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_empty_name = 0;
 223         int has_zero_pad = 0;
 224         int has_bad_modes = 0;
 225         int has_dup_entries = 0;
 226         int not_properly_sorted = 0;
 227         struct tree_desc desc;
 228         unsigned o_mode;
 229         const char *o_name;
 230         const unsigned char *o_sha1;
 231
 232         init_tree_desc(&desc, item->buffer, item->size);
 233
 234         o_mode = 0;
 235         o_name = NULL;
 236         o_sha1 = NULL;
 237         while (desc.size) {
 238                 unsigned mode;
 239                 const char *name;
 240                 const unsigned char *sha1;
 241
 242                 sha1 = tree_entry_extract(&desc, &name, &mode);
 243
 244                 if (strchr(name, '/'))
 245                         has_full_path = 1;
 246                 if (!*name)
 247                         has_empty_name = 1;
 248                 has_zero_pad |= *(char *)desc.buffer == '0';
 249                 update_tree_entry(&desc);
 250
 251                 switch (mode) {
 252                 /*
 253                  * Standard modes..
 254                  */
 255                 case S_IFREG | 0755:
 256                 case S_IFREG | 0644:
 257                 case S_IFLNK:
 258                 case S_IFDIR:
 259                 case S_IFGITLINK:
 260                         break;
 261                 /*
 262                  * This is nonstandard, but we had a few of these
 263                  * early on when we honored the full set of mode
 264                  * bits..
 265                  */
 266                 case S_IFREG | 0664:
 267                         if (!check_strict)
 268                                 break;
 269                 default:
 270                         has_bad_modes = 1;
 271                 }
 272
 273                 if (o_name) {
 274                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 275                         case TREE_UNORDERED:
 276                                 not_properly_sorted = 1;
 277                                 break;
 278                         case TREE_HAS_DUPS:
 279                                 has_dup_entries = 1;
 280                                 break;
 281                         default:
 282                                 break;
 283                         }
 284                 }
 285
 286                 o_mode = mode;
 287                 o_name = name;
 288                 o_sha1 = sha1;
 289         }
 290         free(item->buffer);
 291         item->buffer = NULL;
 292
 293         retval = 0;
 294         if (has_full_path) {
 295                 objwarning(&item->object, "contains full pathnames");
 296         }
 297         if (has_empty_name) {
 298                 objwarning(&item->object, "contains empty pathname");
 299         }
 300         if (has_zero_pad) {
 301                 objwarning(&item->object, "contains zero-padded file modes");
 302         }
 303         if (has_bad_modes) {
 304                 objwarning(&item->object, "contains bad file modes");
 305         }
 306         if (has_dup_entries) {
 307                 retval = objerror(&item->object, "contains duplicate file entries");
 308         }
 309         if (not_properly_sorted) {
 310                 retval = objerror(&item->object, "not properly sorted");
 311         }
 312         return retval;
 313 }
 314
 315 static int fsck_commit(struct commit *commit)
 316 {
 317         char *buffer = commit->buffer;
 318         unsigned char tree_sha1[20], sha1[20];
 319
 320         if (memcmp(buffer, "tree ", 5))
 321                 return objerror(&commit->object, "invalid format - expected 'tree' line");
 322         if (get_sha1_hex(buffer+5, tree_sha1) || buffer[45] != '\n')
 323                 return objerror(&commit->object, "invalid 'tree' line format - bad sha1");
 324         buffer += 46;
 325         while (!memcmp(buffer, "parent ", 7)) {
 326                 if (get_sha1_hex(buffer+7, sha1) || buffer[47] != '\n')
 327                         return objerror(&commit->object, "invalid 'parent' line format - bad sha1");
 328                 buffer += 48;
 329         }
 330         if (memcmp(buffer, "author ", 7))
 331                 return objerror(&commit->object, "invalid format - expected 'author' line");
 332         free(commit->buffer);
 333         commit->buffer = NULL;
 334         if (!commit->tree)
 335                 return objerror(&commit->object, "could not load commit's tree %s", tree_sha1);
 336         if (!commit->parents && show_root)
 337                 printf("root %s\n", sha1_to_hex(commit->object.sha1));
 338         if (!commit->date)
 339                 printf("bad commit date in %s\n",
 340                        sha1_to_hex(commit->object.sha1));
 341         return 0;
 342 }
 343
 344 static int fsck_tag(struct tag *tag)
 345 {
 346         struct object *tagged = tag->tagged;
 347
 348         if (!tagged) {
 349                 return objerror(&tag->object, "could not load tagged object");
 350         }
 351         if (!show_tags)
 352                 return 0;
 353
 354         printf("tagged %s %s", typename(tagged->type), sha1_to_hex(tagged->sha1));
 355         printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
 356         return 0;
 357 }
 358
 359 static int fsck_sha1(const unsigned char *sha1)
 360 {
 361         struct object *obj = parse_object(sha1);
 362         if (!obj) {
 363                 errors_found |= ERROR_OBJECT;
 364                 return error("%s: object corrupt or missing",
 365                              sha1_to_hex(sha1));
 366         }
 367         if (obj->flags & SEEN)
 368                 return 0;
 369         obj->flags |= SEEN;
 370         if (obj->type == OBJ_BLOB)
 371                 return 0;
 372         if (obj->type == OBJ_TREE)
 373                 return fsck_tree((struct tree *) obj);
 374         if (obj->type == OBJ_COMMIT)
 375                 return fsck_commit((struct commit *) obj);
 376         if (obj->type == OBJ_TAG)
 377                 return fsck_tag((struct tag *) obj);
 378
 379         /* By now, parse_object() would've returned NULL instead. */
 380         return objerror(obj, "unknown type '%d' (internal fsck error)",
 381                         obj->type);
 382 }
 383
 384 /*
 385  * This is the sorting chunk size: make it reasonably
 386  * big so that we can sort well..
 387  */
 388 #define MAX_SHA1_ENTRIES (1024)
 389
 390 struct sha1_entry {
 391         unsigned long ino;
 392         unsigned char sha1[20];
 393 };
 394
 395 static struct {
 396         unsigned long nr;
 397         struct sha1_entry *entry[MAX_SHA1_ENTRIES];
 398 } sha1_list;
 399
 400 static int ino_compare(const void *_a, const void *_b)
 401 {
 402         const struct sha1_entry *a = _a, *b = _b;
 403         unsigned long ino1 = a->ino, ino2 = b->ino;
 404         return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0;
 405 }
 406
 407 static void fsck_sha1_list(void)
 408 {
 409         int i, nr = sha1_list.nr;
 410
 411         if (SORT_DIRENT)
 412                 qsort(sha1_list.entry, nr,
 413                       sizeof(struct sha1_entry *), ino_compare);
 414         for (i = 0; i < nr; i++) {
 415                 struct sha1_entry *entry = sha1_list.entry[i];
 416                 unsigned char *sha1 = entry->sha1;
 417
 418                 sha1_list.entry[i] = NULL;
 419                 fsck_sha1(sha1);
 420                 free(entry);
 421         }
 422         sha1_list.nr = 0;
 423 }
 424
 425 static void add_sha1_list(unsigned char *sha1, unsigned long ino)
 426 {
 427         struct sha1_entry *entry = xmalloc(sizeof(*entry));
 428         int nr;
 429
 430         entry->ino = ino;
 431         hashcpy(entry->sha1, sha1);
 432         nr = sha1_list.nr;
 433         if (nr == MAX_SHA1_ENTRIES) {
 434                 fsck_sha1_list();
 435                 nr = 0;
 436         }
 437         sha1_list.entry[nr] = entry;
 438         sha1_list.nr = ++nr;
 439 }
 440
 441 static void fsck_dir(int i, char *path)
 442 {
 443         DIR *dir = opendir(path);
 444         struct dirent *de;
 445
 446         if (!dir)
 447                 return;
 448
 449         while ((de = readdir(dir)) != NULL) {
 450                 char name[100];
 451                 unsigned char sha1[20];
 452                 int len = strlen(de->d_name);
 453
 454                 switch (len) {
 455                 case 2:
 456                         if (de->d_name[1] != '.')
 457                                 break;
 458                 case 1:
 459                         if (de->d_name[0] != '.')
 460                                 break;
 461                         continue;
 462                 case 38:
 463                         sprintf(name, "%02x", i);
 464                         memcpy(name+2, de->d_name, len+1);
 465                         if (get_sha1_hex(name, sha1) < 0)
 466                                 break;
 467                         add_sha1_list(sha1, DIRENT_SORT_HINT(de));
 468                         continue;
 469                 }
 470                 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
 471         }
 472         closedir(dir);
 473 }
 474
 475 static int default_refs;
 476
 477 static int fsck_handle_reflog_ent(unsigned char *osha1, unsigned char *nsha1,
 478                 const char *email, unsigned long timestamp, int tz,
 479                 const char *message, void *cb_data)
 480 {
 481         struct object *obj;
 482
 483         if (!is_null_sha1(osha1)) {
 484                 obj = lookup_object(osha1);
 485                 if (obj) {
 486                         obj->used = 1;
 487                         mark_reachable(obj, REACHABLE);
 488                 }
 489         }
 490         obj = lookup_object(nsha1);
 491         if (obj) {
 492                 obj->used = 1;
 493                 mark_reachable(obj, REACHABLE);
 494         }
 495         return 0;
 496 }
 497
 498 static int fsck_handle_reflog(const char *logname, const unsigned char *sha1, int flag, void *cb_data)
 499 {
 500         for_each_reflog_ent(logname, fsck_handle_reflog_ent, NULL);
 501         return 0;
 502 }
 503
 504 static int fsck_handle_ref(const char *refname, const unsigned char *sha1, int flag, void *cb_data)
 505 {
 506         struct object *obj;
 507
 508         obj = lookup_object(sha1);
 509         if (!obj) {
 510                 if (has_sha1_file(sha1)) {
 511                         default_refs++;
 512                         return 0; /* it is in a pack */
 513                 }
 514                 error("%s: invalid sha1 pointer %s", refname, sha1_to_hex(sha1));
 515                 /* We'll continue with the rest despite the error.. */
 516                 return 0;
 517         }
 518         default_refs++;
 519         obj->used = 1;
 520         mark_reachable(obj, REACHABLE);
 521
 522         return 0;
 523 }
 524
 525 static void get_default_heads(void)
 526 {
 527         for_each_ref(fsck_handle_ref, NULL);
 528         if (include_reflogs)
 529                 for_each_reflog(fsck_handle_reflog, NULL);
 530
 531         /*
 532          * Not having any default heads isn't really fatal, but
 533          * it does mean that "--unreachable" no longer makes any
 534          * sense (since in this case everything will obviously
 535          * be unreachable by definition.
 536          *
 537          * Showing dangling objects is valid, though (as those
 538          * dangling objects are likely lost heads).
 539          *
 540          * So we just print a warning about it, and clear the
 541          * "show_unreachable" flag.
 542          */
 543         if (!default_refs) {
 544                 fprintf(stderr, "notice: No default references\n");
 545                 show_unreachable = 0;
 546         }
 547 }
 548
 549 static void fsck_object_dir(const char *path)
 550 {
 551         int i;
 552         for (i = 0; i < 256; i++) {
 553                 static char dir[4096];
 554                 sprintf(dir, "%s/%02x", path, i);
 555                 fsck_dir(i, dir);
 556         }
 557         fsck_sha1_list();
 558 }
 559
 560 static int fsck_head_link(void)
 561 {
 562         unsigned char sha1[20];
 563         int flag;
 564         int null_is_error = 0;
 565         const char *head_points_at = resolve_ref("HEAD", sha1, 0, &flag);
 566
 567         if (!head_points_at)
 568                 return error("Invalid HEAD");
 569         if (!strcmp(head_points_at, "HEAD"))
 570                 /* detached HEAD */
 571                 null_is_error = 1;
 572         else if (prefixcmp(head_points_at, "refs/heads/"))
 573                 return error("HEAD points to something strange (%s)",
 574                              head_points_at);
 575         if (is_null_sha1(sha1)) {
 576                 if (null_is_error)
 577                         return error("HEAD: detached HEAD points at nothing");
 578                 fprintf(stderr, "notice: HEAD points to an unborn branch (%s)\n",
 579                         head_points_at + 11);
 580         }
 581         return 0;
 582 }
 583
 584 static int fsck_cache_tree(struct cache_tree *it)
 585 {
 586         int i;
 587         int err = 0;
 588
 589         if (0 <= it->entry_count) {
 590                 struct object *obj = parse_object(it->sha1);
 591                 if (!obj) {
 592                         error("%s: invalid sha1 pointer in cache-tree",
 593                               sha1_to_hex(it->sha1));
 594                         return 1;
 595                 }
 596                 mark_reachable(obj, REACHABLE);
 597                 obj->used = 1;
 598                 if (obj->type != OBJ_TREE)
 599                         err |= objerror(obj, "non-tree in cache-tree");
 600         }
 601         for (i = 0; i < it->subtree_nr; i++)
 602                 err |= fsck_cache_tree(it->down[i]->cache_tree);
 603         return err;
 604 }
 605
 606 static const char fsck_usage[] =
 607 "git-fsck [--tags] [--root] [[--unreachable] [--cache] [--full] "
 608 "[--strict] <head-sha1>*]";
 609
 610 int cmd_fsck(int argc, char **argv, const char *prefix)
 611 {
 612         int i, heads;
 613
 614         track_object_refs = 1;
 615         errors_found = 0;
 616
 617         for (i = 1; i < argc; i++) {
 618                 const char *arg = argv[i];
 619
 620                 if (!strcmp(arg, "--unreachable")) {
 621                         show_unreachable = 1;
 622                         continue;
 623                 }
 624                 if (!strcmp(arg, "--tags")) {
 625                         show_tags = 1;
 626                         continue;
 627                 }
 628                 if (!strcmp(arg, "--root")) {
 629                         show_root = 1;
 630                         continue;
 631                 }
 632                 if (!strcmp(arg, "--cache")) {
 633                         keep_cache_objects = 1;
 634                         continue;
 635                 }
 636                 if (!strcmp(arg, "--no-reflogs")) {
 637                         include_reflogs = 0;
 638                         continue;
 639                 }
 640                 if (!strcmp(arg, "--full")) {
 641                         check_full = 1;
 642                         continue;
 643                 }
 644                 if (!strcmp(arg, "--strict")) {
 645                         check_strict = 1;
 646                         continue;
 647                 }
 648                 if (*arg == '-')
 649                         usage(fsck_usage);
 650         }
 651
 652         fsck_head_link();
 653         fsck_object_dir(get_object_directory());
 654         if (check_full) {
 655                 struct alternate_object_database *alt;
 656                 struct packed_git *p;
 657                 prepare_alt_odb();
 658                 for (alt = alt_odb_list; alt; alt = alt->next) {
 659                         char namebuf[PATH_MAX];
 660                         int namelen = alt->name - alt->base;
 661                         memcpy(namebuf, alt->base, namelen);
 662                         namebuf[namelen - 1] = 0;
 663                         fsck_object_dir(namebuf);
 664                 }
 665                 prepare_packed_git();
 666                 for (p = packed_git; p; p = p->next)
 667                         /* verify gives error messages itself */
 668                         verify_pack(p, 0);
 669
 670                 for (p = packed_git; p; p = p->next) {
 671                         uint32_t i, num = p->num_objects;
 672                         for (i = 0; i < num; i++)
 673                                 fsck_sha1(nth_packed_object_sha1(p, i));
 674                 }
 675         }
 676
 677         heads = 0;
 678         for (i = 1; i < argc; i++) {
 679                 const char *arg = argv[i];
 680
 681                 if (*arg == '-')
 682                         continue;
 683
 684                 if (!get_sha1(arg, head_sha1)) {
 685                         struct object *obj = lookup_object(head_sha1);
 686
 687                         /* Error is printed by lookup_object(). */
 688                         if (!obj)
 689                                 continue;
 690
 691                         obj->used = 1;
 692                         mark_reachable(obj, REACHABLE);
 693                         heads++;
 694                         continue;
 695                 }
 696                 error("invalid parameter: expected sha1, got '%s'", arg);
 697         }
 698
 699         /*
 700          * If we've not been given any explicit head information, do the
 701          * default ones from .git/refs. We also consider the index file
 702          * in this case (ie this implies --cache).
 703          */
 704         if (!heads) {
 705                 get_default_heads();
 706                 keep_cache_objects = 1;
 707         }
 708
 709         if (keep_cache_objects) {
 710                 int i;
 711                 read_cache();
 712                 for (i = 0; i < active_nr; i++) {
 713                         unsigned int mode;
 714                         struct blob *blob;
 715                         struct object *obj;
 716
 717                         mode = ntohl(active_cache[i]->ce_mode);
 718                         if (S_ISGITLINK(mode))
 719                                 continue;
 720                         blob = lookup_blob(active_cache[i]->sha1);
 721                         if (!blob)
 722                                 continue;
 723                         obj = &blob->object;
 724                         obj->used = 1;
 725                         mark_reachable(obj, REACHABLE);
 726                 }
 727                 if (active_cache_tree)
 728                         fsck_cache_tree(active_cache_tree);
 729         }
 730
 731         check_connectivity();
 732         return errors_found;
 733 }