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 = kh_init_oid_map();
 319 }
 320
 321 const char *fsck_get_object_name(struct fsck_options *options,
 322                                  const struct object_id *oid)
 323 {
 324         khiter_t pos;
 325         if (!options->object_names)
 326                 return NULL;
 327         pos = kh_get_oid_map(options->object_names, *oid);
 328         if (pos >= kh_end(options->object_names))
 329                 return NULL;
 330         return kh_value(options->object_names, pos);
 331 }
 332
 333 void fsck_put_object_name(struct fsck_options *options,
 334                           const struct object_id *oid,
 335                           const char *fmt, ...)
 336 {
 337         va_list ap;
 338         struct strbuf buf = STRBUF_INIT;
 339         khiter_t pos;
 340         int hashret;
 341
 342         if (!options->object_names)
 343                 return;
 344
 345         pos = kh_put_oid_map(options->object_names, *oid, &hashret);
 346         if (!hashret)
 347                 return;
 348         va_start(ap, fmt);
 349         strbuf_vaddf(&buf, fmt, ap);
 350         kh_value(options->object_names, pos) = strbuf_detach(&buf, NULL);
 351         va_end(ap);
 352 }
 353
 354 const char *fsck_describe_object(struct fsck_options *options,
 355                                  const struct object_id *oid)
 356 {
 357         static struct strbuf bufs[] = {
 358                 STRBUF_INIT, STRBUF_INIT, STRBUF_INIT, STRBUF_INIT
 359         };
 360         static int b = 0;
 361         struct strbuf *buf;
 362         const char *name = fsck_get_object_name(options, oid);
 363
 364         buf = bufs + b;
 365         b = (b + 1) % ARRAY_SIZE(bufs);
 366         strbuf_reset(buf);
 367         strbuf_addstr(buf, oid_to_hex(oid));
 368         if (name)
 369                 strbuf_addf(buf, " (%s)", name);
 370
 371         return buf->buf;
 372 }
 373
 374 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
 375 {
 376         struct tree_desc desc;
 377         struct name_entry entry;
 378         int res = 0;
 379         const char *name;
 380
 381         if (parse_tree(tree))
 382                 return -1;
 383
 384         name = fsck_get_object_name(options, &tree->object.oid);
 385         if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
 386                 return -1;
 387         while (tree_entry_gently(&desc, &entry)) {
 388                 struct object *obj;
 389                 int result;
 390
 391                 if (S_ISGITLINK(entry.mode))
 392                         continue;
 393
 394                 if (S_ISDIR(entry.mode)) {
 395                         obj = (struct object *)lookup_tree(the_repository, &entry.oid);
 396                         if (name && obj)
 397                                 fsck_put_object_name(options, &entry.oid, "%s%s/",
 398                                                      name, entry.path);
 399                         result = options->walk(obj, OBJ_TREE, data, options);
 400                 }
 401                 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
 402                         obj = (struct object *)lookup_blob(the_repository, &entry.oid);
 403                         if (name && obj)
 404                                 fsck_put_object_name(options, &entry.oid, "%s%s",
 405                                                      name, entry.path);
 406                         result = options->walk(obj, OBJ_BLOB, data, options);
 407                 }
 408                 else {
 409                         result = error("in tree %s: entry %s has bad mode %.6o",
 410                                        fsck_describe_object(options, &tree->object.oid),
 411                                        entry.path, entry.mode);
 412                 }
 413                 if (result < 0)
 414                         return result;
 415                 if (!res)
 416                         res = result;
 417         }
 418         return res;
 419 }
 420
 421 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
 422 {
 423         int counter = 0, generation = 0, name_prefix_len = 0;
 424         struct commit_list *parents;
 425         int res;
 426         int result;
 427         const char *name;
 428
 429         if (parse_commit(commit))
 430                 return -1;
 431
 432         name = fsck_get_object_name(options, &commit->object.oid);
 433         if (name)
 434                 fsck_put_object_name(options, get_commit_tree_oid(commit),
 435                                      "%s:", name);
 436
 437         result = options->walk((struct object *)get_commit_tree(commit),
 438                                OBJ_TREE, data, options);
 439         if (result < 0)
 440                 return result;
 441         res = result;
 442
 443         parents = commit->parents;
 444         if (name && parents) {
 445                 int len = strlen(name), power;
 446
 447                 if (len && name[len - 1] == '^') {
 448                         generation = 1;
 449                         name_prefix_len = len - 1;
 450                 }
 451                 else { /* parse ~<generation> suffix */
 452                         for (generation = 0, power = 1;
 453                              len && isdigit(name[len - 1]);
 454                              power *= 10)
 455                                 generation += power * (name[--len] - '0');
 456                         if (power > 1 && len && name[len - 1] == '~')
 457                                 name_prefix_len = len - 1;
 458                 }
 459         }
 460
 461         while (parents) {
 462                 if (name) {
 463                         struct object_id *oid = &parents->item->object.oid;
 464
 465                         if (counter++)
 466                                 fsck_put_object_name(options, oid, "%s^%d",
 467                                                      name, counter);
 468                         else if (generation > 0)
 469                                 fsck_put_object_name(options, oid, "%.*s~%d",
 470                                                      name_prefix_len, name,
 471                                                      generation + 1);
 472                         else
 473                                 fsck_put_object_name(options, oid, "%s^", name);
 474                 }
 475                 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
 476                 if (result < 0)
 477                         return result;
 478                 if (!res)
 479                         res = result;
 480                 parents = parents->next;
 481         }
 482         return res;
 483 }
 484
 485 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
 486 {
 487         const char *name = fsck_get_object_name(options, &tag->object.oid);
 488
 489         if (parse_tag(tag))
 490                 return -1;
 491         if (name)
 492                 fsck_put_object_name(options, &tag->tagged->oid, "%s", name);
 493         return options->walk(tag->tagged, OBJ_ANY, data, options);
 494 }
 495
 496 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
 497 {
 498         if (!obj)
 499                 return -1;
 500
 501         if (obj->type == OBJ_NONE)
 502                 parse_object(the_repository, &obj->oid);
 503
 504         switch (obj->type) {
 505         case OBJ_BLOB:
 506                 return 0;
 507         case OBJ_TREE:
 508                 return fsck_walk_tree((struct tree *)obj, data, options);
 509         case OBJ_COMMIT:
 510                 return fsck_walk_commit((struct commit *)obj, data, options);
 511         case OBJ_TAG:
 512                 return fsck_walk_tag((struct tag *)obj, data, options);
 513         default:
 514                 error("Unknown object type for %s",
 515                       fsck_describe_object(options, &obj->oid));
 516                 return -1;
 517         }
 518 }
 519
 520 /*
 521  * The entries in a tree are ordered in the _path_ order,
 522  * which means that a directory entry is ordered by adding
 523  * a slash to the end of it.
 524  *
 525  * So a directory called "a" is ordered _after_ a file
 526  * called "a.c", because "a/" sorts after "a.c".
 527  */
 528 #define TREE_UNORDERED (-1)
 529 #define TREE_HAS_DUPS  (-2)
 530
 531 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 532 {
 533         int len1 = strlen(name1);
 534         int len2 = strlen(name2);
 535         int len = len1 < len2 ? len1 : len2;
 536         unsigned char c1, c2;
 537         int cmp;
 538
 539         cmp = memcmp(name1, name2, len);
 540         if (cmp < 0)
 541                 return 0;
 542         if (cmp > 0)
 543                 return TREE_UNORDERED;
 544
 545         /*
 546          * Ok, the first <len> characters are the same.
 547          * Now we need to order the next one, but turn
 548          * a '\0' into a '/' for a directory entry.
 549          */
 550         c1 = name1[len];
 551         c2 = name2[len];
 552         if (!c1 && !c2)
 553                 /*
 554                  * git-write-tree used to write out a nonsense tree that has
 555                  * entries with the same name, one blob and one tree.  Make
 556                  * sure we do not have duplicate entries.
 557                  */
 558                 return TREE_HAS_DUPS;
 559         if (!c1 && S_ISDIR(mode1))
 560                 c1 = '/';
 561         if (!c2 && S_ISDIR(mode2))
 562                 c2 = '/';
 563         return c1 < c2 ? 0 : TREE_UNORDERED;
 564 }
 565
 566 static int fsck_tree(struct tree *item,
 567                      const char *buffer, unsigned long size,
 568                      struct fsck_options *options)
 569 {
 570         int retval = 0;
 571         int has_null_sha1 = 0;
 572         int has_full_path = 0;
 573         int has_empty_name = 0;
 574         int has_dot = 0;
 575         int has_dotdot = 0;
 576         int has_dotgit = 0;
 577         int has_zero_pad = 0;
 578         int has_bad_modes = 0;
 579         int has_dup_entries = 0;
 580         int not_properly_sorted = 0;
 581         struct tree_desc desc;
 582         unsigned o_mode;
 583         const char *o_name;
 584
 585         if (init_tree_desc_gently(&desc, buffer, size)) {
 586                 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 587                 return retval;
 588         }
 589
 590         o_mode = 0;
 591         o_name = NULL;
 592
 593         while (desc.size) {
 594                 unsigned short mode;
 595                 const char *name;
 596                 const struct object_id *oid;
 597
 598                 oid = tree_entry_extract(&desc, &name, &mode);
 599
 600                 has_null_sha1 |= is_null_oid(oid);
 601                 has_full_path |= !!strchr(name, '/');
 602                 has_empty_name |= !*name;
 603                 has_dot |= !strcmp(name, ".");
 604                 has_dotdot |= !strcmp(name, "..");
 605                 has_dotgit |= is_hfs_dotgit(name) || is_ntfs_dotgit(name);
 606                 has_zero_pad |= *(char *)desc.buffer == '0';
 607
 608                 if (is_hfs_dotgitmodules(name) || is_ntfs_dotgitmodules(name)) {
 609                         if (!S_ISLNK(mode))
 610                                 oidset_insert(&gitmodules_found, oid);
 611                         else
 612                                 retval += report(options, &item->object,
 613                                                  FSCK_MSG_GITMODULES_SYMLINK,
 614                                                  ".gitmodules is a symbolic link");
 615                 }
 616
 617                 if (update_tree_entry_gently(&desc)) {
 618                         retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 619                         break;
 620                 }
 621
 622                 switch (mode) {
 623                 /*
 624                  * Standard modes..
 625                  */
 626                 case S_IFREG | 0755:
 627                 case S_IFREG | 0644:
 628                 case S_IFLNK:
 629                 case S_IFDIR:
 630                 case S_IFGITLINK:
 631                         break;
 632                 /*
 633                  * This is nonstandard, but we had a few of these
 634                  * early on when we honored the full set of mode
 635                  * bits..
 636                  */
 637                 case S_IFREG | 0664:
 638                         if (!options->strict)
 639                                 break;
 640                         /* fallthrough */
 641                 default:
 642                         has_bad_modes = 1;
 643                 }
 644
 645                 if (o_name) {
 646                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 647                         case TREE_UNORDERED:
 648                                 not_properly_sorted = 1;
 649                                 break;
 650                         case TREE_HAS_DUPS:
 651                                 has_dup_entries = 1;
 652                                 break;
 653                         default:
 654                                 break;
 655                         }
 656                 }
 657
 658                 o_mode = mode;
 659                 o_name = name;
 660         }
 661
 662         if (has_null_sha1)
 663                 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
 664         if (has_full_path)
 665                 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
 666         if (has_empty_name)
 667                 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
 668         if (has_dot)
 669                 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
 670         if (has_dotdot)
 671                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
 672         if (has_dotgit)
 673                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
 674         if (has_zero_pad)
 675                 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
 676         if (has_bad_modes)
 677                 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
 678         if (has_dup_entries)
 679                 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
 680         if (not_properly_sorted)
 681                 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
 682         return retval;
 683 }
 684
 685 static int verify_headers(const void *data, unsigned long size,
 686                           struct object *obj, struct fsck_options *options)
 687 {
 688         const char *buffer = (const char *)data;
 689         unsigned long i;
 690
 691         for (i = 0; i < size; i++) {
 692                 switch (buffer[i]) {
 693                 case '\0':
 694                         return report(options, obj,
 695                                 FSCK_MSG_NUL_IN_HEADER,
 696                                 "unterminated header: NUL at offset %ld", i);
 697                 case '\n':
 698                         if (i + 1 < size && buffer[i + 1] == '\n')
 699                                 return 0;
 700                 }
 701         }
 702
 703         /*
 704          * We did not find double-LF that separates the header
 705          * and the body.  Not having a body is not a crime but
 706          * we do want to see the terminating LF for the last header
 707          * line.
 708          */
 709         if (size && buffer[size - 1] == '\n')
 710                 return 0;
 711
 712         return report(options, obj,
 713                 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
 714 }
 715
 716 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
 717 {
 718         const char *p = *ident;
 719         char *end;
 720
 721         *ident = strchrnul(*ident, '\n');
 722         if (**ident == '\n')
 723                 (*ident)++;
 724
 725         if (*p == '<')
 726                 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 727         p += strcspn(p, "<>\n");
 728         if (*p == '>')
 729                 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
 730         if (*p != '<')
 731                 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
 732         if (p[-1] != ' ')
 733                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 734         p++;
 735         p += strcspn(p, "<>\n");
 736         if (*p != '>')
 737                 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
 738         p++;
 739         if (*p != ' ')
 740                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
 741         p++;
 742         if (*p == '0' && p[1] != ' ')
 743                 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
 744         if (date_overflows(parse_timestamp(p, &end, 10)))
 745                 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
 746         if ((end == p || *end != ' '))
 747                 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
 748         p = end + 1;
 749         if ((*p != '+' && *p != '-') ||
 750             !isdigit(p[1]) ||
 751             !isdigit(p[2]) ||
 752             !isdigit(p[3]) ||
 753             !isdigit(p[4]) ||
 754             (p[5] != '\n'))
 755                 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
 756         p += 6;
 757         return 0;
 758 }
 759
 760 static int fsck_commit(struct commit *commit, const char *buffer,
 761                        unsigned long size, struct fsck_options *options)
 762 {
 763         struct object_id tree_oid, oid;
 764         unsigned author_count;
 765         int err;
 766         const char *buffer_begin = buffer;
 767         const char *p;
 768
 769         if (verify_headers(buffer, size, &commit->object, options))
 770                 return -1;
 771
 772         if (!skip_prefix(buffer, "tree ", &buffer))
 773                 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
 774         if (parse_oid_hex(buffer, &tree_oid, &p) || *p != '\n') {
 775                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
 776                 if (err)
 777                         return err;
 778         }
 779         buffer = p + 1;
 780         while (skip_prefix(buffer, "parent ", &buffer)) {
 781                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 782                         err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
 783                         if (err)
 784                                 return err;
 785                 }
 786                 buffer = p + 1;
 787         }
 788         author_count = 0;
 789         while (skip_prefix(buffer, "author ", &buffer)) {
 790                 author_count++;
 791                 err = fsck_ident(&buffer, &commit->object, options);
 792                 if (err)
 793                         return err;
 794         }
 795         if (author_count < 1)
 796                 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
 797         else if (author_count > 1)
 798                 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
 799         if (err)
 800                 return err;
 801         if (!skip_prefix(buffer, "committer ", &buffer))
 802                 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
 803         err = fsck_ident(&buffer, &commit->object, options);
 804         if (err)
 805                 return err;
 806         if (memchr(buffer_begin, '\0', size)) {
 807                 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
 808                              "NUL byte in the commit object body");
 809                 if (err)
 810                         return err;
 811         }
 812         return 0;
 813 }
 814
 815 static int fsck_tag(struct tag *tag, const char *buffer,
 816                     unsigned long size, struct fsck_options *options)
 817 {
 818         struct object_id oid;
 819         int ret = 0;
 820         char *eol;
 821         struct strbuf sb = STRBUF_INIT;
 822         const char *p;
 823
 824         ret = verify_headers(buffer, size, &tag->object, options);
 825         if (ret)
 826                 goto done;
 827
 828         if (!skip_prefix(buffer, "object ", &buffer)) {
 829                 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
 830                 goto done;
 831         }
 832         if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 833                 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
 834                 if (ret)
 835                         goto done;
 836         }
 837         buffer = p + 1;
 838
 839         if (!skip_prefix(buffer, "type ", &buffer)) {
 840                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
 841                 goto done;
 842         }
 843         eol = strchr(buffer, '\n');
 844         if (!eol) {
 845                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
 846                 goto done;
 847         }
 848         if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
 849                 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
 850         if (ret)
 851                 goto done;
 852         buffer = eol + 1;
 853
 854         if (!skip_prefix(buffer, "tag ", &buffer)) {
 855                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
 856                 goto done;
 857         }
 858         eol = strchr(buffer, '\n');
 859         if (!eol) {
 860                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
 861                 goto done;
 862         }
 863         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
 864         if (check_refname_format(sb.buf, 0)) {
 865                 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
 866                            "invalid 'tag' name: %.*s",
 867                            (int)(eol - buffer), buffer);
 868                 if (ret)
 869                         goto done;
 870         }
 871         buffer = eol + 1;
 872
 873         if (!skip_prefix(buffer, "tagger ", &buffer)) {
 874                 /* early tags do not contain 'tagger' lines; warn only */
 875                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
 876                 if (ret)
 877                         goto done;
 878         }
 879         else
 880                 ret = fsck_ident(&buffer, &tag->object, options);
 881
 882 done:
 883         strbuf_release(&sb);
 884         return ret;
 885 }
 886
 887 struct fsck_gitmodules_data {
 888         struct object *obj;
 889         struct fsck_options *options;
 890         int ret;
 891 };
 892
 893 static int fsck_gitmodules_fn(const char *var, const char *value, void *vdata)
 894 {
 895         struct fsck_gitmodules_data *data = vdata;
 896         const char *subsection, *key;
 897         int subsection_len;
 898         char *name;
 899
 900         if (parse_config_key(var, "submodule", &subsection, &subsection_len, &key) < 0 ||
 901             !subsection)
 902                 return 0;
 903
 904         name = xmemdupz(subsection, subsection_len);
 905         if (check_submodule_name(name) < 0)
 906                 data->ret |= report(data->options, data->obj,
 907                                     FSCK_MSG_GITMODULES_NAME,
 908                                     "disallowed submodule name: %s",
 909                                     name);
 910         if (!strcmp(key, "url") && value &&
 911             looks_like_command_line_option(value))
 912                 data->ret |= report(data->options, data->obj,
 913                                     FSCK_MSG_GITMODULES_URL,
 914                                     "disallowed submodule url: %s",
 915                                     value);
 916         if (!strcmp(key, "path") && value &&
 917             looks_like_command_line_option(value))
 918                 data->ret |= report(data->options, data->obj,
 919                                     FSCK_MSG_GITMODULES_PATH,
 920                                     "disallowed submodule path: %s",
 921                                     value);
 922         free(name);
 923
 924         return 0;
 925 }
 926
 927 static int fsck_blob(struct blob *blob, const char *buf,
 928                      unsigned long size, struct fsck_options *options)
 929 {
 930         struct fsck_gitmodules_data data;
 931         struct config_options config_opts = { 0 };
 932
 933         if (!oidset_contains(&gitmodules_found, &blob->object.oid))
 934                 return 0;
 935         oidset_insert(&gitmodules_done, &blob->object.oid);
 936
 937         if (object_on_skiplist(options, &blob->object))
 938                 return 0;
 939
 940         if (!buf) {
 941                 /*
 942                  * A missing buffer here is a sign that the caller found the
 943                  * blob too gigantic to load into memory. Let's just consider
 944                  * that an error.
 945                  */
 946                 return report(options, &blob->object,
 947                               FSCK_MSG_GITMODULES_LARGE,
 948                               ".gitmodules too large to parse");
 949         }
 950
 951         data.obj = &blob->object;
 952         data.options = options;
 953         data.ret = 0;
 954         config_opts.error_action = CONFIG_ERROR_SILENT;
 955         if (git_config_from_mem(fsck_gitmodules_fn, CONFIG_ORIGIN_BLOB,
 956                                 ".gitmodules", buf, size, &data, &config_opts))
 957                 data.ret |= report(options, &blob->object,
 958                                    FSCK_MSG_GITMODULES_PARSE,
 959                                    "could not parse gitmodules blob");
 960
 961         return data.ret;
 962 }
 963
 964 int fsck_object(struct object *obj, void *data, unsigned long size,
 965         struct fsck_options *options)
 966 {
 967         if (!obj)
 968                 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
 969
 970         if (obj->type == OBJ_BLOB)
 971                 return fsck_blob((struct blob *)obj, data, size, options);
 972         if (obj->type == OBJ_TREE)
 973                 return fsck_tree((struct tree *) obj, data, size, options);
 974         if (obj->type == OBJ_COMMIT)
 975                 return fsck_commit((struct commit *) obj, (const char *) data,
 976                         size, options);
 977         if (obj->type == OBJ_TAG)
 978                 return fsck_tag((struct tag *) obj, (const char *) data,
 979                         size, options);
 980
 981         return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
 982                           obj->type);
 983 }
 984
 985 int fsck_error_function(struct fsck_options *o,
 986         struct object *obj, int msg_type, const char *message)
 987 {
 988         if (msg_type == FSCK_WARN) {
 989                 warning("object %s: %s", fsck_describe_object(o, &obj->oid), message);
 990                 return 0;
 991         }
 992         error("object %s: %s", fsck_describe_object(o, &obj->oid), message);
 993         return 1;
 994 }
 995
 996 int fsck_finish(struct fsck_options *options)
 997 {
 998         int ret = 0;
 999         struct oidset_iter iter;
1000         const struct object_id *oid;
1001
1002         oidset_iter_init(&gitmodules_found, &iter);
1003         while ((oid = oidset_iter_next(&iter))) {
1004                 struct blob *blob;
1005                 enum object_type type;
1006                 unsigned long size;
1007                 char *buf;
1008
1009                 if (oidset_contains(&gitmodules_done, oid))
1010                         continue;
1011
1012                 blob = lookup_blob(the_repository, oid);
1013                 if (!blob) {
1014                         struct object *obj = lookup_unknown_object(oid);
1015                         ret |= report(options, obj,
1016                                       FSCK_MSG_GITMODULES_BLOB,
1017                                       "non-blob found at .gitmodules");
1018                         continue;
1019                 }
1020
1021                 buf = read_object_file(oid, &type, &size);
1022                 if (!buf) {
1023                         if (is_promisor_object(&blob->object.oid))
1024                                 continue;
1025                         ret |= report(options, &blob->object,
1026                                       FSCK_MSG_GITMODULES_MISSING,
1027                                       "unable to read .gitmodules blob");
1028                         continue;
1029                 }
1030
1031                 if (type == OBJ_BLOB)
1032                         ret |= fsck_blob(blob, buf, size, options);
1033                 else
1034                         ret |= report(options, &blob->object,
1035                                       FSCK_MSG_GITMODULES_BLOB,
1036                                       "non-blob found at .gitmodules");
1037                 free(buf);
1038         }
1039
1040
1041         oidset_clear(&gitmodules_found);
1042         oidset_clear(&gitmodules_done);
1043         return ret;
1044 }