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