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 static 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         char header[100];
 675         unsigned char real_sha1[20];
 676         SHA_CTX c;
 677
 678         SHA1_Init(&c);
 679         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 680         SHA1_Update(&c, map, size);
 681         SHA1_Final(real_sha1, &c);
 682         return hashcmp(sha1, real_sha1) ? -1 : 0;
 683 }
 684
 685 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 686 {
 687         struct stat st;
 688         void *map;
 689         int fd;
 690         char *filename = find_sha1_file(sha1, &st);
 691
 692         if (!filename) {
 693                 return NULL;
 694         }
 695
 696         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 697         if (fd < 0) {
 698                 /* See if it works without O_NOATIME */
 699                 switch (sha1_file_open_flag) {
 700                 default:
 701                         fd = open(filename, O_RDONLY);
 702                         if (fd >= 0)
 703                                 break;
 704                 /* Fallthrough */
 705                 case 0:
 706                         return NULL;
 707                 }
 708
 709                 /* If it failed once, it will probably fail again.
 710                  * Stop using O_NOATIME
 711                  */
 712                 sha1_file_open_flag = 0;
 713         }
 714         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 715         close(fd);
 716         if (map == MAP_FAILED)
 717                 return NULL;
 718         *size = st.st_size;
 719         return map;
 720 }
 721
 722 int legacy_loose_object(unsigned char *map)
 723 {
 724         unsigned int word;
 725
 726         /*
 727          * Is it a zlib-compressed buffer? If so, the first byte
 728          * must be 0x78 (15-bit window size, deflated), and the
 729          * first 16-bit word is evenly divisible by 31
 730          */
 731         word = (map[0] << 8) + map[1];
 732         if (map[0] == 0x78 && !(word % 31))
 733                 return 1;
 734         else
 735                 return 0;
 736 }
 737
 738 unsigned long unpack_object_header_gently(const unsigned char *buf, unsigned long len, enum object_type *type, unsigned long *sizep)
 739 {
 740         unsigned shift;
 741         unsigned char c;
 742         unsigned long size;
 743         unsigned long used = 0;
 744
 745         c = buf[used++];
 746         *type = (c >> 4) & 7;
 747         size = c & 15;
 748         shift = 4;
 749         while (c & 0x80) {
 750                 if (len <= used)
 751                         return 0;
 752                 if (sizeof(long) * 8 <= shift)
 753                         return 0;
 754                 c = buf[used++];
 755                 size += (c & 0x7f) << shift;
 756                 shift += 7;
 757         }
 758         *sizep = size;
 759         return used;
 760 }
 761
 762 static int unpack_sha1_header(z_stream *stream, unsigned char *map, unsigned long mapsize, void *buffer, unsigned long bufsiz)
 763 {
 764         unsigned long size, used;
 765         static const char valid_loose_object_type[8] = {
 766                 0, /* OBJ_EXT */
 767                 1, 1, 1, 1, /* "commit", "tree", "blob", "tag" */
 768                 0, /* "delta" and others are invalid in a loose object */
 769         };
 770         enum object_type type;
 771
 772         /* Get the data stream */
 773         memset(stream, 0, sizeof(*stream));
 774         stream->next_in = map;
 775         stream->avail_in = mapsize;
 776         stream->next_out = buffer;
 777         stream->avail_out = bufsiz;
 778
 779         if (legacy_loose_object(map)) {
 780                 inflateInit(stream);
 781                 return inflate(stream, 0);
 782         }
 783
 784         used = unpack_object_header_gently(map, mapsize, &type, &size);
 785         if (!used || !valid_loose_object_type[type])
 786                 return -1;
 787         map += used;
 788         mapsize -= used;
 789
 790         /* Set up the stream for the rest.. */
 791         stream->next_in = map;
 792         stream->avail_in = mapsize;
 793         inflateInit(stream);
 794
 795         /* And generate the fake traditional header */
 796         stream->total_out = 1 + snprintf(buffer, bufsiz, "%s %lu",
 797                                          type_names[type], size);
 798         return 0;
 799 }
 800
 801 static void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 802 {
 803         int bytes = strlen(buffer) + 1;
 804         unsigned char *buf = xmalloc(1+size);
 805         unsigned long n;
 806
 807         n = stream->total_out - bytes;
 808         if (n > size)
 809                 n = size;
 810         memcpy(buf, (char *) buffer + bytes, n);
 811         bytes = n;
 812         if (bytes < size) {
 813                 stream->next_out = buf + bytes;
 814                 stream->avail_out = size - bytes;
 815                 while (inflate(stream, Z_FINISH) == Z_OK)
 816                         /* nothing */;
 817         }
 818         buf[size] = 0;
 819         inflateEnd(stream);
 820         return buf;
 821 }
 822
 823 /*
 824  * We used to just use "sscanf()", but that's actually way
 825  * too permissive for what we want to check. So do an anal
 826  * object header parse by hand.
 827  */
 828 static int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 829 {
 830         int i;
 831         unsigned long size;
 832
 833         /*
 834          * The type can be at most ten bytes (including the
 835          * terminating '\0' that we add), and is followed by
 836          * a space.
 837          */
 838         i = 10;
 839         for (;;) {
 840                 char c = *hdr++;
 841                 if (c == ' ')
 842                         break;
 843                 if (!--i)
 844                         return -1;
 845                 *type++ = c;
 846         }
 847         *type = 0;
 848
 849         /*
 850          * The length must follow immediately, and be in canonical
 851          * decimal format (ie "010" is not valid).
 852          */
 853         size = *hdr++ - '0';
 854         if (size > 9)
 855                 return -1;
 856         if (size) {
 857                 for (;;) {
 858                         unsigned long c = *hdr - '0';
 859                         if (c > 9)
 860                                 break;
 861                         hdr++;
 862                         size = size * 10 + c;
 863                 }
 864         }
 865         *sizep = size;
 866
 867         /*
 868          * The length must be followed by a zero byte
 869          */
 870         return *hdr ? -1 : 0;
 871 }
 872
 873 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 874 {
 875         int ret;
 876         z_stream stream;
 877         char hdr[8192];
 878
 879         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
 880         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
 881                 return NULL;
 882
 883         return unpack_sha1_rest(&stream, hdr, *size);
 884 }
 885
 886 static unsigned long get_delta_base(struct packed_git *p,
 887                                     unsigned long offset,
 888                                     enum object_type kind,
 889                                     unsigned long delta_obj_offset,
 890                                     unsigned long *base_obj_offset)
 891 {
 892         unsigned char *base_info = (unsigned char *) p->pack_base + offset;
 893         unsigned long base_offset;
 894
 895         /* there must be at least 20 bytes left regardless of delta type */
 896         if (p->pack_size <= offset + 20)
 897                 die("truncated pack file");
 898
 899         if (kind == OBJ_OFS_DELTA) {
 900                 unsigned used = 0;
 901                 unsigned char c = base_info[used++];
 902                 base_offset = c & 127;
 903                 while (c & 128) {
 904                         base_offset += 1;
 905                         if (!base_offset || base_offset & ~(~0UL >> 7))
 906                                 die("offset value overflow for delta base object");
 907                         c = base_info[used++];
 908                         base_offset = (base_offset << 7) + (c & 127);
 909                 }
 910                 base_offset = delta_obj_offset - base_offset;
 911                 if (base_offset >= delta_obj_offset)
 912                         die("delta base offset out of bound");
 913                 offset += used;
 914         } else if (kind == OBJ_REF_DELTA) {
 915                 /* The base entry _must_ be in the same pack */
 916                 base_offset = find_pack_entry_one(base_info, p);
 917                 if (!base_offset)
 918                         die("failed to find delta-pack base object %s",
 919                                 sha1_to_hex(base_info));
 920                 offset += 20;
 921         } else
 922                 die("I am totally screwed");
 923         *base_obj_offset = base_offset;
 924         return offset;
 925 }
 926
 927 /* forward declaration for a mutually recursive function */
 928 static int packed_object_info(struct packed_git *p, unsigned long offset,
 929                               char *type, unsigned long *sizep);
 930
 931 static int packed_delta_info(struct packed_git *p,
 932                              unsigned long offset,
 933                              enum object_type kind,
 934                              unsigned long obj_offset,
 935                              char *type,
 936                              unsigned long *sizep)
 937 {
 938         unsigned long base_offset;
 939
 940         offset = get_delta_base(p, offset, kind, obj_offset, &base_offset);
 941
 942         /* We choose to only get the type of the base object and
 943          * ignore potentially corrupt pack file that expects the delta
 944          * based on a base with a wrong size.  This saves tons of
 945          * inflate() calls.
 946          */
 947         if (packed_object_info(p, base_offset, type, NULL))
 948                 die("cannot get info for delta-pack base");
 949
 950         if (sizep) {
 951                 const unsigned char *data;
 952                 unsigned char delta_head[64];
 953                 unsigned long result_size;
 954                 z_stream stream;
 955                 int st;
 956
 957                 memset(&stream, 0, sizeof(stream));
 958
 959                 stream.next_in = (unsigned char *) p->pack_base + offset;
 960                 stream.avail_in = p->pack_size - offset;
 961                 stream.next_out = delta_head;
 962                 stream.avail_out = sizeof(delta_head);
 963
 964                 inflateInit(&stream);
 965                 st = inflate(&stream, Z_FINISH);
 966                 inflateEnd(&stream);
 967                 if ((st != Z_STREAM_END) &&
 968                     stream.total_out != sizeof(delta_head))
 969                         die("delta data unpack-initial failed");
 970
 971                 /* Examine the initial part of the delta to figure out
 972                  * the result size.
 973                  */
 974                 data = delta_head;
 975
 976                 /* ignore base size */
 977                 get_delta_hdr_size(&data, delta_head+sizeof(delta_head));
 978
 979                 /* Read the result size */
 980                 result_size = get_delta_hdr_size(&data, delta_head+sizeof(delta_head));
 981                 *sizep = result_size;
 982         }
 983         return 0;
 984 }
 985
 986 static unsigned long unpack_object_header(struct packed_git *p, unsigned long offset,
 987         enum object_type *type, unsigned long *sizep)
 988 {
 989         unsigned long used;
 990
 991         if (p->pack_size <= offset)
 992                 die("object offset outside of pack file");
 993
 994         used = unpack_object_header_gently((unsigned char *)p->pack_base +
 995                                            offset,
 996                                            p->pack_size - offset, type, sizep);
 997         if (!used)
 998                 die("object offset outside of pack file");
 999
1000         return offset + used;
1001 }
1002
1003 int check_reuse_pack_delta(struct packed_git *p, unsigned long offset,
1004                            unsigned char *base, unsigned long *sizep,
1005                            enum object_type *kindp)
1006 {
1007         unsigned long ptr;
1008         int status = -1;
1009
1010         use_packed_git(p);
1011         ptr = offset;
1012         ptr = unpack_object_header(p, ptr, kindp, sizep);
1013         if (*kindp != OBJ_REF_DELTA)
1014                 goto done;
1015         hashcpy(base, (unsigned char *) p->pack_base + ptr);
1016         status = 0;
1017  done:
1018         unuse_packed_git(p);
1019         return status;
1020 }
1021
1022 void packed_object_info_detail(struct packed_git *p,
1023                                unsigned long offset,
1024                                char *type,
1025                                unsigned long *size,
1026                                unsigned long *store_size,
1027                                unsigned int *delta_chain_length,
1028                                unsigned char *base_sha1)
1029 {
1030         unsigned long obj_offset, val;
1031         unsigned char *next_sha1;
1032         enum object_type kind;
1033
1034         *delta_chain_length = 0;
1035         obj_offset = offset;
1036         offset = unpack_object_header(p, offset, &kind, size);
1037
1038         for (;;) {
1039                 switch (kind) {
1040                 default:
1041                         die("corrupted pack file %s containing object of kind %d",
1042                             p->pack_name, kind);
1043                 case OBJ_COMMIT:
1044                 case OBJ_TREE:
1045                 case OBJ_BLOB:
1046                 case OBJ_TAG:
1047                         strcpy(type, type_names[kind]);
1048                         *store_size = 0; /* notyet */
1049                         return;
1050                 case OBJ_OFS_DELTA:
1051                         get_delta_base(p, offset, kind, obj_offset, &offset);
1052                         if (*delta_chain_length == 0) {
1053                                 /* TODO: find base_sha1 as pointed by offset */
1054                         }
1055                         break;
1056                 case OBJ_REF_DELTA:
1057                         if (p->pack_size <= offset + 20)
1058                                 die("pack file %s records an incomplete delta base",
1059                                     p->pack_name);
1060                         next_sha1 = (unsigned char *) p->pack_base + offset;
1061                         if (*delta_chain_length == 0)
1062                                 hashcpy(base_sha1, next_sha1);
1063                         offset = find_pack_entry_one(next_sha1, p);
1064                         break;
1065                 }
1066                 obj_offset = offset;
1067                 offset = unpack_object_header(p, offset, &kind, &val);
1068                 (*delta_chain_length)++;
1069         }
1070 }
1071
1072 static int packed_object_info(struct packed_git *p, unsigned long offset,
1073                               char *type, unsigned long *sizep)
1074 {
1075         unsigned long size, obj_offset = offset;
1076         enum object_type kind;
1077
1078         offset = unpack_object_header(p, offset, &kind, &size);
1079
1080         switch (kind) {
1081         case OBJ_OFS_DELTA:
1082         case OBJ_REF_DELTA:
1083                 return packed_delta_info(p, offset, kind, obj_offset, type, sizep);
1084         case OBJ_COMMIT:
1085         case OBJ_TREE:
1086         case OBJ_BLOB:
1087         case OBJ_TAG:
1088                 strcpy(type, type_names[kind]);
1089                 break;
1090         default:
1091                 die("corrupted pack file %s containing object of kind %d",
1092                     p->pack_name, kind);
1093         }
1094         if (sizep)
1095                 *sizep = size;
1096         return 0;
1097 }
1098
1099 static void *unpack_compressed_entry(struct packed_git *p,
1100                                     unsigned long offset,
1101                                     unsigned long size)
1102 {
1103         int st;
1104         z_stream stream;
1105         unsigned char *buffer;
1106
1107         buffer = xmalloc(size + 1);
1108         buffer[size] = 0;
1109         memset(&stream, 0, sizeof(stream));
1110         stream.next_in = (unsigned char*)p->pack_base + offset;
1111         stream.avail_in = p->pack_size - offset;
1112         stream.next_out = buffer;
1113         stream.avail_out = size;
1114
1115         inflateInit(&stream);
1116         st = inflate(&stream, Z_FINISH);
1117         inflateEnd(&stream);
1118         if ((st != Z_STREAM_END) || stream.total_out != size) {
1119                 free(buffer);
1120                 return NULL;
1121         }
1122
1123         return buffer;
1124 }
1125
1126 static void *unpack_delta_entry(struct packed_git *p,
1127                                 unsigned long offset,
1128                                 unsigned long delta_size,
1129                                 enum object_type kind,
1130                                 unsigned long obj_offset,
1131                                 char *type,
1132                                 unsigned long *sizep)
1133 {
1134         void *delta_data, *result, *base;
1135         unsigned long result_size, base_size, base_offset;
1136
1137         offset = get_delta_base(p, offset, kind, obj_offset, &base_offset);
1138         base = unpack_entry_gently(p, base_offset, type, &base_size);
1139         if (!base)
1140                 die("failed to read delta base object at %lu from %s",
1141                     base_offset, p->pack_name);
1142
1143         delta_data = unpack_compressed_entry(p, offset, delta_size);
1144         result = patch_delta(base, base_size,
1145                              delta_data, delta_size,
1146                              &result_size);
1147         if (!result)
1148                 die("failed to apply delta");
1149         free(delta_data);
1150         free(base);
1151         *sizep = result_size;
1152         return result;
1153 }
1154
1155 static void *unpack_entry(struct pack_entry *entry,
1156                           char *type, unsigned long *sizep)
1157 {
1158         struct packed_git *p = entry->p;
1159         void *retval;
1160
1161         if (use_packed_git(p))
1162                 die("cannot map packed file");
1163         retval = unpack_entry_gently(p, entry->offset, type, sizep);
1164         unuse_packed_git(p);
1165         if (!retval)
1166                 die("corrupted pack file %s", p->pack_name);
1167         return retval;
1168 }
1169
1170 /* The caller is responsible for use_packed_git()/unuse_packed_git() pair */
1171 void *unpack_entry_gently(struct packed_git *p, unsigned long offset,
1172                           char *type, unsigned long *sizep)
1173 {
1174         unsigned long size, obj_offset = offset;
1175         enum object_type kind;
1176
1177         offset = unpack_object_header(p, offset, &kind, &size);
1178         switch (kind) {
1179         case OBJ_OFS_DELTA:
1180         case OBJ_REF_DELTA:
1181                 return unpack_delta_entry(p, offset, size, kind, obj_offset, type, sizep);
1182         case OBJ_COMMIT:
1183         case OBJ_TREE:
1184         case OBJ_BLOB:
1185         case OBJ_TAG:
1186                 strcpy(type, type_names[kind]);
1187                 *sizep = size;
1188                 return unpack_compressed_entry(p, offset, size);
1189         default:
1190                 return NULL;
1191         }
1192 }
1193
1194 int num_packed_objects(const struct packed_git *p)
1195 {
1196         /* See check_packed_git_idx() */
1197         return (p->index_size - 20 - 20 - 4*256) / 24;
1198 }
1199
1200 int nth_packed_object_sha1(const struct packed_git *p, int n,
1201                            unsigned char* sha1)
1202 {
1203         void *index = p->index_base + 256;
1204         if (n < 0 || num_packed_objects(p) <= n)
1205                 return -1;
1206         hashcpy(sha1, (unsigned char *) index + (24 * n) + 4);
1207         return 0;
1208 }
1209
1210 unsigned long find_pack_entry_one(const unsigned char *sha1,
1211                                   struct packed_git *p)
1212 {
1213         unsigned int *level1_ofs = p->index_base;
1214         int hi = ntohl(level1_ofs[*sha1]);
1215         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
1216         void *index = p->index_base + 256;
1217
1218         do {
1219                 int mi = (lo + hi) / 2;
1220                 int cmp = hashcmp((unsigned char *)index + (24 * mi) + 4, sha1);
1221                 if (!cmp)
1222                         return ntohl(*((unsigned int *) ((char *) index + (24 * mi))));
1223                 if (cmp > 0)
1224                         hi = mi;
1225                 else
1226                         lo = mi+1;
1227         } while (lo < hi);
1228         return 0;
1229 }
1230
1231 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e, const char **ignore_packed)
1232 {
1233         struct packed_git *p;
1234         unsigned long offset;
1235
1236         prepare_packed_git();
1237
1238         for (p = packed_git; p; p = p->next) {
1239                 if (ignore_packed) {
1240                         const char **ig;
1241                         for (ig = ignore_packed; *ig; ig++)
1242                                 if (!strcmp(p->pack_name, *ig))
1243                                         break;
1244                         if (*ig)
1245                                 continue;
1246                 }
1247                 offset = find_pack_entry_one(sha1, p);
1248                 if (offset) {
1249                         e->offset = offset;
1250                         e->p = p;
1251                         hashcpy(e->sha1, sha1);
1252                         return 1;
1253                 }
1254         }
1255         return 0;
1256 }
1257
1258 struct packed_git *find_sha1_pack(const unsigned char *sha1,
1259                                   struct packed_git *packs)
1260 {
1261         struct packed_git *p;
1262
1263         for (p = packs; p; p = p->next) {
1264                 if (find_pack_entry_one(sha1, p))
1265                         return p;
1266         }
1267         return NULL;
1268
1269 }
1270
1271 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
1272 {
1273         int status;
1274         unsigned long mapsize, size;
1275         void *map;
1276         z_stream stream;
1277         char hdr[128];
1278
1279         map = map_sha1_file(sha1, &mapsize);
1280         if (!map) {
1281                 struct pack_entry e;
1282
1283                 if (!find_pack_entry(sha1, &e, NULL)) {
1284                         reprepare_packed_git();
1285                         if (!find_pack_entry(sha1, &e, NULL))
1286                                 return error("unable to find %s", sha1_to_hex(sha1));
1287                 }
1288                 if (use_packed_git(e.p))
1289                         die("cannot map packed file");
1290                 status = packed_object_info(e.p, e.offset, type, sizep);
1291                 unuse_packed_git(e.p);
1292                 return status;
1293         }
1294         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
1295                 status = error("unable to unpack %s header",
1296                                sha1_to_hex(sha1));
1297         if (parse_sha1_header(hdr, type, &size) < 0)
1298                 status = error("unable to parse %s header", sha1_to_hex(sha1));
1299         else {
1300                 status = 0;
1301                 if (sizep)
1302                         *sizep = size;
1303         }
1304         inflateEnd(&stream);
1305         munmap(map, mapsize);
1306         return status;
1307 }
1308
1309 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
1310 {
1311         struct pack_entry e;
1312
1313         if (!find_pack_entry(sha1, &e, NULL)) {
1314                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
1315                 return NULL;
1316         }
1317         return unpack_entry(&e, type, size);
1318 }
1319
1320 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
1321 {
1322         unsigned long mapsize;
1323         void *map, *buf;
1324         struct pack_entry e;
1325
1326         if (find_pack_entry(sha1, &e, NULL))
1327                 return read_packed_sha1(sha1, type, size);
1328         map = map_sha1_file(sha1, &mapsize);
1329         if (map) {
1330                 buf = unpack_sha1_file(map, mapsize, type, size);
1331                 munmap(map, mapsize);
1332                 return buf;
1333         }
1334         reprepare_packed_git();
1335         if (find_pack_entry(sha1, &e, NULL))
1336                 return read_packed_sha1(sha1, type, size);
1337         return NULL;
1338 }
1339
1340 void *read_object_with_reference(const unsigned char *sha1,
1341                                  const char *required_type,
1342                                  unsigned long *size,
1343                                  unsigned char *actual_sha1_return)
1344 {
1345         char type[20];
1346         void *buffer;
1347         unsigned long isize;
1348         unsigned char actual_sha1[20];
1349
1350         hashcpy(actual_sha1, sha1);
1351         while (1) {
1352                 int ref_length = -1;
1353                 const char *ref_type = NULL;
1354
1355                 buffer = read_sha1_file(actual_sha1, type, &isize);
1356                 if (!buffer)
1357                         return NULL;
1358                 if (!strcmp(type, required_type)) {
1359                         *size = isize;
1360                         if (actual_sha1_return)
1361                                 hashcpy(actual_sha1_return, actual_sha1);
1362                         return buffer;
1363                 }
1364                 /* Handle references */
1365                 else if (!strcmp(type, commit_type))
1366                         ref_type = "tree ";
1367                 else if (!strcmp(type, tag_type))
1368                         ref_type = "object ";
1369                 else {
1370                         free(buffer);
1371                         return NULL;
1372                 }
1373                 ref_length = strlen(ref_type);
1374
1375                 if (memcmp(buffer, ref_type, ref_length) ||
1376                     get_sha1_hex((char *) buffer + ref_length, actual_sha1)) {
1377                         free(buffer);
1378                         return NULL;
1379                 }
1380                 free(buffer);
1381                 /* Now we have the ID of the referred-to object in
1382                  * actual_sha1.  Check again. */
1383         }
1384 }
1385
1386 char *write_sha1_file_prepare(void *buf,
1387                               unsigned long len,
1388                               const char *type,
1389                               unsigned char *sha1,
1390                               unsigned char *hdr,
1391                               int *hdrlen)
1392 {
1393         SHA_CTX c;
1394
1395         /* Generate the header */
1396         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
1397
1398         /* Sha1.. */
1399         SHA1_Init(&c);
1400         SHA1_Update(&c, hdr, *hdrlen);
1401         SHA1_Update(&c, buf, len);
1402         SHA1_Final(sha1, &c);
1403
1404         return sha1_file_name(sha1);
1405 }
1406
1407 /*
1408  * Link the tempfile to the final place, possibly creating the
1409  * last directory level as you do so.
1410  *
1411  * Returns the errno on failure, 0 on success.
1412  */
1413 static int link_temp_to_file(const char *tmpfile, const char *filename)
1414 {
1415         int ret;
1416         char *dir;
1417
1418         if (!link(tmpfile, filename))
1419                 return 0;
1420
1421         /*
1422          * Try to mkdir the last path component if that failed.
1423          *
1424          * Re-try the "link()" regardless of whether the mkdir
1425          * succeeds, since a race might mean that somebody
1426          * else succeeded.
1427          */
1428         ret = errno;
1429         dir = strrchr(filename, '/');
1430         if (dir) {
1431                 *dir = 0;
1432                 mkdir(filename, 0777);
1433                 if (adjust_shared_perm(filename))
1434                         return -2;
1435                 *dir = '/';
1436                 if (!link(tmpfile, filename))
1437                         return 0;
1438                 ret = errno;
1439         }
1440         return ret;
1441 }
1442
1443 /*
1444  * Move the just written object into its final resting place
1445  */
1446 int move_temp_to_file(const char *tmpfile, const char *filename)
1447 {
1448         int ret = link_temp_to_file(tmpfile, filename);
1449
1450         /*
1451          * Coda hack - coda doesn't like cross-directory links,
1452          * so we fall back to a rename, which will mean that it
1453          * won't be able to check collisions, but that's not a
1454          * big deal.
1455          *
1456          * The same holds for FAT formatted media.
1457          *
1458          * When this succeeds, we just return 0. We have nothing
1459          * left to unlink.
1460          */
1461         if (ret && ret != EEXIST) {
1462                 if (!rename(tmpfile, filename))
1463                         return 0;
1464                 ret = errno;
1465         }
1466         unlink(tmpfile);
1467         if (ret) {
1468                 if (ret != EEXIST) {
1469                         fprintf(stderr, "unable to write sha1 filename %s: %s\n", filename, strerror(ret));
1470                         return -1;
1471                 }
1472                 /* FIXME!!! Collision check here ? */
1473         }
1474
1475         return 0;
1476 }
1477
1478 static int write_buffer(int fd, const void *buf, size_t len)
1479 {
1480         while (len) {
1481                 ssize_t size;
1482
1483                 size = write(fd, buf, len);
1484                 if (!size)
1485                         return error("file write: disk full");
1486                 if (size < 0) {
1487                         if (errno == EINTR || errno == EAGAIN)
1488                                 continue;
1489                         return error("file write error (%s)", strerror(errno));
1490                 }
1491                 len -= size;
1492                 buf = (char *) buf + size;
1493         }
1494         return 0;
1495 }
1496
1497 static int write_binary_header(unsigned char *hdr, enum object_type type, unsigned long len)
1498 {
1499         int hdr_len;
1500         unsigned char c;
1501
1502         c = (type << 4) | (len & 15);
1503         len >>= 4;
1504         hdr_len = 1;
1505         while (len) {
1506                 *hdr++ = c | 0x80;
1507                 hdr_len++;
1508                 c = (len & 0x7f);
1509                 len >>= 7;
1510         }
1511         *hdr = c;
1512         return hdr_len;
1513 }
1514
1515 static void setup_object_header(z_stream *stream, const char *type, unsigned long len)
1516 {
1517         int obj_type, hdr;
1518
1519         if (use_legacy_headers) {
1520                 while (deflate(stream, 0) == Z_OK)
1521                         /* nothing */;
1522                 return;
1523         }
1524         if (!strcmp(type, blob_type))
1525                 obj_type = OBJ_BLOB;
1526         else if (!strcmp(type, tree_type))
1527                 obj_type = OBJ_TREE;
1528         else if (!strcmp(type, commit_type))
1529                 obj_type = OBJ_COMMIT;
1530         else if (!strcmp(type, tag_type))
1531                 obj_type = OBJ_TAG;
1532         else
1533                 die("trying to generate bogus object of type '%s'", type);
1534         hdr = write_binary_header(stream->next_out, obj_type, len);
1535         stream->total_out = hdr;
1536         stream->next_out += hdr;
1537         stream->avail_out -= hdr;
1538 }
1539
1540 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1541 {
1542         int size;
1543         unsigned char *compressed;
1544         z_stream stream;
1545         unsigned char sha1[20];
1546         char *filename;
1547         static char tmpfile[PATH_MAX];
1548         unsigned char hdr[50];
1549         int fd, hdrlen;
1550
1551         /* Normally if we have it in the pack then we do not bother writing
1552          * it out into .git/objects/??/?{38} file.
1553          */
1554         filename = write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1555         if (returnsha1)
1556                 hashcpy(returnsha1, sha1);
1557         if (has_sha1_file(sha1))
1558                 return 0;
1559         fd = open(filename, O_RDONLY);
1560         if (fd >= 0) {
1561                 /*
1562                  * FIXME!!! We might do collision checking here, but we'd
1563                  * need to uncompress the old file and check it. Later.
1564                  */
1565                 close(fd);
1566                 return 0;
1567         }
1568
1569         if (errno != ENOENT) {
1570                 fprintf(stderr, "sha1 file %s: %s\n", filename, strerror(errno));
1571                 return -1;
1572         }
1573
1574         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1575
1576         fd = mkstemp(tmpfile);
1577         if (fd < 0) {
1578                 fprintf(stderr, "unable to create temporary sha1 filename %s: %s\n", tmpfile, strerror(errno));
1579                 return -1;
1580         }
1581
1582         /* Set it up */
1583         memset(&stream, 0, sizeof(stream));
1584         deflateInit(&stream, zlib_compression_level);
1585         size = 8 + deflateBound(&stream, len+hdrlen);
1586         compressed = xmalloc(size);
1587
1588         /* Compress it */
1589         stream.next_out = compressed;
1590         stream.avail_out = size;
1591
1592         /* First header.. */
1593         stream.next_in = hdr;
1594         stream.avail_in = hdrlen;
1595         setup_object_header(&stream, type, len);
1596
1597         /* Then the data itself.. */
1598         stream.next_in = buf;
1599         stream.avail_in = len;
1600         while (deflate(&stream, Z_FINISH) == Z_OK)
1601                 /* nothing */;
1602         deflateEnd(&stream);
1603         size = stream.total_out;
1604
1605         if (write_buffer(fd, compressed, size) < 0)
1606                 die("unable to write sha1 file");
1607         fchmod(fd, 0444);
1608         close(fd);
1609         free(compressed);
1610
1611         return move_temp_to_file(tmpfile, filename);
1612 }
1613
1614 /*
1615  * We need to unpack and recompress the object for writing
1616  * it out to a different file.
1617  */
1618 static void *repack_object(const unsigned char *sha1, unsigned long *objsize)
1619 {
1620         size_t size;
1621         z_stream stream;
1622         unsigned char *unpacked;
1623         unsigned long len;
1624         char type[20];
1625         char hdr[50];
1626         int hdrlen;
1627         void *buf;
1628
1629         /* need to unpack and recompress it by itself */
1630         unpacked = read_packed_sha1(sha1, type, &len);
1631
1632         hdrlen = sprintf(hdr, "%s %lu", type, len) + 1;
1633
1634         /* Set it up */
1635         memset(&stream, 0, sizeof(stream));
1636         deflateInit(&stream, zlib_compression_level);
1637         size = deflateBound(&stream, len + hdrlen);
1638         buf = xmalloc(size);
1639
1640         /* Compress it */
1641         stream.next_out = buf;
1642         stream.avail_out = size;
1643
1644         /* First header.. */
1645         stream.next_in = (void *)hdr;
1646         stream.avail_in = hdrlen;
1647         while (deflate(&stream, 0) == Z_OK)
1648                 /* nothing */;
1649
1650         /* Then the data itself.. */
1651         stream.next_in = unpacked;
1652         stream.avail_in = len;
1653         while (deflate(&stream, Z_FINISH) == Z_OK)
1654                 /* nothing */;
1655         deflateEnd(&stream);
1656         free(unpacked);
1657
1658         *objsize = stream.total_out;
1659         return buf;
1660 }
1661
1662 int write_sha1_to_fd(int fd, const unsigned char *sha1)
1663 {
1664         int retval;
1665         unsigned long objsize;
1666         void *buf = map_sha1_file(sha1, &objsize);
1667
1668         if (buf) {
1669                 retval = write_buffer(fd, buf, objsize);
1670                 munmap(buf, objsize);
1671                 return retval;
1672         }
1673
1674         buf = repack_object(sha1, &objsize);
1675         retval = write_buffer(fd, buf, objsize);
1676         free(buf);
1677         return retval;
1678 }
1679
1680 int write_sha1_from_fd(const unsigned char *sha1, int fd, char *buffer,
1681                        size_t bufsize, size_t *bufposn)
1682 {
1683         char tmpfile[PATH_MAX];
1684         int local;
1685         z_stream stream;
1686         unsigned char real_sha1[20];
1687         unsigned char discard[4096];
1688         int ret;
1689         SHA_CTX c;
1690
1691         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1692
1693         local = mkstemp(tmpfile);
1694         if (local < 0)
1695                 return error("Couldn't open %s for %s",
1696                              tmpfile, sha1_to_hex(sha1));
1697
1698         memset(&stream, 0, sizeof(stream));
1699
1700         inflateInit(&stream);
1701
1702         SHA1_Init(&c);
1703
1704         do {
1705                 ssize_t size;
1706                 if (*bufposn) {
1707                         stream.avail_in = *bufposn;
1708                         stream.next_in = (unsigned char *) buffer;
1709                         do {
1710                                 stream.next_out = discard;
1711                                 stream.avail_out = sizeof(discard);
1712                                 ret = inflate(&stream, Z_SYNC_FLUSH);
1713                                 SHA1_Update(&c, discard, sizeof(discard) -
1714                                             stream.avail_out);
1715                         } while (stream.avail_in && ret == Z_OK);
1716                         if (write_buffer(local, buffer, *bufposn - stream.avail_in) < 0)
1717                                 die("unable to write sha1 file");
1718                         memmove(buffer, buffer + *bufposn - stream.avail_in,
1719                                 stream.avail_in);
1720                         *bufposn = stream.avail_in;
1721                         if (ret != Z_OK)
1722                                 break;
1723                 }
1724                 size = read(fd, buffer + *bufposn, bufsize - *bufposn);
1725                 if (size <= 0) {
1726                         close(local);
1727                         unlink(tmpfile);
1728                         if (!size)
1729                                 return error("Connection closed?");
1730                         perror("Reading from connection");
1731                         return -1;
1732                 }
1733                 *bufposn += size;
1734         } while (1);
1735         inflateEnd(&stream);
1736
1737         close(local);
1738         SHA1_Final(real_sha1, &c);
1739         if (ret != Z_STREAM_END) {
1740                 unlink(tmpfile);
1741                 return error("File %s corrupted", sha1_to_hex(sha1));
1742         }
1743         if (hashcmp(sha1, real_sha1)) {
1744                 unlink(tmpfile);
1745                 return error("File %s has bad hash", sha1_to_hex(sha1));
1746         }
1747
1748         return move_temp_to_file(tmpfile, sha1_file_name(sha1));
1749 }
1750
1751 int has_pack_index(const unsigned char *sha1)
1752 {
1753         struct stat st;
1754         if (stat(sha1_pack_index_name(sha1), &st))
1755                 return 0;
1756         return 1;
1757 }
1758
1759 int has_pack_file(const unsigned char *sha1)
1760 {
1761         struct stat st;
1762         if (stat(sha1_pack_name(sha1), &st))
1763                 return 0;
1764         return 1;
1765 }
1766
1767 int has_sha1_pack(const unsigned char *sha1, const char **ignore_packed)
1768 {
1769         struct pack_entry e;
1770         return find_pack_entry(sha1, &e, ignore_packed);
1771 }
1772
1773 int has_sha1_file(const unsigned char *sha1)
1774 {
1775         struct stat st;
1776         struct pack_entry e;
1777
1778         if (find_pack_entry(sha1, &e, NULL))
1779                 return 1;
1780         return find_sha1_file(sha1, &st) ? 1 : 0;
1781 }
1782
1783 /*
1784  * reads from fd as long as possible into a supplied buffer of size bytes.
1785  * If necessary the buffer's size is increased using realloc()
1786  *
1787  * returns 0 if anything went fine and -1 otherwise
1788  *
1789  * NOTE: both buf and size may change, but even when -1 is returned
1790  * you still have to free() it yourself.
1791  */
1792 int read_pipe(int fd, char** return_buf, unsigned long* return_size)
1793 {
1794         char* buf = *return_buf;
1795         unsigned long size = *return_size;
1796         int iret;
1797         unsigned long off = 0;
1798
1799         do {
1800                 iret = xread(fd, buf + off, size - off);
1801                 if (iret > 0) {
1802                         off += iret;
1803                         if (off == size) {
1804                                 size *= 2;
1805                                 buf = xrealloc(buf, size);
1806                         }
1807                 }
1808         } while (iret > 0);
1809
1810         *return_buf = buf;
1811         *return_size = off;
1812
1813         if (iret < 0)
1814                 return -1;
1815         return 0;
1816 }
1817
1818 int index_pipe(unsigned char *sha1, int fd, const char *type, int write_object)
1819 {
1820         unsigned long size = 4096;
1821         char *buf = xmalloc(size);
1822         int ret;
1823         unsigned char hdr[50];
1824         int hdrlen;
1825
1826         if (read_pipe(fd, &buf, &size)) {
1827                 free(buf);
1828                 return -1;
1829         }
1830
1831         if (!type)
1832                 type = blob_type;
1833         if (write_object)
1834                 ret = write_sha1_file(buf, size, type, sha1);
1835         else {
1836                 write_sha1_file_prepare(buf, size, type, sha1, hdr, &hdrlen);
1837                 ret = 0;
1838         }
1839         free(buf);
1840         return ret;
1841 }
1842
1843 int index_fd(unsigned char *sha1, int fd, struct stat *st, int write_object, const char *type)
1844 {
1845         unsigned long size = st->st_size;
1846         void *buf;
1847         int ret;
1848         unsigned char hdr[50];
1849         int hdrlen;
1850
1851         buf = "";
1852         if (size)
1853                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1854         close(fd);
1855         if (buf == MAP_FAILED)
1856                 return -1;
1857
1858         if (!type)
1859                 type = blob_type;
1860         if (write_object)
1861                 ret = write_sha1_file(buf, size, type, sha1);
1862         else {
1863                 write_sha1_file_prepare(buf, size, type, sha1, hdr, &hdrlen);
1864                 ret = 0;
1865         }
1866         if (size)
1867                 munmap(buf, size);
1868         return ret;
1869 }
1870
1871 int index_path(unsigned char *sha1, const char *path, struct stat *st, int write_object)
1872 {
1873         int fd;
1874         char *target;
1875
1876         switch (st->st_mode & S_IFMT) {
1877         case S_IFREG:
1878                 fd = open(path, O_RDONLY);
1879                 if (fd < 0)
1880                         return error("open(\"%s\"): %s", path,
1881                                      strerror(errno));
1882                 if (index_fd(sha1, fd, st, write_object, NULL) < 0)
1883                         return error("%s: failed to insert into database",
1884                                      path);
1885                 break;
1886         case S_IFLNK:
1887                 target = xmalloc(st->st_size+1);
1888                 if (readlink(path, target, st->st_size+1) != st->st_size) {
1889                         char *errstr = strerror(errno);
1890                         free(target);
1891                         return error("readlink(\"%s\"): %s", path,
1892                                      errstr);
1893                 }
1894                 if (!write_object) {
1895                         unsigned char hdr[50];
1896                         int hdrlen;
1897                         write_sha1_file_prepare(target, st->st_size, blob_type,
1898                                                 sha1, hdr, &hdrlen);
1899                 } else if (write_sha1_file(target, st->st_size, blob_type, sha1))
1900                         return error("%s: failed to insert into database",
1901                                      path);
1902                 free(target);
1903                 break;
1904         default:
1905                 return error("%s: unsupported file type", path);
1906         }
1907         return 0;
1908 }