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 "help.h"
  14
  15 #define FSCK_FATAL -1
  16 #define FSCK_INFO -2
  17
  18 #define FOREACH_MSG_ID(FUNC) \
  19         /* fatal errors */ \
  20         FUNC(NUL_IN_HEADER, FATAL) \
  21         FUNC(UNTERMINATED_HEADER, FATAL) \
  22         /* errors */ \
  23         FUNC(BAD_DATE, ERROR) \
  24         FUNC(BAD_DATE_OVERFLOW, ERROR) \
  25         FUNC(BAD_EMAIL, ERROR) \
  26         FUNC(BAD_NAME, ERROR) \
  27         FUNC(BAD_OBJECT_SHA1, ERROR) \
  28         FUNC(BAD_PARENT_SHA1, ERROR) \
  29         FUNC(BAD_TAG_OBJECT, ERROR) \
  30         FUNC(BAD_TIMEZONE, ERROR) \
  31         FUNC(BAD_TREE, ERROR) \
  32         FUNC(BAD_TREE_SHA1, ERROR) \
  33         FUNC(BAD_TYPE, ERROR) \
  34         FUNC(DUPLICATE_ENTRIES, ERROR) \
  35         FUNC(MISSING_AUTHOR, ERROR) \
  36         FUNC(MISSING_COMMITTER, ERROR) \
  37         FUNC(MISSING_EMAIL, ERROR) \
  38         FUNC(MISSING_GRAFT, ERROR) \
  39         FUNC(MISSING_NAME_BEFORE_EMAIL, ERROR) \
  40         FUNC(MISSING_OBJECT, ERROR) \
  41         FUNC(MISSING_PARENT, ERROR) \
  42         FUNC(MISSING_SPACE_BEFORE_DATE, ERROR) \
  43         FUNC(MISSING_SPACE_BEFORE_EMAIL, ERROR) \
  44         FUNC(MISSING_TAG, ERROR) \
  45         FUNC(MISSING_TAG_ENTRY, ERROR) \
  46         FUNC(MISSING_TAG_OBJECT, ERROR) \
  47         FUNC(MISSING_TREE, ERROR) \
  48         FUNC(MISSING_TYPE, ERROR) \
  49         FUNC(MISSING_TYPE_ENTRY, ERROR) \
  50         FUNC(MULTIPLE_AUTHORS, ERROR) \
  51         FUNC(TAG_OBJECT_NOT_TAG, ERROR) \
  52         FUNC(TREE_NOT_SORTED, ERROR) \
  53         FUNC(UNKNOWN_TYPE, ERROR) \
  54         FUNC(ZERO_PADDED_DATE, ERROR) \
  55         /* warnings */ \
  56         FUNC(BAD_FILEMODE, WARN) \
  57         FUNC(EMPTY_NAME, WARN) \
  58         FUNC(FULL_PATHNAME, WARN) \
  59         FUNC(HAS_DOT, WARN) \
  60         FUNC(HAS_DOTDOT, WARN) \
  61         FUNC(HAS_DOTGIT, WARN) \
  62         FUNC(NULL_SHA1, WARN) \
  63         FUNC(ZERO_PADDED_FILEMODE, WARN) \
  64         FUNC(NUL_IN_COMMIT, WARN) \
  65         /* infos (reported as warnings, but ignored by default) */ \
  66         FUNC(BAD_TAG_NAME, INFO) \
  67         FUNC(MISSING_TAGGER_ENTRY, INFO)
  68
  69 #define MSG_ID(id, msg_type) FSCK_MSG_##id,
  70 enum fsck_msg_id {
  71         FOREACH_MSG_ID(MSG_ID)
  72         FSCK_MSG_MAX
  73 };
  74 #undef MSG_ID
  75
  76 #define STR(x) #x
  77 #define MSG_ID(id, msg_type) { STR(id), NULL, NULL, FSCK_##msg_type },
  78 static struct {
  79         const char *id_string;
  80         const char *downcased;
  81         const char *camelcased;
  82         int msg_type;
  83 } msg_id_info[FSCK_MSG_MAX + 1] = {
  84         FOREACH_MSG_ID(MSG_ID)
  85         { NULL, NULL, NULL, -1 }
  86 };
  87 #undef MSG_ID
  88
  89 static void prepare_msg_ids(void)
  90 {
  91         int i;
  92
  93         if (msg_id_info[0].downcased)
  94                 return;
  95
  96         /* convert id_string to lower case, without underscores. */
  97         for (i = 0; i < FSCK_MSG_MAX; i++) {
  98                 const char *p = msg_id_info[i].id_string;
  99                 int len = strlen(p);
 100                 char *q = xmalloc(len);
 101
 102                 msg_id_info[i].downcased = q;
 103                 while (*p)
 104                         if (*p == '_')
 105                                 p++;
 106                         else
 107                                 *(q)++ = tolower(*(p)++);
 108                 *q = '\0';
 109
 110                 p = msg_id_info[i].id_string;
 111                 q = xmalloc(len);
 112                 msg_id_info[i].camelcased = q;
 113                 while (*p) {
 114                         if (*p == '_') {
 115                                 p++;
 116                                 if (*p)
 117                                         *q++ = *p++;
 118                         } else {
 119                                 *q++ = tolower(*p++);
 120                         }
 121                 }
 122                 *q = '\0';
 123         }
 124 }
 125
 126 static int parse_msg_id(const char *text)
 127 {
 128         int i;
 129
 130         prepare_msg_ids();
 131
 132         for (i = 0; i < FSCK_MSG_MAX; i++)
 133                 if (!strcmp(text, msg_id_info[i].downcased))
 134                         return i;
 135
 136         return -1;
 137 }
 138
 139 void list_config_fsck_msg_ids(struct string_list *list, const char *prefix)
 140 {
 141         int i;
 142
 143         prepare_msg_ids();
 144
 145         for (i = 0; i < FSCK_MSG_MAX; i++)
 146                 list_config_item(list, prefix, msg_id_info[i].camelcased);
 147 }
 148
 149 static int fsck_msg_type(enum fsck_msg_id msg_id,
 150         struct fsck_options *options)
 151 {
 152         int msg_type;
 153
 154         assert(msg_id >= 0 && msg_id < FSCK_MSG_MAX);
 155
 156         if (options->msg_type)
 157                 msg_type = options->msg_type[msg_id];
 158         else {
 159                 msg_type = msg_id_info[msg_id].msg_type;
 160                 if (options->strict && msg_type == FSCK_WARN)
 161                         msg_type = FSCK_ERROR;
 162         }
 163
 164         return msg_type;
 165 }
 166
 167 static void init_skiplist(struct fsck_options *options, const char *path)
 168 {
 169         static struct oid_array skiplist = OID_ARRAY_INIT;
 170         int sorted, fd;
 171         char buffer[GIT_MAX_HEXSZ + 1];
 172         struct object_id oid;
 173
 174         if (options->skiplist)
 175                 sorted = options->skiplist->sorted;
 176         else {
 177                 sorted = 1;
 178                 options->skiplist = &skiplist;
 179         }
 180
 181         fd = open(path, O_RDONLY);
 182         if (fd < 0)
 183                 die("Could not open skip list: %s", path);
 184         for (;;) {
 185                 const char *p;
 186                 int result = read_in_full(fd, buffer, sizeof(buffer));
 187                 if (result < 0)
 188                         die_errno("Could not read '%s'", path);
 189                 if (!result)
 190                         break;
 191                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n')
 192                         die("Invalid SHA-1: %s", buffer);
 193                 oid_array_append(&skiplist, &oid);
 194                 if (sorted && skiplist.nr > 1 &&
 195                                 oidcmp(&skiplist.oid[skiplist.nr - 2],
 196                                        &oid) > 0)
 197                         sorted = 0;
 198         }
 199         close(fd);
 200
 201         if (sorted)
 202                 skiplist.sorted = 1;
 203 }
 204
 205 static int parse_msg_type(const char *str)
 206 {
 207         if (!strcmp(str, "error"))
 208                 return FSCK_ERROR;
 209         else if (!strcmp(str, "warn"))
 210                 return FSCK_WARN;
 211         else if (!strcmp(str, "ignore"))
 212                 return FSCK_IGNORE;
 213         else
 214                 die("Unknown fsck message type: '%s'", str);
 215 }
 216
 217 int is_valid_msg_type(const char *msg_id, const char *msg_type)
 218 {
 219         if (parse_msg_id(msg_id) < 0)
 220                 return 0;
 221         parse_msg_type(msg_type);
 222         return 1;
 223 }
 224
 225 void fsck_set_msg_type(struct fsck_options *options,
 226                 const char *msg_id, const char *msg_type)
 227 {
 228         int id = parse_msg_id(msg_id), type;
 229
 230         if (id < 0)
 231                 die("Unhandled message id: %s", msg_id);
 232         type = parse_msg_type(msg_type);
 233
 234         if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
 235                 die("Cannot demote %s to %s", msg_id, msg_type);
 236
 237         if (!options->msg_type) {
 238                 int i;
 239                 int *msg_type;
 240                 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
 241                 for (i = 0; i < FSCK_MSG_MAX; i++)
 242                         msg_type[i] = fsck_msg_type(i, options);
 243                 options->msg_type = msg_type;
 244         }
 245
 246         options->msg_type[id] = type;
 247 }
 248
 249 void fsck_set_msg_types(struct fsck_options *options, const char *values)
 250 {
 251         char *buf = xstrdup(values), *to_free = buf;
 252         int done = 0;
 253
 254         while (!done) {
 255                 int len = strcspn(buf, " ,|"), equal;
 256
 257                 done = !buf[len];
 258                 if (!len) {
 259                         buf++;
 260                         continue;
 261                 }
 262                 buf[len] = '\0';
 263
 264                 for (equal = 0;
 265                      equal < len && buf[equal] != '=' && buf[equal] != ':';
 266                      equal++)
 267                         buf[equal] = tolower(buf[equal]);
 268                 buf[equal] = '\0';
 269
 270                 if (!strcmp(buf, "skiplist")) {
 271                         if (equal == len)
 272                                 die("skiplist requires a path");
 273                         init_skiplist(options, buf + equal + 1);
 274                         buf += len + 1;
 275                         continue;
 276                 }
 277
 278                 if (equal == len)
 279                         die("Missing '=': '%s'", buf);
 280
 281                 fsck_set_msg_type(options, buf, buf + equal + 1);
 282                 buf += len + 1;
 283         }
 284         free(to_free);
 285 }
 286
 287 static void append_msg_id(struct strbuf *sb, const char *msg_id)
 288 {
 289         for (;;) {
 290                 char c = *(msg_id)++;
 291
 292                 if (!c)
 293                         break;
 294                 if (c != '_')
 295                         strbuf_addch(sb, tolower(c));
 296                 else {
 297                         assert(*msg_id);
 298                         strbuf_addch(sb, *(msg_id)++);
 299                 }
 300         }
 301
 302         strbuf_addstr(sb, ": ");
 303 }
 304
 305 __attribute__((format (printf, 4, 5)))
 306 static int report(struct fsck_options *options, struct object *object,
 307         enum fsck_msg_id id, const char *fmt, ...)
 308 {
 309         va_list ap;
 310         struct strbuf sb = STRBUF_INIT;
 311         int msg_type = fsck_msg_type(id, options), result;
 312
 313         if (msg_type == FSCK_IGNORE)
 314                 return 0;
 315
 316         if (options->skiplist && object &&
 317                         oid_array_lookup(options->skiplist, &object->oid) >= 0)
 318                 return 0;
 319
 320         if (msg_type == FSCK_FATAL)
 321                 msg_type = FSCK_ERROR;
 322         else if (msg_type == FSCK_INFO)
 323                 msg_type = FSCK_WARN;
 324
 325         append_msg_id(&sb, msg_id_info[id].id_string);
 326
 327         va_start(ap, fmt);
 328         strbuf_vaddf(&sb, fmt, ap);
 329         result = options->error_func(options, object, msg_type, sb.buf);
 330         strbuf_release(&sb);
 331         va_end(ap);
 332
 333         return result;
 334 }
 335
 336 static char *get_object_name(struct fsck_options *options, struct object *obj)
 337 {
 338         if (!options->object_names)
 339                 return NULL;
 340         return lookup_decoration(options->object_names, obj);
 341 }
 342
 343 static void put_object_name(struct fsck_options *options, struct object *obj,
 344         const char *fmt, ...)
 345 {
 346         va_list ap;
 347         struct strbuf buf = STRBUF_INIT;
 348         char *existing;
 349
 350         if (!options->object_names)
 351                 return;
 352         existing = lookup_decoration(options->object_names, obj);
 353         if (existing)
 354                 return;
 355         va_start(ap, fmt);
 356         strbuf_vaddf(&buf, fmt, ap);
 357         add_decoration(options->object_names, obj, strbuf_detach(&buf, NULL));
 358         va_end(ap);
 359 }
 360
 361 static const char *describe_object(struct fsck_options *o, struct object *obj)
 362 {
 363         static struct strbuf buf = STRBUF_INIT;
 364         char *name;
 365
 366         strbuf_reset(&buf);
 367         strbuf_addstr(&buf, oid_to_hex(&obj->oid));
 368         if (o->object_names && (name = lookup_decoration(o->object_names, obj)))
 369                 strbuf_addf(&buf, " (%s)", name);
 370
 371         return buf.buf;
 372 }
 373
 374 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
 375 {
 376         struct tree_desc desc;
 377         struct name_entry entry;
 378         int res = 0;
 379         const char *name;
 380
 381         if (parse_tree(tree))
 382                 return -1;
 383
 384         name = get_object_name(options, &tree->object);
 385         if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
 386                 return -1;
 387         while (tree_entry_gently(&desc, &entry)) {
 388                 struct object *obj;
 389                 int result;
 390
 391                 if (S_ISGITLINK(entry.mode))
 392                         continue;
 393
 394                 if (S_ISDIR(entry.mode)) {
 395                         obj = (struct object *)lookup_tree(entry.oid);
 396                         if (name && obj)
 397                                 put_object_name(options, obj, "%s%s/", name,
 398                                         entry.path);
 399                         result = options->walk(obj, OBJ_TREE, data, options);
 400                 }
 401                 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
 402                         obj = (struct object *)lookup_blob(entry.oid);
 403                         if (name && obj)
 404                                 put_object_name(options, obj, "%s%s", name,
 405                                         entry.path);
 406                         result = options->walk(obj, OBJ_BLOB, data, options);
 407                 }
 408                 else {
 409                         result = error("in tree %s: entry %s has bad mode %.6o",
 410                                         describe_object(options, &tree->object), entry.path, entry.mode);
 411                 }
 412                 if (result < 0)
 413                         return result;
 414                 if (!res)
 415                         res = result;
 416         }
 417         return res;
 418 }
 419
 420 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
 421 {
 422         int counter = 0, generation = 0, name_prefix_len = 0;
 423         struct commit_list *parents;
 424         int res;
 425         int result;
 426         const char *name;
 427
 428         if (parse_commit(commit))
 429                 return -1;
 430
 431         name = get_object_name(options, &commit->object);
 432         if (name)
 433                 put_object_name(options, &get_commit_tree(commit)->object,
 434                                 "%s:", name);
 435
 436         result = options->walk((struct object *)get_commit_tree(commit),
 437                                OBJ_TREE, data, options);
 438         if (result < 0)
 439                 return result;
 440         res = result;
 441
 442         parents = commit->parents;
 443         if (name && parents) {
 444                 int len = strlen(name), power;
 445
 446                 if (len && name[len - 1] == '^') {
 447                         generation = 1;
 448                         name_prefix_len = len - 1;
 449                 }
 450                 else { /* parse ~<generation> suffix */
 451                         for (generation = 0, power = 1;
 452                              len && isdigit(name[len - 1]);
 453                              power *= 10)
 454                                 generation += power * (name[--len] - '0');
 455                         if (power > 1 && len && name[len - 1] == '~')
 456                                 name_prefix_len = len - 1;
 457                 }
 458         }
 459
 460         while (parents) {
 461                 if (name) {
 462                         struct object *obj = &parents->item->object;
 463
 464                         if (++counter > 1)
 465                                 put_object_name(options, obj, "%s^%d",
 466                                         name, counter);
 467                         else if (generation > 0)
 468                                 put_object_name(options, obj, "%.*s~%d",
 469                                         name_prefix_len, name, generation + 1);
 470                         else
 471                                 put_object_name(options, obj, "%s^", name);
 472                 }
 473                 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
 474                 if (result < 0)
 475                         return result;
 476                 if (!res)
 477                         res = result;
 478                 parents = parents->next;
 479         }
 480         return res;
 481 }
 482
 483 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
 484 {
 485         char *name = get_object_name(options, &tag->object);
 486
 487         if (parse_tag(tag))
 488                 return -1;
 489         if (name)
 490                 put_object_name(options, tag->tagged, "%s", name);
 491         return options->walk(tag->tagged, OBJ_ANY, data, options);
 492 }
 493
 494 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
 495 {
 496         if (!obj)
 497                 return -1;
 498
 499         if (obj->type == OBJ_NONE)
 500                 parse_object(&obj->oid);
 501
 502         switch (obj->type) {
 503         case OBJ_BLOB:
 504                 return 0;
 505         case OBJ_TREE:
 506                 return fsck_walk_tree((struct tree *)obj, data, options);
 507         case OBJ_COMMIT:
 508                 return fsck_walk_commit((struct commit *)obj, data, options);
 509         case OBJ_TAG:
 510                 return fsck_walk_tag((struct tag *)obj, data, options);
 511         default:
 512                 error("Unknown object type for %s", describe_object(options, obj));
 513                 return -1;
 514         }
 515 }
 516
 517 /*
 518  * The entries in a tree are ordered in the _path_ order,
 519  * which means that a directory entry is ordered by adding
 520  * a slash to the end of it.
 521  *
 522  * So a directory called "a" is ordered _after_ a file
 523  * called "a.c", because "a/" sorts after "a.c".
 524  */
 525 #define TREE_UNORDERED (-1)
 526 #define TREE_HAS_DUPS  (-2)
 527
 528 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 529 {
 530         int len1 = strlen(name1);
 531         int len2 = strlen(name2);
 532         int len = len1 < len2 ? len1 : len2;
 533         unsigned char c1, c2;
 534         int cmp;
 535
 536         cmp = memcmp(name1, name2, len);
 537         if (cmp < 0)
 538                 return 0;
 539         if (cmp > 0)
 540                 return TREE_UNORDERED;
 541
 542         /*
 543          * Ok, the first <len> characters are the same.
 544          * Now we need to order the next one, but turn
 545          * a '\0' into a '/' for a directory entry.
 546          */
 547         c1 = name1[len];
 548         c2 = name2[len];
 549         if (!c1 && !c2)
 550                 /*
 551                  * git-write-tree used to write out a nonsense tree that has
 552                  * entries with the same name, one blob and one tree.  Make
 553                  * sure we do not have duplicate entries.
 554                  */
 555                 return TREE_HAS_DUPS;
 556         if (!c1 && S_ISDIR(mode1))
 557                 c1 = '/';
 558         if (!c2 && S_ISDIR(mode2))
 559                 c2 = '/';
 560         return c1 < c2 ? 0 : TREE_UNORDERED;
 561 }
 562
 563 static int fsck_tree(struct tree *item, struct fsck_options *options)
 564 {
 565         int retval = 0;
 566         int has_null_sha1 = 0;
 567         int has_full_path = 0;
 568         int has_empty_name = 0;
 569         int has_dot = 0;
 570         int has_dotdot = 0;
 571         int has_dotgit = 0;
 572         int has_zero_pad = 0;
 573         int has_bad_modes = 0;
 574         int has_dup_entries = 0;
 575         int not_properly_sorted = 0;
 576         struct tree_desc desc;
 577         unsigned o_mode;
 578         const char *o_name;
 579
 580         if (init_tree_desc_gently(&desc, item->buffer, item->size)) {
 581                 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 582                 return retval;
 583         }
 584
 585         o_mode = 0;
 586         o_name = NULL;
 587
 588         while (desc.size) {
 589                 unsigned mode;
 590                 const char *name;
 591                 const struct object_id *oid;
 592
 593                 oid = tree_entry_extract(&desc, &name, &mode);
 594
 595                 has_null_sha1 |= is_null_oid(oid);
 596                 has_full_path |= !!strchr(name, '/');
 597                 has_empty_name |= !*name;
 598                 has_dot |= !strcmp(name, ".");
 599                 has_dotdot |= !strcmp(name, "..");
 600                 has_dotgit |= (!strcmp(name, ".git") ||
 601                                is_hfs_dotgit(name) ||
 602                                is_ntfs_dotgit(name));
 603                 has_zero_pad |= *(char *)desc.buffer == '0';
 604                 if (update_tree_entry_gently(&desc)) {
 605                         retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 606                         break;
 607                 }
 608
 609                 switch (mode) {
 610                 /*
 611                  * Standard modes..
 612                  */
 613                 case S_IFREG | 0755:
 614                 case S_IFREG | 0644:
 615                 case S_IFLNK:
 616                 case S_IFDIR:
 617                 case S_IFGITLINK:
 618                         break;
 619                 /*
 620                  * This is nonstandard, but we had a few of these
 621                  * early on when we honored the full set of mode
 622                  * bits..
 623                  */
 624                 case S_IFREG | 0664:
 625                         if (!options->strict)
 626                                 break;
 627                         /* fallthrough */
 628                 default:
 629                         has_bad_modes = 1;
 630                 }
 631
 632                 if (o_name) {
 633                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 634                         case TREE_UNORDERED:
 635                                 not_properly_sorted = 1;
 636                                 break;
 637                         case TREE_HAS_DUPS:
 638                                 has_dup_entries = 1;
 639                                 break;
 640                         default:
 641                                 break;
 642                         }
 643                 }
 644
 645                 o_mode = mode;
 646                 o_name = name;
 647         }
 648
 649         if (has_null_sha1)
 650                 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
 651         if (has_full_path)
 652                 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
 653         if (has_empty_name)
 654                 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
 655         if (has_dot)
 656                 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
 657         if (has_dotdot)
 658                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
 659         if (has_dotgit)
 660                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
 661         if (has_zero_pad)
 662                 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
 663         if (has_bad_modes)
 664                 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
 665         if (has_dup_entries)
 666                 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
 667         if (not_properly_sorted)
 668                 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
 669         return retval;
 670 }
 671
 672 static int verify_headers(const void *data, unsigned long size,
 673                           struct object *obj, struct fsck_options *options)
 674 {
 675         const char *buffer = (const char *)data;
 676         unsigned long i;
 677
 678         for (i = 0; i < size; i++) {
 679                 switch (buffer[i]) {
 680                 case '\0':
 681                         return report(options, obj,
 682                                 FSCK_MSG_NUL_IN_HEADER,
 683                                 "unterminated header: NUL at offset %ld", i);
 684                 case '\n':
 685                         if (i + 1 < size && buffer[i + 1] == '\n')
 686                                 return 0;
 687                 }
 688         }
 689
 690         /*
 691          * We did not find double-LF that separates the header
 692          * and the body.  Not having a body is not a crime but
 693          * we do want to see the terminating LF for the last header
 694          * line.
 695          */
 696         if (size && buffer[size - 1] == '\n')
 697                 return 0;
 698
 699         return report(options, obj,
 700                 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
 701 }
 702
 703 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
 704 {
 705         const char *p = *ident;
 706         char *end;
 707
 708         *ident = strchrnul(*ident, '\n');
 709         if (**ident == '\n')
 710                 (*ident)++;
 711
 712         if (*p == '<')
 713                 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 714         p += strcspn(p, "<>\n");
 715         if (*p == '>')
 716                 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
 717         if (*p != '<')
 718                 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
 719         if (p[-1] != ' ')
 720                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 721         p++;
 722         p += strcspn(p, "<>\n");
 723         if (*p != '>')
 724                 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
 725         p++;
 726         if (*p != ' ')
 727                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
 728         p++;
 729         if (*p == '0' && p[1] != ' ')
 730                 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
 731         if (date_overflows(parse_timestamp(p, &end, 10)))
 732                 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
 733         if ((end == p || *end != ' '))
 734                 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
 735         p = end + 1;
 736         if ((*p != '+' && *p != '-') ||
 737             !isdigit(p[1]) ||
 738             !isdigit(p[2]) ||
 739             !isdigit(p[3]) ||
 740             !isdigit(p[4]) ||
 741             (p[5] != '\n'))
 742                 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
 743         p += 6;
 744         return 0;
 745 }
 746
 747 static int fsck_commit_buffer(struct commit *commit, const char *buffer,
 748         unsigned long size, struct fsck_options *options)
 749 {
 750         unsigned char tree_sha1[20], sha1[20];
 751         struct commit_graft *graft;
 752         unsigned parent_count, parent_line_count = 0, author_count;
 753         int err;
 754         const char *buffer_begin = buffer;
 755
 756         if (verify_headers(buffer, size, &commit->object, options))
 757                 return -1;
 758
 759         if (!skip_prefix(buffer, "tree ", &buffer))
 760                 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
 761         if (get_sha1_hex(buffer, tree_sha1) || buffer[40] != '\n') {
 762                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
 763                 if (err)
 764                         return err;
 765         }
 766         buffer += 41;
 767         while (skip_prefix(buffer, "parent ", &buffer)) {
 768                 if (get_sha1_hex(buffer, sha1) || buffer[40] != '\n') {
 769                         err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
 770                         if (err)
 771                                 return err;
 772                 }
 773                 buffer += 41;
 774                 parent_line_count++;
 775         }
 776         graft = lookup_commit_graft(&commit->object.oid);
 777         parent_count = commit_list_count(commit->parents);
 778         if (graft) {
 779                 if (graft->nr_parent == -1 && !parent_count)
 780                         ; /* shallow commit */
 781                 else if (graft->nr_parent != parent_count) {
 782                         err = report(options, &commit->object, FSCK_MSG_MISSING_GRAFT, "graft objects missing");
 783                         if (err)
 784                                 return err;
 785                 }
 786         } else {
 787                 if (parent_count != parent_line_count) {
 788                         err = report(options, &commit->object, FSCK_MSG_MISSING_PARENT, "parent objects missing");
 789                         if (err)
 790                                 return err;
 791                 }
 792         }
 793         author_count = 0;
 794         while (skip_prefix(buffer, "author ", &buffer)) {
 795                 author_count++;
 796                 err = fsck_ident(&buffer, &commit->object, options);
 797                 if (err)
 798                         return err;
 799         }
 800         if (author_count < 1)
 801                 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
 802         else if (author_count > 1)
 803                 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
 804         if (err)
 805                 return err;
 806         if (!skip_prefix(buffer, "committer ", &buffer))
 807                 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
 808         err = fsck_ident(&buffer, &commit->object, options);
 809         if (err)
 810                 return err;
 811         if (!get_commit_tree(commit)) {
 812                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE, "could not load commit's tree %s", sha1_to_hex(tree_sha1));
 813                 if (err)
 814                         return err;
 815         }
 816         if (memchr(buffer_begin, '\0', size)) {
 817                 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
 818                              "NUL byte in the commit object body");
 819                 if (err)
 820                         return err;
 821         }
 822         return 0;
 823 }
 824
 825 static int fsck_commit(struct commit *commit, const char *data,
 826         unsigned long size, struct fsck_options *options)
 827 {
 828         const char *buffer = data ?  data : get_commit_buffer(commit, &size);
 829         int ret = fsck_commit_buffer(commit, buffer, size, options);
 830         if (!data)
 831                 unuse_commit_buffer(commit, buffer);
 832         return ret;
 833 }
 834
 835 static int fsck_tag_buffer(struct tag *tag, const char *data,
 836         unsigned long size, struct fsck_options *options)
 837 {
 838         unsigned char sha1[20];
 839         int ret = 0;
 840         const char *buffer;
 841         char *to_free = NULL, *eol;
 842         struct strbuf sb = STRBUF_INIT;
 843
 844         if (data)
 845                 buffer = data;
 846         else {
 847                 enum object_type type;
 848
 849                 buffer = to_free =
 850                         read_object_file(&tag->object.oid, &type, &size);
 851                 if (!buffer)
 852                         return report(options, &tag->object,
 853                                 FSCK_MSG_MISSING_TAG_OBJECT,
 854                                 "cannot read tag object");
 855
 856                 if (type != OBJ_TAG) {
 857                         ret = report(options, &tag->object,
 858                                 FSCK_MSG_TAG_OBJECT_NOT_TAG,
 859                                 "expected tag got %s",
 860                             type_name(type));
 861                         goto done;
 862                 }
 863         }
 864
 865         ret = verify_headers(buffer, size, &tag->object, options);
 866         if (ret)
 867                 goto done;
 868
 869         if (!skip_prefix(buffer, "object ", &buffer)) {
 870                 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
 871                 goto done;
 872         }
 873         if (get_sha1_hex(buffer, sha1) || buffer[40] != '\n') {
 874                 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
 875                 if (ret)
 876                         goto done;
 877         }
 878         buffer += 41;
 879
 880         if (!skip_prefix(buffer, "type ", &buffer)) {
 881                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
 882                 goto done;
 883         }
 884         eol = strchr(buffer, '\n');
 885         if (!eol) {
 886                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
 887                 goto done;
 888         }
 889         if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
 890                 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
 891         if (ret)
 892                 goto done;
 893         buffer = eol + 1;
 894
 895         if (!skip_prefix(buffer, "tag ", &buffer)) {
 896                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
 897                 goto done;
 898         }
 899         eol = strchr(buffer, '\n');
 900         if (!eol) {
 901                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
 902                 goto done;
 903         }
 904         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
 905         if (check_refname_format(sb.buf, 0)) {
 906                 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
 907                            "invalid 'tag' name: %.*s",
 908                            (int)(eol - buffer), buffer);
 909                 if (ret)
 910                         goto done;
 911         }
 912         buffer = eol + 1;
 913
 914         if (!skip_prefix(buffer, "tagger ", &buffer)) {
 915                 /* early tags do not contain 'tagger' lines; warn only */
 916                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
 917                 if (ret)
 918                         goto done;
 919         }
 920         else
 921                 ret = fsck_ident(&buffer, &tag->object, options);
 922
 923 done:
 924         strbuf_release(&sb);
 925         free(to_free);
 926         return ret;
 927 }
 928
 929 static int fsck_tag(struct tag *tag, const char *data,
 930         unsigned long size, struct fsck_options *options)
 931 {
 932         struct object *tagged = tag->tagged;
 933
 934         if (!tagged)
 935                 return report(options, &tag->object, FSCK_MSG_BAD_TAG_OBJECT, "could not load tagged object");
 936
 937         return fsck_tag_buffer(tag, data, size, options);
 938 }
 939
 940 int fsck_object(struct object *obj, void *data, unsigned long size,
 941         struct fsck_options *options)
 942 {
 943         if (!obj)
 944                 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
 945
 946         if (obj->type == OBJ_BLOB)
 947                 return 0;
 948         if (obj->type == OBJ_TREE)
 949                 return fsck_tree((struct tree *) obj, options);
 950         if (obj->type == OBJ_COMMIT)
 951                 return fsck_commit((struct commit *) obj, (const char *) data,
 952                         size, options);
 953         if (obj->type == OBJ_TAG)
 954                 return fsck_tag((struct tag *) obj, (const char *) data,
 955                         size, options);
 956
 957         return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
 958                           obj->type);
 959 }
 960
 961 int fsck_error_function(struct fsck_options *o,
 962         struct object *obj, int msg_type, const char *message)
 963 {
 964         if (msg_type == FSCK_WARN) {
 965                 warning("object %s: %s", describe_object(o, obj), message);
 966                 return 0;
 967         }
 968         error("object %s: %s", describe_object(o, obj), message);
 969         return 1;
 970 }