tree: add repository argument to lookup_tree
[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 "sha1-array.h"
14 #include "decorate.h"
15 #include "oidset.h"
16 #include "packfile.h"
17 #include "submodule-config.h"
18 #include "config.h"
19 #include "help.h"
20
21 static struct oidset gitmodules_found = OIDSET_INIT;
22 static struct oidset gitmodules_done = OIDSET_INIT;
23
24 #define FSCK_FATAL -1
25 #define FSCK_INFO -2
26
27 #define FOREACH_MSG_ID(FUNC) \
28 /* fatal errors */ \
29 FUNC(NUL_IN_HEADER, FATAL) \
30 FUNC(UNTERMINATED_HEADER, FATAL) \
31 /* errors */ \
32 FUNC(BAD_DATE, ERROR) \
33 FUNC(BAD_DATE_OVERFLOW, ERROR) \
34 FUNC(BAD_EMAIL, ERROR) \
35 FUNC(BAD_NAME, ERROR) \
36 FUNC(BAD_OBJECT_SHA1, ERROR) \
37 FUNC(BAD_PARENT_SHA1, ERROR) \
38 FUNC(BAD_TAG_OBJECT, ERROR) \
39 FUNC(BAD_TIMEZONE, ERROR) \
40 FUNC(BAD_TREE, ERROR) \
41 FUNC(BAD_TREE_SHA1, ERROR) \
42 FUNC(BAD_TYPE, ERROR) \
43 FUNC(DUPLICATE_ENTRIES, ERROR) \
44 FUNC(MISSING_AUTHOR, ERROR) \
45 FUNC(MISSING_COMMITTER, ERROR) \
46 FUNC(MISSING_EMAIL, ERROR) \
47 FUNC(MISSING_GRAFT, ERROR) \
48 FUNC(MISSING_NAME_BEFORE_EMAIL, ERROR) \
49 FUNC(MISSING_OBJECT, ERROR) \
50 FUNC(MISSING_PARENT, ERROR) \
51 FUNC(MISSING_SPACE_BEFORE_DATE, ERROR) \
52 FUNC(MISSING_SPACE_BEFORE_EMAIL, ERROR) \
53 FUNC(MISSING_TAG, ERROR) \
54 FUNC(MISSING_TAG_ENTRY, ERROR) \
55 FUNC(MISSING_TAG_OBJECT, ERROR) \
56 FUNC(MISSING_TREE, ERROR) \
57 FUNC(MISSING_TREE_OBJECT, ERROR) \
58 FUNC(MISSING_TYPE, ERROR) \
59 FUNC(MISSING_TYPE_ENTRY, ERROR) \
60 FUNC(MULTIPLE_AUTHORS, ERROR) \
61 FUNC(TAG_OBJECT_NOT_TAG, ERROR) \
62 FUNC(TREE_NOT_SORTED, ERROR) \
63 FUNC(UNKNOWN_TYPE, ERROR) \
64 FUNC(ZERO_PADDED_DATE, ERROR) \
65 FUNC(GITMODULES_MISSING, ERROR) \
66 FUNC(GITMODULES_BLOB, ERROR) \
67 FUNC(GITMODULES_PARSE, ERROR) \
68 FUNC(GITMODULES_NAME, ERROR) \
69 FUNC(GITMODULES_SYMLINK, ERROR) \
70 /* warnings */ \
71 FUNC(BAD_FILEMODE, WARN) \
72 FUNC(EMPTY_NAME, WARN) \
73 FUNC(FULL_PATHNAME, WARN) \
74 FUNC(HAS_DOT, WARN) \
75 FUNC(HAS_DOTDOT, WARN) \
76 FUNC(HAS_DOTGIT, WARN) \
77 FUNC(NULL_SHA1, WARN) \
78 FUNC(ZERO_PADDED_FILEMODE, WARN) \
79 FUNC(NUL_IN_COMMIT, WARN) \
80 /* infos (reported as warnings, but ignored by default) */ \
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 static struct oid_array skiplist = OID_ARRAY_INIT;
185 int sorted, fd;
186 char buffer[GIT_MAX_HEXSZ + 1];
187 struct object_id oid;
188
189 if (options->skiplist)
190 sorted = options->skiplist->sorted;
191 else {
192 sorted = 1;
193 options->skiplist = &skiplist;
194 }
195
196 fd = open(path, O_RDONLY);
197 if (fd < 0)
198 die("Could not open skip list: %s", path);
199 for (;;) {
200 const char *p;
201 int result = read_in_full(fd, buffer, sizeof(buffer));
202 if (result < 0)
203 die_errno("Could not read '%s'", path);
204 if (!result)
205 break;
206 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n')
207 die("Invalid SHA-1: %s", buffer);
208 oid_array_append(&skiplist, &oid);
209 if (sorted && skiplist.nr > 1 &&
210 oidcmp(&skiplist.oid[skiplist.nr - 2],
211 &oid) > 0)
212 sorted = 0;
213 }
214 close(fd);
215
216 if (sorted)
217 skiplist.sorted = 1;
218 }
219
220 static int parse_msg_type(const char *str)
221 {
222 if (!strcmp(str, "error"))
223 return FSCK_ERROR;
224 else if (!strcmp(str, "warn"))
225 return FSCK_WARN;
226 else if (!strcmp(str, "ignore"))
227 return FSCK_IGNORE;
228 else
229 die("Unknown fsck message type: '%s'", str);
230 }
231
232 int is_valid_msg_type(const char *msg_id, const char *msg_type)
233 {
234 if (parse_msg_id(msg_id) < 0)
235 return 0;
236 parse_msg_type(msg_type);
237 return 1;
238 }
239
240 void fsck_set_msg_type(struct fsck_options *options,
241 const char *msg_id, const char *msg_type)
242 {
243 int id = parse_msg_id(msg_id), type;
244
245 if (id < 0)
246 die("Unhandled message id: %s", msg_id);
247 type = parse_msg_type(msg_type);
248
249 if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
250 die("Cannot demote %s to %s", msg_id, msg_type);
251
252 if (!options->msg_type) {
253 int i;
254 int *msg_type;
255 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
256 for (i = 0; i < FSCK_MSG_MAX; i++)
257 msg_type[i] = fsck_msg_type(i, options);
258 options->msg_type = msg_type;
259 }
260
261 options->msg_type[id] = type;
262 }
263
264 void fsck_set_msg_types(struct fsck_options *options, const char *values)
265 {
266 char *buf = xstrdup(values), *to_free = buf;
267 int done = 0;
268
269 while (!done) {
270 int len = strcspn(buf, " ,|"), equal;
271
272 done = !buf[len];
273 if (!len) {
274 buf++;
275 continue;
276 }
277 buf[len] = '\0';
278
279 for (equal = 0;
280 equal < len && buf[equal] != '=' && buf[equal] != ':';
281 equal++)
282 buf[equal] = tolower(buf[equal]);
283 buf[equal] = '\0';
284
285 if (!strcmp(buf, "skiplist")) {
286 if (equal == len)
287 die("skiplist requires a path");
288 init_skiplist(options, buf + equal + 1);
289 buf += len + 1;
290 continue;
291 }
292
293 if (equal == len)
294 die("Missing '=': '%s'", buf);
295
296 fsck_set_msg_type(options, buf, buf + equal + 1);
297 buf += len + 1;
298 }
299 free(to_free);
300 }
301
302 static void append_msg_id(struct strbuf *sb, const char *msg_id)
303 {
304 for (;;) {
305 char c = *(msg_id)++;
306
307 if (!c)
308 break;
309 if (c != '_')
310 strbuf_addch(sb, tolower(c));
311 else {
312 assert(*msg_id);
313 strbuf_addch(sb, *(msg_id)++);
314 }
315 }
316
317 strbuf_addstr(sb, ": ");
318 }
319
320 __attribute__((format (printf, 4, 5)))
321 static int report(struct fsck_options *options, struct object *object,
322 enum fsck_msg_id id, const char *fmt, ...)
323 {
324 va_list ap;
325 struct strbuf sb = STRBUF_INIT;
326 int msg_type = fsck_msg_type(id, options), result;
327
328 if (msg_type == FSCK_IGNORE)
329 return 0;
330
331 if (options->skiplist && object &&
332 oid_array_lookup(options->skiplist, &object->oid) >= 0)
333 return 0;
334
335 if (msg_type == FSCK_FATAL)
336 msg_type = FSCK_ERROR;
337 else if (msg_type == FSCK_INFO)
338 msg_type = FSCK_WARN;
339
340 append_msg_id(&sb, msg_id_info[id].id_string);
341
342 va_start(ap, fmt);
343 strbuf_vaddf(&sb, fmt, ap);
344 result = options->error_func(options, object, msg_type, sb.buf);
345 strbuf_release(&sb);
346 va_end(ap);
347
348 return result;
349 }
350
351 static char *get_object_name(struct fsck_options *options, struct object *obj)
352 {
353 if (!options->object_names)
354 return NULL;
355 return lookup_decoration(options->object_names, obj);
356 }
357
358 static void put_object_name(struct fsck_options *options, struct object *obj,
359 const char *fmt, ...)
360 {
361 va_list ap;
362 struct strbuf buf = STRBUF_INIT;
363 char *existing;
364
365 if (!options->object_names)
366 return;
367 existing = lookup_decoration(options->object_names, obj);
368 if (existing)
369 return;
370 va_start(ap, fmt);
371 strbuf_vaddf(&buf, fmt, ap);
372 add_decoration(options->object_names, obj, strbuf_detach(&buf, NULL));
373 va_end(ap);
374 }
375
376 static const char *describe_object(struct fsck_options *o, struct object *obj)
377 {
378 static struct strbuf buf = STRBUF_INIT;
379 char *name;
380
381 strbuf_reset(&buf);
382 strbuf_addstr(&buf, oid_to_hex(&obj->oid));
383 if (o->object_names && (name = lookup_decoration(o->object_names, obj)))
384 strbuf_addf(&buf, " (%s)", name);
385
386 return buf.buf;
387 }
388
389 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
390 {
391 struct tree_desc desc;
392 struct name_entry entry;
393 int res = 0;
394 const char *name;
395
396 if (parse_tree(tree))
397 return -1;
398
399 name = get_object_name(options, &tree->object);
400 if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
401 return -1;
402 while (tree_entry_gently(&desc, &entry)) {
403 struct object *obj;
404 int result;
405
406 if (S_ISGITLINK(entry.mode))
407 continue;
408
409 if (S_ISDIR(entry.mode)) {
410 obj = (struct object *)lookup_tree(the_repository, entry.oid);
411 if (name && obj)
412 put_object_name(options, obj, "%s%s/", name,
413 entry.path);
414 result = options->walk(obj, OBJ_TREE, data, options);
415 }
416 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
417 obj = (struct object *)lookup_blob(the_repository, entry.oid);
418 if (name && obj)
419 put_object_name(options, obj, "%s%s", name,
420 entry.path);
421 result = options->walk(obj, OBJ_BLOB, data, options);
422 }
423 else {
424 result = error("in tree %s: entry %s has bad mode %.6o",
425 describe_object(options, &tree->object), entry.path, entry.mode);
426 }
427 if (result < 0)
428 return result;
429 if (!res)
430 res = result;
431 }
432 return res;
433 }
434
435 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
436 {
437 int counter = 0, generation = 0, name_prefix_len = 0;
438 struct commit_list *parents;
439 int res;
440 int result;
441 const char *name;
442
443 if (parse_commit(commit))
444 return -1;
445
446 name = get_object_name(options, &commit->object);
447 if (name)
448 put_object_name(options, &get_commit_tree(commit)->object,
449 "%s:", name);
450
451 result = options->walk((struct object *)get_commit_tree(commit),
452 OBJ_TREE, data, options);
453 if (result < 0)
454 return result;
455 res = result;
456
457 parents = commit->parents;
458 if (name && parents) {
459 int len = strlen(name), power;
460
461 if (len && name[len - 1] == '^') {
462 generation = 1;
463 name_prefix_len = len - 1;
464 }
465 else { /* parse ~<generation> suffix */
466 for (generation = 0, power = 1;
467 len && isdigit(name[len - 1]);
468 power *= 10)
469 generation += power * (name[--len] - '0');
470 if (power > 1 && len && name[len - 1] == '~')
471 name_prefix_len = len - 1;
472 }
473 }
474
475 while (parents) {
476 if (name) {
477 struct object *obj = &parents->item->object;
478
479 if (++counter > 1)
480 put_object_name(options, obj, "%s^%d",
481 name, counter);
482 else if (generation > 0)
483 put_object_name(options, obj, "%.*s~%d",
484 name_prefix_len, name, generation + 1);
485 else
486 put_object_name(options, obj, "%s^", name);
487 }
488 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
489 if (result < 0)
490 return result;
491 if (!res)
492 res = result;
493 parents = parents->next;
494 }
495 return res;
496 }
497
498 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
499 {
500 char *name = get_object_name(options, &tag->object);
501
502 if (parse_tag(tag))
503 return -1;
504 if (name)
505 put_object_name(options, tag->tagged, "%s", name);
506 return options->walk(tag->tagged, OBJ_ANY, data, options);
507 }
508
509 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
510 {
511 if (!obj)
512 return -1;
513
514 if (obj->type == OBJ_NONE)
515 parse_object(the_repository, &obj->oid);
516
517 switch (obj->type) {
518 case OBJ_BLOB:
519 return 0;
520 case OBJ_TREE:
521 return fsck_walk_tree((struct tree *)obj, data, options);
522 case OBJ_COMMIT:
523 return fsck_walk_commit((struct commit *)obj, data, options);
524 case OBJ_TAG:
525 return fsck_walk_tag((struct tag *)obj, data, options);
526 default:
527 error("Unknown object type for %s", describe_object(options, obj));
528 return -1;
529 }
530 }
531
532 /*
533 * The entries in a tree are ordered in the _path_ order,
534 * which means that a directory entry is ordered by adding
535 * a slash to the end of it.
536 *
537 * So a directory called "a" is ordered _after_ a file
538 * called "a.c", because "a/" sorts after "a.c".
539 */
540 #define TREE_UNORDERED (-1)
541 #define TREE_HAS_DUPS (-2)
542
543 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
544 {
545 int len1 = strlen(name1);
546 int len2 = strlen(name2);
547 int len = len1 < len2 ? len1 : len2;
548 unsigned char c1, c2;
549 int cmp;
550
551 cmp = memcmp(name1, name2, len);
552 if (cmp < 0)
553 return 0;
554 if (cmp > 0)
555 return TREE_UNORDERED;
556
557 /*
558 * Ok, the first <len> characters are the same.
559 * Now we need to order the next one, but turn
560 * a '\0' into a '/' for a directory entry.
561 */
562 c1 = name1[len];
563 c2 = name2[len];
564 if (!c1 && !c2)
565 /*
566 * git-write-tree used to write out a nonsense tree that has
567 * entries with the same name, one blob and one tree. Make
568 * sure we do not have duplicate entries.
569 */
570 return TREE_HAS_DUPS;
571 if (!c1 && S_ISDIR(mode1))
572 c1 = '/';
573 if (!c2 && S_ISDIR(mode2))
574 c2 = '/';
575 return c1 < c2 ? 0 : TREE_UNORDERED;
576 }
577
578 static int fsck_tree(struct tree *item, struct fsck_options *options)
579 {
580 int retval = 0;
581 int has_null_sha1 = 0;
582 int has_full_path = 0;
583 int has_empty_name = 0;
584 int has_dot = 0;
585 int has_dotdot = 0;
586 int has_dotgit = 0;
587 int has_zero_pad = 0;
588 int has_bad_modes = 0;
589 int has_dup_entries = 0;
590 int not_properly_sorted = 0;
591 struct tree_desc desc;
592 unsigned o_mode;
593 const char *o_name;
594
595 if (init_tree_desc_gently(&desc, item->buffer, item->size)) {
596 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
597 return retval;
598 }
599
600 o_mode = 0;
601 o_name = NULL;
602
603 while (desc.size) {
604 unsigned mode;
605 const char *name;
606 const struct object_id *oid;
607
608 oid = tree_entry_extract(&desc, &name, &mode);
609
610 has_null_sha1 |= is_null_oid(oid);
611 has_full_path |= !!strchr(name, '/');
612 has_empty_name |= !*name;
613 has_dot |= !strcmp(name, ".");
614 has_dotdot |= !strcmp(name, "..");
615 has_dotgit |= is_hfs_dotgit(name) || is_ntfs_dotgit(name);
616 has_zero_pad |= *(char *)desc.buffer == '0';
617
618 if (is_hfs_dotgitmodules(name) || is_ntfs_dotgitmodules(name)) {
619 if (!S_ISLNK(mode))
620 oidset_insert(&gitmodules_found, oid);
621 else
622 retval += report(options, &item->object,
623 FSCK_MSG_GITMODULES_SYMLINK,
624 ".gitmodules is a symbolic link");
625 }
626
627 if (update_tree_entry_gently(&desc)) {
628 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
629 break;
630 }
631
632 switch (mode) {
633 /*
634 * Standard modes..
635 */
636 case S_IFREG | 0755:
637 case S_IFREG | 0644:
638 case S_IFLNK:
639 case S_IFDIR:
640 case S_IFGITLINK:
641 break;
642 /*
643 * This is nonstandard, but we had a few of these
644 * early on when we honored the full set of mode
645 * bits..
646 */
647 case S_IFREG | 0664:
648 if (!options->strict)
649 break;
650 /* fallthrough */
651 default:
652 has_bad_modes = 1;
653 }
654
655 if (o_name) {
656 switch (verify_ordered(o_mode, o_name, mode, name)) {
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 if (has_null_sha1)
673 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
674 if (has_full_path)
675 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
676 if (has_empty_name)
677 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
678 if (has_dot)
679 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
680 if (has_dotdot)
681 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
682 if (has_dotgit)
683 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
684 if (has_zero_pad)
685 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
686 if (has_bad_modes)
687 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
688 if (has_dup_entries)
689 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
690 if (not_properly_sorted)
691 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
692 return retval;
693 }
694
695 static int verify_headers(const void *data, unsigned long size,
696 struct object *obj, struct fsck_options *options)
697 {
698 const char *buffer = (const char *)data;
699 unsigned long i;
700
701 for (i = 0; i < size; i++) {
702 switch (buffer[i]) {
703 case '\0':
704 return report(options, obj,
705 FSCK_MSG_NUL_IN_HEADER,
706 "unterminated header: NUL at offset %ld", i);
707 case '\n':
708 if (i + 1 < size && buffer[i + 1] == '\n')
709 return 0;
710 }
711 }
712
713 /*
714 * We did not find double-LF that separates the header
715 * and the body. Not having a body is not a crime but
716 * we do want to see the terminating LF for the last header
717 * line.
718 */
719 if (size && buffer[size - 1] == '\n')
720 return 0;
721
722 return report(options, obj,
723 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
724 }
725
726 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
727 {
728 const char *p = *ident;
729 char *end;
730
731 *ident = strchrnul(*ident, '\n');
732 if (**ident == '\n')
733 (*ident)++;
734
735 if (*p == '<')
736 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
737 p += strcspn(p, "<>\n");
738 if (*p == '>')
739 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
740 if (*p != '<')
741 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
742 if (p[-1] != ' ')
743 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
744 p++;
745 p += strcspn(p, "<>\n");
746 if (*p != '>')
747 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
748 p++;
749 if (*p != ' ')
750 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
751 p++;
752 if (*p == '0' && p[1] != ' ')
753 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
754 if (date_overflows(parse_timestamp(p, &end, 10)))
755 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
756 if ((end == p || *end != ' '))
757 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
758 p = end + 1;
759 if ((*p != '+' && *p != '-') ||
760 !isdigit(p[1]) ||
761 !isdigit(p[2]) ||
762 !isdigit(p[3]) ||
763 !isdigit(p[4]) ||
764 (p[5] != '\n'))
765 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
766 p += 6;
767 return 0;
768 }
769
770 static int fsck_commit_buffer(struct commit *commit, const char *buffer,
771 unsigned long size, struct fsck_options *options)
772 {
773 struct object_id tree_oid, oid;
774 struct commit_graft *graft;
775 unsigned parent_count, parent_line_count = 0, author_count;
776 int err;
777 const char *buffer_begin = buffer;
778 const char *p;
779
780 if (verify_headers(buffer, size, &commit->object, options))
781 return -1;
782
783 if (!skip_prefix(buffer, "tree ", &buffer))
784 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
785 if (parse_oid_hex(buffer, &tree_oid, &p) || *p != '\n') {
786 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
787 if (err)
788 return err;
789 }
790 buffer = p + 1;
791 while (skip_prefix(buffer, "parent ", &buffer)) {
792 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
793 err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
794 if (err)
795 return err;
796 }
797 buffer = p + 1;
798 parent_line_count++;
799 }
800 graft = lookup_commit_graft(the_repository, &commit->object.oid);
801 parent_count = commit_list_count(commit->parents);
802 if (graft) {
803 if (graft->nr_parent == -1 && !parent_count)
804 ; /* shallow commit */
805 else if (graft->nr_parent != parent_count) {
806 err = report(options, &commit->object, FSCK_MSG_MISSING_GRAFT, "graft objects missing");
807 if (err)
808 return err;
809 }
810 } else {
811 if (parent_count != parent_line_count) {
812 err = report(options, &commit->object, FSCK_MSG_MISSING_PARENT, "parent objects missing");
813 if (err)
814 return err;
815 }
816 }
817 author_count = 0;
818 while (skip_prefix(buffer, "author ", &buffer)) {
819 author_count++;
820 err = fsck_ident(&buffer, &commit->object, options);
821 if (err)
822 return err;
823 }
824 if (author_count < 1)
825 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
826 else if (author_count > 1)
827 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
828 if (err)
829 return err;
830 if (!skip_prefix(buffer, "committer ", &buffer))
831 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
832 err = fsck_ident(&buffer, &commit->object, options);
833 if (err)
834 return err;
835 if (!get_commit_tree(commit)) {
836 err = report(options, &commit->object, FSCK_MSG_BAD_TREE, "could not load commit's tree %s", oid_to_hex(&tree_oid));
837 if (err)
838 return err;
839 }
840 if (memchr(buffer_begin, '\0', size)) {
841 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
842 "NUL byte in the commit object body");
843 if (err)
844 return err;
845 }
846 return 0;
847 }
848
849 static int fsck_commit(struct commit *commit, const char *data,
850 unsigned long size, struct fsck_options *options)
851 {
852 const char *buffer = data ? data : get_commit_buffer(commit, &size);
853 int ret = fsck_commit_buffer(commit, buffer, size, options);
854 if (!data)
855 unuse_commit_buffer(commit, buffer);
856 return ret;
857 }
858
859 static int fsck_tag_buffer(struct tag *tag, const char *data,
860 unsigned long size, struct fsck_options *options)
861 {
862 struct object_id oid;
863 int ret = 0;
864 const char *buffer;
865 char *to_free = NULL, *eol;
866 struct strbuf sb = STRBUF_INIT;
867 const char *p;
868
869 if (data)
870 buffer = data;
871 else {
872 enum object_type type;
873
874 buffer = to_free =
875 read_object_file(&tag->object.oid, &type, &size);
876 if (!buffer)
877 return report(options, &tag->object,
878 FSCK_MSG_MISSING_TAG_OBJECT,
879 "cannot read tag object");
880
881 if (type != OBJ_TAG) {
882 ret = report(options, &tag->object,
883 FSCK_MSG_TAG_OBJECT_NOT_TAG,
884 "expected tag got %s",
885 type_name(type));
886 goto done;
887 }
888 }
889
890 ret = verify_headers(buffer, size, &tag->object, options);
891 if (ret)
892 goto done;
893
894 if (!skip_prefix(buffer, "object ", &buffer)) {
895 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
896 goto done;
897 }
898 if (parse_oid_hex(buffer, &oid, &p) || *p != '\n') {
899 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
900 if (ret)
901 goto done;
902 }
903 buffer = p + 1;
904
905 if (!skip_prefix(buffer, "type ", &buffer)) {
906 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
907 goto done;
908 }
909 eol = strchr(buffer, '\n');
910 if (!eol) {
911 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
912 goto done;
913 }
914 if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
915 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
916 if (ret)
917 goto done;
918 buffer = eol + 1;
919
920 if (!skip_prefix(buffer, "tag ", &buffer)) {
921 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
922 goto done;
923 }
924 eol = strchr(buffer, '\n');
925 if (!eol) {
926 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
927 goto done;
928 }
929 strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
930 if (check_refname_format(sb.buf, 0)) {
931 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
932 "invalid 'tag' name: %.*s",
933 (int)(eol - buffer), buffer);
934 if (ret)
935 goto done;
936 }
937 buffer = eol + 1;
938
939 if (!skip_prefix(buffer, "tagger ", &buffer)) {
940 /* early tags do not contain 'tagger' lines; warn only */
941 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
942 if (ret)
943 goto done;
944 }
945 else
946 ret = fsck_ident(&buffer, &tag->object, options);
947
948 done:
949 strbuf_release(&sb);
950 free(to_free);
951 return ret;
952 }
953
954 static int fsck_tag(struct tag *tag, const char *data,
955 unsigned long size, struct fsck_options *options)
956 {
957 struct object *tagged = tag->tagged;
958
959 if (!tagged)
960 return report(options, &tag->object, FSCK_MSG_BAD_TAG_OBJECT, "could not load tagged object");
961
962 return fsck_tag_buffer(tag, data, size, options);
963 }
964
965 struct fsck_gitmodules_data {
966 struct object *obj;
967 struct fsck_options *options;
968 int ret;
969 };
970
971 static int fsck_gitmodules_fn(const char *var, const char *value, void *vdata)
972 {
973 struct fsck_gitmodules_data *data = vdata;
974 const char *subsection, *key;
975 int subsection_len;
976 char *name;
977
978 if (parse_config_key(var, "submodule", &subsection, &subsection_len, &key) < 0 ||
979 !subsection)
980 return 0;
981
982 name = xmemdupz(subsection, subsection_len);
983 if (check_submodule_name(name) < 0)
984 data->ret |= report(data->options, data->obj,
985 FSCK_MSG_GITMODULES_NAME,
986 "disallowed submodule name: %s",
987 name);
988 free(name);
989
990 return 0;
991 }
992
993 static int fsck_blob(struct blob *blob, const char *buf,
994 unsigned long size, struct fsck_options *options)
995 {
996 struct fsck_gitmodules_data data;
997
998 if (!oidset_contains(&gitmodules_found, &blob->object.oid))
999 return 0;
1000 oidset_insert(&gitmodules_done, &blob->object.oid);
1001
1002 if (!buf) {
1003 /*
1004 * A missing buffer here is a sign that the caller found the
1005 * blob too gigantic to load into memory. Let's just consider
1006 * that an error.
1007 */
1008 return report(options, &blob->object,
1009 FSCK_MSG_GITMODULES_PARSE,
1010 ".gitmodules too large to parse");
1011 }
1012
1013 data.obj = &blob->object;
1014 data.options = options;
1015 data.ret = 0;
1016 if (git_config_from_mem(fsck_gitmodules_fn, CONFIG_ORIGIN_BLOB,
1017 ".gitmodules", buf, size, &data))
1018 data.ret |= report(options, &blob->object,
1019 FSCK_MSG_GITMODULES_PARSE,
1020 "could not parse gitmodules blob");
1021
1022 return data.ret;
1023 }
1024
1025 int fsck_object(struct object *obj, void *data, unsigned long size,
1026 struct fsck_options *options)
1027 {
1028 if (!obj)
1029 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
1030
1031 if (obj->type == OBJ_BLOB)
1032 return fsck_blob((struct blob *)obj, data, size, options);
1033 if (obj->type == OBJ_TREE)
1034 return fsck_tree((struct tree *) obj, options);
1035 if (obj->type == OBJ_COMMIT)
1036 return fsck_commit((struct commit *) obj, (const char *) data,
1037 size, options);
1038 if (obj->type == OBJ_TAG)
1039 return fsck_tag((struct tag *) obj, (const char *) data,
1040 size, options);
1041
1042 return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
1043 obj->type);
1044 }
1045
1046 int fsck_error_function(struct fsck_options *o,
1047 struct object *obj, int msg_type, const char *message)
1048 {
1049 if (msg_type == FSCK_WARN) {
1050 warning("object %s: %s", describe_object(o, obj), message);
1051 return 0;
1052 }
1053 error("object %s: %s", describe_object(o, obj), message);
1054 return 1;
1055 }
1056
1057 int fsck_finish(struct fsck_options *options)
1058 {
1059 int ret = 0;
1060 struct oidset_iter iter;
1061 const struct object_id *oid;
1062
1063 oidset_iter_init(&gitmodules_found, &iter);
1064 while ((oid = oidset_iter_next(&iter))) {
1065 struct blob *blob;
1066 enum object_type type;
1067 unsigned long size;
1068 char *buf;
1069
1070 if (oidset_contains(&gitmodules_done, oid))
1071 continue;
1072
1073 blob = lookup_blob(the_repository, oid);
1074 if (!blob) {
1075 struct object *obj = lookup_unknown_object(oid->hash);
1076 ret |= report(options, obj,
1077 FSCK_MSG_GITMODULES_BLOB,
1078 "non-blob found at .gitmodules");
1079 continue;
1080 }
1081
1082 buf = read_object_file(oid, &type, &size);
1083 if (!buf) {
1084 if (is_promisor_object(&blob->object.oid))
1085 continue;
1086 ret |= report(options, &blob->object,
1087 FSCK_MSG_GITMODULES_MISSING,
1088 "unable to read .gitmodules blob");
1089 continue;
1090 }
1091
1092 if (type == OBJ_BLOB)
1093 ret |= fsck_blob(blob, buf, size, options);
1094 else
1095 ret |= report(options, &blob->object,
1096 FSCK_MSG_GITMODULES_BLOB,
1097 "non-blob found at .gitmodules");
1098 free(buf);
1099 }
1100
1101
1102 oidset_clear(&gitmodules_found);
1103 oidset_clear(&gitmodules_done);
1104 return ret;
1105 }