Better error messages for corrupt databases
[git/git.git] / fsck-objects.c
CommitLineData
4b182421 1#include "cache.h"
ff5ebe39
DB
2#include "commit.h"
3#include "tree.h"
4#include "blob.h"
c418eda4 5#include "tag.h"
944d8589 6#include "refs.h"
f9253394 7#include "pack.h"
53dc3f3e 8#include "cache-tree.h"
e9a95bef 9#include "tree-walk.h"
ff5ebe39
DB
10
11#define REACHABLE 0x0001
2d9c58c6 12#define SEEN 0x0002
d9839e03 13
96f1e58f
DR
14static int show_root;
15static int show_tags;
16static int show_unreachable;
17static int check_full;
18static int check_strict;
19static int keep_cache_objects;
d9839e03
LT
20static unsigned char head_sha1[20];
21
962554c6 22#ifdef NO_D_INO_IN_DIRENT
35a730f0
JH
23#define SORT_DIRENT 0
24#define DIRENT_SORT_HINT(de) 0
25#else
26#define SORT_DIRENT 1
27#define DIRENT_SORT_HINT(de) ((de)->d_ino)
28#endif
f1f0d088
PB
29
30static void objreport(struct object *obj, const char *severity,
31 const char *err, va_list params)
32{
33 fprintf(stderr, "%s in %s %s: ",
885a86ab 34 severity, typename(obj->type), sha1_to_hex(obj->sha1));
f1f0d088
PB
35 vfprintf(stderr, err, params);
36 fputs("\n", stderr);
37}
38
a7928f8e 39static int objerror(struct object *obj, const char *err, ...)
f1f0d088
PB
40{
41 va_list params;
42 va_start(params, err);
43 objreport(obj, "error", err, params);
44 va_end(params);
45 return -1;
46}
47
a7928f8e 48static int objwarning(struct object *obj, const char *err, ...)
f1f0d088
PB
49{
50 va_list params;
51 va_start(params, err);
52 objreport(obj, "warning", err, params);
53 va_end(params);
54 return -1;
55}
56
57
8ba0bbb2
LT
58static void check_connectivity(void)
59{
fc046a75 60 int i, max;
8ba0bbb2 61
8ba0bbb2 62 /* Look up all the requirements, warn about missing objects.. */
fc046a75
LT
63 max = get_max_object_index();
64 for (i = 0; i < max; i++) {
3e4339e6 65 const struct object_refs *refs;
fc046a75 66 struct object *obj = get_indexed_object(i);
8ba0bbb2 67
070879ca
JS
68 if (!obj)
69 continue;
70
3a6a23e6 71 if (!obj->parsed) {
7aaa715d 72 if (has_sha1_file(obj->sha1))
8a498a05
JH
73 ; /* it is in pack */
74 else
75 printf("missing %s %s\n",
885a86ab 76 typename(obj->type), sha1_to_hex(obj->sha1));
3a6a23e6
LT
77 continue;
78 }
79
3e4339e6
LT
80 refs = lookup_object_refs(obj);
81 if (refs) {
4a4e6fd7
SV
82 unsigned j;
83 for (j = 0; j < refs->count; j++) {
84 struct object *ref = refs->ref[j];
85 if (ref->parsed ||
7aaa715d 86 (has_sha1_file(ref->sha1)))
4a4e6fd7
SV
87 continue;
88 printf("broken link from %7s %s\n",
885a86ab 89 typename(obj->type), sha1_to_hex(obj->sha1));
4a4e6fd7 90 printf(" to %7s %s\n",
885a86ab 91 typename(ref->type), sha1_to_hex(ref->sha1));
4a4e6fd7 92 }
3a6a23e6
LT
93 }
94
ff5ebe39 95 if (show_unreachable && !(obj->flags & REACHABLE)) {
c4584ae3 96 printf("unreachable %s %s\n",
885a86ab 97 typename(obj->type), sha1_to_hex(obj->sha1));
8ba0bbb2 98 continue;
d9839e03 99 }
8ba0bbb2 100
ff5ebe39 101 if (!obj->used) {
885a86ab 102 printf("dangling %s %s\n", typename(obj->type),
ff5ebe39 103 sha1_to_hex(obj->sha1));
d9839e03 104 }
8ba0bbb2
LT
105 }
106}
107
85003492
LT
108/*
109 * The entries in a tree are ordered in the _path_ order,
110 * which means that a directory entry is ordered by adding
111 * a slash to the end of it.
112 *
113 * So a directory called "a" is ordered _after_ a file
114 * called "a.c", because "a/" sorts after "a.c".
115 */
a4f35a2d
JH
116#define TREE_UNORDERED (-1)
117#define TREE_HAS_DUPS (-2)
118
e9a95bef 119static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
85003492 120{
e9a95bef
LT
121 int len1 = strlen(name1);
122 int len2 = strlen(name2);
85003492
LT
123 int len = len1 < len2 ? len1 : len2;
124 unsigned char c1, c2;
125 int cmp;
126
e9a95bef 127 cmp = memcmp(name1, name2, len);
85003492
LT
128 if (cmp < 0)
129 return 0;
130 if (cmp > 0)
a4f35a2d 131 return TREE_UNORDERED;
85003492
LT
132
133 /*
134 * Ok, the first <len> characters are the same.
135 * Now we need to order the next one, but turn
136 * a '\0' into a '/' for a directory entry.
137 */
e9a95bef
LT
138 c1 = name1[len];
139 c2 = name2[len];
a4f35a2d
JH
140 if (!c1 && !c2)
141 /*
142 * git-write-tree used to write out a nonsense tree that has
143 * entries with the same name, one blob and one tree. Make
144 * sure we do not have duplicate entries.
145 */
146 return TREE_HAS_DUPS;
e9a95bef 147 if (!c1 && S_ISDIR(mode1))
85003492 148 c1 = '/';
e9a95bef 149 if (!c2 && S_ISDIR(mode2))
85003492 150 c2 = '/';
a4f35a2d 151 return c1 < c2 ? 0 : TREE_UNORDERED;
85003492
LT
152}
153
c418eda4 154static int fsck_tree(struct tree *item)
20222118 155{
64071805 156 int retval;
85003492 157 int has_full_path = 0;
64071805
LT
158 int has_zero_pad = 0;
159 int has_bad_modes = 0;
160 int has_dup_entries = 0;
161 int not_properly_sorted = 0;
e9a95bef
LT
162 struct tree_desc desc;
163 unsigned o_mode;
164 const char *o_name;
165 const unsigned char *o_sha1;
85003492 166
e9a95bef
LT
167 desc.buf = item->buffer;
168 desc.size = item->size;
169
170 o_mode = 0;
171 o_name = NULL;
172 o_sha1 = NULL;
173 while (desc.size) {
174 unsigned mode;
175 const char *name;
176 const unsigned char *sha1;
177
178 sha1 = tree_entry_extract(&desc, &name, &mode);
179
180 if (strchr(name, '/'))
85003492 181 has_full_path = 1;
e9a95bef
LT
182 has_zero_pad |= *(char *)desc.buf == '0';
183 update_tree_entry(&desc);
85003492 184
e9a95bef 185 switch (mode) {
42ea9cb2 186 /*
e9a95bef 187 * Standard modes..
42ea9cb2
LT
188 */
189 case S_IFREG | 0755:
190 case S_IFREG | 0644:
191 case S_IFLNK:
192 case S_IFDIR:
193 break;
194 /*
195 * This is nonstandard, but we had a few of these
196 * early on when we honored the full set of mode
197 * bits..
198 */
199 case S_IFREG | 0664:
de2eb7f6
LT
200 if (!check_strict)
201 break;
42ea9cb2 202 default:
64071805 203 has_bad_modes = 1;
42ea9cb2
LT
204 }
205
e9a95bef
LT
206 if (o_name) {
207 switch (verify_ordered(o_mode, o_name, mode, name)) {
a4f35a2d 208 case TREE_UNORDERED:
64071805
LT
209 not_properly_sorted = 1;
210 break;
a4f35a2d 211 case TREE_HAS_DUPS:
64071805
LT
212 has_dup_entries = 1;
213 break;
a4f35a2d
JH
214 default:
215 break;
85003492
LT
216 }
217 }
218
e9a95bef
LT
219 o_mode = mode;
220 o_name = name;
221 o_sha1 = sha1;
85003492 222 }
136f2e54
LT
223 free(item->buffer);
224 item->buffer = NULL;
85003492 225
64071805 226 retval = 0;
85003492 227 if (has_full_path) {
f1f0d088 228 objwarning(&item->object, "contains full pathnames");
1ea34e36 229 }
64071805 230 if (has_zero_pad) {
f1f0d088 231 objwarning(&item->object, "contains zero-padded file modes");
64071805
LT
232 }
233 if (has_bad_modes) {
f1f0d088 234 objwarning(&item->object, "contains bad file modes");
64071805
LT
235 }
236 if (has_dup_entries) {
f1f0d088 237 retval = objerror(&item->object, "contains duplicate file entries");
64071805
LT
238 }
239 if (not_properly_sorted) {
f1f0d088 240 retval = objerror(&item->object, "not properly sorted");
64071805
LT
241 }
242 return retval;
1ea34e36
LT
243}
244
c418eda4 245static int fsck_commit(struct commit *commit)
1ea34e36 246{
de2eb7f6 247 char *buffer = commit->buffer;
f1f0d088 248 unsigned char tree_sha1[20], sha1[20];
de2eb7f6
LT
249
250 if (memcmp(buffer, "tree ", 5))
f1f0d088
PB
251 return objerror(&commit->object, "invalid format - expected 'tree' line");
252 if (get_sha1_hex(buffer+5, tree_sha1) || buffer[45] != '\n')
253 return objerror(&commit->object, "invalid 'tree' line format - bad sha1");
de2eb7f6
LT
254 buffer += 46;
255 while (!memcmp(buffer, "parent ", 7)) {
256 if (get_sha1_hex(buffer+7, sha1) || buffer[47] != '\n')
f1f0d088 257 return objerror(&commit->object, "invalid 'parent' line format - bad sha1");
de2eb7f6
LT
258 buffer += 48;
259 }
260 if (memcmp(buffer, "author ", 7))
f1f0d088 261 return objerror(&commit->object, "invalid format - expected 'author' line");
bd1e17e2
LT
262 free(commit->buffer);
263 commit->buffer = NULL;
ff5ebe39 264 if (!commit->tree)
f1f0d088 265 return objerror(&commit->object, "could not load commit's tree %s", tree_sha1);
ab7df187 266 if (!commit->parents && show_root)
c418eda4 267 printf("root %s\n", sha1_to_hex(commit->object.sha1));
e6948b6d 268 if (!commit->date)
c418eda4
DB
269 printf("bad commit date in %s\n",
270 sha1_to_hex(commit->object.sha1));
4728b861
LT
271 return 0;
272}
273
c418eda4 274static int fsck_tag(struct tag *tag)
ec4465ad 275{
92d4c85d
LT
276 struct object *tagged = tag->tagged;
277
278 if (!tagged) {
f1f0d088 279 return objerror(&tag->object, "could not load tagged object");
92d4c85d 280 }
889262ea
LT
281 if (!show_tags)
282 return 0;
283
885a86ab 284 printf("tagged %s %s", typename(tagged->type), sha1_to_hex(tagged->sha1));
92d4c85d 285 printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
ff5ebe39 286 return 0;
20222118
LT
287}
288
7e8c174a 289static int fsck_sha1(unsigned char *sha1)
20222118 290{
7e8c174a
LT
291 struct object *obj = parse_object(sha1);
292 if (!obj)
9130ac1e 293 return error("%s: object corrupt or missing", sha1_to_hex(sha1));
2d9c58c6
LT
294 if (obj->flags & SEEN)
295 return 0;
296 obj->flags |= SEEN;
1974632c 297 if (obj->type == OBJ_BLOB)
7e8c174a 298 return 0;
1974632c 299 if (obj->type == OBJ_TREE)
7e8c174a 300 return fsck_tree((struct tree *) obj);
1974632c 301 if (obj->type == OBJ_COMMIT)
7e8c174a 302 return fsck_commit((struct commit *) obj);
1974632c 303 if (obj->type == OBJ_TAG)
7e8c174a 304 return fsck_tag((struct tag *) obj);
f1f0d088 305 /* By now, parse_object() would've returned NULL instead. */
885a86ab 306 return objerror(obj, "unknown type '%d' (internal fsck error)", obj->type);
7e8c174a
LT
307}
308
309/*
310 * This is the sorting chunk size: make it reasonably
311 * big so that we can sort well..
312 */
313#define MAX_SHA1_ENTRIES (1024)
314
315struct sha1_entry {
316 unsigned long ino;
20222118 317 unsigned char sha1[20];
7e8c174a
LT
318};
319
320static struct {
321 unsigned long nr;
322 struct sha1_entry *entry[MAX_SHA1_ENTRIES];
323} sha1_list;
324
325static int ino_compare(const void *_a, const void *_b)
326{
327 const struct sha1_entry *a = _a, *b = _b;
328 unsigned long ino1 = a->ino, ino2 = b->ino;
329 return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0;
330}
331
332static void fsck_sha1_list(void)
333{
334 int i, nr = sha1_list.nr;
335
35a730f0
JH
336 if (SORT_DIRENT)
337 qsort(sha1_list.entry, nr,
338 sizeof(struct sha1_entry *), ino_compare);
7e8c174a
LT
339 for (i = 0; i < nr; i++) {
340 struct sha1_entry *entry = sha1_list.entry[i];
341 unsigned char *sha1 = entry->sha1;
342
343 sha1_list.entry[i] = NULL;
f1f0d088 344 fsck_sha1(sha1);
7e8c174a 345 free(entry);
20222118 346 }
7e8c174a
LT
347 sha1_list.nr = 0;
348}
349
350static void add_sha1_list(unsigned char *sha1, unsigned long ino)
351{
352 struct sha1_entry *entry = xmalloc(sizeof(*entry));
353 int nr;
354
355 entry->ino = ino;
e702496e 356 hashcpy(entry->sha1, sha1);
7e8c174a
LT
357 nr = sha1_list.nr;
358 if (nr == MAX_SHA1_ENTRIES) {
359 fsck_sha1_list();
360 nr = 0;
361 }
362 sha1_list.entry[nr] = entry;
363 sha1_list.nr = ++nr;
20222118
LT
364}
365
b5524c82 366static void fsck_dir(int i, char *path)
20222118
LT
367{
368 DIR *dir = opendir(path);
369 struct dirent *de;
370
230f1322 371 if (!dir)
b5524c82 372 return;
20222118
LT
373
374 while ((de = readdir(dir)) != NULL) {
375 char name[100];
7e8c174a 376 unsigned char sha1[20];
20222118
LT
377 int len = strlen(de->d_name);
378
379 switch (len) {
380 case 2:
381 if (de->d_name[1] != '.')
382 break;
383 case 1:
384 if (de->d_name[0] != '.')
385 break;
386 continue;
387 case 38:
388 sprintf(name, "%02x", i);
389 memcpy(name+2, de->d_name, len+1);
7e8c174a
LT
390 if (get_sha1_hex(name, sha1) < 0)
391 break;
35a730f0 392 add_sha1_list(sha1, DIRENT_SORT_HINT(de));
7e8c174a 393 continue;
20222118
LT
394 }
395 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
396 }
397 closedir(dir);
20222118
LT
398}
399
96f1e58f 400static int default_refs;
944d8589 401
883d60fa
JS
402static int fsck_handle_reflog_ent(unsigned char *osha1, unsigned char *nsha1,
403 const char *email, unsigned long timestamp, int tz,
404 const char *message, void *cb_data)
55dd5526
JH
405{
406 struct object *obj;
407
408 if (!is_null_sha1(osha1)) {
409 obj = lookup_object(osha1);
410 if (obj) {
411 obj->used = 1;
412 mark_reachable(obj, REACHABLE);
413 }
414 }
415 obj = lookup_object(nsha1);
416 if (obj) {
417 obj->used = 1;
418 mark_reachable(obj, REACHABLE);
419 }
420 return 0;
421}
422
8da19775 423static int fsck_handle_ref(const char *refname, const unsigned char *sha1, int flag, void *cb_data)
1024932f 424{
1024932f
LT
425 struct object *obj;
426
1024932f 427 obj = lookup_object(sha1);
8a498a05 428 if (!obj) {
7aaa715d 429 if (has_sha1_file(sha1)) {
659cacf5 430 default_refs++;
944d8589 431 return 0; /* it is in a pack */
659cacf5 432 }
944d8589
LT
433 error("%s: invalid sha1 pointer %s", refname, sha1_to_hex(sha1));
434 /* We'll continue with the rest despite the error.. */
435 return 0;
8a498a05 436 }
944d8589 437 default_refs++;
1024932f
LT
438 obj->used = 1;
439 mark_reachable(obj, REACHABLE);
55dd5526
JH
440
441 for_each_reflog_ent(refname, fsck_handle_reflog_ent, NULL);
442
7c4d07c7 443 return 0;
1024932f
LT
444}
445
1024932f
LT
446static void get_default_heads(void)
447{
cb5d709f 448 for_each_ref(fsck_handle_ref, NULL);
071fa89e
LT
449
450 /*
451 * Not having any default heads isn't really fatal, but
452 * it does mean that "--unreachable" no longer makes any
453 * sense (since in this case everything will obviously
454 * be unreachable by definition.
455 *
456 * Showing dangling objects is valid, though (as those
457 * dangling objects are likely lost heads).
458 *
459 * So we just print a warning about it, and clear the
460 * "show_unreachable" flag.
461 */
462 if (!default_refs) {
463 error("No default references");
464 show_unreachable = 0;
465 }
1024932f
LT
466}
467
8a498a05
JH
468static void fsck_object_dir(const char *path)
469{
470 int i;
471 for (i = 0; i < 256; i++) {
472 static char dir[4096];
473 sprintf(dir, "%s/%02x", path, i);
474 fsck_dir(i, dir);
475 }
476 fsck_sha1_list();
477}
478
c3330383
LT
479static int fsck_head_link(void)
480{
c3330383 481 unsigned char sha1[20];
8da19775
JH
482 int flag;
483 const char *head_points_at = resolve_ref("HEAD", sha1, 1, &flag);
8098a178 484
8da19775 485 if (!head_points_at || !(flag & REF_ISSYMREF))
8098a178 486 return error("HEAD is not a symbolic ref");
5b10b091 487 if (strncmp(head_points_at, "refs/heads/", 11))
8098a178 488 return error("HEAD points to something strange (%s)",
5b10b091 489 head_points_at);
0bef57ee 490 if (is_null_sha1(sha1))
c3330383
LT
491 return error("HEAD: not a valid git pointer");
492 return 0;
493}
494
53dc3f3e
JH
495static int fsck_cache_tree(struct cache_tree *it)
496{
497 int i;
498 int err = 0;
499
500 if (0 <= it->entry_count) {
501 struct object *obj = parse_object(it->sha1);
6d60bbef
JH
502 if (!obj) {
503 error("%s: invalid sha1 pointer in cache-tree",
504 sha1_to_hex(it->sha1));
505 return 1;
506 }
cdc08b33
JH
507 mark_reachable(obj, REACHABLE);
508 obj->used = 1;
1974632c 509 if (obj->type != OBJ_TREE)
53dc3f3e
JH
510 err |= objerror(obj, "non-tree in cache-tree");
511 }
512 for (i = 0; i < it->subtree_nr; i++)
513 err |= fsck_cache_tree(it->down[i]->cache_tree);
514 return err;
515}
516
20222118
LT
517int main(int argc, char **argv)
518{
bcee6fd8 519 int i, heads;
20222118 520
3a7c352b 521 track_object_refs = 1;
61e2b015
JH
522 setup_git_directory();
523
889262ea
LT
524 for (i = 1; i < argc; i++) {
525 const char *arg = argv[i];
526
527 if (!strcmp(arg, "--unreachable")) {
528 show_unreachable = 1;
529 continue;
530 }
531 if (!strcmp(arg, "--tags")) {
532 show_tags = 1;
533 continue;
534 }
ab7df187
LT
535 if (!strcmp(arg, "--root")) {
536 show_root = 1;
537 continue;
538 }
ae7c0c92
JH
539 if (!strcmp(arg, "--cache")) {
540 keep_cache_objects = 1;
541 continue;
542 }
8a498a05
JH
543 if (!strcmp(arg, "--full")) {
544 check_full = 1;
545 continue;
546 }
de2eb7f6
LT
547 if (!strcmp(arg, "--strict")) {
548 check_strict = 1;
549 continue;
550 }
889262ea 551 if (*arg == '-')
7aaa715d 552 usage("git-fsck-objects [--tags] [--root] [[--unreachable] [--cache] [--full] [--strict] <head-sha1>*]");
889262ea
LT
553 }
554
c3330383 555 fsck_head_link();
8a498a05
JH
556 fsck_object_dir(get_object_directory());
557 if (check_full) {
d5a63b99 558 struct alternate_object_database *alt;
8a498a05
JH
559 struct packed_git *p;
560 prepare_alt_odb();
d5a63b99 561 for (alt = alt_odb_list; alt; alt = alt->next) {
a3eb250f 562 char namebuf[PATH_MAX];
d5a63b99
JH
563 int namelen = alt->name - alt->base;
564 memcpy(namebuf, alt->base, namelen);
a3eb250f
JH
565 namebuf[namelen - 1] = 0;
566 fsck_object_dir(namebuf);
8a498a05
JH
567 }
568 prepare_packed_git();
f9253394
JH
569 for (p = packed_git; p; p = p->next)
570 /* verify gives error messages itself */
f3bf9224 571 verify_pack(p, 0);
f9253394 572
8a498a05
JH
573 for (p = packed_git; p; p = p->next) {
574 int num = num_packed_objects(p);
575 for (i = 0; i < num; i++) {
576 unsigned char sha1[20];
577 nth_packed_object_sha1(p, i, sha1);
f1f0d088 578 fsck_sha1(sha1);
8a498a05
JH
579 }
580 }
bcee6fd8
LT
581 }
582
583 heads = 0;
d9839e03 584 for (i = 1; i < argc; i++) {
889262ea
LT
585 const char *arg = argv[i];
586
587 if (*arg == '-')
d9839e03 588 continue;
889262ea 589
3c249c95 590 if (!get_sha1(arg, head_sha1)) {
770896e5 591 struct object *obj = lookup_object(head_sha1);
e1a1388d 592
770896e5
LT
593 /* Error is printed by lookup_object(). */
594 if (!obj)
e1a1388d
JF
595 continue;
596
ff5ebe39
DB
597 obj->used = 1;
598 mark_reachable(obj, REACHABLE);
bcee6fd8 599 heads++;
d9839e03
LT
600 continue;
601 }
f1f0d088 602 error("invalid parameter: expected sha1, got '%s'", arg);
d9839e03 603 }
d9839e03 604
1024932f 605 /*
d1af002d 606 * If we've not been given any explicit head information, do the
e7bd907d
LT
607 * default ones from .git/refs. We also consider the index file
608 * in this case (ie this implies --cache).
1024932f 609 */
e7bd907d 610 if (!heads) {
1024932f
LT
611 get_default_heads();
612 keep_cache_objects = 1;
613 }
614
ae7c0c92
JH
615 if (keep_cache_objects) {
616 int i;
617 read_cache();
618 for (i = 0; i < active_nr; i++) {
619 struct blob *blob = lookup_blob(active_cache[i]->sha1);
620 struct object *obj;
621 if (!blob)
622 continue;
623 obj = &blob->object;
624 obj->used = 1;
625 mark_reachable(obj, REACHABLE);
626 }
53dc3f3e
JH
627 if (active_cache_tree)
628 fsck_cache_tree(active_cache_tree);
ae7c0c92
JH
629 }
630
8ba0bbb2 631 check_connectivity();
20222118
LT
632 return 0;
633}