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 <sys/types.h>
  10 #include <dirent.h>
  11 #include "cache.h"
  12 #include "delta.h"
  13 #include "pack.h"
  14
  15 #ifndef O_NOATIME
  16 #if defined(__linux__) && (defined(__i386__) || defined(__PPC__))
  17 #define O_NOATIME 01000000
  18 #else
  19 #define O_NOATIME 0
  20 #endif
  21 #endif
  22
  23 static unsigned int sha1_file_open_flag = O_NOATIME;
  24
  25 static unsigned hexval(char c)
  26 {
  27         if (c >= '0' && c <= '9')
  28                 return c - '0';
  29         if (c >= 'a' && c <= 'f')
  30                 return c - 'a' + 10;
  31         if (c >= 'A' && c <= 'F')
  32                 return c - 'A' + 10;
  33         return ~0;
  34 }
  35
  36 int get_sha1_hex(const char *hex, unsigned char *sha1)
  37 {
  38         int i;
  39         for (i = 0; i < 20; i++) {
  40                 unsigned int val = (hexval(hex[0]) << 4) | hexval(hex[1]);
  41                 if (val & ~0xff)
  42                         return -1;
  43                 *sha1++ = val;
  44                 hex += 2;
  45         }
  46         return 0;
  47 }
  48
  49 static int get_sha1_file(const char *path, unsigned char *result)
  50 {
  51         char buffer[60];
  52         int fd = open(path, O_RDONLY);
  53         int len;
  54
  55         if (fd < 0)
  56                 return -1;
  57         len = read(fd, buffer, sizeof(buffer));
  58         close(fd);
  59         if (len < 40)
  60                 return -1;
  61         return get_sha1_hex(buffer, result);
  62 }
  63
  64 static char *git_dir, *git_object_dir, *git_index_file, *git_refs_dir;
  65 static void setup_git_env(void)
  66 {
  67         git_dir = gitenv(GIT_DIR_ENVIRONMENT);
  68         if (!git_dir)
  69                 git_dir = DEFAULT_GIT_DIR_ENVIRONMENT;
  70         git_object_dir = gitenv(DB_ENVIRONMENT);
  71         if (!git_object_dir) {
  72                 git_object_dir = xmalloc(strlen(git_dir) + 9);
  73                 sprintf(git_object_dir, "%s/objects", git_dir);
  74         }
  75         git_refs_dir = xmalloc(strlen(git_dir) + 6);
  76         sprintf(git_refs_dir, "%s/refs", git_dir);
  77         git_index_file = gitenv(INDEX_ENVIRONMENT);
  78         if (!git_index_file) {
  79                 git_index_file = xmalloc(strlen(git_dir) + 7);
  80                 sprintf(git_index_file, "%s/index", git_dir);
  81         }
  82 }
  83
  84 char *get_object_directory(void)
  85 {
  86         if (!git_object_dir)
  87                 setup_git_env();
  88         return git_object_dir;
  89 }
  90
  91 char *get_refs_directory(void)
  92 {
  93         if (!git_refs_dir)
  94                 setup_git_env();
  95         return git_refs_dir;
  96 }
  97
  98 char *get_index_file(void)
  99 {
 100         if (!git_index_file)
 101                 setup_git_env();
 102         return git_index_file;
 103 }
 104
 105 int get_sha1(const char *str, unsigned char *sha1)
 106 {
 107         static char pathname[PATH_MAX];
 108         static const char *prefix[] = {
 109                 "",
 110                 "refs",
 111                 "refs/tags",
 112                 "refs/heads",
 113                 "refs/snap",
 114                 NULL
 115         };
 116         const char **p;
 117
 118         if (!get_sha1_hex(str, sha1))
 119                 return 0;
 120
 121         if (!git_dir)
 122                 setup_git_env();
 123         for (p = prefix; *p; p++) {
 124                 snprintf(pathname, sizeof(pathname), "%s/%s/%s",
 125                          git_dir, *p, str);
 126                 if (!get_sha1_file(pathname, sha1))
 127                         return 0;
 128         }
 129
 130         return -1;
 131 }
 132
 133 char * sha1_to_hex(const unsigned char *sha1)
 134 {
 135         static char buffer[50];
 136         static const char hex[] = "0123456789abcdef";
 137         char *buf = buffer;
 138         int i;
 139
 140         for (i = 0; i < 20; i++) {
 141                 unsigned int val = *sha1++;
 142                 *buf++ = hex[val >> 4];
 143                 *buf++ = hex[val & 0xf];
 144         }
 145         return buffer;
 146 }
 147
 148 static void fill_sha1_path(char *pathbuf, const unsigned char *sha1)
 149 {
 150         int i;
 151         for (i = 0; i < 20; i++) {
 152                 static char hex[] = "0123456789abcdef";
 153                 unsigned int val = sha1[i];
 154                 char *pos = pathbuf + i*2 + (i > 0);
 155                 *pos++ = hex[val >> 4];
 156                 *pos = hex[val & 0xf];
 157         }
 158 }
 159
 160 /*
 161  * NOTE! This returns a statically allocated buffer, so you have to be
 162  * careful about using it. Do a "strdup()" if you need to save the
 163  * filename.
 164  *
 165  * Also note that this returns the location for creating.  Reading
 166  * SHA1 file can happen from any alternate directory listed in the
 167  * DB_ENVIRONMENT environment variable if it is not found in
 168  * the primary object database.
 169  */
 170 char *sha1_file_name(const unsigned char *sha1)
 171 {
 172         static char *name, *base;
 173
 174         if (!base) {
 175                 const char *sha1_file_directory = get_object_directory();
 176                 int len = strlen(sha1_file_directory);
 177                 base = xmalloc(len + 60);
 178                 memcpy(base, sha1_file_directory, len);
 179                 memset(base+len, 0, 60);
 180                 base[len] = '/';
 181                 base[len+3] = '/';
 182                 name = base + len + 1;
 183         }
 184         fill_sha1_path(name, sha1);
 185         return base;
 186 }
 187
 188 struct alternate_object_database *alt_odb;
 189
 190 /*
 191  * Prepare alternate object database registry.
 192  * alt_odb points at an array of struct alternate_object_database.
 193  * This array is terminated with an element that has both its base
 194  * and name set to NULL.  alt_odb[n] comes from n'th non-empty
 195  * element from colon separated ALTERNATE_DB_ENVIRONMENT environment
 196  * variable, and its base points at a statically allocated buffer
 197  * that contains "/the/directory/corresponding/to/.git/objects/...",
 198  * while its name points just after the slash at the end of
 199  * ".git/objects/" in the example above, and has enough space to hold
 200  * 40-byte hex SHA1, an extra slash for the first level indirection,
 201  * and the terminating NUL.
 202  * This function allocates the alt_odb array and all the strings
 203  * pointed by base fields of the array elements with one xmalloc();
 204  * the string pool immediately follows the array.
 205  */
 206 void prepare_alt_odb(void)
 207 {
 208         int pass, totlen, i;
 209         const char *cp, *last;
 210         char *op = NULL;
 211         const char *alt = gitenv(ALTERNATE_DB_ENVIRONMENT) ? : "";
 212
 213         if (alt_odb)
 214                 return;
 215         /* The first pass counts how large an area to allocate to
 216          * hold the entire alt_odb structure, including array of
 217          * structs and path buffers for them.  The second pass fills
 218          * the structure and prepares the path buffers for use by
 219          * fill_sha1_path().
 220          */
 221         for (totlen = pass = 0; pass < 2; pass++) {
 222                 last = alt;
 223                 i = 0;
 224                 do {
 225                         cp = strchr(last, ':') ? : last + strlen(last);
 226                         if (last != cp) {
 227                                 /* 43 = 40-byte + 2 '/' + terminating NUL */
 228                                 int pfxlen = cp - last;
 229                                 int entlen = pfxlen + 43;
 230                                 if (pass == 0)
 231                                         totlen += entlen;
 232                                 else {
 233                                         alt_odb[i].base = op;
 234                                         alt_odb[i].name = op + pfxlen + 1;
 235                                         memcpy(op, last, pfxlen);
 236                                         op[pfxlen] = op[pfxlen + 3] = '/';
 237                                         op[entlen-1] = 0;
 238                                         op += entlen;
 239                                 }
 240                                 i++;
 241                         }
 242                         while (*cp && *cp == ':')
 243                                 cp++;
 244                         last = cp;
 245                 } while (*cp);
 246                 if (pass)
 247                         break;
 248                 alt_odb = xmalloc(sizeof(*alt_odb) * (i + 1) + totlen);
 249                 alt_odb[i].base = alt_odb[i].name = NULL;
 250                 op = (char*)(&alt_odb[i+1]);
 251         }
 252 }
 253
 254 static char *find_sha1_file(const unsigned char *sha1, struct stat *st)
 255 {
 256         int i;
 257         char *name = sha1_file_name(sha1);
 258
 259         if (!stat(name, st))
 260                 return name;
 261         prepare_alt_odb();
 262         for (i = 0; (name = alt_odb[i].name) != NULL; i++) {
 263                 fill_sha1_path(name, sha1);
 264                 if (!stat(alt_odb[i].base, st))
 265                         return alt_odb[i].base;
 266         }
 267         return NULL;
 268 }
 269
 270 #define PACK_MAX_SZ (1<<26)
 271 static int pack_used_ctr;
 272 static unsigned long pack_mapped;
 273 struct packed_git *packed_git;
 274
 275 static int check_packed_git_idx(const char *path, unsigned long *idx_size_,
 276                                 void **idx_map_)
 277 {
 278         void *idx_map;
 279         unsigned int *index;
 280         unsigned long idx_size;
 281         int nr, i;
 282         int fd = open(path, O_RDONLY);
 283         struct stat st;
 284         if (fd < 0)
 285                 return -1;
 286         if (fstat(fd, &st)) {
 287                 close(fd);
 288                 return -1;
 289         }
 290         idx_size = st.st_size;
 291         idx_map = mmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0);
 292         close(fd);
 293         if (idx_map == MAP_FAILED)
 294                 return -1;
 295
 296         index = idx_map;
 297         *idx_map_ = idx_map;
 298         *idx_size_ = idx_size;
 299
 300         /* check index map */
 301         if (idx_size < 4*256 + 20 + 20)
 302                 return error("index file too small");
 303         nr = 0;
 304         for (i = 0; i < 256; i++) {
 305                 unsigned int n = ntohl(index[i]);
 306                 if (n < nr)
 307                         return error("non-monotonic index");
 308                 nr = n;
 309         }
 310
 311         /*
 312          * Total size:
 313          *  - 256 index entries 4 bytes each
 314          *  - 24-byte entries * nr (20-byte sha1 + 4-byte offset)
 315          *  - 20-byte SHA1 of the packfile
 316          *  - 20-byte SHA1 file checksum
 317          */
 318         if (idx_size != 4*256 + nr * 24 + 20 + 20)
 319                 return error("wrong index file size");
 320
 321         return 0;
 322 }
 323
 324 static int unuse_one_packed_git(void)
 325 {
 326         struct packed_git *p, *lru = NULL;
 327
 328         for (p = packed_git; p; p = p->next) {
 329                 if (p->pack_use_cnt || !p->pack_base)
 330                         continue;
 331                 if (!lru || p->pack_last_used < lru->pack_last_used)
 332                         lru = p;
 333         }
 334         if (!lru)
 335                 return 0;
 336         munmap(lru->pack_base, lru->pack_size);
 337         lru->pack_base = NULL;
 338         return 1;
 339 }
 340
 341 void unuse_packed_git(struct packed_git *p)
 342 {
 343         p->pack_use_cnt--;
 344 }
 345
 346 int use_packed_git(struct packed_git *p)
 347 {
 348         if (!p->pack_base) {
 349                 int fd;
 350                 struct stat st;
 351                 void *map;
 352
 353                 pack_mapped += p->pack_size;
 354                 while (PACK_MAX_SZ < pack_mapped && unuse_one_packed_git())
 355                         ; /* nothing */
 356                 fd = open(p->pack_name, O_RDONLY);
 357                 if (fd < 0)
 358                         die("packfile %s cannot be opened", p->pack_name);
 359                 if (fstat(fd, &st)) {
 360                         close(fd);
 361                         die("packfile %s cannot be opened", p->pack_name);
 362                 }
 363                 if (st.st_size != p->pack_size)
 364                         die("packfile %s size mismatch.", p->pack_name);
 365                 map = mmap(NULL, p->pack_size, PROT_READ, MAP_PRIVATE, fd, 0);
 366                 close(fd);
 367                 if (map == MAP_FAILED)
 368                         die("packfile %s cannot be mapped.", p->pack_name);
 369                 p->pack_base = map;
 370
 371                 /* Check if the pack file matches with the index file.
 372                  * this is cheap.
 373                  */
 374                 if (memcmp((char*)(p->index_base) + p->index_size - 40,
 375                            p->pack_base + p->pack_size - 20, 20))
 376                         die("packfile %s does not match index.", p->pack_name);
 377         }
 378         p->pack_last_used = pack_used_ctr++;
 379         p->pack_use_cnt++;
 380         return 0;
 381 }
 382
 383 struct packed_git *add_packed_git(char *path, int path_len)
 384 {
 385         struct stat st;
 386         struct packed_git *p;
 387         unsigned long idx_size;
 388         void *idx_map;
 389
 390         if (check_packed_git_idx(path, &idx_size, &idx_map))
 391                 return NULL;
 392
 393         /* do we have a corresponding .pack file? */
 394         strcpy(path + path_len - 4, ".pack");
 395         if (stat(path, &st) || !S_ISREG(st.st_mode)) {
 396                 munmap(idx_map, idx_size);
 397                 return NULL;
 398         }
 399         /* ok, it looks sane as far as we can check without
 400          * actually mapping the pack file.
 401          */
 402         p = xmalloc(sizeof(*p) + path_len + 2);
 403         strcpy(p->pack_name, path);
 404         p->index_size = idx_size;
 405         p->pack_size = st.st_size;
 406         p->index_base = idx_map;
 407         p->next = NULL;
 408         p->pack_base = NULL;
 409         p->pack_last_used = 0;
 410         p->pack_use_cnt = 0;
 411         return p;
 412 }
 413
 414 static void prepare_packed_git_one(char *objdir)
 415 {
 416         char path[PATH_MAX];
 417         int len;
 418         DIR *dir;
 419         struct dirent *de;
 420
 421         sprintf(path, "%s/pack", objdir);
 422         len = strlen(path);
 423         dir = opendir(path);
 424         if (!dir)
 425                 return;
 426         path[len++] = '/';
 427         while ((de = readdir(dir)) != NULL) {
 428                 int namelen = strlen(de->d_name);
 429                 struct packed_git *p;
 430
 431                 if (strcmp(de->d_name + namelen - 4, ".idx"))
 432                         continue;
 433
 434                 /* we have .idx.  Is it a file we can map? */
 435                 strcpy(path + len, de->d_name);
 436                 p = add_packed_git(path, len + namelen);
 437                 if (!p)
 438                         continue;
 439                 p->next = packed_git;
 440                 packed_git = p;
 441         }
 442 }
 443
 444 void prepare_packed_git(void)
 445 {
 446         int i;
 447         static int run_once = 0;
 448
 449         if (run_once++)
 450                 return;
 451
 452         prepare_packed_git_one(get_object_directory());
 453         prepare_alt_odb();
 454         for (i = 0; alt_odb[i].base != NULL; i++) {
 455                 alt_odb[i].name[0] = 0;
 456                 prepare_packed_git_one(alt_odb[i].base);
 457         }
 458 }
 459
 460 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 461 {
 462         char header[100];
 463         unsigned char real_sha1[20];
 464         SHA_CTX c;
 465
 466         SHA1_Init(&c);
 467         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 468         SHA1_Update(&c, map, size);
 469         SHA1_Final(real_sha1, &c);
 470         return memcmp(sha1, real_sha1, 20) ? -1 : 0;
 471 }
 472
 473 static void *map_sha1_file_internal(const unsigned char *sha1,
 474                                     unsigned long *size,
 475                                     int say_error)
 476 {
 477         struct stat st;
 478         void *map;
 479         int fd;
 480         char *filename = find_sha1_file(sha1, &st);
 481
 482         if (!filename) {
 483                 if (say_error)
 484                         error("cannot map sha1 file %s", sha1_to_hex(sha1));
 485                 return NULL;
 486         }
 487
 488         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 489         if (fd < 0) {
 490                 /* See if it works without O_NOATIME */
 491                 switch (sha1_file_open_flag) {
 492                 default:
 493                         fd = open(filename, O_RDONLY);
 494                         if (fd >= 0)
 495                                 break;
 496                 /* Fallthrough */
 497                 case 0:
 498                         if (say_error)
 499                                 perror(filename);
 500                         return NULL;
 501                 }
 502
 503                 /* If it failed once, it will probably fail again.
 504                  * Stop using O_NOATIME
 505                  */
 506                 sha1_file_open_flag = 0;
 507         }
 508         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 509         close(fd);
 510         if (-1 == (int)(long)map)
 511                 return NULL;
 512         *size = st.st_size;
 513         return map;
 514 }
 515
 516 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 517 {
 518         return map_sha1_file_internal(sha1, size, 1);
 519 }
 520
 521 int unpack_sha1_header(z_stream *stream, void *map, unsigned long mapsize, void *buffer, unsigned long size)
 522 {
 523         /* Get the data stream */
 524         memset(stream, 0, sizeof(*stream));
 525         stream->next_in = map;
 526         stream->avail_in = mapsize;
 527         stream->next_out = buffer;
 528         stream->avail_out = size;
 529
 530         inflateInit(stream);
 531         return inflate(stream, 0);
 532 }
 533
 534 static void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 535 {
 536         int bytes = strlen(buffer) + 1;
 537         unsigned char *buf = xmalloc(1+size);
 538
 539         memcpy(buf, buffer + bytes, stream->total_out - bytes);
 540         bytes = stream->total_out - bytes;
 541         if (bytes < size) {
 542                 stream->next_out = buf + bytes;
 543                 stream->avail_out = size - bytes;
 544                 while (inflate(stream, Z_FINISH) == Z_OK)
 545                         /* nothing */;
 546         }
 547         buf[size] = 0;
 548         inflateEnd(stream);
 549         return buf;
 550 }
 551
 552 /*
 553  * We used to just use "sscanf()", but that's actually way
 554  * too permissive for what we want to check. So do an anal
 555  * object header parse by hand.
 556  */
 557 int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 558 {
 559         int i;
 560         unsigned long size;
 561
 562         /*
 563          * The type can be at most ten bytes (including the
 564          * terminating '\0' that we add), and is followed by
 565          * a space.
 566          */
 567         i = 10;
 568         for (;;) {
 569                 char c = *hdr++;
 570                 if (c == ' ')
 571                         break;
 572                 if (!--i)
 573                         return -1;
 574                 *type++ = c;
 575         }
 576         *type = 0;
 577
 578         /*
 579          * The length must follow immediately, and be in canonical
 580          * decimal format (ie "010" is not valid).
 581          */
 582         size = *hdr++ - '0';
 583         if (size > 9)
 584                 return -1;
 585         if (size) {
 586                 for (;;) {
 587                         unsigned long c = *hdr - '0';
 588                         if (c > 9)
 589                                 break;
 590                         hdr++;
 591                         size = size * 10 + c;
 592                 }
 593         }
 594         *sizep = size;
 595
 596         /*
 597          * The length must be followed by a zero byte
 598          */
 599         return *hdr ? -1 : 0;
 600 }
 601
 602 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 603 {
 604         int ret;
 605         z_stream stream;
 606         char hdr[8192];
 607
 608         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
 609         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
 610                 return NULL;
 611
 612         return unpack_sha1_rest(&stream, hdr, *size);
 613 }
 614
 615 /* forward declaration for a mutually recursive function */
 616 static int packed_object_info(struct pack_entry *entry,
 617                               char *type, unsigned long *sizep);
 618
 619 static int packed_delta_info(unsigned char *base_sha1,
 620                              unsigned long delta_size,
 621                              unsigned long left,
 622                              char *type,
 623                              unsigned long *sizep,
 624                              struct packed_git *p)
 625 {
 626         struct pack_entry base_ent;
 627
 628         if (left < 20)
 629                 die("truncated pack file");
 630
 631         /* The base entry _must_ be in the same pack */
 632         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 633                 die("failed to find delta-pack base object %s",
 634                     sha1_to_hex(base_sha1));
 635
 636         /* We choose to only get the type of the base object and
 637          * ignore potentially corrupt pack file that expects the delta
 638          * based on a base with a wrong size.  This saves tons of
 639          * inflate() calls.
 640          */
 641
 642         if (packed_object_info(&base_ent, type, NULL))
 643                 die("cannot get info for delta-pack base");
 644
 645         if (sizep) {
 646                 const unsigned char *data;
 647                 unsigned char delta_head[64];
 648                 unsigned long result_size;
 649                 z_stream stream;
 650                 int st;
 651
 652                 memset(&stream, 0, sizeof(stream));
 653
 654                 data = stream.next_in = base_sha1 + 20;
 655                 stream.avail_in = left - 20;
 656                 stream.next_out = delta_head;
 657                 stream.avail_out = sizeof(delta_head);
 658
 659                 inflateInit(&stream);
 660                 st = inflate(&stream, Z_FINISH);
 661                 inflateEnd(&stream);
 662                 if ((st != Z_STREAM_END) &&
 663                     stream.total_out != sizeof(delta_head))
 664                         die("delta data unpack-initial failed");
 665
 666                 /* Examine the initial part of the delta to figure out
 667                  * the result size.
 668                  */
 669                 data = delta_head;
 670                 get_delta_hdr_size(&data); /* ignore base size */
 671
 672                 /* Read the result size */
 673                 result_size = get_delta_hdr_size(&data);
 674                 *sizep = result_size;
 675         }
 676         return 0;
 677 }
 678
 679 static unsigned long unpack_object_header(struct packed_git *p, unsigned long offset,
 680         enum object_type *type, unsigned long *sizep)
 681 {
 682         unsigned shift;
 683         unsigned char *pack, c;
 684         unsigned long size;
 685
 686         if (offset >= p->pack_size)
 687                 die("object offset outside of pack file");
 688
 689         pack =  p->pack_base + offset;
 690         c = *pack++;
 691         offset++;
 692         *type = (c >> 4) & 7;
 693         size = c & 15;
 694         shift = 4;
 695         while (c & 0x80) {
 696                 if (offset >= p->pack_size)
 697                         die("object offset outside of pack file");
 698                 c = *pack++;
 699                 offset++;
 700                 size += (c & 0x7f) << shift;
 701                 shift += 7;
 702         }
 703         *sizep = size;
 704         return offset;
 705 }
 706
 707 void packed_object_info_detail(struct pack_entry *e,
 708                                char *type,
 709                                unsigned long *size,
 710                                unsigned long *store_size,
 711                                int *delta_chain_length,
 712                                unsigned char *base_sha1)
 713 {
 714         struct packed_git *p = e->p;
 715         unsigned long offset, left;
 716         unsigned char *pack;
 717         enum object_type kind;
 718
 719         offset = unpack_object_header(p, e->offset, &kind, size);
 720         pack = p->pack_base + offset;
 721         left = p->pack_size - offset;
 722         if (kind != OBJ_DELTA)
 723                 *delta_chain_length = 0;
 724         else {
 725                 int chain_length = 0;
 726                 memcpy(base_sha1, pack, 20);
 727                 do {
 728                         struct pack_entry base_ent;
 729                         unsigned long junk;
 730
 731                         find_pack_entry_one(pack, &base_ent, p);
 732                         offset = unpack_object_header(p, base_ent.offset,
 733                                                       &kind, &junk);
 734                         pack = p->pack_base + offset;
 735                         chain_length++;
 736                 } while (kind == OBJ_DELTA);
 737                 *delta_chain_length = chain_length;
 738         }
 739         switch (kind) {
 740         case OBJ_COMMIT:
 741                 strcpy(type, "commit");
 742                 break;
 743         case OBJ_TREE:
 744                 strcpy(type, "tree");
 745                 break;
 746         case OBJ_BLOB:
 747                 strcpy(type, "blob");
 748                 break;
 749         case OBJ_TAG:
 750                 strcpy(type, "tag");
 751                 break;
 752         default:
 753                 die("corrupted pack file");
 754         }
 755         *store_size = 0; /* notyet */
 756 }
 757
 758 static int packed_object_info(struct pack_entry *entry,
 759                               char *type, unsigned long *sizep)
 760 {
 761         struct packed_git *p = entry->p;
 762         unsigned long offset, size, left;
 763         unsigned char *pack;
 764         enum object_type kind;
 765         int retval;
 766
 767         if (use_packed_git(p))
 768                 die("cannot map packed file");
 769
 770         offset = unpack_object_header(p, entry->offset, &kind, &size);
 771         pack = p->pack_base + offset;
 772         left = p->pack_size - offset;
 773
 774         switch (kind) {
 775         case OBJ_DELTA:
 776                 retval = packed_delta_info(pack, size, left, type, sizep, p);
 777                 unuse_packed_git(p);
 778                 return retval;
 779         case OBJ_COMMIT:
 780                 strcpy(type, "commit");
 781                 break;
 782         case OBJ_TREE:
 783                 strcpy(type, "tree");
 784                 break;
 785         case OBJ_BLOB:
 786                 strcpy(type, "blob");
 787                 break;
 788         case OBJ_TAG:
 789                 strcpy(type, "tag");
 790                 break;
 791         default:
 792                 die("corrupted pack file");
 793         }
 794         if (sizep)
 795                 *sizep = size;
 796         unuse_packed_git(p);
 797         return 0;
 798 }
 799
 800 /* forward declaration for a mutually recursive function */
 801 static void *unpack_entry(struct pack_entry *, char *, unsigned long *);
 802
 803 static void *unpack_delta_entry(unsigned char *base_sha1,
 804                                 unsigned long delta_size,
 805                                 unsigned long left,
 806                                 char *type,
 807                                 unsigned long *sizep,
 808                                 struct packed_git *p)
 809 {
 810         struct pack_entry base_ent;
 811         void *data, *delta_data, *result, *base;
 812         unsigned long data_size, result_size, base_size;
 813         z_stream stream;
 814         int st;
 815
 816         if (left < 20)
 817                 die("truncated pack file");
 818         data = base_sha1 + 20;
 819         data_size = left - 20;
 820         delta_data = xmalloc(delta_size);
 821
 822         memset(&stream, 0, sizeof(stream));
 823
 824         stream.next_in = data;
 825         stream.avail_in = data_size;
 826         stream.next_out = delta_data;
 827         stream.avail_out = delta_size;
 828
 829         inflateInit(&stream);
 830         st = inflate(&stream, Z_FINISH);
 831         inflateEnd(&stream);
 832         if ((st != Z_STREAM_END) || stream.total_out != delta_size)
 833                 die("delta data unpack failed");
 834
 835         /* The base entry _must_ be in the same pack */
 836         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 837                 die("failed to find delta-pack base object %s",
 838                     sha1_to_hex(base_sha1));
 839         base = unpack_entry_gently(&base_ent, type, &base_size);
 840         if (!base)
 841                 die("failed to read delta-pack base object %s",
 842                     sha1_to_hex(base_sha1));
 843         result = patch_delta(base, base_size,
 844                              delta_data, delta_size,
 845                              &result_size);
 846         if (!result)
 847                 die("failed to apply delta");
 848         free(delta_data);
 849         free(base);
 850         *sizep = result_size;
 851         return result;
 852 }
 853
 854 static void *unpack_non_delta_entry(unsigned char *data,
 855                                     unsigned long size,
 856                                     unsigned long left)
 857 {
 858         int st;
 859         z_stream stream;
 860         unsigned char *buffer;
 861
 862         buffer = xmalloc(size + 1);
 863         buffer[size] = 0;
 864         memset(&stream, 0, sizeof(stream));
 865         stream.next_in = data;
 866         stream.avail_in = left;
 867         stream.next_out = buffer;
 868         stream.avail_out = size;
 869
 870         inflateInit(&stream);
 871         st = inflate(&stream, Z_FINISH);
 872         inflateEnd(&stream);
 873         if ((st != Z_STREAM_END) || stream.total_out != size) {
 874                 free(buffer);
 875                 return NULL;
 876         }
 877
 878         return buffer;
 879 }
 880
 881 static void *unpack_entry(struct pack_entry *entry,
 882                           char *type, unsigned long *sizep)
 883 {
 884         struct packed_git *p = entry->p;
 885         void *retval;
 886
 887         if (use_packed_git(p))
 888                 die("cannot map packed file");
 889         retval = unpack_entry_gently(entry, type, sizep);
 890         unuse_packed_git(p);
 891         if (!retval)
 892                 die("corrupted pack file");
 893         return retval;
 894 }
 895
 896 /* The caller is responsible for use_packed_git()/unuse_packed_git() pair */
 897 void *unpack_entry_gently(struct pack_entry *entry,
 898                           char *type, unsigned long *sizep)
 899 {
 900         struct packed_git *p = entry->p;
 901         unsigned long offset, size, left;
 902         unsigned char *pack;
 903         enum object_type kind;
 904         void *retval;
 905
 906         offset = unpack_object_header(p, entry->offset, &kind, &size);
 907         pack = p->pack_base + offset;
 908         left = p->pack_size - offset;
 909         switch (kind) {
 910         case OBJ_DELTA:
 911                 retval = unpack_delta_entry(pack, size, left, type, sizep, p);
 912                 return retval;
 913         case OBJ_COMMIT:
 914                 strcpy(type, "commit");
 915                 break;
 916         case OBJ_TREE:
 917                 strcpy(type, "tree");
 918                 break;
 919         case OBJ_BLOB:
 920                 strcpy(type, "blob");
 921                 break;
 922         case OBJ_TAG:
 923                 strcpy(type, "tag");
 924                 break;
 925         default:
 926                 return NULL;
 927         }
 928         *sizep = size;
 929         retval = unpack_non_delta_entry(pack, size, left);
 930         return retval;
 931 }
 932
 933 int num_packed_objects(const struct packed_git *p)
 934 {
 935         /* See check_packed_git_idx() */
 936         return (p->index_size - 20 - 20 - 4*256) / 24;
 937 }
 938
 939 int nth_packed_object_sha1(const struct packed_git *p, int n,
 940                            unsigned char* sha1)
 941 {
 942         void *index = p->index_base + 256;
 943         if (n < 0 || num_packed_objects(p) <= n)
 944                 return -1;
 945         memcpy(sha1, (index + 24 * n + 4), 20);
 946         return 0;
 947 }
 948
 949 int find_pack_entry_one(const unsigned char *sha1,
 950                         struct pack_entry *e, struct packed_git *p)
 951 {
 952         unsigned int *level1_ofs = p->index_base;
 953         int hi = ntohl(level1_ofs[*sha1]);
 954         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
 955         void *index = p->index_base + 256;
 956
 957         do {
 958                 int mi = (lo + hi) / 2;
 959                 int cmp = memcmp(index + 24 * mi + 4, sha1, 20);
 960                 if (!cmp) {
 961                         e->offset = ntohl(*((int*)(index + 24 * mi)));
 962                         memcpy(e->sha1, sha1, 20);
 963                         e->p = p;
 964                         return 1;
 965                 }
 966                 if (cmp > 0)
 967                         hi = mi;
 968                 else
 969                         lo = mi+1;
 970         } while (lo < hi);
 971         return 0;
 972 }
 973
 974 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e)
 975 {
 976         struct packed_git *p;
 977         prepare_packed_git();
 978
 979         for (p = packed_git; p; p = p->next) {
 980                 if (find_pack_entry_one(sha1, e, p))
 981                         return 1;
 982         }
 983         return 0;
 984 }
 985
 986 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
 987 {
 988         int status;
 989         unsigned long mapsize, size;
 990         void *map;
 991         z_stream stream;
 992         char hdr[128];
 993
 994         map = map_sha1_file_internal(sha1, &mapsize, 0);
 995         if (!map) {
 996                 struct pack_entry e;
 997
 998                 if (!find_pack_entry(sha1, &e))
 999                         return error("unable to find %s", sha1_to_hex(sha1));
1000                 return packed_object_info(&e, type, sizep);
1001         }
1002         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
1003                 status = error("unable to unpack %s header",
1004                                sha1_to_hex(sha1));
1005         if (parse_sha1_header(hdr, type, &size) < 0)
1006                 status = error("unable to parse %s header", sha1_to_hex(sha1));
1007         else {
1008                 status = 0;
1009                 if (sizep)
1010                         *sizep = size;
1011         }
1012         inflateEnd(&stream);
1013         munmap(map, mapsize);
1014         return status;
1015 }
1016
1017 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
1018 {
1019         struct pack_entry e;
1020
1021         if (!find_pack_entry(sha1, &e)) {
1022                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
1023                 return NULL;
1024         }
1025         return unpack_entry(&e, type, size);
1026 }
1027
1028 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
1029 {
1030         unsigned long mapsize;
1031         void *map, *buf;
1032
1033         map = map_sha1_file_internal(sha1, &mapsize, 0);
1034         if (map) {
1035                 buf = unpack_sha1_file(map, mapsize, type, size);
1036                 munmap(map, mapsize);
1037                 return buf;
1038         }
1039         return read_packed_sha1(sha1, type, size);
1040 }
1041
1042 void *read_object_with_reference(const unsigned char *sha1,
1043                                  const char *required_type,
1044                                  unsigned long *size,
1045                                  unsigned char *actual_sha1_return)
1046 {
1047         char type[20];
1048         void *buffer;
1049         unsigned long isize;
1050         unsigned char actual_sha1[20];
1051
1052         memcpy(actual_sha1, sha1, 20);
1053         while (1) {
1054                 int ref_length = -1;
1055                 const char *ref_type = NULL;
1056
1057                 buffer = read_sha1_file(actual_sha1, type, &isize);
1058                 if (!buffer)
1059                         return NULL;
1060                 if (!strcmp(type, required_type)) {
1061                         *size = isize;
1062                         if (actual_sha1_return)
1063                                 memcpy(actual_sha1_return, actual_sha1, 20);
1064                         return buffer;
1065                 }
1066                 /* Handle references */
1067                 else if (!strcmp(type, "commit"))
1068                         ref_type = "tree ";
1069                 else if (!strcmp(type, "tag"))
1070                         ref_type = "object ";
1071                 else {
1072                         free(buffer);
1073                         return NULL;
1074                 }
1075                 ref_length = strlen(ref_type);
1076
1077                 if (memcmp(buffer, ref_type, ref_length) ||
1078                     get_sha1_hex(buffer + ref_length, actual_sha1)) {
1079                         free(buffer);
1080                         return NULL;
1081                 }
1082                 /* Now we have the ID of the referred-to object in
1083                  * actual_sha1.  Check again. */
1084         }
1085 }
1086
1087 static char *write_sha1_file_prepare(void *buf,
1088                                      unsigned long len,
1089                                      const char *type,
1090                                      unsigned char *sha1,
1091                                      unsigned char *hdr,
1092                                      int *hdrlen)
1093 {
1094         SHA_CTX c;
1095
1096         /* Generate the header */
1097         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
1098
1099         /* Sha1.. */
1100         SHA1_Init(&c);
1101         SHA1_Update(&c, hdr, *hdrlen);
1102         SHA1_Update(&c, buf, len);
1103         SHA1_Final(sha1, &c);
1104
1105         return sha1_file_name(sha1);
1106 }
1107
1108 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1109 {
1110         int size;
1111         unsigned char *compressed;
1112         z_stream stream;
1113         unsigned char sha1[20];
1114         char *filename;
1115         static char tmpfile[PATH_MAX];
1116         unsigned char hdr[50];
1117         int fd, hdrlen, ret;
1118
1119         /* Normally if we have it in the pack then we do not bother writing
1120          * it out into .git/objects/??/?{38} file.
1121          */
1122         filename = write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1123         if (returnsha1)
1124                 memcpy(returnsha1, sha1, 20);
1125         if (has_sha1_file(sha1))
1126                 return 0;
1127         fd = open(filename, O_RDONLY);
1128         if (fd >= 0) {
1129                 /*
1130                  * FIXME!!! We might do collision checking here, but we'd
1131                  * need to uncompress the old file and check it. Later.
1132                  */
1133                 close(fd);
1134                 return 0;
1135         }
1136
1137         if (errno != ENOENT) {
1138                 fprintf(stderr, "sha1 file %s: %s", filename, strerror(errno));
1139                 return -1;
1140         }
1141
1142         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1143
1144         fd = mkstemp(tmpfile);
1145         if (fd < 0) {
1146                 fprintf(stderr, "unable to create temporary sha1 filename %s: %s", tmpfile, strerror(errno));
1147                 return -1;
1148         }
1149
1150         /* Set it up */
1151         memset(&stream, 0, sizeof(stream));
1152         deflateInit(&stream, Z_BEST_COMPRESSION);
1153         size = deflateBound(&stream, len+hdrlen);
1154         compressed = xmalloc(size);
1155
1156         /* Compress it */
1157         stream.next_out = compressed;
1158         stream.avail_out = size;
1159
1160         /* First header.. */
1161         stream.next_in = hdr;
1162         stream.avail_in = hdrlen;
1163         while (deflate(&stream, 0) == Z_OK)
1164                 /* nothing */;
1165
1166         /* Then the data itself.. */
1167         stream.next_in = buf;
1168         stream.avail_in = len;
1169         while (deflate(&stream, Z_FINISH) == Z_OK)
1170                 /* nothing */;
1171         deflateEnd(&stream);
1172         size = stream.total_out;
1173
1174         if (write(fd, compressed, size) != size)
1175                 die("unable to write file");
1176         fchmod(fd, 0444);
1177         close(fd);
1178         free(compressed);
1179
1180         ret = link(tmpfile, filename);
1181         if (ret < 0) {
1182                 ret = errno;
1183
1184                 /*
1185                  * Coda hack - coda doesn't like cross-directory links,
1186                  * so we fall back to a rename, which will mean that it
1187                  * won't be able to check collisions, but that's not a
1188                  * big deal.
1189                  *
1190                  * When this succeeds, we just return 0. We have nothing
1191                  * left to unlink.
1192                  */
1193                 if (ret == EXDEV && !rename(tmpfile, filename))
1194                         return 0;
1195         }
1196         unlink(tmpfile);
1197         if (ret) {
1198                 if (ret != EEXIST) {
1199                         fprintf(stderr, "unable to write sha1 filename %s: %s", filename, strerror(ret));
1200                         return -1;
1201                 }
1202                 /* FIXME!!! Collision check here ? */
1203         }
1204
1205         return 0;
1206 }
1207
1208 int write_sha1_from_fd(const unsigned char *sha1, int fd)
1209 {
1210         char *filename = sha1_file_name(sha1);
1211
1212         int local;
1213         z_stream stream;
1214         unsigned char real_sha1[20];
1215         unsigned char buf[4096];
1216         unsigned char discard[4096];
1217         int ret;
1218         SHA_CTX c;
1219
1220         local = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
1221
1222         if (local < 0)
1223                 return error("Couldn't open %s\n", filename);
1224
1225         memset(&stream, 0, sizeof(stream));
1226
1227         inflateInit(&stream);
1228
1229         SHA1_Init(&c);
1230
1231         do {
1232                 ssize_t size;
1233                 size = read(fd, buf, 4096);
1234                 if (size <= 0) {
1235                         close(local);
1236                         unlink(filename);
1237                         if (!size)
1238                                 return error("Connection closed?");
1239                         perror("Reading from connection");
1240                         return -1;
1241                 }
1242                 write(local, buf, size);
1243                 stream.avail_in = size;
1244                 stream.next_in = buf;
1245                 do {
1246                         stream.next_out = discard;
1247                         stream.avail_out = sizeof(discard);
1248                         ret = inflate(&stream, Z_SYNC_FLUSH);
1249                         SHA1_Update(&c, discard, sizeof(discard) -
1250                                     stream.avail_out);
1251                 } while (stream.avail_in && ret == Z_OK);
1252
1253         } while (ret == Z_OK);
1254         inflateEnd(&stream);
1255
1256         close(local);
1257         SHA1_Final(real_sha1, &c);
1258         if (ret != Z_STREAM_END) {
1259                 unlink(filename);
1260                 return error("File %s corrupted", sha1_to_hex(sha1));
1261         }
1262         if (memcmp(sha1, real_sha1, 20)) {
1263                 unlink(filename);
1264                 return error("File %s has bad hash\n", sha1_to_hex(sha1));
1265         }
1266
1267         return 0;
1268 }
1269
1270 int has_sha1_file(const unsigned char *sha1)
1271 {
1272         struct stat st;
1273         struct pack_entry e;
1274
1275         if (find_sha1_file(sha1, &st))
1276                 return 1;
1277         return find_pack_entry(sha1, &e);
1278 }
1279
1280 int index_fd(unsigned char *sha1, int fd, struct stat *st)
1281 {
1282         unsigned long size = st->st_size;
1283         void *buf;
1284         int ret;
1285
1286         buf = "";
1287         if (size)
1288                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1289         close(fd);
1290         if ((int)(long)buf == -1)
1291                 return -1;
1292
1293         ret = write_sha1_file(buf, size, "blob", sha1);
1294         if (size)
1295                 munmap(buf, size);
1296         return ret;
1297 }