git.oblomov.eu Git - git/blob - 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
  15 static struct oidset gitmodules_found = OIDSET_INIT;
  16 static struct oidset gitmodules_done = OIDSET_INIT;
  17
  18 #define FSCK_FATAL -1
  19 #define FSCK_INFO -2
  20
  21 #define FOREACH_MSG_ID(FUNC) \
  22         /* fatal errors */ \
  23         FUNC(NUL_IN_HEADER, FATAL) \
  24         FUNC(UNTERMINATED_HEADER, FATAL) \
  25         /* errors */ \
  26         FUNC(BAD_DATE, ERROR) \
  27         FUNC(BAD_DATE_OVERFLOW, ERROR) \
  28         FUNC(BAD_EMAIL, ERROR) \
  29         FUNC(BAD_NAME, ERROR) \
  30         FUNC(BAD_OBJECT_SHA1, ERROR) \
  31         FUNC(BAD_PARENT_SHA1, ERROR) \
  32         FUNC(BAD_TAG_OBJECT, ERROR) \
  33         FUNC(BAD_TIMEZONE, ERROR) \
  34         FUNC(BAD_TREE, ERROR) \
  35         FUNC(BAD_TREE_SHA1, ERROR) \
  36         FUNC(BAD_TYPE, ERROR) \
  37         FUNC(DUPLICATE_ENTRIES, ERROR) \
  38         FUNC(MISSING_AUTHOR, ERROR) \
  39         FUNC(MISSING_COMMITTER, ERROR) \
  40         FUNC(MISSING_EMAIL, ERROR) \
  41         FUNC(MISSING_GRAFT, ERROR) \
  42         FUNC(MISSING_NAME_BEFORE_EMAIL, ERROR) \
  43         FUNC(MISSING_OBJECT, ERROR) \
  44         FUNC(MISSING_PARENT, ERROR) \
  45         FUNC(MISSING_SPACE_BEFORE_DATE, ERROR) \
  46         FUNC(MISSING_SPACE_BEFORE_EMAIL, ERROR) \
  47         FUNC(MISSING_TAG, ERROR) \
  48         FUNC(MISSING_TAG_ENTRY, ERROR) \
  49         FUNC(MISSING_TAG_OBJECT, ERROR) \
  50         FUNC(MISSING_TREE, ERROR) \
  51         FUNC(MISSING_TREE_OBJECT, ERROR) \
  52         FUNC(MISSING_TYPE, ERROR) \
  53         FUNC(MISSING_TYPE_ENTRY, ERROR) \
  54         FUNC(MULTIPLE_AUTHORS, ERROR) \
  55         FUNC(TAG_OBJECT_NOT_TAG, ERROR) \
  56         FUNC(TREE_NOT_SORTED, ERROR) \
  57         FUNC(UNKNOWN_TYPE, ERROR) \
  58         FUNC(ZERO_PADDED_DATE, ERROR) \
  59         FUNC(GITMODULES_MISSING, ERROR) \
  60         FUNC(GITMODULES_BLOB, ERROR) \
  61         /* warnings */ \
  62         FUNC(BAD_FILEMODE, WARN) \
  63         FUNC(EMPTY_NAME, WARN) \
  64         FUNC(FULL_PATHNAME, WARN) \
  65         FUNC(HAS_DOT, WARN) \
  66         FUNC(HAS_DOTDOT, WARN) \
  67         FUNC(HAS_DOTGIT, WARN) \
  68         FUNC(NULL_SHA1, WARN) \
  69         FUNC(ZERO_PADDED_FILEMODE, WARN) \
  70         FUNC(NUL_IN_COMMIT, WARN) \
  71         /* infos (reported as warnings, but ignored by default) */ \
  72         FUNC(BAD_TAG_NAME, INFO) \
  73         FUNC(MISSING_TAGGER_ENTRY, INFO)
  74
  75 #define MSG_ID(id, msg_type) FSCK_MSG_##id,
  76 enum fsck_msg_id {
  77         FOREACH_MSG_ID(MSG_ID)
  78         FSCK_MSG_MAX
  79 };
  80 #undef MSG_ID
  81
  82 #define STR(x) #x
  83 #define MSG_ID(id, msg_type) { STR(id), NULL, FSCK_##msg_type },
  84 static struct {
  85         const char *id_string;
  86         const char *downcased;
  87         int msg_type;
  88 } msg_id_info[FSCK_MSG_MAX + 1] = {
  89         FOREACH_MSG_ID(MSG_ID)
  90         { NULL, NULL, -1 }
  91 };
  92 #undef MSG_ID
  93
  94 static int parse_msg_id(const char *text)
  95 {
  96         int i;
  97
  98         if (!msg_id_info[0].downcased) {
  99                 /* convert id_string to lower case, without underscores. */
 100                 for (i = 0; i < FSCK_MSG_MAX; i++) {
 101                         const char *p = msg_id_info[i].id_string;
 102                         int len = strlen(p);
 103                         char *q = xmalloc(len);
 104
 105                         msg_id_info[i].downcased = q;
 106                         while (*p)
 107                                 if (*p == '_')
 108                                         p++;
 109                                 else
 110                                         *(q)++ = tolower(*(p)++);
 111                         *q = '\0';
 112                 }
 113         }
 114
 115         for (i = 0; i < FSCK_MSG_MAX; i++)
 116                 if (!strcmp(text, msg_id_info[i].downcased))
 117                         return i;
 118
 119         return -1;
 120 }
 121
 122 static int fsck_msg_type(enum fsck_msg_id msg_id,
 123         struct fsck_options *options)
 124 {
 125         int msg_type;
 126
 127         assert(msg_id >= 0 && msg_id < FSCK_MSG_MAX);
 128
 129         if (options->msg_type)
 130                 msg_type = options->msg_type[msg_id];
 131         else {
 132                 msg_type = msg_id_info[msg_id].msg_type;
 133                 if (options->strict && msg_type == FSCK_WARN)
 134                         msg_type = FSCK_ERROR;
 135         }
 136
 137         return msg_type;
 138 }
 139
 140 static void init_skiplist(struct fsck_options *options, const char *path)
 141 {
 142         static struct oid_array skiplist = OID_ARRAY_INIT;
 143         int sorted, fd;
 144         char buffer[GIT_MAX_HEXSZ + 1];
 145         struct object_id oid;
 146
 147         if (options->skiplist)
 148                 sorted = options->skiplist->sorted;
 149         else {
 150                 sorted = 1;
 151                 options->skiplist = &skiplist;
 152         }
 153
 154         fd = open(path, O_RDONLY);
 155         if (fd < 0)
 156                 die("Could not open skip list: %s", path);
 157         for (;;) {
 158                 const char *p;
 159                 int result = read_in_full(fd, buffer, sizeof(buffer));
 160                 if (result < 0)
 161                         die_errno("Could not read '%s'", path);
 162                 if (!result)
 163                         break;
 164                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n')
 165                         die("Invalid SHA-1: %s", buffer);
 166                 oid_array_append(&skiplist, &oid);
 167                 if (sorted && skiplist.nr > 1 &&
 168                                 oidcmp(&skiplist.oid[skiplist.nr - 2],
 169                                        &oid) > 0)
 170                         sorted = 0;
 171         }
 172         close(fd);
 173
 174         if (sorted)
 175                 skiplist.sorted = 1;
 176 }
 177
 178 static int parse_msg_type(const char *str)
 179 {
 180         if (!strcmp(str, "error"))
 181                 return FSCK_ERROR;
 182         else if (!strcmp(str, "warn"))
 183                 return FSCK_WARN;
 184         else if (!strcmp(str, "ignore"))
 185                 return FSCK_IGNORE;
 186         else
 187                 die("Unknown fsck message type: '%s'", str);
 188 }
 189
 190 int is_valid_msg_type(const char *msg_id, const char *msg_type)
 191 {
 192         if (parse_msg_id(msg_id) < 0)
 193                 return 0;
 194         parse_msg_type(msg_type);
 195         return 1;
 196 }
 197
 198 void fsck_set_msg_type(struct fsck_options *options,
 199                 const char *msg_id, const char *msg_type)
 200 {
 201         int id = parse_msg_id(msg_id), type;
 202
 203         if (id < 0)
 204                 die("Unhandled message id: %s", msg_id);
 205         type = parse_msg_type(msg_type);
 206
 207         if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
 208                 die("Cannot demote %s to %s", msg_id, msg_type);
 209
 210         if (!options->msg_type) {
 211                 int i;
 212                 int *msg_type;
 213                 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
 214                 for (i = 0; i < FSCK_MSG_MAX; i++)
 215                         msg_type[i] = fsck_msg_type(i, options);
 216                 options->msg_type = msg_type;
 217         }
 218
 219         options->msg_type[id] = type;
 220 }
 221
 222 void fsck_set_msg_types(struct fsck_options *options, const char *values)
 223 {
 224         char *buf = xstrdup(values), *to_free = buf;
 225         int done = 0;
 226
 227         while (!done) {
 228                 int len = strcspn(buf, " ,|"), equal;
 229
 230                 done = !buf[len];
 231                 if (!len) {
 232                         buf++;
 233                         continue;
 234                 }
 235                 buf[len] = '\0';
 236
 237                 for (equal = 0;
 238                      equal < len && buf[equal] != '=' && buf[equal] != ':';
 239                      equal++)
 240                         buf[equal] = tolower(buf[equal]);
 241                 buf[equal] = '\0';
 242
 243                 if (!strcmp(buf, "skiplist")) {
 244                         if (equal == len)
 245                                 die("skiplist requires a path");
 246                         init_skiplist(options, buf + equal + 1);
 247                         buf += len + 1;
 248                         continue;
 249                 }
 250
 251                 if (equal == len)
 252                         die("Missing '=': '%s'", buf);
 253
 254                 fsck_set_msg_type(options, buf, buf + equal + 1);
 255                 buf += len + 1;
 256         }
 257         free(to_free);
 258 }
 259
 260 static void append_msg_id(struct strbuf *sb, const char *msg_id)
 261 {
 262         for (;;) {
 263                 char c = *(msg_id)++;
 264
 265                 if (!c)
 266                         break;
 267                 if (c != '_')
 268                         strbuf_addch(sb, tolower(c));
 269                 else {
 270                         assert(*msg_id);
 271                         strbuf_addch(sb, *(msg_id)++);
 272                 }
 273         }
 274
 275         strbuf_addstr(sb, ": ");
 276 }
 277
 278 __attribute__((format (printf, 4, 5)))
 279 static int report(struct fsck_options *options, struct object *object,
 280         enum fsck_msg_id id, const char *fmt, ...)
 281 {
 282         va_list ap;
 283         struct strbuf sb = STRBUF_INIT;
 284         int msg_type = fsck_msg_type(id, options), result;
 285
 286         if (msg_type == FSCK_IGNORE)
 287                 return 0;
 288
 289         if (options->skiplist && object &&
 290                         oid_array_lookup(options->skiplist, &object->oid) >= 0)
 291                 return 0;
 292
 293         if (msg_type == FSCK_FATAL)
 294                 msg_type = FSCK_ERROR;
 295         else if (msg_type == FSCK_INFO)
 296                 msg_type = FSCK_WARN;
 297
 298         append_msg_id(&sb, msg_id_info[id].id_string);
 299
 300         va_start(ap, fmt);
 301         strbuf_vaddf(&sb, fmt, ap);
 302         result = options->error_func(options, object, msg_type, sb.buf);
 303         strbuf_release(&sb);
 304         va_end(ap);
 305
 306         return result;
 307 }
 308
 309 static char *get_object_name(struct fsck_options *options, struct object *obj)
 310 {
 311         if (!options->object_names)
 312                 return NULL;
 313         return lookup_decoration(options->object_names, obj);
 314 }
 315
 316 static void put_object_name(struct fsck_options *options, struct object *obj,
 317         const char *fmt, ...)
 318 {
 319         va_list ap;
 320         struct strbuf buf = STRBUF_INIT;
 321         char *existing;
 322
 323         if (!options->object_names)
 324                 return;
 325         existing = lookup_decoration(options->object_names, obj);
 326         if (existing)
 327                 return;
 328         va_start(ap, fmt);
 329         strbuf_vaddf(&buf, fmt, ap);
 330         add_decoration(options->object_names, obj, strbuf_detach(&buf, NULL));
 331         va_end(ap);
 332 }
 333
 334 static const char *describe_object(struct fsck_options *o, struct object *obj)
 335 {
 336         static struct strbuf buf = STRBUF_INIT;
 337         char *name;
 338
 339         strbuf_reset(&buf);
 340         strbuf_addstr(&buf, oid_to_hex(&obj->oid));
 341         if (o->object_names && (name = lookup_decoration(o->object_names, obj)))
 342                 strbuf_addf(&buf, " (%s)", name);
 343
 344         return buf.buf;
 345 }
 346
 347 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
 348 {
 349         struct tree_desc desc;
 350         struct name_entry entry;
 351         int res = 0;
 352         const char *name;
 353
 354         if (parse_tree(tree))
 355                 return -1;
 356
 357         name = get_object_name(options, &tree->object);
 358         if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
 359                 return -1;
 360         while (tree_entry_gently(&desc, &entry)) {
 361                 struct object *obj;
 362                 int result;
 363
 364                 if (S_ISGITLINK(entry.mode))
 365                         continue;
 366
 367                 if (S_ISDIR(entry.mode)) {
 368                         obj = (struct object *)lookup_tree(entry.oid);
 369                         if (name && obj)
 370                                 put_object_name(options, obj, "%s%s/", name,
 371                                         entry.path);
 372                         result = options->walk(obj, OBJ_TREE, data, options);
 373                 }
 374                 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
 375                         obj = (struct object *)lookup_blob(entry.oid);
 376                         if (name && obj)
 377                                 put_object_name(options, obj, "%s%s", name,
 378                                         entry.path);
 379                         result = options->walk(obj, OBJ_BLOB, data, options);
 380                 }
 381                 else {
 382                         result = error("in tree %s: entry %s has bad mode %.6o",
 383                                         describe_object(options, &tree->object), entry.path, entry.mode);
 384                 }
 385                 if (result < 0)
 386                         return result;
 387                 if (!res)
 388                         res = result;
 389         }
 390         return res;
 391 }
 392
 393 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
 394 {
 395         int counter = 0, generation = 0, name_prefix_len = 0;
 396         struct commit_list *parents;
 397         int res;
 398         int result;
 399         const char *name;
 400
 401         if (parse_commit(commit))
 402                 return -1;
 403
 404         name = get_object_name(options, &commit->object);
 405         if (name)
 406                 put_object_name(options, &commit->tree->object, "%s:", name);
 407
 408         result = options->walk((struct object *)commit->tree, OBJ_TREE, data, options);
 409         if (result < 0)
 410                 return result;
 411         res = result;
 412
 413         parents = commit->parents;
 414         if (name && parents) {
 415                 int len = strlen(name), power;
 416
 417                 if (len && name[len - 1] == '^') {
 418                         generation = 1;
 419                         name_prefix_len = len - 1;
 420                 }
 421                 else { /* parse ~<generation> suffix */
 422                         for (generation = 0, power = 1;
 423                              len && isdigit(name[len - 1]);
 424                              power *= 10)
 425                                 generation += power * (name[--len] - '0');
 426                         if (power > 1 && len && name[len - 1] == '~')
 427                                 name_prefix_len = len - 1;
 428                 }
 429         }
 430
 431         while (parents) {
 432                 if (name) {
 433                         struct object *obj = &parents->item->object;
 434
 435                         if (++counter > 1)
 436                                 put_object_name(options, obj, "%s^%d",
 437                                         name, counter);
 438                         else if (generation > 0)
 439                                 put_object_name(options, obj, "%.*s~%d",
 440                                         name_prefix_len, name, generation + 1);
 441                         else
 442                                 put_object_name(options, obj, "%s^", name);
 443                 }
 444                 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
 445                 if (result < 0)
 446                         return result;
 447                 if (!res)
 448                         res = result;
 449                 parents = parents->next;
 450         }
 451         return res;
 452 }
 453
 454 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
 455 {
 456         char *name = get_object_name(options, &tag->object);
 457
 458         if (parse_tag(tag))
 459                 return -1;
 460         if (name)
 461                 put_object_name(options, tag->tagged, "%s", name);
 462         return options->walk(tag->tagged, OBJ_ANY, data, options);
 463 }
 464
 465 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
 466 {
 467         if (!obj)
 468                 return -1;
 469
 470         if (obj->type == OBJ_NONE)
 471                 parse_object(&obj->oid);
 472
 473         switch (obj->type) {
 474         case OBJ_BLOB:
 475                 return 0;
 476         case OBJ_TREE:
 477                 return fsck_walk_tree((struct tree *)obj, data, options);
 478         case OBJ_COMMIT:
 479                 return fsck_walk_commit((struct commit *)obj, data, options);
 480         case OBJ_TAG:
 481                 return fsck_walk_tag((struct tag *)obj, data, options);
 482         default:
 483                 error("Unknown object type for %s", describe_object(options, obj));
 484                 return -1;
 485         }
 486 }
 487
 488 /*
 489  * The entries in a tree are ordered in the _path_ order,
 490  * which means that a directory entry is ordered by adding
 491  * a slash to the end of it.
 492  *
 493  * So a directory called "a" is ordered _after_ a file
 494  * called "a.c", because "a/" sorts after "a.c".
 495  */
 496 #define TREE_UNORDERED (-1)
 497 #define TREE_HAS_DUPS  (-2)
 498
 499 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 500 {
 501         int len1 = strlen(name1);
 502         int len2 = strlen(name2);
 503         int len = len1 < len2 ? len1 : len2;
 504         unsigned char c1, c2;
 505         int cmp;
 506
 507         cmp = memcmp(name1, name2, len);
 508         if (cmp < 0)
 509                 return 0;
 510         if (cmp > 0)
 511                 return TREE_UNORDERED;
 512
 513         /*
 514          * Ok, the first <len> characters are the same.
 515          * Now we need to order the next one, but turn
 516          * a '\0' into a '/' for a directory entry.
 517          */
 518         c1 = name1[len];
 519         c2 = name2[len];
 520         if (!c1 && !c2)
 521                 /*
 522                  * git-write-tree used to write out a nonsense tree that has
 523                  * entries with the same name, one blob and one tree.  Make
 524                  * sure we do not have duplicate entries.
 525                  */
 526                 return TREE_HAS_DUPS;
 527         if (!c1 && S_ISDIR(mode1))
 528                 c1 = '/';
 529         if (!c2 && S_ISDIR(mode2))
 530                 c2 = '/';
 531         return c1 < c2 ? 0 : TREE_UNORDERED;
 532 }
 533
 534 static int fsck_tree(struct tree *item, struct fsck_options *options)
 535 {
 536         int retval = 0;
 537         int has_null_sha1 = 0;
 538         int has_full_path = 0;
 539         int has_empty_name = 0;
 540         int has_dot = 0;
 541         int has_dotdot = 0;
 542         int has_dotgit = 0;
 543         int has_zero_pad = 0;
 544         int has_bad_modes = 0;
 545         int has_dup_entries = 0;
 546         int not_properly_sorted = 0;
 547         struct tree_desc desc;
 548         unsigned o_mode;
 549         const char *o_name;
 550
 551         if (init_tree_desc_gently(&desc, item->buffer, item->size)) {
 552                 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 553                 return retval;
 554         }
 555
 556         o_mode = 0;
 557         o_name = NULL;
 558
 559         while (desc.size) {
 560                 unsigned mode;
 561                 const char *name;
 562                 const struct object_id *oid;
 563
 564                 oid = tree_entry_extract(&desc, &name, &mode);
 565
 566                 has_null_sha1 |= is_null_oid(oid);
 567                 has_full_path |= !!strchr(name, '/');
 568                 has_empty_name |= !*name;
 569                 has_dot |= !strcmp(name, ".");
 570                 has_dotdot |= !strcmp(name, "..");
 571                 has_dotgit |= is_hfs_dotgit(name) || is_ntfs_dotgit(name);
 572                 has_zero_pad |= *(char *)desc.buffer == '0';
 573
 574                 if (is_hfs_dotgitmodules(name) || is_ntfs_dotgitmodules(name))
 575                         oidset_insert(&gitmodules_found, oid);
 576
 577                 if (update_tree_entry_gently(&desc)) {
 578                         retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 579                         break;
 580                 }
 581
 582                 switch (mode) {
 583                 /*
 584                  * Standard modes..
 585                  */
 586                 case S_IFREG | 0755:
 587                 case S_IFREG | 0644:
 588                 case S_IFLNK:
 589                 case S_IFDIR:
 590                 case S_IFGITLINK:
 591                         break;
 592                 /*
 593                  * This is nonstandard, but we had a few of these
 594                  * early on when we honored the full set of mode
 595                  * bits..
 596                  */
 597                 case S_IFREG | 0664:
 598                         if (!options->strict)
 599                                 break;
 600                         /* fallthrough */
 601                 default:
 602                         has_bad_modes = 1;
 603                 }
 604
 605                 if (o_name) {
 606                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 607                         case TREE_UNORDERED:
 608                                 not_properly_sorted = 1;
 609                                 break;
 610                         case TREE_HAS_DUPS:
 611                                 has_dup_entries = 1;
 612                                 break;
 613                         default:
 614                                 break;
 615                         }
 616                 }
 617
 618                 o_mode = mode;
 619                 o_name = name;
 620         }
 621
 622         if (has_null_sha1)
 623                 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
 624         if (has_full_path)
 625                 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
 626         if (has_empty_name)
 627                 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
 628         if (has_dot)
 629                 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
 630         if (has_dotdot)
 631                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
 632         if (has_dotgit)
 633                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
 634         if (has_zero_pad)
 635                 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
 636         if (has_bad_modes)
 637                 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
 638         if (has_dup_entries)
 639                 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
 640         if (not_properly_sorted)
 641                 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
 642         return retval;
 643 }
 644
 645 static int verify_headers(const void *data, unsigned long size,
 646                           struct object *obj, struct fsck_options *options)
 647 {
 648         const char *buffer = (const char *)data;
 649         unsigned long i;
 650
 651         for (i = 0; i < size; i++) {
 652                 switch (buffer[i]) {
 653                 case '\0':
 654                         return report(options, obj,
 655                                 FSCK_MSG_NUL_IN_HEADER,
 656                                 "unterminated header: NUL at offset %ld", i);
 657                 case '\n':
 658                         if (i + 1 < size && buffer[i + 1] == '\n')
 659                                 return 0;
 660                 }
 661         }
 662
 663         /*
 664          * We did not find double-LF that separates the header
 665          * and the body.  Not having a body is not a crime but
 666          * we do want to see the terminating LF for the last header
 667          * line.
 668          */
 669         if (size && buffer[size - 1] == '\n')
 670                 return 0;
 671
 672         return report(options, obj,
 673                 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
 674 }
 675
 676 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
 677 {
 678         const char *p = *ident;
 679         char *end;
 680
 681         *ident = strchrnul(*ident, '\n');
 682         if (**ident == '\n')
 683                 (*ident)++;
 684
 685         if (*p == '<')
 686                 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 687         p += strcspn(p, "<>\n");
 688         if (*p == '>')
 689                 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
 690         if (*p != '<')
 691                 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
 692         if (p[-1] != ' ')
 693                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 694         p++;
 695         p += strcspn(p, "<>\n");
 696         if (*p != '>')
 697                 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
 698         p++;
 699         if (*p != ' ')
 700                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
 701         p++;
 702         if (*p == '0' && p[1] != ' ')
 703                 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
 704         if (date_overflows(parse_timestamp(p, &end, 10)))
 705                 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
 706         if ((end == p || *end != ' '))
 707                 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
 708         p = end + 1;
 709         if ((*p != '+' && *p != '-') ||
 710             !isdigit(p[1]) ||
 711             !isdigit(p[2]) ||
 712             !isdigit(p[3]) ||
 713             !isdigit(p[4]) ||
 714             (p[5] != '\n'))
 715                 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
 716         p += 6;
 717         return 0;
 718 }
 719
 720 static int fsck_commit_buffer(struct commit *commit, const char *buffer,
 721         unsigned long size, struct fsck_options *options)
 722 {
 723         unsigned char tree_sha1[20], sha1[20];
 724         struct commit_graft *graft;
 725         unsigned parent_count, parent_line_count = 0, author_count;
 726         int err;
 727         const char *buffer_begin = buffer;
 728
 729         if (verify_headers(buffer, size, &commit->object, options))
 730                 return -1;
 731
 732         if (!skip_prefix(buffer, "tree ", &buffer))
 733                 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
 734         if (get_sha1_hex(buffer, tree_sha1) || buffer[40] != '\n') {
 735                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
 736                 if (err)
 737                         return err;
 738         }
 739         buffer += 41;
 740         while (skip_prefix(buffer, "parent ", &buffer)) {
 741                 if (get_sha1_hex(buffer, sha1) || buffer[40] != '\n') {
 742                         err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
 743                         if (err)
 744                                 return err;
 745                 }
 746                 buffer += 41;
 747                 parent_line_count++;
 748         }
 749         graft = lookup_commit_graft(&commit->object.oid);
 750         parent_count = commit_list_count(commit->parents);
 751         if (graft) {
 752                 if (graft->nr_parent == -1 && !parent_count)
 753                         ; /* shallow commit */
 754                 else if (graft->nr_parent != parent_count) {
 755                         err = report(options, &commit->object, FSCK_MSG_MISSING_GRAFT, "graft objects missing");
 756                         if (err)
 757                                 return err;
 758                 }
 759         } else {
 760                 if (parent_count != parent_line_count) {
 761                         err = report(options, &commit->object, FSCK_MSG_MISSING_PARENT, "parent objects missing");
 762                         if (err)
 763                                 return err;
 764                 }
 765         }
 766         author_count = 0;
 767         while (skip_prefix(buffer, "author ", &buffer)) {
 768                 author_count++;
 769                 err = fsck_ident(&buffer, &commit->object, options);
 770                 if (err)
 771                         return err;
 772         }
 773         if (author_count < 1)
 774                 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
 775         else if (author_count > 1)
 776                 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
 777         if (err)
 778                 return err;
 779         if (!skip_prefix(buffer, "committer ", &buffer))
 780                 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
 781         err = fsck_ident(&buffer, &commit->object, options);
 782         if (err)
 783                 return err;
 784         if (!commit->tree) {
 785                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE, "could not load commit's tree %s", sha1_to_hex(tree_sha1));
 786                 if (err)
 787                         return err;
 788         }
 789         if (memchr(buffer_begin, '\0', size)) {
 790                 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
 791                              "NUL byte in the commit object body");
 792                 if (err)
 793                         return err;
 794         }
 795         return 0;
 796 }
 797
 798 static int fsck_commit(struct commit *commit, const char *data,
 799         unsigned long size, struct fsck_options *options)
 800 {
 801         const char *buffer = data ?  data : get_commit_buffer(commit, &size);
 802         int ret = fsck_commit_buffer(commit, buffer, size, options);
 803         if (!data)
 804                 unuse_commit_buffer(commit, buffer);
 805         return ret;
 806 }
 807
 808 static int fsck_tag_buffer(struct tag *tag, const char *data,
 809         unsigned long size, struct fsck_options *options)
 810 {
 811         unsigned char sha1[20];
 812         int ret = 0;
 813         const char *buffer;
 814         char *to_free = NULL, *eol;
 815         struct strbuf sb = STRBUF_INIT;
 816
 817         if (data)
 818                 buffer = data;
 819         else {
 820                 enum object_type type;
 821
 822                 buffer = to_free =
 823                         read_sha1_file(tag->object.oid.hash, &type, &size);
 824                 if (!buffer)
 825                         return report(options, &tag->object,
 826                                 FSCK_MSG_MISSING_TAG_OBJECT,
 827                                 "cannot read tag object");
 828
 829                 if (type != OBJ_TAG) {
 830                         ret = report(options, &tag->object,
 831                                 FSCK_MSG_TAG_OBJECT_NOT_TAG,
 832                                 "expected tag got %s",
 833                             type_name(type));
 834                         goto done;
 835                 }
 836         }
 837
 838         ret = verify_headers(buffer, size, &tag->object, options);
 839         if (ret)
 840                 goto done;
 841
 842         if (!skip_prefix(buffer, "object ", &buffer)) {
 843                 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
 844                 goto done;
 845         }
 846         if (get_sha1_hex(buffer, sha1) || buffer[40] != '\n') {
 847                 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
 848                 if (ret)
 849                         goto done;
 850         }
 851         buffer += 41;
 852
 853         if (!skip_prefix(buffer, "type ", &buffer)) {
 854                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
 855                 goto done;
 856         }
 857         eol = strchr(buffer, '\n');
 858         if (!eol) {
 859                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
 860                 goto done;
 861         }
 862         if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
 863                 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
 864         if (ret)
 865                 goto done;
 866         buffer = eol + 1;
 867
 868         if (!skip_prefix(buffer, "tag ", &buffer)) {
 869                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
 870                 goto done;
 871         }
 872         eol = strchr(buffer, '\n');
 873         if (!eol) {
 874                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
 875                 goto done;
 876         }
 877         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
 878         if (check_refname_format(sb.buf, 0)) {
 879                 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
 880                            "invalid 'tag' name: %.*s",
 881                            (int)(eol - buffer), buffer);
 882                 if (ret)
 883                         goto done;
 884         }
 885         buffer = eol + 1;
 886
 887         if (!skip_prefix(buffer, "tagger ", &buffer)) {
 888                 /* early tags do not contain 'tagger' lines; warn only */
 889                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
 890                 if (ret)
 891                         goto done;
 892         }
 893         else
 894                 ret = fsck_ident(&buffer, &tag->object, options);
 895
 896 done:
 897         strbuf_release(&sb);
 898         free(to_free);
 899         return ret;
 900 }
 901
 902 static int fsck_tag(struct tag *tag, const char *data,
 903         unsigned long size, struct fsck_options *options)
 904 {
 905         struct object *tagged = tag->tagged;
 906
 907         if (!tagged)
 908                 return report(options, &tag->object, FSCK_MSG_BAD_TAG_OBJECT, "could not load tagged object");
 909
 910         return fsck_tag_buffer(tag, data, size, options);
 911 }
 912
 913 static int fsck_blob(struct blob *blob, const char *buf,
 914                      unsigned long size, struct fsck_options *options)
 915 {
 916         return 0;
 917 }
 918
 919 int fsck_object(struct object *obj, void *data, unsigned long size,
 920         struct fsck_options *options)
 921 {
 922         if (!obj)
 923                 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
 924
 925         if (obj->type == OBJ_BLOB)
 926                 return fsck_blob((struct blob *)obj, data, size, options);
 927         if (obj->type == OBJ_TREE)
 928                 return fsck_tree((struct tree *) obj, options);
 929         if (obj->type == OBJ_COMMIT)
 930                 return fsck_commit((struct commit *) obj, (const char *) data,
 931                         size, options);
 932         if (obj->type == OBJ_TAG)
 933                 return fsck_tag((struct tag *) obj, (const char *) data,
 934                         size, options);
 935
 936         return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
 937                           obj->type);
 938 }
 939
 940 int fsck_error_function(struct fsck_options *o,
 941         struct object *obj, int msg_type, const char *message)
 942 {
 943         if (msg_type == FSCK_WARN) {
 944                 warning("object %s: %s", describe_object(o, obj), message);
 945                 return 0;
 946         }
 947         error("object %s: %s", describe_object(o, obj), message);
 948         return 1;
 949 }
 950
 951 int fsck_finish(struct fsck_options *options)
 952 {
 953         int ret = 0;
 954         struct oidset_iter iter;
 955         const struct object_id *oid;
 956
 957         oidset_iter_init(&gitmodules_found, &iter);
 958         while ((oid = oidset_iter_next(&iter))) {
 959                 struct blob *blob;
 960                 enum object_type type;
 961                 unsigned long size;
 962                 char *buf;
 963
 964                 if (oidset_contains(&gitmodules_done, oid))
 965                         continue;
 966
 967                 blob = lookup_blob(oid);
 968                 if (!blob) {
 969                         ret |= report(options, &blob->object,
 970                                       FSCK_MSG_GITMODULES_BLOB,
 971                                       "non-blob found at .gitmodules");
 972                         continue;
 973                 }
 974
 975                 buf = read_sha1_file(oid->hash, &type, &size);
 976                 if (!buf) {
 977                         ret |= report(options, &blob->object,
 978                                       FSCK_MSG_GITMODULES_MISSING,
 979                                       "unable to read .gitmodules blob");
 980                         continue;
 981                 }
 982
 983                 if (type == OBJ_BLOB)
 984                         ret |= fsck_blob(blob, buf, size, options);
 985                 else
 986                         ret |= report(options, &blob->object,
 987                                       FSCK_MSG_GITMODULES_BLOB,
 988                                       "non-blob found at .gitmodules");
 989                 free(buf);
 990         }
 991
 992
 993         oidset_clear(&gitmodules_found);
 994         oidset_clear(&gitmodules_done);
 995         return ret;
 996 }