git.oblomov.eu Git - git/blob - fsck.c

   1 #include "cache.h"
   2 #include "object-store.h"
   3 #include "repository.h"
   4 #include "object.h"
   5 #include "blob.h"
   6 #include "tree.h"
   7 #include "tree-walk.h"
   8 #include "commit.h"
   9 #include "tag.h"
  10 #include "fsck.h"
  11 #include "refs.h"
  12 #include "utf8.h"
  13 #include "decorate.h"
  14 #include "oidset.h"
  15 #include "packfile.h"
  16 #include "submodule-config.h"
  17 #include "config.h"
  18 #include "help.h"
  19
  20 static struct oidset gitmodules_found = OIDSET_INIT;
  21 static struct oidset gitmodules_done = OIDSET_INIT;
  22
  23 #define FSCK_FATAL -1
  24 #define FSCK_INFO -2
  25
  26 #define FOREACH_MSG_ID(FUNC) \
  27         /* fatal errors */ \
  28         FUNC(NUL_IN_HEADER, FATAL) \
  29         FUNC(UNTERMINATED_HEADER, FATAL) \
  30         /* errors */ \
  31         FUNC(BAD_DATE, ERROR) \
  32         FUNC(BAD_DATE_OVERFLOW, ERROR) \
  33         FUNC(BAD_EMAIL, ERROR) \
  34         FUNC(BAD_NAME, ERROR) \
  35         FUNC(BAD_OBJECT_SHA1, ERROR) \
  36         FUNC(BAD_PARENT_SHA1, ERROR) \
  37         FUNC(BAD_TAG_OBJECT, ERROR) \
  38         FUNC(BAD_TIMEZONE, ERROR) \
  39         FUNC(BAD_TREE, ERROR) \
  40         FUNC(BAD_TREE_SHA1, ERROR) \
  41         FUNC(BAD_TYPE, ERROR) \
  42         FUNC(DUPLICATE_ENTRIES, ERROR) \
  43         FUNC(MISSING_AUTHOR, ERROR) \
  44         FUNC(MISSING_COMMITTER, ERROR) \
  45         FUNC(MISSING_EMAIL, ERROR) \
  46         FUNC(MISSING_NAME_BEFORE_EMAIL, ERROR) \
  47         FUNC(MISSING_OBJECT, ERROR) \
  48         FUNC(MISSING_SPACE_BEFORE_DATE, ERROR) \
  49         FUNC(MISSING_SPACE_BEFORE_EMAIL, ERROR) \
  50         FUNC(MISSING_TAG, ERROR) \
  51         FUNC(MISSING_TAG_ENTRY, ERROR) \
  52         FUNC(MISSING_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_LARGE, 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(GITMODULES_PARSE, INFO) \
  81         FUNC(BAD_TAG_NAME, INFO) \
  82         FUNC(MISSING_TAGGER_ENTRY, INFO)
  83
  84 #define MSG_ID(id, msg_type) FSCK_MSG_##id,
  85 enum fsck_msg_id {
  86         FOREACH_MSG_ID(MSG_ID)
  87         FSCK_MSG_MAX
  88 };
  89 #undef MSG_ID
  90
  91 #define STR(x) #x
  92 #define MSG_ID(id, msg_type) { STR(id), NULL, NULL, FSCK_##msg_type },
  93 static struct {
  94         const char *id_string;
  95         const char *downcased;
  96         const char *camelcased;
  97         int msg_type;
  98 } msg_id_info[FSCK_MSG_MAX + 1] = {
  99         FOREACH_MSG_ID(MSG_ID)
 100         { NULL, NULL, NULL, -1 }
 101 };
 102 #undef MSG_ID
 103
 104 static void prepare_msg_ids(void)
 105 {
 106         int i;
 107
 108         if (msg_id_info[0].downcased)
 109                 return;
 110
 111         /* convert id_string to lower case, without underscores. */
 112         for (i = 0; i < FSCK_MSG_MAX; i++) {
 113                 const char *p = msg_id_info[i].id_string;
 114                 int len = strlen(p);
 115                 char *q = xmalloc(len);
 116
 117                 msg_id_info[i].downcased = q;
 118                 while (*p)
 119                         if (*p == '_')
 120                                 p++;
 121                         else
 122                                 *(q)++ = tolower(*(p)++);
 123                 *q = '\0';
 124
 125                 p = msg_id_info[i].id_string;
 126                 q = xmalloc(len);
 127                 msg_id_info[i].camelcased = q;
 128                 while (*p) {
 129                         if (*p == '_') {
 130                                 p++;
 131                                 if (*p)
 132                                         *q++ = *p++;
 133                         } else {
 134                                 *q++ = tolower(*p++);
 135                         }
 136                 }
 137                 *q = '\0';
 138         }
 139 }
 140
 141 static int parse_msg_id(const char *text)
 142 {
 143         int i;
 144
 145         prepare_msg_ids();
 146
 147         for (i = 0; i < FSCK_MSG_MAX; i++)
 148                 if (!strcmp(text, msg_id_info[i].downcased))
 149                         return i;
 150
 151         return -1;
 152 }
 153
 154 void list_config_fsck_msg_ids(struct string_list *list, const char *prefix)
 155 {
 156         int i;
 157
 158         prepare_msg_ids();
 159
 160         for (i = 0; i < FSCK_MSG_MAX; i++)
 161                 list_config_item(list, prefix, msg_id_info[i].camelcased);
 162 }
 163
 164 static int fsck_msg_type(enum fsck_msg_id msg_id,
 165         struct fsck_options *options)
 166 {
 167         int msg_type;
 168
 169         assert(msg_id >= 0 && msg_id < FSCK_MSG_MAX);
 170
 171         if (options->msg_type)
 172                 msg_type = options->msg_type[msg_id];
 173         else {
 174                 msg_type = msg_id_info[msg_id].msg_type;
 175                 if (options->strict && msg_type == FSCK_WARN)
 176                         msg_type = FSCK_ERROR;
 177         }
 178
 179         return msg_type;
 180 }
 181
 182 static int parse_msg_type(const char *str)
 183 {
 184         if (!strcmp(str, "error"))
 185                 return FSCK_ERROR;
 186         else if (!strcmp(str, "warn"))
 187                 return FSCK_WARN;
 188         else if (!strcmp(str, "ignore"))
 189                 return FSCK_IGNORE;
 190         else
 191                 die("Unknown fsck message type: '%s'", str);
 192 }
 193
 194 int is_valid_msg_type(const char *msg_id, const char *msg_type)
 195 {
 196         if (parse_msg_id(msg_id) < 0)
 197                 return 0;
 198         parse_msg_type(msg_type);
 199         return 1;
 200 }
 201
 202 void fsck_set_msg_type(struct fsck_options *options,
 203                 const char *msg_id, const char *msg_type)
 204 {
 205         int id = parse_msg_id(msg_id), type;
 206
 207         if (id < 0)
 208                 die("Unhandled message id: %s", msg_id);
 209         type = parse_msg_type(msg_type);
 210
 211         if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
 212                 die("Cannot demote %s to %s", msg_id, msg_type);
 213
 214         if (!options->msg_type) {
 215                 int i;
 216                 int *msg_type;
 217                 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
 218                 for (i = 0; i < FSCK_MSG_MAX; i++)
 219                         msg_type[i] = fsck_msg_type(i, options);
 220                 options->msg_type = msg_type;
 221         }
 222
 223         options->msg_type[id] = type;
 224 }
 225
 226 void fsck_set_msg_types(struct fsck_options *options, const char *values)
 227 {
 228         char *buf = xstrdup(values), *to_free = buf;
 229         int done = 0;
 230
 231         while (!done) {
 232                 int len = strcspn(buf, " ,|"), equal;
 233
 234                 done = !buf[len];
 235                 if (!len) {
 236                         buf++;
 237                         continue;
 238                 }
 239                 buf[len] = '\0';
 240
 241                 for (equal = 0;
 242                      equal < len && buf[equal] != '=' && buf[equal] != ':';
 243                      equal++)
 244                         buf[equal] = tolower(buf[equal]);
 245                 buf[equal] = '\0';
 246
 247                 if (!strcmp(buf, "skiplist")) {
 248                         if (equal == len)
 249                                 die("skiplist requires a path");
 250                         oidset_parse_file(&options->skiplist, buf + equal + 1);
 251                         buf += len + 1;
 252                         continue;
 253                 }
 254
 255                 if (equal == len)
 256                         die("Missing '=': '%s'", buf);
 257
 258                 fsck_set_msg_type(options, buf, buf + equal + 1);
 259                 buf += len + 1;
 260         }
 261         free(to_free);
 262 }
 263
 264 static void append_msg_id(struct strbuf *sb, const char *msg_id)
 265 {
 266         for (;;) {
 267                 char c = *(msg_id)++;
 268
 269                 if (!c)
 270                         break;
 271                 if (c != '_')
 272                         strbuf_addch(sb, tolower(c));
 273                 else {
 274                         assert(*msg_id);
 275                         strbuf_addch(sb, *(msg_id)++);
 276                 }
 277         }
 278
 279         strbuf_addstr(sb, ": ");
 280 }
 281
 282 static int object_on_skiplist(struct fsck_options *opts, struct object *obj)
 283 {
 284         return opts && obj && oidset_contains(&opts->skiplist, &obj->oid);
 285 }
 286
 287 __attribute__((format (printf, 4, 5)))
 288 static int report(struct fsck_options *options, struct object *object,
 289         enum fsck_msg_id id, const char *fmt, ...)
 290 {
 291         va_list ap;
 292         struct strbuf sb = STRBUF_INIT;
 293         int msg_type = fsck_msg_type(id, options), result;
 294
 295         if (msg_type == FSCK_IGNORE)
 296                 return 0;
 297
 298         if (object_on_skiplist(options, object))
 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(the_repository, &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(the_repository, &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++)
 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(the_repository, &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 short 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         unsigned author_count;
 741         int err;
 742         const char *buffer_begin = buffer;
 743         const char *p;
 744
 745         if (verify_headers(buffer, size, &commit->object, options))
 746                 return -1;
 747
 748         if (!skip_prefix(buffer, "tree ", &buffer))
 749                 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
 750         if (parse_oid_hex(buffer, &tree_oid, &p) || *p != '\n') {
 751                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
 752                 if (err)
 753                         return err;
 754         }
 755         buffer = p + 1;
 756         while (skip_prefix(buffer, "parent ", &buffer)) {
 757                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 758                         err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
 759                         if (err)
 760                                 return err;
 761                 }
 762                 buffer = p + 1;
 763         }
 764         author_count = 0;
 765         while (skip_prefix(buffer, "author ", &buffer)) {
 766                 author_count++;
 767                 err = fsck_ident(&buffer, &commit->object, options);
 768                 if (err)
 769                         return err;
 770         }
 771         if (author_count < 1)
 772                 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
 773         else if (author_count > 1)
 774                 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
 775         if (err)
 776                 return err;
 777         if (!skip_prefix(buffer, "committer ", &buffer))
 778                 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
 779         err = fsck_ident(&buffer, &commit->object, options);
 780         if (err)
 781                 return err;
 782         if (memchr(buffer_begin, '\0', size)) {
 783                 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
 784                              "NUL byte in the commit object body");
 785                 if (err)
 786                         return err;
 787         }
 788         return 0;
 789 }
 790
 791 static int fsck_commit(struct commit *commit, const char *data,
 792         unsigned long size, struct fsck_options *options)
 793 {
 794         const char *buffer = data ?  data : get_commit_buffer(commit, &size);
 795         int ret = fsck_commit_buffer(commit, buffer, size, options);
 796         if (!data)
 797                 unuse_commit_buffer(commit, buffer);
 798         return ret;
 799 }
 800
 801 static int fsck_tag(struct tag *tag, const char *data,
 802                     unsigned long size, struct fsck_options *options)
 803 {
 804         struct object_id oid;
 805         int ret = 0;
 806         const char *buffer;
 807         char *to_free = NULL, *eol;
 808         struct strbuf sb = STRBUF_INIT;
 809         const char *p;
 810
 811         if (data)
 812                 buffer = data;
 813         else {
 814                 enum object_type type;
 815
 816                 buffer = to_free =
 817                         read_object_file(&tag->object.oid, &type, &size);
 818                 if (!buffer)
 819                         return report(options, &tag->object,
 820                                 FSCK_MSG_MISSING_TAG_OBJECT,
 821                                 "cannot read tag object");
 822
 823                 if (type != OBJ_TAG) {
 824                         ret = report(options, &tag->object,
 825                                 FSCK_MSG_TAG_OBJECT_NOT_TAG,
 826                                 "expected tag got %s",
 827                             type_name(type));
 828                         goto done;
 829                 }
 830         }
 831
 832         ret = verify_headers(buffer, size, &tag->object, options);
 833         if (ret)
 834                 goto done;
 835
 836         if (!skip_prefix(buffer, "object ", &buffer)) {
 837                 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
 838                 goto done;
 839         }
 840         if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
 841                 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
 842                 if (ret)
 843                         goto done;
 844         }
 845         buffer = p + 1;
 846
 847         if (!skip_prefix(buffer, "type ", &buffer)) {
 848                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
 849                 goto done;
 850         }
 851         eol = strchr(buffer, '\n');
 852         if (!eol) {
 853                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
 854                 goto done;
 855         }
 856         if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
 857                 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
 858         if (ret)
 859                 goto done;
 860         buffer = eol + 1;
 861
 862         if (!skip_prefix(buffer, "tag ", &buffer)) {
 863                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
 864                 goto done;
 865         }
 866         eol = strchr(buffer, '\n');
 867         if (!eol) {
 868                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
 869                 goto done;
 870         }
 871         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
 872         if (check_refname_format(sb.buf, 0)) {
 873                 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
 874                            "invalid 'tag' name: %.*s",
 875                            (int)(eol - buffer), buffer);
 876                 if (ret)
 877                         goto done;
 878         }
 879         buffer = eol + 1;
 880
 881         if (!skip_prefix(buffer, "tagger ", &buffer)) {
 882                 /* early tags do not contain 'tagger' lines; warn only */
 883                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
 884                 if (ret)
 885                         goto done;
 886         }
 887         else
 888                 ret = fsck_ident(&buffer, &tag->object, options);
 889
 890 done:
 891         strbuf_release(&sb);
 892         free(to_free);
 893         return ret;
 894 }
 895
 896 struct fsck_gitmodules_data {
 897         struct object *obj;
 898         struct fsck_options *options;
 899         int ret;
 900 };
 901
 902 static int fsck_gitmodules_fn(const char *var, const char *value, void *vdata)
 903 {
 904         struct fsck_gitmodules_data *data = vdata;
 905         const char *subsection, *key;
 906         int subsection_len;
 907         char *name;
 908
 909         if (parse_config_key(var, "submodule", &subsection, &subsection_len, &key) < 0 ||
 910             !subsection)
 911                 return 0;
 912
 913         name = xmemdupz(subsection, subsection_len);
 914         if (check_submodule_name(name) < 0)
 915                 data->ret |= report(data->options, data->obj,
 916                                     FSCK_MSG_GITMODULES_NAME,
 917                                     "disallowed submodule name: %s",
 918                                     name);
 919         if (!strcmp(key, "url") && value &&
 920             looks_like_command_line_option(value))
 921                 data->ret |= report(data->options, data->obj,
 922                                     FSCK_MSG_GITMODULES_URL,
 923                                     "disallowed submodule url: %s",
 924                                     value);
 925         if (!strcmp(key, "path") && value &&
 926             looks_like_command_line_option(value))
 927                 data->ret |= report(data->options, data->obj,
 928                                     FSCK_MSG_GITMODULES_PATH,
 929                                     "disallowed submodule path: %s",
 930                                     value);
 931         free(name);
 932
 933         return 0;
 934 }
 935
 936 static int fsck_blob(struct blob *blob, const char *buf,
 937                      unsigned long size, struct fsck_options *options)
 938 {
 939         struct fsck_gitmodules_data data;
 940         struct config_options config_opts = { 0 };
 941
 942         if (!oidset_contains(&gitmodules_found, &blob->object.oid))
 943                 return 0;
 944         oidset_insert(&gitmodules_done, &blob->object.oid);
 945
 946         if (object_on_skiplist(options, &blob->object))
 947                 return 0;
 948
 949         if (!buf) {
 950                 /*
 951                  * A missing buffer here is a sign that the caller found the
 952                  * blob too gigantic to load into memory. Let's just consider
 953                  * that an error.
 954                  */
 955                 return report(options, &blob->object,
 956                               FSCK_MSG_GITMODULES_LARGE,
 957                               ".gitmodules too large to parse");
 958         }
 959
 960         data.obj = &blob->object;
 961         data.options = options;
 962         data.ret = 0;
 963         config_opts.error_action = CONFIG_ERROR_SILENT;
 964         if (git_config_from_mem(fsck_gitmodules_fn, CONFIG_ORIGIN_BLOB,
 965                                 ".gitmodules", buf, size, &data, &config_opts))
 966                 data.ret |= report(options, &blob->object,
 967                                    FSCK_MSG_GITMODULES_PARSE,
 968                                    "could not parse gitmodules blob");
 969
 970         return data.ret;
 971 }
 972
 973 int fsck_object(struct object *obj, void *data, unsigned long size,
 974         struct fsck_options *options)
 975 {
 976         if (!obj)
 977                 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
 978
 979         if (obj->type == OBJ_BLOB)
 980                 return fsck_blob((struct blob *)obj, data, size, options);
 981         if (obj->type == OBJ_TREE)
 982                 return fsck_tree((struct tree *) obj, options);
 983         if (obj->type == OBJ_COMMIT)
 984                 return fsck_commit((struct commit *) obj, (const char *) data,
 985                         size, options);
 986         if (obj->type == OBJ_TAG)
 987                 return fsck_tag((struct tag *) obj, (const char *) data,
 988                         size, options);
 989
 990         return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
 991                           obj->type);
 992 }
 993
 994 int fsck_error_function(struct fsck_options *o,
 995         struct object *obj, int msg_type, const char *message)
 996 {
 997         if (msg_type == FSCK_WARN) {
 998                 warning("object %s: %s", describe_object(o, obj), message);
 999                 return 0;
1000         }
1001         error("object %s: %s", describe_object(o, obj), message);
1002         return 1;
1003 }
1004
1005 int fsck_finish(struct fsck_options *options)
1006 {
1007         int ret = 0;
1008         struct oidset_iter iter;
1009         const struct object_id *oid;
1010
1011         oidset_iter_init(&gitmodules_found, &iter);
1012         while ((oid = oidset_iter_next(&iter))) {
1013                 struct blob *blob;
1014                 enum object_type type;
1015                 unsigned long size;
1016                 char *buf;
1017
1018                 if (oidset_contains(&gitmodules_done, oid))
1019                         continue;
1020
1021                 blob = lookup_blob(the_repository, oid);
1022                 if (!blob) {
1023                         struct object *obj = lookup_unknown_object(oid);
1024                         ret |= report(options, obj,
1025                                       FSCK_MSG_GITMODULES_BLOB,
1026                                       "non-blob found at .gitmodules");
1027                         continue;
1028                 }
1029
1030                 buf = read_object_file(oid, &type, &size);
1031                 if (!buf) {
1032                         if (is_promisor_object(&blob->object.oid))
1033                                 continue;
1034                         ret |= report(options, &blob->object,
1035                                       FSCK_MSG_GITMODULES_MISSING,
1036                                       "unable to read .gitmodules blob");
1037                         continue;
1038                 }
1039
1040                 if (type == OBJ_BLOB)
1041                         ret |= fsck_blob(blob, buf, size, options);
1042                 else
1043                         ret |= report(options, &blob->object,
1044                                       FSCK_MSG_GITMODULES_BLOB,
1045                                       "non-blob found at .gitmodules");
1046                 free(buf);
1047         }
1048
1049
1050         oidset_clear(&gitmodules_found);
1051         oidset_clear(&gitmodules_done);
1052         return ret;
1053 }