git.oblomov.eu Git - git/blob - fsck.c

   1 #include "cache.h"
   2 #include "object-store.h"
   3 #include "repository.h"
   4 #include "object.h"
   5 #include "blob.h"
   6 #include "tree.h"
   7 #include "tree-walk.h"
   8 #include "commit.h"
   9 #include "tag.h"
  10 #include "fsck.h"
  11 #include "refs.h"
  12 #include "utf8.h"
  13 #include "decorate.h"
  14 #include "oidset.h"
  15 #include "packfile.h"
  16 #include "submodule-config.h"
  17 #include "config.h"
  18 #include "help.h"
  19
  20 static struct oidset gitmodules_found = OIDSET_INIT;
  21 static struct oidset gitmodules_done = OIDSET_INIT;
  22
  23 #define FSCK_FATAL -1
  24 #define FSCK_INFO -2
  25
  26 #define FOREACH_MSG_ID(FUNC) \
  27         /* fatal errors */ \
  28         FUNC(NUL_IN_HEADER, FATAL) \
  29         FUNC(UNTERMINATED_HEADER, FATAL) \
  30         /* errors */ \
  31         FUNC(BAD_DATE, ERROR) \
  32         FUNC(BAD_DATE_OVERFLOW, ERROR) \
  33         FUNC(BAD_EMAIL, ERROR) \
  34         FUNC(BAD_NAME, ERROR) \
  35         FUNC(BAD_OBJECT_SHA1, ERROR) \
  36         FUNC(BAD_PARENT_SHA1, ERROR) \
  37         FUNC(BAD_TAG_OBJECT, ERROR) \
  38         FUNC(BAD_TIMEZONE, ERROR) \
  39         FUNC(BAD_TREE, ERROR) \
  40         FUNC(BAD_TREE_SHA1, ERROR) \
  41         FUNC(BAD_TYPE, ERROR) \
  42         FUNC(DUPLICATE_ENTRIES, ERROR) \
  43         FUNC(MISSING_AUTHOR, ERROR) \
  44         FUNC(MISSING_COMMITTER, ERROR) \
  45         FUNC(MISSING_EMAIL, ERROR) \
  46         FUNC(MISSING_NAME_BEFORE_EMAIL, ERROR) \
  47         FUNC(MISSING_OBJECT, ERROR) \
  48         FUNC(MISSING_SPACE_BEFORE_DATE, ERROR) \
  49         FUNC(MISSING_SPACE_BEFORE_EMAIL, ERROR) \
  50         FUNC(MISSING_TAG, ERROR) \
  51         FUNC(MISSING_TAG_ENTRY, ERROR) \
  52         FUNC(MISSING_TREE, ERROR) \
  53         FUNC(MISSING_TREE_OBJECT, ERROR) \
  54         FUNC(MISSING_TYPE, ERROR) \
  55         FUNC(MISSING_TYPE_ENTRY, ERROR) \
  56         FUNC(MULTIPLE_AUTHORS, ERROR) \
  57         FUNC(TREE_NOT_SORTED, ERROR) \
  58         FUNC(UNKNOWN_TYPE, ERROR) \
  59         FUNC(ZERO_PADDED_DATE, ERROR) \
  60         FUNC(GITMODULES_MISSING, ERROR) \
  61         FUNC(GITMODULES_BLOB, ERROR) \
  62         FUNC(GITMODULES_LARGE, ERROR) \
  63         FUNC(GITMODULES_NAME, ERROR) \
  64         FUNC(GITMODULES_SYMLINK, ERROR) \
  65         FUNC(GITMODULES_URL, ERROR) \
  66         FUNC(GITMODULES_PATH, ERROR) \
  67         /* warnings */ \
  68         FUNC(BAD_FILEMODE, WARN) \
  69         FUNC(EMPTY_NAME, WARN) \
  70         FUNC(FULL_PATHNAME, WARN) \
  71         FUNC(HAS_DOT, WARN) \
  72         FUNC(HAS_DOTDOT, WARN) \
  73         FUNC(HAS_DOTGIT, WARN) \
  74         FUNC(NULL_SHA1, WARN) \
  75         FUNC(ZERO_PADDED_FILEMODE, WARN) \
  76         FUNC(NUL_IN_COMMIT, WARN) \
  77         /* infos (reported as warnings, but ignored by default) */ \
  78         FUNC(GITMODULES_PARSE, INFO) \
  79         FUNC(BAD_TAG_NAME, INFO) \
  80         FUNC(MISSING_TAGGER_ENTRY, INFO)
  81
  82 #define MSG_ID(id, msg_type) FSCK_MSG_##id,
  83 enum fsck_msg_id {
  84         FOREACH_MSG_ID(MSG_ID)
  85         FSCK_MSG_MAX
  86 };
  87 #undef MSG_ID
  88
  89 #define STR(x) #x
  90 #define MSG_ID(id, msg_type) { STR(id), NULL, NULL, FSCK_##msg_type },
  91 static struct {
  92         const char *id_string;
  93         const char *downcased;
  94         const char *camelcased;
  95         int msg_type;
  96 } msg_id_info[FSCK_MSG_MAX + 1] = {
  97         FOREACH_MSG_ID(MSG_ID)
  98         { NULL, NULL, NULL, -1 }
  99 };
 100 #undef MSG_ID
 101
 102 static void prepare_msg_ids(void)
 103 {
 104         int i;
 105
 106         if (msg_id_info[0].downcased)
 107                 return;
 108
 109         /* convert id_string to lower case, without underscores. */
 110         for (i = 0; i < FSCK_MSG_MAX; i++) {
 111                 const char *p = msg_id_info[i].id_string;
 112                 int len = strlen(p);
 113                 char *q = xmalloc(len);
 114
 115                 msg_id_info[i].downcased = q;
 116                 while (*p)
 117                         if (*p == '_')
 118                                 p++;
 119                         else
 120                                 *(q)++ = tolower(*(p)++);
 121                 *q = '\0';
 122
 123                 p = msg_id_info[i].id_string;
 124                 q = xmalloc(len);
 125                 msg_id_info[i].camelcased = q;
 126                 while (*p) {
 127                         if (*p == '_') {
 128                                 p++;
 129                                 if (*p)
 130                                         *q++ = *p++;
 131                         } else {
 132                                 *q++ = tolower(*p++);
 133                         }
 134                 }
 135                 *q = '\0';
 136         }
 137 }
 138
 139 static int parse_msg_id(const char *text)
 140 {
 141         int i;
 142
 143         prepare_msg_ids();
 144
 145         for (i = 0; i < FSCK_MSG_MAX; i++)
 146                 if (!strcmp(text, msg_id_info[i].downcased))
 147                         return i;
 148
 149         return -1;
 150 }
 151
 152 void list_config_fsck_msg_ids(struct string_list *list, const char *prefix)
 153 {
 154         int i;
 155
 156         prepare_msg_ids();
 157
 158         for (i = 0; i < FSCK_MSG_MAX; i++)
 159                 list_config_item(list, prefix, msg_id_info[i].camelcased);
 160 }
 161
 162 static int fsck_msg_type(enum fsck_msg_id msg_id,
 163         struct fsck_options *options)
 164 {
 165         int msg_type;
 166
 167         assert(msg_id >= 0 && msg_id < FSCK_MSG_MAX);
 168
 169         if (options->msg_type)
 170                 msg_type = options->msg_type[msg_id];
 171         else {
 172                 msg_type = msg_id_info[msg_id].msg_type;
 173                 if (options->strict && msg_type == FSCK_WARN)
 174                         msg_type = FSCK_ERROR;
 175         }
 176
 177         return msg_type;
 178 }
 179
 180 static int parse_msg_type(const char *str)
 181 {
 182         if (!strcmp(str, "error"))
 183                 return FSCK_ERROR;
 184         else if (!strcmp(str, "warn"))
 185                 return FSCK_WARN;
 186         else if (!strcmp(str, "ignore"))
 187                 return FSCK_IGNORE;
 188         else
 189                 die("Unknown fsck message type: '%s'", str);
 190 }
 191
 192 int is_valid_msg_type(const char *msg_id, const char *msg_type)
 193 {
 194         if (parse_msg_id(msg_id) < 0)
 195                 return 0;
 196         parse_msg_type(msg_type);
 197         return 1;
 198 }
 199
 200 void fsck_set_msg_type(struct fsck_options *options,
 201                 const char *msg_id, const char *msg_type)
 202 {
 203         int id = parse_msg_id(msg_id), type;
 204
 205         if (id < 0)
 206                 die("Unhandled message id: %s", msg_id);
 207         type = parse_msg_type(msg_type);
 208
 209         if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
 210                 die("Cannot demote %s to %s", msg_id, msg_type);
 211
 212         if (!options->msg_type) {
 213                 int i;
 214                 int *msg_type;
 215                 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
 216                 for (i = 0; i < FSCK_MSG_MAX; i++)
 217                         msg_type[i] = fsck_msg_type(i, options);
 218                 options->msg_type = msg_type;
 219         }
 220
 221         options->msg_type[id] = type;
 222 }
 223
 224 void fsck_set_msg_types(struct fsck_options *options, const char *values)
 225 {
 226         char *buf = xstrdup(values), *to_free = buf;
 227         int done = 0;
 228
 229         while (!done) {
 230                 int len = strcspn(buf, " ,|"), equal;
 231
 232                 done = !buf[len];
 233                 if (!len) {
 234                         buf++;
 235                         continue;
 236                 }
 237                 buf[len] = '\0';
 238
 239                 for (equal = 0;
 240                      equal < len && buf[equal] != '=' && buf[equal] != ':';
 241                      equal++)
 242                         buf[equal] = tolower(buf[equal]);
 243                 buf[equal] = '\0';
 244
 245                 if (!strcmp(buf, "skiplist")) {
 246                         if (equal == len)
 247                                 die("skiplist requires a path");
 248                         oidset_parse_file(&options->skiplist, buf + equal + 1);
 249                         buf += len + 1;
 250                         continue;
 251                 }
 252
 253                 if (equal == len)
 254                         die("Missing '=': '%s'", buf);
 255
 256                 fsck_set_msg_type(options, buf, buf + equal + 1);
 257                 buf += len + 1;
 258         }
 259         free(to_free);
 260 }
 261
 262 static void append_msg_id(struct strbuf *sb, const char *msg_id)
 263 {
 264         for (;;) {
 265                 char c = *(msg_id)++;
 266
 267                 if (!c)
 268                         break;
 269                 if (c != '_')
 270                         strbuf_addch(sb, tolower(c));
 271                 else {
 272                         assert(*msg_id);
 273                         strbuf_addch(sb, *(msg_id)++);
 274                 }
 275         }
 276
 277         strbuf_addstr(sb, ": ");
 278 }
 279
 280 static int object_on_skiplist(struct fsck_options *opts,
 281                               const struct object_id *oid)
 282 {
 283         return opts && oid && oidset_contains(&opts->skiplist, oid);
 284 }
 285
 286 __attribute__((format (printf, 4, 5)))
 287 static int report(struct fsck_options *options, struct object *object,
 288         enum fsck_msg_id id, const char *fmt, ...)
 289 {
 290         va_list ap;
 291         struct strbuf sb = STRBUF_INIT;
 292         int msg_type = fsck_msg_type(id, options), result;
 293
 294         if (msg_type == FSCK_IGNORE)
 295                 return 0;
 296
 297         if (object_on_skiplist(options, &object->oid))
 298                 return 0;
 299
 300         if (msg_type == FSCK_FATAL)
 301                 msg_type = FSCK_ERROR;
 302         else if (msg_type == FSCK_INFO)
 303                 msg_type = FSCK_WARN;
 304
 305         append_msg_id(&sb, msg_id_info[id].id_string);
 306
 307         va_start(ap, fmt);
 308         strbuf_vaddf(&sb, fmt, ap);
 309         result = options->error_func(options, &object->oid, object->type,
 310                                      msg_type, sb.buf);
 311         strbuf_release(&sb);
 312         va_end(ap);
 313
 314         return result;
 315 }
 316
 317 void fsck_enable_object_names(struct fsck_options *options)
 318 {
 319         if (!options->object_names)
 320                 options->object_names = kh_init_oid_map();
 321 }
 322
 323 const char *fsck_get_object_name(struct fsck_options *options,
 324                                  const struct object_id *oid)
 325 {
 326         khiter_t pos;
 327         if (!options->object_names)
 328                 return NULL;
 329         pos = kh_get_oid_map(options->object_names, *oid);
 330         if (pos >= kh_end(options->object_names))
 331                 return NULL;
 332         return kh_value(options->object_names, pos);
 333 }
 334
 335 void fsck_put_object_name(struct fsck_options *options,
 336                           const struct object_id *oid,
 337                           const char *fmt, ...)
 338 {
 339         va_list ap;
 340         struct strbuf buf = STRBUF_INIT;
 341         khiter_t pos;
 342         int hashret;
 343
 344         if (!options->object_names)
 345                 return;
 346
 347         pos = kh_put_oid_map(options->object_names, *oid, &hashret);
 348         if (!hashret)
 349                 return;
 350         va_start(ap, fmt);
 351         strbuf_vaddf(&buf, fmt, ap);
 352         kh_value(options->object_names, pos) = strbuf_detach(&buf, NULL);
 353         va_end(ap);
 354 }
 355
 356 const char *fsck_describe_object(struct fsck_options *options,
 357                                  const struct object_id *oid)
 358 {
 359         static struct strbuf bufs[] = {
 360                 STRBUF_INIT, STRBUF_INIT, STRBUF_INIT, STRBUF_INIT
 361         };
 362         static int b = 0;
 363         struct strbuf *buf;
 364         const char *name = fsck_get_object_name(options, oid);
 365
 366         buf = bufs + b;
 367         b = (b + 1) % ARRAY_SIZE(bufs);
 368         strbuf_reset(buf);
 369         strbuf_addstr(buf, oid_to_hex(oid));
 370         if (name)
 371                 strbuf_addf(buf, " (%s)", name);
 372
 373         return buf->buf;
 374 }
 375
 376 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
 377 {
 378         struct tree_desc desc;
 379         struct name_entry entry;
 380         int res = 0;
 381         const char *name;
 382
 383         if (parse_tree(tree))
 384                 return -1;
 385
 386         name = fsck_get_object_name(options, &tree->object.oid);
 387         if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
 388                 return -1;
 389         while (tree_entry_gently(&desc, &entry)) {
 390                 struct object *obj;
 391                 int result;
 392
 393                 if (S_ISGITLINK(entry.mode))
 394                         continue;
 395
 396                 if (S_ISDIR(entry.mode)) {
 397                         obj = (struct object *)lookup_tree(the_repository, &entry.oid);
 398                         if (name && obj)
 399                                 fsck_put_object_name(options, &entry.oid, "%s%s/",
 400                                                      name, entry.path);
 401                         result = options->walk(obj, OBJ_TREE, data, options);
 402                 }
 403                 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
 404                         obj = (struct object *)lookup_blob(the_repository, &entry.oid);
 405                         if (name && obj)
 406                                 fsck_put_object_name(options, &entry.oid, "%s%s",
 407                                                      name, entry.path);
 408                         result = options->walk(obj, OBJ_BLOB, data, options);
 409                 }
 410                 else {
 411                         result = error("in tree %s: entry %s has bad mode %.6o",
 412                                        fsck_describe_object(options, &tree->object.oid),
 413                                        entry.path, entry.mode);
 414                 }
 415                 if (result < 0)
 416                         return result;
 417                 if (!res)
 418                         res = result;
 419         }
 420         return res;
 421 }
 422
 423 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
 424 {
 425         int counter = 0, generation = 0, name_prefix_len = 0;
 426         struct commit_list *parents;
 427         int res;
 428         int result;
 429         const char *name;
 430
 431         if (parse_commit(commit))
 432                 return -1;
 433
 434         name = fsck_get_object_name(options, &commit->object.oid);
 435         if (name)
 436                 fsck_put_object_name(options, get_commit_tree_oid(commit),
 437                                      "%s:", name);
 438
 439         result = options->walk((struct object *)get_commit_tree(commit),
 440                                OBJ_TREE, data, options);
 441         if (result < 0)
 442                 return result;
 443         res = result;
 444
 445         parents = commit->parents;
 446         if (name && parents) {
 447                 int len = strlen(name), power;
 448
 449                 if (len && name[len - 1] == '^') {
 450                         generation = 1;
 451                         name_prefix_len = len - 1;
 452                 }
 453                 else { /* parse ~<generation> suffix */
 454                         for (generation = 0, power = 1;
 455                              len && isdigit(name[len - 1]);
 456                              power *= 10)
 457                                 generation += power * (name[--len] - '0');
 458                         if (power > 1 && len && name[len - 1] == '~')
 459                                 name_prefix_len = len - 1;
 460                 }
 461         }
 462
 463         while (parents) {
 464                 if (name) {
 465                         struct object_id *oid = &parents->item->object.oid;
 466
 467                         if (counter++)
 468                                 fsck_put_object_name(options, oid, "%s^%d",
 469                                                      name, counter);
 470                         else if (generation > 0)
 471                                 fsck_put_object_name(options, oid, "%.*s~%d",
 472                                                      name_prefix_len, name,
 473                                                      generation + 1);
 474                         else
 475                                 fsck_put_object_name(options, oid, "%s^", name);
 476                 }
 477                 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
 478                 if (result < 0)
 479                         return result;
 480                 if (!res)
 481                         res = result;
 482                 parents = parents->next;
 483         }
 484         return res;
 485 }
 486
 487 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
 488 {
 489         const char *name = fsck_get_object_name(options, &tag->object.oid);
 490
 491         if (parse_tag(tag))
 492                 return -1;
 493         if (name)
 494                 fsck_put_object_name(options, &tag->tagged->oid, "%s", name);
 495         return options->walk(tag->tagged, OBJ_ANY, data, options);
 496 }
 497
 498 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
 499 {
 500         if (!obj)
 501                 return -1;
 502
 503         if (obj->type == OBJ_NONE)
 504                 parse_object(the_repository, &obj->oid);
 505
 506         switch (obj->type) {
 507         case OBJ_BLOB:
 508                 return 0;
 509         case OBJ_TREE:
 510                 return fsck_walk_tree((struct tree *)obj, data, options);
 511         case OBJ_COMMIT:
 512                 return fsck_walk_commit((struct commit *)obj, data, options);
 513         case OBJ_TAG:
 514                 return fsck_walk_tag((struct tag *)obj, data, options);
 515         default:
 516                 error("Unknown object type for %s",
 517                       fsck_describe_object(options, &obj->oid));
 518                 return -1;
 519         }
 520 }
 521
 522 /*
 523  * The entries in a tree are ordered in the _path_ order,
 524  * which means that a directory entry is ordered by adding
 525  * a slash to the end of it.
 526  *
 527  * So a directory called "a" is ordered _after_ a file
 528  * called "a.c", because "a/" sorts after "a.c".
 529  */
 530 #define TREE_UNORDERED (-1)
 531 #define TREE_HAS_DUPS  (-2)
 532
 533 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 534 {
 535         int len1 = strlen(name1);
 536         int len2 = strlen(name2);
 537         int len = len1 < len2 ? len1 : len2;
 538         unsigned char c1, c2;
 539         int cmp;
 540
 541         cmp = memcmp(name1, name2, len);
 542         if (cmp < 0)
 543                 return 0;
 544         if (cmp > 0)
 545                 return TREE_UNORDERED;
 546
 547         /*
 548          * Ok, the first <len> characters are the same.
 549          * Now we need to order the next one, but turn
 550          * a '\0' into a '/' for a directory entry.
 551          */
 552         c1 = name1[len];
 553         c2 = name2[len];
 554         if (!c1 && !c2)
 555                 /*
 556                  * git-write-tree used to write out a nonsense tree that has
 557                  * entries with the same name, one blob and one tree.  Make
 558                  * sure we do not have duplicate entries.
 559                  */
 560                 return TREE_HAS_DUPS;
 561         if (!c1 && S_ISDIR(mode1))
 562                 c1 = '/';
 563         if (!c2 && S_ISDIR(mode2))
 564                 c2 = '/';
 565         return c1 < c2 ? 0 : TREE_UNORDERED;
 566 }
 567
 568 static int fsck_tree(struct tree *item,
 569                      const char *buffer, unsigned long size,
 570                      struct fsck_options *options)
 571 {
 572         int retval = 0;
 573         int has_null_sha1 = 0;
 574         int has_full_path = 0;
 575         int has_empty_name = 0;
 576         int has_dot = 0;
 577         int has_dotdot = 0;
 578         int has_dotgit = 0;
 579         int has_zero_pad = 0;
 580         int has_bad_modes = 0;
 581         int has_dup_entries = 0;
 582         int not_properly_sorted = 0;
 583         struct tree_desc desc;
 584         unsigned o_mode;
 585         const char *o_name;
 586
 587         if (init_tree_desc_gently(&desc, buffer, size)) {
 588                 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 589                 return retval;
 590         }
 591
 592         o_mode = 0;
 593         o_name = NULL;
 594
 595         while (desc.size) {
 596                 unsigned short mode;
 597                 const char *name;
 598                 const struct object_id *oid;
 599
 600                 oid = tree_entry_extract(&desc, &name, &mode);
 601
 602                 has_null_sha1 |= is_null_oid(oid);
 603                 has_full_path |= !!strchr(name, '/');
 604                 has_empty_name |= !*name;
 605                 has_dot |= !strcmp(name, ".");
 606                 has_dotdot |= !strcmp(name, "..");
 607                 has_dotgit |= is_hfs_dotgit(name) || is_ntfs_dotgit(name);
 608                 has_zero_pad |= *(char *)desc.buffer == '0';
 609
 610                 if (is_hfs_dotgitmodules(name) || is_ntfs_dotgitmodules(name)) {
 611                         if (!S_ISLNK(mode))
 612                                 oidset_insert(&gitmodules_found, oid);
 613                         else
 614                                 retval += report(options, &item->object,
 615                                                  FSCK_MSG_GITMODULES_SYMLINK,
 616                                                  ".gitmodules is a symbolic link");
 617                 }
 618
 619                 if (update_tree_entry_gently(&desc)) {
 620                         retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 621                         break;
 622                 }
 623
 624                 switch (mode) {
 625                 /*
 626                  * Standard modes..
 627                  */
 628                 case S_IFREG | 0755:
 629                 case S_IFREG | 0644:
 630                 case S_IFLNK:
 631                 case S_IFDIR:
 632                 case S_IFGITLINK:
 633                         break;
 634                 /*
 635                  * This is nonstandard, but we had a few of these
 636                  * early on when we honored the full set of mode
 637                  * bits..
 638                  */
 639                 case S_IFREG | 0664:
 640                         if (!options->strict)
 641                                 break;
 642                         /* fallthrough */
 643                 default:
 644                         has_bad_modes = 1;
 645                 }
 646
 647                 if (o_name) {
 648                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 649                         case TREE_UNORDERED:
 650                                 not_properly_sorted = 1;
 651                                 break;
 652                         case TREE_HAS_DUPS:
 653                                 has_dup_entries = 1;
 654                                 break;
 655                         default:
 656                                 break;
 657                         }
 658                 }
 659
 660                 o_mode = mode;
 661                 o_name = name;
 662         }
 663
 664         if (has_null_sha1)
 665                 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
 666         if (has_full_path)
 667                 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
 668         if (has_empty_name)
 669                 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
 670         if (has_dot)
 671                 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
 672         if (has_dotdot)
 673                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
 674         if (has_dotgit)
 675                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
 676         if (has_zero_pad)
 677                 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
 678         if (has_bad_modes)
 679                 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
 680         if (has_dup_entries)
 681                 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
 682         if (not_properly_sorted)
 683                 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
 684         return retval;
 685 }
 686
 687 static int verify_headers(const void *data, unsigned long size,
 688                           struct object *obj, struct fsck_options *options)
 689 {
 690         const char *buffer = (const char *)data;
 691         unsigned long i;
 692
 693         for (i = 0; i < size; i++) {
 694                 switch (buffer[i]) {
 695                 case '\0':
 696                         return report(options, obj,
 697                                 FSCK_MSG_NUL_IN_HEADER,
 698                                 "unterminated header: NUL at offset %ld", i);
 699                 case '\n':
 700                         if (i + 1 < size && buffer[i + 1] == '\n')
 701                                 return 0;
 702                 }
 703         }
 704
 705         /*
 706          * We did not find double-LF that separates the header
 707          * and the body.  Not having a body is not a crime but
 708          * we do want to see the terminating LF for the last header
 709          * line.
 710          */
 711         if (size && buffer[size - 1] == '\n')
 712                 return 0;
 713
 714         return report(options, obj,
 715                 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
 716 }
 717
 718 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
 719 {
 720         const char *p = *ident;
 721         char *end;
 722
 723         *ident = strchrnul(*ident, '\n');
 724         if (**ident == '\n')
 725                 (*ident)++;
 726
 727         if (*p == '<')
 728                 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 729         p += strcspn(p, "<>\n");
 730         if (*p == '>')
 731                 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
 732         if (*p != '<')
 733                 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
 734         if (p[-1] != ' ')
 735                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 736         p++;
 737         p += strcspn(p, "<>\n");
 738         if (*p != '>')
 739                 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
 740         p++;
 741         if (*p != ' ')
 742                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
 743         p++;
 744         if (*p == '0' && p[1] != ' ')
 745                 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
 746         if (date_overflows(parse_timestamp(p, &end, 10)))
 747                 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
 748         if ((end == p || *end != ' '))
 749                 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
 750         p = end + 1;
 751         if ((*p != '+' && *p != '-') ||
 752             !isdigit(p[1]) ||
 753             !isdigit(p[2]) ||
 754             !isdigit(p[3]) ||
 755             !isdigit(p[4]) ||
 756             (p[5] != '\n'))
 757                 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
 758         p += 6;
 759         return 0;
 760 }
 761
 762 static int fsck_commit(struct commit *commit, const char *buffer,
 763                        unsigned long size, struct fsck_options *options)
 764 {
 765         struct object_id tree_oid, oid;
 766         unsigned author_count;
 767         int err;
 768         const char *buffer_begin = buffer;
 769         const char *p;
 770
 771         if (verify_headers(buffer, size, &commit->object, options))
 772                 return -1;
 773
 774         if (!skip_prefix(buffer, "tree ", &buffer))
 775                 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
 776         if (parse_oid_hex(buffer, &tree_oid, &p) || *p != '\n') {
 777                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
 778                 if (err)
 779                         return err;
 780         }
 781         buffer = p + 1;
 782         while (skip_prefix(buffer, "parent ", &buffer)) {
 783                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 784                         err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
 785                         if (err)
 786                                 return err;
 787                 }
 788                 buffer = p + 1;
 789         }
 790         author_count = 0;
 791         while (skip_prefix(buffer, "author ", &buffer)) {
 792                 author_count++;
 793                 err = fsck_ident(&buffer, &commit->object, options);
 794                 if (err)
 795                         return err;
 796         }
 797         if (author_count < 1)
 798                 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
 799         else if (author_count > 1)
 800                 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
 801         if (err)
 802                 return err;
 803         if (!skip_prefix(buffer, "committer ", &buffer))
 804                 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
 805         err = fsck_ident(&buffer, &commit->object, options);
 806         if (err)
 807                 return err;
 808         if (memchr(buffer_begin, '\0', size)) {
 809                 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
 810                              "NUL byte in the commit object body");
 811                 if (err)
 812                         return err;
 813         }
 814         return 0;
 815 }
 816
 817 static int fsck_tag(struct tag *tag, const char *buffer,
 818                     unsigned long size, struct fsck_options *options)
 819 {
 820         struct object_id oid;
 821         int ret = 0;
 822         char *eol;
 823         struct strbuf sb = STRBUF_INIT;
 824         const char *p;
 825
 826         ret = verify_headers(buffer, size, &tag->object, options);
 827         if (ret)
 828                 goto done;
 829
 830         if (!skip_prefix(buffer, "object ", &buffer)) {
 831                 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
 832                 goto done;
 833         }
 834         if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 835                 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
 836                 if (ret)
 837                         goto done;
 838         }
 839         buffer = p + 1;
 840
 841         if (!skip_prefix(buffer, "type ", &buffer)) {
 842                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
 843                 goto done;
 844         }
 845         eol = strchr(buffer, '\n');
 846         if (!eol) {
 847                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
 848                 goto done;
 849         }
 850         if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
 851                 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
 852         if (ret)
 853                 goto done;
 854         buffer = eol + 1;
 855
 856         if (!skip_prefix(buffer, "tag ", &buffer)) {
 857                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
 858                 goto done;
 859         }
 860         eol = strchr(buffer, '\n');
 861         if (!eol) {
 862                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
 863                 goto done;
 864         }
 865         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
 866         if (check_refname_format(sb.buf, 0)) {
 867                 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
 868                            "invalid 'tag' name: %.*s",
 869                            (int)(eol - buffer), buffer);
 870                 if (ret)
 871                         goto done;
 872         }
 873         buffer = eol + 1;
 874
 875         if (!skip_prefix(buffer, "tagger ", &buffer)) {
 876                 /* early tags do not contain 'tagger' lines; warn only */
 877                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
 878                 if (ret)
 879                         goto done;
 880         }
 881         else
 882                 ret = fsck_ident(&buffer, &tag->object, options);
 883
 884 done:
 885         strbuf_release(&sb);
 886         return ret;
 887 }
 888
 889 struct fsck_gitmodules_data {
 890         struct object *obj;
 891         struct fsck_options *options;
 892         int ret;
 893 };
 894
 895 static int fsck_gitmodules_fn(const char *var, const char *value, void *vdata)
 896 {
 897         struct fsck_gitmodules_data *data = vdata;
 898         const char *subsection, *key;
 899         int subsection_len;
 900         char *name;
 901
 902         if (parse_config_key(var, "submodule", &subsection, &subsection_len, &key) < 0 ||
 903             !subsection)
 904                 return 0;
 905
 906         name = xmemdupz(subsection, subsection_len);
 907         if (check_submodule_name(name) < 0)
 908                 data->ret |= report(data->options, data->obj,
 909                                     FSCK_MSG_GITMODULES_NAME,
 910                                     "disallowed submodule name: %s",
 911                                     name);
 912         if (!strcmp(key, "url") && value &&
 913             looks_like_command_line_option(value))
 914                 data->ret |= report(data->options, data->obj,
 915                                     FSCK_MSG_GITMODULES_URL,
 916                                     "disallowed submodule url: %s",
 917                                     value);
 918         if (!strcmp(key, "path") && value &&
 919             looks_like_command_line_option(value))
 920                 data->ret |= report(data->options, data->obj,
 921                                     FSCK_MSG_GITMODULES_PATH,
 922                                     "disallowed submodule path: %s",
 923                                     value);
 924         free(name);
 925
 926         return 0;
 927 }
 928
 929 static int fsck_blob(struct blob *blob, const char *buf,
 930                      unsigned long size, struct fsck_options *options)
 931 {
 932         struct fsck_gitmodules_data data;
 933         struct config_options config_opts = { 0 };
 934
 935         if (!oidset_contains(&gitmodules_found, &blob->object.oid))
 936                 return 0;
 937         oidset_insert(&gitmodules_done, &blob->object.oid);
 938
 939         if (object_on_skiplist(options, &blob->object.oid))
 940                 return 0;
 941
 942         if (!buf) {
 943                 /*
 944                  * A missing buffer here is a sign that the caller found the
 945                  * blob too gigantic to load into memory. Let's just consider
 946                  * that an error.
 947                  */
 948                 return report(options, &blob->object,
 949                               FSCK_MSG_GITMODULES_LARGE,
 950                               ".gitmodules too large to parse");
 951         }
 952
 953         data.obj = &blob->object;
 954         data.options = options;
 955         data.ret = 0;
 956         config_opts.error_action = CONFIG_ERROR_SILENT;
 957         if (git_config_from_mem(fsck_gitmodules_fn, CONFIG_ORIGIN_BLOB,
 958                                 ".gitmodules", buf, size, &data, &config_opts))
 959                 data.ret |= report(options, &blob->object,
 960                                    FSCK_MSG_GITMODULES_PARSE,
 961                                    "could not parse gitmodules blob");
 962
 963         return data.ret;
 964 }
 965
 966 int fsck_object(struct object *obj, void *data, unsigned long size,
 967         struct fsck_options *options)
 968 {
 969         if (!obj)
 970                 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
 971
 972         if (obj->type == OBJ_BLOB)
 973                 return fsck_blob((struct blob *)obj, data, size, options);
 974         if (obj->type == OBJ_TREE)
 975                 return fsck_tree((struct tree *) obj, data, size, options);
 976         if (obj->type == OBJ_COMMIT)
 977                 return fsck_commit((struct commit *) obj, (const char *) data,
 978                         size, options);
 979         if (obj->type == OBJ_TAG)
 980                 return fsck_tag((struct tag *) obj, (const char *) data,
 981                         size, options);
 982
 983         return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
 984                           obj->type);
 985 }
 986
 987 int fsck_error_function(struct fsck_options *o,
 988                         const struct object_id *oid,
 989                         enum object_type object_type,
 990                         int msg_type, const char *message)
 991 {
 992         if (msg_type == FSCK_WARN) {
 993                 warning("object %s: %s", fsck_describe_object(o, oid), message);
 994                 return 0;
 995         }
 996         error("object %s: %s", fsck_describe_object(o, oid), message);
 997         return 1;
 998 }
 999
1000 int fsck_finish(struct fsck_options *options)
1001 {
1002         int ret = 0;
1003         struct oidset_iter iter;
1004         const struct object_id *oid;
1005
1006         oidset_iter_init(&gitmodules_found, &iter);
1007         while ((oid = oidset_iter_next(&iter))) {
1008                 struct blob *blob;
1009                 enum object_type type;
1010                 unsigned long size;
1011                 char *buf;
1012
1013                 if (oidset_contains(&gitmodules_done, oid))
1014                         continue;
1015
1016                 blob = lookup_blob(the_repository, oid);
1017                 if (!blob) {
1018                         struct object *obj = lookup_unknown_object(oid);
1019                         ret |= report(options, obj,
1020                                       FSCK_MSG_GITMODULES_BLOB,
1021                                       "non-blob found at .gitmodules");
1022                         continue;
1023                 }
1024
1025                 buf = read_object_file(oid, &type, &size);
1026                 if (!buf) {
1027                         if (is_promisor_object(&blob->object.oid))
1028                                 continue;
1029                         ret |= report(options, &blob->object,
1030                                       FSCK_MSG_GITMODULES_MISSING,
1031                                       "unable to read .gitmodules blob");
1032                         continue;
1033                 }
1034
1035                 if (type == OBJ_BLOB)
1036                         ret |= fsck_blob(blob, buf, size, options);
1037                 else
1038                         ret |= report(options, &blob->object,
1039                                       FSCK_MSG_GITMODULES_BLOB,
1040                                       "non-blob found at .gitmodules");
1041                 free(buf);
1042         }
1043
1044
1045         oidset_clear(&gitmodules_found);
1046         oidset_clear(&gitmodules_done);
1047         return ret;
1048 }