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], *ret;
 108         va_list args;
 109         int len;
 110
 111         if (!git_dir)
 112                 setup_git_env();
 113         len = strlen(git_dir);
 114         if (len > PATH_MAX-100)
 115                 return "pad-path";
 116         memcpy(pathname, git_dir, len);
 117         if (len && git_dir[len-1] != '/')
 118                 pathname[len++] = '/';
 119         va_start(args, fmt);
 120         vsnprintf(pathname + len, sizeof(pathname) - len, fmt, args);
 121         va_end(args);
 122         ret = pathname;
 123
 124         /* Clean it up */
 125         if (!memcmp(pathname, "./", 2)) {
 126                 ret += 2;
 127                 while (*ret == '/')
 128                         ret++;
 129         }
 130         return ret;
 131 }
 132
 133 int get_sha1(const char *str, unsigned char *sha1)
 134 {
 135         static const char *prefix[] = {
 136                 "",
 137                 "refs",
 138                 "refs/tags",
 139                 "refs/heads",
 140                 "refs/snap",
 141                 NULL
 142         };
 143         const char **p;
 144
 145         if (!get_sha1_hex(str, sha1))
 146                 return 0;
 147
 148         for (p = prefix; *p; p++) {
 149                 char * pathname = git_path("%s/%s", *p, str);
 150                 if (!get_sha1_file(pathname, sha1))
 151                         return 0;
 152         }
 153
 154         return -1;
 155 }
 156
 157 char * sha1_to_hex(const unsigned char *sha1)
 158 {
 159         static char buffer[50];
 160         static const char hex[] = "0123456789abcdef";
 161         char *buf = buffer;
 162         int i;
 163
 164         for (i = 0; i < 20; i++) {
 165                 unsigned int val = *sha1++;
 166                 *buf++ = hex[val >> 4];
 167                 *buf++ = hex[val & 0xf];
 168         }
 169         return buffer;
 170 }
 171
 172 static void fill_sha1_path(char *pathbuf, const unsigned char *sha1)
 173 {
 174         int i;
 175         for (i = 0; i < 20; i++) {
 176                 static char hex[] = "0123456789abcdef";
 177                 unsigned int val = sha1[i];
 178                 char *pos = pathbuf + i*2 + (i > 0);
 179                 *pos++ = hex[val >> 4];
 180                 *pos = hex[val & 0xf];
 181         }
 182 }
 183
 184 /*
 185  * NOTE! This returns a statically allocated buffer, so you have to be
 186  * careful about using it. Do a "strdup()" if you need to save the
 187  * filename.
 188  *
 189  * Also note that this returns the location for creating.  Reading
 190  * SHA1 file can happen from any alternate directory listed in the
 191  * DB_ENVIRONMENT environment variable if it is not found in
 192  * the primary object database.
 193  */
 194 char *sha1_file_name(const unsigned char *sha1)
 195 {
 196         static char *name, *base;
 197
 198         if (!base) {
 199                 const char *sha1_file_directory = get_object_directory();
 200                 int len = strlen(sha1_file_directory);
 201                 base = xmalloc(len + 60);
 202                 memcpy(base, sha1_file_directory, len);
 203                 memset(base+len, 0, 60);
 204                 base[len] = '/';
 205                 base[len+3] = '/';
 206                 name = base + len + 1;
 207         }
 208         fill_sha1_path(name, sha1);
 209         return base;
 210 }
 211
 212 struct alternate_object_database *alt_odb;
 213
 214 /*
 215  * Prepare alternate object database registry.
 216  * alt_odb points at an array of struct alternate_object_database.
 217  * This array is terminated with an element that has both its base
 218  * and name set to NULL.  alt_odb[n] comes from n'th non-empty
 219  * element from colon separated ALTERNATE_DB_ENVIRONMENT environment
 220  * variable, and its base points at a statically allocated buffer
 221  * that contains "/the/directory/corresponding/to/.git/objects/...",
 222  * while its name points just after the slash at the end of
 223  * ".git/objects/" in the example above, and has enough space to hold
 224  * 40-byte hex SHA1, an extra slash for the first level indirection,
 225  * and the terminating NUL.
 226  * This function allocates the alt_odb array and all the strings
 227  * pointed by base fields of the array elements with one xmalloc();
 228  * the string pool immediately follows the array.
 229  */
 230 void prepare_alt_odb(void)
 231 {
 232         int pass, totlen, i;
 233         const char *cp, *last;
 234         char *op = NULL;
 235         const char *alt = gitenv(ALTERNATE_DB_ENVIRONMENT) ? : "";
 236
 237         if (alt_odb)
 238                 return;
 239         /* The first pass counts how large an area to allocate to
 240          * hold the entire alt_odb structure, including array of
 241          * structs and path buffers for them.  The second pass fills
 242          * the structure and prepares the path buffers for use by
 243          * fill_sha1_path().
 244          */
 245         for (totlen = pass = 0; pass < 2; pass++) {
 246                 last = alt;
 247                 i = 0;
 248                 do {
 249                         cp = strchr(last, ':') ? : last + strlen(last);
 250                         if (last != cp) {
 251                                 /* 43 = 40-byte + 2 '/' + terminating NUL */
 252                                 int pfxlen = cp - last;
 253                                 int entlen = pfxlen + 43;
 254                                 if (pass == 0)
 255                                         totlen += entlen;
 256                                 else {
 257                                         alt_odb[i].base = op;
 258                                         alt_odb[i].name = op + pfxlen + 1;
 259                                         memcpy(op, last, pfxlen);
 260                                         op[pfxlen] = op[pfxlen + 3] = '/';
 261                                         op[entlen-1] = 0;
 262                                         op += entlen;
 263                                 }
 264                                 i++;
 265                         }
 266                         while (*cp && *cp == ':')
 267                                 cp++;
 268                         last = cp;
 269                 } while (*cp);
 270                 if (pass)
 271                         break;
 272                 alt_odb = xmalloc(sizeof(*alt_odb) * (i + 1) + totlen);
 273                 alt_odb[i].base = alt_odb[i].name = NULL;
 274                 op = (char*)(&alt_odb[i+1]);
 275         }
 276 }
 277
 278 static char *find_sha1_file(const unsigned char *sha1, struct stat *st)
 279 {
 280         int i;
 281         char *name = sha1_file_name(sha1);
 282
 283         if (!stat(name, st))
 284                 return name;
 285         prepare_alt_odb();
 286         for (i = 0; (name = alt_odb[i].name) != NULL; i++) {
 287                 fill_sha1_path(name, sha1);
 288                 if (!stat(alt_odb[i].base, st))
 289                         return alt_odb[i].base;
 290         }
 291         return NULL;
 292 }
 293
 294 #define PACK_MAX_SZ (1<<26)
 295 static int pack_used_ctr;
 296 static unsigned long pack_mapped;
 297 struct packed_git *packed_git;
 298
 299 static int check_packed_git_idx(const char *path, unsigned long *idx_size_,
 300                                 void **idx_map_)
 301 {
 302         void *idx_map;
 303         unsigned int *index;
 304         unsigned long idx_size;
 305         int nr, i;
 306         int fd = open(path, O_RDONLY);
 307         struct stat st;
 308         if (fd < 0)
 309                 return -1;
 310         if (fstat(fd, &st)) {
 311                 close(fd);
 312                 return -1;
 313         }
 314         idx_size = st.st_size;
 315         idx_map = mmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0);
 316         close(fd);
 317         if (idx_map == MAP_FAILED)
 318                 return -1;
 319
 320         index = idx_map;
 321         *idx_map_ = idx_map;
 322         *idx_size_ = idx_size;
 323
 324         /* check index map */
 325         if (idx_size < 4*256 + 20 + 20)
 326                 return error("index file too small");
 327         nr = 0;
 328         for (i = 0; i < 256; i++) {
 329                 unsigned int n = ntohl(index[i]);
 330                 if (n < nr)
 331                         return error("non-monotonic index");
 332                 nr = n;
 333         }
 334
 335         /*
 336          * Total size:
 337          *  - 256 index entries 4 bytes each
 338          *  - 24-byte entries * nr (20-byte sha1 + 4-byte offset)
 339          *  - 20-byte SHA1 of the packfile
 340          *  - 20-byte SHA1 file checksum
 341          */
 342         if (idx_size != 4*256 + nr * 24 + 20 + 20)
 343                 return error("wrong index file size");
 344
 345         return 0;
 346 }
 347
 348 static int unuse_one_packed_git(void)
 349 {
 350         struct packed_git *p, *lru = NULL;
 351
 352         for (p = packed_git; p; p = p->next) {
 353                 if (p->pack_use_cnt || !p->pack_base)
 354                         continue;
 355                 if (!lru || p->pack_last_used < lru->pack_last_used)
 356                         lru = p;
 357         }
 358         if (!lru)
 359                 return 0;
 360         munmap(lru->pack_base, lru->pack_size);
 361         lru->pack_base = NULL;
 362         return 1;
 363 }
 364
 365 void unuse_packed_git(struct packed_git *p)
 366 {
 367         p->pack_use_cnt--;
 368 }
 369
 370 int use_packed_git(struct packed_git *p)
 371 {
 372         if (!p->pack_base) {
 373                 int fd;
 374                 struct stat st;
 375                 void *map;
 376
 377                 pack_mapped += p->pack_size;
 378                 while (PACK_MAX_SZ < pack_mapped && unuse_one_packed_git())
 379                         ; /* nothing */
 380                 fd = open(p->pack_name, O_RDONLY);
 381                 if (fd < 0)
 382                         die("packfile %s cannot be opened", p->pack_name);
 383                 if (fstat(fd, &st)) {
 384                         close(fd);
 385                         die("packfile %s cannot be opened", p->pack_name);
 386                 }
 387                 if (st.st_size != p->pack_size)
 388                         die("packfile %s size mismatch.", p->pack_name);
 389                 map = mmap(NULL, p->pack_size, PROT_READ, MAP_PRIVATE, fd, 0);
 390                 close(fd);
 391                 if (map == MAP_FAILED)
 392                         die("packfile %s cannot be mapped.", p->pack_name);
 393                 p->pack_base = map;
 394
 395                 /* Check if the pack file matches with the index file.
 396                  * this is cheap.
 397                  */
 398                 if (memcmp((char*)(p->index_base) + p->index_size - 40,
 399                            p->pack_base + p->pack_size - 20, 20))
 400                         die("packfile %s does not match index.", p->pack_name);
 401         }
 402         p->pack_last_used = pack_used_ctr++;
 403         p->pack_use_cnt++;
 404         return 0;
 405 }
 406
 407 struct packed_git *add_packed_git(char *path, int path_len)
 408 {
 409         struct stat st;
 410         struct packed_git *p;
 411         unsigned long idx_size;
 412         void *idx_map;
 413
 414         if (check_packed_git_idx(path, &idx_size, &idx_map))
 415                 return NULL;
 416
 417         /* do we have a corresponding .pack file? */
 418         strcpy(path + path_len - 4, ".pack");
 419         if (stat(path, &st) || !S_ISREG(st.st_mode)) {
 420                 munmap(idx_map, idx_size);
 421                 return NULL;
 422         }
 423         /* ok, it looks sane as far as we can check without
 424          * actually mapping the pack file.
 425          */
 426         p = xmalloc(sizeof(*p) + path_len + 2);
 427         strcpy(p->pack_name, path);
 428         p->index_size = idx_size;
 429         p->pack_size = st.st_size;
 430         p->index_base = idx_map;
 431         p->next = NULL;
 432         p->pack_base = NULL;
 433         p->pack_last_used = 0;
 434         p->pack_use_cnt = 0;
 435         return p;
 436 }
 437
 438 static void prepare_packed_git_one(char *objdir)
 439 {
 440         char path[PATH_MAX];
 441         int len;
 442         DIR *dir;
 443         struct dirent *de;
 444
 445         sprintf(path, "%s/pack", objdir);
 446         len = strlen(path);
 447         dir = opendir(path);
 448         if (!dir)
 449                 return;
 450         path[len++] = '/';
 451         while ((de = readdir(dir)) != NULL) {
 452                 int namelen = strlen(de->d_name);
 453                 struct packed_git *p;
 454
 455                 if (strcmp(de->d_name + namelen - 4, ".idx"))
 456                         continue;
 457
 458                 /* we have .idx.  Is it a file we can map? */
 459                 strcpy(path + len, de->d_name);
 460                 p = add_packed_git(path, len + namelen);
 461                 if (!p)
 462                         continue;
 463                 p->next = packed_git;
 464                 packed_git = p;
 465         }
 466         closedir(dir);
 467 }
 468
 469 void prepare_packed_git(void)
 470 {
 471         int i;
 472         static int run_once = 0;
 473
 474         if (run_once++)
 475                 return;
 476
 477         prepare_packed_git_one(get_object_directory());
 478         prepare_alt_odb();
 479         for (i = 0; alt_odb[i].base != NULL; i++) {
 480                 alt_odb[i].name[0] = 0;
 481                 prepare_packed_git_one(alt_odb[i].base);
 482         }
 483 }
 484
 485 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 486 {
 487         char header[100];
 488         unsigned char real_sha1[20];
 489         SHA_CTX c;
 490
 491         SHA1_Init(&c);
 492         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 493         SHA1_Update(&c, map, size);
 494         SHA1_Final(real_sha1, &c);
 495         return memcmp(sha1, real_sha1, 20) ? -1 : 0;
 496 }
 497
 498 static void *map_sha1_file_internal(const unsigned char *sha1,
 499                                     unsigned long *size,
 500                                     int say_error)
 501 {
 502         struct stat st;
 503         void *map;
 504         int fd;
 505         char *filename = find_sha1_file(sha1, &st);
 506
 507         if (!filename) {
 508                 if (say_error)
 509                         error("cannot map sha1 file %s", sha1_to_hex(sha1));
 510                 return NULL;
 511         }
 512
 513         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 514         if (fd < 0) {
 515                 /* See if it works without O_NOATIME */
 516                 switch (sha1_file_open_flag) {
 517                 default:
 518                         fd = open(filename, O_RDONLY);
 519                         if (fd >= 0)
 520                                 break;
 521                 /* Fallthrough */
 522                 case 0:
 523                         if (say_error)
 524                                 perror(filename);
 525                         return NULL;
 526                 }
 527
 528                 /* If it failed once, it will probably fail again.
 529                  * Stop using O_NOATIME
 530                  */
 531                 sha1_file_open_flag = 0;
 532         }
 533         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 534         close(fd);
 535         if (-1 == (int)(long)map)
 536                 return NULL;
 537         *size = st.st_size;
 538         return map;
 539 }
 540
 541 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 542 {
 543         return map_sha1_file_internal(sha1, size, 1);
 544 }
 545
 546 int unpack_sha1_header(z_stream *stream, void *map, unsigned long mapsize, void *buffer, unsigned long size)
 547 {
 548         /* Get the data stream */
 549         memset(stream, 0, sizeof(*stream));
 550         stream->next_in = map;
 551         stream->avail_in = mapsize;
 552         stream->next_out = buffer;
 553         stream->avail_out = size;
 554
 555         inflateInit(stream);
 556         return inflate(stream, 0);
 557 }
 558
 559 static void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 560 {
 561         int bytes = strlen(buffer) + 1;
 562         unsigned char *buf = xmalloc(1+size);
 563
 564         memcpy(buf, buffer + bytes, stream->total_out - bytes);
 565         bytes = stream->total_out - bytes;
 566         if (bytes < size) {
 567                 stream->next_out = buf + bytes;
 568                 stream->avail_out = size - bytes;
 569                 while (inflate(stream, Z_FINISH) == Z_OK)
 570                         /* nothing */;
 571         }
 572         buf[size] = 0;
 573         inflateEnd(stream);
 574         return buf;
 575 }
 576
 577 /*
 578  * We used to just use "sscanf()", but that's actually way
 579  * too permissive for what we want to check. So do an anal
 580  * object header parse by hand.
 581  */
 582 int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 583 {
 584         int i;
 585         unsigned long size;
 586
 587         /*
 588          * The type can be at most ten bytes (including the
 589          * terminating '\0' that we add), and is followed by
 590          * a space.
 591          */
 592         i = 10;
 593         for (;;) {
 594                 char c = *hdr++;
 595                 if (c == ' ')
 596                         break;
 597                 if (!--i)
 598                         return -1;
 599                 *type++ = c;
 600         }
 601         *type = 0;
 602
 603         /*
 604          * The length must follow immediately, and be in canonical
 605          * decimal format (ie "010" is not valid).
 606          */
 607         size = *hdr++ - '0';
 608         if (size > 9)
 609                 return -1;
 610         if (size) {
 611                 for (;;) {
 612                         unsigned long c = *hdr - '0';
 613                         if (c > 9)
 614                                 break;
 615                         hdr++;
 616                         size = size * 10 + c;
 617                 }
 618         }
 619         *sizep = size;
 620
 621         /*
 622          * The length must be followed by a zero byte
 623          */
 624         return *hdr ? -1 : 0;
 625 }
 626
 627 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 628 {
 629         int ret;
 630         z_stream stream;
 631         char hdr[8192];
 632
 633         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
 634         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
 635                 return NULL;
 636
 637         return unpack_sha1_rest(&stream, hdr, *size);
 638 }
 639
 640 /* forward declaration for a mutually recursive function */
 641 static int packed_object_info(struct pack_entry *entry,
 642                               char *type, unsigned long *sizep);
 643
 644 static int packed_delta_info(unsigned char *base_sha1,
 645                              unsigned long delta_size,
 646                              unsigned long left,
 647                              char *type,
 648                              unsigned long *sizep,
 649                              struct packed_git *p)
 650 {
 651         struct pack_entry base_ent;
 652
 653         if (left < 20)
 654                 die("truncated pack file");
 655
 656         /* The base entry _must_ be in the same pack */
 657         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 658                 die("failed to find delta-pack base object %s",
 659                     sha1_to_hex(base_sha1));
 660
 661         /* We choose to only get the type of the base object and
 662          * ignore potentially corrupt pack file that expects the delta
 663          * based on a base with a wrong size.  This saves tons of
 664          * inflate() calls.
 665          */
 666
 667         if (packed_object_info(&base_ent, type, NULL))
 668                 die("cannot get info for delta-pack base");
 669
 670         if (sizep) {
 671                 const unsigned char *data;
 672                 unsigned char delta_head[64];
 673                 unsigned long result_size;
 674                 z_stream stream;
 675                 int st;
 676
 677                 memset(&stream, 0, sizeof(stream));
 678
 679                 data = stream.next_in = base_sha1 + 20;
 680                 stream.avail_in = left - 20;
 681                 stream.next_out = delta_head;
 682                 stream.avail_out = sizeof(delta_head);
 683
 684                 inflateInit(&stream);
 685                 st = inflate(&stream, Z_FINISH);
 686                 inflateEnd(&stream);
 687                 if ((st != Z_STREAM_END) &&
 688                     stream.total_out != sizeof(delta_head))
 689                         die("delta data unpack-initial failed");
 690
 691                 /* Examine the initial part of the delta to figure out
 692                  * the result size.
 693                  */
 694                 data = delta_head;
 695                 get_delta_hdr_size(&data); /* ignore base size */
 696
 697                 /* Read the result size */
 698                 result_size = get_delta_hdr_size(&data);
 699                 *sizep = result_size;
 700         }
 701         return 0;
 702 }
 703
 704 static unsigned long unpack_object_header(struct packed_git *p, unsigned long offset,
 705         enum object_type *type, unsigned long *sizep)
 706 {
 707         unsigned shift;
 708         unsigned char *pack, c;
 709         unsigned long size;
 710
 711         if (offset >= p->pack_size)
 712                 die("object offset outside of pack file");
 713
 714         pack =  p->pack_base + offset;
 715         c = *pack++;
 716         offset++;
 717         *type = (c >> 4) & 7;
 718         size = c & 15;
 719         shift = 4;
 720         while (c & 0x80) {
 721                 if (offset >= p->pack_size)
 722                         die("object offset outside of pack file");
 723                 c = *pack++;
 724                 offset++;
 725                 size += (c & 0x7f) << shift;
 726                 shift += 7;
 727         }
 728         *sizep = size;
 729         return offset;
 730 }
 731
 732 void packed_object_info_detail(struct pack_entry *e,
 733                                char *type,
 734                                unsigned long *size,
 735                                unsigned long *store_size,
 736                                int *delta_chain_length,
 737                                unsigned char *base_sha1)
 738 {
 739         struct packed_git *p = e->p;
 740         unsigned long offset, left;
 741         unsigned char *pack;
 742         enum object_type kind;
 743
 744         offset = unpack_object_header(p, e->offset, &kind, size);
 745         pack = p->pack_base + offset;
 746         left = p->pack_size - offset;
 747         if (kind != OBJ_DELTA)
 748                 *delta_chain_length = 0;
 749         else {
 750                 int chain_length = 0;
 751                 memcpy(base_sha1, pack, 20);
 752                 do {
 753                         struct pack_entry base_ent;
 754                         unsigned long junk;
 755
 756                         find_pack_entry_one(pack, &base_ent, p);
 757                         offset = unpack_object_header(p, base_ent.offset,
 758                                                       &kind, &junk);
 759                         pack = p->pack_base + offset;
 760                         chain_length++;
 761                 } while (kind == OBJ_DELTA);
 762                 *delta_chain_length = chain_length;
 763         }
 764         switch (kind) {
 765         case OBJ_COMMIT:
 766                 strcpy(type, "commit");
 767                 break;
 768         case OBJ_TREE:
 769                 strcpy(type, "tree");
 770                 break;
 771         case OBJ_BLOB:
 772                 strcpy(type, "blob");
 773                 break;
 774         case OBJ_TAG:
 775                 strcpy(type, "tag");
 776                 break;
 777         default:
 778                 die("corrupted pack file");
 779         }
 780         *store_size = 0; /* notyet */
 781 }
 782
 783 static int packed_object_info(struct pack_entry *entry,
 784                               char *type, unsigned long *sizep)
 785 {
 786         struct packed_git *p = entry->p;
 787         unsigned long offset, size, left;
 788         unsigned char *pack;
 789         enum object_type kind;
 790         int retval;
 791
 792         if (use_packed_git(p))
 793                 die("cannot map packed file");
 794
 795         offset = unpack_object_header(p, entry->offset, &kind, &size);
 796         pack = p->pack_base + offset;
 797         left = p->pack_size - offset;
 798
 799         switch (kind) {
 800         case OBJ_DELTA:
 801                 retval = packed_delta_info(pack, size, left, type, sizep, p);
 802                 unuse_packed_git(p);
 803                 return retval;
 804         case OBJ_COMMIT:
 805                 strcpy(type, "commit");
 806                 break;
 807         case OBJ_TREE:
 808                 strcpy(type, "tree");
 809                 break;
 810         case OBJ_BLOB:
 811                 strcpy(type, "blob");
 812                 break;
 813         case OBJ_TAG:
 814                 strcpy(type, "tag");
 815                 break;
 816         default:
 817                 die("corrupted pack file");
 818         }
 819         if (sizep)
 820                 *sizep = size;
 821         unuse_packed_git(p);
 822         return 0;
 823 }
 824
 825 /* forward declaration for a mutually recursive function */
 826 static void *unpack_entry(struct pack_entry *, char *, unsigned long *);
 827
 828 static void *unpack_delta_entry(unsigned char *base_sha1,
 829                                 unsigned long delta_size,
 830                                 unsigned long left,
 831                                 char *type,
 832                                 unsigned long *sizep,
 833                                 struct packed_git *p)
 834 {
 835         struct pack_entry base_ent;
 836         void *data, *delta_data, *result, *base;
 837         unsigned long data_size, result_size, base_size;
 838         z_stream stream;
 839         int st;
 840
 841         if (left < 20)
 842                 die("truncated pack file");
 843         data = base_sha1 + 20;
 844         data_size = left - 20;
 845         delta_data = xmalloc(delta_size);
 846
 847         memset(&stream, 0, sizeof(stream));
 848
 849         stream.next_in = data;
 850         stream.avail_in = data_size;
 851         stream.next_out = delta_data;
 852         stream.avail_out = delta_size;
 853
 854         inflateInit(&stream);
 855         st = inflate(&stream, Z_FINISH);
 856         inflateEnd(&stream);
 857         if ((st != Z_STREAM_END) || stream.total_out != delta_size)
 858                 die("delta data unpack failed");
 859
 860         /* The base entry _must_ be in the same pack */
 861         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 862                 die("failed to find delta-pack base object %s",
 863                     sha1_to_hex(base_sha1));
 864         base = unpack_entry_gently(&base_ent, type, &base_size);
 865         if (!base)
 866                 die("failed to read delta-pack base object %s",
 867                     sha1_to_hex(base_sha1));
 868         result = patch_delta(base, base_size,
 869                              delta_data, delta_size,
 870                              &result_size);
 871         if (!result)
 872                 die("failed to apply delta");
 873         free(delta_data);
 874         free(base);
 875         *sizep = result_size;
 876         return result;
 877 }
 878
 879 static void *unpack_non_delta_entry(unsigned char *data,
 880                                     unsigned long size,
 881                                     unsigned long left)
 882 {
 883         int st;
 884         z_stream stream;
 885         unsigned char *buffer;
 886
 887         buffer = xmalloc(size + 1);
 888         buffer[size] = 0;
 889         memset(&stream, 0, sizeof(stream));
 890         stream.next_in = data;
 891         stream.avail_in = left;
 892         stream.next_out = buffer;
 893         stream.avail_out = size;
 894
 895         inflateInit(&stream);
 896         st = inflate(&stream, Z_FINISH);
 897         inflateEnd(&stream);
 898         if ((st != Z_STREAM_END) || stream.total_out != size) {
 899                 free(buffer);
 900                 return NULL;
 901         }
 902
 903         return buffer;
 904 }
 905
 906 static void *unpack_entry(struct pack_entry *entry,
 907                           char *type, unsigned long *sizep)
 908 {
 909         struct packed_git *p = entry->p;
 910         void *retval;
 911
 912         if (use_packed_git(p))
 913                 die("cannot map packed file");
 914         retval = unpack_entry_gently(entry, type, sizep);
 915         unuse_packed_git(p);
 916         if (!retval)
 917                 die("corrupted pack file");
 918         return retval;
 919 }
 920
 921 /* The caller is responsible for use_packed_git()/unuse_packed_git() pair */
 922 void *unpack_entry_gently(struct pack_entry *entry,
 923                           char *type, unsigned long *sizep)
 924 {
 925         struct packed_git *p = entry->p;
 926         unsigned long offset, size, left;
 927         unsigned char *pack;
 928         enum object_type kind;
 929         void *retval;
 930
 931         offset = unpack_object_header(p, entry->offset, &kind, &size);
 932         pack = p->pack_base + offset;
 933         left = p->pack_size - offset;
 934         switch (kind) {
 935         case OBJ_DELTA:
 936                 retval = unpack_delta_entry(pack, size, left, type, sizep, p);
 937                 return retval;
 938         case OBJ_COMMIT:
 939                 strcpy(type, "commit");
 940                 break;
 941         case OBJ_TREE:
 942                 strcpy(type, "tree");
 943                 break;
 944         case OBJ_BLOB:
 945                 strcpy(type, "blob");
 946                 break;
 947         case OBJ_TAG:
 948                 strcpy(type, "tag");
 949                 break;
 950         default:
 951                 return NULL;
 952         }
 953         *sizep = size;
 954         retval = unpack_non_delta_entry(pack, size, left);
 955         return retval;
 956 }
 957
 958 int num_packed_objects(const struct packed_git *p)
 959 {
 960         /* See check_packed_git_idx() */
 961         return (p->index_size - 20 - 20 - 4*256) / 24;
 962 }
 963
 964 int nth_packed_object_sha1(const struct packed_git *p, int n,
 965                            unsigned char* sha1)
 966 {
 967         void *index = p->index_base + 256;
 968         if (n < 0 || num_packed_objects(p) <= n)
 969                 return -1;
 970         memcpy(sha1, (index + 24 * n + 4), 20);
 971         return 0;
 972 }
 973
 974 int find_pack_entry_one(const unsigned char *sha1,
 975                         struct pack_entry *e, struct packed_git *p)
 976 {
 977         unsigned int *level1_ofs = p->index_base;
 978         int hi = ntohl(level1_ofs[*sha1]);
 979         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
 980         void *index = p->index_base + 256;
 981
 982         do {
 983                 int mi = (lo + hi) / 2;
 984                 int cmp = memcmp(index + 24 * mi + 4, sha1, 20);
 985                 if (!cmp) {
 986                         e->offset = ntohl(*((int*)(index + 24 * mi)));
 987                         memcpy(e->sha1, sha1, 20);
 988                         e->p = p;
 989                         return 1;
 990                 }
 991                 if (cmp > 0)
 992                         hi = mi;
 993                 else
 994                         lo = mi+1;
 995         } while (lo < hi);
 996         return 0;
 997 }
 998
 999 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e)
1000 {
1001         struct packed_git *p;
1002         prepare_packed_git();
1003
1004         for (p = packed_git; p; p = p->next) {
1005                 if (find_pack_entry_one(sha1, e, p))
1006                         return 1;
1007         }
1008         return 0;
1009 }
1010
1011 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
1012 {
1013         int status;
1014         unsigned long mapsize, size;
1015         void *map;
1016         z_stream stream;
1017         char hdr[128];
1018
1019         map = map_sha1_file_internal(sha1, &mapsize, 0);
1020         if (!map) {
1021                 struct pack_entry e;
1022
1023                 if (!find_pack_entry(sha1, &e))
1024                         return error("unable to find %s", sha1_to_hex(sha1));
1025                 return packed_object_info(&e, type, sizep);
1026         }
1027         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
1028                 status = error("unable to unpack %s header",
1029                                sha1_to_hex(sha1));
1030         if (parse_sha1_header(hdr, type, &size) < 0)
1031                 status = error("unable to parse %s header", sha1_to_hex(sha1));
1032         else {
1033                 status = 0;
1034                 if (sizep)
1035                         *sizep = size;
1036         }
1037         inflateEnd(&stream);
1038         munmap(map, mapsize);
1039         return status;
1040 }
1041
1042 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
1043 {
1044         struct pack_entry e;
1045
1046         if (!find_pack_entry(sha1, &e)) {
1047                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
1048                 return NULL;
1049         }
1050         return unpack_entry(&e, type, size);
1051 }
1052
1053 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
1054 {
1055         unsigned long mapsize;
1056         void *map, *buf;
1057
1058         map = map_sha1_file_internal(sha1, &mapsize, 0);
1059         if (map) {
1060                 buf = unpack_sha1_file(map, mapsize, type, size);
1061                 munmap(map, mapsize);
1062                 return buf;
1063         }
1064         return read_packed_sha1(sha1, type, size);
1065 }
1066
1067 void *read_object_with_reference(const unsigned char *sha1,
1068                                  const char *required_type,
1069                                  unsigned long *size,
1070                                  unsigned char *actual_sha1_return)
1071 {
1072         char type[20];
1073         void *buffer;
1074         unsigned long isize;
1075         unsigned char actual_sha1[20];
1076
1077         memcpy(actual_sha1, sha1, 20);
1078         while (1) {
1079                 int ref_length = -1;
1080                 const char *ref_type = NULL;
1081
1082                 buffer = read_sha1_file(actual_sha1, type, &isize);
1083                 if (!buffer)
1084                         return NULL;
1085                 if (!strcmp(type, required_type)) {
1086                         *size = isize;
1087                         if (actual_sha1_return)
1088                                 memcpy(actual_sha1_return, actual_sha1, 20);
1089                         return buffer;
1090                 }
1091                 /* Handle references */
1092                 else if (!strcmp(type, "commit"))
1093                         ref_type = "tree ";
1094                 else if (!strcmp(type, "tag"))
1095                         ref_type = "object ";
1096                 else {
1097                         free(buffer);
1098                         return NULL;
1099                 }
1100                 ref_length = strlen(ref_type);
1101
1102                 if (memcmp(buffer, ref_type, ref_length) ||
1103                     get_sha1_hex(buffer + ref_length, actual_sha1)) {
1104                         free(buffer);
1105                         return NULL;
1106                 }
1107                 /* Now we have the ID of the referred-to object in
1108                  * actual_sha1.  Check again. */
1109         }
1110 }
1111
1112 static char *write_sha1_file_prepare(void *buf,
1113                                      unsigned long len,
1114                                      const char *type,
1115                                      unsigned char *sha1,
1116                                      unsigned char *hdr,
1117                                      int *hdrlen)
1118 {
1119         SHA_CTX c;
1120
1121         /* Generate the header */
1122         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
1123
1124         /* Sha1.. */
1125         SHA1_Init(&c);
1126         SHA1_Update(&c, hdr, *hdrlen);
1127         SHA1_Update(&c, buf, len);
1128         SHA1_Final(sha1, &c);
1129
1130         return sha1_file_name(sha1);
1131 }
1132
1133 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1134 {
1135         int size;
1136         unsigned char *compressed;
1137         z_stream stream;
1138         unsigned char sha1[20];
1139         char *filename;
1140         static char tmpfile[PATH_MAX];
1141         unsigned char hdr[50];
1142         int fd, hdrlen, ret;
1143
1144         /* Normally if we have it in the pack then we do not bother writing
1145          * it out into .git/objects/??/?{38} file.
1146          */
1147         filename = write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1148         if (returnsha1)
1149                 memcpy(returnsha1, sha1, 20);
1150         if (has_sha1_file(sha1))
1151                 return 0;
1152         fd = open(filename, O_RDONLY);
1153         if (fd >= 0) {
1154                 /*
1155                  * FIXME!!! We might do collision checking here, but we'd
1156                  * need to uncompress the old file and check it. Later.
1157                  */
1158                 close(fd);
1159                 return 0;
1160         }
1161
1162         if (errno != ENOENT) {
1163                 fprintf(stderr, "sha1 file %s: %s", filename, strerror(errno));
1164                 return -1;
1165         }
1166
1167         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1168
1169         fd = mkstemp(tmpfile);
1170         if (fd < 0) {
1171                 fprintf(stderr, "unable to create temporary sha1 filename %s: %s", tmpfile, strerror(errno));
1172                 return -1;
1173         }
1174
1175         /* Set it up */
1176         memset(&stream, 0, sizeof(stream));
1177         deflateInit(&stream, Z_BEST_COMPRESSION);
1178         size = deflateBound(&stream, len+hdrlen);
1179         compressed = xmalloc(size);
1180
1181         /* Compress it */
1182         stream.next_out = compressed;
1183         stream.avail_out = size;
1184
1185         /* First header.. */
1186         stream.next_in = hdr;
1187         stream.avail_in = hdrlen;
1188         while (deflate(&stream, 0) == Z_OK)
1189                 /* nothing */;
1190
1191         /* Then the data itself.. */
1192         stream.next_in = buf;
1193         stream.avail_in = len;
1194         while (deflate(&stream, Z_FINISH) == Z_OK)
1195                 /* nothing */;
1196         deflateEnd(&stream);
1197         size = stream.total_out;
1198
1199         if (write(fd, compressed, size) != size)
1200                 die("unable to write file");
1201         fchmod(fd, 0444);
1202         close(fd);
1203         free(compressed);
1204
1205         ret = link(tmpfile, filename);
1206         if (ret < 0) {
1207                 ret = errno;
1208
1209                 /*
1210                  * Coda hack - coda doesn't like cross-directory links,
1211                  * so we fall back to a rename, which will mean that it
1212                  * won't be able to check collisions, but that's not a
1213                  * big deal.
1214                  *
1215                  * When this succeeds, we just return 0. We have nothing
1216                  * left to unlink.
1217                  */
1218                 if (ret == EXDEV && !rename(tmpfile, filename))
1219                         return 0;
1220         }
1221         unlink(tmpfile);
1222         if (ret) {
1223                 if (ret != EEXIST) {
1224                         fprintf(stderr, "unable to write sha1 filename %s: %s", filename, strerror(ret));
1225                         return -1;
1226                 }
1227                 /* FIXME!!! Collision check here ? */
1228         }
1229
1230         return 0;
1231 }
1232
1233 int write_sha1_from_fd(const unsigned char *sha1, int fd)
1234 {
1235         char *filename = sha1_file_name(sha1);
1236
1237         int local;
1238         z_stream stream;
1239         unsigned char real_sha1[20];
1240         unsigned char buf[4096];
1241         unsigned char discard[4096];
1242         int ret;
1243         SHA_CTX c;
1244
1245         local = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
1246
1247         if (local < 0)
1248                 return error("Couldn't open %s\n", filename);
1249
1250         memset(&stream, 0, sizeof(stream));
1251
1252         inflateInit(&stream);
1253
1254         SHA1_Init(&c);
1255
1256         do {
1257                 ssize_t size;
1258                 size = read(fd, buf, 4096);
1259                 if (size <= 0) {
1260                         close(local);
1261                         unlink(filename);
1262                         if (!size)
1263                                 return error("Connection closed?");
1264                         perror("Reading from connection");
1265                         return -1;
1266                 }
1267                 write(local, buf, size);
1268                 stream.avail_in = size;
1269                 stream.next_in = buf;
1270                 do {
1271                         stream.next_out = discard;
1272                         stream.avail_out = sizeof(discard);
1273                         ret = inflate(&stream, Z_SYNC_FLUSH);
1274                         SHA1_Update(&c, discard, sizeof(discard) -
1275                                     stream.avail_out);
1276                 } while (stream.avail_in && ret == Z_OK);
1277
1278         } while (ret == Z_OK);
1279         inflateEnd(&stream);
1280
1281         close(local);
1282         SHA1_Final(real_sha1, &c);
1283         if (ret != Z_STREAM_END) {
1284                 unlink(filename);
1285                 return error("File %s corrupted", sha1_to_hex(sha1));
1286         }
1287         if (memcmp(sha1, real_sha1, 20)) {
1288                 unlink(filename);
1289                 return error("File %s has bad hash\n", sha1_to_hex(sha1));
1290         }
1291
1292         return 0;
1293 }
1294
1295 int has_sha1_pack(const unsigned char *sha1)
1296 {
1297         struct pack_entry e;
1298         return find_pack_entry(sha1, &e);
1299 }
1300
1301 int has_sha1_file(const unsigned char *sha1)
1302 {
1303         struct stat st;
1304         struct pack_entry e;
1305
1306         if (find_sha1_file(sha1, &st))
1307                 return 1;
1308         return find_pack_entry(sha1, &e);
1309 }
1310
1311 int index_fd(unsigned char *sha1, int fd, struct stat *st)
1312 {
1313         unsigned long size = st->st_size;
1314         void *buf;
1315         int ret;
1316
1317         buf = "";
1318         if (size)
1319                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1320         close(fd);
1321         if ((int)(long)buf == -1)
1322                 return -1;
1323
1324         ret = write_sha1_file(buf, size, "blob", sha1);
1325         if (size)
1326                 munmap(buf, size);
1327         return ret;
1328 }