fsck: use oidset instead of oid_array for skipList
[git/git.git] / 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 "utf8.h"
13 #include "decorate.h"
14 #include "oidset.h"
15 #include "packfile.h"
16 #include "submodule-config.h"
17 #include "config.h"
18 #include "help.h"
19
20 static struct oidset gitmodules_found = OIDSET_INIT;
21 static struct oidset gitmodules_done = OIDSET_INIT;
22
23 #define FSCK_FATAL -1
24 #define FSCK_INFO -2
25
26 #define FOREACH_MSG_ID(FUNC) \
27 /* fatal errors */ \
28 FUNC(NUL_IN_HEADER, FATAL) \
29 FUNC(UNTERMINATED_HEADER, FATAL) \
30 /* errors */ \
31 FUNC(BAD_DATE, ERROR) \
32 FUNC(BAD_DATE_OVERFLOW, ERROR) \
33 FUNC(BAD_EMAIL, ERROR) \
34 FUNC(BAD_NAME, ERROR) \
35 FUNC(BAD_OBJECT_SHA1, ERROR) \
36 FUNC(BAD_PARENT_SHA1, ERROR) \
37 FUNC(BAD_TAG_OBJECT, ERROR) \
38 FUNC(BAD_TIMEZONE, ERROR) \
39 FUNC(BAD_TREE, ERROR) \
40 FUNC(BAD_TREE_SHA1, ERROR) \
41 FUNC(BAD_TYPE, ERROR) \
42 FUNC(DUPLICATE_ENTRIES, ERROR) \
43 FUNC(MISSING_AUTHOR, ERROR) \
44 FUNC(MISSING_COMMITTER, ERROR) \
45 FUNC(MISSING_EMAIL, ERROR) \
46 FUNC(MISSING_GRAFT, ERROR) \
47 FUNC(MISSING_NAME_BEFORE_EMAIL, ERROR) \
48 FUNC(MISSING_OBJECT, ERROR) \
49 FUNC(MISSING_PARENT, ERROR) \
50 FUNC(MISSING_SPACE_BEFORE_DATE, ERROR) \
51 FUNC(MISSING_SPACE_BEFORE_EMAIL, ERROR) \
52 FUNC(MISSING_TAG, ERROR) \
53 FUNC(MISSING_TAG_ENTRY, ERROR) \
54 FUNC(MISSING_TAG_OBJECT, ERROR) \
55 FUNC(MISSING_TREE, ERROR) \
56 FUNC(MISSING_TREE_OBJECT, ERROR) \
57 FUNC(MISSING_TYPE, ERROR) \
58 FUNC(MISSING_TYPE_ENTRY, ERROR) \
59 FUNC(MULTIPLE_AUTHORS, ERROR) \
60 FUNC(TAG_OBJECT_NOT_TAG, ERROR) \
61 FUNC(TREE_NOT_SORTED, ERROR) \
62 FUNC(UNKNOWN_TYPE, ERROR) \
63 FUNC(ZERO_PADDED_DATE, ERROR) \
64 FUNC(GITMODULES_MISSING, ERROR) \
65 FUNC(GITMODULES_BLOB, ERROR) \
66 FUNC(GITMODULES_LARGE, ERROR) \
67 FUNC(GITMODULES_NAME, ERROR) \
68 FUNC(GITMODULES_SYMLINK, ERROR) \
69 /* warnings */ \
70 FUNC(BAD_FILEMODE, WARN) \
71 FUNC(EMPTY_NAME, WARN) \
72 FUNC(FULL_PATHNAME, WARN) \
73 FUNC(HAS_DOT, WARN) \
74 FUNC(HAS_DOTDOT, WARN) \
75 FUNC(HAS_DOTGIT, WARN) \
76 FUNC(NULL_SHA1, WARN) \
77 FUNC(ZERO_PADDED_FILEMODE, WARN) \
78 FUNC(NUL_IN_COMMIT, WARN) \
79 /* infos (reported as warnings, but ignored by default) */ \
80 FUNC(GITMODULES_PARSE, INFO) \
81 FUNC(BAD_TAG_NAME, INFO) \
82 FUNC(MISSING_TAGGER_ENTRY, INFO)
83
84 #define MSG_ID(id, msg_type) FSCK_MSG_##id,
85 enum fsck_msg_id {
86 FOREACH_MSG_ID(MSG_ID)
87 FSCK_MSG_MAX
88 };
89 #undef MSG_ID
90
91 #define STR(x) #x
92 #define MSG_ID(id, msg_type) { STR(id), NULL, NULL, FSCK_##msg_type },
93 static struct {
94 const char *id_string;
95 const char *downcased;
96 const char *camelcased;
97 int msg_type;
98 } msg_id_info[FSCK_MSG_MAX + 1] = {
99 FOREACH_MSG_ID(MSG_ID)
100 { NULL, NULL, NULL, -1 }
101 };
102 #undef MSG_ID
103
104 static void prepare_msg_ids(void)
105 {
106 int i;
107
108 if (msg_id_info[0].downcased)
109 return;
110
111 /* convert id_string to lower case, without underscores. */
112 for (i = 0; i < FSCK_MSG_MAX; i++) {
113 const char *p = msg_id_info[i].id_string;
114 int len = strlen(p);
115 char *q = xmalloc(len);
116
117 msg_id_info[i].downcased = q;
118 while (*p)
119 if (*p == '_')
120 p++;
121 else
122 *(q)++ = tolower(*(p)++);
123 *q = '\0';
124
125 p = msg_id_info[i].id_string;
126 q = xmalloc(len);
127 msg_id_info[i].camelcased = q;
128 while (*p) {
129 if (*p == '_') {
130 p++;
131 if (*p)
132 *q++ = *p++;
133 } else {
134 *q++ = tolower(*p++);
135 }
136 }
137 *q = '\0';
138 }
139 }
140
141 static int parse_msg_id(const char *text)
142 {
143 int i;
144
145 prepare_msg_ids();
146
147 for (i = 0; i < FSCK_MSG_MAX; i++)
148 if (!strcmp(text, msg_id_info[i].downcased))
149 return i;
150
151 return -1;
152 }
153
154 void list_config_fsck_msg_ids(struct string_list *list, const char *prefix)
155 {
156 int i;
157
158 prepare_msg_ids();
159
160 for (i = 0; i < FSCK_MSG_MAX; i++)
161 list_config_item(list, prefix, msg_id_info[i].camelcased);
162 }
163
164 static int fsck_msg_type(enum fsck_msg_id msg_id,
165 struct fsck_options *options)
166 {
167 int msg_type;
168
169 assert(msg_id >= 0 && msg_id < FSCK_MSG_MAX);
170
171 if (options->msg_type)
172 msg_type = options->msg_type[msg_id];
173 else {
174 msg_type = msg_id_info[msg_id].msg_type;
175 if (options->strict && msg_type == FSCK_WARN)
176 msg_type = FSCK_ERROR;
177 }
178
179 return msg_type;
180 }
181
182 static void init_skiplist(struct fsck_options *options, const char *path)
183 {
184 FILE *fp;
185 struct strbuf sb = STRBUF_INIT;
186 struct object_id oid;
187
188 fp = fopen(path, "r");
189 if (!fp)
190 die("Could not open skip list: %s", path);
191 while (!strbuf_getline(&sb, fp)) {
192 const char *p;
193 if (parse_oid_hex(sb.buf, &oid, &p) || *p != '\0')
194 die("Invalid SHA-1: %s", sb.buf);
195 oidset_insert(&options->skiplist, &oid);
196 }
197 if (ferror(fp))
198 die_errno("Could not read '%s'", path);
199 fclose(fp);
200 strbuf_release(&sb);
201 }
202
203 static int parse_msg_type(const char *str)
204 {
205 if (!strcmp(str, "error"))
206 return FSCK_ERROR;
207 else if (!strcmp(str, "warn"))
208 return FSCK_WARN;
209 else if (!strcmp(str, "ignore"))
210 return FSCK_IGNORE;
211 else
212 die("Unknown fsck message type: '%s'", str);
213 }
214
215 int is_valid_msg_type(const char *msg_id, const char *msg_type)
216 {
217 if (parse_msg_id(msg_id) < 0)
218 return 0;
219 parse_msg_type(msg_type);
220 return 1;
221 }
222
223 void fsck_set_msg_type(struct fsck_options *options,
224 const char *msg_id, const char *msg_type)
225 {
226 int id = parse_msg_id(msg_id), type;
227
228 if (id < 0)
229 die("Unhandled message id: %s", msg_id);
230 type = parse_msg_type(msg_type);
231
232 if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
233 die("Cannot demote %s to %s", msg_id, msg_type);
234
235 if (!options->msg_type) {
236 int i;
237 int *msg_type;
238 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
239 for (i = 0; i < FSCK_MSG_MAX; i++)
240 msg_type[i] = fsck_msg_type(i, options);
241 options->msg_type = msg_type;
242 }
243
244 options->msg_type[id] = type;
245 }
246
247 void fsck_set_msg_types(struct fsck_options *options, const char *values)
248 {
249 char *buf = xstrdup(values), *to_free = buf;
250 int done = 0;
251
252 while (!done) {
253 int len = strcspn(buf, " ,|"), equal;
254
255 done = !buf[len];
256 if (!len) {
257 buf++;
258 continue;
259 }
260 buf[len] = '\0';
261
262 for (equal = 0;
263 equal < len && buf[equal] != '=' && buf[equal] != ':';
264 equal++)
265 buf[equal] = tolower(buf[equal]);
266 buf[equal] = '\0';
267
268 if (!strcmp(buf, "skiplist")) {
269 if (equal == len)
270 die("skiplist requires a path");
271 init_skiplist(options, buf + equal + 1);
272 buf += len + 1;
273 continue;
274 }
275
276 if (equal == len)
277 die("Missing '=': '%s'", buf);
278
279 fsck_set_msg_type(options, buf, buf + equal + 1);
280 buf += len + 1;
281 }
282 free(to_free);
283 }
284
285 static void append_msg_id(struct strbuf *sb, const char *msg_id)
286 {
287 for (;;) {
288 char c = *(msg_id)++;
289
290 if (!c)
291 break;
292 if (c != '_')
293 strbuf_addch(sb, tolower(c));
294 else {
295 assert(*msg_id);
296 strbuf_addch(sb, *(msg_id)++);
297 }
298 }
299
300 strbuf_addstr(sb, ": ");
301 }
302
303 static int object_on_skiplist(struct fsck_options *opts, struct object *obj)
304 {
305 return opts && obj && oidset_contains(&opts->skiplist, &obj->oid);
306 }
307
308 __attribute__((format (printf, 4, 5)))
309 static int report(struct fsck_options *options, struct object *object,
310 enum fsck_msg_id id, const char *fmt, ...)
311 {
312 va_list ap;
313 struct strbuf sb = STRBUF_INIT;
314 int msg_type = fsck_msg_type(id, options), result;
315
316 if (msg_type == FSCK_IGNORE)
317 return 0;
318
319 if (object_on_skiplist(options, object))
320 return 0;
321
322 if (msg_type == FSCK_FATAL)
323 msg_type = FSCK_ERROR;
324 else if (msg_type == FSCK_INFO)
325 msg_type = FSCK_WARN;
326
327 append_msg_id(&sb, msg_id_info[id].id_string);
328
329 va_start(ap, fmt);
330 strbuf_vaddf(&sb, fmt, ap);
331 result = options->error_func(options, object, msg_type, sb.buf);
332 strbuf_release(&sb);
333 va_end(ap);
334
335 return result;
336 }
337
338 static char *get_object_name(struct fsck_options *options, struct object *obj)
339 {
340 if (!options->object_names)
341 return NULL;
342 return lookup_decoration(options->object_names, obj);
343 }
344
345 static void put_object_name(struct fsck_options *options, struct object *obj,
346 const char *fmt, ...)
347 {
348 va_list ap;
349 struct strbuf buf = STRBUF_INIT;
350 char *existing;
351
352 if (!options->object_names)
353 return;
354 existing = lookup_decoration(options->object_names, obj);
355 if (existing)
356 return;
357 va_start(ap, fmt);
358 strbuf_vaddf(&buf, fmt, ap);
359 add_decoration(options->object_names, obj, strbuf_detach(&buf, NULL));
360 va_end(ap);
361 }
362
363 static const char *describe_object(struct fsck_options *o, struct object *obj)
364 {
365 static struct strbuf buf = STRBUF_INIT;
366 char *name;
367
368 strbuf_reset(&buf);
369 strbuf_addstr(&buf, oid_to_hex(&obj->oid));
370 if (o->object_names && (name = lookup_decoration(o->object_names, obj)))
371 strbuf_addf(&buf, " (%s)", name);
372
373 return buf.buf;
374 }
375
376 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
377 {
378 struct tree_desc desc;
379 struct name_entry entry;
380 int res = 0;
381 const char *name;
382
383 if (parse_tree(tree))
384 return -1;
385
386 name = get_object_name(options, &tree->object);
387 if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
388 return -1;
389 while (tree_entry_gently(&desc, &entry)) {
390 struct object *obj;
391 int result;
392
393 if (S_ISGITLINK(entry.mode))
394 continue;
395
396 if (S_ISDIR(entry.mode)) {
397 obj = (struct object *)lookup_tree(the_repository, entry.oid);
398 if (name && obj)
399 put_object_name(options, obj, "%s%s/", name,
400 entry.path);
401 result = options->walk(obj, OBJ_TREE, data, options);
402 }
403 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
404 obj = (struct object *)lookup_blob(the_repository, entry.oid);
405 if (name && obj)
406 put_object_name(options, obj, "%s%s", name,
407 entry.path);
408 result = options->walk(obj, OBJ_BLOB, data, options);
409 }
410 else {
411 result = error("in tree %s: entry %s has bad mode %.6o",
412 describe_object(options, &tree->object), entry.path, entry.mode);
413 }
414 if (result < 0)
415 return result;
416 if (!res)
417 res = result;
418 }
419 return res;
420 }
421
422 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
423 {
424 int counter = 0, generation = 0, name_prefix_len = 0;
425 struct commit_list *parents;
426 int res;
427 int result;
428 const char *name;
429
430 if (parse_commit(commit))
431 return -1;
432
433 name = get_object_name(options, &commit->object);
434 if (name)
435 put_object_name(options, &get_commit_tree(commit)->object,
436 "%s:", name);
437
438 result = options->walk((struct object *)get_commit_tree(commit),
439 OBJ_TREE, data, options);
440 if (result < 0)
441 return result;
442 res = result;
443
444 parents = commit->parents;
445 if (name && parents) {
446 int len = strlen(name), power;
447
448 if (len && name[len - 1] == '^') {
449 generation = 1;
450 name_prefix_len = len - 1;
451 }
452 else { /* parse ~<generation> suffix */
453 for (generation = 0, power = 1;
454 len && isdigit(name[len - 1]);
455 power *= 10)
456 generation += power * (name[--len] - '0');
457 if (power > 1 && len && name[len - 1] == '~')
458 name_prefix_len = len - 1;
459 }
460 }
461
462 while (parents) {
463 if (name) {
464 struct object *obj = &parents->item->object;
465
466 if (++counter > 1)
467 put_object_name(options, obj, "%s^%d",
468 name, counter);
469 else if (generation > 0)
470 put_object_name(options, obj, "%.*s~%d",
471 name_prefix_len, name, generation + 1);
472 else
473 put_object_name(options, obj, "%s^", name);
474 }
475 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
476 if (result < 0)
477 return result;
478 if (!res)
479 res = result;
480 parents = parents->next;
481 }
482 return res;
483 }
484
485 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
486 {
487 char *name = get_object_name(options, &tag->object);
488
489 if (parse_tag(tag))
490 return -1;
491 if (name)
492 put_object_name(options, tag->tagged, "%s", name);
493 return options->walk(tag->tagged, OBJ_ANY, data, options);
494 }
495
496 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
497 {
498 if (!obj)
499 return -1;
500
501 if (obj->type == OBJ_NONE)
502 parse_object(the_repository, &obj->oid);
503
504 switch (obj->type) {
505 case OBJ_BLOB:
506 return 0;
507 case OBJ_TREE:
508 return fsck_walk_tree((struct tree *)obj, data, options);
509 case OBJ_COMMIT:
510 return fsck_walk_commit((struct commit *)obj, data, options);
511 case OBJ_TAG:
512 return fsck_walk_tag((struct tag *)obj, data, options);
513 default:
514 error("Unknown object type for %s", describe_object(options, obj));
515 return -1;
516 }
517 }
518
519 /*
520 * The entries in a tree are ordered in the _path_ order,
521 * which means that a directory entry is ordered by adding
522 * a slash to the end of it.
523 *
524 * So a directory called "a" is ordered _after_ a file
525 * called "a.c", because "a/" sorts after "a.c".
526 */
527 #define TREE_UNORDERED (-1)
528 #define TREE_HAS_DUPS (-2)
529
530 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
531 {
532 int len1 = strlen(name1);
533 int len2 = strlen(name2);
534 int len = len1 < len2 ? len1 : len2;
535 unsigned char c1, c2;
536 int cmp;
537
538 cmp = memcmp(name1, name2, len);
539 if (cmp < 0)
540 return 0;
541 if (cmp > 0)
542 return TREE_UNORDERED;
543
544 /*
545 * Ok, the first <len> characters are the same.
546 * Now we need to order the next one, but turn
547 * a '\0' into a '/' for a directory entry.
548 */
549 c1 = name1[len];
550 c2 = name2[len];
551 if (!c1 && !c2)
552 /*
553 * git-write-tree used to write out a nonsense tree that has
554 * entries with the same name, one blob and one tree. Make
555 * sure we do not have duplicate entries.
556 */
557 return TREE_HAS_DUPS;
558 if (!c1 && S_ISDIR(mode1))
559 c1 = '/';
560 if (!c2 && S_ISDIR(mode2))
561 c2 = '/';
562 return c1 < c2 ? 0 : TREE_UNORDERED;
563 }
564
565 static int fsck_tree(struct tree *item, struct fsck_options *options)
566 {
567 int retval = 0;
568 int has_null_sha1 = 0;
569 int has_full_path = 0;
570 int has_empty_name = 0;
571 int has_dot = 0;
572 int has_dotdot = 0;
573 int has_dotgit = 0;
574 int has_zero_pad = 0;
575 int has_bad_modes = 0;
576 int has_dup_entries = 0;
577 int not_properly_sorted = 0;
578 struct tree_desc desc;
579 unsigned o_mode;
580 const char *o_name;
581
582 if (init_tree_desc_gently(&desc, item->buffer, item->size)) {
583 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
584 return retval;
585 }
586
587 o_mode = 0;
588 o_name = NULL;
589
590 while (desc.size) {
591 unsigned mode;
592 const char *name;
593 const struct object_id *oid;
594
595 oid = tree_entry_extract(&desc, &name, &mode);
596
597 has_null_sha1 |= is_null_oid(oid);
598 has_full_path |= !!strchr(name, '/');
599 has_empty_name |= !*name;
600 has_dot |= !strcmp(name, ".");
601 has_dotdot |= !strcmp(name, "..");
602 has_dotgit |= is_hfs_dotgit(name) || is_ntfs_dotgit(name);
603 has_zero_pad |= *(char *)desc.buffer == '0';
604
605 if (is_hfs_dotgitmodules(name) || is_ntfs_dotgitmodules(name)) {
606 if (!S_ISLNK(mode))
607 oidset_insert(&gitmodules_found, oid);
608 else
609 retval += report(options, &item->object,
610 FSCK_MSG_GITMODULES_SYMLINK,
611 ".gitmodules is a symbolic link");
612 }
613
614 if (update_tree_entry_gently(&desc)) {
615 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
616 break;
617 }
618
619 switch (mode) {
620 /*
621 * Standard modes..
622 */
623 case S_IFREG | 0755:
624 case S_IFREG | 0644:
625 case S_IFLNK:
626 case S_IFDIR:
627 case S_IFGITLINK:
628 break;
629 /*
630 * This is nonstandard, but we had a few of these
631 * early on when we honored the full set of mode
632 * bits..
633 */
634 case S_IFREG | 0664:
635 if (!options->strict)
636 break;
637 /* fallthrough */
638 default:
639 has_bad_modes = 1;
640 }
641
642 if (o_name) {
643 switch (verify_ordered(o_mode, o_name, mode, name)) {
644 case TREE_UNORDERED:
645 not_properly_sorted = 1;
646 break;
647 case TREE_HAS_DUPS:
648 has_dup_entries = 1;
649 break;
650 default:
651 break;
652 }
653 }
654
655 o_mode = mode;
656 o_name = name;
657 }
658
659 if (has_null_sha1)
660 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
661 if (has_full_path)
662 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
663 if (has_empty_name)
664 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
665 if (has_dot)
666 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
667 if (has_dotdot)
668 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
669 if (has_dotgit)
670 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
671 if (has_zero_pad)
672 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
673 if (has_bad_modes)
674 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
675 if (has_dup_entries)
676 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
677 if (not_properly_sorted)
678 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
679 return retval;
680 }
681
682 static int verify_headers(const void *data, unsigned long size,
683 struct object *obj, struct fsck_options *options)
684 {
685 const char *buffer = (const char *)data;
686 unsigned long i;
687
688 for (i = 0; i < size; i++) {
689 switch (buffer[i]) {
690 case '\0':
691 return report(options, obj,
692 FSCK_MSG_NUL_IN_HEADER,
693 "unterminated header: NUL at offset %ld", i);
694 case '\n':
695 if (i + 1 < size && buffer[i + 1] == '\n')
696 return 0;
697 }
698 }
699
700 /*
701 * We did not find double-LF that separates the header
702 * and the body. Not having a body is not a crime but
703 * we do want to see the terminating LF for the last header
704 * line.
705 */
706 if (size && buffer[size - 1] == '\n')
707 return 0;
708
709 return report(options, obj,
710 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
711 }
712
713 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
714 {
715 const char *p = *ident;
716 char *end;
717
718 *ident = strchrnul(*ident, '\n');
719 if (**ident == '\n')
720 (*ident)++;
721
722 if (*p == '<')
723 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
724 p += strcspn(p, "<>\n");
725 if (*p == '>')
726 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
727 if (*p != '<')
728 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
729 if (p[-1] != ' ')
730 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
731 p++;
732 p += strcspn(p, "<>\n");
733 if (*p != '>')
734 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
735 p++;
736 if (*p != ' ')
737 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
738 p++;
739 if (*p == '0' && p[1] != ' ')
740 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
741 if (date_overflows(parse_timestamp(p, &end, 10)))
742 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
743 if ((end == p || *end != ' '))
744 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
745 p = end + 1;
746 if ((*p != '+' && *p != '-') ||
747 !isdigit(p[1]) ||
748 !isdigit(p[2]) ||
749 !isdigit(p[3]) ||
750 !isdigit(p[4]) ||
751 (p[5] != '\n'))
752 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
753 p += 6;
754 return 0;
755 }
756
757 static int fsck_commit_buffer(struct commit *commit, const char *buffer,
758 unsigned long size, struct fsck_options *options)
759 {
760 struct object_id tree_oid, oid;
761 struct commit_graft *graft;
762 unsigned parent_count, parent_line_count = 0, author_count;
763 int err;
764 const char *buffer_begin = buffer;
765 const char *p;
766
767 if (verify_headers(buffer, size, &commit->object, options))
768 return -1;
769
770 if (!skip_prefix(buffer, "tree ", &buffer))
771 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
772 if (parse_oid_hex(buffer, &tree_oid, &p) || *p != '\n') {
773 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
774 if (err)
775 return err;
776 }
777 buffer = p + 1;
778 while (skip_prefix(buffer, "parent ", &buffer)) {
779 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
780 err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
781 if (err)
782 return err;
783 }
784 buffer = p + 1;
785 parent_line_count++;
786 }
787 graft = lookup_commit_graft(the_repository, &commit->object.oid);
788 parent_count = commit_list_count(commit->parents);
789 if (graft) {
790 if (graft->nr_parent == -1 && !parent_count)
791 ; /* shallow commit */
792 else if (graft->nr_parent != parent_count) {
793 err = report(options, &commit->object, FSCK_MSG_MISSING_GRAFT, "graft objects missing");
794 if (err)
795 return err;
796 }
797 } else {
798 if (parent_count != parent_line_count) {
799 err = report(options, &commit->object, FSCK_MSG_MISSING_PARENT, "parent objects missing");
800 if (err)
801 return err;
802 }
803 }
804 author_count = 0;
805 while (skip_prefix(buffer, "author ", &buffer)) {
806 author_count++;
807 err = fsck_ident(&buffer, &commit->object, options);
808 if (err)
809 return err;
810 }
811 if (author_count < 1)
812 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
813 else if (author_count > 1)
814 err = report(options, &commit->object, 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, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
819 err = fsck_ident(&buffer, &commit->object, options);
820 if (err)
821 return err;
822 if (!get_commit_tree(commit)) {
823 err = report(options, &commit->object, FSCK_MSG_BAD_TREE, "could not load commit's tree %s", oid_to_hex(&tree_oid));
824 if (err)
825 return err;
826 }
827 if (memchr(buffer_begin, '\0', size)) {
828 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
829 "NUL byte in the commit object body");
830 if (err)
831 return err;
832 }
833 return 0;
834 }
835
836 static int fsck_commit(struct commit *commit, const char *data,
837 unsigned long size, struct fsck_options *options)
838 {
839 const char *buffer = data ? data : get_commit_buffer(commit, &size);
840 int ret = fsck_commit_buffer(commit, buffer, size, options);
841 if (!data)
842 unuse_commit_buffer(commit, buffer);
843 return ret;
844 }
845
846 static int fsck_tag_buffer(struct tag *tag, const char *data,
847 unsigned long size, struct fsck_options *options)
848 {
849 struct object_id oid;
850 int ret = 0;
851 const char *buffer;
852 char *to_free = NULL, *eol;
853 struct strbuf sb = STRBUF_INIT;
854 const char *p;
855
856 if (data)
857 buffer = data;
858 else {
859 enum object_type type;
860
861 buffer = to_free =
862 read_object_file(&tag->object.oid, &type, &size);
863 if (!buffer)
864 return report(options, &tag->object,
865 FSCK_MSG_MISSING_TAG_OBJECT,
866 "cannot read tag object");
867
868 if (type != OBJ_TAG) {
869 ret = report(options, &tag->object,
870 FSCK_MSG_TAG_OBJECT_NOT_TAG,
871 "expected tag got %s",
872 type_name(type));
873 goto done;
874 }
875 }
876
877 ret = verify_headers(buffer, size, &tag->object, options);
878 if (ret)
879 goto done;
880
881 if (!skip_prefix(buffer, "object ", &buffer)) {
882 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
883 goto done;
884 }
885 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
886 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
887 if (ret)
888 goto done;
889 }
890 buffer = p + 1;
891
892 if (!skip_prefix(buffer, "type ", &buffer)) {
893 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
894 goto done;
895 }
896 eol = strchr(buffer, '\n');
897 if (!eol) {
898 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
899 goto done;
900 }
901 if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
902 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
903 if (ret)
904 goto done;
905 buffer = eol + 1;
906
907 if (!skip_prefix(buffer, "tag ", &buffer)) {
908 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
909 goto done;
910 }
911 eol = strchr(buffer, '\n');
912 if (!eol) {
913 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
914 goto done;
915 }
916 strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
917 if (check_refname_format(sb.buf, 0)) {
918 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
919 "invalid 'tag' name: %.*s",
920 (int)(eol - buffer), buffer);
921 if (ret)
922 goto done;
923 }
924 buffer = eol + 1;
925
926 if (!skip_prefix(buffer, "tagger ", &buffer)) {
927 /* early tags do not contain 'tagger' lines; warn only */
928 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
929 if (ret)
930 goto done;
931 }
932 else
933 ret = fsck_ident(&buffer, &tag->object, options);
934
935 done:
936 strbuf_release(&sb);
937 free(to_free);
938 return ret;
939 }
940
941 static int fsck_tag(struct tag *tag, const char *data,
942 unsigned long size, struct fsck_options *options)
943 {
944 struct object *tagged = tag->tagged;
945
946 if (!tagged)
947 return report(options, &tag->object, FSCK_MSG_BAD_TAG_OBJECT, "could not load tagged object");
948
949 return fsck_tag_buffer(tag, data, size, options);
950 }
951
952 struct fsck_gitmodules_data {
953 struct object *obj;
954 struct fsck_options *options;
955 int ret;
956 };
957
958 static int fsck_gitmodules_fn(const char *var, const char *value, void *vdata)
959 {
960 struct fsck_gitmodules_data *data = vdata;
961 const char *subsection, *key;
962 int subsection_len;
963 char *name;
964
965 if (parse_config_key(var, "submodule", &subsection, &subsection_len, &key) < 0 ||
966 !subsection)
967 return 0;
968
969 name = xmemdupz(subsection, subsection_len);
970 if (check_submodule_name(name) < 0)
971 data->ret |= report(data->options, data->obj,
972 FSCK_MSG_GITMODULES_NAME,
973 "disallowed submodule name: %s",
974 name);
975 free(name);
976
977 return 0;
978 }
979
980 static int fsck_blob(struct blob *blob, const char *buf,
981 unsigned long size, struct fsck_options *options)
982 {
983 struct fsck_gitmodules_data data;
984 struct config_options config_opts = { 0 };
985
986 if (!oidset_contains(&gitmodules_found, &blob->object.oid))
987 return 0;
988 oidset_insert(&gitmodules_done, &blob->object.oid);
989
990 if (object_on_skiplist(options, &blob->object))
991 return 0;
992
993 if (!buf) {
994 /*
995 * A missing buffer here is a sign that the caller found the
996 * blob too gigantic to load into memory. Let's just consider
997 * that an error.
998 */
999 return report(options, &blob->object,
1000 FSCK_MSG_GITMODULES_LARGE,
1001 ".gitmodules too large to parse");
1002 }
1003
1004 data.obj = &blob->object;
1005 data.options = options;
1006 data.ret = 0;
1007 config_opts.error_action = CONFIG_ERROR_SILENT;
1008 if (git_config_from_mem(fsck_gitmodules_fn, CONFIG_ORIGIN_BLOB,
1009 ".gitmodules", buf, size, &data, &config_opts))
1010 data.ret |= report(options, &blob->object,
1011 FSCK_MSG_GITMODULES_PARSE,
1012 "could not parse gitmodules blob");
1013
1014 return data.ret;
1015 }
1016
1017 int fsck_object(struct object *obj, void *data, unsigned long size,
1018 struct fsck_options *options)
1019 {
1020 if (!obj)
1021 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
1022
1023 if (obj->type == OBJ_BLOB)
1024 return fsck_blob((struct blob *)obj, data, size, options);
1025 if (obj->type == OBJ_TREE)
1026 return fsck_tree((struct tree *) obj, options);
1027 if (obj->type == OBJ_COMMIT)
1028 return fsck_commit((struct commit *) obj, (const char *) data,
1029 size, options);
1030 if (obj->type == OBJ_TAG)
1031 return fsck_tag((struct tag *) obj, (const char *) data,
1032 size, options);
1033
1034 return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
1035 obj->type);
1036 }
1037
1038 int fsck_error_function(struct fsck_options *o,
1039 struct object *obj, int msg_type, const char *message)
1040 {
1041 if (msg_type == FSCK_WARN) {
1042 warning("object %s: %s", describe_object(o, obj), message);
1043 return 0;
1044 }
1045 error("object %s: %s", describe_object(o, obj), message);
1046 return 1;
1047 }
1048
1049 int fsck_finish(struct fsck_options *options)
1050 {
1051 int ret = 0;
1052 struct oidset_iter iter;
1053 const struct object_id *oid;
1054
1055 oidset_iter_init(&gitmodules_found, &iter);
1056 while ((oid = oidset_iter_next(&iter))) {
1057 struct blob *blob;
1058 enum object_type type;
1059 unsigned long size;
1060 char *buf;
1061
1062 if (oidset_contains(&gitmodules_done, oid))
1063 continue;
1064
1065 blob = lookup_blob(the_repository, oid);
1066 if (!blob) {
1067 struct object *obj = lookup_unknown_object(oid->hash);
1068 ret |= report(options, obj,
1069 FSCK_MSG_GITMODULES_BLOB,
1070 "non-blob found at .gitmodules");
1071 continue;
1072 }
1073
1074 buf = read_object_file(oid, &type, &size);
1075 if (!buf) {
1076 if (is_promisor_object(&blob->object.oid))
1077 continue;
1078 ret |= report(options, &blob->object,
1079 FSCK_MSG_GITMODULES_MISSING,
1080 "unable to read .gitmodules blob");
1081 continue;
1082 }
1083
1084 if (type == OBJ_BLOB)
1085 ret |= fsck_blob(blob, buf, size, options);
1086 else
1087 ret |= report(options, &blob->object,
1088 FSCK_MSG_GITMODULES_BLOB,
1089 "non-blob found at .gitmodules");
1090 free(buf);
1091 }
1092
1093
1094 oidset_clear(&gitmodules_found);
1095 oidset_clear(&gitmodules_done);
1096 return ret;
1097 }