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 int get_sha1(const char *str, unsigned char *sha1)
 106 {
 107         static char pathname[PATH_MAX];
 108         static const char *prefix[] = {
 109                 "",
 110                 "refs",
 111                 "refs/tags",
 112                 "refs/heads",
 113                 "refs/snap",
 114                 NULL
 115         };
 116         const char **p;
 117
 118         if (!get_sha1_hex(str, sha1))
 119                 return 0;
 120
 121         if (!git_dir)
 122                 setup_git_env();
 123         for (p = prefix; *p; p++) {
 124                 snprintf(pathname, sizeof(pathname), "%s/%s/%s",
 125                          git_dir, *p, str);
 126                 if (!get_sha1_file(pathname, sha1))
 127                         return 0;
 128         }
 129
 130         return -1;
 131 }
 132
 133 char * sha1_to_hex(const unsigned char *sha1)
 134 {
 135         static char buffer[50];
 136         static const char hex[] = "0123456789abcdef";
 137         char *buf = buffer;
 138         int i;
 139
 140         for (i = 0; i < 20; i++) {
 141                 unsigned int val = *sha1++;
 142                 *buf++ = hex[val >> 4];
 143                 *buf++ = hex[val & 0xf];
 144         }
 145         return buffer;
 146 }
 147
 148 static void fill_sha1_path(char *pathbuf, const unsigned char *sha1)
 149 {
 150         int i;
 151         for (i = 0; i < 20; i++) {
 152                 static char hex[] = "0123456789abcdef";
 153                 unsigned int val = sha1[i];
 154                 char *pos = pathbuf + i*2 + (i > 0);
 155                 *pos++ = hex[val >> 4];
 156                 *pos = hex[val & 0xf];
 157         }
 158 }
 159
 160 /*
 161  * NOTE! This returns a statically allocated buffer, so you have to be
 162  * careful about using it. Do a "strdup()" if you need to save the
 163  * filename.
 164  *
 165  * Also note that this returns the location for creating.  Reading
 166  * SHA1 file can happen from any alternate directory listed in the
 167  * DB_ENVIRONMENT environment variable if it is not found in
 168  * the primary object database.
 169  */
 170 char *sha1_file_name(const unsigned char *sha1)
 171 {
 172         static char *name, *base;
 173
 174         if (!base) {
 175                 const char *sha1_file_directory = get_object_directory();
 176                 int len = strlen(sha1_file_directory);
 177                 base = xmalloc(len + 60);
 178                 memcpy(base, sha1_file_directory, len);
 179                 memset(base+len, 0, 60);
 180                 base[len] = '/';
 181                 base[len+3] = '/';
 182                 name = base + len + 1;
 183         }
 184         fill_sha1_path(name, sha1);
 185         return base;
 186 }
 187
 188 struct alternate_object_database *alt_odb;
 189
 190 /*
 191  * Prepare alternate object database registry.
 192  * alt_odb points at an array of struct alternate_object_database.
 193  * This array is terminated with an element that has both its base
 194  * and name set to NULL.  alt_odb[n] comes from n'th non-empty
 195  * element from colon separated ALTERNATE_DB_ENVIRONMENT environment
 196  * variable, and its base points at a statically allocated buffer
 197  * that contains "/the/directory/corresponding/to/.git/objects/...",
 198  * while its name points just after the slash at the end of
 199  * ".git/objects/" in the example above, and has enough space to hold
 200  * 40-byte hex SHA1, an extra slash for the first level indirection,
 201  * and the terminating NUL.
 202  * This function allocates the alt_odb array and all the strings
 203  * pointed by base fields of the array elements with one xmalloc();
 204  * the string pool immediately follows the array.
 205  */
 206 void prepare_alt_odb(void)
 207 {
 208         int pass, totlen, i;
 209         const char *cp, *last;
 210         char *op = NULL;
 211         const char *alt = gitenv(ALTERNATE_DB_ENVIRONMENT) ? : "";
 212
 213         if (alt_odb)
 214                 return;
 215         /* The first pass counts how large an area to allocate to
 216          * hold the entire alt_odb structure, including array of
 217          * structs and path buffers for them.  The second pass fills
 218          * the structure and prepares the path buffers for use by
 219          * fill_sha1_path().
 220          */
 221         for (totlen = pass = 0; pass < 2; pass++) {
 222                 last = alt;
 223                 i = 0;
 224                 do {
 225                         cp = strchr(last, ':') ? : last + strlen(last);
 226                         if (last != cp) {
 227                                 /* 43 = 40-byte + 2 '/' + terminating NUL */
 228                                 int pfxlen = cp - last;
 229                                 int entlen = pfxlen + 43;
 230                                 if (pass == 0)
 231                                         totlen += entlen;
 232                                 else {
 233                                         alt_odb[i].base = op;
 234                                         alt_odb[i].name = op + pfxlen + 1;
 235                                         memcpy(op, last, pfxlen);
 236                                         op[pfxlen] = op[pfxlen + 3] = '/';
 237                                         op[entlen-1] = 0;
 238                                         op += entlen;
 239                                 }
 240                                 i++;
 241                         }
 242                         while (*cp && *cp == ':')
 243                                 cp++;
 244                         last = cp;
 245                 } while (*cp);
 246                 if (pass)
 247                         break;
 248                 alt_odb = xmalloc(sizeof(*alt_odb) * (i + 1) + totlen);
 249                 alt_odb[i].base = alt_odb[i].name = NULL;
 250                 op = (char*)(&alt_odb[i+1]);
 251         }
 252 }
 253
 254 static char *find_sha1_file(const unsigned char *sha1, struct stat *st)
 255 {
 256         int i;
 257         char *name = sha1_file_name(sha1);
 258
 259         if (!stat(name, st))
 260                 return name;
 261         prepare_alt_odb();
 262         for (i = 0; (name = alt_odb[i].name) != NULL; i++) {
 263                 fill_sha1_path(name, sha1);
 264                 if (!stat(alt_odb[i].base, st))
 265                         return alt_odb[i].base;
 266         }
 267         return NULL;
 268 }
 269
 270 #define PACK_MAX_SZ (1<<26)
 271 static int pack_used_ctr;
 272 static unsigned long pack_mapped;
 273 struct packed_git *packed_git;
 274
 275 struct pack_entry {
 276         unsigned int offset;
 277         unsigned char sha1[20];
 278         struct packed_git *p;
 279 };
 280
 281 static int check_packed_git_idx(const char *path, unsigned long *idx_size_,
 282                                 void **idx_map_)
 283 {
 284         void *idx_map;
 285         unsigned int *index;
 286         unsigned long idx_size;
 287         int nr, i;
 288         int fd = open(path, O_RDONLY);
 289         struct stat st;
 290         if (fd < 0)
 291                 return -1;
 292         if (fstat(fd, &st)) {
 293                 close(fd);
 294                 return -1;
 295         }
 296         idx_size = st.st_size;
 297         idx_map = mmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0);
 298         close(fd);
 299         if (idx_map == MAP_FAILED)
 300                 return -1;
 301
 302         index = idx_map;
 303
 304         /* check index map */
 305         if (idx_size < 4*256 + 20 + 20)
 306                 return error("index file too small");
 307         nr = 0;
 308         for (i = 0; i < 256; i++) {
 309                 unsigned int n = ntohl(index[i]);
 310                 if (n < nr)
 311                         return error("non-monotonic index");
 312                 nr = n;
 313         }
 314
 315         /*
 316          * Total size:
 317          *  - 256 index entries 4 bytes each
 318          *  - 24-byte entries * nr (20-byte sha1 + 4-byte offset)
 319          *  - 20-byte SHA1 of the packfile
 320          *  - 20-byte SHA1 file checksum
 321          */
 322         if (idx_size != 4*256 + nr * 24 + 20 + 20)
 323                 return error("wrong index file size");
 324
 325         *idx_map_ = idx_map;
 326         *idx_size_ = idx_size;
 327         return 0;
 328 }
 329
 330 static int unuse_one_packed_git(void)
 331 {
 332         struct packed_git *p, *lru = NULL;
 333
 334         for (p = packed_git; p; p = p->next) {
 335                 if (p->pack_use_cnt || !p->pack_base)
 336                         continue;
 337                 if (!lru || p->pack_last_used < lru->pack_last_used)
 338                         lru = p;
 339         }
 340         if (!lru)
 341                 return 0;
 342         munmap(lru->pack_base, lru->pack_size);
 343         lru->pack_base = NULL;
 344         return 1;
 345 }
 346
 347 void unuse_packed_git(struct packed_git *p)
 348 {
 349         p->pack_use_cnt--;
 350 }
 351
 352 int use_packed_git(struct packed_git *p)
 353 {
 354         if (!p->pack_base) {
 355                 int fd;
 356                 struct stat st;
 357                 void *map;
 358
 359                 pack_mapped += p->pack_size;
 360                 while (PACK_MAX_SZ < pack_mapped && unuse_one_packed_git())
 361                         ; /* nothing */
 362                 fd = open(p->pack_name, O_RDONLY);
 363                 if (fd < 0)
 364                         die("packfile %s cannot be opened", p->pack_name);
 365                 if (fstat(fd, &st)) {
 366                         close(fd);
 367                         die("packfile %s cannot be opened", p->pack_name);
 368                 }
 369                 if (st.st_size != p->pack_size)
 370                         die("packfile %s size mismatch.", p->pack_name);
 371                 map = mmap(NULL, p->pack_size, PROT_READ, MAP_PRIVATE, fd, 0);
 372                 close(fd);
 373                 if (map == MAP_FAILED)
 374                         die("packfile %s cannot be mapped.", p->pack_name);
 375                 p->pack_base = map;
 376
 377                 /* Check if the pack file matches with the index file.
 378                  * this is cheap.
 379                  */
 380                 if (memcmp((char*)(p->index_base) + p->index_size - 40,
 381                            p->pack_base + p->pack_size - 20, 20))
 382                         die("packfile %s does not match index.", p->pack_name);
 383         }
 384         p->pack_last_used = pack_used_ctr++;
 385         p->pack_use_cnt++;
 386         return 0;
 387 }
 388
 389 struct packed_git *add_packed_git(char *path, int path_len)
 390 {
 391         struct stat st;
 392         struct packed_git *p;
 393         unsigned long idx_size;
 394         void *idx_map;
 395
 396         if (check_packed_git_idx(path, &idx_size, &idx_map))
 397                 return NULL;
 398
 399         /* do we have a corresponding .pack file? */
 400         strcpy(path + path_len - 4, ".pack");
 401         if (stat(path, &st) || !S_ISREG(st.st_mode)) {
 402                 munmap(idx_map, idx_size);
 403                 return NULL;
 404         }
 405         /* ok, it looks sane as far as we can check without
 406          * actually mapping the pack file.
 407          */
 408         p = xmalloc(sizeof(*p) + path_len + 2);
 409         strcpy(p->pack_name, path);
 410         p->index_size = idx_size;
 411         p->pack_size = st.st_size;
 412         p->index_base = idx_map;
 413         p->next = NULL;
 414         p->pack_base = NULL;
 415         p->pack_last_used = 0;
 416         p->pack_use_cnt = 0;
 417         return p;
 418 }
 419
 420 static void prepare_packed_git_one(char *objdir)
 421 {
 422         char path[PATH_MAX];
 423         int len;
 424         DIR *dir;
 425         struct dirent *de;
 426
 427         sprintf(path, "%s/pack", objdir);
 428         len = strlen(path);
 429         dir = opendir(path);
 430         if (!dir)
 431                 return;
 432         path[len++] = '/';
 433         while ((de = readdir(dir)) != NULL) {
 434                 int namelen = strlen(de->d_name);
 435                 struct packed_git *p;
 436
 437                 if (strcmp(de->d_name + namelen - 4, ".idx"))
 438                         continue;
 439
 440                 /* we have .idx.  Is it a file we can map? */
 441                 strcpy(path + len, de->d_name);
 442                 p = add_packed_git(path, len + namelen);
 443                 if (!p)
 444                         continue;
 445                 p->next = packed_git;
 446                 packed_git = p;
 447         }
 448 }
 449
 450 void prepare_packed_git(void)
 451 {
 452         int i;
 453         static int run_once = 0;
 454
 455         if (run_once++)
 456                 return;
 457
 458         prepare_packed_git_one(get_object_directory());
 459         prepare_alt_odb();
 460         for (i = 0; alt_odb[i].base != NULL; i++) {
 461                 alt_odb[i].name[0] = 0;
 462                 prepare_packed_git_one(alt_odb[i].base);
 463         }
 464 }
 465
 466 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 467 {
 468         char header[100];
 469         unsigned char real_sha1[20];
 470         SHA_CTX c;
 471
 472         SHA1_Init(&c);
 473         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 474         SHA1_Update(&c, map, size);
 475         SHA1_Final(real_sha1, &c);
 476         return memcmp(sha1, real_sha1, 20) ? -1 : 0;
 477 }
 478
 479 static void *map_sha1_file_internal(const unsigned char *sha1,
 480                                     unsigned long *size,
 481                                     int say_error)
 482 {
 483         struct stat st;
 484         void *map;
 485         int fd;
 486         char *filename = find_sha1_file(sha1, &st);
 487
 488         if (!filename) {
 489                 if (say_error)
 490                         error("cannot map sha1 file %s", sha1_to_hex(sha1));
 491                 return NULL;
 492         }
 493
 494         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 495         if (fd < 0) {
 496                 /* See if it works without O_NOATIME */
 497                 switch (sha1_file_open_flag) {
 498                 default:
 499                         fd = open(filename, O_RDONLY);
 500                         if (fd >= 0)
 501                                 break;
 502                 /* Fallthrough */
 503                 case 0:
 504                         if (say_error)
 505                                 perror(filename);
 506                         return NULL;
 507                 }
 508
 509                 /* If it failed once, it will probably fail again.
 510                  * Stop using O_NOATIME
 511                  */
 512                 sha1_file_open_flag = 0;
 513         }
 514         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 515         close(fd);
 516         if (-1 == (int)(long)map)
 517                 return NULL;
 518         *size = st.st_size;
 519         return map;
 520 }
 521
 522 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 523 {
 524         return map_sha1_file_internal(sha1, size, 1);
 525 }
 526
 527 int unpack_sha1_header(z_stream *stream, void *map, unsigned long mapsize, void *buffer, unsigned long size)
 528 {
 529         /* Get the data stream */
 530         memset(stream, 0, sizeof(*stream));
 531         stream->next_in = map;
 532         stream->avail_in = mapsize;
 533         stream->next_out = buffer;
 534         stream->avail_out = size;
 535
 536         inflateInit(stream);
 537         return inflate(stream, 0);
 538 }
 539
 540 void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 541 {
 542         int bytes = strlen(buffer) + 1;
 543         unsigned char *buf = xmalloc(1+size);
 544
 545         memcpy(buf, buffer + bytes, stream->total_out - bytes);
 546         bytes = stream->total_out - bytes;
 547         if (bytes < size) {
 548                 stream->next_out = buf + bytes;
 549                 stream->avail_out = size - bytes;
 550                 while (inflate(stream, Z_FINISH) == Z_OK)
 551                         /* nothing */;
 552         }
 553         buf[size] = 0;
 554         inflateEnd(stream);
 555         return buf;
 556 }
 557
 558 /*
 559  * We used to just use "sscanf()", but that's actually way
 560  * too permissive for what we want to check. So do an anal
 561  * object header parse by hand.
 562  */
 563 int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 564 {
 565         int i;
 566         unsigned long size;
 567
 568         /*
 569          * The type can be at most ten bytes (including the
 570          * terminating '\0' that we add), and is followed by
 571          * a space.
 572          */
 573         i = 10;
 574         for (;;) {
 575                 char c = *hdr++;
 576                 if (c == ' ')
 577                         break;
 578                 if (!--i)
 579                         return -1;
 580                 *type++ = c;
 581         }
 582         *type = 0;
 583
 584         /*
 585          * The length must follow immediately, and be in canonical
 586          * decimal format (ie "010" is not valid).
 587          */
 588         size = *hdr++ - '0';
 589         if (size > 9)
 590                 return -1;
 591         if (size) {
 592                 for (;;) {
 593                         unsigned long c = *hdr - '0';
 594                         if (c > 9)
 595                                 break;
 596                         hdr++;
 597                         size = size * 10 + c;
 598                 }
 599         }
 600         *sizep = size;
 601
 602         /*
 603          * The length must be followed by a zero byte
 604          */
 605         return *hdr ? -1 : 0;
 606 }
 607
 608 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 609 {
 610         int ret;
 611         z_stream stream;
 612         char hdr[8192];
 613
 614         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
 615         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
 616                 return NULL;
 617
 618         return unpack_sha1_rest(&stream, hdr, *size);
 619 }
 620
 621 static int packed_delta_info(unsigned char *base_sha1,
 622                              unsigned long delta_size,
 623                              unsigned long left,
 624                              char *type,
 625                              unsigned long *sizep)
 626 {
 627         if (left < 20)
 628                 die("truncated pack file");
 629
 630         /* We choose to only get the type of the base object and
 631          * ignore potentially corrupt pack file that expects the delta
 632          * based on a base with a wrong size.  This saves tons of
 633          * inflate() calls.
 634          */
 635
 636         if (sha1_object_info(base_sha1, type, NULL))
 637                 die("cannot get info for delta-pack base");
 638
 639         if (sizep) {
 640                 const unsigned char *data;
 641                 unsigned char delta_head[64];
 642                 unsigned long result_size;
 643                 z_stream stream;
 644                 int st;
 645
 646                 memset(&stream, 0, sizeof(stream));
 647
 648                 data = stream.next_in = base_sha1 + 20;
 649                 stream.avail_in = left - 20;
 650                 stream.next_out = delta_head;
 651                 stream.avail_out = sizeof(delta_head);
 652
 653                 inflateInit(&stream);
 654                 st = inflate(&stream, Z_FINISH);
 655                 inflateEnd(&stream);
 656                 if ((st != Z_STREAM_END) &&
 657                     stream.total_out != sizeof(delta_head))
 658                         die("delta data unpack-initial failed");
 659
 660                 /* Examine the initial part of the delta to figure out
 661                  * the result size.
 662                  */
 663                 data = delta_head;
 664                 get_delta_hdr_size(&data); /* ignore base size */
 665
 666                 /* Read the result size */
 667                 result_size = get_delta_hdr_size(&data);
 668                 *sizep = result_size;
 669         }
 670         return 0;
 671 }
 672
 673 static unsigned long unpack_object_header(struct packed_git *p, unsigned long offset,
 674         enum object_type *type, unsigned long *sizep)
 675 {
 676         unsigned shift;
 677         unsigned char *pack, c;
 678         unsigned long size;
 679
 680         if (offset >= p->pack_size)
 681                 die("object offset outside of pack file");
 682
 683         pack =  p->pack_base + offset;
 684         c = *pack++;
 685         offset++;
 686         *type = (c >> 4) & 7;
 687         size = c & 15;
 688         shift = 4;
 689         while (c & 0x80) {
 690                 if (offset >= p->pack_size)
 691                         die("object offset outside of pack file");
 692                 c = *pack++;
 693                 offset++;
 694                 size += (c & 0x7f) << shift;
 695                 shift += 7;
 696         }
 697         *sizep = size;
 698         return offset;
 699 }
 700
 701 static int packed_object_info(struct pack_entry *entry,
 702                               char *type, unsigned long *sizep)
 703 {
 704         struct packed_git *p = entry->p;
 705         unsigned long offset, size, left;
 706         unsigned char *pack;
 707         enum object_type kind;
 708         int retval;
 709
 710         if (use_packed_git(p))
 711                 die("cannot map packed file");
 712
 713         offset = unpack_object_header(p, entry->offset, &kind, &size);
 714         pack = p->pack_base + offset;
 715         left = p->pack_size - offset;
 716
 717         switch (kind) {
 718         case OBJ_DELTA:
 719                 retval = packed_delta_info(pack, size, left, type, sizep);
 720                 unuse_packed_git(p);
 721                 return retval;
 722         case OBJ_COMMIT:
 723                 strcpy(type, "commit");
 724                 break;
 725         case OBJ_TREE:
 726                 strcpy(type, "tree");
 727                 break;
 728         case OBJ_BLOB:
 729                 strcpy(type, "blob");
 730                 break;
 731         case OBJ_TAG:
 732                 strcpy(type, "tag");
 733                 break;
 734         default:
 735                 die("corrupted pack file");
 736         }
 737         if (sizep)
 738                 *sizep = size;
 739         unuse_packed_git(p);
 740         return 0;
 741 }
 742
 743 /* forward declaration for a mutually recursive function */
 744 static void *unpack_entry(struct pack_entry *, char *, unsigned long *);
 745
 746 static void *unpack_delta_entry(unsigned char *base_sha1,
 747                                 unsigned long delta_size,
 748                                 unsigned long left,
 749                                 char *type,
 750                                 unsigned long *sizep)
 751 {
 752         void *data, *delta_data, *result, *base;
 753         unsigned long data_size, result_size, base_size;
 754         z_stream stream;
 755         int st;
 756
 757         if (left < 20)
 758                 die("truncated pack file");
 759         data = base_sha1 + 20;
 760         data_size = left - 20;
 761         delta_data = xmalloc(delta_size);
 762
 763         memset(&stream, 0, sizeof(stream));
 764
 765         stream.next_in = data;
 766         stream.avail_in = data_size;
 767         stream.next_out = delta_data;
 768         stream.avail_out = delta_size;
 769
 770         inflateInit(&stream);
 771         st = inflate(&stream, Z_FINISH);
 772         inflateEnd(&stream);
 773         if ((st != Z_STREAM_END) || stream.total_out != delta_size)
 774                 die("delta data unpack failed");
 775
 776         /* This may recursively unpack the base, which is what we want */
 777         base = read_sha1_file(base_sha1, type, &base_size);
 778         if (!base)
 779                 die("failed to read delta-pack base object %s",
 780                     sha1_to_hex(base_sha1));
 781         result = patch_delta(base, base_size,
 782                              delta_data, delta_size,
 783                              &result_size);
 784         if (!result)
 785                 die("failed to apply delta");
 786         free(delta_data);
 787         free(base);
 788         *sizep = result_size;
 789         return result;
 790 }
 791
 792 static void *unpack_non_delta_entry(unsigned char *data,
 793                                     unsigned long size,
 794                                     unsigned long left)
 795 {
 796         int st;
 797         z_stream stream;
 798         char *buffer;
 799
 800         buffer = xmalloc(size + 1);
 801         buffer[size] = 0;
 802         memset(&stream, 0, sizeof(stream));
 803         stream.next_in = data;
 804         stream.avail_in = left;
 805         stream.next_out = buffer;
 806         stream.avail_out = size;
 807
 808         inflateInit(&stream);
 809         st = inflate(&stream, Z_FINISH);
 810         inflateEnd(&stream);
 811         if ((st != Z_STREAM_END) || stream.total_out != size) {
 812                 free(buffer);
 813                 return NULL;
 814         }
 815
 816         return buffer;
 817 }
 818
 819 static void *unpack_entry(struct pack_entry *entry,
 820                           char *type, unsigned long *sizep)
 821 {
 822         struct packed_git *p = entry->p;
 823         unsigned long offset, size, left;
 824         unsigned char *pack;
 825         enum object_type kind;
 826         void *retval;
 827
 828         if (use_packed_git(p))
 829                 die("cannot map packed file");
 830
 831         offset = unpack_object_header(p, entry->offset, &kind, &size);
 832         pack = p->pack_base + offset;
 833         left = p->pack_size - offset;
 834         switch (kind) {
 835         case OBJ_DELTA:
 836                 retval = unpack_delta_entry(pack, size, left, type, sizep);
 837                 unuse_packed_git(p);
 838                 return retval;
 839         case OBJ_COMMIT:
 840                 strcpy(type, "commit");
 841                 break;
 842         case OBJ_TREE:
 843                 strcpy(type, "tree");
 844                 break;
 845         case OBJ_BLOB:
 846                 strcpy(type, "blob");
 847                 break;
 848         case OBJ_TAG:
 849                 strcpy(type, "tag");
 850                 break;
 851         default:
 852                 die("corrupted pack file");
 853         }
 854         *sizep = size;
 855         retval = unpack_non_delta_entry(pack, size, left);
 856         unuse_packed_git(p);
 857         return retval;
 858 }
 859
 860 int num_packed_objects(const struct packed_git *p)
 861 {
 862         /* See check_packed_git_idx() */
 863         return (p->index_size - 20 - 20 - 4*256) / 24;
 864 }
 865
 866 int nth_packed_object_sha1(const struct packed_git *p, int n,
 867                            unsigned char* sha1)
 868 {
 869         void *index = p->index_base + 256;
 870         if (n < 0 || num_packed_objects(p) <= n)
 871                 return -1;
 872         memcpy(sha1, (index + 24 * n + 4), 20);
 873         return 0;
 874 }
 875
 876 static int find_pack_entry_1(const unsigned char *sha1,
 877                              struct pack_entry *e, struct packed_git *p)
 878 {
 879         int *level1_ofs = p->index_base;
 880         int hi = ntohl(level1_ofs[*sha1]);
 881         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
 882         void *index = p->index_base + 256;
 883
 884         do {
 885                 int mi = (lo + hi) / 2;
 886                 int cmp = memcmp(index + 24 * mi + 4, sha1, 20);
 887                 if (!cmp) {
 888                         e->offset = ntohl(*((int*)(index + 24 * mi)));
 889                         memcpy(e->sha1, sha1, 20);
 890                         e->p = p;
 891                         return 1;
 892                 }
 893                 if (cmp > 0)
 894                         hi = mi;
 895                 else
 896                         lo = mi+1;
 897         } while (lo < hi);
 898         return 0;
 899 }
 900
 901 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e)
 902 {
 903         struct packed_git *p;
 904         prepare_packed_git();
 905
 906         for (p = packed_git; p; p = p->next) {
 907                 if (find_pack_entry_1(sha1, e, p))
 908                         return 1;
 909         }
 910         return 0;
 911 }
 912
 913 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
 914 {
 915         int status;
 916         unsigned long mapsize, size;
 917         void *map;
 918         z_stream stream;
 919         char hdr[128];
 920
 921         map = map_sha1_file_internal(sha1, &mapsize, 0);
 922         if (!map) {
 923                 struct pack_entry e;
 924
 925                 if (!find_pack_entry(sha1, &e))
 926                         return error("unable to find %s", sha1_to_hex(sha1));
 927                 return packed_object_info(&e, type, sizep);
 928         }
 929         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
 930                 status = error("unable to unpack %s header",
 931                                sha1_to_hex(sha1));
 932         if (parse_sha1_header(hdr, type, &size) < 0)
 933                 status = error("unable to parse %s header", sha1_to_hex(sha1));
 934         else {
 935                 status = 0;
 936                 if (sizep)
 937                         *sizep = size;
 938         }
 939         inflateEnd(&stream);
 940         munmap(map, mapsize);
 941         return status;
 942 }
 943
 944 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
 945 {
 946         struct pack_entry e;
 947
 948         if (!find_pack_entry(sha1, &e)) {
 949                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
 950                 return NULL;
 951         }
 952         return unpack_entry(&e, type, size);
 953 }
 954
 955 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
 956 {
 957         unsigned long mapsize;
 958         void *map, *buf;
 959
 960         map = map_sha1_file_internal(sha1, &mapsize, 0);
 961         if (map) {
 962                 buf = unpack_sha1_file(map, mapsize, type, size);
 963                 munmap(map, mapsize);
 964                 return buf;
 965         }
 966         return read_packed_sha1(sha1, type, size);
 967 }
 968
 969 void *read_object_with_reference(const unsigned char *sha1,
 970                                  const char *required_type,
 971                                  unsigned long *size,
 972                                  unsigned char *actual_sha1_return)
 973 {
 974         char type[20];
 975         void *buffer;
 976         unsigned long isize;
 977         unsigned char actual_sha1[20];
 978
 979         memcpy(actual_sha1, sha1, 20);
 980         while (1) {
 981                 int ref_length = -1;
 982                 const char *ref_type = NULL;
 983
 984                 buffer = read_sha1_file(actual_sha1, type, &isize);
 985                 if (!buffer)
 986                         return NULL;
 987                 if (!strcmp(type, required_type)) {
 988                         *size = isize;
 989                         if (actual_sha1_return)
 990                                 memcpy(actual_sha1_return, actual_sha1, 20);
 991                         return buffer;
 992                 }
 993                 /* Handle references */
 994                 else if (!strcmp(type, "commit"))
 995                         ref_type = "tree ";
 996                 else if (!strcmp(type, "tag"))
 997                         ref_type = "object ";
 998                 else {
 999                         free(buffer);
1000                         return NULL;
1001                 }
1002                 ref_length = strlen(ref_type);
1003
1004                 if (memcmp(buffer, ref_type, ref_length) ||
1005                     get_sha1_hex(buffer + ref_length, actual_sha1)) {
1006                         free(buffer);
1007                         return NULL;
1008                 }
1009                 /* Now we have the ID of the referred-to object in
1010                  * actual_sha1.  Check again. */
1011         }
1012 }
1013
1014 static char *write_sha1_file_prepare(void *buf,
1015                                      unsigned long len,
1016                                      const char *type,
1017                                      unsigned char *sha1,
1018                                      unsigned char *hdr,
1019                                      int *hdrlen)
1020 {
1021         SHA_CTX c;
1022
1023         /* Generate the header */
1024         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
1025
1026         /* Sha1.. */
1027         SHA1_Init(&c);
1028         SHA1_Update(&c, hdr, *hdrlen);
1029         SHA1_Update(&c, buf, len);
1030         SHA1_Final(sha1, &c);
1031
1032         return sha1_file_name(sha1);
1033 }
1034
1035 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1036 {
1037         int size;
1038         unsigned char *compressed;
1039         z_stream stream;
1040         unsigned char sha1[20];
1041         char *filename;
1042         static char tmpfile[PATH_MAX];
1043         unsigned char hdr[50];
1044         int fd, hdrlen, ret;
1045
1046         /* Normally if we have it in the pack then we do not bother writing
1047          * it out into .git/objects/??/?{38} file.
1048          */
1049         filename = write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1050         if (returnsha1)
1051                 memcpy(returnsha1, sha1, 20);
1052         if (has_sha1_file(sha1))
1053                 return 0;
1054         fd = open(filename, O_RDONLY);
1055         if (fd >= 0) {
1056                 /*
1057                  * FIXME!!! We might do collision checking here, but we'd
1058                  * need to uncompress the old file and check it. Later.
1059                  */
1060                 close(fd);
1061                 return 0;
1062         }
1063
1064         if (errno != ENOENT) {
1065                 fprintf(stderr, "sha1 file %s: %s", filename, strerror(errno));
1066                 return -1;
1067         }
1068
1069         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1070
1071         fd = mkstemp(tmpfile);
1072         if (fd < 0) {
1073                 fprintf(stderr, "unable to create temporary sha1 filename %s: %s", tmpfile, strerror(errno));
1074                 return -1;
1075         }
1076
1077         /* Set it up */
1078         memset(&stream, 0, sizeof(stream));
1079         deflateInit(&stream, Z_BEST_COMPRESSION);
1080         size = deflateBound(&stream, len+hdrlen);
1081         compressed = xmalloc(size);
1082
1083         /* Compress it */
1084         stream.next_out = compressed;
1085         stream.avail_out = size;
1086
1087         /* First header.. */
1088         stream.next_in = hdr;
1089         stream.avail_in = hdrlen;
1090         while (deflate(&stream, 0) == Z_OK)
1091                 /* nothing */;
1092
1093         /* Then the data itself.. */
1094         stream.next_in = buf;
1095         stream.avail_in = len;
1096         while (deflate(&stream, Z_FINISH) == Z_OK)
1097                 /* nothing */;
1098         deflateEnd(&stream);
1099         size = stream.total_out;
1100
1101         if (write(fd, compressed, size) != size)
1102                 die("unable to write file");
1103         fchmod(fd, 0444);
1104         close(fd);
1105         free(compressed);
1106
1107         ret = link(tmpfile, filename);
1108         if (ret < 0) {
1109                 ret = errno;
1110
1111                 /*
1112                  * Coda hack - coda doesn't like cross-directory links,
1113                  * so we fall back to a rename, which will mean that it
1114                  * won't be able to check collisions, but that's not a
1115                  * big deal.
1116                  *
1117                  * When this succeeds, we just return 0. We have nothing
1118                  * left to unlink.
1119                  */
1120                 if (ret == EXDEV && !rename(tmpfile, filename))
1121                         return 0;
1122         }
1123         unlink(tmpfile);
1124         if (ret) {
1125                 if (ret != EEXIST) {
1126                         fprintf(stderr, "unable to write sha1 filename %s: %s", filename, strerror(ret));
1127                         return -1;
1128                 }
1129                 /* FIXME!!! Collision check here ? */
1130         }
1131
1132         return 0;
1133 }
1134
1135 int write_sha1_from_fd(const unsigned char *sha1, int fd)
1136 {
1137         char *filename = sha1_file_name(sha1);
1138
1139         int local;
1140         z_stream stream;
1141         unsigned char real_sha1[20];
1142         unsigned char buf[4096];
1143         unsigned char discard[4096];
1144         int ret;
1145         SHA_CTX c;
1146
1147         local = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
1148
1149         if (local < 0)
1150                 return error("Couldn't open %s\n", filename);
1151
1152         memset(&stream, 0, sizeof(stream));
1153
1154         inflateInit(&stream);
1155
1156         SHA1_Init(&c);
1157
1158         do {
1159                 ssize_t size;
1160                 size = read(fd, buf, 4096);
1161                 if (size <= 0) {
1162                         close(local);
1163                         unlink(filename);
1164                         if (!size)
1165                                 return error("Connection closed?");
1166                         perror("Reading from connection");
1167                         return -1;
1168                 }
1169                 write(local, buf, size);
1170                 stream.avail_in = size;
1171                 stream.next_in = buf;
1172                 do {
1173                         stream.next_out = discard;
1174                         stream.avail_out = sizeof(discard);
1175                         ret = inflate(&stream, Z_SYNC_FLUSH);
1176                         SHA1_Update(&c, discard, sizeof(discard) -
1177                                     stream.avail_out);
1178                 } while (stream.avail_in && ret == Z_OK);
1179
1180         } while (ret == Z_OK);
1181         inflateEnd(&stream);
1182
1183         close(local);
1184         SHA1_Final(real_sha1, &c);
1185         if (ret != Z_STREAM_END) {
1186                 unlink(filename);
1187                 return error("File %s corrupted", sha1_to_hex(sha1));
1188         }
1189         if (memcmp(sha1, real_sha1, 20)) {
1190                 unlink(filename);
1191                 return error("File %s has bad hash\n", sha1_to_hex(sha1));
1192         }
1193
1194         return 0;
1195 }
1196
1197 int has_sha1_file(const unsigned char *sha1)
1198 {
1199         struct stat st;
1200         struct pack_entry e;
1201
1202         if (find_sha1_file(sha1, &st))
1203                 return 1;
1204         return find_pack_entry(sha1, &e);
1205 }
1206
1207 int index_fd(unsigned char *sha1, int fd, struct stat *st)
1208 {
1209         unsigned long size = st->st_size;
1210         void *buf;
1211         int ret;
1212
1213         buf = "";
1214         if (size)
1215                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1216         close(fd);
1217         if ((int)(long)buf == -1)
1218                 return -1;
1219
1220         ret = write_sha1_file(buf, size, "blob", sha1);
1221         if (size)
1222                 munmap(buf, size);
1223         return ret;
1224 }