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