11 #include "sha1-array.h"
17 #define FOREACH_MSG_ID(FUNC) \
19 FUNC(NUL_IN_HEADER, FATAL) \
20 FUNC(UNTERMINATED_HEADER, FATAL) \
22 FUNC(BAD_DATE, ERROR) \
23 FUNC(BAD_DATE_OVERFLOW, ERROR) \
24 FUNC(BAD_EMAIL, ERROR) \
25 FUNC(BAD_NAME, ERROR) \
26 FUNC(BAD_OBJECT_SHA1, ERROR) \
27 FUNC(BAD_PARENT_SHA1, ERROR) \
28 FUNC(BAD_TAG_OBJECT, ERROR) \
29 FUNC(BAD_TIMEZONE, ERROR) \
30 FUNC(BAD_TREE, ERROR) \
31 FUNC(BAD_TREE_SHA1, ERROR) \
32 FUNC(BAD_TYPE, ERROR) \
33 FUNC(DUPLICATE_ENTRIES, ERROR) \
34 FUNC(MISSING_AUTHOR, ERROR) \
35 FUNC(MISSING_COMMITTER, ERROR) \
36 FUNC(MISSING_EMAIL, ERROR) \
37 FUNC(MISSING_GRAFT, ERROR) \
38 FUNC(MISSING_NAME_BEFORE_EMAIL, ERROR) \
39 FUNC(MISSING_OBJECT, ERROR) \
40 FUNC(MISSING_PARENT, ERROR) \
41 FUNC(MISSING_SPACE_BEFORE_DATE, ERROR) \
42 FUNC(MISSING_SPACE_BEFORE_EMAIL, ERROR) \
43 FUNC(MISSING_TAG, ERROR) \
44 FUNC(MISSING_TAG_ENTRY, ERROR) \
45 FUNC(MISSING_TAG_OBJECT, ERROR) \
46 FUNC(MISSING_TREE, ERROR) \
47 FUNC(MISSING_TYPE, ERROR) \
48 FUNC(MISSING_TYPE_ENTRY, ERROR) \
49 FUNC(MULTIPLE_AUTHORS, ERROR) \
50 FUNC(TAG_OBJECT_NOT_TAG, ERROR) \
51 FUNC(TREE_NOT_SORTED, ERROR) \
52 FUNC(UNKNOWN_TYPE, ERROR) \
53 FUNC(ZERO_PADDED_DATE, ERROR) \
55 FUNC(BAD_FILEMODE, WARN) \
56 FUNC(EMPTY_NAME, WARN) \
57 FUNC(FULL_PATHNAME, WARN) \
59 FUNC(HAS_DOTDOT, WARN) \
60 FUNC(HAS_DOTGIT, WARN) \
61 FUNC(NULL_SHA1, WARN) \
62 FUNC(ZERO_PADDED_FILEMODE, WARN) \
63 FUNC(NUL_IN_COMMIT, WARN) \
64 /* infos (reported as warnings, but ignored by default) */ \
65 FUNC(BAD_TAG_NAME, INFO) \
66 FUNC(MISSING_TAGGER_ENTRY, INFO)
68 #define MSG_ID(id, msg_type) FSCK_MSG_##id,
70 FOREACH_MSG_ID(MSG_ID)
76 #define MSG_ID(id, msg_type) { STR(id), NULL, FSCK_##msg_type },
78 const char *id_string;
79 const char *downcased;
81 } msg_id_info[FSCK_MSG_MAX + 1] = {
82 FOREACH_MSG_ID(MSG_ID)
87 static int parse_msg_id(const char *text)
91 if (!msg_id_info[0].downcased) {
92 /* convert id_string to lower case, without underscores. */
93 for (i = 0; i < FSCK_MSG_MAX; i++) {
94 const char *p = msg_id_info[i].id_string;
96 char *q = xmalloc(len);
98 msg_id_info[i].downcased = q;
103 *(q)++ = tolower(*(p)++);
108 for (i = 0; i < FSCK_MSG_MAX; i++)
109 if (!strcmp(text, msg_id_info[i].downcased))
115 static int fsck_msg_type(enum fsck_msg_id msg_id,
116 struct fsck_options *options)
120 assert(msg_id >= 0 && msg_id < FSCK_MSG_MAX);
122 if (options->msg_type)
123 msg_type = options->msg_type[msg_id];
125 msg_type = msg_id_info[msg_id].msg_type;
126 if (options->strict && msg_type == FSCK_WARN)
127 msg_type = FSCK_ERROR;
133 static void init_skiplist(struct fsck_options *options, const char *path)
135 static struct sha1_array skiplist = SHA1_ARRAY_INIT;
138 unsigned char sha1[20];
140 if (options->skiplist)
141 sorted = options->skiplist->sorted;
144 options->skiplist = &skiplist;
147 fd = open(path, O_RDONLY);
149 die("Could not open skip list: %s", path);
151 int result = read_in_full(fd, buffer, sizeof(buffer));
153 die_errno("Could not read '%s'", path);
156 if (get_sha1_hex(buffer, sha1) || buffer[40] != '\n')
157 die("Invalid SHA-1: %s", buffer);
158 sha1_array_append(&skiplist, sha1);
159 if (sorted && skiplist.nr > 1 &&
160 hashcmp(skiplist.sha1[skiplist.nr - 2],
170 static int parse_msg_type(const char *str)
172 if (!strcmp(str, "error"))
174 else if (!strcmp(str, "warn"))
176 else if (!strcmp(str, "ignore"))
179 die("Unknown fsck message type: '%s'", str);
182 int is_valid_msg_type(const char *msg_id, const char *msg_type)
184 if (parse_msg_id(msg_id) < 0)
186 parse_msg_type(msg_type);
190 void fsck_set_msg_type(struct fsck_options *options,
191 const char *msg_id, const char *msg_type)
193 int id = parse_msg_id(msg_id), type;
196 die("Unhandled message id: %s", msg_id);
197 type = parse_msg_type(msg_type);
199 if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
200 die("Cannot demote %s to %s", msg_id, msg_type);
202 if (!options->msg_type) {
205 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
206 for (i = 0; i < FSCK_MSG_MAX; i++)
207 msg_type[i] = fsck_msg_type(i, options);
208 options->msg_type = msg_type;
211 options->msg_type[id] = type;
214 void fsck_set_msg_types(struct fsck_options *options, const char *values)
216 char *buf = xstrdup(values), *to_free = buf;
220 int len = strcspn(buf, " ,|"), equal;
230 equal < len && buf[equal] != '=' && buf[equal] != ':';
232 buf[equal] = tolower(buf[equal]);
235 if (!strcmp(buf, "skiplist")) {
237 die("skiplist requires a path");
238 init_skiplist(options, buf + equal + 1);
244 die("Missing '=': '%s'", buf);
246 fsck_set_msg_type(options, buf, buf + equal + 1);
252 static void append_msg_id(struct strbuf *sb, const char *msg_id)
255 char c = *(msg_id)++;
260 strbuf_addch(sb, tolower(c));
263 strbuf_addch(sb, *(msg_id)++);
267 strbuf_addstr(sb, ": ");
270 __attribute__((format (printf, 4, 5)))
271 static int report(struct fsck_options *options, struct object *object,
272 enum fsck_msg_id id, const char *fmt, ...)
275 struct strbuf sb = STRBUF_INIT;
276 int msg_type = fsck_msg_type(id, options), result;
278 if (msg_type == FSCK_IGNORE)
281 if (options->skiplist && object &&
282 sha1_array_lookup(options->skiplist, object->oid.hash) >= 0)
285 if (msg_type == FSCK_FATAL)
286 msg_type = FSCK_ERROR;
287 else if (msg_type == FSCK_INFO)
288 msg_type = FSCK_WARN;
290 append_msg_id(&sb, msg_id_info[id].id_string);
293 strbuf_vaddf(&sb, fmt, ap);
294 result = options->error_func(options, object, msg_type, sb.buf);
301 static char *get_object_name(struct fsck_options *options, struct object *obj)
303 if (!options->object_names)
305 return lookup_decoration(options->object_names, obj);
308 static void put_object_name(struct fsck_options *options, struct object *obj,
309 const char *fmt, ...)
312 struct strbuf buf = STRBUF_INIT;
315 if (!options->object_names)
317 existing = lookup_decoration(options->object_names, obj);
321 strbuf_vaddf(&buf, fmt, ap);
322 add_decoration(options->object_names, obj, strbuf_detach(&buf, NULL));
326 static const char *describe_object(struct fsck_options *o, struct object *obj)
328 static struct strbuf buf = STRBUF_INIT;
332 strbuf_addstr(&buf, oid_to_hex(&obj->oid));
333 if (o->object_names && (name = lookup_decoration(o->object_names, obj)))
334 strbuf_addf(&buf, " (%s)", name);
339 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
341 struct tree_desc desc;
342 struct name_entry entry;
346 if (parse_tree(tree))
349 name = get_object_name(options, &tree->object);
350 if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
352 while (tree_entry_gently(&desc, &entry)) {
356 if (S_ISGITLINK(entry.mode))
359 if (S_ISDIR(entry.mode)) {
360 obj = &lookup_tree(entry.oid->hash)->object;
362 put_object_name(options, obj, "%s%s/", name,
364 result = options->walk(obj, OBJ_TREE, data, options);
366 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
367 obj = &lookup_blob(entry.oid->hash)->object;
369 put_object_name(options, obj, "%s%s", name,
371 result = options->walk(obj, OBJ_BLOB, data, options);
374 result = error("in tree %s: entry %s has bad mode %.6o",
375 describe_object(options, &tree->object), entry.path, entry.mode);
385 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
387 int counter = 0, generation = 0, name_prefix_len = 0;
388 struct commit_list *parents;
393 if (parse_commit(commit))
396 name = get_object_name(options, &commit->object);
398 put_object_name(options, &commit->tree->object, "%s:", name);
400 result = options->walk((struct object *)commit->tree, OBJ_TREE, data, options);
405 parents = commit->parents;
406 if (name && parents) {
407 int len = strlen(name), power;
409 if (len && name[len - 1] == '^') {
411 name_prefix_len = len - 1;
413 else { /* parse ~<generation> suffix */
414 for (generation = 0, power = 1;
415 len && isdigit(name[len - 1]);
417 generation += power * (name[--len] - '0');
418 if (power > 1 && len && name[len - 1] == '~')
419 name_prefix_len = len - 1;
425 struct object *obj = &parents->item->object;
428 put_object_name(options, obj, "%s^%d",
430 else if (generation > 0)
431 put_object_name(options, obj, "%.*s~%d",
432 name_prefix_len, name, generation + 1);
434 put_object_name(options, obj, "%s^", name);
436 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
441 parents = parents->next;
446 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
448 char *name = get_object_name(options, &tag->object);
453 put_object_name(options, tag->tagged, "%s", name);
454 return options->walk(tag->tagged, OBJ_ANY, data, options);
457 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
465 return fsck_walk_tree((struct tree *)obj, data, options);
467 return fsck_walk_commit((struct commit *)obj, data, options);
469 return fsck_walk_tag((struct tag *)obj, data, options);
471 error("Unknown object type for %s", describe_object(options, obj));
477 * The entries in a tree are ordered in the _path_ order,
478 * which means that a directory entry is ordered by adding
479 * a slash to the end of it.
481 * So a directory called "a" is ordered _after_ a file
482 * called "a.c", because "a/" sorts after "a.c".
484 #define TREE_UNORDERED (-1)
485 #define TREE_HAS_DUPS (-2)
487 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
489 int len1 = strlen(name1);
490 int len2 = strlen(name2);
491 int len = len1 < len2 ? len1 : len2;
492 unsigned char c1, c2;
495 cmp = memcmp(name1, name2, len);
499 return TREE_UNORDERED;
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.
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.
514 return TREE_HAS_DUPS;
515 if (!c1 && S_ISDIR(mode1))
517 if (!c2 && S_ISDIR(mode2))
519 return c1 < c2 ? 0 : TREE_UNORDERED;
522 static int fsck_tree(struct tree *item, struct fsck_options *options)
525 int has_null_sha1 = 0;
526 int has_full_path = 0;
527 int has_empty_name = 0;
531 int has_zero_pad = 0;
532 int has_bad_modes = 0;
533 int has_dup_entries = 0;
534 int not_properly_sorted = 0;
535 struct tree_desc desc;
539 if (init_tree_desc_gently(&desc, item->buffer, item->size)) {
540 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
550 const struct object_id *oid;
552 oid = tree_entry_extract(&desc, &name, &mode);
554 has_null_sha1 |= is_null_oid(oid);
555 has_full_path |= !!strchr(name, '/');
556 has_empty_name |= !*name;
557 has_dot |= !strcmp(name, ".");
558 has_dotdot |= !strcmp(name, "..");
559 has_dotgit |= (!strcmp(name, ".git") ||
560 is_hfs_dotgit(name) ||
561 is_ntfs_dotgit(name));
562 has_zero_pad |= *(char *)desc.buffer == '0';
563 if (update_tree_entry_gently(&desc)) {
564 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
579 * This is nonstandard, but we had a few of these
580 * early on when we honored the full set of mode
584 if (!options->strict)
591 switch (verify_ordered(o_mode, o_name, mode, name)) {
593 not_properly_sorted = 1;
608 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
610 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
612 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
614 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
616 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
618 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
620 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
622 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
624 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
625 if (not_properly_sorted)
626 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
630 static int verify_headers(const void *data, unsigned long size,
631 struct object *obj, struct fsck_options *options)
633 const char *buffer = (const char *)data;
636 for (i = 0; i < size; i++) {
639 return report(options, obj,
640 FSCK_MSG_NUL_IN_HEADER,
641 "unterminated header: NUL at offset %ld", i);
643 if (i + 1 < size && buffer[i + 1] == '\n')
649 * We did not find double-LF that separates the header
650 * and the body. Not having a body is not a crime but
651 * we do want to see the terminating LF for the last header
654 if (size && buffer[size - 1] == '\n')
657 return report(options, obj,
658 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
661 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
663 const char *p = *ident;
666 *ident = strchrnul(*ident, '\n');
671 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
672 p += strcspn(p, "<>\n");
674 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
676 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
678 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
680 p += strcspn(p, "<>\n");
682 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
685 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
687 if (*p == '0' && p[1] != ' ')
688 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
689 if (date_overflows(strtoul(p, &end, 10)))
690 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
691 if ((end == p || *end != ' '))
692 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
694 if ((*p != '+' && *p != '-') ||
700 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
705 static int fsck_commit_buffer(struct commit *commit, const char *buffer,
706 unsigned long size, struct fsck_options *options)
708 unsigned char tree_sha1[20], sha1[20];
709 struct commit_graft *graft;
710 unsigned parent_count, parent_line_count = 0, author_count;
712 const char *buffer_begin = buffer;
714 if (verify_headers(buffer, size, &commit->object, options))
717 if (!skip_prefix(buffer, "tree ", &buffer))
718 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
719 if (get_sha1_hex(buffer, tree_sha1) || buffer[40] != '\n') {
720 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
725 while (skip_prefix(buffer, "parent ", &buffer)) {
726 if (get_sha1_hex(buffer, sha1) || buffer[40] != '\n') {
727 err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
734 graft = lookup_commit_graft(commit->object.oid.hash);
735 parent_count = commit_list_count(commit->parents);
737 if (graft->nr_parent == -1 && !parent_count)
738 ; /* shallow commit */
739 else if (graft->nr_parent != parent_count) {
740 err = report(options, &commit->object, FSCK_MSG_MISSING_GRAFT, "graft objects missing");
745 if (parent_count != parent_line_count) {
746 err = report(options, &commit->object, FSCK_MSG_MISSING_PARENT, "parent objects missing");
752 while (skip_prefix(buffer, "author ", &buffer)) {
754 err = fsck_ident(&buffer, &commit->object, options);
758 if (author_count < 1)
759 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
760 else if (author_count > 1)
761 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
764 if (!skip_prefix(buffer, "committer ", &buffer))
765 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
766 err = fsck_ident(&buffer, &commit->object, options);
770 err = report(options, &commit->object, FSCK_MSG_BAD_TREE, "could not load commit's tree %s", sha1_to_hex(tree_sha1));
774 if (memchr(buffer_begin, '\0', size)) {
775 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
776 "NUL byte in the commit object body");
783 static int fsck_commit(struct commit *commit, const char *data,
784 unsigned long size, struct fsck_options *options)
786 const char *buffer = data ? data : get_commit_buffer(commit, &size);
787 int ret = fsck_commit_buffer(commit, buffer, size, options);
789 unuse_commit_buffer(commit, buffer);
793 static int fsck_tag_buffer(struct tag *tag, const char *data,
794 unsigned long size, struct fsck_options *options)
796 unsigned char sha1[20];
799 char *to_free = NULL, *eol;
800 struct strbuf sb = STRBUF_INIT;
805 enum object_type type;
808 read_sha1_file(tag->object.oid.hash, &type, &size);
810 return report(options, &tag->object,
811 FSCK_MSG_MISSING_TAG_OBJECT,
812 "cannot read tag object");
814 if (type != OBJ_TAG) {
815 ret = report(options, &tag->object,
816 FSCK_MSG_TAG_OBJECT_NOT_TAG,
817 "expected tag got %s",
823 ret = verify_headers(buffer, size, &tag->object, options);
827 if (!skip_prefix(buffer, "object ", &buffer)) {
828 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
831 if (get_sha1_hex(buffer, sha1) || buffer[40] != '\n') {
832 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
838 if (!skip_prefix(buffer, "type ", &buffer)) {
839 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
842 eol = strchr(buffer, '\n');
844 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
847 if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
848 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
853 if (!skip_prefix(buffer, "tag ", &buffer)) {
854 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
857 eol = strchr(buffer, '\n');
859 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
862 strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
863 if (check_refname_format(sb.buf, 0)) {
864 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
865 "invalid 'tag' name: %.*s",
866 (int)(eol - buffer), buffer);
872 if (!skip_prefix(buffer, "tagger ", &buffer)) {
873 /* early tags do not contain 'tagger' lines; warn only */
874 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
879 ret = fsck_ident(&buffer, &tag->object, options);
887 static int fsck_tag(struct tag *tag, const char *data,
888 unsigned long size, struct fsck_options *options)
890 struct object *tagged = tag->tagged;
893 return report(options, &tag->object, FSCK_MSG_BAD_TAG_OBJECT, "could not load tagged object");
895 return fsck_tag_buffer(tag, data, size, options);
898 int fsck_object(struct object *obj, void *data, unsigned long size,
899 struct fsck_options *options)
902 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
904 if (obj->type == OBJ_BLOB)
906 if (obj->type == OBJ_TREE)
907 return fsck_tree((struct tree *) obj, options);
908 if (obj->type == OBJ_COMMIT)
909 return fsck_commit((struct commit *) obj, (const char *) data,
911 if (obj->type == OBJ_TAG)
912 return fsck_tag((struct tag *) obj, (const char *) data,
915 return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
919 int fsck_error_function(struct fsck_options *o,
920 struct object *obj, int msg_type, const char *message)
922 if (msg_type == FSCK_WARN) {
923 warning("object %s: %s", describe_object(o, obj), message);
926 error("object %s: %s", describe_object(o, obj), message);