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 safe_create_leading_directories(char *path)
 106 {
 107         char *pos = path;
 108
 109         while (pos) {
 110                 pos = strchr(pos, '/');
 111                 if (!pos)
 112                         break;
 113                 *pos = 0;
 114                 if (mkdir(path, 0777) < 0)
 115                         if (errno != EEXIST) {
 116                                 *pos = '/';
 117                                 return -1;
 118                         }
 119                 *pos++ = '/';
 120         }
 121         return 0;
 122 }
 123
 124 int get_sha1(const char *str, unsigned char *sha1)
 125 {
 126         static const char *prefix[] = {
 127                 "",
 128                 "refs",
 129                 "refs/tags",
 130                 "refs/heads",
 131                 "refs/snap",
 132                 NULL
 133         };
 134         const char **p;
 135
 136         if (!get_sha1_hex(str, sha1))
 137                 return 0;
 138
 139         for (p = prefix; *p; p++) {
 140                 char * pathname = git_path("%s/%s", *p, str);
 141                 if (!get_sha1_file(pathname, sha1))
 142                         return 0;
 143         }
 144
 145         return -1;
 146 }
 147
 148 char * sha1_to_hex(const unsigned char *sha1)
 149 {
 150         static char buffer[50];
 151         static const char hex[] = "0123456789abcdef";
 152         char *buf = buffer;
 153         int i;
 154
 155         for (i = 0; i < 20; i++) {
 156                 unsigned int val = *sha1++;
 157                 *buf++ = hex[val >> 4];
 158                 *buf++ = hex[val & 0xf];
 159         }
 160         return buffer;
 161 }
 162
 163 static void fill_sha1_path(char *pathbuf, const unsigned char *sha1)
 164 {
 165         int i;
 166         for (i = 0; i < 20; i++) {
 167                 static char hex[] = "0123456789abcdef";
 168                 unsigned int val = sha1[i];
 169                 char *pos = pathbuf + i*2 + (i > 0);
 170                 *pos++ = hex[val >> 4];
 171                 *pos = hex[val & 0xf];
 172         }
 173 }
 174
 175 /*
 176  * NOTE! This returns a statically allocated buffer, so you have to be
 177  * careful about using it. Do a "strdup()" if you need to save the
 178  * filename.
 179  *
 180  * Also note that this returns the location for creating.  Reading
 181  * SHA1 file can happen from any alternate directory listed in the
 182  * DB_ENVIRONMENT environment variable if it is not found in
 183  * the primary object database.
 184  */
 185 char *sha1_file_name(const unsigned char *sha1)
 186 {
 187         static char *name, *base;
 188
 189         if (!base) {
 190                 const char *sha1_file_directory = get_object_directory();
 191                 int len = strlen(sha1_file_directory);
 192                 base = xmalloc(len + 60);
 193                 memcpy(base, sha1_file_directory, len);
 194                 memset(base+len, 0, 60);
 195                 base[len] = '/';
 196                 base[len+3] = '/';
 197                 name = base + len + 1;
 198         }
 199         fill_sha1_path(name, sha1);
 200         return base;
 201 }
 202
 203 struct alternate_object_database *alt_odb;
 204
 205 /*
 206  * Prepare alternate object database registry.
 207  * alt_odb points at an array of struct alternate_object_database.
 208  * This array is terminated with an element that has both its base
 209  * and name set to NULL.  alt_odb[n] comes from n'th non-empty
 210  * element from colon separated ALTERNATE_DB_ENVIRONMENT environment
 211  * variable, and its base points at a statically allocated buffer
 212  * that contains "/the/directory/corresponding/to/.git/objects/...",
 213  * while its name points just after the slash at the end of
 214  * ".git/objects/" in the example above, and has enough space to hold
 215  * 40-byte hex SHA1, an extra slash for the first level indirection,
 216  * and the terminating NUL.
 217  * This function allocates the alt_odb array and all the strings
 218  * pointed by base fields of the array elements with one xmalloc();
 219  * the string pool immediately follows the array.
 220  */
 221 void prepare_alt_odb(void)
 222 {
 223         int pass, totlen, i;
 224         const char *cp, *last;
 225         char *op = NULL;
 226         const char *alt = gitenv(ALTERNATE_DB_ENVIRONMENT) ? : "";
 227
 228         if (alt_odb)
 229                 return;
 230         /* The first pass counts how large an area to allocate to
 231          * hold the entire alt_odb structure, including array of
 232          * structs and path buffers for them.  The second pass fills
 233          * the structure and prepares the path buffers for use by
 234          * fill_sha1_path().
 235          */
 236         for (totlen = pass = 0; pass < 2; pass++) {
 237                 last = alt;
 238                 i = 0;
 239                 do {
 240                         cp = strchr(last, ':') ? : last + strlen(last);
 241                         if (last != cp) {
 242                                 /* 43 = 40-byte + 2 '/' + terminating NUL */
 243                                 int pfxlen = cp - last;
 244                                 int entlen = pfxlen + 43;
 245                                 if (pass == 0)
 246                                         totlen += entlen;
 247                                 else {
 248                                         alt_odb[i].base = op;
 249                                         alt_odb[i].name = op + pfxlen + 1;
 250                                         memcpy(op, last, pfxlen);
 251                                         op[pfxlen] = op[pfxlen + 3] = '/';
 252                                         op[entlen-1] = 0;
 253                                         op += entlen;
 254                                 }
 255                                 i++;
 256                         }
 257                         while (*cp && *cp == ':')
 258                                 cp++;
 259                         last = cp;
 260                 } while (*cp);
 261                 if (pass)
 262                         break;
 263                 alt_odb = xmalloc(sizeof(*alt_odb) * (i + 1) + totlen);
 264                 alt_odb[i].base = alt_odb[i].name = NULL;
 265                 op = (char*)(&alt_odb[i+1]);
 266         }
 267 }
 268
 269 static char *find_sha1_file(const unsigned char *sha1, struct stat *st)
 270 {
 271         int i;
 272         char *name = sha1_file_name(sha1);
 273
 274         if (!stat(name, st))
 275                 return name;
 276         prepare_alt_odb();
 277         for (i = 0; (name = alt_odb[i].name) != NULL; i++) {
 278                 fill_sha1_path(name, sha1);
 279                 if (!stat(alt_odb[i].base, st))
 280                         return alt_odb[i].base;
 281         }
 282         return NULL;
 283 }
 284
 285 #define PACK_MAX_SZ (1<<26)
 286 static int pack_used_ctr;
 287 static unsigned long pack_mapped;
 288 struct packed_git *packed_git;
 289
 290 static int check_packed_git_idx(const char *path, unsigned long *idx_size_,
 291                                 void **idx_map_)
 292 {
 293         void *idx_map;
 294         unsigned int *index;
 295         unsigned long idx_size;
 296         int nr, i;
 297         int fd = open(path, O_RDONLY);
 298         struct stat st;
 299         if (fd < 0)
 300                 return -1;
 301         if (fstat(fd, &st)) {
 302                 close(fd);
 303                 return -1;
 304         }
 305         idx_size = st.st_size;
 306         idx_map = mmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0);
 307         close(fd);
 308         if (idx_map == MAP_FAILED)
 309                 return -1;
 310
 311         index = idx_map;
 312         *idx_map_ = idx_map;
 313         *idx_size_ = idx_size;
 314
 315         /* check index map */
 316         if (idx_size < 4*256 + 20 + 20)
 317                 return error("index file too small");
 318         nr = 0;
 319         for (i = 0; i < 256; i++) {
 320                 unsigned int n = ntohl(index[i]);
 321                 if (n < nr)
 322                         return error("non-monotonic index");
 323                 nr = n;
 324         }
 325
 326         /*
 327          * Total size:
 328          *  - 256 index entries 4 bytes each
 329          *  - 24-byte entries * nr (20-byte sha1 + 4-byte offset)
 330          *  - 20-byte SHA1 of the packfile
 331          *  - 20-byte SHA1 file checksum
 332          */
 333         if (idx_size != 4*256 + nr * 24 + 20 + 20)
 334                 return error("wrong index file size");
 335
 336         return 0;
 337 }
 338
 339 static int unuse_one_packed_git(void)
 340 {
 341         struct packed_git *p, *lru = NULL;
 342
 343         for (p = packed_git; p; p = p->next) {
 344                 if (p->pack_use_cnt || !p->pack_base)
 345                         continue;
 346                 if (!lru || p->pack_last_used < lru->pack_last_used)
 347                         lru = p;
 348         }
 349         if (!lru)
 350                 return 0;
 351         munmap(lru->pack_base, lru->pack_size);
 352         lru->pack_base = NULL;
 353         return 1;
 354 }
 355
 356 void unuse_packed_git(struct packed_git *p)
 357 {
 358         p->pack_use_cnt--;
 359 }
 360
 361 int use_packed_git(struct packed_git *p)
 362 {
 363         if (!p->pack_base) {
 364                 int fd;
 365                 struct stat st;
 366                 void *map;
 367
 368                 pack_mapped += p->pack_size;
 369                 while (PACK_MAX_SZ < pack_mapped && unuse_one_packed_git())
 370                         ; /* nothing */
 371                 fd = open(p->pack_name, O_RDONLY);
 372                 if (fd < 0)
 373                         die("packfile %s cannot be opened", p->pack_name);
 374                 if (fstat(fd, &st)) {
 375                         close(fd);
 376                         die("packfile %s cannot be opened", p->pack_name);
 377                 }
 378                 if (st.st_size != p->pack_size)
 379                         die("packfile %s size mismatch.", p->pack_name);
 380                 map = mmap(NULL, p->pack_size, PROT_READ, MAP_PRIVATE, fd, 0);
 381                 close(fd);
 382                 if (map == MAP_FAILED)
 383                         die("packfile %s cannot be mapped.", p->pack_name);
 384                 p->pack_base = map;
 385
 386                 /* Check if the pack file matches with the index file.
 387                  * this is cheap.
 388                  */
 389                 if (memcmp((char*)(p->index_base) + p->index_size - 40,
 390                            p->pack_base + p->pack_size - 20, 20))
 391                         die("packfile %s does not match index.", p->pack_name);
 392         }
 393         p->pack_last_used = pack_used_ctr++;
 394         p->pack_use_cnt++;
 395         return 0;
 396 }
 397
 398 struct packed_git *add_packed_git(char *path, int path_len)
 399 {
 400         struct stat st;
 401         struct packed_git *p;
 402         unsigned long idx_size;
 403         void *idx_map;
 404
 405         if (check_packed_git_idx(path, &idx_size, &idx_map))
 406                 return NULL;
 407
 408         /* do we have a corresponding .pack file? */
 409         strcpy(path + path_len - 4, ".pack");
 410         if (stat(path, &st) || !S_ISREG(st.st_mode)) {
 411                 munmap(idx_map, idx_size);
 412                 return NULL;
 413         }
 414         /* ok, it looks sane as far as we can check without
 415          * actually mapping the pack file.
 416          */
 417         p = xmalloc(sizeof(*p) + path_len + 2);
 418         strcpy(p->pack_name, path);
 419         p->index_size = idx_size;
 420         p->pack_size = st.st_size;
 421         p->index_base = idx_map;
 422         p->next = NULL;
 423         p->pack_base = NULL;
 424         p->pack_last_used = 0;
 425         p->pack_use_cnt = 0;
 426         return p;
 427 }
 428
 429 static void prepare_packed_git_one(char *objdir)
 430 {
 431         char path[PATH_MAX];
 432         int len;
 433         DIR *dir;
 434         struct dirent *de;
 435
 436         sprintf(path, "%s/pack", objdir);
 437         len = strlen(path);
 438         dir = opendir(path);
 439         if (!dir)
 440                 return;
 441         path[len++] = '/';
 442         while ((de = readdir(dir)) != NULL) {
 443                 int namelen = strlen(de->d_name);
 444                 struct packed_git *p;
 445
 446                 if (strcmp(de->d_name + namelen - 4, ".idx"))
 447                         continue;
 448
 449                 /* we have .idx.  Is it a file we can map? */
 450                 strcpy(path + len, de->d_name);
 451                 p = add_packed_git(path, len + namelen);
 452                 if (!p)
 453                         continue;
 454                 p->next = packed_git;
 455                 packed_git = p;
 456         }
 457         closedir(dir);
 458 }
 459
 460 void prepare_packed_git(void)
 461 {
 462         int i;
 463         static int run_once = 0;
 464
 465         if (run_once++)
 466                 return;
 467
 468         prepare_packed_git_one(get_object_directory());
 469         prepare_alt_odb();
 470         for (i = 0; alt_odb[i].base != NULL; i++) {
 471                 alt_odb[i].name[0] = 0;
 472                 prepare_packed_git_one(alt_odb[i].base);
 473         }
 474 }
 475
 476 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 477 {
 478         char header[100];
 479         unsigned char real_sha1[20];
 480         SHA_CTX c;
 481
 482         SHA1_Init(&c);
 483         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 484         SHA1_Update(&c, map, size);
 485         SHA1_Final(real_sha1, &c);
 486         return memcmp(sha1, real_sha1, 20) ? -1 : 0;
 487 }
 488
 489 static void *map_sha1_file_internal(const unsigned char *sha1,
 490                                     unsigned long *size)
 491 {
 492         struct stat st;
 493         void *map;
 494         int fd;
 495         char *filename = find_sha1_file(sha1, &st);
 496
 497         if (!filename) {
 498                 return NULL;
 499         }
 500
 501         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 502         if (fd < 0) {
 503                 /* See if it works without O_NOATIME */
 504                 switch (sha1_file_open_flag) {
 505                 default:
 506                         fd = open(filename, O_RDONLY);
 507                         if (fd >= 0)
 508                                 break;
 509                 /* Fallthrough */
 510                 case 0:
 511                         return NULL;
 512                 }
 513
 514                 /* If it failed once, it will probably fail again.
 515                  * Stop using O_NOATIME
 516                  */
 517                 sha1_file_open_flag = 0;
 518         }
 519         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 520         close(fd);
 521         if (-1 == (int)(long)map)
 522                 return NULL;
 523         *size = st.st_size;
 524         return map;
 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 static 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 /* forward declaration for a mutually recursive function */
 622 static int packed_object_info(struct pack_entry *entry,
 623                               char *type, unsigned long *sizep);
 624
 625 static int packed_delta_info(unsigned char *base_sha1,
 626                              unsigned long delta_size,
 627                              unsigned long left,
 628                              char *type,
 629                              unsigned long *sizep,
 630                              struct packed_git *p)
 631 {
 632         struct pack_entry base_ent;
 633
 634         if (left < 20)
 635                 die("truncated pack file");
 636
 637         /* The base entry _must_ be in the same pack */
 638         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 639                 die("failed to find delta-pack base object %s",
 640                     sha1_to_hex(base_sha1));
 641
 642         /* We choose to only get the type of the base object and
 643          * ignore potentially corrupt pack file that expects the delta
 644          * based on a base with a wrong size.  This saves tons of
 645          * inflate() calls.
 646          */
 647
 648         if (packed_object_info(&base_ent, type, NULL))
 649                 die("cannot get info for delta-pack base");
 650
 651         if (sizep) {
 652                 const unsigned char *data;
 653                 unsigned char delta_head[64];
 654                 unsigned long result_size;
 655                 z_stream stream;
 656                 int st;
 657
 658                 memset(&stream, 0, sizeof(stream));
 659
 660                 data = stream.next_in = base_sha1 + 20;
 661                 stream.avail_in = left - 20;
 662                 stream.next_out = delta_head;
 663                 stream.avail_out = sizeof(delta_head);
 664
 665                 inflateInit(&stream);
 666                 st = inflate(&stream, Z_FINISH);
 667                 inflateEnd(&stream);
 668                 if ((st != Z_STREAM_END) &&
 669                     stream.total_out != sizeof(delta_head))
 670                         die("delta data unpack-initial failed");
 671
 672                 /* Examine the initial part of the delta to figure out
 673                  * the result size.
 674                  */
 675                 data = delta_head;
 676                 get_delta_hdr_size(&data); /* ignore base size */
 677
 678                 /* Read the result size */
 679                 result_size = get_delta_hdr_size(&data);
 680                 *sizep = result_size;
 681         }
 682         return 0;
 683 }
 684
 685 static unsigned long unpack_object_header(struct packed_git *p, unsigned long offset,
 686         enum object_type *type, unsigned long *sizep)
 687 {
 688         unsigned shift;
 689         unsigned char *pack, c;
 690         unsigned long size;
 691
 692         if (offset >= p->pack_size)
 693                 die("object offset outside of pack file");
 694
 695         pack =  p->pack_base + offset;
 696         c = *pack++;
 697         offset++;
 698         *type = (c >> 4) & 7;
 699         size = c & 15;
 700         shift = 4;
 701         while (c & 0x80) {
 702                 if (offset >= p->pack_size)
 703                         die("object offset outside of pack file");
 704                 c = *pack++;
 705                 offset++;
 706                 size += (c & 0x7f) << shift;
 707                 shift += 7;
 708         }
 709         *sizep = size;
 710         return offset;
 711 }
 712
 713 void packed_object_info_detail(struct pack_entry *e,
 714                                char *type,
 715                                unsigned long *size,
 716                                unsigned long *store_size,
 717                                int *delta_chain_length,
 718                                unsigned char *base_sha1)
 719 {
 720         struct packed_git *p = e->p;
 721         unsigned long offset, left;
 722         unsigned char *pack;
 723         enum object_type kind;
 724
 725         offset = unpack_object_header(p, e->offset, &kind, size);
 726         pack = p->pack_base + offset;
 727         left = p->pack_size - offset;
 728         if (kind != OBJ_DELTA)
 729                 *delta_chain_length = 0;
 730         else {
 731                 int chain_length = 0;
 732                 memcpy(base_sha1, pack, 20);
 733                 do {
 734                         struct pack_entry base_ent;
 735                         unsigned long junk;
 736
 737                         find_pack_entry_one(pack, &base_ent, p);
 738                         offset = unpack_object_header(p, base_ent.offset,
 739                                                       &kind, &junk);
 740                         pack = p->pack_base + offset;
 741                         chain_length++;
 742                 } while (kind == OBJ_DELTA);
 743                 *delta_chain_length = chain_length;
 744         }
 745         switch (kind) {
 746         case OBJ_COMMIT:
 747                 strcpy(type, "commit");
 748                 break;
 749         case OBJ_TREE:
 750                 strcpy(type, "tree");
 751                 break;
 752         case OBJ_BLOB:
 753                 strcpy(type, "blob");
 754                 break;
 755         case OBJ_TAG:
 756                 strcpy(type, "tag");
 757                 break;
 758         default:
 759                 die("corrupted pack file");
 760         }
 761         *store_size = 0; /* notyet */
 762 }
 763
 764 static int packed_object_info(struct pack_entry *entry,
 765                               char *type, unsigned long *sizep)
 766 {
 767         struct packed_git *p = entry->p;
 768         unsigned long offset, size, left;
 769         unsigned char *pack;
 770         enum object_type kind;
 771         int retval;
 772
 773         if (use_packed_git(p))
 774                 die("cannot map packed file");
 775
 776         offset = unpack_object_header(p, entry->offset, &kind, &size);
 777         pack = p->pack_base + offset;
 778         left = p->pack_size - offset;
 779
 780         switch (kind) {
 781         case OBJ_DELTA:
 782                 retval = packed_delta_info(pack, size, left, type, sizep, p);
 783                 unuse_packed_git(p);
 784                 return retval;
 785         case OBJ_COMMIT:
 786                 strcpy(type, "commit");
 787                 break;
 788         case OBJ_TREE:
 789                 strcpy(type, "tree");
 790                 break;
 791         case OBJ_BLOB:
 792                 strcpy(type, "blob");
 793                 break;
 794         case OBJ_TAG:
 795                 strcpy(type, "tag");
 796                 break;
 797         default:
 798                 die("corrupted pack file");
 799         }
 800         if (sizep)
 801                 *sizep = size;
 802         unuse_packed_git(p);
 803         return 0;
 804 }
 805
 806 /* forward declaration for a mutually recursive function */
 807 static void *unpack_entry(struct pack_entry *, char *, unsigned long *);
 808
 809 static void *unpack_delta_entry(unsigned char *base_sha1,
 810                                 unsigned long delta_size,
 811                                 unsigned long left,
 812                                 char *type,
 813                                 unsigned long *sizep,
 814                                 struct packed_git *p)
 815 {
 816         struct pack_entry base_ent;
 817         void *data, *delta_data, *result, *base;
 818         unsigned long data_size, result_size, base_size;
 819         z_stream stream;
 820         int st;
 821
 822         if (left < 20)
 823                 die("truncated pack file");
 824         data = base_sha1 + 20;
 825         data_size = left - 20;
 826         delta_data = xmalloc(delta_size);
 827
 828         memset(&stream, 0, sizeof(stream));
 829
 830         stream.next_in = data;
 831         stream.avail_in = data_size;
 832         stream.next_out = delta_data;
 833         stream.avail_out = delta_size;
 834
 835         inflateInit(&stream);
 836         st = inflate(&stream, Z_FINISH);
 837         inflateEnd(&stream);
 838         if ((st != Z_STREAM_END) || stream.total_out != delta_size)
 839                 die("delta data unpack failed");
 840
 841         /* The base entry _must_ be in the same pack */
 842         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 843                 die("failed to find delta-pack base object %s",
 844                     sha1_to_hex(base_sha1));
 845         base = unpack_entry_gently(&base_ent, type, &base_size);
 846         if (!base)
 847                 die("failed to read delta-pack base object %s",
 848                     sha1_to_hex(base_sha1));
 849         result = patch_delta(base, base_size,
 850                              delta_data, delta_size,
 851                              &result_size);
 852         if (!result)
 853                 die("failed to apply delta");
 854         free(delta_data);
 855         free(base);
 856         *sizep = result_size;
 857         return result;
 858 }
 859
 860 static void *unpack_non_delta_entry(unsigned char *data,
 861                                     unsigned long size,
 862                                     unsigned long left)
 863 {
 864         int st;
 865         z_stream stream;
 866         unsigned char *buffer;
 867
 868         buffer = xmalloc(size + 1);
 869         buffer[size] = 0;
 870         memset(&stream, 0, sizeof(stream));
 871         stream.next_in = data;
 872         stream.avail_in = left;
 873         stream.next_out = buffer;
 874         stream.avail_out = size;
 875
 876         inflateInit(&stream);
 877         st = inflate(&stream, Z_FINISH);
 878         inflateEnd(&stream);
 879         if ((st != Z_STREAM_END) || stream.total_out != size) {
 880                 free(buffer);
 881                 return NULL;
 882         }
 883
 884         return buffer;
 885 }
 886
 887 static void *unpack_entry(struct pack_entry *entry,
 888                           char *type, unsigned long *sizep)
 889 {
 890         struct packed_git *p = entry->p;
 891         void *retval;
 892
 893         if (use_packed_git(p))
 894                 die("cannot map packed file");
 895         retval = unpack_entry_gently(entry, type, sizep);
 896         unuse_packed_git(p);
 897         if (!retval)
 898                 die("corrupted pack file");
 899         return retval;
 900 }
 901
 902 /* The caller is responsible for use_packed_git()/unuse_packed_git() pair */
 903 void *unpack_entry_gently(struct pack_entry *entry,
 904                           char *type, unsigned long *sizep)
 905 {
 906         struct packed_git *p = entry->p;
 907         unsigned long offset, size, left;
 908         unsigned char *pack;
 909         enum object_type kind;
 910         void *retval;
 911
 912         offset = unpack_object_header(p, entry->offset, &kind, &size);
 913         pack = p->pack_base + offset;
 914         left = p->pack_size - offset;
 915         switch (kind) {
 916         case OBJ_DELTA:
 917                 retval = unpack_delta_entry(pack, size, left, type, sizep, p);
 918                 return retval;
 919         case OBJ_COMMIT:
 920                 strcpy(type, "commit");
 921                 break;
 922         case OBJ_TREE:
 923                 strcpy(type, "tree");
 924                 break;
 925         case OBJ_BLOB:
 926                 strcpy(type, "blob");
 927                 break;
 928         case OBJ_TAG:
 929                 strcpy(type, "tag");
 930                 break;
 931         default:
 932                 return NULL;
 933         }
 934         *sizep = size;
 935         retval = unpack_non_delta_entry(pack, size, left);
 936         return retval;
 937 }
 938
 939 int num_packed_objects(const struct packed_git *p)
 940 {
 941         /* See check_packed_git_idx() */
 942         return (p->index_size - 20 - 20 - 4*256) / 24;
 943 }
 944
 945 int nth_packed_object_sha1(const struct packed_git *p, int n,
 946                            unsigned char* sha1)
 947 {
 948         void *index = p->index_base + 256;
 949         if (n < 0 || num_packed_objects(p) <= n)
 950                 return -1;
 951         memcpy(sha1, (index + 24 * n + 4), 20);
 952         return 0;
 953 }
 954
 955 int find_pack_entry_one(const unsigned char *sha1,
 956                         struct pack_entry *e, struct packed_git *p)
 957 {
 958         unsigned int *level1_ofs = p->index_base;
 959         int hi = ntohl(level1_ofs[*sha1]);
 960         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
 961         void *index = p->index_base + 256;
 962
 963         do {
 964                 int mi = (lo + hi) / 2;
 965                 int cmp = memcmp(index + 24 * mi + 4, sha1, 20);
 966                 if (!cmp) {
 967                         e->offset = ntohl(*((int*)(index + 24 * mi)));
 968                         memcpy(e->sha1, sha1, 20);
 969                         e->p = p;
 970                         return 1;
 971                 }
 972                 if (cmp > 0)
 973                         hi = mi;
 974                 else
 975                         lo = mi+1;
 976         } while (lo < hi);
 977         return 0;
 978 }
 979
 980 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e)
 981 {
 982         struct packed_git *p;
 983         prepare_packed_git();
 984
 985         for (p = packed_git; p; p = p->next) {
 986                 if (find_pack_entry_one(sha1, e, p))
 987                         return 1;
 988         }
 989         return 0;
 990 }
 991
 992 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
 993 {
 994         int status;
 995         unsigned long mapsize, size;
 996         void *map;
 997         z_stream stream;
 998         char hdr[128];
 999
1000         map = map_sha1_file_internal(sha1, &mapsize);
1001         if (!map) {
1002                 struct pack_entry e;
1003
1004                 if (!find_pack_entry(sha1, &e))
1005                         return error("unable to find %s", sha1_to_hex(sha1));
1006                 return packed_object_info(&e, type, sizep);
1007         }
1008         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
1009                 status = error("unable to unpack %s header",
1010                                sha1_to_hex(sha1));
1011         if (parse_sha1_header(hdr, type, &size) < 0)
1012                 status = error("unable to parse %s header", sha1_to_hex(sha1));
1013         else {
1014                 status = 0;
1015                 if (sizep)
1016                         *sizep = size;
1017         }
1018         inflateEnd(&stream);
1019         munmap(map, mapsize);
1020         return status;
1021 }
1022
1023 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
1024 {
1025         struct pack_entry e;
1026
1027         if (!find_pack_entry(sha1, &e)) {
1028                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
1029                 return NULL;
1030         }
1031         return unpack_entry(&e, type, size);
1032 }
1033
1034 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
1035 {
1036         unsigned long mapsize;
1037         void *map, *buf;
1038         struct pack_entry e;
1039
1040         if (find_pack_entry(sha1, &e))
1041                 return read_packed_sha1(sha1, type, size);
1042         map = map_sha1_file_internal(sha1, &mapsize);
1043         if (map) {
1044                 buf = unpack_sha1_file(map, mapsize, type, size);
1045                 munmap(map, mapsize);
1046                 return buf;
1047         }
1048         return NULL;
1049 }
1050
1051 void *read_object_with_reference(const unsigned char *sha1,
1052                                  const char *required_type,
1053                                  unsigned long *size,
1054                                  unsigned char *actual_sha1_return)
1055 {
1056         char type[20];
1057         void *buffer;
1058         unsigned long isize;
1059         unsigned char actual_sha1[20];
1060
1061         memcpy(actual_sha1, sha1, 20);
1062         while (1) {
1063                 int ref_length = -1;
1064                 const char *ref_type = NULL;
1065
1066                 buffer = read_sha1_file(actual_sha1, type, &isize);
1067                 if (!buffer)
1068                         return NULL;
1069                 if (!strcmp(type, required_type)) {
1070                         *size = isize;
1071                         if (actual_sha1_return)
1072                                 memcpy(actual_sha1_return, actual_sha1, 20);
1073                         return buffer;
1074                 }
1075                 /* Handle references */
1076                 else if (!strcmp(type, "commit"))
1077                         ref_type = "tree ";
1078                 else if (!strcmp(type, "tag"))
1079                         ref_type = "object ";
1080                 else {
1081                         free(buffer);
1082                         return NULL;
1083                 }
1084                 ref_length = strlen(ref_type);
1085
1086                 if (memcmp(buffer, ref_type, ref_length) ||
1087                     get_sha1_hex(buffer + ref_length, actual_sha1)) {
1088                         free(buffer);
1089                         return NULL;
1090                 }
1091                 /* Now we have the ID of the referred-to object in
1092                  * actual_sha1.  Check again. */
1093         }
1094 }
1095
1096 char *write_sha1_file_prepare(void *buf,
1097                               unsigned long len,
1098                               const char *type,
1099                               unsigned char *sha1,
1100                               unsigned char *hdr,
1101                               int *hdrlen)
1102 {
1103         SHA_CTX c;
1104
1105         /* Generate the header */
1106         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
1107
1108         /* Sha1.. */
1109         SHA1_Init(&c);
1110         SHA1_Update(&c, hdr, *hdrlen);
1111         SHA1_Update(&c, buf, len);
1112         SHA1_Final(sha1, &c);
1113
1114         return sha1_file_name(sha1);
1115 }
1116
1117 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1118 {
1119         int size;
1120         unsigned char *compressed;
1121         z_stream stream;
1122         unsigned char sha1[20];
1123         char *filename;
1124         static char tmpfile[PATH_MAX];
1125         unsigned char hdr[50];
1126         int fd, hdrlen, ret;
1127
1128         /* Normally if we have it in the pack then we do not bother writing
1129          * it out into .git/objects/??/?{38} file.
1130          */
1131         filename = write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1132         if (returnsha1)
1133                 memcpy(returnsha1, sha1, 20);
1134         if (has_sha1_file(sha1))
1135                 return 0;
1136         fd = open(filename, O_RDONLY);
1137         if (fd >= 0) {
1138                 /*
1139                  * FIXME!!! We might do collision checking here, but we'd
1140                  * need to uncompress the old file and check it. Later.
1141                  */
1142                 close(fd);
1143                 return 0;
1144         }
1145
1146         if (errno != ENOENT) {
1147                 fprintf(stderr, "sha1 file %s: %s", filename, strerror(errno));
1148                 return -1;
1149         }
1150
1151         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1152
1153         fd = mkstemp(tmpfile);
1154         if (fd < 0) {
1155                 fprintf(stderr, "unable to create temporary sha1 filename %s: %s", tmpfile, strerror(errno));
1156                 return -1;
1157         }
1158
1159         /* Set it up */
1160         memset(&stream, 0, sizeof(stream));
1161         deflateInit(&stream, Z_BEST_COMPRESSION);
1162         size = deflateBound(&stream, len+hdrlen);
1163         compressed = xmalloc(size);
1164
1165         /* Compress it */
1166         stream.next_out = compressed;
1167         stream.avail_out = size;
1168
1169         /* First header.. */
1170         stream.next_in = hdr;
1171         stream.avail_in = hdrlen;
1172         while (deflate(&stream, 0) == Z_OK)
1173                 /* nothing */;
1174
1175         /* Then the data itself.. */
1176         stream.next_in = buf;
1177         stream.avail_in = len;
1178         while (deflate(&stream, Z_FINISH) == Z_OK)
1179                 /* nothing */;
1180         deflateEnd(&stream);
1181         size = stream.total_out;
1182
1183         if (write(fd, compressed, size) != size)
1184                 die("unable to write file");
1185         fchmod(fd, 0444);
1186         close(fd);
1187         free(compressed);
1188
1189         ret = link(tmpfile, filename);
1190         if (ret < 0) {
1191                 ret = errno;
1192
1193                 /*
1194                  * Coda hack - coda doesn't like cross-directory links,
1195                  * so we fall back to a rename, which will mean that it
1196                  * won't be able to check collisions, but that's not a
1197                  * big deal.
1198                  *
1199                  * When this succeeds, we just return 0. We have nothing
1200                  * left to unlink.
1201                  */
1202                 if (ret == EXDEV && !rename(tmpfile, filename))
1203                         return 0;
1204         }
1205         unlink(tmpfile);
1206         if (ret) {
1207                 if (ret != EEXIST) {
1208                         fprintf(stderr, "unable to write sha1 filename %s: %s", filename, strerror(ret));
1209                         return -1;
1210                 }
1211                 /* FIXME!!! Collision check here ? */
1212         }
1213
1214         return 0;
1215 }
1216
1217 int write_sha1_to_fd(int fd, const unsigned char *sha1)
1218 {
1219         ssize_t size;
1220         unsigned long objsize;
1221         int posn = 0;
1222         void *buf = map_sha1_file_internal(sha1, &objsize);
1223         z_stream stream;
1224         if (!buf) {
1225                 unsigned char *unpacked;
1226                 unsigned long len;
1227                 char type[20];
1228                 char hdr[50];
1229                 int hdrlen;
1230                 // need to unpack and recompress it by itself
1231                 unpacked = read_packed_sha1(sha1, type, &len);
1232
1233                 hdrlen = sprintf(hdr, "%s %lu", type, len) + 1;
1234
1235                 /* Set it up */
1236                 memset(&stream, 0, sizeof(stream));
1237                 deflateInit(&stream, Z_BEST_COMPRESSION);
1238                 size = deflateBound(&stream, len + hdrlen);
1239                 buf = xmalloc(size);
1240
1241                 /* Compress it */
1242                 stream.next_out = buf;
1243                 stream.avail_out = size;
1244
1245                 /* First header.. */
1246                 stream.next_in = (void *)hdr;
1247                 stream.avail_in = hdrlen;
1248                 while (deflate(&stream, 0) == Z_OK)
1249                         /* nothing */;
1250
1251                 /* Then the data itself.. */
1252                 stream.next_in = unpacked;
1253                 stream.avail_in = len;
1254                 while (deflate(&stream, Z_FINISH) == Z_OK)
1255                         /* nothing */;
1256                 deflateEnd(&stream);
1257
1258                 objsize = stream.total_out;
1259         }
1260
1261         do {
1262                 size = write(fd, buf + posn, objsize - posn);
1263                 if (size <= 0) {
1264                         if (!size) {
1265                                 fprintf(stderr, "write closed");
1266                         } else {
1267                                 perror("write ");
1268                         }
1269                         return -1;
1270                 }
1271                 posn += size;
1272         } while (posn < objsize);
1273         return 0;
1274 }
1275
1276 int write_sha1_from_fd(const unsigned char *sha1, int fd)
1277 {
1278         char *filename = sha1_file_name(sha1);
1279
1280         int local;
1281         z_stream stream;
1282         unsigned char real_sha1[20];
1283         unsigned char buf[4096];
1284         unsigned char discard[4096];
1285         int ret;
1286         SHA_CTX c;
1287
1288         local = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
1289
1290         if (local < 0)
1291                 return error("Couldn't open %s\n", filename);
1292
1293         memset(&stream, 0, sizeof(stream));
1294
1295         inflateInit(&stream);
1296
1297         SHA1_Init(&c);
1298
1299         do {
1300                 ssize_t size;
1301                 size = read(fd, buf, 4096);
1302                 if (size <= 0) {
1303                         close(local);
1304                         unlink(filename);
1305                         if (!size)
1306                                 return error("Connection closed?");
1307                         perror("Reading from connection");
1308                         return -1;
1309                 }
1310                 write(local, buf, size);
1311                 stream.avail_in = size;
1312                 stream.next_in = buf;
1313                 do {
1314                         stream.next_out = discard;
1315                         stream.avail_out = sizeof(discard);
1316                         ret = inflate(&stream, Z_SYNC_FLUSH);
1317                         SHA1_Update(&c, discard, sizeof(discard) -
1318                                     stream.avail_out);
1319                 } while (stream.avail_in && ret == Z_OK);
1320
1321         } while (ret == Z_OK);
1322         inflateEnd(&stream);
1323
1324         close(local);
1325         SHA1_Final(real_sha1, &c);
1326         if (ret != Z_STREAM_END) {
1327                 unlink(filename);
1328                 return error("File %s corrupted", sha1_to_hex(sha1));
1329         }
1330         if (memcmp(sha1, real_sha1, 20)) {
1331                 unlink(filename);
1332                 return error("File %s has bad hash\n", sha1_to_hex(sha1));
1333         }
1334
1335         return 0;
1336 }
1337
1338 int has_sha1_pack(const unsigned char *sha1)
1339 {
1340         struct pack_entry e;
1341         return find_pack_entry(sha1, &e);
1342 }
1343
1344 int has_sha1_file(const unsigned char *sha1)
1345 {
1346         struct stat st;
1347         struct pack_entry e;
1348
1349         if (find_pack_entry(sha1, &e))
1350                 return 1;
1351         return find_sha1_file(sha1, &st) ? 1 : 0;
1352 }
1353
1354 int index_fd(unsigned char *sha1, int fd, struct stat *st, int write_object, const char *type)
1355 {
1356         unsigned long size = st->st_size;
1357         void *buf;
1358         int ret;
1359         unsigned char hdr[50];
1360         int hdrlen;
1361
1362         buf = "";
1363         if (size)
1364                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1365         close(fd);
1366         if ((int)(long)buf == -1)
1367                 return -1;
1368
1369         if (!type)
1370                 type = "blob";
1371         if (write_object)
1372                 ret = write_sha1_file(buf, size, type, sha1);
1373         else {
1374                 write_sha1_file_prepare(buf, size, type, sha1, hdr, &hdrlen);
1375                 ret = 0;
1376         }
1377         if (size)
1378                 munmap(buf, size);
1379         return ret;
1380 }