fsck.c

   1 #include "cache.h"
   2 #include "object.h"
   3 #include "blob.h"
   4 #include "tree.h"
   5 #include "tree-walk.h"
   6 #include "commit.h"
   7 #include "tag.h"
   8 #include "fsck.h"
   9 #include "refs.h"
  10 #include "utf8.h"
  11 #include "sha1-array.h"
  12 #include "decorate.h"
  13 #include "oidset.h"
  14 #include "packfile.h"
  15 #include "submodule-config.h"
  16 #include "config.h"
  17
  18 static struct oidset gitmodules_found = OIDSET_INIT;
  19 static struct oidset gitmodules_done = OIDSET_INIT;
  20
  21 #define FSCK_FATAL -1
  22 #define FSCK_INFO -2
  23
  24 #define FOREACH_MSG_ID(FUNC) \
  25         /* fatal errors */ \
  26         FUNC(NUL_IN_HEADER, FATAL) \
  27         FUNC(UNTERMINATED_HEADER, FATAL) \
  28         /* errors */ \
  29         FUNC(BAD_DATE, ERROR) \
  30         FUNC(BAD_DATE_OVERFLOW, ERROR) \
  31         FUNC(BAD_EMAIL, ERROR) \
  32         FUNC(BAD_NAME, ERROR) \
  33         FUNC(BAD_OBJECT_SHA1, ERROR) \
  34         FUNC(BAD_PARENT_SHA1, ERROR) \
  35         FUNC(BAD_TAG_OBJECT, ERROR) \
  36         FUNC(BAD_TIMEZONE, ERROR) \
  37         FUNC(BAD_TREE, ERROR) \
  38         FUNC(BAD_TREE_SHA1, ERROR) \
  39         FUNC(BAD_TYPE, ERROR) \
  40         FUNC(DUPLICATE_ENTRIES, ERROR) \
  41         FUNC(MISSING_AUTHOR, ERROR) \
  42         FUNC(MISSING_COMMITTER, ERROR) \
  43         FUNC(MISSING_EMAIL, ERROR) \
  44         FUNC(MISSING_GRAFT, ERROR) \
  45         FUNC(MISSING_NAME_BEFORE_EMAIL, ERROR) \
  46         FUNC(MISSING_OBJECT, ERROR) \
  47         FUNC(MISSING_PARENT, 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_TAG_OBJECT, ERROR) \
  53         FUNC(MISSING_TREE, ERROR) \
  54         FUNC(MISSING_TREE_OBJECT, ERROR) \
  55         FUNC(MISSING_TYPE, ERROR) \
  56         FUNC(MISSING_TYPE_ENTRY, ERROR) \
  57         FUNC(MULTIPLE_AUTHORS, ERROR) \
  58         FUNC(TAG_OBJECT_NOT_TAG, ERROR) \
  59         FUNC(TREE_NOT_SORTED, ERROR) \
  60         FUNC(UNKNOWN_TYPE, ERROR) \
  61         FUNC(ZERO_PADDED_DATE, ERROR) \
  62         FUNC(GITMODULES_MISSING, ERROR) \
  63         FUNC(GITMODULES_BLOB, ERROR) \
  64         FUNC(GITMODULES_PARSE, ERROR) \
  65         FUNC(GITMODULES_NAME, ERROR) \
  66         FUNC(GITMODULES_SYMLINK, ERROR) \
  67         FUNC(GITMODULES_URL, ERROR) \
  68         FUNC(GITMODULES_PATH, ERROR) \
  69         FUNC(GITMODULES_UPDATE, ERROR) \
  70         /* warnings */ \
  71         FUNC(BAD_FILEMODE, WARN) \
  72         FUNC(EMPTY_NAME, WARN) \
  73         FUNC(FULL_PATHNAME, WARN) \
  74         FUNC(HAS_DOT, WARN) \
  75         FUNC(HAS_DOTDOT, WARN) \
  76         FUNC(HAS_DOTGIT, WARN) \
  77         FUNC(NULL_SHA1, WARN) \
  78         FUNC(ZERO_PADDED_FILEMODE, WARN) \
  79         FUNC(NUL_IN_COMMIT, WARN) \
  80         /* infos (reported as warnings, but ignored by default) */ \
  81         FUNC(BAD_TAG_NAME, INFO) \
  82         FUNC(MISSING_TAGGER_ENTRY, INFO)
  83
  84 #define MSG_ID(id, msg_type) FSCK_MSG_##id,
  85 enum fsck_msg_id {
  86         FOREACH_MSG_ID(MSG_ID)
  87         FSCK_MSG_MAX
  88 };
  89 #undef MSG_ID
  90
  91 #define STR(x) #x
  92 #define MSG_ID(id, msg_type) { STR(id), NULL, FSCK_##msg_type },
  93 static struct {
  94         const char *id_string;
  95         const char *downcased;
  96         int msg_type;
  97 } msg_id_info[FSCK_MSG_MAX + 1] = {
  98         FOREACH_MSG_ID(MSG_ID)
  99         { NULL, NULL, -1 }
 100 };
 101 #undef MSG_ID
 102
 103 static int parse_msg_id(const char *text)
 104 {
 105         int i;
 106
 107         if (!msg_id_info[0].downcased) {
 108                 /* convert id_string to lower case, without underscores. */
 109                 for (i = 0; i < FSCK_MSG_MAX; i++) {
 110                         const char *p = msg_id_info[i].id_string;
 111                         int len = strlen(p);
 112                         char *q = xmalloc(len);
 113
 114                         msg_id_info[i].downcased = q;
 115                         while (*p)
 116                                 if (*p == '_')
 117                                         p++;
 118                                 else
 119                                         *(q)++ = tolower(*(p)++);
 120                         *q = '\0';
 121                 }
 122         }
 123
 124         for (i = 0; i < FSCK_MSG_MAX; i++)
 125                 if (!strcmp(text, msg_id_info[i].downcased))
 126                         return i;
 127
 128         return -1;
 129 }
 130
 131 static int fsck_msg_type(enum fsck_msg_id msg_id,
 132         struct fsck_options *options)
 133 {
 134         int msg_type;
 135
 136         assert(msg_id >= 0 && msg_id < FSCK_MSG_MAX);
 137
 138         if (options->msg_type)
 139                 msg_type = options->msg_type[msg_id];
 140         else {
 141                 msg_type = msg_id_info[msg_id].msg_type;
 142                 if (options->strict && msg_type == FSCK_WARN)
 143                         msg_type = FSCK_ERROR;
 144         }
 145
 146         return msg_type;
 147 }
 148
 149 static void init_skiplist(struct fsck_options *options, const char *path)
 150 {
 151         static struct oid_array skiplist = OID_ARRAY_INIT;
 152         int sorted, fd;
 153         char buffer[GIT_MAX_HEXSZ + 1];
 154         struct object_id oid;
 155
 156         if (options->skiplist)
 157                 sorted = options->skiplist->sorted;
 158         else {
 159                 sorted = 1;
 160                 options->skiplist = &skiplist;
 161         }
 162
 163         fd = open(path, O_RDONLY);
 164         if (fd < 0)
 165                 die("Could not open skip list: %s", path);
 166         for (;;) {
 167                 const char *p;
 168                 int result = read_in_full(fd, buffer, sizeof(buffer));
 169                 if (result < 0)
 170                         die_errno("Could not read '%s'", path);
 171                 if (!result)
 172                         break;
 173                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n')
 174                         die("Invalid SHA-1: %s", buffer);
 175                 oid_array_append(&skiplist, &oid);
 176                 if (sorted && skiplist.nr > 1 &&
 177                                 oidcmp(&skiplist.oid[skiplist.nr - 2],
 178                                        &oid) > 0)
 179                         sorted = 0;
 180         }
 181         close(fd);
 182
 183         if (sorted)
 184                 skiplist.sorted = 1;
 185 }
 186
 187 static int parse_msg_type(const char *str)
 188 {
 189         if (!strcmp(str, "error"))
 190                 return FSCK_ERROR;
 191         else if (!strcmp(str, "warn"))
 192                 return FSCK_WARN;
 193         else if (!strcmp(str, "ignore"))
 194                 return FSCK_IGNORE;
 195         else
 196                 die("Unknown fsck message type: '%s'", str);
 197 }
 198
 199 int is_valid_msg_type(const char *msg_id, const char *msg_type)
 200 {
 201         if (parse_msg_id(msg_id) < 0)
 202                 return 0;
 203         parse_msg_type(msg_type);
 204         return 1;
 205 }
 206
 207 void fsck_set_msg_type(struct fsck_options *options,
 208                 const char *msg_id, const char *msg_type)
 209 {
 210         int id = parse_msg_id(msg_id), type;
 211
 212         if (id < 0)
 213                 die("Unhandled message id: %s", msg_id);
 214         type = parse_msg_type(msg_type);
 215
 216         if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
 217                 die("Cannot demote %s to %s", msg_id, msg_type);
 218
 219         if (!options->msg_type) {
 220                 int i;
 221                 int *msg_type;
 222                 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
 223                 for (i = 0; i < FSCK_MSG_MAX; i++)
 224                         msg_type[i] = fsck_msg_type(i, options);
 225                 options->msg_type = msg_type;
 226         }
 227
 228         options->msg_type[id] = type;
 229 }
 230
 231 void fsck_set_msg_types(struct fsck_options *options, const char *values)
 232 {
 233         char *buf = xstrdup(values), *to_free = buf;
 234         int done = 0;
 235
 236         while (!done) {
 237                 int len = strcspn(buf, " ,|"), equal;
 238
 239                 done = !buf[len];
 240                 if (!len) {
 241                         buf++;
 242                         continue;
 243                 }
 244                 buf[len] = '\0';
 245
 246                 for (equal = 0;
 247                      equal < len && buf[equal] != '=' && buf[equal] != ':';
 248                      equal++)
 249                         buf[equal] = tolower(buf[equal]);
 250                 buf[equal] = '\0';
 251
 252                 if (!strcmp(buf, "skiplist")) {
 253                         if (equal == len)
 254                                 die("skiplist requires a path");
 255                         init_skiplist(options, buf + equal + 1);
 256                         buf += len + 1;
 257                         continue;
 258                 }
 259
 260                 if (equal == len)
 261                         die("Missing '=': '%s'", buf);
 262
 263                 fsck_set_msg_type(options, buf, buf + equal + 1);
 264                 buf += len + 1;
 265         }
 266         free(to_free);
 267 }
 268
 269 static void append_msg_id(struct strbuf *sb, const char *msg_id)
 270 {
 271         for (;;) {
 272                 char c = *(msg_id)++;
 273
 274                 if (!c)
 275                         break;
 276                 if (c != '_')
 277                         strbuf_addch(sb, tolower(c));
 278                 else {
 279                         assert(*msg_id);
 280                         strbuf_addch(sb, *(msg_id)++);
 281                 }
 282         }
 283
 284         strbuf_addstr(sb, ": ");
 285 }
 286
 287 __attribute__((format (printf, 4, 5)))
 288 static int report(struct fsck_options *options, struct object *object,
 289         enum fsck_msg_id id, const char *fmt, ...)
 290 {
 291         va_list ap;
 292         struct strbuf sb = STRBUF_INIT;
 293         int msg_type = fsck_msg_type(id, options), result;
 294
 295         if (msg_type == FSCK_IGNORE)
 296                 return 0;
 297
 298         if (options->skiplist && object &&
 299                         oid_array_lookup(options->skiplist, &object->oid) >= 0)
 300                 return 0;
 301
 302         if (msg_type == FSCK_FATAL)
 303                 msg_type = FSCK_ERROR;
 304         else if (msg_type == FSCK_INFO)
 305                 msg_type = FSCK_WARN;
 306
 307         append_msg_id(&sb, msg_id_info[id].id_string);
 308
 309         va_start(ap, fmt);
 310         strbuf_vaddf(&sb, fmt, ap);
 311         result = options->error_func(options, object, msg_type, sb.buf);
 312         strbuf_release(&sb);
 313         va_end(ap);
 314
 315         return result;
 316 }
 317
 318 static char *get_object_name(struct fsck_options *options, struct object *obj)
 319 {
 320         if (!options->object_names)
 321                 return NULL;
 322         return lookup_decoration(options->object_names, obj);
 323 }
 324
 325 static void put_object_name(struct fsck_options *options, struct object *obj,
 326         const char *fmt, ...)
 327 {
 328         va_list ap;
 329         struct strbuf buf = STRBUF_INIT;
 330         char *existing;
 331
 332         if (!options->object_names)
 333                 return;
 334         existing = lookup_decoration(options->object_names, obj);
 335         if (existing)
 336                 return;
 337         va_start(ap, fmt);
 338         strbuf_vaddf(&buf, fmt, ap);
 339         add_decoration(options->object_names, obj, strbuf_detach(&buf, NULL));
 340         va_end(ap);
 341 }
 342
 343 static const char *describe_object(struct fsck_options *o, struct object *obj)
 344 {
 345         static struct strbuf buf = STRBUF_INIT;
 346         char *name;
 347
 348         strbuf_reset(&buf);
 349         strbuf_addstr(&buf, oid_to_hex(&obj->oid));
 350         if (o->object_names && (name = lookup_decoration(o->object_names, obj)))
 351                 strbuf_addf(&buf, " (%s)", name);
 352
 353         return buf.buf;
 354 }
 355
 356 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
 357 {
 358         struct tree_desc desc;
 359         struct name_entry entry;
 360         int res = 0;
 361         const char *name;
 362
 363         if (parse_tree(tree))
 364                 return -1;
 365
 366         name = get_object_name(options, &tree->object);
 367         if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
 368                 return -1;
 369         while (tree_entry_gently(&desc, &entry)) {
 370                 struct object *obj;
 371                 int result;
 372
 373                 if (S_ISGITLINK(entry.mode))
 374                         continue;
 375
 376                 if (S_ISDIR(entry.mode)) {
 377                         obj = (struct object *)lookup_tree(entry.oid);
 378                         if (name && obj)
 379                                 put_object_name(options, obj, "%s%s/", name,
 380                                         entry.path);
 381                         result = options->walk(obj, OBJ_TREE, data, options);
 382                 }
 383                 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
 384                         obj = (struct object *)lookup_blob(entry.oid);
 385                         if (name && obj)
 386                                 put_object_name(options, obj, "%s%s", name,
 387                                         entry.path);
 388                         result = options->walk(obj, OBJ_BLOB, data, options);
 389                 }
 390                 else {
 391                         result = error("in tree %s: entry %s has bad mode %.6o",
 392                                         describe_object(options, &tree->object), entry.path, entry.mode);
 393                 }
 394                 if (result < 0)
 395                         return result;
 396                 if (!res)
 397                         res = result;
 398         }
 399         return res;
 400 }
 401
 402 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
 403 {
 404         int counter = 0, generation = 0, name_prefix_len = 0;
 405         struct commit_list *parents;
 406         int res;
 407         int result;
 408         const char *name;
 409
 410         if (parse_commit(commit))
 411                 return -1;
 412
 413         name = get_object_name(options, &commit->object);
 414         if (name)
 415                 put_object_name(options, &get_commit_tree(commit)->object,
 416                                 "%s:", name);
 417
 418         result = options->walk((struct object *)get_commit_tree(commit),
 419                                OBJ_TREE, data, options);
 420         if (result < 0)
 421                 return result;
 422         res = result;
 423
 424         parents = commit->parents;
 425         if (name && parents) {
 426                 int len = strlen(name), power;
 427
 428                 if (len && name[len - 1] == '^') {
 429                         generation = 1;
 430                         name_prefix_len = len - 1;
 431                 }
 432                 else { /* parse ~<generation> suffix */
 433                         for (generation = 0, power = 1;
 434                              len && isdigit(name[len - 1]);
 435                              power *= 10)
 436                                 generation += power * (name[--len] - '0');
 437                         if (power > 1 && len && name[len - 1] == '~')
 438                                 name_prefix_len = len - 1;
 439                 }
 440         }
 441
 442         while (parents) {
 443                 if (name) {
 444                         struct object *obj = &parents->item->object;
 445
 446                         if (++counter > 1)
 447                                 put_object_name(options, obj, "%s^%d",
 448                                         name, counter);
 449                         else if (generation > 0)
 450                                 put_object_name(options, obj, "%.*s~%d",
 451                                         name_prefix_len, name, generation + 1);
 452                         else
 453                                 put_object_name(options, obj, "%s^", name);
 454                 }
 455                 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
 456                 if (result < 0)
 457                         return result;
 458                 if (!res)
 459                         res = result;
 460                 parents = parents->next;
 461         }
 462         return res;
 463 }
 464
 465 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
 466 {
 467         char *name = get_object_name(options, &tag->object);
 468
 469         if (parse_tag(tag))
 470                 return -1;
 471         if (name)
 472                 put_object_name(options, tag->tagged, "%s", name);
 473         return options->walk(tag->tagged, OBJ_ANY, data, options);
 474 }
 475
 476 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
 477 {
 478         if (!obj)
 479                 return -1;
 480
 481         if (obj->type == OBJ_NONE)
 482                 parse_object(&obj->oid);
 483
 484         switch (obj->type) {
 485         case OBJ_BLOB:
 486                 return 0;
 487         case OBJ_TREE:
 488                 return fsck_walk_tree((struct tree *)obj, data, options);
 489         case OBJ_COMMIT:
 490                 return fsck_walk_commit((struct commit *)obj, data, options);
 491         case OBJ_TAG:
 492                 return fsck_walk_tag((struct tag *)obj, data, options);
 493         default:
 494                 error("Unknown object type for %s", describe_object(options, obj));
 495                 return -1;
 496         }
 497 }
 498
 499 /*
 500  * The entries in a tree are ordered in the _path_ order,
 501  * which means that a directory entry is ordered by adding
 502  * a slash to the end of it.
 503  *
 504  * So a directory called "a" is ordered _after_ a file
 505  * called "a.c", because "a/" sorts after "a.c".
 506  */
 507 #define TREE_UNORDERED (-1)
 508 #define TREE_HAS_DUPS  (-2)
 509
 510 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 511 {
 512         int len1 = strlen(name1);
 513         int len2 = strlen(name2);
 514         int len = len1 < len2 ? len1 : len2;
 515         unsigned char c1, c2;
 516         int cmp;
 517
 518         cmp = memcmp(name1, name2, len);
 519         if (cmp < 0)
 520                 return 0;
 521         if (cmp > 0)
 522                 return TREE_UNORDERED;
 523
 524         /*
 525          * Ok, the first <len> characters are the same.
 526          * Now we need to order the next one, but turn
 527          * a '\0' into a '/' for a directory entry.
 528          */
 529         c1 = name1[len];
 530         c2 = name2[len];
 531         if (!c1 && !c2)
 532                 /*
 533                  * git-write-tree used to write out a nonsense tree that has
 534                  * entries with the same name, one blob and one tree.  Make
 535                  * sure we do not have duplicate entries.
 536                  */
 537                 return TREE_HAS_DUPS;
 538         if (!c1 && S_ISDIR(mode1))
 539                 c1 = '/';
 540         if (!c2 && S_ISDIR(mode2))
 541                 c2 = '/';
 542         return c1 < c2 ? 0 : TREE_UNORDERED;
 543 }
 544
 545 static int fsck_tree(struct tree *item, struct fsck_options *options)
 546 {
 547         int retval = 0;
 548         int has_null_sha1 = 0;
 549         int has_full_path = 0;
 550         int has_empty_name = 0;
 551         int has_dot = 0;
 552         int has_dotdot = 0;
 553         int has_dotgit = 0;
 554         int has_zero_pad = 0;
 555         int has_bad_modes = 0;
 556         int has_dup_entries = 0;
 557         int not_properly_sorted = 0;
 558         struct tree_desc desc;
 559         unsigned o_mode;
 560         const char *o_name;
 561
 562         if (init_tree_desc_gently(&desc, item->buffer, item->size)) {
 563                 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 564                 return retval;
 565         }
 566
 567         o_mode = 0;
 568         o_name = NULL;
 569
 570         while (desc.size) {
 571                 unsigned mode;
 572                 const char *name, *backslash;
 573                 const struct object_id *oid;
 574
 575                 oid = tree_entry_extract(&desc, &name, &mode);
 576
 577                 has_null_sha1 |= is_null_oid(oid);
 578                 has_full_path |= !!strchr(name, '/');
 579                 has_empty_name |= !*name;
 580                 has_dot |= !strcmp(name, ".");
 581                 has_dotdot |= !strcmp(name, "..");
 582                 has_dotgit |= is_hfs_dotgit(name) || is_ntfs_dotgit(name);
 583                 has_zero_pad |= *(char *)desc.buffer == '0';
 584
 585                 if (is_hfs_dotgitmodules(name) || is_ntfs_dotgitmodules(name)) {
 586                         if (!S_ISLNK(mode))
 587                                 oidset_insert(&gitmodules_found, oid);
 588                         else
 589                                 retval += report(options, &item->object,
 590                                                  FSCK_MSG_GITMODULES_SYMLINK,
 591                                                  ".gitmodules is a symbolic link");
 592                 }
 593
 594                 if ((backslash = strchr(name, '\\'))) {
 595                         while (backslash) {
 596                                 backslash++;
 597                                 has_dotgit |= is_ntfs_dotgit(backslash);
 598                                 if (is_ntfs_dotgitmodules(backslash)) {
 599                                         if (!S_ISLNK(mode))
 600                                                 oidset_insert(&gitmodules_found, oid);
 601                                         else
 602                                                 retval += report(options, &item->object,
 603                                                                  FSCK_MSG_GITMODULES_SYMLINK,
 604                                                                  ".gitmodules is a symbolic link");
 605                                 }
 606                                 backslash = strchr(backslash, '\\');
 607                         }
 608                 }
 609
 610                 if (update_tree_entry_gently(&desc)) {
 611                         retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 612                         break;
 613                 }
 614
 615                 switch (mode) {
 616                 /*
 617                  * Standard modes..
 618                  */
 619                 case S_IFREG | 0755:
 620                 case S_IFREG | 0644:
 621                 case S_IFLNK:
 622                 case S_IFDIR:
 623                 case S_IFGITLINK:
 624                         break;
 625                 /*
 626                  * This is nonstandard, but we had a few of these
 627                  * early on when we honored the full set of mode
 628                  * bits..
 629                  */
 630                 case S_IFREG | 0664:
 631                         if (!options->strict)
 632                                 break;
 633                         /* fallthrough */
 634                 default:
 635                         has_bad_modes = 1;
 636                 }
 637
 638                 if (o_name) {
 639                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 640                         case TREE_UNORDERED:
 641                                 not_properly_sorted = 1;
 642                                 break;
 643                         case TREE_HAS_DUPS:
 644                                 has_dup_entries = 1;
 645                                 break;
 646                         default:
 647                                 break;
 648                         }
 649                 }
 650
 651                 o_mode = mode;
 652                 o_name = name;
 653         }
 654
 655         if (has_null_sha1)
 656                 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
 657         if (has_full_path)
 658                 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
 659         if (has_empty_name)
 660                 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
 661         if (has_dot)
 662                 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
 663         if (has_dotdot)
 664                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
 665         if (has_dotgit)
 666                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
 667         if (has_zero_pad)
 668                 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
 669         if (has_bad_modes)
 670                 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
 671         if (has_dup_entries)
 672                 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
 673         if (not_properly_sorted)
 674                 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
 675         return retval;
 676 }
 677
 678 static int verify_headers(const void *data, unsigned long size,
 679                           struct object *obj, struct fsck_options *options)
 680 {
 681         const char *buffer = (const char *)data;
 682         unsigned long i;
 683
 684         for (i = 0; i < size; i++) {
 685                 switch (buffer[i]) {
 686                 case '\0':
 687                         return report(options, obj,
 688                                 FSCK_MSG_NUL_IN_HEADER,
 689                                 "unterminated header: NUL at offset %ld", i);
 690                 case '\n':
 691                         if (i + 1 < size && buffer[i + 1] == '\n')
 692                                 return 0;
 693                 }
 694         }
 695
 696         /*
 697          * We did not find double-LF that separates the header
 698          * and the body.  Not having a body is not a crime but
 699          * we do want to see the terminating LF for the last header
 700          * line.
 701          */
 702         if (size && buffer[size - 1] == '\n')
 703                 return 0;
 704
 705         return report(options, obj,
 706                 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
 707 }
 708
 709 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
 710 {
 711         const char *p = *ident;
 712         char *end;
 713
 714         *ident = strchrnul(*ident, '\n');
 715         if (**ident == '\n')
 716                 (*ident)++;
 717
 718         if (*p == '<')
 719                 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 720         p += strcspn(p, "<>\n");
 721         if (*p == '>')
 722                 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
 723         if (*p != '<')
 724                 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
 725         if (p[-1] != ' ')
 726                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 727         p++;
 728         p += strcspn(p, "<>\n");
 729         if (*p != '>')
 730                 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
 731         p++;
 732         if (*p != ' ')
 733                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
 734         p++;
 735         if (*p == '0' && p[1] != ' ')
 736                 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
 737         if (date_overflows(parse_timestamp(p, &end, 10)))
 738                 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
 739         if ((end == p || *end != ' '))
 740                 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
 741         p = end + 1;
 742         if ((*p != '+' && *p != '-') ||
 743             !isdigit(p[1]) ||
 744             !isdigit(p[2]) ||
 745             !isdigit(p[3]) ||
 746             !isdigit(p[4]) ||
 747             (p[5] != '\n'))
 748                 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
 749         p += 6;
 750         return 0;
 751 }
 752
 753 static int fsck_commit_buffer(struct commit *commit, const char *buffer,
 754         unsigned long size, struct fsck_options *options)
 755 {
 756         struct object_id tree_oid, oid;
 757         struct commit_graft *graft;
 758         unsigned parent_count, parent_line_count = 0, author_count;
 759         int err;
 760         const char *buffer_begin = buffer;
 761         const char *p;
 762
 763         if (verify_headers(buffer, size, &commit->object, options))
 764                 return -1;
 765
 766         if (!skip_prefix(buffer, "tree ", &buffer))
 767                 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
 768         if (parse_oid_hex(buffer, &tree_oid, &p) || *p != '\n') {
 769                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
 770                 if (err)
 771                         return err;
 772         }
 773         buffer = p + 1;
 774         while (skip_prefix(buffer, "parent ", &buffer)) {
 775                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 776                         err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
 777                         if (err)
 778                                 return err;
 779                 }
 780                 buffer = p + 1;
 781                 parent_line_count++;
 782         }
 783         graft = lookup_commit_graft(&commit->object.oid);
 784         parent_count = commit_list_count(commit->parents);
 785         if (graft) {
 786                 if (graft->nr_parent == -1 && !parent_count)
 787                         ; /* shallow commit */
 788                 else if (graft->nr_parent != parent_count) {
 789                         err = report(options, &commit->object, FSCK_MSG_MISSING_GRAFT, "graft objects missing");
 790                         if (err)
 791                                 return err;
 792                 }
 793         } else {
 794                 if (parent_count != parent_line_count) {
 795                         err = report(options, &commit->object, FSCK_MSG_MISSING_PARENT, "parent objects missing");
 796                         if (err)
 797                                 return err;
 798                 }
 799         }
 800         author_count = 0;
 801         while (skip_prefix(buffer, "author ", &buffer)) {
 802                 author_count++;
 803                 err = fsck_ident(&buffer, &commit->object, options);
 804                 if (err)
 805                         return err;
 806         }
 807         if (author_count < 1)
 808                 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
 809         else if (author_count > 1)
 810                 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
 811         if (err)
 812                 return err;
 813         if (!skip_prefix(buffer, "committer ", &buffer))
 814                 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
 815         err = fsck_ident(&buffer, &commit->object, options);
 816         if (err)
 817                 return err;
 818         if (!get_commit_tree(commit)) {
 819                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE, "could not load commit's tree %s", oid_to_hex(&tree_oid));
 820                 if (err)
 821                         return err;
 822         }
 823         if (memchr(buffer_begin, '\0', size)) {
 824                 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
 825                              "NUL byte in the commit object body");
 826                 if (err)
 827                         return err;
 828         }
 829         return 0;
 830 }
 831
 832 static int fsck_commit(struct commit *commit, const char *data,
 833         unsigned long size, struct fsck_options *options)
 834 {
 835         const char *buffer = data ?  data : get_commit_buffer(commit, &size);
 836         int ret = fsck_commit_buffer(commit, buffer, size, options);
 837         if (!data)
 838                 unuse_commit_buffer(commit, buffer);
 839         return ret;
 840 }
 841
 842 static int fsck_tag_buffer(struct tag *tag, const char *data,
 843         unsigned long size, struct fsck_options *options)
 844 {
 845         struct object_id oid;
 846         int ret = 0;
 847         const char *buffer;
 848         char *to_free = NULL, *eol;
 849         struct strbuf sb = STRBUF_INIT;
 850         const char *p;
 851
 852         if (data)
 853                 buffer = data;
 854         else {
 855                 enum object_type type;
 856
 857                 buffer = to_free =
 858                         read_object_file(&tag->object.oid, &type, &size);
 859                 if (!buffer)
 860                         return report(options, &tag->object,
 861                                 FSCK_MSG_MISSING_TAG_OBJECT,
 862                                 "cannot read tag object");
 863
 864                 if (type != OBJ_TAG) {
 865                         ret = report(options, &tag->object,
 866                                 FSCK_MSG_TAG_OBJECT_NOT_TAG,
 867                                 "expected tag got %s",
 868                             type_name(type));
 869                         goto done;
 870                 }
 871         }
 872
 873         ret = verify_headers(buffer, size, &tag->object, options);
 874         if (ret)
 875                 goto done;
 876
 877         if (!skip_prefix(buffer, "object ", &buffer)) {
 878                 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
 879                 goto done;
 880         }
 881         if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 882                 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
 883                 if (ret)
 884                         goto done;
 885         }
 886         buffer = p + 1;
 887
 888         if (!skip_prefix(buffer, "type ", &buffer)) {
 889                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
 890                 goto done;
 891         }
 892         eol = strchr(buffer, '\n');
 893         if (!eol) {
 894                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
 895                 goto done;
 896         }
 897         if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
 898                 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
 899         if (ret)
 900                 goto done;
 901         buffer = eol + 1;
 902
 903         if (!skip_prefix(buffer, "tag ", &buffer)) {
 904                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
 905                 goto done;
 906         }
 907         eol = strchr(buffer, '\n');
 908         if (!eol) {
 909                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
 910                 goto done;
 911         }
 912         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
 913         if (check_refname_format(sb.buf, 0)) {
 914                 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
 915                            "invalid 'tag' name: %.*s",
 916                            (int)(eol - buffer), buffer);
 917                 if (ret)
 918                         goto done;
 919         }
 920         buffer = eol + 1;
 921
 922         if (!skip_prefix(buffer, "tagger ", &buffer)) {
 923                 /* early tags do not contain 'tagger' lines; warn only */
 924                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
 925                 if (ret)
 926                         goto done;
 927         }
 928         else
 929                 ret = fsck_ident(&buffer, &tag->object, options);
 930
 931 done:
 932         strbuf_release(&sb);
 933         free(to_free);
 934         return ret;
 935 }
 936
 937 static int fsck_tag(struct tag *tag, const char *data,
 938         unsigned long size, struct fsck_options *options)
 939 {
 940         struct object *tagged = tag->tagged;
 941
 942         if (!tagged)
 943                 return report(options, &tag->object, FSCK_MSG_BAD_TAG_OBJECT, "could not load tagged object");
 944
 945         return fsck_tag_buffer(tag, data, size, options);
 946 }
 947
 948 struct fsck_gitmodules_data {
 949         struct object *obj;
 950         struct fsck_options *options;
 951         int ret;
 952 };
 953
 954 static int fsck_gitmodules_fn(const char *var, const char *value, void *vdata)
 955 {
 956         struct fsck_gitmodules_data *data = vdata;
 957         const char *subsection, *key;
 958         int subsection_len;
 959         char *name;
 960
 961         if (parse_config_key(var, "submodule", &subsection, &subsection_len, &key) < 0 ||
 962             !subsection)
 963                 return 0;
 964
 965         name = xmemdupz(subsection, subsection_len);
 966         if (check_submodule_name(name) < 0)
 967                 data->ret |= report(data->options, data->obj,
 968                                     FSCK_MSG_GITMODULES_NAME,
 969                                     "disallowed submodule name: %s",
 970                                     name);
 971         if (!strcmp(key, "url") && value &&
 972             looks_like_command_line_option(value))
 973                 data->ret |= report(data->options, data->obj,
 974                                     FSCK_MSG_GITMODULES_URL,
 975                                     "disallowed submodule url: %s",
 976                                     value);
 977         if (!strcmp(key, "path") && value &&
 978             looks_like_command_line_option(value))
 979                 data->ret |= report(data->options, data->obj,
 980                                     FSCK_MSG_GITMODULES_PATH,
 981                                     "disallowed submodule path: %s",
 982                                     value);
 983         if (!strcmp(key, "update") && value &&
 984             parse_submodule_update_type(value) == SM_UPDATE_COMMAND)
 985                 data->ret |= report(data->options, data->obj,
 986                                     FSCK_MSG_GITMODULES_UPDATE,
 987                                     "disallowed submodule update setting: %s",
 988                                     value);
 989         free(name);
 990
 991         return 0;
 992 }
 993
 994 static int fsck_blob(struct blob *blob, const char *buf,
 995                      unsigned long size, struct fsck_options *options)
 996 {
 997         struct fsck_gitmodules_data data;
 998
 999         if (!oidset_contains(&gitmodules_found, &blob->object.oid))
1000                 return 0;
1001         oidset_insert(&gitmodules_done, &blob->object.oid);
1002
1003         if (!buf) {
1004                 /*
1005                  * A missing buffer here is a sign that the caller found the
1006                  * blob too gigantic to load into memory. Let's just consider
1007                  * that an error.
1008                  */
1009                 return report(options, &blob->object,
1010                               FSCK_MSG_GITMODULES_PARSE,
1011                               ".gitmodules too large to parse");
1012         }
1013
1014         data.obj = &blob->object;
1015         data.options = options;
1016         data.ret = 0;
1017         if (git_config_from_mem(fsck_gitmodules_fn, CONFIG_ORIGIN_BLOB,
1018                                 ".gitmodules", buf, size, &data))
1019                 data.ret |= report(options, &blob->object,
1020                                    FSCK_MSG_GITMODULES_PARSE,
1021                                    "could not parse gitmodules blob");
1022
1023         return data.ret;
1024 }
1025
1026 int fsck_object(struct object *obj, void *data, unsigned long size,
1027         struct fsck_options *options)
1028 {
1029         if (!obj)
1030                 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
1031
1032         if (obj->type == OBJ_BLOB)
1033                 return fsck_blob((struct blob *)obj, data, size, options);
1034         if (obj->type == OBJ_TREE)
1035                 return fsck_tree((struct tree *) obj, options);
1036         if (obj->type == OBJ_COMMIT)
1037                 return fsck_commit((struct commit *) obj, (const char *) data,
1038                         size, options);
1039         if (obj->type == OBJ_TAG)
1040                 return fsck_tag((struct tag *) obj, (const char *) data,
1041                         size, options);
1042
1043         return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
1044                           obj->type);
1045 }
1046
1047 int fsck_error_function(struct fsck_options *o,
1048         struct object *obj, int msg_type, const char *message)
1049 {
1050         if (msg_type == FSCK_WARN) {
1051                 warning("object %s: %s", describe_object(o, obj), message);
1052                 return 0;
1053         }
1054         error("object %s: %s", describe_object(o, obj), message);
1055         return 1;
1056 }
1057
1058 int fsck_finish(struct fsck_options *options)
1059 {
1060         int ret = 0;
1061         struct oidset_iter iter;
1062         const struct object_id *oid;
1063
1064         oidset_iter_init(&gitmodules_found, &iter);
1065         while ((oid = oidset_iter_next(&iter))) {
1066                 struct blob *blob;
1067                 enum object_type type;
1068                 unsigned long size;
1069                 char *buf;
1070
1071                 if (oidset_contains(&gitmodules_done, oid))
1072                         continue;
1073
1074                 blob = lookup_blob(oid);
1075                 if (!blob) {
1076                         struct object *obj = lookup_unknown_object(oid->hash);
1077                         ret |= report(options, obj,
1078                                       FSCK_MSG_GITMODULES_BLOB,
1079                                       "non-blob found at .gitmodules");
1080                         continue;
1081                 }
1082
1083                 buf = read_object_file(oid, &type, &size);
1084                 if (!buf) {
1085                         if (is_promisor_object(&blob->object.oid))
1086                                 continue;
1087                         ret |= report(options, &blob->object,
1088                                       FSCK_MSG_GITMODULES_MISSING,
1089                                       "unable to read .gitmodules blob");
1090                         continue;
1091                 }
1092
1093                 if (type == OBJ_BLOB)
1094                         ret |= fsck_blob(blob, buf, size, options);
1095                 else
1096                         ret |= report(options, &blob->object,
1097                                       FSCK_MSG_GITMODULES_BLOB,
1098                                       "non-blob found at .gitmodules");
1099                 free(buf);
1100         }
1101
1102
1103         oidset_clear(&gitmodules_found);
1104         oidset_clear(&gitmodules_done);
1105         return ret;
1106 }