git.oblomov.eu Git - git/blob - 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 "cache.h"
  10 #include "config.h"
  11 #include "string-list.h"
  12 #include "lockfile.h"
  13 #include "delta.h"
  14 #include "pack.h"
  15 #include "blob.h"
  16 #include "commit.h"
  17 #include "run-command.h"
  18 #include "tag.h"
  19 #include "tree.h"
  20 #include "tree-walk.h"
  21 #include "refs.h"
  22 #include "pack-revindex.h"
  23 #include "sha1-lookup.h"
  24 #include "bulk-checkin.h"
  25 #include "repository.h"
  26 #include "streaming.h"
  27 #include "dir.h"
  28 #include "list.h"
  29 #include "mergesort.h"
  30 #include "quote.h"
  31 #include "packfile.h"
  32 #include "fetch-object.h"
  33 #include "object-store.h"
  34
  35 const unsigned char null_sha1[GIT_MAX_RAWSZ];
  36 const struct object_id null_oid;
  37 const struct object_id empty_tree_oid = {
  38         EMPTY_TREE_SHA1_BIN_LITERAL
  39 };
  40 const struct object_id empty_blob_oid = {
  41         EMPTY_BLOB_SHA1_BIN_LITERAL
  42 };
  43
  44 static void git_hash_sha1_init(void *ctx)
  45 {
  46         git_SHA1_Init((git_SHA_CTX *)ctx);
  47 }
  48
  49 static void git_hash_sha1_update(void *ctx, const void *data, size_t len)
  50 {
  51         git_SHA1_Update((git_SHA_CTX *)ctx, data, len);
  52 }
  53
  54 static void git_hash_sha1_final(unsigned char *hash, void *ctx)
  55 {
  56         git_SHA1_Final(hash, (git_SHA_CTX *)ctx);
  57 }
  58
  59 static void git_hash_unknown_init(void *ctx)
  60 {
  61         die("trying to init unknown hash");
  62 }
  63
  64 static void git_hash_unknown_update(void *ctx, const void *data, size_t len)
  65 {
  66         die("trying to update unknown hash");
  67 }
  68
  69 static void git_hash_unknown_final(unsigned char *hash, void *ctx)
  70 {
  71         die("trying to finalize unknown hash");
  72 }
  73
  74 const struct git_hash_algo hash_algos[GIT_HASH_NALGOS] = {
  75         {
  76                 NULL,
  77                 0x00000000,
  78                 0,
  79                 0,
  80                 0,
  81                 git_hash_unknown_init,
  82                 git_hash_unknown_update,
  83                 git_hash_unknown_final,
  84                 NULL,
  85                 NULL,
  86         },
  87         {
  88                 "sha-1",
  89                 /* "sha1", big-endian */
  90                 0x73686131,
  91                 sizeof(git_SHA_CTX),
  92                 GIT_SHA1_RAWSZ,
  93                 GIT_SHA1_HEXSZ,
  94                 git_hash_sha1_init,
  95                 git_hash_sha1_update,
  96                 git_hash_sha1_final,
  97                 &empty_tree_oid,
  98                 &empty_blob_oid,
  99         },
 100 };
 101
 102 /*
 103  * This is meant to hold a *small* number of objects that you would
 104  * want read_sha1_file() to be able to return, but yet you do not want
 105  * to write them into the object store (e.g. a browse-only
 106  * application).
 107  */
 108 static struct cached_object {
 109         unsigned char sha1[20];
 110         enum object_type type;
 111         void *buf;
 112         unsigned long size;
 113 } *cached_objects;
 114 static int cached_object_nr, cached_object_alloc;
 115
 116 static struct cached_object empty_tree = {
 117         EMPTY_TREE_SHA1_BIN_LITERAL,
 118         OBJ_TREE,
 119         "",
 120         0
 121 };
 122
 123 static struct cached_object *find_cached_object(const unsigned char *sha1)
 124 {
 125         int i;
 126         struct cached_object *co = cached_objects;
 127
 128         for (i = 0; i < cached_object_nr; i++, co++) {
 129                 if (!hashcmp(co->sha1, sha1))
 130                         return co;
 131         }
 132         if (!hashcmp(sha1, empty_tree.sha1))
 133                 return &empty_tree;
 134         return NULL;
 135 }
 136
 137
 138 static int get_conv_flags(unsigned flags)
 139 {
 140         if (flags & HASH_RENORMALIZE)
 141                 return CONV_EOL_RENORMALIZE;
 142         else if (flags & HASH_WRITE_OBJECT)
 143           return global_conv_flags_eol;
 144         else
 145                 return 0;
 146 }
 147
 148
 149 int mkdir_in_gitdir(const char *path)
 150 {
 151         if (mkdir(path, 0777)) {
 152                 int saved_errno = errno;
 153                 struct stat st;
 154                 struct strbuf sb = STRBUF_INIT;
 155
 156                 if (errno != EEXIST)
 157                         return -1;
 158                 /*
 159                  * Are we looking at a path in a symlinked worktree
 160                  * whose original repository does not yet have it?
 161                  * e.g. .git/rr-cache pointing at its original
 162                  * repository in which the user hasn't performed any
 163                  * conflict resolution yet?
 164                  */
 165                 if (lstat(path, &st) || !S_ISLNK(st.st_mode) ||
 166                     strbuf_readlink(&sb, path, st.st_size) ||
 167                     !is_absolute_path(sb.buf) ||
 168                     mkdir(sb.buf, 0777)) {
 169                         strbuf_release(&sb);
 170                         errno = saved_errno;
 171                         return -1;
 172                 }
 173                 strbuf_release(&sb);
 174         }
 175         return adjust_shared_perm(path);
 176 }
 177
 178 enum scld_error safe_create_leading_directories(char *path)
 179 {
 180         char *next_component = path + offset_1st_component(path);
 181         enum scld_error ret = SCLD_OK;
 182
 183         while (ret == SCLD_OK && next_component) {
 184                 struct stat st;
 185                 char *slash = next_component, slash_character;
 186
 187                 while (*slash && !is_dir_sep(*slash))
 188                         slash++;
 189
 190                 if (!*slash)
 191                         break;
 192
 193                 next_component = slash + 1;
 194                 while (is_dir_sep(*next_component))
 195                         next_component++;
 196                 if (!*next_component)
 197                         break;
 198
 199                 slash_character = *slash;
 200                 *slash = '\0';
 201                 if (!stat(path, &st)) {
 202                         /* path exists */
 203                         if (!S_ISDIR(st.st_mode)) {
 204                                 errno = ENOTDIR;
 205                                 ret = SCLD_EXISTS;
 206                         }
 207                 } else if (mkdir(path, 0777)) {
 208                         if (errno == EEXIST &&
 209                             !stat(path, &st) && S_ISDIR(st.st_mode))
 210                                 ; /* somebody created it since we checked */
 211                         else if (errno == ENOENT)
 212                                 /*
 213                                  * Either mkdir() failed because
 214                                  * somebody just pruned the containing
 215                                  * directory, or stat() failed because
 216                                  * the file that was in our way was
 217                                  * just removed.  Either way, inform
 218                                  * the caller that it might be worth
 219                                  * trying again:
 220                                  */
 221                                 ret = SCLD_VANISHED;
 222                         else
 223                                 ret = SCLD_FAILED;
 224                 } else if (adjust_shared_perm(path)) {
 225                         ret = SCLD_PERMS;
 226                 }
 227                 *slash = slash_character;
 228         }
 229         return ret;
 230 }
 231
 232 enum scld_error safe_create_leading_directories_const(const char *path)
 233 {
 234         int save_errno;
 235         /* path points to cache entries, so xstrdup before messing with it */
 236         char *buf = xstrdup(path);
 237         enum scld_error result = safe_create_leading_directories(buf);
 238
 239         save_errno = errno;
 240         free(buf);
 241         errno = save_errno;
 242         return result;
 243 }
 244
 245 int raceproof_create_file(const char *path, create_file_fn fn, void *cb)
 246 {
 247         /*
 248          * The number of times we will try to remove empty directories
 249          * in the way of path. This is only 1 because if another
 250          * process is racily creating directories that conflict with
 251          * us, we don't want to fight against them.
 252          */
 253         int remove_directories_remaining = 1;
 254
 255         /*
 256          * The number of times that we will try to create the
 257          * directories containing path. We are willing to attempt this
 258          * more than once, because another process could be trying to
 259          * clean up empty directories at the same time as we are
 260          * trying to create them.
 261          */
 262         int create_directories_remaining = 3;
 263
 264         /* A scratch copy of path, filled lazily if we need it: */
 265         struct strbuf path_copy = STRBUF_INIT;
 266
 267         int ret, save_errno;
 268
 269         /* Sanity check: */
 270         assert(*path);
 271
 272 retry_fn:
 273         ret = fn(path, cb);
 274         save_errno = errno;
 275         if (!ret)
 276                 goto out;
 277
 278         if (errno == EISDIR && remove_directories_remaining-- > 0) {
 279                 /*
 280                  * A directory is in the way. Maybe it is empty; try
 281                  * to remove it:
 282                  */
 283                 if (!path_copy.len)
 284                         strbuf_addstr(&path_copy, path);
 285
 286                 if (!remove_dir_recursively(&path_copy, REMOVE_DIR_EMPTY_ONLY))
 287                         goto retry_fn;
 288         } else if (errno == ENOENT && create_directories_remaining-- > 0) {
 289                 /*
 290                  * Maybe the containing directory didn't exist, or
 291                  * maybe it was just deleted by a process that is
 292                  * racing with us to clean up empty directories. Try
 293                  * to create it:
 294                  */
 295                 enum scld_error scld_result;
 296
 297                 if (!path_copy.len)
 298                         strbuf_addstr(&path_copy, path);
 299
 300                 do {
 301                         scld_result = safe_create_leading_directories(path_copy.buf);
 302                         if (scld_result == SCLD_OK)
 303                                 goto retry_fn;
 304                 } while (scld_result == SCLD_VANISHED && create_directories_remaining-- > 0);
 305         }
 306
 307 out:
 308         strbuf_release(&path_copy);
 309         errno = save_errno;
 310         return ret;
 311 }
 312
 313 static void fill_sha1_path(struct strbuf *buf, const unsigned char *sha1)
 314 {
 315         int i;
 316         for (i = 0; i < 20; i++) {
 317                 static char hex[] = "0123456789abcdef";
 318                 unsigned int val = sha1[i];
 319                 strbuf_addch(buf, hex[val >> 4]);
 320                 strbuf_addch(buf, hex[val & 0xf]);
 321                 if (!i)
 322                         strbuf_addch(buf, '/');
 323         }
 324 }
 325
 326 void sha1_file_name_the_repository(struct strbuf *buf, const unsigned char *sha1)
 327 {
 328         strbuf_addstr(buf, get_object_directory());
 329         strbuf_addch(buf, '/');
 330         fill_sha1_path(buf, sha1);
 331 }
 332
 333 struct strbuf *alt_scratch_buf(struct alternate_object_database *alt)
 334 {
 335         strbuf_setlen(&alt->scratch, alt->base_len);
 336         return &alt->scratch;
 337 }
 338
 339 static const char *alt_sha1_path(struct alternate_object_database *alt,
 340                                  const unsigned char *sha1)
 341 {
 342         struct strbuf *buf = alt_scratch_buf(alt);
 343         fill_sha1_path(buf, sha1);
 344         return buf->buf;
 345 }
 346
 347 /*
 348  * Return non-zero iff the path is usable as an alternate object database.
 349  */
 350 static int alt_odb_usable(struct raw_object_store *o,
 351                           struct strbuf *path,
 352                           const char *normalized_objdir)
 353 {
 354         struct alternate_object_database *alt;
 355
 356         /* Detect cases where alternate disappeared */
 357         if (!is_directory(path->buf)) {
 358                 error("object directory %s does not exist; "
 359                       "check .git/objects/info/alternates.",
 360                       path->buf);
 361                 return 0;
 362         }
 363
 364         /*
 365          * Prevent the common mistake of listing the same
 366          * thing twice, or object directory itself.
 367          */
 368         for (alt = o->alt_odb_list; alt; alt = alt->next) {
 369                 if (!fspathcmp(path->buf, alt->path))
 370                         return 0;
 371         }
 372         if (!fspathcmp(path->buf, normalized_objdir))
 373                 return 0;
 374
 375         return 1;
 376 }
 377
 378 /*
 379  * Prepare alternate object database registry.
 380  *
 381  * The variable alt_odb_list points at the list of struct
 382  * alternate_object_database.  The elements on this list come from
 383  * non-empty elements from colon separated ALTERNATE_DB_ENVIRONMENT
 384  * environment variable, and $GIT_OBJECT_DIRECTORY/info/alternates,
 385  * whose contents is similar to that environment variable but can be
 386  * LF separated.  Its base points at a statically allocated buffer that
 387  * contains "/the/directory/corresponding/to/.git/objects/...", while
 388  * its name points just after the slash at the end of ".git/objects/"
 389  * in the example above, and has enough space to hold 40-byte hex
 390  * SHA1, an extra slash for the first level indirection, and the
 391  * terminating NUL.
 392  */
 393 static void read_info_alternates(struct repository *r,
 394                                  const char *relative_base,
 395                                  int depth);
 396 static int link_alt_odb_entry(struct repository *r, const char *entry,
 397         const char *relative_base, int depth, const char *normalized_objdir)
 398 {
 399         struct alternate_object_database *ent;
 400         struct strbuf pathbuf = STRBUF_INIT;
 401
 402         if (!is_absolute_path(entry) && relative_base) {
 403                 strbuf_realpath(&pathbuf, relative_base, 1);
 404                 strbuf_addch(&pathbuf, '/');
 405         }
 406         strbuf_addstr(&pathbuf, entry);
 407
 408         if (strbuf_normalize_path(&pathbuf) < 0 && relative_base) {
 409                 error("unable to normalize alternate object path: %s",
 410                       pathbuf.buf);
 411                 strbuf_release(&pathbuf);
 412                 return -1;
 413         }
 414
 415         /*
 416          * The trailing slash after the directory name is given by
 417          * this function at the end. Remove duplicates.
 418          */
 419         while (pathbuf.len && pathbuf.buf[pathbuf.len - 1] == '/')
 420                 strbuf_setlen(&pathbuf, pathbuf.len - 1);
 421
 422         if (!alt_odb_usable(r->objects, &pathbuf, normalized_objdir)) {
 423                 strbuf_release(&pathbuf);
 424                 return -1;
 425         }
 426
 427         ent = alloc_alt_odb(pathbuf.buf);
 428
 429         /* add the alternate entry */
 430         *r->objects->alt_odb_tail = ent;
 431         r->objects->alt_odb_tail = &(ent->next);
 432         ent->next = NULL;
 433
 434         /* recursively add alternates */
 435         read_info_alternates(r, pathbuf.buf, depth + 1);
 436
 437         strbuf_release(&pathbuf);
 438         return 0;
 439 }
 440
 441 static const char *parse_alt_odb_entry(const char *string,
 442                                        int sep,
 443                                        struct strbuf *out)
 444 {
 445         const char *end;
 446
 447         strbuf_reset(out);
 448
 449         if (*string == '#') {
 450                 /* comment; consume up to next separator */
 451                 end = strchrnul(string, sep);
 452         } else if (*string == '"' && !unquote_c_style(out, string, &end)) {
 453                 /*
 454                  * quoted path; unquote_c_style has copied the
 455                  * data for us and set "end". Broken quoting (e.g.,
 456                  * an entry that doesn't end with a quote) falls
 457                  * back to the unquoted case below.
 458                  */
 459         } else {
 460                 /* normal, unquoted path */
 461                 end = strchrnul(string, sep);
 462                 strbuf_add(out, string, end - string);
 463         }
 464
 465         if (*end)
 466                 end++;
 467         return end;
 468 }
 469
 470 static void link_alt_odb_entries(struct repository *r, const char *alt,
 471                                  int sep, const char *relative_base, int depth)
 472 {
 473         struct strbuf objdirbuf = STRBUF_INIT;
 474         struct strbuf entry = STRBUF_INIT;
 475
 476         if (!alt || !*alt)
 477                 return;
 478
 479         if (depth > 5) {
 480                 error("%s: ignoring alternate object stores, nesting too deep.",
 481                                 relative_base);
 482                 return;
 483         }
 484
 485         strbuf_add_absolute_path(&objdirbuf, r->objects->objectdir);
 486         if (strbuf_normalize_path(&objdirbuf) < 0)
 487                 die("unable to normalize object directory: %s",
 488                     objdirbuf.buf);
 489
 490         while (*alt) {
 491                 alt = parse_alt_odb_entry(alt, sep, &entry);
 492                 if (!entry.len)
 493                         continue;
 494                 link_alt_odb_entry(r, entry.buf,
 495                                    relative_base, depth, objdirbuf.buf);
 496         }
 497         strbuf_release(&entry);
 498         strbuf_release(&objdirbuf);
 499 }
 500
 501 static void read_info_alternates(struct repository *r,
 502                                  const char *relative_base,
 503                                  int depth)
 504 {
 505         char *path;
 506         struct strbuf buf = STRBUF_INIT;
 507
 508         path = xstrfmt("%s/info/alternates", relative_base);
 509         if (strbuf_read_file(&buf, path, 1024) < 0) {
 510                 warn_on_fopen_errors(path);
 511                 free(path);
 512                 return;
 513         }
 514
 515         link_alt_odb_entries(r, buf.buf, '\n', relative_base, depth);
 516         strbuf_release(&buf);
 517         free(path);
 518 }
 519
 520 struct alternate_object_database *alloc_alt_odb(const char *dir)
 521 {
 522         struct alternate_object_database *ent;
 523
 524         FLEX_ALLOC_STR(ent, path, dir);
 525         strbuf_init(&ent->scratch, 0);
 526         strbuf_addf(&ent->scratch, "%s/", dir);
 527         ent->base_len = ent->scratch.len;
 528
 529         return ent;
 530 }
 531
 532 void add_to_alternates_file(const char *reference)
 533 {
 534         struct lock_file lock = LOCK_INIT;
 535         char *alts = git_pathdup("objects/info/alternates");
 536         FILE *in, *out;
 537         int found = 0;
 538
 539         hold_lock_file_for_update(&lock, alts, LOCK_DIE_ON_ERROR);
 540         out = fdopen_lock_file(&lock, "w");
 541         if (!out)
 542                 die_errno("unable to fdopen alternates lockfile");
 543
 544         in = fopen(alts, "r");
 545         if (in) {
 546                 struct strbuf line = STRBUF_INIT;
 547
 548                 while (strbuf_getline(&line, in) != EOF) {
 549                         if (!strcmp(reference, line.buf)) {
 550                                 found = 1;
 551                                 break;
 552                         }
 553                         fprintf_or_die(out, "%s\n", line.buf);
 554                 }
 555
 556                 strbuf_release(&line);
 557                 fclose(in);
 558         }
 559         else if (errno != ENOENT)
 560                 die_errno("unable to read alternates file");
 561
 562         if (found) {
 563                 rollback_lock_file(&lock);
 564         } else {
 565                 fprintf_or_die(out, "%s\n", reference);
 566                 if (commit_lock_file(&lock))
 567                         die_errno("unable to move new alternates file into place");
 568                 if (the_repository->objects->alt_odb_tail)
 569                         link_alt_odb_entries(the_repository, reference,
 570                                              '\n', NULL, 0);
 571         }
 572         free(alts);
 573 }
 574
 575 void add_to_alternates_memory(const char *reference)
 576 {
 577         /*
 578          * Make sure alternates are initialized, or else our entry may be
 579          * overwritten when they are.
 580          */
 581         prepare_alt_odb(the_repository);
 582
 583         link_alt_odb_entries(the_repository, reference,
 584                              '\n', NULL, 0);
 585 }
 586
 587 /*
 588  * Compute the exact path an alternate is at and returns it. In case of
 589  * error NULL is returned and the human readable error is added to `err`
 590  * `path` may be relative and should point to $GITDIR.
 591  * `err` must not be null.
 592  */
 593 char *compute_alternate_path(const char *path, struct strbuf *err)
 594 {
 595         char *ref_git = NULL;
 596         const char *repo, *ref_git_s;
 597         int seen_error = 0;
 598
 599         ref_git_s = real_path_if_valid(path);
 600         if (!ref_git_s) {
 601                 seen_error = 1;
 602                 strbuf_addf(err, _("path '%s' does not exist"), path);
 603                 goto out;
 604         } else
 605                 /*
 606                  * Beware: read_gitfile(), real_path() and mkpath()
 607                  * return static buffer
 608                  */
 609                 ref_git = xstrdup(ref_git_s);
 610
 611         repo = read_gitfile(ref_git);
 612         if (!repo)
 613                 repo = read_gitfile(mkpath("%s/.git", ref_git));
 614         if (repo) {
 615                 free(ref_git);
 616                 ref_git = xstrdup(repo);
 617         }
 618
 619         if (!repo && is_directory(mkpath("%s/.git/objects", ref_git))) {
 620                 char *ref_git_git = mkpathdup("%s/.git", ref_git);
 621                 free(ref_git);
 622                 ref_git = ref_git_git;
 623         } else if (!is_directory(mkpath("%s/objects", ref_git))) {
 624                 struct strbuf sb = STRBUF_INIT;
 625                 seen_error = 1;
 626                 if (get_common_dir(&sb, ref_git)) {
 627                         strbuf_addf(err,
 628                                     _("reference repository '%s' as a linked "
 629                                       "checkout is not supported yet."),
 630                                     path);
 631                         goto out;
 632                 }
 633
 634                 strbuf_addf(err, _("reference repository '%s' is not a "
 635                                         "local repository."), path);
 636                 goto out;
 637         }
 638
 639         if (!access(mkpath("%s/shallow", ref_git), F_OK)) {
 640                 strbuf_addf(err, _("reference repository '%s' is shallow"),
 641                             path);
 642                 seen_error = 1;
 643                 goto out;
 644         }
 645
 646         if (!access(mkpath("%s/info/grafts", ref_git), F_OK)) {
 647                 strbuf_addf(err,
 648                             _("reference repository '%s' is grafted"),
 649                             path);
 650                 seen_error = 1;
 651                 goto out;
 652         }
 653
 654 out:
 655         if (seen_error) {
 656                 FREE_AND_NULL(ref_git);
 657         }
 658
 659         return ref_git;
 660 }
 661
 662 int foreach_alt_odb(alt_odb_fn fn, void *cb)
 663 {
 664         struct alternate_object_database *ent;
 665         int r = 0;
 666
 667         prepare_alt_odb(the_repository);
 668         for (ent = the_repository->objects->alt_odb_list; ent; ent = ent->next) {
 669                 r = fn(ent, cb);
 670                 if (r)
 671                         break;
 672         }
 673         return r;
 674 }
 675
 676 void prepare_alt_odb(struct repository *r)
 677 {
 678         if (r->objects->alt_odb_tail)
 679                 return;
 680
 681         r->objects->alt_odb_tail = &r->objects->alt_odb_list;
 682         link_alt_odb_entries(r, r->objects->alternate_db, PATH_SEP, NULL, 0);
 683
 684         read_info_alternates(r, r->objects->objectdir, 0);
 685 }
 686
 687 /* Returns 1 if we have successfully freshened the file, 0 otherwise. */
 688 static int freshen_file(const char *fn)
 689 {
 690         struct utimbuf t;
 691         t.actime = t.modtime = time(NULL);
 692         return !utime(fn, &t);
 693 }
 694
 695 /*
 696  * All of the check_and_freshen functions return 1 if the file exists and was
 697  * freshened (if freshening was requested), 0 otherwise. If they return
 698  * 0, you should not assume that it is safe to skip a write of the object (it
 699  * either does not exist on disk, or has a stale mtime and may be subject to
 700  * pruning).
 701  */
 702 int check_and_freshen_file(const char *fn, int freshen)
 703 {
 704         if (access(fn, F_OK))
 705                 return 0;
 706         if (freshen && !freshen_file(fn))
 707                 return 0;
 708         return 1;
 709 }
 710
 711 static int check_and_freshen_local(const unsigned char *sha1, int freshen)
 712 {
 713         static struct strbuf buf = STRBUF_INIT;
 714
 715         strbuf_reset(&buf);
 716         sha1_file_name(the_repository, &buf, sha1);
 717
 718         return check_and_freshen_file(buf.buf, freshen);
 719 }
 720
 721 static int check_and_freshen_nonlocal(const unsigned char *sha1, int freshen)
 722 {
 723         struct alternate_object_database *alt;
 724         prepare_alt_odb(the_repository);
 725         for (alt = the_repository->objects->alt_odb_list; alt; alt = alt->next) {
 726                 const char *path = alt_sha1_path(alt, sha1);
 727                 if (check_and_freshen_file(path, freshen))
 728                         return 1;
 729         }
 730         return 0;
 731 }
 732
 733 static int check_and_freshen(const unsigned char *sha1, int freshen)
 734 {
 735         return check_and_freshen_local(sha1, freshen) ||
 736                check_and_freshen_nonlocal(sha1, freshen);
 737 }
 738
 739 int has_loose_object_nonlocal(const unsigned char *sha1)
 740 {
 741         return check_and_freshen_nonlocal(sha1, 0);
 742 }
 743
 744 static int has_loose_object(const unsigned char *sha1)
 745 {
 746         return check_and_freshen(sha1, 0);
 747 }
 748
 749 static void mmap_limit_check(size_t length)
 750 {
 751         static size_t limit = 0;
 752         if (!limit) {
 753                 limit = git_env_ulong("GIT_MMAP_LIMIT", 0);
 754                 if (!limit)
 755                         limit = SIZE_MAX;
 756         }
 757         if (length > limit)
 758                 die("attempting to mmap %"PRIuMAX" over limit %"PRIuMAX,
 759                     (uintmax_t)length, (uintmax_t)limit);
 760 }
 761
 762 void *xmmap_gently(void *start, size_t length,
 763                   int prot, int flags, int fd, off_t offset)
 764 {
 765         void *ret;
 766
 767         mmap_limit_check(length);
 768         ret = mmap(start, length, prot, flags, fd, offset);
 769         if (ret == MAP_FAILED) {
 770                 if (!length)
 771                         return NULL;
 772                 release_pack_memory(length);
 773                 ret = mmap(start, length, prot, flags, fd, offset);
 774         }
 775         return ret;
 776 }
 777
 778 void *xmmap(void *start, size_t length,
 779         int prot, int flags, int fd, off_t offset)
 780 {
 781         void *ret = xmmap_gently(start, length, prot, flags, fd, offset);
 782         if (ret == MAP_FAILED)
 783                 die_errno("mmap failed");
 784         return ret;
 785 }
 786
 787 /*
 788  * With an in-core object data in "map", rehash it to make sure the
 789  * object name actually matches "sha1" to detect object corruption.
 790  * With "map" == NULL, try reading the object named with "sha1" using
 791  * the streaming interface and rehash it to do the same.
 792  */
 793 int check_sha1_signature(const unsigned char *sha1, void *map,
 794                          unsigned long size, const char *type)
 795 {
 796         unsigned char real_sha1[20];
 797         enum object_type obj_type;
 798         struct git_istream *st;
 799         git_SHA_CTX c;
 800         char hdr[32];
 801         int hdrlen;
 802
 803         if (map) {
 804                 hash_sha1_file(map, size, type, real_sha1);
 805                 return hashcmp(sha1, real_sha1) ? -1 : 0;
 806         }
 807
 808         st = open_istream(sha1, &obj_type, &size, NULL);
 809         if (!st)
 810                 return -1;
 811
 812         /* Generate the header */
 813         hdrlen = xsnprintf(hdr, sizeof(hdr), "%s %lu", typename(obj_type), size) + 1;
 814
 815         /* Sha1.. */
 816         git_SHA1_Init(&c);
 817         git_SHA1_Update(&c, hdr, hdrlen);
 818         for (;;) {
 819                 char buf[1024 * 16];
 820                 ssize_t readlen = read_istream(st, buf, sizeof(buf));
 821
 822                 if (readlen < 0) {
 823                         close_istream(st);
 824                         return -1;
 825                 }
 826                 if (!readlen)
 827                         break;
 828                 git_SHA1_Update(&c, buf, readlen);
 829         }
 830         git_SHA1_Final(real_sha1, &c);
 831         close_istream(st);
 832         return hashcmp(sha1, real_sha1) ? -1 : 0;
 833 }
 834
 835 int git_open_cloexec(const char *name, int flags)
 836 {
 837         int fd;
 838         static int o_cloexec = O_CLOEXEC;
 839
 840         fd = open(name, flags | o_cloexec);
 841         if ((o_cloexec & O_CLOEXEC) && fd < 0 && errno == EINVAL) {
 842                 /* Try again w/o O_CLOEXEC: the kernel might not support it */
 843                 o_cloexec &= ~O_CLOEXEC;
 844                 fd = open(name, flags | o_cloexec);
 845         }
 846
 847 #if defined(F_GETFD) && defined(F_SETFD) && defined(FD_CLOEXEC)
 848         {
 849                 static int fd_cloexec = FD_CLOEXEC;
 850
 851                 if (!o_cloexec && 0 <= fd && fd_cloexec) {
 852                         /* Opened w/o O_CLOEXEC?  try with fcntl(2) to add it */
 853                         int flags = fcntl(fd, F_GETFD);
 854                         if (fcntl(fd, F_SETFD, flags | fd_cloexec))
 855                                 fd_cloexec = 0;
 856                 }
 857         }
 858 #endif
 859         return fd;
 860 }
 861
 862 /*
 863  * Find "sha1" as a loose object in the local repository or in an alternate.
 864  * Returns 0 on success, negative on failure.
 865  *
 866  * The "path" out-parameter will give the path of the object we found (if any).
 867  * Note that it may point to static storage and is only valid until another
 868  * call to sha1_file_name(), etc.
 869  */
 870 static int stat_sha1_file(const unsigned char *sha1, struct stat *st,
 871                           const char **path)
 872 {
 873         struct alternate_object_database *alt;
 874         static struct strbuf buf = STRBUF_INIT;
 875
 876         strbuf_reset(&buf);
 877         sha1_file_name(the_repository, &buf, sha1);
 878         *path = buf.buf;
 879
 880         if (!lstat(*path, st))
 881                 return 0;
 882
 883         prepare_alt_odb(the_repository);
 884         errno = ENOENT;
 885         for (alt = the_repository->objects->alt_odb_list; alt; alt = alt->next) {
 886                 *path = alt_sha1_path(alt, sha1);
 887                 if (!lstat(*path, st))
 888                         return 0;
 889         }
 890
 891         return -1;
 892 }
 893
 894 /*
 895  * Like stat_sha1_file(), but actually open the object and return the
 896  * descriptor. See the caveats on the "path" parameter above.
 897  */
 898 static int open_sha1_file(const unsigned char *sha1, const char **path)
 899 {
 900         int fd;
 901         struct alternate_object_database *alt;
 902         int most_interesting_errno;
 903         static struct strbuf buf = STRBUF_INIT;
 904
 905         strbuf_reset(&buf);
 906         sha1_file_name(the_repository, &buf, sha1);
 907         *path = buf.buf;
 908
 909         fd = git_open(*path);
 910         if (fd >= 0)
 911                 return fd;
 912         most_interesting_errno = errno;
 913
 914         prepare_alt_odb(the_repository);
 915         for (alt = the_repository->objects->alt_odb_list; alt; alt = alt->next) {
 916                 *path = alt_sha1_path(alt, sha1);
 917                 fd = git_open(*path);
 918                 if (fd >= 0)
 919                         return fd;
 920                 if (most_interesting_errno == ENOENT)
 921                         most_interesting_errno = errno;
 922         }
 923         errno = most_interesting_errno;
 924         return -1;
 925 }
 926
 927 /*
 928  * Map the loose object at "path" if it is not NULL, or the path found by
 929  * searching for a loose object named "sha1".
 930  */
 931 static void *map_sha1_file_1(const char *path,
 932                              const unsigned char *sha1,
 933                              unsigned long *size)
 934 {
 935         void *map;
 936         int fd;
 937
 938         if (path)
 939                 fd = git_open(path);
 940         else
 941                 fd = open_sha1_file(sha1, &path);
 942         map = NULL;
 943         if (fd >= 0) {
 944                 struct stat st;
 945
 946                 if (!fstat(fd, &st)) {
 947                         *size = xsize_t(st.st_size);
 948                         if (!*size) {
 949                                 /* mmap() is forbidden on empty files */
 950                                 error("object file %s is empty", path);
 951                                 return NULL;
 952                         }
 953                         map = xmmap(NULL, *size, PROT_READ, MAP_PRIVATE, fd, 0);
 954                 }
 955                 close(fd);
 956         }
 957         return map;
 958 }
 959
 960 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 961 {
 962         return map_sha1_file_1(NULL, sha1, size);
 963 }
 964
 965 static int unpack_sha1_short_header(git_zstream *stream,
 966                                     unsigned char *map, unsigned long mapsize,
 967                                     void *buffer, unsigned long bufsiz)
 968 {
 969         /* Get the data stream */
 970         memset(stream, 0, sizeof(*stream));
 971         stream->next_in = map;
 972         stream->avail_in = mapsize;
 973         stream->next_out = buffer;
 974         stream->avail_out = bufsiz;
 975
 976         git_inflate_init(stream);
 977         return git_inflate(stream, 0);
 978 }
 979
 980 int unpack_sha1_header(git_zstream *stream,
 981                        unsigned char *map, unsigned long mapsize,
 982                        void *buffer, unsigned long bufsiz)
 983 {
 984         int status = unpack_sha1_short_header(stream, map, mapsize,
 985                                               buffer, bufsiz);
 986
 987         if (status < Z_OK)
 988                 return status;
 989
 990         /* Make sure we have the terminating NUL */
 991         if (!memchr(buffer, '\0', stream->next_out - (unsigned char *)buffer))
 992                 return -1;
 993         return 0;
 994 }
 995
 996 static int unpack_sha1_header_to_strbuf(git_zstream *stream, unsigned char *map,
 997                                         unsigned long mapsize, void *buffer,
 998                                         unsigned long bufsiz, struct strbuf *header)
 999 {
1000         int status;
1001
1002         status = unpack_sha1_short_header(stream, map, mapsize, buffer, bufsiz);
1003         if (status < Z_OK)
1004                 return -1;
1005
1006         /*
1007          * Check if entire header is unpacked in the first iteration.
1008          */
1009         if (memchr(buffer, '\0', stream->next_out - (unsigned char *)buffer))
1010                 return 0;
1011
1012         /*
1013          * buffer[0..bufsiz] was not large enough.  Copy the partial
1014          * result out to header, and then append the result of further
1015          * reading the stream.
1016          */
1017         strbuf_add(header, buffer, stream->next_out - (unsigned char *)buffer);
1018         stream->next_out = buffer;
1019         stream->avail_out = bufsiz;
1020
1021         do {
1022                 status = git_inflate(stream, 0);
1023                 strbuf_add(header, buffer, stream->next_out - (unsigned char *)buffer);
1024                 if (memchr(buffer, '\0', stream->next_out - (unsigned char *)buffer))
1025                         return 0;
1026                 stream->next_out = buffer;
1027                 stream->avail_out = bufsiz;
1028         } while (status != Z_STREAM_END);
1029         return -1;
1030 }
1031
1032 static void *unpack_sha1_rest(git_zstream *stream, void *buffer, unsigned long size, const unsigned char *sha1)
1033 {
1034         int bytes = strlen(buffer) + 1;
1035         unsigned char *buf = xmallocz(size);
1036         unsigned long n;
1037         int status = Z_OK;
1038
1039         n = stream->total_out - bytes;
1040         if (n > size)
1041                 n = size;
1042         memcpy(buf, (char *) buffer + bytes, n);
1043         bytes = n;
1044         if (bytes <= size) {
1045                 /*
1046                  * The above condition must be (bytes <= size), not
1047                  * (bytes < size).  In other words, even though we
1048                  * expect no more output and set avail_out to zero,
1049                  * the input zlib stream may have bytes that express
1050                  * "this concludes the stream", and we *do* want to
1051                  * eat that input.
1052                  *
1053                  * Otherwise we would not be able to test that we
1054                  * consumed all the input to reach the expected size;
1055                  * we also want to check that zlib tells us that all
1056                  * went well with status == Z_STREAM_END at the end.
1057                  */
1058                 stream->next_out = buf + bytes;
1059                 stream->avail_out = size - bytes;
1060                 while (status == Z_OK)
1061                         status = git_inflate(stream, Z_FINISH);
1062         }
1063         if (status == Z_STREAM_END && !stream->avail_in) {
1064                 git_inflate_end(stream);
1065                 return buf;
1066         }
1067
1068         if (status < 0)
1069                 error("corrupt loose object '%s'", sha1_to_hex(sha1));
1070         else if (stream->avail_in)
1071                 error("garbage at end of loose object '%s'",
1072                       sha1_to_hex(sha1));
1073         free(buf);
1074         return NULL;
1075 }
1076
1077 /*
1078  * We used to just use "sscanf()", but that's actually way
1079  * too permissive for what we want to check. So do an anal
1080  * object header parse by hand.
1081  */
1082 static int parse_sha1_header_extended(const char *hdr, struct object_info *oi,
1083                                unsigned int flags)
1084 {
1085         const char *type_buf = hdr;
1086         unsigned long size;
1087         int type, type_len = 0;
1088
1089         /*
1090          * The type can be of any size but is followed by
1091          * a space.
1092          */
1093         for (;;) {
1094                 char c = *hdr++;
1095                 if (!c)
1096                         return -1;
1097                 if (c == ' ')
1098                         break;
1099                 type_len++;
1100         }
1101
1102         type = type_from_string_gently(type_buf, type_len, 1);
1103         if (oi->typename)
1104                 strbuf_add(oi->typename, type_buf, type_len);
1105         /*
1106          * Set type to 0 if its an unknown object and
1107          * we're obtaining the type using '--allow-unknown-type'
1108          * option.
1109          */
1110         if ((flags & OBJECT_INFO_ALLOW_UNKNOWN_TYPE) && (type < 0))
1111                 type = 0;
1112         else if (type < 0)
1113                 die("invalid object type");
1114         if (oi->typep)
1115                 *oi->typep = type;
1116
1117         /*
1118          * The length must follow immediately, and be in canonical
1119          * decimal format (ie "010" is not valid).
1120          */
1121         size = *hdr++ - '0';
1122         if (size > 9)
1123                 return -1;
1124         if (size) {
1125                 for (;;) {
1126                         unsigned long c = *hdr - '0';
1127                         if (c > 9)
1128                                 break;
1129                         hdr++;
1130                         size = size * 10 + c;
1131                 }
1132         }
1133
1134         if (oi->sizep)
1135                 *oi->sizep = size;
1136
1137         /*
1138          * The length must be followed by a zero byte
1139          */
1140         return *hdr ? -1 : type;
1141 }
1142
1143 int parse_sha1_header(const char *hdr, unsigned long *sizep)
1144 {
1145         struct object_info oi = OBJECT_INFO_INIT;
1146
1147         oi.sizep = sizep;
1148         return parse_sha1_header_extended(hdr, &oi, 0);
1149 }
1150
1151 static int sha1_loose_object_info(const unsigned char *sha1,
1152                                   struct object_info *oi,
1153                                   int flags)
1154 {
1155         int status = 0;
1156         unsigned long mapsize;
1157         void *map;
1158         git_zstream stream;
1159         char hdr[32];
1160         struct strbuf hdrbuf = STRBUF_INIT;
1161         unsigned long size_scratch;
1162
1163         if (oi->delta_base_sha1)
1164                 hashclr(oi->delta_base_sha1);
1165
1166         /*
1167          * If we don't care about type or size, then we don't
1168          * need to look inside the object at all. Note that we
1169          * do not optimize out the stat call, even if the
1170          * caller doesn't care about the disk-size, since our
1171          * return value implicitly indicates whether the
1172          * object even exists.
1173          */
1174         if (!oi->typep && !oi->typename && !oi->sizep && !oi->contentp) {
1175                 const char *path;
1176                 struct stat st;
1177                 if (stat_sha1_file(sha1, &st, &path) < 0)
1178                         return -1;
1179                 if (oi->disk_sizep)
1180                         *oi->disk_sizep = st.st_size;
1181                 return 0;
1182         }
1183
1184         map = map_sha1_file(sha1, &mapsize);
1185         if (!map)
1186                 return -1;
1187
1188         if (!oi->sizep)
1189                 oi->sizep = &size_scratch;
1190
1191         if (oi->disk_sizep)
1192                 *oi->disk_sizep = mapsize;
1193         if ((flags & OBJECT_INFO_ALLOW_UNKNOWN_TYPE)) {
1194                 if (unpack_sha1_header_to_strbuf(&stream, map, mapsize, hdr, sizeof(hdr), &hdrbuf) < 0)
1195                         status = error("unable to unpack %s header with --allow-unknown-type",
1196                                        sha1_to_hex(sha1));
1197         } else if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
1198                 status = error("unable to unpack %s header",
1199                                sha1_to_hex(sha1));
1200         if (status < 0)
1201                 ; /* Do nothing */
1202         else if (hdrbuf.len) {
1203                 if ((status = parse_sha1_header_extended(hdrbuf.buf, oi, flags)) < 0)
1204                         status = error("unable to parse %s header with --allow-unknown-type",
1205                                        sha1_to_hex(sha1));
1206         } else if ((status = parse_sha1_header_extended(hdr, oi, flags)) < 0)
1207                 status = error("unable to parse %s header", sha1_to_hex(sha1));
1208
1209         if (status >= 0 && oi->contentp) {
1210                 *oi->contentp = unpack_sha1_rest(&stream, hdr,
1211                                                  *oi->sizep, sha1);
1212                 if (!*oi->contentp) {
1213                         git_inflate_end(&stream);
1214                         status = -1;
1215                 }
1216         } else
1217                 git_inflate_end(&stream);
1218
1219         munmap(map, mapsize);
1220         if (status && oi->typep)
1221                 *oi->typep = status;
1222         if (oi->sizep == &size_scratch)
1223                 oi->sizep = NULL;
1224         strbuf_release(&hdrbuf);
1225         oi->whence = OI_LOOSE;
1226         return (status < 0) ? status : 0;
1227 }
1228
1229 int fetch_if_missing = 1;
1230
1231 int sha1_object_info_extended(const unsigned char *sha1, struct object_info *oi, unsigned flags)
1232 {
1233         static struct object_info blank_oi = OBJECT_INFO_INIT;
1234         struct pack_entry e;
1235         int rtype;
1236         const unsigned char *real = (flags & OBJECT_INFO_LOOKUP_REPLACE) ?
1237                                     lookup_replace_object(sha1) :
1238                                     sha1;
1239         int already_retried = 0;
1240
1241         if (is_null_sha1(real))
1242                 return -1;
1243
1244         if (!oi)
1245                 oi = &blank_oi;
1246
1247         if (!(flags & OBJECT_INFO_SKIP_CACHED)) {
1248                 struct cached_object *co = find_cached_object(real);
1249                 if (co) {
1250                         if (oi->typep)
1251                                 *(oi->typep) = co->type;
1252                         if (oi->sizep)
1253                                 *(oi->sizep) = co->size;
1254                         if (oi->disk_sizep)
1255                                 *(oi->disk_sizep) = 0;
1256                         if (oi->delta_base_sha1)
1257                                 hashclr(oi->delta_base_sha1);
1258                         if (oi->typename)
1259                                 strbuf_addstr(oi->typename, typename(co->type));
1260                         if (oi->contentp)
1261                                 *oi->contentp = xmemdupz(co->buf, co->size);
1262                         oi->whence = OI_CACHED;
1263                         return 0;
1264                 }
1265         }
1266
1267         while (1) {
1268                 if (find_pack_entry(real, &e))
1269                         break;
1270
1271                 /* Most likely it's a loose object. */
1272                 if (!sha1_loose_object_info(real, oi, flags))
1273                         return 0;
1274
1275                 /* Not a loose object; someone else may have just packed it. */
1276                 reprepare_packed_git();
1277                 if (find_pack_entry(real, &e))
1278                         break;
1279
1280                 /* Check if it is a missing object */
1281                 if (fetch_if_missing && repository_format_partial_clone &&
1282                     !already_retried) {
1283                         /*
1284                          * TODO Investigate haveing fetch_object() return
1285                          * TODO error/success and stopping the music here.
1286                          */
1287                         fetch_object(repository_format_partial_clone, real);
1288                         already_retried = 1;
1289                         continue;
1290                 }
1291
1292                 return -1;
1293         }
1294
1295         if (oi == &blank_oi)
1296                 /*
1297                  * We know that the caller doesn't actually need the
1298                  * information below, so return early.
1299                  */
1300                 return 0;
1301         rtype = packed_object_info(e.p, e.offset, oi);
1302         if (rtype < 0) {
1303                 mark_bad_packed_object(e.p, real);
1304                 return sha1_object_info_extended(real, oi, 0);
1305         } else if (oi->whence == OI_PACKED) {
1306                 oi->u.packed.offset = e.offset;
1307                 oi->u.packed.pack = e.p;
1308                 oi->u.packed.is_delta = (rtype == OBJ_REF_DELTA ||
1309                                          rtype == OBJ_OFS_DELTA);
1310         }
1311
1312         return 0;
1313 }
1314
1315 /* returns enum object_type or negative */
1316 int sha1_object_info(const unsigned char *sha1, unsigned long *sizep)
1317 {
1318         enum object_type type;
1319         struct object_info oi = OBJECT_INFO_INIT;
1320
1321         oi.typep = &type;
1322         oi.sizep = sizep;
1323         if (sha1_object_info_extended(sha1, &oi,
1324                                       OBJECT_INFO_LOOKUP_REPLACE) < 0)
1325                 return -1;
1326         return type;
1327 }
1328
1329 static void *read_object(const unsigned char *sha1, enum object_type *type,
1330                          unsigned long *size)
1331 {
1332         struct object_info oi = OBJECT_INFO_INIT;
1333         void *content;
1334         oi.typep = type;
1335         oi.sizep = size;
1336         oi.contentp = &content;
1337
1338         if (sha1_object_info_extended(sha1, &oi, 0) < 0)
1339                 return NULL;
1340         return content;
1341 }
1342
1343 int pretend_sha1_file(void *buf, unsigned long len, enum object_type type,
1344                       unsigned char *sha1)
1345 {
1346         struct cached_object *co;
1347
1348         hash_sha1_file(buf, len, typename(type), sha1);
1349         if (has_sha1_file(sha1) || find_cached_object(sha1))
1350                 return 0;
1351         ALLOC_GROW(cached_objects, cached_object_nr + 1, cached_object_alloc);
1352         co = &cached_objects[cached_object_nr++];
1353         co->size = len;
1354         co->type = type;
1355         co->buf = xmalloc(len);
1356         memcpy(co->buf, buf, len);
1357         hashcpy(co->sha1, sha1);
1358         return 0;
1359 }
1360
1361 /*
1362  * This function dies on corrupt objects; the callers who want to
1363  * deal with them should arrange to call read_object() and give error
1364  * messages themselves.
1365  */
1366 void *read_sha1_file_extended(const unsigned char *sha1,
1367                               enum object_type *type,
1368                               unsigned long *size,
1369                               int lookup_replace)
1370 {
1371         void *data;
1372         const struct packed_git *p;
1373         const char *path;
1374         struct stat st;
1375         const unsigned char *repl = lookup_replace ? lookup_replace_object(sha1)
1376                                                    : sha1;
1377
1378         errno = 0;
1379         data = read_object(repl, type, size);
1380         if (data)
1381                 return data;
1382
1383         if (errno && errno != ENOENT)
1384                 die_errno("failed to read object %s", sha1_to_hex(sha1));
1385
1386         /* die if we replaced an object with one that does not exist */
1387         if (repl != sha1)
1388                 die("replacement %s not found for %s",
1389                     sha1_to_hex(repl), sha1_to_hex(sha1));
1390
1391         if (!stat_sha1_file(repl, &st, &path))
1392                 die("loose object %s (stored in %s) is corrupt",
1393                     sha1_to_hex(repl), path);
1394
1395         if ((p = has_packed_and_bad(repl)) != NULL)
1396                 die("packed object %s (stored in %s) is corrupt",
1397                     sha1_to_hex(repl), p->pack_name);
1398
1399         return NULL;
1400 }
1401
1402 void *read_object_with_reference(const unsigned char *sha1,
1403                                  const char *required_type_name,
1404                                  unsigned long *size,
1405                                  unsigned char *actual_sha1_return)
1406 {
1407         enum object_type type, required_type;
1408         void *buffer;
1409         unsigned long isize;
1410         unsigned char actual_sha1[20];
1411
1412         required_type = type_from_string(required_type_name);
1413         hashcpy(actual_sha1, sha1);
1414         while (1) {
1415                 int ref_length = -1;
1416                 const char *ref_type = NULL;
1417
1418                 buffer = read_sha1_file(actual_sha1, &type, &isize);
1419                 if (!buffer)
1420                         return NULL;
1421                 if (type == required_type) {
1422                         *size = isize;
1423                         if (actual_sha1_return)
1424                                 hashcpy(actual_sha1_return, actual_sha1);
1425                         return buffer;
1426                 }
1427                 /* Handle references */
1428                 else if (type == OBJ_COMMIT)
1429                         ref_type = "tree ";
1430                 else if (type == OBJ_TAG)
1431                         ref_type = "object ";
1432                 else {
1433                         free(buffer);
1434                         return NULL;
1435                 }
1436                 ref_length = strlen(ref_type);
1437
1438                 if (ref_length + 40 > isize ||
1439                     memcmp(buffer, ref_type, ref_length) ||
1440                     get_sha1_hex((char *) buffer + ref_length, actual_sha1)) {
1441                         free(buffer);
1442                         return NULL;
1443                 }
1444                 free(buffer);
1445                 /* Now we have the ID of the referred-to object in
1446                  * actual_sha1.  Check again. */
1447         }
1448 }
1449
1450 static void write_sha1_file_prepare(const void *buf, unsigned long len,
1451                                     const char *type, unsigned char *sha1,
1452                                     char *hdr, int *hdrlen)
1453 {
1454         git_SHA_CTX c;
1455
1456         /* Generate the header */
1457         *hdrlen = xsnprintf(hdr, *hdrlen, "%s %lu", type, len)+1;
1458
1459         /* Sha1.. */
1460         git_SHA1_Init(&c);
1461         git_SHA1_Update(&c, hdr, *hdrlen);
1462         git_SHA1_Update(&c, buf, len);
1463         git_SHA1_Final(sha1, &c);
1464 }
1465
1466 /*
1467  * Move the just written object into its final resting place.
1468  */
1469 int finalize_object_file(const char *tmpfile, const char *filename)
1470 {
1471         int ret = 0;
1472
1473         if (object_creation_mode == OBJECT_CREATION_USES_RENAMES)
1474                 goto try_rename;
1475         else if (link(tmpfile, filename))
1476                 ret = errno;
1477
1478         /*
1479          * Coda hack - coda doesn't like cross-directory links,
1480          * so we fall back to a rename, which will mean that it
1481          * won't be able to check collisions, but that's not a
1482          * big deal.
1483          *
1484          * The same holds for FAT formatted media.
1485          *
1486          * When this succeeds, we just return.  We have nothing
1487          * left to unlink.
1488          */
1489         if (ret && ret != EEXIST) {
1490         try_rename:
1491                 if (!rename(tmpfile, filename))
1492                         goto out;
1493                 ret = errno;
1494         }
1495         unlink_or_warn(tmpfile);
1496         if (ret) {
1497                 if (ret != EEXIST) {
1498                         return error_errno("unable to write sha1 filename %s", filename);
1499                 }
1500                 /* FIXME!!! Collision check here ? */
1501         }
1502
1503 out:
1504         if (adjust_shared_perm(filename))
1505                 return error("unable to set permission to '%s'", filename);
1506         return 0;
1507 }
1508
1509 static int write_buffer(int fd, const void *buf, size_t len)
1510 {
1511         if (write_in_full(fd, buf, len) < 0)
1512                 return error_errno("file write error");
1513         return 0;
1514 }
1515
1516 int hash_sha1_file(const void *buf, unsigned long len, const char *type,
1517                    unsigned char *sha1)
1518 {
1519         char hdr[32];
1520         int hdrlen = sizeof(hdr);
1521         write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1522         return 0;
1523 }
1524
1525 /* Finalize a file on disk, and close it. */
1526 static void close_sha1_file(int fd)
1527 {
1528         if (fsync_object_files)
1529                 fsync_or_die(fd, "sha1 file");
1530         if (close(fd) != 0)
1531                 die_errno("error when closing sha1 file");
1532 }
1533
1534 /* Size of directory component, including the ending '/' */
1535 static inline int directory_size(const char *filename)
1536 {
1537         const char *s = strrchr(filename, '/');
1538         if (!s)
1539                 return 0;
1540         return s - filename + 1;
1541 }
1542
1543 /*
1544  * This creates a temporary file in the same directory as the final
1545  * 'filename'
1546  *
1547  * We want to avoid cross-directory filename renames, because those
1548  * can have problems on various filesystems (FAT, NFS, Coda).
1549  */
1550 static int create_tmpfile(struct strbuf *tmp, const char *filename)
1551 {
1552         int fd, dirlen = directory_size(filename);
1553
1554         strbuf_reset(tmp);
1555         strbuf_add(tmp, filename, dirlen);
1556         strbuf_addstr(tmp, "tmp_obj_XXXXXX");
1557         fd = git_mkstemp_mode(tmp->buf, 0444);
1558         if (fd < 0 && dirlen && errno == ENOENT) {
1559                 /*
1560                  * Make sure the directory exists; note that the contents
1561                  * of the buffer are undefined after mkstemp returns an
1562                  * error, so we have to rewrite the whole buffer from
1563                  * scratch.
1564                  */
1565                 strbuf_reset(tmp);
1566                 strbuf_add(tmp, filename, dirlen - 1);
1567                 if (mkdir(tmp->buf, 0777) && errno != EEXIST)
1568                         return -1;
1569                 if (adjust_shared_perm(tmp->buf))
1570                         return -1;
1571
1572                 /* Try again */
1573                 strbuf_addstr(tmp, "/tmp_obj_XXXXXX");
1574                 fd = git_mkstemp_mode(tmp->buf, 0444);
1575         }
1576         return fd;
1577 }
1578
1579 static int write_loose_object(const unsigned char *sha1, char *hdr, int hdrlen,
1580                               const void *buf, unsigned long len, time_t mtime)
1581 {
1582         int fd, ret;
1583         unsigned char compressed[4096];
1584         git_zstream stream;
1585         git_SHA_CTX c;
1586         unsigned char parano_sha1[20];
1587         static struct strbuf tmp_file = STRBUF_INIT;
1588         static struct strbuf filename = STRBUF_INIT;
1589
1590         strbuf_reset(&filename);
1591         sha1_file_name(the_repository, &filename, sha1);
1592
1593         fd = create_tmpfile(&tmp_file, filename.buf);
1594         if (fd < 0) {
1595                 if (errno == EACCES)
1596                         return error("insufficient permission for adding an object to repository database %s", get_object_directory());
1597                 else
1598                         return error_errno("unable to create temporary file");
1599         }
1600
1601         /* Set it up */
1602         git_deflate_init(&stream, zlib_compression_level);
1603         stream.next_out = compressed;
1604         stream.avail_out = sizeof(compressed);
1605         git_SHA1_Init(&c);
1606
1607         /* First header.. */
1608         stream.next_in = (unsigned char *)hdr;
1609         stream.avail_in = hdrlen;
1610         while (git_deflate(&stream, 0) == Z_OK)
1611                 ; /* nothing */
1612         git_SHA1_Update(&c, hdr, hdrlen);
1613
1614         /* Then the data itself.. */
1615         stream.next_in = (void *)buf;
1616         stream.avail_in = len;
1617         do {
1618                 unsigned char *in0 = stream.next_in;
1619                 ret = git_deflate(&stream, Z_FINISH);
1620                 git_SHA1_Update(&c, in0, stream.next_in - in0);
1621                 if (write_buffer(fd, compressed, stream.next_out - compressed) < 0)
1622                         die("unable to write sha1 file");
1623                 stream.next_out = compressed;
1624                 stream.avail_out = sizeof(compressed);
1625         } while (ret == Z_OK);
1626
1627         if (ret != Z_STREAM_END)
1628                 die("unable to deflate new object %s (%d)", sha1_to_hex(sha1), ret);
1629         ret = git_deflate_end_gently(&stream);
1630         if (ret != Z_OK)
1631                 die("deflateEnd on object %s failed (%d)", sha1_to_hex(sha1), ret);
1632         git_SHA1_Final(parano_sha1, &c);
1633         if (hashcmp(sha1, parano_sha1) != 0)
1634                 die("confused by unstable object source data for %s", sha1_to_hex(sha1));
1635
1636         close_sha1_file(fd);
1637
1638         if (mtime) {
1639                 struct utimbuf utb;
1640                 utb.actime = mtime;
1641                 utb.modtime = mtime;
1642                 if (utime(tmp_file.buf, &utb) < 0)
1643                         warning_errno("failed utime() on %s", tmp_file.buf);
1644         }
1645
1646         return finalize_object_file(tmp_file.buf, filename.buf);
1647 }
1648
1649 static int freshen_loose_object(const unsigned char *sha1)
1650 {
1651         return check_and_freshen(sha1, 1);
1652 }
1653
1654 static int freshen_packed_object(const unsigned char *sha1)
1655 {
1656         struct pack_entry e;
1657         if (!find_pack_entry(sha1, &e))
1658                 return 0;
1659         if (e.p->freshened)
1660                 return 1;
1661         if (!freshen_file(e.p->pack_name))
1662                 return 0;
1663         e.p->freshened = 1;
1664         return 1;
1665 }
1666
1667 int write_sha1_file(const void *buf, unsigned long len, const char *type, unsigned char *sha1)
1668 {
1669         char hdr[32];
1670         int hdrlen = sizeof(hdr);
1671
1672         /* Normally if we have it in the pack then we do not bother writing
1673          * it out into .git/objects/??/?{38} file.
1674          */
1675         write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1676         if (freshen_packed_object(sha1) || freshen_loose_object(sha1))
1677                 return 0;
1678         return write_loose_object(sha1, hdr, hdrlen, buf, len, 0);
1679 }
1680
1681 int hash_sha1_file_literally(const void *buf, unsigned long len, const char *type,
1682                              struct object_id *oid, unsigned flags)
1683 {
1684         char *header;
1685         int hdrlen, status = 0;
1686
1687         /* type string, SP, %lu of the length plus NUL must fit this */
1688         hdrlen = strlen(type) + 32;
1689         header = xmalloc(hdrlen);
1690         write_sha1_file_prepare(buf, len, type, oid->hash, header, &hdrlen);
1691
1692         if (!(flags & HASH_WRITE_OBJECT))
1693                 goto cleanup;
1694         if (freshen_packed_object(oid->hash) || freshen_loose_object(oid->hash))
1695                 goto cleanup;
1696         status = write_loose_object(oid->hash, header, hdrlen, buf, len, 0);
1697
1698 cleanup:
1699         free(header);
1700         return status;
1701 }
1702
1703 int force_object_loose(const unsigned char *sha1, time_t mtime)
1704 {
1705         void *buf;
1706         unsigned long len;
1707         enum object_type type;
1708         char hdr[32];
1709         int hdrlen;
1710         int ret;
1711
1712         if (has_loose_object(sha1))
1713                 return 0;
1714         buf = read_object(sha1, &type, &len);
1715         if (!buf)
1716                 return error("cannot read sha1_file for %s", sha1_to_hex(sha1));
1717         hdrlen = xsnprintf(hdr, sizeof(hdr), "%s %lu", typename(type), len) + 1;
1718         ret = write_loose_object(sha1, hdr, hdrlen, buf, len, mtime);
1719         free(buf);
1720
1721         return ret;
1722 }
1723
1724 int has_sha1_file_with_flags(const unsigned char *sha1, int flags)
1725 {
1726         if (!startup_info->have_repository)
1727                 return 0;
1728         return sha1_object_info_extended(sha1, NULL,
1729                                          flags | OBJECT_INFO_SKIP_CACHED) >= 0;
1730 }
1731
1732 int has_object_file(const struct object_id *oid)
1733 {
1734         return has_sha1_file(oid->hash);
1735 }
1736
1737 int has_object_file_with_flags(const struct object_id *oid, int flags)
1738 {
1739         return has_sha1_file_with_flags(oid->hash, flags);
1740 }
1741
1742 static void check_tree(const void *buf, size_t size)
1743 {
1744         struct tree_desc desc;
1745         struct name_entry entry;
1746
1747         init_tree_desc(&desc, buf, size);
1748         while (tree_entry(&desc, &entry))
1749                 /* do nothing
1750                  * tree_entry() will die() on malformed entries */
1751                 ;
1752 }
1753
1754 static void check_commit(const void *buf, size_t size)
1755 {
1756         struct commit c;
1757         memset(&c, 0, sizeof(c));
1758         if (parse_commit_buffer(&c, buf, size))
1759                 die("corrupt commit");
1760 }
1761
1762 static void check_tag(const void *buf, size_t size)
1763 {
1764         struct tag t;
1765         memset(&t, 0, sizeof(t));
1766         if (parse_tag_buffer(&t, buf, size))
1767                 die("corrupt tag");
1768 }
1769
1770 static int index_mem(struct object_id *oid, void *buf, size_t size,
1771                      enum object_type type,
1772                      const char *path, unsigned flags)
1773 {
1774         int ret, re_allocated = 0;
1775         int write_object = flags & HASH_WRITE_OBJECT;
1776
1777         if (!type)
1778                 type = OBJ_BLOB;
1779
1780         /*
1781          * Convert blobs to git internal format
1782          */
1783         if ((type == OBJ_BLOB) && path) {
1784                 struct strbuf nbuf = STRBUF_INIT;
1785                 if (convert_to_git(&the_index, path, buf, size, &nbuf,
1786                                    get_conv_flags(flags))) {
1787                         buf = strbuf_detach(&nbuf, &size);
1788                         re_allocated = 1;
1789                 }
1790         }
1791         if (flags & HASH_FORMAT_CHECK) {
1792                 if (type == OBJ_TREE)
1793                         check_tree(buf, size);
1794                 if (type == OBJ_COMMIT)
1795                         check_commit(buf, size);
1796                 if (type == OBJ_TAG)
1797                         check_tag(buf, size);
1798         }
1799
1800         if (write_object)
1801                 ret = write_sha1_file(buf, size, typename(type), oid->hash);
1802         else
1803                 ret = hash_sha1_file(buf, size, typename(type), oid->hash);
1804         if (re_allocated)
1805                 free(buf);
1806         return ret;
1807 }
1808
1809 static int index_stream_convert_blob(struct object_id *oid, int fd,
1810                                      const char *path, unsigned flags)
1811 {
1812         int ret;
1813         const int write_object = flags & HASH_WRITE_OBJECT;
1814         struct strbuf sbuf = STRBUF_INIT;
1815
1816         assert(path);
1817         assert(would_convert_to_git_filter_fd(path));
1818
1819         convert_to_git_filter_fd(&the_index, path, fd, &sbuf,
1820                                  get_conv_flags(flags));
1821
1822         if (write_object)
1823                 ret = write_sha1_file(sbuf.buf, sbuf.len, typename(OBJ_BLOB),
1824                                       oid->hash);
1825         else
1826                 ret = hash_sha1_file(sbuf.buf, sbuf.len, typename(OBJ_BLOB),
1827                                      oid->hash);
1828         strbuf_release(&sbuf);
1829         return ret;
1830 }
1831
1832 static int index_pipe(struct object_id *oid, int fd, enum object_type type,
1833                       const char *path, unsigned flags)
1834 {
1835         struct strbuf sbuf = STRBUF_INIT;
1836         int ret;
1837
1838         if (strbuf_read(&sbuf, fd, 4096) >= 0)
1839                 ret = index_mem(oid, sbuf.buf, sbuf.len, type, path, flags);
1840         else
1841                 ret = -1;
1842         strbuf_release(&sbuf);
1843         return ret;
1844 }
1845
1846 #define SMALL_FILE_SIZE (32*1024)
1847
1848 static int index_core(struct object_id *oid, int fd, size_t size,
1849                       enum object_type type, const char *path,
1850                       unsigned flags)
1851 {
1852         int ret;
1853
1854         if (!size) {
1855                 ret = index_mem(oid, "", size, type, path, flags);
1856         } else if (size <= SMALL_FILE_SIZE) {
1857                 char *buf = xmalloc(size);
1858                 ssize_t read_result = read_in_full(fd, buf, size);
1859                 if (read_result < 0)
1860                         ret = error_errno("read error while indexing %s",
1861                                           path ? path : "<unknown>");
1862                 else if (read_result != size)
1863                         ret = error("short read while indexing %s",
1864                                     path ? path : "<unknown>");
1865                 else
1866                         ret = index_mem(oid, buf, size, type, path, flags);
1867                 free(buf);
1868         } else {
1869                 void *buf = xmmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1870                 ret = index_mem(oid, buf, size, type, path, flags);
1871                 munmap(buf, size);
1872         }
1873         return ret;
1874 }
1875
1876 /*
1877  * This creates one packfile per large blob unless bulk-checkin
1878  * machinery is "plugged".
1879  *
1880  * This also bypasses the usual "convert-to-git" dance, and that is on
1881  * purpose. We could write a streaming version of the converting
1882  * functions and insert that before feeding the data to fast-import
1883  * (or equivalent in-core API described above). However, that is
1884  * somewhat complicated, as we do not know the size of the filter
1885  * result, which we need to know beforehand when writing a git object.
1886  * Since the primary motivation for trying to stream from the working
1887  * tree file and to avoid mmaping it in core is to deal with large
1888  * binary blobs, they generally do not want to get any conversion, and
1889  * callers should avoid this code path when filters are requested.
1890  */
1891 static int index_stream(struct object_id *oid, int fd, size_t size,
1892                         enum object_type type, const char *path,
1893                         unsigned flags)
1894 {
1895         return index_bulk_checkin(oid->hash, fd, size, type, path, flags);
1896 }
1897
1898 int index_fd(struct object_id *oid, int fd, struct stat *st,
1899              enum object_type type, const char *path, unsigned flags)
1900 {
1901         int ret;
1902
1903         /*
1904          * Call xsize_t() only when needed to avoid potentially unnecessary
1905          * die() for large files.
1906          */
1907         if (type == OBJ_BLOB && path && would_convert_to_git_filter_fd(path))
1908                 ret = index_stream_convert_blob(oid, fd, path, flags);
1909         else if (!S_ISREG(st->st_mode))
1910                 ret = index_pipe(oid, fd, type, path, flags);
1911         else if (st->st_size <= big_file_threshold || type != OBJ_BLOB ||
1912                  (path && would_convert_to_git(&the_index, path)))
1913                 ret = index_core(oid, fd, xsize_t(st->st_size), type, path,
1914                                  flags);
1915         else
1916                 ret = index_stream(oid, fd, xsize_t(st->st_size), type, path,
1917                                    flags);
1918         close(fd);
1919         return ret;
1920 }
1921
1922 int index_path(struct object_id *oid, const char *path, struct stat *st, unsigned flags)
1923 {
1924         int fd;
1925         struct strbuf sb = STRBUF_INIT;
1926         int rc = 0;
1927
1928         switch (st->st_mode & S_IFMT) {
1929         case S_IFREG:
1930                 fd = open(path, O_RDONLY);
1931                 if (fd < 0)
1932                         return error_errno("open(\"%s\")", path);
1933                 if (index_fd(oid, fd, st, OBJ_BLOB, path, flags) < 0)
1934                         return error("%s: failed to insert into database",
1935                                      path);
1936                 break;
1937         case S_IFLNK:
1938                 if (strbuf_readlink(&sb, path, st->st_size))
1939                         return error_errno("readlink(\"%s\")", path);
1940                 if (!(flags & HASH_WRITE_OBJECT))
1941                         hash_sha1_file(sb.buf, sb.len, blob_type, oid->hash);
1942                 else if (write_sha1_file(sb.buf, sb.len, blob_type, oid->hash))
1943                         rc = error("%s: failed to insert into database", path);
1944                 strbuf_release(&sb);
1945                 break;
1946         case S_IFDIR:
1947                 return resolve_gitlink_ref(path, "HEAD", oid);
1948         default:
1949                 return error("%s: unsupported file type", path);
1950         }
1951         return rc;
1952 }
1953
1954 int read_pack_header(int fd, struct pack_header *header)
1955 {
1956         if (read_in_full(fd, header, sizeof(*header)) != sizeof(*header))
1957                 /* "eof before pack header was fully read" */
1958                 return PH_ERROR_EOF;
1959
1960         if (header->hdr_signature != htonl(PACK_SIGNATURE))
1961                 /* "protocol error (pack signature mismatch detected)" */
1962                 return PH_ERROR_PACK_SIGNATURE;
1963         if (!pack_version_ok(header->hdr_version))
1964                 /* "protocol error (pack version unsupported)" */
1965                 return PH_ERROR_PROTOCOL;
1966         return 0;
1967 }
1968
1969 void assert_sha1_type(const unsigned char *sha1, enum object_type expect)
1970 {
1971         enum object_type type = sha1_object_info(sha1, NULL);
1972         if (type < 0)
1973                 die("%s is not a valid object", sha1_to_hex(sha1));
1974         if (type != expect)
1975                 die("%s is not a valid '%s' object", sha1_to_hex(sha1),
1976                     typename(expect));
1977 }
1978
1979 int for_each_file_in_obj_subdir(unsigned int subdir_nr,
1980                                 struct strbuf *path,
1981                                 each_loose_object_fn obj_cb,
1982                                 each_loose_cruft_fn cruft_cb,
1983                                 each_loose_subdir_fn subdir_cb,
1984                                 void *data)
1985 {
1986         size_t origlen, baselen;
1987         DIR *dir;
1988         struct dirent *de;
1989         int r = 0;
1990         struct object_id oid;
1991
1992         if (subdir_nr > 0xff)
1993                 BUG("invalid loose object subdirectory: %x", subdir_nr);
1994
1995         origlen = path->len;
1996         strbuf_complete(path, '/');
1997         strbuf_addf(path, "%02x", subdir_nr);
1998
1999         dir = opendir(path->buf);
2000         if (!dir) {
2001                 if (errno != ENOENT)
2002                         r = error_errno("unable to open %s", path->buf);
2003                 strbuf_setlen(path, origlen);
2004                 return r;
2005         }
2006
2007         oid.hash[0] = subdir_nr;
2008         strbuf_addch(path, '/');
2009         baselen = path->len;
2010
2011         while ((de = readdir(dir))) {
2012                 size_t namelen;
2013                 if (is_dot_or_dotdot(de->d_name))
2014                         continue;
2015
2016                 namelen = strlen(de->d_name);
2017                 strbuf_setlen(path, baselen);
2018                 strbuf_add(path, de->d_name, namelen);
2019                 if (namelen == GIT_SHA1_HEXSZ - 2 &&
2020                     !hex_to_bytes(oid.hash + 1, de->d_name,
2021                                   GIT_SHA1_RAWSZ - 1)) {
2022                         if (obj_cb) {
2023                                 r = obj_cb(&oid, path->buf, data);
2024                                 if (r)
2025                                         break;
2026                         }
2027                         continue;
2028                 }
2029
2030                 if (cruft_cb) {
2031                         r = cruft_cb(de->d_name, path->buf, data);
2032                         if (r)
2033                                 break;
2034                 }
2035         }
2036         closedir(dir);
2037
2038         strbuf_setlen(path, baselen - 1);
2039         if (!r && subdir_cb)
2040                 r = subdir_cb(subdir_nr, path->buf, data);
2041
2042         strbuf_setlen(path, origlen);
2043
2044         return r;
2045 }
2046
2047 int for_each_loose_file_in_objdir_buf(struct strbuf *path,
2048                             each_loose_object_fn obj_cb,
2049                             each_loose_cruft_fn cruft_cb,
2050                             each_loose_subdir_fn subdir_cb,
2051                             void *data)
2052 {
2053         int r = 0;
2054         int i;
2055
2056         for (i = 0; i < 256; i++) {
2057                 r = for_each_file_in_obj_subdir(i, path, obj_cb, cruft_cb,
2058                                                 subdir_cb, data);
2059                 if (r)
2060                         break;
2061         }
2062
2063         return r;
2064 }
2065
2066 int for_each_loose_file_in_objdir(const char *path,
2067                                   each_loose_object_fn obj_cb,
2068                                   each_loose_cruft_fn cruft_cb,
2069                                   each_loose_subdir_fn subdir_cb,
2070                                   void *data)
2071 {
2072         struct strbuf buf = STRBUF_INIT;
2073         int r;
2074
2075         strbuf_addstr(&buf, path);
2076         r = for_each_loose_file_in_objdir_buf(&buf, obj_cb, cruft_cb,
2077                                               subdir_cb, data);
2078         strbuf_release(&buf);
2079
2080         return r;
2081 }
2082
2083 struct loose_alt_odb_data {
2084         each_loose_object_fn *cb;
2085         void *data;
2086 };
2087
2088 static int loose_from_alt_odb(struct alternate_object_database *alt,
2089                               void *vdata)
2090 {
2091         struct loose_alt_odb_data *data = vdata;
2092         struct strbuf buf = STRBUF_INIT;
2093         int r;
2094
2095         strbuf_addstr(&buf, alt->path);
2096         r = for_each_loose_file_in_objdir_buf(&buf,
2097                                               data->cb, NULL, NULL,
2098                                               data->data);
2099         strbuf_release(&buf);
2100         return r;
2101 }
2102
2103 int for_each_loose_object(each_loose_object_fn cb, void *data, unsigned flags)
2104 {
2105         struct loose_alt_odb_data alt;
2106         int r;
2107
2108         r = for_each_loose_file_in_objdir(get_object_directory(),
2109                                           cb, NULL, NULL, data);
2110         if (r)
2111                 return r;
2112
2113         if (flags & FOR_EACH_OBJECT_LOCAL_ONLY)
2114                 return 0;
2115
2116         alt.cb = cb;
2117         alt.data = data;
2118         return foreach_alt_odb(loose_from_alt_odb, &alt);
2119 }
2120
2121 static int check_stream_sha1(git_zstream *stream,
2122                              const char *hdr,
2123                              unsigned long size,
2124                              const char *path,
2125                              const unsigned char *expected_sha1)
2126 {
2127         git_SHA_CTX c;
2128         unsigned char real_sha1[GIT_MAX_RAWSZ];
2129         unsigned char buf[4096];
2130         unsigned long total_read;
2131         int status = Z_OK;
2132
2133         git_SHA1_Init(&c);
2134         git_SHA1_Update(&c, hdr, stream->total_out);
2135
2136         /*
2137          * We already read some bytes into hdr, but the ones up to the NUL
2138          * do not count against the object's content size.
2139          */
2140         total_read = stream->total_out - strlen(hdr) - 1;
2141
2142         /*
2143          * This size comparison must be "<=" to read the final zlib packets;
2144          * see the comment in unpack_sha1_rest for details.
2145          */
2146         while (total_read <= size &&
2147                (status == Z_OK || status == Z_BUF_ERROR)) {
2148                 stream->next_out = buf;
2149                 stream->avail_out = sizeof(buf);
2150                 if (size - total_read < stream->avail_out)
2151                         stream->avail_out = size - total_read;
2152                 status = git_inflate(stream, Z_FINISH);
2153                 git_SHA1_Update(&c, buf, stream->next_out - buf);
2154                 total_read += stream->next_out - buf;
2155         }
2156         git_inflate_end(stream);
2157
2158         if (status != Z_STREAM_END) {
2159                 error("corrupt loose object '%s'", sha1_to_hex(expected_sha1));
2160                 return -1;
2161         }
2162         if (stream->avail_in) {
2163                 error("garbage at end of loose object '%s'",
2164                       sha1_to_hex(expected_sha1));
2165                 return -1;
2166         }
2167
2168         git_SHA1_Final(real_sha1, &c);
2169         if (hashcmp(expected_sha1, real_sha1)) {
2170                 error("sha1 mismatch for %s (expected %s)", path,
2171                       sha1_to_hex(expected_sha1));
2172                 return -1;
2173         }
2174
2175         return 0;
2176 }
2177
2178 int read_loose_object(const char *path,
2179                       const unsigned char *expected_sha1,
2180                       enum object_type *type,
2181                       unsigned long *size,
2182                       void **contents)
2183 {
2184         int ret = -1;
2185         void *map = NULL;
2186         unsigned long mapsize;
2187         git_zstream stream;
2188         char hdr[32];
2189
2190         *contents = NULL;
2191
2192         map = map_sha1_file_1(path, NULL, &mapsize);
2193         if (!map) {
2194                 error_errno("unable to mmap %s", path);
2195                 goto out;
2196         }
2197
2198         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0) {
2199                 error("unable to unpack header of %s", path);
2200                 goto out;
2201         }
2202
2203         *type = parse_sha1_header(hdr, size);
2204         if (*type < 0) {
2205                 error("unable to parse header of %s", path);
2206                 git_inflate_end(&stream);
2207                 goto out;
2208         }
2209
2210         if (*type == OBJ_BLOB) {
2211                 if (check_stream_sha1(&stream, hdr, *size, path, expected_sha1) < 0)
2212                         goto out;
2213         } else {
2214                 *contents = unpack_sha1_rest(&stream, hdr, *size, expected_sha1);
2215                 if (!*contents) {
2216                         error("unable to unpack contents of %s", path);
2217                         git_inflate_end(&stream);
2218                         goto out;
2219                 }
2220                 if (check_sha1_signature(expected_sha1, *contents,
2221                                          *size, typename(*type))) {
2222                         error("sha1 mismatch for %s (expected %s)", path,
2223                               sha1_to_hex(expected_sha1));
2224                         free(*contents);
2225                         goto out;
2226                 }
2227         }
2228
2229         ret = 0; /* everything checks out */
2230
2231 out:
2232         if (map)
2233                 munmap(map, mapsize);
2234         return ret;
2235 }