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