fsck-objects.c

   1 #include <sys/types.h>
   2 #include <dirent.h>
   3
   4 #include "cache.h"
   5 #include "commit.h"
   6 #include "tree.h"
   7 #include "blob.h"
   8 #include "tag.h"
   9 #include "refs.h"
  10 #include "pack.h"
  11 #include "cache-tree.h"
  12
  13 #define REACHABLE 0x0001
  14
  15 static int show_root = 0;
  16 static int show_tags = 0;
  17 static int show_unreachable = 0;
  18 static int check_full = 0;
  19 static int check_strict = 0;
  20 static int keep_cache_objects = 0;
  21 static unsigned char head_sha1[20];
  22
  23 #ifdef NO_D_INO_IN_DIRENT
  24 #define SORT_DIRENT 0
  25 #define DIRENT_SORT_HINT(de) 0
  26 #else
  27 #define SORT_DIRENT 1
  28 #define DIRENT_SORT_HINT(de) ((de)->d_ino)
  29 #endif
  30
  31 static void objreport(struct object *obj, const char *severity,
  32                       const char *err, va_list params)
  33 {
  34         fprintf(stderr, "%s in %s %s: ",
  35                 severity, obj->type, sha1_to_hex(obj->sha1));
  36         vfprintf(stderr, err, params);
  37         fputs("\n", stderr);
  38 }
  39
  40 static int objerror(struct object *obj, const char *err, ...)
  41 {
  42         va_list params;
  43         va_start(params, err);
  44         objreport(obj, "error", err, params);
  45         va_end(params);
  46         return -1;
  47 }
  48
  49 static int objwarning(struct object *obj, const char *err, ...)
  50 {
  51         va_list params;
  52         va_start(params, err);
  53         objreport(obj, "warning", err, params);
  54         va_end(params);
  55         return -1;
  56 }
  57
  58
  59 static void check_connectivity(void)
  60 {
  61         int i;
  62
  63         /* Look up all the requirements, warn about missing objects.. */
  64         for (i = 0; i < obj_allocs; i++) {
  65                 struct object *obj = objs[i];
  66
  67                 if (!obj)
  68                         continue;
  69
  70                 if (!obj->parsed) {
  71                         if (has_sha1_file(obj->sha1))
  72                                 ; /* it is in pack */
  73                         else
  74                                 printf("missing %s %s\n",
  75                                        obj->type, sha1_to_hex(obj->sha1));
  76                         continue;
  77                 }
  78
  79                 if (obj->refs) {
  80                         const struct object_refs *refs = obj->refs;
  81                         unsigned j;
  82                         for (j = 0; j < refs->count; j++) {
  83                                 struct object *ref = refs->ref[j];
  84                                 if (ref->parsed ||
  85                                     (has_sha1_file(ref->sha1)))
  86                                         continue;
  87                                 printf("broken link from %7s %s\n",
  88                                        obj->type, sha1_to_hex(obj->sha1));
  89                                 printf("              to %7s %s\n",
  90                                        ref->type, sha1_to_hex(ref->sha1));
  91                         }
  92                 }
  93
  94                 if (show_unreachable && !(obj->flags & REACHABLE)) {
  95                         printf("unreachable %s %s\n",
  96                                obj->type, sha1_to_hex(obj->sha1));
  97                         continue;
  98                 }
  99
 100                 if (!obj->used) {
 101                         printf("dangling %s %s\n", obj->type,
 102                                sha1_to_hex(obj->sha1));
 103                 }
 104         }
 105 }
 106
 107 /*
 108  * The entries in a tree are ordered in the _path_ order,
 109  * which means that a directory entry is ordered by adding
 110  * a slash to the end of it.
 111  *
 112  * So a directory called "a" is ordered _after_ a file
 113  * called "a.c", because "a/" sorts after "a.c".
 114  */
 115 #define TREE_UNORDERED (-1)
 116 #define TREE_HAS_DUPS  (-2)
 117
 118 static int verify_ordered(struct tree_entry_list *a, struct tree_entry_list *b)
 119 {
 120         int len1 = strlen(a->name);
 121         int len2 = strlen(b->name);
 122         int len = len1 < len2 ? len1 : len2;
 123         unsigned char c1, c2;
 124         int cmp;
 125
 126         cmp = memcmp(a->name, b->name, len);
 127         if (cmp < 0)
 128                 return 0;
 129         if (cmp > 0)
 130                 return TREE_UNORDERED;
 131
 132         /*
 133          * Ok, the first <len> characters are the same.
 134          * Now we need to order the next one, but turn
 135          * a '\0' into a '/' for a directory entry.
 136          */
 137         c1 = a->name[len];
 138         c2 = b->name[len];
 139         if (!c1 && !c2)
 140                 /*
 141                  * git-write-tree used to write out a nonsense tree that has
 142                  * entries with the same name, one blob and one tree.  Make
 143                  * sure we do not have duplicate entries.
 144                  */
 145                 return TREE_HAS_DUPS;
 146         if (!c1 && a->directory)
 147                 c1 = '/';
 148         if (!c2 && b->directory)
 149                 c2 = '/';
 150         return c1 < c2 ? 0 : TREE_UNORDERED;
 151 }
 152
 153 static int fsck_tree(struct tree *item)
 154 {
 155         int retval;
 156         int has_full_path = 0;
 157         int has_zero_pad = 0;
 158         int has_bad_modes = 0;
 159         int has_dup_entries = 0;
 160         int not_properly_sorted = 0;
 161         struct tree_entry_list *entry, *last;
 162
 163         last = NULL;
 164         for (entry = item->entries; entry; entry = entry->next) {
 165                 if (strchr(entry->name, '/'))
 166                         has_full_path = 1;
 167                 has_zero_pad |= entry->zeropad;
 168
 169                 switch (entry->mode) {
 170                 /*
 171                  * Standard modes..
 172                  */
 173                 case S_IFREG | 0755:
 174                 case S_IFREG | 0644:
 175                 case S_IFLNK:
 176                 case S_IFDIR:
 177                         break;
 178                 /*
 179                  * This is nonstandard, but we had a few of these
 180                  * early on when we honored the full set of mode
 181                  * bits..
 182                  */
 183                 case S_IFREG | 0664:
 184                         if (!check_strict)
 185                                 break;
 186                 default:
 187                         has_bad_modes = 1;
 188                 }
 189
 190                 if (last) {
 191                         switch (verify_ordered(last, entry)) {
 192                         case TREE_UNORDERED:
 193                                 not_properly_sorted = 1;
 194                                 break;
 195                         case TREE_HAS_DUPS:
 196                                 has_dup_entries = 1;
 197                                 break;
 198                         default:
 199                                 break;
 200                         }
 201                         free(last->name);
 202                         free(last);
 203                 }
 204
 205                 last = entry;
 206         }
 207         if (last) {
 208                 free(last->name);
 209                 free(last);
 210         }
 211         item->entries = NULL;
 212
 213         retval = 0;
 214         if (has_full_path) {
 215                 objwarning(&item->object, "contains full pathnames");
 216         }
 217         if (has_zero_pad) {
 218                 objwarning(&item->object, "contains zero-padded file modes");
 219         }
 220         if (has_bad_modes) {
 221                 objwarning(&item->object, "contains bad file modes");
 222         }
 223         if (has_dup_entries) {
 224                 retval = objerror(&item->object, "contains duplicate file entries");
 225         }
 226         if (not_properly_sorted) {
 227                 retval = objerror(&item->object, "not properly sorted");
 228         }
 229         return retval;
 230 }
 231
 232 static int fsck_commit(struct commit *commit)
 233 {
 234         char *buffer = commit->buffer;
 235         unsigned char tree_sha1[20], sha1[20];
 236
 237         if (memcmp(buffer, "tree ", 5))
 238                 return objerror(&commit->object, "invalid format - expected 'tree' line");
 239         if (get_sha1_hex(buffer+5, tree_sha1) || buffer[45] != '\n')
 240                 return objerror(&commit->object, "invalid 'tree' line format - bad sha1");
 241         buffer += 46;
 242         while (!memcmp(buffer, "parent ", 7)) {
 243                 if (get_sha1_hex(buffer+7, sha1) || buffer[47] != '\n')
 244                         return objerror(&commit->object, "invalid 'parent' line format - bad sha1");
 245                 buffer += 48;
 246         }
 247         if (memcmp(buffer, "author ", 7))
 248                 return objerror(&commit->object, "invalid format - expected 'author' line");
 249         free(commit->buffer);
 250         commit->buffer = NULL;
 251         if (!commit->tree)
 252                 return objerror(&commit->object, "could not load commit's tree %s", tree_sha1);
 253         if (!commit->parents && show_root)
 254                 printf("root %s\n", sha1_to_hex(commit->object.sha1));
 255         if (!commit->date)
 256                 printf("bad commit date in %s\n",
 257                        sha1_to_hex(commit->object.sha1));
 258         return 0;
 259 }
 260
 261 static int fsck_tag(struct tag *tag)
 262 {
 263         struct object *tagged = tag->tagged;
 264
 265         if (!tagged) {
 266                 return objerror(&tag->object, "could not load tagged object");
 267         }
 268         if (!show_tags)
 269                 return 0;
 270
 271         printf("tagged %s %s", tagged->type, sha1_to_hex(tagged->sha1));
 272         printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
 273         return 0;
 274 }
 275
 276 static int fsck_sha1(unsigned char *sha1)
 277 {
 278         struct object *obj = parse_object(sha1);
 279         if (!obj)
 280                 return error("%s: object not found", sha1_to_hex(sha1));
 281         if (obj->type == blob_type)
 282                 return 0;
 283         if (obj->type == tree_type)
 284                 return fsck_tree((struct tree *) obj);
 285         if (obj->type == commit_type)
 286                 return fsck_commit((struct commit *) obj);
 287         if (obj->type == tag_type)
 288                 return fsck_tag((struct tag *) obj);
 289         /* By now, parse_object() would've returned NULL instead. */
 290         return objerror(obj, "unknown type '%s' (internal fsck error)", obj->type);
 291 }
 292
 293 /*
 294  * This is the sorting chunk size: make it reasonably
 295  * big so that we can sort well..
 296  */
 297 #define MAX_SHA1_ENTRIES (1024)
 298
 299 struct sha1_entry {
 300         unsigned long ino;
 301         unsigned char sha1[20];
 302 };
 303
 304 static struct {
 305         unsigned long nr;
 306         struct sha1_entry *entry[MAX_SHA1_ENTRIES];
 307 } sha1_list;
 308
 309 static int ino_compare(const void *_a, const void *_b)
 310 {
 311         const struct sha1_entry *a = _a, *b = _b;
 312         unsigned long ino1 = a->ino, ino2 = b->ino;
 313         return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0;
 314 }
 315
 316 static void fsck_sha1_list(void)
 317 {
 318         int i, nr = sha1_list.nr;
 319
 320         if (SORT_DIRENT)
 321                 qsort(sha1_list.entry, nr,
 322                       sizeof(struct sha1_entry *), ino_compare);
 323         for (i = 0; i < nr; i++) {
 324                 struct sha1_entry *entry = sha1_list.entry[i];
 325                 unsigned char *sha1 = entry->sha1;
 326
 327                 sha1_list.entry[i] = NULL;
 328                 fsck_sha1(sha1);
 329                 free(entry);
 330         }
 331         sha1_list.nr = 0;
 332 }
 333
 334 static void add_sha1_list(unsigned char *sha1, unsigned long ino)
 335 {
 336         struct sha1_entry *entry = xmalloc(sizeof(*entry));
 337         int nr;
 338
 339         entry->ino = ino;
 340         memcpy(entry->sha1, sha1, 20);
 341         nr = sha1_list.nr;
 342         if (nr == MAX_SHA1_ENTRIES) {
 343                 fsck_sha1_list();
 344                 nr = 0;
 345         }
 346         sha1_list.entry[nr] = entry;
 347         sha1_list.nr = ++nr;
 348 }
 349
 350 static int fsck_dir(int i, char *path)
 351 {
 352         DIR *dir = opendir(path);
 353         struct dirent *de;
 354
 355         if (!dir)
 356                 return 0;
 357
 358         while ((de = readdir(dir)) != NULL) {
 359                 char name[100];
 360                 unsigned char sha1[20];
 361                 int len = strlen(de->d_name);
 362
 363                 switch (len) {
 364                 case 2:
 365                         if (de->d_name[1] != '.')
 366                                 break;
 367                 case 1:
 368                         if (de->d_name[0] != '.')
 369                                 break;
 370                         continue;
 371                 case 38:
 372                         sprintf(name, "%02x", i);
 373                         memcpy(name+2, de->d_name, len+1);
 374                         if (get_sha1_hex(name, sha1) < 0)
 375                                 break;
 376                         add_sha1_list(sha1, DIRENT_SORT_HINT(de));
 377                         continue;
 378                 }
 379                 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
 380         }
 381         closedir(dir);
 382         return 0;
 383 }
 384
 385 static int default_refs = 0;
 386
 387 static int fsck_handle_ref(const char *refname, const unsigned char *sha1)
 388 {
 389         struct object *obj;
 390
 391         obj = lookup_object(sha1);
 392         if (!obj) {
 393                 if (has_sha1_file(sha1)) {
 394                         default_refs++;
 395                         return 0; /* it is in a pack */
 396                 }
 397                 error("%s: invalid sha1 pointer %s", refname, sha1_to_hex(sha1));
 398                 /* We'll continue with the rest despite the error.. */
 399                 return 0;
 400         }
 401         default_refs++;
 402         obj->used = 1;
 403         mark_reachable(obj, REACHABLE);
 404         return 0;
 405 }
 406
 407 static void get_default_heads(void)
 408 {
 409         for_each_ref(fsck_handle_ref);
 410         if (!default_refs)
 411                 die("No default references");
 412 }
 413
 414 static void fsck_object_dir(const char *path)
 415 {
 416         int i;
 417         for (i = 0; i < 256; i++) {
 418                 static char dir[4096];
 419                 sprintf(dir, "%s/%02x", path, i);
 420                 fsck_dir(i, dir);
 421         }
 422         fsck_sha1_list();
 423 }
 424
 425 static int fsck_head_link(void)
 426 {
 427         unsigned char sha1[20];
 428         const char *git_HEAD = strdup(git_path("HEAD"));
 429         const char *git_refs_heads_master = resolve_ref(git_HEAD, sha1, 1);
 430         int pfxlen = strlen(git_HEAD) - 4; /* strip .../.git/ part */
 431
 432         if (!git_refs_heads_master)
 433                 return error("HEAD is not a symbolic ref");
 434         if (strncmp(git_refs_heads_master + pfxlen, "refs/heads/", 11))
 435                 return error("HEAD points to something strange (%s)",
 436                              git_refs_heads_master + pfxlen);
 437         if (!memcmp(null_sha1, sha1, 20))
 438                 return error("HEAD: not a valid git pointer");
 439         return 0;
 440 }
 441
 442 static int fsck_cache_tree(struct cache_tree *it)
 443 {
 444         int i;
 445         int err = 0;
 446
 447         if (0 <= it->entry_count) {
 448                 struct object *obj = parse_object(it->sha1);
 449                 if (obj->type != tree_type)
 450                         err |= objerror(obj, "non-tree in cache-tree");
 451         }
 452         for (i = 0; i < it->subtree_nr; i++)
 453                 err |= fsck_cache_tree(it->down[i]->cache_tree);
 454         return err;
 455 }
 456
 457 int main(int argc, char **argv)
 458 {
 459         int i, heads;
 460
 461         setup_git_directory();
 462
 463         for (i = 1; i < argc; i++) {
 464                 const char *arg = argv[i];
 465
 466                 if (!strcmp(arg, "--unreachable")) {
 467                         show_unreachable = 1;
 468                         continue;
 469                 }
 470                 if (!strcmp(arg, "--tags")) {
 471                         show_tags = 1;
 472                         continue;
 473                 }
 474                 if (!strcmp(arg, "--root")) {
 475                         show_root = 1;
 476                         continue;
 477                 }
 478                 if (!strcmp(arg, "--cache")) {
 479                         keep_cache_objects = 1;
 480                         continue;
 481                 }
 482                 if (!strcmp(arg, "--full")) {
 483                         check_full = 1;
 484                         continue;
 485                 }
 486                 if (!strcmp(arg, "--strict")) {
 487                         check_strict = 1;
 488                         continue;
 489                 }
 490                 if (*arg == '-')
 491                         usage("git-fsck-objects [--tags] [--root] [[--unreachable] [--cache] [--full] [--strict] <head-sha1>*]");
 492         }
 493
 494         fsck_head_link();
 495         fsck_object_dir(get_object_directory());
 496         if (check_full) {
 497                 struct alternate_object_database *alt;
 498                 struct packed_git *p;
 499                 prepare_alt_odb();
 500                 for (alt = alt_odb_list; alt; alt = alt->next) {
 501                         char namebuf[PATH_MAX];
 502                         int namelen = alt->name - alt->base;
 503                         memcpy(namebuf, alt->base, namelen);
 504                         namebuf[namelen - 1] = 0;
 505                         fsck_object_dir(namebuf);
 506                 }
 507                 prepare_packed_git();
 508                 for (p = packed_git; p; p = p->next)
 509                         /* verify gives error messages itself */
 510                         verify_pack(p, 0);
 511
 512                 for (p = packed_git; p; p = p->next) {
 513                         int num = num_packed_objects(p);
 514                         for (i = 0; i < num; i++) {
 515                                 unsigned char sha1[20];
 516                                 nth_packed_object_sha1(p, i, sha1);
 517                                 fsck_sha1(sha1);
 518                         }
 519                 }
 520         }
 521
 522         heads = 0;
 523         for (i = 1; i < argc; i++) {
 524                 const char *arg = argv[i];
 525
 526                 if (*arg == '-')
 527                         continue;
 528
 529                 if (!get_sha1(arg, head_sha1)) {
 530                         struct object *obj = lookup_object(head_sha1);
 531
 532                         /* Error is printed by lookup_object(). */
 533                         if (!obj)
 534                                 continue;
 535
 536                         obj->used = 1;
 537                         mark_reachable(obj, REACHABLE);
 538                         heads++;
 539                         continue;
 540                 }
 541                 error("invalid parameter: expected sha1, got '%s'", arg);
 542         }
 543
 544         /*
 545          * If we've not been given any explicit head information, do the
 546          * default ones from .git/refs. We also consider the index file
 547          * in this case (ie this implies --cache).
 548          */
 549         if (!heads) {
 550                 get_default_heads();
 551                 keep_cache_objects = 1;
 552         }
 553
 554         if (keep_cache_objects) {
 555                 int i;
 556                 read_cache();
 557                 for (i = 0; i < active_nr; i++) {
 558                         struct blob *blob = lookup_blob(active_cache[i]->sha1);
 559                         struct object *obj;
 560                         if (!blob)
 561                                 continue;
 562                         obj = &blob->object;
 563                         obj->used = 1;
 564                         mark_reachable(obj, REACHABLE);
 565                 }
 566                 if (active_cache_tree)
 567                         fsck_cache_tree(active_cache_tree);
 568         }
 569
 570         check_connectivity();
 571         return 0;
 572 }