builtin/fsck.c

   1 #include "builtin.h"
   2 #include "cache.h"
   3 #include "commit.h"
   4 #include "tree.h"
   5 #include "blob.h"
   6 #include "tag.h"
   7 #include "refs.h"
   8 #include "pack.h"
   9 #include "cache-tree.h"
  10 #include "tree-walk.h"
  11 #include "fsck.h"
  12 #include "parse-options.h"
  13 #include "dir.h"
  14 #include "progress.h"
  15 #include "streaming.h"
  16
  17 #define REACHABLE 0x0001
  18 #define SEEN      0x0002
  19 #define HAS_OBJ   0x0004
  20
  21 static int show_root;
  22 static int show_tags;
  23 static int show_unreachable;
  24 static int include_reflogs = 1;
  25 static int check_full = 1;
  26 static int connectivity_only;
  27 static int check_strict;
  28 static int keep_cache_objects;
  29 static struct fsck_options fsck_walk_options = FSCK_OPTIONS_DEFAULT;
  30 static struct fsck_options fsck_obj_options = FSCK_OPTIONS_DEFAULT;
  31 static struct object_id head_oid;
  32 static const char *head_points_at;
  33 static int errors_found;
  34 static int write_lost_and_found;
  35 static int verbose;
  36 static int show_progress = -1;
  37 static int show_dangling = 1;
  38 #define ERROR_OBJECT 01
  39 #define ERROR_REACHABLE 02
  40 #define ERROR_PACK 04
  41
  42 static int fsck_config(const char *var, const char *value, void *cb)
  43 {
  44         if (strcmp(var, "fsck.skiplist") == 0) {
  45                 const char *path;
  46                 struct strbuf sb = STRBUF_INIT;
  47
  48                 if (git_config_pathname(&path, var, value))
  49                         return 1;
  50                 strbuf_addf(&sb, "skiplist=%s", path);
  51                 free((char *)path);
  52                 fsck_set_msg_types(&fsck_obj_options, sb.buf);
  53                 strbuf_release(&sb);
  54                 return 0;
  55         }
  56
  57         if (skip_prefix(var, "fsck.", &var)) {
  58                 fsck_set_msg_type(&fsck_obj_options, var, value);
  59                 return 0;
  60         }
  61
  62         return git_default_config(var, value, cb);
  63 }
  64
  65 static void objreport(struct object *obj, const char *msg_type,
  66                         const char *err)
  67 {
  68         fprintf(stderr, "%s in %s %s: %s\n",
  69                 msg_type, typename(obj->type), sha1_to_hex(obj->sha1), err);
  70 }
  71
  72 static int objerror(struct object *obj, const char *err)
  73 {
  74         errors_found |= ERROR_OBJECT;
  75         objreport(obj, "error", err);
  76         return -1;
  77 }
  78
  79 static int fsck_error_func(struct object *obj, int type, const char *message)
  80 {
  81         objreport(obj, (type == FSCK_WARN) ? "warning" : "error", message);
  82         return (type == FSCK_WARN) ? 0 : 1;
  83 }
  84
  85 static struct object_array pending;
  86
  87 static int mark_object(struct object *obj, int type, void *data, struct fsck_options *options)
  88 {
  89         struct object *parent = data;
  90
  91         /*
  92          * The only case data is NULL or type is OBJ_ANY is when
  93          * mark_object_reachable() calls us.  All the callers of
  94          * that function has non-NULL obj hence ...
  95          */
  96         if (!obj) {
  97                 /* ... these references to parent->fld are safe here */
  98                 printf("broken link from %7s %s\n",
  99                            typename(parent->type), sha1_to_hex(parent->sha1));
 100                 printf("broken link from %7s %s\n",
 101                            (type == OBJ_ANY ? "unknown" : typename(type)), "unknown");
 102                 errors_found |= ERROR_REACHABLE;
 103                 return 1;
 104         }
 105
 106         if (type != OBJ_ANY && obj->type != type)
 107                 /* ... and the reference to parent is safe here */
 108                 objerror(parent, "wrong object type in link");
 109
 110         if (obj->flags & REACHABLE)
 111                 return 0;
 112         obj->flags |= REACHABLE;
 113         if (!(obj->flags & HAS_OBJ)) {
 114                 if (parent && !has_sha1_file(obj->sha1)) {
 115                         printf("broken link from %7s %s\n",
 116                                  typename(parent->type), sha1_to_hex(parent->sha1));
 117                         printf("              to %7s %s\n",
 118                                  typename(obj->type), sha1_to_hex(obj->sha1));
 119                         errors_found |= ERROR_REACHABLE;
 120                 }
 121                 return 1;
 122         }
 123
 124         add_object_array(obj, NULL, &pending);
 125         return 0;
 126 }
 127
 128 static void mark_object_reachable(struct object *obj)
 129 {
 130         mark_object(obj, OBJ_ANY, NULL, NULL);
 131 }
 132
 133 static int traverse_one_object(struct object *obj)
 134 {
 135         int result;
 136         struct tree *tree = NULL;
 137
 138         if (obj->type == OBJ_TREE) {
 139                 tree = (struct tree *)obj;
 140                 if (parse_tree(tree) < 0)
 141                         return 1; /* error already displayed */
 142         }
 143         result = fsck_walk(obj, obj, &fsck_walk_options);
 144         if (tree)
 145                 free_tree_buffer(tree);
 146         return result;
 147 }
 148
 149 static int traverse_reachable(void)
 150 {
 151         struct progress *progress = NULL;
 152         unsigned int nr = 0;
 153         int result = 0;
 154         if (show_progress)
 155                 progress = start_progress_delay(_("Checking connectivity"), 0, 0, 2);
 156         while (pending.nr) {
 157                 struct object_array_entry *entry;
 158                 struct object *obj;
 159
 160                 entry = pending.objects + --pending.nr;
 161                 obj = entry->item;
 162                 result |= traverse_one_object(obj);
 163                 display_progress(progress, ++nr);
 164         }
 165         stop_progress(&progress);
 166         return !!result;
 167 }
 168
 169 static int mark_used(struct object *obj, int type, void *data, struct fsck_options *options)
 170 {
 171         if (!obj)
 172                 return 1;
 173         obj->used = 1;
 174         return 0;
 175 }
 176
 177 /*
 178  * Check a single reachable object
 179  */
 180 static void check_reachable_object(struct object *obj)
 181 {
 182         /*
 183          * We obviously want the object to be parsed,
 184          * except if it was in a pack-file and we didn't
 185          * do a full fsck
 186          */
 187         if (!(obj->flags & HAS_OBJ)) {
 188                 if (has_sha1_pack(obj->sha1))
 189                         return; /* it is in pack - forget about it */
 190                 if (connectivity_only && has_sha1_file(obj->sha1))
 191                         return;
 192                 printf("missing %s %s\n", typename(obj->type), sha1_to_hex(obj->sha1));
 193                 errors_found |= ERROR_REACHABLE;
 194                 return;
 195         }
 196 }
 197
 198 /*
 199  * Check a single unreachable object
 200  */
 201 static void check_unreachable_object(struct object *obj)
 202 {
 203         /*
 204          * Missing unreachable object? Ignore it. It's not like
 205          * we miss it (since it can't be reached), nor do we want
 206          * to complain about it being unreachable (since it does
 207          * not exist).
 208          */
 209         if (!obj->parsed)
 210                 return;
 211
 212         /*
 213          * Unreachable object that exists? Show it if asked to,
 214          * since this is something that is prunable.
 215          */
 216         if (show_unreachable) {
 217                 printf("unreachable %s %s\n", typename(obj->type), sha1_to_hex(obj->sha1));
 218                 return;
 219         }
 220
 221         /*
 222          * "!used" means that nothing at all points to it, including
 223          * other unreachable objects. In other words, it's the "tip"
 224          * of some set of unreachable objects, usually a commit that
 225          * got dropped.
 226          *
 227          * Such starting points are more interesting than some random
 228          * set of unreachable objects, so we show them even if the user
 229          * hasn't asked for _all_ unreachable objects. If you have
 230          * deleted a branch by mistake, this is a prime candidate to
 231          * start looking at, for example.
 232          */
 233         if (!obj->used) {
 234                 if (show_dangling)
 235                         printf("dangling %s %s\n", typename(obj->type),
 236                                sha1_to_hex(obj->sha1));
 237                 if (write_lost_and_found) {
 238                         char *filename = git_pathdup("lost-found/%s/%s",
 239                                 obj->type == OBJ_COMMIT ? "commit" : "other",
 240                                 sha1_to_hex(obj->sha1));
 241                         FILE *f;
 242
 243                         if (safe_create_leading_directories_const(filename)) {
 244                                 error("Could not create lost-found");
 245                                 free(filename);
 246                                 return;
 247                         }
 248                         if (!(f = fopen(filename, "w")))
 249                                 die_errno("Could not open '%s'", filename);
 250                         if (obj->type == OBJ_BLOB) {
 251                                 if (stream_blob_to_fd(fileno(f), obj->sha1, NULL, 1))
 252                                         die_errno("Could not write '%s'", filename);
 253                         } else
 254                                 fprintf(f, "%s\n", sha1_to_hex(obj->sha1));
 255                         if (fclose(f))
 256                                 die_errno("Could not finish '%s'",
 257                                           filename);
 258                         free(filename);
 259                 }
 260                 return;
 261         }
 262
 263         /*
 264          * Otherwise? It's there, it's unreachable, and some other unreachable
 265          * object points to it. Ignore it - it's not interesting, and we showed
 266          * all the interesting cases above.
 267          */
 268 }
 269
 270 static void check_object(struct object *obj)
 271 {
 272         if (verbose)
 273                 fprintf(stderr, "Checking %s\n", sha1_to_hex(obj->sha1));
 274
 275         if (obj->flags & REACHABLE)
 276                 check_reachable_object(obj);
 277         else
 278                 check_unreachable_object(obj);
 279 }
 280
 281 static void check_connectivity(void)
 282 {
 283         int i, max;
 284
 285         /* Traverse the pending reachable objects */
 286         traverse_reachable();
 287
 288         /* Look up all the requirements, warn about missing objects.. */
 289         max = get_max_object_index();
 290         if (verbose)
 291                 fprintf(stderr, "Checking connectivity (%d objects)\n", max);
 292
 293         for (i = 0; i < max; i++) {
 294                 struct object *obj = get_indexed_object(i);
 295
 296                 if (obj)
 297                         check_object(obj);
 298         }
 299 }
 300
 301 static int fsck_obj(struct object *obj)
 302 {
 303         if (obj->flags & SEEN)
 304                 return 0;
 305         obj->flags |= SEEN;
 306
 307         if (verbose)
 308                 fprintf(stderr, "Checking %s %s\n",
 309                         typename(obj->type), sha1_to_hex(obj->sha1));
 310
 311         if (fsck_walk(obj, NULL, &fsck_obj_options))
 312                 objerror(obj, "broken links");
 313         if (fsck_object(obj, NULL, 0, &fsck_obj_options))
 314                 return -1;
 315
 316         if (obj->type == OBJ_TREE) {
 317                 struct tree *item = (struct tree *) obj;
 318
 319                 free_tree_buffer(item);
 320         }
 321
 322         if (obj->type == OBJ_COMMIT) {
 323                 struct commit *commit = (struct commit *) obj;
 324
 325                 free_commit_buffer(commit);
 326
 327                 if (!commit->parents && show_root)
 328                         printf("root %s\n", sha1_to_hex(commit->object.sha1));
 329         }
 330
 331         if (obj->type == OBJ_TAG) {
 332                 struct tag *tag = (struct tag *) obj;
 333
 334                 if (show_tags && tag->tagged) {
 335                         printf("tagged %s %s", typename(tag->tagged->type), sha1_to_hex(tag->tagged->sha1));
 336                         printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
 337                 }
 338         }
 339
 340         return 0;
 341 }
 342
 343 static int fsck_sha1(const unsigned char *sha1)
 344 {
 345         struct object *obj = parse_object(sha1);
 346         if (!obj) {
 347                 errors_found |= ERROR_OBJECT;
 348                 return error("%s: object corrupt or missing",
 349                              sha1_to_hex(sha1));
 350         }
 351         obj->flags |= HAS_OBJ;
 352         return fsck_obj(obj);
 353 }
 354
 355 static int fsck_obj_buffer(const unsigned char *sha1, enum object_type type,
 356                            unsigned long size, void *buffer, int *eaten)
 357 {
 358         struct object *obj;
 359         obj = parse_object_buffer(sha1, type, size, buffer, eaten);
 360         if (!obj) {
 361                 errors_found |= ERROR_OBJECT;
 362                 return error("%s: object corrupt or missing", sha1_to_hex(sha1));
 363         }
 364         obj->flags = HAS_OBJ;
 365         return fsck_obj(obj);
 366 }
 367
 368 static inline int is_loose_object_file(struct dirent *de,
 369                                        char *name, unsigned char *sha1)
 370 {
 371         if (strlen(de->d_name) != 38)
 372                 return 0;
 373         memcpy(name + 2, de->d_name, 39);
 374         return !get_sha1_hex(name, sha1);
 375 }
 376
 377 static void fsck_dir(int i, char *path)
 378 {
 379         DIR *dir = opendir(path);
 380         struct dirent *de;
 381         char name[100];
 382
 383         if (!dir)
 384                 return;
 385
 386         if (verbose)
 387                 fprintf(stderr, "Checking directory %s\n", path);
 388
 389         sprintf(name, "%02x", i);
 390         while ((de = readdir(dir)) != NULL) {
 391                 unsigned char sha1[20];
 392
 393                 if (is_dot_or_dotdot(de->d_name))
 394                         continue;
 395                 if (is_loose_object_file(de, name, sha1)) {
 396                         if (fsck_sha1(sha1))
 397                                 errors_found |= ERROR_OBJECT;
 398                         continue;
 399                 }
 400                 if (starts_with(de->d_name, "tmp_obj_"))
 401                         continue;
 402                 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
 403         }
 404         closedir(dir);
 405 }
 406
 407 static int default_refs;
 408
 409 static void fsck_handle_reflog_sha1(const char *refname, unsigned char *sha1)
 410 {
 411         struct object *obj;
 412
 413         if (!is_null_sha1(sha1)) {
 414                 obj = lookup_object(sha1);
 415                 if (obj) {
 416                         obj->used = 1;
 417                         mark_object_reachable(obj);
 418                 } else {
 419                         error("%s: invalid reflog entry %s", refname, sha1_to_hex(sha1));
 420                         errors_found |= ERROR_REACHABLE;
 421                 }
 422         }
 423 }
 424
 425 static int fsck_handle_reflog_ent(unsigned char *osha1, unsigned char *nsha1,
 426                 const char *email, unsigned long timestamp, int tz,
 427                 const char *message, void *cb_data)
 428 {
 429         const char *refname = cb_data;
 430
 431         if (verbose)
 432                 fprintf(stderr, "Checking reflog %s->%s\n",
 433                         sha1_to_hex(osha1), sha1_to_hex(nsha1));
 434
 435         fsck_handle_reflog_sha1(refname, osha1);
 436         fsck_handle_reflog_sha1(refname, nsha1);
 437         return 0;
 438 }
 439
 440 static int fsck_handle_reflog(const char *logname, const struct object_id *oid,
 441                               int flag, void *cb_data)
 442 {
 443         for_each_reflog_ent(logname, fsck_handle_reflog_ent, (void *)logname);
 444         return 0;
 445 }
 446
 447 static int fsck_handle_ref(const char *refname, const struct object_id *oid,
 448                            int flag, void *cb_data)
 449 {
 450         struct object *obj;
 451
 452         obj = parse_object(oid->hash);
 453         if (!obj) {
 454                 error("%s: invalid sha1 pointer %s", refname, oid_to_hex(oid));
 455                 errors_found |= ERROR_REACHABLE;
 456                 /* We'll continue with the rest despite the error.. */
 457                 return 0;
 458         }
 459         if (obj->type != OBJ_COMMIT && is_branch(refname))
 460                 error("%s: not a commit", refname);
 461         default_refs++;
 462         obj->used = 1;
 463         mark_object_reachable(obj);
 464
 465         return 0;
 466 }
 467
 468 static void get_default_heads(void)
 469 {
 470         if (head_points_at && !is_null_oid(&head_oid))
 471                 fsck_handle_ref("HEAD", &head_oid, 0, NULL);
 472         for_each_rawref(fsck_handle_ref, NULL);
 473         if (include_reflogs)
 474                 for_each_reflog(fsck_handle_reflog, NULL);
 475
 476         /*
 477          * Not having any default heads isn't really fatal, but
 478          * it does mean that "--unreachable" no longer makes any
 479          * sense (since in this case everything will obviously
 480          * be unreachable by definition.
 481          *
 482          * Showing dangling objects is valid, though (as those
 483          * dangling objects are likely lost heads).
 484          *
 485          * So we just print a warning about it, and clear the
 486          * "show_unreachable" flag.
 487          */
 488         if (!default_refs) {
 489                 fprintf(stderr, "notice: No default references\n");
 490                 show_unreachable = 0;
 491         }
 492 }
 493
 494 static void fsck_object_dir(const char *path)
 495 {
 496         int i;
 497         struct progress *progress = NULL;
 498
 499         if (verbose)
 500                 fprintf(stderr, "Checking object directory\n");
 501
 502         if (show_progress)
 503                 progress = start_progress(_("Checking object directories"), 256);
 504         for (i = 0; i < 256; i++) {
 505                 static char dir[4096];
 506                 sprintf(dir, "%s/%02x", path, i);
 507                 fsck_dir(i, dir);
 508                 display_progress(progress, i+1);
 509         }
 510         stop_progress(&progress);
 511 }
 512
 513 static int fsck_head_link(void)
 514 {
 515         int flag;
 516         int null_is_error = 0;
 517
 518         if (verbose)
 519                 fprintf(stderr, "Checking HEAD link\n");
 520
 521         head_points_at = resolve_ref_unsafe("HEAD", 0, head_oid.hash, &flag);
 522         if (!head_points_at)
 523                 return error("Invalid HEAD");
 524         if (!strcmp(head_points_at, "HEAD"))
 525                 /* detached HEAD */
 526                 null_is_error = 1;
 527         else if (!starts_with(head_points_at, "refs/heads/"))
 528                 return error("HEAD points to something strange (%s)",
 529                              head_points_at);
 530         if (is_null_oid(&head_oid)) {
 531                 if (null_is_error)
 532                         return error("HEAD: detached HEAD points at nothing");
 533                 fprintf(stderr, "notice: HEAD points to an unborn branch (%s)\n",
 534                         head_points_at + 11);
 535         }
 536         return 0;
 537 }
 538
 539 static int fsck_cache_tree(struct cache_tree *it)
 540 {
 541         int i;
 542         int err = 0;
 543
 544         if (verbose)
 545                 fprintf(stderr, "Checking cache tree\n");
 546
 547         if (0 <= it->entry_count) {
 548                 struct object *obj = parse_object(it->sha1);
 549                 if (!obj) {
 550                         error("%s: invalid sha1 pointer in cache-tree",
 551                               sha1_to_hex(it->sha1));
 552                         return 1;
 553                 }
 554                 obj->used = 1;
 555                 mark_object_reachable(obj);
 556                 if (obj->type != OBJ_TREE)
 557                         err |= objerror(obj, "non-tree in cache-tree");
 558         }
 559         for (i = 0; i < it->subtree_nr; i++)
 560                 err |= fsck_cache_tree(it->down[i]->cache_tree);
 561         return err;
 562 }
 563
 564 static char const * const fsck_usage[] = {
 565         N_("git fsck [<options>] [<object>...]"),
 566         NULL
 567 };
 568
 569 static struct option fsck_opts[] = {
 570         OPT__VERBOSE(&verbose, N_("be verbose")),
 571         OPT_BOOL(0, "unreachable", &show_unreachable, N_("show unreachable objects")),
 572         OPT_BOOL(0, "dangling", &show_dangling, N_("show dangling objects")),
 573         OPT_BOOL(0, "tags", &show_tags, N_("report tags")),
 574         OPT_BOOL(0, "root", &show_root, N_("report root nodes")),
 575         OPT_BOOL(0, "cache", &keep_cache_objects, N_("make index objects head nodes")),
 576         OPT_BOOL(0, "reflogs", &include_reflogs, N_("make reflogs head nodes (default)")),
 577         OPT_BOOL(0, "full", &check_full, N_("also consider packs and alternate objects")),
 578         OPT_BOOL(0, "connectivity-only", &connectivity_only, N_("check only connectivity")),
 579         OPT_BOOL(0, "strict", &check_strict, N_("enable more strict checking")),
 580         OPT_BOOL(0, "lost-found", &write_lost_and_found,
 581                                 N_("write dangling objects in .git/lost-found")),
 582         OPT_BOOL(0, "progress", &show_progress, N_("show progress")),
 583         OPT_END(),
 584 };
 585
 586 int cmd_fsck(int argc, const char **argv, const char *prefix)
 587 {
 588         int i, heads;
 589         struct alternate_object_database *alt;
 590
 591         errors_found = 0;
 592         check_replace_refs = 0;
 593
 594         argc = parse_options(argc, argv, prefix, fsck_opts, fsck_usage, 0);
 595
 596         fsck_walk_options.walk = mark_object;
 597         fsck_obj_options.walk = mark_used;
 598         fsck_obj_options.error_func = fsck_error_func;
 599         if (check_strict)
 600                 fsck_obj_options.strict = 1;
 601
 602         if (show_progress == -1)
 603                 show_progress = isatty(2);
 604         if (verbose)
 605                 show_progress = 0;
 606
 607         if (write_lost_and_found) {
 608                 check_full = 1;
 609                 include_reflogs = 0;
 610         }
 611
 612         git_config(fsck_config, NULL);
 613
 614         fsck_head_link();
 615         if (!connectivity_only) {
 616                 fsck_object_dir(get_object_directory());
 617
 618                 prepare_alt_odb();
 619                 for (alt = alt_odb_list; alt; alt = alt->next) {
 620                         /* directory name, minus trailing slash */
 621                         size_t namelen = alt->name - alt->base - 1;
 622                         struct strbuf name = STRBUF_INIT;
 623                         strbuf_add(&name, alt->base, namelen);
 624                         fsck_object_dir(name.buf);
 625                         strbuf_release(&name);
 626                 }
 627         }
 628
 629         if (check_full) {
 630                 struct packed_git *p;
 631                 uint32_t total = 0, count = 0;
 632                 struct progress *progress = NULL;
 633
 634                 prepare_packed_git();
 635
 636                 if (show_progress) {
 637                         for (p = packed_git; p; p = p->next) {
 638                                 if (open_pack_index(p))
 639                                         continue;
 640                                 total += p->num_objects;
 641                         }
 642
 643                         progress = start_progress(_("Checking objects"), total);
 644                 }
 645                 for (p = packed_git; p; p = p->next) {
 646                         /* verify gives error messages itself */
 647                         if (verify_pack(p, fsck_obj_buffer,
 648                                         progress, count))
 649                                 errors_found |= ERROR_PACK;
 650                         count += p->num_objects;
 651                 }
 652                 stop_progress(&progress);
 653         }
 654
 655         heads = 0;
 656         for (i = 0; i < argc; i++) {
 657                 const char *arg = argv[i];
 658                 unsigned char sha1[20];
 659                 if (!get_sha1(arg, sha1)) {
 660                         struct object *obj = lookup_object(sha1);
 661
 662                         /* Error is printed by lookup_object(). */
 663                         if (!obj)
 664                                 continue;
 665
 666                         obj->used = 1;
 667                         mark_object_reachable(obj);
 668                         heads++;
 669                         continue;
 670                 }
 671                 error("invalid parameter: expected sha1, got '%s'", arg);
 672         }
 673
 674         /*
 675          * If we've not been given any explicit head information, do the
 676          * default ones from .git/refs. We also consider the index file
 677          * in this case (ie this implies --cache).
 678          */
 679         if (!heads) {
 680                 get_default_heads();
 681                 keep_cache_objects = 1;
 682         }
 683
 684         if (keep_cache_objects) {
 685                 read_cache();
 686                 for (i = 0; i < active_nr; i++) {
 687                         unsigned int mode;
 688                         struct blob *blob;
 689                         struct object *obj;
 690
 691                         mode = active_cache[i]->ce_mode;
 692                         if (S_ISGITLINK(mode))
 693                                 continue;
 694                         blob = lookup_blob(active_cache[i]->sha1);
 695                         if (!blob)
 696                                 continue;
 697                         obj = &blob->object;
 698                         obj->used = 1;
 699                         mark_object_reachable(obj);
 700                 }
 701                 if (active_cache_tree)
 702                         fsck_cache_tree(active_cache_tree);
 703         }
 704
 705         check_connectivity();
 706         return errors_found;
 707 }