]>
Commit | Line | Data |
---|---|---|
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 FSCK_FATAL -1 | |
26 | #define FSCK_INFO -2 | |
27 | ||
28 | #define FOREACH_MSG_ID(FUNC) \ | |
29 | /* fatal errors */ \ | |
30 | FUNC(NUL_IN_HEADER, FATAL) \ | |
31 | FUNC(UNTERMINATED_HEADER, FATAL) \ | |
32 | /* errors */ \ | |
33 | FUNC(BAD_DATE, ERROR) \ | |
34 | FUNC(BAD_DATE_OVERFLOW, ERROR) \ | |
35 | FUNC(BAD_EMAIL, ERROR) \ | |
36 | FUNC(BAD_NAME, ERROR) \ | |
37 | FUNC(BAD_OBJECT_SHA1, ERROR) \ | |
38 | FUNC(BAD_PARENT_SHA1, ERROR) \ | |
39 | FUNC(BAD_TAG_OBJECT, ERROR) \ | |
40 | FUNC(BAD_TIMEZONE, ERROR) \ | |
41 | FUNC(BAD_TREE, ERROR) \ | |
42 | FUNC(BAD_TREE_SHA1, ERROR) \ | |
43 | FUNC(BAD_TYPE, ERROR) \ | |
44 | FUNC(DUPLICATE_ENTRIES, ERROR) \ | |
45 | FUNC(MISSING_AUTHOR, ERROR) \ | |
46 | FUNC(MISSING_COMMITTER, ERROR) \ | |
47 | FUNC(MISSING_EMAIL, ERROR) \ | |
48 | FUNC(MISSING_NAME_BEFORE_EMAIL, ERROR) \ | |
49 | FUNC(MISSING_OBJECT, 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_TREE, ERROR) \ | |
55 | FUNC(MISSING_TREE_OBJECT, ERROR) \ | |
56 | FUNC(MISSING_TYPE, ERROR) \ | |
57 | FUNC(MISSING_TYPE_ENTRY, ERROR) \ | |
58 | FUNC(MULTIPLE_AUTHORS, ERROR) \ | |
59 | FUNC(TREE_NOT_SORTED, ERROR) \ | |
60 | FUNC(UNKNOWN_TYPE, ERROR) \ | |
61 | FUNC(ZERO_PADDED_DATE, ERROR) \ | |
62 | FUNC(GITMODULES_MISSING, ERROR) \ | |
63 | FUNC(GITMODULES_BLOB, ERROR) \ | |
64 | FUNC(GITMODULES_LARGE, ERROR) \ | |
65 | FUNC(GITMODULES_NAME, ERROR) \ | |
66 | FUNC(GITMODULES_SYMLINK, ERROR) \ | |
67 | FUNC(GITMODULES_URL, ERROR) \ | |
68 | FUNC(GITMODULES_PATH, ERROR) \ | |
69 | FUNC(GITMODULES_UPDATE, 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(GITMODULES_PARSE, INFO) \ | |
82 | FUNC(BAD_TAG_NAME, INFO) \ | |
83 | FUNC(MISSING_TAGGER_ENTRY, INFO) | |
84 | ||
85 | #define MSG_ID(id, msg_type) FSCK_MSG_##id, | |
86 | enum fsck_msg_id { | |
87 | FOREACH_MSG_ID(MSG_ID) | |
88 | FSCK_MSG_MAX | |
89 | }; | |
90 | #undef MSG_ID | |
91 | ||
92 | #define STR(x) #x | |
93 | #define MSG_ID(id, msg_type) { STR(id), NULL, NULL, FSCK_##msg_type }, | |
94 | static struct { | |
95 | const char *id_string; | |
96 | const char *downcased; | |
97 | const char *camelcased; | |
98 | int msg_type; | |
99 | } msg_id_info[FSCK_MSG_MAX + 1] = { | |
100 | FOREACH_MSG_ID(MSG_ID) | |
101 | { NULL, NULL, NULL, -1 } | |
102 | }; | |
103 | #undef MSG_ID | |
104 | ||
105 | static void prepare_msg_ids(void) | |
106 | { | |
107 | int i; | |
108 | ||
109 | if (msg_id_info[0].downcased) | |
110 | return; | |
111 | ||
112 | /* convert id_string to lower case, without underscores. */ | |
113 | for (i = 0; i < FSCK_MSG_MAX; i++) { | |
114 | const char *p = msg_id_info[i].id_string; | |
115 | int len = strlen(p); | |
116 | char *q = xmalloc(len); | |
117 | ||
118 | msg_id_info[i].downcased = q; | |
119 | while (*p) | |
120 | if (*p == '_') | |
121 | p++; | |
122 | else | |
123 | *(q)++ = tolower(*(p)++); | |
124 | *q = '\0'; | |
125 | ||
126 | p = msg_id_info[i].id_string; | |
127 | q = xmalloc(len); | |
128 | msg_id_info[i].camelcased = q; | |
129 | while (*p) { | |
130 | if (*p == '_') { | |
131 | p++; | |
132 | if (*p) | |
133 | *q++ = *p++; | |
134 | } else { | |
135 | *q++ = tolower(*p++); | |
136 | } | |
137 | } | |
138 | *q = '\0'; | |
139 | } | |
140 | } | |
141 | ||
142 | static int parse_msg_id(const char *text) | |
143 | { | |
144 | int i; | |
145 | ||
146 | prepare_msg_ids(); | |
147 | ||
148 | for (i = 0; i < FSCK_MSG_MAX; i++) | |
149 | if (!strcmp(text, msg_id_info[i].downcased)) | |
150 | return i; | |
151 | ||
152 | return -1; | |
153 | } | |
154 | ||
155 | void list_config_fsck_msg_ids(struct string_list *list, const char *prefix) | |
156 | { | |
157 | int i; | |
158 | ||
159 | prepare_msg_ids(); | |
160 | ||
161 | for (i = 0; i < FSCK_MSG_MAX; i++) | |
162 | list_config_item(list, prefix, msg_id_info[i].camelcased); | |
163 | } | |
164 | ||
165 | static int fsck_msg_type(enum fsck_msg_id msg_id, | |
166 | struct fsck_options *options) | |
167 | { | |
168 | int msg_type; | |
169 | ||
170 | assert(msg_id >= 0 && msg_id < FSCK_MSG_MAX); | |
171 | ||
172 | if (options->msg_type) | |
173 | msg_type = options->msg_type[msg_id]; | |
174 | else { | |
175 | msg_type = msg_id_info[msg_id].msg_type; | |
176 | if (options->strict && msg_type == FSCK_WARN) | |
177 | msg_type = FSCK_ERROR; | |
178 | } | |
179 | ||
180 | return msg_type; | |
181 | } | |
182 | ||
183 | static int parse_msg_type(const char *str) | |
184 | { | |
185 | if (!strcmp(str, "error")) | |
186 | return FSCK_ERROR; | |
187 | else if (!strcmp(str, "warn")) | |
188 | return FSCK_WARN; | |
189 | else if (!strcmp(str, "ignore")) | |
190 | return FSCK_IGNORE; | |
191 | else | |
192 | die("Unknown fsck message type: '%s'", str); | |
193 | } | |
194 | ||
195 | int is_valid_msg_type(const char *msg_id, const char *msg_type) | |
196 | { | |
197 | if (parse_msg_id(msg_id) < 0) | |
198 | return 0; | |
199 | parse_msg_type(msg_type); | |
200 | return 1; | |
201 | } | |
202 | ||
203 | void fsck_set_msg_type(struct fsck_options *options, | |
204 | const char *msg_id, const char *msg_type) | |
205 | { | |
206 | int id = parse_msg_id(msg_id), type; | |
207 | ||
208 | if (id < 0) | |
209 | die("Unhandled message id: %s", msg_id); | |
210 | type = parse_msg_type(msg_type); | |
211 | ||
212 | if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL) | |
213 | die("Cannot demote %s to %s", msg_id, msg_type); | |
214 | ||
215 | if (!options->msg_type) { | |
216 | int i; | |
217 | int *msg_type; | |
218 | ALLOC_ARRAY(msg_type, FSCK_MSG_MAX); | |
219 | for (i = 0; i < FSCK_MSG_MAX; i++) | |
220 | msg_type[i] = fsck_msg_type(i, options); | |
221 | options->msg_type = msg_type; | |
222 | } | |
223 | ||
224 | options->msg_type[id] = type; | |
225 | } | |
226 | ||
227 | void fsck_set_msg_types(struct fsck_options *options, const char *values) | |
228 | { | |
229 | char *buf = xstrdup(values), *to_free = buf; | |
230 | int done = 0; | |
231 | ||
232 | while (!done) { | |
233 | int len = strcspn(buf, " ,|"), equal; | |
234 | ||
235 | done = !buf[len]; | |
236 | if (!len) { | |
237 | buf++; | |
238 | continue; | |
239 | } | |
240 | buf[len] = '\0'; | |
241 | ||
242 | for (equal = 0; | |
243 | equal < len && buf[equal] != '=' && buf[equal] != ':'; | |
244 | equal++) | |
245 | buf[equal] = tolower(buf[equal]); | |
246 | buf[equal] = '\0'; | |
247 | ||
248 | if (!strcmp(buf, "skiplist")) { | |
249 | if (equal == len) | |
250 | die("skiplist requires a path"); | |
251 | oidset_parse_file(&options->skiplist, buf + equal + 1); | |
252 | buf += len + 1; | |
253 | continue; | |
254 | } | |
255 | ||
256 | if (equal == len) | |
257 | die("Missing '=': '%s'", buf); | |
258 | ||
259 | fsck_set_msg_type(options, buf, buf + equal + 1); | |
260 | buf += len + 1; | |
261 | } | |
262 | free(to_free); | |
263 | } | |
264 | ||
265 | static void append_msg_id(struct strbuf *sb, const char *msg_id) | |
266 | { | |
267 | for (;;) { | |
268 | char c = *(msg_id)++; | |
269 | ||
270 | if (!c) | |
271 | break; | |
272 | if (c != '_') | |
273 | strbuf_addch(sb, tolower(c)); | |
274 | else { | |
275 | assert(*msg_id); | |
276 | strbuf_addch(sb, *(msg_id)++); | |
277 | } | |
278 | } | |
279 | ||
280 | strbuf_addstr(sb, ": "); | |
281 | } | |
282 | ||
283 | static int object_on_skiplist(struct fsck_options *opts, | |
284 | const struct object_id *oid) | |
285 | { | |
286 | return opts && oid && oidset_contains(&opts->skiplist, oid); | |
287 | } | |
288 | ||
289 | __attribute__((format (printf, 5, 6))) | |
290 | static int report(struct fsck_options *options, | |
291 | const struct object_id *oid, enum object_type object_type, | |
292 | enum fsck_msg_id id, const char *fmt, ...) | |
293 | { | |
294 | va_list ap; | |
295 | struct strbuf sb = STRBUF_INIT; | |
296 | int msg_type = fsck_msg_type(id, options), result; | |
297 | ||
298 | if (msg_type == FSCK_IGNORE) | |
299 | return 0; | |
300 | ||
301 | if (object_on_skiplist(options, oid)) | |
302 | return 0; | |
303 | ||
304 | if (msg_type == FSCK_FATAL) | |
305 | msg_type = FSCK_ERROR; | |
306 | else if (msg_type == FSCK_INFO) | |
307 | msg_type = FSCK_WARN; | |
308 | ||
309 | append_msg_id(&sb, msg_id_info[id].id_string); | |
310 | ||
311 | va_start(ap, fmt); | |
312 | strbuf_vaddf(&sb, fmt, ap); | |
313 | result = options->error_func(options, oid, object_type, | |
314 | msg_type, sb.buf); | |
315 | strbuf_release(&sb); | |
316 | va_end(ap); | |
317 | ||
318 | return result; | |
319 | } | |
320 | ||
321 | void fsck_enable_object_names(struct fsck_options *options) | |
322 | { | |
323 | if (!options->object_names) | |
324 | options->object_names = kh_init_oid_map(); | |
325 | } | |
326 | ||
327 | const char *fsck_get_object_name(struct fsck_options *options, | |
328 | const struct object_id *oid) | |
329 | { | |
330 | khiter_t pos; | |
331 | if (!options->object_names) | |
332 | return NULL; | |
333 | pos = kh_get_oid_map(options->object_names, *oid); | |
334 | if (pos >= kh_end(options->object_names)) | |
335 | return NULL; | |
336 | return kh_value(options->object_names, pos); | |
337 | } | |
338 | ||
339 | void fsck_put_object_name(struct fsck_options *options, | |
340 | const struct object_id *oid, | |
341 | const char *fmt, ...) | |
342 | { | |
343 | va_list ap; | |
344 | struct strbuf buf = STRBUF_INIT; | |
345 | khiter_t pos; | |
346 | int hashret; | |
347 | ||
348 | if (!options->object_names) | |
349 | return; | |
350 | ||
351 | pos = kh_put_oid_map(options->object_names, *oid, &hashret); | |
352 | if (!hashret) | |
353 | return; | |
354 | va_start(ap, fmt); | |
355 | strbuf_vaddf(&buf, fmt, ap); | |
356 | kh_value(options->object_names, pos) = strbuf_detach(&buf, NULL); | |
357 | va_end(ap); | |
358 | } | |
359 | ||
360 | const char *fsck_describe_object(struct fsck_options *options, | |
361 | const struct object_id *oid) | |
362 | { | |
363 | static struct strbuf bufs[] = { | |
364 | STRBUF_INIT, STRBUF_INIT, STRBUF_INIT, STRBUF_INIT | |
365 | }; | |
366 | static int b = 0; | |
367 | struct strbuf *buf; | |
368 | const char *name = fsck_get_object_name(options, oid); | |
369 | ||
370 | buf = bufs + b; | |
371 | b = (b + 1) % ARRAY_SIZE(bufs); | |
372 | strbuf_reset(buf); | |
373 | strbuf_addstr(buf, oid_to_hex(oid)); | |
374 | if (name) | |
375 | strbuf_addf(buf, " (%s)", name); | |
376 | ||
377 | return buf->buf; | |
378 | } | |
379 | ||
380 | static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options) | |
381 | { | |
382 | struct tree_desc desc; | |
383 | struct name_entry entry; | |
384 | int res = 0; | |
385 | const char *name; | |
386 | ||
387 | if (parse_tree(tree)) | |
388 | return -1; | |
389 | ||
390 | name = fsck_get_object_name(options, &tree->object.oid); | |
391 | if (init_tree_desc_gently(&desc, tree->buffer, tree->size)) | |
392 | return -1; | |
393 | while (tree_entry_gently(&desc, &entry)) { | |
394 | struct object *obj; | |
395 | int result; | |
396 | ||
397 | if (S_ISGITLINK(entry.mode)) | |
398 | continue; | |
399 | ||
400 | if (S_ISDIR(entry.mode)) { | |
401 | obj = (struct object *)lookup_tree(the_repository, &entry.oid); | |
402 | if (name && obj) | |
403 | fsck_put_object_name(options, &entry.oid, "%s%s/", | |
404 | name, entry.path); | |
405 | result = options->walk(obj, OBJ_TREE, data, options); | |
406 | } | |
407 | else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) { | |
408 | obj = (struct object *)lookup_blob(the_repository, &entry.oid); | |
409 | if (name && obj) | |
410 | fsck_put_object_name(options, &entry.oid, "%s%s", | |
411 | name, entry.path); | |
412 | result = options->walk(obj, OBJ_BLOB, data, options); | |
413 | } | |
414 | else { | |
415 | result = error("in tree %s: entry %s has bad mode %.6o", | |
416 | fsck_describe_object(options, &tree->object.oid), | |
417 | entry.path, entry.mode); | |
418 | } | |
419 | if (result < 0) | |
420 | return result; | |
421 | if (!res) | |
422 | res = result; | |
423 | } | |
424 | return res; | |
425 | } | |
426 | ||
427 | static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options) | |
428 | { | |
429 | int counter = 0, generation = 0, name_prefix_len = 0; | |
430 | struct commit_list *parents; | |
431 | int res; | |
432 | int result; | |
433 | const char *name; | |
434 | ||
435 | if (parse_commit(commit)) | |
436 | return -1; | |
437 | ||
438 | name = fsck_get_object_name(options, &commit->object.oid); | |
439 | if (name) | |
440 | fsck_put_object_name(options, get_commit_tree_oid(commit), | |
441 | "%s:", name); | |
442 | ||
443 | result = options->walk((struct object *)get_commit_tree(commit), | |
444 | OBJ_TREE, data, options); | |
445 | if (result < 0) | |
446 | return result; | |
447 | res = result; | |
448 | ||
449 | parents = commit->parents; | |
450 | if (name && parents) { | |
451 | int len = strlen(name), power; | |
452 | ||
453 | if (len && name[len - 1] == '^') { | |
454 | generation = 1; | |
455 | name_prefix_len = len - 1; | |
456 | } | |
457 | else { /* parse ~<generation> suffix */ | |
458 | for (generation = 0, power = 1; | |
459 | len && isdigit(name[len - 1]); | |
460 | power *= 10) | |
461 | generation += power * (name[--len] - '0'); | |
462 | if (power > 1 && len && name[len - 1] == '~') | |
463 | name_prefix_len = len - 1; | |
464 | } | |
465 | } | |
466 | ||
467 | while (parents) { | |
468 | if (name) { | |
469 | struct object_id *oid = &parents->item->object.oid; | |
470 | ||
471 | if (counter++) | |
472 | fsck_put_object_name(options, oid, "%s^%d", | |
473 | name, counter); | |
474 | else if (generation > 0) | |
475 | fsck_put_object_name(options, oid, "%.*s~%d", | |
476 | name_prefix_len, name, | |
477 | generation + 1); | |
478 | else | |
479 | fsck_put_object_name(options, oid, "%s^", name); | |
480 | } | |
481 | result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options); | |
482 | if (result < 0) | |
483 | return result; | |
484 | if (!res) | |
485 | res = result; | |
486 | parents = parents->next; | |
487 | } | |
488 | return res; | |
489 | } | |
490 | ||
491 | static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options) | |
492 | { | |
493 | const char *name = fsck_get_object_name(options, &tag->object.oid); | |
494 | ||
495 | if (parse_tag(tag)) | |
496 | return -1; | |
497 | if (name) | |
498 | fsck_put_object_name(options, &tag->tagged->oid, "%s", name); | |
499 | return options->walk(tag->tagged, OBJ_ANY, data, options); | |
500 | } | |
501 | ||
502 | int fsck_walk(struct object *obj, void *data, struct fsck_options *options) | |
503 | { | |
504 | if (!obj) | |
505 | return -1; | |
506 | ||
507 | if (obj->type == OBJ_NONE) | |
508 | parse_object(the_repository, &obj->oid); | |
509 | ||
510 | switch (obj->type) { | |
511 | case OBJ_BLOB: | |
512 | return 0; | |
513 | case OBJ_TREE: | |
514 | return fsck_walk_tree((struct tree *)obj, data, options); | |
515 | case OBJ_COMMIT: | |
516 | return fsck_walk_commit((struct commit *)obj, data, options); | |
517 | case OBJ_TAG: | |
518 | return fsck_walk_tag((struct tag *)obj, data, options); | |
519 | default: | |
520 | error("Unknown object type for %s", | |
521 | fsck_describe_object(options, &obj->oid)); | |
522 | return -1; | |
523 | } | |
524 | } | |
525 | ||
526 | struct name_stack { | |
527 | const char **names; | |
528 | size_t nr, alloc; | |
529 | }; | |
530 | ||
531 | static void name_stack_push(struct name_stack *stack, const char *name) | |
532 | { | |
533 | ALLOC_GROW(stack->names, stack->nr + 1, stack->alloc); | |
534 | stack->names[stack->nr++] = name; | |
535 | } | |
536 | ||
537 | static const char *name_stack_pop(struct name_stack *stack) | |
538 | { | |
539 | return stack->nr ? stack->names[--stack->nr] : NULL; | |
540 | } | |
541 | ||
542 | static void name_stack_clear(struct name_stack *stack) | |
543 | { | |
544 | FREE_AND_NULL(stack->names); | |
545 | stack->nr = stack->alloc = 0; | |
546 | } | |
547 | ||
548 | /* | |
549 | * The entries in a tree are ordered in the _path_ order, | |
550 | * which means that a directory entry is ordered by adding | |
551 | * a slash to the end of it. | |
552 | * | |
553 | * So a directory called "a" is ordered _after_ a file | |
554 | * called "a.c", because "a/" sorts after "a.c". | |
555 | */ | |
556 | #define TREE_UNORDERED (-1) | |
557 | #define TREE_HAS_DUPS (-2) | |
558 | ||
559 | static int is_less_than_slash(unsigned char c) | |
560 | { | |
561 | return '\0' < c && c < '/'; | |
562 | } | |
563 | ||
564 | static int verify_ordered(unsigned mode1, const char *name1, | |
565 | unsigned mode2, const char *name2, | |
566 | struct name_stack *candidates) | |
567 | { | |
568 | int len1 = strlen(name1); | |
569 | int len2 = strlen(name2); | |
570 | int len = len1 < len2 ? len1 : len2; | |
571 | unsigned char c1, c2; | |
572 | int cmp; | |
573 | ||
574 | cmp = memcmp(name1, name2, len); | |
575 | if (cmp < 0) | |
576 | return 0; | |
577 | if (cmp > 0) | |
578 | return TREE_UNORDERED; | |
579 | ||
580 | /* | |
581 | * Ok, the first <len> characters are the same. | |
582 | * Now we need to order the next one, but turn | |
583 | * a '\0' into a '/' for a directory entry. | |
584 | */ | |
585 | c1 = name1[len]; | |
586 | c2 = name2[len]; | |
587 | if (!c1 && !c2) | |
588 | /* | |
589 | * git-write-tree used to write out a nonsense tree that has | |
590 | * entries with the same name, one blob and one tree. Make | |
591 | * sure we do not have duplicate entries. | |
592 | */ | |
593 | return TREE_HAS_DUPS; | |
594 | if (!c1 && S_ISDIR(mode1)) | |
595 | c1 = '/'; | |
596 | if (!c2 && S_ISDIR(mode2)) | |
597 | c2 = '/'; | |
598 | ||
599 | /* | |
600 | * There can be non-consecutive duplicates due to the implicitly | |
601 | * added slash, e.g.: | |
602 | * | |
603 | * foo | |
604 | * foo.bar | |
605 | * foo.bar.baz | |
606 | * foo.bar/ | |
607 | * foo/ | |
608 | * | |
609 | * Record non-directory candidates (like "foo" and "foo.bar" in | |
610 | * the example) on a stack and check directory candidates (like | |
611 | * foo/" and "foo.bar/") against that stack. | |
612 | */ | |
613 | if (!c1 && is_less_than_slash(c2)) { | |
614 | name_stack_push(candidates, name1); | |
615 | } else if (c2 == '/' && is_less_than_slash(c1)) { | |
616 | for (;;) { | |
617 | const char *p; | |
618 | const char *f_name = name_stack_pop(candidates); | |
619 | ||
620 | if (!f_name) | |
621 | break; | |
622 | if (!skip_prefix(name2, f_name, &p)) | |
623 | continue; | |
624 | if (!*p) | |
625 | return TREE_HAS_DUPS; | |
626 | if (is_less_than_slash(*p)) { | |
627 | name_stack_push(candidates, f_name); | |
628 | break; | |
629 | } | |
630 | } | |
631 | } | |
632 | ||
633 | return c1 < c2 ? 0 : TREE_UNORDERED; | |
634 | } | |
635 | ||
636 | static int fsck_tree(const struct object_id *oid, | |
637 | const char *buffer, unsigned long size, | |
638 | struct fsck_options *options) | |
639 | { | |
640 | int retval = 0; | |
641 | int has_null_sha1 = 0; | |
642 | int has_full_path = 0; | |
643 | int has_empty_name = 0; | |
644 | int has_dot = 0; | |
645 | int has_dotdot = 0; | |
646 | int has_dotgit = 0; | |
647 | int has_zero_pad = 0; | |
648 | int has_bad_modes = 0; | |
649 | int has_dup_entries = 0; | |
650 | int not_properly_sorted = 0; | |
651 | struct tree_desc desc; | |
652 | unsigned o_mode; | |
653 | const char *o_name; | |
654 | struct name_stack df_dup_candidates = { NULL }; | |
655 | ||
656 | if (init_tree_desc_gently(&desc, buffer, size)) { | |
657 | retval += report(options, oid, OBJ_TREE, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree"); | |
658 | return retval; | |
659 | } | |
660 | ||
661 | o_mode = 0; | |
662 | o_name = NULL; | |
663 | ||
664 | while (desc.size) { | |
665 | unsigned short mode; | |
666 | const char *name, *backslash; | |
667 | const struct object_id *oid; | |
668 | ||
669 | oid = tree_entry_extract(&desc, &name, &mode); | |
670 | ||
671 | has_null_sha1 |= is_null_oid(oid); | |
672 | has_full_path |= !!strchr(name, '/'); | |
673 | has_empty_name |= !*name; | |
674 | has_dot |= !strcmp(name, "."); | |
675 | has_dotdot |= !strcmp(name, ".."); | |
676 | has_dotgit |= is_hfs_dotgit(name) || is_ntfs_dotgit(name); | |
677 | has_zero_pad |= *(char *)desc.buffer == '0'; | |
678 | ||
679 | if (is_hfs_dotgitmodules(name) || is_ntfs_dotgitmodules(name)) { | |
680 | if (!S_ISLNK(mode)) | |
681 | oidset_insert(&gitmodules_found, oid); | |
682 | else | |
683 | retval += report(options, | |
684 | oid, OBJ_TREE, | |
685 | FSCK_MSG_GITMODULES_SYMLINK, | |
686 | ".gitmodules is a symbolic link"); | |
687 | } | |
688 | ||
689 | if ((backslash = strchr(name, '\\'))) { | |
690 | while (backslash) { | |
691 | backslash++; | |
692 | has_dotgit |= is_ntfs_dotgit(backslash); | |
693 | if (is_ntfs_dotgitmodules(backslash)) { | |
694 | if (!S_ISLNK(mode)) | |
695 | oidset_insert(&gitmodules_found, oid); | |
696 | else | |
697 | retval += report(options, oid, OBJ_TREE, | |
698 | FSCK_MSG_GITMODULES_SYMLINK, | |
699 | ".gitmodules is a symbolic link"); | |
700 | } | |
701 | backslash = strchr(backslash, '\\'); | |
702 | } | |
703 | } | |
704 | ||
705 | if (update_tree_entry_gently(&desc)) { | |
706 | retval += report(options, oid, OBJ_TREE, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree"); | |
707 | break; | |
708 | } | |
709 | ||
710 | switch (mode) { | |
711 | /* | |
712 | * Standard modes.. | |
713 | */ | |
714 | case S_IFREG | 0755: | |
715 | case S_IFREG | 0644: | |
716 | case S_IFLNK: | |
717 | case S_IFDIR: | |
718 | case S_IFGITLINK: | |
719 | break; | |
720 | /* | |
721 | * This is nonstandard, but we had a few of these | |
722 | * early on when we honored the full set of mode | |
723 | * bits.. | |
724 | */ | |
725 | case S_IFREG | 0664: | |
726 | if (!options->strict) | |
727 | break; | |
728 | /* fallthrough */ | |
729 | default: | |
730 | has_bad_modes = 1; | |
731 | } | |
732 | ||
733 | if (o_name) { | |
734 | switch (verify_ordered(o_mode, o_name, mode, name, | |
735 | &df_dup_candidates)) { | |
736 | case TREE_UNORDERED: | |
737 | not_properly_sorted = 1; | |
738 | break; | |
739 | case TREE_HAS_DUPS: | |
740 | has_dup_entries = 1; | |
741 | break; | |
742 | default: | |
743 | break; | |
744 | } | |
745 | } | |
746 | ||
747 | o_mode = mode; | |
748 | o_name = name; | |
749 | } | |
750 | ||
751 | name_stack_clear(&df_dup_candidates); | |
752 | ||
753 | if (has_null_sha1) | |
754 | retval += report(options, oid, OBJ_TREE, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1"); | |
755 | if (has_full_path) | |
756 | retval += report(options, oid, OBJ_TREE, FSCK_MSG_FULL_PATHNAME, "contains full pathnames"); | |
757 | if (has_empty_name) | |
758 | retval += report(options, oid, OBJ_TREE, FSCK_MSG_EMPTY_NAME, "contains empty pathname"); | |
759 | if (has_dot) | |
760 | retval += report(options, oid, OBJ_TREE, FSCK_MSG_HAS_DOT, "contains '.'"); | |
761 | if (has_dotdot) | |
762 | retval += report(options, oid, OBJ_TREE, FSCK_MSG_HAS_DOTDOT, "contains '..'"); | |
763 | if (has_dotgit) | |
764 | retval += report(options, oid, OBJ_TREE, FSCK_MSG_HAS_DOTGIT, "contains '.git'"); | |
765 | if (has_zero_pad) | |
766 | retval += report(options, oid, OBJ_TREE, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes"); | |
767 | if (has_bad_modes) | |
768 | retval += report(options, oid, OBJ_TREE, FSCK_MSG_BAD_FILEMODE, "contains bad file modes"); | |
769 | if (has_dup_entries) | |
770 | retval += report(options, oid, OBJ_TREE, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries"); | |
771 | if (not_properly_sorted) | |
772 | retval += report(options, oid, OBJ_TREE, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted"); | |
773 | return retval; | |
774 | } | |
775 | ||
776 | static int verify_headers(const void *data, unsigned long size, | |
777 | const struct object_id *oid, enum object_type type, | |
778 | struct fsck_options *options) | |
779 | { | |
780 | const char *buffer = (const char *)data; | |
781 | unsigned long i; | |
782 | ||
783 | for (i = 0; i < size; i++) { | |
784 | switch (buffer[i]) { | |
785 | case '\0': | |
786 | return report(options, oid, type, | |
787 | FSCK_MSG_NUL_IN_HEADER, | |
788 | "unterminated header: NUL at offset %ld", i); | |
789 | case '\n': | |
790 | if (i + 1 < size && buffer[i + 1] == '\n') | |
791 | return 0; | |
792 | } | |
793 | } | |
794 | ||
795 | /* | |
796 | * We did not find double-LF that separates the header | |
797 | * and the body. Not having a body is not a crime but | |
798 | * we do want to see the terminating LF for the last header | |
799 | * line. | |
800 | */ | |
801 | if (size && buffer[size - 1] == '\n') | |
802 | return 0; | |
803 | ||
804 | return report(options, oid, type, | |
805 | FSCK_MSG_UNTERMINATED_HEADER, "unterminated header"); | |
806 | } | |
807 | ||
808 | static int fsck_ident(const char **ident, | |
809 | const struct object_id *oid, enum object_type type, | |
810 | struct fsck_options *options) | |
811 | { | |
812 | const char *p = *ident; | |
813 | char *end; | |
814 | ||
815 | *ident = strchrnul(*ident, '\n'); | |
816 | if (**ident == '\n') | |
817 | (*ident)++; | |
818 | ||
819 | if (*p == '<') | |
820 | return report(options, oid, type, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email"); | |
821 | p += strcspn(p, "<>\n"); | |
822 | if (*p == '>') | |
823 | return report(options, oid, type, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name"); | |
824 | if (*p != '<') | |
825 | return report(options, oid, type, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email"); | |
826 | if (p[-1] != ' ') | |
827 | return report(options, oid, type, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email"); | |
828 | p++; | |
829 | p += strcspn(p, "<>\n"); | |
830 | if (*p != '>') | |
831 | return report(options, oid, type, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email"); | |
832 | p++; | |
833 | if (*p != ' ') | |
834 | return report(options, oid, type, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date"); | |
835 | p++; | |
836 | if (*p == '0' && p[1] != ' ') | |
837 | return report(options, oid, type, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date"); | |
838 | if (date_overflows(parse_timestamp(p, &end, 10))) | |
839 | return report(options, oid, type, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow"); | |
840 | if ((end == p || *end != ' ')) | |
841 | return report(options, oid, type, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date"); | |
842 | p = end + 1; | |
843 | if ((*p != '+' && *p != '-') || | |
844 | !isdigit(p[1]) || | |
845 | !isdigit(p[2]) || | |
846 | !isdigit(p[3]) || | |
847 | !isdigit(p[4]) || | |
848 | (p[5] != '\n')) | |
849 | return report(options, oid, type, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone"); | |
850 | p += 6; | |
851 | return 0; | |
852 | } | |
853 | ||
854 | static int fsck_commit(const struct object_id *oid, | |
855 | const char *buffer, unsigned long size, | |
856 | struct fsck_options *options) | |
857 | { | |
858 | struct object_id tree_oid, parent_oid; | |
859 | unsigned author_count; | |
860 | int err; | |
861 | const char *buffer_begin = buffer; | |
862 | const char *p; | |
863 | ||
864 | if (verify_headers(buffer, size, oid, OBJ_COMMIT, options)) | |
865 | return -1; | |
866 | ||
867 | if (!skip_prefix(buffer, "tree ", &buffer)) | |
868 | return report(options, oid, OBJ_COMMIT, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line"); | |
869 | if (parse_oid_hex(buffer, &tree_oid, &p) || *p != '\n') { | |
870 | err = report(options, oid, OBJ_COMMIT, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1"); | |
871 | if (err) | |
872 | return err; | |
873 | } | |
874 | buffer = p + 1; | |
875 | while (skip_prefix(buffer, "parent ", &buffer)) { | |
876 | if (parse_oid_hex(buffer, &parent_oid, &p) || *p != '\n') { | |
877 | err = report(options, oid, OBJ_COMMIT, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1"); | |
878 | if (err) | |
879 | return err; | |
880 | } | |
881 | buffer = p + 1; | |
882 | } | |
883 | author_count = 0; | |
884 | while (skip_prefix(buffer, "author ", &buffer)) { | |
885 | author_count++; | |
886 | err = fsck_ident(&buffer, oid, OBJ_COMMIT, options); | |
887 | if (err) | |
888 | return err; | |
889 | } | |
890 | if (author_count < 1) | |
891 | err = report(options, oid, OBJ_COMMIT, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line"); | |
892 | else if (author_count > 1) | |
893 | err = report(options, oid, OBJ_COMMIT, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines"); | |
894 | if (err) | |
895 | return err; | |
896 | if (!skip_prefix(buffer, "committer ", &buffer)) | |
897 | return report(options, oid, OBJ_COMMIT, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line"); | |
898 | err = fsck_ident(&buffer, oid, OBJ_COMMIT, options); | |
899 | if (err) | |
900 | return err; | |
901 | if (memchr(buffer_begin, '\0', size)) { | |
902 | err = report(options, oid, OBJ_COMMIT, FSCK_MSG_NUL_IN_COMMIT, | |
903 | "NUL byte in the commit object body"); | |
904 | if (err) | |
905 | return err; | |
906 | } | |
907 | return 0; | |
908 | } | |
909 | ||
910 | static int fsck_tag(const struct object_id *oid, const char *buffer, | |
911 | unsigned long size, struct fsck_options *options) | |
912 | { | |
913 | struct object_id tagged_oid; | |
914 | int ret = 0; | |
915 | char *eol; | |
916 | struct strbuf sb = STRBUF_INIT; | |
917 | const char *p; | |
918 | ||
919 | ret = verify_headers(buffer, size, oid, OBJ_TAG, options); | |
920 | if (ret) | |
921 | goto done; | |
922 | ||
923 | if (!skip_prefix(buffer, "object ", &buffer)) { | |
924 | ret = report(options, oid, OBJ_TAG, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line"); | |
925 | goto done; | |
926 | } | |
927 | if (parse_oid_hex(buffer, &tagged_oid, &p) || *p != '\n') { | |
928 | ret = report(options, oid, OBJ_TAG, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1"); | |
929 | if (ret) | |
930 | goto done; | |
931 | } | |
932 | buffer = p + 1; | |
933 | ||
934 | if (!skip_prefix(buffer, "type ", &buffer)) { | |
935 | ret = report(options, oid, OBJ_TAG, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line"); | |
936 | goto done; | |
937 | } | |
938 | eol = strchr(buffer, '\n'); | |
939 | if (!eol) { | |
940 | ret = report(options, oid, OBJ_TAG, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line"); | |
941 | goto done; | |
942 | } | |
943 | if (type_from_string_gently(buffer, eol - buffer, 1) < 0) | |
944 | ret = report(options, oid, OBJ_TAG, FSCK_MSG_BAD_TYPE, "invalid 'type' value"); | |
945 | if (ret) | |
946 | goto done; | |
947 | buffer = eol + 1; | |
948 | ||
949 | if (!skip_prefix(buffer, "tag ", &buffer)) { | |
950 | ret = report(options, oid, OBJ_TAG, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line"); | |
951 | goto done; | |
952 | } | |
953 | eol = strchr(buffer, '\n'); | |
954 | if (!eol) { | |
955 | ret = report(options, oid, OBJ_TAG, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line"); | |
956 | goto done; | |
957 | } | |
958 | strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer); | |
959 | if (check_refname_format(sb.buf, 0)) { | |
960 | ret = report(options, oid, OBJ_TAG, | |
961 | FSCK_MSG_BAD_TAG_NAME, | |
962 | "invalid 'tag' name: %.*s", | |
963 | (int)(eol - buffer), buffer); | |
964 | if (ret) | |
965 | goto done; | |
966 | } | |
967 | buffer = eol + 1; | |
968 | ||
969 | if (!skip_prefix(buffer, "tagger ", &buffer)) { | |
970 | /* early tags do not contain 'tagger' lines; warn only */ | |
971 | ret = report(options, oid, OBJ_TAG, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line"); | |
972 | if (ret) | |
973 | goto done; | |
974 | } | |
975 | else | |
976 | ret = fsck_ident(&buffer, oid, OBJ_TAG, options); | |
977 | ||
978 | done: | |
979 | strbuf_release(&sb); | |
980 | return ret; | |
981 | } | |
982 | ||
983 | /* | |
984 | * Like builtin/submodule--helper.c's starts_with_dot_slash, but without | |
985 | * relying on the platform-dependent is_dir_sep helper. | |
986 | * | |
987 | * This is for use in checking whether a submodule URL is interpreted as | |
988 | * relative to the current directory on any platform, since \ is a | |
989 | * directory separator on Windows but not on other platforms. | |
990 | */ | |
991 | static int starts_with_dot_slash(const char *str) | |
992 | { | |
993 | return str[0] == '.' && (str[1] == '/' || str[1] == '\\'); | |
994 | } | |
995 | ||
996 | /* | |
997 | * Like starts_with_dot_slash, this is a variant of submodule--helper's | |
998 | * helper of the same name with the twist that it accepts backslash as a | |
999 | * directory separator even on non-Windows platforms. | |
1000 | */ | |
1001 | static int starts_with_dot_dot_slash(const char *str) | |
1002 | { | |
1003 | return str[0] == '.' && starts_with_dot_slash(str + 1); | |
1004 | } | |
1005 | ||
1006 | static int submodule_url_is_relative(const char *url) | |
1007 | { | |
1008 | return starts_with_dot_slash(url) || starts_with_dot_dot_slash(url); | |
1009 | } | |
1010 | ||
1011 | /* | |
1012 | * Count directory components that a relative submodule URL should chop | |
1013 | * from the remote_url it is to be resolved against. | |
1014 | * | |
1015 | * In other words, this counts "../" components at the start of a | |
1016 | * submodule URL. | |
1017 | * | |
1018 | * Returns the number of directory components to chop and writes a | |
1019 | * pointer to the next character of url after all leading "./" and | |
1020 | * "../" components to out. | |
1021 | */ | |
1022 | static int count_leading_dotdots(const char *url, const char **out) | |
1023 | { | |
1024 | int result = 0; | |
1025 | while (1) { | |
1026 | if (starts_with_dot_dot_slash(url)) { | |
1027 | result++; | |
1028 | url += strlen("../"); | |
1029 | continue; | |
1030 | } | |
1031 | if (starts_with_dot_slash(url)) { | |
1032 | url += strlen("./"); | |
1033 | continue; | |
1034 | } | |
1035 | *out = url; | |
1036 | return result; | |
1037 | } | |
1038 | } | |
1039 | /* | |
1040 | * Check whether a transport is implemented by git-remote-curl. | |
1041 | * | |
1042 | * If it is, returns 1 and writes the URL that would be passed to | |
1043 | * git-remote-curl to the "out" parameter. | |
1044 | * | |
1045 | * Otherwise, returns 0 and leaves "out" untouched. | |
1046 | * | |
1047 | * Examples: | |
1048 | * http::https://example.com/repo.git -> 1, https://example.com/repo.git | |
1049 | * https://example.com/repo.git -> 1, https://example.com/repo.git | |
1050 | * git://example.com/repo.git -> 0 | |
1051 | * | |
1052 | * This is for use in checking for previously exploitable bugs that | |
1053 | * required a submodule URL to be passed to git-remote-curl. | |
1054 | */ | |
1055 | static int url_to_curl_url(const char *url, const char **out) | |
1056 | { | |
1057 | /* | |
1058 | * We don't need to check for case-aliases, "http.exe", and so | |
1059 | * on because in the default configuration, is_transport_allowed | |
1060 | * prevents URLs with those schemes from being cloned | |
1061 | * automatically. | |
1062 | */ | |
1063 | if (skip_prefix(url, "http::", out) || | |
1064 | skip_prefix(url, "https::", out) || | |
1065 | skip_prefix(url, "ftp::", out) || | |
1066 | skip_prefix(url, "ftps::", out)) | |
1067 | return 1; | |
1068 | if (starts_with(url, "http://") || | |
1069 | starts_with(url, "https://") || | |
1070 | starts_with(url, "ftp://") || | |
1071 | starts_with(url, "ftps://")) { | |
1072 | *out = url; | |
1073 | return 1; | |
1074 | } | |
1075 | return 0; | |
1076 | } | |
1077 | ||
1078 | static int check_submodule_url(const char *url) | |
1079 | { | |
1080 | const char *curl_url; | |
1081 | ||
1082 | if (looks_like_command_line_option(url)) | |
1083 | return -1; | |
1084 | ||
1085 | if (submodule_url_is_relative(url)) { | |
1086 | char *decoded; | |
1087 | const char *next; | |
1088 | int has_nl; | |
1089 | ||
1090 | /* | |
1091 | * This could be appended to an http URL and url-decoded; | |
1092 | * check for malicious characters. | |
1093 | */ | |
1094 | decoded = url_decode(url); | |
1095 | has_nl = !!strchr(decoded, '\n'); | |
1096 | ||
1097 | free(decoded); | |
1098 | if (has_nl) | |
1099 | return -1; | |
1100 | ||
1101 | /* | |
1102 | * URLs which escape their root via "../" can overwrite | |
1103 | * the host field and previous components, resolving to | |
1104 | * URLs like https::example.com/submodule.git and | |
1105 | * https:///example.com/submodule.git that were | |
1106 | * susceptible to CVE-2020-11008. | |
1107 | */ | |
1108 | if (count_leading_dotdots(url, &next) > 0 && | |
1109 | (*next == ':' || *next == '/')) | |
1110 | return -1; | |
1111 | } | |
1112 | ||
1113 | else if (url_to_curl_url(url, &curl_url)) { | |
1114 | struct credential c = CREDENTIAL_INIT; | |
1115 | int ret = 0; | |
1116 | if (credential_from_url_gently(&c, curl_url, 1) || | |
1117 | !*c.host) | |
1118 | ret = -1; | |
1119 | credential_clear(&c); | |
1120 | return ret; | |
1121 | } | |
1122 | ||
1123 | return 0; | |
1124 | } | |
1125 | ||
1126 | struct fsck_gitmodules_data { | |
1127 | const struct object_id *oid; | |
1128 | struct fsck_options *options; | |
1129 | int ret; | |
1130 | }; | |
1131 | ||
1132 | static int fsck_gitmodules_fn(const char *var, const char *value, void *vdata) | |
1133 | { | |
1134 | struct fsck_gitmodules_data *data = vdata; | |
1135 | const char *subsection, *key; | |
1136 | size_t subsection_len; | |
1137 | char *name; | |
1138 | ||
1139 | if (parse_config_key(var, "submodule", &subsection, &subsection_len, &key) < 0 || | |
1140 | !subsection) | |
1141 | return 0; | |
1142 | ||
1143 | name = xmemdupz(subsection, subsection_len); | |
1144 | if (check_submodule_name(name) < 0) | |
1145 | data->ret |= report(data->options, | |
1146 | data->oid, OBJ_BLOB, | |
1147 | FSCK_MSG_GITMODULES_NAME, | |
1148 | "disallowed submodule name: %s", | |
1149 | name); | |
1150 | if (!strcmp(key, "url") && value && | |
1151 | check_submodule_url(value) < 0) | |
1152 | data->ret |= report(data->options, | |
1153 | data->oid, OBJ_BLOB, | |
1154 | FSCK_MSG_GITMODULES_URL, | |
1155 | "disallowed submodule url: %s", | |
1156 | value); | |
1157 | if (!strcmp(key, "path") && value && | |
1158 | looks_like_command_line_option(value)) | |
1159 | data->ret |= report(data->options, | |
1160 | data->oid, OBJ_BLOB, | |
1161 | FSCK_MSG_GITMODULES_PATH, | |
1162 | "disallowed submodule path: %s", | |
1163 | value); | |
1164 | if (!strcmp(key, "update") && value && | |
1165 | parse_submodule_update_type(value) == SM_UPDATE_COMMAND) | |
1166 | data->ret |= report(data->options, data->oid, OBJ_BLOB, | |
1167 | FSCK_MSG_GITMODULES_UPDATE, | |
1168 | "disallowed submodule update setting: %s", | |
1169 | value); | |
1170 | free(name); | |
1171 | ||
1172 | return 0; | |
1173 | } | |
1174 | ||
1175 | static int fsck_blob(const struct object_id *oid, const char *buf, | |
1176 | unsigned long size, struct fsck_options *options) | |
1177 | { | |
1178 | struct fsck_gitmodules_data data; | |
1179 | struct config_options config_opts = { 0 }; | |
1180 | ||
1181 | if (!oidset_contains(&gitmodules_found, oid)) | |
1182 | return 0; | |
1183 | oidset_insert(&gitmodules_done, oid); | |
1184 | ||
1185 | if (object_on_skiplist(options, oid)) | |
1186 | return 0; | |
1187 | ||
1188 | if (!buf) { | |
1189 | /* | |
1190 | * A missing buffer here is a sign that the caller found the | |
1191 | * blob too gigantic to load into memory. Let's just consider | |
1192 | * that an error. | |
1193 | */ | |
1194 | return report(options, oid, OBJ_BLOB, | |
1195 | FSCK_MSG_GITMODULES_LARGE, | |
1196 | ".gitmodules too large to parse"); | |
1197 | } | |
1198 | ||
1199 | data.oid = oid; | |
1200 | data.options = options; | |
1201 | data.ret = 0; | |
1202 | config_opts.error_action = CONFIG_ERROR_SILENT; | |
1203 | if (git_config_from_mem(fsck_gitmodules_fn, CONFIG_ORIGIN_BLOB, | |
1204 | ".gitmodules", buf, size, &data, &config_opts)) | |
1205 | data.ret |= report(options, oid, OBJ_BLOB, | |
1206 | FSCK_MSG_GITMODULES_PARSE, | |
1207 | "could not parse gitmodules blob"); | |
1208 | ||
1209 | return data.ret; | |
1210 | } | |
1211 | ||
1212 | int fsck_object(struct object *obj, void *data, unsigned long size, | |
1213 | struct fsck_options *options) | |
1214 | { | |
1215 | if (!obj) | |
1216 | return report(options, NULL, OBJ_NONE, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck"); | |
1217 | ||
1218 | if (obj->type == OBJ_BLOB) | |
1219 | return fsck_blob(&obj->oid, data, size, options); | |
1220 | if (obj->type == OBJ_TREE) | |
1221 | return fsck_tree(&obj->oid, data, size, options); | |
1222 | if (obj->type == OBJ_COMMIT) | |
1223 | return fsck_commit(&obj->oid, data, size, options); | |
1224 | if (obj->type == OBJ_TAG) | |
1225 | return fsck_tag(&obj->oid, data, size, options); | |
1226 | ||
1227 | return report(options, &obj->oid, obj->type, | |
1228 | FSCK_MSG_UNKNOWN_TYPE, | |
1229 | "unknown type '%d' (internal fsck error)", | |
1230 | obj->type); | |
1231 | } | |
1232 | ||
1233 | int fsck_error_function(struct fsck_options *o, | |
1234 | const struct object_id *oid, | |
1235 | enum object_type object_type, | |
1236 | int msg_type, const char *message) | |
1237 | { | |
1238 | if (msg_type == FSCK_WARN) { | |
1239 | warning("object %s: %s", fsck_describe_object(o, oid), message); | |
1240 | return 0; | |
1241 | } | |
1242 | error("object %s: %s", fsck_describe_object(o, oid), message); | |
1243 | return 1; | |
1244 | } | |
1245 | ||
1246 | int fsck_finish(struct fsck_options *options) | |
1247 | { | |
1248 | int ret = 0; | |
1249 | struct oidset_iter iter; | |
1250 | const struct object_id *oid; | |
1251 | ||
1252 | oidset_iter_init(&gitmodules_found, &iter); | |
1253 | while ((oid = oidset_iter_next(&iter))) { | |
1254 | enum object_type type; | |
1255 | unsigned long size; | |
1256 | char *buf; | |
1257 | ||
1258 | if (oidset_contains(&gitmodules_done, oid)) | |
1259 | continue; | |
1260 | ||
1261 | buf = read_object_file(oid, &type, &size); | |
1262 | if (!buf) { | |
1263 | if (is_promisor_object(oid)) | |
1264 | continue; | |
1265 | ret |= report(options, | |
1266 | oid, OBJ_BLOB, | |
1267 | FSCK_MSG_GITMODULES_MISSING, | |
1268 | "unable to read .gitmodules blob"); | |
1269 | continue; | |
1270 | } | |
1271 | ||
1272 | if (type == OBJ_BLOB) | |
1273 | ret |= fsck_blob(oid, buf, size, options); | |
1274 | else | |
1275 | ret |= report(options, | |
1276 | oid, type, | |
1277 | FSCK_MSG_GITMODULES_BLOB, | |
1278 | "non-blob found at .gitmodules"); | |
1279 | free(buf); | |
1280 | } | |
1281 | ||
1282 | ||
1283 | oidset_clear(&gitmodules_found); | |
1284 | oidset_clear(&gitmodules_done); | |
1285 | return ret; | |
1286 | } |