git.oblomov.eu Git - git/blob - fsck.c

   1 #include "cache.h"
   2 #include "object-store.h"
   3 #include "repository.h"
   4 #include "object.h"
   5 #include "blob.h"
   6 #include "tree.h"
   7 #include "tree-walk.h"
   8 #include "commit.h"
   9 #include "tag.h"
  10 #include "fsck.h"
  11 #include "refs.h"
  12 #include "utf8.h"
  13 #include "decorate.h"
  14 #include "oidset.h"
  15 #include "packfile.h"
  16 #include "submodule-config.h"
  17 #include "config.h"
  18 #include "help.h"
  19
  20 static struct oidset gitmodules_found = OIDSET_INIT;
  21 static struct oidset gitmodules_done = OIDSET_INIT;
  22
  23 #define FSCK_FATAL -1
  24 #define FSCK_INFO -2
  25
  26 #define FOREACH_MSG_ID(FUNC) \
  27         /* fatal errors */ \
  28         FUNC(NUL_IN_HEADER, FATAL) \
  29         FUNC(UNTERMINATED_HEADER, FATAL) \
  30         /* errors */ \
  31         FUNC(BAD_DATE, ERROR) \
  32         FUNC(BAD_DATE_OVERFLOW, ERROR) \
  33         FUNC(BAD_EMAIL, ERROR) \
  34         FUNC(BAD_NAME, ERROR) \
  35         FUNC(BAD_OBJECT_SHA1, ERROR) \
  36         FUNC(BAD_PARENT_SHA1, ERROR) \
  37         FUNC(BAD_TAG_OBJECT, ERROR) \
  38         FUNC(BAD_TIMEZONE, ERROR) \
  39         FUNC(BAD_TREE, ERROR) \
  40         FUNC(BAD_TREE_SHA1, ERROR) \
  41         FUNC(BAD_TYPE, ERROR) \
  42         FUNC(DUPLICATE_ENTRIES, ERROR) \
  43         FUNC(MISSING_AUTHOR, ERROR) \
  44         FUNC(MISSING_COMMITTER, ERROR) \
  45         FUNC(MISSING_EMAIL, ERROR) \
  46         FUNC(MISSING_NAME_BEFORE_EMAIL, ERROR) \
  47         FUNC(MISSING_OBJECT, ERROR) \
  48         FUNC(MISSING_SPACE_BEFORE_DATE, ERROR) \
  49         FUNC(MISSING_SPACE_BEFORE_EMAIL, ERROR) \
  50         FUNC(MISSING_TAG, ERROR) \
  51         FUNC(MISSING_TAG_ENTRY, ERROR) \
  52         FUNC(MISSING_TREE, ERROR) \
  53         FUNC(MISSING_TREE_OBJECT, ERROR) \
  54         FUNC(MISSING_TYPE, ERROR) \
  55         FUNC(MISSING_TYPE_ENTRY, ERROR) \
  56         FUNC(MULTIPLE_AUTHORS, ERROR) \
  57         FUNC(TREE_NOT_SORTED, ERROR) \
  58         FUNC(UNKNOWN_TYPE, ERROR) \
  59         FUNC(ZERO_PADDED_DATE, ERROR) \
  60         FUNC(GITMODULES_MISSING, ERROR) \
  61         FUNC(GITMODULES_BLOB, ERROR) \
  62         FUNC(GITMODULES_LARGE, ERROR) \
  63         FUNC(GITMODULES_NAME, ERROR) \
  64         FUNC(GITMODULES_SYMLINK, ERROR) \
  65         FUNC(GITMODULES_URL, ERROR) \
  66         FUNC(GITMODULES_PATH, ERROR) \
  67         /* warnings */ \
  68         FUNC(BAD_FILEMODE, WARN) \
  69         FUNC(EMPTY_NAME, WARN) \
  70         FUNC(FULL_PATHNAME, WARN) \
  71         FUNC(HAS_DOT, WARN) \
  72         FUNC(HAS_DOTDOT, WARN) \
  73         FUNC(HAS_DOTGIT, WARN) \
  74         FUNC(NULL_SHA1, WARN) \
  75         FUNC(ZERO_PADDED_FILEMODE, WARN) \
  76         FUNC(NUL_IN_COMMIT, WARN) \
  77         /* infos (reported as warnings, but ignored by default) */ \
  78         FUNC(GITMODULES_PARSE, INFO) \
  79         FUNC(BAD_TAG_NAME, INFO) \
  80         FUNC(MISSING_TAGGER_ENTRY, INFO)
  81
  82 #define MSG_ID(id, msg_type) FSCK_MSG_##id,
  83 enum fsck_msg_id {
  84         FOREACH_MSG_ID(MSG_ID)
  85         FSCK_MSG_MAX
  86 };
  87 #undef MSG_ID
  88
  89 #define STR(x) #x
  90 #define MSG_ID(id, msg_type) { STR(id), NULL, NULL, FSCK_##msg_type },
  91 static struct {
  92         const char *id_string;
  93         const char *downcased;
  94         const char *camelcased;
  95         int msg_type;
  96 } msg_id_info[FSCK_MSG_MAX + 1] = {
  97         FOREACH_MSG_ID(MSG_ID)
  98         { NULL, NULL, NULL, -1 }
  99 };
 100 #undef MSG_ID
 101
 102 static void prepare_msg_ids(void)
 103 {
 104         int i;
 105
 106         if (msg_id_info[0].downcased)
 107                 return;
 108
 109         /* convert id_string to lower case, without underscores. */
 110         for (i = 0; i < FSCK_MSG_MAX; i++) {
 111                 const char *p = msg_id_info[i].id_string;
 112                 int len = strlen(p);
 113                 char *q = xmalloc(len);
 114
 115                 msg_id_info[i].downcased = q;
 116                 while (*p)
 117                         if (*p == '_')
 118                                 p++;
 119                         else
 120                                 *(q)++ = tolower(*(p)++);
 121                 *q = '\0';
 122
 123                 p = msg_id_info[i].id_string;
 124                 q = xmalloc(len);
 125                 msg_id_info[i].camelcased = q;
 126                 while (*p) {
 127                         if (*p == '_') {
 128                                 p++;
 129                                 if (*p)
 130                                         *q++ = *p++;
 131                         } else {
 132                                 *q++ = tolower(*p++);
 133                         }
 134                 }
 135                 *q = '\0';
 136         }
 137 }
 138
 139 static int parse_msg_id(const char *text)
 140 {
 141         int i;
 142
 143         prepare_msg_ids();
 144
 145         for (i = 0; i < FSCK_MSG_MAX; i++)
 146                 if (!strcmp(text, msg_id_info[i].downcased))
 147                         return i;
 148
 149         return -1;
 150 }
 151
 152 void list_config_fsck_msg_ids(struct string_list *list, const char *prefix)
 153 {
 154         int i;
 155
 156         prepare_msg_ids();
 157
 158         for (i = 0; i < FSCK_MSG_MAX; i++)
 159                 list_config_item(list, prefix, msg_id_info[i].camelcased);
 160 }
 161
 162 static int fsck_msg_type(enum fsck_msg_id msg_id,
 163         struct fsck_options *options)
 164 {
 165         int msg_type;
 166
 167         assert(msg_id >= 0 && msg_id < FSCK_MSG_MAX);
 168
 169         if (options->msg_type)
 170                 msg_type = options->msg_type[msg_id];
 171         else {
 172                 msg_type = msg_id_info[msg_id].msg_type;
 173                 if (options->strict && msg_type == FSCK_WARN)
 174                         msg_type = FSCK_ERROR;
 175         }
 176
 177         return msg_type;
 178 }
 179
 180 static int parse_msg_type(const char *str)
 181 {
 182         if (!strcmp(str, "error"))
 183                 return FSCK_ERROR;
 184         else if (!strcmp(str, "warn"))
 185                 return FSCK_WARN;
 186         else if (!strcmp(str, "ignore"))
 187                 return FSCK_IGNORE;
 188         else
 189                 die("Unknown fsck message type: '%s'", str);
 190 }
 191
 192 int is_valid_msg_type(const char *msg_id, const char *msg_type)
 193 {
 194         if (parse_msg_id(msg_id) < 0)
 195                 return 0;
 196         parse_msg_type(msg_type);
 197         return 1;
 198 }
 199
 200 void fsck_set_msg_type(struct fsck_options *options,
 201                 const char *msg_id, const char *msg_type)
 202 {
 203         int id = parse_msg_id(msg_id), type;
 204
 205         if (id < 0)
 206                 die("Unhandled message id: %s", msg_id);
 207         type = parse_msg_type(msg_type);
 208
 209         if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
 210                 die("Cannot demote %s to %s", msg_id, msg_type);
 211
 212         if (!options->msg_type) {
 213                 int i;
 214                 int *msg_type;
 215                 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
 216                 for (i = 0; i < FSCK_MSG_MAX; i++)
 217                         msg_type[i] = fsck_msg_type(i, options);
 218                 options->msg_type = msg_type;
 219         }
 220
 221         options->msg_type[id] = type;
 222 }
 223
 224 void fsck_set_msg_types(struct fsck_options *options, const char *values)
 225 {
 226         char *buf = xstrdup(values), *to_free = buf;
 227         int done = 0;
 228
 229         while (!done) {
 230                 int len = strcspn(buf, " ,|"), equal;
 231
 232                 done = !buf[len];
 233                 if (!len) {
 234                         buf++;
 235                         continue;
 236                 }
 237                 buf[len] = '\0';
 238
 239                 for (equal = 0;
 240                      equal < len && buf[equal] != '=' && buf[equal] != ':';
 241                      equal++)
 242                         buf[equal] = tolower(buf[equal]);
 243                 buf[equal] = '\0';
 244
 245                 if (!strcmp(buf, "skiplist")) {
 246                         if (equal == len)
 247                                 die("skiplist requires a path");
 248                         oidset_parse_file(&options->skiplist, buf + equal + 1);
 249                         buf += len + 1;
 250                         continue;
 251                 }
 252
 253                 if (equal == len)
 254                         die("Missing '=': '%s'", buf);
 255
 256                 fsck_set_msg_type(options, buf, buf + equal + 1);
 257                 buf += len + 1;
 258         }
 259         free(to_free);
 260 }
 261
 262 static void append_msg_id(struct strbuf *sb, const char *msg_id)
 263 {
 264         for (;;) {
 265                 char c = *(msg_id)++;
 266
 267                 if (!c)
 268                         break;
 269                 if (c != '_')
 270                         strbuf_addch(sb, tolower(c));
 271                 else {
 272                         assert(*msg_id);
 273                         strbuf_addch(sb, *(msg_id)++);
 274                 }
 275         }
 276
 277         strbuf_addstr(sb, ": ");
 278 }
 279
 280 static int object_on_skiplist(struct fsck_options *opts, struct object *obj)
 281 {
 282         return opts && obj && oidset_contains(&opts->skiplist, &obj->oid);
 283 }
 284
 285 __attribute__((format (printf, 4, 5)))
 286 static int report(struct fsck_options *options, struct object *object,
 287         enum fsck_msg_id id, const char *fmt, ...)
 288 {
 289         va_list ap;
 290         struct strbuf sb = STRBUF_INIT;
 291         int msg_type = fsck_msg_type(id, options), result;
 292
 293         if (msg_type == FSCK_IGNORE)
 294                 return 0;
 295
 296         if (object_on_skiplist(options, object))
 297                 return 0;
 298
 299         if (msg_type == FSCK_FATAL)
 300                 msg_type = FSCK_ERROR;
 301         else if (msg_type == FSCK_INFO)
 302                 msg_type = FSCK_WARN;
 303
 304         append_msg_id(&sb, msg_id_info[id].id_string);
 305
 306         va_start(ap, fmt);
 307         strbuf_vaddf(&sb, fmt, ap);
 308         result = options->error_func(options, object, msg_type, sb.buf);
 309         strbuf_release(&sb);
 310         va_end(ap);
 311
 312         return result;
 313 }
 314
 315 void fsck_enable_object_names(struct fsck_options *options)
 316 {
 317         if (!options->object_names)
 318                 options->object_names = xcalloc(1, sizeof(struct decoration));
 319 }
 320
 321 const char *fsck_get_object_name(struct fsck_options *options, struct object *obj)
 322 {
 323         if (!options->object_names)
 324                 return NULL;
 325         return lookup_decoration(options->object_names, obj);
 326 }
 327
 328 void fsck_put_object_name(struct fsck_options *options, struct object *obj,
 329                           const char *fmt, ...)
 330 {
 331         va_list ap;
 332         struct strbuf buf = STRBUF_INIT;
 333         char *existing;
 334
 335         if (!options->object_names)
 336                 return;
 337         existing = lookup_decoration(options->object_names, obj);
 338         if (existing)
 339                 return;
 340         va_start(ap, fmt);
 341         strbuf_vaddf(&buf, fmt, ap);
 342         add_decoration(options->object_names, obj, strbuf_detach(&buf, NULL));
 343         va_end(ap);
 344 }
 345
 346 const char *fsck_describe_object(struct fsck_options *options,
 347                                  struct object *obj)
 348 {
 349         static struct strbuf bufs[] = {
 350                 STRBUF_INIT, STRBUF_INIT, STRBUF_INIT, STRBUF_INIT
 351         };
 352         static int b = 0;
 353         struct strbuf *buf;
 354         char *name = NULL;
 355
 356         if (options->object_names)
 357                 name = lookup_decoration(options->object_names, obj);
 358
 359         buf = bufs + b;
 360         b = (b + 1) % ARRAY_SIZE(bufs);
 361         strbuf_reset(buf);
 362         strbuf_addstr(buf, oid_to_hex(&obj->oid));
 363         if (name)
 364                 strbuf_addf(buf, " (%s)", name);
 365
 366         return buf->buf;
 367 }
 368
 369 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
 370 {
 371         struct tree_desc desc;
 372         struct name_entry entry;
 373         int res = 0;
 374         const char *name;
 375
 376         if (parse_tree(tree))
 377                 return -1;
 378
 379         name = fsck_get_object_name(options, &tree->object);
 380         if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
 381                 return -1;
 382         while (tree_entry_gently(&desc, &entry)) {
 383                 struct object *obj;
 384                 int result;
 385
 386                 if (S_ISGITLINK(entry.mode))
 387                         continue;
 388
 389                 if (S_ISDIR(entry.mode)) {
 390                         obj = (struct object *)lookup_tree(the_repository, &entry.oid);
 391                         if (name && obj)
 392                                 fsck_put_object_name(options, obj, "%s%s/",
 393                                                      name, entry.path);
 394                         result = options->walk(obj, OBJ_TREE, data, options);
 395                 }
 396                 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
 397                         obj = (struct object *)lookup_blob(the_repository, &entry.oid);
 398                         if (name && obj)
 399                                 fsck_put_object_name(options, obj, "%s%s",
 400                                                      name, entry.path);
 401                         result = options->walk(obj, OBJ_BLOB, data, options);
 402                 }
 403                 else {
 404                         result = error("in tree %s: entry %s has bad mode %.6o",
 405                                        fsck_describe_object(options, &tree->object),
 406                                        entry.path, entry.mode);
 407                 }
 408                 if (result < 0)
 409                         return result;
 410                 if (!res)
 411                         res = result;
 412         }
 413         return res;
 414 }
 415
 416 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
 417 {
 418         int counter = 0, generation = 0, name_prefix_len = 0;
 419         struct commit_list *parents;
 420         int res;
 421         int result;
 422         const char *name;
 423
 424         if (parse_commit(commit))
 425                 return -1;
 426
 427         name = fsck_get_object_name(options, &commit->object);
 428         if (name)
 429                 fsck_put_object_name(options, &get_commit_tree(commit)->object,
 430                                      "%s:", name);
 431
 432         result = options->walk((struct object *)get_commit_tree(commit),
 433                                OBJ_TREE, data, options);
 434         if (result < 0)
 435                 return result;
 436         res = result;
 437
 438         parents = commit->parents;
 439         if (name && parents) {
 440                 int len = strlen(name), power;
 441
 442                 if (len && name[len - 1] == '^') {
 443                         generation = 1;
 444                         name_prefix_len = len - 1;
 445                 }
 446                 else { /* parse ~<generation> suffix */
 447                         for (generation = 0, power = 1;
 448                              len && isdigit(name[len - 1]);
 449                              power *= 10)
 450                                 generation += power * (name[--len] - '0');
 451                         if (power > 1 && len && name[len - 1] == '~')
 452                                 name_prefix_len = len - 1;
 453                 }
 454         }
 455
 456         while (parents) {
 457                 if (name) {
 458                         struct object *obj = &parents->item->object;
 459
 460                         if (counter++)
 461                                 fsck_put_object_name(options, obj, "%s^%d",
 462                                                      name, counter);
 463                         else if (generation > 0)
 464                                 fsck_put_object_name(options, obj, "%.*s~%d",
 465                                                      name_prefix_len, name,
 466                                                      generation + 1);
 467                         else
 468                                 fsck_put_object_name(options, obj, "%s^",
 469                                                      name);
 470                 }
 471                 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
 472                 if (result < 0)
 473                         return result;
 474                 if (!res)
 475                         res = result;
 476                 parents = parents->next;
 477         }
 478         return res;
 479 }
 480
 481 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
 482 {
 483         const char *name = fsck_get_object_name(options, &tag->object);
 484
 485         if (parse_tag(tag))
 486                 return -1;
 487         if (name)
 488                 fsck_put_object_name(options, tag->tagged, "%s", name);
 489         return options->walk(tag->tagged, OBJ_ANY, data, options);
 490 }
 491
 492 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
 493 {
 494         if (!obj)
 495                 return -1;
 496
 497         if (obj->type == OBJ_NONE)
 498                 parse_object(the_repository, &obj->oid);
 499
 500         switch (obj->type) {
 501         case OBJ_BLOB:
 502                 return 0;
 503         case OBJ_TREE:
 504                 return fsck_walk_tree((struct tree *)obj, data, options);
 505         case OBJ_COMMIT:
 506                 return fsck_walk_commit((struct commit *)obj, data, options);
 507         case OBJ_TAG:
 508                 return fsck_walk_tag((struct tag *)obj, data, options);
 509         default:
 510                 error("Unknown object type for %s",
 511                       fsck_describe_object(options, obj));
 512                 return -1;
 513         }
 514 }
 515
 516 /*
 517  * The entries in a tree are ordered in the _path_ order,
 518  * which means that a directory entry is ordered by adding
 519  * a slash to the end of it.
 520  *
 521  * So a directory called "a" is ordered _after_ a file
 522  * called "a.c", because "a/" sorts after "a.c".
 523  */
 524 #define TREE_UNORDERED (-1)
 525 #define TREE_HAS_DUPS  (-2)
 526
 527 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 528 {
 529         int len1 = strlen(name1);
 530         int len2 = strlen(name2);
 531         int len = len1 < len2 ? len1 : len2;
 532         unsigned char c1, c2;
 533         int cmp;
 534
 535         cmp = memcmp(name1, name2, len);
 536         if (cmp < 0)
 537                 return 0;
 538         if (cmp > 0)
 539                 return TREE_UNORDERED;
 540
 541         /*
 542          * Ok, the first <len> characters are the same.
 543          * Now we need to order the next one, but turn
 544          * a '\0' into a '/' for a directory entry.
 545          */
 546         c1 = name1[len];
 547         c2 = name2[len];
 548         if (!c1 && !c2)
 549                 /*
 550                  * git-write-tree used to write out a nonsense tree that has
 551                  * entries with the same name, one blob and one tree.  Make
 552                  * sure we do not have duplicate entries.
 553                  */
 554                 return TREE_HAS_DUPS;
 555         if (!c1 && S_ISDIR(mode1))
 556                 c1 = '/';
 557         if (!c2 && S_ISDIR(mode2))
 558                 c2 = '/';
 559         return c1 < c2 ? 0 : TREE_UNORDERED;
 560 }
 561
 562 static int fsck_tree(struct tree *item,
 563                      const char *buffer, unsigned long size,
 564                      struct fsck_options *options)
 565 {
 566         int retval = 0;
 567         int has_null_sha1 = 0;
 568         int has_full_path = 0;
 569         int has_empty_name = 0;
 570         int has_dot = 0;
 571         int has_dotdot = 0;
 572         int has_dotgit = 0;
 573         int has_zero_pad = 0;
 574         int has_bad_modes = 0;
 575         int has_dup_entries = 0;
 576         int not_properly_sorted = 0;
 577         struct tree_desc desc;
 578         unsigned o_mode;
 579         const char *o_name;
 580
 581         if (init_tree_desc_gently(&desc, buffer, size)) {
 582                 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 583                 return retval;
 584         }
 585
 586         o_mode = 0;
 587         o_name = NULL;
 588
 589         while (desc.size) {
 590                 unsigned short mode;
 591                 const char *name;
 592                 const struct object_id *oid;
 593
 594                 oid = tree_entry_extract(&desc, &name, &mode);
 595
 596                 has_null_sha1 |= is_null_oid(oid);
 597                 has_full_path |= !!strchr(name, '/');
 598                 has_empty_name |= !*name;
 599                 has_dot |= !strcmp(name, ".");
 600                 has_dotdot |= !strcmp(name, "..");
 601                 has_dotgit |= is_hfs_dotgit(name) || is_ntfs_dotgit(name);
 602                 has_zero_pad |= *(char *)desc.buffer == '0';
 603
 604                 if (is_hfs_dotgitmodules(name) || is_ntfs_dotgitmodules(name)) {
 605                         if (!S_ISLNK(mode))
 606                                 oidset_insert(&gitmodules_found, oid);
 607                         else
 608                                 retval += report(options, &item->object,
 609                                                  FSCK_MSG_GITMODULES_SYMLINK,
 610                                                  ".gitmodules is a symbolic link");
 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(struct commit *commit, const char *buffer,
 757                        unsigned long size, struct fsck_options *options)
 758 {
 759         struct object_id tree_oid, oid;
 760         unsigned author_count;
 761         int err;
 762         const char *buffer_begin = buffer;
 763         const char *p;
 764
 765         if (verify_headers(buffer, size, &commit->object, options))
 766                 return -1;
 767
 768         if (!skip_prefix(buffer, "tree ", &buffer))
 769                 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
 770         if (parse_oid_hex(buffer, &tree_oid, &p) || *p != '\n') {
 771                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
 772                 if (err)
 773                         return err;
 774         }
 775         buffer = p + 1;
 776         while (skip_prefix(buffer, "parent ", &buffer)) {
 777                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 778                         err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
 779                         if (err)
 780                                 return err;
 781                 }
 782                 buffer = p + 1;
 783         }
 784         author_count = 0;
 785         while (skip_prefix(buffer, "author ", &buffer)) {
 786                 author_count++;
 787                 err = fsck_ident(&buffer, &commit->object, options);
 788                 if (err)
 789                         return err;
 790         }
 791         if (author_count < 1)
 792                 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
 793         else if (author_count > 1)
 794                 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
 795         if (err)
 796                 return err;
 797         if (!skip_prefix(buffer, "committer ", &buffer))
 798                 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
 799         err = fsck_ident(&buffer, &commit->object, options);
 800         if (err)
 801                 return err;
 802         if (memchr(buffer_begin, '\0', size)) {
 803                 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
 804                              "NUL byte in the commit object body");
 805                 if (err)
 806                         return err;
 807         }
 808         return 0;
 809 }
 810
 811 static int fsck_tag(struct tag *tag, const char *buffer,
 812                     unsigned long size, struct fsck_options *options)
 813 {
 814         struct object_id oid;
 815         int ret = 0;
 816         char *eol;
 817         struct strbuf sb = STRBUF_INIT;
 818         const char *p;
 819
 820         ret = verify_headers(buffer, size, &tag->object, options);
 821         if (ret)
 822                 goto done;
 823
 824         if (!skip_prefix(buffer, "object ", &buffer)) {
 825                 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
 826                 goto done;
 827         }
 828         if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 829                 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
 830                 if (ret)
 831                         goto done;
 832         }
 833         buffer = p + 1;
 834
 835         if (!skip_prefix(buffer, "type ", &buffer)) {
 836                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
 837                 goto done;
 838         }
 839         eol = strchr(buffer, '\n');
 840         if (!eol) {
 841                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
 842                 goto done;
 843         }
 844         if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
 845                 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
 846         if (ret)
 847                 goto done;
 848         buffer = eol + 1;
 849
 850         if (!skip_prefix(buffer, "tag ", &buffer)) {
 851                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
 852                 goto done;
 853         }
 854         eol = strchr(buffer, '\n');
 855         if (!eol) {
 856                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
 857                 goto done;
 858         }
 859         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
 860         if (check_refname_format(sb.buf, 0)) {
 861                 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
 862                            "invalid 'tag' name: %.*s",
 863                            (int)(eol - buffer), buffer);
 864                 if (ret)
 865                         goto done;
 866         }
 867         buffer = eol + 1;
 868
 869         if (!skip_prefix(buffer, "tagger ", &buffer)) {
 870                 /* early tags do not contain 'tagger' lines; warn only */
 871                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
 872                 if (ret)
 873                         goto done;
 874         }
 875         else
 876                 ret = fsck_ident(&buffer, &tag->object, options);
 877
 878 done:
 879         strbuf_release(&sb);
 880         return ret;
 881 }
 882
 883 struct fsck_gitmodules_data {
 884         struct object *obj;
 885         struct fsck_options *options;
 886         int ret;
 887 };
 888
 889 static int fsck_gitmodules_fn(const char *var, const char *value, void *vdata)
 890 {
 891         struct fsck_gitmodules_data *data = vdata;
 892         const char *subsection, *key;
 893         int subsection_len;
 894         char *name;
 895
 896         if (parse_config_key(var, "submodule", &subsection, &subsection_len, &key) < 0 ||
 897             !subsection)
 898                 return 0;
 899
 900         name = xmemdupz(subsection, subsection_len);
 901         if (check_submodule_name(name) < 0)
 902                 data->ret |= report(data->options, data->obj,
 903                                     FSCK_MSG_GITMODULES_NAME,
 904                                     "disallowed submodule name: %s",
 905                                     name);
 906         if (!strcmp(key, "url") && value &&
 907             looks_like_command_line_option(value))
 908                 data->ret |= report(data->options, data->obj,
 909                                     FSCK_MSG_GITMODULES_URL,
 910                                     "disallowed submodule url: %s",
 911                                     value);
 912         if (!strcmp(key, "path") && value &&
 913             looks_like_command_line_option(value))
 914                 data->ret |= report(data->options, data->obj,
 915                                     FSCK_MSG_GITMODULES_PATH,
 916                                     "disallowed submodule path: %s",
 917                                     value);
 918         free(name);
 919
 920         return 0;
 921 }
 922
 923 static int fsck_blob(struct blob *blob, const char *buf,
 924                      unsigned long size, struct fsck_options *options)
 925 {
 926         struct fsck_gitmodules_data data;
 927         struct config_options config_opts = { 0 };
 928
 929         if (!oidset_contains(&gitmodules_found, &blob->object.oid))
 930                 return 0;
 931         oidset_insert(&gitmodules_done, &blob->object.oid);
 932
 933         if (object_on_skiplist(options, &blob->object))
 934                 return 0;
 935
 936         if (!buf) {
 937                 /*
 938                  * A missing buffer here is a sign that the caller found the
 939                  * blob too gigantic to load into memory. Let's just consider
 940                  * that an error.
 941                  */
 942                 return report(options, &blob->object,
 943                               FSCK_MSG_GITMODULES_LARGE,
 944                               ".gitmodules too large to parse");
 945         }
 946
 947         data.obj = &blob->object;
 948         data.options = options;
 949         data.ret = 0;
 950         config_opts.error_action = CONFIG_ERROR_SILENT;
 951         if (git_config_from_mem(fsck_gitmodules_fn, CONFIG_ORIGIN_BLOB,
 952                                 ".gitmodules", buf, size, &data, &config_opts))
 953                 data.ret |= report(options, &blob->object,
 954                                    FSCK_MSG_GITMODULES_PARSE,
 955                                    "could not parse gitmodules blob");
 956
 957         return data.ret;
 958 }
 959
 960 int fsck_object(struct object *obj, void *data, unsigned long size,
 961         struct fsck_options *options)
 962 {
 963         if (!obj)
 964                 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
 965
 966         if (obj->type == OBJ_BLOB)
 967                 return fsck_blob((struct blob *)obj, data, size, options);
 968         if (obj->type == OBJ_TREE)
 969                 return fsck_tree((struct tree *) obj, data, size, options);
 970         if (obj->type == OBJ_COMMIT)
 971                 return fsck_commit((struct commit *) obj, (const char *) data,
 972                         size, options);
 973         if (obj->type == OBJ_TAG)
 974                 return fsck_tag((struct tag *) obj, (const char *) data,
 975                         size, options);
 976
 977         return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
 978                           obj->type);
 979 }
 980
 981 int fsck_error_function(struct fsck_options *o,
 982         struct object *obj, int msg_type, const char *message)
 983 {
 984         if (msg_type == FSCK_WARN) {
 985                 warning("object %s: %s", fsck_describe_object(o, obj), message);
 986                 return 0;
 987         }
 988         error("object %s: %s", fsck_describe_object(o, obj), message);
 989         return 1;
 990 }
 991
 992 int fsck_finish(struct fsck_options *options)
 993 {
 994         int ret = 0;
 995         struct oidset_iter iter;
 996         const struct object_id *oid;
 997
 998         oidset_iter_init(&gitmodules_found, &iter);
 999         while ((oid = oidset_iter_next(&iter))) {
1000                 struct blob *blob;
1001                 enum object_type type;
1002                 unsigned long size;
1003                 char *buf;
1004
1005                 if (oidset_contains(&gitmodules_done, oid))
1006                         continue;
1007
1008                 blob = lookup_blob(the_repository, oid);
1009                 if (!blob) {
1010                         struct object *obj = lookup_unknown_object(oid);
1011                         ret |= report(options, obj,
1012                                       FSCK_MSG_GITMODULES_BLOB,
1013                                       "non-blob found at .gitmodules");
1014                         continue;
1015                 }
1016
1017                 buf = read_object_file(oid, &type, &size);
1018                 if (!buf) {
1019                         if (is_promisor_object(&blob->object.oid))
1020                                 continue;
1021                         ret |= report(options, &blob->object,
1022                                       FSCK_MSG_GITMODULES_MISSING,
1023                                       "unable to read .gitmodules blob");
1024                         continue;
1025                 }
1026
1027                 if (type == OBJ_BLOB)
1028                         ret |= fsck_blob(blob, buf, size, options);
1029                 else
1030                         ret |= report(options, &blob->object,
1031                                       FSCK_MSG_GITMODULES_BLOB,
1032                                       "non-blob found at .gitmodules");
1033                 free(buf);
1034         }
1035
1036
1037         oidset_clear(&gitmodules_found);
1038         oidset_clear(&gitmodules_done);
1039         return ret;
1040 }