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 static struct alternate_object_database {
 189         char *base;
 190         char *name;
 191 } *alt_odb;
 192
 193 /*
 194  * Prepare alternate object database registry.
 195  * alt_odb points at an array of struct alternate_object_database.
 196  * This array is terminated with an element that has both its base
 197  * and name set to NULL.  alt_odb[n] comes from n'th non-empty
 198  * element from colon separated ALTERNATE_DB_ENVIRONMENT environment
 199  * variable, and its base points at a statically allocated buffer
 200  * that contains "/the/directory/corresponding/to/.git/objects/...",
 201  * while its name points just after the slash at the end of
 202  * ".git/objects/" in the example above, and has enough space to hold
 203  * 40-byte hex SHA1, an extra slash for the first level indirection,
 204  * and the terminating NUL.
 205  * This function allocates the alt_odb array and all the strings
 206  * pointed by base fields of the array elements with one xmalloc();
 207  * the string pool immediately follows the array.
 208  */
 209 static void prepare_alt_odb(void)
 210 {
 211         int pass, totlen, i;
 212         const char *cp, *last;
 213         char *op = NULL;
 214         const char *alt = gitenv(ALTERNATE_DB_ENVIRONMENT) ? : "";
 215
 216         /* The first pass counts how large an area to allocate to
 217          * hold the entire alt_odb structure, including array of
 218          * structs and path buffers for them.  The second pass fills
 219          * the structure and prepares the path buffers for use by
 220          * fill_sha1_path().
 221          */
 222         for (totlen = pass = 0; pass < 2; pass++) {
 223                 last = alt;
 224                 i = 0;
 225                 do {
 226                         cp = strchr(last, ':') ? : last + strlen(last);
 227                         if (last != cp) {
 228                                 /* 43 = 40-byte + 2 '/' + terminating NUL */
 229                                 int pfxlen = cp - last;
 230                                 int entlen = pfxlen + 43;
 231                                 if (pass == 0)
 232                                         totlen += entlen;
 233                                 else {
 234                                         alt_odb[i].base = op;
 235                                         alt_odb[i].name = op + pfxlen + 1;
 236                                         memcpy(op, last, pfxlen);
 237                                         op[pfxlen] = op[pfxlen + 3] = '/';
 238                                         op[entlen-1] = 0;
 239                                         op += entlen;
 240                                 }
 241                                 i++;
 242                         }
 243                         while (*cp && *cp == ':')
 244                                 cp++;
 245                         last = cp;
 246                 } while (*cp);
 247                 if (pass)
 248                         break;
 249                 alt_odb = xmalloc(sizeof(*alt_odb) * (i + 1) + totlen);
 250                 alt_odb[i].base = alt_odb[i].name = NULL;
 251                 op = (char*)(&alt_odb[i+1]);
 252         }
 253 }
 254
 255 static char *find_sha1_file(const unsigned char *sha1, struct stat *st)
 256 {
 257         int i;
 258         char *name = sha1_file_name(sha1);
 259
 260         if (!stat(name, st))
 261                 return name;
 262         if (!alt_odb)
 263                 prepare_alt_odb();
 264         for (i = 0; (name = alt_odb[i].name) != NULL; i++) {
 265                 fill_sha1_path(name, sha1);
 266                 if (!stat(alt_odb[i].base, st))
 267                         return alt_odb[i].base;
 268         }
 269         return NULL;
 270 }
 271
 272 #define PACK_MAX_SZ (1<<26)
 273 static int pack_used_ctr;
 274 static unsigned long pack_mapped;
 275 static struct packed_git {
 276         struct packed_git *next;
 277         unsigned long index_size;
 278         unsigned long pack_size;
 279         unsigned int *index_base;
 280         void *pack_base;
 281         unsigned int pack_last_used;
 282         char pack_name[0]; /* something like ".git/objects/pack/xxxxx.pack" */
 283 } *packed_git;
 284
 285 struct pack_entry {
 286         unsigned int offset;
 287         unsigned char sha1[20];
 288         struct packed_git *p;
 289 };
 290
 291 static int check_packed_git_idx(const char *path, unsigned long *idx_size_,
 292                                 void **idx_map_)
 293 {
 294         void *idx_map;
 295         unsigned int *index;
 296         unsigned long idx_size;
 297         int nr, i;
 298         int fd = open(path, O_RDONLY);
 299         struct stat st;
 300         if (fd < 0)
 301                 return -1;
 302         if (fstat(fd, &st)) {
 303                 close(fd);
 304                 return -1;
 305         }
 306         idx_size = st.st_size;
 307         idx_map = mmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0);
 308         close(fd);
 309         if (idx_map == MAP_FAILED)
 310                 return -1;
 311
 312         index = idx_map;
 313
 314         /* check index map */
 315         if (idx_size < 4*256 + 20)
 316                 return error("index file too small");
 317         nr = 0;
 318         for (i = 0; i < 256; i++) {
 319                 unsigned int n = ntohl(index[i]);
 320                 if (n < nr)
 321                         return error("non-monotonic index");
 322                 nr = n;
 323         }
 324
 325         /*
 326          * Total size:
 327          *  - 256 index entries 4 bytes each
 328          *  - 24-byte entries * nr (20-byte sha1 + 4-byte offset)
 329          *  - 20-byte SHA1 of the packfile
 330          *  - 20-byte SHA1 file checksum
 331          */
 332         if (idx_size != 4*256 + nr * 24 + 20 + 20)
 333                 return error("wrong index file size");
 334
 335         *idx_map_ = idx_map;
 336         *idx_size_ = idx_size;
 337         return 0;
 338 }
 339
 340 static void unuse_one_packed_git(void)
 341 {
 342         /* NOTYET */
 343 }
 344
 345 static int use_packed_git(struct packed_git *p)
 346 {
 347         if (!p->pack_base) {
 348                 int fd;
 349                 struct stat st;
 350                 void *map;
 351
 352                 pack_mapped += p->pack_size;
 353                 while (PACK_MAX_SZ < pack_mapped)
 354                         unuse_one_packed_git();
 355                 fd = open(p->pack_name, O_RDONLY);
 356                 if (fd < 0)
 357                         return -1;
 358                 if (fstat(fd, &st)) {
 359                         close(fd);
 360                         return -1;
 361                 }
 362                 if (st.st_size != p->pack_size)
 363                         return -1;
 364                 map = mmap(NULL, p->pack_size, PROT_READ, MAP_PRIVATE, fd, 0);
 365                 close(fd);
 366                 if (map == MAP_FAILED)
 367                         return -1;
 368                 p->pack_base = map;
 369         }
 370         p->pack_last_used = pack_used_ctr++;
 371         return 0;
 372 }
 373
 374 static struct packed_git *add_packed_git(char *path, int path_len)
 375 {
 376         struct stat st;
 377         struct packed_git *p;
 378         unsigned long idx_size;
 379         void *idx_map;
 380
 381         if (check_packed_git_idx(path, &idx_size, &idx_map))
 382                 return NULL;
 383
 384         /* do we have a corresponding .pack file? */
 385         strcpy(path + path_len - 4, ".pack");
 386         if (stat(path, &st) || !S_ISREG(st.st_mode)) {
 387                 munmap(idx_map, idx_size);
 388                 return NULL;
 389         }
 390         /* ok, it looks sane as far as we can check without
 391          * actually mapping the pack file.
 392          */
 393         p = xmalloc(sizeof(*p) + path_len + 2);
 394         strcpy(p->pack_name, path);
 395         p->index_size = idx_size;
 396         p->pack_size = st.st_size;
 397         p->index_base = idx_map;
 398         p->next = NULL;
 399         p->pack_base = NULL;
 400         p->pack_last_used = 0;
 401         return p;
 402 }
 403
 404 static void prepare_packed_git_one(char *objdir)
 405 {
 406         char path[PATH_MAX];
 407         int len;
 408         DIR *dir;
 409         struct dirent *de;
 410
 411         sprintf(path, "%s/pack", objdir);
 412         len = strlen(path);
 413         dir = opendir(path);
 414         if (!dir)
 415                 return;
 416         path[len++] = '/';
 417         while ((de = readdir(dir)) != NULL) {
 418                 int namelen = strlen(de->d_name);
 419                 struct packed_git *p;
 420
 421                 if (strcmp(de->d_name + namelen - 4, ".idx"))
 422                         continue;
 423
 424                 /* we have .idx.  Is it a file we can map? */
 425                 strcpy(path + len, de->d_name);
 426                 p = add_packed_git(path, len + namelen);
 427                 if (!p)
 428                         continue;
 429                 p->next = packed_git;
 430                 packed_git = p;
 431         }
 432 }
 433
 434 static void prepare_packed_git(void)
 435 {
 436         int i;
 437         static int run_once = 0;
 438
 439         if (run_once++)
 440                 return;
 441
 442         prepare_packed_git_one(get_object_directory());
 443         if (!alt_odb)
 444                 prepare_alt_odb();
 445         for (i = 0; alt_odb[i].base != NULL; i++) {
 446                 alt_odb[i].name[0] = 0;
 447                 prepare_packed_git_one(alt_odb[i].base);
 448         }
 449 }
 450
 451 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 452 {
 453         char header[100];
 454         unsigned char real_sha1[20];
 455         SHA_CTX c;
 456
 457         SHA1_Init(&c);
 458         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 459         SHA1_Update(&c, map, size);
 460         SHA1_Final(real_sha1, &c);
 461         return memcmp(sha1, real_sha1, 20) ? -1 : 0;
 462 }
 463
 464 static void *map_sha1_file_internal(const unsigned char *sha1,
 465                                     unsigned long *size,
 466                                     int say_error)
 467 {
 468         struct stat st;
 469         void *map;
 470         int fd;
 471         char *filename = find_sha1_file(sha1, &st);
 472
 473         if (!filename) {
 474                 if (say_error)
 475                         error("cannot map sha1 file %s", sha1_to_hex(sha1));
 476                 return NULL;
 477         }
 478
 479         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 480         if (fd < 0) {
 481                 /* See if it works without O_NOATIME */
 482                 switch (sha1_file_open_flag) {
 483                 default:
 484                         fd = open(filename, O_RDONLY);
 485                         if (fd >= 0)
 486                                 break;
 487                 /* Fallthrough */
 488                 case 0:
 489                         if (say_error)
 490                                 perror(filename);
 491                         return NULL;
 492                 }
 493
 494                 /* If it failed once, it will probably fail again.
 495                  * Stop using O_NOATIME
 496                  */
 497                 sha1_file_open_flag = 0;
 498         }
 499         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 500         close(fd);
 501         if (-1 == (int)(long)map)
 502                 return NULL;
 503         *size = st.st_size;
 504         return map;
 505 }
 506
 507 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 508 {
 509         return map_sha1_file_internal(sha1, size, 1);
 510 }
 511
 512 int unpack_sha1_header(z_stream *stream, void *map, unsigned long mapsize, void *buffer, unsigned long size)
 513 {
 514         /* Get the data stream */
 515         memset(stream, 0, sizeof(*stream));
 516         stream->next_in = map;
 517         stream->avail_in = mapsize;
 518         stream->next_out = buffer;
 519         stream->avail_out = size;
 520
 521         inflateInit(stream);
 522         return inflate(stream, 0);
 523 }
 524
 525 void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 526 {
 527         int bytes = strlen(buffer) + 1;
 528         unsigned char *buf = xmalloc(1+size);
 529
 530         memcpy(buf, buffer + bytes, stream->total_out - bytes);
 531         bytes = stream->total_out - bytes;
 532         if (bytes < size) {
 533                 stream->next_out = buf + bytes;
 534                 stream->avail_out = size - bytes;
 535                 while (inflate(stream, Z_FINISH) == Z_OK)
 536                         /* nothing */;
 537         }
 538         buf[size] = 0;
 539         inflateEnd(stream);
 540         return buf;
 541 }
 542
 543 /*
 544  * We used to just use "sscanf()", but that's actually way
 545  * too permissive for what we want to check. So do an anal
 546  * object header parse by hand.
 547  */
 548 int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 549 {
 550         int i;
 551         unsigned long size;
 552
 553         /*
 554          * The type can be at most ten bytes (including the
 555          * terminating '\0' that we add), and is followed by
 556          * a space.
 557          */
 558         i = 10;
 559         for (;;) {
 560                 char c = *hdr++;
 561                 if (c == ' ')
 562                         break;
 563                 if (!--i)
 564                         return -1;
 565                 *type++ = c;
 566         }
 567         *type = 0;
 568
 569         /*
 570          * The length must follow immediately, and be in canonical
 571          * decimal format (ie "010" is not valid).
 572          */
 573         size = *hdr++ - '0';
 574         if (size > 9)
 575                 return -1;
 576         if (size) {
 577                 for (;;) {
 578                         unsigned long c = *hdr - '0';
 579                         if (c > 9)
 580                                 break;
 581                         hdr++;
 582                         size = size * 10 + c;
 583                 }
 584         }
 585         *sizep = size;
 586
 587         /*
 588          * The length must be followed by a zero byte
 589          */
 590         return *hdr ? -1 : 0;
 591 }
 592
 593 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 594 {
 595         int ret;
 596         z_stream stream;
 597         char hdr[8192];
 598
 599         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
 600         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
 601                 return NULL;
 602
 603         return unpack_sha1_rest(&stream, hdr, *size);
 604 }
 605
 606 static int packed_delta_info(unsigned char *base_sha1,
 607                              unsigned long delta_size,
 608                              unsigned long left,
 609                              char *type,
 610                              unsigned long *sizep)
 611 {
 612         unsigned char *data;
 613         unsigned char delta_head[64];
 614         int i;
 615         unsigned char cmd;
 616         unsigned long data_size, result_size, base_size, verify_base_size;
 617         z_stream stream;
 618         int st;
 619
 620         if (left < 20)
 621                 die("truncated pack file");
 622         if (sha1_object_info(base_sha1, type, &base_size))
 623                 die("cannot get info for delta-pack base");
 624
 625         data = base_sha1 + 20;
 626         data_size = left - 20;
 627
 628         memset(&stream, 0, sizeof(stream));
 629
 630         stream.next_in = data;
 631         stream.avail_in = data_size;
 632         stream.next_out = delta_head;
 633         stream.avail_out = sizeof(delta_head);
 634
 635         inflateInit(&stream);
 636         st = inflate(&stream, Z_FINISH);
 637         inflateEnd(&stream);
 638         if ((st != Z_STREAM_END) && stream.total_out != sizeof(delta_head))
 639                 die("delta data unpack-initial failed");
 640
 641         /* Examine the initial part of the delta to figure out
 642          * the result size.  Verify the base size while we are at it.
 643          */
 644         data = delta_head;
 645         verify_base_size = i = 0;
 646         cmd = *data++;
 647         while (cmd) {
 648                 if (cmd & 1)
 649                         verify_base_size |= *data++ << i;
 650                 i += 8;
 651                 cmd >>= 1;
 652         }
 653
 654         /* Read the result size */
 655         result_size = i = 0;
 656         cmd = *data++;
 657         while (cmd) {
 658                 if (cmd & 1)
 659                         result_size |= *data++ << i;
 660                 i += 8;
 661                 cmd >>= 1;
 662         }
 663         if (verify_base_size != base_size)
 664                 die("delta base size mismatch");
 665
 666         *sizep = result_size;
 667         return 0;
 668 }
 669
 670 static unsigned long unpack_object_header(struct packed_git *p, unsigned long offset,
 671         enum object_type *type, unsigned long *sizep)
 672 {
 673         unsigned char *pack, c;
 674         unsigned long size;
 675
 676         if (offset >= p->pack_size)
 677                 die("object offset outside of pack file");
 678
 679         pack =  p->pack_base + offset;
 680         c = *pack++;
 681         offset++;
 682         *type = (c >> 4) & 7;
 683         size = c & 15;
 684         while (c & 0x80) {
 685                 if (offset >= p->pack_size)
 686                         die("object offset outside of pack file");
 687                 c = *pack++;
 688                 offset++;
 689                 size = (size << 7) | (c & 0x7f);
 690         }
 691         *sizep = size;
 692         return offset;
 693 }
 694
 695 static int packed_object_info(struct pack_entry *entry,
 696                               char *type, unsigned long *sizep)
 697 {
 698         struct packed_git *p = entry->p;
 699         unsigned long offset, size, left;
 700         unsigned char *pack;
 701         enum object_type kind;
 702
 703         if (use_packed_git(p))
 704                 die("cannot map packed file");
 705
 706         offset = unpack_object_header(p, entry->offset, &kind, &size);
 707         pack = p->pack_base + offset;
 708         left = p->pack_size - offset;
 709
 710         switch (kind) {
 711         case OBJ_DELTA:
 712                 return packed_delta_info(pack, size, left, type, sizep);
 713                 break;
 714         case OBJ_COMMIT:
 715                 strcpy(type, "commit");
 716                 break;
 717         case OBJ_TREE:
 718                 strcpy(type, "tree");
 719                 break;
 720         case OBJ_BLOB:
 721                 strcpy(type, "blob");
 722                 break;
 723         case OBJ_TAG:
 724                 strcpy(type, "tag");
 725                 break;
 726         default:
 727                 die("corrupted pack file");
 728         }
 729         *sizep = size;
 730         return 0;
 731 }
 732
 733 /* forward declaration for a mutually recursive function */
 734 static void *unpack_entry(struct pack_entry *, char *, unsigned long *);
 735
 736 static void *unpack_delta_entry(unsigned char *base_sha1,
 737                                 unsigned long delta_size,
 738                                 unsigned long left,
 739                                 char *type,
 740                                 unsigned long *sizep)
 741 {
 742         void *data, *delta_data, *result, *base;
 743         unsigned long data_size, result_size, base_size;
 744         z_stream stream;
 745         int st;
 746
 747         if (left < 20)
 748                 die("truncated pack file");
 749         data = base_sha1 + 20;
 750         data_size = left - 20;
 751         delta_data = xmalloc(delta_size);
 752
 753         memset(&stream, 0, sizeof(stream));
 754
 755         stream.next_in = data;
 756         stream.avail_in = data_size;
 757         stream.next_out = delta_data;
 758         stream.avail_out = delta_size;
 759
 760         inflateInit(&stream);
 761         st = inflate(&stream, Z_FINISH);
 762         inflateEnd(&stream);
 763         if ((st != Z_STREAM_END) || stream.total_out != delta_size)
 764                 die("delta data unpack failed");
 765
 766         /* This may recursively unpack the base, which is what we want */
 767         base = read_sha1_file(base_sha1, type, &base_size);
 768         if (!base)
 769                 die("failed to read delta-pack base object %s",
 770                     sha1_to_hex(base_sha1));
 771         result = patch_delta(base, base_size,
 772                              delta_data, delta_size,
 773                              &result_size);
 774         if (!result)
 775                 die("failed to apply delta");
 776         free(delta_data);
 777         free(base);
 778         *sizep = result_size;
 779         return result;
 780 }
 781
 782 static void *unpack_non_delta_entry(unsigned char *data,
 783                                     unsigned long size,
 784                                     unsigned long left)
 785 {
 786         int st;
 787         z_stream stream;
 788         char *buffer;
 789
 790         buffer = xmalloc(size + 1);
 791         buffer[size] = 0;
 792         memset(&stream, 0, sizeof(stream));
 793         stream.next_in = data;
 794         stream.avail_in = left;
 795         stream.next_out = buffer;
 796         stream.avail_out = size;
 797
 798         inflateInit(&stream);
 799         st = inflate(&stream, Z_FINISH);
 800         inflateEnd(&stream);
 801         if ((st != Z_STREAM_END) || stream.total_out != size) {
 802                 free(buffer);
 803                 return NULL;
 804         }
 805
 806         return buffer;
 807 }
 808
 809 static void *unpack_entry(struct pack_entry *entry,
 810                           char *type, unsigned long *sizep)
 811 {
 812         struct packed_git *p = entry->p;
 813         unsigned long offset, size, left;
 814         unsigned char *pack;
 815         enum object_type kind;
 816
 817         if (use_packed_git(p))
 818                 die("cannot map packed file");
 819
 820         offset = unpack_object_header(p, entry->offset, &kind, &size);
 821         pack = p->pack_base + offset;
 822         left = p->pack_size - offset;
 823         switch (kind) {
 824         case OBJ_DELTA:
 825                 return unpack_delta_entry(pack, size, left, type, sizep);
 826         case OBJ_COMMIT:
 827                 strcpy(type, "commit");
 828                 break;
 829         case OBJ_TREE:
 830                 strcpy(type, "tree");
 831                 break;
 832         case OBJ_BLOB:
 833                 strcpy(type, "blob");
 834                 break;
 835         case OBJ_TAG:
 836                 strcpy(type, "tag");
 837                 break;
 838         default:
 839                 die("corrupted pack file");
 840         }
 841         *sizep = size;
 842         return unpack_non_delta_entry(pack, size, left);
 843 }
 844
 845 static int find_pack_entry_1(const unsigned char *sha1,
 846                              struct pack_entry *e, struct packed_git *p)
 847 {
 848         int *level1_ofs = p->index_base;
 849         int hi = ntohl(level1_ofs[*sha1]);
 850         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
 851         void *index = p->index_base + 256;
 852
 853         do {
 854                 int mi = (lo + hi) / 2;
 855                 int cmp = memcmp(index + 24 * mi + 4, sha1, 20);
 856                 if (!cmp) {
 857                         e->offset = ntohl(*((int*)(index + 24 * mi)));
 858                         memcpy(e->sha1, sha1, 20);
 859                         e->p = p;
 860                         return 1;
 861                 }
 862                 if (cmp > 0)
 863                         hi = mi;
 864                 else
 865                         lo = mi+1;
 866         } while (lo < hi);
 867         return 0;
 868 }
 869
 870 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e)
 871 {
 872         struct packed_git *p;
 873         prepare_packed_git();
 874
 875         for (p = packed_git; p; p = p->next) {
 876                 if (find_pack_entry_1(sha1, e, p))
 877                         return 1;
 878         }
 879         return 0;
 880 }
 881
 882 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
 883 {
 884         int status;
 885         unsigned long mapsize, size;
 886         void *map;
 887         z_stream stream;
 888         char hdr[128];
 889
 890         map = map_sha1_file_internal(sha1, &mapsize, 0);
 891         if (!map) {
 892                 struct pack_entry e;
 893
 894                 if (!find_pack_entry(sha1, &e))
 895                         return error("unable to find %s", sha1_to_hex(sha1));
 896                 if (!packed_object_info(&e, type, sizep))
 897                         return 0;
 898                 /* sheesh */
 899                 map = unpack_entry(&e, type, sizep);
 900                 free(map);
 901                 return (map == NULL) ? 0 : -1;
 902         }
 903         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
 904                 status = error("unable to unpack %s header",
 905                                sha1_to_hex(sha1));
 906         if (parse_sha1_header(hdr, type, &size) < 0)
 907                 status = error("unable to parse %s header", sha1_to_hex(sha1));
 908         else {
 909                 status = 0;
 910                 *sizep = size;
 911         }
 912         inflateEnd(&stream);
 913         munmap(map, mapsize);
 914         return status;
 915 }
 916
 917 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
 918 {
 919         struct pack_entry e;
 920
 921         if (!find_pack_entry(sha1, &e)) {
 922                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
 923                 return NULL;
 924         }
 925         return unpack_entry(&e, type, size);
 926 }
 927
 928 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
 929 {
 930         unsigned long mapsize;
 931         void *map, *buf;
 932
 933         map = map_sha1_file_internal(sha1, &mapsize, 0);
 934         if (map) {
 935                 buf = unpack_sha1_file(map, mapsize, type, size);
 936                 munmap(map, mapsize);
 937                 return buf;
 938         }
 939         return read_packed_sha1(sha1, type, size);
 940 }
 941
 942 void *read_object_with_reference(const unsigned char *sha1,
 943                                  const char *required_type,
 944                                  unsigned long *size,
 945                                  unsigned char *actual_sha1_return)
 946 {
 947         char type[20];
 948         void *buffer;
 949         unsigned long isize;
 950         unsigned char actual_sha1[20];
 951
 952         memcpy(actual_sha1, sha1, 20);
 953         while (1) {
 954                 int ref_length = -1;
 955                 const char *ref_type = NULL;
 956
 957                 buffer = read_sha1_file(actual_sha1, type, &isize);
 958                 if (!buffer)
 959                         return NULL;
 960                 if (!strcmp(type, required_type)) {
 961                         *size = isize;
 962                         if (actual_sha1_return)
 963                                 memcpy(actual_sha1_return, actual_sha1, 20);
 964                         return buffer;
 965                 }
 966                 /* Handle references */
 967                 else if (!strcmp(type, "commit"))
 968                         ref_type = "tree ";
 969                 else if (!strcmp(type, "tag"))
 970                         ref_type = "object ";
 971                 else {
 972                         free(buffer);
 973                         return NULL;
 974                 }
 975                 ref_length = strlen(ref_type);
 976
 977                 if (memcmp(buffer, ref_type, ref_length) ||
 978                     get_sha1_hex(buffer + ref_length, actual_sha1)) {
 979                         free(buffer);
 980                         return NULL;
 981                 }
 982                 /* Now we have the ID of the referred-to object in
 983                  * actual_sha1.  Check again. */
 984         }
 985 }
 986
 987 static char *write_sha1_file_prepare(void *buf,
 988                                      unsigned long len,
 989                                      const char *type,
 990                                      unsigned char *sha1,
 991                                      unsigned char *hdr,
 992                                      int *hdrlen)
 993 {
 994         SHA_CTX c;
 995
 996         /* Generate the header */
 997         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
 998
 999         /* Sha1.. */
1000         SHA1_Init(&c);
1001         SHA1_Update(&c, hdr, *hdrlen);
1002         SHA1_Update(&c, buf, len);
1003         SHA1_Final(sha1, &c);
1004
1005         return sha1_file_name(sha1);
1006 }
1007
1008 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1009 {
1010         int size;
1011         unsigned char *compressed;
1012         z_stream stream;
1013         unsigned char sha1[20];
1014         char *filename;
1015         static char tmpfile[PATH_MAX];
1016         unsigned char hdr[50];
1017         int fd, hdrlen, ret;
1018
1019         /* Normally if we have it in the pack then we do not bother writing
1020          * it out into .git/objects/??/?{38} file.
1021          */
1022         filename = write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1023         if (returnsha1)
1024                 memcpy(returnsha1, sha1, 20);
1025         if (has_sha1_file(sha1))
1026                 return 0;
1027         fd = open(filename, O_RDONLY);
1028         if (fd >= 0) {
1029                 /*
1030                  * FIXME!!! We might do collision checking here, but we'd
1031                  * need to uncompress the old file and check it. Later.
1032                  */
1033                 close(fd);
1034                 return 0;
1035         }
1036
1037         if (errno != ENOENT) {
1038                 fprintf(stderr, "sha1 file %s: %s", filename, strerror(errno));
1039                 return -1;
1040         }
1041
1042         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1043
1044         fd = mkstemp(tmpfile);
1045         if (fd < 0) {
1046                 fprintf(stderr, "unable to create temporary sha1 filename %s: %s", tmpfile, strerror(errno));
1047                 return -1;
1048         }
1049
1050         /* Set it up */
1051         memset(&stream, 0, sizeof(stream));
1052         deflateInit(&stream, Z_BEST_COMPRESSION);
1053         size = deflateBound(&stream, len+hdrlen);
1054         compressed = xmalloc(size);
1055
1056         /* Compress it */
1057         stream.next_out = compressed;
1058         stream.avail_out = size;
1059
1060         /* First header.. */
1061         stream.next_in = hdr;
1062         stream.avail_in = hdrlen;
1063         while (deflate(&stream, 0) == Z_OK)
1064                 /* nothing */;
1065
1066         /* Then the data itself.. */
1067         stream.next_in = buf;
1068         stream.avail_in = len;
1069         while (deflate(&stream, Z_FINISH) == Z_OK)
1070                 /* nothing */;
1071         deflateEnd(&stream);
1072         size = stream.total_out;
1073
1074         if (write(fd, compressed, size) != size)
1075                 die("unable to write file");
1076         fchmod(fd, 0444);
1077         close(fd);
1078         free(compressed);
1079
1080         ret = link(tmpfile, filename);
1081         if (ret < 0) {
1082                 ret = errno;
1083
1084                 /*
1085                  * Coda hack - coda doesn't like cross-directory links,
1086                  * so we fall back to a rename, which will mean that it
1087                  * won't be able to check collisions, but that's not a
1088                  * big deal.
1089                  *
1090                  * When this succeeds, we just return 0. We have nothing
1091                  * left to unlink.
1092                  */
1093                 if (ret == EXDEV && !rename(tmpfile, filename))
1094                         return 0;
1095         }
1096         unlink(tmpfile);
1097         if (ret) {
1098                 if (ret != EEXIST) {
1099                         fprintf(stderr, "unable to write sha1 filename %s: %s", filename, strerror(ret));
1100                         return -1;
1101                 }
1102                 /* FIXME!!! Collision check here ? */
1103         }
1104
1105         return 0;
1106 }
1107
1108 int write_sha1_from_fd(const unsigned char *sha1, int fd)
1109 {
1110         char *filename = sha1_file_name(sha1);
1111
1112         int local;
1113         z_stream stream;
1114         unsigned char real_sha1[20];
1115         unsigned char buf[4096];
1116         unsigned char discard[4096];
1117         int ret;
1118         SHA_CTX c;
1119
1120         local = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
1121
1122         if (local < 0)
1123                 return error("Couldn't open %s\n", filename);
1124
1125         memset(&stream, 0, sizeof(stream));
1126
1127         inflateInit(&stream);
1128
1129         SHA1_Init(&c);
1130
1131         do {
1132                 ssize_t size;
1133                 size = read(fd, buf, 4096);
1134                 if (size <= 0) {
1135                         close(local);
1136                         unlink(filename);
1137                         if (!size)
1138                                 return error("Connection closed?");
1139                         perror("Reading from connection");
1140                         return -1;
1141                 }
1142                 write(local, buf, size);
1143                 stream.avail_in = size;
1144                 stream.next_in = buf;
1145                 do {
1146                         stream.next_out = discard;
1147                         stream.avail_out = sizeof(discard);
1148                         ret = inflate(&stream, Z_SYNC_FLUSH);
1149                         SHA1_Update(&c, discard, sizeof(discard) -
1150                                     stream.avail_out);
1151                 } while (stream.avail_in && ret == Z_OK);
1152
1153         } while (ret == Z_OK);
1154         inflateEnd(&stream);
1155
1156         close(local);
1157         SHA1_Final(real_sha1, &c);
1158         if (ret != Z_STREAM_END) {
1159                 unlink(filename);
1160                 return error("File %s corrupted", sha1_to_hex(sha1));
1161         }
1162         if (memcmp(sha1, real_sha1, 20)) {
1163                 unlink(filename);
1164                 return error("File %s has bad hash\n", sha1_to_hex(sha1));
1165         }
1166
1167         return 0;
1168 }
1169
1170 int has_sha1_file(const unsigned char *sha1)
1171 {
1172         struct stat st;
1173         struct pack_entry e;
1174
1175         if (find_sha1_file(sha1, &st))
1176                 return 1;
1177         return find_pack_entry(sha1, &e);
1178 }
1179
1180 int index_fd(unsigned char *sha1, int fd, struct stat *st)
1181 {
1182         unsigned long size = st->st_size;
1183         void *buf;
1184         int ret;
1185
1186         buf = "";
1187         if (size)
1188                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1189         close(fd);
1190         if ((int)(long)buf == -1)
1191                 return -1;
1192
1193         ret = write_sha1_file(buf, size, "blob", sha1);
1194         if (size)
1195                 munmap(buf, size);
1196         return ret;
1197 }