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