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 "sha1-array.h"
  14 #include "decorate.h"
  15 #include "oidset.h"
  16 #include "packfile.h"
  17 #include "submodule-config.h"
  18 #include "config.h"
  19 #include "help.h"
  20
  21 static struct oidset gitmodules_found = OIDSET_INIT;
  22 static struct oidset gitmodules_done = OIDSET_INIT;
  23
  24 #define FSCK_FATAL -1
  25 #define FSCK_INFO -2
  26
  27 #define FOREACH_MSG_ID(FUNC) \
  28         /* fatal errors */ \
  29         FUNC(NUL_IN_HEADER, FATAL) \
  30         FUNC(UNTERMINATED_HEADER, FATAL) \
  31         /* errors */ \
  32         FUNC(BAD_DATE, ERROR) \
  33         FUNC(BAD_DATE_OVERFLOW, ERROR) \
  34         FUNC(BAD_EMAIL, ERROR) \
  35         FUNC(BAD_NAME, ERROR) \
  36         FUNC(BAD_OBJECT_SHA1, ERROR) \
  37         FUNC(BAD_PARENT_SHA1, ERROR) \
  38         FUNC(BAD_TAG_OBJECT, ERROR) \
  39         FUNC(BAD_TIMEZONE, ERROR) \
  40         FUNC(BAD_TREE, ERROR) \
  41         FUNC(BAD_TREE_SHA1, ERROR) \
  42         FUNC(BAD_TYPE, ERROR) \
  43         FUNC(DUPLICATE_ENTRIES, ERROR) \
  44         FUNC(MISSING_AUTHOR, ERROR) \
  45         FUNC(MISSING_COMMITTER, ERROR) \
  46         FUNC(MISSING_EMAIL, ERROR) \
  47         FUNC(MISSING_GRAFT, ERROR) \
  48         FUNC(MISSING_NAME_BEFORE_EMAIL, ERROR) \
  49         FUNC(MISSING_OBJECT, ERROR) \
  50         FUNC(MISSING_PARENT, ERROR) \
  51         FUNC(MISSING_SPACE_BEFORE_DATE, ERROR) \
  52         FUNC(MISSING_SPACE_BEFORE_EMAIL, ERROR) \
  53         FUNC(MISSING_TAG, ERROR) \
  54         FUNC(MISSING_TAG_ENTRY, ERROR) \
  55         FUNC(MISSING_TAG_OBJECT, ERROR) \
  56         FUNC(MISSING_TREE, ERROR) \
  57         FUNC(MISSING_TREE_OBJECT, ERROR) \
  58         FUNC(MISSING_TYPE, ERROR) \
  59         FUNC(MISSING_TYPE_ENTRY, ERROR) \
  60         FUNC(MULTIPLE_AUTHORS, ERROR) \
  61         FUNC(TAG_OBJECT_NOT_TAG, ERROR) \
  62         FUNC(TREE_NOT_SORTED, ERROR) \
  63         FUNC(UNKNOWN_TYPE, ERROR) \
  64         FUNC(ZERO_PADDED_DATE, ERROR) \
  65         FUNC(GITMODULES_MISSING, ERROR) \
  66         FUNC(GITMODULES_BLOB, ERROR) \
  67         FUNC(GITMODULES_LARGE, ERROR) \
  68         FUNC(GITMODULES_NAME, ERROR) \
  69         FUNC(GITMODULES_SYMLINK, ERROR) \
  70         FUNC(GITMODULES_URL, ERROR) \
  71         FUNC(GITMODULES_PATH, ERROR) \
  72         FUNC(GITMODULES_UPDATE, ERROR) \
  73         /* warnings */ \
  74         FUNC(BAD_FILEMODE, WARN) \
  75         FUNC(EMPTY_NAME, WARN) \
  76         FUNC(FULL_PATHNAME, WARN) \
  77         FUNC(HAS_DOT, WARN) \
  78         FUNC(HAS_DOTDOT, WARN) \
  79         FUNC(HAS_DOTGIT, WARN) \
  80         FUNC(NULL_SHA1, WARN) \
  81         FUNC(ZERO_PADDED_FILEMODE, WARN) \
  82         FUNC(NUL_IN_COMMIT, WARN) \
  83         /* infos (reported as warnings, but ignored by default) */ \
  84         FUNC(GITMODULES_PARSE, INFO) \
  85         FUNC(BAD_TAG_NAME, INFO) \
  86         FUNC(MISSING_TAGGER_ENTRY, INFO)
  87
  88 #define MSG_ID(id, msg_type) FSCK_MSG_##id,
  89 enum fsck_msg_id {
  90         FOREACH_MSG_ID(MSG_ID)
  91         FSCK_MSG_MAX
  92 };
  93 #undef MSG_ID
  94
  95 #define STR(x) #x
  96 #define MSG_ID(id, msg_type) { STR(id), NULL, NULL, FSCK_##msg_type },
  97 static struct {
  98         const char *id_string;
  99         const char *downcased;
 100         const char *camelcased;
 101         int msg_type;
 102 } msg_id_info[FSCK_MSG_MAX + 1] = {
 103         FOREACH_MSG_ID(MSG_ID)
 104         { NULL, NULL, NULL, -1 }
 105 };
 106 #undef MSG_ID
 107
 108 static void prepare_msg_ids(void)
 109 {
 110         int i;
 111
 112         if (msg_id_info[0].downcased)
 113                 return;
 114
 115         /* convert id_string to lower case, without underscores. */
 116         for (i = 0; i < FSCK_MSG_MAX; i++) {
 117                 const char *p = msg_id_info[i].id_string;
 118                 int len = strlen(p);
 119                 char *q = xmalloc(len);
 120
 121                 msg_id_info[i].downcased = q;
 122                 while (*p)
 123                         if (*p == '_')
 124                                 p++;
 125                         else
 126                                 *(q)++ = tolower(*(p)++);
 127                 *q = '\0';
 128
 129                 p = msg_id_info[i].id_string;
 130                 q = xmalloc(len);
 131                 msg_id_info[i].camelcased = q;
 132                 while (*p) {
 133                         if (*p == '_') {
 134                                 p++;
 135                                 if (*p)
 136                                         *q++ = *p++;
 137                         } else {
 138                                 *q++ = tolower(*p++);
 139                         }
 140                 }
 141                 *q = '\0';
 142         }
 143 }
 144
 145 static int parse_msg_id(const char *text)
 146 {
 147         int i;
 148
 149         prepare_msg_ids();
 150
 151         for (i = 0; i < FSCK_MSG_MAX; i++)
 152                 if (!strcmp(text, msg_id_info[i].downcased))
 153                         return i;
 154
 155         return -1;
 156 }
 157
 158 void list_config_fsck_msg_ids(struct string_list *list, const char *prefix)
 159 {
 160         int i;
 161
 162         prepare_msg_ids();
 163
 164         for (i = 0; i < FSCK_MSG_MAX; i++)
 165                 list_config_item(list, prefix, msg_id_info[i].camelcased);
 166 }
 167
 168 static int fsck_msg_type(enum fsck_msg_id msg_id,
 169         struct fsck_options *options)
 170 {
 171         int msg_type;
 172
 173         assert(msg_id >= 0 && msg_id < FSCK_MSG_MAX);
 174
 175         if (options->msg_type)
 176                 msg_type = options->msg_type[msg_id];
 177         else {
 178                 msg_type = msg_id_info[msg_id].msg_type;
 179                 if (options->strict && msg_type == FSCK_WARN)
 180                         msg_type = FSCK_ERROR;
 181         }
 182
 183         return msg_type;
 184 }
 185
 186 static void init_skiplist(struct fsck_options *options, const char *path)
 187 {
 188         static struct oid_array skiplist = OID_ARRAY_INIT;
 189         int sorted, fd;
 190         char buffer[GIT_MAX_HEXSZ + 1];
 191         struct object_id oid;
 192
 193         if (options->skiplist)
 194                 sorted = options->skiplist->sorted;
 195         else {
 196                 sorted = 1;
 197                 options->skiplist = &skiplist;
 198         }
 199
 200         fd = open(path, O_RDONLY);
 201         if (fd < 0)
 202                 die("Could not open skip list: %s", path);
 203         for (;;) {
 204                 const char *p;
 205                 int result = read_in_full(fd, buffer, sizeof(buffer));
 206                 if (result < 0)
 207                         die_errno("Could not read '%s'", path);
 208                 if (!result)
 209                         break;
 210                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n')
 211                         die("Invalid SHA-1: %s", buffer);
 212                 oid_array_append(&skiplist, &oid);
 213                 if (sorted && skiplist.nr > 1 &&
 214                                 oidcmp(&skiplist.oid[skiplist.nr - 2],
 215                                        &oid) > 0)
 216                         sorted = 0;
 217         }
 218         close(fd);
 219
 220         if (sorted)
 221                 skiplist.sorted = 1;
 222 }
 223
 224 static int parse_msg_type(const char *str)
 225 {
 226         if (!strcmp(str, "error"))
 227                 return FSCK_ERROR;
 228         else if (!strcmp(str, "warn"))
 229                 return FSCK_WARN;
 230         else if (!strcmp(str, "ignore"))
 231                 return FSCK_IGNORE;
 232         else
 233                 die("Unknown fsck message type: '%s'", str);
 234 }
 235
 236 int is_valid_msg_type(const char *msg_id, const char *msg_type)
 237 {
 238         if (parse_msg_id(msg_id) < 0)
 239                 return 0;
 240         parse_msg_type(msg_type);
 241         return 1;
 242 }
 243
 244 void fsck_set_msg_type(struct fsck_options *options,
 245                 const char *msg_id, const char *msg_type)
 246 {
 247         int id = parse_msg_id(msg_id), type;
 248
 249         if (id < 0)
 250                 die("Unhandled message id: %s", msg_id);
 251         type = parse_msg_type(msg_type);
 252
 253         if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
 254                 die("Cannot demote %s to %s", msg_id, msg_type);
 255
 256         if (!options->msg_type) {
 257                 int i;
 258                 int *msg_type;
 259                 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
 260                 for (i = 0; i < FSCK_MSG_MAX; i++)
 261                         msg_type[i] = fsck_msg_type(i, options);
 262                 options->msg_type = msg_type;
 263         }
 264
 265         options->msg_type[id] = type;
 266 }
 267
 268 void fsck_set_msg_types(struct fsck_options *options, const char *values)
 269 {
 270         char *buf = xstrdup(values), *to_free = buf;
 271         int done = 0;
 272
 273         while (!done) {
 274                 int len = strcspn(buf, " ,|"), equal;
 275
 276                 done = !buf[len];
 277                 if (!len) {
 278                         buf++;
 279                         continue;
 280                 }
 281                 buf[len] = '\0';
 282
 283                 for (equal = 0;
 284                      equal < len && buf[equal] != '=' && buf[equal] != ':';
 285                      equal++)
 286                         buf[equal] = tolower(buf[equal]);
 287                 buf[equal] = '\0';
 288
 289                 if (!strcmp(buf, "skiplist")) {
 290                         if (equal == len)
 291                                 die("skiplist requires a path");
 292                         init_skiplist(options, buf + equal + 1);
 293                         buf += len + 1;
 294                         continue;
 295                 }
 296
 297                 if (equal == len)
 298                         die("Missing '=': '%s'", buf);
 299
 300                 fsck_set_msg_type(options, buf, buf + equal + 1);
 301                 buf += len + 1;
 302         }
 303         free(to_free);
 304 }
 305
 306 static void append_msg_id(struct strbuf *sb, const char *msg_id)
 307 {
 308         for (;;) {
 309                 char c = *(msg_id)++;
 310
 311                 if (!c)
 312                         break;
 313                 if (c != '_')
 314                         strbuf_addch(sb, tolower(c));
 315                 else {
 316                         assert(*msg_id);
 317                         strbuf_addch(sb, *(msg_id)++);
 318                 }
 319         }
 320
 321         strbuf_addstr(sb, ": ");
 322 }
 323
 324 static int object_on_skiplist(struct fsck_options *opts, struct object *obj)
 325 {
 326         if (opts && opts->skiplist && obj)
 327                 return oid_array_lookup(opts->skiplist, &obj->oid) >= 0;
 328         return 0;
 329 }
 330
 331 __attribute__((format (printf, 4, 5)))
 332 static int report(struct fsck_options *options, struct object *object,
 333         enum fsck_msg_id id, const char *fmt, ...)
 334 {
 335         va_list ap;
 336         struct strbuf sb = STRBUF_INIT;
 337         int msg_type = fsck_msg_type(id, options), result;
 338
 339         if (msg_type == FSCK_IGNORE)
 340                 return 0;
 341
 342         if (object_on_skiplist(options, object))
 343                 return 0;
 344
 345         if (msg_type == FSCK_FATAL)
 346                 msg_type = FSCK_ERROR;
 347         else if (msg_type == FSCK_INFO)
 348                 msg_type = FSCK_WARN;
 349
 350         append_msg_id(&sb, msg_id_info[id].id_string);
 351
 352         va_start(ap, fmt);
 353         strbuf_vaddf(&sb, fmt, ap);
 354         result = options->error_func(options, object, msg_type, sb.buf);
 355         strbuf_release(&sb);
 356         va_end(ap);
 357
 358         return result;
 359 }
 360
 361 static char *get_object_name(struct fsck_options *options, struct object *obj)
 362 {
 363         if (!options->object_names)
 364                 return NULL;
 365         return lookup_decoration(options->object_names, obj);
 366 }
 367
 368 static void put_object_name(struct fsck_options *options, struct object *obj,
 369         const char *fmt, ...)
 370 {
 371         va_list ap;
 372         struct strbuf buf = STRBUF_INIT;
 373         char *existing;
 374
 375         if (!options->object_names)
 376                 return;
 377         existing = lookup_decoration(options->object_names, obj);
 378         if (existing)
 379                 return;
 380         va_start(ap, fmt);
 381         strbuf_vaddf(&buf, fmt, ap);
 382         add_decoration(options->object_names, obj, strbuf_detach(&buf, NULL));
 383         va_end(ap);
 384 }
 385
 386 static const char *describe_object(struct fsck_options *o, struct object *obj)
 387 {
 388         static struct strbuf buf = STRBUF_INIT;
 389         char *name;
 390
 391         strbuf_reset(&buf);
 392         strbuf_addstr(&buf, oid_to_hex(&obj->oid));
 393         if (o->object_names && (name = lookup_decoration(o->object_names, obj)))
 394                 strbuf_addf(&buf, " (%s)", name);
 395
 396         return buf.buf;
 397 }
 398
 399 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
 400 {
 401         struct tree_desc desc;
 402         struct name_entry entry;
 403         int res = 0;
 404         const char *name;
 405
 406         if (parse_tree(tree))
 407                 return -1;
 408
 409         name = get_object_name(options, &tree->object);
 410         if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
 411                 return -1;
 412         while (tree_entry_gently(&desc, &entry)) {
 413                 struct object *obj;
 414                 int result;
 415
 416                 if (S_ISGITLINK(entry.mode))
 417                         continue;
 418
 419                 if (S_ISDIR(entry.mode)) {
 420                         obj = (struct object *)lookup_tree(the_repository, entry.oid);
 421                         if (name && obj)
 422                                 put_object_name(options, obj, "%s%s/", name,
 423                                         entry.path);
 424                         result = options->walk(obj, OBJ_TREE, data, options);
 425                 }
 426                 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
 427                         obj = (struct object *)lookup_blob(the_repository, entry.oid);
 428                         if (name && obj)
 429                                 put_object_name(options, obj, "%s%s", name,
 430                                         entry.path);
 431                         result = options->walk(obj, OBJ_BLOB, data, options);
 432                 }
 433                 else {
 434                         result = error("in tree %s: entry %s has bad mode %.6o",
 435                                         describe_object(options, &tree->object), entry.path, entry.mode);
 436                 }
 437                 if (result < 0)
 438                         return result;
 439                 if (!res)
 440                         res = result;
 441         }
 442         return res;
 443 }
 444
 445 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
 446 {
 447         int counter = 0, generation = 0, name_prefix_len = 0;
 448         struct commit_list *parents;
 449         int res;
 450         int result;
 451         const char *name;
 452
 453         if (parse_commit(commit))
 454                 return -1;
 455
 456         name = get_object_name(options, &commit->object);
 457         if (name)
 458                 put_object_name(options, &get_commit_tree(commit)->object,
 459                                 "%s:", name);
 460
 461         result = options->walk((struct object *)get_commit_tree(commit),
 462                                OBJ_TREE, data, options);
 463         if (result < 0)
 464                 return result;
 465         res = result;
 466
 467         parents = commit->parents;
 468         if (name && parents) {
 469                 int len = strlen(name), power;
 470
 471                 if (len && name[len - 1] == '^') {
 472                         generation = 1;
 473                         name_prefix_len = len - 1;
 474                 }
 475                 else { /* parse ~<generation> suffix */
 476                         for (generation = 0, power = 1;
 477                              len && isdigit(name[len - 1]);
 478                              power *= 10)
 479                                 generation += power * (name[--len] - '0');
 480                         if (power > 1 && len && name[len - 1] == '~')
 481                                 name_prefix_len = len - 1;
 482                 }
 483         }
 484
 485         while (parents) {
 486                 if (name) {
 487                         struct object *obj = &parents->item->object;
 488
 489                         if (counter++)
 490                                 put_object_name(options, obj, "%s^%d",
 491                                         name, counter);
 492                         else if (generation > 0)
 493                                 put_object_name(options, obj, "%.*s~%d",
 494                                         name_prefix_len, name, generation + 1);
 495                         else
 496                                 put_object_name(options, obj, "%s^", name);
 497                 }
 498                 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
 499                 if (result < 0)
 500                         return result;
 501                 if (!res)
 502                         res = result;
 503                 parents = parents->next;
 504         }
 505         return res;
 506 }
 507
 508 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
 509 {
 510         char *name = get_object_name(options, &tag->object);
 511
 512         if (parse_tag(tag))
 513                 return -1;
 514         if (name)
 515                 put_object_name(options, tag->tagged, "%s", name);
 516         return options->walk(tag->tagged, OBJ_ANY, data, options);
 517 }
 518
 519 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
 520 {
 521         if (!obj)
 522                 return -1;
 523
 524         if (obj->type == OBJ_NONE)
 525                 parse_object(the_repository, &obj->oid);
 526
 527         switch (obj->type) {
 528         case OBJ_BLOB:
 529                 return 0;
 530         case OBJ_TREE:
 531                 return fsck_walk_tree((struct tree *)obj, data, options);
 532         case OBJ_COMMIT:
 533                 return fsck_walk_commit((struct commit *)obj, data, options);
 534         case OBJ_TAG:
 535                 return fsck_walk_tag((struct tag *)obj, data, options);
 536         default:
 537                 error("Unknown object type for %s", describe_object(options, obj));
 538                 return -1;
 539         }
 540 }
 541
 542 /*
 543  * The entries in a tree are ordered in the _path_ order,
 544  * which means that a directory entry is ordered by adding
 545  * a slash to the end of it.
 546  *
 547  * So a directory called "a" is ordered _after_ a file
 548  * called "a.c", because "a/" sorts after "a.c".
 549  */
 550 #define TREE_UNORDERED (-1)
 551 #define TREE_HAS_DUPS  (-2)
 552
 553 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 554 {
 555         int len1 = strlen(name1);
 556         int len2 = strlen(name2);
 557         int len = len1 < len2 ? len1 : len2;
 558         unsigned char c1, c2;
 559         int cmp;
 560
 561         cmp = memcmp(name1, name2, len);
 562         if (cmp < 0)
 563                 return 0;
 564         if (cmp > 0)
 565                 return TREE_UNORDERED;
 566
 567         /*
 568          * Ok, the first <len> characters are the same.
 569          * Now we need to order the next one, but turn
 570          * a '\0' into a '/' for a directory entry.
 571          */
 572         c1 = name1[len];
 573         c2 = name2[len];
 574         if (!c1 && !c2)
 575                 /*
 576                  * git-write-tree used to write out a nonsense tree that has
 577                  * entries with the same name, one blob and one tree.  Make
 578                  * sure we do not have duplicate entries.
 579                  */
 580                 return TREE_HAS_DUPS;
 581         if (!c1 && S_ISDIR(mode1))
 582                 c1 = '/';
 583         if (!c2 && S_ISDIR(mode2))
 584                 c2 = '/';
 585         return c1 < c2 ? 0 : TREE_UNORDERED;
 586 }
 587
 588 static int fsck_tree(struct tree *item, struct fsck_options *options)
 589 {
 590         int retval = 0;
 591         int has_null_sha1 = 0;
 592         int has_full_path = 0;
 593         int has_empty_name = 0;
 594         int has_dot = 0;
 595         int has_dotdot = 0;
 596         int has_dotgit = 0;
 597         int has_zero_pad = 0;
 598         int has_bad_modes = 0;
 599         int has_dup_entries = 0;
 600         int not_properly_sorted = 0;
 601         struct tree_desc desc;
 602         unsigned o_mode;
 603         const char *o_name;
 604
 605         if (init_tree_desc_gently(&desc, item->buffer, item->size)) {
 606                 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 607                 return retval;
 608         }
 609
 610         o_mode = 0;
 611         o_name = NULL;
 612
 613         while (desc.size) {
 614                 unsigned mode;
 615                 const char *name, *backslash;
 616                 const struct object_id *oid;
 617
 618                 oid = tree_entry_extract(&desc, &name, &mode);
 619
 620                 has_null_sha1 |= is_null_oid(oid);
 621                 has_full_path |= !!strchr(name, '/');
 622                 has_empty_name |= !*name;
 623                 has_dot |= !strcmp(name, ".");
 624                 has_dotdot |= !strcmp(name, "..");
 625                 has_dotgit |= is_hfs_dotgit(name) || is_ntfs_dotgit(name);
 626                 has_zero_pad |= *(char *)desc.buffer == '0';
 627
 628                 if (is_hfs_dotgitmodules(name) || is_ntfs_dotgitmodules(name)) {
 629                         if (!S_ISLNK(mode))
 630                                 oidset_insert(&gitmodules_found, oid);
 631                         else
 632                                 retval += report(options, &item->object,
 633                                                  FSCK_MSG_GITMODULES_SYMLINK,
 634                                                  ".gitmodules is a symbolic link");
 635                 }
 636
 637                 if ((backslash = strchr(name, '\\'))) {
 638                         while (backslash) {
 639                                 backslash++;
 640                                 has_dotgit |= is_ntfs_dotgit(backslash);
 641                                 if (is_ntfs_dotgitmodules(backslash)) {
 642                                         if (!S_ISLNK(mode))
 643                                                 oidset_insert(&gitmodules_found, oid);
 644                                         else
 645                                                 retval += report(options, &item->object,
 646                                                                  FSCK_MSG_GITMODULES_SYMLINK,
 647                                                                  ".gitmodules is a symbolic link");
 648                                 }
 649                                 backslash = strchr(backslash, '\\');
 650                         }
 651                 }
 652
 653                 if (update_tree_entry_gently(&desc)) {
 654                         retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 655                         break;
 656                 }
 657
 658                 switch (mode) {
 659                 /*
 660                  * Standard modes..
 661                  */
 662                 case S_IFREG | 0755:
 663                 case S_IFREG | 0644:
 664                 case S_IFLNK:
 665                 case S_IFDIR:
 666                 case S_IFGITLINK:
 667                         break;
 668                 /*
 669                  * This is nonstandard, but we had a few of these
 670                  * early on when we honored the full set of mode
 671                  * bits..
 672                  */
 673                 case S_IFREG | 0664:
 674                         if (!options->strict)
 675                                 break;
 676                         /* fallthrough */
 677                 default:
 678                         has_bad_modes = 1;
 679                 }
 680
 681                 if (o_name) {
 682                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 683                         case TREE_UNORDERED:
 684                                 not_properly_sorted = 1;
 685                                 break;
 686                         case TREE_HAS_DUPS:
 687                                 has_dup_entries = 1;
 688                                 break;
 689                         default:
 690                                 break;
 691                         }
 692                 }
 693
 694                 o_mode = mode;
 695                 o_name = name;
 696         }
 697
 698         if (has_null_sha1)
 699                 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
 700         if (has_full_path)
 701                 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
 702         if (has_empty_name)
 703                 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
 704         if (has_dot)
 705                 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
 706         if (has_dotdot)
 707                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
 708         if (has_dotgit)
 709                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
 710         if (has_zero_pad)
 711                 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
 712         if (has_bad_modes)
 713                 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
 714         if (has_dup_entries)
 715                 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
 716         if (not_properly_sorted)
 717                 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
 718         return retval;
 719 }
 720
 721 static int verify_headers(const void *data, unsigned long size,
 722                           struct object *obj, struct fsck_options *options)
 723 {
 724         const char *buffer = (const char *)data;
 725         unsigned long i;
 726
 727         for (i = 0; i < size; i++) {
 728                 switch (buffer[i]) {
 729                 case '\0':
 730                         return report(options, obj,
 731                                 FSCK_MSG_NUL_IN_HEADER,
 732                                 "unterminated header: NUL at offset %ld", i);
 733                 case '\n':
 734                         if (i + 1 < size && buffer[i + 1] == '\n')
 735                                 return 0;
 736                 }
 737         }
 738
 739         /*
 740          * We did not find double-LF that separates the header
 741          * and the body.  Not having a body is not a crime but
 742          * we do want to see the terminating LF for the last header
 743          * line.
 744          */
 745         if (size && buffer[size - 1] == '\n')
 746                 return 0;
 747
 748         return report(options, obj,
 749                 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
 750 }
 751
 752 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
 753 {
 754         const char *p = *ident;
 755         char *end;
 756
 757         *ident = strchrnul(*ident, '\n');
 758         if (**ident == '\n')
 759                 (*ident)++;
 760
 761         if (*p == '<')
 762                 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 763         p += strcspn(p, "<>\n");
 764         if (*p == '>')
 765                 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
 766         if (*p != '<')
 767                 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
 768         if (p[-1] != ' ')
 769                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 770         p++;
 771         p += strcspn(p, "<>\n");
 772         if (*p != '>')
 773                 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
 774         p++;
 775         if (*p != ' ')
 776                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
 777         p++;
 778         if (*p == '0' && p[1] != ' ')
 779                 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
 780         if (date_overflows(parse_timestamp(p, &end, 10)))
 781                 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
 782         if ((end == p || *end != ' '))
 783                 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
 784         p = end + 1;
 785         if ((*p != '+' && *p != '-') ||
 786             !isdigit(p[1]) ||
 787             !isdigit(p[2]) ||
 788             !isdigit(p[3]) ||
 789             !isdigit(p[4]) ||
 790             (p[5] != '\n'))
 791                 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
 792         p += 6;
 793         return 0;
 794 }
 795
 796 static int fsck_commit_buffer(struct commit *commit, const char *buffer,
 797         unsigned long size, struct fsck_options *options)
 798 {
 799         struct object_id tree_oid, oid;
 800         struct commit_graft *graft;
 801         unsigned parent_count, parent_line_count = 0, author_count;
 802         int err;
 803         const char *buffer_begin = buffer;
 804         const char *p;
 805
 806         if (verify_headers(buffer, size, &commit->object, options))
 807                 return -1;
 808
 809         if (!skip_prefix(buffer, "tree ", &buffer))
 810                 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
 811         if (parse_oid_hex(buffer, &tree_oid, &p) || *p != '\n') {
 812                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
 813                 if (err)
 814                         return err;
 815         }
 816         buffer = p + 1;
 817         while (skip_prefix(buffer, "parent ", &buffer)) {
 818                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 819                         err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
 820                         if (err)
 821                                 return err;
 822                 }
 823                 buffer = p + 1;
 824                 parent_line_count++;
 825         }
 826         graft = lookup_commit_graft(the_repository, &commit->object.oid);
 827         parent_count = commit_list_count(commit->parents);
 828         if (graft) {
 829                 if (graft->nr_parent == -1 && !parent_count)
 830                         ; /* shallow commit */
 831                 else if (graft->nr_parent != parent_count) {
 832                         err = report(options, &commit->object, FSCK_MSG_MISSING_GRAFT, "graft objects missing");
 833                         if (err)
 834                                 return err;
 835                 }
 836         } else {
 837                 if (parent_count != parent_line_count) {
 838                         err = report(options, &commit->object, FSCK_MSG_MISSING_PARENT, "parent objects missing");
 839                         if (err)
 840                                 return err;
 841                 }
 842         }
 843         author_count = 0;
 844         while (skip_prefix(buffer, "author ", &buffer)) {
 845                 author_count++;
 846                 err = fsck_ident(&buffer, &commit->object, options);
 847                 if (err)
 848                         return err;
 849         }
 850         if (author_count < 1)
 851                 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
 852         else if (author_count > 1)
 853                 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
 854         if (err)
 855                 return err;
 856         if (!skip_prefix(buffer, "committer ", &buffer))
 857                 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
 858         err = fsck_ident(&buffer, &commit->object, options);
 859         if (err)
 860                 return err;
 861         if (!get_commit_tree(commit)) {
 862                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE, "could not load commit's tree %s", oid_to_hex(&tree_oid));
 863                 if (err)
 864                         return err;
 865         }
 866         if (memchr(buffer_begin, '\0', size)) {
 867                 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
 868                              "NUL byte in the commit object body");
 869                 if (err)
 870                         return err;
 871         }
 872         return 0;
 873 }
 874
 875 static int fsck_commit(struct commit *commit, const char *data,
 876         unsigned long size, struct fsck_options *options)
 877 {
 878         const char *buffer = data ?  data : get_commit_buffer(commit, &size);
 879         int ret = fsck_commit_buffer(commit, buffer, size, options);
 880         if (!data)
 881                 unuse_commit_buffer(commit, buffer);
 882         return ret;
 883 }
 884
 885 static int fsck_tag_buffer(struct tag *tag, const char *data,
 886         unsigned long size, struct fsck_options *options)
 887 {
 888         struct object_id oid;
 889         int ret = 0;
 890         const char *buffer;
 891         char *to_free = NULL, *eol;
 892         struct strbuf sb = STRBUF_INIT;
 893         const char *p;
 894
 895         if (data)
 896                 buffer = data;
 897         else {
 898                 enum object_type type;
 899
 900                 buffer = to_free =
 901                         read_object_file(&tag->object.oid, &type, &size);
 902                 if (!buffer)
 903                         return report(options, &tag->object,
 904                                 FSCK_MSG_MISSING_TAG_OBJECT,
 905                                 "cannot read tag object");
 906
 907                 if (type != OBJ_TAG) {
 908                         ret = report(options, &tag->object,
 909                                 FSCK_MSG_TAG_OBJECT_NOT_TAG,
 910                                 "expected tag got %s",
 911                             type_name(type));
 912                         goto done;
 913                 }
 914         }
 915
 916         ret = verify_headers(buffer, size, &tag->object, options);
 917         if (ret)
 918                 goto done;
 919
 920         if (!skip_prefix(buffer, "object ", &buffer)) {
 921                 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
 922                 goto done;
 923         }
 924         if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 925                 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
 926                 if (ret)
 927                         goto done;
 928         }
 929         buffer = p + 1;
 930
 931         if (!skip_prefix(buffer, "type ", &buffer)) {
 932                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
 933                 goto done;
 934         }
 935         eol = strchr(buffer, '\n');
 936         if (!eol) {
 937                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
 938                 goto done;
 939         }
 940         if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
 941                 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
 942         if (ret)
 943                 goto done;
 944         buffer = eol + 1;
 945
 946         if (!skip_prefix(buffer, "tag ", &buffer)) {
 947                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
 948                 goto done;
 949         }
 950         eol = strchr(buffer, '\n');
 951         if (!eol) {
 952                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
 953                 goto done;
 954         }
 955         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
 956         if (check_refname_format(sb.buf, 0)) {
 957                 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
 958                            "invalid 'tag' name: %.*s",
 959                            (int)(eol - buffer), buffer);
 960                 if (ret)
 961                         goto done;
 962         }
 963         buffer = eol + 1;
 964
 965         if (!skip_prefix(buffer, "tagger ", &buffer)) {
 966                 /* early tags do not contain 'tagger' lines; warn only */
 967                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
 968                 if (ret)
 969                         goto done;
 970         }
 971         else
 972                 ret = fsck_ident(&buffer, &tag->object, options);
 973
 974 done:
 975         strbuf_release(&sb);
 976         free(to_free);
 977         return ret;
 978 }
 979
 980 static int fsck_tag(struct tag *tag, const char *data,
 981         unsigned long size, struct fsck_options *options)
 982 {
 983         struct object *tagged = tag->tagged;
 984
 985         if (!tagged)
 986                 return report(options, &tag->object, FSCK_MSG_BAD_TAG_OBJECT, "could not load tagged object");
 987
 988         return fsck_tag_buffer(tag, data, size, options);
 989 }
 990
 991 struct fsck_gitmodules_data {
 992         struct object *obj;
 993         struct fsck_options *options;
 994         int ret;
 995 };
 996
 997 static int fsck_gitmodules_fn(const char *var, const char *value, void *vdata)
 998 {
 999         struct fsck_gitmodules_data *data = vdata;
1000         const char *subsection, *key;
1001         int subsection_len;
1002         char *name;
1003
1004         if (parse_config_key(var, "submodule", &subsection, &subsection_len, &key) < 0 ||
1005             !subsection)
1006                 return 0;
1007
1008         name = xmemdupz(subsection, subsection_len);
1009         if (check_submodule_name(name) < 0)
1010                 data->ret |= report(data->options, data->obj,
1011                                     FSCK_MSG_GITMODULES_NAME,
1012                                     "disallowed submodule name: %s",
1013                                     name);
1014         if (!strcmp(key, "url") && value &&
1015             looks_like_command_line_option(value))
1016                 data->ret |= report(data->options, data->obj,
1017                                     FSCK_MSG_GITMODULES_URL,
1018                                     "disallowed submodule url: %s",
1019                                     value);
1020         if (!strcmp(key, "path") && value &&
1021             looks_like_command_line_option(value))
1022                 data->ret |= report(data->options, data->obj,
1023                                     FSCK_MSG_GITMODULES_PATH,
1024                                     "disallowed submodule path: %s",
1025                                     value);
1026         if (!strcmp(key, "update") && value &&
1027             parse_submodule_update_type(value) == SM_UPDATE_COMMAND)
1028                 data->ret |= report(data->options, data->obj,
1029                                     FSCK_MSG_GITMODULES_UPDATE,
1030                                     "disallowed submodule update setting: %s",
1031                                     value);
1032         free(name);
1033
1034         return 0;
1035 }
1036
1037 static int fsck_blob(struct blob *blob, const char *buf,
1038                      unsigned long size, struct fsck_options *options)
1039 {
1040         struct fsck_gitmodules_data data;
1041         struct config_options config_opts = { 0 };
1042
1043         if (!oidset_contains(&gitmodules_found, &blob->object.oid))
1044                 return 0;
1045         oidset_insert(&gitmodules_done, &blob->object.oid);
1046
1047         if (object_on_skiplist(options, &blob->object))
1048                 return 0;
1049
1050         if (!buf) {
1051                 /*
1052                  * A missing buffer here is a sign that the caller found the
1053                  * blob too gigantic to load into memory. Let's just consider
1054                  * that an error.
1055                  */
1056                 return report(options, &blob->object,
1057                               FSCK_MSG_GITMODULES_LARGE,
1058                               ".gitmodules too large to parse");
1059         }
1060
1061         data.obj = &blob->object;
1062         data.options = options;
1063         data.ret = 0;
1064         config_opts.error_action = CONFIG_ERROR_SILENT;
1065         if (git_config_from_mem(fsck_gitmodules_fn, CONFIG_ORIGIN_BLOB,
1066                                 ".gitmodules", buf, size, &data, &config_opts))
1067                 data.ret |= report(options, &blob->object,
1068                                    FSCK_MSG_GITMODULES_PARSE,
1069                                    "could not parse gitmodules blob");
1070
1071         return data.ret;
1072 }
1073
1074 int fsck_object(struct object *obj, void *data, unsigned long size,
1075         struct fsck_options *options)
1076 {
1077         if (!obj)
1078                 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
1079
1080         if (obj->type == OBJ_BLOB)
1081                 return fsck_blob((struct blob *)obj, data, size, options);
1082         if (obj->type == OBJ_TREE)
1083                 return fsck_tree((struct tree *) obj, options);
1084         if (obj->type == OBJ_COMMIT)
1085                 return fsck_commit((struct commit *) obj, (const char *) data,
1086                         size, options);
1087         if (obj->type == OBJ_TAG)
1088                 return fsck_tag((struct tag *) obj, (const char *) data,
1089                         size, options);
1090
1091         return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
1092                           obj->type);
1093 }
1094
1095 int fsck_error_function(struct fsck_options *o,
1096         struct object *obj, int msg_type, const char *message)
1097 {
1098         if (msg_type == FSCK_WARN) {
1099                 warning("object %s: %s", describe_object(o, obj), message);
1100                 return 0;
1101         }
1102         error("object %s: %s", describe_object(o, obj), message);
1103         return 1;
1104 }
1105
1106 int fsck_finish(struct fsck_options *options)
1107 {
1108         int ret = 0;
1109         struct oidset_iter iter;
1110         const struct object_id *oid;
1111
1112         oidset_iter_init(&gitmodules_found, &iter);
1113         while ((oid = oidset_iter_next(&iter))) {
1114                 struct blob *blob;
1115                 enum object_type type;
1116                 unsigned long size;
1117                 char *buf;
1118
1119                 if (oidset_contains(&gitmodules_done, oid))
1120                         continue;
1121
1122                 blob = lookup_blob(the_repository, oid);
1123                 if (!blob) {
1124                         struct object *obj = lookup_unknown_object(oid->hash);
1125                         ret |= report(options, obj,
1126                                       FSCK_MSG_GITMODULES_BLOB,
1127                                       "non-blob found at .gitmodules");
1128                         continue;
1129                 }
1130
1131                 buf = read_object_file(oid, &type, &size);
1132                 if (!buf) {
1133                         if (is_promisor_object(&blob->object.oid))
1134                                 continue;
1135                         ret |= report(options, &blob->object,
1136                                       FSCK_MSG_GITMODULES_MISSING,
1137                                       "unable to read .gitmodules blob");
1138                         continue;
1139                 }
1140
1141                 if (type == OBJ_BLOB)
1142                         ret |= fsck_blob(blob, buf, size, options);
1143                 else
1144                         ret |= report(options, &blob->object,
1145                                       FSCK_MSG_GITMODULES_BLOB,
1146                                       "non-blob found at .gitmodules");
1147                 free(buf);
1148         }
1149
1150
1151         oidset_clear(&gitmodules_found);
1152         oidset_clear(&gitmodules_done);
1153         return ret;
1154 }