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_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         FILE *fp;
 189         struct strbuf sb = STRBUF_INIT;
 190         struct object_id oid;
 191
 192         fp = fopen(path, "r");
 193         if (!fp)
 194                 die("Could not open skip list: %s", path);
 195         while (!strbuf_getline(&sb, fp)) {
 196                 const char *p;
 197                 const char *hash;
 198
 199                 /*
 200                  * Allow trailing comments, leading whitespace
 201                  * (including before commits), and empty or whitespace
 202                  * only lines.
 203                  */
 204                 hash = strchr(sb.buf, '#');
 205                 if (hash)
 206                         strbuf_setlen(&sb, hash - sb.buf);
 207                 strbuf_trim(&sb);
 208                 if (!sb.len)
 209                         continue;
 210
 211                 if (parse_oid_hex(sb.buf, &oid, &p) || *p != '\0')
 212                         die("Invalid SHA-1: %s", sb.buf);
 213                 oidset_insert(&options->skiplist, &oid);
 214         }
 215         if (ferror(fp))
 216                 die_errno("Could not read '%s'", path);
 217         fclose(fp);
 218         strbuf_release(&sb);
 219 }
 220
 221 static int parse_msg_type(const char *str)
 222 {
 223         if (!strcmp(str, "error"))
 224                 return FSCK_ERROR;
 225         else if (!strcmp(str, "warn"))
 226                 return FSCK_WARN;
 227         else if (!strcmp(str, "ignore"))
 228                 return FSCK_IGNORE;
 229         else
 230                 die("Unknown fsck message type: '%s'", str);
 231 }
 232
 233 int is_valid_msg_type(const char *msg_id, const char *msg_type)
 234 {
 235         if (parse_msg_id(msg_id) < 0)
 236                 return 0;
 237         parse_msg_type(msg_type);
 238         return 1;
 239 }
 240
 241 void fsck_set_msg_type(struct fsck_options *options,
 242                 const char *msg_id, const char *msg_type)
 243 {
 244         int id = parse_msg_id(msg_id), type;
 245
 246         if (id < 0)
 247                 die("Unhandled message id: %s", msg_id);
 248         type = parse_msg_type(msg_type);
 249
 250         if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
 251                 die("Cannot demote %s to %s", msg_id, msg_type);
 252
 253         if (!options->msg_type) {
 254                 int i;
 255                 int *msg_type;
 256                 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
 257                 for (i = 0; i < FSCK_MSG_MAX; i++)
 258                         msg_type[i] = fsck_msg_type(i, options);
 259                 options->msg_type = msg_type;
 260         }
 261
 262         options->msg_type[id] = type;
 263 }
 264
 265 void fsck_set_msg_types(struct fsck_options *options, const char *values)
 266 {
 267         char *buf = xstrdup(values), *to_free = buf;
 268         int done = 0;
 269
 270         while (!done) {
 271                 int len = strcspn(buf, " ,|"), equal;
 272
 273                 done = !buf[len];
 274                 if (!len) {
 275                         buf++;
 276                         continue;
 277                 }
 278                 buf[len] = '\0';
 279
 280                 for (equal = 0;
 281                      equal < len && buf[equal] != '=' && buf[equal] != ':';
 282                      equal++)
 283                         buf[equal] = tolower(buf[equal]);
 284                 buf[equal] = '\0';
 285
 286                 if (!strcmp(buf, "skiplist")) {
 287                         if (equal == len)
 288                                 die("skiplist requires a path");
 289                         init_skiplist(options, buf + equal + 1);
 290                         buf += len + 1;
 291                         continue;
 292                 }
 293
 294                 if (equal == len)
 295                         die("Missing '=': '%s'", buf);
 296
 297                 fsck_set_msg_type(options, buf, buf + equal + 1);
 298                 buf += len + 1;
 299         }
 300         free(to_free);
 301 }
 302
 303 static void append_msg_id(struct strbuf *sb, const char *msg_id)
 304 {
 305         for (;;) {
 306                 char c = *(msg_id)++;
 307
 308                 if (!c)
 309                         break;
 310                 if (c != '_')
 311                         strbuf_addch(sb, tolower(c));
 312                 else {
 313                         assert(*msg_id);
 314                         strbuf_addch(sb, *(msg_id)++);
 315                 }
 316         }
 317
 318         strbuf_addstr(sb, ": ");
 319 }
 320
 321 static int object_on_skiplist(struct fsck_options *opts, struct object *obj)
 322 {
 323         return opts && obj && oidset_contains(&opts->skiplist, &obj->oid);
 324 }
 325
 326 __attribute__((format (printf, 4, 5)))
 327 static int report(struct fsck_options *options, struct object *object,
 328         enum fsck_msg_id id, const char *fmt, ...)
 329 {
 330         va_list ap;
 331         struct strbuf sb = STRBUF_INIT;
 332         int msg_type = fsck_msg_type(id, options), result;
 333
 334         if (msg_type == FSCK_IGNORE)
 335                 return 0;
 336
 337         if (object_on_skiplist(options, object))
 338                 return 0;
 339
 340         if (msg_type == FSCK_FATAL)
 341                 msg_type = FSCK_ERROR;
 342         else if (msg_type == FSCK_INFO)
 343                 msg_type = FSCK_WARN;
 344
 345         append_msg_id(&sb, msg_id_info[id].id_string);
 346
 347         va_start(ap, fmt);
 348         strbuf_vaddf(&sb, fmt, ap);
 349         result = options->error_func(options, object, msg_type, sb.buf);
 350         strbuf_release(&sb);
 351         va_end(ap);
 352
 353         return result;
 354 }
 355
 356 static char *get_object_name(struct fsck_options *options, struct object *obj)
 357 {
 358         if (!options->object_names)
 359                 return NULL;
 360         return lookup_decoration(options->object_names, obj);
 361 }
 362
 363 static void put_object_name(struct fsck_options *options, struct object *obj,
 364         const char *fmt, ...)
 365 {
 366         va_list ap;
 367         struct strbuf buf = STRBUF_INIT;
 368         char *existing;
 369
 370         if (!options->object_names)
 371                 return;
 372         existing = lookup_decoration(options->object_names, obj);
 373         if (existing)
 374                 return;
 375         va_start(ap, fmt);
 376         strbuf_vaddf(&buf, fmt, ap);
 377         add_decoration(options->object_names, obj, strbuf_detach(&buf, NULL));
 378         va_end(ap);
 379 }
 380
 381 static const char *describe_object(struct fsck_options *o, struct object *obj)
 382 {
 383         static struct strbuf buf = STRBUF_INIT;
 384         char *name;
 385
 386         strbuf_reset(&buf);
 387         strbuf_addstr(&buf, oid_to_hex(&obj->oid));
 388         if (o->object_names && (name = lookup_decoration(o->object_names, obj)))
 389                 strbuf_addf(&buf, " (%s)", name);
 390
 391         return buf.buf;
 392 }
 393
 394 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
 395 {
 396         struct tree_desc desc;
 397         struct name_entry entry;
 398         int res = 0;
 399         const char *name;
 400
 401         if (parse_tree(tree))
 402                 return -1;
 403
 404         name = get_object_name(options, &tree->object);
 405         if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
 406                 return -1;
 407         while (tree_entry_gently(&desc, &entry)) {
 408                 struct object *obj;
 409                 int result;
 410
 411                 if (S_ISGITLINK(entry.mode))
 412                         continue;
 413
 414                 if (S_ISDIR(entry.mode)) {
 415                         obj = (struct object *)lookup_tree(the_repository, &entry.oid);
 416                         if (name && obj)
 417                                 put_object_name(options, obj, "%s%s/", name,
 418                                         entry.path);
 419                         result = options->walk(obj, OBJ_TREE, data, options);
 420                 }
 421                 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
 422                         obj = (struct object *)lookup_blob(the_repository, &entry.oid);
 423                         if (name && obj)
 424                                 put_object_name(options, obj, "%s%s", name,
 425                                         entry.path);
 426                         result = options->walk(obj, OBJ_BLOB, data, options);
 427                 }
 428                 else {
 429                         result = error("in tree %s: entry %s has bad mode %.6o",
 430                                         describe_object(options, &tree->object), entry.path, entry.mode);
 431                 }
 432                 if (result < 0)
 433                         return result;
 434                 if (!res)
 435                         res = result;
 436         }
 437         return res;
 438 }
 439
 440 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
 441 {
 442         int counter = 0, generation = 0, name_prefix_len = 0;
 443         struct commit_list *parents;
 444         int res;
 445         int result;
 446         const char *name;
 447
 448         if (parse_commit(commit))
 449                 return -1;
 450
 451         name = get_object_name(options, &commit->object);
 452         if (name)
 453                 put_object_name(options, &get_commit_tree(commit)->object,
 454                                 "%s:", name);
 455
 456         result = options->walk((struct object *)get_commit_tree(commit),
 457                                OBJ_TREE, data, options);
 458         if (result < 0)
 459                 return result;
 460         res = result;
 461
 462         parents = commit->parents;
 463         if (name && parents) {
 464                 int len = strlen(name), power;
 465
 466                 if (len && name[len - 1] == '^') {
 467                         generation = 1;
 468                         name_prefix_len = len - 1;
 469                 }
 470                 else { /* parse ~<generation> suffix */
 471                         for (generation = 0, power = 1;
 472                              len && isdigit(name[len - 1]);
 473                              power *= 10)
 474                                 generation += power * (name[--len] - '0');
 475                         if (power > 1 && len && name[len - 1] == '~')
 476                                 name_prefix_len = len - 1;
 477                 }
 478         }
 479
 480         while (parents) {
 481                 if (name) {
 482                         struct object *obj = &parents->item->object;
 483
 484                         if (counter++)
 485                                 put_object_name(options, obj, "%s^%d",
 486                                         name, counter);
 487                         else if (generation > 0)
 488                                 put_object_name(options, obj, "%.*s~%d",
 489                                         name_prefix_len, name, generation + 1);
 490                         else
 491                                 put_object_name(options, obj, "%s^", name);
 492                 }
 493                 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
 494                 if (result < 0)
 495                         return result;
 496                 if (!res)
 497                         res = result;
 498                 parents = parents->next;
 499         }
 500         return res;
 501 }
 502
 503 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
 504 {
 505         char *name = get_object_name(options, &tag->object);
 506
 507         if (parse_tag(tag))
 508                 return -1;
 509         if (name)
 510                 put_object_name(options, tag->tagged, "%s", name);
 511         return options->walk(tag->tagged, OBJ_ANY, data, options);
 512 }
 513
 514 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
 515 {
 516         if (!obj)
 517                 return -1;
 518
 519         if (obj->type == OBJ_NONE)
 520                 parse_object(the_repository, &obj->oid);
 521
 522         switch (obj->type) {
 523         case OBJ_BLOB:
 524                 return 0;
 525         case OBJ_TREE:
 526                 return fsck_walk_tree((struct tree *)obj, data, options);
 527         case OBJ_COMMIT:
 528                 return fsck_walk_commit((struct commit *)obj, data, options);
 529         case OBJ_TAG:
 530                 return fsck_walk_tag((struct tag *)obj, data, options);
 531         default:
 532                 error("Unknown object type for %s", describe_object(options, obj));
 533                 return -1;
 534         }
 535 }
 536
 537 /*
 538  * The entries in a tree are ordered in the _path_ order,
 539  * which means that a directory entry is ordered by adding
 540  * a slash to the end of it.
 541  *
 542  * So a directory called "a" is ordered _after_ a file
 543  * called "a.c", because "a/" sorts after "a.c".
 544  */
 545 #define TREE_UNORDERED (-1)
 546 #define TREE_HAS_DUPS  (-2)
 547
 548 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 549 {
 550         int len1 = strlen(name1);
 551         int len2 = strlen(name2);
 552         int len = len1 < len2 ? len1 : len2;
 553         unsigned char c1, c2;
 554         int cmp;
 555
 556         cmp = memcmp(name1, name2, len);
 557         if (cmp < 0)
 558                 return 0;
 559         if (cmp > 0)
 560                 return TREE_UNORDERED;
 561
 562         /*
 563          * Ok, the first <len> characters are the same.
 564          * Now we need to order the next one, but turn
 565          * a '\0' into a '/' for a directory entry.
 566          */
 567         c1 = name1[len];
 568         c2 = name2[len];
 569         if (!c1 && !c2)
 570                 /*
 571                  * git-write-tree used to write out a nonsense tree that has
 572                  * entries with the same name, one blob and one tree.  Make
 573                  * sure we do not have duplicate entries.
 574                  */
 575                 return TREE_HAS_DUPS;
 576         if (!c1 && S_ISDIR(mode1))
 577                 c1 = '/';
 578         if (!c2 && S_ISDIR(mode2))
 579                 c2 = '/';
 580         return c1 < c2 ? 0 : TREE_UNORDERED;
 581 }
 582
 583 static int fsck_tree(struct tree *item, struct fsck_options *options)
 584 {
 585         int retval = 0;
 586         int has_null_sha1 = 0;
 587         int has_full_path = 0;
 588         int has_empty_name = 0;
 589         int has_dot = 0;
 590         int has_dotdot = 0;
 591         int has_dotgit = 0;
 592         int has_zero_pad = 0;
 593         int has_bad_modes = 0;
 594         int has_dup_entries = 0;
 595         int not_properly_sorted = 0;
 596         struct tree_desc desc;
 597         unsigned o_mode;
 598         const char *o_name;
 599
 600         if (init_tree_desc_gently(&desc, item->buffer, item->size)) {
 601                 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 602                 return retval;
 603         }
 604
 605         o_mode = 0;
 606         o_name = NULL;
 607
 608         while (desc.size) {
 609                 unsigned short mode;
 610                 const char *name, *backslash;
 611                 const struct object_id *oid;
 612
 613                 oid = tree_entry_extract(&desc, &name, &mode);
 614
 615                 has_null_sha1 |= is_null_oid(oid);
 616                 has_full_path |= !!strchr(name, '/');
 617                 has_empty_name |= !*name;
 618                 has_dot |= !strcmp(name, ".");
 619                 has_dotdot |= !strcmp(name, "..");
 620                 has_dotgit |= is_hfs_dotgit(name) || is_ntfs_dotgit(name);
 621                 has_zero_pad |= *(char *)desc.buffer == '0';
 622
 623                 if (is_hfs_dotgitmodules(name) || is_ntfs_dotgitmodules(name)) {
 624                         if (!S_ISLNK(mode))
 625                                 oidset_insert(&gitmodules_found, oid);
 626                         else
 627                                 retval += report(options, &item->object,
 628                                                  FSCK_MSG_GITMODULES_SYMLINK,
 629                                                  ".gitmodules is a symbolic link");
 630                 }
 631
 632                 if ((backslash = strchr(name, '\\'))) {
 633                         while (backslash) {
 634                                 backslash++;
 635                                 has_dotgit |= is_ntfs_dotgit(backslash);
 636                                 if (is_ntfs_dotgitmodules(backslash)) {
 637                                         if (!S_ISLNK(mode))
 638                                                 oidset_insert(&gitmodules_found, oid);
 639                                         else
 640                                                 retval += report(options, &item->object,
 641                                                                  FSCK_MSG_GITMODULES_SYMLINK,
 642                                                                  ".gitmodules is a symbolic link");
 643                                 }
 644                                 backslash = strchr(backslash, '\\');
 645                         }
 646                 }
 647
 648                 if (update_tree_entry_gently(&desc)) {
 649                         retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 650                         break;
 651                 }
 652
 653                 switch (mode) {
 654                 /*
 655                  * Standard modes..
 656                  */
 657                 case S_IFREG | 0755:
 658                 case S_IFREG | 0644:
 659                 case S_IFLNK:
 660                 case S_IFDIR:
 661                 case S_IFGITLINK:
 662                         break;
 663                 /*
 664                  * This is nonstandard, but we had a few of these
 665                  * early on when we honored the full set of mode
 666                  * bits..
 667                  */
 668                 case S_IFREG | 0664:
 669                         if (!options->strict)
 670                                 break;
 671                         /* fallthrough */
 672                 default:
 673                         has_bad_modes = 1;
 674                 }
 675
 676                 if (o_name) {
 677                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 678                         case TREE_UNORDERED:
 679                                 not_properly_sorted = 1;
 680                                 break;
 681                         case TREE_HAS_DUPS:
 682                                 has_dup_entries = 1;
 683                                 break;
 684                         default:
 685                                 break;
 686                         }
 687                 }
 688
 689                 o_mode = mode;
 690                 o_name = name;
 691         }
 692
 693         if (has_null_sha1)
 694                 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
 695         if (has_full_path)
 696                 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
 697         if (has_empty_name)
 698                 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
 699         if (has_dot)
 700                 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
 701         if (has_dotdot)
 702                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
 703         if (has_dotgit)
 704                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
 705         if (has_zero_pad)
 706                 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
 707         if (has_bad_modes)
 708                 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
 709         if (has_dup_entries)
 710                 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
 711         if (not_properly_sorted)
 712                 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
 713         return retval;
 714 }
 715
 716 static int verify_headers(const void *data, unsigned long size,
 717                           struct object *obj, struct fsck_options *options)
 718 {
 719         const char *buffer = (const char *)data;
 720         unsigned long i;
 721
 722         for (i = 0; i < size; i++) {
 723                 switch (buffer[i]) {
 724                 case '\0':
 725                         return report(options, obj,
 726                                 FSCK_MSG_NUL_IN_HEADER,
 727                                 "unterminated header: NUL at offset %ld", i);
 728                 case '\n':
 729                         if (i + 1 < size && buffer[i + 1] == '\n')
 730                                 return 0;
 731                 }
 732         }
 733
 734         /*
 735          * We did not find double-LF that separates the header
 736          * and the body.  Not having a body is not a crime but
 737          * we do want to see the terminating LF for the last header
 738          * line.
 739          */
 740         if (size && buffer[size - 1] == '\n')
 741                 return 0;
 742
 743         return report(options, obj,
 744                 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
 745 }
 746
 747 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
 748 {
 749         const char *p = *ident;
 750         char *end;
 751
 752         *ident = strchrnul(*ident, '\n');
 753         if (**ident == '\n')
 754                 (*ident)++;
 755
 756         if (*p == '<')
 757                 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 758         p += strcspn(p, "<>\n");
 759         if (*p == '>')
 760                 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
 761         if (*p != '<')
 762                 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
 763         if (p[-1] != ' ')
 764                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 765         p++;
 766         p += strcspn(p, "<>\n");
 767         if (*p != '>')
 768                 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
 769         p++;
 770         if (*p != ' ')
 771                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
 772         p++;
 773         if (*p == '0' && p[1] != ' ')
 774                 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
 775         if (date_overflows(parse_timestamp(p, &end, 10)))
 776                 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
 777         if ((end == p || *end != ' '))
 778                 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
 779         p = end + 1;
 780         if ((*p != '+' && *p != '-') ||
 781             !isdigit(p[1]) ||
 782             !isdigit(p[2]) ||
 783             !isdigit(p[3]) ||
 784             !isdigit(p[4]) ||
 785             (p[5] != '\n'))
 786                 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
 787         p += 6;
 788         return 0;
 789 }
 790
 791 static int fsck_commit_buffer(struct commit *commit, const char *buffer,
 792         unsigned long size, struct fsck_options *options)
 793 {
 794         struct object_id tree_oid, oid;
 795         struct commit_graft *graft;
 796         unsigned parent_count, parent_line_count = 0, author_count;
 797         int err;
 798         const char *buffer_begin = buffer;
 799         const char *p;
 800
 801         if (verify_headers(buffer, size, &commit->object, options))
 802                 return -1;
 803
 804         if (!skip_prefix(buffer, "tree ", &buffer))
 805                 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
 806         if (parse_oid_hex(buffer, &tree_oid, &p) || *p != '\n') {
 807                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
 808                 if (err)
 809                         return err;
 810         }
 811         buffer = p + 1;
 812         while (skip_prefix(buffer, "parent ", &buffer)) {
 813                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 814                         err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
 815                         if (err)
 816                                 return err;
 817                 }
 818                 buffer = p + 1;
 819                 parent_line_count++;
 820         }
 821         graft = lookup_commit_graft(the_repository, &commit->object.oid);
 822         parent_count = commit_list_count(commit->parents);
 823         if (graft) {
 824                 if (graft->nr_parent == -1 && !parent_count)
 825                         ; /* shallow commit */
 826                 else if (graft->nr_parent != parent_count) {
 827                         err = report(options, &commit->object, FSCK_MSG_MISSING_GRAFT, "graft objects missing");
 828                         if (err)
 829                                 return err;
 830                 }
 831         } else {
 832                 if (parent_count != parent_line_count) {
 833                         err = report(options, &commit->object, FSCK_MSG_MISSING_PARENT, "parent objects missing");
 834                         if (err)
 835                                 return err;
 836                 }
 837         }
 838         author_count = 0;
 839         while (skip_prefix(buffer, "author ", &buffer)) {
 840                 author_count++;
 841                 err = fsck_ident(&buffer, &commit->object, options);
 842                 if (err)
 843                         return err;
 844         }
 845         if (author_count < 1)
 846                 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
 847         else if (author_count > 1)
 848                 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
 849         if (err)
 850                 return err;
 851         if (!skip_prefix(buffer, "committer ", &buffer))
 852                 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
 853         err = fsck_ident(&buffer, &commit->object, options);
 854         if (err)
 855                 return err;
 856         if (!get_commit_tree(commit)) {
 857                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE, "could not load commit's tree %s", oid_to_hex(&tree_oid));
 858                 if (err)
 859                         return err;
 860         }
 861         if (memchr(buffer_begin, '\0', size)) {
 862                 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
 863                              "NUL byte in the commit object body");
 864                 if (err)
 865                         return err;
 866         }
 867         return 0;
 868 }
 869
 870 static int fsck_commit(struct commit *commit, const char *data,
 871         unsigned long size, struct fsck_options *options)
 872 {
 873         const char *buffer = data ?  data : get_commit_buffer(commit, &size);
 874         int ret = fsck_commit_buffer(commit, buffer, size, options);
 875         if (!data)
 876                 unuse_commit_buffer(commit, buffer);
 877         return ret;
 878 }
 879
 880 static int fsck_tag_buffer(struct tag *tag, const char *data,
 881         unsigned long size, struct fsck_options *options)
 882 {
 883         struct object_id oid;
 884         int ret = 0;
 885         const char *buffer;
 886         char *to_free = NULL, *eol;
 887         struct strbuf sb = STRBUF_INIT;
 888         const char *p;
 889
 890         if (data)
 891                 buffer = data;
 892         else {
 893                 enum object_type type;
 894
 895                 buffer = to_free =
 896                         read_object_file(&tag->object.oid, &type, &size);
 897                 if (!buffer)
 898                         return report(options, &tag->object,
 899                                 FSCK_MSG_MISSING_TAG_OBJECT,
 900                                 "cannot read tag object");
 901
 902                 if (type != OBJ_TAG) {
 903                         ret = report(options, &tag->object,
 904                                 FSCK_MSG_TAG_OBJECT_NOT_TAG,
 905                                 "expected tag got %s",
 906                             type_name(type));
 907                         goto done;
 908                 }
 909         }
 910
 911         ret = verify_headers(buffer, size, &tag->object, options);
 912         if (ret)
 913                 goto done;
 914
 915         if (!skip_prefix(buffer, "object ", &buffer)) {
 916                 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
 917                 goto done;
 918         }
 919         if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 920                 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
 921                 if (ret)
 922                         goto done;
 923         }
 924         buffer = p + 1;
 925
 926         if (!skip_prefix(buffer, "type ", &buffer)) {
 927                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
 928                 goto done;
 929         }
 930         eol = strchr(buffer, '\n');
 931         if (!eol) {
 932                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
 933                 goto done;
 934         }
 935         if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
 936                 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
 937         if (ret)
 938                 goto done;
 939         buffer = eol + 1;
 940
 941         if (!skip_prefix(buffer, "tag ", &buffer)) {
 942                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
 943                 goto done;
 944         }
 945         eol = strchr(buffer, '\n');
 946         if (!eol) {
 947                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
 948                 goto done;
 949         }
 950         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
 951         if (check_refname_format(sb.buf, 0)) {
 952                 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
 953                            "invalid 'tag' name: %.*s",
 954                            (int)(eol - buffer), buffer);
 955                 if (ret)
 956                         goto done;
 957         }
 958         buffer = eol + 1;
 959
 960         if (!skip_prefix(buffer, "tagger ", &buffer)) {
 961                 /* early tags do not contain 'tagger' lines; warn only */
 962                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
 963                 if (ret)
 964                         goto done;
 965         }
 966         else
 967                 ret = fsck_ident(&buffer, &tag->object, options);
 968
 969 done:
 970         strbuf_release(&sb);
 971         free(to_free);
 972         return ret;
 973 }
 974
 975 static int fsck_tag(struct tag *tag, const char *data,
 976         unsigned long size, struct fsck_options *options)
 977 {
 978         struct object *tagged = tag->tagged;
 979
 980         if (!tagged)
 981                 return report(options, &tag->object, FSCK_MSG_BAD_TAG_OBJECT, "could not load tagged object");
 982
 983         return fsck_tag_buffer(tag, data, size, options);
 984 }
 985
 986 static int check_submodule_url(const char *url)
 987 {
 988         struct credential c = CREDENTIAL_INIT;
 989         int ret;
 990
 991         if (looks_like_command_line_option(url))
 992                 return -1;
 993
 994         ret = credential_from_url_gently(&c, url, 1);
 995         credential_clear(&c);
 996         return ret;
 997 }
 998
 999 struct fsck_gitmodules_data {
1000         struct object *obj;
1001         struct fsck_options *options;
1002         int ret;
1003 };
1004
1005 static int fsck_gitmodules_fn(const char *var, const char *value, void *vdata)
1006 {
1007         struct fsck_gitmodules_data *data = vdata;
1008         const char *subsection, *key;
1009         int subsection_len;
1010         char *name;
1011
1012         if (parse_config_key(var, "submodule", &subsection, &subsection_len, &key) < 0 ||
1013             !subsection)
1014                 return 0;
1015
1016         name = xmemdupz(subsection, subsection_len);
1017         if (check_submodule_name(name) < 0)
1018                 data->ret |= report(data->options, data->obj,
1019                                     FSCK_MSG_GITMODULES_NAME,
1020                                     "disallowed submodule name: %s",
1021                                     name);
1022         if (!strcmp(key, "url") && value &&
1023             check_submodule_url(value) < 0)
1024                 data->ret |= report(data->options, data->obj,
1025                                     FSCK_MSG_GITMODULES_URL,
1026                                     "disallowed submodule url: %s",
1027                                     value);
1028         if (!strcmp(key, "path") && value &&
1029             looks_like_command_line_option(value))
1030                 data->ret |= report(data->options, data->obj,
1031                                     FSCK_MSG_GITMODULES_PATH,
1032                                     "disallowed submodule path: %s",
1033                                     value);
1034         if (!strcmp(key, "update") && value &&
1035             parse_submodule_update_type(value) == SM_UPDATE_COMMAND)
1036                 data->ret |= report(data->options, data->obj,
1037                                     FSCK_MSG_GITMODULES_UPDATE,
1038                                     "disallowed submodule update setting: %s",
1039                                     value);
1040         free(name);
1041
1042         return 0;
1043 }
1044
1045 static int fsck_blob(struct blob *blob, const char *buf,
1046                      unsigned long size, struct fsck_options *options)
1047 {
1048         struct fsck_gitmodules_data data;
1049         struct config_options config_opts = { 0 };
1050
1051         if (!oidset_contains(&gitmodules_found, &blob->object.oid))
1052                 return 0;
1053         oidset_insert(&gitmodules_done, &blob->object.oid);
1054
1055         if (object_on_skiplist(options, &blob->object))
1056                 return 0;
1057
1058         if (!buf) {
1059                 /*
1060                  * A missing buffer here is a sign that the caller found the
1061                  * blob too gigantic to load into memory. Let's just consider
1062                  * that an error.
1063                  */
1064                 return report(options, &blob->object,
1065                               FSCK_MSG_GITMODULES_LARGE,
1066                               ".gitmodules too large to parse");
1067         }
1068
1069         data.obj = &blob->object;
1070         data.options = options;
1071         data.ret = 0;
1072         config_opts.error_action = CONFIG_ERROR_SILENT;
1073         if (git_config_from_mem(fsck_gitmodules_fn, CONFIG_ORIGIN_BLOB,
1074                                 ".gitmodules", buf, size, &data, &config_opts))
1075                 data.ret |= report(options, &blob->object,
1076                                    FSCK_MSG_GITMODULES_PARSE,
1077                                    "could not parse gitmodules blob");
1078
1079         return data.ret;
1080 }
1081
1082 int fsck_object(struct object *obj, void *data, unsigned long size,
1083         struct fsck_options *options)
1084 {
1085         if (!obj)
1086                 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
1087
1088         if (obj->type == OBJ_BLOB)
1089                 return fsck_blob((struct blob *)obj, data, size, options);
1090         if (obj->type == OBJ_TREE)
1091                 return fsck_tree((struct tree *) obj, options);
1092         if (obj->type == OBJ_COMMIT)
1093                 return fsck_commit((struct commit *) obj, (const char *) data,
1094                         size, options);
1095         if (obj->type == OBJ_TAG)
1096                 return fsck_tag((struct tag *) obj, (const char *) data,
1097                         size, options);
1098
1099         return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
1100                           obj->type);
1101 }
1102
1103 int fsck_error_function(struct fsck_options *o,
1104         struct object *obj, int msg_type, const char *message)
1105 {
1106         if (msg_type == FSCK_WARN) {
1107                 warning("object %s: %s", describe_object(o, obj), message);
1108                 return 0;
1109         }
1110         error("object %s: %s", describe_object(o, obj), message);
1111         return 1;
1112 }
1113
1114 int fsck_finish(struct fsck_options *options)
1115 {
1116         int ret = 0;
1117         struct oidset_iter iter;
1118         const struct object_id *oid;
1119
1120         oidset_iter_init(&gitmodules_found, &iter);
1121         while ((oid = oidset_iter_next(&iter))) {
1122                 struct blob *blob;
1123                 enum object_type type;
1124                 unsigned long size;
1125                 char *buf;
1126
1127                 if (oidset_contains(&gitmodules_done, oid))
1128                         continue;
1129
1130                 blob = lookup_blob(the_repository, oid);
1131                 if (!blob) {
1132                         struct object *obj = lookup_unknown_object(oid->hash);
1133                         ret |= report(options, obj,
1134                                       FSCK_MSG_GITMODULES_BLOB,
1135                                       "non-blob found at .gitmodules");
1136                         continue;
1137                 }
1138
1139                 buf = read_object_file(oid, &type, &size);
1140                 if (!buf) {
1141                         if (is_promisor_object(&blob->object.oid))
1142                                 continue;
1143                         ret |= report(options, &blob->object,
1144                                       FSCK_MSG_GITMODULES_MISSING,
1145                                       "unable to read .gitmodules blob");
1146                         continue;
1147                 }
1148
1149                 if (type == OBJ_BLOB)
1150                         ret |= fsck_blob(blob, buf, size, options);
1151                 else
1152                         ret |= report(options, &blob->object,
1153                                       FSCK_MSG_GITMODULES_BLOB,
1154                                       "non-blob found at .gitmodules");
1155                 free(buf);
1156         }
1157
1158
1159         oidset_clear(&gitmodules_found);
1160         oidset_clear(&gitmodules_done);
1161         return ret;
1162 }