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
  12 #define REACHABLE 0x0001
  13 #define SEEN      0x0002
  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 = create_tree_entry_list(item); 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);
 202                 }
 203
 204                 last = entry;
 205         }
 206         if (last)
 207                 free(last);
 208         free(item->buffer);
 209         item->buffer = NULL;
 210
 211         retval = 0;
 212         if (has_full_path) {
 213                 objwarning(&item->object, "contains full pathnames");
 214         }
 215         if (has_zero_pad) {
 216                 objwarning(&item->object, "contains zero-padded file modes");
 217         }
 218         if (has_bad_modes) {
 219                 objwarning(&item->object, "contains bad file modes");
 220         }
 221         if (has_dup_entries) {
 222                 retval = objerror(&item->object, "contains duplicate file entries");
 223         }
 224         if (not_properly_sorted) {
 225                 retval = objerror(&item->object, "not properly sorted");
 226         }
 227         return retval;
 228 }
 229
 230 static int fsck_commit(struct commit *commit)
 231 {
 232         char *buffer = commit->buffer;
 233         unsigned char tree_sha1[20], sha1[20];
 234
 235         if (memcmp(buffer, "tree ", 5))
 236                 return objerror(&commit->object, "invalid format - expected 'tree' line");
 237         if (get_sha1_hex(buffer+5, tree_sha1) || buffer[45] != '\n')
 238                 return objerror(&commit->object, "invalid 'tree' line format - bad sha1");
 239         buffer += 46;
 240         while (!memcmp(buffer, "parent ", 7)) {
 241                 if (get_sha1_hex(buffer+7, sha1) || buffer[47] != '\n')
 242                         return objerror(&commit->object, "invalid 'parent' line format - bad sha1");
 243                 buffer += 48;
 244         }
 245         if (memcmp(buffer, "author ", 7))
 246                 return objerror(&commit->object, "invalid format - expected 'author' line");
 247         free(commit->buffer);
 248         commit->buffer = NULL;
 249         if (!commit->tree)
 250                 return objerror(&commit->object, "could not load commit's tree %s", tree_sha1);
 251         if (!commit->parents && show_root)
 252                 printf("root %s\n", sha1_to_hex(commit->object.sha1));
 253         if (!commit->date)
 254                 printf("bad commit date in %s\n",
 255                        sha1_to_hex(commit->object.sha1));
 256         return 0;
 257 }
 258
 259 static int fsck_tag(struct tag *tag)
 260 {
 261         struct object *tagged = tag->tagged;
 262
 263         if (!tagged) {
 264                 return objerror(&tag->object, "could not load tagged object");
 265         }
 266         if (!show_tags)
 267                 return 0;
 268
 269         printf("tagged %s %s", tagged->type, sha1_to_hex(tagged->sha1));
 270         printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
 271         return 0;
 272 }
 273
 274 static int fsck_sha1(unsigned char *sha1)
 275 {
 276         struct object *obj = parse_object(sha1);
 277         if (!obj)
 278                 return error("%s: object not found", sha1_to_hex(sha1));
 279         if (obj->flags & SEEN)
 280                 return 0;
 281         obj->flags |= SEEN;
 282         if (obj->type == blob_type)
 283                 return 0;
 284         if (obj->type == tree_type)
 285                 return fsck_tree((struct tree *) obj);
 286         if (obj->type == commit_type)
 287                 return fsck_commit((struct commit *) obj);
 288         if (obj->type == tag_type)
 289                 return fsck_tag((struct tag *) obj);
 290         /* By now, parse_object() would've returned NULL instead. */
 291         return objerror(obj, "unknown type '%s' (internal fsck error)", obj->type);
 292 }
 293
 294 /*
 295  * This is the sorting chunk size: make it reasonably
 296  * big so that we can sort well..
 297  */
 298 #define MAX_SHA1_ENTRIES (1024)
 299
 300 struct sha1_entry {
 301         unsigned long ino;
 302         unsigned char sha1[20];
 303 };
 304
 305 static struct {
 306         unsigned long nr;
 307         struct sha1_entry *entry[MAX_SHA1_ENTRIES];
 308 } sha1_list;
 309
 310 static int ino_compare(const void *_a, const void *_b)
 311 {
 312         const struct sha1_entry *a = _a, *b = _b;
 313         unsigned long ino1 = a->ino, ino2 = b->ino;
 314         return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0;
 315 }
 316
 317 static void fsck_sha1_list(void)
 318 {
 319         int i, nr = sha1_list.nr;
 320
 321         if (SORT_DIRENT)
 322                 qsort(sha1_list.entry, nr,
 323                       sizeof(struct sha1_entry *), ino_compare);
 324         for (i = 0; i < nr; i++) {
 325                 struct sha1_entry *entry = sha1_list.entry[i];
 326                 unsigned char *sha1 = entry->sha1;
 327
 328                 sha1_list.entry[i] = NULL;
 329                 fsck_sha1(sha1);
 330                 free(entry);
 331         }
 332         sha1_list.nr = 0;
 333 }
 334
 335 static void add_sha1_list(unsigned char *sha1, unsigned long ino)
 336 {
 337         struct sha1_entry *entry = xmalloc(sizeof(*entry));
 338         int nr;
 339
 340         entry->ino = ino;
 341         memcpy(entry->sha1, sha1, 20);
 342         nr = sha1_list.nr;
 343         if (nr == MAX_SHA1_ENTRIES) {
 344                 fsck_sha1_list();
 345                 nr = 0;
 346         }
 347         sha1_list.entry[nr] = entry;
 348         sha1_list.nr = ++nr;
 349 }
 350
 351 static int fsck_dir(int i, char *path)
 352 {
 353         DIR *dir = opendir(path);
 354         struct dirent *de;
 355
 356         if (!dir)
 357                 return 0;
 358
 359         while ((de = readdir(dir)) != NULL) {
 360                 char name[100];
 361                 unsigned char sha1[20];
 362                 int len = strlen(de->d_name);
 363
 364                 switch (len) {
 365                 case 2:
 366                         if (de->d_name[1] != '.')
 367                                 break;
 368                 case 1:
 369                         if (de->d_name[0] != '.')
 370                                 break;
 371                         continue;
 372                 case 38:
 373                         sprintf(name, "%02x", i);
 374                         memcpy(name+2, de->d_name, len+1);
 375                         if (get_sha1_hex(name, sha1) < 0)
 376                                 break;
 377                         add_sha1_list(sha1, DIRENT_SORT_HINT(de));
 378                         continue;
 379                 }
 380                 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
 381         }
 382         closedir(dir);
 383         return 0;
 384 }
 385
 386 static int default_refs = 0;
 387
 388 static int fsck_handle_ref(const char *refname, const unsigned char *sha1)
 389 {
 390         struct object *obj;
 391
 392         obj = lookup_object(sha1);
 393         if (!obj) {
 394                 if (has_sha1_file(sha1)) {
 395                         default_refs++;
 396                         return 0; /* it is in a pack */
 397                 }
 398                 error("%s: invalid sha1 pointer %s", refname, sha1_to_hex(sha1));
 399                 /* We'll continue with the rest despite the error.. */
 400                 return 0;
 401         }
 402         default_refs++;
 403         obj->used = 1;
 404         mark_reachable(obj, REACHABLE);
 405         return 0;
 406 }
 407
 408 static void get_default_heads(void)
 409 {
 410         for_each_ref(fsck_handle_ref);
 411         if (!default_refs)
 412                 die("No default references");
 413 }
 414
 415 static void fsck_object_dir(const char *path)
 416 {
 417         int i;
 418         for (i = 0; i < 256; i++) {
 419                 static char dir[4096];
 420                 sprintf(dir, "%s/%02x", path, i);
 421                 fsck_dir(i, dir);
 422         }
 423         fsck_sha1_list();
 424 }
 425
 426 static int fsck_head_link(void)
 427 {
 428         unsigned char sha1[20];
 429         const char *git_HEAD = strdup(git_path("HEAD"));
 430         const char *git_refs_heads_master = resolve_ref(git_HEAD, sha1, 1);
 431         int pfxlen = strlen(git_HEAD) - 4; /* strip .../.git/ part */
 432
 433         if (!git_refs_heads_master)
 434                 return error("HEAD is not a symbolic ref");
 435         if (strncmp(git_refs_heads_master + pfxlen, "refs/heads/", 11))
 436                 return error("HEAD points to something strange (%s)",
 437                              git_refs_heads_master + pfxlen);
 438         if (!memcmp(null_sha1, sha1, 20))
 439                 return error("HEAD: not a valid git pointer");
 440         return 0;
 441 }
 442
 443 int main(int argc, char **argv)
 444 {
 445         int i, heads;
 446
 447         track_object_refs = 1;
 448         setup_git_directory();
 449
 450         for (i = 1; i < argc; i++) {
 451                 const char *arg = argv[i];
 452
 453                 if (!strcmp(arg, "--unreachable")) {
 454                         show_unreachable = 1;
 455                         continue;
 456                 }
 457                 if (!strcmp(arg, "--tags")) {
 458                         show_tags = 1;
 459                         continue;
 460                 }
 461                 if (!strcmp(arg, "--root")) {
 462                         show_root = 1;
 463                         continue;
 464                 }
 465                 if (!strcmp(arg, "--cache")) {
 466                         keep_cache_objects = 1;
 467                         continue;
 468                 }
 469                 if (!strcmp(arg, "--full")) {
 470                         check_full = 1;
 471                         continue;
 472                 }
 473                 if (!strcmp(arg, "--strict")) {
 474                         check_strict = 1;
 475                         continue;
 476                 }
 477                 if (*arg == '-')
 478                         usage("git-fsck-objects [--tags] [--root] [[--unreachable] [--cache] [--full] [--strict] <head-sha1>*]");
 479         }
 480
 481         fsck_head_link();
 482         fsck_object_dir(get_object_directory());
 483         if (check_full) {
 484                 struct alternate_object_database *alt;
 485                 struct packed_git *p;
 486                 prepare_alt_odb();
 487                 for (alt = alt_odb_list; alt; alt = alt->next) {
 488                         char namebuf[PATH_MAX];
 489                         int namelen = alt->name - alt->base;
 490                         memcpy(namebuf, alt->base, namelen);
 491                         namebuf[namelen - 1] = 0;
 492                         fsck_object_dir(namebuf);
 493                 }
 494                 prepare_packed_git();
 495                 for (p = packed_git; p; p = p->next)
 496                         /* verify gives error messages itself */
 497                         verify_pack(p, 0);
 498
 499                 for (p = packed_git; p; p = p->next) {
 500                         int num = num_packed_objects(p);
 501                         for (i = 0; i < num; i++) {
 502                                 unsigned char sha1[20];
 503                                 nth_packed_object_sha1(p, i, sha1);
 504                                 fsck_sha1(sha1);
 505                         }
 506                 }
 507         }
 508
 509         heads = 0;
 510         for (i = 1; i < argc; i++) {
 511                 const char *arg = argv[i];
 512
 513                 if (*arg == '-')
 514                         continue;
 515
 516                 if (!get_sha1(arg, head_sha1)) {
 517                         struct object *obj = lookup_object(head_sha1);
 518
 519                         /* Error is printed by lookup_object(). */
 520                         if (!obj)
 521                                 continue;
 522
 523                         obj->used = 1;
 524                         mark_reachable(obj, REACHABLE);
 525                         heads++;
 526                         continue;
 527                 }
 528                 error("invalid parameter: expected sha1, got '%s'", arg);
 529         }
 530
 531         /*
 532          * If we've not been given any explicit head information, do the
 533          * default ones from .git/refs. We also consider the index file
 534          * in this case (ie this implies --cache).
 535          */
 536         if (!heads) {
 537                 get_default_heads();
 538                 keep_cache_objects = 1;
 539         }
 540
 541         if (keep_cache_objects) {
 542                 int i;
 543                 read_cache();
 544                 for (i = 0; i < active_nr; i++) {
 545                         struct blob *blob = lookup_blob(active_cache[i]->sha1);
 546                         struct object *obj;
 547                         if (!blob)
 548                                 continue;
 549                         obj = &blob->object;
 550                         obj->used = 1;
 551                         mark_reachable(obj, REACHABLE);
 552                 }
 553         }
 554
 555         check_connectivity();
 556         return 0;
 557 }