fsck-objects.c

   1 #include "cache.h"
   2 #include "commit.h"
   3 #include "tree.h"
   4 #include "blob.h"
   5 #include "tag.h"
   6 #include "refs.h"
   7 #include "pack.h"
   8 #include "cache-tree.h"
   9 #include "tree-walk.h"
  10
  11 #define REACHABLE 0x0001
  12 #define SEEN      0x0002
  13
  14 static int show_root;
  15 static int show_tags;
  16 static int show_unreachable;
  17 static int check_full;
  18 static int check_strict;
  19 static int keep_cache_objects;
  20 static unsigned char head_sha1[20];
  21
  22 #ifdef NO_D_INO_IN_DIRENT
  23 #define SORT_DIRENT 0
  24 #define DIRENT_SORT_HINT(de) 0
  25 #else
  26 #define SORT_DIRENT 1
  27 #define DIRENT_SORT_HINT(de) ((de)->d_ino)
  28 #endif
  29
  30 static void objreport(struct object *obj, const char *severity,
  31                       const char *err, va_list params)
  32 {
  33         fprintf(stderr, "%s in %s %s: ",
  34                 severity, typename(obj->type), sha1_to_hex(obj->sha1));
  35         vfprintf(stderr, err, params);
  36         fputs("\n", stderr);
  37 }
  38
  39 static int objerror(struct object *obj, const char *err, ...)
  40 {
  41         va_list params;
  42         va_start(params, err);
  43         objreport(obj, "error", err, params);
  44         va_end(params);
  45         return -1;
  46 }
  47
  48 static int objwarning(struct object *obj, const char *err, ...)
  49 {
  50         va_list params;
  51         va_start(params, err);
  52         objreport(obj, "warning", err, params);
  53         va_end(params);
  54         return -1;
  55 }
  56
  57 /*
  58  * Check a single reachable object
  59  */
  60 static void check_reachable_object(struct object *obj)
  61 {
  62         const struct object_refs *refs;
  63
  64         /*
  65          * We obviously want the object to be parsed,
  66          * except if it was in a pack-file and we didn't
  67          * do a full fsck
  68          */
  69         if (!obj->parsed) {
  70                 if (has_sha1_file(obj->sha1))
  71                         return; /* it is in pack - forget about it */
  72                 printf("missing %s %s\n", typename(obj->type), sha1_to_hex(obj->sha1));
  73                 return;
  74         }
  75
  76         /*
  77          * Check that everything that we try to reference is also good.
  78          */
  79         refs = lookup_object_refs(obj);
  80         if (refs) {
  81                 unsigned j;
  82                 for (j = 0; j < refs->count; j++) {
  83                         struct object *ref = refs->ref[j];
  84                         if (ref->parsed ||
  85                             (has_sha1_file(ref->sha1)))
  86                                 continue;
  87                         printf("broken link from %7s %s\n",
  88                                typename(obj->type), sha1_to_hex(obj->sha1));
  89                         printf("              to %7s %s\n",
  90                                typename(ref->type), sha1_to_hex(ref->sha1));
  91                 }
  92         }
  93 }
  94
  95 /*
  96  * Check a single unreachable object
  97  */
  98 static void check_unreachable_object(struct object *obj)
  99 {
 100         /*
 101          * Missing unreachable object? Ignore it. It's not like
 102          * we miss it (since it can't be reached), nor do we want
 103          * to complain about it being unreachable (since it does
 104          * not exist).
 105          */
 106         if (!obj->parsed)
 107                 return;
 108
 109         /*
 110          * Unreachable object that exists? Show it if asked to,
 111          * since this is something that is prunable.
 112          */
 113         if (show_unreachable) {
 114                 printf("unreachable %s %s\n", typename(obj->type), sha1_to_hex(obj->sha1));
 115                 return;
 116         }
 117
 118         /*
 119          * "!used" means that nothing at all points to it, including
 120          * other unreacahble objects. In other words, it's the "tip"
 121          * of some set of unreachable objects, usually a commit that
 122          * got dropped.
 123          *
 124          * Such starting points are more interesting than some random
 125          * set of unreachable objects, so we show them even if the user
 126          * hasn't asked for _all_ unreachable objects. If you have
 127          * deleted a branch by mistake, this is a prime candidate to
 128          * start looking at, for example.
 129          */
 130         if (!obj->used) {
 131                 printf("dangling %s %s\n", typename(obj->type),
 132                        sha1_to_hex(obj->sha1));
 133                 return;
 134         }
 135
 136         /*
 137          * Otherwise? It's there, it's unreachable, and some other unreachable
 138          * object points to it. Ignore it - it's not interesting, and we showed
 139          * all the interesting cases above.
 140          */
 141 }
 142
 143 static void check_object(struct object *obj)
 144 {
 145         if (obj->flags & REACHABLE)
 146                 check_reachable_object(obj);
 147         else
 148                 check_unreachable_object(obj);
 149 }
 150
 151 static void check_connectivity(void)
 152 {
 153         int i, max;
 154
 155         /* Look up all the requirements, warn about missing objects.. */
 156         max = get_max_object_index();
 157         for (i = 0; i < max; i++) {
 158                 struct object *obj = get_indexed_object(i);
 159
 160                 if (obj)
 161                         check_object(obj);
 162         }
 163 }
 164
 165 /*
 166  * The entries in a tree are ordered in the _path_ order,
 167  * which means that a directory entry is ordered by adding
 168  * a slash to the end of it.
 169  *
 170  * So a directory called "a" is ordered _after_ a file
 171  * called "a.c", because "a/" sorts after "a.c".
 172  */
 173 #define TREE_UNORDERED (-1)
 174 #define TREE_HAS_DUPS  (-2)
 175
 176 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 177 {
 178         int len1 = strlen(name1);
 179         int len2 = strlen(name2);
 180         int len = len1 < len2 ? len1 : len2;
 181         unsigned char c1, c2;
 182         int cmp;
 183
 184         cmp = memcmp(name1, name2, len);
 185         if (cmp < 0)
 186                 return 0;
 187         if (cmp > 0)
 188                 return TREE_UNORDERED;
 189
 190         /*
 191          * Ok, the first <len> characters are the same.
 192          * Now we need to order the next one, but turn
 193          * a '\0' into a '/' for a directory entry.
 194          */
 195         c1 = name1[len];
 196         c2 = name2[len];
 197         if (!c1 && !c2)
 198                 /*
 199                  * git-write-tree used to write out a nonsense tree that has
 200                  * entries with the same name, one blob and one tree.  Make
 201                  * sure we do not have duplicate entries.
 202                  */
 203                 return TREE_HAS_DUPS;
 204         if (!c1 && S_ISDIR(mode1))
 205                 c1 = '/';
 206         if (!c2 && S_ISDIR(mode2))
 207                 c2 = '/';
 208         return c1 < c2 ? 0 : TREE_UNORDERED;
 209 }
 210
 211 static int fsck_tree(struct tree *item)
 212 {
 213         int retval;
 214         int has_full_path = 0;
 215         int has_zero_pad = 0;
 216         int has_bad_modes = 0;
 217         int has_dup_entries = 0;
 218         int not_properly_sorted = 0;
 219         struct tree_desc desc;
 220         unsigned o_mode;
 221         const char *o_name;
 222         const unsigned char *o_sha1;
 223
 224         desc.buf = item->buffer;
 225         desc.size = item->size;
 226
 227         o_mode = 0;
 228         o_name = NULL;
 229         o_sha1 = NULL;
 230         while (desc.size) {
 231                 unsigned mode;
 232                 const char *name;
 233                 const unsigned char *sha1;
 234
 235                 sha1 = tree_entry_extract(&desc, &name, &mode);
 236
 237                 if (strchr(name, '/'))
 238                         has_full_path = 1;
 239                 has_zero_pad |= *(char *)desc.buf == '0';
 240                 update_tree_entry(&desc);
 241
 242                 switch (mode) {
 243                 /*
 244                  * Standard modes..
 245                  */
 246                 case S_IFREG | 0755:
 247                 case S_IFREG | 0644:
 248                 case S_IFLNK:
 249                 case S_IFDIR:
 250                         break;
 251                 /*
 252                  * This is nonstandard, but we had a few of these
 253                  * early on when we honored the full set of mode
 254                  * bits..
 255                  */
 256                 case S_IFREG | 0664:
 257                         if (!check_strict)
 258                                 break;
 259                 default:
 260                         has_bad_modes = 1;
 261                 }
 262
 263                 if (o_name) {
 264                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 265                         case TREE_UNORDERED:
 266                                 not_properly_sorted = 1;
 267                                 break;
 268                         case TREE_HAS_DUPS:
 269                                 has_dup_entries = 1;
 270                                 break;
 271                         default:
 272                                 break;
 273                         }
 274                 }
 275
 276                 o_mode = mode;
 277                 o_name = name;
 278                 o_sha1 = sha1;
 279         }
 280         free(item->buffer);
 281         item->buffer = NULL;
 282
 283         retval = 0;
 284         if (has_full_path) {
 285                 objwarning(&item->object, "contains full pathnames");
 286         }
 287         if (has_zero_pad) {
 288                 objwarning(&item->object, "contains zero-padded file modes");
 289         }
 290         if (has_bad_modes) {
 291                 objwarning(&item->object, "contains bad file modes");
 292         }
 293         if (has_dup_entries) {
 294                 retval = objerror(&item->object, "contains duplicate file entries");
 295         }
 296         if (not_properly_sorted) {
 297                 retval = objerror(&item->object, "not properly sorted");
 298         }
 299         return retval;
 300 }
 301
 302 static int fsck_commit(struct commit *commit)
 303 {
 304         char *buffer = commit->buffer;
 305         unsigned char tree_sha1[20], sha1[20];
 306
 307         if (memcmp(buffer, "tree ", 5))
 308                 return objerror(&commit->object, "invalid format - expected 'tree' line");
 309         if (get_sha1_hex(buffer+5, tree_sha1) || buffer[45] != '\n')
 310                 return objerror(&commit->object, "invalid 'tree' line format - bad sha1");
 311         buffer += 46;
 312         while (!memcmp(buffer, "parent ", 7)) {
 313                 if (get_sha1_hex(buffer+7, sha1) || buffer[47] != '\n')
 314                         return objerror(&commit->object, "invalid 'parent' line format - bad sha1");
 315                 buffer += 48;
 316         }
 317         if (memcmp(buffer, "author ", 7))
 318                 return objerror(&commit->object, "invalid format - expected 'author' line");
 319         free(commit->buffer);
 320         commit->buffer = NULL;
 321         if (!commit->tree)
 322                 return objerror(&commit->object, "could not load commit's tree %s", tree_sha1);
 323         if (!commit->parents && show_root)
 324                 printf("root %s\n", sha1_to_hex(commit->object.sha1));
 325         if (!commit->date)
 326                 printf("bad commit date in %s\n",
 327                        sha1_to_hex(commit->object.sha1));
 328         return 0;
 329 }
 330
 331 static int fsck_tag(struct tag *tag)
 332 {
 333         struct object *tagged = tag->tagged;
 334
 335         if (!tagged) {
 336                 return objerror(&tag->object, "could not load tagged object");
 337         }
 338         if (!show_tags)
 339                 return 0;
 340
 341         printf("tagged %s %s", typename(tagged->type), sha1_to_hex(tagged->sha1));
 342         printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
 343         return 0;
 344 }
 345
 346 static int fsck_sha1(unsigned char *sha1)
 347 {
 348         struct object *obj = parse_object(sha1);
 349         if (!obj)
 350                 return error("%s: object corrupt or missing", sha1_to_hex(sha1));
 351         if (obj->flags & SEEN)
 352                 return 0;
 353         obj->flags |= SEEN;
 354         if (obj->type == OBJ_BLOB)
 355                 return 0;
 356         if (obj->type == OBJ_TREE)
 357                 return fsck_tree((struct tree *) obj);
 358         if (obj->type == OBJ_COMMIT)
 359                 return fsck_commit((struct commit *) obj);
 360         if (obj->type == OBJ_TAG)
 361                 return fsck_tag((struct tag *) obj);
 362         /* By now, parse_object() would've returned NULL instead. */
 363         return objerror(obj, "unknown type '%d' (internal fsck error)", obj->type);
 364 }
 365
 366 /*
 367  * This is the sorting chunk size: make it reasonably
 368  * big so that we can sort well..
 369  */
 370 #define MAX_SHA1_ENTRIES (1024)
 371
 372 struct sha1_entry {
 373         unsigned long ino;
 374         unsigned char sha1[20];
 375 };
 376
 377 static struct {
 378         unsigned long nr;
 379         struct sha1_entry *entry[MAX_SHA1_ENTRIES];
 380 } sha1_list;
 381
 382 static int ino_compare(const void *_a, const void *_b)
 383 {
 384         const struct sha1_entry *a = _a, *b = _b;
 385         unsigned long ino1 = a->ino, ino2 = b->ino;
 386         return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0;
 387 }
 388
 389 static void fsck_sha1_list(void)
 390 {
 391         int i, nr = sha1_list.nr;
 392
 393         if (SORT_DIRENT)
 394                 qsort(sha1_list.entry, nr,
 395                       sizeof(struct sha1_entry *), ino_compare);
 396         for (i = 0; i < nr; i++) {
 397                 struct sha1_entry *entry = sha1_list.entry[i];
 398                 unsigned char *sha1 = entry->sha1;
 399
 400                 sha1_list.entry[i] = NULL;
 401                 fsck_sha1(sha1);
 402                 free(entry);
 403         }
 404         sha1_list.nr = 0;
 405 }
 406
 407 static void add_sha1_list(unsigned char *sha1, unsigned long ino)
 408 {
 409         struct sha1_entry *entry = xmalloc(sizeof(*entry));
 410         int nr;
 411
 412         entry->ino = ino;
 413         hashcpy(entry->sha1, sha1);
 414         nr = sha1_list.nr;
 415         if (nr == MAX_SHA1_ENTRIES) {
 416                 fsck_sha1_list();
 417                 nr = 0;
 418         }
 419         sha1_list.entry[nr] = entry;
 420         sha1_list.nr = ++nr;
 421 }
 422
 423 static void fsck_dir(int i, char *path)
 424 {
 425         DIR *dir = opendir(path);
 426         struct dirent *de;
 427
 428         if (!dir)
 429                 return;
 430
 431         while ((de = readdir(dir)) != NULL) {
 432                 char name[100];
 433                 unsigned char sha1[20];
 434                 int len = strlen(de->d_name);
 435
 436                 switch (len) {
 437                 case 2:
 438                         if (de->d_name[1] != '.')
 439                                 break;
 440                 case 1:
 441                         if (de->d_name[0] != '.')
 442                                 break;
 443                         continue;
 444                 case 38:
 445                         sprintf(name, "%02x", i);
 446                         memcpy(name+2, de->d_name, len+1);
 447                         if (get_sha1_hex(name, sha1) < 0)
 448                                 break;
 449                         add_sha1_list(sha1, DIRENT_SORT_HINT(de));
 450                         continue;
 451                 }
 452                 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
 453         }
 454         closedir(dir);
 455 }
 456
 457 static int default_refs;
 458
 459 static int fsck_handle_reflog_ent(unsigned char *osha1, unsigned char *nsha1,
 460                 const char *email, unsigned long timestamp, int tz,
 461                 const char *message, void *cb_data)
 462 {
 463         struct object *obj;
 464
 465         if (!is_null_sha1(osha1)) {
 466                 obj = lookup_object(osha1);
 467                 if (obj) {
 468                         obj->used = 1;
 469                         mark_reachable(obj, REACHABLE);
 470                 }
 471         }
 472         obj = lookup_object(nsha1);
 473         if (obj) {
 474                 obj->used = 1;
 475                 mark_reachable(obj, REACHABLE);
 476         }
 477         return 0;
 478 }
 479
 480 static int fsck_handle_ref(const char *refname, const unsigned char *sha1, int flag, void *cb_data)
 481 {
 482         struct object *obj;
 483
 484         obj = lookup_object(sha1);
 485         if (!obj) {
 486                 if (has_sha1_file(sha1)) {
 487                         default_refs++;
 488                         return 0; /* it is in a pack */
 489                 }
 490                 error("%s: invalid sha1 pointer %s", refname, sha1_to_hex(sha1));
 491                 /* We'll continue with the rest despite the error.. */
 492                 return 0;
 493         }
 494         default_refs++;
 495         obj->used = 1;
 496         mark_reachable(obj, REACHABLE);
 497
 498         for_each_reflog_ent(refname, fsck_handle_reflog_ent, NULL);
 499
 500         return 0;
 501 }
 502
 503 static void get_default_heads(void)
 504 {
 505         for_each_ref(fsck_handle_ref, NULL);
 506
 507         /*
 508          * Not having any default heads isn't really fatal, but
 509          * it does mean that "--unreachable" no longer makes any
 510          * sense (since in this case everything will obviously
 511          * be unreachable by definition.
 512          *
 513          * Showing dangling objects is valid, though (as those
 514          * dangling objects are likely lost heads).
 515          *
 516          * So we just print a warning about it, and clear the
 517          * "show_unreachable" flag.
 518          */
 519         if (!default_refs) {
 520                 error("No default references");
 521                 show_unreachable = 0;
 522         }
 523 }
 524
 525 static void fsck_object_dir(const char *path)
 526 {
 527         int i;
 528         for (i = 0; i < 256; i++) {
 529                 static char dir[4096];
 530                 sprintf(dir, "%s/%02x", path, i);
 531                 fsck_dir(i, dir);
 532         }
 533         fsck_sha1_list();
 534 }
 535
 536 static int fsck_head_link(void)
 537 {
 538         unsigned char sha1[20];
 539         int flag;
 540         const char *head_points_at = resolve_ref("HEAD", sha1, 1, &flag);
 541
 542         if (!head_points_at || !(flag & REF_ISSYMREF))
 543                 return error("HEAD is not a symbolic ref");
 544         if (strncmp(head_points_at, "refs/heads/", 11))
 545                 return error("HEAD points to something strange (%s)",
 546                              head_points_at);
 547         if (is_null_sha1(sha1))
 548                 return error("HEAD: not a valid git pointer");
 549         return 0;
 550 }
 551
 552 static int fsck_cache_tree(struct cache_tree *it)
 553 {
 554         int i;
 555         int err = 0;
 556
 557         if (0 <= it->entry_count) {
 558                 struct object *obj = parse_object(it->sha1);
 559                 if (!obj) {
 560                         error("%s: invalid sha1 pointer in cache-tree",
 561                               sha1_to_hex(it->sha1));
 562                         return 1;
 563                 }
 564                 mark_reachable(obj, REACHABLE);
 565                 obj->used = 1;
 566                 if (obj->type != OBJ_TREE)
 567                         err |= objerror(obj, "non-tree in cache-tree");
 568         }
 569         for (i = 0; i < it->subtree_nr; i++)
 570                 err |= fsck_cache_tree(it->down[i]->cache_tree);
 571         return err;
 572 }
 573
 574 int main(int argc, char **argv)
 575 {
 576         int i, heads;
 577
 578         track_object_refs = 1;
 579         setup_git_directory();
 580
 581         for (i = 1; i < argc; i++) {
 582                 const char *arg = argv[i];
 583
 584                 if (!strcmp(arg, "--unreachable")) {
 585                         show_unreachable = 1;
 586                         continue;
 587                 }
 588                 if (!strcmp(arg, "--tags")) {
 589                         show_tags = 1;
 590                         continue;
 591                 }
 592                 if (!strcmp(arg, "--root")) {
 593                         show_root = 1;
 594                         continue;
 595                 }
 596                 if (!strcmp(arg, "--cache")) {
 597                         keep_cache_objects = 1;
 598                         continue;
 599                 }
 600                 if (!strcmp(arg, "--full")) {
 601                         check_full = 1;
 602                         continue;
 603                 }
 604                 if (!strcmp(arg, "--strict")) {
 605                         check_strict = 1;
 606                         continue;
 607                 }
 608                 if (*arg == '-')
 609                         usage("git-fsck-objects [--tags] [--root] [[--unreachable] [--cache] [--full] [--strict] <head-sha1>*]");
 610         }
 611
 612         fsck_head_link();
 613         fsck_object_dir(get_object_directory());
 614         if (check_full) {
 615                 struct alternate_object_database *alt;
 616                 struct packed_git *p;
 617                 prepare_alt_odb();
 618                 for (alt = alt_odb_list; alt; alt = alt->next) {
 619                         char namebuf[PATH_MAX];
 620                         int namelen = alt->name - alt->base;
 621                         memcpy(namebuf, alt->base, namelen);
 622                         namebuf[namelen - 1] = 0;
 623                         fsck_object_dir(namebuf);
 624                 }
 625                 prepare_packed_git();
 626                 for (p = packed_git; p; p = p->next)
 627                         /* verify gives error messages itself */
 628                         verify_pack(p, 0);
 629
 630                 for (p = packed_git; p; p = p->next) {
 631                         int num = num_packed_objects(p);
 632                         for (i = 0; i < num; i++) {
 633                                 unsigned char sha1[20];
 634                                 nth_packed_object_sha1(p, i, sha1);
 635                                 fsck_sha1(sha1);
 636                         }
 637                 }
 638         }
 639
 640         heads = 0;
 641         for (i = 1; i < argc; i++) {
 642                 const char *arg = argv[i];
 643
 644                 if (*arg == '-')
 645                         continue;
 646
 647                 if (!get_sha1(arg, head_sha1)) {
 648                         struct object *obj = lookup_object(head_sha1);
 649
 650                         /* Error is printed by lookup_object(). */
 651                         if (!obj)
 652                                 continue;
 653
 654                         obj->used = 1;
 655                         mark_reachable(obj, REACHABLE);
 656                         heads++;
 657                         continue;
 658                 }
 659                 error("invalid parameter: expected sha1, got '%s'", arg);
 660         }
 661
 662         /*
 663          * If we've not been given any explicit head information, do the
 664          * default ones from .git/refs. We also consider the index file
 665          * in this case (ie this implies --cache).
 666          */
 667         if (!heads) {
 668                 get_default_heads();
 669                 keep_cache_objects = 1;
 670         }
 671
 672         if (keep_cache_objects) {
 673                 int i;
 674                 read_cache();
 675                 for (i = 0; i < active_nr; i++) {
 676                         struct blob *blob = lookup_blob(active_cache[i]->sha1);
 677                         struct object *obj;
 678                         if (!blob)
 679                                 continue;
 680                         obj = &blob->object;
 681                         obj->used = 1;
 682                         mark_reachable(obj, REACHABLE);
 683                 }
 684                 if (active_cache_tree)
 685                         fsck_cache_tree(active_cache_tree);
 686         }
 687
 688         check_connectivity();
 689         return 0;
 690 }