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_last_used = 0;
 400         return p;
 401 }
 402
 403 static void prepare_packed_git_one(char *objdir)
 404 {
 405         char path[PATH_MAX];
 406         int len;
 407         DIR *dir;
 408         struct dirent *de;
 409
 410         sprintf(path, "%s/pack", objdir);
 411         len = strlen(path);
 412         dir = opendir(path);
 413         if (!dir)
 414                 return;
 415         path[len++] = '/';
 416         while ((de = readdir(dir)) != NULL) {
 417                 int namelen = strlen(de->d_name);
 418                 struct packed_git *p;
 419
 420                 if (strcmp(de->d_name + namelen - 4, ".idx"))
 421                         continue;
 422
 423                 /* we have .idx.  Is it a file we can map? */
 424                 strcpy(path + len, de->d_name);
 425                 p = add_packed_git(path, len + namelen);
 426                 if (!p)
 427                         continue;
 428                 p->next = packed_git;
 429                 packed_git = p;
 430         }
 431 }
 432
 433 static void prepare_packed_git(void)
 434 {
 435         int i;
 436         static int run_once = 0;
 437
 438         if (run_once++)
 439                 return;
 440
 441         prepare_packed_git_one(get_object_directory());
 442         if (!alt_odb)
 443                 prepare_alt_odb();
 444         for (i = 0; alt_odb[i].base != NULL; i++) {
 445                 alt_odb[i].name[0] = 0;
 446                 prepare_packed_git_one(alt_odb[i].base);
 447         }
 448 }
 449
 450 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 451 {
 452         char header[100];
 453         unsigned char real_sha1[20];
 454         SHA_CTX c;
 455
 456         SHA1_Init(&c);
 457         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 458         SHA1_Update(&c, map, size);
 459         SHA1_Final(real_sha1, &c);
 460         return memcmp(sha1, real_sha1, 20) ? -1 : 0;
 461 }
 462
 463 static void *map_sha1_file_internal(const unsigned char *sha1,
 464                                     unsigned long *size,
 465                                     int say_error)
 466 {
 467         struct stat st;
 468         void *map;
 469         int fd;
 470         char *filename = find_sha1_file(sha1, &st);
 471
 472         if (!filename) {
 473                 if (say_error)
 474                         error("cannot map sha1 file %s", sha1_to_hex(sha1));
 475                 return NULL;
 476         }
 477
 478         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 479         if (fd < 0) {
 480                 /* See if it works without O_NOATIME */
 481                 switch (sha1_file_open_flag) {
 482                 default:
 483                         fd = open(filename, O_RDONLY);
 484                         if (fd >= 0)
 485                                 break;
 486                 /* Fallthrough */
 487                 case 0:
 488                         if (say_error)
 489                                 perror(filename);
 490                         return NULL;
 491                 }
 492
 493                 /* If it failed once, it will probably fail again.
 494                  * Stop using O_NOATIME
 495                  */
 496                 sha1_file_open_flag = 0;
 497         }
 498         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 499         close(fd);
 500         if (-1 == (int)(long)map)
 501                 return NULL;
 502         *size = st.st_size;
 503         return map;
 504 }
 505
 506 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 507 {
 508         return map_sha1_file_internal(sha1, size, 1);
 509 }
 510
 511 int unpack_sha1_header(z_stream *stream, void *map, unsigned long mapsize, void *buffer, unsigned long size)
 512 {
 513         /* Get the data stream */
 514         memset(stream, 0, sizeof(*stream));
 515         stream->next_in = map;
 516         stream->avail_in = mapsize;
 517         stream->next_out = buffer;
 518         stream->avail_out = size;
 519
 520         inflateInit(stream);
 521         return inflate(stream, 0);
 522 }
 523
 524 void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 525 {
 526         int bytes = strlen(buffer) + 1;
 527         unsigned char *buf = xmalloc(1+size);
 528
 529         memcpy(buf, buffer + bytes, stream->total_out - bytes);
 530         bytes = stream->total_out - bytes;
 531         if (bytes < size) {
 532                 stream->next_out = buf + bytes;
 533                 stream->avail_out = size - bytes;
 534                 while (inflate(stream, Z_FINISH) == Z_OK)
 535                         /* nothing */;
 536         }
 537         buf[size] = 0;
 538         inflateEnd(stream);
 539         return buf;
 540 }
 541
 542 /*
 543  * We used to just use "sscanf()", but that's actually way
 544  * too permissive for what we want to check. So do an anal
 545  * object header parse by hand.
 546  */
 547 int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 548 {
 549         int i;
 550         unsigned long size;
 551
 552         /*
 553          * The type can be at most ten bytes (including the
 554          * terminating '\0' that we add), and is followed by
 555          * a space.
 556          */
 557         i = 10;
 558         for (;;) {
 559                 char c = *hdr++;
 560                 if (c == ' ')
 561                         break;
 562                 if (!--i)
 563                         return -1;
 564                 *type++ = c;
 565         }
 566         *type = 0;
 567
 568         /*
 569          * The length must follow immediately, and be in canonical
 570          * decimal format (ie "010" is not valid).
 571          */
 572         size = *hdr++ - '0';
 573         if (size > 9)
 574                 return -1;
 575         if (size) {
 576                 for (;;) {
 577                         unsigned long c = *hdr - '0';
 578                         if (c > 9)
 579                                 break;
 580                         hdr++;
 581                         size = size * 10 + c;
 582                 }
 583         }
 584         *sizep = size;
 585
 586         /*
 587          * The length must be followed by a zero byte
 588          */
 589         return *hdr ? -1 : 0;
 590 }
 591
 592 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 593 {
 594         int ret;
 595         z_stream stream;
 596         char hdr[8192];
 597
 598         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
 599         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
 600                 return NULL;
 601
 602         return unpack_sha1_rest(&stream, hdr, *size);
 603 }
 604
 605 static int packed_delta_info(unsigned char *base_sha1,
 606                              unsigned long delta_size,
 607                              unsigned long left,
 608                              char *type,
 609                              unsigned long *sizep)
 610 {
 611         unsigned char *data;
 612         unsigned char delta_head[64];
 613         int i;
 614         unsigned char cmd;
 615         unsigned long data_size, result_size, base_size, verify_base_size;
 616         z_stream stream;
 617         int st;
 618
 619         if (left < 20)
 620                 die("truncated pack file");
 621         if (sha1_object_info(base_sha1, type, &base_size))
 622                 die("cannot get info for delta-pack base");
 623
 624         data = base_sha1 + 20;
 625         data_size = left - 20;
 626
 627         memset(&stream, 0, sizeof(stream));
 628
 629         stream.next_in = data;
 630         stream.avail_in = data_size;
 631         stream.next_out = delta_head;
 632         stream.avail_out = sizeof(delta_head);
 633
 634         inflateInit(&stream);
 635         st = inflate(&stream, Z_FINISH);
 636         inflateEnd(&stream);
 637         if ((st != Z_STREAM_END) && stream.total_out != sizeof(delta_head))
 638                 die("delta data unpack-initial failed");
 639
 640         /* Examine the initial part of the delta to figure out
 641          * the result size.  Verify the base size while we are at it.
 642          */
 643         data = delta_head;
 644         verify_base_size = i = 0;
 645         cmd = *data++;
 646         while (cmd) {
 647                 if (cmd & 1)
 648                         verify_base_size |= *data++ << i;
 649                 i += 8;
 650                 cmd >>= 1;
 651         }
 652
 653         /* Read the result size */
 654         result_size = i = 0;
 655         cmd = *data++;
 656         while (cmd) {
 657                 if (cmd & 1)
 658                         result_size |= *data++ << i;
 659                 i += 8;
 660                 cmd >>= 1;
 661         }
 662         if (verify_base_size != base_size)
 663                 die("delta base size mismatch");
 664
 665         *sizep = result_size;
 666         return 0;
 667 }
 668
 669 static unsigned long unpack_object_header(struct packed_git *p, unsigned long offset,
 670         enum object_type *type, unsigned long *sizep)
 671 {
 672         unsigned char *pack, c;
 673         unsigned long size;
 674
 675         if (offset >= p->pack_size)
 676                 die("object offset outside of pack file");
 677
 678         pack =  p->pack_base + offset;
 679         c = *pack++;
 680         offset++;
 681         *type = (c >> 4) & 7;
 682         size = c & 15;
 683         while (c & 0x80) {
 684                 if (offset >= p->pack_size)
 685                         die("object offset outside of pack file");
 686                 c = *pack++;
 687                 offset++;
 688                 size = (size << 7) | (c & 0x7f);
 689         }
 690         *sizep = size;
 691         return offset;
 692 }
 693
 694 static int packed_object_info(struct pack_entry *entry,
 695                               char *type, unsigned long *sizep)
 696 {
 697         struct packed_git *p = entry->p;
 698         unsigned long offset, size, left;
 699         unsigned char *pack;
 700         enum object_type kind;
 701
 702         if (use_packed_git(p))
 703                 die("cannot map packed file");
 704
 705         offset = unpack_object_header(p, entry->offset, &kind, &size);
 706         pack = p->pack_base + offset;
 707         left = p->pack_size - offset;
 708
 709         switch (kind) {
 710         case OBJ_DELTA:
 711                 return packed_delta_info(pack, size, left, type, sizep);
 712                 break;
 713         case OBJ_COMMIT:
 714                 strcpy(type, "commit");
 715                 break;
 716         case OBJ_TREE:
 717                 strcpy(type, "tree");
 718                 break;
 719         case OBJ_BLOB:
 720                 strcpy(type, "blob");
 721                 break;
 722         case OBJ_TAG:
 723                 strcpy(type, "tag");
 724                 break;
 725         default:
 726                 die("corrupted pack file");
 727         }
 728         *sizep = size;
 729         return 0;
 730 }
 731
 732 /* forward declaration for a mutually recursive function */
 733 static void *unpack_entry(struct pack_entry *, char *, unsigned long *);
 734
 735 static void *unpack_delta_entry(unsigned char *base_sha1,
 736                                 unsigned long delta_size,
 737                                 unsigned long left,
 738                                 char *type,
 739                                 unsigned long *sizep)
 740 {
 741         void *data, *delta_data, *result, *base;
 742         unsigned long data_size, result_size, base_size;
 743         z_stream stream;
 744         int st;
 745
 746         if (left < 20)
 747                 die("truncated pack file");
 748         data = base_sha1 + 20;
 749         data_size = left - 20;
 750         delta_data = xmalloc(delta_size);
 751
 752         memset(&stream, 0, sizeof(stream));
 753
 754         stream.next_in = data;
 755         stream.avail_in = data_size;
 756         stream.next_out = delta_data;
 757         stream.avail_out = delta_size;
 758
 759         inflateInit(&stream);
 760         st = inflate(&stream, Z_FINISH);
 761         inflateEnd(&stream);
 762         if ((st != Z_STREAM_END) || stream.total_out != delta_size)
 763                 die("delta data unpack failed");
 764
 765         /* This may recursively unpack the base, which is what we want */
 766         base = read_sha1_file(base_sha1, type, &base_size);
 767         if (!base)
 768                 die("failed to read delta-pack base object %s",
 769                     sha1_to_hex(base_sha1));
 770         result = patch_delta(base, base_size,
 771                              delta_data, delta_size,
 772                              &result_size);
 773         if (!result)
 774                 die("failed to apply delta");
 775         free(delta_data);
 776         free(base);
 777         *sizep = result_size;
 778         return result;
 779 }
 780
 781 static void *unpack_non_delta_entry(unsigned char *data,
 782                                     unsigned long size,
 783                                     unsigned long left)
 784 {
 785         int st;
 786         z_stream stream;
 787         char *buffer;
 788
 789         buffer = xmalloc(size + 1);
 790         buffer[size] = 0;
 791         memset(&stream, 0, sizeof(stream));
 792         stream.next_in = data;
 793         stream.avail_in = left;
 794         stream.next_out = buffer;
 795         stream.avail_out = size;
 796
 797         inflateInit(&stream);
 798         st = inflate(&stream, Z_FINISH);
 799         inflateEnd(&stream);
 800         if ((st != Z_STREAM_END) || stream.total_out != size) {
 801                 free(buffer);
 802                 return NULL;
 803         }
 804
 805         return buffer;
 806 }
 807
 808 static void *unpack_entry(struct pack_entry *entry,
 809                           char *type, unsigned long *sizep)
 810 {
 811         struct packed_git *p = entry->p;
 812         unsigned long offset, size, left;
 813         unsigned char *pack;
 814         enum object_type kind;
 815
 816         if (use_packed_git(p))
 817                 die("cannot map packed file");
 818
 819         offset = unpack_object_header(p, entry->offset, &kind, &size);
 820         pack = p->pack_base + offset;
 821         left = p->pack_size - offset;
 822         switch (kind) {
 823         case OBJ_DELTA:
 824                 return unpack_delta_entry(pack, size, left, type, sizep);
 825         case OBJ_COMMIT:
 826                 strcpy(type, "commit");
 827                 break;
 828         case OBJ_TREE:
 829                 strcpy(type, "tree");
 830                 break;
 831         case OBJ_BLOB:
 832                 strcpy(type, "blob");
 833                 break;
 834         case OBJ_TAG:
 835                 strcpy(type, "tag");
 836                 break;
 837         default:
 838                 die("corrupted pack file");
 839         }
 840         *sizep = size;
 841         return unpack_non_delta_entry(pack, size, left);
 842 }
 843
 844 static int find_pack_entry_1(const unsigned char *sha1,
 845                              struct pack_entry *e, struct packed_git *p)
 846 {
 847         int *level1_ofs = p->index_base;
 848         int hi = ntohl(level1_ofs[*sha1]);
 849         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
 850         void *index = p->index_base + 256;
 851
 852         do {
 853                 int mi = (lo + hi) / 2;
 854                 int cmp = memcmp(index + 24 * mi + 4, sha1, 20);
 855                 if (!cmp) {
 856                         e->offset = ntohl(*((int*)(index + 24 * mi)));
 857                         memcpy(e->sha1, sha1, 20);
 858                         e->p = p;
 859                         return 1;
 860                 }
 861                 if (cmp > 0)
 862                         hi = mi;
 863                 else
 864                         lo = mi+1;
 865         } while (lo < hi);
 866         return 0;
 867 }
 868
 869 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e)
 870 {
 871         struct packed_git *p;
 872         prepare_packed_git();
 873
 874         for (p = packed_git; p; p = p->next) {
 875                 if (find_pack_entry_1(sha1, e, p))
 876                         return 1;
 877         }
 878         return 0;
 879 }
 880
 881 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
 882 {
 883         int status;
 884         unsigned long mapsize, size;
 885         void *map;
 886         z_stream stream;
 887         char hdr[128];
 888
 889         map = map_sha1_file_internal(sha1, &mapsize, 0);
 890         if (!map) {
 891                 struct pack_entry e;
 892
 893                 if (!find_pack_entry(sha1, &e))
 894                         return error("unable to find %s", sha1_to_hex(sha1));
 895                 if (!packed_object_info(&e, type, sizep))
 896                         return 0;
 897                 /* sheesh */
 898                 map = unpack_entry(&e, type, sizep);
 899                 free(map);
 900                 return (map == NULL) ? 0 : -1;
 901         }
 902         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
 903                 status = error("unable to unpack %s header",
 904                                sha1_to_hex(sha1));
 905         if (parse_sha1_header(hdr, type, &size) < 0)
 906                 status = error("unable to parse %s header", sha1_to_hex(sha1));
 907         else {
 908                 status = 0;
 909                 *sizep = size;
 910         }
 911         inflateEnd(&stream);
 912         munmap(map, mapsize);
 913         return status;
 914 }
 915
 916 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
 917 {
 918         struct pack_entry e;
 919
 920         if (!find_pack_entry(sha1, &e)) {
 921                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
 922                 return NULL;
 923         }
 924         return unpack_entry(&e, type, size);
 925 }
 926
 927 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
 928 {
 929         unsigned long mapsize;
 930         void *map, *buf;
 931
 932         map = map_sha1_file_internal(sha1, &mapsize, 0);
 933         if (map) {
 934                 buf = unpack_sha1_file(map, mapsize, type, size);
 935                 munmap(map, mapsize);
 936                 return buf;
 937         }
 938         return read_packed_sha1(sha1, type, size);
 939 }
 940
 941 void *read_object_with_reference(const unsigned char *sha1,
 942                                  const char *required_type,
 943                                  unsigned long *size,
 944                                  unsigned char *actual_sha1_return)
 945 {
 946         char type[20];
 947         void *buffer;
 948         unsigned long isize;
 949         unsigned char actual_sha1[20];
 950
 951         memcpy(actual_sha1, sha1, 20);
 952         while (1) {
 953                 int ref_length = -1;
 954                 const char *ref_type = NULL;
 955
 956                 buffer = read_sha1_file(actual_sha1, type, &isize);
 957                 if (!buffer)
 958                         return NULL;
 959                 if (!strcmp(type, required_type)) {
 960                         *size = isize;
 961                         if (actual_sha1_return)
 962                                 memcpy(actual_sha1_return, actual_sha1, 20);
 963                         return buffer;
 964                 }
 965                 /* Handle references */
 966                 else if (!strcmp(type, "commit"))
 967                         ref_type = "tree ";
 968                 else if (!strcmp(type, "tag"))
 969                         ref_type = "object ";
 970                 else {
 971                         free(buffer);
 972                         return NULL;
 973                 }
 974                 ref_length = strlen(ref_type);
 975
 976                 if (memcmp(buffer, ref_type, ref_length) ||
 977                     get_sha1_hex(buffer + ref_length, actual_sha1)) {
 978                         free(buffer);
 979                         return NULL;
 980                 }
 981                 /* Now we have the ID of the referred-to object in
 982                  * actual_sha1.  Check again. */
 983         }
 984 }
 985
 986 static char *write_sha1_file_prepare(void *buf,
 987                                      unsigned long len,
 988                                      const char *type,
 989                                      unsigned char *sha1,
 990                                      unsigned char *hdr,
 991                                      int *hdrlen)
 992 {
 993         SHA_CTX c;
 994
 995         /* Generate the header */
 996         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
 997
 998         /* Sha1.. */
 999         SHA1_Init(&c);
1000         SHA1_Update(&c, hdr, *hdrlen);
1001         SHA1_Update(&c, buf, len);
1002         SHA1_Final(sha1, &c);
1003
1004         return sha1_file_name(sha1);
1005 }
1006
1007 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1008 {
1009         int size;
1010         unsigned char *compressed;
1011         z_stream stream;
1012         unsigned char sha1[20];
1013         char *filename;
1014         static char tmpfile[PATH_MAX];
1015         unsigned char hdr[50];
1016         int fd, hdrlen, ret;
1017
1018         /* Normally if we have it in the pack then we do not bother writing
1019          * it out into .git/objects/??/?{38} file.
1020          */
1021         filename = write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1022         if (returnsha1)
1023                 memcpy(returnsha1, sha1, 20);
1024         if (has_sha1_file(sha1))
1025                 return 0;
1026         fd = open(filename, O_RDONLY);
1027         if (fd >= 0) {
1028                 /*
1029                  * FIXME!!! We might do collision checking here, but we'd
1030                  * need to uncompress the old file and check it. Later.
1031                  */
1032                 close(fd);
1033                 return 0;
1034         }
1035
1036         if (errno != ENOENT) {
1037                 fprintf(stderr, "sha1 file %s: %s", filename, strerror(errno));
1038                 return -1;
1039         }
1040
1041         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1042
1043         fd = mkstemp(tmpfile);
1044         if (fd < 0) {
1045                 fprintf(stderr, "unable to create temporary sha1 filename %s: %s", tmpfile, strerror(errno));
1046                 return -1;
1047         }
1048
1049         /* Set it up */
1050         memset(&stream, 0, sizeof(stream));
1051         deflateInit(&stream, Z_BEST_COMPRESSION);
1052         size = deflateBound(&stream, len+hdrlen);
1053         compressed = xmalloc(size);
1054
1055         /* Compress it */
1056         stream.next_out = compressed;
1057         stream.avail_out = size;
1058
1059         /* First header.. */
1060         stream.next_in = hdr;
1061         stream.avail_in = hdrlen;
1062         while (deflate(&stream, 0) == Z_OK)
1063                 /* nothing */;
1064
1065         /* Then the data itself.. */
1066         stream.next_in = buf;
1067         stream.avail_in = len;
1068         while (deflate(&stream, Z_FINISH) == Z_OK)
1069                 /* nothing */;
1070         deflateEnd(&stream);
1071         size = stream.total_out;
1072
1073         if (write(fd, compressed, size) != size)
1074                 die("unable to write file");
1075         fchmod(fd, 0444);
1076         close(fd);
1077         free(compressed);
1078
1079         ret = link(tmpfile, filename);
1080         if (ret < 0) {
1081                 ret = errno;
1082
1083                 /*
1084                  * Coda hack - coda doesn't like cross-directory links,
1085                  * so we fall back to a rename, which will mean that it
1086                  * won't be able to check collisions, but that's not a
1087                  * big deal.
1088                  *
1089                  * When this succeeds, we just return 0. We have nothing
1090                  * left to unlink.
1091                  */
1092                 if (ret == EXDEV && !rename(tmpfile, filename))
1093                         return 0;
1094         }
1095         unlink(tmpfile);
1096         if (ret) {
1097                 if (ret != EEXIST) {
1098                         fprintf(stderr, "unable to write sha1 filename %s: %s", filename, strerror(ret));
1099                         return -1;
1100                 }
1101                 /* FIXME!!! Collision check here ? */
1102         }
1103
1104         return 0;
1105 }
1106
1107 int write_sha1_from_fd(const unsigned char *sha1, int fd)
1108 {
1109         char *filename = sha1_file_name(sha1);
1110
1111         int local;
1112         z_stream stream;
1113         unsigned char real_sha1[20];
1114         unsigned char buf[4096];
1115         unsigned char discard[4096];
1116         int ret;
1117         SHA_CTX c;
1118
1119         local = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
1120
1121         if (local < 0)
1122                 return error("Couldn't open %s\n", filename);
1123
1124         memset(&stream, 0, sizeof(stream));
1125
1126         inflateInit(&stream);
1127
1128         SHA1_Init(&c);
1129
1130         do {
1131                 ssize_t size;
1132                 size = read(fd, buf, 4096);
1133                 if (size <= 0) {
1134                         close(local);
1135                         unlink(filename);
1136                         if (!size)
1137                                 return error("Connection closed?");
1138                         perror("Reading from connection");
1139                         return -1;
1140                 }
1141                 write(local, buf, size);
1142                 stream.avail_in = size;
1143                 stream.next_in = buf;
1144                 do {
1145                         stream.next_out = discard;
1146                         stream.avail_out = sizeof(discard);
1147                         ret = inflate(&stream, Z_SYNC_FLUSH);
1148                         SHA1_Update(&c, discard, sizeof(discard) -
1149                                     stream.avail_out);
1150                 } while (stream.avail_in && ret == Z_OK);
1151
1152         } while (ret == Z_OK);
1153         inflateEnd(&stream);
1154
1155         close(local);
1156         SHA1_Final(real_sha1, &c);
1157         if (ret != Z_STREAM_END) {
1158                 unlink(filename);
1159                 return error("File %s corrupted", sha1_to_hex(sha1));
1160         }
1161         if (memcmp(sha1, real_sha1, 20)) {
1162                 unlink(filename);
1163                 return error("File %s has bad hash\n", sha1_to_hex(sha1));
1164         }
1165
1166         return 0;
1167 }
1168
1169 int has_sha1_file(const unsigned char *sha1)
1170 {
1171         struct stat st;
1172         struct pack_entry e;
1173
1174         if (find_sha1_file(sha1, &st))
1175                 return 1;
1176         return find_pack_entry(sha1, &e);
1177 }
1178
1179 int index_fd(unsigned char *sha1, int fd, struct stat *st)
1180 {
1181         unsigned long size = st->st_size;
1182         void *buf;
1183         int ret;
1184
1185         buf = "";
1186         if (size)
1187                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1188         close(fd);
1189         if ((int)(long)buf == -1)
1190                 return -1;
1191
1192         ret = write_sha1_file(buf, size, "blob", sha1);
1193         if (size)
1194                 munmap(buf, size);
1195         return ret;
1196 }