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 #define SEEN      0x0002
  15
  16 static int show_root = 0;
  17 static int show_tags = 0;
  18 static int show_unreachable = 0;
  19 static int check_full = 0;
  20 static int check_strict = 0;
  21 static int keep_cache_objects = 0;
  22 static unsigned char head_sha1[20];
  23
  24 #ifdef NO_D_INO_IN_DIRENT
  25 #define SORT_DIRENT 0
  26 #define DIRENT_SORT_HINT(de) 0
  27 #else
  28 #define SORT_DIRENT 1
  29 #define DIRENT_SORT_HINT(de) ((de)->d_ino)
  30 #endif
  31
  32 static void objreport(struct object *obj, const char *severity,
  33                       const char *err, va_list params)
  34 {
  35         fprintf(stderr, "%s in %s %s: ",
  36                 severity, obj->type, sha1_to_hex(obj->sha1));
  37         vfprintf(stderr, err, params);
  38         fputs("\n", stderr);
  39 }
  40
  41 static int objerror(struct object *obj, const char *err, ...)
  42 {
  43         va_list params;
  44         va_start(params, err);
  45         objreport(obj, "error", err, params);
  46         va_end(params);
  47         return -1;
  48 }
  49
  50 static int objwarning(struct object *obj, const char *err, ...)
  51 {
  52         va_list params;
  53         va_start(params, err);
  54         objreport(obj, "warning", err, params);
  55         va_end(params);
  56         return -1;
  57 }
  58
  59
  60 static void check_connectivity(void)
  61 {
  62         int i;
  63
  64         /* Look up all the requirements, warn about missing objects.. */
  65         for (i = 0; i < obj_allocs; i++) {
  66                 struct object *obj = objs[i];
  67
  68                 if (!obj)
  69                         continue;
  70
  71                 if (!obj->parsed) {
  72                         if (has_sha1_file(obj->sha1))
  73                                 ; /* it is in pack */
  74                         else
  75                                 printf("missing %s %s\n",
  76                                        obj->type, sha1_to_hex(obj->sha1));
  77                         continue;
  78                 }
  79
  80                 if (obj->refs) {
  81                         const struct object_refs *refs = obj->refs;
  82                         unsigned j;
  83                         for (j = 0; j < refs->count; j++) {
  84                                 struct object *ref = refs->ref[j];
  85                                 if (ref->parsed ||
  86                                     (has_sha1_file(ref->sha1)))
  87                                         continue;
  88                                 printf("broken link from %7s %s\n",
  89                                        obj->type, sha1_to_hex(obj->sha1));
  90                                 printf("              to %7s %s\n",
  91                                        ref->type, sha1_to_hex(ref->sha1));
  92                         }
  93                 }
  94
  95                 if (show_unreachable && !(obj->flags & REACHABLE)) {
  96                         printf("unreachable %s %s\n",
  97                                obj->type, sha1_to_hex(obj->sha1));
  98                         continue;
  99                 }
 100
 101                 if (!obj->used) {
 102                         printf("dangling %s %s\n", obj->type,
 103                                sha1_to_hex(obj->sha1));
 104                 }
 105         }
 106 }
 107
 108 /*
 109  * The entries in a tree are ordered in the _path_ order,
 110  * which means that a directory entry is ordered by adding
 111  * a slash to the end of it.
 112  *
 113  * So a directory called "a" is ordered _after_ a file
 114  * called "a.c", because "a/" sorts after "a.c".
 115  */
 116 #define TREE_UNORDERED (-1)
 117 #define TREE_HAS_DUPS  (-2)
 118
 119 static int verify_ordered(struct tree_entry_list *a, struct tree_entry_list *b)
 120 {
 121         int len1 = strlen(a->name);
 122         int len2 = strlen(b->name);
 123         int len = len1 < len2 ? len1 : len2;
 124         unsigned char c1, c2;
 125         int cmp;
 126
 127         cmp = memcmp(a->name, b->name, len);
 128         if (cmp < 0)
 129                 return 0;
 130         if (cmp > 0)
 131                 return TREE_UNORDERED;
 132
 133         /*
 134          * Ok, the first <len> characters are the same.
 135          * Now we need to order the next one, but turn
 136          * a '\0' into a '/' for a directory entry.
 137          */
 138         c1 = a->name[len];
 139         c2 = b->name[len];
 140         if (!c1 && !c2)
 141                 /*
 142                  * git-write-tree used to write out a nonsense tree that has
 143                  * entries with the same name, one blob and one tree.  Make
 144                  * sure we do not have duplicate entries.
 145                  */
 146                 return TREE_HAS_DUPS;
 147         if (!c1 && a->directory)
 148                 c1 = '/';
 149         if (!c2 && b->directory)
 150                 c2 = '/';
 151         return c1 < c2 ? 0 : TREE_UNORDERED;
 152 }
 153
 154 static int fsck_tree(struct tree *item)
 155 {
 156         int retval;
 157         int has_full_path = 0;
 158         int has_zero_pad = 0;
 159         int has_bad_modes = 0;
 160         int has_dup_entries = 0;
 161         int not_properly_sorted = 0;
 162         struct tree_entry_list *entry, *last;
 163
 164         last = NULL;
 165         for (entry = create_tree_entry_list(item); entry; entry = entry->next) {
 166                 if (strchr(entry->name, '/'))
 167                         has_full_path = 1;
 168                 has_zero_pad |= entry->zeropad;
 169
 170                 switch (entry->mode) {
 171                 /*
 172                  * Standard modes..
 173                  */
 174                 case S_IFREG | 0755:
 175                 case S_IFREG | 0644:
 176                 case S_IFLNK:
 177                 case S_IFDIR:
 178                         break;
 179                 /*
 180                  * This is nonstandard, but we had a few of these
 181                  * early on when we honored the full set of mode
 182                  * bits..
 183                  */
 184                 case S_IFREG | 0664:
 185                         if (!check_strict)
 186                                 break;
 187                 default:
 188                         has_bad_modes = 1;
 189                 }
 190
 191                 if (last) {
 192                         switch (verify_ordered(last, entry)) {
 193                         case TREE_UNORDERED:
 194                                 not_properly_sorted = 1;
 195                                 break;
 196                         case TREE_HAS_DUPS:
 197                                 has_dup_entries = 1;
 198                                 break;
 199                         default:
 200                                 break;
 201                         }
 202                         free(last);
 203                 }
 204
 205                 last = entry;
 206         }
 207         if (last)
 208                 free(last);
 209         free(item->buffer);
 210         item->buffer = NULL;
 211
 212         retval = 0;
 213         if (has_full_path) {
 214                 objwarning(&item->object, "contains full pathnames");
 215         }
 216         if (has_zero_pad) {
 217                 objwarning(&item->object, "contains zero-padded file modes");
 218         }
 219         if (has_bad_modes) {
 220                 objwarning(&item->object, "contains bad file modes");
 221         }
 222         if (has_dup_entries) {
 223                 retval = objerror(&item->object, "contains duplicate file entries");
 224         }
 225         if (not_properly_sorted) {
 226                 retval = objerror(&item->object, "not properly sorted");
 227         }
 228         return retval;
 229 }
 230
 231 static int fsck_commit(struct commit *commit)
 232 {
 233         char *buffer = commit->buffer;
 234         unsigned char tree_sha1[20], sha1[20];
 235
 236         if (memcmp(buffer, "tree ", 5))
 237                 return objerror(&commit->object, "invalid format - expected 'tree' line");
 238         if (get_sha1_hex(buffer+5, tree_sha1) || buffer[45] != '\n')
 239                 return objerror(&commit->object, "invalid 'tree' line format - bad sha1");
 240         buffer += 46;
 241         while (!memcmp(buffer, "parent ", 7)) {
 242                 if (get_sha1_hex(buffer+7, sha1) || buffer[47] != '\n')
 243                         return objerror(&commit->object, "invalid 'parent' line format - bad sha1");
 244                 buffer += 48;
 245         }
 246         if (memcmp(buffer, "author ", 7))
 247                 return objerror(&commit->object, "invalid format - expected 'author' line");
 248         free(commit->buffer);
 249         commit->buffer = NULL;
 250         if (!commit->tree)
 251                 return objerror(&commit->object, "could not load commit's tree %s", tree_sha1);
 252         if (!commit->parents && show_root)
 253                 printf("root %s\n", sha1_to_hex(commit->object.sha1));
 254         if (!commit->date)
 255                 printf("bad commit date in %s\n",
 256                        sha1_to_hex(commit->object.sha1));
 257         return 0;
 258 }
 259
 260 static int fsck_tag(struct tag *tag)
 261 {
 262         struct object *tagged = tag->tagged;
 263
 264         if (!tagged) {
 265                 return objerror(&tag->object, "could not load tagged object");
 266         }
 267         if (!show_tags)
 268                 return 0;
 269
 270         printf("tagged %s %s", tagged->type, sha1_to_hex(tagged->sha1));
 271         printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
 272         return 0;
 273 }
 274
 275 static int fsck_sha1(unsigned char *sha1)
 276 {
 277         struct object *obj = parse_object(sha1);
 278         if (!obj)
 279                 return error("%s: object not found", sha1_to_hex(sha1));
 280         if (obj->flags & SEEN)
 281                 return 0;
 282         obj->flags |= SEEN;
 283         if (obj->type == blob_type)
 284                 return 0;
 285         if (obj->type == tree_type)
 286                 return fsck_tree((struct tree *) obj);
 287         if (obj->type == commit_type)
 288                 return fsck_commit((struct commit *) obj);
 289         if (obj->type == tag_type)
 290                 return fsck_tag((struct tag *) obj);
 291         /* By now, parse_object() would've returned NULL instead. */
 292         return objerror(obj, "unknown type '%s' (internal fsck error)", obj->type);
 293 }
 294
 295 /*
 296  * This is the sorting chunk size: make it reasonably
 297  * big so that we can sort well..
 298  */
 299 #define MAX_SHA1_ENTRIES (1024)
 300
 301 struct sha1_entry {
 302         unsigned long ino;
 303         unsigned char sha1[20];
 304 };
 305
 306 static struct {
 307         unsigned long nr;
 308         struct sha1_entry *entry[MAX_SHA1_ENTRIES];
 309 } sha1_list;
 310
 311 static int ino_compare(const void *_a, const void *_b)
 312 {
 313         const struct sha1_entry *a = _a, *b = _b;
 314         unsigned long ino1 = a->ino, ino2 = b->ino;
 315         return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0;
 316 }
 317
 318 static void fsck_sha1_list(void)
 319 {
 320         int i, nr = sha1_list.nr;
 321
 322         if (SORT_DIRENT)
 323                 qsort(sha1_list.entry, nr,
 324                       sizeof(struct sha1_entry *), ino_compare);
 325         for (i = 0; i < nr; i++) {
 326                 struct sha1_entry *entry = sha1_list.entry[i];
 327                 unsigned char *sha1 = entry->sha1;
 328
 329                 sha1_list.entry[i] = NULL;
 330                 fsck_sha1(sha1);
 331                 free(entry);
 332         }
 333         sha1_list.nr = 0;
 334 }
 335
 336 static void add_sha1_list(unsigned char *sha1, unsigned long ino)
 337 {
 338         struct sha1_entry *entry = xmalloc(sizeof(*entry));
 339         int nr;
 340
 341         entry->ino = ino;
 342         memcpy(entry->sha1, sha1, 20);
 343         nr = sha1_list.nr;
 344         if (nr == MAX_SHA1_ENTRIES) {
 345                 fsck_sha1_list();
 346                 nr = 0;
 347         }
 348         sha1_list.entry[nr] = entry;
 349         sha1_list.nr = ++nr;
 350 }
 351
 352 static int fsck_dir(int i, char *path)
 353 {
 354         DIR *dir = opendir(path);
 355         struct dirent *de;
 356
 357         if (!dir)
 358                 return 0;
 359
 360         while ((de = readdir(dir)) != NULL) {
 361                 char name[100];
 362                 unsigned char sha1[20];
 363                 int len = strlen(de->d_name);
 364
 365                 switch (len) {
 366                 case 2:
 367                         if (de->d_name[1] != '.')
 368                                 break;
 369                 case 1:
 370                         if (de->d_name[0] != '.')
 371                                 break;
 372                         continue;
 373                 case 38:
 374                         sprintf(name, "%02x", i);
 375                         memcpy(name+2, de->d_name, len+1);
 376                         if (get_sha1_hex(name, sha1) < 0)
 377                                 break;
 378                         add_sha1_list(sha1, DIRENT_SORT_HINT(de));
 379                         continue;
 380                 }
 381                 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
 382         }
 383         closedir(dir);
 384         return 0;
 385 }
 386
 387 static int default_refs = 0;
 388
 389 static int fsck_handle_ref(const char *refname, const unsigned char *sha1)
 390 {
 391         struct object *obj;
 392
 393         obj = lookup_object(sha1);
 394         if (!obj) {
 395                 if (has_sha1_file(sha1)) {
 396                         default_refs++;
 397                         return 0; /* it is in a pack */
 398                 }
 399                 error("%s: invalid sha1 pointer %s", refname, sha1_to_hex(sha1));
 400                 /* We'll continue with the rest despite the error.. */
 401                 return 0;
 402         }
 403         default_refs++;
 404         obj->used = 1;
 405         mark_reachable(obj, REACHABLE);
 406         return 0;
 407 }
 408
 409 static void get_default_heads(void)
 410 {
 411         for_each_ref(fsck_handle_ref);
 412         if (!default_refs)
 413                 die("No default references");
 414 }
 415
 416 static void fsck_object_dir(const char *path)
 417 {
 418         int i;
 419         for (i = 0; i < 256; i++) {
 420                 static char dir[4096];
 421                 sprintf(dir, "%s/%02x", path, i);
 422                 fsck_dir(i, dir);
 423         }
 424         fsck_sha1_list();
 425 }
 426
 427 static int fsck_head_link(void)
 428 {
 429         unsigned char sha1[20];
 430         const char *git_HEAD = strdup(git_path("HEAD"));
 431         const char *git_refs_heads_master = resolve_ref(git_HEAD, sha1, 1);
 432         int pfxlen = strlen(git_HEAD) - 4; /* strip .../.git/ part */
 433
 434         if (!git_refs_heads_master)
 435                 return error("HEAD is not a symbolic ref");
 436         if (strncmp(git_refs_heads_master + pfxlen, "refs/heads/", 11))
 437                 return error("HEAD points to something strange (%s)",
 438                              git_refs_heads_master + pfxlen);
 439         if (!memcmp(null_sha1, sha1, 20))
 440                 return error("HEAD: not a valid git pointer");
 441         return 0;
 442 }
 443
 444 static int fsck_cache_tree(struct cache_tree *it)
 445 {
 446         int i;
 447         int err = 0;
 448
 449         if (0 <= it->entry_count) {
 450                 struct object *obj = parse_object(it->sha1);
 451                 if (!obj) {
 452                         error("%s: invalid sha1 pointer in cache-tree",
 453                               sha1_to_hex(it->sha1));
 454                         return 1;
 455                 }
 456                 mark_reachable(obj, REACHABLE);
 457                 obj->used = 1;
 458                 if (obj->type != tree_type)
 459                         err |= objerror(obj, "non-tree in cache-tree");
 460         }
 461         for (i = 0; i < it->subtree_nr; i++)
 462                 err |= fsck_cache_tree(it->down[i]->cache_tree);
 463         return err;
 464 }
 465
 466 int main(int argc, char **argv)
 467 {
 468         int i, heads;
 469
 470         track_object_refs = 1;
 471         setup_git_directory();
 472
 473         for (i = 1; i < argc; i++) {
 474                 const char *arg = argv[i];
 475
 476                 if (!strcmp(arg, "--unreachable")) {
 477                         show_unreachable = 1;
 478                         continue;
 479                 }
 480                 if (!strcmp(arg, "--tags")) {
 481                         show_tags = 1;
 482                         continue;
 483                 }
 484                 if (!strcmp(arg, "--root")) {
 485                         show_root = 1;
 486                         continue;
 487                 }
 488                 if (!strcmp(arg, "--cache")) {
 489                         keep_cache_objects = 1;
 490                         continue;
 491                 }
 492                 if (!strcmp(arg, "--full")) {
 493                         check_full = 1;
 494                         continue;
 495                 }
 496                 if (!strcmp(arg, "--strict")) {
 497                         check_strict = 1;
 498                         continue;
 499                 }
 500                 if (*arg == '-')
 501                         usage("git-fsck-objects [--tags] [--root] [[--unreachable] [--cache] [--full] [--strict] <head-sha1>*]");
 502         }
 503
 504         fsck_head_link();
 505         fsck_object_dir(get_object_directory());
 506         if (check_full) {
 507                 struct alternate_object_database *alt;
 508                 struct packed_git *p;
 509                 prepare_alt_odb();
 510                 for (alt = alt_odb_list; alt; alt = alt->next) {
 511                         char namebuf[PATH_MAX];
 512                         int namelen = alt->name - alt->base;
 513                         memcpy(namebuf, alt->base, namelen);
 514                         namebuf[namelen - 1] = 0;
 515                         fsck_object_dir(namebuf);
 516                 }
 517                 prepare_packed_git();
 518                 for (p = packed_git; p; p = p->next)
 519                         /* verify gives error messages itself */
 520                         verify_pack(p, 0);
 521
 522                 for (p = packed_git; p; p = p->next) {
 523                         int num = num_packed_objects(p);
 524                         for (i = 0; i < num; i++) {
 525                                 unsigned char sha1[20];
 526                                 nth_packed_object_sha1(p, i, sha1);
 527                                 fsck_sha1(sha1);
 528                         }
 529                 }
 530         }
 531
 532         heads = 0;
 533         for (i = 1; i < argc; i++) {
 534                 const char *arg = argv[i];
 535
 536                 if (*arg == '-')
 537                         continue;
 538
 539                 if (!get_sha1(arg, head_sha1)) {
 540                         struct object *obj = lookup_object(head_sha1);
 541
 542                         /* Error is printed by lookup_object(). */
 543                         if (!obj)
 544                                 continue;
 545
 546                         obj->used = 1;
 547                         mark_reachable(obj, REACHABLE);
 548                         heads++;
 549                         continue;
 550                 }
 551                 error("invalid parameter: expected sha1, got '%s'", arg);
 552         }
 553
 554         /*
 555          * If we've not been given any explicit head information, do the
 556          * default ones from .git/refs. We also consider the index file
 557          * in this case (ie this implies --cache).
 558          */
 559         if (!heads) {
 560                 get_default_heads();
 561                 keep_cache_objects = 1;
 562         }
 563
 564         if (keep_cache_objects) {
 565                 int i;
 566                 read_cache();
 567                 for (i = 0; i < active_nr; i++) {
 568                         struct blob *blob = lookup_blob(active_cache[i]->sha1);
 569                         struct object *obj;
 570                         if (!blob)
 571                                 continue;
 572                         obj = &blob->object;
 573                         obj->used = 1;
 574                         mark_reachable(obj, REACHABLE);
 575                 }
 576                 if (active_cache_tree)
 577                         fsck_cache_tree(active_cache_tree);
 578         }
 579
 580         check_connectivity();
 581         return 0;
 582 }