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 "cache.h"
  10 #include "delta.h"
  11 #include "pack.h"
  12 #include "blob.h"
  13 #include "commit.h"
  14 #include "tag.h"
  15 #include "tree.h"
  16
  17 #ifndef O_NOATIME
  18 #if defined(__linux__) && (defined(__i386__) || defined(__PPC__))
  19 #define O_NOATIME 01000000
  20 #else
  21 #define O_NOATIME 0
  22 #endif
  23 #endif
  24
  25 const unsigned char null_sha1[20];
  26
  27 static unsigned int sha1_file_open_flag = O_NOATIME;
  28
  29 static unsigned hexval(char c)
  30 {
  31         if (c >= '0' && c <= '9')
  32                 return c - '0';
  33         if (c >= 'a' && c <= 'f')
  34                 return c - 'a' + 10;
  35         if (c >= 'A' && c <= 'F')
  36                 return c - 'A' + 10;
  37         return ~0;
  38 }
  39
  40 int get_sha1_hex(const char *hex, unsigned char *sha1)
  41 {
  42         int i;
  43         for (i = 0; i < 20; i++) {
  44                 unsigned int val = (hexval(hex[0]) << 4) | hexval(hex[1]);
  45                 if (val & ~0xff)
  46                         return -1;
  47                 *sha1++ = val;
  48                 hex += 2;
  49         }
  50         return 0;
  51 }
  52
  53 int safe_create_leading_directories(char *path)
  54 {
  55         char *pos = path;
  56         struct stat st;
  57
  58         if (*pos == '/')
  59                 pos++;
  60
  61         while (pos) {
  62                 pos = strchr(pos, '/');
  63                 if (!pos)
  64                         break;
  65                 *pos = 0;
  66                 if (!stat(path, &st)) {
  67                         /* path exists */
  68                         if (!S_ISDIR(st.st_mode)) {
  69                                 *pos = '/';
  70                                 return -3;
  71                         }
  72                 }
  73                 else if (mkdir(path, 0777)) {
  74                         *pos = '/';
  75                         return -1;
  76                 }
  77                 else if (adjust_shared_perm(path)) {
  78                         *pos = '/';
  79                         return -2;
  80                 }
  81                 *pos++ = '/';
  82         }
  83         return 0;
  84 }
  85
  86 char * sha1_to_hex(const unsigned char *sha1)
  87 {
  88         static int bufno;
  89         static char hexbuffer[4][50];
  90         static const char hex[] = "0123456789abcdef";
  91         char *buffer = hexbuffer[3 & ++bufno], *buf = buffer;
  92         int i;
  93
  94         for (i = 0; i < 20; i++) {
  95                 unsigned int val = *sha1++;
  96                 *buf++ = hex[val >> 4];
  97                 *buf++ = hex[val & 0xf];
  98         }
  99         *buf = '\0';
 100
 101         return buffer;
 102 }
 103
 104 static void fill_sha1_path(char *pathbuf, const unsigned char *sha1)
 105 {
 106         int i;
 107         for (i = 0; i < 20; i++) {
 108                 static char hex[] = "0123456789abcdef";
 109                 unsigned int val = sha1[i];
 110                 char *pos = pathbuf + i*2 + (i > 0);
 111                 *pos++ = hex[val >> 4];
 112                 *pos = hex[val & 0xf];
 113         }
 114 }
 115
 116 /*
 117  * NOTE! This returns a statically allocated buffer, so you have to be
 118  * careful about using it. Do a "xstrdup()" if you need to save the
 119  * filename.
 120  *
 121  * Also note that this returns the location for creating.  Reading
 122  * SHA1 file can happen from any alternate directory listed in the
 123  * DB_ENVIRONMENT environment variable if it is not found in
 124  * the primary object database.
 125  */
 126 char *sha1_file_name(const unsigned char *sha1)
 127 {
 128         static char *name, *base;
 129
 130         if (!base) {
 131                 const char *sha1_file_directory = get_object_directory();
 132                 int len = strlen(sha1_file_directory);
 133                 base = xmalloc(len + 60);
 134                 memcpy(base, sha1_file_directory, len);
 135                 memset(base+len, 0, 60);
 136                 base[len] = '/';
 137                 base[len+3] = '/';
 138                 name = base + len + 1;
 139         }
 140         fill_sha1_path(name, sha1);
 141         return base;
 142 }
 143
 144 char *sha1_pack_name(const unsigned char *sha1)
 145 {
 146         static const char hex[] = "0123456789abcdef";
 147         static char *name, *base, *buf;
 148         int i;
 149
 150         if (!base) {
 151                 const char *sha1_file_directory = get_object_directory();
 152                 int len = strlen(sha1_file_directory);
 153                 base = xmalloc(len + 60);
 154                 sprintf(base, "%s/pack/pack-1234567890123456789012345678901234567890.pack", sha1_file_directory);
 155                 name = base + len + 11;
 156         }
 157
 158         buf = name;
 159
 160         for (i = 0; i < 20; i++) {
 161                 unsigned int val = *sha1++;
 162                 *buf++ = hex[val >> 4];
 163                 *buf++ = hex[val & 0xf];
 164         }
 165
 166         return base;
 167 }
 168
 169 char *sha1_pack_index_name(const unsigned char *sha1)
 170 {
 171         static const char hex[] = "0123456789abcdef";
 172         static char *name, *base, *buf;
 173         int i;
 174
 175         if (!base) {
 176                 const char *sha1_file_directory = get_object_directory();
 177                 int len = strlen(sha1_file_directory);
 178                 base = xmalloc(len + 60);
 179                 sprintf(base, "%s/pack/pack-1234567890123456789012345678901234567890.idx", sha1_file_directory);
 180                 name = base + len + 11;
 181         }
 182
 183         buf = name;
 184
 185         for (i = 0; i < 20; i++) {
 186                 unsigned int val = *sha1++;
 187                 *buf++ = hex[val >> 4];
 188                 *buf++ = hex[val & 0xf];
 189         }
 190
 191         return base;
 192 }
 193
 194 struct alternate_object_database *alt_odb_list;
 195 static struct alternate_object_database **alt_odb_tail;
 196
 197 static void read_info_alternates(const char * alternates, int depth);
 198
 199 /*
 200  * Prepare alternate object database registry.
 201  *
 202  * The variable alt_odb_list points at the list of struct
 203  * alternate_object_database.  The elements on this list come from
 204  * non-empty elements from colon separated ALTERNATE_DB_ENVIRONMENT
 205  * environment variable, and $GIT_OBJECT_DIRECTORY/info/alternates,
 206  * whose contents is similar to that environment variable but can be
 207  * LF separated.  Its base points at a statically allocated buffer that
 208  * contains "/the/directory/corresponding/to/.git/objects/...", while
 209  * its name points just after the slash at the end of ".git/objects/"
 210  * in the example above, and has enough space to hold 40-byte hex
 211  * SHA1, an extra slash for the first level indirection, and the
 212  * terminating NUL.
 213  */
 214 static int link_alt_odb_entry(const char * entry, int len, const char * relative_base, int depth)
 215 {
 216         struct stat st;
 217         const char *objdir = get_object_directory();
 218         struct alternate_object_database *ent;
 219         struct alternate_object_database *alt;
 220         /* 43 = 40-byte + 2 '/' + terminating NUL */
 221         int pfxlen = len;
 222         int entlen = pfxlen + 43;
 223         int base_len = -1;
 224
 225         if (*entry != '/' && relative_base) {
 226                 /* Relative alt-odb */
 227                 if (base_len < 0)
 228                         base_len = strlen(relative_base) + 1;
 229                 entlen += base_len;
 230                 pfxlen += base_len;
 231         }
 232         ent = xmalloc(sizeof(*ent) + entlen);
 233
 234         if (*entry != '/' && relative_base) {
 235                 memcpy(ent->base, relative_base, base_len - 1);
 236                 ent->base[base_len - 1] = '/';
 237                 memcpy(ent->base + base_len, entry, len);
 238         }
 239         else
 240                 memcpy(ent->base, entry, pfxlen);
 241
 242         ent->name = ent->base + pfxlen + 1;
 243         ent->base[pfxlen + 3] = '/';
 244         ent->base[pfxlen] = ent->base[entlen-1] = 0;
 245
 246         /* Detect cases where alternate disappeared */
 247         if (stat(ent->base, &st) || !S_ISDIR(st.st_mode)) {
 248                 error("object directory %s does not exist; "
 249                       "check .git/objects/info/alternates.",
 250                       ent->base);
 251                 free(ent);
 252                 return -1;
 253         }
 254
 255         /* Prevent the common mistake of listing the same
 256          * thing twice, or object directory itself.
 257          */
 258         for (alt = alt_odb_list; alt; alt = alt->next) {
 259                 if (!memcmp(ent->base, alt->base, pfxlen)) {
 260                         free(ent);
 261                         return -1;
 262                 }
 263         }
 264         if (!memcmp(ent->base, objdir, pfxlen)) {
 265                 free(ent);
 266                 return -1;
 267         }
 268
 269         /* add the alternate entry */
 270         *alt_odb_tail = ent;
 271         alt_odb_tail = &(ent->next);
 272         ent->next = NULL;
 273
 274         /* recursively add alternates */
 275         read_info_alternates(ent->base, depth + 1);
 276
 277         ent->base[pfxlen] = '/';
 278
 279         return 0;
 280 }
 281
 282 static void link_alt_odb_entries(const char *alt, const char *ep, int sep,
 283                                  const char *relative_base, int depth)
 284 {
 285         const char *cp, *last;
 286
 287         if (depth > 5) {
 288                 error("%s: ignoring alternate object stores, nesting too deep.",
 289                                 relative_base);
 290                 return;
 291         }
 292
 293         last = alt;
 294         while (last < ep) {
 295                 cp = last;
 296                 if (cp < ep && *cp == '#') {
 297                         while (cp < ep && *cp != sep)
 298                                 cp++;
 299                         last = cp + 1;
 300                         continue;
 301                 }
 302                 while (cp < ep && *cp != sep)
 303                         cp++;
 304                 if (last != cp) {
 305                         if ((*last != '/') && depth) {
 306                                 error("%s: ignoring relative alternate object store %s",
 307                                                 relative_base, last);
 308                         } else {
 309                                 link_alt_odb_entry(last, cp - last,
 310                                                 relative_base, depth);
 311                         }
 312                 }
 313                 while (cp < ep && *cp == sep)
 314                         cp++;
 315                 last = cp;
 316         }
 317 }
 318
 319 static void read_info_alternates(const char * relative_base, int depth)
 320 {
 321         char *map;
 322         struct stat st;
 323         char path[PATH_MAX];
 324         int fd;
 325
 326         sprintf(path, "%s/info/alternates", relative_base);
 327         fd = open(path, O_RDONLY);
 328         if (fd < 0)
 329                 return;
 330         if (fstat(fd, &st) || (st.st_size == 0)) {
 331                 close(fd);
 332                 return;
 333         }
 334         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 335         close(fd);
 336         if (map == MAP_FAILED)
 337                 return;
 338
 339         link_alt_odb_entries(map, map + st.st_size, '\n', relative_base, depth);
 340
 341         munmap(map, st.st_size);
 342 }
 343
 344 void prepare_alt_odb(void)
 345 {
 346         const char *alt;
 347
 348         alt = getenv(ALTERNATE_DB_ENVIRONMENT);
 349         if (!alt) alt = "";
 350
 351         if (alt_odb_tail)
 352                 return;
 353         alt_odb_tail = &alt_odb_list;
 354         link_alt_odb_entries(alt, alt + strlen(alt), ':', NULL, 0);
 355
 356         read_info_alternates(get_object_directory(), 0);
 357 }
 358
 359 static char *find_sha1_file(const unsigned char *sha1, struct stat *st)
 360 {
 361         char *name = sha1_file_name(sha1);
 362         struct alternate_object_database *alt;
 363
 364         if (!stat(name, st))
 365                 return name;
 366         prepare_alt_odb();
 367         for (alt = alt_odb_list; alt; alt = alt->next) {
 368                 name = alt->name;
 369                 fill_sha1_path(name, sha1);
 370                 if (!stat(alt->base, st))
 371                         return alt->base;
 372         }
 373         return NULL;
 374 }
 375
 376 #define PACK_MAX_SZ (1<<26)
 377 static int pack_used_ctr;
 378 static unsigned long pack_mapped;
 379 struct packed_git *packed_git;
 380
 381 static int check_packed_git_idx(const char *path, unsigned long *idx_size_,
 382                                 void **idx_map_)
 383 {
 384         void *idx_map;
 385         unsigned int *index;
 386         unsigned long idx_size;
 387         int nr, i;
 388         int fd = open(path, O_RDONLY);
 389         struct stat st;
 390         if (fd < 0)
 391                 return -1;
 392         if (fstat(fd, &st)) {
 393                 close(fd);
 394                 return -1;
 395         }
 396         idx_size = st.st_size;
 397         idx_map = mmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0);
 398         close(fd);
 399         if (idx_map == MAP_FAILED)
 400                 return -1;
 401
 402         index = idx_map;
 403         *idx_map_ = idx_map;
 404         *idx_size_ = idx_size;
 405
 406         /* check index map */
 407         if (idx_size < 4*256 + 20 + 20)
 408                 return error("index file too small");
 409         nr = 0;
 410         for (i = 0; i < 256; i++) {
 411                 unsigned int n = ntohl(index[i]);
 412                 if (n < nr)
 413                         return error("non-monotonic index");
 414                 nr = n;
 415         }
 416
 417         /*
 418          * Total size:
 419          *  - 256 index entries 4 bytes each
 420          *  - 24-byte entries * nr (20-byte sha1 + 4-byte offset)
 421          *  - 20-byte SHA1 of the packfile
 422          *  - 20-byte SHA1 file checksum
 423          */
 424         if (idx_size != 4*256 + nr * 24 + 20 + 20)
 425                 return error("wrong index file size");
 426
 427         return 0;
 428 }
 429
 430 static int unuse_one_packed_git(void)
 431 {
 432         struct packed_git *p, *lru = NULL;
 433
 434         for (p = packed_git; p; p = p->next) {
 435                 if (p->pack_use_cnt || !p->pack_base)
 436                         continue;
 437                 if (!lru || p->pack_last_used < lru->pack_last_used)
 438                         lru = p;
 439         }
 440         if (!lru)
 441                 return 0;
 442         munmap(lru->pack_base, lru->pack_size);
 443         lru->pack_base = NULL;
 444         return 1;
 445 }
 446
 447 void unuse_packed_git(struct packed_git *p)
 448 {
 449         p->pack_use_cnt--;
 450 }
 451
 452 int use_packed_git(struct packed_git *p)
 453 {
 454         if (!p->pack_size) {
 455                 struct stat st;
 456                 /* We created the struct before we had the pack */
 457                 stat(p->pack_name, &st);
 458                 if (!S_ISREG(st.st_mode))
 459                         die("packfile %s not a regular file", p->pack_name);
 460                 p->pack_size = st.st_size;
 461         }
 462         if (!p->pack_base) {
 463                 int fd;
 464                 struct stat st;
 465                 void *map;
 466                 struct pack_header *hdr;
 467
 468                 pack_mapped += p->pack_size;
 469                 while (PACK_MAX_SZ < pack_mapped && unuse_one_packed_git())
 470                         ; /* nothing */
 471                 fd = open(p->pack_name, O_RDONLY);
 472                 if (fd < 0)
 473                         die("packfile %s cannot be opened", p->pack_name);
 474                 if (fstat(fd, &st)) {
 475                         close(fd);
 476                         die("packfile %s cannot be opened", p->pack_name);
 477                 }
 478                 if (st.st_size != p->pack_size)
 479                         die("packfile %s size mismatch.", p->pack_name);
 480                 map = mmap(NULL, p->pack_size, PROT_READ, MAP_PRIVATE, fd, 0);
 481                 close(fd);
 482                 if (map == MAP_FAILED)
 483                         die("packfile %s cannot be mapped.", p->pack_name);
 484                 p->pack_base = map;
 485
 486                 /* Check if we understand this pack file.  If we don't we're
 487                  * likely too old to handle it.
 488                  */
 489                 hdr = map;
 490                 if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 491                         die("packfile %s isn't actually a pack.", p->pack_name);
 492                 if (!pack_version_ok(hdr->hdr_version))
 493                         die("packfile %s is version %i and not supported"
 494                                 " (try upgrading GIT to a newer version)",
 495                                 p->pack_name, ntohl(hdr->hdr_version));
 496
 497                 /* Check if the pack file matches with the index file.
 498                  * this is cheap.
 499                  */
 500                 if (hashcmp((unsigned char *)(p->index_base) +
 501                             p->index_size - 40,
 502                             (unsigned char *)p->pack_base +
 503                             p->pack_size - 20)) {
 504                         die("packfile %s does not match index.", p->pack_name);
 505                 }
 506         }
 507         p->pack_last_used = pack_used_ctr++;
 508         p->pack_use_cnt++;
 509         return 0;
 510 }
 511
 512 struct packed_git *add_packed_git(char *path, int path_len, int local)
 513 {
 514         struct stat st;
 515         struct packed_git *p;
 516         unsigned long idx_size;
 517         void *idx_map;
 518         unsigned char sha1[20];
 519
 520         if (check_packed_git_idx(path, &idx_size, &idx_map))
 521                 return NULL;
 522
 523         /* do we have a corresponding .pack file? */
 524         strcpy(path + path_len - 4, ".pack");
 525         if (stat(path, &st) || !S_ISREG(st.st_mode)) {
 526                 munmap(idx_map, idx_size);
 527                 return NULL;
 528         }
 529         /* ok, it looks sane as far as we can check without
 530          * actually mapping the pack file.
 531          */
 532         p = xmalloc(sizeof(*p) + path_len + 2);
 533         strcpy(p->pack_name, path);
 534         p->index_size = idx_size;
 535         p->pack_size = st.st_size;
 536         p->index_base = idx_map;
 537         p->next = NULL;
 538         p->pack_base = NULL;
 539         p->pack_last_used = 0;
 540         p->pack_use_cnt = 0;
 541         p->pack_local = local;
 542         if ((path_len > 44) && !get_sha1_hex(path + path_len - 44, sha1))
 543                 hashcpy(p->sha1, sha1);
 544         return p;
 545 }
 546
 547 struct packed_git *parse_pack_index(unsigned char *sha1)
 548 {
 549         char *path = sha1_pack_index_name(sha1);
 550         return parse_pack_index_file(sha1, path);
 551 }
 552
 553 struct packed_git *parse_pack_index_file(const unsigned char *sha1, char *idx_path)
 554 {
 555         struct packed_git *p;
 556         unsigned long idx_size;
 557         void *idx_map;
 558         char *path;
 559
 560         if (check_packed_git_idx(idx_path, &idx_size, &idx_map))
 561                 return NULL;
 562
 563         path = sha1_pack_name(sha1);
 564
 565         p = xmalloc(sizeof(*p) + strlen(path) + 2);
 566         strcpy(p->pack_name, path);
 567         p->index_size = idx_size;
 568         p->pack_size = 0;
 569         p->index_base = idx_map;
 570         p->next = NULL;
 571         p->pack_base = NULL;
 572         p->pack_last_used = 0;
 573         p->pack_use_cnt = 0;
 574         hashcpy(p->sha1, sha1);
 575         return p;
 576 }
 577
 578 void install_packed_git(struct packed_git *pack)
 579 {
 580         pack->next = packed_git;
 581         packed_git = pack;
 582 }
 583
 584 static void prepare_packed_git_one(char *objdir, int local)
 585 {
 586         char path[PATH_MAX];
 587         int len;
 588         DIR *dir;
 589         struct dirent *de;
 590
 591         sprintf(path, "%s/pack", objdir);
 592         len = strlen(path);
 593         dir = opendir(path);
 594         if (!dir) {
 595                 if (errno != ENOENT)
 596                         error("unable to open object pack directory: %s: %s",
 597                               path, strerror(errno));
 598                 return;
 599         }
 600         path[len++] = '/';
 601         while ((de = readdir(dir)) != NULL) {
 602                 int namelen = strlen(de->d_name);
 603                 struct packed_git *p;
 604
 605                 if (!has_extension(de->d_name, ".idx"))
 606                         continue;
 607
 608                 /* we have .idx.  Is it a file we can map? */
 609                 strcpy(path + len, de->d_name);
 610                 for (p = packed_git; p; p = p->next) {
 611                         if (!memcmp(path, p->pack_name, len + namelen - 4))
 612                                 break;
 613                 }
 614                 if (p)
 615                         continue;
 616                 p = add_packed_git(path, len + namelen, local);
 617                 if (!p)
 618                         continue;
 619                 p->next = packed_git;
 620                 packed_git = p;
 621         }
 622         closedir(dir);
 623 }
 624
 625 static int prepare_packed_git_run_once = 0;
 626 void prepare_packed_git(void)
 627 {
 628         struct alternate_object_database *alt;
 629
 630         if (prepare_packed_git_run_once)
 631                 return;
 632         prepare_packed_git_one(get_object_directory(), 1);
 633         prepare_alt_odb();
 634         for (alt = alt_odb_list; alt; alt = alt->next) {
 635                 alt->name[-1] = 0;
 636                 prepare_packed_git_one(alt->base, 0);
 637                 alt->name[-1] = '/';
 638         }
 639         prepare_packed_git_run_once = 1;
 640 }
 641
 642 static void reprepare_packed_git(void)
 643 {
 644         prepare_packed_git_run_once = 0;
 645         prepare_packed_git();
 646 }
 647
 648 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 649 {
 650         char header[100];
 651         unsigned char real_sha1[20];
 652         SHA_CTX c;
 653
 654         SHA1_Init(&c);
 655         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 656         SHA1_Update(&c, map, size);
 657         SHA1_Final(real_sha1, &c);
 658         return hashcmp(sha1, real_sha1) ? -1 : 0;
 659 }
 660
 661 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 662 {
 663         struct stat st;
 664         void *map;
 665         int fd;
 666         char *filename = find_sha1_file(sha1, &st);
 667
 668         if (!filename) {
 669                 return NULL;
 670         }
 671
 672         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 673         if (fd < 0) {
 674                 /* See if it works without O_NOATIME */
 675                 switch (sha1_file_open_flag) {
 676                 default:
 677                         fd = open(filename, O_RDONLY);
 678                         if (fd >= 0)
 679                                 break;
 680                 /* Fallthrough */
 681                 case 0:
 682                         return NULL;
 683                 }
 684
 685                 /* If it failed once, it will probably fail again.
 686                  * Stop using O_NOATIME
 687                  */
 688                 sha1_file_open_flag = 0;
 689         }
 690         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 691         close(fd);
 692         if (map == MAP_FAILED)
 693                 return NULL;
 694         *size = st.st_size;
 695         return map;
 696 }
 697
 698 int legacy_loose_object(unsigned char *map)
 699 {
 700         unsigned int word;
 701
 702         /*
 703          * Is it a zlib-compressed buffer? If so, the first byte
 704          * must be 0x78 (15-bit window size, deflated), and the
 705          * first 16-bit word is evenly divisible by 31
 706          */
 707         word = (map[0] << 8) + map[1];
 708         if (map[0] == 0x78 && !(word % 31))
 709                 return 1;
 710         else
 711                 return 0;
 712 }
 713
 714 static unsigned long unpack_object_header_gently(const unsigned char *buf, unsigned long len, enum object_type *type, unsigned long *sizep)
 715 {
 716         unsigned shift;
 717         unsigned char c;
 718         unsigned long size;
 719         unsigned long used = 0;
 720
 721         c = buf[used++];
 722         *type = (c >> 4) & 7;
 723         size = c & 15;
 724         shift = 4;
 725         while (c & 0x80) {
 726                 if (len <= used)
 727                         return 0;
 728                 if (sizeof(long) * 8 <= shift)
 729                         return 0;
 730                 c = buf[used++];
 731                 size += (c & 0x7f) << shift;
 732                 shift += 7;
 733         }
 734         *sizep = size;
 735         return used;
 736 }
 737
 738 static int unpack_sha1_header(z_stream *stream, unsigned char *map, unsigned long mapsize, void *buffer, unsigned long bufsiz)
 739 {
 740         unsigned long size, used;
 741         static const char valid_loose_object_type[8] = {
 742                 0, /* OBJ_EXT */
 743                 1, 1, 1, 1, /* "commit", "tree", "blob", "tag" */
 744                 0, /* "delta" and others are invalid in a loose object */
 745         };
 746         enum object_type type;
 747
 748         /* Get the data stream */
 749         memset(stream, 0, sizeof(*stream));
 750         stream->next_in = map;
 751         stream->avail_in = mapsize;
 752         stream->next_out = buffer;
 753         stream->avail_out = bufsiz;
 754
 755         if (legacy_loose_object(map)) {
 756                 inflateInit(stream);
 757                 return inflate(stream, 0);
 758         }
 759
 760         used = unpack_object_header_gently(map, mapsize, &type, &size);
 761         if (!used || !valid_loose_object_type[type])
 762                 return -1;
 763         map += used;
 764         mapsize -= used;
 765
 766         /* Set up the stream for the rest.. */
 767         stream->next_in = map;
 768         stream->avail_in = mapsize;
 769         inflateInit(stream);
 770
 771         /* And generate the fake traditional header */
 772         stream->total_out = 1 + snprintf(buffer, bufsiz, "%s %lu",
 773                                          type_names[type], size);
 774         return 0;
 775 }
 776
 777 static void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 778 {
 779         int bytes = strlen(buffer) + 1;
 780         unsigned char *buf = xmalloc(1+size);
 781         unsigned long n;
 782
 783         n = stream->total_out - bytes;
 784         if (n > size)
 785                 n = size;
 786         memcpy(buf, (char *) buffer + bytes, n);
 787         bytes = n;
 788         if (bytes < size) {
 789                 stream->next_out = buf + bytes;
 790                 stream->avail_out = size - bytes;
 791                 while (inflate(stream, Z_FINISH) == Z_OK)
 792                         /* nothing */;
 793         }
 794         buf[size] = 0;
 795         inflateEnd(stream);
 796         return buf;
 797 }
 798
 799 /*
 800  * We used to just use "sscanf()", but that's actually way
 801  * too permissive for what we want to check. So do an anal
 802  * object header parse by hand.
 803  */
 804 static int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 805 {
 806         int i;
 807         unsigned long size;
 808
 809         /*
 810          * The type can be at most ten bytes (including the
 811          * terminating '\0' that we add), and is followed by
 812          * a space.
 813          */
 814         i = 10;
 815         for (;;) {
 816                 char c = *hdr++;
 817                 if (c == ' ')
 818                         break;
 819                 if (!--i)
 820                         return -1;
 821                 *type++ = c;
 822         }
 823         *type = 0;
 824
 825         /*
 826          * The length must follow immediately, and be in canonical
 827          * decimal format (ie "010" is not valid).
 828          */
 829         size = *hdr++ - '0';
 830         if (size > 9)
 831                 return -1;
 832         if (size) {
 833                 for (;;) {
 834                         unsigned long c = *hdr - '0';
 835                         if (c > 9)
 836                                 break;
 837                         hdr++;
 838                         size = size * 10 + c;
 839                 }
 840         }
 841         *sizep = size;
 842
 843         /*
 844          * The length must be followed by a zero byte
 845          */
 846         return *hdr ? -1 : 0;
 847 }
 848
 849 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 850 {
 851         int ret;
 852         z_stream stream;
 853         char hdr[8192];
 854
 855         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
 856         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
 857                 return NULL;
 858
 859         return unpack_sha1_rest(&stream, hdr, *size);
 860 }
 861
 862 /* forward declaration for a mutually recursive function */
 863 static int packed_object_info(struct pack_entry *entry,
 864                               char *type, unsigned long *sizep);
 865
 866 static int packed_delta_info(unsigned char *base_sha1,
 867                              unsigned long delta_size,
 868                              unsigned long left,
 869                              char *type,
 870                              unsigned long *sizep,
 871                              struct packed_git *p)
 872 {
 873         struct pack_entry base_ent;
 874
 875         if (left < 20)
 876                 die("truncated pack file");
 877
 878         /* The base entry _must_ be in the same pack */
 879         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 880                 die("failed to find delta-pack base object %s",
 881                     sha1_to_hex(base_sha1));
 882
 883         /* We choose to only get the type of the base object and
 884          * ignore potentially corrupt pack file that expects the delta
 885          * based on a base with a wrong size.  This saves tons of
 886          * inflate() calls.
 887          */
 888
 889         if (packed_object_info(&base_ent, type, NULL))
 890                 die("cannot get info for delta-pack base");
 891
 892         if (sizep) {
 893                 const unsigned char *data;
 894                 unsigned char delta_head[64];
 895                 unsigned long result_size;
 896                 z_stream stream;
 897                 int st;
 898
 899                 memset(&stream, 0, sizeof(stream));
 900
 901                 data = stream.next_in = base_sha1 + 20;
 902                 stream.avail_in = left - 20;
 903                 stream.next_out = delta_head;
 904                 stream.avail_out = sizeof(delta_head);
 905
 906                 inflateInit(&stream);
 907                 st = inflate(&stream, Z_FINISH);
 908                 inflateEnd(&stream);
 909                 if ((st != Z_STREAM_END) &&
 910                     stream.total_out != sizeof(delta_head))
 911                         die("delta data unpack-initial failed");
 912
 913                 /* Examine the initial part of the delta to figure out
 914                  * the result size.
 915                  */
 916                 data = delta_head;
 917
 918                 /* ignore base size */
 919                 get_delta_hdr_size(&data, delta_head+sizeof(delta_head));
 920
 921                 /* Read the result size */
 922                 result_size = get_delta_hdr_size(&data, delta_head+sizeof(delta_head));
 923                 *sizep = result_size;
 924         }
 925         return 0;
 926 }
 927
 928 static unsigned long unpack_object_header(struct packed_git *p, unsigned long offset,
 929         enum object_type *type, unsigned long *sizep)
 930 {
 931         unsigned long used;
 932
 933         if (p->pack_size <= offset)
 934                 die("object offset outside of pack file");
 935
 936         used = unpack_object_header_gently((unsigned char *)p->pack_base +
 937                                            offset,
 938                                            p->pack_size - offset, type, sizep);
 939         if (!used)
 940                 die("object offset outside of pack file");
 941
 942         return offset + used;
 943 }
 944
 945 int check_reuse_pack_delta(struct packed_git *p, unsigned long offset,
 946                            unsigned char *base, unsigned long *sizep,
 947                            enum object_type *kindp)
 948 {
 949         unsigned long ptr;
 950         int status = -1;
 951
 952         use_packed_git(p);
 953         ptr = offset;
 954         ptr = unpack_object_header(p, ptr, kindp, sizep);
 955         if (*kindp != OBJ_DELTA)
 956                 goto done;
 957         hashcpy(base, (unsigned char *) p->pack_base + ptr);
 958         status = 0;
 959  done:
 960         unuse_packed_git(p);
 961         return status;
 962 }
 963
 964 void packed_object_info_detail(struct pack_entry *e,
 965                                char *type,
 966                                unsigned long *size,
 967                                unsigned long *store_size,
 968                                unsigned int *delta_chain_length,
 969                                unsigned char *base_sha1)
 970 {
 971         struct packed_git *p = e->p;
 972         unsigned long offset;
 973         unsigned char *pack;
 974         enum object_type kind;
 975
 976         offset = unpack_object_header(p, e->offset, &kind, size);
 977         pack = (unsigned char *) p->pack_base + offset;
 978         if (kind != OBJ_DELTA)
 979                 *delta_chain_length = 0;
 980         else {
 981                 unsigned int chain_length = 0;
 982                 if (p->pack_size <= offset + 20)
 983                         die("pack file %s records an incomplete delta base",
 984                             p->pack_name);
 985                 hashcpy(base_sha1, pack);
 986                 do {
 987                         struct pack_entry base_ent;
 988                         unsigned long junk;
 989
 990                         find_pack_entry_one(pack, &base_ent, p);
 991                         offset = unpack_object_header(p, base_ent.offset,
 992                                                       &kind, &junk);
 993                         pack = (unsigned char *) p->pack_base + offset;
 994                         chain_length++;
 995                 } while (kind == OBJ_DELTA);
 996                 *delta_chain_length = chain_length;
 997         }
 998         switch (kind) {
 999         case OBJ_COMMIT:
1000         case OBJ_TREE:
1001         case OBJ_BLOB:
1002         case OBJ_TAG:
1003                 strcpy(type, type_names[kind]);
1004                 break;
1005         default:
1006                 die("corrupted pack file %s containing object of kind %d",
1007                     p->pack_name, kind);
1008         }
1009         *store_size = 0; /* notyet */
1010 }
1011
1012 static int packed_object_info(struct pack_entry *entry,
1013                               char *type, unsigned long *sizep)
1014 {
1015         struct packed_git *p = entry->p;
1016         unsigned long offset, size, left;
1017         unsigned char *pack;
1018         enum object_type kind;
1019         int retval;
1020
1021         if (use_packed_git(p))
1022                 die("cannot map packed file");
1023
1024         offset = unpack_object_header(p, entry->offset, &kind, &size);
1025         pack = (unsigned char *) p->pack_base + offset;
1026         left = p->pack_size - offset;
1027
1028         switch (kind) {
1029         case OBJ_DELTA:
1030                 retval = packed_delta_info(pack, size, left, type, sizep, p);
1031                 unuse_packed_git(p);
1032                 return retval;
1033         case OBJ_COMMIT:
1034         case OBJ_TREE:
1035         case OBJ_BLOB:
1036         case OBJ_TAG:
1037                 strcpy(type, type_names[kind]);
1038                 break;
1039         default:
1040                 die("corrupted pack file %s containing object of kind %d",
1041                     p->pack_name, kind);
1042         }
1043         if (sizep)
1044                 *sizep = size;
1045         unuse_packed_git(p);
1046         return 0;
1047 }
1048
1049 static void *unpack_compressed_entry(struct packed_git *p,
1050                                     unsigned long offset,
1051                                     unsigned long size)
1052 {
1053         int st;
1054         z_stream stream;
1055         unsigned char *buffer;
1056
1057         buffer = xmalloc(size + 1);
1058         buffer[size] = 0;
1059         memset(&stream, 0, sizeof(stream));
1060         stream.next_in = (unsigned char*)p->pack_base + offset;
1061         stream.avail_in = p->pack_size - offset;
1062         stream.next_out = buffer;
1063         stream.avail_out = size;
1064
1065         inflateInit(&stream);
1066         st = inflate(&stream, Z_FINISH);
1067         inflateEnd(&stream);
1068         if ((st != Z_STREAM_END) || stream.total_out != size) {
1069                 free(buffer);
1070                 return NULL;
1071         }
1072
1073         return buffer;
1074 }
1075
1076 static void *unpack_delta_entry(struct packed_git *p,
1077                                 unsigned long offset,
1078                                 unsigned long delta_size,
1079                                 char *type,
1080                                 unsigned long *sizep)
1081 {
1082         struct pack_entry base_ent;
1083         void *delta_data, *result, *base;
1084         unsigned long result_size, base_size;
1085         unsigned char* base_sha1;
1086
1087         if ((offset + 20) >= p->pack_size)
1088                 die("truncated pack file");
1089
1090         /* The base entry _must_ be in the same pack */
1091         base_sha1 = (unsigned char*)p->pack_base + offset;
1092         if (!find_pack_entry_one(base_sha1, &base_ent, p))
1093                 die("failed to find delta-pack base object %s",
1094                     sha1_to_hex(base_sha1));
1095         base = unpack_entry_gently(&base_ent, type, &base_size);
1096         if (!base)
1097                 die("failed to read delta-pack base object %s",
1098                     sha1_to_hex(base_sha1));
1099
1100         delta_data = unpack_compressed_entry(p, offset + 20, delta_size);
1101         result = patch_delta(base, base_size,
1102                              delta_data, delta_size,
1103                              &result_size);
1104         if (!result)
1105                 die("failed to apply delta");
1106         free(delta_data);
1107         free(base);
1108         *sizep = result_size;
1109         return result;
1110 }
1111
1112 static void *unpack_entry(struct pack_entry *entry,
1113                           char *type, unsigned long *sizep)
1114 {
1115         struct packed_git *p = entry->p;
1116         void *retval;
1117
1118         if (use_packed_git(p))
1119                 die("cannot map packed file");
1120         retval = unpack_entry_gently(entry, type, sizep);
1121         unuse_packed_git(p);
1122         if (!retval)
1123                 die("corrupted pack file %s", p->pack_name);
1124         return retval;
1125 }
1126
1127 /* The caller is responsible for use_packed_git()/unuse_packed_git() pair */
1128 void *unpack_entry_gently(struct pack_entry *entry,
1129                           char *type, unsigned long *sizep)
1130 {
1131         struct packed_git *p = entry->p;
1132         unsigned long offset, size;
1133         enum object_type kind;
1134
1135         offset = unpack_object_header(p, entry->offset, &kind, &size);
1136         switch (kind) {
1137         case OBJ_DELTA:
1138                 return unpack_delta_entry(p, offset, size, type, sizep);
1139         case OBJ_COMMIT:
1140         case OBJ_TREE:
1141         case OBJ_BLOB:
1142         case OBJ_TAG:
1143                 strcpy(type, type_names[kind]);
1144                 *sizep = size;
1145                 return unpack_compressed_entry(p, offset, size);
1146         default:
1147                 return NULL;
1148         }
1149 }
1150
1151 int num_packed_objects(const struct packed_git *p)
1152 {
1153         /* See check_packed_git_idx() */
1154         return (p->index_size - 20 - 20 - 4*256) / 24;
1155 }
1156
1157 int nth_packed_object_sha1(const struct packed_git *p, int n,
1158                            unsigned char* sha1)
1159 {
1160         void *index = p->index_base + 256;
1161         if (n < 0 || num_packed_objects(p) <= n)
1162                 return -1;
1163         hashcpy(sha1, (unsigned char *) index + (24 * n) + 4);
1164         return 0;
1165 }
1166
1167 int find_pack_entry_one(const unsigned char *sha1,
1168                         struct pack_entry *e, struct packed_git *p)
1169 {
1170         unsigned int *level1_ofs = p->index_base;
1171         int hi = ntohl(level1_ofs[*sha1]);
1172         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
1173         void *index = p->index_base + 256;
1174
1175         do {
1176                 int mi = (lo + hi) / 2;
1177                 int cmp = hashcmp((unsigned char *)index + (24 * mi) + 4, sha1);
1178                 if (!cmp) {
1179                         e->offset = ntohl(*((unsigned int *) ((char *) index + (24 * mi))));
1180                         hashcpy(e->sha1, sha1);
1181                         e->p = p;
1182                         return 1;
1183                 }
1184                 if (cmp > 0)
1185                         hi = mi;
1186                 else
1187                         lo = mi+1;
1188         } while (lo < hi);
1189         return 0;
1190 }
1191
1192 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e)
1193 {
1194         struct packed_git *p;
1195         prepare_packed_git();
1196
1197         for (p = packed_git; p; p = p->next) {
1198                 if (find_pack_entry_one(sha1, e, p))
1199                         return 1;
1200         }
1201         return 0;
1202 }
1203
1204 struct packed_git *find_sha1_pack(const unsigned char *sha1,
1205                                   struct packed_git *packs)
1206 {
1207         struct packed_git *p;
1208         struct pack_entry e;
1209
1210         for (p = packs; p; p = p->next) {
1211                 if (find_pack_entry_one(sha1, &e, p))
1212                         return p;
1213         }
1214         return NULL;
1215
1216 }
1217
1218 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
1219 {
1220         int status;
1221         unsigned long mapsize, size;
1222         void *map;
1223         z_stream stream;
1224         char hdr[128];
1225
1226         map = map_sha1_file(sha1, &mapsize);
1227         if (!map) {
1228                 struct pack_entry e;
1229
1230                 if (find_pack_entry(sha1, &e))
1231                         return packed_object_info(&e, type, sizep);
1232                 reprepare_packed_git();
1233                 if (find_pack_entry(sha1, &e))
1234                         return packed_object_info(&e, type, sizep);
1235                 return error("unable to find %s", sha1_to_hex(sha1));
1236         }
1237         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
1238                 status = error("unable to unpack %s header",
1239                                sha1_to_hex(sha1));
1240         if (parse_sha1_header(hdr, type, &size) < 0)
1241                 status = error("unable to parse %s header", sha1_to_hex(sha1));
1242         else {
1243                 status = 0;
1244                 if (sizep)
1245                         *sizep = size;
1246         }
1247         inflateEnd(&stream);
1248         munmap(map, mapsize);
1249         return status;
1250 }
1251
1252 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
1253 {
1254         struct pack_entry e;
1255
1256         if (!find_pack_entry(sha1, &e)) {
1257                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
1258                 return NULL;
1259         }
1260         return unpack_entry(&e, type, size);
1261 }
1262
1263 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
1264 {
1265         unsigned long mapsize;
1266         void *map, *buf;
1267         struct pack_entry e;
1268
1269         if (find_pack_entry(sha1, &e))
1270                 return read_packed_sha1(sha1, type, size);
1271         map = map_sha1_file(sha1, &mapsize);
1272         if (map) {
1273                 buf = unpack_sha1_file(map, mapsize, type, size);
1274                 munmap(map, mapsize);
1275                 return buf;
1276         }
1277         reprepare_packed_git();
1278         if (find_pack_entry(sha1, &e))
1279                 return read_packed_sha1(sha1, type, size);
1280         return NULL;
1281 }
1282
1283 void *read_object_with_reference(const unsigned char *sha1,
1284                                  const char *required_type,
1285                                  unsigned long *size,
1286                                  unsigned char *actual_sha1_return)
1287 {
1288         char type[20];
1289         void *buffer;
1290         unsigned long isize;
1291         unsigned char actual_sha1[20];
1292
1293         hashcpy(actual_sha1, sha1);
1294         while (1) {
1295                 int ref_length = -1;
1296                 const char *ref_type = NULL;
1297
1298                 buffer = read_sha1_file(actual_sha1, type, &isize);
1299                 if (!buffer)
1300                         return NULL;
1301                 if (!strcmp(type, required_type)) {
1302                         *size = isize;
1303                         if (actual_sha1_return)
1304                                 hashcpy(actual_sha1_return, actual_sha1);
1305                         return buffer;
1306                 }
1307                 /* Handle references */
1308                 else if (!strcmp(type, commit_type))
1309                         ref_type = "tree ";
1310                 else if (!strcmp(type, tag_type))
1311                         ref_type = "object ";
1312                 else {
1313                         free(buffer);
1314                         return NULL;
1315                 }
1316                 ref_length = strlen(ref_type);
1317
1318                 if (memcmp(buffer, ref_type, ref_length) ||
1319                     get_sha1_hex((char *) buffer + ref_length, actual_sha1)) {
1320                         free(buffer);
1321                         return NULL;
1322                 }
1323                 free(buffer);
1324                 /* Now we have the ID of the referred-to object in
1325                  * actual_sha1.  Check again. */
1326         }
1327 }
1328
1329 char *write_sha1_file_prepare(void *buf,
1330                               unsigned long len,
1331                               const char *type,
1332                               unsigned char *sha1,
1333                               unsigned char *hdr,
1334                               int *hdrlen)
1335 {
1336         SHA_CTX c;
1337
1338         /* Generate the header */
1339         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
1340
1341         /* Sha1.. */
1342         SHA1_Init(&c);
1343         SHA1_Update(&c, hdr, *hdrlen);
1344         SHA1_Update(&c, buf, len);
1345         SHA1_Final(sha1, &c);
1346
1347         return sha1_file_name(sha1);
1348 }
1349
1350 /*
1351  * Link the tempfile to the final place, possibly creating the
1352  * last directory level as you do so.
1353  *
1354  * Returns the errno on failure, 0 on success.
1355  */
1356 static int link_temp_to_file(const char *tmpfile, const char *filename)
1357 {
1358         int ret;
1359         char *dir;
1360
1361         if (!link(tmpfile, filename))
1362                 return 0;
1363
1364         /*
1365          * Try to mkdir the last path component if that failed.
1366          *
1367          * Re-try the "link()" regardless of whether the mkdir
1368          * succeeds, since a race might mean that somebody
1369          * else succeeded.
1370          */
1371         ret = errno;
1372         dir = strrchr(filename, '/');
1373         if (dir) {
1374                 *dir = 0;
1375                 mkdir(filename, 0777);
1376                 if (adjust_shared_perm(filename))
1377                         return -2;
1378                 *dir = '/';
1379                 if (!link(tmpfile, filename))
1380                         return 0;
1381                 ret = errno;
1382         }
1383         return ret;
1384 }
1385
1386 /*
1387  * Move the just written object into its final resting place
1388  */
1389 int move_temp_to_file(const char *tmpfile, const char *filename)
1390 {
1391         int ret = link_temp_to_file(tmpfile, filename);
1392
1393         /*
1394          * Coda hack - coda doesn't like cross-directory links,
1395          * so we fall back to a rename, which will mean that it
1396          * won't be able to check collisions, but that's not a
1397          * big deal.
1398          *
1399          * The same holds for FAT formatted media.
1400          *
1401          * When this succeeds, we just return 0. We have nothing
1402          * left to unlink.
1403          */
1404         if (ret && ret != EEXIST) {
1405                 if (!rename(tmpfile, filename))
1406                         return 0;
1407                 ret = errno;
1408         }
1409         unlink(tmpfile);
1410         if (ret) {
1411                 if (ret != EEXIST) {
1412                         fprintf(stderr, "unable to write sha1 filename %s: %s\n", filename, strerror(ret));
1413                         return -1;
1414                 }
1415                 /* FIXME!!! Collision check here ? */
1416         }
1417
1418         return 0;
1419 }
1420
1421 static int write_buffer(int fd, const void *buf, size_t len)
1422 {
1423         while (len) {
1424                 ssize_t size;
1425
1426                 size = write(fd, buf, len);
1427                 if (!size)
1428                         return error("file write: disk full");
1429                 if (size < 0) {
1430                         if (errno == EINTR || errno == EAGAIN)
1431                                 continue;
1432                         return error("file write error (%s)", strerror(errno));
1433                 }
1434                 len -= size;
1435                 buf = (char *) buf + size;
1436         }
1437         return 0;
1438 }
1439
1440 static int write_binary_header(unsigned char *hdr, enum object_type type, unsigned long len)
1441 {
1442         int hdr_len;
1443         unsigned char c;
1444
1445         c = (type << 4) | (len & 15);
1446         len >>= 4;
1447         hdr_len = 1;
1448         while (len) {
1449                 *hdr++ = c | 0x80;
1450                 hdr_len++;
1451                 c = (len & 0x7f);
1452                 len >>= 7;
1453         }
1454         *hdr = c;
1455         return hdr_len;
1456 }
1457
1458 static void setup_object_header(z_stream *stream, const char *type, unsigned long len)
1459 {
1460         int obj_type, hdr;
1461
1462         if (use_legacy_headers) {
1463                 while (deflate(stream, 0) == Z_OK)
1464                         /* nothing */;
1465                 return;
1466         }
1467         if (!strcmp(type, blob_type))
1468                 obj_type = OBJ_BLOB;
1469         else if (!strcmp(type, tree_type))
1470                 obj_type = OBJ_TREE;
1471         else if (!strcmp(type, commit_type))
1472                 obj_type = OBJ_COMMIT;
1473         else if (!strcmp(type, tag_type))
1474                 obj_type = OBJ_TAG;
1475         else
1476                 die("trying to generate bogus object of type '%s'", type);
1477         hdr = write_binary_header(stream->next_out, obj_type, len);
1478         stream->total_out = hdr;
1479         stream->next_out += hdr;
1480         stream->avail_out -= hdr;
1481 }
1482
1483 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1484 {
1485         int size;
1486         unsigned char *compressed;
1487         z_stream stream;
1488         unsigned char sha1[20];
1489         char *filename;
1490         static char tmpfile[PATH_MAX];
1491         unsigned char hdr[50];
1492         int fd, hdrlen;
1493
1494         /* Normally if we have it in the pack then we do not bother writing
1495          * it out into .git/objects/??/?{38} file.
1496          */
1497         filename = write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1498         if (returnsha1)
1499                 hashcpy(returnsha1, sha1);
1500         if (has_sha1_file(sha1))
1501                 return 0;
1502         fd = open(filename, O_RDONLY);
1503         if (fd >= 0) {
1504                 /*
1505                  * FIXME!!! We might do collision checking here, but we'd
1506                  * need to uncompress the old file and check it. Later.
1507                  */
1508                 close(fd);
1509                 return 0;
1510         }
1511
1512         if (errno != ENOENT) {
1513                 fprintf(stderr, "sha1 file %s: %s\n", filename, strerror(errno));
1514                 return -1;
1515         }
1516
1517         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1518
1519         fd = mkstemp(tmpfile);
1520         if (fd < 0) {
1521                 fprintf(stderr, "unable to create temporary sha1 filename %s: %s\n", tmpfile, strerror(errno));
1522                 return -1;
1523         }
1524
1525         /* Set it up */
1526         memset(&stream, 0, sizeof(stream));
1527         deflateInit(&stream, zlib_compression_level);
1528         size = 8 + deflateBound(&stream, len+hdrlen);
1529         compressed = xmalloc(size);
1530
1531         /* Compress it */
1532         stream.next_out = compressed;
1533         stream.avail_out = size;
1534
1535         /* First header.. */
1536         stream.next_in = hdr;
1537         stream.avail_in = hdrlen;
1538         setup_object_header(&stream, type, len);
1539
1540         /* Then the data itself.. */
1541         stream.next_in = buf;
1542         stream.avail_in = len;
1543         while (deflate(&stream, Z_FINISH) == Z_OK)
1544                 /* nothing */;
1545         deflateEnd(&stream);
1546         size = stream.total_out;
1547
1548         if (write_buffer(fd, compressed, size) < 0)
1549                 die("unable to write sha1 file");
1550         fchmod(fd, 0444);
1551         close(fd);
1552         free(compressed);
1553
1554         return move_temp_to_file(tmpfile, filename);
1555 }
1556
1557 /*
1558  * We need to unpack and recompress the object for writing
1559  * it out to a different file.
1560  */
1561 static void *repack_object(const unsigned char *sha1, unsigned long *objsize)
1562 {
1563         size_t size;
1564         z_stream stream;
1565         unsigned char *unpacked;
1566         unsigned long len;
1567         char type[20];
1568         char hdr[50];
1569         int hdrlen;
1570         void *buf;
1571
1572         /* need to unpack and recompress it by itself */
1573         unpacked = read_packed_sha1(sha1, type, &len);
1574
1575         hdrlen = sprintf(hdr, "%s %lu", type, len) + 1;
1576
1577         /* Set it up */
1578         memset(&stream, 0, sizeof(stream));
1579         deflateInit(&stream, zlib_compression_level);
1580         size = deflateBound(&stream, len + hdrlen);
1581         buf = xmalloc(size);
1582
1583         /* Compress it */
1584         stream.next_out = buf;
1585         stream.avail_out = size;
1586
1587         /* First header.. */
1588         stream.next_in = (void *)hdr;
1589         stream.avail_in = hdrlen;
1590         while (deflate(&stream, 0) == Z_OK)
1591                 /* nothing */;
1592
1593         /* Then the data itself.. */
1594         stream.next_in = unpacked;
1595         stream.avail_in = len;
1596         while (deflate(&stream, Z_FINISH) == Z_OK)
1597                 /* nothing */;
1598         deflateEnd(&stream);
1599         free(unpacked);
1600
1601         *objsize = stream.total_out;
1602         return buf;
1603 }
1604
1605 int write_sha1_to_fd(int fd, const unsigned char *sha1)
1606 {
1607         int retval;
1608         unsigned long objsize;
1609         void *buf = map_sha1_file(sha1, &objsize);
1610
1611         if (buf) {
1612                 retval = write_buffer(fd, buf, objsize);
1613                 munmap(buf, objsize);
1614                 return retval;
1615         }
1616
1617         buf = repack_object(sha1, &objsize);
1618         retval = write_buffer(fd, buf, objsize);
1619         free(buf);
1620         return retval;
1621 }
1622
1623 int write_sha1_from_fd(const unsigned char *sha1, int fd, char *buffer,
1624                        size_t bufsize, size_t *bufposn)
1625 {
1626         char tmpfile[PATH_MAX];
1627         int local;
1628         z_stream stream;
1629         unsigned char real_sha1[20];
1630         unsigned char discard[4096];
1631         int ret;
1632         SHA_CTX c;
1633
1634         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1635
1636         local = mkstemp(tmpfile);
1637         if (local < 0)
1638                 return error("Couldn't open %s for %s",
1639                              tmpfile, sha1_to_hex(sha1));
1640
1641         memset(&stream, 0, sizeof(stream));
1642
1643         inflateInit(&stream);
1644
1645         SHA1_Init(&c);
1646
1647         do {
1648                 ssize_t size;
1649                 if (*bufposn) {
1650                         stream.avail_in = *bufposn;
1651                         stream.next_in = (unsigned char *) buffer;
1652                         do {
1653                                 stream.next_out = discard;
1654                                 stream.avail_out = sizeof(discard);
1655                                 ret = inflate(&stream, Z_SYNC_FLUSH);
1656                                 SHA1_Update(&c, discard, sizeof(discard) -
1657                                             stream.avail_out);
1658                         } while (stream.avail_in && ret == Z_OK);
1659                         if (write_buffer(local, buffer, *bufposn - stream.avail_in) < 0)
1660                                 die("unable to write sha1 file");
1661                         memmove(buffer, buffer + *bufposn - stream.avail_in,
1662                                 stream.avail_in);
1663                         *bufposn = stream.avail_in;
1664                         if (ret != Z_OK)
1665                                 break;
1666                 }
1667                 size = read(fd, buffer + *bufposn, bufsize - *bufposn);
1668                 if (size <= 0) {
1669                         close(local);
1670                         unlink(tmpfile);
1671                         if (!size)
1672                                 return error("Connection closed?");
1673                         perror("Reading from connection");
1674                         return -1;
1675                 }
1676                 *bufposn += size;
1677         } while (1);
1678         inflateEnd(&stream);
1679
1680         close(local);
1681         SHA1_Final(real_sha1, &c);
1682         if (ret != Z_STREAM_END) {
1683                 unlink(tmpfile);
1684                 return error("File %s corrupted", sha1_to_hex(sha1));
1685         }
1686         if (hashcmp(sha1, real_sha1)) {
1687                 unlink(tmpfile);
1688                 return error("File %s has bad hash", sha1_to_hex(sha1));
1689         }
1690
1691         return move_temp_to_file(tmpfile, sha1_file_name(sha1));
1692 }
1693
1694 int has_pack_index(const unsigned char *sha1)
1695 {
1696         struct stat st;
1697         if (stat(sha1_pack_index_name(sha1), &st))
1698                 return 0;
1699         return 1;
1700 }
1701
1702 int has_pack_file(const unsigned char *sha1)
1703 {
1704         struct stat st;
1705         if (stat(sha1_pack_name(sha1), &st))
1706                 return 0;
1707         return 1;
1708 }
1709
1710 int has_sha1_pack(const unsigned char *sha1)
1711 {
1712         struct pack_entry e;
1713         return find_pack_entry(sha1, &e);
1714 }
1715
1716 int has_sha1_file(const unsigned char *sha1)
1717 {
1718         struct stat st;
1719         struct pack_entry e;
1720
1721         if (find_pack_entry(sha1, &e))
1722                 return 1;
1723         return find_sha1_file(sha1, &st) ? 1 : 0;
1724 }
1725
1726 /*
1727  * reads from fd as long as possible into a supplied buffer of size bytes.
1728  * If necessary the buffer's size is increased using realloc()
1729  *
1730  * returns 0 if anything went fine and -1 otherwise
1731  *
1732  * NOTE: both buf and size may change, but even when -1 is returned
1733  * you still have to free() it yourself.
1734  */
1735 int read_pipe(int fd, char** return_buf, unsigned long* return_size)
1736 {
1737         char* buf = *return_buf;
1738         unsigned long size = *return_size;
1739         int iret;
1740         unsigned long off = 0;
1741
1742         do {
1743                 iret = xread(fd, buf + off, size - off);
1744                 if (iret > 0) {
1745                         off += iret;
1746                         if (off == size) {
1747                                 size *= 2;
1748                                 buf = xrealloc(buf, size);
1749                         }
1750                 }
1751         } while (iret > 0);
1752
1753         *return_buf = buf;
1754         *return_size = off;
1755
1756         if (iret < 0)
1757                 return -1;
1758         return 0;
1759 }
1760
1761 int index_pipe(unsigned char *sha1, int fd, const char *type, int write_object)
1762 {
1763         unsigned long size = 4096;
1764         char *buf = xmalloc(size);
1765         int ret;
1766         unsigned char hdr[50];
1767         int hdrlen;
1768
1769         if (read_pipe(fd, &buf, &size)) {
1770                 free(buf);
1771                 return -1;
1772         }
1773
1774         if (!type)
1775                 type = blob_type;
1776         if (write_object)
1777                 ret = write_sha1_file(buf, size, type, sha1);
1778         else {
1779                 write_sha1_file_prepare(buf, size, type, sha1, hdr, &hdrlen);
1780                 ret = 0;
1781         }
1782         free(buf);
1783         return ret;
1784 }
1785
1786 int index_fd(unsigned char *sha1, int fd, struct stat *st, int write_object, const char *type)
1787 {
1788         unsigned long size = st->st_size;
1789         void *buf;
1790         int ret;
1791         unsigned char hdr[50];
1792         int hdrlen;
1793
1794         buf = "";
1795         if (size)
1796                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1797         close(fd);
1798         if (buf == MAP_FAILED)
1799                 return -1;
1800
1801         if (!type)
1802                 type = blob_type;
1803         if (write_object)
1804                 ret = write_sha1_file(buf, size, type, sha1);
1805         else {
1806                 write_sha1_file_prepare(buf, size, type, sha1, hdr, &hdrlen);
1807                 ret = 0;
1808         }
1809         if (size)
1810                 munmap(buf, size);
1811         return ret;
1812 }
1813
1814 int index_path(unsigned char *sha1, const char *path, struct stat *st, int write_object)
1815 {
1816         int fd;
1817         char *target;
1818
1819         switch (st->st_mode & S_IFMT) {
1820         case S_IFREG:
1821                 fd = open(path, O_RDONLY);
1822                 if (fd < 0)
1823                         return error("open(\"%s\"): %s", path,
1824                                      strerror(errno));
1825                 if (index_fd(sha1, fd, st, write_object, NULL) < 0)
1826                         return error("%s: failed to insert into database",
1827                                      path);
1828                 break;
1829         case S_IFLNK:
1830                 target = xmalloc(st->st_size+1);
1831                 if (readlink(path, target, st->st_size+1) != st->st_size) {
1832                         char *errstr = strerror(errno);
1833                         free(target);
1834                         return error("readlink(\"%s\"): %s", path,
1835                                      errstr);
1836                 }
1837                 if (!write_object) {
1838                         unsigned char hdr[50];
1839                         int hdrlen;
1840                         write_sha1_file_prepare(target, st->st_size, blob_type,
1841                                                 sha1, hdr, &hdrlen);
1842                 } else if (write_sha1_file(target, st->st_size, blob_type, sha1))
1843                         return error("%s: failed to insert into database",
1844                                      path);
1845                 free(target);
1846                 break;
1847         default:
1848                 return error("%s: unsupported file type", path);
1849         }
1850         return 0;
1851 }