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 #include "credential.h"
  18
  19 static struct oidset gitmodules_found = OIDSET_INIT;
  20 static struct oidset gitmodules_done = OIDSET_INIT;
  21
  22 #define FSCK_FATAL -1
  23 #define FSCK_INFO -2
  24
  25 #define FOREACH_MSG_ID(FUNC) \
  26         /* fatal errors */ \
  27         FUNC(NUL_IN_HEADER, FATAL) \
  28         FUNC(UNTERMINATED_HEADER, FATAL) \
  29         /* errors */ \
  30         FUNC(BAD_DATE, ERROR) \
  31         FUNC(BAD_DATE_OVERFLOW, ERROR) \
  32         FUNC(BAD_EMAIL, ERROR) \
  33         FUNC(BAD_NAME, ERROR) \
  34         FUNC(BAD_OBJECT_SHA1, ERROR) \
  35         FUNC(BAD_PARENT_SHA1, ERROR) \
  36         FUNC(BAD_TAG_OBJECT, ERROR) \
  37         FUNC(BAD_TIMEZONE, ERROR) \
  38         FUNC(BAD_TREE, ERROR) \
  39         FUNC(BAD_TREE_SHA1, ERROR) \
  40         FUNC(BAD_TYPE, ERROR) \
  41         FUNC(DUPLICATE_ENTRIES, ERROR) \
  42         FUNC(MISSING_AUTHOR, ERROR) \
  43         FUNC(MISSING_COMMITTER, ERROR) \
  44         FUNC(MISSING_EMAIL, ERROR) \
  45         FUNC(MISSING_GRAFT, ERROR) \
  46         FUNC(MISSING_NAME_BEFORE_EMAIL, ERROR) \
  47         FUNC(MISSING_OBJECT, ERROR) \
  48         FUNC(MISSING_PARENT, ERROR) \
  49         FUNC(MISSING_SPACE_BEFORE_DATE, ERROR) \
  50         FUNC(MISSING_SPACE_BEFORE_EMAIL, ERROR) \
  51         FUNC(MISSING_TAG, ERROR) \
  52         FUNC(MISSING_TAG_ENTRY, ERROR) \
  53         FUNC(MISSING_TAG_OBJECT, ERROR) \
  54         FUNC(MISSING_TREE, ERROR) \
  55         FUNC(MISSING_TREE_OBJECT, ERROR) \
  56         FUNC(MISSING_TYPE, ERROR) \
  57         FUNC(MISSING_TYPE_ENTRY, ERROR) \
  58         FUNC(MULTIPLE_AUTHORS, ERROR) \
  59         FUNC(TAG_OBJECT_NOT_TAG, ERROR) \
  60         FUNC(TREE_NOT_SORTED, ERROR) \
  61         FUNC(UNKNOWN_TYPE, ERROR) \
  62         FUNC(ZERO_PADDED_DATE, ERROR) \
  63         FUNC(GITMODULES_MISSING, ERROR) \
  64         FUNC(GITMODULES_BLOB, ERROR) \
  65         FUNC(GITMODULES_PARSE, ERROR) \
  66         FUNC(GITMODULES_NAME, ERROR) \
  67         FUNC(GITMODULES_SYMLINK, ERROR) \
  68         FUNC(GITMODULES_URL, ERROR) \
  69         FUNC(GITMODULES_PATH, ERROR) \
  70         FUNC(GITMODULES_UPDATE, ERROR) \
  71         /* warnings */ \
  72         FUNC(BAD_FILEMODE, WARN) \
  73         FUNC(EMPTY_NAME, WARN) \
  74         FUNC(FULL_PATHNAME, WARN) \
  75         FUNC(HAS_DOT, WARN) \
  76         FUNC(HAS_DOTDOT, WARN) \
  77         FUNC(HAS_DOTGIT, WARN) \
  78         FUNC(NULL_SHA1, WARN) \
  79         FUNC(ZERO_PADDED_FILEMODE, WARN) \
  80         FUNC(NUL_IN_COMMIT, WARN) \
  81         /* infos (reported as warnings, but ignored by default) */ \
  82         FUNC(BAD_TAG_NAME, INFO) \
  83         FUNC(MISSING_TAGGER_ENTRY, INFO)
  84
  85 #define MSG_ID(id, msg_type) FSCK_MSG_##id,
  86 enum fsck_msg_id {
  87         FOREACH_MSG_ID(MSG_ID)
  88         FSCK_MSG_MAX
  89 };
  90 #undef MSG_ID
  91
  92 #define STR(x) #x
  93 #define MSG_ID(id, msg_type) { STR(id), NULL, FSCK_##msg_type },
  94 static struct {
  95         const char *id_string;
  96         const char *downcased;
  97         int msg_type;
  98 } msg_id_info[FSCK_MSG_MAX + 1] = {
  99         FOREACH_MSG_ID(MSG_ID)
 100         { NULL, NULL, -1 }
 101 };
 102 #undef MSG_ID
 103
 104 static int parse_msg_id(const char *text)
 105 {
 106         int i;
 107
 108         if (!msg_id_info[0].downcased) {
 109                 /* convert id_string to lower case, without underscores. */
 110                 for (i = 0; i < FSCK_MSG_MAX; i++) {
 111                         const char *p = msg_id_info[i].id_string;
 112                         int len = strlen(p);
 113                         char *q = xmalloc(len);
 114
 115                         msg_id_info[i].downcased = q;
 116                         while (*p)
 117                                 if (*p == '_')
 118                                         p++;
 119                                 else
 120                                         *(q)++ = tolower(*(p)++);
 121                         *q = '\0';
 122                 }
 123         }
 124
 125         for (i = 0; i < FSCK_MSG_MAX; i++)
 126                 if (!strcmp(text, msg_id_info[i].downcased))
 127                         return i;
 128
 129         return -1;
 130 }
 131
 132 static int fsck_msg_type(enum fsck_msg_id msg_id,
 133         struct fsck_options *options)
 134 {
 135         int msg_type;
 136
 137         assert(msg_id >= 0 && msg_id < FSCK_MSG_MAX);
 138
 139         if (options->msg_type)
 140                 msg_type = options->msg_type[msg_id];
 141         else {
 142                 msg_type = msg_id_info[msg_id].msg_type;
 143                 if (options->strict && msg_type == FSCK_WARN)
 144                         msg_type = FSCK_ERROR;
 145         }
 146
 147         return msg_type;
 148 }
 149
 150 static void init_skiplist(struct fsck_options *options, const char *path)
 151 {
 152         static struct oid_array skiplist = OID_ARRAY_INIT;
 153         int sorted, fd;
 154         char buffer[GIT_MAX_HEXSZ + 1];
 155         struct object_id oid;
 156
 157         if (options->skiplist)
 158                 sorted = options->skiplist->sorted;
 159         else {
 160                 sorted = 1;
 161                 options->skiplist = &skiplist;
 162         }
 163
 164         fd = open(path, O_RDONLY);
 165         if (fd < 0)
 166                 die("Could not open skip list: %s", path);
 167         for (;;) {
 168                 const char *p;
 169                 int result = read_in_full(fd, buffer, sizeof(buffer));
 170                 if (result < 0)
 171                         die_errno("Could not read '%s'", path);
 172                 if (!result)
 173                         break;
 174                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n')
 175                         die("Invalid SHA-1: %s", buffer);
 176                 oid_array_append(&skiplist, &oid);
 177                 if (sorted && skiplist.nr > 1 &&
 178                                 oidcmp(&skiplist.oid[skiplist.nr - 2],
 179                                        &oid) > 0)
 180                         sorted = 0;
 181         }
 182         close(fd);
 183
 184         if (sorted)
 185                 skiplist.sorted = 1;
 186 }
 187
 188 static int parse_msg_type(const char *str)
 189 {
 190         if (!strcmp(str, "error"))
 191                 return FSCK_ERROR;
 192         else if (!strcmp(str, "warn"))
 193                 return FSCK_WARN;
 194         else if (!strcmp(str, "ignore"))
 195                 return FSCK_IGNORE;
 196         else
 197                 die("Unknown fsck message type: '%s'", str);
 198 }
 199
 200 int is_valid_msg_type(const char *msg_id, const char *msg_type)
 201 {
 202         if (parse_msg_id(msg_id) < 0)
 203                 return 0;
 204         parse_msg_type(msg_type);
 205         return 1;
 206 }
 207
 208 void fsck_set_msg_type(struct fsck_options *options,
 209                 const char *msg_id, const char *msg_type)
 210 {
 211         int id = parse_msg_id(msg_id), type;
 212
 213         if (id < 0)
 214                 die("Unhandled message id: %s", msg_id);
 215         type = parse_msg_type(msg_type);
 216
 217         if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
 218                 die("Cannot demote %s to %s", msg_id, msg_type);
 219
 220         if (!options->msg_type) {
 221                 int i;
 222                 int *msg_type;
 223                 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
 224                 for (i = 0; i < FSCK_MSG_MAX; i++)
 225                         msg_type[i] = fsck_msg_type(i, options);
 226                 options->msg_type = msg_type;
 227         }
 228
 229         options->msg_type[id] = type;
 230 }
 231
 232 void fsck_set_msg_types(struct fsck_options *options, const char *values)
 233 {
 234         char *buf = xstrdup(values), *to_free = buf;
 235         int done = 0;
 236
 237         while (!done) {
 238                 int len = strcspn(buf, " ,|"), equal;
 239
 240                 done = !buf[len];
 241                 if (!len) {
 242                         buf++;
 243                         continue;
 244                 }
 245                 buf[len] = '\0';
 246
 247                 for (equal = 0;
 248                      equal < len && buf[equal] != '=' && buf[equal] != ':';
 249                      equal++)
 250                         buf[equal] = tolower(buf[equal]);
 251                 buf[equal] = '\0';
 252
 253                 if (!strcmp(buf, "skiplist")) {
 254                         if (equal == len)
 255                                 die("skiplist requires a path");
 256                         init_skiplist(options, buf + equal + 1);
 257                         buf += len + 1;
 258                         continue;
 259                 }
 260
 261                 if (equal == len)
 262                         die("Missing '=': '%s'", buf);
 263
 264                 fsck_set_msg_type(options, buf, buf + equal + 1);
 265                 buf += len + 1;
 266         }
 267         free(to_free);
 268 }
 269
 270 static void append_msg_id(struct strbuf *sb, const char *msg_id)
 271 {
 272         for (;;) {
 273                 char c = *(msg_id)++;
 274
 275                 if (!c)
 276                         break;
 277                 if (c != '_')
 278                         strbuf_addch(sb, tolower(c));
 279                 else {
 280                         assert(*msg_id);
 281                         strbuf_addch(sb, *(msg_id)++);
 282                 }
 283         }
 284
 285         strbuf_addstr(sb, ": ");
 286 }
 287
 288 __attribute__((format (printf, 4, 5)))
 289 static int report(struct fsck_options *options, struct object *object,
 290         enum fsck_msg_id id, const char *fmt, ...)
 291 {
 292         va_list ap;
 293         struct strbuf sb = STRBUF_INIT;
 294         int msg_type = fsck_msg_type(id, options), result;
 295
 296         if (msg_type == FSCK_IGNORE)
 297                 return 0;
 298
 299         if (options->skiplist && object &&
 300                         oid_array_lookup(options->skiplist, &object->oid) >= 0)
 301                 return 0;
 302
 303         if (msg_type == FSCK_FATAL)
 304                 msg_type = FSCK_ERROR;
 305         else if (msg_type == FSCK_INFO)
 306                 msg_type = FSCK_WARN;
 307
 308         append_msg_id(&sb, msg_id_info[id].id_string);
 309
 310         va_start(ap, fmt);
 311         strbuf_vaddf(&sb, fmt, ap);
 312         result = options->error_func(options, object, msg_type, sb.buf);
 313         strbuf_release(&sb);
 314         va_end(ap);
 315
 316         return result;
 317 }
 318
 319 static char *get_object_name(struct fsck_options *options, struct object *obj)
 320 {
 321         if (!options->object_names)
 322                 return NULL;
 323         return lookup_decoration(options->object_names, obj);
 324 }
 325
 326 static void put_object_name(struct fsck_options *options, struct object *obj,
 327         const char *fmt, ...)
 328 {
 329         va_list ap;
 330         struct strbuf buf = STRBUF_INIT;
 331         char *existing;
 332
 333         if (!options->object_names)
 334                 return;
 335         existing = lookup_decoration(options->object_names, obj);
 336         if (existing)
 337                 return;
 338         va_start(ap, fmt);
 339         strbuf_vaddf(&buf, fmt, ap);
 340         add_decoration(options->object_names, obj, strbuf_detach(&buf, NULL));
 341         va_end(ap);
 342 }
 343
 344 static const char *describe_object(struct fsck_options *o, struct object *obj)
 345 {
 346         static struct strbuf buf = STRBUF_INIT;
 347         char *name;
 348
 349         strbuf_reset(&buf);
 350         strbuf_addstr(&buf, oid_to_hex(&obj->oid));
 351         if (o->object_names && (name = lookup_decoration(o->object_names, obj)))
 352                 strbuf_addf(&buf, " (%s)", name);
 353
 354         return buf.buf;
 355 }
 356
 357 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
 358 {
 359         struct tree_desc desc;
 360         struct name_entry entry;
 361         int res = 0;
 362         const char *name;
 363
 364         if (parse_tree(tree))
 365                 return -1;
 366
 367         name = get_object_name(options, &tree->object);
 368         if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
 369                 return -1;
 370         while (tree_entry_gently(&desc, &entry)) {
 371                 struct object *obj;
 372                 int result;
 373
 374                 if (S_ISGITLINK(entry.mode))
 375                         continue;
 376
 377                 if (S_ISDIR(entry.mode)) {
 378                         obj = (struct object *)lookup_tree(entry.oid);
 379                         if (name && obj)
 380                                 put_object_name(options, obj, "%s%s/", name,
 381                                         entry.path);
 382                         result = options->walk(obj, OBJ_TREE, data, options);
 383                 }
 384                 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
 385                         obj = (struct object *)lookup_blob(entry.oid);
 386                         if (name && obj)
 387                                 put_object_name(options, obj, "%s%s", name,
 388                                         entry.path);
 389                         result = options->walk(obj, OBJ_BLOB, data, options);
 390                 }
 391                 else {
 392                         result = error("in tree %s: entry %s has bad mode %.6o",
 393                                         describe_object(options, &tree->object), entry.path, entry.mode);
 394                 }
 395                 if (result < 0)
 396                         return result;
 397                 if (!res)
 398                         res = result;
 399         }
 400         return res;
 401 }
 402
 403 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
 404 {
 405         int counter = 0, generation = 0, name_prefix_len = 0;
 406         struct commit_list *parents;
 407         int res;
 408         int result;
 409         const char *name;
 410
 411         if (parse_commit(commit))
 412                 return -1;
 413
 414         name = get_object_name(options, &commit->object);
 415         if (name)
 416                 put_object_name(options, &get_commit_tree(commit)->object,
 417                                 "%s:", name);
 418
 419         result = options->walk((struct object *)get_commit_tree(commit),
 420                                OBJ_TREE, data, options);
 421         if (result < 0)
 422                 return result;
 423         res = result;
 424
 425         parents = commit->parents;
 426         if (name && parents) {
 427                 int len = strlen(name), power;
 428
 429                 if (len && name[len - 1] == '^') {
 430                         generation = 1;
 431                         name_prefix_len = len - 1;
 432                 }
 433                 else { /* parse ~<generation> suffix */
 434                         for (generation = 0, power = 1;
 435                              len && isdigit(name[len - 1]);
 436                              power *= 10)
 437                                 generation += power * (name[--len] - '0');
 438                         if (power > 1 && len && name[len - 1] == '~')
 439                                 name_prefix_len = len - 1;
 440                 }
 441         }
 442
 443         while (parents) {
 444                 if (name) {
 445                         struct object *obj = &parents->item->object;
 446
 447                         if (++counter > 1)
 448                                 put_object_name(options, obj, "%s^%d",
 449                                         name, counter);
 450                         else if (generation > 0)
 451                                 put_object_name(options, obj, "%.*s~%d",
 452                                         name_prefix_len, name, generation + 1);
 453                         else
 454                                 put_object_name(options, obj, "%s^", name);
 455                 }
 456                 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
 457                 if (result < 0)
 458                         return result;
 459                 if (!res)
 460                         res = result;
 461                 parents = parents->next;
 462         }
 463         return res;
 464 }
 465
 466 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
 467 {
 468         char *name = get_object_name(options, &tag->object);
 469
 470         if (parse_tag(tag))
 471                 return -1;
 472         if (name)
 473                 put_object_name(options, tag->tagged, "%s", name);
 474         return options->walk(tag->tagged, OBJ_ANY, data, options);
 475 }
 476
 477 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
 478 {
 479         if (!obj)
 480                 return -1;
 481
 482         if (obj->type == OBJ_NONE)
 483                 parse_object(&obj->oid);
 484
 485         switch (obj->type) {
 486         case OBJ_BLOB:
 487                 return 0;
 488         case OBJ_TREE:
 489                 return fsck_walk_tree((struct tree *)obj, data, options);
 490         case OBJ_COMMIT:
 491                 return fsck_walk_commit((struct commit *)obj, data, options);
 492         case OBJ_TAG:
 493                 return fsck_walk_tag((struct tag *)obj, data, options);
 494         default:
 495                 error("Unknown object type for %s", describe_object(options, obj));
 496                 return -1;
 497         }
 498 }
 499
 500 /*
 501  * The entries in a tree are ordered in the _path_ order,
 502  * which means that a directory entry is ordered by adding
 503  * a slash to the end of it.
 504  *
 505  * So a directory called "a" is ordered _after_ a file
 506  * called "a.c", because "a/" sorts after "a.c".
 507  */
 508 #define TREE_UNORDERED (-1)
 509 #define TREE_HAS_DUPS  (-2)
 510
 511 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 512 {
 513         int len1 = strlen(name1);
 514         int len2 = strlen(name2);
 515         int len = len1 < len2 ? len1 : len2;
 516         unsigned char c1, c2;
 517         int cmp;
 518
 519         cmp = memcmp(name1, name2, len);
 520         if (cmp < 0)
 521                 return 0;
 522         if (cmp > 0)
 523                 return TREE_UNORDERED;
 524
 525         /*
 526          * Ok, the first <len> characters are the same.
 527          * Now we need to order the next one, but turn
 528          * a '\0' into a '/' for a directory entry.
 529          */
 530         c1 = name1[len];
 531         c2 = name2[len];
 532         if (!c1 && !c2)
 533                 /*
 534                  * git-write-tree used to write out a nonsense tree that has
 535                  * entries with the same name, one blob and one tree.  Make
 536                  * sure we do not have duplicate entries.
 537                  */
 538                 return TREE_HAS_DUPS;
 539         if (!c1 && S_ISDIR(mode1))
 540                 c1 = '/';
 541         if (!c2 && S_ISDIR(mode2))
 542                 c2 = '/';
 543         return c1 < c2 ? 0 : TREE_UNORDERED;
 544 }
 545
 546 static int fsck_tree(struct tree *item, struct fsck_options *options)
 547 {
 548         int retval = 0;
 549         int has_null_sha1 = 0;
 550         int has_full_path = 0;
 551         int has_empty_name = 0;
 552         int has_dot = 0;
 553         int has_dotdot = 0;
 554         int has_dotgit = 0;
 555         int has_zero_pad = 0;
 556         int has_bad_modes = 0;
 557         int has_dup_entries = 0;
 558         int not_properly_sorted = 0;
 559         struct tree_desc desc;
 560         unsigned o_mode;
 561         const char *o_name;
 562
 563         if (init_tree_desc_gently(&desc, item->buffer, item->size)) {
 564                 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 565                 return retval;
 566         }
 567
 568         o_mode = 0;
 569         o_name = NULL;
 570
 571         while (desc.size) {
 572                 unsigned mode;
 573                 const char *name, *backslash;
 574                 const struct object_id *oid;
 575
 576                 oid = tree_entry_extract(&desc, &name, &mode);
 577
 578                 has_null_sha1 |= is_null_oid(oid);
 579                 has_full_path |= !!strchr(name, '/');
 580                 has_empty_name |= !*name;
 581                 has_dot |= !strcmp(name, ".");
 582                 has_dotdot |= !strcmp(name, "..");
 583                 has_dotgit |= is_hfs_dotgit(name) || is_ntfs_dotgit(name);
 584                 has_zero_pad |= *(char *)desc.buffer == '0';
 585
 586                 if (is_hfs_dotgitmodules(name) || is_ntfs_dotgitmodules(name)) {
 587                         if (!S_ISLNK(mode))
 588                                 oidset_insert(&gitmodules_found, oid);
 589                         else
 590                                 retval += report(options, &item->object,
 591                                                  FSCK_MSG_GITMODULES_SYMLINK,
 592                                                  ".gitmodules is a symbolic link");
 593                 }
 594
 595                 if ((backslash = strchr(name, '\\'))) {
 596                         while (backslash) {
 597                                 backslash++;
 598                                 has_dotgit |= is_ntfs_dotgit(backslash);
 599                                 if (is_ntfs_dotgitmodules(backslash)) {
 600                                         if (!S_ISLNK(mode))
 601                                                 oidset_insert(&gitmodules_found, oid);
 602                                         else
 603                                                 retval += report(options, &item->object,
 604                                                                  FSCK_MSG_GITMODULES_SYMLINK,
 605                                                                  ".gitmodules is a symbolic link");
 606                                 }
 607                                 backslash = strchr(backslash, '\\');
 608                         }
 609                 }
 610
 611                 if (update_tree_entry_gently(&desc)) {
 612                         retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 613                         break;
 614                 }
 615
 616                 switch (mode) {
 617                 /*
 618                  * Standard modes..
 619                  */
 620                 case S_IFREG | 0755:
 621                 case S_IFREG | 0644:
 622                 case S_IFLNK:
 623                 case S_IFDIR:
 624                 case S_IFGITLINK:
 625                         break;
 626                 /*
 627                  * This is nonstandard, but we had a few of these
 628                  * early on when we honored the full set of mode
 629                  * bits..
 630                  */
 631                 case S_IFREG | 0664:
 632                         if (!options->strict)
 633                                 break;
 634                         /* fallthrough */
 635                 default:
 636                         has_bad_modes = 1;
 637                 }
 638
 639                 if (o_name) {
 640                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 641                         case TREE_UNORDERED:
 642                                 not_properly_sorted = 1;
 643                                 break;
 644                         case TREE_HAS_DUPS:
 645                                 has_dup_entries = 1;
 646                                 break;
 647                         default:
 648                                 break;
 649                         }
 650                 }
 651
 652                 o_mode = mode;
 653                 o_name = name;
 654         }
 655
 656         if (has_null_sha1)
 657                 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
 658         if (has_full_path)
 659                 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
 660         if (has_empty_name)
 661                 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
 662         if (has_dot)
 663                 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
 664         if (has_dotdot)
 665                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
 666         if (has_dotgit)
 667                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
 668         if (has_zero_pad)
 669                 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
 670         if (has_bad_modes)
 671                 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
 672         if (has_dup_entries)
 673                 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
 674         if (not_properly_sorted)
 675                 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
 676         return retval;
 677 }
 678
 679 static int verify_headers(const void *data, unsigned long size,
 680                           struct object *obj, struct fsck_options *options)
 681 {
 682         const char *buffer = (const char *)data;
 683         unsigned long i;
 684
 685         for (i = 0; i < size; i++) {
 686                 switch (buffer[i]) {
 687                 case '\0':
 688                         return report(options, obj,
 689                                 FSCK_MSG_NUL_IN_HEADER,
 690                                 "unterminated header: NUL at offset %ld", i);
 691                 case '\n':
 692                         if (i + 1 < size && buffer[i + 1] == '\n')
 693                                 return 0;
 694                 }
 695         }
 696
 697         /*
 698          * We did not find double-LF that separates the header
 699          * and the body.  Not having a body is not a crime but
 700          * we do want to see the terminating LF for the last header
 701          * line.
 702          */
 703         if (size && buffer[size - 1] == '\n')
 704                 return 0;
 705
 706         return report(options, obj,
 707                 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
 708 }
 709
 710 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
 711 {
 712         const char *p = *ident;
 713         char *end;
 714
 715         *ident = strchrnul(*ident, '\n');
 716         if (**ident == '\n')
 717                 (*ident)++;
 718
 719         if (*p == '<')
 720                 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 721         p += strcspn(p, "<>\n");
 722         if (*p == '>')
 723                 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
 724         if (*p != '<')
 725                 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
 726         if (p[-1] != ' ')
 727                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 728         p++;
 729         p += strcspn(p, "<>\n");
 730         if (*p != '>')
 731                 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
 732         p++;
 733         if (*p != ' ')
 734                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
 735         p++;
 736         if (*p == '0' && p[1] != ' ')
 737                 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
 738         if (date_overflows(parse_timestamp(p, &end, 10)))
 739                 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
 740         if ((end == p || *end != ' '))
 741                 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
 742         p = end + 1;
 743         if ((*p != '+' && *p != '-') ||
 744             !isdigit(p[1]) ||
 745             !isdigit(p[2]) ||
 746             !isdigit(p[3]) ||
 747             !isdigit(p[4]) ||
 748             (p[5] != '\n'))
 749                 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
 750         p += 6;
 751         return 0;
 752 }
 753
 754 static int fsck_commit_buffer(struct commit *commit, const char *buffer,
 755         unsigned long size, struct fsck_options *options)
 756 {
 757         struct object_id tree_oid, oid;
 758         struct commit_graft *graft;
 759         unsigned parent_count, parent_line_count = 0, author_count;
 760         int err;
 761         const char *buffer_begin = buffer;
 762         const char *p;
 763
 764         if (verify_headers(buffer, size, &commit->object, options))
 765                 return -1;
 766
 767         if (!skip_prefix(buffer, "tree ", &buffer))
 768                 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
 769         if (parse_oid_hex(buffer, &tree_oid, &p) || *p != '\n') {
 770                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
 771                 if (err)
 772                         return err;
 773         }
 774         buffer = p + 1;
 775         while (skip_prefix(buffer, "parent ", &buffer)) {
 776                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 777                         err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
 778                         if (err)
 779                                 return err;
 780                 }
 781                 buffer = p + 1;
 782                 parent_line_count++;
 783         }
 784         graft = lookup_commit_graft(&commit->object.oid);
 785         parent_count = commit_list_count(commit->parents);
 786         if (graft) {
 787                 if (graft->nr_parent == -1 && !parent_count)
 788                         ; /* shallow commit */
 789                 else if (graft->nr_parent != parent_count) {
 790                         err = report(options, &commit->object, FSCK_MSG_MISSING_GRAFT, "graft objects missing");
 791                         if (err)
 792                                 return err;
 793                 }
 794         } else {
 795                 if (parent_count != parent_line_count) {
 796                         err = report(options, &commit->object, FSCK_MSG_MISSING_PARENT, "parent objects missing");
 797                         if (err)
 798                                 return err;
 799                 }
 800         }
 801         author_count = 0;
 802         while (skip_prefix(buffer, "author ", &buffer)) {
 803                 author_count++;
 804                 err = fsck_ident(&buffer, &commit->object, options);
 805                 if (err)
 806                         return err;
 807         }
 808         if (author_count < 1)
 809                 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
 810         else if (author_count > 1)
 811                 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
 812         if (err)
 813                 return err;
 814         if (!skip_prefix(buffer, "committer ", &buffer))
 815                 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
 816         err = fsck_ident(&buffer, &commit->object, options);
 817         if (err)
 818                 return err;
 819         if (!get_commit_tree(commit)) {
 820                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE, "could not load commit's tree %s", oid_to_hex(&tree_oid));
 821                 if (err)
 822                         return err;
 823         }
 824         if (memchr(buffer_begin, '\0', size)) {
 825                 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
 826                              "NUL byte in the commit object body");
 827                 if (err)
 828                         return err;
 829         }
 830         return 0;
 831 }
 832
 833 static int fsck_commit(struct commit *commit, const char *data,
 834         unsigned long size, struct fsck_options *options)
 835 {
 836         const char *buffer = data ?  data : get_commit_buffer(commit, &size);
 837         int ret = fsck_commit_buffer(commit, buffer, size, options);
 838         if (!data)
 839                 unuse_commit_buffer(commit, buffer);
 840         return ret;
 841 }
 842
 843 static int fsck_tag_buffer(struct tag *tag, const char *data,
 844         unsigned long size, struct fsck_options *options)
 845 {
 846         struct object_id oid;
 847         int ret = 0;
 848         const char *buffer;
 849         char *to_free = NULL, *eol;
 850         struct strbuf sb = STRBUF_INIT;
 851         const char *p;
 852
 853         if (data)
 854                 buffer = data;
 855         else {
 856                 enum object_type type;
 857
 858                 buffer = to_free =
 859                         read_object_file(&tag->object.oid, &type, &size);
 860                 if (!buffer)
 861                         return report(options, &tag->object,
 862                                 FSCK_MSG_MISSING_TAG_OBJECT,
 863                                 "cannot read tag object");
 864
 865                 if (type != OBJ_TAG) {
 866                         ret = report(options, &tag->object,
 867                                 FSCK_MSG_TAG_OBJECT_NOT_TAG,
 868                                 "expected tag got %s",
 869                             type_name(type));
 870                         goto done;
 871                 }
 872         }
 873
 874         ret = verify_headers(buffer, size, &tag->object, options);
 875         if (ret)
 876                 goto done;
 877
 878         if (!skip_prefix(buffer, "object ", &buffer)) {
 879                 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
 880                 goto done;
 881         }
 882         if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 883                 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
 884                 if (ret)
 885                         goto done;
 886         }
 887         buffer = p + 1;
 888
 889         if (!skip_prefix(buffer, "type ", &buffer)) {
 890                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
 891                 goto done;
 892         }
 893         eol = strchr(buffer, '\n');
 894         if (!eol) {
 895                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
 896                 goto done;
 897         }
 898         if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
 899                 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
 900         if (ret)
 901                 goto done;
 902         buffer = eol + 1;
 903
 904         if (!skip_prefix(buffer, "tag ", &buffer)) {
 905                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
 906                 goto done;
 907         }
 908         eol = strchr(buffer, '\n');
 909         if (!eol) {
 910                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
 911                 goto done;
 912         }
 913         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
 914         if (check_refname_format(sb.buf, 0)) {
 915                 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
 916                            "invalid 'tag' name: %.*s",
 917                            (int)(eol - buffer), buffer);
 918                 if (ret)
 919                         goto done;
 920         }
 921         buffer = eol + 1;
 922
 923         if (!skip_prefix(buffer, "tagger ", &buffer)) {
 924                 /* early tags do not contain 'tagger' lines; warn only */
 925                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
 926                 if (ret)
 927                         goto done;
 928         }
 929         else
 930                 ret = fsck_ident(&buffer, &tag->object, options);
 931
 932 done:
 933         strbuf_release(&sb);
 934         free(to_free);
 935         return ret;
 936 }
 937
 938 static int fsck_tag(struct tag *tag, const char *data,
 939         unsigned long size, struct fsck_options *options)
 940 {
 941         struct object *tagged = tag->tagged;
 942
 943         if (!tagged)
 944                 return report(options, &tag->object, FSCK_MSG_BAD_TAG_OBJECT, "could not load tagged object");
 945
 946         return fsck_tag_buffer(tag, data, size, options);
 947 }
 948
 949 static int check_submodule_url(const char *url)
 950 {
 951         struct credential c = CREDENTIAL_INIT;
 952         int ret;
 953
 954         if (looks_like_command_line_option(url))
 955                 return -1;
 956
 957         ret = credential_from_url_gently(&c, url, 1);
 958         credential_clear(&c);
 959         return ret;
 960 }
 961
 962 struct fsck_gitmodules_data {
 963         struct object *obj;
 964         struct fsck_options *options;
 965         int ret;
 966 };
 967
 968 static int fsck_gitmodules_fn(const char *var, const char *value, void *vdata)
 969 {
 970         struct fsck_gitmodules_data *data = vdata;
 971         const char *subsection, *key;
 972         int subsection_len;
 973         char *name;
 974
 975         if (parse_config_key(var, "submodule", &subsection, &subsection_len, &key) < 0 ||
 976             !subsection)
 977                 return 0;
 978
 979         name = xmemdupz(subsection, subsection_len);
 980         if (check_submodule_name(name) < 0)
 981                 data->ret |= report(data->options, data->obj,
 982                                     FSCK_MSG_GITMODULES_NAME,
 983                                     "disallowed submodule name: %s",
 984                                     name);
 985         if (!strcmp(key, "url") && value &&
 986             check_submodule_url(value) < 0)
 987                 data->ret |= report(data->options, data->obj,
 988                                     FSCK_MSG_GITMODULES_URL,
 989                                     "disallowed submodule url: %s",
 990                                     value);
 991         if (!strcmp(key, "path") && value &&
 992             looks_like_command_line_option(value))
 993                 data->ret |= report(data->options, data->obj,
 994                                     FSCK_MSG_GITMODULES_PATH,
 995                                     "disallowed submodule path: %s",
 996                                     value);
 997         if (!strcmp(key, "update") && value &&
 998             parse_submodule_update_type(value) == SM_UPDATE_COMMAND)
 999                 data->ret |= report(data->options, data->obj,
1000                                     FSCK_MSG_GITMODULES_UPDATE,
1001                                     "disallowed submodule update setting: %s",
1002                                     value);
1003         free(name);
1004
1005         return 0;
1006 }
1007
1008 static int fsck_blob(struct blob *blob, const char *buf,
1009                      unsigned long size, struct fsck_options *options)
1010 {
1011         struct fsck_gitmodules_data data;
1012
1013         if (!oidset_contains(&gitmodules_found, &blob->object.oid))
1014                 return 0;
1015         oidset_insert(&gitmodules_done, &blob->object.oid);
1016
1017         if (!buf) {
1018                 /*
1019                  * A missing buffer here is a sign that the caller found the
1020                  * blob too gigantic to load into memory. Let's just consider
1021                  * that an error.
1022                  */
1023                 return report(options, &blob->object,
1024                               FSCK_MSG_GITMODULES_PARSE,
1025                               ".gitmodules too large to parse");
1026         }
1027
1028         data.obj = &blob->object;
1029         data.options = options;
1030         data.ret = 0;
1031         if (git_config_from_mem(fsck_gitmodules_fn, CONFIG_ORIGIN_BLOB,
1032                                 ".gitmodules", buf, size, &data))
1033                 data.ret |= report(options, &blob->object,
1034                                    FSCK_MSG_GITMODULES_PARSE,
1035                                    "could not parse gitmodules blob");
1036
1037         return data.ret;
1038 }
1039
1040 int fsck_object(struct object *obj, void *data, unsigned long size,
1041         struct fsck_options *options)
1042 {
1043         if (!obj)
1044                 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
1045
1046         if (obj->type == OBJ_BLOB)
1047                 return fsck_blob((struct blob *)obj, data, size, options);
1048         if (obj->type == OBJ_TREE)
1049                 return fsck_tree((struct tree *) obj, options);
1050         if (obj->type == OBJ_COMMIT)
1051                 return fsck_commit((struct commit *) obj, (const char *) data,
1052                         size, options);
1053         if (obj->type == OBJ_TAG)
1054                 return fsck_tag((struct tag *) obj, (const char *) data,
1055                         size, options);
1056
1057         return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
1058                           obj->type);
1059 }
1060
1061 int fsck_error_function(struct fsck_options *o,
1062         struct object *obj, int msg_type, const char *message)
1063 {
1064         if (msg_type == FSCK_WARN) {
1065                 warning("object %s: %s", describe_object(o, obj), message);
1066                 return 0;
1067         }
1068         error("object %s: %s", describe_object(o, obj), message);
1069         return 1;
1070 }
1071
1072 int fsck_finish(struct fsck_options *options)
1073 {
1074         int ret = 0;
1075         struct oidset_iter iter;
1076         const struct object_id *oid;
1077
1078         oidset_iter_init(&gitmodules_found, &iter);
1079         while ((oid = oidset_iter_next(&iter))) {
1080                 struct blob *blob;
1081                 enum object_type type;
1082                 unsigned long size;
1083                 char *buf;
1084
1085                 if (oidset_contains(&gitmodules_done, oid))
1086                         continue;
1087
1088                 blob = lookup_blob(oid);
1089                 if (!blob) {
1090                         struct object *obj = lookup_unknown_object(oid->hash);
1091                         ret |= report(options, obj,
1092                                       FSCK_MSG_GITMODULES_BLOB,
1093                                       "non-blob found at .gitmodules");
1094                         continue;
1095                 }
1096
1097                 buf = read_object_file(oid, &type, &size);
1098                 if (!buf) {
1099                         if (is_promisor_object(&blob->object.oid))
1100                                 continue;
1101                         ret |= report(options, &blob->object,
1102                                       FSCK_MSG_GITMODULES_MISSING,
1103                                       "unable to read .gitmodules blob");
1104                         continue;
1105                 }
1106
1107                 if (type == OBJ_BLOB)
1108                         ret |= fsck_blob(blob, buf, size, options);
1109                 else
1110                         ret |= report(options, &blob->object,
1111                                       FSCK_MSG_GITMODULES_BLOB,
1112                                       "non-blob found at .gitmodules");
1113                 free(buf);
1114         }
1115
1116
1117         oidset_clear(&gitmodules_found);
1118         oidset_clear(&gitmodules_done);
1119         return ret;
1120 }