sha1_file.c

   1 /*
   2  * GIT - The information manager from hell
   3  *
   4  * Copyright (C) Linus Torvalds, 2005
   5  *
   6  * This handles basic git sha1 object files - packing, unpacking,
   7  * creation etc.
   8  */
   9 #include <sys/types.h>
  10 #include <dirent.h>
  11 #include "cache.h"
  12 #include "delta.h"
  13 #include "pack.h"
  14
  15 #ifndef O_NOATIME
  16 #if defined(__linux__) && (defined(__i386__) || defined(__PPC__))
  17 #define O_NOATIME 01000000
  18 #else
  19 #define O_NOATIME 0
  20 #endif
  21 #endif
  22
  23 static unsigned int sha1_file_open_flag = O_NOATIME;
  24
  25 static unsigned hexval(char c)
  26 {
  27         if (c >= '0' && c <= '9')
  28                 return c - '0';
  29         if (c >= 'a' && c <= 'f')
  30                 return c - 'a' + 10;
  31         if (c >= 'A' && c <= 'F')
  32                 return c - 'A' + 10;
  33         return ~0;
  34 }
  35
  36 int get_sha1_hex(const char *hex, unsigned char *sha1)
  37 {
  38         int i;
  39         for (i = 0; i < 20; i++) {
  40                 unsigned int val = (hexval(hex[0]) << 4) | hexval(hex[1]);
  41                 if (val & ~0xff)
  42                         return -1;
  43                 *sha1++ = val;
  44                 hex += 2;
  45         }
  46         return 0;
  47 }
  48
  49 static int get_sha1_file(const char *path, unsigned char *result)
  50 {
  51         char buffer[60];
  52         int fd = open(path, O_RDONLY);
  53         int len;
  54
  55         if (fd < 0)
  56                 return -1;
  57         len = read(fd, buffer, sizeof(buffer));
  58         close(fd);
  59         if (len < 40)
  60                 return -1;
  61         return get_sha1_hex(buffer, result);
  62 }
  63
  64 static char *git_dir, *git_object_dir, *git_index_file, *git_refs_dir;
  65 static void setup_git_env(void)
  66 {
  67         git_dir = gitenv(GIT_DIR_ENVIRONMENT);
  68         if (!git_dir)
  69                 git_dir = DEFAULT_GIT_DIR_ENVIRONMENT;
  70         git_object_dir = gitenv(DB_ENVIRONMENT);
  71         if (!git_object_dir) {
  72                 git_object_dir = xmalloc(strlen(git_dir) + 9);
  73                 sprintf(git_object_dir, "%s/objects", git_dir);
  74         }
  75         git_refs_dir = xmalloc(strlen(git_dir) + 6);
  76         sprintf(git_refs_dir, "%s/refs", git_dir);
  77         git_index_file = gitenv(INDEX_ENVIRONMENT);
  78         if (!git_index_file) {
  79                 git_index_file = xmalloc(strlen(git_dir) + 7);
  80                 sprintf(git_index_file, "%s/index", git_dir);
  81         }
  82 }
  83
  84 char *get_object_directory(void)
  85 {
  86         if (!git_object_dir)
  87                 setup_git_env();
  88         return git_object_dir;
  89 }
  90
  91 char *get_refs_directory(void)
  92 {
  93         if (!git_refs_dir)
  94                 setup_git_env();
  95         return git_refs_dir;
  96 }
  97
  98 char *get_index_file(void)
  99 {
 100         if (!git_index_file)
 101                 setup_git_env();
 102         return git_index_file;
 103 }
 104
 105 char *git_path(const char *fmt, ...)
 106 {
 107         static char pathname[PATH_MAX];
 108         va_list args;
 109         int len;
 110
 111         if (!git_dir)
 112                 setup_git_env();
 113         len = strlen(git_dir);
 114         if (len == 1 && *git_dir == '.')
 115                 len = 0;
 116         if (len > PATH_MAX-100)
 117                 return "pad-path";
 118         memcpy(pathname, git_dir, len);
 119         if (len && git_dir[len-1] != '/')
 120                 pathname[len++] = '/';
 121         va_start(args, fmt);
 122         vsnprintf(pathname + len, sizeof(pathname) - len, fmt, args);
 123         va_end(args);
 124         return pathname;
 125 }
 126
 127 int get_sha1(const char *str, unsigned char *sha1)
 128 {
 129         static const char *prefix[] = {
 130                 "",
 131                 "refs",
 132                 "refs/tags",
 133                 "refs/heads",
 134                 "refs/snap",
 135                 NULL
 136         };
 137         const char **p;
 138
 139         if (!get_sha1_hex(str, sha1))
 140                 return 0;
 141
 142         for (p = prefix; *p; p++) {
 143                 char * pathname = git_path("%s/%s", *p, str);
 144                 if (!get_sha1_file(pathname, sha1))
 145                         return 0;
 146         }
 147
 148         return -1;
 149 }
 150
 151 char * sha1_to_hex(const unsigned char *sha1)
 152 {
 153         static char buffer[50];
 154         static const char hex[] = "0123456789abcdef";
 155         char *buf = buffer;
 156         int i;
 157
 158         for (i = 0; i < 20; i++) {
 159                 unsigned int val = *sha1++;
 160                 *buf++ = hex[val >> 4];
 161                 *buf++ = hex[val & 0xf];
 162         }
 163         return buffer;
 164 }
 165
 166 static void fill_sha1_path(char *pathbuf, const unsigned char *sha1)
 167 {
 168         int i;
 169         for (i = 0; i < 20; i++) {
 170                 static char hex[] = "0123456789abcdef";
 171                 unsigned int val = sha1[i];
 172                 char *pos = pathbuf + i*2 + (i > 0);
 173                 *pos++ = hex[val >> 4];
 174                 *pos = hex[val & 0xf];
 175         }
 176 }
 177
 178 /*
 179  * NOTE! This returns a statically allocated buffer, so you have to be
 180  * careful about using it. Do a "strdup()" if you need to save the
 181  * filename.
 182  *
 183  * Also note that this returns the location for creating.  Reading
 184  * SHA1 file can happen from any alternate directory listed in the
 185  * DB_ENVIRONMENT environment variable if it is not found in
 186  * the primary object database.
 187  */
 188 char *sha1_file_name(const unsigned char *sha1)
 189 {
 190         static char *name, *base;
 191
 192         if (!base) {
 193                 const char *sha1_file_directory = get_object_directory();
 194                 int len = strlen(sha1_file_directory);
 195                 base = xmalloc(len + 60);
 196                 memcpy(base, sha1_file_directory, len);
 197                 memset(base+len, 0, 60);
 198                 base[len] = '/';
 199                 base[len+3] = '/';
 200                 name = base + len + 1;
 201         }
 202         fill_sha1_path(name, sha1);
 203         return base;
 204 }
 205
 206 struct alternate_object_database *alt_odb;
 207
 208 /*
 209  * Prepare alternate object database registry.
 210  * alt_odb points at an array of struct alternate_object_database.
 211  * This array is terminated with an element that has both its base
 212  * and name set to NULL.  alt_odb[n] comes from n'th non-empty
 213  * element from colon separated ALTERNATE_DB_ENVIRONMENT environment
 214  * variable, and its base points at a statically allocated buffer
 215  * that contains "/the/directory/corresponding/to/.git/objects/...",
 216  * while its name points just after the slash at the end of
 217  * ".git/objects/" in the example above, and has enough space to hold
 218  * 40-byte hex SHA1, an extra slash for the first level indirection,
 219  * and the terminating NUL.
 220  * This function allocates the alt_odb array and all the strings
 221  * pointed by base fields of the array elements with one xmalloc();
 222  * the string pool immediately follows the array.
 223  */
 224 void prepare_alt_odb(void)
 225 {
 226         int pass, totlen, i;
 227         const char *cp, *last;
 228         char *op = NULL;
 229         const char *alt = gitenv(ALTERNATE_DB_ENVIRONMENT) ? : "";
 230
 231         if (alt_odb)
 232                 return;
 233         /* The first pass counts how large an area to allocate to
 234          * hold the entire alt_odb structure, including array of
 235          * structs and path buffers for them.  The second pass fills
 236          * the structure and prepares the path buffers for use by
 237          * fill_sha1_path().
 238          */
 239         for (totlen = pass = 0; pass < 2; pass++) {
 240                 last = alt;
 241                 i = 0;
 242                 do {
 243                         cp = strchr(last, ':') ? : last + strlen(last);
 244                         if (last != cp) {
 245                                 /* 43 = 40-byte + 2 '/' + terminating NUL */
 246                                 int pfxlen = cp - last;
 247                                 int entlen = pfxlen + 43;
 248                                 if (pass == 0)
 249                                         totlen += entlen;
 250                                 else {
 251                                         alt_odb[i].base = op;
 252                                         alt_odb[i].name = op + pfxlen + 1;
 253                                         memcpy(op, last, pfxlen);
 254                                         op[pfxlen] = op[pfxlen + 3] = '/';
 255                                         op[entlen-1] = 0;
 256                                         op += entlen;
 257                                 }
 258                                 i++;
 259                         }
 260                         while (*cp && *cp == ':')
 261                                 cp++;
 262                         last = cp;
 263                 } while (*cp);
 264                 if (pass)
 265                         break;
 266                 alt_odb = xmalloc(sizeof(*alt_odb) * (i + 1) + totlen);
 267                 alt_odb[i].base = alt_odb[i].name = NULL;
 268                 op = (char*)(&alt_odb[i+1]);
 269         }
 270 }
 271
 272 static char *find_sha1_file(const unsigned char *sha1, struct stat *st)
 273 {
 274         int i;
 275         char *name = sha1_file_name(sha1);
 276
 277         if (!stat(name, st))
 278                 return name;
 279         prepare_alt_odb();
 280         for (i = 0; (name = alt_odb[i].name) != NULL; i++) {
 281                 fill_sha1_path(name, sha1);
 282                 if (!stat(alt_odb[i].base, st))
 283                         return alt_odb[i].base;
 284         }
 285         return NULL;
 286 }
 287
 288 #define PACK_MAX_SZ (1<<26)
 289 static int pack_used_ctr;
 290 static unsigned long pack_mapped;
 291 struct packed_git *packed_git;
 292
 293 static int check_packed_git_idx(const char *path, unsigned long *idx_size_,
 294                                 void **idx_map_)
 295 {
 296         void *idx_map;
 297         unsigned int *index;
 298         unsigned long idx_size;
 299         int nr, i;
 300         int fd = open(path, O_RDONLY);
 301         struct stat st;
 302         if (fd < 0)
 303                 return -1;
 304         if (fstat(fd, &st)) {
 305                 close(fd);
 306                 return -1;
 307         }
 308         idx_size = st.st_size;
 309         idx_map = mmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0);
 310         close(fd);
 311         if (idx_map == MAP_FAILED)
 312                 return -1;
 313
 314         index = idx_map;
 315         *idx_map_ = idx_map;
 316         *idx_size_ = idx_size;
 317
 318         /* check index map */
 319         if (idx_size < 4*256 + 20 + 20)
 320                 return error("index file too small");
 321         nr = 0;
 322         for (i = 0; i < 256; i++) {
 323                 unsigned int n = ntohl(index[i]);
 324                 if (n < nr)
 325                         return error("non-monotonic index");
 326                 nr = n;
 327         }
 328
 329         /*
 330          * Total size:
 331          *  - 256 index entries 4 bytes each
 332          *  - 24-byte entries * nr (20-byte sha1 + 4-byte offset)
 333          *  - 20-byte SHA1 of the packfile
 334          *  - 20-byte SHA1 file checksum
 335          */
 336         if (idx_size != 4*256 + nr * 24 + 20 + 20)
 337                 return error("wrong index file size");
 338
 339         return 0;
 340 }
 341
 342 static int unuse_one_packed_git(void)
 343 {
 344         struct packed_git *p, *lru = NULL;
 345
 346         for (p = packed_git; p; p = p->next) {
 347                 if (p->pack_use_cnt || !p->pack_base)
 348                         continue;
 349                 if (!lru || p->pack_last_used < lru->pack_last_used)
 350                         lru = p;
 351         }
 352         if (!lru)
 353                 return 0;
 354         munmap(lru->pack_base, lru->pack_size);
 355         lru->pack_base = NULL;
 356         return 1;
 357 }
 358
 359 void unuse_packed_git(struct packed_git *p)
 360 {
 361         p->pack_use_cnt--;
 362 }
 363
 364 int use_packed_git(struct packed_git *p)
 365 {
 366         if (!p->pack_base) {
 367                 int fd;
 368                 struct stat st;
 369                 void *map;
 370
 371                 pack_mapped += p->pack_size;
 372                 while (PACK_MAX_SZ < pack_mapped && unuse_one_packed_git())
 373                         ; /* nothing */
 374                 fd = open(p->pack_name, O_RDONLY);
 375                 if (fd < 0)
 376                         die("packfile %s cannot be opened", p->pack_name);
 377                 if (fstat(fd, &st)) {
 378                         close(fd);
 379                         die("packfile %s cannot be opened", p->pack_name);
 380                 }
 381                 if (st.st_size != p->pack_size)
 382                         die("packfile %s size mismatch.", p->pack_name);
 383                 map = mmap(NULL, p->pack_size, PROT_READ, MAP_PRIVATE, fd, 0);
 384                 close(fd);
 385                 if (map == MAP_FAILED)
 386                         die("packfile %s cannot be mapped.", p->pack_name);
 387                 p->pack_base = map;
 388
 389                 /* Check if the pack file matches with the index file.
 390                  * this is cheap.
 391                  */
 392                 if (memcmp((char*)(p->index_base) + p->index_size - 40,
 393                            p->pack_base + p->pack_size - 20, 20))
 394                         die("packfile %s does not match index.", p->pack_name);
 395         }
 396         p->pack_last_used = pack_used_ctr++;
 397         p->pack_use_cnt++;
 398         return 0;
 399 }
 400
 401 struct packed_git *add_packed_git(char *path, int path_len)
 402 {
 403         struct stat st;
 404         struct packed_git *p;
 405         unsigned long idx_size;
 406         void *idx_map;
 407
 408         if (check_packed_git_idx(path, &idx_size, &idx_map))
 409                 return NULL;
 410
 411         /* do we have a corresponding .pack file? */
 412         strcpy(path + path_len - 4, ".pack");
 413         if (stat(path, &st) || !S_ISREG(st.st_mode)) {
 414                 munmap(idx_map, idx_size);
 415                 return NULL;
 416         }
 417         /* ok, it looks sane as far as we can check without
 418          * actually mapping the pack file.
 419          */
 420         p = xmalloc(sizeof(*p) + path_len + 2);
 421         strcpy(p->pack_name, path);
 422         p->index_size = idx_size;
 423         p->pack_size = st.st_size;
 424         p->index_base = idx_map;
 425         p->next = NULL;
 426         p->pack_base = NULL;
 427         p->pack_last_used = 0;
 428         p->pack_use_cnt = 0;
 429         return p;
 430 }
 431
 432 static void prepare_packed_git_one(char *objdir)
 433 {
 434         char path[PATH_MAX];
 435         int len;
 436         DIR *dir;
 437         struct dirent *de;
 438
 439         sprintf(path, "%s/pack", objdir);
 440         len = strlen(path);
 441         dir = opendir(path);
 442         if (!dir)
 443                 return;
 444         path[len++] = '/';
 445         while ((de = readdir(dir)) != NULL) {
 446                 int namelen = strlen(de->d_name);
 447                 struct packed_git *p;
 448
 449                 if (strcmp(de->d_name + namelen - 4, ".idx"))
 450                         continue;
 451
 452                 /* we have .idx.  Is it a file we can map? */
 453                 strcpy(path + len, de->d_name);
 454                 p = add_packed_git(path, len + namelen);
 455                 if (!p)
 456                         continue;
 457                 p->next = packed_git;
 458                 packed_git = p;
 459         }
 460 }
 461
 462 void prepare_packed_git(void)
 463 {
 464         int i;
 465         static int run_once = 0;
 466
 467         if (run_once++)
 468                 return;
 469
 470         prepare_packed_git_one(get_object_directory());
 471         prepare_alt_odb();
 472         for (i = 0; alt_odb[i].base != NULL; i++) {
 473                 alt_odb[i].name[0] = 0;
 474                 prepare_packed_git_one(alt_odb[i].base);
 475         }
 476 }
 477
 478 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 479 {
 480         char header[100];
 481         unsigned char real_sha1[20];
 482         SHA_CTX c;
 483
 484         SHA1_Init(&c);
 485         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 486         SHA1_Update(&c, map, size);
 487         SHA1_Final(real_sha1, &c);
 488         return memcmp(sha1, real_sha1, 20) ? -1 : 0;
 489 }
 490
 491 static void *map_sha1_file_internal(const unsigned char *sha1,
 492                                     unsigned long *size,
 493                                     int say_error)
 494 {
 495         struct stat st;
 496         void *map;
 497         int fd;
 498         char *filename = find_sha1_file(sha1, &st);
 499
 500         if (!filename) {
 501                 if (say_error)
 502                         error("cannot map sha1 file %s", sha1_to_hex(sha1));
 503                 return NULL;
 504         }
 505
 506         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 507         if (fd < 0) {
 508                 /* See if it works without O_NOATIME */
 509                 switch (sha1_file_open_flag) {
 510                 default:
 511                         fd = open(filename, O_RDONLY);
 512                         if (fd >= 0)
 513                                 break;
 514                 /* Fallthrough */
 515                 case 0:
 516                         if (say_error)
 517                                 perror(filename);
 518                         return NULL;
 519                 }
 520
 521                 /* If it failed once, it will probably fail again.
 522                  * Stop using O_NOATIME
 523                  */
 524                 sha1_file_open_flag = 0;
 525         }
 526         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 527         close(fd);
 528         if (-1 == (int)(long)map)
 529                 return NULL;
 530         *size = st.st_size;
 531         return map;
 532 }
 533
 534 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 535 {
 536         return map_sha1_file_internal(sha1, size, 1);
 537 }
 538
 539 int unpack_sha1_header(z_stream *stream, void *map, unsigned long mapsize, void *buffer, unsigned long size)
 540 {
 541         /* Get the data stream */
 542         memset(stream, 0, sizeof(*stream));
 543         stream->next_in = map;
 544         stream->avail_in = mapsize;
 545         stream->next_out = buffer;
 546         stream->avail_out = size;
 547
 548         inflateInit(stream);
 549         return inflate(stream, 0);
 550 }
 551
 552 static void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 553 {
 554         int bytes = strlen(buffer) + 1;
 555         unsigned char *buf = xmalloc(1+size);
 556
 557         memcpy(buf, buffer + bytes, stream->total_out - bytes);
 558         bytes = stream->total_out - bytes;
 559         if (bytes < size) {
 560                 stream->next_out = buf + bytes;
 561                 stream->avail_out = size - bytes;
 562                 while (inflate(stream, Z_FINISH) == Z_OK)
 563                         /* nothing */;
 564         }
 565         buf[size] = 0;
 566         inflateEnd(stream);
 567         return buf;
 568 }
 569
 570 /*
 571  * We used to just use "sscanf()", but that's actually way
 572  * too permissive for what we want to check. So do an anal
 573  * object header parse by hand.
 574  */
 575 int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 576 {
 577         int i;
 578         unsigned long size;
 579
 580         /*
 581          * The type can be at most ten bytes (including the
 582          * terminating '\0' that we add), and is followed by
 583          * a space.
 584          */
 585         i = 10;
 586         for (;;) {
 587                 char c = *hdr++;
 588                 if (c == ' ')
 589                         break;
 590                 if (!--i)
 591                         return -1;
 592                 *type++ = c;
 593         }
 594         *type = 0;
 595
 596         /*
 597          * The length must follow immediately, and be in canonical
 598          * decimal format (ie "010" is not valid).
 599          */
 600         size = *hdr++ - '0';
 601         if (size > 9)
 602                 return -1;
 603         if (size) {
 604                 for (;;) {
 605                         unsigned long c = *hdr - '0';
 606                         if (c > 9)
 607                                 break;
 608                         hdr++;
 609                         size = size * 10 + c;
 610                 }
 611         }
 612         *sizep = size;
 613
 614         /*
 615          * The length must be followed by a zero byte
 616          */
 617         return *hdr ? -1 : 0;
 618 }
 619
 620 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 621 {
 622         int ret;
 623         z_stream stream;
 624         char hdr[8192];
 625
 626         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
 627         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
 628                 return NULL;
 629
 630         return unpack_sha1_rest(&stream, hdr, *size);
 631 }
 632
 633 /* forward declaration for a mutually recursive function */
 634 static int packed_object_info(struct pack_entry *entry,
 635                               char *type, unsigned long *sizep);
 636
 637 static int packed_delta_info(unsigned char *base_sha1,
 638                              unsigned long delta_size,
 639                              unsigned long left,
 640                              char *type,
 641                              unsigned long *sizep,
 642                              struct packed_git *p)
 643 {
 644         struct pack_entry base_ent;
 645
 646         if (left < 20)
 647                 die("truncated pack file");
 648
 649         /* The base entry _must_ be in the same pack */
 650         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 651                 die("failed to find delta-pack base object %s",
 652                     sha1_to_hex(base_sha1));
 653
 654         /* We choose to only get the type of the base object and
 655          * ignore potentially corrupt pack file that expects the delta
 656          * based on a base with a wrong size.  This saves tons of
 657          * inflate() calls.
 658          */
 659
 660         if (packed_object_info(&base_ent, type, NULL))
 661                 die("cannot get info for delta-pack base");
 662
 663         if (sizep) {
 664                 const unsigned char *data;
 665                 unsigned char delta_head[64];
 666                 unsigned long result_size;
 667                 z_stream stream;
 668                 int st;
 669
 670                 memset(&stream, 0, sizeof(stream));
 671
 672                 data = stream.next_in = base_sha1 + 20;
 673                 stream.avail_in = left - 20;
 674                 stream.next_out = delta_head;
 675                 stream.avail_out = sizeof(delta_head);
 676
 677                 inflateInit(&stream);
 678                 st = inflate(&stream, Z_FINISH);
 679                 inflateEnd(&stream);
 680                 if ((st != Z_STREAM_END) &&
 681                     stream.total_out != sizeof(delta_head))
 682                         die("delta data unpack-initial failed");
 683
 684                 /* Examine the initial part of the delta to figure out
 685                  * the result size.
 686                  */
 687                 data = delta_head;
 688                 get_delta_hdr_size(&data); /* ignore base size */
 689
 690                 /* Read the result size */
 691                 result_size = get_delta_hdr_size(&data);
 692                 *sizep = result_size;
 693         }
 694         return 0;
 695 }
 696
 697 static unsigned long unpack_object_header(struct packed_git *p, unsigned long offset,
 698         enum object_type *type, unsigned long *sizep)
 699 {
 700         unsigned shift;
 701         unsigned char *pack, c;
 702         unsigned long size;
 703
 704         if (offset >= p->pack_size)
 705                 die("object offset outside of pack file");
 706
 707         pack =  p->pack_base + offset;
 708         c = *pack++;
 709         offset++;
 710         *type = (c >> 4) & 7;
 711         size = c & 15;
 712         shift = 4;
 713         while (c & 0x80) {
 714                 if (offset >= p->pack_size)
 715                         die("object offset outside of pack file");
 716                 c = *pack++;
 717                 offset++;
 718                 size += (c & 0x7f) << shift;
 719                 shift += 7;
 720         }
 721         *sizep = size;
 722         return offset;
 723 }
 724
 725 void packed_object_info_detail(struct pack_entry *e,
 726                                char *type,
 727                                unsigned long *size,
 728                                unsigned long *store_size,
 729                                int *delta_chain_length,
 730                                unsigned char *base_sha1)
 731 {
 732         struct packed_git *p = e->p;
 733         unsigned long offset, left;
 734         unsigned char *pack;
 735         enum object_type kind;
 736
 737         offset = unpack_object_header(p, e->offset, &kind, size);
 738         pack = p->pack_base + offset;
 739         left = p->pack_size - offset;
 740         if (kind != OBJ_DELTA)
 741                 *delta_chain_length = 0;
 742         else {
 743                 int chain_length = 0;
 744                 memcpy(base_sha1, pack, 20);
 745                 do {
 746                         struct pack_entry base_ent;
 747                         unsigned long junk;
 748
 749                         find_pack_entry_one(pack, &base_ent, p);
 750                         offset = unpack_object_header(p, base_ent.offset,
 751                                                       &kind, &junk);
 752                         pack = p->pack_base + offset;
 753                         chain_length++;
 754                 } while (kind == OBJ_DELTA);
 755                 *delta_chain_length = chain_length;
 756         }
 757         switch (kind) {
 758         case OBJ_COMMIT:
 759                 strcpy(type, "commit");
 760                 break;
 761         case OBJ_TREE:
 762                 strcpy(type, "tree");
 763                 break;
 764         case OBJ_BLOB:
 765                 strcpy(type, "blob");
 766                 break;
 767         case OBJ_TAG:
 768                 strcpy(type, "tag");
 769                 break;
 770         default:
 771                 die("corrupted pack file");
 772         }
 773         *store_size = 0; /* notyet */
 774 }
 775
 776 static int packed_object_info(struct pack_entry *entry,
 777                               char *type, unsigned long *sizep)
 778 {
 779         struct packed_git *p = entry->p;
 780         unsigned long offset, size, left;
 781         unsigned char *pack;
 782         enum object_type kind;
 783         int retval;
 784
 785         if (use_packed_git(p))
 786                 die("cannot map packed file");
 787
 788         offset = unpack_object_header(p, entry->offset, &kind, &size);
 789         pack = p->pack_base + offset;
 790         left = p->pack_size - offset;
 791
 792         switch (kind) {
 793         case OBJ_DELTA:
 794                 retval = packed_delta_info(pack, size, left, type, sizep, p);
 795                 unuse_packed_git(p);
 796                 return retval;
 797         case OBJ_COMMIT:
 798                 strcpy(type, "commit");
 799                 break;
 800         case OBJ_TREE:
 801                 strcpy(type, "tree");
 802                 break;
 803         case OBJ_BLOB:
 804                 strcpy(type, "blob");
 805                 break;
 806         case OBJ_TAG:
 807                 strcpy(type, "tag");
 808                 break;
 809         default:
 810                 die("corrupted pack file");
 811         }
 812         if (sizep)
 813                 *sizep = size;
 814         unuse_packed_git(p);
 815         return 0;
 816 }
 817
 818 /* forward declaration for a mutually recursive function */
 819 static void *unpack_entry(struct pack_entry *, char *, unsigned long *);
 820
 821 static void *unpack_delta_entry(unsigned char *base_sha1,
 822                                 unsigned long delta_size,
 823                                 unsigned long left,
 824                                 char *type,
 825                                 unsigned long *sizep,
 826                                 struct packed_git *p)
 827 {
 828         struct pack_entry base_ent;
 829         void *data, *delta_data, *result, *base;
 830         unsigned long data_size, result_size, base_size;
 831         z_stream stream;
 832         int st;
 833
 834         if (left < 20)
 835                 die("truncated pack file");
 836         data = base_sha1 + 20;
 837         data_size = left - 20;
 838         delta_data = xmalloc(delta_size);
 839
 840         memset(&stream, 0, sizeof(stream));
 841
 842         stream.next_in = data;
 843         stream.avail_in = data_size;
 844         stream.next_out = delta_data;
 845         stream.avail_out = delta_size;
 846
 847         inflateInit(&stream);
 848         st = inflate(&stream, Z_FINISH);
 849         inflateEnd(&stream);
 850         if ((st != Z_STREAM_END) || stream.total_out != delta_size)
 851                 die("delta data unpack failed");
 852
 853         /* The base entry _must_ be in the same pack */
 854         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 855                 die("failed to find delta-pack base object %s",
 856                     sha1_to_hex(base_sha1));
 857         base = unpack_entry_gently(&base_ent, type, &base_size);
 858         if (!base)
 859                 die("failed to read delta-pack base object %s",
 860                     sha1_to_hex(base_sha1));
 861         result = patch_delta(base, base_size,
 862                              delta_data, delta_size,
 863                              &result_size);
 864         if (!result)
 865                 die("failed to apply delta");
 866         free(delta_data);
 867         free(base);
 868         *sizep = result_size;
 869         return result;
 870 }
 871
 872 static void *unpack_non_delta_entry(unsigned char *data,
 873                                     unsigned long size,
 874                                     unsigned long left)
 875 {
 876         int st;
 877         z_stream stream;
 878         unsigned char *buffer;
 879
 880         buffer = xmalloc(size + 1);
 881         buffer[size] = 0;
 882         memset(&stream, 0, sizeof(stream));
 883         stream.next_in = data;
 884         stream.avail_in = left;
 885         stream.next_out = buffer;
 886         stream.avail_out = size;
 887
 888         inflateInit(&stream);
 889         st = inflate(&stream, Z_FINISH);
 890         inflateEnd(&stream);
 891         if ((st != Z_STREAM_END) || stream.total_out != size) {
 892                 free(buffer);
 893                 return NULL;
 894         }
 895
 896         return buffer;
 897 }
 898
 899 static void *unpack_entry(struct pack_entry *entry,
 900                           char *type, unsigned long *sizep)
 901 {
 902         struct packed_git *p = entry->p;
 903         void *retval;
 904
 905         if (use_packed_git(p))
 906                 die("cannot map packed file");
 907         retval = unpack_entry_gently(entry, type, sizep);
 908         unuse_packed_git(p);
 909         if (!retval)
 910                 die("corrupted pack file");
 911         return retval;
 912 }
 913
 914 /* The caller is responsible for use_packed_git()/unuse_packed_git() pair */
 915 void *unpack_entry_gently(struct pack_entry *entry,
 916                           char *type, unsigned long *sizep)
 917 {
 918         struct packed_git *p = entry->p;
 919         unsigned long offset, size, left;
 920         unsigned char *pack;
 921         enum object_type kind;
 922         void *retval;
 923
 924         offset = unpack_object_header(p, entry->offset, &kind, &size);
 925         pack = p->pack_base + offset;
 926         left = p->pack_size - offset;
 927         switch (kind) {
 928         case OBJ_DELTA:
 929                 retval = unpack_delta_entry(pack, size, left, type, sizep, p);
 930                 return retval;
 931         case OBJ_COMMIT:
 932                 strcpy(type, "commit");
 933                 break;
 934         case OBJ_TREE:
 935                 strcpy(type, "tree");
 936                 break;
 937         case OBJ_BLOB:
 938                 strcpy(type, "blob");
 939                 break;
 940         case OBJ_TAG:
 941                 strcpy(type, "tag");
 942                 break;
 943         default:
 944                 return NULL;
 945         }
 946         *sizep = size;
 947         retval = unpack_non_delta_entry(pack, size, left);
 948         return retval;
 949 }
 950
 951 int num_packed_objects(const struct packed_git *p)
 952 {
 953         /* See check_packed_git_idx() */
 954         return (p->index_size - 20 - 20 - 4*256) / 24;
 955 }
 956
 957 int nth_packed_object_sha1(const struct packed_git *p, int n,
 958                            unsigned char* sha1)
 959 {
 960         void *index = p->index_base + 256;
 961         if (n < 0 || num_packed_objects(p) <= n)
 962                 return -1;
 963         memcpy(sha1, (index + 24 * n + 4), 20);
 964         return 0;
 965 }
 966
 967 int find_pack_entry_one(const unsigned char *sha1,
 968                         struct pack_entry *e, struct packed_git *p)
 969 {
 970         unsigned int *level1_ofs = p->index_base;
 971         int hi = ntohl(level1_ofs[*sha1]);
 972         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
 973         void *index = p->index_base + 256;
 974
 975         do {
 976                 int mi = (lo + hi) / 2;
 977                 int cmp = memcmp(index + 24 * mi + 4, sha1, 20);
 978                 if (!cmp) {
 979                         e->offset = ntohl(*((int*)(index + 24 * mi)));
 980                         memcpy(e->sha1, sha1, 20);
 981                         e->p = p;
 982                         return 1;
 983                 }
 984                 if (cmp > 0)
 985                         hi = mi;
 986                 else
 987                         lo = mi+1;
 988         } while (lo < hi);
 989         return 0;
 990 }
 991
 992 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e)
 993 {
 994         struct packed_git *p;
 995         prepare_packed_git();
 996
 997         for (p = packed_git; p; p = p->next) {
 998                 if (find_pack_entry_one(sha1, e, p))
 999                         return 1;
1000         }
1001         return 0;
1002 }
1003
1004 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
1005 {
1006         int status;
1007         unsigned long mapsize, size;
1008         void *map;
1009         z_stream stream;
1010         char hdr[128];
1011
1012         map = map_sha1_file_internal(sha1, &mapsize, 0);
1013         if (!map) {
1014                 struct pack_entry e;
1015
1016                 if (!find_pack_entry(sha1, &e))
1017                         return error("unable to find %s", sha1_to_hex(sha1));
1018                 return packed_object_info(&e, type, sizep);
1019         }
1020         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
1021                 status = error("unable to unpack %s header",
1022                                sha1_to_hex(sha1));
1023         if (parse_sha1_header(hdr, type, &size) < 0)
1024                 status = error("unable to parse %s header", sha1_to_hex(sha1));
1025         else {
1026                 status = 0;
1027                 if (sizep)
1028                         *sizep = size;
1029         }
1030         inflateEnd(&stream);
1031         munmap(map, mapsize);
1032         return status;
1033 }
1034
1035 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
1036 {
1037         struct pack_entry e;
1038
1039         if (!find_pack_entry(sha1, &e)) {
1040                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
1041                 return NULL;
1042         }
1043         return unpack_entry(&e, type, size);
1044 }
1045
1046 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
1047 {
1048         unsigned long mapsize;
1049         void *map, *buf;
1050
1051         map = map_sha1_file_internal(sha1, &mapsize, 0);
1052         if (map) {
1053                 buf = unpack_sha1_file(map, mapsize, type, size);
1054                 munmap(map, mapsize);
1055                 return buf;
1056         }
1057         return read_packed_sha1(sha1, type, size);
1058 }
1059
1060 void *read_object_with_reference(const unsigned char *sha1,
1061                                  const char *required_type,
1062                                  unsigned long *size,
1063                                  unsigned char *actual_sha1_return)
1064 {
1065         char type[20];
1066         void *buffer;
1067         unsigned long isize;
1068         unsigned char actual_sha1[20];
1069
1070         memcpy(actual_sha1, sha1, 20);
1071         while (1) {
1072                 int ref_length = -1;
1073                 const char *ref_type = NULL;
1074
1075                 buffer = read_sha1_file(actual_sha1, type, &isize);
1076                 if (!buffer)
1077                         return NULL;
1078                 if (!strcmp(type, required_type)) {
1079                         *size = isize;
1080                         if (actual_sha1_return)
1081                                 memcpy(actual_sha1_return, actual_sha1, 20);
1082                         return buffer;
1083                 }
1084                 /* Handle references */
1085                 else if (!strcmp(type, "commit"))
1086                         ref_type = "tree ";
1087                 else if (!strcmp(type, "tag"))
1088                         ref_type = "object ";
1089                 else {
1090                         free(buffer);
1091                         return NULL;
1092                 }
1093                 ref_length = strlen(ref_type);
1094
1095                 if (memcmp(buffer, ref_type, ref_length) ||
1096                     get_sha1_hex(buffer + ref_length, actual_sha1)) {
1097                         free(buffer);
1098                         return NULL;
1099                 }
1100                 /* Now we have the ID of the referred-to object in
1101                  * actual_sha1.  Check again. */
1102         }
1103 }
1104
1105 static char *write_sha1_file_prepare(void *buf,
1106                                      unsigned long len,
1107                                      const char *type,
1108                                      unsigned char *sha1,
1109                                      unsigned char *hdr,
1110                                      int *hdrlen)
1111 {
1112         SHA_CTX c;
1113
1114         /* Generate the header */
1115         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
1116
1117         /* Sha1.. */
1118         SHA1_Init(&c);
1119         SHA1_Update(&c, hdr, *hdrlen);
1120         SHA1_Update(&c, buf, len);
1121         SHA1_Final(sha1, &c);
1122
1123         return sha1_file_name(sha1);
1124 }
1125
1126 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1127 {
1128         int size;
1129         unsigned char *compressed;
1130         z_stream stream;
1131         unsigned char sha1[20];
1132         char *filename;
1133         static char tmpfile[PATH_MAX];
1134         unsigned char hdr[50];
1135         int fd, hdrlen, ret;
1136
1137         /* Normally if we have it in the pack then we do not bother writing
1138          * it out into .git/objects/??/?{38} file.
1139          */
1140         filename = write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1141         if (returnsha1)
1142                 memcpy(returnsha1, sha1, 20);
1143         if (has_sha1_file(sha1))
1144                 return 0;
1145         fd = open(filename, O_RDONLY);
1146         if (fd >= 0) {
1147                 /*
1148                  * FIXME!!! We might do collision checking here, but we'd
1149                  * need to uncompress the old file and check it. Later.
1150                  */
1151                 close(fd);
1152                 return 0;
1153         }
1154
1155         if (errno != ENOENT) {
1156                 fprintf(stderr, "sha1 file %s: %s", filename, strerror(errno));
1157                 return -1;
1158         }
1159
1160         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1161
1162         fd = mkstemp(tmpfile);
1163         if (fd < 0) {
1164                 fprintf(stderr, "unable to create temporary sha1 filename %s: %s", tmpfile, strerror(errno));
1165                 return -1;
1166         }
1167
1168         /* Set it up */
1169         memset(&stream, 0, sizeof(stream));
1170         deflateInit(&stream, Z_BEST_COMPRESSION);
1171         size = deflateBound(&stream, len+hdrlen);
1172         compressed = xmalloc(size);
1173
1174         /* Compress it */
1175         stream.next_out = compressed;
1176         stream.avail_out = size;
1177
1178         /* First header.. */
1179         stream.next_in = hdr;
1180         stream.avail_in = hdrlen;
1181         while (deflate(&stream, 0) == Z_OK)
1182                 /* nothing */;
1183
1184         /* Then the data itself.. */
1185         stream.next_in = buf;
1186         stream.avail_in = len;
1187         while (deflate(&stream, Z_FINISH) == Z_OK)
1188                 /* nothing */;
1189         deflateEnd(&stream);
1190         size = stream.total_out;
1191
1192         if (write(fd, compressed, size) != size)
1193                 die("unable to write file");
1194         fchmod(fd, 0444);
1195         close(fd);
1196         free(compressed);
1197
1198         ret = link(tmpfile, filename);
1199         if (ret < 0) {
1200                 ret = errno;
1201
1202                 /*
1203                  * Coda hack - coda doesn't like cross-directory links,
1204                  * so we fall back to a rename, which will mean that it
1205                  * won't be able to check collisions, but that's not a
1206                  * big deal.
1207                  *
1208                  * When this succeeds, we just return 0. We have nothing
1209                  * left to unlink.
1210                  */
1211                 if (ret == EXDEV && !rename(tmpfile, filename))
1212                         return 0;
1213         }
1214         unlink(tmpfile);
1215         if (ret) {
1216                 if (ret != EEXIST) {
1217                         fprintf(stderr, "unable to write sha1 filename %s: %s", filename, strerror(ret));
1218                         return -1;
1219                 }
1220                 /* FIXME!!! Collision check here ? */
1221         }
1222
1223         return 0;
1224 }
1225
1226 int write_sha1_from_fd(const unsigned char *sha1, int fd)
1227 {
1228         char *filename = sha1_file_name(sha1);
1229
1230         int local;
1231         z_stream stream;
1232         unsigned char real_sha1[20];
1233         unsigned char buf[4096];
1234         unsigned char discard[4096];
1235         int ret;
1236         SHA_CTX c;
1237
1238         local = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
1239
1240         if (local < 0)
1241                 return error("Couldn't open %s\n", filename);
1242
1243         memset(&stream, 0, sizeof(stream));
1244
1245         inflateInit(&stream);
1246
1247         SHA1_Init(&c);
1248
1249         do {
1250                 ssize_t size;
1251                 size = read(fd, buf, 4096);
1252                 if (size <= 0) {
1253                         close(local);
1254                         unlink(filename);
1255                         if (!size)
1256                                 return error("Connection closed?");
1257                         perror("Reading from connection");
1258                         return -1;
1259                 }
1260                 write(local, buf, size);
1261                 stream.avail_in = size;
1262                 stream.next_in = buf;
1263                 do {
1264                         stream.next_out = discard;
1265                         stream.avail_out = sizeof(discard);
1266                         ret = inflate(&stream, Z_SYNC_FLUSH);
1267                         SHA1_Update(&c, discard, sizeof(discard) -
1268                                     stream.avail_out);
1269                 } while (stream.avail_in && ret == Z_OK);
1270
1271         } while (ret == Z_OK);
1272         inflateEnd(&stream);
1273
1274         close(local);
1275         SHA1_Final(real_sha1, &c);
1276         if (ret != Z_STREAM_END) {
1277                 unlink(filename);
1278                 return error("File %s corrupted", sha1_to_hex(sha1));
1279         }
1280         if (memcmp(sha1, real_sha1, 20)) {
1281                 unlink(filename);
1282                 return error("File %s has bad hash\n", sha1_to_hex(sha1));
1283         }
1284
1285         return 0;
1286 }
1287
1288 int has_sha1_pack(const unsigned char *sha1)
1289 {
1290         struct pack_entry e;
1291         return find_pack_entry(sha1, &e);
1292 }
1293
1294 int has_sha1_file(const unsigned char *sha1)
1295 {
1296         struct stat st;
1297         struct pack_entry e;
1298
1299         if (find_sha1_file(sha1, &st))
1300                 return 1;
1301         return find_pack_entry(sha1, &e);
1302 }
1303
1304 int index_fd(unsigned char *sha1, int fd, struct stat *st)
1305 {
1306         unsigned long size = st->st_size;
1307         void *buf;
1308         int ret;
1309
1310         buf = "";
1311         if (size)
1312                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1313         close(fd);
1314         if ((int)(long)buf == -1)
1315                 return -1;
1316
1317         ret = write_sha1_file(buf, size, "blob", sha1);
1318         if (size)
1319                 munmap(buf, size);
1320         return ret;
1321 }