330
 331         /* Then the data itself.. */
 332         stream.next_in = buf;
 333         stream.avail_in = len;
 334         while (deflate(&stream, Z_FINISH) == Z_OK)
 335                 /* nothing */;
 336         deflateEnd(&stream);
 337         size = stream.total_out;
 338
 339         if (write(fd, compressed, size) != size)
 340                 die("unable to write file");
 341         close(fd);
 342
 343         return 0;
 344 }
 345
 346 static inline int collision_check(char *filename, void *buf, unsigned int size)
 347 {
 348 #ifdef COLLISION_CHECK
 349         void *map;
 350         int fd = open(filename, O_RDONLY);
 351         struct stat st;
 352         int cmp;
 353
 354         /* Unreadable object, or object went away? Strange. */
 355         if (fd < 0)
 356                 return -1;
 357
 358         if (fstat(fd, &st) < 0 || size != st.st_size)
 359                 return -1;
 360
 361         map = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
 362         close(fd);
 363         if (map == MAP_FAILED)
 364                 return -1;
 365         cmp = memcmp(buf, map, size);
 366         munmap(map, size);
 367         if (cmp)
 368                 return -1;
 369 #endif
 370         return 0;
 371 }
 372
 373 int write_sha1_buffer(const unsigned char *sha1, void *buf, unsigned int size)
 374 {
 375         char *filename = sha1_file_name(sha1);
 376         int fd;
 377
 378         fd = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
 379         if (fd < 0) {
 380                 if (errno != EEXIST)
 381                         return -1;
 382                 if (collision_check(filename, buf, size))
 383                         return error("SHA1 collision detected!"
 384                                         " This is bad, bad, BAD!\a\n");
 385                 return 0;
 386         }
 387         write(fd, buf, size);
 388         close(fd);
 389         return 0;
 390 }
 391
 392 int write_sha1_from_fd(const unsigned char *sha1, int fd)
 393 {
 394         char *filename = sha1_file_name(sha1);
 395
 396         int local;
 397         z_stream stream;
 398         unsigned char real_sha1[20];
 399         char buf[4096];
 400         char discard[4096];
 401         int ret;
 402         SHA_CTX c;
 403
 404         local = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
 405
 406         if (local < 0)
 407                 return error("Couldn't open %s\n", filename);
 408
 409         memset(&stream, 0, sizeof(stream));
 410
 411         inflateInit(&stream);
 412
 413         SHA1_Init(&c);
 414
 415         do {
 416                 ssize_t size;
 417                 size = read(fd, buf, 4096);
 418                 if (size <= 0) {
 419                         close(local);
 420                         unlink(filename);
 421                         if (!size)
 422                                 return error("Connection closed?");
 423                         perror("Reading from connection");
 424                         return -1;
 425                 }
 426                 write(local, buf, size);
 427                 stream.avail_in = size;
 428                 stream.next_in = buf;
 429                 do {
 430                         stream.next_out = discard;
 431                         stream.avail_out = sizeof(discard);
 432                         ret = inflate(&stream, Z_SYNC_FLUSH);
 433                         SHA1_Update(&c, discard, sizeof(discard) -
 434                                     stream.avail_out);
 435                 } while (stream.avail_in && ret == Z_OK);
 436
 437         } while (ret == Z_OK);
 438         inflateEnd(&stream);
 439
 440         close(local);
 441         SHA1_Final(real_sha1, &c);
 442         if (ret != Z_STREAM_END) {
 443                 unlink(filename);
 444                 return error("File %s corrupted", sha1_to_hex(sha1));
 445         }
 446         if (memcmp(sha1, real_sha1, 20)) {
 447                 unlink(filename);
 448                 return error("File %s has bad hash\n", sha1_to_hex(sha1));
 449         }
 450
 451         return 0;
 452 }
 453
 454 int has_sha1_file(const unsigned char *sha1)
 455 {
 456         char *filename = sha1_file_name(sha1);
 457         struct stat st;
 458
 459         if (!stat(filename, &st))
 460                 return 1;
 461         return 0;
 462 }
 463
 464 int index_fd(unsigned char *sha1, int fd, struct stat *st)
 465 {
 466         z_stream stream;
 467         unsigned long size = st->st_size;
 468         int max_out_bytes = size + 200;
 469         void *out = xmalloc(max_out_bytes);
 470         void *metadata = xmalloc(200);
 471         int metadata_size;
 472         void *in;
 473         SHA_CTX c;
 474
 475         in = "";
 476         if (size)
 477                 in = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
 478         close(fd);
 479         if (!out || (int)(long)in == -1)
 480                 return -1;
 481
 482         metadata_size = 1+sprintf(metadata, "blob %lu", size);
 483
 484         SHA1_Init(&c);
 485         SHA1_Update(&c, metadata, metadata_size);
 486         SHA1_Update(&c, in, size);
 487         SHA1_Final(sha1, &c);
 488
 489         memset(&stream, 0, sizeof(stream));
 490         deflateInit(&stream, Z_BEST_COMPRESSION);
 491
 492         /*
 493          * ASCII size + nul byte
 494          */
 495         stream.next_in = metadata;
 496         stream.avail_in = metadata_size;
 497         stream.next_out = out;
 498         stream.avail_out = max_out_bytes;
 499         while (deflate(&stream, 0) == Z_OK)
 500                 /* nothing */;
 501
 502         /*
 503          * File content
 504          */
 505         stream.next_in = in;
 506         stream.avail_in = size;
 507         while (deflate(&stream, Z_FINISH) == Z_OK)
 508                 /*nothing */;
 509
 510         deflateEnd(&stream);
 511
 512         return write_sha1_buffer(sha1, out, stream.total_out);
 513 }