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 signed char hexval_table[256] = {
  30          -1, -1, -1, -1, -1, -1, -1, -1,                /* 00-07 */
  31          -1, -1, -1, -1, -1, -1, -1, -1,                /* 08-0f */
  32          -1, -1, -1, -1, -1, -1, -1, -1,                /* 10-17 */
  33          -1, -1, -1, -1, -1, -1, -1, -1,                /* 18-1f */
  34          -1, -1, -1, -1, -1, -1, -1, -1,                /* 20-27 */
  35          -1, -1, -1, -1, -1, -1, -1, -1,                /* 28-2f */
  36           0,  1,  2,  3,  4,  5,  6,  7,                /* 30-37 */
  37           8,  9, -1, -1, -1, -1, -1, -1,                /* 38-3f */
  38          -1, 10, 11, 12, 13, 14, 15, -1,                /* 40-47 */
  39          -1, -1, -1, -1, -1, -1, -1, -1,                /* 48-4f */
  40          -1, -1, -1, -1, -1, -1, -1, -1,                /* 50-57 */
  41          -1, -1, -1, -1, -1, -1, -1, -1,                /* 58-5f */
  42          -1, 10, 11, 12, 13, 14, 15, -1,                /* 60-67 */
  43          -1, -1, -1, -1, -1, -1, -1, -1,                /* 68-67 */
  44          -1, -1, -1, -1, -1, -1, -1, -1,                /* 70-77 */
  45          -1, -1, -1, -1, -1, -1, -1, -1,                /* 78-7f */
  46          -1, -1, -1, -1, -1, -1, -1, -1,                /* 80-87 */
  47          -1, -1, -1, -1, -1, -1, -1, -1,                /* 88-8f */
  48          -1, -1, -1, -1, -1, -1, -1, -1,                /* 90-97 */
  49          -1, -1, -1, -1, -1, -1, -1, -1,                /* 98-9f */
  50          -1, -1, -1, -1, -1, -1, -1, -1,                /* a0-a7 */
  51          -1, -1, -1, -1, -1, -1, -1, -1,                /* a8-af */
  52          -1, -1, -1, -1, -1, -1, -1, -1,                /* b0-b7 */
  53          -1, -1, -1, -1, -1, -1, -1, -1,                /* b8-bf */
  54          -1, -1, -1, -1, -1, -1, -1, -1,                /* c0-c7 */
  55          -1, -1, -1, -1, -1, -1, -1, -1,                /* c8-cf */
  56          -1, -1, -1, -1, -1, -1, -1, -1,                /* d0-d7 */
  57          -1, -1, -1, -1, -1, -1, -1, -1,                /* d8-df */
  58          -1, -1, -1, -1, -1, -1, -1, -1,                /* e0-e7 */
  59          -1, -1, -1, -1, -1, -1, -1, -1,                /* e8-ef */
  60          -1, -1, -1, -1, -1, -1, -1, -1,                /* f0-f7 */
  61          -1, -1, -1, -1, -1, -1, -1, -1,                /* f8-ff */
  62 };
  63
  64 int get_sha1_hex(const char *hex, unsigned char *sha1)
  65 {
  66         int i;
  67         for (i = 0; i < 20; i++) {
  68                 unsigned int val = (hexval(hex[0]) << 4) | hexval(hex[1]);
  69                 if (val & ~0xff)
  70                         return -1;
  71                 *sha1++ = val;
  72                 hex += 2;
  73         }
  74         return 0;
  75 }
  76
  77 int safe_create_leading_directories(char *path)
  78 {
  79         char *pos = path;
  80         struct stat st;
  81
  82         if (*pos == '/')
  83                 pos++;
  84
  85         while (pos) {
  86                 pos = strchr(pos, '/');
  87                 if (!pos)
  88                         break;
  89                 *pos = 0;
  90                 if (!stat(path, &st)) {
  91                         /* path exists */
  92                         if (!S_ISDIR(st.st_mode)) {
  93                                 *pos = '/';
  94                                 return -3;
  95                         }
  96                 }
  97                 else if (mkdir(path, 0777)) {
  98                         *pos = '/';
  99                         return -1;
 100                 }
 101                 else if (adjust_shared_perm(path)) {
 102                         *pos = '/';
 103                         return -2;
 104                 }
 105                 *pos++ = '/';
 106         }
 107         return 0;
 108 }
 109
 110 char * sha1_to_hex(const unsigned char *sha1)
 111 {
 112         static int bufno;
 113         static char hexbuffer[4][50];
 114         static const char hex[] = "0123456789abcdef";
 115         char *buffer = hexbuffer[3 & ++bufno], *buf = buffer;
 116         int i;
 117
 118         for (i = 0; i < 20; i++) {
 119                 unsigned int val = *sha1++;
 120                 *buf++ = hex[val >> 4];
 121                 *buf++ = hex[val & 0xf];
 122         }
 123         *buf = '\0';
 124
 125         return buffer;
 126 }
 127
 128 static void fill_sha1_path(char *pathbuf, const unsigned char *sha1)
 129 {
 130         int i;
 131         for (i = 0; i < 20; i++) {
 132                 static char hex[] = "0123456789abcdef";
 133                 unsigned int val = sha1[i];
 134                 char *pos = pathbuf + i*2 + (i > 0);
 135                 *pos++ = hex[val >> 4];
 136                 *pos = hex[val & 0xf];
 137         }
 138 }
 139
 140 /*
 141  * NOTE! This returns a statically allocated buffer, so you have to be
 142  * careful about using it. Do a "xstrdup()" if you need to save the
 143  * filename.
 144  *
 145  * Also note that this returns the location for creating.  Reading
 146  * SHA1 file can happen from any alternate directory listed in the
 147  * DB_ENVIRONMENT environment variable if it is not found in
 148  * the primary object database.
 149  */
 150 char *sha1_file_name(const unsigned char *sha1)
 151 {
 152         static char *name, *base;
 153
 154         if (!base) {
 155                 const char *sha1_file_directory = get_object_directory();
 156                 int len = strlen(sha1_file_directory);
 157                 base = xmalloc(len + 60);
 158                 memcpy(base, sha1_file_directory, len);
 159                 memset(base+len, 0, 60);
 160                 base[len] = '/';
 161                 base[len+3] = '/';
 162                 name = base + len + 1;
 163         }
 164         fill_sha1_path(name, sha1);
 165         return base;
 166 }
 167
 168 char *sha1_pack_name(const unsigned char *sha1)
 169 {
 170         static const char hex[] = "0123456789abcdef";
 171         static char *name, *base, *buf;
 172         int i;
 173
 174         if (!base) {
 175                 const char *sha1_file_directory = get_object_directory();
 176                 int len = strlen(sha1_file_directory);
 177                 base = xmalloc(len + 60);
 178                 sprintf(base, "%s/pack/pack-1234567890123456789012345678901234567890.pack", sha1_file_directory);
 179                 name = base + len + 11;
 180         }
 181
 182         buf = name;
 183
 184         for (i = 0; i < 20; i++) {
 185                 unsigned int val = *sha1++;
 186                 *buf++ = hex[val >> 4];
 187                 *buf++ = hex[val & 0xf];
 188         }
 189
 190         return base;
 191 }
 192
 193 char *sha1_pack_index_name(const unsigned char *sha1)
 194 {
 195         static const char hex[] = "0123456789abcdef";
 196         static char *name, *base, *buf;
 197         int i;
 198
 199         if (!base) {
 200                 const char *sha1_file_directory = get_object_directory();
 201                 int len = strlen(sha1_file_directory);
 202                 base = xmalloc(len + 60);
 203                 sprintf(base, "%s/pack/pack-1234567890123456789012345678901234567890.idx", sha1_file_directory);
 204                 name = base + len + 11;
 205         }
 206
 207         buf = name;
 208
 209         for (i = 0; i < 20; i++) {
 210                 unsigned int val = *sha1++;
 211                 *buf++ = hex[val >> 4];
 212                 *buf++ = hex[val & 0xf];
 213         }
 214
 215         return base;
 216 }
 217
 218 struct alternate_object_database *alt_odb_list;
 219 static struct alternate_object_database **alt_odb_tail;
 220
 221 static void read_info_alternates(const char * alternates, int depth);
 222
 223 /*
 224  * Prepare alternate object database registry.
 225  *
 226  * The variable alt_odb_list points at the list of struct
 227  * alternate_object_database.  The elements on this list come from
 228  * non-empty elements from colon separated ALTERNATE_DB_ENVIRONMENT
 229  * environment variable, and $GIT_OBJECT_DIRECTORY/info/alternates,
 230  * whose contents is similar to that environment variable but can be
 231  * LF separated.  Its base points at a statically allocated buffer that
 232  * contains "/the/directory/corresponding/to/.git/objects/...", while
 233  * its name points just after the slash at the end of ".git/objects/"
 234  * in the example above, and has enough space to hold 40-byte hex
 235  * SHA1, an extra slash for the first level indirection, and the
 236  * terminating NUL.
 237  */
 238 static int link_alt_odb_entry(const char * entry, int len, const char * relative_base, int depth)
 239 {
 240         struct stat st;
 241         const char *objdir = get_object_directory();
 242         struct alternate_object_database *ent;
 243         struct alternate_object_database *alt;
 244         /* 43 = 40-byte + 2 '/' + terminating NUL */
 245         int pfxlen = len;
 246         int entlen = pfxlen + 43;
 247         int base_len = -1;
 248
 249         if (*entry != '/' && relative_base) {
 250                 /* Relative alt-odb */
 251                 if (base_len < 0)
 252                         base_len = strlen(relative_base) + 1;
 253                 entlen += base_len;
 254                 pfxlen += base_len;
 255         }
 256         ent = xmalloc(sizeof(*ent) + entlen);
 257
 258         if (*entry != '/' && relative_base) {
 259                 memcpy(ent->base, relative_base, base_len - 1);
 260                 ent->base[base_len - 1] = '/';
 261                 memcpy(ent->base + base_len, entry, len);
 262         }
 263         else
 264                 memcpy(ent->base, entry, pfxlen);
 265
 266         ent->name = ent->base + pfxlen + 1;
 267         ent->base[pfxlen + 3] = '/';
 268         ent->base[pfxlen] = ent->base[entlen-1] = 0;
 269
 270         /* Detect cases where alternate disappeared */
 271         if (stat(ent->base, &st) || !S_ISDIR(st.st_mode)) {
 272                 error("object directory %s does not exist; "
 273                       "check .git/objects/info/alternates.",
 274                       ent->base);
 275                 free(ent);
 276                 return -1;
 277         }
 278
 279         /* Prevent the common mistake of listing the same
 280          * thing twice, or object directory itself.
 281          */
 282         for (alt = alt_odb_list; alt; alt = alt->next) {
 283                 if (!memcmp(ent->base, alt->base, pfxlen)) {
 284                         free(ent);
 285                         return -1;
 286                 }
 287         }
 288         if (!memcmp(ent->base, objdir, pfxlen)) {
 289                 free(ent);
 290                 return -1;
 291         }
 292
 293         /* add the alternate entry */
 294         *alt_odb_tail = ent;
 295         alt_odb_tail = &(ent->next);
 296         ent->next = NULL;
 297
 298         /* recursively add alternates */
 299         read_info_alternates(ent->base, depth + 1);
 300
 301         ent->base[pfxlen] = '/';
 302
 303         return 0;
 304 }
 305
 306 static void link_alt_odb_entries(const char *alt, const char *ep, int sep,
 307                                  const char *relative_base, int depth)
 308 {
 309         const char *cp, *last;
 310
 311         if (depth > 5) {
 312                 error("%s: ignoring alternate object stores, nesting too deep.",
 313                                 relative_base);
 314                 return;
 315         }
 316
 317         last = alt;
 318         while (last < ep) {
 319                 cp = last;
 320                 if (cp < ep && *cp == '#') {
 321                         while (cp < ep && *cp != sep)
 322                                 cp++;
 323                         last = cp + 1;
 324                         continue;
 325                 }
 326                 while (cp < ep && *cp != sep)
 327                         cp++;
 328                 if (last != cp) {
 329                         if ((*last != '/') && depth) {
 330                                 error("%s: ignoring relative alternate object store %s",
 331                                                 relative_base, last);
 332                         } else {
 333                                 link_alt_odb_entry(last, cp - last,
 334                                                 relative_base, depth);
 335                         }
 336                 }
 337                 while (cp < ep && *cp == sep)
 338                         cp++;
 339                 last = cp;
 340         }
 341 }
 342
 343 static void read_info_alternates(const char * relative_base, int depth)
 344 {
 345         char *map;
 346         struct stat st;
 347         char path[PATH_MAX];
 348         int fd;
 349
 350         sprintf(path, "%s/info/alternates", relative_base);
 351         fd = open(path, O_RDONLY);
 352         if (fd < 0)
 353                 return;
 354         if (fstat(fd, &st) || (st.st_size == 0)) {
 355                 close(fd);
 356                 return;
 357         }
 358         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 359         close(fd);
 360         if (map == MAP_FAILED)
 361                 return;
 362
 363         link_alt_odb_entries(map, map + st.st_size, '\n', relative_base, depth);
 364
 365         munmap(map, st.st_size);
 366 }
 367
 368 void prepare_alt_odb(void)
 369 {
 370         const char *alt;
 371
 372         alt = getenv(ALTERNATE_DB_ENVIRONMENT);
 373         if (!alt) alt = "";
 374
 375         if (alt_odb_tail)
 376                 return;
 377         alt_odb_tail = &alt_odb_list;
 378         link_alt_odb_entries(alt, alt + strlen(alt), ':', NULL, 0);
 379
 380         read_info_alternates(get_object_directory(), 0);
 381 }
 382
 383 static char *find_sha1_file(const unsigned char *sha1, struct stat *st)
 384 {
 385         char *name = sha1_file_name(sha1);
 386         struct alternate_object_database *alt;
 387
 388         if (!stat(name, st))
 389                 return name;
 390         prepare_alt_odb();
 391         for (alt = alt_odb_list; alt; alt = alt->next) {
 392                 name = alt->name;
 393                 fill_sha1_path(name, sha1);
 394                 if (!stat(alt->base, st))
 395                         return alt->base;
 396         }
 397         return NULL;
 398 }
 399
 400 #define PACK_MAX_SZ (1<<26)
 401 static int pack_used_ctr;
 402 static unsigned long pack_mapped;
 403 struct packed_git *packed_git;
 404
 405 static int check_packed_git_idx(const char *path, unsigned long *idx_size_,
 406                                 void **idx_map_)
 407 {
 408         void *idx_map;
 409         unsigned int *index;
 410         unsigned long idx_size;
 411         int nr, i;
 412         int fd = open(path, O_RDONLY);
 413         struct stat st;
 414         if (fd < 0)
 415                 return -1;
 416         if (fstat(fd, &st)) {
 417                 close(fd);
 418                 return -1;
 419         }
 420         idx_size = st.st_size;
 421         idx_map = mmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0);
 422         close(fd);
 423         if (idx_map == MAP_FAILED)
 424                 return -1;
 425
 426         index = idx_map;
 427         *idx_map_ = idx_map;
 428         *idx_size_ = idx_size;
 429
 430         /* check index map */
 431         if (idx_size < 4*256 + 20 + 20)
 432                 return error("index file too small");
 433         nr = 0;
 434         for (i = 0; i < 256; i++) {
 435                 unsigned int n = ntohl(index[i]);
 436                 if (n < nr)
 437                         return error("non-monotonic index");
 438                 nr = n;
 439         }
 440
 441         /*
 442          * Total size:
 443          *  - 256 index entries 4 bytes each
 444          *  - 24-byte entries * nr (20-byte sha1 + 4-byte offset)
 445          *  - 20-byte SHA1 of the packfile
 446          *  - 20-byte SHA1 file checksum
 447          */
 448         if (idx_size != 4*256 + nr * 24 + 20 + 20)
 449                 return error("wrong index file size");
 450
 451         return 0;
 452 }
 453
 454 static int unuse_one_packed_git(void)
 455 {
 456         struct packed_git *p, *lru = NULL;
 457
 458         for (p = packed_git; p; p = p->next) {
 459                 if (p->pack_use_cnt || !p->pack_base)
 460                         continue;
 461                 if (!lru || p->pack_last_used < lru->pack_last_used)
 462                         lru = p;
 463         }
 464         if (!lru)
 465                 return 0;
 466         munmap(lru->pack_base, lru->pack_size);
 467         lru->pack_base = NULL;
 468         return 1;
 469 }
 470
 471 void unuse_packed_git(struct packed_git *p)
 472 {
 473         p->pack_use_cnt--;
 474 }
 475
 476 int use_packed_git(struct packed_git *p)
 477 {
 478         if (!p->pack_size) {
 479                 struct stat st;
 480                 /* We created the struct before we had the pack */
 481                 stat(p->pack_name, &st);
 482                 if (!S_ISREG(st.st_mode))
 483                         die("packfile %s not a regular file", p->pack_name);
 484                 p->pack_size = st.st_size;
 485         }
 486         if (!p->pack_base) {
 487                 int fd;
 488                 struct stat st;
 489                 void *map;
 490                 struct pack_header *hdr;
 491
 492                 pack_mapped += p->pack_size;
 493                 while (PACK_MAX_SZ < pack_mapped && unuse_one_packed_git())
 494                         ; /* nothing */
 495                 fd = open(p->pack_name, O_RDONLY);
 496                 if (fd < 0)
 497                         die("packfile %s cannot be opened", p->pack_name);
 498                 if (fstat(fd, &st)) {
 499                         close(fd);
 500                         die("packfile %s cannot be opened", p->pack_name);
 501                 }
 502                 if (st.st_size != p->pack_size)
 503                         die("packfile %s size mismatch.", p->pack_name);
 504                 map = mmap(NULL, p->pack_size, PROT_READ, MAP_PRIVATE, fd, 0);
 505                 close(fd);
 506                 if (map == MAP_FAILED)
 507                         die("packfile %s cannot be mapped.", p->pack_name);
 508                 p->pack_base = map;
 509
 510                 /* Check if we understand this pack file.  If we don't we're
 511                  * likely too old to handle it.
 512                  */
 513                 hdr = map;
 514                 if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 515                         die("packfile %s isn't actually a pack.", p->pack_name);
 516                 if (!pack_version_ok(hdr->hdr_version))
 517                         die("packfile %s is version %i and not supported"
 518                                 " (try upgrading GIT to a newer version)",
 519                                 p->pack_name, ntohl(hdr->hdr_version));
 520
 521                 /* Check if the pack file matches with the index file.
 522                  * this is cheap.
 523                  */
 524                 if (hashcmp((unsigned char *)(p->index_base) +
 525                             p->index_size - 40,
 526                             (unsigned char *)p->pack_base +
 527                             p->pack_size - 20)) {
 528                         die("packfile %s does not match index.", p->pack_name);
 529                 }
 530         }
 531         p->pack_last_used = pack_used_ctr++;
 532         p->pack_use_cnt++;
 533         return 0;
 534 }
 535
 536 struct packed_git *add_packed_git(char *path, int path_len, int local)
 537 {
 538         struct stat st;
 539         struct packed_git *p;
 540         unsigned long idx_size;
 541         void *idx_map;
 542         unsigned char sha1[20];
 543
 544         if (check_packed_git_idx(path, &idx_size, &idx_map))
 545                 return NULL;
 546
 547         /* do we have a corresponding .pack file? */
 548         strcpy(path + path_len - 4, ".pack");
 549         if (stat(path, &st) || !S_ISREG(st.st_mode)) {
 550                 munmap(idx_map, idx_size);
 551                 return NULL;
 552         }
 553         /* ok, it looks sane as far as we can check without
 554          * actually mapping the pack file.
 555          */
 556         p = xmalloc(sizeof(*p) + path_len + 2);
 557         strcpy(p->pack_name, path);
 558         p->index_size = idx_size;
 559         p->pack_size = st.st_size;
 560         p->index_base = idx_map;
 561         p->next = NULL;
 562         p->pack_base = NULL;
 563         p->pack_last_used = 0;
 564         p->pack_use_cnt = 0;
 565         p->pack_local = local;
 566         if ((path_len > 44) && !get_sha1_hex(path + path_len - 44, sha1))
 567                 hashcpy(p->sha1, sha1);
 568         return p;
 569 }
 570
 571 struct packed_git *parse_pack_index(unsigned char *sha1)
 572 {
 573         char *path = sha1_pack_index_name(sha1);
 574         return parse_pack_index_file(sha1, path);
 575 }
 576
 577 struct packed_git *parse_pack_index_file(const unsigned char *sha1, char *idx_path)
 578 {
 579         struct packed_git *p;
 580         unsigned long idx_size;
 581         void *idx_map;
 582         char *path;
 583
 584         if (check_packed_git_idx(idx_path, &idx_size, &idx_map))
 585                 return NULL;
 586
 587         path = sha1_pack_name(sha1);
 588
 589         p = xmalloc(sizeof(*p) + strlen(path) + 2);
 590         strcpy(p->pack_name, path);
 591         p->index_size = idx_size;
 592         p->pack_size = 0;
 593         p->index_base = idx_map;
 594         p->next = NULL;
 595         p->pack_base = NULL;
 596         p->pack_last_used = 0;
 597         p->pack_use_cnt = 0;
 598         hashcpy(p->sha1, sha1);
 599         return p;
 600 }
 601
 602 void install_packed_git(struct packed_git *pack)
 603 {
 604         pack->next = packed_git;
 605         packed_git = pack;
 606 }
 607
 608 static void prepare_packed_git_one(char *objdir, int local)
 609 {
 610         char path[PATH_MAX];
 611         int len;
 612         DIR *dir;
 613         struct dirent *de;
 614
 615         sprintf(path, "%s/pack", objdir);
 616         len = strlen(path);
 617         dir = opendir(path);
 618         if (!dir) {
 619                 if (errno != ENOENT)
 620                         error("unable to open object pack directory: %s: %s",
 621                               path, strerror(errno));
 622                 return;
 623         }
 624         path[len++] = '/';
 625         while ((de = readdir(dir)) != NULL) {
 626                 int namelen = strlen(de->d_name);
 627                 struct packed_git *p;
 628
 629                 if (!has_extension(de->d_name, ".idx"))
 630                         continue;
 631
 632                 /* we have .idx.  Is it a file we can map? */
 633                 strcpy(path + len, de->d_name);
 634                 for (p = packed_git; p; p = p->next) {
 635                         if (!memcmp(path, p->pack_name, len + namelen - 4))
 636                                 break;
 637                 }
 638                 if (p)
 639                         continue;
 640                 p = add_packed_git(path, len + namelen, local);
 641                 if (!p)
 642                         continue;
 643                 p->next = packed_git;
 644                 packed_git = p;
 645         }
 646         closedir(dir);
 647 }
 648
 649 static int prepare_packed_git_run_once = 0;
 650 void prepare_packed_git(void)
 651 {
 652         struct alternate_object_database *alt;
 653
 654         if (prepare_packed_git_run_once)
 655                 return;
 656         prepare_packed_git_one(get_object_directory(), 1);
 657         prepare_alt_odb();
 658         for (alt = alt_odb_list; alt; alt = alt->next) {
 659                 alt->name[-1] = 0;
 660                 prepare_packed_git_one(alt->base, 0);
 661                 alt->name[-1] = '/';
 662         }
 663         prepare_packed_git_run_once = 1;
 664 }
 665
 666 void reprepare_packed_git(void)
 667 {
 668         prepare_packed_git_run_once = 0;
 669         prepare_packed_git();
 670 }
 671
 672 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 673 {
 674         unsigned char real_sha1[20];
 675         hash_sha1_file(map, size, type, real_sha1);
 676         return hashcmp(sha1, real_sha1) ? -1 : 0;
 677 }
 678
 679 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 680 {
 681         struct stat st;
 682         void *map;
 683         int fd;
 684         char *filename = find_sha1_file(sha1, &st);
 685
 686         if (!filename) {
 687                 return NULL;
 688         }
 689
 690         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 691         if (fd < 0) {
 692                 /* See if it works without O_NOATIME */
 693                 switch (sha1_file_open_flag) {
 694                 default:
 695                         fd = open(filename, O_RDONLY);
 696                         if (fd >= 0)
 697                                 break;
 698                 /* Fallthrough */
 699                 case 0:
 700                         return NULL;
 701                 }
 702
 703                 /* If it failed once, it will probably fail again.
 704                  * Stop using O_NOATIME
 705                  */
 706                 sha1_file_open_flag = 0;
 707         }
 708         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 709         close(fd);
 710         if (map == MAP_FAILED)
 711                 return NULL;
 712         *size = st.st_size;
 713         return map;
 714 }
 715
 716 int legacy_loose_object(unsigned char *map)
 717 {
 718         unsigned int word;
 719
 720         /*
 721          * Is it a zlib-compressed buffer? If so, the first byte
 722          * must be 0x78 (15-bit window size, deflated), and the
 723          * first 16-bit word is evenly divisible by 31
 724          */
 725         word = (map[0] << 8) + map[1];
 726         if (map[0] == 0x78 && !(word % 31))
 727                 return 1;
 728         else
 729                 return 0;
 730 }
 731
 732 unsigned long unpack_object_header_gently(const unsigned char *buf, unsigned long len, enum object_type *type, unsigned long *sizep)
 733 {
 734         unsigned shift;
 735         unsigned char c;
 736         unsigned long size;
 737         unsigned long used = 0;
 738
 739         c = buf[used++];
 740         *type = (c >> 4) & 7;
 741         size = c & 15;
 742         shift = 4;
 743         while (c & 0x80) {
 744                 if (len <= used)
 745                         return 0;
 746                 if (sizeof(long) * 8 <= shift)
 747                         return 0;
 748                 c = buf[used++];
 749                 size += (c & 0x7f) << shift;
 750                 shift += 7;
 751         }
 752         *sizep = size;
 753         return used;
 754 }
 755
 756 static int unpack_sha1_header(z_stream *stream, unsigned char *map, unsigned long mapsize, void *buffer, unsigned long bufsiz)
 757 {
 758         unsigned long size, used;
 759         static const char valid_loose_object_type[8] = {
 760                 0, /* OBJ_EXT */
 761                 1, 1, 1, 1, /* "commit", "tree", "blob", "tag" */
 762                 0, /* "delta" and others are invalid in a loose object */
 763         };
 764         enum object_type type;
 765
 766         /* Get the data stream */
 767         memset(stream, 0, sizeof(*stream));
 768         stream->next_in = map;
 769         stream->avail_in = mapsize;
 770         stream->next_out = buffer;
 771         stream->avail_out = bufsiz;
 772
 773         if (legacy_loose_object(map)) {
 774                 inflateInit(stream);
 775                 return inflate(stream, 0);
 776         }
 777
 778         used = unpack_object_header_gently(map, mapsize, &type, &size);
 779         if (!used || !valid_loose_object_type[type])
 780                 return -1;
 781         map += used;
 782         mapsize -= used;
 783
 784         /* Set up the stream for the rest.. */
 785         stream->next_in = map;
 786         stream->avail_in = mapsize;
 787         inflateInit(stream);
 788
 789         /* And generate the fake traditional header */
 790         stream->total_out = 1 + snprintf(buffer, bufsiz, "%s %lu",
 791                                          type_names[type], size);
 792         return 0;
 793 }
 794
 795 static void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 796 {
 797         int bytes = strlen(buffer) + 1;
 798         unsigned char *buf = xmalloc(1+size);
 799         unsigned long n;
 800
 801         n = stream->total_out - bytes;
 802         if (n > size)
 803                 n = size;
 804         memcpy(buf, (char *) buffer + bytes, n);
 805         bytes = n;
 806         if (bytes < size) {
 807                 stream->next_out = buf + bytes;
 808                 stream->avail_out = size - bytes;
 809                 while (inflate(stream, Z_FINISH) == Z_OK)
 810                         /* nothing */;
 811         }
 812         buf[size] = 0;
 813         inflateEnd(stream);
 814         return buf;
 815 }
 816
 817 /*
 818  * We used to just use "sscanf()", but that's actually way
 819  * too permissive for what we want to check. So do an anal
 820  * object header parse by hand.
 821  */
 822 static int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 823 {
 824         int i;
 825         unsigned long size;
 826
 827         /*
 828          * The type can be at most ten bytes (including the
 829          * terminating '\0' that we add), and is followed by
 830          * a space.
 831          */
 832         i = 10;
 833         for (;;) {
 834                 char c = *hdr++;
 835                 if (c == ' ')
 836                         break;
 837                 if (!--i)
 838                         return -1;
 839                 *type++ = c;
 840         }
 841         *type = 0;
 842
 843         /*
 844          * The length must follow immediately, and be in canonical
 845          * decimal format (ie "010" is not valid).
 846          */
 847         size = *hdr++ - '0';
 848         if (size > 9)
 849                 return -1;
 850         if (size) {
 851                 for (;;) {
 852                         unsigned long c = *hdr - '0';
 853                         if (c > 9)
 854                                 break;
 855                         hdr++;
 856                         size = size * 10 + c;
 857                 }
 858         }
 859         *sizep = size;
 860
 861         /*
 862          * The length must be followed by a zero byte
 863          */
 864         return *hdr ? -1 : 0;
 865 }
 866
 867 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 868 {
 869         int ret;
 870         z_stream stream;
 871         char hdr[8192];
 872
 873         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
 874         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
 875                 return NULL;
 876
 877         return unpack_sha1_rest(&stream, hdr, *size);
 878 }
 879
 880 static unsigned long get_delta_base(struct packed_git *p,
 881                                     unsigned long offset,
 882                                     enum object_type kind,
 883                                     unsigned long delta_obj_offset,
 884                                     unsigned long *base_obj_offset)
 885 {
 886         unsigned char *base_info = (unsigned char *) p->pack_base + offset;
 887         unsigned long base_offset;
 888
 889         /* there must be at least 20 bytes left regardless of delta type */
 890         if (p->pack_size <= offset + 20)
 891                 die("truncated pack file");
 892
 893         if (kind == OBJ_OFS_DELTA) {
 894                 unsigned used = 0;
 895                 unsigned char c = base_info[used++];
 896                 base_offset = c & 127;
 897                 while (c & 128) {
 898                         base_offset += 1;
 899                         if (!base_offset || base_offset & ~(~0UL >> 7))
 900                                 die("offset value overflow for delta base object");
 901                         c = base_info[used++];
 902                         base_offset = (base_offset << 7) + (c & 127);
 903                 }
 904                 base_offset = delta_obj_offset - base_offset;
 905                 if (base_offset >= delta_obj_offset)
 906                         die("delta base offset out of bound");
 907                 offset += used;
 908         } else if (kind == OBJ_REF_DELTA) {
 909                 /* The base entry _must_ be in the same pack */
 910                 base_offset = find_pack_entry_one(base_info, p);
 911                 if (!base_offset)
 912                         die("failed to find delta-pack base object %s",
 913                                 sha1_to_hex(base_info));
 914                 offset += 20;
 915         } else
 916                 die("I am totally screwed");
 917         *base_obj_offset = base_offset;
 918         return offset;
 919 }
 920
 921 /* forward declaration for a mutually recursive function */
 922 static int packed_object_info(struct packed_git *p, unsigned long offset,
 923                               char *type, unsigned long *sizep);
 924
 925 static int packed_delta_info(struct packed_git *p,
 926                              unsigned long offset,
 927                              enum object_type kind,
 928                              unsigned long obj_offset,
 929                              char *type,
 930                              unsigned long *sizep)
 931 {
 932         unsigned long base_offset;
 933
 934         offset = get_delta_base(p, offset, kind, obj_offset, &base_offset);
 935
 936         /* We choose to only get the type of the base object and
 937          * ignore potentially corrupt pack file that expects the delta
 938          * based on a base with a wrong size.  This saves tons of
 939          * inflate() calls.
 940          */
 941         if (packed_object_info(p, base_offset, type, NULL))
 942                 die("cannot get info for delta-pack base");
 943
 944         if (sizep) {
 945                 const unsigned char *data;
 946                 unsigned char delta_head[20];
 947                 unsigned long result_size;
 948                 z_stream stream;
 949                 int st;
 950
 951                 memset(&stream, 0, sizeof(stream));
 952
 953                 stream.next_in = (unsigned char *) p->pack_base + offset;
 954                 stream.avail_in = p->pack_size - offset;
 955                 stream.next_out = delta_head;
 956                 stream.avail_out = sizeof(delta_head);
 957
 958                 inflateInit(&stream);
 959                 st = inflate(&stream, Z_FINISH);
 960                 inflateEnd(&stream);
 961                 if ((st != Z_STREAM_END) &&
 962                     stream.total_out != sizeof(delta_head))
 963                         die("delta data unpack-initial failed");
 964
 965                 /* Examine the initial part of the delta to figure out
 966                  * the result size.
 967                  */
 968                 data = delta_head;
 969
 970                 /* ignore base size */
 971                 get_delta_hdr_size(&data, delta_head+sizeof(delta_head));
 972
 973                 /* Read the result size */
 974                 result_size = get_delta_hdr_size(&data, delta_head+sizeof(delta_head));
 975                 *sizep = result_size;
 976         }
 977         return 0;
 978 }
 979
 980 static unsigned long unpack_object_header(struct packed_git *p, unsigned long offset,
 981         enum object_type *type, unsigned long *sizep)
 982 {
 983         unsigned long used;
 984
 985         if (p->pack_size <= offset)
 986                 die("object offset outside of pack file");
 987
 988         used = unpack_object_header_gently((unsigned char *)p->pack_base +
 989                                            offset,
 990                                            p->pack_size - offset, type, sizep);
 991         if (!used)
 992                 die("object offset outside of pack file");
 993
 994         return offset + used;
 995 }
 996
 997 void packed_object_info_detail(struct packed_git *p,
 998                                unsigned long offset,
 999                                char *type,
1000                                unsigned long *size,
1001                                unsigned long *store_size,
1002                                unsigned int *delta_chain_length,
1003                                unsigned char *base_sha1)
1004 {
1005         unsigned long obj_offset, val;
1006         unsigned char *next_sha1;
1007         enum object_type kind;
1008
1009         *delta_chain_length = 0;
1010         obj_offset = offset;
1011         offset = unpack_object_header(p, offset, &kind, size);
1012
1013         for (;;) {
1014                 switch (kind) {
1015                 default:
1016                         die("corrupted pack file %s containing object of kind %d",
1017                             p->pack_name, kind);
1018                 case OBJ_COMMIT:
1019                 case OBJ_TREE:
1020                 case OBJ_BLOB:
1021                 case OBJ_TAG:
1022                         strcpy(type, type_names[kind]);
1023                         *store_size = 0; /* notyet */
1024                         return;
1025                 case OBJ_OFS_DELTA:
1026                         get_delta_base(p, offset, kind, obj_offset, &offset);
1027                         if (*delta_chain_length == 0) {
1028                                 /* TODO: find base_sha1 as pointed by offset */
1029                         }
1030                         break;
1031                 case OBJ_REF_DELTA:
1032                         if (p->pack_size <= offset + 20)
1033                                 die("pack file %s records an incomplete delta base",
1034                                     p->pack_name);
1035                         next_sha1 = (unsigned char *) p->pack_base + offset;
1036                         if (*delta_chain_length == 0)
1037                                 hashcpy(base_sha1, next_sha1);
1038                         offset = find_pack_entry_one(next_sha1, p);
1039                         break;
1040                 }
1041                 obj_offset = offset;
1042                 offset = unpack_object_header(p, offset, &kind, &val);
1043                 (*delta_chain_length)++;
1044         }
1045 }
1046
1047 static int packed_object_info(struct packed_git *p, unsigned long offset,
1048                               char *type, unsigned long *sizep)
1049 {
1050         unsigned long size, obj_offset = offset;
1051         enum object_type kind;
1052
1053         offset = unpack_object_header(p, offset, &kind, &size);
1054
1055         switch (kind) {
1056         case OBJ_OFS_DELTA:
1057         case OBJ_REF_DELTA:
1058                 return packed_delta_info(p, offset, kind, obj_offset, type, sizep);
1059         case OBJ_COMMIT:
1060         case OBJ_TREE:
1061         case OBJ_BLOB:
1062         case OBJ_TAG:
1063                 strcpy(type, type_names[kind]);
1064                 break;
1065         default:
1066                 die("corrupted pack file %s containing object of kind %d",
1067                     p->pack_name, kind);
1068         }
1069         if (sizep)
1070                 *sizep = size;
1071         return 0;
1072 }
1073
1074 static void *unpack_compressed_entry(struct packed_git *p,
1075                                     unsigned long offset,
1076                                     unsigned long size)
1077 {
1078         int st;
1079         z_stream stream;
1080         unsigned char *buffer;
1081
1082         buffer = xmalloc(size + 1);
1083         buffer[size] = 0;
1084         memset(&stream, 0, sizeof(stream));
1085         stream.next_in = (unsigned char*)p->pack_base + offset;
1086         stream.avail_in = p->pack_size - offset;
1087         stream.next_out = buffer;
1088         stream.avail_out = size;
1089
1090         inflateInit(&stream);
1091         st = inflate(&stream, Z_FINISH);
1092         inflateEnd(&stream);
1093         if ((st != Z_STREAM_END) || stream.total_out != size) {
1094                 free(buffer);
1095                 return NULL;
1096         }
1097
1098         return buffer;
1099 }
1100
1101 static void *unpack_delta_entry(struct packed_git *p,
1102                                 unsigned long offset,
1103                                 unsigned long delta_size,
1104                                 enum object_type kind,
1105                                 unsigned long obj_offset,
1106                                 char *type,
1107                                 unsigned long *sizep)
1108 {
1109         void *delta_data, *result, *base;
1110         unsigned long result_size, base_size, base_offset;
1111
1112         offset = get_delta_base(p, offset, kind, obj_offset, &base_offset);
1113         base = unpack_entry_gently(p, base_offset, type, &base_size);
1114         if (!base)
1115                 die("failed to read delta base object at %lu from %s",
1116                     base_offset, p->pack_name);
1117
1118         delta_data = unpack_compressed_entry(p, offset, delta_size);
1119         result = patch_delta(base, base_size,
1120                              delta_data, delta_size,
1121                              &result_size);
1122         if (!result)
1123                 die("failed to apply delta");
1124         free(delta_data);
1125         free(base);
1126         *sizep = result_size;
1127         return result;
1128 }
1129
1130 static void *unpack_entry(struct pack_entry *entry,
1131                           char *type, unsigned long *sizep)
1132 {
1133         struct packed_git *p = entry->p;
1134         void *retval;
1135
1136         if (use_packed_git(p))
1137                 die("cannot map packed file");
1138         retval = unpack_entry_gently(p, entry->offset, type, sizep);
1139         unuse_packed_git(p);
1140         if (!retval)
1141                 die("corrupted pack file %s", p->pack_name);
1142         return retval;
1143 }
1144
1145 /* The caller is responsible for use_packed_git()/unuse_packed_git() pair */
1146 void *unpack_entry_gently(struct packed_git *p, unsigned long offset,
1147                           char *type, unsigned long *sizep)
1148 {
1149         unsigned long size, obj_offset = offset;
1150         enum object_type kind;
1151
1152         offset = unpack_object_header(p, offset, &kind, &size);
1153         switch (kind) {
1154         case OBJ_OFS_DELTA:
1155         case OBJ_REF_DELTA:
1156                 return unpack_delta_entry(p, offset, size, kind, obj_offset, type, sizep);
1157         case OBJ_COMMIT:
1158         case OBJ_TREE:
1159         case OBJ_BLOB:
1160         case OBJ_TAG:
1161                 strcpy(type, type_names[kind]);
1162                 *sizep = size;
1163                 return unpack_compressed_entry(p, offset, size);
1164         default:
1165                 return NULL;
1166         }
1167 }
1168
1169 int num_packed_objects(const struct packed_git *p)
1170 {
1171         /* See check_packed_git_idx() */
1172         return (p->index_size - 20 - 20 - 4*256) / 24;
1173 }
1174
1175 int nth_packed_object_sha1(const struct packed_git *p, int n,
1176                            unsigned char* sha1)
1177 {
1178         void *index = p->index_base + 256;
1179         if (n < 0 || num_packed_objects(p) <= n)
1180                 return -1;
1181         hashcpy(sha1, (unsigned char *) index + (24 * n) + 4);
1182         return 0;
1183 }
1184
1185 unsigned long find_pack_entry_one(const unsigned char *sha1,
1186                                   struct packed_git *p)
1187 {
1188         unsigned int *level1_ofs = p->index_base;
1189         int hi = ntohl(level1_ofs[*sha1]);
1190         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
1191         void *index = p->index_base + 256;
1192
1193         do {
1194                 int mi = (lo + hi) / 2;
1195                 int cmp = hashcmp((unsigned char *)index + (24 * mi) + 4, sha1);
1196                 if (!cmp)
1197                         return ntohl(*((unsigned int *) ((char *) index + (24 * mi))));
1198                 if (cmp > 0)
1199                         hi = mi;
1200                 else
1201                         lo = mi+1;
1202         } while (lo < hi);
1203         return 0;
1204 }
1205
1206 static int matches_pack_name(struct packed_git *p, const char *ig)
1207 {
1208         const char *last_c, *c;
1209
1210         if (!strcmp(p->pack_name, ig))
1211                 return 0;
1212
1213         for (c = p->pack_name, last_c = c; *c;)
1214                 if (*c == '/')
1215                         last_c = ++c;
1216                 else
1217                         ++c;
1218         if (!strcmp(last_c, ig))
1219                 return 0;
1220
1221         return 1;
1222 }
1223
1224 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e, const char **ignore_packed)
1225 {
1226         struct packed_git *p;
1227         unsigned long offset;
1228
1229         prepare_packed_git();
1230
1231         for (p = packed_git; p; p = p->next) {
1232                 if (ignore_packed) {
1233                         const char **ig;
1234                         for (ig = ignore_packed; *ig; ig++)
1235                                 if (!matches_pack_name(p, *ig))
1236                                         break;
1237                         if (*ig)
1238                                 continue;
1239                 }
1240                 offset = find_pack_entry_one(sha1, p);
1241                 if (offset) {
1242                         e->offset = offset;
1243                         e->p = p;
1244                         hashcpy(e->sha1, sha1);
1245                         return 1;
1246                 }
1247         }
1248         return 0;
1249 }
1250
1251 struct packed_git *find_sha1_pack(const unsigned char *sha1,
1252                                   struct packed_git *packs)
1253 {
1254         struct packed_git *p;
1255
1256         for (p = packs; p; p = p->next) {
1257                 if (find_pack_entry_one(sha1, p))
1258                         return p;
1259         }
1260         return NULL;
1261
1262 }
1263
1264 static int sha1_loose_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
1265 {
1266         int status;
1267         unsigned long mapsize, size;
1268         void *map;
1269         z_stream stream;
1270         char hdr[128];
1271
1272         map = map_sha1_file(sha1, &mapsize);
1273         if (!map)
1274                 return error("unable to find %s", sha1_to_hex(sha1));
1275         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
1276                 status = error("unable to unpack %s header",
1277                                sha1_to_hex(sha1));
1278         if (parse_sha1_header(hdr, type, &size) < 0)
1279                 status = error("unable to parse %s header", sha1_to_hex(sha1));
1280         else {
1281                 status = 0;
1282                 if (sizep)
1283                         *sizep = size;
1284         }
1285         inflateEnd(&stream);
1286         munmap(map, mapsize);
1287         return status;
1288 }
1289
1290 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
1291 {
1292         int status;
1293         struct pack_entry e;
1294
1295         if (!find_pack_entry(sha1, &e, NULL)) {
1296                 reprepare_packed_git();
1297                 if (!find_pack_entry(sha1, &e, NULL))
1298                         return sha1_loose_object_info(sha1, type, sizep);
1299         }
1300         if (use_packed_git(e.p))
1301                 die("cannot map packed file");
1302         status = packed_object_info(e.p, e.offset, type, sizep);
1303         unuse_packed_git(e.p);
1304         return status;
1305 }
1306
1307 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
1308 {
1309         struct pack_entry e;
1310
1311         if (!find_pack_entry(sha1, &e, NULL)) {
1312                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
1313                 return NULL;
1314         }
1315         return unpack_entry(&e, type, size);
1316 }
1317
1318 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
1319 {
1320         unsigned long mapsize;
1321         void *map, *buf;
1322         struct pack_entry e;
1323
1324         if (find_pack_entry(sha1, &e, NULL))
1325                 return read_packed_sha1(sha1, type, size);
1326         map = map_sha1_file(sha1, &mapsize);
1327         if (map) {
1328                 buf = unpack_sha1_file(map, mapsize, type, size);
1329                 munmap(map, mapsize);
1330                 return buf;
1331         }
1332         reprepare_packed_git();
1333         if (find_pack_entry(sha1, &e, NULL))
1334                 return read_packed_sha1(sha1, type, size);
1335         return NULL;
1336 }
1337
1338 void *read_object_with_reference(const unsigned char *sha1,
1339                                  const char *required_type,
1340                                  unsigned long *size,
1341                                  unsigned char *actual_sha1_return)
1342 {
1343         char type[20];
1344         void *buffer;
1345         unsigned long isize;
1346         unsigned char actual_sha1[20];
1347
1348         hashcpy(actual_sha1, sha1);
1349         while (1) {
1350                 int ref_length = -1;
1351                 const char *ref_type = NULL;
1352
1353                 buffer = read_sha1_file(actual_sha1, type, &isize);
1354                 if (!buffer)
1355                         return NULL;
1356                 if (!strcmp(type, required_type)) {
1357                         *size = isize;
1358                         if (actual_sha1_return)
1359                                 hashcpy(actual_sha1_return, actual_sha1);
1360                         return buffer;
1361                 }
1362                 /* Handle references */
1363                 else if (!strcmp(type, commit_type))
1364                         ref_type = "tree ";
1365                 else if (!strcmp(type, tag_type))
1366                         ref_type = "object ";
1367                 else {
1368                         free(buffer);
1369                         return NULL;
1370                 }
1371                 ref_length = strlen(ref_type);
1372
1373                 if (memcmp(buffer, ref_type, ref_length) ||
1374                     get_sha1_hex((char *) buffer + ref_length, actual_sha1)) {
1375                         free(buffer);
1376                         return NULL;
1377                 }
1378                 free(buffer);
1379                 /* Now we have the ID of the referred-to object in
1380                  * actual_sha1.  Check again. */
1381         }
1382 }
1383
1384 static void write_sha1_file_prepare(void *buf, unsigned long len,
1385                                     const char *type, unsigned char *sha1,
1386                                     unsigned char *hdr, int *hdrlen)
1387 {
1388         SHA_CTX c;
1389
1390         /* Generate the header */
1391         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
1392
1393         /* Sha1.. */
1394         SHA1_Init(&c);
1395         SHA1_Update(&c, hdr, *hdrlen);
1396         SHA1_Update(&c, buf, len);
1397         SHA1_Final(sha1, &c);
1398 }
1399
1400 /*
1401  * Link the tempfile to the final place, possibly creating the
1402  * last directory level as you do so.
1403  *
1404  * Returns the errno on failure, 0 on success.
1405  */
1406 static int link_temp_to_file(const char *tmpfile, const char *filename)
1407 {
1408         int ret;
1409         char *dir;
1410
1411         if (!link(tmpfile, filename))
1412                 return 0;
1413
1414         /*
1415          * Try to mkdir the last path component if that failed.
1416          *
1417          * Re-try the "link()" regardless of whether the mkdir
1418          * succeeds, since a race might mean that somebody
1419          * else succeeded.
1420          */
1421         ret = errno;
1422         dir = strrchr(filename, '/');
1423         if (dir) {
1424                 *dir = 0;
1425                 if (!mkdir(filename, 0777) && adjust_shared_perm(filename)) {
1426                         *dir = '/';
1427                         return -2;
1428                 }
1429                 *dir = '/';
1430                 if (!link(tmpfile, filename))
1431                         return 0;
1432                 ret = errno;
1433         }
1434         return ret;
1435 }
1436
1437 /*
1438  * Move the just written object into its final resting place
1439  */
1440 int move_temp_to_file(const char *tmpfile, const char *filename)
1441 {
1442         int ret = link_temp_to_file(tmpfile, filename);
1443
1444         /*
1445          * Coda hack - coda doesn't like cross-directory links,
1446          * so we fall back to a rename, which will mean that it
1447          * won't be able to check collisions, but that's not a
1448          * big deal.
1449          *
1450          * The same holds for FAT formatted media.
1451          *
1452          * When this succeeds, we just return 0. We have nothing
1453          * left to unlink.
1454          */
1455         if (ret && ret != EEXIST) {
1456                 if (!rename(tmpfile, filename))
1457                         return 0;
1458                 ret = errno;
1459         }
1460         unlink(tmpfile);
1461         if (ret) {
1462                 if (ret != EEXIST) {
1463                         return error("unable to write sha1 filename %s: %s\n", filename, strerror(ret));
1464                 }
1465                 /* FIXME!!! Collision check here ? */
1466         }
1467
1468         return 0;
1469 }
1470
1471 static int write_buffer(int fd, const void *buf, size_t len)
1472 {
1473         while (len) {
1474                 ssize_t size;
1475
1476                 size = write(fd, buf, len);
1477                 if (!size)
1478                         return error("file write: disk full");
1479                 if (size < 0) {
1480                         if (errno == EINTR || errno == EAGAIN)
1481                                 continue;
1482                         return error("file write error (%s)", strerror(errno));
1483                 }
1484                 len -= size;
1485                 buf = (char *) buf + size;
1486         }
1487         return 0;
1488 }
1489
1490 static int write_binary_header(unsigned char *hdr, enum object_type type, unsigned long len)
1491 {
1492         int hdr_len;
1493         unsigned char c;
1494
1495         c = (type << 4) | (len & 15);
1496         len >>= 4;
1497         hdr_len = 1;
1498         while (len) {
1499                 *hdr++ = c | 0x80;
1500                 hdr_len++;
1501                 c = (len & 0x7f);
1502                 len >>= 7;
1503         }
1504         *hdr = c;
1505         return hdr_len;
1506 }
1507
1508 static void setup_object_header(z_stream *stream, const char *type, unsigned long len)
1509 {
1510         int obj_type, hdr;
1511
1512         if (use_legacy_headers) {
1513                 while (deflate(stream, 0) == Z_OK)
1514                         /* nothing */;
1515                 return;
1516         }
1517         if (!strcmp(type, blob_type))
1518                 obj_type = OBJ_BLOB;
1519         else if (!strcmp(type, tree_type))
1520                 obj_type = OBJ_TREE;
1521         else if (!strcmp(type, commit_type))
1522                 obj_type = OBJ_COMMIT;
1523         else if (!strcmp(type, tag_type))
1524                 obj_type = OBJ_TAG;
1525         else
1526                 die("trying to generate bogus object of type '%s'", type);
1527         hdr = write_binary_header(stream->next_out, obj_type, len);
1528         stream->total_out = hdr;
1529         stream->next_out += hdr;
1530         stream->avail_out -= hdr;
1531 }
1532
1533 int hash_sha1_file(void *buf, unsigned long len, const char *type,
1534                    unsigned char *sha1)
1535 {
1536         unsigned char hdr[50];
1537         int hdrlen;
1538         write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1539         return 0;
1540 }
1541
1542 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1543 {
1544         int size;
1545         unsigned char *compressed;
1546         z_stream stream;
1547         unsigned char sha1[20];
1548         char *filename;
1549         static char tmpfile[PATH_MAX];
1550         unsigned char hdr[50];
1551         int fd, hdrlen;
1552
1553         /* Normally if we have it in the pack then we do not bother writing
1554          * it out into .git/objects/??/?{38} file.
1555          */
1556         write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1557         filename = sha1_file_name(sha1);
1558         if (returnsha1)
1559                 hashcpy(returnsha1, sha1);
1560         if (has_sha1_file(sha1))
1561                 return 0;
1562         fd = open(filename, O_RDONLY);
1563         if (fd >= 0) {
1564                 /*
1565                  * FIXME!!! We might do collision checking here, but we'd
1566                  * need to uncompress the old file and check it. Later.
1567                  */
1568                 close(fd);
1569                 return 0;
1570         }
1571
1572         if (errno != ENOENT) {
1573                 return error("sha1 file %s: %s\n", filename, strerror(errno));
1574         }
1575
1576         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1577
1578         fd = mkstemp(tmpfile);
1579         if (fd < 0) {
1580                 if (errno == EPERM)
1581                         return error("insufficient permission for adding an object to repository database %s\n", get_object_directory());
1582                 else
1583                         return error("unable to create temporary sha1 filename %s: %s\n", tmpfile, strerror(errno));
1584         }
1585
1586         /* Set it up */
1587         memset(&stream, 0, sizeof(stream));
1588         deflateInit(&stream, zlib_compression_level);
1589         size = 8 + deflateBound(&stream, len+hdrlen);
1590         compressed = xmalloc(size);
1591
1592         /* Compress it */
1593         stream.next_out = compressed;
1594         stream.avail_out = size;
1595
1596         /* First header.. */
1597         stream.next_in = hdr;
1598         stream.avail_in = hdrlen;
1599         setup_object_header(&stream, type, len);
1600
1601         /* Then the data itself.. */
1602         stream.next_in = buf;
1603         stream.avail_in = len;
1604         while (deflate(&stream, Z_FINISH) == Z_OK)
1605                 /* nothing */;
1606         deflateEnd(&stream);
1607         size = stream.total_out;
1608
1609         if (write_buffer(fd, compressed, size) < 0)
1610                 die("unable to write sha1 file");
1611         fchmod(fd, 0444);
1612         close(fd);
1613         free(compressed);
1614
1615         return move_temp_to_file(tmpfile, filename);
1616 }
1617
1618 /*
1619  * We need to unpack and recompress the object for writing
1620  * it out to a different file.
1621  */
1622 static void *repack_object(const unsigned char *sha1, unsigned long *objsize)
1623 {
1624         size_t size;
1625         z_stream stream;
1626         unsigned char *unpacked;
1627         unsigned long len;
1628         char type[20];
1629         char hdr[50];
1630         int hdrlen;
1631         void *buf;
1632
1633         /* need to unpack and recompress it by itself */
1634         unpacked = read_packed_sha1(sha1, type, &len);
1635
1636         hdrlen = sprintf(hdr, "%s %lu", type, len) + 1;
1637
1638         /* Set it up */
1639         memset(&stream, 0, sizeof(stream));
1640         deflateInit(&stream, zlib_compression_level);
1641         size = deflateBound(&stream, len + hdrlen);
1642         buf = xmalloc(size);
1643
1644         /* Compress it */
1645         stream.next_out = buf;
1646         stream.avail_out = size;
1647
1648         /* First header.. */
1649         stream.next_in = (void *)hdr;
1650         stream.avail_in = hdrlen;
1651         while (deflate(&stream, 0) == Z_OK)
1652                 /* nothing */;
1653
1654         /* Then the data itself.. */
1655         stream.next_in = unpacked;
1656         stream.avail_in = len;
1657         while (deflate(&stream, Z_FINISH) == Z_OK)
1658                 /* nothing */;
1659         deflateEnd(&stream);
1660         free(unpacked);
1661
1662         *objsize = stream.total_out;
1663         return buf;
1664 }
1665
1666 int write_sha1_to_fd(int fd, const unsigned char *sha1)
1667 {
1668         int retval;
1669         unsigned long objsize;
1670         void *buf = map_sha1_file(sha1, &objsize);
1671
1672         if (buf) {
1673                 retval = write_buffer(fd, buf, objsize);
1674                 munmap(buf, objsize);
1675                 return retval;
1676         }
1677
1678         buf = repack_object(sha1, &objsize);
1679         retval = write_buffer(fd, buf, objsize);
1680         free(buf);
1681         return retval;
1682 }
1683
1684 int write_sha1_from_fd(const unsigned char *sha1, int fd, char *buffer,
1685                        size_t bufsize, size_t *bufposn)
1686 {
1687         char tmpfile[PATH_MAX];
1688         int local;
1689         z_stream stream;
1690         unsigned char real_sha1[20];
1691         unsigned char discard[4096];
1692         int ret;
1693         SHA_CTX c;
1694
1695         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1696
1697         local = mkstemp(tmpfile);
1698         if (local < 0) {
1699                 if (errno == EPERM)
1700                         return error("insufficient permission for adding an object to repository database %s\n", get_object_directory());
1701                 else
1702                         return error("unable to create temporary sha1 filename %s: %s\n", tmpfile, strerror(errno));
1703         }
1704
1705         memset(&stream, 0, sizeof(stream));
1706
1707         inflateInit(&stream);
1708
1709         SHA1_Init(&c);
1710
1711         do {
1712                 ssize_t size;
1713                 if (*bufposn) {
1714                         stream.avail_in = *bufposn;
1715                         stream.next_in = (unsigned char *) buffer;
1716                         do {
1717                                 stream.next_out = discard;
1718                                 stream.avail_out = sizeof(discard);
1719                                 ret = inflate(&stream, Z_SYNC_FLUSH);
1720                                 SHA1_Update(&c, discard, sizeof(discard) -
1721                                             stream.avail_out);
1722                         } while (stream.avail_in && ret == Z_OK);
1723                         if (write_buffer(local, buffer, *bufposn - stream.avail_in) < 0)
1724                                 die("unable to write sha1 file");
1725                         memmove(buffer, buffer + *bufposn - stream.avail_in,
1726                                 stream.avail_in);
1727                         *bufposn = stream.avail_in;
1728                         if (ret != Z_OK)
1729                                 break;
1730                 }
1731                 size = read(fd, buffer + *bufposn, bufsize - *bufposn);
1732                 if (size <= 0) {
1733                         close(local);
1734                         unlink(tmpfile);
1735                         if (!size)
1736                                 return error("Connection closed?");
1737                         perror("Reading from connection");
1738                         return -1;
1739                 }
1740                 *bufposn += size;
1741         } while (1);
1742         inflateEnd(&stream);
1743
1744         close(local);
1745         SHA1_Final(real_sha1, &c);
1746         if (ret != Z_STREAM_END) {
1747                 unlink(tmpfile);
1748                 return error("File %s corrupted", sha1_to_hex(sha1));
1749         }
1750         if (hashcmp(sha1, real_sha1)) {
1751                 unlink(tmpfile);
1752                 return error("File %s has bad hash", sha1_to_hex(sha1));
1753         }
1754
1755         return move_temp_to_file(tmpfile, sha1_file_name(sha1));
1756 }
1757
1758 int has_pack_index(const unsigned char *sha1)
1759 {
1760         struct stat st;
1761         if (stat(sha1_pack_index_name(sha1), &st))
1762                 return 0;
1763         return 1;
1764 }
1765
1766 int has_pack_file(const unsigned char *sha1)
1767 {
1768         struct stat st;
1769         if (stat(sha1_pack_name(sha1), &st))
1770                 return 0;
1771         return 1;
1772 }
1773
1774 int has_sha1_pack(const unsigned char *sha1, const char **ignore_packed)
1775 {
1776         struct pack_entry e;
1777         return find_pack_entry(sha1, &e, ignore_packed);
1778 }
1779
1780 int has_sha1_file(const unsigned char *sha1)
1781 {
1782         struct stat st;
1783         struct pack_entry e;
1784
1785         if (find_pack_entry(sha1, &e, NULL))
1786                 return 1;
1787         return find_sha1_file(sha1, &st) ? 1 : 0;
1788 }
1789
1790 /*
1791  * reads from fd as long as possible into a supplied buffer of size bytes.
1792  * If necessary the buffer's size is increased using realloc()
1793  *
1794  * returns 0 if anything went fine and -1 otherwise
1795  *
1796  * NOTE: both buf and size may change, but even when -1 is returned
1797  * you still have to free() it yourself.
1798  */
1799 int read_pipe(int fd, char** return_buf, unsigned long* return_size)
1800 {
1801         char* buf = *return_buf;
1802         unsigned long size = *return_size;
1803         int iret;
1804         unsigned long off = 0;
1805
1806         do {
1807                 iret = xread(fd, buf + off, size - off);
1808                 if (iret > 0) {
1809                         off += iret;
1810                         if (off == size) {
1811                                 size *= 2;
1812                                 buf = xrealloc(buf, size);
1813                         }
1814                 }
1815         } while (iret > 0);
1816
1817         *return_buf = buf;
1818         *return_size = off;
1819
1820         if (iret < 0)
1821                 return -1;
1822         return 0;
1823 }
1824
1825 int index_pipe(unsigned char *sha1, int fd, const char *type, int write_object)
1826 {
1827         unsigned long size = 4096;
1828         char *buf = xmalloc(size);
1829         int ret;
1830
1831         if (read_pipe(fd, &buf, &size)) {
1832                 free(buf);
1833                 return -1;
1834         }
1835
1836         if (!type)
1837                 type = blob_type;
1838         if (write_object)
1839                 ret = write_sha1_file(buf, size, type, sha1);
1840         else
1841                 ret = hash_sha1_file(buf, size, type, sha1);
1842         free(buf);
1843         return ret;
1844 }
1845
1846 int index_fd(unsigned char *sha1, int fd, struct stat *st, int write_object, const char *type)
1847 {
1848         unsigned long size = st->st_size;
1849         void *buf;
1850         int ret;
1851
1852         buf = "";
1853         if (size)
1854                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1855         close(fd);
1856         if (buf == MAP_FAILED)
1857                 return -1;
1858
1859         if (!type)
1860                 type = blob_type;
1861         if (write_object)
1862                 ret = write_sha1_file(buf, size, type, sha1);
1863         else
1864                 ret = hash_sha1_file(buf, size, type, sha1);
1865         if (size)
1866                 munmap(buf, size);
1867         return ret;
1868 }
1869
1870 int index_path(unsigned char *sha1, const char *path, struct stat *st, int write_object)
1871 {
1872         int fd;
1873         char *target;
1874
1875         switch (st->st_mode & S_IFMT) {
1876         case S_IFREG:
1877                 fd = open(path, O_RDONLY);
1878                 if (fd < 0)
1879                         return error("open(\"%s\"): %s", path,
1880                                      strerror(errno));
1881                 if (index_fd(sha1, fd, st, write_object, NULL) < 0)
1882                         return error("%s: failed to insert into database",
1883                                      path);
1884                 break;
1885         case S_IFLNK:
1886                 target = xmalloc(st->st_size+1);
1887                 if (readlink(path, target, st->st_size+1) != st->st_size) {
1888                         char *errstr = strerror(errno);
1889                         free(target);
1890                         return error("readlink(\"%s\"): %s", path,
1891                                      errstr);
1892                 }
1893                 if (!write_object)
1894                         hash_sha1_file(target, st->st_size, blob_type, sha1);
1895                 else if (write_sha1_file(target, st->st_size, blob_type, sha1))
1896                         return error("%s: failed to insert into database",
1897                                      path);
1898                 free(target);
1899                 break;
1900         default:
1901                 return error("%s: unsupported file type", path);
1902         }
1903         return 0;
1904 }