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         *git_graft_file;
  66 static void setup_git_env(void)
  67 {
  68         git_dir = gitenv(GIT_DIR_ENVIRONMENT);
  69         if (!git_dir)
  70                 git_dir = DEFAULT_GIT_DIR_ENVIRONMENT;
  71         git_object_dir = gitenv(DB_ENVIRONMENT);
  72         if (!git_object_dir) {
  73                 git_object_dir = xmalloc(strlen(git_dir) + 9);
  74                 sprintf(git_object_dir, "%s/objects", git_dir);
  75         }
  76         git_refs_dir = xmalloc(strlen(git_dir) + 6);
  77         sprintf(git_refs_dir, "%s/refs", git_dir);
  78         git_index_file = gitenv(INDEX_ENVIRONMENT);
  79         if (!git_index_file) {
  80                 git_index_file = xmalloc(strlen(git_dir) + 7);
  81                 sprintf(git_index_file, "%s/index", git_dir);
  82         }
  83         git_graft_file = gitenv(GRAFT_ENVIRONMENT);
  84         if (!git_graft_file)
  85                 git_graft_file = strdup(git_path("info/grafts"));
  86 }
  87
  88 char *get_object_directory(void)
  89 {
  90         if (!git_object_dir)
  91                 setup_git_env();
  92         return git_object_dir;
  93 }
  94
  95 char *get_refs_directory(void)
  96 {
  97         if (!git_refs_dir)
  98                 setup_git_env();
  99         return git_refs_dir;
 100 }
 101
 102 char *get_index_file(void)
 103 {
 104         if (!git_index_file)
 105                 setup_git_env();
 106         return git_index_file;
 107 }
 108
 109 char *get_graft_file(void)
 110 {
 111         if (!git_graft_file)
 112                 setup_git_env();
 113         return git_graft_file;
 114 }
 115
 116 int safe_create_leading_directories(char *path)
 117 {
 118         char *pos = path;
 119
 120         while (pos) {
 121                 pos = strchr(pos, '/');
 122                 if (!pos)
 123                         break;
 124                 *pos = 0;
 125                 if (mkdir(path, 0777) < 0)
 126                         if (errno != EEXIST) {
 127                                 *pos = '/';
 128                                 return -1;
 129                         }
 130                 *pos++ = '/';
 131         }
 132         return 0;
 133 }
 134
 135 int get_sha1(const char *str, unsigned char *sha1)
 136 {
 137         static const char *prefix[] = {
 138                 "",
 139                 "refs",
 140                 "refs/tags",
 141                 "refs/heads",
 142                 "refs/snap",
 143                 NULL
 144         };
 145         const char **p;
 146
 147         if (!get_sha1_hex(str, sha1))
 148                 return 0;
 149
 150         for (p = prefix; *p; p++) {
 151                 char * pathname = git_path("%s/%s", *p, str);
 152                 if (!get_sha1_file(pathname, sha1))
 153                         return 0;
 154         }
 155
 156         return -1;
 157 }
 158
 159 char * sha1_to_hex(const unsigned char *sha1)
 160 {
 161         static char buffer[50];
 162         static const char hex[] = "0123456789abcdef";
 163         char *buf = buffer;
 164         int i;
 165
 166         for (i = 0; i < 20; i++) {
 167                 unsigned int val = *sha1++;
 168                 *buf++ = hex[val >> 4];
 169                 *buf++ = hex[val & 0xf];
 170         }
 171         return buffer;
 172 }
 173
 174 static void fill_sha1_path(char *pathbuf, const unsigned char *sha1)
 175 {
 176         int i;
 177         for (i = 0; i < 20; i++) {
 178                 static char hex[] = "0123456789abcdef";
 179                 unsigned int val = sha1[i];
 180                 char *pos = pathbuf + i*2 + (i > 0);
 181                 *pos++ = hex[val >> 4];
 182                 *pos = hex[val & 0xf];
 183         }
 184 }
 185
 186 /*
 187  * NOTE! This returns a statically allocated buffer, so you have to be
 188  * careful about using it. Do a "strdup()" if you need to save the
 189  * filename.
 190  *
 191  * Also note that this returns the location for creating.  Reading
 192  * SHA1 file can happen from any alternate directory listed in the
 193  * DB_ENVIRONMENT environment variable if it is not found in
 194  * the primary object database.
 195  */
 196 char *sha1_file_name(const unsigned char *sha1)
 197 {
 198         static char *name, *base;
 199
 200         if (!base) {
 201                 const char *sha1_file_directory = get_object_directory();
 202                 int len = strlen(sha1_file_directory);
 203                 base = xmalloc(len + 60);
 204                 memcpy(base, sha1_file_directory, len);
 205                 memset(base+len, 0, 60);
 206                 base[len] = '/';
 207                 base[len+3] = '/';
 208                 name = base + len + 1;
 209         }
 210         fill_sha1_path(name, sha1);
 211         return base;
 212 }
 213
 214 struct alternate_object_database *alt_odb;
 215
 216 /*
 217  * Prepare alternate object database registry.
 218  * alt_odb points at an array of struct alternate_object_database.
 219  * This array is terminated with an element that has both its base
 220  * and name set to NULL.  alt_odb[n] comes from n'th non-empty
 221  * element from colon separated ALTERNATE_DB_ENVIRONMENT environment
 222  * variable, and its base points at a statically allocated buffer
 223  * that contains "/the/directory/corresponding/to/.git/objects/...",
 224  * while its name points just after the slash at the end of
 225  * ".git/objects/" in the example above, and has enough space to hold
 226  * 40-byte hex SHA1, an extra slash for the first level indirection,
 227  * and the terminating NUL.
 228  * This function allocates the alt_odb array and all the strings
 229  * pointed by base fields of the array elements with one xmalloc();
 230  * the string pool immediately follows the array.
 231  */
 232 void prepare_alt_odb(void)
 233 {
 234         int pass, totlen, i;
 235         const char *cp, *last;
 236         char *op = NULL;
 237         const char *alt = gitenv(ALTERNATE_DB_ENVIRONMENT) ? : "";
 238
 239         if (alt_odb)
 240                 return;
 241         /* The first pass counts how large an area to allocate to
 242          * hold the entire alt_odb structure, including array of
 243          * structs and path buffers for them.  The second pass fills
 244          * the structure and prepares the path buffers for use by
 245          * fill_sha1_path().
 246          */
 247         for (totlen = pass = 0; pass < 2; pass++) {
 248                 last = alt;
 249                 i = 0;
 250                 do {
 251                         cp = strchr(last, ':') ? : last + strlen(last);
 252                         if (last != cp) {
 253                                 /* 43 = 40-byte + 2 '/' + terminating NUL */
 254                                 int pfxlen = cp - last;
 255                                 int entlen = pfxlen + 43;
 256                                 if (pass == 0)
 257                                         totlen += entlen;
 258                                 else {
 259                                         alt_odb[i].base = op;
 260                                         alt_odb[i].name = op + pfxlen + 1;
 261                                         memcpy(op, last, pfxlen);
 262                                         op[pfxlen] = op[pfxlen + 3] = '/';
 263                                         op[entlen-1] = 0;
 264                                         op += entlen;
 265                                 }
 266                                 i++;
 267                         }
 268                         while (*cp && *cp == ':')
 269                                 cp++;
 270                         last = cp;
 271                 } while (*cp);
 272                 if (pass)
 273                         break;
 274                 alt_odb = xmalloc(sizeof(*alt_odb) * (i + 1) + totlen);
 275                 alt_odb[i].base = alt_odb[i].name = NULL;
 276                 op = (char*)(&alt_odb[i+1]);
 277         }
 278 }
 279
 280 static char *find_sha1_file(const unsigned char *sha1, struct stat *st)
 281 {
 282         int i;
 283         char *name = sha1_file_name(sha1);
 284
 285         if (!stat(name, st))
 286                 return name;
 287         prepare_alt_odb();
 288         for (i = 0; (name = alt_odb[i].name) != NULL; i++) {
 289                 fill_sha1_path(name, sha1);
 290                 if (!stat(alt_odb[i].base, st))
 291                         return alt_odb[i].base;
 292         }
 293         return NULL;
 294 }
 295
 296 #define PACK_MAX_SZ (1<<26)
 297 static int pack_used_ctr;
 298 static unsigned long pack_mapped;
 299 struct packed_git *packed_git;
 300
 301 static int check_packed_git_idx(const char *path, unsigned long *idx_size_,
 302                                 void **idx_map_)
 303 {
 304         void *idx_map;
 305         unsigned int *index;
 306         unsigned long idx_size;
 307         int nr, i;
 308         int fd = open(path, O_RDONLY);
 309         struct stat st;
 310         if (fd < 0)
 311                 return -1;
 312         if (fstat(fd, &st)) {
 313                 close(fd);
 314                 return -1;
 315         }
 316         idx_size = st.st_size;
 317         idx_map = mmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0);
 318         close(fd);
 319         if (idx_map == MAP_FAILED)
 320                 return -1;
 321
 322         index = idx_map;
 323         *idx_map_ = idx_map;
 324         *idx_size_ = idx_size;
 325
 326         /* check index map */
 327         if (idx_size < 4*256 + 20 + 20)
 328                 return error("index file too small");
 329         nr = 0;
 330         for (i = 0; i < 256; i++) {
 331                 unsigned int n = ntohl(index[i]);
 332                 if (n < nr)
 333                         return error("non-monotonic index");
 334                 nr = n;
 335         }
 336
 337         /*
 338          * Total size:
 339          *  - 256 index entries 4 bytes each
 340          *  - 24-byte entries * nr (20-byte sha1 + 4-byte offset)
 341          *  - 20-byte SHA1 of the packfile
 342          *  - 20-byte SHA1 file checksum
 343          */
 344         if (idx_size != 4*256 + nr * 24 + 20 + 20)
 345                 return error("wrong index file size");
 346
 347         return 0;
 348 }
 349
 350 static int unuse_one_packed_git(void)
 351 {
 352         struct packed_git *p, *lru = NULL;
 353
 354         for (p = packed_git; p; p = p->next) {
 355                 if (p->pack_use_cnt || !p->pack_base)
 356                         continue;
 357                 if (!lru || p->pack_last_used < lru->pack_last_used)
 358                         lru = p;
 359         }
 360         if (!lru)
 361                 return 0;
 362         munmap(lru->pack_base, lru->pack_size);
 363         lru->pack_base = NULL;
 364         return 1;
 365 }
 366
 367 void unuse_packed_git(struct packed_git *p)
 368 {
 369         p->pack_use_cnt--;
 370 }
 371
 372 int use_packed_git(struct packed_git *p)
 373 {
 374         if (!p->pack_base) {
 375                 int fd;
 376                 struct stat st;
 377                 void *map;
 378
 379                 pack_mapped += p->pack_size;
 380                 while (PACK_MAX_SZ < pack_mapped && unuse_one_packed_git())
 381                         ; /* nothing */
 382                 fd = open(p->pack_name, O_RDONLY);
 383                 if (fd < 0)
 384                         die("packfile %s cannot be opened", p->pack_name);
 385                 if (fstat(fd, &st)) {
 386                         close(fd);
 387                         die("packfile %s cannot be opened", p->pack_name);
 388                 }
 389                 if (st.st_size != p->pack_size)
 390                         die("packfile %s size mismatch.", p->pack_name);
 391                 map = mmap(NULL, p->pack_size, PROT_READ, MAP_PRIVATE, fd, 0);
 392                 close(fd);
 393                 if (map == MAP_FAILED)
 394                         die("packfile %s cannot be mapped.", p->pack_name);
 395                 p->pack_base = map;
 396
 397                 /* Check if the pack file matches with the index file.
 398                  * this is cheap.
 399                  */
 400                 if (memcmp((char*)(p->index_base) + p->index_size - 40,
 401                            p->pack_base + p->pack_size - 20, 20))
 402                         die("packfile %s does not match index.", p->pack_name);
 403         }
 404         p->pack_last_used = pack_used_ctr++;
 405         p->pack_use_cnt++;
 406         return 0;
 407 }
 408
 409 struct packed_git *add_packed_git(char *path, int path_len)
 410 {
 411         struct stat st;
 412         struct packed_git *p;
 413         unsigned long idx_size;
 414         void *idx_map;
 415
 416         if (check_packed_git_idx(path, &idx_size, &idx_map))
 417                 return NULL;
 418
 419         /* do we have a corresponding .pack file? */
 420         strcpy(path + path_len - 4, ".pack");
 421         if (stat(path, &st) || !S_ISREG(st.st_mode)) {
 422                 munmap(idx_map, idx_size);
 423                 return NULL;
 424         }
 425         /* ok, it looks sane as far as we can check without
 426          * actually mapping the pack file.
 427          */
 428         p = xmalloc(sizeof(*p) + path_len + 2);
 429         strcpy(p->pack_name, path);
 430         p->index_size = idx_size;
 431         p->pack_size = st.st_size;
 432         p->index_base = idx_map;
 433         p->next = NULL;
 434         p->pack_base = NULL;
 435         p->pack_last_used = 0;
 436         p->pack_use_cnt = 0;
 437         return p;
 438 }
 439
 440 static void prepare_packed_git_one(char *objdir)
 441 {
 442         char path[PATH_MAX];
 443         int len;
 444         DIR *dir;
 445         struct dirent *de;
 446
 447         sprintf(path, "%s/pack", objdir);
 448         len = strlen(path);
 449         dir = opendir(path);
 450         if (!dir)
 451                 return;
 452         path[len++] = '/';
 453         while ((de = readdir(dir)) != NULL) {
 454                 int namelen = strlen(de->d_name);
 455                 struct packed_git *p;
 456
 457                 if (strcmp(de->d_name + namelen - 4, ".idx"))
 458                         continue;
 459
 460                 /* we have .idx.  Is it a file we can map? */
 461                 strcpy(path + len, de->d_name);
 462                 p = add_packed_git(path, len + namelen);
 463                 if (!p)
 464                         continue;
 465                 p->next = packed_git;
 466                 packed_git = p;
 467         }
 468         closedir(dir);
 469 }
 470
 471 void prepare_packed_git(void)
 472 {
 473         int i;
 474         static int run_once = 0;
 475
 476         if (run_once++)
 477                 return;
 478
 479         prepare_packed_git_one(get_object_directory());
 480         prepare_alt_odb();
 481         for (i = 0; alt_odb[i].base != NULL; i++) {
 482                 alt_odb[i].name[0] = 0;
 483                 prepare_packed_git_one(alt_odb[i].base);
 484         }
 485 }
 486
 487 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 488 {
 489         char header[100];
 490         unsigned char real_sha1[20];
 491         SHA_CTX c;
 492
 493         SHA1_Init(&c);
 494         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 495         SHA1_Update(&c, map, size);
 496         SHA1_Final(real_sha1, &c);
 497         return memcmp(sha1, real_sha1, 20) ? -1 : 0;
 498 }
 499
 500 static void *map_sha1_file_internal(const unsigned char *sha1,
 501                                     unsigned long *size)
 502 {
 503         struct stat st;
 504         void *map;
 505         int fd;
 506         char *filename = find_sha1_file(sha1, &st);
 507
 508         if (!filename) {
 509                 return NULL;
 510         }
 511
 512         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 513         if (fd < 0) {
 514                 /* See if it works without O_NOATIME */
 515                 switch (sha1_file_open_flag) {
 516                 default:
 517                         fd = open(filename, O_RDONLY);
 518                         if (fd >= 0)
 519                                 break;
 520                 /* Fallthrough */
 521                 case 0:
 522                         return NULL;
 523                 }
 524
 525                 /* If it failed once, it will probably fail again.
 526                  * Stop using O_NOATIME
 527                  */
 528                 sha1_file_open_flag = 0;
 529         }
 530         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 531         close(fd);
 532         if (map == MAP_FAILED)
 533                 return NULL;
 534         *size = st.st_size;
 535         return map;
 536 }
 537
 538 int unpack_sha1_header(z_stream *stream, void *map, unsigned long mapsize, void *buffer, unsigned long size)
 539 {
 540         /* Get the data stream */
 541         memset(stream, 0, sizeof(*stream));
 542         stream->next_in = map;
 543         stream->avail_in = mapsize;
 544         stream->next_out = buffer;
 545         stream->avail_out = size;
 546
 547         inflateInit(stream);
 548         return inflate(stream, 0);
 549 }
 550
 551 static void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 552 {
 553         int bytes = strlen(buffer) + 1;
 554         unsigned char *buf = xmalloc(1+size);
 555
 556         memcpy(buf, buffer + bytes, stream->total_out - bytes);
 557         bytes = stream->total_out - bytes;
 558         if (bytes < size) {
 559                 stream->next_out = buf + bytes;
 560                 stream->avail_out = size - bytes;
 561                 while (inflate(stream, Z_FINISH) == Z_OK)
 562                         /* nothing */;
 563         }
 564         buf[size] = 0;
 565         inflateEnd(stream);
 566         return buf;
 567 }
 568
 569 /*
 570  * We used to just use "sscanf()", but that's actually way
 571  * too permissive for what we want to check. So do an anal
 572  * object header parse by hand.
 573  */
 574 int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 575 {
 576         int i;
 577         unsigned long size;
 578
 579         /*
 580          * The type can be at most ten bytes (including the
 581          * terminating '\0' that we add), and is followed by
 582          * a space.
 583          */
 584         i = 10;
 585         for (;;) {
 586                 char c = *hdr++;
 587                 if (c == ' ')
 588                         break;
 589                 if (!--i)
 590                         return -1;
 591                 *type++ = c;
 592         }
 593         *type = 0;
 594
 595         /*
 596          * The length must follow immediately, and be in canonical
 597          * decimal format (ie "010" is not valid).
 598          */
 599         size = *hdr++ - '0';
 600         if (size > 9)
 601                 return -1;
 602         if (size) {
 603                 for (;;) {
 604                         unsigned long c = *hdr - '0';
 605                         if (c > 9)
 606                                 break;
 607                         hdr++;
 608                         size = size * 10 + c;
 609                 }
 610         }
 611         *sizep = size;
 612
 613         /*
 614          * The length must be followed by a zero byte
 615          */
 616         return *hdr ? -1 : 0;
 617 }
 618
 619 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 620 {
 621         int ret;
 622         z_stream stream;
 623         char hdr[8192];
 624
 625         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
 626         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
 627                 return NULL;
 628
 629         return unpack_sha1_rest(&stream, hdr, *size);
 630 }
 631
 632 /* forward declaration for a mutually recursive function */
 633 static int packed_object_info(struct pack_entry *entry,
 634                               char *type, unsigned long *sizep);
 635
 636 static int packed_delta_info(unsigned char *base_sha1,
 637                              unsigned long delta_size,
 638                              unsigned long left,
 639                              char *type,
 640                              unsigned long *sizep,
 641                              struct packed_git *p)
 642 {
 643         struct pack_entry base_ent;
 644
 645         if (left < 20)
 646                 die("truncated pack file");
 647
 648         /* The base entry _must_ be in the same pack */
 649         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 650                 die("failed to find delta-pack base object %s",
 651                     sha1_to_hex(base_sha1));
 652
 653         /* We choose to only get the type of the base object and
 654          * ignore potentially corrupt pack file that expects the delta
 655          * based on a base with a wrong size.  This saves tons of
 656          * inflate() calls.
 657          */
 658
 659         if (packed_object_info(&base_ent, type, NULL))
 660                 die("cannot get info for delta-pack base");
 661
 662         if (sizep) {
 663                 const unsigned char *data;
 664                 unsigned char delta_head[64];
 665                 unsigned long result_size;
 666                 z_stream stream;
 667                 int st;
 668
 669                 memset(&stream, 0, sizeof(stream));
 670
 671                 data = stream.next_in = base_sha1 + 20;
 672                 stream.avail_in = left - 20;
 673                 stream.next_out = delta_head;
 674                 stream.avail_out = sizeof(delta_head);
 675
 676                 inflateInit(&stream);
 677                 st = inflate(&stream, Z_FINISH);
 678                 inflateEnd(&stream);
 679                 if ((st != Z_STREAM_END) &&
 680                     stream.total_out != sizeof(delta_head))
 681                         die("delta data unpack-initial failed");
 682
 683                 /* Examine the initial part of the delta to figure out
 684                  * the result size.
 685                  */
 686                 data = delta_head;
 687                 get_delta_hdr_size(&data); /* ignore base size */
 688
 689                 /* Read the result size */
 690                 result_size = get_delta_hdr_size(&data);
 691                 *sizep = result_size;
 692         }
 693         return 0;
 694 }
 695
 696 static unsigned long unpack_object_header(struct packed_git *p, unsigned long offset,
 697         enum object_type *type, unsigned long *sizep)
 698 {
 699         unsigned shift;
 700         unsigned char *pack, c;
 701         unsigned long size;
 702
 703         if (offset >= p->pack_size)
 704                 die("object offset outside of pack file");
 705
 706         pack =  p->pack_base + offset;
 707         c = *pack++;
 708         offset++;
 709         *type = (c >> 4) & 7;
 710         size = c & 15;
 711         shift = 4;
 712         while (c & 0x80) {
 713                 if (offset >= p->pack_size)
 714                         die("object offset outside of pack file");
 715                 c = *pack++;
 716                 offset++;
 717                 size += (c & 0x7f) << shift;
 718                 shift += 7;
 719         }
 720         *sizep = size;
 721         return offset;
 722 }
 723
 724 void packed_object_info_detail(struct pack_entry *e,
 725                                char *type,
 726                                unsigned long *size,
 727                                unsigned long *store_size,
 728                                int *delta_chain_length,
 729                                unsigned char *base_sha1)
 730 {
 731         struct packed_git *p = e->p;
 732         unsigned long offset, left;
 733         unsigned char *pack;
 734         enum object_type kind;
 735
 736         offset = unpack_object_header(p, e->offset, &kind, size);
 737         pack = p->pack_base + offset;
 738         left = p->pack_size - offset;
 739         if (kind != OBJ_DELTA)
 740                 *delta_chain_length = 0;
 741         else {
 742                 int chain_length = 0;
 743                 memcpy(base_sha1, pack, 20);
 744                 do {
 745                         struct pack_entry base_ent;
 746                         unsigned long junk;
 747
 748                         find_pack_entry_one(pack, &base_ent, p);
 749                         offset = unpack_object_header(p, base_ent.offset,
 750                                                       &kind, &junk);
 751                         pack = p->pack_base + offset;
 752                         chain_length++;
 753                 } while (kind == OBJ_DELTA);
 754                 *delta_chain_length = chain_length;
 755         }
 756         switch (kind) {
 757         case OBJ_COMMIT:
 758                 strcpy(type, "commit");
 759                 break;
 760         case OBJ_TREE:
 761                 strcpy(type, "tree");
 762                 break;
 763         case OBJ_BLOB:
 764                 strcpy(type, "blob");
 765                 break;
 766         case OBJ_TAG:
 767                 strcpy(type, "tag");
 768                 break;
 769         default:
 770                 die("corrupted pack file");
 771         }
 772         *store_size = 0; /* notyet */
 773 }
 774
 775 static int packed_object_info(struct pack_entry *entry,
 776                               char *type, unsigned long *sizep)
 777 {
 778         struct packed_git *p = entry->p;
 779         unsigned long offset, size, left;
 780         unsigned char *pack;
 781         enum object_type kind;
 782         int retval;
 783
 784         if (use_packed_git(p))
 785                 die("cannot map packed file");
 786
 787         offset = unpack_object_header(p, entry->offset, &kind, &size);
 788         pack = p->pack_base + offset;
 789         left = p->pack_size - offset;
 790
 791         switch (kind) {
 792         case OBJ_DELTA:
 793                 retval = packed_delta_info(pack, size, left, type, sizep, p);
 794                 unuse_packed_git(p);
 795                 return retval;
 796         case OBJ_COMMIT:
 797                 strcpy(type, "commit");
 798                 break;
 799         case OBJ_TREE:
 800                 strcpy(type, "tree");
 801                 break;
 802         case OBJ_BLOB:
 803                 strcpy(type, "blob");
 804                 break;
 805         case OBJ_TAG:
 806                 strcpy(type, "tag");
 807                 break;
 808         default:
 809                 die("corrupted pack file");
 810         }
 811         if (sizep)
 812                 *sizep = size;
 813         unuse_packed_git(p);
 814         return 0;
 815 }
 816
 817 /* forward declaration for a mutually recursive function */
 818 static void *unpack_entry(struct pack_entry *, char *, unsigned long *);
 819
 820 static void *unpack_delta_entry(unsigned char *base_sha1,
 821                                 unsigned long delta_size,
 822                                 unsigned long left,
 823                                 char *type,
 824                                 unsigned long *sizep,
 825                                 struct packed_git *p)
 826 {
 827         struct pack_entry base_ent;
 828         void *data, *delta_data, *result, *base;
 829         unsigned long data_size, result_size, base_size;
 830         z_stream stream;
 831         int st;
 832
 833         if (left < 20)
 834                 die("truncated pack file");
 835         data = base_sha1 + 20;
 836         data_size = left - 20;
 837         delta_data = xmalloc(delta_size);
 838
 839         memset(&stream, 0, sizeof(stream));
 840
 841         stream.next_in = data;
 842         stream.avail_in = data_size;
 843         stream.next_out = delta_data;
 844         stream.avail_out = delta_size;
 845
 846         inflateInit(&stream);
 847         st = inflate(&stream, Z_FINISH);
 848         inflateEnd(&stream);
 849         if ((st != Z_STREAM_END) || stream.total_out != delta_size)
 850                 die("delta data unpack failed");
 851
 852         /* The base entry _must_ be in the same pack */
 853         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 854                 die("failed to find delta-pack base object %s",
 855                     sha1_to_hex(base_sha1));
 856         base = unpack_entry_gently(&base_ent, type, &base_size);
 857         if (!base)
 858                 die("failed to read delta-pack base object %s",
 859                     sha1_to_hex(base_sha1));
 860         result = patch_delta(base, base_size,
 861                              delta_data, delta_size,
 862                              &result_size);
 863         if (!result)
 864                 die("failed to apply delta");
 865         free(delta_data);
 866         free(base);
 867         *sizep = result_size;
 868         return result;
 869 }
 870
 871 static void *unpack_non_delta_entry(unsigned char *data,
 872                                     unsigned long size,
 873                                     unsigned long left)
 874 {
 875         int st;
 876         z_stream stream;
 877         unsigned char *buffer;
 878
 879         buffer = xmalloc(size + 1);
 880         buffer[size] = 0;
 881         memset(&stream, 0, sizeof(stream));
 882         stream.next_in = data;
 883         stream.avail_in = left;
 884         stream.next_out = buffer;
 885         stream.avail_out = size;
 886
 887         inflateInit(&stream);
 888         st = inflate(&stream, Z_FINISH);
 889         inflateEnd(&stream);
 890         if ((st != Z_STREAM_END) || stream.total_out != size) {
 891                 free(buffer);
 892                 return NULL;
 893         }
 894
 895         return buffer;
 896 }
 897
 898 static void *unpack_entry(struct pack_entry *entry,
 899                           char *type, unsigned long *sizep)
 900 {
 901         struct packed_git *p = entry->p;
 902         void *retval;
 903
 904         if (use_packed_git(p))
 905                 die("cannot map packed file");
 906         retval = unpack_entry_gently(entry, type, sizep);
 907         unuse_packed_git(p);
 908         if (!retval)
 909                 die("corrupted pack file");
 910         return retval;
 911 }
 912
 913 /* The caller is responsible for use_packed_git()/unuse_packed_git() pair */
 914 void *unpack_entry_gently(struct pack_entry *entry,
 915                           char *type, unsigned long *sizep)
 916 {
 917         struct packed_git *p = entry->p;
 918         unsigned long offset, size, left;
 919         unsigned char *pack;
 920         enum object_type kind;
 921         void *retval;
 922
 923         offset = unpack_object_header(p, entry->offset, &kind, &size);
 924         pack = p->pack_base + offset;
 925         left = p->pack_size - offset;
 926         switch (kind) {
 927         case OBJ_DELTA:
 928                 retval = unpack_delta_entry(pack, size, left, type, sizep, p);
 929                 return retval;
 930         case OBJ_COMMIT:
 931                 strcpy(type, "commit");
 932                 break;
 933         case OBJ_TREE:
 934                 strcpy(type, "tree");
 935                 break;
 936         case OBJ_BLOB:
 937                 strcpy(type, "blob");
 938                 break;
 939         case OBJ_TAG:
 940                 strcpy(type, "tag");
 941                 break;
 942         default:
 943                 return NULL;
 944         }
 945         *sizep = size;
 946         retval = unpack_non_delta_entry(pack, size, left);
 947         return retval;
 948 }
 949
 950 int num_packed_objects(const struct packed_git *p)
 951 {
 952         /* See check_packed_git_idx() */
 953         return (p->index_size - 20 - 20 - 4*256) / 24;
 954 }
 955
 956 int nth_packed_object_sha1(const struct packed_git *p, int n,
 957                            unsigned char* sha1)
 958 {
 959         void *index = p->index_base + 256;
 960         if (n < 0 || num_packed_objects(p) <= n)
 961                 return -1;
 962         memcpy(sha1, (index + 24 * n + 4), 20);
 963         return 0;
 964 }
 965
 966 int find_pack_entry_one(const unsigned char *sha1,
 967                         struct pack_entry *e, struct packed_git *p)
 968 {
 969         unsigned int *level1_ofs = p->index_base;
 970         int hi = ntohl(level1_ofs[*sha1]);
 971         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
 972         void *index = p->index_base + 256;
 973
 974         do {
 975                 int mi = (lo + hi) / 2;
 976                 int cmp = memcmp(index + 24 * mi + 4, sha1, 20);
 977                 if (!cmp) {
 978                         e->offset = ntohl(*((int*)(index + 24 * mi)));
 979                         memcpy(e->sha1, sha1, 20);
 980                         e->p = p;
 981                         return 1;
 982                 }
 983                 if (cmp > 0)
 984                         hi = mi;
 985                 else
 986                         lo = mi+1;
 987         } while (lo < hi);
 988         return 0;
 989 }
 990
 991 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e)
 992 {
 993         struct packed_git *p;
 994         prepare_packed_git();
 995
 996         for (p = packed_git; p; p = p->next) {
 997                 if (find_pack_entry_one(sha1, e, p))
 998                         return 1;
 999         }
1000         return 0;
1001 }
1002
1003 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
1004 {
1005         int status;
1006         unsigned long mapsize, size;
1007         void *map;
1008         z_stream stream;
1009         char hdr[128];
1010
1011         map = map_sha1_file_internal(sha1, &mapsize);
1012         if (!map) {
1013                 struct pack_entry e;
1014
1015                 if (!find_pack_entry(sha1, &e))
1016                         return error("unable to find %s", sha1_to_hex(sha1));
1017                 return packed_object_info(&e, type, sizep);
1018         }
1019         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
1020                 status = error("unable to unpack %s header",
1021                                sha1_to_hex(sha1));
1022         if (parse_sha1_header(hdr, type, &size) < 0)
1023                 status = error("unable to parse %s header", sha1_to_hex(sha1));
1024         else {
1025                 status = 0;
1026                 if (sizep)
1027                         *sizep = size;
1028         }
1029         inflateEnd(&stream);
1030         munmap(map, mapsize);
1031         return status;
1032 }
1033
1034 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
1035 {
1036         struct pack_entry e;
1037
1038         if (!find_pack_entry(sha1, &e)) {
1039                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
1040                 return NULL;
1041         }
1042         return unpack_entry(&e, type, size);
1043 }
1044
1045 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
1046 {
1047         unsigned long mapsize;
1048         void *map, *buf;
1049         struct pack_entry e;
1050
1051         if (find_pack_entry(sha1, &e))
1052                 return read_packed_sha1(sha1, type, size);
1053         map = map_sha1_file_internal(sha1, &mapsize);
1054         if (map) {
1055                 buf = unpack_sha1_file(map, mapsize, type, size);
1056                 munmap(map, mapsize);
1057                 return buf;
1058         }
1059         return NULL;
1060 }
1061
1062 void *read_object_with_reference(const unsigned char *sha1,
1063                                  const char *required_type,
1064                                  unsigned long *size,
1065                                  unsigned char *actual_sha1_return)
1066 {
1067         char type[20];
1068         void *buffer;
1069         unsigned long isize;
1070         unsigned char actual_sha1[20];
1071
1072         memcpy(actual_sha1, sha1, 20);
1073         while (1) {
1074                 int ref_length = -1;
1075                 const char *ref_type = NULL;
1076
1077                 buffer = read_sha1_file(actual_sha1, type, &isize);
1078                 if (!buffer)
1079                         return NULL;
1080                 if (!strcmp(type, required_type)) {
1081                         *size = isize;
1082                         if (actual_sha1_return)
1083                                 memcpy(actual_sha1_return, actual_sha1, 20);
1084                         return buffer;
1085                 }
1086                 /* Handle references */
1087                 else if (!strcmp(type, "commit"))
1088                         ref_type = "tree ";
1089                 else if (!strcmp(type, "tag"))
1090                         ref_type = "object ";
1091                 else {
1092                         free(buffer);
1093                         return NULL;
1094                 }
1095                 ref_length = strlen(ref_type);
1096
1097                 if (memcmp(buffer, ref_type, ref_length) ||
1098                     get_sha1_hex(buffer + ref_length, actual_sha1)) {
1099                         free(buffer);
1100                         return NULL;
1101                 }
1102                 /* Now we have the ID of the referred-to object in
1103                  * actual_sha1.  Check again. */
1104         }
1105 }
1106
1107 char *write_sha1_file_prepare(void *buf,
1108                               unsigned long len,
1109                               const char *type,
1110                               unsigned char *sha1,
1111                               unsigned char *hdr,
1112                               int *hdrlen)
1113 {
1114         SHA_CTX c;
1115
1116         /* Generate the header */
1117         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
1118
1119         /* Sha1.. */
1120         SHA1_Init(&c);
1121         SHA1_Update(&c, hdr, *hdrlen);
1122         SHA1_Update(&c, buf, len);
1123         SHA1_Final(sha1, &c);
1124
1125         return sha1_file_name(sha1);
1126 }
1127
1128 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1129 {
1130         int size;
1131         unsigned char *compressed;
1132         z_stream stream;
1133         unsigned char sha1[20];
1134         char *filename;
1135         static char tmpfile[PATH_MAX];
1136         unsigned char hdr[50];
1137         int fd, hdrlen, ret;
1138
1139         /* Normally if we have it in the pack then we do not bother writing
1140          * it out into .git/objects/??/?{38} file.
1141          */
1142         filename = write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1143         if (returnsha1)
1144                 memcpy(returnsha1, sha1, 20);
1145         if (has_sha1_file(sha1))
1146                 return 0;
1147         fd = open(filename, O_RDONLY);
1148         if (fd >= 0) {
1149                 /*
1150                  * FIXME!!! We might do collision checking here, but we'd
1151                  * need to uncompress the old file and check it. Later.
1152                  */
1153                 close(fd);
1154                 return 0;
1155         }
1156
1157         if (errno != ENOENT) {
1158                 fprintf(stderr, "sha1 file %s: %s", filename, strerror(errno));
1159                 return -1;
1160         }
1161
1162         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1163
1164         fd = mkstemp(tmpfile);
1165         if (fd < 0) {
1166                 fprintf(stderr, "unable to create temporary sha1 filename %s: %s", tmpfile, strerror(errno));
1167                 return -1;
1168         }
1169
1170         /* Set it up */
1171         memset(&stream, 0, sizeof(stream));
1172         deflateInit(&stream, Z_BEST_COMPRESSION);
1173         size = deflateBound(&stream, len+hdrlen);
1174         compressed = xmalloc(size);
1175
1176         /* Compress it */
1177         stream.next_out = compressed;
1178         stream.avail_out = size;
1179
1180         /* First header.. */
1181         stream.next_in = hdr;
1182         stream.avail_in = hdrlen;
1183         while (deflate(&stream, 0) == Z_OK)
1184                 /* nothing */;
1185
1186         /* Then the data itself.. */
1187         stream.next_in = buf;
1188         stream.avail_in = len;
1189         while (deflate(&stream, Z_FINISH) == Z_OK)
1190                 /* nothing */;
1191         deflateEnd(&stream);
1192         size = stream.total_out;
1193
1194         if (write(fd, compressed, size) != size)
1195                 die("unable to write file");
1196         fchmod(fd, 0444);
1197         close(fd);
1198         free(compressed);
1199
1200         ret = link(tmpfile, filename);
1201         if (ret < 0) {
1202                 ret = errno;
1203
1204                 /*
1205                  * Coda hack - coda doesn't like cross-directory links,
1206                  * so we fall back to a rename, which will mean that it
1207                  * won't be able to check collisions, but that's not a
1208                  * big deal.
1209                  *
1210                  * When this succeeds, we just return 0. We have nothing
1211                  * left to unlink.
1212                  */
1213                 if (ret == EXDEV && !rename(tmpfile, filename))
1214                         return 0;
1215         }
1216         unlink(tmpfile);
1217         if (ret) {
1218                 if (ret != EEXIST) {
1219                         fprintf(stderr, "unable to write sha1 filename %s: %s", filename, strerror(ret));
1220                         return -1;
1221                 }
1222                 /* FIXME!!! Collision check here ? */
1223         }
1224
1225         return 0;
1226 }
1227
1228 int write_sha1_to_fd(int fd, const unsigned char *sha1)
1229 {
1230         ssize_t size;
1231         unsigned long objsize;
1232         int posn = 0;
1233         void *buf = map_sha1_file_internal(sha1, &objsize);
1234         z_stream stream;
1235         if (!buf) {
1236                 unsigned char *unpacked;
1237                 unsigned long len;
1238                 char type[20];
1239                 char hdr[50];
1240                 int hdrlen;
1241                 // need to unpack and recompress it by itself
1242                 unpacked = read_packed_sha1(sha1, type, &len);
1243
1244                 hdrlen = sprintf(hdr, "%s %lu", type, len) + 1;
1245
1246                 /* Set it up */
1247                 memset(&stream, 0, sizeof(stream));
1248                 deflateInit(&stream, Z_BEST_COMPRESSION);
1249                 size = deflateBound(&stream, len + hdrlen);
1250                 buf = xmalloc(size);
1251
1252                 /* Compress it */
1253                 stream.next_out = buf;
1254                 stream.avail_out = size;
1255
1256                 /* First header.. */
1257                 stream.next_in = (void *)hdr;
1258                 stream.avail_in = hdrlen;
1259                 while (deflate(&stream, 0) == Z_OK)
1260                         /* nothing */;
1261
1262                 /* Then the data itself.. */
1263                 stream.next_in = unpacked;
1264                 stream.avail_in = len;
1265                 while (deflate(&stream, Z_FINISH) == Z_OK)
1266                         /* nothing */;
1267                 deflateEnd(&stream);
1268
1269                 objsize = stream.total_out;
1270         }
1271
1272         do {
1273                 size = write(fd, buf + posn, objsize - posn);
1274                 if (size <= 0) {
1275                         if (!size) {
1276                                 fprintf(stderr, "write closed");
1277                         } else {
1278                                 perror("write ");
1279                         }
1280                         return -1;
1281                 }
1282                 posn += size;
1283         } while (posn < objsize);
1284         return 0;
1285 }
1286
1287 int write_sha1_from_fd(const unsigned char *sha1, int fd)
1288 {
1289         char *filename = sha1_file_name(sha1);
1290
1291         int local;
1292         z_stream stream;
1293         unsigned char real_sha1[20];
1294         unsigned char buf[4096];
1295         unsigned char discard[4096];
1296         int ret;
1297         SHA_CTX c;
1298
1299         local = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
1300
1301         if (local < 0)
1302                 return error("Couldn't open %s\n", filename);
1303
1304         memset(&stream, 0, sizeof(stream));
1305
1306         inflateInit(&stream);
1307
1308         SHA1_Init(&c);
1309
1310         do {
1311                 ssize_t size;
1312                 size = read(fd, buf, 4096);
1313                 if (size <= 0) {
1314                         close(local);
1315                         unlink(filename);
1316                         if (!size)
1317                                 return error("Connection closed?");
1318                         perror("Reading from connection");
1319                         return -1;
1320                 }
1321                 write(local, buf, size);
1322                 stream.avail_in = size;
1323                 stream.next_in = buf;
1324                 do {
1325                         stream.next_out = discard;
1326                         stream.avail_out = sizeof(discard);
1327                         ret = inflate(&stream, Z_SYNC_FLUSH);
1328                         SHA1_Update(&c, discard, sizeof(discard) -
1329                                     stream.avail_out);
1330                 } while (stream.avail_in && ret == Z_OK);
1331
1332         } while (ret == Z_OK);
1333         inflateEnd(&stream);
1334
1335         close(local);
1336         SHA1_Final(real_sha1, &c);
1337         if (ret != Z_STREAM_END) {
1338                 unlink(filename);
1339                 return error("File %s corrupted", sha1_to_hex(sha1));
1340         }
1341         if (memcmp(sha1, real_sha1, 20)) {
1342                 unlink(filename);
1343                 return error("File %s has bad hash\n", sha1_to_hex(sha1));
1344         }
1345
1346         return 0;
1347 }
1348
1349 int has_sha1_pack(const unsigned char *sha1)
1350 {
1351         struct pack_entry e;
1352         return find_pack_entry(sha1, &e);
1353 }
1354
1355 int has_sha1_file(const unsigned char *sha1)
1356 {
1357         struct stat st;
1358         struct pack_entry e;
1359
1360         if (find_pack_entry(sha1, &e))
1361                 return 1;
1362         return find_sha1_file(sha1, &st) ? 1 : 0;
1363 }
1364
1365 int index_fd(unsigned char *sha1, int fd, struct stat *st, int write_object, const char *type)
1366 {
1367         unsigned long size = st->st_size;
1368         void *buf;
1369         int ret;
1370         unsigned char hdr[50];
1371         int hdrlen;
1372
1373         buf = "";
1374         if (size)
1375                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1376         close(fd);
1377         if (buf == MAP_FAILED)
1378                 return -1;
1379
1380         if (!type)
1381                 type = "blob";
1382         if (write_object)
1383                 ret = write_sha1_file(buf, size, type, sha1);
1384         else {
1385                 write_sha1_file_prepare(buf, size, type, sha1, hdr, &hdrlen);
1386                 ret = 0;
1387         }
1388         if (size)
1389                 munmap(buf, size);
1390         return ret;
1391 }