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 <limits.h>
  11 #include "cache.h"
  12
  13 #ifndef O_NOATIME
  14 #if defined(__linux__) && (defined(__i386__) || defined(__PPC__))
  15 #define O_NOATIME 01000000
  16 #else
  17 #define O_NOATIME 0
  18 #endif
  19 #endif
  20
  21 static unsigned int sha1_file_open_flag = O_NOATIME;
  22
  23 static unsigned hexval(char c)
  24 {
  25         if (c >= '0' && c <= '9')
  26                 return c - '0';
  27         if (c >= 'a' && c <= 'f')
  28                 return c - 'a' + 10;
  29         if (c >= 'A' && c <= 'F')
  30                 return c - 'A' + 10;
  31         return ~0;
  32 }
  33
  34 int get_sha1_hex(const char *hex, unsigned char *sha1)
  35 {
  36         int i;
  37         for (i = 0; i < 20; i++) {
  38                 unsigned int val = (hexval(hex[0]) << 4) | hexval(hex[1]);
  39                 if (val & ~0xff)
  40                         return -1;
  41                 *sha1++ = val;
  42                 hex += 2;
  43         }
  44         return 0;
  45 }
  46
  47 int get_sha1_file(const char *path, unsigned char *result)
  48 {
  49         char buffer[60];
  50         int fd = open(path, O_RDONLY);
  51         int len;
  52
  53         if (fd < 0)
  54                 return -1;
  55         len = read(fd, buffer, sizeof(buffer));
  56         close(fd);
  57         if (len < 40)
  58                 return -1;
  59         return get_sha1_hex(buffer, result);
  60 }
  61
  62 int get_sha1(const char *str, unsigned char *sha1)
  63 {
  64         static char pathname[PATH_MAX];
  65         static const char *prefix[] = {
  66                 "",
  67                 "refs",
  68                 "refs/tags",
  69                 "refs/heads",
  70                 "refs/snap",
  71                 NULL
  72         };
  73         const char *gitdir;
  74         const char **p;
  75
  76         if (!get_sha1_hex(str, sha1))
  77                 return 0;
  78
  79         gitdir = ".git";
  80         for (p = prefix; *p; p++) {
  81                 snprintf(pathname, sizeof(pathname), "%s/%s/%s", gitdir, *p, str);
  82                 if (!get_sha1_file(pathname, sha1))
  83                         return 0;
  84         }
  85
  86         return -1;
  87 }
  88
  89 char * sha1_to_hex(const unsigned char *sha1)
  90 {
  91         static char buffer[50];
  92         static const char hex[] = "0123456789abcdef";
  93         char *buf = buffer;
  94         int i;
  95
  96         for (i = 0; i < 20; i++) {
  97                 unsigned int val = *sha1++;
  98                 *buf++ = hex[val >> 4];
  99                 *buf++ = hex[val & 0xf];
 100         }
 101         return buffer;
 102 }
 103
 104 static void fill_sha1_path(char *pathbuf, const unsigned char *sha1)
 105 {
 106         int i;
 107         for (i = 0; i < 20; i++) {
 108                 static char hex[] = "0123456789abcdef";
 109                 unsigned int val = sha1[i];
 110                 char *pos = pathbuf + i*2 + (i > 0);
 111                 *pos++ = hex[val >> 4];
 112                 *pos = hex[val & 0xf];
 113         }
 114 }
 115
 116 /*
 117  * NOTE! This returns a statically allocated buffer, so you have to be
 118  * careful about using it. Do a "strdup()" if you need to save the
 119  * filename.
 120  *
 121  * Also note that this returns the location for creating.  Reading
 122  * SHA1 file can happen from any alternate directory listed in the
 123  * SHA1_FILE_DIRECTORIES environment variable if it is not found in
 124  * the primary object database.
 125  */
 126 char *sha1_file_name(const unsigned char *sha1)
 127 {
 128         static char *name, *base;
 129
 130         if (!base) {
 131                 char *sha1_file_directory = get_object_directory();
 132                 int len = strlen(sha1_file_directory);
 133                 base = xmalloc(len + 60);
 134                 memcpy(base, sha1_file_directory, len);
 135                 memset(base+len, 0, 60);
 136                 base[len] = '/';
 137                 base[len+3] = '/';
 138                 name = base + len + 1;
 139         }
 140         fill_sha1_path(name, sha1);
 141         return base;
 142 }
 143
 144 static struct alternate_object_database
 145 {
 146         char *base;
 147         char *name;
 148 } *alt_odb;
 149
 150 static void prepare_alt_odb(void)
 151 {
 152         int pass, totlen, i;
 153         void *buf;
 154         const char *cp, *last;
 155         char *op = 0;
 156         const char *alt = getenv(ALTERNATE_DB_ENVIRONMENT) ? : "";
 157
 158         for (totlen = pass = 0; pass < 2; pass++) {
 159                 last = alt;
 160                 i = 0;
 161                 do {
 162                         cp = strchr(last, ':') ? : last + strlen(last);
 163                         if (last != cp) {
 164                                 /* 43 = 40-byte + 2 '/' + terminating NUL */
 165                                 int pfxlen = cp - last;
 166                                 int entlen = pfxlen + 43;
 167                                 if (pass == 0)
 168                                         totlen += entlen;
 169                                 else {
 170                                         alt_odb[i].base = op;
 171                                         alt_odb[i].name = op + pfxlen + 1;
 172                                         memcpy(op, last, pfxlen);
 173                                         op[pfxlen] = op[pfxlen + 3] = '/';
 174                                         op[entlen-1] = 0;
 175                                         op += entlen;
 176                                 }
 177                                 i++;
 178                         }
 179                         while (*cp && *cp == ':')
 180                                 cp++;
 181                         last = cp;
 182                 } while (*cp);
 183                 if (pass)
 184                         break;
 185                 alt_odb = buf = xmalloc(sizeof(*alt_odb) * (i + 1) + totlen);
 186                 alt_odb[i].base = alt_odb[i].name = 0;
 187                 op = (char*)(&alt_odb[i+1]);
 188         }
 189 }
 190
 191 static char *find_sha1_file(const unsigned char *sha1, struct stat *st)
 192 {
 193         int i;
 194         char *name = sha1_file_name(sha1);
 195
 196         if (!stat(name, st))
 197                 return name;
 198         if (!alt_odb)
 199                 prepare_alt_odb();
 200         for (i = 0; (name = alt_odb[i].name) != NULL; i++) {
 201                 fill_sha1_path(name, sha1);
 202                 if (!stat(alt_odb[i].base, st))
 203                         return alt_odb[i].base;
 204         }
 205         return NULL;
 206 }
 207
 208 int check_sha1_signature(unsigned char *sha1, void *map, unsigned long size, const char *type)
 209 {
 210         char header[100];
 211         unsigned char real_sha1[20];
 212         SHA_CTX c;
 213
 214         SHA1_Init(&c);
 215         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 216         SHA1_Update(&c, map, size);
 217         SHA1_Final(real_sha1, &c);
 218         return memcmp(sha1, real_sha1, 20) ? -1 : 0;
 219 }
 220
 221 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 222 {
 223         struct stat st;
 224         void *map;
 225         int fd;
 226         char *filename = find_sha1_file(sha1, &st);
 227
 228         if (!filename) {
 229                 error("cannot map sha1 file %s", sha1_to_hex(sha1));
 230                 return NULL;
 231         }
 232
 233         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 234         if (fd < 0) {
 235                 /* See if it works without O_NOATIME */
 236                 switch (sha1_file_open_flag) {
 237                 default:
 238                         fd = open(filename, O_RDONLY);
 239                         if (fd >= 0)
 240                                 break;
 241                 /* Fallthrough */
 242                 case 0:
 243                         perror(filename);
 244                         return NULL;
 245                 }
 246
 247                 /* If it failed once, it will probably fail again. Stop using O_NOATIME */
 248                 sha1_file_open_flag = 0;
 249         }
 250         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 251         close(fd);
 252         if (-1 == (int)(long)map)
 253                 return NULL;
 254         *size = st.st_size;
 255         return map;
 256 }
 257
 258 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 259 {
 260         int ret, bytes;
 261         z_stream stream;
 262         char buffer[8192];
 263         char *buf;
 264
 265         /* Get the data stream */
 266         memset(&stream, 0, sizeof(stream));
 267         stream.next_in = map;
 268         stream.avail_in = mapsize;
 269         stream.next_out = buffer;
 270         stream.avail_out = sizeof(buffer);
 271
 272         inflateInit(&stream);
 273         ret = inflate(&stream, 0);
 274         if (ret < Z_OK)
 275                 return NULL;
 276         if (sscanf(buffer, "%10s %lu", type, size) != 2)
 277                 return NULL;
 278
 279         bytes = strlen(buffer) + 1;
 280         buf = xmalloc(*size);
 281
 282         memcpy(buf, buffer + bytes, stream.total_out - bytes);
 283         bytes = stream.total_out - bytes;
 284         if (bytes < *size && ret == Z_OK) {
 285                 stream.next_out = buf + bytes;
 286                 stream.avail_out = *size - bytes;
 287                 while (inflate(&stream, Z_FINISH) == Z_OK)
 288                         /* nothing */;
 289         }
 290         inflateEnd(&stream);
 291         return buf;
 292 }
 293
 294 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
 295 {
 296         unsigned long mapsize;
 297         void *map, *buf;
 298
 299         map = map_sha1_file(sha1, &mapsize);
 300         if (map) {
 301                 buf = unpack_sha1_file(map, mapsize, type, size);
 302                 munmap(map, mapsize);
 303                 return buf;
 304         }
 305         return NULL;
 306 }
 307
 308 void *read_object_with_reference(const unsigned char *sha1,
 309                                  const unsigned char *required_type,
 310                                  unsigned long *size,
 311                                  unsigned char *actual_sha1_return)
 312 {
 313         char type[20];
 314         void *buffer;
 315         unsigned long isize;
 316         unsigned char actual_sha1[20];
 317
 318         memcpy(actual_sha1, sha1, 20);
 319         while (1) {
 320                 int ref_length = -1;
 321                 const char *ref_type = NULL;
 322
 323                 buffer = read_sha1_file(actual_sha1, type, &isize);
 324                 if (!buffer)
 325                         return NULL;
 326                 if (!strcmp(type, required_type)) {
 327                         *size = isize;
 328                         if (actual_sha1_return)
 329                                 memcpy(actual_sha1_return, actual_sha1, 20);
 330                         return buffer;
 331                 }
 332                 /* Handle references */
 333                 else if (!strcmp(type, "commit"))
 334                         ref_type = "tree ";
 335                 else if (!strcmp(type, "tag"))
 336                         ref_type = "object ";
 337                 else {
 338                         free(buffer);
 339                         return NULL;
 340                 }
 341                 ref_length = strlen(ref_type);
 342
 343                 if (memcmp(buffer, ref_type, ref_length) ||
 344                     get_sha1_hex(buffer + ref_length, actual_sha1)) {
 345                         free(buffer);
 346                         return NULL;
 347                 }
 348                 /* Now we have the ID of the referred-to object in
 349                  * actual_sha1.  Check again. */
 350         }
 351 }
 352
 353 int write_sha1_file(char *buf, unsigned long len, const char *type, unsigned char *returnsha1)
 354 {
 355         int size;
 356         char *compressed;
 357         z_stream stream;
 358         unsigned char sha1[20];
 359         SHA_CTX c;
 360         char *filename;
 361         static char tmpfile[PATH_MAX];
 362         char hdr[50];
 363         int fd, hdrlen, ret;
 364
 365         /* Generate the header */
 366         hdrlen = sprintf(hdr, "%s %lu", type, len)+1;
 367
 368         /* Sha1.. */
 369         SHA1_Init(&c);
 370         SHA1_Update(&c, hdr, hdrlen);
 371         SHA1_Update(&c, buf, len);
 372         SHA1_Final(sha1, &c);
 373
 374         if (returnsha1)
 375                 memcpy(returnsha1, sha1, 20);
 376
 377         filename = sha1_file_name(sha1);
 378         fd = open(filename, O_RDONLY);
 379         if (fd >= 0) {
 380                 /*
 381                  * FIXME!!! We might do collision checking here, but we'd
 382                  * need to uncompress the old file and check it. Later.
 383                  */
 384                 close(fd);
 385                 return 0;
 386         }
 387
 388         if (errno != ENOENT) {
 389                 fprintf(stderr, "sha1 file %s: %s", filename, strerror(errno));
 390                 return -1;
 391         }
 392
 393         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
 394
 395         fd = mkstemp(tmpfile);
 396         if (fd < 0) {
 397                 fprintf(stderr, "unable to create temporary sha1 filename %s: %s", tmpfile, strerror(errno));
 398                 return -1;
 399         }
 400
 401         /* Set it up */
 402         memset(&stream, 0, sizeof(stream));
 403         deflateInit(&stream, Z_BEST_COMPRESSION);
 404         size = deflateBound(&stream, len+hdrlen);
 405         compressed = xmalloc(size);
 406
 407         /* Compress it */
 408         stream.next_out = compressed;
 409         stream.avail_out = size;
 410
 411         /* First header.. */
 412         stream.next_in = hdr;
 413         stream.avail_in = hdrlen;
 414         while (deflate(&stream, 0) == Z_OK)
 415                 /* nothing */
 416
 417         /* Then the data itself.. */
 418         stream.next_in = buf;
 419         stream.avail_in = len;
 420         while (deflate(&stream, Z_FINISH) == Z_OK)
 421                 /* nothing */;
 422         deflateEnd(&stream);
 423         size = stream.total_out;
 424
 425         if (write(fd, compressed, size) != size)
 426                 die("unable to write file");
 427         fchmod(fd, 0444);
 428         close(fd);
 429         free(compressed);
 430
 431         ret = link(tmpfile, filename);
 432         if (ret < 0) {
 433                 ret = errno;
 434
 435                 /*
 436                  * Coda hack - coda doesn't like cross-directory links,
 437                  * so we fall back to a rename, which will mean that it
 438                  * won't be able to check collisions, but that's not a
 439                  * big deal.
 440                  *
 441                  * When this succeeds, we just return 0. We have nothing
 442                  * left to unlink.
 443                  */
 444                 if (ret == EXDEV && !rename(tmpfile, filename))
 445                         return 0;
 446         }
 447         unlink(tmpfile);
 448         if (ret) {
 449                 if (ret != EEXIST) {
 450                         fprintf(stderr, "unable to write sha1 filename %s: %s", filename, strerror(ret));
 451                         return -1;
 452                 }
 453                 /* FIXME!!! Collision check here ? */
 454         }
 455
 456         return 0;
 457 }
 458
 459 int write_sha1_from_fd(const unsigned char *sha1, int fd)
 460 {
 461         char *filename = sha1_file_name(sha1);
 462
 463         int local;
 464         z_stream stream;
 465         unsigned char real_sha1[20];
 466         char buf[4096];
 467         char discard[4096];
 468         int ret;
 469         SHA_CTX c;
 470
 471         local = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
 472
 473         if (local < 0)
 474                 return error("Couldn't open %s\n", filename);
 475
 476         memset(&stream, 0, sizeof(stream));
 477
 478         inflateInit(&stream);
 479
 480         SHA1_Init(&c);
 481
 482         do {
 483                 ssize_t size;
 484                 size = read(fd, buf, 4096);
 485                 if (size <= 0) {
 486                         close(local);
 487                         unlink(filename);
 488                         if (!size)
 489                                 return error("Connection closed?");
 490                         perror("Reading from connection");
 491                         return -1;
 492                 }
 493                 write(local, buf, size);
 494                 stream.avail_in = size;
 495                 stream.next_in = buf;
 496                 do {
 497                         stream.next_out = discard;
 498                         stream.avail_out = sizeof(discard);
 499                         ret = inflate(&stream, Z_SYNC_FLUSH);
 500                         SHA1_Update(&c, discard, sizeof(discard) -
 501                                     stream.avail_out);
 502                 } while (stream.avail_in && ret == Z_OK);
 503
 504         } while (ret == Z_OK);
 505         inflateEnd(&stream);
 506
 507         close(local);
 508         SHA1_Final(real_sha1, &c);
 509         if (ret != Z_STREAM_END) {
 510                 unlink(filename);
 511                 return error("File %s corrupted", sha1_to_hex(sha1));
 512         }
 513         if (memcmp(sha1, real_sha1, 20)) {
 514                 unlink(filename);
 515                 return error("File %s has bad hash\n", sha1_to_hex(sha1));
 516         }
 517
 518         return 0;
 519 }
 520
 521 int has_sha1_file(const unsigned char *sha1)
 522 {
 523         struct stat st;
 524         return !!find_sha1_file(sha1, &st);
 525 }
 526
 527 int index_fd(unsigned char *sha1, int fd, struct stat *st)
 528 {
 529         unsigned long size = st->st_size;
 530         void *buf;
 531         int ret;
 532
 533         buf = "";
 534         if (size)
 535                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
 536         close(fd);
 537         if ((int)(long)buf == -1)
 538                 return -1;
 539
 540         ret = write_sha1_file(buf, size, "blob", sha1);
 541         if (size)
 542                 munmap(buf, size);
 543         return ret;
 544 }