]> git.ipfire.org Git - thirdparty/git.git/blob - builtin/fsck.c
fsck: parse loose object paths directly
[thirdparty/git.git] / builtin / fsck.c
1 #include "builtin.h"
2 #include "cache.h"
3 #include "commit.h"
4 #include "tree.h"
5 #include "blob.h"
6 #include "tag.h"
7 #include "refs.h"
8 #include "pack.h"
9 #include "cache-tree.h"
10 #include "tree-walk.h"
11 #include "fsck.h"
12 #include "parse-options.h"
13 #include "dir.h"
14 #include "progress.h"
15 #include "streaming.h"
16 #include "decorate.h"
17
18 #define REACHABLE 0x0001
19 #define SEEN 0x0002
20 #define HAS_OBJ 0x0004
21
22 static int show_root;
23 static int show_tags;
24 static int show_unreachable;
25 static int include_reflogs = 1;
26 static int check_full = 1;
27 static int connectivity_only;
28 static int check_strict;
29 static int keep_cache_objects;
30 static struct fsck_options fsck_walk_options = FSCK_OPTIONS_DEFAULT;
31 static struct fsck_options fsck_obj_options = FSCK_OPTIONS_DEFAULT;
32 static struct object_id head_oid;
33 static const char *head_points_at;
34 static int errors_found;
35 static int write_lost_and_found;
36 static int verbose;
37 static int show_progress = -1;
38 static int show_dangling = 1;
39 static int name_objects;
40 #define ERROR_OBJECT 01
41 #define ERROR_REACHABLE 02
42 #define ERROR_PACK 04
43 #define ERROR_REFS 010
44
45 static const char *describe_object(struct object *obj)
46 {
47 static struct strbuf buf = STRBUF_INIT;
48 char *name = name_objects ?
49 lookup_decoration(fsck_walk_options.object_names, obj) : NULL;
50
51 strbuf_reset(&buf);
52 strbuf_addstr(&buf, oid_to_hex(&obj->oid));
53 if (name)
54 strbuf_addf(&buf, " (%s)", name);
55
56 return buf.buf;
57 }
58
59 static int fsck_config(const char *var, const char *value, void *cb)
60 {
61 if (strcmp(var, "fsck.skiplist") == 0) {
62 const char *path;
63 struct strbuf sb = STRBUF_INIT;
64
65 if (git_config_pathname(&path, var, value))
66 return 1;
67 strbuf_addf(&sb, "skiplist=%s", path);
68 free((char *)path);
69 fsck_set_msg_types(&fsck_obj_options, sb.buf);
70 strbuf_release(&sb);
71 return 0;
72 }
73
74 if (skip_prefix(var, "fsck.", &var)) {
75 fsck_set_msg_type(&fsck_obj_options, var, value);
76 return 0;
77 }
78
79 return git_default_config(var, value, cb);
80 }
81
82 static void objreport(struct object *obj, const char *msg_type,
83 const char *err)
84 {
85 fprintf(stderr, "%s in %s %s: %s\n",
86 msg_type, typename(obj->type), describe_object(obj), err);
87 }
88
89 static int objerror(struct object *obj, const char *err)
90 {
91 errors_found |= ERROR_OBJECT;
92 objreport(obj, "error", err);
93 return -1;
94 }
95
96 static int fsck_error_func(struct fsck_options *o,
97 struct object *obj, int type, const char *message)
98 {
99 objreport(obj, (type == FSCK_WARN) ? "warning" : "error", message);
100 return (type == FSCK_WARN) ? 0 : 1;
101 }
102
103 static struct object_array pending;
104
105 static int mark_object(struct object *obj, int type, void *data, struct fsck_options *options)
106 {
107 struct object *parent = data;
108
109 /*
110 * The only case data is NULL or type is OBJ_ANY is when
111 * mark_object_reachable() calls us. All the callers of
112 * that function has non-NULL obj hence ...
113 */
114 if (!obj) {
115 /* ... these references to parent->fld are safe here */
116 printf("broken link from %7s %s\n",
117 typename(parent->type), describe_object(parent));
118 printf("broken link from %7s %s\n",
119 (type == OBJ_ANY ? "unknown" : typename(type)), "unknown");
120 errors_found |= ERROR_REACHABLE;
121 return 1;
122 }
123
124 if (type != OBJ_ANY && obj->type != type)
125 /* ... and the reference to parent is safe here */
126 objerror(parent, "wrong object type in link");
127
128 if (obj->flags & REACHABLE)
129 return 0;
130 obj->flags |= REACHABLE;
131 if (!(obj->flags & HAS_OBJ)) {
132 if (parent && !has_object_file(&obj->oid)) {
133 printf("broken link from %7s %s\n",
134 typename(parent->type), describe_object(parent));
135 printf(" to %7s %s\n",
136 typename(obj->type), describe_object(obj));
137 errors_found |= ERROR_REACHABLE;
138 }
139 return 1;
140 }
141
142 add_object_array(obj, NULL, &pending);
143 return 0;
144 }
145
146 static void mark_object_reachable(struct object *obj)
147 {
148 mark_object(obj, OBJ_ANY, NULL, NULL);
149 }
150
151 static int traverse_one_object(struct object *obj)
152 {
153 int result;
154 struct tree *tree = NULL;
155
156 if (obj->type == OBJ_TREE) {
157 tree = (struct tree *)obj;
158 if (parse_tree(tree) < 0)
159 return 1; /* error already displayed */
160 }
161 result = fsck_walk(obj, obj, &fsck_walk_options);
162 if (tree)
163 free_tree_buffer(tree);
164 return result;
165 }
166
167 static int traverse_reachable(void)
168 {
169 struct progress *progress = NULL;
170 unsigned int nr = 0;
171 int result = 0;
172 if (show_progress)
173 progress = start_progress_delay(_("Checking connectivity"), 0, 0, 2);
174 while (pending.nr) {
175 struct object_array_entry *entry;
176 struct object *obj;
177
178 entry = pending.objects + --pending.nr;
179 obj = entry->item;
180 result |= traverse_one_object(obj);
181 display_progress(progress, ++nr);
182 }
183 stop_progress(&progress);
184 return !!result;
185 }
186
187 static int mark_used(struct object *obj, int type, void *data, struct fsck_options *options)
188 {
189 if (!obj)
190 return 1;
191 obj->used = 1;
192 return 0;
193 }
194
195 /*
196 * Check a single reachable object
197 */
198 static void check_reachable_object(struct object *obj)
199 {
200 /*
201 * We obviously want the object to be parsed,
202 * except if it was in a pack-file and we didn't
203 * do a full fsck
204 */
205 if (!(obj->flags & HAS_OBJ)) {
206 if (has_sha1_pack(obj->oid.hash))
207 return; /* it is in pack - forget about it */
208 if (connectivity_only && has_object_file(&obj->oid))
209 return;
210 printf("missing %s %s\n", typename(obj->type),
211 describe_object(obj));
212 errors_found |= ERROR_REACHABLE;
213 return;
214 }
215 }
216
217 /*
218 * Check a single unreachable object
219 */
220 static void check_unreachable_object(struct object *obj)
221 {
222 /*
223 * Missing unreachable object? Ignore it. It's not like
224 * we miss it (since it can't be reached), nor do we want
225 * to complain about it being unreachable (since it does
226 * not exist).
227 */
228 if (!obj->parsed)
229 return;
230
231 /*
232 * Unreachable object that exists? Show it if asked to,
233 * since this is something that is prunable.
234 */
235 if (show_unreachable) {
236 printf("unreachable %s %s\n", typename(obj->type),
237 describe_object(obj));
238 return;
239 }
240
241 /*
242 * "!used" means that nothing at all points to it, including
243 * other unreachable objects. In other words, it's the "tip"
244 * of some set of unreachable objects, usually a commit that
245 * got dropped.
246 *
247 * Such starting points are more interesting than some random
248 * set of unreachable objects, so we show them even if the user
249 * hasn't asked for _all_ unreachable objects. If you have
250 * deleted a branch by mistake, this is a prime candidate to
251 * start looking at, for example.
252 */
253 if (!obj->used) {
254 if (show_dangling)
255 printf("dangling %s %s\n", typename(obj->type),
256 describe_object(obj));
257 if (write_lost_and_found) {
258 char *filename = git_pathdup("lost-found/%s/%s",
259 obj->type == OBJ_COMMIT ? "commit" : "other",
260 describe_object(obj));
261 FILE *f;
262
263 if (safe_create_leading_directories_const(filename)) {
264 error("Could not create lost-found");
265 free(filename);
266 return;
267 }
268 if (!(f = fopen(filename, "w")))
269 die_errno("Could not open '%s'", filename);
270 if (obj->type == OBJ_BLOB) {
271 if (stream_blob_to_fd(fileno(f), &obj->oid, NULL, 1))
272 die_errno("Could not write '%s'", filename);
273 } else
274 fprintf(f, "%s\n", describe_object(obj));
275 if (fclose(f))
276 die_errno("Could not finish '%s'",
277 filename);
278 free(filename);
279 }
280 return;
281 }
282
283 /*
284 * Otherwise? It's there, it's unreachable, and some other unreachable
285 * object points to it. Ignore it - it's not interesting, and we showed
286 * all the interesting cases above.
287 */
288 }
289
290 static void check_object(struct object *obj)
291 {
292 if (verbose)
293 fprintf(stderr, "Checking %s\n", describe_object(obj));
294
295 if (obj->flags & REACHABLE)
296 check_reachable_object(obj);
297 else
298 check_unreachable_object(obj);
299 }
300
301 static void check_connectivity(void)
302 {
303 int i, max;
304
305 /* Traverse the pending reachable objects */
306 traverse_reachable();
307
308 /* Look up all the requirements, warn about missing objects.. */
309 max = get_max_object_index();
310 if (verbose)
311 fprintf(stderr, "Checking connectivity (%d objects)\n", max);
312
313 for (i = 0; i < max; i++) {
314 struct object *obj = get_indexed_object(i);
315
316 if (obj)
317 check_object(obj);
318 }
319 }
320
321 static int fsck_obj(struct object *obj)
322 {
323 if (obj->flags & SEEN)
324 return 0;
325 obj->flags |= SEEN;
326
327 if (verbose)
328 fprintf(stderr, "Checking %s %s\n",
329 typename(obj->type), describe_object(obj));
330
331 if (fsck_walk(obj, NULL, &fsck_obj_options))
332 objerror(obj, "broken links");
333 if (fsck_object(obj, NULL, 0, &fsck_obj_options))
334 return -1;
335
336 if (obj->type == OBJ_TREE) {
337 struct tree *item = (struct tree *) obj;
338
339 free_tree_buffer(item);
340 }
341
342 if (obj->type == OBJ_COMMIT) {
343 struct commit *commit = (struct commit *) obj;
344
345 free_commit_buffer(commit);
346
347 if (!commit->parents && show_root)
348 printf("root %s\n", describe_object(&commit->object));
349 }
350
351 if (obj->type == OBJ_TAG) {
352 struct tag *tag = (struct tag *) obj;
353
354 if (show_tags && tag->tagged) {
355 printf("tagged %s %s", typename(tag->tagged->type),
356 describe_object(tag->tagged));
357 printf(" (%s) in %s\n", tag->tag,
358 describe_object(&tag->object));
359 }
360 }
361
362 return 0;
363 }
364
365 static int fsck_obj_buffer(const unsigned char *sha1, enum object_type type,
366 unsigned long size, void *buffer, int *eaten)
367 {
368 /*
369 * Note, buffer may be NULL if type is OBJ_BLOB. See
370 * verify_packfile(), data_valid variable for details.
371 */
372 struct object *obj;
373 obj = parse_object_buffer(sha1, type, size, buffer, eaten);
374 if (!obj) {
375 errors_found |= ERROR_OBJECT;
376 return error("%s: object corrupt or missing", sha1_to_hex(sha1));
377 }
378 obj->flags = HAS_OBJ;
379 return fsck_obj(obj);
380 }
381
382 static int default_refs;
383
384 static void fsck_handle_reflog_sha1(const char *refname, unsigned char *sha1,
385 unsigned long timestamp)
386 {
387 struct object *obj;
388
389 if (!is_null_sha1(sha1)) {
390 obj = lookup_object(sha1);
391 if (obj) {
392 if (timestamp && name_objects)
393 add_decoration(fsck_walk_options.object_names,
394 obj,
395 xstrfmt("%s@{%ld}", refname, timestamp));
396 obj->used = 1;
397 mark_object_reachable(obj);
398 } else {
399 error("%s: invalid reflog entry %s", refname, sha1_to_hex(sha1));
400 errors_found |= ERROR_REACHABLE;
401 }
402 }
403 }
404
405 static int fsck_handle_reflog_ent(unsigned char *osha1, unsigned char *nsha1,
406 const char *email, unsigned long timestamp, int tz,
407 const char *message, void *cb_data)
408 {
409 const char *refname = cb_data;
410
411 if (verbose)
412 fprintf(stderr, "Checking reflog %s->%s\n",
413 sha1_to_hex(osha1), sha1_to_hex(nsha1));
414
415 fsck_handle_reflog_sha1(refname, osha1, 0);
416 fsck_handle_reflog_sha1(refname, nsha1, timestamp);
417 return 0;
418 }
419
420 static int fsck_handle_reflog(const char *logname, const struct object_id *oid,
421 int flag, void *cb_data)
422 {
423 for_each_reflog_ent(logname, fsck_handle_reflog_ent, (void *)logname);
424 return 0;
425 }
426
427 static int fsck_handle_ref(const char *refname, const struct object_id *oid,
428 int flag, void *cb_data)
429 {
430 struct object *obj;
431
432 obj = parse_object(oid->hash);
433 if (!obj) {
434 error("%s: invalid sha1 pointer %s", refname, oid_to_hex(oid));
435 errors_found |= ERROR_REACHABLE;
436 /* We'll continue with the rest despite the error.. */
437 return 0;
438 }
439 if (obj->type != OBJ_COMMIT && is_branch(refname)) {
440 error("%s: not a commit", refname);
441 errors_found |= ERROR_REFS;
442 }
443 default_refs++;
444 obj->used = 1;
445 if (name_objects)
446 add_decoration(fsck_walk_options.object_names,
447 obj, xstrdup(refname));
448 mark_object_reachable(obj);
449
450 return 0;
451 }
452
453 static void get_default_heads(void)
454 {
455 if (head_points_at && !is_null_oid(&head_oid))
456 fsck_handle_ref("HEAD", &head_oid, 0, NULL);
457 for_each_rawref(fsck_handle_ref, NULL);
458 if (include_reflogs)
459 for_each_reflog(fsck_handle_reflog, NULL);
460
461 /*
462 * Not having any default heads isn't really fatal, but
463 * it does mean that "--unreachable" no longer makes any
464 * sense (since in this case everything will obviously
465 * be unreachable by definition.
466 *
467 * Showing dangling objects is valid, though (as those
468 * dangling objects are likely lost heads).
469 *
470 * So we just print a warning about it, and clear the
471 * "show_unreachable" flag.
472 */
473 if (!default_refs) {
474 fprintf(stderr, "notice: No default references\n");
475 show_unreachable = 0;
476 }
477 }
478
479 static struct object *parse_loose_object(const unsigned char *sha1,
480 const char *path)
481 {
482 struct object *obj;
483 void *contents;
484 enum object_type type;
485 unsigned long size;
486 int eaten;
487
488 if (read_loose_object(path, sha1, &type, &size, &contents) < 0)
489 return NULL;
490
491 if (!contents && type != OBJ_BLOB)
492 die("BUG: read_loose_object streamed a non-blob");
493
494 obj = parse_object_buffer(sha1, type, size, contents, &eaten);
495
496 if (!eaten)
497 free(contents);
498 return obj;
499 }
500
501 static int fsck_loose(const unsigned char *sha1, const char *path, void *data)
502 {
503 struct object *obj = parse_loose_object(sha1, path);
504
505 if (!obj) {
506 errors_found |= ERROR_OBJECT;
507 error("%s: object corrupt or missing: %s",
508 sha1_to_hex(sha1), path);
509 return 0; /* keep checking other objects */
510 }
511
512 obj->flags = HAS_OBJ;
513 if (fsck_obj(obj))
514 errors_found |= ERROR_OBJECT;
515 return 0;
516 }
517
518 static int fsck_cruft(const char *basename, const char *path, void *data)
519 {
520 if (!starts_with(basename, "tmp_obj_"))
521 fprintf(stderr, "bad sha1 file: %s\n", path);
522 return 0;
523 }
524
525 static int fsck_subdir(int nr, const char *path, void *progress)
526 {
527 display_progress(progress, nr + 1);
528 return 0;
529 }
530
531 static void fsck_object_dir(const char *path)
532 {
533 struct progress *progress = NULL;
534
535 if (verbose)
536 fprintf(stderr, "Checking object directory\n");
537
538 if (show_progress)
539 progress = start_progress(_("Checking object directories"), 256);
540
541 for_each_loose_file_in_objdir(path, fsck_loose, fsck_cruft, fsck_subdir,
542 progress);
543 display_progress(progress, 256);
544 stop_progress(&progress);
545 }
546
547 static int fsck_head_link(void)
548 {
549 int null_is_error = 0;
550
551 if (verbose)
552 fprintf(stderr, "Checking HEAD link\n");
553
554 head_points_at = resolve_ref_unsafe("HEAD", 0, head_oid.hash, NULL);
555 if (!head_points_at) {
556 errors_found |= ERROR_REFS;
557 return error("Invalid HEAD");
558 }
559 if (!strcmp(head_points_at, "HEAD"))
560 /* detached HEAD */
561 null_is_error = 1;
562 else if (!starts_with(head_points_at, "refs/heads/")) {
563 errors_found |= ERROR_REFS;
564 return error("HEAD points to something strange (%s)",
565 head_points_at);
566 }
567 if (is_null_oid(&head_oid)) {
568 if (null_is_error) {
569 errors_found |= ERROR_REFS;
570 return error("HEAD: detached HEAD points at nothing");
571 }
572 fprintf(stderr, "notice: HEAD points to an unborn branch (%s)\n",
573 head_points_at + 11);
574 }
575 return 0;
576 }
577
578 static int fsck_cache_tree(struct cache_tree *it)
579 {
580 int i;
581 int err = 0;
582
583 if (verbose)
584 fprintf(stderr, "Checking cache tree\n");
585
586 if (0 <= it->entry_count) {
587 struct object *obj = parse_object(it->sha1);
588 if (!obj) {
589 error("%s: invalid sha1 pointer in cache-tree",
590 sha1_to_hex(it->sha1));
591 errors_found |= ERROR_REFS;
592 return 1;
593 }
594 obj->used = 1;
595 if (name_objects)
596 add_decoration(fsck_walk_options.object_names,
597 obj, xstrdup(":"));
598 mark_object_reachable(obj);
599 if (obj->type != OBJ_TREE)
600 err |= objerror(obj, "non-tree in cache-tree");
601 }
602 for (i = 0; i < it->subtree_nr; i++)
603 err |= fsck_cache_tree(it->down[i]->cache_tree);
604 return err;
605 }
606
607 static char const * const fsck_usage[] = {
608 N_("git fsck [<options>] [<object>...]"),
609 NULL
610 };
611
612 static struct option fsck_opts[] = {
613 OPT__VERBOSE(&verbose, N_("be verbose")),
614 OPT_BOOL(0, "unreachable", &show_unreachable, N_("show unreachable objects")),
615 OPT_BOOL(0, "dangling", &show_dangling, N_("show dangling objects")),
616 OPT_BOOL(0, "tags", &show_tags, N_("report tags")),
617 OPT_BOOL(0, "root", &show_root, N_("report root nodes")),
618 OPT_BOOL(0, "cache", &keep_cache_objects, N_("make index objects head nodes")),
619 OPT_BOOL(0, "reflogs", &include_reflogs, N_("make reflogs head nodes (default)")),
620 OPT_BOOL(0, "full", &check_full, N_("also consider packs and alternate objects")),
621 OPT_BOOL(0, "connectivity-only", &connectivity_only, N_("check only connectivity")),
622 OPT_BOOL(0, "strict", &check_strict, N_("enable more strict checking")),
623 OPT_BOOL(0, "lost-found", &write_lost_and_found,
624 N_("write dangling objects in .git/lost-found")),
625 OPT_BOOL(0, "progress", &show_progress, N_("show progress")),
626 OPT_BOOL(0, "name-objects", &name_objects, N_("show verbose names for reachable objects")),
627 OPT_END(),
628 };
629
630 int cmd_fsck(int argc, const char **argv, const char *prefix)
631 {
632 int i, heads;
633 struct alternate_object_database *alt;
634
635 errors_found = 0;
636 check_replace_refs = 0;
637
638 argc = parse_options(argc, argv, prefix, fsck_opts, fsck_usage, 0);
639
640 fsck_walk_options.walk = mark_object;
641 fsck_obj_options.walk = mark_used;
642 fsck_obj_options.error_func = fsck_error_func;
643 if (check_strict)
644 fsck_obj_options.strict = 1;
645
646 if (show_progress == -1)
647 show_progress = isatty(2);
648 if (verbose)
649 show_progress = 0;
650
651 if (write_lost_and_found) {
652 check_full = 1;
653 include_reflogs = 0;
654 }
655
656 if (name_objects)
657 fsck_walk_options.object_names =
658 xcalloc(1, sizeof(struct decoration));
659
660 git_config(fsck_config, NULL);
661
662 fsck_head_link();
663 if (!connectivity_only) {
664 fsck_object_dir(get_object_directory());
665
666 prepare_alt_odb();
667 for (alt = alt_odb_list; alt; alt = alt->next)
668 fsck_object_dir(alt->path);
669 }
670
671 if (check_full) {
672 struct packed_git *p;
673 uint32_t total = 0, count = 0;
674 struct progress *progress = NULL;
675
676 prepare_packed_git();
677
678 if (show_progress) {
679 for (p = packed_git; p; p = p->next) {
680 if (open_pack_index(p))
681 continue;
682 total += p->num_objects;
683 }
684
685 progress = start_progress(_("Checking objects"), total);
686 }
687 for (p = packed_git; p; p = p->next) {
688 /* verify gives error messages itself */
689 if (verify_pack(p, fsck_obj_buffer,
690 progress, count))
691 errors_found |= ERROR_PACK;
692 count += p->num_objects;
693 }
694 stop_progress(&progress);
695 }
696
697 heads = 0;
698 for (i = 0; i < argc; i++) {
699 const char *arg = argv[i];
700 unsigned char sha1[20];
701 if (!get_sha1(arg, sha1)) {
702 struct object *obj = lookup_object(sha1);
703
704 /* Error is printed by lookup_object(). */
705 if (!obj)
706 continue;
707
708 obj->used = 1;
709 if (name_objects)
710 add_decoration(fsck_walk_options.object_names,
711 obj, xstrdup(arg));
712 mark_object_reachable(obj);
713 heads++;
714 continue;
715 }
716 error("invalid parameter: expected sha1, got '%s'", arg);
717 }
718
719 /*
720 * If we've not been given any explicit head information, do the
721 * default ones from .git/refs. We also consider the index file
722 * in this case (ie this implies --cache).
723 */
724 if (!heads) {
725 get_default_heads();
726 keep_cache_objects = 1;
727 }
728
729 if (keep_cache_objects) {
730 read_cache();
731 for (i = 0; i < active_nr; i++) {
732 unsigned int mode;
733 struct blob *blob;
734 struct object *obj;
735
736 mode = active_cache[i]->ce_mode;
737 if (S_ISGITLINK(mode))
738 continue;
739 blob = lookup_blob(active_cache[i]->oid.hash);
740 if (!blob)
741 continue;
742 obj = &blob->object;
743 obj->used = 1;
744 if (name_objects)
745 add_decoration(fsck_walk_options.object_names,
746 obj,
747 xstrfmt(":%s", active_cache[i]->name));
748 mark_object_reachable(obj);
749 }
750 if (active_cache_tree)
751 fsck_cache_tree(active_cache_tree);
752 }
753
754 check_connectivity();
755 return errors_found;
756 }