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 safe_create_leading_directories(char *path)
 106 {
 107         char *pos = path;
 108
 109         while (pos) {
 110                 pos = strchr(pos, '/');
 111                 if (!pos)
 112                         break;
 113                 *pos = 0;
 114                 if (mkdir(path, 0777) < 0)
 115                         if (errno != EEXIST) {
 116                                 *pos = '/';
 117                                 return -1;
 118                         }
 119                 *pos++ = '/';
 120         }
 121         return 0;
 122 }
 123
 124 int get_sha1(const char *str, unsigned char *sha1)
 125 {
 126         static const char *prefix[] = {
 127                 "",
 128                 "refs",
 129                 "refs/tags",
 130                 "refs/heads",
 131                 "refs/snap",
 132                 NULL
 133         };
 134         const char **p;
 135
 136         if (!get_sha1_hex(str, sha1))
 137                 return 0;
 138
 139         for (p = prefix; *p; p++) {
 140                 char * pathname = git_path("%s/%s", *p, str);
 141                 if (!get_sha1_file(pathname, sha1))
 142                         return 0;
 143         }
 144
 145         return -1;
 146 }
 147
 148 char * sha1_to_hex(const unsigned char *sha1)
 149 {
 150         static char buffer[50];
 151         static const char hex[] = "0123456789abcdef";
 152         char *buf = buffer;
 153         int i;
 154
 155         for (i = 0; i < 20; i++) {
 156                 unsigned int val = *sha1++;
 157                 *buf++ = hex[val >> 4];
 158                 *buf++ = hex[val & 0xf];
 159         }
 160         return buffer;
 161 }
 162
 163 static void fill_sha1_path(char *pathbuf, const unsigned char *sha1)
 164 {
 165         int i;
 166         for (i = 0; i < 20; i++) {
 167                 static char hex[] = "0123456789abcdef";
 168                 unsigned int val = sha1[i];
 169                 char *pos = pathbuf + i*2 + (i > 0);
 170                 *pos++ = hex[val >> 4];
 171                 *pos = hex[val & 0xf];
 172         }
 173 }
 174
 175 /*
 176  * NOTE! This returns a statically allocated buffer, so you have to be
 177  * careful about using it. Do a "strdup()" if you need to save the
 178  * filename.
 179  *
 180  * Also note that this returns the location for creating.  Reading
 181  * SHA1 file can happen from any alternate directory listed in the
 182  * DB_ENVIRONMENT environment variable if it is not found in
 183  * the primary object database.
 184  */
 185 char *sha1_file_name(const unsigned char *sha1)
 186 {
 187         static char *name, *base;
 188
 189         if (!base) {
 190                 const char *sha1_file_directory = get_object_directory();
 191                 int len = strlen(sha1_file_directory);
 192                 base = xmalloc(len + 60);
 193                 memcpy(base, sha1_file_directory, len);
 194                 memset(base+len, 0, 60);
 195                 base[len] = '/';
 196                 base[len+3] = '/';
 197                 name = base + len + 1;
 198         }
 199         fill_sha1_path(name, sha1);
 200         return base;
 201 }
 202
 203 struct alternate_object_database *alt_odb;
 204
 205 /*
 206  * Prepare alternate object database registry.
 207  * alt_odb points at an array of struct alternate_object_database.
 208  * This array is terminated with an element that has both its base
 209  * and name set to NULL.  alt_odb[n] comes from n'th non-empty
 210  * element from colon separated ALTERNATE_DB_ENVIRONMENT environment
 211  * variable, and its base points at a statically allocated buffer
 212  * that contains "/the/directory/corresponding/to/.git/objects/...",
 213  * while its name points just after the slash at the end of
 214  * ".git/objects/" in the example above, and has enough space to hold
 215  * 40-byte hex SHA1, an extra slash for the first level indirection,
 216  * and the terminating NUL.
 217  * This function allocates the alt_odb array and all the strings
 218  * pointed by base fields of the array elements with one xmalloc();
 219  * the string pool immediately follows the array.
 220  */
 221 void prepare_alt_odb(void)
 222 {
 223         int pass, totlen, i;
 224         const char *cp, *last;
 225         char *op = NULL;
 226         const char *alt = gitenv(ALTERNATE_DB_ENVIRONMENT) ? : "";
 227
 228         if (alt_odb)
 229                 return;
 230         /* The first pass counts how large an area to allocate to
 231          * hold the entire alt_odb structure, including array of
 232          * structs and path buffers for them.  The second pass fills
 233          * the structure and prepares the path buffers for use by
 234          * fill_sha1_path().
 235          */
 236         for (totlen = pass = 0; pass < 2; pass++) {
 237                 last = alt;
 238                 i = 0;
 239                 do {
 240                         cp = strchr(last, ':') ? : last + strlen(last);
 241                         if (last != cp) {
 242                                 /* 43 = 40-byte + 2 '/' + terminating NUL */
 243                                 int pfxlen = cp - last;
 244                                 int entlen = pfxlen + 43;
 245                                 if (pass == 0)
 246                                         totlen += entlen;
 247                                 else {
 248                                         alt_odb[i].base = op;
 249                                         alt_odb[i].name = op + pfxlen + 1;
 250                                         memcpy(op, last, pfxlen);
 251                                         op[pfxlen] = op[pfxlen + 3] = '/';
 252                                         op[entlen-1] = 0;
 253                                         op += entlen;
 254                                 }
 255                                 i++;
 256                         }
 257                         while (*cp && *cp == ':')
 258                                 cp++;
 259                         last = cp;
 260                 } while (*cp);
 261                 if (pass)
 262                         break;
 263                 alt_odb = xmalloc(sizeof(*alt_odb) * (i + 1) + totlen);
 264                 alt_odb[i].base = alt_odb[i].name = NULL;
 265                 op = (char*)(&alt_odb[i+1]);
 266         }
 267 }
 268
 269 static char *find_sha1_file(const unsigned char *sha1, struct stat *st)
 270 {
 271         int i;
 272         char *name = sha1_file_name(sha1);
 273
 274         if (!stat(name, st))
 275                 return name;
 276         prepare_alt_odb();
 277         for (i = 0; (name = alt_odb[i].name) != NULL; i++) {
 278                 fill_sha1_path(name, sha1);
 279                 if (!stat(alt_odb[i].base, st))
 280                         return alt_odb[i].base;
 281         }
 282         return NULL;
 283 }
 284
 285 #define PACK_MAX_SZ (1<<26)
 286 static int pack_used_ctr;
 287 static unsigned long pack_mapped;
 288 struct packed_git *packed_git;
 289
 290 static int check_packed_git_idx(const char *path, unsigned long *idx_size_,
 291                                 void **idx_map_)
 292 {
 293         void *idx_map;
 294         unsigned int *index;
 295         unsigned long idx_size;
 296         int nr, i;
 297         int fd = open(path, O_RDONLY);
 298         struct stat st;
 299         if (fd < 0)
 300                 return -1;
 301         if (fstat(fd, &st)) {
 302                 close(fd);
 303                 return -1;
 304         }
 305         idx_size = st.st_size;
 306         idx_map = mmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0);
 307         close(fd);
 308         if (idx_map == MAP_FAILED)
 309                 return -1;
 310
 311         index = idx_map;
 312         *idx_map_ = idx_map;
 313         *idx_size_ = idx_size;
 314
 315         /* check index map */
 316         if (idx_size < 4*256 + 20 + 20)
 317                 return error("index file too small");
 318         nr = 0;
 319         for (i = 0; i < 256; i++) {
 320                 unsigned int n = ntohl(index[i]);
 321                 if (n < nr)
 322                         return error("non-monotonic index");
 323                 nr = n;
 324         }
 325
 326         /*
 327          * Total size:
 328          *  - 256 index entries 4 bytes each
 329          *  - 24-byte entries * nr (20-byte sha1 + 4-byte offset)
 330          *  - 20-byte SHA1 of the packfile
 331          *  - 20-byte SHA1 file checksum
 332          */
 333         if (idx_size != 4*256 + nr * 24 + 20 + 20)
 334                 return error("wrong index file size");
 335
 336         return 0;
 337 }
 338
 339 static int unuse_one_packed_git(void)
 340 {
 341         struct packed_git *p, *lru = NULL;
 342
 343         for (p = packed_git; p; p = p->next) {
 344                 if (p->pack_use_cnt || !p->pack_base)
 345                         continue;
 346                 if (!lru || p->pack_last_used < lru->pack_last_used)
 347                         lru = p;
 348         }
 349         if (!lru)
 350                 return 0;
 351         munmap(lru->pack_base, lru->pack_size);
 352         lru->pack_base = NULL;
 353         return 1;
 354 }
 355
 356 void unuse_packed_git(struct packed_git *p)
 357 {
 358         p->pack_use_cnt--;
 359 }
 360
 361 int use_packed_git(struct packed_git *p)
 362 {
 363         if (!p->pack_base) {
 364                 int fd;
 365                 struct stat st;
 366                 void *map;
 367
 368                 pack_mapped += p->pack_size;
 369                 while (PACK_MAX_SZ < pack_mapped && unuse_one_packed_git())
 370                         ; /* nothing */
 371                 fd = open(p->pack_name, O_RDONLY);
 372                 if (fd < 0)
 373                         die("packfile %s cannot be opened", p->pack_name);
 374                 if (fstat(fd, &st)) {
 375                         close(fd);
 376                         die("packfile %s cannot be opened", p->pack_name);
 377                 }
 378                 if (st.st_size != p->pack_size)
 379                         die("packfile %s size mismatch.", p->pack_name);
 380                 map = mmap(NULL, p->pack_size, PROT_READ, MAP_PRIVATE, fd, 0);
 381                 close(fd);
 382                 if (map == MAP_FAILED)
 383                         die("packfile %s cannot be mapped.", p->pack_name);
 384                 p->pack_base = map;
 385
 386                 /* Check if the pack file matches with the index file.
 387                  * this is cheap.
 388                  */
 389                 if (memcmp((char*)(p->index_base) + p->index_size - 40,
 390                            p->pack_base + p->pack_size - 20, 20))
 391                         die("packfile %s does not match index.", p->pack_name);
 392         }
 393         p->pack_last_used = pack_used_ctr++;
 394         p->pack_use_cnt++;
 395         return 0;
 396 }
 397
 398 struct packed_git *add_packed_git(char *path, int path_len)
 399 {
 400         struct stat st;
 401         struct packed_git *p;
 402         unsigned long idx_size;
 403         void *idx_map;
 404
 405         if (check_packed_git_idx(path, &idx_size, &idx_map))
 406                 return NULL;
 407
 408         /* do we have a corresponding .pack file? */
 409         strcpy(path + path_len - 4, ".pack");
 410         if (stat(path, &st) || !S_ISREG(st.st_mode)) {
 411                 munmap(idx_map, idx_size);
 412                 return NULL;
 413         }
 414         /* ok, it looks sane as far as we can check without
 415          * actually mapping the pack file.
 416          */
 417         p = xmalloc(sizeof(*p) + path_len + 2);
 418         strcpy(p->pack_name, path);
 419         p->index_size = idx_size;
 420         p->pack_size = st.st_size;
 421         p->index_base = idx_map;
 422         p->next = NULL;
 423         p->pack_base = NULL;
 424         p->pack_last_used = 0;
 425         p->pack_use_cnt = 0;
 426         return p;
 427 }
 428
 429 static void prepare_packed_git_one(char *objdir)
 430 {
 431         char path[PATH_MAX];
 432         int len;
 433         DIR *dir;
 434         struct dirent *de;
 435
 436         sprintf(path, "%s/pack", objdir);
 437         len = strlen(path);
 438         dir = opendir(path);
 439         if (!dir)
 440                 return;
 441         path[len++] = '/';
 442         while ((de = readdir(dir)) != NULL) {
 443                 int namelen = strlen(de->d_name);
 444                 struct packed_git *p;
 445
 446                 if (strcmp(de->d_name + namelen - 4, ".idx"))
 447                         continue;
 448
 449                 /* we have .idx.  Is it a file we can map? */
 450                 strcpy(path + len, de->d_name);
 451                 p = add_packed_git(path, len + namelen);
 452                 if (!p)
 453                         continue;
 454                 p->next = packed_git;
 455                 packed_git = p;
 456         }
 457         closedir(dir);
 458 }
 459
 460 void prepare_packed_git(void)
 461 {
 462         int i;
 463         static int run_once = 0;
 464
 465         if (run_once++)
 466                 return;
 467
 468         prepare_packed_git_one(get_object_directory());
 469         prepare_alt_odb();
 470         for (i = 0; alt_odb[i].base != NULL; i++) {
 471                 alt_odb[i].name[0] = 0;
 472                 prepare_packed_git_one(alt_odb[i].base);
 473         }
 474 }
 475
 476 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 477 {
 478         char header[100];
 479         unsigned char real_sha1[20];
 480         SHA_CTX c;
 481
 482         SHA1_Init(&c);
 483         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 484         SHA1_Update(&c, map, size);
 485         SHA1_Final(real_sha1, &c);
 486         return memcmp(sha1, real_sha1, 20) ? -1 : 0;
 487 }
 488
 489 static void *map_sha1_file_internal(const unsigned char *sha1,
 490                                     unsigned long *size,
 491                                     int say_error)
 492 {
 493         struct stat st;
 494         void *map;
 495         int fd;
 496         char *filename = find_sha1_file(sha1, &st);
 497
 498         if (!filename) {
 499                 if (say_error)
 500                         error("cannot map sha1 file %s", sha1_to_hex(sha1));
 501                 return NULL;
 502         }
 503
 504         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 505         if (fd < 0) {
 506                 /* See if it works without O_NOATIME */
 507                 switch (sha1_file_open_flag) {
 508                 default:
 509                         fd = open(filename, O_RDONLY);
 510                         if (fd >= 0)
 511                                 break;
 512                 /* Fallthrough */
 513                 case 0:
 514                         if (say_error)
 515                                 perror(filename);
 516                         return NULL;
 517                 }
 518
 519                 /* If it failed once, it will probably fail again.
 520                  * Stop using O_NOATIME
 521                  */
 522                 sha1_file_open_flag = 0;
 523         }
 524         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 525         close(fd);
 526         if (-1 == (int)(long)map)
 527                 return NULL;
 528         *size = st.st_size;
 529         return map;
 530 }
 531
 532 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 533 {
 534         return map_sha1_file_internal(sha1, size, 1);
 535 }
 536
 537 int unpack_sha1_header(z_stream *stream, void *map, unsigned long mapsize, void *buffer, unsigned long size)
 538 {
 539         /* Get the data stream */
 540         memset(stream, 0, sizeof(*stream));
 541         stream->next_in = map;
 542         stream->avail_in = mapsize;
 543         stream->next_out = buffer;
 544         stream->avail_out = size;
 545
 546         inflateInit(stream);
 547         return inflate(stream, 0);
 548 }
 549
 550 static void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 551 {
 552         int bytes = strlen(buffer) + 1;
 553         unsigned char *buf = xmalloc(1+size);
 554
 555         memcpy(buf, buffer + bytes, stream->total_out - bytes);
 556         bytes = stream->total_out - bytes;
 557         if (bytes < size) {
 558                 stream->next_out = buf + bytes;
 559                 stream->avail_out = size - bytes;
 560                 while (inflate(stream, Z_FINISH) == Z_OK)
 561                         /* nothing */;
 562         }
 563         buf[size] = 0;
 564         inflateEnd(stream);
 565         return buf;
 566 }
 567
 568 /*
 569  * We used to just use "sscanf()", but that's actually way
 570  * too permissive for what we want to check. So do an anal
 571  * object header parse by hand.
 572  */
 573 int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 574 {
 575         int i;
 576         unsigned long size;
 577
 578         /*
 579          * The type can be at most ten bytes (including the
 580          * terminating '\0' that we add), and is followed by
 581          * a space.
 582          */
 583         i = 10;
 584         for (;;) {
 585                 char c = *hdr++;
 586                 if (c == ' ')
 587                         break;
 588                 if (!--i)
 589                         return -1;
 590                 *type++ = c;
 591         }
 592         *type = 0;
 593
 594         /*
 595          * The length must follow immediately, and be in canonical
 596          * decimal format (ie "010" is not valid).
 597          */
 598         size = *hdr++ - '0';
 599         if (size > 9)
 600                 return -1;
 601         if (size) {
 602                 for (;;) {
 603                         unsigned long c = *hdr - '0';
 604                         if (c > 9)
 605                                 break;
 606                         hdr++;
 607                         size = size * 10 + c;
 608                 }
 609         }
 610         *sizep = size;
 611
 612         /*
 613          * The length must be followed by a zero byte
 614          */
 615         return *hdr ? -1 : 0;
 616 }
 617
 618 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 619 {
 620         int ret;
 621         z_stream stream;
 622         char hdr[8192];
 623
 624         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
 625         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
 626                 return NULL;
 627
 628         return unpack_sha1_rest(&stream, hdr, *size);
 629 }
 630
 631 /* forward declaration for a mutually recursive function */
 632 static int packed_object_info(struct pack_entry *entry,
 633                               char *type, unsigned long *sizep);
 634
 635 static int packed_delta_info(unsigned char *base_sha1,
 636                              unsigned long delta_size,
 637                              unsigned long left,
 638                              char *type,
 639                              unsigned long *sizep,
 640                              struct packed_git *p)
 641 {
 642         struct pack_entry base_ent;
 643
 644         if (left < 20)
 645                 die("truncated pack file");
 646
 647         /* The base entry _must_ be in the same pack */
 648         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 649                 die("failed to find delta-pack base object %s",
 650                     sha1_to_hex(base_sha1));
 651
 652         /* We choose to only get the type of the base object and
 653          * ignore potentially corrupt pack file that expects the delta
 654          * based on a base with a wrong size.  This saves tons of
 655          * inflate() calls.
 656          */
 657
 658         if (packed_object_info(&base_ent, type, NULL))
 659                 die("cannot get info for delta-pack base");
 660
 661         if (sizep) {
 662                 const unsigned char *data;
 663                 unsigned char delta_head[64];
 664                 unsigned long result_size;
 665                 z_stream stream;
 666                 int st;
 667
 668                 memset(&stream, 0, sizeof(stream));
 669
 670                 data = stream.next_in = base_sha1 + 20;
 671                 stream.avail_in = left - 20;
 672                 stream.next_out = delta_head;
 673                 stream.avail_out = sizeof(delta_head);
 674
 675                 inflateInit(&stream);
 676                 st = inflate(&stream, Z_FINISH);
 677                 inflateEnd(&stream);
 678                 if ((st != Z_STREAM_END) &&
 679                     stream.total_out != sizeof(delta_head))
 680                         die("delta data unpack-initial failed");
 681
 682                 /* Examine the initial part of the delta to figure out
 683                  * the result size.
 684                  */
 685                 data = delta_head;
 686                 get_delta_hdr_size(&data); /* ignore base size */
 687
 688                 /* Read the result size */
 689                 result_size = get_delta_hdr_size(&data);
 690                 *sizep = result_size;
 691         }
 692         return 0;
 693 }
 694
 695 static unsigned long unpack_object_header(struct packed_git *p, unsigned long offset,
 696         enum object_type *type, unsigned long *sizep)
 697 {
 698         unsigned shift;
 699         unsigned char *pack, c;
 700         unsigned long size;
 701
 702         if (offset >= p->pack_size)
 703                 die("object offset outside of pack file");
 704
 705         pack =  p->pack_base + offset;
 706         c = *pack++;
 707         offset++;
 708         *type = (c >> 4) & 7;
 709         size = c & 15;
 710         shift = 4;
 711         while (c & 0x80) {
 712                 if (offset >= p->pack_size)
 713                         die("object offset outside of pack file");
 714                 c = *pack++;
 715                 offset++;
 716                 size += (c & 0x7f) << shift;
 717                 shift += 7;
 718         }
 719         *sizep = size;
 720         return offset;
 721 }
 722
 723 void packed_object_info_detail(struct pack_entry *e,
 724                                char *type,
 725                                unsigned long *size,
 726                                unsigned long *store_size,
 727                                int *delta_chain_length,
 728                                unsigned char *base_sha1)
 729 {
 730         struct packed_git *p = e->p;
 731         unsigned long offset, left;
 732         unsigned char *pack;
 733         enum object_type kind;
 734
 735         offset = unpack_object_header(p, e->offset, &kind, size);
 736         pack = p->pack_base + offset;
 737         left = p->pack_size - offset;
 738         if (kind != OBJ_DELTA)
 739                 *delta_chain_length = 0;
 740         else {
 741                 int chain_length = 0;
 742                 memcpy(base_sha1, pack, 20);
 743                 do {
 744                         struct pack_entry base_ent;
 745                         unsigned long junk;
 746
 747                         find_pack_entry_one(pack, &base_ent, p);
 748                         offset = unpack_object_header(p, base_ent.offset,
 749                                                       &kind, &junk);
 750                         pack = p->pack_base + offset;
 751                         chain_length++;
 752                 } while (kind == OBJ_DELTA);
 753                 *delta_chain_length = chain_length;
 754         }
 755         switch (kind) {
 756         case OBJ_COMMIT:
 757                 strcpy(type, "commit");
 758                 break;
 759         case OBJ_TREE:
 760                 strcpy(type, "tree");
 761                 break;
 762         case OBJ_BLOB:
 763                 strcpy(type, "blob");
 764                 break;
 765         case OBJ_TAG:
 766                 strcpy(type, "tag");
 767                 break;
 768         default:
 769                 die("corrupted pack file");
 770         }
 771         *store_size = 0; /* notyet */
 772 }
 773
 774 static int packed_object_info(struct pack_entry *entry,
 775                               char *type, unsigned long *sizep)
 776 {
 777         struct packed_git *p = entry->p;
 778         unsigned long offset, size, left;
 779         unsigned char *pack;
 780         enum object_type kind;
 781         int retval;
 782
 783         if (use_packed_git(p))
 784                 die("cannot map packed file");
 785
 786         offset = unpack_object_header(p, entry->offset, &kind, &size);
 787         pack = p->pack_base + offset;
 788         left = p->pack_size - offset;
 789
 790         switch (kind) {
 791         case OBJ_DELTA:
 792                 retval = packed_delta_info(pack, size, left, type, sizep, p);
 793                 unuse_packed_git(p);
 794                 return retval;
 795         case OBJ_COMMIT:
 796                 strcpy(type, "commit");
 797                 break;
 798         case OBJ_TREE:
 799                 strcpy(type, "tree");
 800                 break;
 801         case OBJ_BLOB:
 802                 strcpy(type, "blob");
 803                 break;
 804         case OBJ_TAG:
 805                 strcpy(type, "tag");
 806                 break;
 807         default:
 808                 die("corrupted pack file");
 809         }
 810         if (sizep)
 811                 *sizep = size;
 812         unuse_packed_git(p);
 813         return 0;
 814 }
 815
 816 /* forward declaration for a mutually recursive function */
 817 static void *unpack_entry(struct pack_entry *, char *, unsigned long *);
 818
 819 static void *unpack_delta_entry(unsigned char *base_sha1,
 820                                 unsigned long delta_size,
 821                                 unsigned long left,
 822                                 char *type,
 823                                 unsigned long *sizep,
 824                                 struct packed_git *p)
 825 {
 826         struct pack_entry base_ent;
 827         void *data, *delta_data, *result, *base;
 828         unsigned long data_size, result_size, base_size;
 829         z_stream stream;
 830         int st;
 831
 832         if (left < 20)
 833                 die("truncated pack file");
 834         data = base_sha1 + 20;
 835         data_size = left - 20;
 836         delta_data = xmalloc(delta_size);
 837
 838         memset(&stream, 0, sizeof(stream));
 839
 840         stream.next_in = data;
 841         stream.avail_in = data_size;
 842         stream.next_out = delta_data;
 843         stream.avail_out = delta_size;
 844
 845         inflateInit(&stream);
 846         st = inflate(&stream, Z_FINISH);
 847         inflateEnd(&stream);
 848         if ((st != Z_STREAM_END) || stream.total_out != delta_size)
 849                 die("delta data unpack failed");
 850
 851         /* The base entry _must_ be in the same pack */
 852         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 853                 die("failed to find delta-pack base object %s",
 854                     sha1_to_hex(base_sha1));
 855         base = unpack_entry_gently(&base_ent, type, &base_size);
 856         if (!base)
 857                 die("failed to read delta-pack base object %s",
 858                     sha1_to_hex(base_sha1));
 859         result = patch_delta(base, base_size,
 860                              delta_data, delta_size,
 861                              &result_size);
 862         if (!result)
 863                 die("failed to apply delta");
 864         free(delta_data);
 865         free(base);
 866         *sizep = result_size;
 867         return result;
 868 }
 869
 870 static void *unpack_non_delta_entry(unsigned char *data,
 871                                     unsigned long size,
 872                                     unsigned long left)
 873 {
 874         int st;
 875         z_stream stream;
 876         unsigned char *buffer;
 877
 878         buffer = xmalloc(size + 1);
 879         buffer[size] = 0;
 880         memset(&stream, 0, sizeof(stream));
 881         stream.next_in = data;
 882         stream.avail_in = left;
 883         stream.next_out = buffer;
 884         stream.avail_out = size;
 885
 886         inflateInit(&stream);
 887         st = inflate(&stream, Z_FINISH);
 888         inflateEnd(&stream);
 889         if ((st != Z_STREAM_END) || stream.total_out != size) {
 890                 free(buffer);
 891                 return NULL;
 892         }
 893
 894         return buffer;
 895 }
 896
 897 static void *unpack_entry(struct pack_entry *entry,
 898                           char *type, unsigned long *sizep)
 899 {
 900         struct packed_git *p = entry->p;
 901         void *retval;
 902
 903         if (use_packed_git(p))
 904                 die("cannot map packed file");
 905         retval = unpack_entry_gently(entry, type, sizep);
 906         unuse_packed_git(p);
 907         if (!retval)
 908                 die("corrupted pack file");
 909         return retval;
 910 }
 911
 912 /* The caller is responsible for use_packed_git()/unuse_packed_git() pair */
 913 void *unpack_entry_gently(struct pack_entry *entry,
 914                           char *type, unsigned long *sizep)
 915 {
 916         struct packed_git *p = entry->p;
 917         unsigned long offset, size, left;
 918         unsigned char *pack;
 919         enum object_type kind;
 920         void *retval;
 921
 922         offset = unpack_object_header(p, entry->offset, &kind, &size);
 923         pack = p->pack_base + offset;
 924         left = p->pack_size - offset;
 925         switch (kind) {
 926         case OBJ_DELTA:
 927                 retval = unpack_delta_entry(pack, size, left, type, sizep, p);
 928                 return retval;
 929         case OBJ_COMMIT:
 930                 strcpy(type, "commit");
 931                 break;
 932         case OBJ_TREE:
 933                 strcpy(type, "tree");
 934                 break;
 935         case OBJ_BLOB:
 936                 strcpy(type, "blob");
 937                 break;
 938         case OBJ_TAG:
 939                 strcpy(type, "tag");
 940                 break;
 941         default:
 942                 return NULL;
 943         }
 944         *sizep = size;
 945         retval = unpack_non_delta_entry(pack, size, left);
 946         return retval;
 947 }
 948
 949 int num_packed_objects(const struct packed_git *p)
 950 {
 951         /* See check_packed_git_idx() */
 952         return (p->index_size - 20 - 20 - 4*256) / 24;
 953 }
 954
 955 int nth_packed_object_sha1(const struct packed_git *p, int n,
 956                            unsigned char* sha1)
 957 {
 958         void *index = p->index_base + 256;
 959         if (n < 0 || num_packed_objects(p) <= n)
 960                 return -1;
 961         memcpy(sha1, (index + 24 * n + 4), 20);
 962         return 0;
 963 }
 964
 965 int find_pack_entry_one(const unsigned char *sha1,
 966                         struct pack_entry *e, struct packed_git *p)
 967 {
 968         unsigned int *level1_ofs = p->index_base;
 969         int hi = ntohl(level1_ofs[*sha1]);
 970         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
 971         void *index = p->index_base + 256;
 972
 973         do {
 974                 int mi = (lo + hi) / 2;
 975                 int cmp = memcmp(index + 24 * mi + 4, sha1, 20);
 976                 if (!cmp) {
 977                         e->offset = ntohl(*((int*)(index + 24 * mi)));
 978                         memcpy(e->sha1, sha1, 20);
 979                         e->p = p;
 980                         return 1;
 981                 }
 982                 if (cmp > 0)
 983                         hi = mi;
 984                 else
 985                         lo = mi+1;
 986         } while (lo < hi);
 987         return 0;
 988 }
 989
 990 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e)
 991 {
 992         struct packed_git *p;
 993         prepare_packed_git();
 994
 995         for (p = packed_git; p; p = p->next) {
 996                 if (find_pack_entry_one(sha1, e, p))
 997                         return 1;
 998         }
 999         return 0;
1000 }
1001
1002 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
1003 {
1004         int status;
1005         unsigned long mapsize, size;
1006         void *map;
1007         z_stream stream;
1008         char hdr[128];
1009
1010         map = map_sha1_file_internal(sha1, &mapsize, 0);
1011         if (!map) {
1012                 struct pack_entry e;
1013
1014                 if (!find_pack_entry(sha1, &e))
1015                         return error("unable to find %s", sha1_to_hex(sha1));
1016                 return packed_object_info(&e, type, sizep);
1017         }
1018         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
1019                 status = error("unable to unpack %s header",
1020                                sha1_to_hex(sha1));
1021         if (parse_sha1_header(hdr, type, &size) < 0)
1022                 status = error("unable to parse %s header", sha1_to_hex(sha1));
1023         else {
1024                 status = 0;
1025                 if (sizep)
1026                         *sizep = size;
1027         }
1028         inflateEnd(&stream);
1029         munmap(map, mapsize);
1030         return status;
1031 }
1032
1033 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
1034 {
1035         struct pack_entry e;
1036
1037         if (!find_pack_entry(sha1, &e)) {
1038                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
1039                 return NULL;
1040         }
1041         return unpack_entry(&e, type, size);
1042 }
1043
1044 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
1045 {
1046         unsigned long mapsize;
1047         void *map, *buf;
1048
1049         map = map_sha1_file_internal(sha1, &mapsize, 0);
1050         if (map) {
1051                 buf = unpack_sha1_file(map, mapsize, type, size);
1052                 munmap(map, mapsize);
1053                 return buf;
1054         }
1055         return read_packed_sha1(sha1, type, size);
1056 }
1057
1058 void *read_object_with_reference(const unsigned char *sha1,
1059                                  const char *required_type,
1060                                  unsigned long *size,
1061                                  unsigned char *actual_sha1_return)
1062 {
1063         char type[20];
1064         void *buffer;
1065         unsigned long isize;
1066         unsigned char actual_sha1[20];
1067
1068         memcpy(actual_sha1, sha1, 20);
1069         while (1) {
1070                 int ref_length = -1;
1071                 const char *ref_type = NULL;
1072
1073                 buffer = read_sha1_file(actual_sha1, type, &isize);
1074                 if (!buffer)
1075                         return NULL;
1076                 if (!strcmp(type, required_type)) {
1077                         *size = isize;
1078                         if (actual_sha1_return)
1079                                 memcpy(actual_sha1_return, actual_sha1, 20);
1080                         return buffer;
1081                 }
1082                 /* Handle references */
1083                 else if (!strcmp(type, "commit"))
1084                         ref_type = "tree ";
1085                 else if (!strcmp(type, "tag"))
1086                         ref_type = "object ";
1087                 else {
1088                         free(buffer);
1089                         return NULL;
1090                 }
1091                 ref_length = strlen(ref_type);
1092
1093                 if (memcmp(buffer, ref_type, ref_length) ||
1094                     get_sha1_hex(buffer + ref_length, actual_sha1)) {
1095                         free(buffer);
1096                         return NULL;
1097                 }
1098                 /* Now we have the ID of the referred-to object in
1099                  * actual_sha1.  Check again. */
1100         }
1101 }
1102
1103 char *write_sha1_file_prepare(void *buf,
1104                               unsigned long len,
1105                               const char *type,
1106                               unsigned char *sha1,
1107                               unsigned char *hdr,
1108                               int *hdrlen)
1109 {
1110         SHA_CTX c;
1111
1112         /* Generate the header */
1113         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
1114
1115         /* Sha1.. */
1116         SHA1_Init(&c);
1117         SHA1_Update(&c, hdr, *hdrlen);
1118         SHA1_Update(&c, buf, len);
1119         SHA1_Final(sha1, &c);
1120
1121         return sha1_file_name(sha1);
1122 }
1123
1124 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1125 {
1126         int size;
1127         unsigned char *compressed;
1128         z_stream stream;
1129         unsigned char sha1[20];
1130         char *filename;
1131         static char tmpfile[PATH_MAX];
1132         unsigned char hdr[50];
1133         int fd, hdrlen, ret;
1134
1135         /* Normally if we have it in the pack then we do not bother writing
1136          * it out into .git/objects/??/?{38} file.
1137          */
1138         filename = write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1139         if (returnsha1)
1140                 memcpy(returnsha1, sha1, 20);
1141         if (has_sha1_file(sha1))
1142                 return 0;
1143         fd = open(filename, O_RDONLY);
1144         if (fd >= 0) {
1145                 /*
1146                  * FIXME!!! We might do collision checking here, but we'd
1147                  * need to uncompress the old file and check it. Later.
1148                  */
1149                 close(fd);
1150                 return 0;
1151         }
1152
1153         if (errno != ENOENT) {
1154                 fprintf(stderr, "sha1 file %s: %s", filename, strerror(errno));
1155                 return -1;
1156         }
1157
1158         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1159
1160         fd = mkstemp(tmpfile);
1161         if (fd < 0) {
1162                 fprintf(stderr, "unable to create temporary sha1 filename %s: %s", tmpfile, strerror(errno));
1163                 return -1;
1164         }
1165
1166         /* Set it up */
1167         memset(&stream, 0, sizeof(stream));
1168         deflateInit(&stream, Z_BEST_COMPRESSION);
1169         size = deflateBound(&stream, len+hdrlen);
1170         compressed = xmalloc(size);
1171
1172         /* Compress it */
1173         stream.next_out = compressed;
1174         stream.avail_out = size;
1175
1176         /* First header.. */
1177         stream.next_in = hdr;
1178         stream.avail_in = hdrlen;
1179         while (deflate(&stream, 0) == Z_OK)
1180                 /* nothing */;
1181
1182         /* Then the data itself.. */
1183         stream.next_in = buf;
1184         stream.avail_in = len;
1185         while (deflate(&stream, Z_FINISH) == Z_OK)
1186                 /* nothing */;
1187         deflateEnd(&stream);
1188         size = stream.total_out;
1189
1190         if (write(fd, compressed, size) != size)
1191                 die("unable to write file");
1192         fchmod(fd, 0444);
1193         close(fd);
1194         free(compressed);
1195
1196         ret = link(tmpfile, filename);
1197         if (ret < 0) {
1198                 ret = errno;
1199
1200                 /*
1201                  * Coda hack - coda doesn't like cross-directory links,
1202                  * so we fall back to a rename, which will mean that it
1203                  * won't be able to check collisions, but that's not a
1204                  * big deal.
1205                  *
1206                  * When this succeeds, we just return 0. We have nothing
1207                  * left to unlink.
1208                  */
1209                 if (ret == EXDEV && !rename(tmpfile, filename))
1210                         return 0;
1211         }
1212         unlink(tmpfile);
1213         if (ret) {
1214                 if (ret != EEXIST) {
1215                         fprintf(stderr, "unable to write sha1 filename %s: %s", filename, strerror(ret));
1216                         return -1;
1217                 }
1218                 /* FIXME!!! Collision check here ? */
1219         }
1220
1221         return 0;
1222 }
1223
1224 int write_sha1_from_fd(const unsigned char *sha1, int fd)
1225 {
1226         char *filename = sha1_file_name(sha1);
1227
1228         int local;
1229         z_stream stream;
1230         unsigned char real_sha1[20];
1231         unsigned char buf[4096];
1232         unsigned char discard[4096];
1233         int ret;
1234         SHA_CTX c;
1235
1236         local = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
1237
1238         if (local < 0)
1239                 return error("Couldn't open %s\n", filename);
1240
1241         memset(&stream, 0, sizeof(stream));
1242
1243         inflateInit(&stream);
1244
1245         SHA1_Init(&c);
1246
1247         do {
1248                 ssize_t size;
1249                 size = read(fd, buf, 4096);
1250                 if (size <= 0) {
1251                         close(local);
1252                         unlink(filename);
1253                         if (!size)
1254                                 return error("Connection closed?");
1255                         perror("Reading from connection");
1256                         return -1;
1257                 }
1258                 write(local, buf, size);
1259                 stream.avail_in = size;
1260                 stream.next_in = buf;
1261                 do {
1262                         stream.next_out = discard;
1263                         stream.avail_out = sizeof(discard);
1264                         ret = inflate(&stream, Z_SYNC_FLUSH);
1265                         SHA1_Update(&c, discard, sizeof(discard) -
1266                                     stream.avail_out);
1267                 } while (stream.avail_in && ret == Z_OK);
1268
1269         } while (ret == Z_OK);
1270         inflateEnd(&stream);
1271
1272         close(local);
1273         SHA1_Final(real_sha1, &c);
1274         if (ret != Z_STREAM_END) {
1275                 unlink(filename);
1276                 return error("File %s corrupted", sha1_to_hex(sha1));
1277         }
1278         if (memcmp(sha1, real_sha1, 20)) {
1279                 unlink(filename);
1280                 return error("File %s has bad hash\n", sha1_to_hex(sha1));
1281         }
1282
1283         return 0;
1284 }
1285
1286 int has_sha1_pack(const unsigned char *sha1)
1287 {
1288         struct pack_entry e;
1289         return find_pack_entry(sha1, &e);
1290 }
1291
1292 int has_sha1_file(const unsigned char *sha1)
1293 {
1294         struct stat st;
1295         struct pack_entry e;
1296
1297         if (find_sha1_file(sha1, &st))
1298                 return 1;
1299         return find_pack_entry(sha1, &e);
1300 }
1301
1302 int index_fd(unsigned char *sha1, int fd, struct stat *st, int write_object, const char *type)
1303 {
1304         unsigned long size = st->st_size;
1305         void *buf;
1306         int ret;
1307         unsigned char hdr[50];
1308         int hdrlen;
1309
1310         buf = "";
1311         if (size)
1312                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1313         close(fd);
1314         if ((int)(long)buf == -1)
1315                 return -1;
1316
1317         if (!type)
1318                 type = "blob";
1319         if (write_object)
1320                 ret = write_sha1_file(buf, size, type, sha1);
1321         else {
1322                 write_sha1_file_prepare(buf, size, type, sha1, hdr, &hdrlen);
1323                 ret = 0;
1324         }
1325         if (size)
1326                 munmap(buf, size);
1327         return ret;
1328 }