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