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