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 include_reflogs = 1;
  18 static int check_full;
  19 static int check_strict;
  20 static int keep_cache_objects;
  21 static unsigned char head_sha1[20];
  22 static int errors_found;
  23 #define ERROR_OBJECT 01
  24 #define ERROR_REACHABLE 02
  25
  26 #ifdef NO_D_INO_IN_DIRENT
  27 #define SORT_DIRENT 0
  28 #define DIRENT_SORT_HINT(de) 0
  29 #else
  30 #define SORT_DIRENT 1
  31 #define DIRENT_SORT_HINT(de) ((de)->d_ino)
  32 #endif
  33
  34 static void objreport(struct object *obj, const char *severity,
  35                       const char *err, va_list params)
  36 {
  37         fprintf(stderr, "%s in %s %s: ",
  38                 severity, typename(obj->type), sha1_to_hex(obj->sha1));
  39         vfprintf(stderr, err, params);
  40         fputs("\n", stderr);
  41 }
  42
  43 static int objerror(struct object *obj, const char *err, ...)
  44 {
  45         va_list params;
  46         va_start(params, err);
  47         errors_found |= ERROR_OBJECT;
  48         objreport(obj, "error", err, params);
  49         va_end(params);
  50         return -1;
  51 }
  52
  53 static int objwarning(struct object *obj, const char *err, ...)
  54 {
  55         va_list params;
  56         va_start(params, err);
  57         objreport(obj, "warning", err, params);
  58         va_end(params);
  59         return -1;
  60 }
  61
  62 /*
  63  * Check a single reachable object
  64  */
  65 static void check_reachable_object(struct object *obj)
  66 {
  67         const struct object_refs *refs;
  68
  69         /*
  70          * We obviously want the object to be parsed,
  71          * except if it was in a pack-file and we didn't
  72          * do a full fsck
  73          */
  74         if (!obj->parsed) {
  75                 if (has_sha1_pack(obj->sha1, NULL))
  76                         return; /* it is in pack - forget about it */
  77                 printf("missing %s %s\n", typename(obj->type), sha1_to_hex(obj->sha1));
  78                 errors_found |= ERROR_REACHABLE;
  79                 return;
  80         }
  81
  82         /*
  83          * Check that everything that we try to reference is also good.
  84          */
  85         refs = lookup_object_refs(obj);
  86         if (refs) {
  87                 unsigned j;
  88                 for (j = 0; j < refs->count; j++) {
  89                         struct object *ref = refs->ref[j];
  90                         if (ref->parsed ||
  91                             (has_sha1_file(ref->sha1)))
  92                                 continue;
  93                         printf("broken link from %7s %s\n",
  94                                typename(obj->type), sha1_to_hex(obj->sha1));
  95                         printf("              to %7s %s\n",
  96                                typename(ref->type), sha1_to_hex(ref->sha1));
  97                         errors_found |= ERROR_REACHABLE;
  98                 }
  99         }
 100 }
 101
 102 /*
 103  * Check a single unreachable object
 104  */
 105 static void check_unreachable_object(struct object *obj)
 106 {
 107         /*
 108          * Missing unreachable object? Ignore it. It's not like
 109          * we miss it (since it can't be reached), nor do we want
 110          * to complain about it being unreachable (since it does
 111          * not exist).
 112          */
 113         if (!obj->parsed)
 114                 return;
 115
 116         /*
 117          * Unreachable object that exists? Show it if asked to,
 118          * since this is something that is prunable.
 119          */
 120         if (show_unreachable) {
 121                 printf("unreachable %s %s\n", typename(obj->type), sha1_to_hex(obj->sha1));
 122                 return;
 123         }
 124
 125         /*
 126          * "!used" means that nothing at all points to it, including
 127          * other unreachable objects. In other words, it's the "tip"
 128          * of some set of unreachable objects, usually a commit that
 129          * got dropped.
 130          *
 131          * Such starting points are more interesting than some random
 132          * set of unreachable objects, so we show them even if the user
 133          * hasn't asked for _all_ unreachable objects. If you have
 134          * deleted a branch by mistake, this is a prime candidate to
 135          * start looking at, for example.
 136          */
 137         if (!obj->used) {
 138                 printf("dangling %s %s\n", typename(obj->type),
 139                        sha1_to_hex(obj->sha1));
 140                 return;
 141         }
 142
 143         /*
 144          * Otherwise? It's there, it's unreachable, and some other unreachable
 145          * object points to it. Ignore it - it's not interesting, and we showed
 146          * all the interesting cases above.
 147          */
 148 }
 149
 150 static void check_object(struct object *obj)
 151 {
 152         if (obj->flags & REACHABLE)
 153                 check_reachable_object(obj);
 154         else
 155                 check_unreachable_object(obj);
 156 }
 157
 158 static void check_connectivity(void)
 159 {
 160         int i, max;
 161
 162         /* Look up all the requirements, warn about missing objects.. */
 163         max = get_max_object_index();
 164         for (i = 0; i < max; i++) {
 165                 struct object *obj = get_indexed_object(i);
 166
 167                 if (obj)
 168                         check_object(obj);
 169         }
 170 }
 171
 172 /*
 173  * The entries in a tree are ordered in the _path_ order,
 174  * which means that a directory entry is ordered by adding
 175  * a slash to the end of it.
 176  *
 177  * So a directory called "a" is ordered _after_ a file
 178  * called "a.c", because "a/" sorts after "a.c".
 179  */
 180 #define TREE_UNORDERED (-1)
 181 #define TREE_HAS_DUPS  (-2)
 182
 183 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 184 {
 185         int len1 = strlen(name1);
 186         int len2 = strlen(name2);
 187         int len = len1 < len2 ? len1 : len2;
 188         unsigned char c1, c2;
 189         int cmp;
 190
 191         cmp = memcmp(name1, name2, len);
 192         if (cmp < 0)
 193                 return 0;
 194         if (cmp > 0)
 195                 return TREE_UNORDERED;
 196
 197         /*
 198          * Ok, the first <len> characters are the same.
 199          * Now we need to order the next one, but turn
 200          * a '\0' into a '/' for a directory entry.
 201          */
 202         c1 = name1[len];
 203         c2 = name2[len];
 204         if (!c1 && !c2)
 205                 /*
 206                  * git-write-tree used to write out a nonsense tree that has
 207                  * entries with the same name, one blob and one tree.  Make
 208                  * sure we do not have duplicate entries.
 209                  */
 210                 return TREE_HAS_DUPS;
 211         if (!c1 && S_ISDIR(mode1))
 212                 c1 = '/';
 213         if (!c2 && S_ISDIR(mode2))
 214                 c2 = '/';
 215         return c1 < c2 ? 0 : TREE_UNORDERED;
 216 }
 217
 218 static int fsck_tree(struct tree *item)
 219 {
 220         int retval;
 221         int has_full_path = 0;
 222         int has_zero_pad = 0;
 223         int has_bad_modes = 0;
 224         int has_dup_entries = 0;
 225         int not_properly_sorted = 0;
 226         struct tree_desc desc;
 227         unsigned o_mode;
 228         const char *o_name;
 229         const unsigned char *o_sha1;
 230
 231         init_tree_desc(&desc, item->buffer, 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.buffer == '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                 case S_IFDIRLNK:
 257                         break;
 258                 /*
 259                  * This is nonstandard, but we had a few of these
 260                  * early on when we honored the full set of mode
 261                  * bits..
 262                  */
 263                 case S_IFREG | 0664:
 264                         if (!check_strict)
 265                                 break;
 266                 default:
 267                         has_bad_modes = 1;
 268                 }
 269
 270                 if (o_name) {
 271                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 272                         case TREE_UNORDERED:
 273                                 not_properly_sorted = 1;
 274                                 break;
 275                         case TREE_HAS_DUPS:
 276                                 has_dup_entries = 1;
 277                                 break;
 278                         default:
 279                                 break;
 280                         }
 281                 }
 282
 283                 o_mode = mode;
 284                 o_name = name;
 285                 o_sha1 = sha1;
 286         }
 287         free(item->buffer);
 288         item->buffer = NULL;
 289
 290         retval = 0;
 291         if (has_full_path) {
 292                 objwarning(&item->object, "contains full pathnames");
 293         }
 294         if (has_zero_pad) {
 295                 objwarning(&item->object, "contains zero-padded file modes");
 296         }
 297         if (has_bad_modes) {
 298                 objwarning(&item->object, "contains bad file modes");
 299         }
 300         if (has_dup_entries) {
 301                 retval = objerror(&item->object, "contains duplicate file entries");
 302         }
 303         if (not_properly_sorted) {
 304                 retval = objerror(&item->object, "not properly sorted");
 305         }
 306         return retval;
 307 }
 308
 309 static int fsck_commit(struct commit *commit)
 310 {
 311         char *buffer = commit->buffer;
 312         unsigned char tree_sha1[20], sha1[20];
 313
 314         if (memcmp(buffer, "tree ", 5))
 315                 return objerror(&commit->object, "invalid format - expected 'tree' line");
 316         if (get_sha1_hex(buffer+5, tree_sha1) || buffer[45] != '\n')
 317                 return objerror(&commit->object, "invalid 'tree' line format - bad sha1");
 318         buffer += 46;
 319         while (!memcmp(buffer, "parent ", 7)) {
 320                 if (get_sha1_hex(buffer+7, sha1) || buffer[47] != '\n')
 321                         return objerror(&commit->object, "invalid 'parent' line format - bad sha1");
 322                 buffer += 48;
 323         }
 324         if (memcmp(buffer, "author ", 7))
 325                 return objerror(&commit->object, "invalid format - expected 'author' line");
 326         free(commit->buffer);
 327         commit->buffer = NULL;
 328         if (!commit->tree)
 329                 return objerror(&commit->object, "could not load commit's tree %s", tree_sha1);
 330         if (!commit->parents && show_root)
 331                 printf("root %s\n", sha1_to_hex(commit->object.sha1));
 332         if (!commit->date)
 333                 printf("bad commit date in %s\n",
 334                        sha1_to_hex(commit->object.sha1));
 335         return 0;
 336 }
 337
 338 static int fsck_tag(struct tag *tag)
 339 {
 340         struct object *tagged = tag->tagged;
 341
 342         if (!tagged) {
 343                 return objerror(&tag->object, "could not load tagged object");
 344         }
 345         if (!show_tags)
 346                 return 0;
 347
 348         printf("tagged %s %s", typename(tagged->type), sha1_to_hex(tagged->sha1));
 349         printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
 350         return 0;
 351 }
 352
 353 static int fsck_sha1(const unsigned char *sha1)
 354 {
 355         struct object *obj = parse_object(sha1);
 356         if (!obj) {
 357                 errors_found |= ERROR_OBJECT;
 358                 return error("%s: object corrupt or missing",
 359                              sha1_to_hex(sha1));
 360         }
 361         if (obj->flags & SEEN)
 362                 return 0;
 363         obj->flags |= SEEN;
 364         if (obj->type == OBJ_BLOB)
 365                 return 0;
 366         if (obj->type == OBJ_TREE)
 367                 return fsck_tree((struct tree *) obj);
 368         if (obj->type == OBJ_COMMIT)
 369                 return fsck_commit((struct commit *) obj);
 370         if (obj->type == OBJ_TAG)
 371                 return fsck_tag((struct tag *) obj);
 372
 373         /* By now, parse_object() would've returned NULL instead. */
 374         return objerror(obj, "unknown type '%d' (internal fsck error)",
 375                         obj->type);
 376 }
 377
 378 /*
 379  * This is the sorting chunk size: make it reasonably
 380  * big so that we can sort well..
 381  */
 382 #define MAX_SHA1_ENTRIES (1024)
 383
 384 struct sha1_entry {
 385         unsigned long ino;
 386         unsigned char sha1[20];
 387 };
 388
 389 static struct {
 390         unsigned long nr;
 391         struct sha1_entry *entry[MAX_SHA1_ENTRIES];
 392 } sha1_list;
 393
 394 static int ino_compare(const void *_a, const void *_b)
 395 {
 396         const struct sha1_entry *a = _a, *b = _b;
 397         unsigned long ino1 = a->ino, ino2 = b->ino;
 398         return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0;
 399 }
 400
 401 static void fsck_sha1_list(void)
 402 {
 403         int i, nr = sha1_list.nr;
 404
 405         if (SORT_DIRENT)
 406                 qsort(sha1_list.entry, nr,
 407                       sizeof(struct sha1_entry *), ino_compare);
 408         for (i = 0; i < nr; i++) {
 409                 struct sha1_entry *entry = sha1_list.entry[i];
 410                 unsigned char *sha1 = entry->sha1;
 411
 412                 sha1_list.entry[i] = NULL;
 413                 fsck_sha1(sha1);
 414                 free(entry);
 415         }
 416         sha1_list.nr = 0;
 417 }
 418
 419 static void add_sha1_list(unsigned char *sha1, unsigned long ino)
 420 {
 421         struct sha1_entry *entry = xmalloc(sizeof(*entry));
 422         int nr;
 423
 424         entry->ino = ino;
 425         hashcpy(entry->sha1, sha1);
 426         nr = sha1_list.nr;
 427         if (nr == MAX_SHA1_ENTRIES) {
 428                 fsck_sha1_list();
 429                 nr = 0;
 430         }
 431         sha1_list.entry[nr] = entry;
 432         sha1_list.nr = ++nr;
 433 }
 434
 435 static void fsck_dir(int i, char *path)
 436 {
 437         DIR *dir = opendir(path);
 438         struct dirent *de;
 439
 440         if (!dir)
 441                 return;
 442
 443         while ((de = readdir(dir)) != NULL) {
 444                 char name[100];
 445                 unsigned char sha1[20];
 446                 int len = strlen(de->d_name);
 447
 448                 switch (len) {
 449                 case 2:
 450                         if (de->d_name[1] != '.')
 451                                 break;
 452                 case 1:
 453                         if (de->d_name[0] != '.')
 454                                 break;
 455                         continue;
 456                 case 38:
 457                         sprintf(name, "%02x", i);
 458                         memcpy(name+2, de->d_name, len+1);
 459                         if (get_sha1_hex(name, sha1) < 0)
 460                                 break;
 461                         add_sha1_list(sha1, DIRENT_SORT_HINT(de));
 462                         continue;
 463                 }
 464                 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
 465         }
 466         closedir(dir);
 467 }
 468
 469 static int default_refs;
 470
 471 static int fsck_handle_reflog_ent(unsigned char *osha1, unsigned char *nsha1,
 472                 const char *email, unsigned long timestamp, int tz,
 473                 const char *message, void *cb_data)
 474 {
 475         struct object *obj;
 476
 477         if (!is_null_sha1(osha1)) {
 478                 obj = lookup_object(osha1);
 479                 if (obj) {
 480                         obj->used = 1;
 481                         mark_reachable(obj, REACHABLE);
 482                 }
 483         }
 484         obj = lookup_object(nsha1);
 485         if (obj) {
 486                 obj->used = 1;
 487                 mark_reachable(obj, REACHABLE);
 488         }
 489         return 0;
 490 }
 491
 492 static int fsck_handle_reflog(const char *logname, const unsigned char *sha1, int flag, void *cb_data)
 493 {
 494         for_each_reflog_ent(logname, fsck_handle_reflog_ent, NULL);
 495         return 0;
 496 }
 497
 498 static int fsck_handle_ref(const char *refname, const unsigned char *sha1, int flag, void *cb_data)
 499 {
 500         struct object *obj;
 501
 502         obj = lookup_object(sha1);
 503         if (!obj) {
 504                 if (has_sha1_file(sha1)) {
 505                         default_refs++;
 506                         return 0; /* it is in a pack */
 507                 }
 508                 error("%s: invalid sha1 pointer %s", refname, sha1_to_hex(sha1));
 509                 /* We'll continue with the rest despite the error.. */
 510                 return 0;
 511         }
 512         default_refs++;
 513         obj->used = 1;
 514         mark_reachable(obj, REACHABLE);
 515
 516         return 0;
 517 }
 518
 519 static void get_default_heads(void)
 520 {
 521         for_each_ref(fsck_handle_ref, NULL);
 522         if (include_reflogs)
 523                 for_each_reflog(fsck_handle_reflog, NULL);
 524
 525         /*
 526          * Not having any default heads isn't really fatal, but
 527          * it does mean that "--unreachable" no longer makes any
 528          * sense (since in this case everything will obviously
 529          * be unreachable by definition.
 530          *
 531          * Showing dangling objects is valid, though (as those
 532          * dangling objects are likely lost heads).
 533          *
 534          * So we just print a warning about it, and clear the
 535          * "show_unreachable" flag.
 536          */
 537         if (!default_refs) {
 538                 error("No default references");
 539                 show_unreachable = 0;
 540         }
 541 }
 542
 543 static void fsck_object_dir(const char *path)
 544 {
 545         int i;
 546         for (i = 0; i < 256; i++) {
 547                 static char dir[4096];
 548                 sprintf(dir, "%s/%02x", path, i);
 549                 fsck_dir(i, dir);
 550         }
 551         fsck_sha1_list();
 552 }
 553
 554 static int fsck_head_link(void)
 555 {
 556         unsigned char sha1[20];
 557         int flag;
 558         const char *head_points_at = resolve_ref("HEAD", sha1, 1, &flag);
 559
 560         if (!head_points_at || !(flag & REF_ISSYMREF))
 561                 return error("HEAD is not a symbolic ref");
 562         if (prefixcmp(head_points_at, "refs/heads/"))
 563                 return error("HEAD points to something strange (%s)",
 564                              head_points_at);
 565         if (is_null_sha1(sha1))
 566                 return error("HEAD: not a valid git pointer");
 567         return 0;
 568 }
 569
 570 static int fsck_cache_tree(struct cache_tree *it)
 571 {
 572         int i;
 573         int err = 0;
 574
 575         if (0 <= it->entry_count) {
 576                 struct object *obj = parse_object(it->sha1);
 577                 if (!obj) {
 578                         error("%s: invalid sha1 pointer in cache-tree",
 579                               sha1_to_hex(it->sha1));
 580                         return 1;
 581                 }
 582                 mark_reachable(obj, REACHABLE);
 583                 obj->used = 1;
 584                 if (obj->type != OBJ_TREE)
 585                         err |= objerror(obj, "non-tree in cache-tree");
 586         }
 587         for (i = 0; i < it->subtree_nr; i++)
 588                 err |= fsck_cache_tree(it->down[i]->cache_tree);
 589         return err;
 590 }
 591
 592 static const char fsck_usage[] =
 593 "git-fsck [--tags] [--root] [[--unreachable] [--cache] [--full] "
 594 "[--strict] <head-sha1>*]";
 595
 596 int cmd_fsck(int argc, char **argv, const char *prefix)
 597 {
 598         int i, heads;
 599
 600         track_object_refs = 1;
 601         errors_found = 0;
 602
 603         for (i = 1; i < argc; i++) {
 604                 const char *arg = argv[i];
 605
 606                 if (!strcmp(arg, "--unreachable")) {
 607                         show_unreachable = 1;
 608                         continue;
 609                 }
 610                 if (!strcmp(arg, "--tags")) {
 611                         show_tags = 1;
 612                         continue;
 613                 }
 614                 if (!strcmp(arg, "--root")) {
 615                         show_root = 1;
 616                         continue;
 617                 }
 618                 if (!strcmp(arg, "--cache")) {
 619                         keep_cache_objects = 1;
 620                         continue;
 621                 }
 622                 if (!strcmp(arg, "--no-reflogs")) {
 623                         include_reflogs = 0;
 624                         continue;
 625                 }
 626                 if (!strcmp(arg, "--full")) {
 627                         check_full = 1;
 628                         continue;
 629                 }
 630                 if (!strcmp(arg, "--strict")) {
 631                         check_strict = 1;
 632                         continue;
 633                 }
 634                 if (*arg == '-')
 635                         usage(fsck_usage);
 636         }
 637
 638         fsck_head_link();
 639         fsck_object_dir(get_object_directory());
 640         if (check_full) {
 641                 struct alternate_object_database *alt;
 642                 struct packed_git *p;
 643                 prepare_alt_odb();
 644                 for (alt = alt_odb_list; alt; alt = alt->next) {
 645                         char namebuf[PATH_MAX];
 646                         int namelen = alt->name - alt->base;
 647                         memcpy(namebuf, alt->base, namelen);
 648                         namebuf[namelen - 1] = 0;
 649                         fsck_object_dir(namebuf);
 650                 }
 651                 prepare_packed_git();
 652                 for (p = packed_git; p; p = p->next)
 653                         /* verify gives error messages itself */
 654                         verify_pack(p, 0);
 655
 656                 for (p = packed_git; p; p = p->next) {
 657                         uint32_t i, num = num_packed_objects(p);
 658                         for (i = 0; i < num; i++)
 659                                 fsck_sha1(nth_packed_object_sha1(p, i));
 660                 }
 661         }
 662
 663         heads = 0;
 664         for (i = 1; i < argc; i++) {
 665                 const char *arg = argv[i];
 666
 667                 if (*arg == '-')
 668                         continue;
 669
 670                 if (!get_sha1(arg, head_sha1)) {
 671                         struct object *obj = lookup_object(head_sha1);
 672
 673                         /* Error is printed by lookup_object(). */
 674                         if (!obj)
 675                                 continue;
 676
 677                         obj->used = 1;
 678                         mark_reachable(obj, REACHABLE);
 679                         heads++;
 680                         continue;
 681                 }
 682                 error("invalid parameter: expected sha1, got '%s'", arg);
 683         }
 684
 685         /*
 686          * If we've not been given any explicit head information, do the
 687          * default ones from .git/refs. We also consider the index file
 688          * in this case (ie this implies --cache).
 689          */
 690         if (!heads) {
 691                 get_default_heads();
 692                 keep_cache_objects = 1;
 693         }
 694
 695         if (keep_cache_objects) {
 696                 int i;
 697                 read_cache();
 698                 for (i = 0; i < active_nr; i++) {
 699                         unsigned int mode;
 700                         struct blob *blob;
 701                         struct object *obj;
 702
 703                         mode = ntohl(active_cache[i]->ce_mode);
 704                         if (S_ISDIRLNK(mode))
 705                                 continue;
 706                         blob = lookup_blob(active_cache[i]->sha1);
 707                         if (!blob)
 708                                 continue;
 709                         obj = &blob->object;
 710                         obj->used = 1;
 711                         mark_reachable(obj, REACHABLE);
 712                 }
 713                 if (active_cache_tree)
 714                         fsck_cache_tree(active_cache_tree);
 715         }
 716
 717         check_connectivity();
 718         return errors_found;
 719 }