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 int parse_msg_type(const char *str)
 187 {
 188         if (!strcmp(str, "error"))
 189                 return FSCK_ERROR;
 190         else if (!strcmp(str, "warn"))
 191                 return FSCK_WARN;
 192         else if (!strcmp(str, "ignore"))
 193                 return FSCK_IGNORE;
 194         else
 195                 die("Unknown fsck message type: '%s'", str);
 196 }
 197
 198 int is_valid_msg_type(const char *msg_id, const char *msg_type)
 199 {
 200         if (parse_msg_id(msg_id) < 0)
 201                 return 0;
 202         parse_msg_type(msg_type);
 203         return 1;
 204 }
 205
 206 void fsck_set_msg_type(struct fsck_options *options,
 207                 const char *msg_id, const char *msg_type)
 208 {
 209         int id = parse_msg_id(msg_id), type;
 210
 211         if (id < 0)
 212                 die("Unhandled message id: %s", msg_id);
 213         type = parse_msg_type(msg_type);
 214
 215         if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
 216                 die("Cannot demote %s to %s", msg_id, msg_type);
 217
 218         if (!options->msg_type) {
 219                 int i;
 220                 int *msg_type;
 221                 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
 222                 for (i = 0; i < FSCK_MSG_MAX; i++)
 223                         msg_type[i] = fsck_msg_type(i, options);
 224                 options->msg_type = msg_type;
 225         }
 226
 227         options->msg_type[id] = type;
 228 }
 229
 230 void fsck_set_msg_types(struct fsck_options *options, const char *values)
 231 {
 232         char *buf = xstrdup(values), *to_free = buf;
 233         int done = 0;
 234
 235         while (!done) {
 236                 int len = strcspn(buf, " ,|"), equal;
 237
 238                 done = !buf[len];
 239                 if (!len) {
 240                         buf++;
 241                         continue;
 242                 }
 243                 buf[len] = '\0';
 244
 245                 for (equal = 0;
 246                      equal < len && buf[equal] != '=' && buf[equal] != ':';
 247                      equal++)
 248                         buf[equal] = tolower(buf[equal]);
 249                 buf[equal] = '\0';
 250
 251                 if (!strcmp(buf, "skiplist")) {
 252                         if (equal == len)
 253                                 die("skiplist requires a path");
 254                         oidset_parse_file(&options->skiplist, buf + equal + 1);
 255                         buf += len + 1;
 256                         continue;
 257                 }
 258
 259                 if (equal == len)
 260                         die("Missing '=': '%s'", buf);
 261
 262                 fsck_set_msg_type(options, buf, buf + equal + 1);
 263                 buf += len + 1;
 264         }
 265         free(to_free);
 266 }
 267
 268 static void append_msg_id(struct strbuf *sb, const char *msg_id)
 269 {
 270         for (;;) {
 271                 char c = *(msg_id)++;
 272
 273                 if (!c)
 274                         break;
 275                 if (c != '_')
 276                         strbuf_addch(sb, tolower(c));
 277                 else {
 278                         assert(*msg_id);
 279                         strbuf_addch(sb, *(msg_id)++);
 280                 }
 281         }
 282
 283         strbuf_addstr(sb, ": ");
 284 }
 285
 286 static int object_on_skiplist(struct fsck_options *opts, struct object *obj)
 287 {
 288         return opts && obj && oidset_contains(&opts->skiplist, &obj->oid);
 289 }
 290
 291 __attribute__((format (printf, 4, 5)))
 292 static int report(struct fsck_options *options, struct object *object,
 293         enum fsck_msg_id id, const char *fmt, ...)
 294 {
 295         va_list ap;
 296         struct strbuf sb = STRBUF_INIT;
 297         int msg_type = fsck_msg_type(id, options), result;
 298
 299         if (msg_type == FSCK_IGNORE)
 300                 return 0;
 301
 302         if (object_on_skiplist(options, object))
 303                 return 0;
 304
 305         if (msg_type == FSCK_FATAL)
 306                 msg_type = FSCK_ERROR;
 307         else if (msg_type == FSCK_INFO)
 308                 msg_type = FSCK_WARN;
 309
 310         append_msg_id(&sb, msg_id_info[id].id_string);
 311
 312         va_start(ap, fmt);
 313         strbuf_vaddf(&sb, fmt, ap);
 314         result = options->error_func(options, object, msg_type, sb.buf);
 315         strbuf_release(&sb);
 316         va_end(ap);
 317
 318         return result;
 319 }
 320
 321 static char *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 static void 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 static const char *describe_object(struct fsck_options *o, struct object *obj)
 347 {
 348         static struct strbuf buf = STRBUF_INIT;
 349         char *name;
 350
 351         strbuf_reset(&buf);
 352         strbuf_addstr(&buf, oid_to_hex(&obj->oid));
 353         if (o->object_names && (name = lookup_decoration(o->object_names, obj)))
 354                 strbuf_addf(&buf, " (%s)", name);
 355
 356         return buf.buf;
 357 }
 358
 359 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
 360 {
 361         struct tree_desc desc;
 362         struct name_entry entry;
 363         int res = 0;
 364         const char *name;
 365
 366         if (parse_tree(tree))
 367                 return -1;
 368
 369         name = get_object_name(options, &tree->object);
 370         if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
 371                 return -1;
 372         while (tree_entry_gently(&desc, &entry)) {
 373                 struct object *obj;
 374                 int result;
 375
 376                 if (S_ISGITLINK(entry.mode))
 377                         continue;
 378
 379                 if (S_ISDIR(entry.mode)) {
 380                         obj = (struct object *)lookup_tree(the_repository, &entry.oid);
 381                         if (name && obj)
 382                                 put_object_name(options, obj, "%s%s/", name,
 383                                         entry.path);
 384                         result = options->walk(obj, OBJ_TREE, data, options);
 385                 }
 386                 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
 387                         obj = (struct object *)lookup_blob(the_repository, &entry.oid);
 388                         if (name && obj)
 389                                 put_object_name(options, obj, "%s%s", name,
 390                                         entry.path);
 391                         result = options->walk(obj, OBJ_BLOB, data, options);
 392                 }
 393                 else {
 394                         result = error("in tree %s: entry %s has bad mode %.6o",
 395                                         describe_object(options, &tree->object), entry.path, entry.mode);
 396                 }
 397                 if (result < 0)
 398                         return result;
 399                 if (!res)
 400                         res = result;
 401         }
 402         return res;
 403 }
 404
 405 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
 406 {
 407         int counter = 0, generation = 0, name_prefix_len = 0;
 408         struct commit_list *parents;
 409         int res;
 410         int result;
 411         const char *name;
 412
 413         if (parse_commit(commit))
 414                 return -1;
 415
 416         name = get_object_name(options, &commit->object);
 417         if (name)
 418                 put_object_name(options, &get_commit_tree(commit)->object,
 419                                 "%s:", name);
 420
 421         result = options->walk((struct object *)get_commit_tree(commit),
 422                                OBJ_TREE, data, options);
 423         if (result < 0)
 424                 return result;
 425         res = result;
 426
 427         parents = commit->parents;
 428         if (name && parents) {
 429                 int len = strlen(name), power;
 430
 431                 if (len && name[len - 1] == '^') {
 432                         generation = 1;
 433                         name_prefix_len = len - 1;
 434                 }
 435                 else { /* parse ~<generation> suffix */
 436                         for (generation = 0, power = 1;
 437                              len && isdigit(name[len - 1]);
 438                              power *= 10)
 439                                 generation += power * (name[--len] - '0');
 440                         if (power > 1 && len && name[len - 1] == '~')
 441                                 name_prefix_len = len - 1;
 442                 }
 443         }
 444
 445         while (parents) {
 446                 if (name) {
 447                         struct object *obj = &parents->item->object;
 448
 449                         if (counter++)
 450                                 put_object_name(options, obj, "%s^%d",
 451                                         name, counter);
 452                         else if (generation > 0)
 453                                 put_object_name(options, obj, "%.*s~%d",
 454                                         name_prefix_len, name, generation + 1);
 455                         else
 456                                 put_object_name(options, obj, "%s^", name);
 457                 }
 458                 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
 459                 if (result < 0)
 460                         return result;
 461                 if (!res)
 462                         res = result;
 463                 parents = parents->next;
 464         }
 465         return res;
 466 }
 467
 468 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
 469 {
 470         char *name = get_object_name(options, &tag->object);
 471
 472         if (parse_tag(tag))
 473                 return -1;
 474         if (name)
 475                 put_object_name(options, tag->tagged, "%s", name);
 476         return options->walk(tag->tagged, OBJ_ANY, data, options);
 477 }
 478
 479 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
 480 {
 481         if (!obj)
 482                 return -1;
 483
 484         if (obj->type == OBJ_NONE)
 485                 parse_object(the_repository, &obj->oid);
 486
 487         switch (obj->type) {
 488         case OBJ_BLOB:
 489                 return 0;
 490         case OBJ_TREE:
 491                 return fsck_walk_tree((struct tree *)obj, data, options);
 492         case OBJ_COMMIT:
 493                 return fsck_walk_commit((struct commit *)obj, data, options);
 494         case OBJ_TAG:
 495                 return fsck_walk_tag((struct tag *)obj, data, options);
 496         default:
 497                 error("Unknown object type for %s", describe_object(options, obj));
 498                 return -1;
 499         }
 500 }
 501
 502 /*
 503  * The entries in a tree are ordered in the _path_ order,
 504  * which means that a directory entry is ordered by adding
 505  * a slash to the end of it.
 506  *
 507  * So a directory called "a" is ordered _after_ a file
 508  * called "a.c", because "a/" sorts after "a.c".
 509  */
 510 #define TREE_UNORDERED (-1)
 511 #define TREE_HAS_DUPS  (-2)
 512
 513 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 514 {
 515         int len1 = strlen(name1);
 516         int len2 = strlen(name2);
 517         int len = len1 < len2 ? len1 : len2;
 518         unsigned char c1, c2;
 519         int cmp;
 520
 521         cmp = memcmp(name1, name2, len);
 522         if (cmp < 0)
 523                 return 0;
 524         if (cmp > 0)
 525                 return TREE_UNORDERED;
 526
 527         /*
 528          * Ok, the first <len> characters are the same.
 529          * Now we need to order the next one, but turn
 530          * a '\0' into a '/' for a directory entry.
 531          */
 532         c1 = name1[len];
 533         c2 = name2[len];
 534         if (!c1 && !c2)
 535                 /*
 536                  * git-write-tree used to write out a nonsense tree that has
 537                  * entries with the same name, one blob and one tree.  Make
 538                  * sure we do not have duplicate entries.
 539                  */
 540                 return TREE_HAS_DUPS;
 541         if (!c1 && S_ISDIR(mode1))
 542                 c1 = '/';
 543         if (!c2 && S_ISDIR(mode2))
 544                 c2 = '/';
 545         return c1 < c2 ? 0 : TREE_UNORDERED;
 546 }
 547
 548 static int fsck_tree(struct tree *item, struct fsck_options *options)
 549 {
 550         int retval = 0;
 551         int has_null_sha1 = 0;
 552         int has_full_path = 0;
 553         int has_empty_name = 0;
 554         int has_dot = 0;
 555         int has_dotdot = 0;
 556         int has_dotgit = 0;
 557         int has_zero_pad = 0;
 558         int has_bad_modes = 0;
 559         int has_dup_entries = 0;
 560         int not_properly_sorted = 0;
 561         struct tree_desc desc;
 562         unsigned o_mode;
 563         const char *o_name;
 564
 565         if (init_tree_desc_gently(&desc, item->buffer, item->size)) {
 566                 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 567                 return retval;
 568         }
 569
 570         o_mode = 0;
 571         o_name = NULL;
 572
 573         while (desc.size) {
 574                 unsigned short mode;
 575                 const char *name, *backslash;
 576                 const struct object_id *oid;
 577
 578                 oid = tree_entry_extract(&desc, &name, &mode);
 579
 580                 has_null_sha1 |= is_null_oid(oid);
 581                 has_full_path |= !!strchr(name, '/');
 582                 has_empty_name |= !*name;
 583                 has_dot |= !strcmp(name, ".");
 584                 has_dotdot |= !strcmp(name, "..");
 585                 has_dotgit |= is_hfs_dotgit(name) || is_ntfs_dotgit(name);
 586                 has_zero_pad |= *(char *)desc.buffer == '0';
 587
 588                 if (is_hfs_dotgitmodules(name) || is_ntfs_dotgitmodules(name)) {
 589                         if (!S_ISLNK(mode))
 590                                 oidset_insert(&gitmodules_found, oid);
 591                         else
 592                                 retval += report(options, &item->object,
 593                                                  FSCK_MSG_GITMODULES_SYMLINK,
 594                                                  ".gitmodules is a symbolic link");
 595                 }
 596
 597                 if ((backslash = strchr(name, '\\'))) {
 598                         while (backslash) {
 599                                 backslash++;
 600                                 has_dotgit |= is_ntfs_dotgit(backslash);
 601                                 if (is_ntfs_dotgitmodules(backslash)) {
 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                                 backslash = strchr(backslash, '\\');
 610                         }
 611                 }
 612
 613                 if (update_tree_entry_gently(&desc)) {
 614                         retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 615                         break;
 616                 }
 617
 618                 switch (mode) {
 619                 /*
 620                  * Standard modes..
 621                  */
 622                 case S_IFREG | 0755:
 623                 case S_IFREG | 0644:
 624                 case S_IFLNK:
 625                 case S_IFDIR:
 626                 case S_IFGITLINK:
 627                         break;
 628                 /*
 629                  * This is nonstandard, but we had a few of these
 630                  * early on when we honored the full set of mode
 631                  * bits..
 632                  */
 633                 case S_IFREG | 0664:
 634                         if (!options->strict)
 635                                 break;
 636                         /* fallthrough */
 637                 default:
 638                         has_bad_modes = 1;
 639                 }
 640
 641                 if (o_name) {
 642                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 643                         case TREE_UNORDERED:
 644                                 not_properly_sorted = 1;
 645                                 break;
 646                         case TREE_HAS_DUPS:
 647                                 has_dup_entries = 1;
 648                                 break;
 649                         default:
 650                                 break;
 651                         }
 652                 }
 653
 654                 o_mode = mode;
 655                 o_name = name;
 656         }
 657
 658         if (has_null_sha1)
 659                 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
 660         if (has_full_path)
 661                 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
 662         if (has_empty_name)
 663                 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
 664         if (has_dot)
 665                 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
 666         if (has_dotdot)
 667                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
 668         if (has_dotgit)
 669                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
 670         if (has_zero_pad)
 671                 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
 672         if (has_bad_modes)
 673                 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
 674         if (has_dup_entries)
 675                 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
 676         if (not_properly_sorted)
 677                 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
 678         return retval;
 679 }
 680
 681 static int verify_headers(const void *data, unsigned long size,
 682                           struct object *obj, struct fsck_options *options)
 683 {
 684         const char *buffer = (const char *)data;
 685         unsigned long i;
 686
 687         for (i = 0; i < size; i++) {
 688                 switch (buffer[i]) {
 689                 case '\0':
 690                         return report(options, obj,
 691                                 FSCK_MSG_NUL_IN_HEADER,
 692                                 "unterminated header: NUL at offset %ld", i);
 693                 case '\n':
 694                         if (i + 1 < size && buffer[i + 1] == '\n')
 695                                 return 0;
 696                 }
 697         }
 698
 699         /*
 700          * We did not find double-LF that separates the header
 701          * and the body.  Not having a body is not a crime but
 702          * we do want to see the terminating LF for the last header
 703          * line.
 704          */
 705         if (size && buffer[size - 1] == '\n')
 706                 return 0;
 707
 708         return report(options, obj,
 709                 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
 710 }
 711
 712 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
 713 {
 714         const char *p = *ident;
 715         char *end;
 716
 717         *ident = strchrnul(*ident, '\n');
 718         if (**ident == '\n')
 719                 (*ident)++;
 720
 721         if (*p == '<')
 722                 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 723         p += strcspn(p, "<>\n");
 724         if (*p == '>')
 725                 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
 726         if (*p != '<')
 727                 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
 728         if (p[-1] != ' ')
 729                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 730         p++;
 731         p += strcspn(p, "<>\n");
 732         if (*p != '>')
 733                 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
 734         p++;
 735         if (*p != ' ')
 736                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
 737         p++;
 738         if (*p == '0' && p[1] != ' ')
 739                 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
 740         if (date_overflows(parse_timestamp(p, &end, 10)))
 741                 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
 742         if ((end == p || *end != ' '))
 743                 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
 744         p = end + 1;
 745         if ((*p != '+' && *p != '-') ||
 746             !isdigit(p[1]) ||
 747             !isdigit(p[2]) ||
 748             !isdigit(p[3]) ||
 749             !isdigit(p[4]) ||
 750             (p[5] != '\n'))
 751                 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
 752         p += 6;
 753         return 0;
 754 }
 755
 756 static int fsck_commit_buffer(struct commit *commit, const char *buffer,
 757         unsigned long size, struct fsck_options *options)
 758 {
 759         struct object_id tree_oid, oid;
 760         struct commit_graft *graft;
 761         unsigned parent_count, parent_line_count = 0, author_count;
 762         int err;
 763         const char *buffer_begin = buffer;
 764         const char *p;
 765
 766         if (verify_headers(buffer, size, &commit->object, options))
 767                 return -1;
 768
 769         if (!skip_prefix(buffer, "tree ", &buffer))
 770                 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
 771         if (parse_oid_hex(buffer, &tree_oid, &p) || *p != '\n') {
 772                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
 773                 if (err)
 774                         return err;
 775         }
 776         buffer = p + 1;
 777         while (skip_prefix(buffer, "parent ", &buffer)) {
 778                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 779                         err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
 780                         if (err)
 781                                 return err;
 782                 }
 783                 buffer = p + 1;
 784                 parent_line_count++;
 785         }
 786         graft = lookup_commit_graft(the_repository, &commit->object.oid);
 787         parent_count = commit_list_count(commit->parents);
 788         if (graft) {
 789                 if (graft->nr_parent == -1 && !parent_count)
 790                         ; /* shallow commit */
 791                 else if (graft->nr_parent != parent_count) {
 792                         err = report(options, &commit->object, FSCK_MSG_MISSING_GRAFT, "graft objects missing");
 793                         if (err)
 794                                 return err;
 795                 }
 796         } else {
 797                 if (parent_count != parent_line_count) {
 798                         err = report(options, &commit->object, FSCK_MSG_MISSING_PARENT, "parent objects missing");
 799                         if (err)
 800                                 return err;
 801                 }
 802         }
 803         author_count = 0;
 804         while (skip_prefix(buffer, "author ", &buffer)) {
 805                 author_count++;
 806                 err = fsck_ident(&buffer, &commit->object, options);
 807                 if (err)
 808                         return err;
 809         }
 810         if (author_count < 1)
 811                 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
 812         else if (author_count > 1)
 813                 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
 814         if (err)
 815                 return err;
 816         if (!skip_prefix(buffer, "committer ", &buffer))
 817                 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
 818         err = fsck_ident(&buffer, &commit->object, options);
 819         if (err)
 820                 return err;
 821         if (!get_commit_tree(commit)) {
 822                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE, "could not load commit's tree %s", oid_to_hex(&tree_oid));
 823                 if (err)
 824                         return err;
 825         }
 826         if (memchr(buffer_begin, '\0', size)) {
 827                 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
 828                              "NUL byte in the commit object body");
 829                 if (err)
 830                         return err;
 831         }
 832         return 0;
 833 }
 834
 835 static int fsck_commit(struct commit *commit, const char *data,
 836         unsigned long size, struct fsck_options *options)
 837 {
 838         const char *buffer = data ?  data : get_commit_buffer(commit, &size);
 839         int ret = fsck_commit_buffer(commit, buffer, size, options);
 840         if (!data)
 841                 unuse_commit_buffer(commit, buffer);
 842         return ret;
 843 }
 844
 845 static int fsck_tag_buffer(struct tag *tag, const char *data,
 846         unsigned long size, struct fsck_options *options)
 847 {
 848         struct object_id oid;
 849         int ret = 0;
 850         const char *buffer;
 851         char *to_free = NULL, *eol;
 852         struct strbuf sb = STRBUF_INIT;
 853         const char *p;
 854
 855         if (data)
 856                 buffer = data;
 857         else {
 858                 enum object_type type;
 859
 860                 buffer = to_free =
 861                         read_object_file(&tag->object.oid, &type, &size);
 862                 if (!buffer)
 863                         return report(options, &tag->object,
 864                                 FSCK_MSG_MISSING_TAG_OBJECT,
 865                                 "cannot read tag object");
 866
 867                 if (type != OBJ_TAG) {
 868                         ret = report(options, &tag->object,
 869                                 FSCK_MSG_TAG_OBJECT_NOT_TAG,
 870                                 "expected tag got %s",
 871                             type_name(type));
 872                         goto done;
 873                 }
 874         }
 875
 876         ret = verify_headers(buffer, size, &tag->object, options);
 877         if (ret)
 878                 goto done;
 879
 880         if (!skip_prefix(buffer, "object ", &buffer)) {
 881                 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
 882                 goto done;
 883         }
 884         if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 885                 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
 886                 if (ret)
 887                         goto done;
 888         }
 889         buffer = p + 1;
 890
 891         if (!skip_prefix(buffer, "type ", &buffer)) {
 892                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
 893                 goto done;
 894         }
 895         eol = strchr(buffer, '\n');
 896         if (!eol) {
 897                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
 898                 goto done;
 899         }
 900         if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
 901                 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
 902         if (ret)
 903                 goto done;
 904         buffer = eol + 1;
 905
 906         if (!skip_prefix(buffer, "tag ", &buffer)) {
 907                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
 908                 goto done;
 909         }
 910         eol = strchr(buffer, '\n');
 911         if (!eol) {
 912                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
 913                 goto done;
 914         }
 915         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
 916         if (check_refname_format(sb.buf, 0)) {
 917                 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
 918                            "invalid 'tag' name: %.*s",
 919                            (int)(eol - buffer), buffer);
 920                 if (ret)
 921                         goto done;
 922         }
 923         buffer = eol + 1;
 924
 925         if (!skip_prefix(buffer, "tagger ", &buffer)) {
 926                 /* early tags do not contain 'tagger' lines; warn only */
 927                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
 928                 if (ret)
 929                         goto done;
 930         }
 931         else
 932                 ret = fsck_ident(&buffer, &tag->object, options);
 933
 934 done:
 935         strbuf_release(&sb);
 936         free(to_free);
 937         return ret;
 938 }
 939
 940 static int fsck_tag(struct tag *tag, const char *data,
 941         unsigned long size, struct fsck_options *options)
 942 {
 943         struct object *tagged = tag->tagged;
 944
 945         if (!tagged)
 946                 return report(options, &tag->object, FSCK_MSG_BAD_TAG_OBJECT, "could not load tagged object");
 947
 948         return fsck_tag_buffer(tag, data, size, options);
 949 }
 950
 951 static int check_submodule_url(const char *url)
 952 {
 953         struct credential c = CREDENTIAL_INIT;
 954         int ret;
 955
 956         if (looks_like_command_line_option(url))
 957                 return -1;
 958
 959         ret = credential_from_url_gently(&c, url, 1);
 960         credential_clear(&c);
 961         return ret;
 962 }
 963
 964 struct fsck_gitmodules_data {
 965         struct object *obj;
 966         struct fsck_options *options;
 967         int ret;
 968 };
 969
 970 static int fsck_gitmodules_fn(const char *var, const char *value, void *vdata)
 971 {
 972         struct fsck_gitmodules_data *data = vdata;
 973         const char *subsection, *key;
 974         int subsection_len;
 975         char *name;
 976
 977         if (parse_config_key(var, "submodule", &subsection, &subsection_len, &key) < 0 ||
 978             !subsection)
 979                 return 0;
 980
 981         name = xmemdupz(subsection, subsection_len);
 982         if (check_submodule_name(name) < 0)
 983                 data->ret |= report(data->options, data->obj,
 984                                     FSCK_MSG_GITMODULES_NAME,
 985                                     "disallowed submodule name: %s",
 986                                     name);
 987         if (!strcmp(key, "url") && value &&
 988             check_submodule_url(value) < 0)
 989                 data->ret |= report(data->options, data->obj,
 990                                     FSCK_MSG_GITMODULES_URL,
 991                                     "disallowed submodule url: %s",
 992                                     value);
 993         if (!strcmp(key, "path") && value &&
 994             looks_like_command_line_option(value))
 995                 data->ret |= report(data->options, data->obj,
 996                                     FSCK_MSG_GITMODULES_PATH,
 997                                     "disallowed submodule path: %s",
 998                                     value);
 999         if (!strcmp(key, "update") && value &&
1000             parse_submodule_update_type(value) == SM_UPDATE_COMMAND)
1001                 data->ret |= report(data->options, data->obj,
1002                                     FSCK_MSG_GITMODULES_UPDATE,
1003                                     "disallowed submodule update setting: %s",
1004                                     value);
1005         free(name);
1006
1007         return 0;
1008 }
1009
1010 static int fsck_blob(struct blob *blob, const char *buf,
1011                      unsigned long size, struct fsck_options *options)
1012 {
1013         struct fsck_gitmodules_data data;
1014         struct config_options config_opts = { 0 };
1015
1016         if (!oidset_contains(&gitmodules_found, &blob->object.oid))
1017                 return 0;
1018         oidset_insert(&gitmodules_done, &blob->object.oid);
1019
1020         if (object_on_skiplist(options, &blob->object))
1021                 return 0;
1022
1023         if (!buf) {
1024                 /*
1025                  * A missing buffer here is a sign that the caller found the
1026                  * blob too gigantic to load into memory. Let's just consider
1027                  * that an error.
1028                  */
1029                 return report(options, &blob->object,
1030                               FSCK_MSG_GITMODULES_LARGE,
1031                               ".gitmodules too large to parse");
1032         }
1033
1034         data.obj = &blob->object;
1035         data.options = options;
1036         data.ret = 0;
1037         config_opts.error_action = CONFIG_ERROR_SILENT;
1038         if (git_config_from_mem(fsck_gitmodules_fn, CONFIG_ORIGIN_BLOB,
1039                                 ".gitmodules", buf, size, &data, &config_opts))
1040                 data.ret |= report(options, &blob->object,
1041                                    FSCK_MSG_GITMODULES_PARSE,
1042                                    "could not parse gitmodules blob");
1043
1044         return data.ret;
1045 }
1046
1047 int fsck_object(struct object *obj, void *data, unsigned long size,
1048         struct fsck_options *options)
1049 {
1050         if (!obj)
1051                 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
1052
1053         if (obj->type == OBJ_BLOB)
1054                 return fsck_blob((struct blob *)obj, data, size, options);
1055         if (obj->type == OBJ_TREE)
1056                 return fsck_tree((struct tree *) obj, options);
1057         if (obj->type == OBJ_COMMIT)
1058                 return fsck_commit((struct commit *) obj, (const char *) data,
1059                         size, options);
1060         if (obj->type == OBJ_TAG)
1061                 return fsck_tag((struct tag *) obj, (const char *) data,
1062                         size, options);
1063
1064         return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
1065                           obj->type);
1066 }
1067
1068 int fsck_error_function(struct fsck_options *o,
1069         struct object *obj, int msg_type, const char *message)
1070 {
1071         if (msg_type == FSCK_WARN) {
1072                 warning("object %s: %s", describe_object(o, obj), message);
1073                 return 0;
1074         }
1075         error("object %s: %s", describe_object(o, obj), message);
1076         return 1;
1077 }
1078
1079 int fsck_finish(struct fsck_options *options)
1080 {
1081         int ret = 0;
1082         struct oidset_iter iter;
1083         const struct object_id *oid;
1084
1085         oidset_iter_init(&gitmodules_found, &iter);
1086         while ((oid = oidset_iter_next(&iter))) {
1087                 struct blob *blob;
1088                 enum object_type type;
1089                 unsigned long size;
1090                 char *buf;
1091
1092                 if (oidset_contains(&gitmodules_done, oid))
1093                         continue;
1094
1095                 blob = lookup_blob(the_repository, oid);
1096                 if (!blob) {
1097                         struct object *obj = lookup_unknown_object(oid);
1098                         ret |= report(options, obj,
1099                                       FSCK_MSG_GITMODULES_BLOB,
1100                                       "non-blob found at .gitmodules");
1101                         continue;
1102                 }
1103
1104                 buf = read_object_file(oid, &type, &size);
1105                 if (!buf) {
1106                         if (is_promisor_object(&blob->object.oid))
1107                                 continue;
1108                         ret |= report(options, &blob->object,
1109                                       FSCK_MSG_GITMODULES_MISSING,
1110                                       "unable to read .gitmodules blob");
1111                         continue;
1112                 }
1113
1114                 if (type == OBJ_BLOB)
1115                         ret |= fsck_blob(blob, buf, size, options);
1116                 else
1117                         ret |= report(options, &blob->object,
1118                                       FSCK_MSG_GITMODULES_BLOB,
1119                                       "non-blob found at .gitmodules");
1120                 free(buf);
1121         }
1122
1123
1124         oidset_clear(&gitmodules_found);
1125         oidset_clear(&gitmodules_done);
1126         return ret;
1127 }