builtin-fsck.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 static int errors_found;
  22 #define ERROR_OBJECT 01
  23 #define ERROR_REACHABLE 02
  24
  25 #ifdef NO_D_INO_IN_DIRENT
  26 #define SORT_DIRENT 0
  27 #define DIRENT_SORT_HINT(de) 0
  28 #else
  29 #define SORT_DIRENT 1
  30 #define DIRENT_SORT_HINT(de) ((de)->d_ino)
  31 #endif
  32
  33 static void objreport(struct object *obj, const char *severity,
  34                       const char *err, va_list params)
  35 {
  36         fprintf(stderr, "%s in %s %s: ",
  37                 severity, typename(obj->type), sha1_to_hex(obj->sha1));
  38         vfprintf(stderr, err, params);
  39         fputs("\n", stderr);
  40 }
  41
  42 static int objerror(struct object *obj, const char *err, ...)
  43 {
  44         va_list params;
  45         va_start(params, err);
  46         errors_found |= ERROR_OBJECT;
  47         objreport(obj, "error", err, params);
  48         va_end(params);
  49         return -1;
  50 }
  51
  52 static int objwarning(struct object *obj, const char *err, ...)
  53 {
  54         va_list params;
  55         va_start(params, err);
  56         objreport(obj, "warning", err, params);
  57         va_end(params);
  58         return -1;
  59 }
  60
  61 /*
  62  * Check a single reachable object
  63  */
  64 static void check_reachable_object(struct object *obj)
  65 {
  66         const struct object_refs *refs;
  67
  68         /*
  69          * We obviously want the object to be parsed,
  70          * except if it was in a pack-file and we didn't
  71          * do a full fsck
  72          */
  73         if (!obj->parsed) {
  74                 if (has_sha1_pack(obj->sha1, NULL))
  75                         return; /* it is in pack - forget about it */
  76                 printf("missing %s %s\n", typename(obj->type), sha1_to_hex(obj->sha1));
  77                 errors_found |= ERROR_REACHABLE;
  78                 return;
  79         }
  80
  81         /*
  82          * Check that everything that we try to reference is also good.
  83          */
  84         refs = lookup_object_refs(obj);
  85         if (refs) {
  86                 unsigned j;
  87                 for (j = 0; j < refs->count; j++) {
  88                         struct object *ref = refs->ref[j];
  89                         if (ref->parsed ||
  90                             (has_sha1_file(ref->sha1)))
  91                                 continue;
  92                         printf("broken link from %7s %s\n",
  93                                typename(obj->type), sha1_to_hex(obj->sha1));
  94                         printf("              to %7s %s\n",
  95                                typename(ref->type), sha1_to_hex(ref->sha1));
  96                         errors_found |= ERROR_REACHABLE;
  97                 }
  98         }
  99 }
 100
 101 /*
 102  * Check a single unreachable object
 103  */
 104 static void check_unreachable_object(struct object *obj)
 105 {
 106         /*
 107          * Missing unreachable object? Ignore it. It's not like
 108          * we miss it (since it can't be reached), nor do we want
 109          * to complain about it being unreachable (since it does
 110          * not exist).
 111          */
 112         if (!obj->parsed)
 113                 return;
 114
 115         /*
 116          * Unreachable object that exists? Show it if asked to,
 117          * since this is something that is prunable.
 118          */
 119         if (show_unreachable) {
 120                 printf("unreachable %s %s\n", typename(obj->type), sha1_to_hex(obj->sha1));
 121                 return;
 122         }
 123
 124         /*
 125          * "!used" means that nothing at all points to it, including
 126          * other unreachable objects. In other words, it's the "tip"
 127          * of some set of unreachable objects, usually a commit that
 128          * got dropped.
 129          *
 130          * Such starting points are more interesting than some random
 131          * set of unreachable objects, so we show them even if the user
 132          * hasn't asked for _all_ unreachable objects. If you have
 133          * deleted a branch by mistake, this is a prime candidate to
 134          * start looking at, for example.
 135          */
 136         if (!obj->used) {
 137                 printf("dangling %s %s\n", typename(obj->type),
 138                        sha1_to_hex(obj->sha1));
 139                 return;
 140         }
 141
 142         /*
 143          * Otherwise? It's there, it's unreachable, and some other unreachable
 144          * object points to it. Ignore it - it's not interesting, and we showed
 145          * all the interesting cases above.
 146          */
 147 }
 148
 149 static void check_object(struct object *obj)
 150 {
 151         if (obj->flags & REACHABLE)
 152                 check_reachable_object(obj);
 153         else
 154                 check_unreachable_object(obj);
 155 }
 156
 157 static void check_connectivity(void)
 158 {
 159         int i, max;
 160
 161         /* Look up all the requirements, warn about missing objects.. */
 162         max = get_max_object_index();
 163         for (i = 0; i < max; i++) {
 164                 struct object *obj = get_indexed_object(i);
 165
 166                 if (obj)
 167                         check_object(obj);
 168         }
 169 }
 170
 171 /*
 172  * The entries in a tree are ordered in the _path_ order,
 173  * which means that a directory entry is ordered by adding
 174  * a slash to the end of it.
 175  *
 176  * So a directory called "a" is ordered _after_ a file
 177  * called "a.c", because "a/" sorts after "a.c".
 178  */
 179 #define TREE_UNORDERED (-1)
 180 #define TREE_HAS_DUPS  (-2)
 181
 182 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 183 {
 184         int len1 = strlen(name1);
 185         int len2 = strlen(name2);
 186         int len = len1 < len2 ? len1 : len2;
 187         unsigned char c1, c2;
 188         int cmp;
 189
 190         cmp = memcmp(name1, name2, len);
 191         if (cmp < 0)
 192                 return 0;
 193         if (cmp > 0)
 194                 return TREE_UNORDERED;
 195
 196         /*
 197          * Ok, the first <len> characters are the same.
 198          * Now we need to order the next one, but turn
 199          * a '\0' into a '/' for a directory entry.
 200          */
 201         c1 = name1[len];
 202         c2 = name2[len];
 203         if (!c1 && !c2)
 204                 /*
 205                  * git-write-tree used to write out a nonsense tree that has
 206                  * entries with the same name, one blob and one tree.  Make
 207                  * sure we do not have duplicate entries.
 208                  */
 209                 return TREE_HAS_DUPS;
 210         if (!c1 && S_ISDIR(mode1))
 211                 c1 = '/';
 212         if (!c2 && S_ISDIR(mode2))
 213                 c2 = '/';
 214         return c1 < c2 ? 0 : TREE_UNORDERED;
 215 }
 216
 217 static int fsck_tree(struct tree *item)
 218 {
 219         int retval;
 220         int has_full_path = 0;
 221         int has_zero_pad = 0;
 222         int has_bad_modes = 0;
 223         int has_dup_entries = 0;
 224         int not_properly_sorted = 0;
 225         struct tree_desc desc;
 226         unsigned o_mode;
 227         const char *o_name;
 228         const unsigned char *o_sha1;
 229
 230         desc.buf = item->buffer;
 231         desc.size = item->size;
 232
 233         o_mode = 0;
 234         o_name = NULL;
 235         o_sha1 = NULL;
 236         while (desc.size) {
 237                 unsigned mode;
 238                 const char *name;
 239                 const unsigned char *sha1;
 240
 241                 sha1 = tree_entry_extract(&desc, &name, &mode);
 242
 243                 if (strchr(name, '/'))
 244                         has_full_path = 1;
 245                 has_zero_pad |= *(char *)desc.buf == '0';
 246                 update_tree_entry(&desc);
 247
 248                 switch (mode) {
 249                 /*
 250                  * Standard modes..
 251                  */
 252                 case S_IFREG | 0755:
 253                 case S_IFREG | 0644:
 254                 case S_IFLNK:
 255                 case S_IFDIR:
 256                         break;
 257                 /*
 258                  * This is nonstandard, but we had a few of these
 259                  * early on when we honored the full set of mode
 260                  * bits..
 261                  */
 262                 case S_IFREG | 0664:
 263                         if (!check_strict)
 264                                 break;
 265                 default:
 266                         has_bad_modes = 1;
 267                 }
 268
 269                 if (o_name) {
 270                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 271                         case TREE_UNORDERED:
 272                                 not_properly_sorted = 1;
 273                                 break;
 274                         case TREE_HAS_DUPS:
 275                                 has_dup_entries = 1;
 276                                 break;
 277                         default:
 278                                 break;
 279                         }
 280                 }
 281
 282                 o_mode = mode;
 283                 o_name = name;
 284                 o_sha1 = sha1;
 285         }
 286         free(item->buffer);
 287         item->buffer = NULL;
 288
 289         retval = 0;
 290         if (has_full_path) {
 291                 objwarning(&item->object, "contains full pathnames");
 292         }
 293         if (has_zero_pad) {
 294                 objwarning(&item->object, "contains zero-padded file modes");
 295         }
 296         if (has_bad_modes) {
 297                 objwarning(&item->object, "contains bad file modes");
 298         }
 299         if (has_dup_entries) {
 300                 retval = objerror(&item->object, "contains duplicate file entries");
 301         }
 302         if (not_properly_sorted) {
 303                 retval = objerror(&item->object, "not properly sorted");
 304         }
 305         return retval;
 306 }
 307
 308 static int fsck_commit(struct commit *commit)
 309 {
 310         char *buffer = commit->buffer;
 311         unsigned char tree_sha1[20], sha1[20];
 312
 313         if (memcmp(buffer, "tree ", 5))
 314                 return objerror(&commit->object, "invalid format - expected 'tree' line");
 315         if (get_sha1_hex(buffer+5, tree_sha1) || buffer[45] != '\n')
 316                 return objerror(&commit->object, "invalid 'tree' line format - bad sha1");
 317         buffer += 46;
 318         while (!memcmp(buffer, "parent ", 7)) {
 319                 if (get_sha1_hex(buffer+7, sha1) || buffer[47] != '\n')
 320                         return objerror(&commit->object, "invalid 'parent' line format - bad sha1");
 321                 buffer += 48;
 322         }
 323         if (memcmp(buffer, "author ", 7))
 324                 return objerror(&commit->object, "invalid format - expected 'author' line");
 325         free(commit->buffer);
 326         commit->buffer = NULL;
 327         if (!commit->tree)
 328                 return objerror(&commit->object, "could not load commit's tree %s", tree_sha1);
 329         if (!commit->parents && show_root)
 330                 printf("root %s\n", sha1_to_hex(commit->object.sha1));
 331         if (!commit->date)
 332                 printf("bad commit date in %s\n",
 333                        sha1_to_hex(commit->object.sha1));
 334         return 0;
 335 }
 336
 337 static int fsck_tag(struct tag *tag)
 338 {
 339         struct object *tagged = tag->tagged;
 340
 341         if (!tagged) {
 342                 return objerror(&tag->object, "could not load tagged object");
 343         }
 344         if (!show_tags)
 345                 return 0;
 346
 347         printf("tagged %s %s", typename(tagged->type), sha1_to_hex(tagged->sha1));
 348         printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
 349         return 0;
 350 }
 351
 352 static int fsck_sha1(unsigned char *sha1)
 353 {
 354         struct object *obj = parse_object(sha1);
 355         if (!obj) {
 356                 errors_found |= ERROR_OBJECT;
 357                 return error("%s: object corrupt or missing",
 358                              sha1_to_hex(sha1));
 359         }
 360         if (obj->flags & SEEN)
 361                 return 0;
 362         obj->flags |= SEEN;
 363         if (obj->type == OBJ_BLOB)
 364                 return 0;
 365         if (obj->type == OBJ_TREE)
 366                 return fsck_tree((struct tree *) obj);
 367         if (obj->type == OBJ_COMMIT)
 368                 return fsck_commit((struct commit *) obj);
 369         if (obj->type == OBJ_TAG)
 370                 return fsck_tag((struct tag *) obj);
 371
 372         /* By now, parse_object() would've returned NULL instead. */
 373         return objerror(obj, "unknown type '%d' (internal fsck error)",
 374                         obj->type);
 375 }
 376
 377 /*
 378  * This is the sorting chunk size: make it reasonably
 379  * big so that we can sort well..
 380  */
 381 #define MAX_SHA1_ENTRIES (1024)
 382
 383 struct sha1_entry {
 384         unsigned long ino;
 385         unsigned char sha1[20];
 386 };
 387
 388 static struct {
 389         unsigned long nr;
 390         struct sha1_entry *entry[MAX_SHA1_ENTRIES];
 391 } sha1_list;
 392
 393 static int ino_compare(const void *_a, const void *_b)
 394 {
 395         const struct sha1_entry *a = _a, *b = _b;
 396         unsigned long ino1 = a->ino, ino2 = b->ino;
 397         return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0;
 398 }
 399
 400 static void fsck_sha1_list(void)
 401 {
 402         int i, nr = sha1_list.nr;
 403
 404         if (SORT_DIRENT)
 405                 qsort(sha1_list.entry, nr,
 406                       sizeof(struct sha1_entry *), ino_compare);
 407         for (i = 0; i < nr; i++) {
 408                 struct sha1_entry *entry = sha1_list.entry[i];
 409                 unsigned char *sha1 = entry->sha1;
 410
 411                 sha1_list.entry[i] = NULL;
 412                 fsck_sha1(sha1);
 413                 free(entry);
 414         }
 415         sha1_list.nr = 0;
 416 }
 417
 418 static void add_sha1_list(unsigned char *sha1, unsigned long ino)
 419 {
 420         struct sha1_entry *entry = xmalloc(sizeof(*entry));
 421         int nr;
 422
 423         entry->ino = ino;
 424         hashcpy(entry->sha1, sha1);
 425         nr = sha1_list.nr;
 426         if (nr == MAX_SHA1_ENTRIES) {
 427                 fsck_sha1_list();
 428                 nr = 0;
 429         }
 430         sha1_list.entry[nr] = entry;
 431         sha1_list.nr = ++nr;
 432 }
 433
 434 static void fsck_dir(int i, char *path)
 435 {
 436         DIR *dir = opendir(path);
 437         struct dirent *de;
 438
 439         if (!dir)
 440                 return;
 441
 442         while ((de = readdir(dir)) != NULL) {
 443                 char name[100];
 444                 unsigned char sha1[20];
 445                 int len = strlen(de->d_name);
 446
 447                 switch (len) {
 448                 case 2:
 449                         if (de->d_name[1] != '.')
 450                                 break;
 451                 case 1:
 452                         if (de->d_name[0] != '.')
 453                                 break;
 454                         continue;
 455                 case 38:
 456                         sprintf(name, "%02x", i);
 457                         memcpy(name+2, de->d_name, len+1);
 458                         if (get_sha1_hex(name, sha1) < 0)
 459                                 break;
 460                         add_sha1_list(sha1, DIRENT_SORT_HINT(de));
 461                         continue;
 462                 }
 463                 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
 464         }
 465         closedir(dir);
 466 }
 467
 468 static int default_refs;
 469
 470 static int fsck_handle_reflog_ent(unsigned char *osha1, unsigned char *nsha1,
 471                 const char *email, unsigned long timestamp, int tz,
 472                 const char *message, void *cb_data)
 473 {
 474         struct object *obj;
 475
 476         if (!is_null_sha1(osha1)) {
 477                 obj = lookup_object(osha1);
 478                 if (obj) {
 479                         obj->used = 1;
 480                         mark_reachable(obj, REACHABLE);
 481                 }
 482         }
 483         obj = lookup_object(nsha1);
 484         if (obj) {
 485                 obj->used = 1;
 486                 mark_reachable(obj, REACHABLE);
 487         }
 488         return 0;
 489 }
 490
 491 static int fsck_handle_reflog(const char *logname, const unsigned char *sha1, int flag, void *cb_data)
 492 {
 493         for_each_reflog_ent(logname, fsck_handle_reflog_ent, NULL);
 494         return 0;
 495 }
 496
 497 static int fsck_handle_ref(const char *refname, const unsigned char *sha1, int flag, void *cb_data)
 498 {
 499         struct object *obj;
 500
 501         obj = lookup_object(sha1);
 502         if (!obj) {
 503                 if (has_sha1_file(sha1)) {
 504                         default_refs++;
 505                         return 0; /* it is in a pack */
 506                 }
 507                 error("%s: invalid sha1 pointer %s", refname, sha1_to_hex(sha1));
 508                 /* We'll continue with the rest despite the error.. */
 509                 return 0;
 510         }
 511         default_refs++;
 512         obj->used = 1;
 513         mark_reachable(obj, REACHABLE);
 514
 515         return 0;
 516 }
 517
 518 static void get_default_heads(void)
 519 {
 520         for_each_ref(fsck_handle_ref, NULL);
 521         for_each_reflog(fsck_handle_reflog, NULL);
 522
 523         /*
 524          * Not having any default heads isn't really fatal, but
 525          * it does mean that "--unreachable" no longer makes any
 526          * sense (since in this case everything will obviously
 527          * be unreachable by definition.
 528          *
 529          * Showing dangling objects is valid, though (as those
 530          * dangling objects are likely lost heads).
 531          *
 532          * So we just print a warning about it, and clear the
 533          * "show_unreachable" flag.
 534          */
 535         if (!default_refs) {
 536                 error("No default references");
 537                 show_unreachable = 0;
 538         }
 539 }
 540
 541 static void fsck_object_dir(const char *path)
 542 {
 543         int i;
 544         for (i = 0; i < 256; i++) {
 545                 static char dir[4096];
 546                 sprintf(dir, "%s/%02x", path, i);
 547                 fsck_dir(i, dir);
 548         }
 549         fsck_sha1_list();
 550 }
 551
 552 static int fsck_head_link(void)
 553 {
 554         unsigned char sha1[20];
 555         int flag;
 556         const char *head_points_at = resolve_ref("HEAD", sha1, 1, &flag);
 557
 558         if (!head_points_at || !(flag & REF_ISSYMREF))
 559                 return error("HEAD is not a symbolic ref");
 560         if (prefixcmp(head_points_at, "refs/heads/"))
 561                 return error("HEAD points to something strange (%s)",
 562                              head_points_at);
 563         if (is_null_sha1(sha1))
 564                 return error("HEAD: not a valid git pointer");
 565         return 0;
 566 }
 567
 568 static int fsck_cache_tree(struct cache_tree *it)
 569 {
 570         int i;
 571         int err = 0;
 572
 573         if (0 <= it->entry_count) {
 574                 struct object *obj = parse_object(it->sha1);
 575                 if (!obj) {
 576                         error("%s: invalid sha1 pointer in cache-tree",
 577                               sha1_to_hex(it->sha1));
 578                         return 1;
 579                 }
 580                 mark_reachable(obj, REACHABLE);
 581                 obj->used = 1;
 582                 if (obj->type != OBJ_TREE)
 583                         err |= objerror(obj, "non-tree in cache-tree");
 584         }
 585         for (i = 0; i < it->subtree_nr; i++)
 586                 err |= fsck_cache_tree(it->down[i]->cache_tree);
 587         return err;
 588 }
 589
 590 static const char fsck_usage[] =
 591 "git-fsck [--tags] [--root] [[--unreachable] [--cache] [--full] "
 592 "[--strict] <head-sha1>*]";
 593
 594 int cmd_fsck(int argc, char **argv, const char *prefix)
 595 {
 596         int i, heads;
 597
 598         track_object_refs = 1;
 599         errors_found = 0;
 600
 601         for (i = 1; i < argc; i++) {
 602                 const char *arg = argv[i];
 603
 604                 if (!strcmp(arg, "--unreachable")) {
 605                         show_unreachable = 1;
 606                         continue;
 607                 }
 608                 if (!strcmp(arg, "--tags")) {
 609                         show_tags = 1;
 610                         continue;
 611                 }
 612                 if (!strcmp(arg, "--root")) {
 613                         show_root = 1;
 614                         continue;
 615                 }
 616                 if (!strcmp(arg, "--cache")) {
 617                         keep_cache_objects = 1;
 618                         continue;
 619                 }
 620                 if (!strcmp(arg, "--full")) {
 621                         check_full = 1;
 622                         continue;
 623                 }
 624                 if (!strcmp(arg, "--strict")) {
 625                         check_strict = 1;
 626                         continue;
 627                 }
 628                 if (*arg == '-')
 629                         usage(fsck_usage);
 630         }
 631
 632         fsck_head_link();
 633         fsck_object_dir(get_object_directory());
 634         if (check_full) {
 635                 struct alternate_object_database *alt;
 636                 struct packed_git *p;
 637                 prepare_alt_odb();
 638                 for (alt = alt_odb_list; alt; alt = alt->next) {
 639                         char namebuf[PATH_MAX];
 640                         int namelen = alt->name - alt->base;
 641                         memcpy(namebuf, alt->base, namelen);
 642                         namebuf[namelen - 1] = 0;
 643                         fsck_object_dir(namebuf);
 644                 }
 645                 prepare_packed_git();
 646                 for (p = packed_git; p; p = p->next)
 647                         /* verify gives error messages itself */
 648                         verify_pack(p, 0);
 649
 650                 for (p = packed_git; p; p = p->next) {
 651                         uint32_t i, num = num_packed_objects(p);
 652                         for (i = 0; i < num; i++) {
 653                                 unsigned char sha1[20];
 654                                 nth_packed_object_sha1(p, i, sha1);
 655                                 fsck_sha1(sha1);
 656                         }
 657                 }
 658         }
 659
 660         heads = 0;
 661         for (i = 1; i < argc; i++) {
 662                 const char *arg = argv[i];
 663
 664                 if (*arg == '-')
 665                         continue;
 666
 667                 if (!get_sha1(arg, head_sha1)) {
 668                         struct object *obj = lookup_object(head_sha1);
 669
 670                         /* Error is printed by lookup_object(). */
 671                         if (!obj)
 672                                 continue;
 673
 674                         obj->used = 1;
 675                         mark_reachable(obj, REACHABLE);
 676                         heads++;
 677                         continue;
 678                 }
 679                 error("invalid parameter: expected sha1, got '%s'", arg);
 680         }
 681
 682         /*
 683          * If we've not been given any explicit head information, do the
 684          * default ones from .git/refs. We also consider the index file
 685          * in this case (ie this implies --cache).
 686          */
 687         if (!heads) {
 688                 get_default_heads();
 689                 keep_cache_objects = 1;
 690         }
 691
 692         if (keep_cache_objects) {
 693                 int i;
 694                 read_cache();
 695                 for (i = 0; i < active_nr; i++) {
 696                         struct blob *blob = lookup_blob(active_cache[i]->sha1);
 697                         struct object *obj;
 698                         if (!blob)
 699                                 continue;
 700                         obj = &blob->object;
 701                         obj->used = 1;
 702                         mark_reachable(obj, REACHABLE);
 703                 }
 704                 if (active_cache_tree)
 705                         fsck_cache_tree(active_cache_tree);
 706         }
 707
 708         check_connectivity();
 709         return errors_found;
 710 }