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