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

   1 #include "cache.h"
   2 #include "object-store.h"
   3 #include "repository.h"
   4 #include "object.h"
   5 #include "blob.h"
   6 #include "tree.h"
   7 #include "tree-walk.h"
   8 #include "commit.h"
   9 #include "tag.h"
  10 #include "fsck.h"
  11 #include "refs.h"
  12 #include "url.h"
  13 #include "utf8.h"
  14 #include "decorate.h"
  15 #include "oidset.h"
  16 #include "packfile.h"
  17 #include "submodule-config.h"
  18 #include "config.h"
  19 #include "credential.h"
  20 #include "help.h"
  21
  22 static struct oidset gitmodules_found = OIDSET_INIT;
  23 static struct oidset gitmodules_done = OIDSET_INIT;
  24
  25 #define STR(x) #x
  26 #define MSG_ID(id, msg_type) { STR(id), NULL, NULL, FSCK_##msg_type },
  27 static struct {
  28         const char *id_string;
  29         const char *downcased;
  30         const char *camelcased;
  31         enum fsck_msg_type msg_type;
  32 } msg_id_info[FSCK_MSG_MAX + 1] = {
  33         FOREACH_FSCK_MSG_ID(MSG_ID)
  34         { NULL, NULL, NULL, -1 }
  35 };
  36 #undef MSG_ID
  37 #undef STR
  38
  39 static void prepare_msg_ids(void)
  40 {
  41         int i;
  42
  43         if (msg_id_info[0].downcased)
  44                 return;
  45
  46         /* convert id_string to lower case, without underscores. */
  47         for (i = 0; i < FSCK_MSG_MAX; i++) {
  48                 const char *p = msg_id_info[i].id_string;
  49                 int len = strlen(p);
  50                 char *q = xmalloc(len);
  51
  52                 msg_id_info[i].downcased = q;
  53                 while (*p)
  54                         if (*p == '_')
  55                                 p++;
  56                         else
  57                                 *(q)++ = tolower(*(p)++);
  58                 *q = '\0';
  59
  60                 p = msg_id_info[i].id_string;
  61                 q = xmalloc(len);
  62                 msg_id_info[i].camelcased = q;
  63                 while (*p) {
  64                         if (*p == '_') {
  65                                 p++;
  66                                 if (*p)
  67                                         *q++ = *p++;
  68                         } else {
  69                                 *q++ = tolower(*p++);
  70                         }
  71                 }
  72                 *q = '\0';
  73         }
  74 }
  75
  76 static int parse_msg_id(const char *text)
  77 {
  78         int i;
  79
  80         prepare_msg_ids();
  81
  82         for (i = 0; i < FSCK_MSG_MAX; i++)
  83                 if (!strcmp(text, msg_id_info[i].downcased))
  84                         return i;
  85
  86         return -1;
  87 }
  88
  89 void list_config_fsck_msg_ids(struct string_list *list, const char *prefix)
  90 {
  91         int i;
  92
  93         prepare_msg_ids();
  94
  95         for (i = 0; i < FSCK_MSG_MAX; i++)
  96                 list_config_item(list, prefix, msg_id_info[i].camelcased);
  97 }
  98
  99 static enum fsck_msg_type fsck_msg_type(enum fsck_msg_id msg_id,
 100         struct fsck_options *options)
 101 {
 102         assert(msg_id >= 0 && msg_id < FSCK_MSG_MAX);
 103
 104         if (!options->msg_type) {
 105                 enum fsck_msg_type msg_type = msg_id_info[msg_id].msg_type;
 106
 107                 if (options->strict && msg_type == FSCK_WARN)
 108                         msg_type = FSCK_ERROR;
 109                 return msg_type;
 110         }
 111
 112         return options->msg_type[msg_id];
 113 }
 114
 115 static enum fsck_msg_type parse_msg_type(const char *str)
 116 {
 117         if (!strcmp(str, "error"))
 118                 return FSCK_ERROR;
 119         else if (!strcmp(str, "warn"))
 120                 return FSCK_WARN;
 121         else if (!strcmp(str, "ignore"))
 122                 return FSCK_IGNORE;
 123         else
 124                 die("Unknown fsck message type: '%s'", str);
 125 }
 126
 127 int is_valid_msg_type(const char *msg_id, const char *msg_type)
 128 {
 129         if (parse_msg_id(msg_id) < 0)
 130                 return 0;
 131         parse_msg_type(msg_type);
 132         return 1;
 133 }
 134
 135 void fsck_set_msg_type(struct fsck_options *options,
 136                        const char *msg_id_str, const char *msg_type_str)
 137 {
 138         int msg_id = parse_msg_id(msg_id_str);
 139         enum fsck_msg_type msg_type = parse_msg_type(msg_type_str);
 140
 141         if (msg_id < 0)
 142                 die("Unhandled message id: %s", msg_id_str);
 143
 144         if (msg_type != FSCK_ERROR && msg_id_info[msg_id].msg_type == FSCK_FATAL)
 145                 die("Cannot demote %s to %s", msg_id_str, msg_type_str);
 146
 147         if (!options->msg_type) {
 148                 int i;
 149                 enum fsck_msg_type *severity;
 150                 ALLOC_ARRAY(severity, FSCK_MSG_MAX);
 151                 for (i = 0; i < FSCK_MSG_MAX; i++)
 152                         severity[i] = fsck_msg_type(i, options);
 153                 options->msg_type = severity;
 154         }
 155
 156         options->msg_type[msg_id] = msg_type;
 157 }
 158
 159 void fsck_set_msg_types(struct fsck_options *options, const char *values)
 160 {
 161         char *buf = xstrdup(values), *to_free = buf;
 162         int done = 0;
 163
 164         while (!done) {
 165                 int len = strcspn(buf, " ,|"), equal;
 166
 167                 done = !buf[len];
 168                 if (!len) {
 169                         buf++;
 170                         continue;
 171                 }
 172                 buf[len] = '\0';
 173
 174                 for (equal = 0;
 175                      equal < len && buf[equal] != '=' && buf[equal] != ':';
 176                      equal++)
 177                         buf[equal] = tolower(buf[equal]);
 178                 buf[equal] = '\0';
 179
 180                 if (!strcmp(buf, "skiplist")) {
 181                         if (equal == len)
 182                                 die("skiplist requires a path");
 183                         oidset_parse_file(&options->skiplist, buf + equal + 1);
 184                         buf += len + 1;
 185                         continue;
 186                 }
 187
 188                 if (equal == len)
 189                         die("Missing '=': '%s'", buf);
 190
 191                 fsck_set_msg_type(options, buf, buf + equal + 1);
 192                 buf += len + 1;
 193         }
 194         free(to_free);
 195 }
 196
 197 static int object_on_skiplist(struct fsck_options *opts,
 198                               const struct object_id *oid)
 199 {
 200         return opts && oid && oidset_contains(&opts->skiplist, oid);
 201 }
 202
 203 __attribute__((format (printf, 5, 6)))
 204 static int report(struct fsck_options *options,
 205                   const struct object_id *oid, enum object_type object_type,
 206                   enum fsck_msg_id msg_id, const char *fmt, ...)
 207 {
 208         va_list ap;
 209         struct strbuf sb = STRBUF_INIT;
 210         enum fsck_msg_type msg_type = fsck_msg_type(msg_id, options);
 211         int result;
 212
 213         if (msg_type == FSCK_IGNORE)
 214                 return 0;
 215
 216         if (object_on_skiplist(options, oid))
 217                 return 0;
 218
 219         if (msg_type == FSCK_FATAL)
 220                 msg_type = FSCK_ERROR;
 221         else if (msg_type == FSCK_INFO)
 222                 msg_type = FSCK_WARN;
 223
 224         prepare_msg_ids();
 225         strbuf_addf(&sb, "%s: ", msg_id_info[msg_id].camelcased);
 226
 227         va_start(ap, fmt);
 228         strbuf_vaddf(&sb, fmt, ap);
 229         result = options->error_func(options, oid, object_type,
 230                                      msg_type, sb.buf);
 231         strbuf_release(&sb);
 232         va_end(ap);
 233
 234         return result;
 235 }
 236
 237 void fsck_enable_object_names(struct fsck_options *options)
 238 {
 239         if (!options->object_names)
 240                 options->object_names = kh_init_oid_map();
 241 }
 242
 243 const char *fsck_get_object_name(struct fsck_options *options,
 244                                  const struct object_id *oid)
 245 {
 246         khiter_t pos;
 247         if (!options->object_names)
 248                 return NULL;
 249         pos = kh_get_oid_map(options->object_names, *oid);
 250         if (pos >= kh_end(options->object_names))
 251                 return NULL;
 252         return kh_value(options->object_names, pos);
 253 }
 254
 255 void fsck_put_object_name(struct fsck_options *options,
 256                           const struct object_id *oid,
 257                           const char *fmt, ...)
 258 {
 259         va_list ap;
 260         struct strbuf buf = STRBUF_INIT;
 261         khiter_t pos;
 262         int hashret;
 263
 264         if (!options->object_names)
 265                 return;
 266
 267         pos = kh_put_oid_map(options->object_names, *oid, &hashret);
 268         if (!hashret)
 269                 return;
 270         va_start(ap, fmt);
 271         strbuf_vaddf(&buf, fmt, ap);
 272         kh_value(options->object_names, pos) = strbuf_detach(&buf, NULL);
 273         va_end(ap);
 274 }
 275
 276 const char *fsck_describe_object(struct fsck_options *options,
 277                                  const struct object_id *oid)
 278 {
 279         static struct strbuf bufs[] = {
 280                 STRBUF_INIT, STRBUF_INIT, STRBUF_INIT, STRBUF_INIT
 281         };
 282         static int b = 0;
 283         struct strbuf *buf;
 284         const char *name = fsck_get_object_name(options, oid);
 285
 286         buf = bufs + b;
 287         b = (b + 1) % ARRAY_SIZE(bufs);
 288         strbuf_reset(buf);
 289         strbuf_addstr(buf, oid_to_hex(oid));
 290         if (name)
 291                 strbuf_addf(buf, " (%s)", name);
 292
 293         return buf->buf;
 294 }
 295
 296 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
 297 {
 298         struct tree_desc desc;
 299         struct name_entry entry;
 300         int res = 0;
 301         const char *name;
 302
 303         if (parse_tree(tree))
 304                 return -1;
 305
 306         name = fsck_get_object_name(options, &tree->object.oid);
 307         if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
 308                 return -1;
 309         while (tree_entry_gently(&desc, &entry)) {
 310                 struct object *obj;
 311                 int result;
 312
 313                 if (S_ISGITLINK(entry.mode))
 314                         continue;
 315
 316                 if (S_ISDIR(entry.mode)) {
 317                         obj = (struct object *)lookup_tree(the_repository, &entry.oid);
 318                         if (name && obj)
 319                                 fsck_put_object_name(options, &entry.oid, "%s%s/",
 320                                                      name, entry.path);
 321                         result = options->walk(obj, OBJ_TREE, data, options);
 322                 }
 323                 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
 324                         obj = (struct object *)lookup_blob(the_repository, &entry.oid);
 325                         if (name && obj)
 326                                 fsck_put_object_name(options, &entry.oid, "%s%s",
 327                                                      name, entry.path);
 328                         result = options->walk(obj, OBJ_BLOB, data, options);
 329                 }
 330                 else {
 331                         result = error("in tree %s: entry %s has bad mode %.6o",
 332                                        fsck_describe_object(options, &tree->object.oid),
 333                                        entry.path, entry.mode);
 334                 }
 335                 if (result < 0)
 336                         return result;
 337                 if (!res)
 338                         res = result;
 339         }
 340         return res;
 341 }
 342
 343 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
 344 {
 345         int counter = 0, generation = 0, name_prefix_len = 0;
 346         struct commit_list *parents;
 347         int res;
 348         int result;
 349         const char *name;
 350
 351         if (parse_commit(commit))
 352                 return -1;
 353
 354         name = fsck_get_object_name(options, &commit->object.oid);
 355         if (name)
 356                 fsck_put_object_name(options, get_commit_tree_oid(commit),
 357                                      "%s:", name);
 358
 359         result = options->walk((struct object *)get_commit_tree(commit),
 360                                OBJ_TREE, data, options);
 361         if (result < 0)
 362                 return result;
 363         res = result;
 364
 365         parents = commit->parents;
 366         if (name && parents) {
 367                 int len = strlen(name), power;
 368
 369                 if (len && name[len - 1] == '^') {
 370                         generation = 1;
 371                         name_prefix_len = len - 1;
 372                 }
 373                 else { /* parse ~<generation> suffix */
 374                         for (generation = 0, power = 1;
 375                              len && isdigit(name[len - 1]);
 376                              power *= 10)
 377                                 generation += power * (name[--len] - '0');
 378                         if (power > 1 && len && name[len - 1] == '~')
 379                                 name_prefix_len = len - 1;
 380                         else {
 381                                 /* Maybe a non-first parent, e.g. HEAD^2 */
 382                                 generation = 0;
 383                                 name_prefix_len = len;
 384                         }
 385                 }
 386         }
 387
 388         while (parents) {
 389                 if (name) {
 390                         struct object_id *oid = &parents->item->object.oid;
 391
 392                         if (counter++)
 393                                 fsck_put_object_name(options, oid, "%s^%d",
 394                                                      name, counter);
 395                         else if (generation > 0)
 396                                 fsck_put_object_name(options, oid, "%.*s~%d",
 397                                                      name_prefix_len, name,
 398                                                      generation + 1);
 399                         else
 400                                 fsck_put_object_name(options, oid, "%s^", name);
 401                 }
 402                 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
 403                 if (result < 0)
 404                         return result;
 405                 if (!res)
 406                         res = result;
 407                 parents = parents->next;
 408         }
 409         return res;
 410 }
 411
 412 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
 413 {
 414         const char *name = fsck_get_object_name(options, &tag->object.oid);
 415
 416         if (parse_tag(tag))
 417                 return -1;
 418         if (name)
 419                 fsck_put_object_name(options, &tag->tagged->oid, "%s", name);
 420         return options->walk(tag->tagged, OBJ_ANY, data, options);
 421 }
 422
 423 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
 424 {
 425         if (!obj)
 426                 return -1;
 427
 428         if (obj->type == OBJ_NONE)
 429                 parse_object(the_repository, &obj->oid);
 430
 431         switch (obj->type) {
 432         case OBJ_BLOB:
 433                 return 0;
 434         case OBJ_TREE:
 435                 return fsck_walk_tree((struct tree *)obj, data, options);
 436         case OBJ_COMMIT:
 437                 return fsck_walk_commit((struct commit *)obj, data, options);
 438         case OBJ_TAG:
 439                 return fsck_walk_tag((struct tag *)obj, data, options);
 440         default:
 441                 error("Unknown object type for %s",
 442                       fsck_describe_object(options, &obj->oid));
 443                 return -1;
 444         }
 445 }
 446
 447 struct name_stack {
 448         const char **names;
 449         size_t nr, alloc;
 450 };
 451
 452 static void name_stack_push(struct name_stack *stack, const char *name)
 453 {
 454         ALLOC_GROW(stack->names, stack->nr + 1, stack->alloc);
 455         stack->names[stack->nr++] = name;
 456 }
 457
 458 static const char *name_stack_pop(struct name_stack *stack)
 459 {
 460         return stack->nr ? stack->names[--stack->nr] : NULL;
 461 }
 462
 463 static void name_stack_clear(struct name_stack *stack)
 464 {
 465         FREE_AND_NULL(stack->names);
 466         stack->nr = stack->alloc = 0;
 467 }
 468
 469 /*
 470  * The entries in a tree are ordered in the _path_ order,
 471  * which means that a directory entry is ordered by adding
 472  * a slash to the end of it.
 473  *
 474  * So a directory called "a" is ordered _after_ a file
 475  * called "a.c", because "a/" sorts after "a.c".
 476  */
 477 #define TREE_UNORDERED (-1)
 478 #define TREE_HAS_DUPS  (-2)
 479
 480 static int is_less_than_slash(unsigned char c)
 481 {
 482         return '\0' < c && c < '/';
 483 }
 484
 485 static int verify_ordered(unsigned mode1, const char *name1,
 486                           unsigned mode2, const char *name2,
 487                           struct name_stack *candidates)
 488 {
 489         int len1 = strlen(name1);
 490         int len2 = strlen(name2);
 491         int len = len1 < len2 ? len1 : len2;
 492         unsigned char c1, c2;
 493         int cmp;
 494
 495         cmp = memcmp(name1, name2, len);
 496         if (cmp < 0)
 497                 return 0;
 498         if (cmp > 0)
 499                 return TREE_UNORDERED;
 500
 501         /*
 502          * Ok, the first <len> characters are the same.
 503          * Now we need to order the next one, but turn
 504          * a '\0' into a '/' for a directory entry.
 505          */
 506         c1 = name1[len];
 507         c2 = name2[len];
 508         if (!c1 && !c2)
 509                 /*
 510                  * git-write-tree used to write out a nonsense tree that has
 511                  * entries with the same name, one blob and one tree.  Make
 512                  * sure we do not have duplicate entries.
 513                  */
 514                 return TREE_HAS_DUPS;
 515         if (!c1 && S_ISDIR(mode1))
 516                 c1 = '/';
 517         if (!c2 && S_ISDIR(mode2))
 518                 c2 = '/';
 519
 520         /*
 521          * There can be non-consecutive duplicates due to the implicitly
 522          * added slash, e.g.:
 523          *
 524          *   foo
 525          *   foo.bar
 526          *   foo.bar.baz
 527          *   foo.bar/
 528          *   foo/
 529          *
 530          * Record non-directory candidates (like "foo" and "foo.bar" in
 531          * the example) on a stack and check directory candidates (like
 532          * foo/" and "foo.bar/") against that stack.
 533          */
 534         if (!c1 && is_less_than_slash(c2)) {
 535                 name_stack_push(candidates, name1);
 536         } else if (c2 == '/' && is_less_than_slash(c1)) {
 537                 for (;;) {
 538                         const char *p;
 539                         const char *f_name = name_stack_pop(candidates);
 540
 541                         if (!f_name)
 542                                 break;
 543                         if (!skip_prefix(name2, f_name, &p))
 544                                 continue;
 545                         if (!*p)
 546                                 return TREE_HAS_DUPS;
 547                         if (is_less_than_slash(*p)) {
 548                                 name_stack_push(candidates, f_name);
 549                                 break;
 550                         }
 551                 }
 552         }
 553
 554         return c1 < c2 ? 0 : TREE_UNORDERED;
 555 }
 556
 557 static int fsck_tree(const struct object_id *oid,
 558                      const char *buffer, unsigned long size,
 559                      struct fsck_options *options)
 560 {
 561         int retval = 0;
 562         int has_null_sha1 = 0;
 563         int has_full_path = 0;
 564         int has_empty_name = 0;
 565         int has_dot = 0;
 566         int has_dotdot = 0;
 567         int has_dotgit = 0;
 568         int has_zero_pad = 0;
 569         int has_bad_modes = 0;
 570         int has_dup_entries = 0;
 571         int not_properly_sorted = 0;
 572         struct tree_desc desc;
 573         unsigned o_mode;
 574         const char *o_name;
 575         struct name_stack df_dup_candidates = { NULL };
 576
 577         if (init_tree_desc_gently(&desc, buffer, size)) {
 578                 retval += report(options, oid, OBJ_TREE, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 579                 return retval;
 580         }
 581
 582         o_mode = 0;
 583         o_name = NULL;
 584
 585         while (desc.size) {
 586                 unsigned short mode;
 587                 const char *name, *backslash;
 588                 const struct object_id *oid;
 589
 590                 oid = tree_entry_extract(&desc, &name, &mode);
 591
 592                 has_null_sha1 |= is_null_oid(oid);
 593                 has_full_path |= !!strchr(name, '/');
 594                 has_empty_name |= !*name;
 595                 has_dot |= !strcmp(name, ".");
 596                 has_dotdot |= !strcmp(name, "..");
 597                 has_dotgit |= is_hfs_dotgit(name) || is_ntfs_dotgit(name);
 598                 has_zero_pad |= *(char *)desc.buffer == '0';
 599
 600                 if (is_hfs_dotgitmodules(name) || is_ntfs_dotgitmodules(name)) {
 601                         if (!S_ISLNK(mode))
 602                                 oidset_insert(&gitmodules_found, oid);
 603                         else
 604                                 retval += report(options,
 605                                                  oid, OBJ_TREE,
 606                                                  FSCK_MSG_GITMODULES_SYMLINK,
 607                                                  ".gitmodules is a symbolic link");
 608                 }
 609
 610                 if ((backslash = strchr(name, '\\'))) {
 611                         while (backslash) {
 612                                 backslash++;
 613                                 has_dotgit |= is_ntfs_dotgit(backslash);
 614                                 if (is_ntfs_dotgitmodules(backslash)) {
 615                                         if (!S_ISLNK(mode))
 616                                                 oidset_insert(&gitmodules_found, oid);
 617                                         else
 618                                                 retval += report(options, oid, OBJ_TREE,
 619                                                                  FSCK_MSG_GITMODULES_SYMLINK,
 620                                                                  ".gitmodules is a symbolic link");
 621                                 }
 622                                 backslash = strchr(backslash, '\\');
 623                         }
 624                 }
 625
 626                 if (update_tree_entry_gently(&desc)) {
 627                         retval += report(options, oid, OBJ_TREE, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
 628                         break;
 629                 }
 630
 631                 switch (mode) {
 632                 /*
 633                  * Standard modes..
 634                  */
 635                 case S_IFREG | 0755:
 636                 case S_IFREG | 0644:
 637                 case S_IFLNK:
 638                 case S_IFDIR:
 639                 case S_IFGITLINK:
 640                         break;
 641                 /*
 642                  * This is nonstandard, but we had a few of these
 643                  * early on when we honored the full set of mode
 644                  * bits..
 645                  */
 646                 case S_IFREG | 0664:
 647                         if (!options->strict)
 648                                 break;
 649                         /* fallthrough */
 650                 default:
 651                         has_bad_modes = 1;
 652                 }
 653
 654                 if (o_name) {
 655                         switch (verify_ordered(o_mode, o_name, mode, name,
 656                                                &df_dup_candidates)) {
 657                         case TREE_UNORDERED:
 658                                 not_properly_sorted = 1;
 659                                 break;
 660                         case TREE_HAS_DUPS:
 661                                 has_dup_entries = 1;
 662                                 break;
 663                         default:
 664                                 break;
 665                         }
 666                 }
 667
 668                 o_mode = mode;
 669                 o_name = name;
 670         }
 671
 672         name_stack_clear(&df_dup_candidates);
 673
 674         if (has_null_sha1)
 675                 retval += report(options, oid, OBJ_TREE, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
 676         if (has_full_path)
 677                 retval += report(options, oid, OBJ_TREE, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
 678         if (has_empty_name)
 679                 retval += report(options, oid, OBJ_TREE, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
 680         if (has_dot)
 681                 retval += report(options, oid, OBJ_TREE, FSCK_MSG_HAS_DOT, "contains '.'");
 682         if (has_dotdot)
 683                 retval += report(options, oid, OBJ_TREE, FSCK_MSG_HAS_DOTDOT, "contains '..'");
 684         if (has_dotgit)
 685                 retval += report(options, oid, OBJ_TREE, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
 686         if (has_zero_pad)
 687                 retval += report(options, oid, OBJ_TREE, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
 688         if (has_bad_modes)
 689                 retval += report(options, oid, OBJ_TREE, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
 690         if (has_dup_entries)
 691                 retval += report(options, oid, OBJ_TREE, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
 692         if (not_properly_sorted)
 693                 retval += report(options, oid, OBJ_TREE, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
 694         return retval;
 695 }
 696
 697 static int verify_headers(const void *data, unsigned long size,
 698                           const struct object_id *oid, enum object_type type,
 699                           struct fsck_options *options)
 700 {
 701         const char *buffer = (const char *)data;
 702         unsigned long i;
 703
 704         for (i = 0; i < size; i++) {
 705                 switch (buffer[i]) {
 706                 case '\0':
 707                         return report(options, oid, type,
 708                                 FSCK_MSG_NUL_IN_HEADER,
 709                                 "unterminated header: NUL at offset %ld", i);
 710                 case '\n':
 711                         if (i + 1 < size && buffer[i + 1] == '\n')
 712                                 return 0;
 713                 }
 714         }
 715
 716         /*
 717          * We did not find double-LF that separates the header
 718          * and the body.  Not having a body is not a crime but
 719          * we do want to see the terminating LF for the last header
 720          * line.
 721          */
 722         if (size && buffer[size - 1] == '\n')
 723                 return 0;
 724
 725         return report(options, oid, type,
 726                 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
 727 }
 728
 729 static int fsck_ident(const char **ident,
 730                       const struct object_id *oid, enum object_type type,
 731                       struct fsck_options *options)
 732 {
 733         const char *p = *ident;
 734         char *end;
 735
 736         *ident = strchrnul(*ident, '\n');
 737         if (**ident == '\n')
 738                 (*ident)++;
 739
 740         if (*p == '<')
 741                 return report(options, oid, type, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 742         p += strcspn(p, "<>\n");
 743         if (*p == '>')
 744                 return report(options, oid, type, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
 745         if (*p != '<')
 746                 return report(options, oid, type, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
 747         if (p[-1] != ' ')
 748                 return report(options, oid, type, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
 749         p++;
 750         p += strcspn(p, "<>\n");
 751         if (*p != '>')
 752                 return report(options, oid, type, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
 753         p++;
 754         if (*p != ' ')
 755                 return report(options, oid, type, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
 756         p++;
 757         if (*p == '0' && p[1] != ' ')
 758                 return report(options, oid, type, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
 759         if (date_overflows(parse_timestamp(p, &end, 10)))
 760                 return report(options, oid, type, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
 761         if ((end == p || *end != ' '))
 762                 return report(options, oid, type, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
 763         p = end + 1;
 764         if ((*p != '+' && *p != '-') ||
 765             !isdigit(p[1]) ||
 766             !isdigit(p[2]) ||
 767             !isdigit(p[3]) ||
 768             !isdigit(p[4]) ||
 769             (p[5] != '\n'))
 770                 return report(options, oid, type, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
 771         p += 6;
 772         return 0;
 773 }
 774
 775 static int fsck_commit(const struct object_id *oid,
 776                        const char *buffer, unsigned long size,
 777                        struct fsck_options *options)
 778 {
 779         struct object_id tree_oid, parent_oid;
 780         unsigned author_count;
 781         int err;
 782         const char *buffer_begin = buffer;
 783         const char *p;
 784
 785         if (verify_headers(buffer, size, oid, OBJ_COMMIT, options))
 786                 return -1;
 787
 788         if (!skip_prefix(buffer, "tree ", &buffer))
 789                 return report(options, oid, OBJ_COMMIT, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
 790         if (parse_oid_hex(buffer, &tree_oid, &p) || *p != '\n') {
 791                 err = report(options, oid, OBJ_COMMIT, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
 792                 if (err)
 793                         return err;
 794         }
 795         buffer = p + 1;
 796         while (skip_prefix(buffer, "parent ", &buffer)) {
 797                 if (parse_oid_hex(buffer, &parent_oid, &p) || *p != '\n') {
 798                         err = report(options, oid, OBJ_COMMIT, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
 799                         if (err)
 800                                 return err;
 801                 }
 802                 buffer = p + 1;
 803         }
 804         author_count = 0;
 805         while (skip_prefix(buffer, "author ", &buffer)) {
 806                 author_count++;
 807                 err = fsck_ident(&buffer, oid, OBJ_COMMIT, options);
 808                 if (err)
 809                         return err;
 810         }
 811         if (author_count < 1)
 812                 err = report(options, oid, OBJ_COMMIT, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
 813         else if (author_count > 1)
 814                 err = report(options, oid, OBJ_COMMIT, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
 815         if (err)
 816                 return err;
 817         if (!skip_prefix(buffer, "committer ", &buffer))
 818                 return report(options, oid, OBJ_COMMIT, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
 819         err = fsck_ident(&buffer, oid, OBJ_COMMIT, options);
 820         if (err)
 821                 return err;
 822         if (memchr(buffer_begin, '\0', size)) {
 823                 err = report(options, oid, OBJ_COMMIT, FSCK_MSG_NUL_IN_COMMIT,
 824                              "NUL byte in the commit object body");
 825                 if (err)
 826                         return err;
 827         }
 828         return 0;
 829 }
 830
 831 static int fsck_tag(const struct object_id *oid, const char *buffer,
 832                     unsigned long size, struct fsck_options *options)
 833 {
 834         struct object_id tagged_oid;
 835         int tagged_type;
 836         return fsck_tag_standalone(oid, buffer, size, options, &tagged_oid,
 837                                    &tagged_type);
 838 }
 839
 840 int fsck_tag_standalone(const struct object_id *oid, const char *buffer,
 841                         unsigned long size, struct fsck_options *options,
 842                         struct object_id *tagged_oid,
 843                         int *tagged_type)
 844 {
 845         int ret = 0;
 846         char *eol;
 847         struct strbuf sb = STRBUF_INIT;
 848         const char *p;
 849
 850         ret = verify_headers(buffer, size, oid, OBJ_TAG, options);
 851         if (ret)
 852                 goto done;
 853
 854         if (!skip_prefix(buffer, "object ", &buffer)) {
 855                 ret = report(options, oid, OBJ_TAG, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
 856                 goto done;
 857         }
 858         if (parse_oid_hex(buffer, tagged_oid, &p) || *p != '\n') {
 859                 ret = report(options, oid, OBJ_TAG, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
 860                 if (ret)
 861                         goto done;
 862         }
 863         buffer = p + 1;
 864
 865         if (!skip_prefix(buffer, "type ", &buffer)) {
 866                 ret = report(options, oid, OBJ_TAG, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
 867                 goto done;
 868         }
 869         eol = strchr(buffer, '\n');
 870         if (!eol) {
 871                 ret = report(options, oid, OBJ_TAG, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
 872                 goto done;
 873         }
 874         *tagged_type = type_from_string_gently(buffer, eol - buffer, 1);
 875         if (*tagged_type < 0)
 876                 ret = report(options, oid, OBJ_TAG, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
 877         if (ret)
 878                 goto done;
 879         buffer = eol + 1;
 880
 881         if (!skip_prefix(buffer, "tag ", &buffer)) {
 882                 ret = report(options, oid, OBJ_TAG, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
 883                 goto done;
 884         }
 885         eol = strchr(buffer, '\n');
 886         if (!eol) {
 887                 ret = report(options, oid, OBJ_TAG, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
 888                 goto done;
 889         }
 890         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
 891         if (check_refname_format(sb.buf, 0)) {
 892                 ret = report(options, oid, OBJ_TAG,
 893                              FSCK_MSG_BAD_TAG_NAME,
 894                              "invalid 'tag' name: %.*s",
 895                              (int)(eol - buffer), buffer);
 896                 if (ret)
 897                         goto done;
 898         }
 899         buffer = eol + 1;
 900
 901         if (!skip_prefix(buffer, "tagger ", &buffer)) {
 902                 /* early tags do not contain 'tagger' lines; warn only */
 903                 ret = report(options, oid, OBJ_TAG, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
 904                 if (ret)
 905                         goto done;
 906         }
 907         else
 908                 ret = fsck_ident(&buffer, oid, OBJ_TAG, options);
 909         if (!*buffer)
 910                 goto done;
 911
 912         if (!starts_with(buffer, "\n")) {
 913                 /*
 914                  * The verify_headers() check will allow
 915                  * e.g. "[...]tagger <tagger>\nsome
 916                  * garbage\n\nmessage" to pass, thinking "some
 917                  * garbage" could be a custom header. E.g. "mktag"
 918                  * doesn't want any unknown headers.
 919                  */
 920                 ret = report(options, oid, OBJ_TAG, FSCK_MSG_EXTRA_HEADER_ENTRY, "invalid format - extra header(s) after 'tagger'");
 921                 if (ret)
 922                         goto done;
 923         }
 924
 925 done:
 926         strbuf_release(&sb);
 927         return ret;
 928 }
 929
 930 /*
 931  * Like builtin/submodule--helper.c's starts_with_dot_slash, but without
 932  * relying on the platform-dependent is_dir_sep helper.
 933  *
 934  * This is for use in checking whether a submodule URL is interpreted as
 935  * relative to the current directory on any platform, since \ is a
 936  * directory separator on Windows but not on other platforms.
 937  */
 938 static int starts_with_dot_slash(const char *str)
 939 {
 940         return str[0] == '.' && (str[1] == '/' || str[1] == '\\');
 941 }
 942
 943 /*
 944  * Like starts_with_dot_slash, this is a variant of submodule--helper's
 945  * helper of the same name with the twist that it accepts backslash as a
 946  * directory separator even on non-Windows platforms.
 947  */
 948 static int starts_with_dot_dot_slash(const char *str)
 949 {
 950         return str[0] == '.' && starts_with_dot_slash(str + 1);
 951 }
 952
 953 static int submodule_url_is_relative(const char *url)
 954 {
 955         return starts_with_dot_slash(url) || starts_with_dot_dot_slash(url);
 956 }
 957
 958 /*
 959  * Count directory components that a relative submodule URL should chop
 960  * from the remote_url it is to be resolved against.
 961  *
 962  * In other words, this counts "../" components at the start of a
 963  * submodule URL.
 964  *
 965  * Returns the number of directory components to chop and writes a
 966  * pointer to the next character of url after all leading "./" and
 967  * "../" components to out.
 968  */
 969 static int count_leading_dotdots(const char *url, const char **out)
 970 {
 971         int result = 0;
 972         while (1) {
 973                 if (starts_with_dot_dot_slash(url)) {
 974                         result++;
 975                         url += strlen("../");
 976                         continue;
 977                 }
 978                 if (starts_with_dot_slash(url)) {
 979                         url += strlen("./");
 980                         continue;
 981                 }
 982                 *out = url;
 983                 return result;
 984         }
 985 }
 986 /*
 987  * Check whether a transport is implemented by git-remote-curl.
 988  *
 989  * If it is, returns 1 and writes the URL that would be passed to
 990  * git-remote-curl to the "out" parameter.
 991  *
 992  * Otherwise, returns 0 and leaves "out" untouched.
 993  *
 994  * Examples:
 995  *   http::https://example.com/repo.git -> 1, https://example.com/repo.git
 996  *   https://example.com/repo.git -> 1, https://example.com/repo.git
 997  *   git://example.com/repo.git -> 0
 998  *
 999  * This is for use in checking for previously exploitable bugs that
1000  * required a submodule URL to be passed to git-remote-curl.
1001  */
1002 static int url_to_curl_url(const char *url, const char **out)
1003 {
1004         /*
1005          * We don't need to check for case-aliases, "http.exe", and so
1006          * on because in the default configuration, is_transport_allowed
1007          * prevents URLs with those schemes from being cloned
1008          * automatically.
1009          */
1010         if (skip_prefix(url, "http::", out) ||
1011             skip_prefix(url, "https::", out) ||
1012             skip_prefix(url, "ftp::", out) ||
1013             skip_prefix(url, "ftps::", out))
1014                 return 1;
1015         if (starts_with(url, "http://") ||
1016             starts_with(url, "https://") ||
1017             starts_with(url, "ftp://") ||
1018             starts_with(url, "ftps://")) {
1019                 *out = url;
1020                 return 1;
1021         }
1022         return 0;
1023 }
1024
1025 static int check_submodule_url(const char *url)
1026 {
1027         const char *curl_url;
1028
1029         if (looks_like_command_line_option(url))
1030                 return -1;
1031
1032         if (submodule_url_is_relative(url) || starts_with(url, "git://")) {
1033                 char *decoded;
1034                 const char *next;
1035                 int has_nl;
1036
1037                 /*
1038                  * This could be appended to an http URL and url-decoded;
1039                  * check for malicious characters.
1040                  */
1041                 decoded = url_decode(url);
1042                 has_nl = !!strchr(decoded, '\n');
1043
1044                 free(decoded);
1045                 if (has_nl)
1046                         return -1;
1047
1048                 /*
1049                  * URLs which escape their root via "../" can overwrite
1050                  * the host field and previous components, resolving to
1051                  * URLs like https::example.com/submodule.git and
1052                  * https:///example.com/submodule.git that were
1053                  * susceptible to CVE-2020-11008.
1054                  */
1055                 if (count_leading_dotdots(url, &next) > 0 &&
1056                     (*next == ':' || *next == '/'))
1057                         return -1;
1058         }
1059
1060         else if (url_to_curl_url(url, &curl_url)) {
1061                 struct credential c = CREDENTIAL_INIT;
1062                 int ret = 0;
1063                 if (credential_from_url_gently(&c, curl_url, 1) ||
1064                     !*c.host)
1065                         ret = -1;
1066                 credential_clear(&c);
1067                 return ret;
1068         }
1069
1070         return 0;
1071 }
1072
1073 struct fsck_gitmodules_data {
1074         const struct object_id *oid;
1075         struct fsck_options *options;
1076         int ret;
1077 };
1078
1079 static int fsck_gitmodules_fn(const char *var, const char *value, void *vdata)
1080 {
1081         struct fsck_gitmodules_data *data = vdata;
1082         const char *subsection, *key;
1083         size_t subsection_len;
1084         char *name;
1085
1086         if (parse_config_key(var, "submodule", &subsection, &subsection_len, &key) < 0 ||
1087             !subsection)
1088                 return 0;
1089
1090         name = xmemdupz(subsection, subsection_len);
1091         if (check_submodule_name(name) < 0)
1092                 data->ret |= report(data->options,
1093                                     data->oid, OBJ_BLOB,
1094                                     FSCK_MSG_GITMODULES_NAME,
1095                                     "disallowed submodule name: %s",
1096                                     name);
1097         if (!strcmp(key, "url") && value &&
1098             check_submodule_url(value) < 0)
1099                 data->ret |= report(data->options,
1100                                     data->oid, OBJ_BLOB,
1101                                     FSCK_MSG_GITMODULES_URL,
1102                                     "disallowed submodule url: %s",
1103                                     value);
1104         if (!strcmp(key, "path") && value &&
1105             looks_like_command_line_option(value))
1106                 data->ret |= report(data->options,
1107                                     data->oid, OBJ_BLOB,
1108                                     FSCK_MSG_GITMODULES_PATH,
1109                                     "disallowed submodule path: %s",
1110                                     value);
1111         if (!strcmp(key, "update") && value &&
1112             parse_submodule_update_type(value) == SM_UPDATE_COMMAND)
1113                 data->ret |= report(data->options, data->oid, OBJ_BLOB,
1114                                     FSCK_MSG_GITMODULES_UPDATE,
1115                                     "disallowed submodule update setting: %s",
1116                                     value);
1117         free(name);
1118
1119         return 0;
1120 }
1121
1122 static int fsck_blob(const struct object_id *oid, const char *buf,
1123                      unsigned long size, struct fsck_options *options)
1124 {
1125         struct fsck_gitmodules_data data;
1126         struct config_options config_opts = { 0 };
1127
1128         if (!oidset_contains(&gitmodules_found, oid))
1129                 return 0;
1130         oidset_insert(&gitmodules_done, oid);
1131
1132         if (object_on_skiplist(options, oid))
1133                 return 0;
1134
1135         if (!buf) {
1136                 /*
1137                  * A missing buffer here is a sign that the caller found the
1138                  * blob too gigantic to load into memory. Let's just consider
1139                  * that an error.
1140                  */
1141                 return report(options, oid, OBJ_BLOB,
1142                               FSCK_MSG_GITMODULES_LARGE,
1143                               ".gitmodules too large to parse");
1144         }
1145
1146         data.oid = oid;
1147         data.options = options;
1148         data.ret = 0;
1149         config_opts.error_action = CONFIG_ERROR_SILENT;
1150         if (git_config_from_mem(fsck_gitmodules_fn, CONFIG_ORIGIN_BLOB,
1151                                 ".gitmodules", buf, size, &data, &config_opts))
1152                 data.ret |= report(options, oid, OBJ_BLOB,
1153                                    FSCK_MSG_GITMODULES_PARSE,
1154                                    "could not parse gitmodules blob");
1155
1156         return data.ret;
1157 }
1158
1159 int fsck_object(struct object *obj, void *data, unsigned long size,
1160         struct fsck_options *options)
1161 {
1162         if (!obj)
1163                 return report(options, NULL, OBJ_NONE, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
1164
1165         if (obj->type == OBJ_BLOB)
1166                 return fsck_blob(&obj->oid, data, size, options);
1167         if (obj->type == OBJ_TREE)
1168                 return fsck_tree(&obj->oid, data, size, options);
1169         if (obj->type == OBJ_COMMIT)
1170                 return fsck_commit(&obj->oid, data, size, options);
1171         if (obj->type == OBJ_TAG)
1172                 return fsck_tag(&obj->oid, data, size, options);
1173
1174         return report(options, &obj->oid, obj->type,
1175                       FSCK_MSG_UNKNOWN_TYPE,
1176                       "unknown type '%d' (internal fsck error)",
1177                       obj->type);
1178 }
1179
1180 int fsck_error_function(struct fsck_options *o,
1181                         const struct object_id *oid,
1182                         enum object_type object_type,
1183                         enum fsck_msg_type msg_type, const char *message)
1184 {
1185         if (msg_type == FSCK_WARN) {
1186                 warning("object %s: %s", fsck_describe_object(o, oid), message);
1187                 return 0;
1188         }
1189         error("object %s: %s", fsck_describe_object(o, oid), message);
1190         return 1;
1191 }
1192
1193 void register_found_gitmodules(const struct object_id *oid)
1194 {
1195         oidset_insert(&gitmodules_found, oid);
1196 }
1197
1198 int fsck_finish(struct fsck_options *options)
1199 {
1200         int ret = 0;
1201         struct oidset_iter iter;
1202         const struct object_id *oid;
1203
1204         oidset_iter_init(&gitmodules_found, &iter);
1205         while ((oid = oidset_iter_next(&iter))) {
1206                 enum object_type type;
1207                 unsigned long size;
1208                 char *buf;
1209
1210                 if (oidset_contains(&gitmodules_done, oid))
1211                         continue;
1212
1213                 buf = read_object_file(oid, &type, &size);
1214                 if (!buf) {
1215                         if (is_promisor_object(oid))
1216                                 continue;
1217                         ret |= report(options,
1218                                       oid, OBJ_BLOB,
1219                                       FSCK_MSG_GITMODULES_MISSING,
1220                                       "unable to read .gitmodules blob");
1221                         continue;
1222                 }
1223
1224                 if (type == OBJ_BLOB)
1225                         ret |= fsck_blob(oid, buf, size, options);
1226                 else
1227                         ret |= report(options,
1228                                       oid, type,
1229                                       FSCK_MSG_GITMODULES_BLOB,
1230                                       "non-blob found at .gitmodules");
1231                 free(buf);
1232         }
1233
1234
1235         oidset_clear(&gitmodules_found);
1236         oidset_clear(&gitmodules_done);
1237         return ret;
1238 }
1239
1240 int git_fsck_config(const char *var, const char *value, void *cb)
1241 {
1242         struct fsck_options *options = cb;
1243         if (strcmp(var, "fsck.skiplist") == 0) {
1244                 const char *path;
1245                 struct strbuf sb = STRBUF_INIT;
1246
1247                 if (git_config_pathname(&path, var, value))
1248                         return 1;
1249                 strbuf_addf(&sb, "skiplist=%s", path);
1250                 free((char *)path);
1251                 fsck_set_msg_types(options, sb.buf);
1252                 strbuf_release(&sb);
1253                 return 0;
1254         }
1255
1256         if (skip_prefix(var, "fsck.", &var)) {
1257                 fsck_set_msg_type(options, var, value);
1258                 return 0;
1259         }
1260
1261         return git_default_config(var, value, cb);
1262 }