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