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