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, struct object *obj)
 281 {
 282         return opts && obj && oidset_contains(&opts->skiplist, &obj->oid);
 283 }
 284
 285 __attribute__((format (printf, 4, 5)))
 286 static int report(struct fsck_options *options, struct object *object,
 287         enum fsck_msg_id id, const char *fmt, ...)
 288 {
 289         va_list ap;
 290         struct strbuf sb = STRBUF_INIT;
 291         int msg_type = fsck_msg_type(id, options), result;
 292
 293         if (msg_type == FSCK_IGNORE)
 294                 return 0;
 295
 296         if (object_on_skiplist(options, object))
 297                 return 0;
 298
 299         if (msg_type == FSCK_FATAL)
 300                 msg_type = FSCK_ERROR;
 301         else if (msg_type == FSCK_INFO)
 302                 msg_type = FSCK_WARN;
 303
 304         append_msg_id(&sb, msg_id_info[id].id_string);
 305
 306         va_start(ap, fmt);
 307         strbuf_vaddf(&sb, fmt, ap);
 308         result = options->error_func(options, object, msg_type, sb.buf);
 309         strbuf_release(&sb);
 310         va_end(ap);
 311
 312         return result;
 313 }
 314
 315 void fsck_enable_object_names(struct fsck_options *options)
 316 {
 317         if (!options->object_names)
 318                 options->object_names = xcalloc(1, sizeof(struct decoration));
 319 }
 320
 321 const char *fsck_get_object_name(struct fsck_options *options, struct object *obj)
 322 {
 323         if (!options->object_names)
 324                 return NULL;
 325         return lookup_decoration(options->object_names, obj);
 326 }
 327
 328 void fsck_put_object_name(struct fsck_options *options, struct object *obj,
 329                           const char *fmt, ...)
 330 {
 331         va_list ap;
 332         struct strbuf buf = STRBUF_INIT;
 333         char *existing;
 334
 335         if (!options->object_names)
 336                 return;
 337         existing = lookup_decoration(options->object_names, obj);
 338         if (existing)
 339                 return;
 340         va_start(ap, fmt);
 341         strbuf_vaddf(&buf, fmt, ap);
 342         add_decoration(options->object_names, obj, strbuf_detach(&buf, NULL));
 343         va_end(ap);
 344 }
 345
 346 const char *fsck_describe_object(struct fsck_options *options,
 347                                  struct object *obj)
 348 {
 349         static struct strbuf bufs[] = {
 350                 STRBUF_INIT, STRBUF_INIT, STRBUF_INIT, STRBUF_INIT
 351         };
 352         static int b = 0;
 353         struct strbuf *buf;
 354         const char *name = fsck_get_object_name(options, obj);
 355
 356         buf = bufs + b;
 357         b = (b + 1) % ARRAY_SIZE(bufs);
 358         strbuf_reset(buf);
 359         strbuf_addstr(buf, oid_to_hex(&obj->oid));
 360         if (name)
 361                 strbuf_addf(buf, " (%s)", name);
 362
 363         return buf->buf;
 364 }
 365
 366 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
 367 {
 368         struct tree_desc desc;
 369         struct name_entry entry;
 370         int res = 0;
 371         const char *name;
 372
 373         if (parse_tree(tree))
 374                 return -1;
 375
 376         name = fsck_get_object_name(options, &tree->object);
 377         if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
 378                 return -1;
 379         while (tree_entry_gently(&desc, &entry)) {
 380                 struct object *obj;
 381                 int result;
 382
 383                 if (S_ISGITLINK(entry.mode))
 384                         continue;
 385
 386                 if (S_ISDIR(entry.mode)) {
 387                         obj = (struct object *)lookup_tree(the_repository, &entry.oid);
 388                         if (name && obj)
 389                                 fsck_put_object_name(options, obj, "%s%s/",
 390                                                      name, entry.path);
 391                         result = options->walk(obj, OBJ_TREE, data, options);
 392                 }
 393                 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
 394                         obj = (struct object *)lookup_blob(the_repository, &entry.oid);
 395                         if (name && obj)
 396                                 fsck_put_object_name(options, obj, "%s%s",
 397                                                      name, entry.path);
 398                         result = options->walk(obj, OBJ_BLOB, data, options);
 399                 }
 400                 else {
 401                         result = error("in tree %s: entry %s has bad mode %.6o",
 402                                        fsck_describe_object(options, &tree->object),
 403                                        entry.path, entry.mode);
 404                 }
 405                 if (result < 0)
 406                         return result;
 407                 if (!res)
 408                         res = result;
 409         }
 410         return res;
 411 }
 412
 413 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
 414 {
 415         int counter = 0, generation = 0, name_prefix_len = 0;
 416         struct commit_list *parents;
 417         int res;
 418         int result;
 419         const char *name;
 420
 421         if (parse_commit(commit))
 422                 return -1;
 423
 424         name = fsck_get_object_name(options, &commit->object);
 425         if (name)
 426                 fsck_put_object_name(options, &get_commit_tree(commit)->object,
 427                                      "%s:", name);
 428
 429         result = options->walk((struct object *)get_commit_tree(commit),
 430                                OBJ_TREE, data, options);
 431         if (result < 0)
 432                 return result;
 433         res = result;
 434
 435         parents = commit->parents;
 436         if (name && parents) {
 437                 int len = strlen(name), power;
 438
 439                 if (len && name[len - 1] == '^') {
 440                         generation = 1;
 441                         name_prefix_len = len - 1;
 442                 }
 443                 else { /* parse ~<generation> suffix */
 444                         for (generation = 0, power = 1;
 445                              len && isdigit(name[len - 1]);
 446                              power *= 10)
 447                                 generation += power * (name[--len] - '0');
 448                         if (power > 1 && len && name[len - 1] == '~')
 449                                 name_prefix_len = len - 1;
 450                 }
 451         }
 452
 453         while (parents) {
 454                 if (name) {
 455                         struct object *obj = &parents->item->object;
 456
 457                         if (counter++)
 458                                 fsck_put_object_name(options, obj, "%s^%d",
 459                                                      name, counter);
 460                         else if (generation > 0)
 461                                 fsck_put_object_name(options, obj, "%.*s~%d",
 462                                                      name_prefix_len, name,
 463                                                      generation + 1);
 464                         else
 465                                 fsck_put_object_name(options, obj, "%s^",
 466                                                      name);
 467                 }
 468                 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
 469                 if (result < 0)
 470                         return result;
 471                 if (!res)
 472                         res = result;
 473                 parents = parents->next;
 474         }
 475         return res;
 476 }
 477
 478 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
 479 {
 480         const char *name = fsck_get_object_name(options, &tag->object);
 481
 482         if (parse_tag(tag))
 483                 return -1;
 484         if (name)
 485                 fsck_put_object_name(options, tag->tagged, "%s", name);
 486         return options->walk(tag->tagged, OBJ_ANY, data, options);
 487 }
 488
 489 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
 490 {
 491         if (!obj)
 492                 return -1;
 493
 494         if (obj->type == OBJ_NONE)
 495                 parse_object(the_repository, &obj->oid);
 496
 497         switch (obj->type) {
 498         case OBJ_BLOB:
 499                 return 0;
 500         case OBJ_TREE:
 501                 return fsck_walk_tree((struct tree *)obj, data, options);
 502         case OBJ_COMMIT:
 503                 return fsck_walk_commit((struct commit *)obj, data, options);
 504         case OBJ_TAG:
 505                 return fsck_walk_tag((struct tag *)obj, data, options);
 506         default:
 507                 error("Unknown object type for %s",
 508                       fsck_describe_object(options, obj));
 509                 return -1;
 510         }
 511 }
 512
 513 /*
 514  * The entries in a tree are ordered in the _path_ order,
 515  * which means that a directory entry is ordered by adding
 516  * a slash to the end of it.
 517  *
 518  * So a directory called "a" is ordered _after_ a file
 519  * called "a.c", because "a/" sorts after "a.c".
 520  */
 521 #define TREE_UNORDERED (-1)
 522 #define TREE_HAS_DUPS  (-2)
 523
 524 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 525 {
 526         int len1 = strlen(name1);
 527         int len2 = strlen(name2);
 528         int len = len1 < len2 ? len1 : len2;
 529         unsigned char c1, c2;
 530         int cmp;
 531
 532         cmp = memcmp(name1, name2, len);
 533         if (cmp < 0)
 534                 return 0;
 535         if (cmp > 0)
 536                 return TREE_UNORDERED;
 537
 538         /*
 539          * Ok, the first <len> characters are the same.
 540          * Now we need to order the next one, but turn
 541          * a '\0' into a '/' for a directory entry.
 542          */
 543         c1 = name1[len];
 544         c2 = name2[len];
 545         if (!c1 && !c2)
 546                 /*
 547                  * git-write-tree used to write out a nonsense tree that has
 548                  * entries with the same name, one blob and one tree.  Make
 549                  * sure we do not have duplicate entries.
 550                  */
 551                 return TREE_HAS_DUPS;
 552         if (!c1 && S_ISDIR(mode1))
 553                 c1 = '/';
 554         if (!c2 && S_ISDIR(mode2))
 555                 c2 = '/';
 556         return c1 < c2 ? 0 : TREE_UNORDERED;
 557 }
 558
 559 static int fsck_tree(struct tree *item,
 560                      const char *buffer, unsigned long size,
 561                      struct fsck_options *options)
 562 {
 563         int retval = 0;
 564         int has_null_sha1 = 0;
 565         int has_full_path = 0;
 566         int has_empty_name = 0;
 567         int has_dot = 0;
 568         int has_dotdot = 0;
 569         int has_dotgit = 0;
 570         int has_zero_pad = 0;
 571         int has_bad_modes = 0;
 572         int has_dup_entries = 0;
 573         int not_properly_sorted = 0;
 574         struct tree_desc desc;
 575         unsigned o_mode;
 576         const char *o_name;
 577
 578         if (init_tree_desc_gently(&desc, buffer, size)) {
 579                 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 580                 return retval;
 581         }
 582
 583         o_mode = 0;
 584         o_name = NULL;
 585
 586         while (desc.size) {
 587                 unsigned short mode;
 588                 const char *name;
 589                 const struct object_id *oid;
 590
 591                 oid = tree_entry_extract(&desc, &name, &mode);
 592
 593                 has_null_sha1 |= is_null_oid(oid);
 594                 has_full_path |= !!strchr(name, '/');
 595                 has_empty_name |= !*name;
 596                 has_dot |= !strcmp(name, ".");
 597                 has_dotdot |= !strcmp(name, "..");
 598                 has_dotgit |= is_hfs_dotgit(name) || is_ntfs_dotgit(name);
 599                 has_zero_pad |= *(char *)desc.buffer == '0';
 600
 601                 if (is_hfs_dotgitmodules(name) || is_ntfs_dotgitmodules(name)) {
 602                         if (!S_ISLNK(mode))
 603                                 oidset_insert(&gitmodules_found, oid);
 604                         else
 605                                 retval += report(options, &item->object,
 606                                                  FSCK_MSG_GITMODULES_SYMLINK,
 607                                                  ".gitmodules is a symbolic link");
 608                 }
 609
 610                 if (update_tree_entry_gently(&desc)) {
 611                         retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 612                         break;
 613                 }
 614
 615                 switch (mode) {
 616                 /*
 617                  * Standard modes..
 618                  */
 619                 case S_IFREG | 0755:
 620                 case S_IFREG | 0644:
 621                 case S_IFLNK:
 622                 case S_IFDIR:
 623                 case S_IFGITLINK:
 624                         break;
 625                 /*
 626                  * This is nonstandard, but we had a few of these
 627                  * early on when we honored the full set of mode
 628                  * bits..
 629                  */
 630                 case S_IFREG | 0664:
 631                         if (!options->strict)
 632                                 break;
 633                         /* fallthrough */
 634                 default:
 635                         has_bad_modes = 1;
 636                 }
 637
 638                 if (o_name) {
 639                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 640                         case TREE_UNORDERED:
 641                                 not_properly_sorted = 1;
 642                                 break;
 643                         case TREE_HAS_DUPS:
 644                                 has_dup_entries = 1;
 645                                 break;
 646                         default:
 647                                 break;
 648                         }
 649                 }
 650
 651                 o_mode = mode;
 652                 o_name = name;
 653         }
 654
 655         if (has_null_sha1)
 656                 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
 657         if (has_full_path)
 658                 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
 659         if (has_empty_name)
 660                 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
 661         if (has_dot)
 662                 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
 663         if (has_dotdot)
 664                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
 665         if (has_dotgit)
 666                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
 667         if (has_zero_pad)
 668                 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
 669         if (has_bad_modes)
 670                 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
 671         if (has_dup_entries)
 672                 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
 673         if (not_properly_sorted)
 674                 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
 675         return retval;
 676 }
 677
 678 static int verify_headers(const void *data, unsigned long size,
 679                           struct object *obj, struct fsck_options *options)
 680 {
 681         const char *buffer = (const char *)data;
 682         unsigned long i;
 683
 684         for (i = 0; i < size; i++) {
 685                 switch (buffer[i]) {
 686                 case '\0':
 687                         return report(options, obj,
 688                                 FSCK_MSG_NUL_IN_HEADER,
 689                                 "unterminated header: NUL at offset %ld", i);
 690                 case '\n':
 691                         if (i + 1 < size && buffer[i + 1] == '\n')
 692                                 return 0;
 693                 }
 694         }
 695
 696         /*
 697          * We did not find double-LF that separates the header
 698          * and the body.  Not having a body is not a crime but
 699          * we do want to see the terminating LF for the last header
 700          * line.
 701          */
 702         if (size && buffer[size - 1] == '\n')
 703                 return 0;
 704
 705         return report(options, obj,
 706                 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
 707 }
 708
 709 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
 710 {
 711         const char *p = *ident;
 712         char *end;
 713
 714         *ident = strchrnul(*ident, '\n');
 715         if (**ident == '\n')
 716                 (*ident)++;
 717
 718         if (*p == '<')
 719                 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 720         p += strcspn(p, "<>\n");
 721         if (*p == '>')
 722                 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
 723         if (*p != '<')
 724                 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
 725         if (p[-1] != ' ')
 726                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 727         p++;
 728         p += strcspn(p, "<>\n");
 729         if (*p != '>')
 730                 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
 731         p++;
 732         if (*p != ' ')
 733                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
 734         p++;
 735         if (*p == '0' && p[1] != ' ')
 736                 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
 737         if (date_overflows(parse_timestamp(p, &end, 10)))
 738                 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
 739         if ((end == p || *end != ' '))
 740                 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
 741         p = end + 1;
 742         if ((*p != '+' && *p != '-') ||
 743             !isdigit(p[1]) ||
 744             !isdigit(p[2]) ||
 745             !isdigit(p[3]) ||
 746             !isdigit(p[4]) ||
 747             (p[5] != '\n'))
 748                 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
 749         p += 6;
 750         return 0;
 751 }
 752
 753 static int fsck_commit(struct commit *commit, const char *buffer,
 754                        unsigned long size, struct fsck_options *options)
 755 {
 756         struct object_id tree_oid, oid;
 757         unsigned author_count;
 758         int err;
 759         const char *buffer_begin = buffer;
 760         const char *p;
 761
 762         if (verify_headers(buffer, size, &commit->object, options))
 763                 return -1;
 764
 765         if (!skip_prefix(buffer, "tree ", &buffer))
 766                 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
 767         if (parse_oid_hex(buffer, &tree_oid, &p) || *p != '\n') {
 768                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
 769                 if (err)
 770                         return err;
 771         }
 772         buffer = p + 1;
 773         while (skip_prefix(buffer, "parent ", &buffer)) {
 774                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 775                         err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
 776                         if (err)
 777                                 return err;
 778                 }
 779                 buffer = p + 1;
 780         }
 781         author_count = 0;
 782         while (skip_prefix(buffer, "author ", &buffer)) {
 783                 author_count++;
 784                 err = fsck_ident(&buffer, &commit->object, options);
 785                 if (err)
 786                         return err;
 787         }
 788         if (author_count < 1)
 789                 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
 790         else if (author_count > 1)
 791                 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
 792         if (err)
 793                 return err;
 794         if (!skip_prefix(buffer, "committer ", &buffer))
 795                 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
 796         err = fsck_ident(&buffer, &commit->object, options);
 797         if (err)
 798                 return err;
 799         if (memchr(buffer_begin, '\0', size)) {
 800                 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
 801                              "NUL byte in the commit object body");
 802                 if (err)
 803                         return err;
 804         }
 805         return 0;
 806 }
 807
 808 static int fsck_tag(struct tag *tag, const char *buffer,
 809                     unsigned long size, struct fsck_options *options)
 810 {
 811         struct object_id oid;
 812         int ret = 0;
 813         char *eol;
 814         struct strbuf sb = STRBUF_INIT;
 815         const char *p;
 816
 817         ret = verify_headers(buffer, size, &tag->object, options);
 818         if (ret)
 819                 goto done;
 820
 821         if (!skip_prefix(buffer, "object ", &buffer)) {
 822                 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
 823                 goto done;
 824         }
 825         if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 826                 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
 827                 if (ret)
 828                         goto done;
 829         }
 830         buffer = p + 1;
 831
 832         if (!skip_prefix(buffer, "type ", &buffer)) {
 833                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
 834                 goto done;
 835         }
 836         eol = strchr(buffer, '\n');
 837         if (!eol) {
 838                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
 839                 goto done;
 840         }
 841         if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
 842                 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
 843         if (ret)
 844                 goto done;
 845         buffer = eol + 1;
 846
 847         if (!skip_prefix(buffer, "tag ", &buffer)) {
 848                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
 849                 goto done;
 850         }
 851         eol = strchr(buffer, '\n');
 852         if (!eol) {
 853                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
 854                 goto done;
 855         }
 856         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
 857         if (check_refname_format(sb.buf, 0)) {
 858                 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
 859                            "invalid 'tag' name: %.*s",
 860                            (int)(eol - buffer), buffer);
 861                 if (ret)
 862                         goto done;
 863         }
 864         buffer = eol + 1;
 865
 866         if (!skip_prefix(buffer, "tagger ", &buffer)) {
 867                 /* early tags do not contain 'tagger' lines; warn only */
 868                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
 869                 if (ret)
 870                         goto done;
 871         }
 872         else
 873                 ret = fsck_ident(&buffer, &tag->object, options);
 874
 875 done:
 876         strbuf_release(&sb);
 877         return ret;
 878 }
 879
 880 struct fsck_gitmodules_data {
 881         struct object *obj;
 882         struct fsck_options *options;
 883         int ret;
 884 };
 885
 886 static int fsck_gitmodules_fn(const char *var, const char *value, void *vdata)
 887 {
 888         struct fsck_gitmodules_data *data = vdata;
 889         const char *subsection, *key;
 890         int subsection_len;
 891         char *name;
 892
 893         if (parse_config_key(var, "submodule", &subsection, &subsection_len, &key) < 0 ||
 894             !subsection)
 895                 return 0;
 896
 897         name = xmemdupz(subsection, subsection_len);
 898         if (check_submodule_name(name) < 0)
 899                 data->ret |= report(data->options, data->obj,
 900                                     FSCK_MSG_GITMODULES_NAME,
 901                                     "disallowed submodule name: %s",
 902                                     name);
 903         if (!strcmp(key, "url") && value &&
 904             looks_like_command_line_option(value))
 905                 data->ret |= report(data->options, data->obj,
 906                                     FSCK_MSG_GITMODULES_URL,
 907                                     "disallowed submodule url: %s",
 908                                     value);
 909         if (!strcmp(key, "path") && value &&
 910             looks_like_command_line_option(value))
 911                 data->ret |= report(data->options, data->obj,
 912                                     FSCK_MSG_GITMODULES_PATH,
 913                                     "disallowed submodule path: %s",
 914                                     value);
 915         free(name);
 916
 917         return 0;
 918 }
 919
 920 static int fsck_blob(struct blob *blob, const char *buf,
 921                      unsigned long size, struct fsck_options *options)
 922 {
 923         struct fsck_gitmodules_data data;
 924         struct config_options config_opts = { 0 };
 925
 926         if (!oidset_contains(&gitmodules_found, &blob->object.oid))
 927                 return 0;
 928         oidset_insert(&gitmodules_done, &blob->object.oid);
 929
 930         if (object_on_skiplist(options, &blob->object))
 931                 return 0;
 932
 933         if (!buf) {
 934                 /*
 935                  * A missing buffer here is a sign that the caller found the
 936                  * blob too gigantic to load into memory. Let's just consider
 937                  * that an error.
 938                  */
 939                 return report(options, &blob->object,
 940                               FSCK_MSG_GITMODULES_LARGE,
 941                               ".gitmodules too large to parse");
 942         }
 943
 944         data.obj = &blob->object;
 945         data.options = options;
 946         data.ret = 0;
 947         config_opts.error_action = CONFIG_ERROR_SILENT;
 948         if (git_config_from_mem(fsck_gitmodules_fn, CONFIG_ORIGIN_BLOB,
 949                                 ".gitmodules", buf, size, &data, &config_opts))
 950                 data.ret |= report(options, &blob->object,
 951                                    FSCK_MSG_GITMODULES_PARSE,
 952                                    "could not parse gitmodules blob");
 953
 954         return data.ret;
 955 }
 956
 957 int fsck_object(struct object *obj, void *data, unsigned long size,
 958         struct fsck_options *options)
 959 {
 960         if (!obj)
 961                 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
 962
 963         if (obj->type == OBJ_BLOB)
 964                 return fsck_blob((struct blob *)obj, data, size, options);
 965         if (obj->type == OBJ_TREE)
 966                 return fsck_tree((struct tree *) obj, data, size, options);
 967         if (obj->type == OBJ_COMMIT)
 968                 return fsck_commit((struct commit *) obj, (const char *) data,
 969                         size, options);
 970         if (obj->type == OBJ_TAG)
 971                 return fsck_tag((struct tag *) obj, (const char *) data,
 972                         size, options);
 973
 974         return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
 975                           obj->type);
 976 }
 977
 978 int fsck_error_function(struct fsck_options *o,
 979         struct object *obj, int msg_type, const char *message)
 980 {
 981         if (msg_type == FSCK_WARN) {
 982                 warning("object %s: %s", fsck_describe_object(o, obj), message);
 983                 return 0;
 984         }
 985         error("object %s: %s", fsck_describe_object(o, obj), message);
 986         return 1;
 987 }
 988
 989 int fsck_finish(struct fsck_options *options)
 990 {
 991         int ret = 0;
 992         struct oidset_iter iter;
 993         const struct object_id *oid;
 994
 995         oidset_iter_init(&gitmodules_found, &iter);
 996         while ((oid = oidset_iter_next(&iter))) {
 997                 struct blob *blob;
 998                 enum object_type type;
 999                 unsigned long size;
1000                 char *buf;
1001
1002                 if (oidset_contains(&gitmodules_done, oid))
1003                         continue;
1004
1005                 blob = lookup_blob(the_repository, oid);
1006                 if (!blob) {
1007                         struct object *obj = lookup_unknown_object(oid);
1008                         ret |= report(options, obj,
1009                                       FSCK_MSG_GITMODULES_BLOB,
1010                                       "non-blob found at .gitmodules");
1011                         continue;
1012                 }
1013
1014                 buf = read_object_file(oid, &type, &size);
1015                 if (!buf) {
1016                         if (is_promisor_object(&blob->object.oid))
1017                                 continue;
1018                         ret |= report(options, &blob->object,
1019                                       FSCK_MSG_GITMODULES_MISSING,
1020                                       "unable to read .gitmodules blob");
1021                         continue;
1022                 }
1023
1024                 if (type == OBJ_BLOB)
1025                         ret |= fsck_blob(blob, buf, size, options);
1026                 else
1027                         ret |= report(options, &blob->object,
1028                                       FSCK_MSG_GITMODULES_BLOB,
1029                                       "non-blob found at .gitmodules");
1030                 free(buf);
1031         }
1032
1033
1034         oidset_clear(&gitmodules_found);
1035         oidset_clear(&gitmodules_done);
1036         return ret;
1037 }