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
 181         if (!write_lost_and_found)
 182                 goto report_and_exit;
 183
 184         filename = git_path("lost-found/%s/%s",
 185                             obj->type == OBJ_COMMIT ? "commit" : "other",
 186                             sha1_to_hex(obj->sha1));
 187
 188         if (safe_create_leading_directories(filename)) {
 189                 error("Could not create lost-found");
 190                 return;
 191         }
 192         if (!(f = fopen(filename, "w")))
 193                 die_errno("Could not open '%s'", filename);
 194         if (obj->type == OBJ_BLOB || obj->type == OBJ_COMMIT)
 195                 buf = read_sha1_file(obj->sha1, &type, &size);
 196
 197         if (obj->type == OBJ_BLOB) {
 198                 if (buf) {
 199                         if (fwrite(buf, size, 1, f) != 1)
 200                                 die_errno("Could not write '%s'", filename);
 201                         free(buf);
 202                 }
 203         } else
 204                 fprintf(f, "%s\n", sha1_to_hex(obj->sha1));
 205         if (fclose(f))
 206                 die_errno("Could not finish '%s'", filename);
 207
 208         if (obj->type == OBJ_COMMIT) {
 209                 struct strbuf sb = STRBUF_INIT;
 210                 struct commit *commit = lookup_commit(obj->sha1);
 211                 int reported = 0;
 212
 213                 if (!commit->buffer)
 214                         commit->buffer = buf;
 215                 if (commit->buffer) {
 216                         parse_commit(commit);
 217                         pretty_print_commit(CMIT_FMT_ONELINE, commit, &sb,
 218                                             0, NULL, NULL, 0, 0);
 219                         printf("dangling commit %s (%s)\n",
 220                                sha1_to_hex(obj->sha1), sb.buf);
 221                         reported = 1;
 222                 }
 223                 strbuf_release(&sb);
 224                 free(commit->buffer);
 225                 if (buf && commit->buffer != buf)
 226                         free(buf);
 227                 commit->buffer = NULL;
 228                 if (reported)
 229                         return;
 230         }
 231
 232  report_and_exit:
 233         printf("dangling %s %s\n", typename(obj->type),
 234                sha1_to_hex(obj->sha1));
 235 }
 236
 237 /*
 238  * Check a single unreachable object
 239  */
 240 static void check_unreachable_object(struct object *obj)
 241 {
 242         /*
 243          * Missing unreachable object? Ignore it. It's not like
 244          * we miss it (since it can't be reached), nor do we want
 245          * to complain about it being unreachable (since it does
 246          * not exist).
 247          */
 248         if (!obj->parsed)
 249                 return;
 250
 251         /*
 252          * Unreachable object that exists? Show it if asked to,
 253          * since this is something that is prunable.
 254          */
 255         if (show_unreachable) {
 256                 printf("unreachable %s %s\n", typename(obj->type), sha1_to_hex(obj->sha1));
 257                 return;
 258         }
 259
 260         /*
 261          * "!used" means that nothing at all points to it, including
 262          * other unreachable objects. In other words, it's the "tip"
 263          * of some set of unreachable objects, usually a commit that
 264          * got dropped.
 265          *
 266          * Such starting points are more interesting than some random
 267          * set of unreachable objects, so we show them even if the user
 268          * hasn't asked for _all_ unreachable objects. If you have
 269          * deleted a branch by mistake, this is a prime candidate to
 270          * start looking at, for example.
 271          */
 272         if (!obj->used)
 273                 dangling_object(obj);
 274
 275         /*
 276          * Otherwise? It's there, it's unreachable, and some other unreachable
 277          * object points to it. Ignore it - it's not interesting, and we showed
 278          * all the interesting cases above.
 279          */
 280 }
 281
 282 static void check_object(struct object *obj)
 283 {
 284         if (verbose)
 285                 fprintf(stderr, "Checking %s\n", sha1_to_hex(obj->sha1));
 286
 287         if (obj->flags & REACHABLE)
 288                 check_reachable_object(obj);
 289         else
 290                 check_unreachable_object(obj);
 291 }
 292
 293 static void check_connectivity(void)
 294 {
 295         int i, max;
 296
 297         /* Traverse the pending reachable objects */
 298         traverse_reachable();
 299
 300         /* Look up all the requirements, warn about missing objects.. */
 301         max = get_max_object_index();
 302         if (verbose)
 303                 fprintf(stderr, "Checking connectivity (%d objects)\n", max);
 304
 305         for (i = 0; i < max; i++) {
 306                 struct object *obj = get_indexed_object(i);
 307
 308                 if (obj)
 309                         check_object(obj);
 310         }
 311 }
 312
 313 static int fsck_sha1(const unsigned char *sha1)
 314 {
 315         struct object *obj = parse_object(sha1);
 316         if (!obj) {
 317                 errors_found |= ERROR_OBJECT;
 318                 return error("%s: object corrupt or missing",
 319                              sha1_to_hex(sha1));
 320         }
 321         if (obj->flags & SEEN)
 322                 return 0;
 323         obj->flags |= SEEN;
 324
 325         if (verbose)
 326                 fprintf(stderr, "Checking %s %s\n",
 327                         typename(obj->type), sha1_to_hex(obj->sha1));
 328
 329         if (fsck_walk(obj, mark_used, NULL))
 330                 objerror(obj, "broken links");
 331         if (fsck_object(obj, check_strict, fsck_error_func))
 332                 return -1;
 333
 334         if (obj->type == OBJ_TREE) {
 335                 struct tree *item = (struct tree *) obj;
 336
 337                 free(item->buffer);
 338                 item->buffer = NULL;
 339         }
 340
 341         if (obj->type == OBJ_COMMIT) {
 342                 struct commit *commit = (struct commit *) obj;
 343
 344                 free(commit->buffer);
 345                 commit->buffer = NULL;
 346
 347                 if (!commit->parents && show_root)
 348                         printf("root %s\n", sha1_to_hex(commit->object.sha1));
 349         }
 350
 351         if (obj->type == OBJ_TAG) {
 352                 struct tag *tag = (struct tag *) obj;
 353
 354                 if (show_tags && tag->tagged) {
 355                         printf("tagged %s %s", typename(tag->tagged->type), sha1_to_hex(tag->tagged->sha1));
 356                         printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
 357                 }
 358         }
 359
 360         return 0;
 361 }
 362
 363 /*
 364  * This is the sorting chunk size: make it reasonably
 365  * big so that we can sort well..
 366  */
 367 #define MAX_SHA1_ENTRIES (1024)
 368
 369 struct sha1_entry {
 370         unsigned long ino;
 371         unsigned char sha1[20];
 372 };
 373
 374 static struct {
 375         unsigned long nr;
 376         struct sha1_entry *entry[MAX_SHA1_ENTRIES];
 377 } sha1_list;
 378
 379 static int ino_compare(const void *_a, const void *_b)
 380 {
 381         const struct sha1_entry *a = _a, *b = _b;
 382         unsigned long ino1 = a->ino, ino2 = b->ino;
 383         return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0;
 384 }
 385
 386 static void fsck_sha1_list(void)
 387 {
 388         int i, nr = sha1_list.nr;
 389
 390         if (SORT_DIRENT)
 391                 qsort(sha1_list.entry, nr,
 392                       sizeof(struct sha1_entry *), ino_compare);
 393         for (i = 0; i < nr; i++) {
 394                 struct sha1_entry *entry = sha1_list.entry[i];
 395                 unsigned char *sha1 = entry->sha1;
 396
 397                 sha1_list.entry[i] = NULL;
 398                 fsck_sha1(sha1);
 399                 free(entry);
 400         }
 401         sha1_list.nr = 0;
 402 }
 403
 404 static void add_sha1_list(unsigned char *sha1, unsigned long ino)
 405 {
 406         struct sha1_entry *entry = xmalloc(sizeof(*entry));
 407         int nr;
 408
 409         entry->ino = ino;
 410         hashcpy(entry->sha1, sha1);
 411         nr = sha1_list.nr;
 412         if (nr == MAX_SHA1_ENTRIES) {
 413                 fsck_sha1_list();
 414                 nr = 0;
 415         }
 416         sha1_list.entry[nr] = entry;
 417         sha1_list.nr = ++nr;
 418 }
 419
 420 static void fsck_dir(int i, char *path)
 421 {
 422         DIR *dir = opendir(path);
 423         struct dirent *de;
 424
 425         if (!dir)
 426                 return;
 427
 428         if (verbose)
 429                 fprintf(stderr, "Checking directory %s\n", path);
 430
 431         while ((de = readdir(dir)) != NULL) {
 432                 char name[100];
 433                 unsigned char sha1[20];
 434
 435                 if (is_dot_or_dotdot(de->d_name))
 436                         continue;
 437                 if (strlen(de->d_name) == 38) {
 438                         sprintf(name, "%02x", i);
 439                         memcpy(name+2, de->d_name, 39);
 440                         if (get_sha1_hex(name, sha1) < 0)
 441                                 break;
 442                         add_sha1_list(sha1, DIRENT_SORT_HINT(de));
 443                         continue;
 444                 }
 445                 if (!prefixcmp(de->d_name, "tmp_obj_"))
 446                         continue;
 447                 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
 448         }
 449         closedir(dir);
 450 }
 451
 452 static int default_refs;
 453
 454 static int fsck_handle_reflog_ent(unsigned char *osha1, unsigned char *nsha1,
 455                 const char *email, unsigned long timestamp, int tz,
 456                 const char *message, void *cb_data)
 457 {
 458         struct object *obj;
 459
 460         if (verbose)
 461                 fprintf(stderr, "Checking reflog %s->%s\n",
 462                         sha1_to_hex(osha1), sha1_to_hex(nsha1));
 463
 464         if (!is_null_sha1(osha1)) {
 465                 obj = lookup_object(osha1);
 466                 if (obj) {
 467                         obj->used = 1;
 468                         mark_object_reachable(obj);
 469                 }
 470         }
 471         obj = lookup_object(nsha1);
 472         if (obj) {
 473                 obj->used = 1;
 474                 mark_object_reachable(obj);
 475         }
 476         return 0;
 477 }
 478
 479 static int fsck_handle_reflog(const char *logname, const unsigned char *sha1, int flag, void *cb_data)
 480 {
 481         for_each_reflog_ent(logname, fsck_handle_reflog_ent, NULL);
 482         return 0;
 483 }
 484
 485 static int is_branch(const char *refname)
 486 {
 487         return !strcmp(refname, "HEAD") || !prefixcmp(refname, "refs/heads/");
 488 }
 489
 490 static int fsck_handle_ref(const char *refname, const unsigned char *sha1, int flag, void *cb_data)
 491 {
 492         struct object *obj;
 493
 494         obj = parse_object(sha1);
 495         if (!obj) {
 496                 error("%s: invalid sha1 pointer %s", refname, sha1_to_hex(sha1));
 497                 /* We'll continue with the rest despite the error.. */
 498                 return 0;
 499         }
 500         if (obj->type != OBJ_COMMIT && is_branch(refname))
 501                 error("%s: not a commit", refname);
 502         default_refs++;
 503         obj->used = 1;
 504         mark_object_reachable(obj);
 505
 506         return 0;
 507 }
 508
 509 static void get_default_heads(void)
 510 {
 511         if (head_points_at && !is_null_sha1(head_sha1))
 512                 fsck_handle_ref("HEAD", head_sha1, 0, NULL);
 513         for_each_ref(fsck_handle_ref, NULL);
 514         if (include_reflogs)
 515                 for_each_reflog(fsck_handle_reflog, NULL);
 516
 517         /*
 518          * Not having any default heads isn't really fatal, but
 519          * it does mean that "--unreachable" no longer makes any
 520          * sense (since in this case everything will obviously
 521          * be unreachable by definition.
 522          *
 523          * Showing dangling objects is valid, though (as those
 524          * dangling objects are likely lost heads).
 525          *
 526          * So we just print a warning about it, and clear the
 527          * "show_unreachable" flag.
 528          */
 529         if (!default_refs) {
 530                 fprintf(stderr, "notice: No default references\n");
 531                 show_unreachable = 0;
 532         }
 533 }
 534
 535 static void fsck_object_dir(const char *path)
 536 {
 537         int i;
 538
 539         if (verbose)
 540                 fprintf(stderr, "Checking object directory\n");
 541
 542         for (i = 0; i < 256; i++) {
 543                 static char dir[4096];
 544                 sprintf(dir, "%s/%02x", path, i);
 545                 fsck_dir(i, dir);
 546         }
 547         fsck_sha1_list();
 548 }
 549
 550 static int fsck_head_link(void)
 551 {
 552         int flag;
 553         int null_is_error = 0;
 554
 555         if (verbose)
 556                 fprintf(stderr, "Checking HEAD link\n");
 557
 558         head_points_at = resolve_ref("HEAD", head_sha1, 0, &flag);
 559         if (!head_points_at)
 560                 return error("Invalid HEAD");
 561         if (!strcmp(head_points_at, "HEAD"))
 562                 /* detached HEAD */
 563                 null_is_error = 1;
 564         else if (prefixcmp(head_points_at, "refs/heads/"))
 565                 return error("HEAD points to something strange (%s)",
 566                              head_points_at);
 567         if (is_null_sha1(head_sha1)) {
 568                 if (null_is_error)
 569                         return error("HEAD: detached HEAD points at nothing");
 570                 fprintf(stderr, "notice: HEAD points to an unborn branch (%s)\n",
 571                         head_points_at + 11);
 572         }
 573         return 0;
 574 }
 575
 576 static int fsck_cache_tree(struct cache_tree *it)
 577 {
 578         int i;
 579         int err = 0;
 580
 581         if (verbose)
 582                 fprintf(stderr, "Checking cache tree\n");
 583
 584         if (0 <= it->entry_count) {
 585                 struct object *obj = parse_object(it->sha1);
 586                 if (!obj) {
 587                         error("%s: invalid sha1 pointer in cache-tree",
 588                               sha1_to_hex(it->sha1));
 589                         return 1;
 590                 }
 591                 mark_object_reachable(obj);
 592                 obj->used = 1;
 593                 if (obj->type != OBJ_TREE)
 594                         err |= objerror(obj, "non-tree in cache-tree");
 595         }
 596         for (i = 0; i < it->subtree_nr; i++)
 597                 err |= fsck_cache_tree(it->down[i]->cache_tree);
 598         return err;
 599 }
 600
 601 static char const * const fsck_usage[] = {
 602         "git fsck [options] [<object>...]",
 603         NULL
 604 };
 605
 606 static struct option fsck_opts[] = {
 607         OPT__VERBOSE(&verbose),
 608         OPT_BOOLEAN(0, "unreachable", &show_unreachable, "show unreachable objects"),
 609         OPT_BOOLEAN(0, "tags", &show_tags, "report tags"),
 610         OPT_BOOLEAN(0, "root", &show_root, "report root nodes"),
 611         OPT_BOOLEAN(0, "cache", &keep_cache_objects, "make index objects head nodes"),
 612         OPT_BOOLEAN(0, "reflogs", &include_reflogs, "make reflogs head nodes (default)"),
 613         OPT_BOOLEAN(0, "full", &check_full, "also consider packs and alternate objects"),
 614         OPT_BOOLEAN(0, "strict", &check_strict, "enable more strict checking"),
 615         OPT_BOOLEAN(0, "lost-found", &write_lost_and_found,
 616                                 "write dangling objects in .git/lost-found"),
 617         OPT_END(),
 618 };
 619
 620 int cmd_fsck(int argc, const char **argv, const char *prefix)
 621 {
 622         int i, heads;
 623         struct alternate_object_database *alt;
 624
 625         errors_found = 0;
 626         read_replace_refs = 0;
 627
 628         argc = parse_options(argc, argv, prefix, fsck_opts, fsck_usage, 0);
 629         if (write_lost_and_found) {
 630                 check_full = 1;
 631                 include_reflogs = 0;
 632         }
 633
 634         fsck_head_link();
 635         fsck_object_dir(get_object_directory());
 636
 637         prepare_alt_odb();
 638         for (alt = alt_odb_list; alt; alt = alt->next) {
 639                 char namebuf[PATH_MAX];
 640                 int namelen = alt->name - alt->base;
 641                 memcpy(namebuf, alt->base, namelen);
 642                 namebuf[namelen - 1] = 0;
 643                 fsck_object_dir(namebuf);
 644         }
 645
 646         if (check_full) {
 647                 struct packed_git *p;
 648
 649                 prepare_packed_git();
 650                 for (p = packed_git; p; p = p->next)
 651                         /* verify gives error messages itself */
 652                         verify_pack(p);
 653
 654                 for (p = packed_git; p; p = p->next) {
 655                         uint32_t j, num;
 656                         if (open_pack_index(p))
 657                                 continue;
 658                         num = p->num_objects;
 659                         for (j = 0; j < num; j++)
 660                                 fsck_sha1(nth_packed_object_sha1(p, j));
 661                 }
 662         }
 663
 664         heads = 0;
 665         for (i = 0; i < argc; i++) {
 666                 const char *arg = argv[i];
 667                 unsigned char sha1[20];
 668                 if (!get_sha1(arg, sha1)) {
 669                         struct object *obj = lookup_object(sha1);
 670
 671                         /* Error is printed by lookup_object(). */
 672                         if (!obj)
 673                                 continue;
 674
 675                         obj->used = 1;
 676                         mark_object_reachable(obj);
 677                         heads++;
 678                         continue;
 679                 }
 680                 error("invalid parameter: expected sha1, got '%s'", arg);
 681         }
 682
 683         /*
 684          * If we've not been given any explicit head information, do the
 685          * default ones from .git/refs. We also consider the index file
 686          * in this case (ie this implies --cache).
 687          */
 688         if (!heads) {
 689                 get_default_heads();
 690                 keep_cache_objects = 1;
 691         }
 692
 693         if (keep_cache_objects) {
 694                 read_cache();
 695                 for (i = 0; i < active_nr; i++) {
 696                         unsigned int mode;
 697                         struct blob *blob;
 698                         struct object *obj;
 699
 700                         mode = active_cache[i]->ce_mode;
 701                         if (S_ISGITLINK(mode))
 702                                 continue;
 703                         blob = lookup_blob(active_cache[i]->sha1);
 704                         if (!blob)
 705                                 continue;
 706                         obj = &blob->object;
 707                         obj->used = 1;
 708                         mark_object_reachable(obj);
 709                 }
 710                 if (active_cache_tree)
 711                         fsck_cache_tree(active_cache_tree);
 712         }
 713
 714         check_connectivity();
 715         return errors_found;
 716 }