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

   1 #include "cache.h"
   2 #include "object-store.h"
   3 #include "object.h"
   4 #include "blob.h"
   5 #include "tree.h"
   6 #include "tree-walk.h"
   7 #include "commit.h"
   8 #include "tag.h"
   9 #include "fsck.h"
  10 #include "refs.h"
  11 #include "utf8.h"
  12 #include "sha1-array.h"
  13 #include "decorate.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, FSCK_##msg_type },
  78 static struct {
  79         const char *id_string;
  80         const char *downcased;
  81         int msg_type;
  82 } msg_id_info[FSCK_MSG_MAX + 1] = {
  83         FOREACH_MSG_ID(MSG_ID)
  84         { NULL, NULL, -1 }
  85 };
  86 #undef MSG_ID
  87
  88 static int parse_msg_id(const char *text)
  89 {
  90         int i;
  91
  92         if (!msg_id_info[0].downcased) {
  93                 /* convert id_string to lower case, without underscores. */
  94                 for (i = 0; i < FSCK_MSG_MAX; i++) {
  95                         const char *p = msg_id_info[i].id_string;
  96                         int len = strlen(p);
  97                         char *q = xmalloc(len);
  98
  99                         msg_id_info[i].downcased = q;
 100                         while (*p)
 101                                 if (*p == '_')
 102                                         p++;
 103                                 else
 104                                         *(q)++ = tolower(*(p)++);
 105                         *q = '\0';
 106                 }
 107         }
 108
 109         for (i = 0; i < FSCK_MSG_MAX; i++)
 110                 if (!strcmp(text, msg_id_info[i].downcased))
 111                         return i;
 112
 113         return -1;
 114 }
 115
 116 static int fsck_msg_type(enum fsck_msg_id msg_id,
 117         struct fsck_options *options)
 118 {
 119         int msg_type;
 120
 121         assert(msg_id >= 0 && msg_id < FSCK_MSG_MAX);
 122
 123         if (options->msg_type)
 124                 msg_type = options->msg_type[msg_id];
 125         else {
 126                 msg_type = msg_id_info[msg_id].msg_type;
 127                 if (options->strict && msg_type == FSCK_WARN)
 128                         msg_type = FSCK_ERROR;
 129         }
 130
 131         return msg_type;
 132 }
 133
 134 static void init_skiplist(struct fsck_options *options, const char *path)
 135 {
 136         static struct oid_array skiplist = OID_ARRAY_INIT;
 137         int sorted, fd;
 138         char buffer[GIT_MAX_HEXSZ + 1];
 139         struct object_id oid;
 140
 141         if (options->skiplist)
 142                 sorted = options->skiplist->sorted;
 143         else {
 144                 sorted = 1;
 145                 options->skiplist = &skiplist;
 146         }
 147
 148         fd = open(path, O_RDONLY);
 149         if (fd < 0)
 150                 die("Could not open skip list: %s", path);
 151         for (;;) {
 152                 const char *p;
 153                 int result = read_in_full(fd, buffer, sizeof(buffer));
 154                 if (result < 0)
 155                         die_errno("Could not read '%s'", path);
 156                 if (!result)
 157                         break;
 158                 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n')
 159                         die("Invalid SHA-1: %s", buffer);
 160                 oid_array_append(&skiplist, &oid);
 161                 if (sorted && skiplist.nr > 1 &&
 162                                 oidcmp(&skiplist.oid[skiplist.nr - 2],
 163                                        &oid) > 0)
 164                         sorted = 0;
 165         }
 166         close(fd);
 167
 168         if (sorted)
 169                 skiplist.sorted = 1;
 170 }
 171
 172 static int parse_msg_type(const char *str)
 173 {
 174         if (!strcmp(str, "error"))
 175                 return FSCK_ERROR;
 176         else if (!strcmp(str, "warn"))
 177                 return FSCK_WARN;
 178         else if (!strcmp(str, "ignore"))
 179                 return FSCK_IGNORE;
 180         else
 181                 die("Unknown fsck message type: '%s'", str);
 182 }
 183
 184 int is_valid_msg_type(const char *msg_id, const char *msg_type)
 185 {
 186         if (parse_msg_id(msg_id) < 0)
 187                 return 0;
 188         parse_msg_type(msg_type);
 189         return 1;
 190 }
 191
 192 void fsck_set_msg_type(struct fsck_options *options,
 193                 const char *msg_id, const char *msg_type)
 194 {
 195         int id = parse_msg_id(msg_id), type;
 196
 197         if (id < 0)
 198                 die("Unhandled message id: %s", msg_id);
 199         type = parse_msg_type(msg_type);
 200
 201         if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
 202                 die("Cannot demote %s to %s", msg_id, msg_type);
 203
 204         if (!options->msg_type) {
 205                 int i;
 206                 int *msg_type;
 207                 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
 208                 for (i = 0; i < FSCK_MSG_MAX; i++)
 209                         msg_type[i] = fsck_msg_type(i, options);
 210                 options->msg_type = msg_type;
 211         }
 212
 213         options->msg_type[id] = type;
 214 }
 215
 216 void fsck_set_msg_types(struct fsck_options *options, const char *values)
 217 {
 218         char *buf = xstrdup(values), *to_free = buf;
 219         int done = 0;
 220
 221         while (!done) {
 222                 int len = strcspn(buf, " ,|"), equal;
 223
 224                 done = !buf[len];
 225                 if (!len) {
 226                         buf++;
 227                         continue;
 228                 }
 229                 buf[len] = '\0';
 230
 231                 for (equal = 0;
 232                      equal < len && buf[equal] != '=' && buf[equal] != ':';
 233                      equal++)
 234                         buf[equal] = tolower(buf[equal]);
 235                 buf[equal] = '\0';
 236
 237                 if (!strcmp(buf, "skiplist")) {
 238                         if (equal == len)
 239                                 die("skiplist requires a path");
 240                         init_skiplist(options, buf + equal + 1);
 241                         buf += len + 1;
 242                         continue;
 243                 }
 244
 245                 if (equal == len)
 246                         die("Missing '=': '%s'", buf);
 247
 248                 fsck_set_msg_type(options, buf, buf + equal + 1);
 249                 buf += len + 1;
 250         }
 251         free(to_free);
 252 }
 253
 254 static void append_msg_id(struct strbuf *sb, const char *msg_id)
 255 {
 256         for (;;) {
 257                 char c = *(msg_id)++;
 258
 259                 if (!c)
 260                         break;
 261                 if (c != '_')
 262                         strbuf_addch(sb, tolower(c));
 263                 else {
 264                         assert(*msg_id);
 265                         strbuf_addch(sb, *(msg_id)++);
 266                 }
 267         }
 268
 269         strbuf_addstr(sb, ": ");
 270 }
 271
 272 __attribute__((format (printf, 4, 5)))
 273 static int report(struct fsck_options *options, struct object *object,
 274         enum fsck_msg_id id, const char *fmt, ...)
 275 {
 276         va_list ap;
 277         struct strbuf sb = STRBUF_INIT;
 278         int msg_type = fsck_msg_type(id, options), result;
 279
 280         if (msg_type == FSCK_IGNORE)
 281                 return 0;
 282
 283         if (options->skiplist && object &&
 284                         oid_array_lookup(options->skiplist, &object->oid) >= 0)
 285                 return 0;
 286
 287         if (msg_type == FSCK_FATAL)
 288                 msg_type = FSCK_ERROR;
 289         else if (msg_type == FSCK_INFO)
 290                 msg_type = FSCK_WARN;
 291
 292         append_msg_id(&sb, msg_id_info[id].id_string);
 293
 294         va_start(ap, fmt);
 295         strbuf_vaddf(&sb, fmt, ap);
 296         result = options->error_func(options, object, msg_type, sb.buf);
 297         strbuf_release(&sb);
 298         va_end(ap);
 299
 300         return result;
 301 }
 302
 303 static char *get_object_name(struct fsck_options *options, struct object *obj)
 304 {
 305         if (!options->object_names)
 306                 return NULL;
 307         return lookup_decoration(options->object_names, obj);
 308 }
 309
 310 static void put_object_name(struct fsck_options *options, struct object *obj,
 311         const char *fmt, ...)
 312 {
 313         va_list ap;
 314         struct strbuf buf = STRBUF_INIT;
 315         char *existing;
 316
 317         if (!options->object_names)
 318                 return;
 319         existing = lookup_decoration(options->object_names, obj);
 320         if (existing)
 321                 return;
 322         va_start(ap, fmt);
 323         strbuf_vaddf(&buf, fmt, ap);
 324         add_decoration(options->object_names, obj, strbuf_detach(&buf, NULL));
 325         va_end(ap);
 326 }
 327
 328 static const char *describe_object(struct fsck_options *o, struct object *obj)
 329 {
 330         static struct strbuf buf = STRBUF_INIT;
 331         char *name;
 332
 333         strbuf_reset(&buf);
 334         strbuf_addstr(&buf, oid_to_hex(&obj->oid));
 335         if (o->object_names && (name = lookup_decoration(o->object_names, obj)))
 336                 strbuf_addf(&buf, " (%s)", name);
 337
 338         return buf.buf;
 339 }
 340
 341 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
 342 {
 343         struct tree_desc desc;
 344         struct name_entry entry;
 345         int res = 0;
 346         const char *name;
 347
 348         if (parse_tree(tree))
 349                 return -1;
 350
 351         name = get_object_name(options, &tree->object);
 352         if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
 353                 return -1;
 354         while (tree_entry_gently(&desc, &entry)) {
 355                 struct object *obj;
 356                 int result;
 357
 358                 if (S_ISGITLINK(entry.mode))
 359                         continue;
 360
 361                 if (S_ISDIR(entry.mode)) {
 362                         obj = (struct object *)lookup_tree(entry.oid);
 363                         if (name && obj)
 364                                 put_object_name(options, obj, "%s%s/", name,
 365                                         entry.path);
 366                         result = options->walk(obj, OBJ_TREE, data, options);
 367                 }
 368                 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
 369                         obj = (struct object *)lookup_blob(entry.oid);
 370                         if (name && obj)
 371                                 put_object_name(options, obj, "%s%s", name,
 372                                         entry.path);
 373                         result = options->walk(obj, OBJ_BLOB, data, options);
 374                 }
 375                 else {
 376                         result = error("in tree %s: entry %s has bad mode %.6o",
 377                                         describe_object(options, &tree->object), entry.path, entry.mode);
 378                 }
 379                 if (result < 0)
 380                         return result;
 381                 if (!res)
 382                         res = result;
 383         }
 384         return res;
 385 }
 386
 387 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
 388 {
 389         int counter = 0, generation = 0, name_prefix_len = 0;
 390         struct commit_list *parents;
 391         int res;
 392         int result;
 393         const char *name;
 394
 395         if (parse_commit(commit))
 396                 return -1;
 397
 398         name = get_object_name(options, &commit->object);
 399         if (name)
 400                 put_object_name(options, &commit->tree->object, "%s:", name);
 401
 402         result = options->walk((struct object *)commit->tree, OBJ_TREE, data, options);
 403         if (result < 0)
 404                 return result;
 405         res = result;
 406
 407         parents = commit->parents;
 408         if (name && parents) {
 409                 int len = strlen(name), power;
 410
 411                 if (len && name[len - 1] == '^') {
 412                         generation = 1;
 413                         name_prefix_len = len - 1;
 414                 }
 415                 else { /* parse ~<generation> suffix */
 416                         for (generation = 0, power = 1;
 417                              len && isdigit(name[len - 1]);
 418                              power *= 10)
 419                                 generation += power * (name[--len] - '0');
 420                         if (power > 1 && len && name[len - 1] == '~')
 421                                 name_prefix_len = len - 1;
 422                 }
 423         }
 424
 425         while (parents) {
 426                 if (name) {
 427                         struct object *obj = &parents->item->object;
 428
 429                         if (++counter > 1)
 430                                 put_object_name(options, obj, "%s^%d",
 431                                         name, counter);
 432                         else if (generation > 0)
 433                                 put_object_name(options, obj, "%.*s~%d",
 434                                         name_prefix_len, name, generation + 1);
 435                         else
 436                                 put_object_name(options, obj, "%s^", name);
 437                 }
 438                 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
 439                 if (result < 0)
 440                         return result;
 441                 if (!res)
 442                         res = result;
 443                 parents = parents->next;
 444         }
 445         return res;
 446 }
 447
 448 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
 449 {
 450         char *name = get_object_name(options, &tag->object);
 451
 452         if (parse_tag(tag))
 453                 return -1;
 454         if (name)
 455                 put_object_name(options, tag->tagged, "%s", name);
 456         return options->walk(tag->tagged, OBJ_ANY, data, options);
 457 }
 458
 459 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
 460 {
 461         if (!obj)
 462                 return -1;
 463
 464         if (obj->type == OBJ_NONE)
 465                 parse_object(&obj->oid);
 466
 467         switch (obj->type) {
 468         case OBJ_BLOB:
 469                 return 0;
 470         case OBJ_TREE:
 471                 return fsck_walk_tree((struct tree *)obj, data, options);
 472         case OBJ_COMMIT:
 473                 return fsck_walk_commit((struct commit *)obj, data, options);
 474         case OBJ_TAG:
 475                 return fsck_walk_tag((struct tag *)obj, data, options);
 476         default:
 477                 error("Unknown object type for %s", describe_object(options, obj));
 478                 return -1;
 479         }
 480 }
 481
 482 /*
 483  * The entries in a tree are ordered in the _path_ order,
 484  * which means that a directory entry is ordered by adding
 485  * a slash to the end of it.
 486  *
 487  * So a directory called "a" is ordered _after_ a file
 488  * called "a.c", because "a/" sorts after "a.c".
 489  */
 490 #define TREE_UNORDERED (-1)
 491 #define TREE_HAS_DUPS  (-2)
 492
 493 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
 494 {
 495         int len1 = strlen(name1);
 496         int len2 = strlen(name2);
 497         int len = len1 < len2 ? len1 : len2;
 498         unsigned char c1, c2;
 499         int cmp;
 500
 501         cmp = memcmp(name1, name2, len);
 502         if (cmp < 0)
 503                 return 0;
 504         if (cmp > 0)
 505                 return TREE_UNORDERED;
 506
 507         /*
 508          * Ok, the first <len> characters are the same.
 509          * Now we need to order the next one, but turn
 510          * a '\0' into a '/' for a directory entry.
 511          */
 512         c1 = name1[len];
 513         c2 = name2[len];
 514         if (!c1 && !c2)
 515                 /*
 516                  * git-write-tree used to write out a nonsense tree that has
 517                  * entries with the same name, one blob and one tree.  Make
 518                  * sure we do not have duplicate entries.
 519                  */
 520                 return TREE_HAS_DUPS;
 521         if (!c1 && S_ISDIR(mode1))
 522                 c1 = '/';
 523         if (!c2 && S_ISDIR(mode2))
 524                 c2 = '/';
 525         return c1 < c2 ? 0 : TREE_UNORDERED;
 526 }
 527
 528 static int fsck_tree(struct tree *item, struct fsck_options *options)
 529 {
 530         int retval = 0;
 531         int has_null_sha1 = 0;
 532         int has_full_path = 0;
 533         int has_empty_name = 0;
 534         int has_dot = 0;
 535         int has_dotdot = 0;
 536         int has_dotgit = 0;
 537         int has_zero_pad = 0;
 538         int has_bad_modes = 0;
 539         int has_dup_entries = 0;
 540         int not_properly_sorted = 0;
 541         struct tree_desc desc;
 542         unsigned o_mode;
 543         const char *o_name;
 544
 545         if (init_tree_desc_gently(&desc, item->buffer, item->size)) {
 546                 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 547                 return retval;
 548         }
 549
 550         o_mode = 0;
 551         o_name = NULL;
 552
 553         while (desc.size) {
 554                 unsigned mode;
 555                 const char *name;
 556                 const struct object_id *oid;
 557
 558                 oid = tree_entry_extract(&desc, &name, &mode);
 559
 560                 has_null_sha1 |= is_null_oid(oid);
 561                 has_full_path |= !!strchr(name, '/');
 562                 has_empty_name |= !*name;
 563                 has_dot |= !strcmp(name, ".");
 564                 has_dotdot |= !strcmp(name, "..");
 565                 has_dotgit |= (!strcmp(name, ".git") ||
 566                                is_hfs_dotgit(name) ||
 567                                is_ntfs_dotgit(name));
 568                 has_zero_pad |= *(char *)desc.buffer == '0';
 569                 if (update_tree_entry_gently(&desc)) {
 570                         retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 571                         break;
 572                 }
 573
 574                 switch (mode) {
 575                 /*
 576                  * Standard modes..
 577                  */
 578                 case S_IFREG | 0755:
 579                 case S_IFREG | 0644:
 580                 case S_IFLNK:
 581                 case S_IFDIR:
 582                 case S_IFGITLINK:
 583                         break;
 584                 /*
 585                  * This is nonstandard, but we had a few of these
 586                  * early on when we honored the full set of mode
 587                  * bits..
 588                  */
 589                 case S_IFREG | 0664:
 590                         if (!options->strict)
 591                                 break;
 592                         /* fallthrough */
 593                 default:
 594                         has_bad_modes = 1;
 595                 }
 596
 597                 if (o_name) {
 598                         switch (verify_ordered(o_mode, o_name, mode, name)) {
 599                         case TREE_UNORDERED:
 600                                 not_properly_sorted = 1;
 601                                 break;
 602                         case TREE_HAS_DUPS:
 603                                 has_dup_entries = 1;
 604                                 break;
 605                         default:
 606                                 break;
 607                         }
 608                 }
 609
 610                 o_mode = mode;
 611                 o_name = name;
 612         }
 613
 614         if (has_null_sha1)
 615                 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
 616         if (has_full_path)
 617                 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
 618         if (has_empty_name)
 619                 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
 620         if (has_dot)
 621                 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
 622         if (has_dotdot)
 623                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
 624         if (has_dotgit)
 625                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
 626         if (has_zero_pad)
 627                 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
 628         if (has_bad_modes)
 629                 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
 630         if (has_dup_entries)
 631                 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
 632         if (not_properly_sorted)
 633                 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
 634         return retval;
 635 }
 636
 637 static int verify_headers(const void *data, unsigned long size,
 638                           struct object *obj, struct fsck_options *options)
 639 {
 640         const char *buffer = (const char *)data;
 641         unsigned long i;
 642
 643         for (i = 0; i < size; i++) {
 644                 switch (buffer[i]) {
 645                 case '\0':
 646                         return report(options, obj,
 647                                 FSCK_MSG_NUL_IN_HEADER,
 648                                 "unterminated header: NUL at offset %ld", i);
 649                 case '\n':
 650                         if (i + 1 < size && buffer[i + 1] == '\n')
 651                                 return 0;
 652                 }
 653         }
 654
 655         /*
 656          * We did not find double-LF that separates the header
 657          * and the body.  Not having a body is not a crime but
 658          * we do want to see the terminating LF for the last header
 659          * line.
 660          */
 661         if (size && buffer[size - 1] == '\n')
 662                 return 0;
 663
 664         return report(options, obj,
 665                 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
 666 }
 667
 668 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
 669 {
 670         const char *p = *ident;
 671         char *end;
 672
 673         *ident = strchrnul(*ident, '\n');
 674         if (**ident == '\n')
 675                 (*ident)++;
 676
 677         if (*p == '<')
 678                 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 679         p += strcspn(p, "<>\n");
 680         if (*p == '>')
 681                 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
 682         if (*p != '<')
 683                 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
 684         if (p[-1] != ' ')
 685                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 686         p++;
 687         p += strcspn(p, "<>\n");
 688         if (*p != '>')
 689                 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
 690         p++;
 691         if (*p != ' ')
 692                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
 693         p++;
 694         if (*p == '0' && p[1] != ' ')
 695                 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
 696         if (date_overflows(parse_timestamp(p, &end, 10)))
 697                 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
 698         if ((end == p || *end != ' '))
 699                 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
 700         p = end + 1;
 701         if ((*p != '+' && *p != '-') ||
 702             !isdigit(p[1]) ||
 703             !isdigit(p[2]) ||
 704             !isdigit(p[3]) ||
 705             !isdigit(p[4]) ||
 706             (p[5] != '\n'))
 707                 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
 708         p += 6;
 709         return 0;
 710 }
 711
 712 static int fsck_commit_buffer(struct commit *commit, const char *buffer,
 713         unsigned long size, struct fsck_options *options)
 714 {
 715         unsigned char tree_sha1[20], sha1[20];
 716         struct commit_graft *graft;
 717         unsigned parent_count, parent_line_count = 0, author_count;
 718         int err;
 719         const char *buffer_begin = buffer;
 720
 721         if (verify_headers(buffer, size, &commit->object, options))
 722                 return -1;
 723
 724         if (!skip_prefix(buffer, "tree ", &buffer))
 725                 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
 726         if (get_sha1_hex(buffer, tree_sha1) || buffer[40] != '\n') {
 727                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
 728                 if (err)
 729                         return err;
 730         }
 731         buffer += 41;
 732         while (skip_prefix(buffer, "parent ", &buffer)) {
 733                 if (get_sha1_hex(buffer, sha1) || buffer[40] != '\n') {
 734                         err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
 735                         if (err)
 736                                 return err;
 737                 }
 738                 buffer += 41;
 739                 parent_line_count++;
 740         }
 741         graft = lookup_commit_graft(the_repository, &commit->object.oid);
 742         parent_count = commit_list_count(commit->parents);
 743         if (graft) {
 744                 if (graft->nr_parent == -1 && !parent_count)
 745                         ; /* shallow commit */
 746                 else if (graft->nr_parent != parent_count) {
 747                         err = report(options, &commit->object, FSCK_MSG_MISSING_GRAFT, "graft objects missing");
 748                         if (err)
 749                                 return err;
 750                 }
 751         } else {
 752                 if (parent_count != parent_line_count) {
 753                         err = report(options, &commit->object, FSCK_MSG_MISSING_PARENT, "parent objects missing");
 754                         if (err)
 755                                 return err;
 756                 }
 757         }
 758         author_count = 0;
 759         while (skip_prefix(buffer, "author ", &buffer)) {
 760                 author_count++;
 761                 err = fsck_ident(&buffer, &commit->object, options);
 762                 if (err)
 763                         return err;
 764         }
 765         if (author_count < 1)
 766                 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
 767         else if (author_count > 1)
 768                 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
 769         if (err)
 770                 return err;
 771         if (!skip_prefix(buffer, "committer ", &buffer))
 772                 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
 773         err = fsck_ident(&buffer, &commit->object, options);
 774         if (err)
 775                 return err;
 776         if (!commit->tree) {
 777                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE, "could not load commit's tree %s", sha1_to_hex(tree_sha1));
 778                 if (err)
 779                         return err;
 780         }
 781         if (memchr(buffer_begin, '\0', size)) {
 782                 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
 783                              "NUL byte in the commit object body");
 784                 if (err)
 785                         return err;
 786         }
 787         return 0;
 788 }
 789
 790 static int fsck_commit(struct commit *commit, const char *data,
 791         unsigned long size, struct fsck_options *options)
 792 {
 793         const char *buffer = data ?  data : get_commit_buffer(commit, &size);
 794         int ret = fsck_commit_buffer(commit, buffer, size, options);
 795         if (!data)
 796                 unuse_commit_buffer(commit, buffer);
 797         return ret;
 798 }
 799
 800 static int fsck_tag_buffer(struct tag *tag, const char *data,
 801         unsigned long size, struct fsck_options *options)
 802 {
 803         unsigned char sha1[20];
 804         int ret = 0;
 805         const char *buffer;
 806         char *to_free = NULL, *eol;
 807         struct strbuf sb = STRBUF_INIT;
 808
 809         if (data)
 810                 buffer = data;
 811         else {
 812                 enum object_type type;
 813
 814                 buffer = to_free =
 815                         read_object_file(&tag->object.oid, &type, &size);
 816                 if (!buffer)
 817                         return report(options, &tag->object,
 818                                 FSCK_MSG_MISSING_TAG_OBJECT,
 819                                 "cannot read tag object");
 820
 821                 if (type != OBJ_TAG) {
 822                         ret = report(options, &tag->object,
 823                                 FSCK_MSG_TAG_OBJECT_NOT_TAG,
 824                                 "expected tag got %s",
 825                             type_name(type));
 826                         goto done;
 827                 }
 828         }
 829
 830         ret = verify_headers(buffer, size, &tag->object, options);
 831         if (ret)
 832                 goto done;
 833
 834         if (!skip_prefix(buffer, "object ", &buffer)) {
 835                 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
 836                 goto done;
 837         }
 838         if (get_sha1_hex(buffer, sha1) || buffer[40] != '\n') {
 839                 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
 840                 if (ret)
 841                         goto done;
 842         }
 843         buffer += 41;
 844
 845         if (!skip_prefix(buffer, "type ", &buffer)) {
 846                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
 847                 goto done;
 848         }
 849         eol = strchr(buffer, '\n');
 850         if (!eol) {
 851                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
 852                 goto done;
 853         }
 854         if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
 855                 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
 856         if (ret)
 857                 goto done;
 858         buffer = eol + 1;
 859
 860         if (!skip_prefix(buffer, "tag ", &buffer)) {
 861                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
 862                 goto done;
 863         }
 864         eol = strchr(buffer, '\n');
 865         if (!eol) {
 866                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
 867                 goto done;
 868         }
 869         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
 870         if (check_refname_format(sb.buf, 0)) {
 871                 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
 872                            "invalid 'tag' name: %.*s",
 873                            (int)(eol - buffer), buffer);
 874                 if (ret)
 875                         goto done;
 876         }
 877         buffer = eol + 1;
 878
 879         if (!skip_prefix(buffer, "tagger ", &buffer)) {
 880                 /* early tags do not contain 'tagger' lines; warn only */
 881                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
 882                 if (ret)
 883                         goto done;
 884         }
 885         else
 886                 ret = fsck_ident(&buffer, &tag->object, options);
 887
 888 done:
 889         strbuf_release(&sb);
 890         free(to_free);
 891         return ret;
 892 }
 893
 894 static int fsck_tag(struct tag *tag, const char *data,
 895         unsigned long size, struct fsck_options *options)
 896 {
 897         struct object *tagged = tag->tagged;
 898
 899         if (!tagged)
 900                 return report(options, &tag->object, FSCK_MSG_BAD_TAG_OBJECT, "could not load tagged object");
 901
 902         return fsck_tag_buffer(tag, data, size, options);
 903 }
 904
 905 int fsck_object(struct object *obj, void *data, unsigned long size,
 906         struct fsck_options *options)
 907 {
 908         if (!obj)
 909                 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
 910
 911         if (obj->type == OBJ_BLOB)
 912                 return 0;
 913         if (obj->type == OBJ_TREE)
 914                 return fsck_tree((struct tree *) obj, options);
 915         if (obj->type == OBJ_COMMIT)
 916                 return fsck_commit((struct commit *) obj, (const char *) data,
 917                         size, options);
 918         if (obj->type == OBJ_TAG)
 919                 return fsck_tag((struct tag *) obj, (const char *) data,
 920                         size, options);
 921
 922         return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
 923                           obj->type);
 924 }
 925
 926 int fsck_error_function(struct fsck_options *o,
 927         struct object *obj, int msg_type, const char *message)
 928 {
 929         if (msg_type == FSCK_WARN) {
 930                 warning("object %s: %s", describe_object(o, obj), message);
 931                 return 0;
 932         }
 933         error("object %s: %s", describe_object(o, obj), message);
 934         return 1;
 935 }