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
  15 #define REACHABLE 0x0001
  16 #define SEEN      0x0002
  17
  18 static int show_root;
  19 static int show_tags;
  20 static int show_unreachable;
  21 static int include_reflogs = 1;
  22 static int check_full = 1;
  23 static int check_strict;
  24 static int keep_cache_objects;
  25 static unsigned char head_sha1[20];
  26 static const char *head_points_at;
  27 static int errors_found;
  28 static int write_lost_and_found;
  29 static int verbose;
  30 #define ERROR_OBJECT 01
  31 #define ERROR_REACHABLE 02
  32
  33 #ifdef NO_D_INO_IN_DIRENT
  34 #define SORT_DIRENT 0
  35 #define DIRENT_SORT_HINT(de) 0
  36 #else
  37 #define SORT_DIRENT 1
  38 #define DIRENT_SORT_HINT(de) ((de)->d_ino)
  39 #endif
  40
  41 static void objreport(struct object *obj, const char *severity,
  42                       const char *err, va_list params)
  43 {
  44         fprintf(stderr, "%s in %s %s: ",
  45                 severity, typename(obj->type), sha1_to_hex(obj->sha1));
  46         vfprintf(stderr, err, params);
  47         fputs("\n", stderr);
  48 }
  49
  50 __attribute__((format (printf, 2, 3)))
  51 static int objerror(struct object *obj, const char *err, ...)
  52 {
  53         va_list params;
  54         va_start(params, err);
  55         errors_found |= ERROR_OBJECT;
  56         objreport(obj, "error", err, params);
  57         va_end(params);
  58         return -1;
  59 }
  60
  61 __attribute__((format (printf, 3, 4)))
  62 static int fsck_error_func(struct object *obj, int type, const char *err, ...)
  63 {
  64         va_list params;
  65         va_start(params, err);
  66         objreport(obj, (type == FSCK_WARN) ? "warning" : "error", err, params);
  67         va_end(params);
  68         return (type == FSCK_WARN) ? 0 : 1;
  69 }
  70
  71 static struct object_array pending;
  72
  73 static int mark_object(struct object *obj, int type, void *data)
  74 {
  75         struct object *parent = data;
  76
  77         if (!obj) {
  78                 printf("broken link from %7s %s\n",
  79                            typename(parent->type), sha1_to_hex(parent->sha1));
  80                 printf("broken link from %7s %s\n",
  81                            (type == OBJ_ANY ? "unknown" : typename(type)), "unknown");
  82                 errors_found |= ERROR_REACHABLE;
  83                 return 1;
  84         }
  85
  86         if (type != OBJ_ANY && obj->type != type)
  87                 objerror(parent, "wrong object type in link");
  88
  89         if (obj->flags & REACHABLE)
  90                 return 0;
  91         obj->flags |= REACHABLE;
  92         if (!obj->parsed) {
  93                 if (parent && !has_sha1_file(obj->sha1)) {
  94                         printf("broken link from %7s %s\n",
  95                                  typename(parent->type), sha1_to_hex(parent->sha1));
  96                         printf("              to %7s %s\n",
  97                                  typename(obj->type), sha1_to_hex(obj->sha1));
  98                         errors_found |= ERROR_REACHABLE;
  99                 }
 100                 return 1;
 101         }
 102
 103         add_object_array(obj, (void *) parent, &pending);
 104         return 0;
 105 }
 106
 107 static void mark_object_reachable(struct object *obj)
 108 {
 109         mark_object(obj, OBJ_ANY, NULL);
 110 }
 111
 112 static int traverse_one_object(struct object *obj, struct object *parent)
 113 {
 114         int result;
 115         struct tree *tree = NULL;
 116
 117         if (obj->type == OBJ_TREE) {
 118                 obj->parsed = 0;
 119                 tree = (struct tree *)obj;
 120                 if (parse_tree(tree) < 0)
 121                         return 1; /* error already displayed */
 122         }
 123         result = fsck_walk(obj, mark_object, obj);
 124         if (tree) {
 125                 free(tree->buffer);
 126                 tree->buffer = NULL;
 127         }
 128         return result;
 129 }
 130
 131 static int traverse_reachable(void)
 132 {
 133         int result = 0;
 134         while (pending.nr) {
 135                 struct object_array_entry *entry;
 136                 struct object *obj, *parent;
 137
 138                 entry = pending.objects + --pending.nr;
 139                 obj = entry->item;
 140                 parent = (struct object *) entry->name;
 141                 result |= traverse_one_object(obj, parent);
 142         }
 143         return !!result;
 144 }
 145
 146 static int mark_used(struct object *obj, int type, void *data)
 147 {
 148         if (!obj)
 149                 return 1;
 150         obj->used = 1;
 151         return 0;
 152 }
 153
 154 /*
 155  * Check a single reachable object
 156  */
 157 static void check_reachable_object(struct object *obj)
 158 {
 159         /*
 160          * We obviously want the object to be parsed,
 161          * except if it was in a pack-file and we didn't
 162          * do a full fsck
 163          */
 164         if (!obj->parsed) {
 165                 if (has_sha1_pack(obj->sha1))
 166                         return; /* it is in pack - forget about it */
 167                 printf("missing %s %s\n", typename(obj->type), sha1_to_hex(obj->sha1));
 168                 errors_found |= ERROR_REACHABLE;
 169                 return;
 170         }
 171 }
 172
 173 static void dangling_object(struct object *obj)
 174 {
 175         char *filename;
 176         FILE *f;
 177         enum object_type type;
 178         unsigned long size;
 179         char *buf = NULL;
 180         struct pretty_print_context pp_context;
 181
 182         if (!write_lost_and_found)
 183                 goto report_and_exit;
 184
 185         memset(&pp_context, 0, sizeof(pp_context));
 186
 187         filename = git_path("lost-found/%s/%s",
 188                             obj->type == OBJ_COMMIT ? "commit" : "other",
 189                             sha1_to_hex(obj->sha1));
 190
 191         if (safe_create_leading_directories(filename)) {
 192                 error("Could not create lost-found");
 193                 return;
 194         }
 195         if (!(f = fopen(filename, "w")))
 196                 die_errno("Could not open '%s'", filename);
 197         if (obj->type == OBJ_BLOB || obj->type == OBJ_COMMIT)
 198                 buf = read_sha1_file(obj->sha1, &type, &size);
 199
 200         if (obj->type == OBJ_BLOB) {
 201                 if (buf) {
 202                         if (fwrite(buf, size, 1, f) != 1)
 203                                 die_errno("Could not write '%s'", filename);
 204                         free(buf);
 205                 }
 206         } else
 207                 fprintf(f, "%s\n", sha1_to_hex(obj->sha1));
 208         if (fclose(f))
 209                 die_errno("Could not finish '%s'", filename);
 210
 211         if (obj->type == OBJ_COMMIT) {
 212                 struct strbuf sb = STRBUF_INIT;
 213                 struct commit *commit = lookup_commit(obj->sha1);
 214                 int reported = 0;
 215
 216                 if (!commit->buffer)
 217                         commit->buffer = buf;
 218                 if (commit->buffer) {
 219                         parse_commit(commit);
 220                         pretty_print_commit(CMIT_FMT_ONELINE, commit, &sb,
 221                                             &pp_context);
 222                         printf("dangling commit %s (%s)\n",
 223                                sha1_to_hex(obj->sha1), sb.buf);
 224                         reported = 1;
 225                 }
 226                 strbuf_release(&sb);
 227                 free(commit->buffer);
 228                 if (buf && commit->buffer != buf)
 229                         free(buf);
 230                 commit->buffer = NULL;
 231                 if (reported)
 232                         return;
 233         }
 234
 235  report_and_exit:
 236         printf("dangling %s %s\n", typename(obj->type),
 237                sha1_to_hex(obj->sha1));
 238 }
 239
 240 /*
 241  * Check a single unreachable object
 242  */
 243 static void check_unreachable_object(struct object *obj)
 244 {
 245         /*
 246          * Missing unreachable object? Ignore it. It's not like
 247          * we miss it (since it can't be reached), nor do we want
 248          * to complain about it being unreachable (since it does
 249          * not exist).
 250          */
 251         if (!obj->parsed)
 252                 return;
 253
 254         /*
 255          * Unreachable object that exists? Show it if asked to,
 256          * since this is something that is prunable.
 257          */
 258         if (show_unreachable) {
 259                 printf("unreachable %s %s\n", typename(obj->type), sha1_to_hex(obj->sha1));
 260                 return;
 261         }
 262
 263         /*
 264          * "!used" means that nothing at all points to it, including
 265          * other unreachable objects. In other words, it's the "tip"
 266          * of some set of unreachable objects, usually a commit that
 267          * got dropped.
 268          *
 269          * Such starting points are more interesting than some random
 270          * set of unreachable objects, so we show them even if the user
 271          * hasn't asked for _all_ unreachable objects. If you have
 272          * deleted a branch by mistake, this is a prime candidate to
 273          * start looking at, for example.
 274          */
 275         if (!obj->used)
 276                 dangling_object(obj);
 277
 278         /*
 279          * Otherwise? It's there, it's unreachable, and some other unreachable
 280          * object points to it. Ignore it - it's not interesting, and we showed
 281          * all the interesting cases above.
 282          */
 283 }
 284
 285 static void check_object(struct object *obj)
 286 {
 287         if (verbose)
 288                 fprintf(stderr, "Checking %s\n", sha1_to_hex(obj->sha1));
 289
 290         if (obj->flags & REACHABLE)
 291                 check_reachable_object(obj);
 292         else
 293                 check_unreachable_object(obj);
 294 }
 295
 296 static void check_connectivity(void)
 297 {
 298         int i, max;
 299
 300         /* Traverse the pending reachable objects */
 301         traverse_reachable();
 302
 303         /* Look up all the requirements, warn about missing objects.. */
 304         max = get_max_object_index();
 305         if (verbose)
 306                 fprintf(stderr, "Checking connectivity (%d objects)\n", max);
 307
 308         for (i = 0; i < max; i++) {
 309                 struct object *obj = get_indexed_object(i);
 310
 311                 if (obj)
 312                         check_object(obj);
 313         }
 314 }
 315
 316 static int fsck_sha1(const unsigned char *sha1)
 317 {
 318         struct object *obj = parse_object(sha1);
 319         if (!obj) {
 320                 errors_found |= ERROR_OBJECT;
 321                 return error("%s: object corrupt or missing",
 322                              sha1_to_hex(sha1));
 323         }
 324         if (obj->flags & SEEN)
 325                 return 0;
 326         obj->flags |= SEEN;
 327
 328         if (verbose)
 329                 fprintf(stderr, "Checking %s %s\n",
 330                         typename(obj->type), sha1_to_hex(obj->sha1));
 331
 332         if (fsck_walk(obj, mark_used, NULL))
 333                 objerror(obj, "broken links");
 334         if (fsck_object(obj, check_strict, fsck_error_func))
 335                 return -1;
 336
 337         if (obj->type == OBJ_TREE) {
 338                 struct tree *item = (struct tree *) obj;
 339
 340                 free(item->buffer);
 341                 item->buffer = NULL;
 342         }
 343
 344         if (obj->type == OBJ_COMMIT) {
 345                 struct commit *commit = (struct commit *) obj;
 346
 347                 free(commit->buffer);
 348                 commit->buffer = NULL;
 349
 350                 if (!commit->parents && show_root)
 351                         printf("root %s\n", sha1_to_hex(commit->object.sha1));
 352         }
 353
 354         if (obj->type == OBJ_TAG) {
 355                 struct tag *tag = (struct tag *) obj;
 356
 357                 if (show_tags && tag->tagged) {
 358                         printf("tagged %s %s", typename(tag->tagged->type), sha1_to_hex(tag->tagged->sha1));
 359                         printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
 360                 }
 361         }
 362
 363         return 0;
 364 }
 365
 366 /*
 367  * This is the sorting chunk size: make it reasonably
 368  * big so that we can sort well..
 369  */
 370 #define MAX_SHA1_ENTRIES (1024)
 371
 372 struct sha1_entry {
 373         unsigned long ino;
 374         unsigned char sha1[20];
 375 };
 376
 377 static struct {
 378         unsigned long nr;
 379         struct sha1_entry *entry[MAX_SHA1_ENTRIES];
 380 } sha1_list;
 381
 382 static int ino_compare(const void *_a, const void *_b)
 383 {
 384         const struct sha1_entry *a = _a, *b = _b;
 385         unsigned long ino1 = a->ino, ino2 = b->ino;
 386         return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0;
 387 }
 388
 389 static void fsck_sha1_list(void)
 390 {
 391         int i, nr = sha1_list.nr;
 392
 393         if (SORT_DIRENT)
 394                 qsort(sha1_list.entry, nr,
 395                       sizeof(struct sha1_entry *), ino_compare);
 396         for (i = 0; i < nr; i++) {
 397                 struct sha1_entry *entry = sha1_list.entry[i];
 398                 unsigned char *sha1 = entry->sha1;
 399
 400                 sha1_list.entry[i] = NULL;
 401                 fsck_sha1(sha1);
 402                 free(entry);
 403         }
 404         sha1_list.nr = 0;
 405 }
 406
 407 static void add_sha1_list(unsigned char *sha1, unsigned long ino)
 408 {
 409         struct sha1_entry *entry = xmalloc(sizeof(*entry));
 410         int nr;
 411
 412         entry->ino = ino;
 413         hashcpy(entry->sha1, sha1);
 414         nr = sha1_list.nr;
 415         if (nr == MAX_SHA1_ENTRIES) {
 416                 fsck_sha1_list();
 417                 nr = 0;
 418         }
 419         sha1_list.entry[nr] = entry;
 420         sha1_list.nr = ++nr;
 421 }
 422
 423 static void fsck_dir(int i, char *path)
 424 {
 425         DIR *dir = opendir(path);
 426         struct dirent *de;
 427
 428         if (!dir)
 429                 return;
 430
 431         if (verbose)
 432                 fprintf(stderr, "Checking directory %s\n", path);
 433
 434         while ((de = readdir(dir)) != NULL) {
 435                 char name[100];
 436                 unsigned char sha1[20];
 437
 438                 if (is_dot_or_dotdot(de->d_name))
 439                         continue;
 440                 if (strlen(de->d_name) == 38) {
 441                         sprintf(name, "%02x", i);
 442                         memcpy(name+2, de->d_name, 39);
 443                         if (get_sha1_hex(name, sha1) < 0)
 444                                 break;
 445                         add_sha1_list(sha1, DIRENT_SORT_HINT(de));
 446                         continue;
 447                 }
 448                 if (!prefixcmp(de->d_name, "tmp_obj_"))
 449                         continue;
 450                 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
 451         }
 452         closedir(dir);
 453 }
 454
 455 static int default_refs;
 456
 457 static int fsck_handle_reflog_ent(unsigned char *osha1, unsigned char *nsha1,
 458                 const char *email, unsigned long timestamp, int tz,
 459                 const char *message, void *cb_data)
 460 {
 461         struct object *obj;
 462
 463         if (verbose)
 464                 fprintf(stderr, "Checking reflog %s->%s\n",
 465                         sha1_to_hex(osha1), sha1_to_hex(nsha1));
 466
 467         if (!is_null_sha1(osha1)) {
 468                 obj = lookup_object(osha1);
 469                 if (obj) {
 470                         obj->used = 1;
 471                         mark_object_reachable(obj);
 472                 }
 473         }
 474         obj = lookup_object(nsha1);
 475         if (obj) {
 476                 obj->used = 1;
 477                 mark_object_reachable(obj);
 478         }
 479         return 0;
 480 }
 481
 482 static int fsck_handle_reflog(const char *logname, const unsigned char *sha1, int flag, void *cb_data)
 483 {
 484         for_each_reflog_ent(logname, fsck_handle_reflog_ent, NULL);
 485         return 0;
 486 }
 487
 488 static int is_branch(const char *refname)
 489 {
 490         return !strcmp(refname, "HEAD") || !prefixcmp(refname, "refs/heads/");
 491 }
 492
 493 static int fsck_handle_ref(const char *refname, const unsigned char *sha1, int flag, void *cb_data)
 494 {
 495         struct object *obj;
 496
 497         obj = parse_object(sha1);
 498         if (!obj) {
 499                 error("%s: invalid sha1 pointer %s", refname, sha1_to_hex(sha1));
 500                 /* We'll continue with the rest despite the error.. */
 501                 return 0;
 502         }
 503         if (obj->type != OBJ_COMMIT && is_branch(refname))
 504                 error("%s: not a commit", refname);
 505         default_refs++;
 506         obj->used = 1;
 507         mark_object_reachable(obj);
 508
 509         return 0;
 510 }
 511
 512 static void get_default_heads(void)
 513 {
 514         if (head_points_at && !is_null_sha1(head_sha1))
 515                 fsck_handle_ref("HEAD", head_sha1, 0, NULL);
 516         for_each_ref(fsck_handle_ref, NULL);
 517         if (include_reflogs)
 518                 for_each_reflog(fsck_handle_reflog, NULL);
 519
 520         /*
 521          * Not having any default heads isn't really fatal, but
 522          * it does mean that "--unreachable" no longer makes any
 523          * sense (since in this case everything will obviously
 524          * be unreachable by definition.
 525          *
 526          * Showing dangling objects is valid, though (as those
 527          * dangling objects are likely lost heads).
 528          *
 529          * So we just print a warning about it, and clear the
 530          * "show_unreachable" flag.
 531          */
 532         if (!default_refs) {
 533                 fprintf(stderr, "notice: No default references\n");
 534                 show_unreachable = 0;
 535         }
 536 }
 537
 538 static void fsck_object_dir(const char *path)
 539 {
 540         int i;
 541
 542         if (verbose)
 543                 fprintf(stderr, "Checking object directory\n");
 544
 545         for (i = 0; i < 256; i++) {
 546                 static char dir[4096];
 547                 sprintf(dir, "%s/%02x", path, i);
 548                 fsck_dir(i, dir);
 549         }
 550         fsck_sha1_list();
 551 }
 552
 553 static int fsck_head_link(void)
 554 {
 555         int flag;
 556         int null_is_error = 0;
 557
 558         if (verbose)
 559                 fprintf(stderr, "Checking HEAD link\n");
 560
 561         head_points_at = resolve_ref("HEAD", head_sha1, 0, &flag);
 562         if (!head_points_at)
 563                 return error("Invalid HEAD");
 564         if (!strcmp(head_points_at, "HEAD"))
 565                 /* detached HEAD */
 566                 null_is_error = 1;
 567         else if (prefixcmp(head_points_at, "refs/heads/"))
 568                 return error("HEAD points to something strange (%s)",
 569                              head_points_at);
 570         if (is_null_sha1(head_sha1)) {
 571                 if (null_is_error)
 572                         return error("HEAD: detached HEAD points at nothing");
 573                 fprintf(stderr, "notice: HEAD points to an unborn branch (%s)\n",
 574                         head_points_at + 11);
 575         }
 576         return 0;
 577 }
 578
 579 static int fsck_cache_tree(struct cache_tree *it)
 580 {
 581         int i;
 582         int err = 0;
 583
 584         if (verbose)
 585                 fprintf(stderr, "Checking cache tree\n");
 586
 587         if (0 <= it->entry_count) {
 588                 struct object *obj = parse_object(it->sha1);
 589                 if (!obj) {
 590                         error("%s: invalid sha1 pointer in cache-tree",
 591                               sha1_to_hex(it->sha1));
 592                         return 1;
 593                 }
 594                 mark_object_reachable(obj);
 595                 obj->used = 1;
 596                 if (obj->type != OBJ_TREE)
 597                         err |= objerror(obj, "non-tree in cache-tree");
 598         }
 599         for (i = 0; i < it->subtree_nr; i++)
 600                 err |= fsck_cache_tree(it->down[i]->cache_tree);
 601         return err;
 602 }
 603
 604 static char const * const fsck_usage[] = {
 605         "git fsck [options] [<object>...]",
 606         NULL
 607 };
 608
 609 static struct option fsck_opts[] = {
 610         OPT__VERBOSE(&verbose),
 611         OPT_BOOLEAN(0, "unreachable", &show_unreachable, "show unreachable objects"),
 612         OPT_BOOLEAN(0, "tags", &show_tags, "report tags"),
 613         OPT_BOOLEAN(0, "root", &show_root, "report root nodes"),
 614         OPT_BOOLEAN(0, "cache", &keep_cache_objects, "make index objects head nodes"),
 615         OPT_BOOLEAN(0, "reflogs", &include_reflogs, "make reflogs head nodes (default)"),
 616         OPT_BOOLEAN(0, "full", &check_full, "also consider packs and alternate objects"),
 617         OPT_BOOLEAN(0, "strict", &check_strict, "enable more strict checking"),
 618         OPT_BOOLEAN(0, "lost-found", &write_lost_and_found,
 619                                 "write dangling objects in .git/lost-found"),
 620         OPT_END(),
 621 };
 622
 623 int cmd_fsck(int argc, const char **argv, const char *prefix)
 624 {
 625         int i, heads;
 626         struct alternate_object_database *alt;
 627
 628         errors_found = 0;
 629         read_replace_refs = 0;
 630
 631         argc = parse_options(argc, argv, prefix, fsck_opts, fsck_usage, 0);
 632         if (write_lost_and_found) {
 633                 check_full = 1;
 634                 include_reflogs = 0;
 635         }
 636
 637         fsck_head_link();
 638         fsck_object_dir(get_object_directory());
 639
 640         prepare_alt_odb();
 641         for (alt = alt_odb_list; alt; alt = alt->next) {
 642                 char namebuf[PATH_MAX];
 643                 int namelen = alt->name - alt->base;
 644                 memcpy(namebuf, alt->base, namelen);
 645                 namebuf[namelen - 1] = 0;
 646                 fsck_object_dir(namebuf);
 647         }
 648
 649         if (check_full) {
 650                 struct packed_git *p;
 651
 652                 prepare_packed_git();
 653                 for (p = packed_git; p; p = p->next)
 654                         /* verify gives error messages itself */
 655                         verify_pack(p);
 656
 657                 for (p = packed_git; p; p = p->next) {
 658                         uint32_t j, num;
 659                         if (open_pack_index(p))
 660                                 continue;
 661                         num = p->num_objects;
 662                         for (j = 0; j < num; j++)
 663                                 fsck_sha1(nth_packed_object_sha1(p, j));
 664                 }
 665         }
 666
 667         heads = 0;
 668         for (i = 0; i < argc; i++) {
 669                 const char *arg = argv[i];
 670                 unsigned char sha1[20];
 671                 if (!get_sha1(arg, sha1)) {
 672                         struct object *obj = lookup_object(sha1);
 673
 674                         /* Error is printed by lookup_object(). */
 675                         if (!obj)
 676                                 continue;
 677
 678                         obj->used = 1;
 679                         mark_object_reachable(obj);
 680                         heads++;
 681                         continue;
 682                 }
 683                 error("invalid parameter: expected sha1, got '%s'", arg);
 684         }
 685
 686         /*
 687          * If we've not been given any explicit head information, do the
 688          * default ones from .git/refs. We also consider the index file
 689          * in this case (ie this implies --cache).
 690          */
 691         if (!heads) {
 692                 get_default_heads();
 693                 keep_cache_objects = 1;
 694         }
 695
 696         if (keep_cache_objects) {
 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 = active_cache[i]->ce_mode;
 704                         if (S_ISGITLINK(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_object_reachable(obj);
 712                 }
 713                 if (active_cache_tree)
 714                         fsck_cache_tree(active_cache_tree);
 715         }
 716
 717         check_connectivity();
 718         return errors_found;
 719 }