]> git.ipfire.org Git - thirdparty/git.git/blame_incremental - builtin/fsck.c
The sixth batch
[thirdparty/git.git] / builtin / fsck.c
... / ...
CommitLineData
1#define USE_THE_REPOSITORY_VARIABLE
2#include "builtin.h"
3#include "gettext.h"
4#include "hex.h"
5#include "config.h"
6#include "commit.h"
7#include "tree.h"
8#include "blob.h"
9#include "tag.h"
10#include "refs.h"
11#include "pack.h"
12#include "cache-tree.h"
13#include "fsck.h"
14#include "parse-options.h"
15#include "progress.h"
16#include "streaming.h"
17#include "packfile.h"
18#include "object-file.h"
19#include "object-name.h"
20#include "object-store.h"
21#include "path.h"
22#include "read-cache-ll.h"
23#include "replace-object.h"
24#include "resolve-undo.h"
25#include "run-command.h"
26#include "sparse-index.h"
27#include "worktree.h"
28#include "pack-revindex.h"
29#include "pack-bitmap.h"
30
31#define REACHABLE 0x0001
32#define SEEN 0x0002
33#define HAS_OBJ 0x0004
34/* This flag is set if something points to this object. */
35#define USED 0x0008
36
37static int show_root;
38static int show_tags;
39static int show_unreachable;
40static int include_reflogs = 1;
41static int check_full = 1;
42static int connectivity_only;
43static int check_strict;
44static int keep_cache_objects;
45static struct fsck_options fsck_walk_options = FSCK_OPTIONS_DEFAULT;
46static struct fsck_options fsck_obj_options = FSCK_OPTIONS_DEFAULT;
47static int errors_found;
48static int write_lost_and_found;
49static int verbose;
50static int show_progress = -1;
51static int show_dangling = 1;
52static int name_objects;
53static int check_references = 1;
54#define ERROR_OBJECT 01
55#define ERROR_REACHABLE 02
56#define ERROR_PACK 04
57#define ERROR_REFS 010
58#define ERROR_COMMIT_GRAPH 020
59#define ERROR_MULTI_PACK_INDEX 040
60#define ERROR_PACK_REV_INDEX 0100
61#define ERROR_BITMAP 0200
62
63static const char *describe_object(const struct object_id *oid)
64{
65 return fsck_describe_object(&fsck_walk_options, oid);
66}
67
68static const char *printable_type(const struct object_id *oid,
69 enum object_type type)
70{
71 const char *ret;
72
73 if (type == OBJ_NONE)
74 type = oid_object_info(the_repository, oid, NULL);
75
76 ret = type_name(type);
77 if (!ret)
78 ret = _("unknown");
79
80 return ret;
81}
82
83static int objerror(struct object *obj, const char *err)
84{
85 errors_found |= ERROR_OBJECT;
86 /* TRANSLATORS: e.g. error in tree 01bfda: <more explanation> */
87 fprintf_ln(stderr, _("error in %s %s: %s"),
88 printable_type(&obj->oid, obj->type),
89 describe_object(&obj->oid), err);
90 return -1;
91}
92
93static int fsck_objects_error_func(struct fsck_options *o UNUSED,
94 void *fsck_report,
95 enum fsck_msg_type msg_type,
96 enum fsck_msg_id msg_id UNUSED,
97 const char *message)
98{
99 struct fsck_object_report *report = fsck_report;
100 const struct object_id *oid = report->oid;
101 enum object_type object_type = report->object_type;
102
103 switch (msg_type) {
104 case FSCK_WARN:
105 /* TRANSLATORS: e.g. warning in tree 01bfda: <more explanation> */
106 fprintf_ln(stderr, _("warning in %s %s: %s"),
107 printable_type(oid, object_type),
108 describe_object(oid), message);
109 return 0;
110 case FSCK_ERROR:
111 /* TRANSLATORS: e.g. error in tree 01bfda: <more explanation> */
112 fprintf_ln(stderr, _("error in %s %s: %s"),
113 printable_type(oid, object_type),
114 describe_object(oid), message);
115 return 1;
116 default:
117 BUG("%d (FSCK_IGNORE?) should never trigger this callback",
118 msg_type);
119 }
120}
121
122static struct object_array pending;
123
124static int mark_object(struct object *obj, enum object_type type,
125 void *data, struct fsck_options *options UNUSED)
126{
127 struct object *parent = data;
128
129 /*
130 * The only case data is NULL or type is OBJ_ANY is when
131 * mark_object_reachable() calls us. All the callers of
132 * that function has non-NULL obj hence ...
133 */
134 if (!obj) {
135 /* ... these references to parent->fld are safe here */
136 printf_ln(_("broken link from %7s %s"),
137 printable_type(&parent->oid, parent->type),
138 describe_object(&parent->oid));
139 printf_ln(_("broken link from %7s %s"),
140 (type == OBJ_ANY ? _("unknown") : type_name(type)),
141 _("unknown"));
142 errors_found |= ERROR_REACHABLE;
143 return 1;
144 }
145
146 if (type != OBJ_ANY && obj->type != type)
147 /* ... and the reference to parent is safe here */
148 objerror(parent, _("wrong object type in link"));
149
150 if (obj->flags & REACHABLE)
151 return 0;
152 obj->flags |= REACHABLE;
153
154 if (is_promisor_object(the_repository, &obj->oid))
155 /*
156 * Further recursion does not need to be performed on this
157 * object since it is a promisor object (so it does not need to
158 * be added to "pending").
159 */
160 return 0;
161
162 if (!(obj->flags & HAS_OBJ)) {
163 if (parent && !has_object(the_repository, &obj->oid, 1)) {
164 printf_ln(_("broken link from %7s %s\n"
165 " to %7s %s"),
166 printable_type(&parent->oid, parent->type),
167 describe_object(&parent->oid),
168 printable_type(&obj->oid, obj->type),
169 describe_object(&obj->oid));
170 errors_found |= ERROR_REACHABLE;
171 }
172 return 1;
173 }
174
175 add_object_array(obj, NULL, &pending);
176 return 0;
177}
178
179static void mark_object_reachable(struct object *obj)
180{
181 mark_object(obj, OBJ_ANY, NULL, NULL);
182}
183
184static int traverse_one_object(struct object *obj)
185{
186 int result = fsck_walk(obj, obj, &fsck_walk_options);
187
188 if (obj->type == OBJ_TREE) {
189 struct tree *tree = (struct tree *)obj;
190 free_tree_buffer(tree);
191 }
192 return result;
193}
194
195static int traverse_reachable(void)
196{
197 struct progress *progress = NULL;
198 unsigned int nr = 0;
199 int result = 0;
200 if (show_progress)
201 progress = start_delayed_progress(the_repository,
202 _("Checking connectivity"), 0);
203 while (pending.nr) {
204 result |= traverse_one_object(object_array_pop(&pending));
205 display_progress(progress, ++nr);
206 }
207 stop_progress(&progress);
208 return !!result;
209}
210
211static int mark_used(struct object *obj, enum object_type type UNUSED,
212 void *data UNUSED, struct fsck_options *options UNUSED)
213{
214 if (!obj)
215 return 1;
216 obj->flags |= USED;
217 return 0;
218}
219
220static void mark_unreachable_referents(const struct object_id *oid)
221{
222 struct fsck_options options = FSCK_OPTIONS_DEFAULT;
223 struct object *obj = lookup_object(the_repository, oid);
224
225 if (!obj || !(obj->flags & HAS_OBJ))
226 return; /* not part of our original set */
227 if (obj->flags & REACHABLE)
228 return; /* reachable objects already traversed */
229
230 /*
231 * Avoid passing OBJ_NONE to fsck_walk, which will parse the object
232 * (and we want to avoid parsing blobs).
233 */
234 if (obj->type == OBJ_NONE) {
235 enum object_type type = oid_object_info(the_repository,
236 &obj->oid, NULL);
237 if (type > 0)
238 object_as_type(obj, type, 0);
239 }
240
241 options.walk = mark_used;
242 fsck_walk(obj, NULL, &options);
243 if (obj->type == OBJ_TREE)
244 free_tree_buffer((struct tree *)obj);
245}
246
247static int mark_loose_unreachable_referents(const struct object_id *oid,
248 const char *path UNUSED,
249 void *data UNUSED)
250{
251 mark_unreachable_referents(oid);
252 return 0;
253}
254
255static int mark_packed_unreachable_referents(const struct object_id *oid,
256 struct packed_git *pack UNUSED,
257 uint32_t pos UNUSED,
258 void *data UNUSED)
259{
260 mark_unreachable_referents(oid);
261 return 0;
262}
263
264/*
265 * Check a single reachable object
266 */
267static void check_reachable_object(struct object *obj)
268{
269 /*
270 * We obviously want the object to be parsed,
271 * except if it was in a pack-file and we didn't
272 * do a full fsck
273 */
274 if (!(obj->flags & HAS_OBJ)) {
275 if (is_promisor_object(the_repository, &obj->oid))
276 return;
277 if (has_object_pack(the_repository, &obj->oid))
278 return; /* it is in pack - forget about it */
279 printf_ln(_("missing %s %s"),
280 printable_type(&obj->oid, obj->type),
281 describe_object(&obj->oid));
282 errors_found |= ERROR_REACHABLE;
283 return;
284 }
285}
286
287/*
288 * Check a single unreachable object
289 */
290static void check_unreachable_object(struct object *obj)
291{
292 /*
293 * Missing unreachable object? Ignore it. It's not like
294 * we miss it (since it can't be reached), nor do we want
295 * to complain about it being unreachable (since it does
296 * not exist).
297 */
298 if (!(obj->flags & HAS_OBJ))
299 return;
300
301 /*
302 * Unreachable object that exists? Show it if asked to,
303 * since this is something that is prunable.
304 */
305 if (show_unreachable) {
306 printf_ln(_("unreachable %s %s"),
307 printable_type(&obj->oid, obj->type),
308 describe_object(&obj->oid));
309 return;
310 }
311
312 /*
313 * "!USED" means that nothing at all points to it, including
314 * other unreachable objects. In other words, it's the "tip"
315 * of some set of unreachable objects, usually a commit that
316 * got dropped.
317 *
318 * Such starting points are more interesting than some random
319 * set of unreachable objects, so we show them even if the user
320 * hasn't asked for _all_ unreachable objects. If you have
321 * deleted a branch by mistake, this is a prime candidate to
322 * start looking at, for example.
323 */
324 if (!(obj->flags & USED)) {
325 if (show_dangling)
326 printf_ln(_("dangling %s %s"),
327 printable_type(&obj->oid, obj->type),
328 describe_object(&obj->oid));
329 if (write_lost_and_found) {
330 char *filename = repo_git_path(the_repository, "lost-found/%s/%s",
331 obj->type == OBJ_COMMIT ? "commit" : "other",
332 describe_object(&obj->oid));
333 FILE *f;
334
335 if (safe_create_leading_directories_const(the_repository, filename)) {
336 error(_("could not create lost-found"));
337 free(filename);
338 return;
339 }
340 f = xfopen(filename, "w");
341 if (obj->type == OBJ_BLOB) {
342 if (stream_blob_to_fd(fileno(f), &obj->oid, NULL, 1))
343 die_errno(_("could not write '%s'"), filename);
344 } else
345 fprintf(f, "%s\n", describe_object(&obj->oid));
346 if (fclose(f))
347 die_errno(_("could not finish '%s'"),
348 filename);
349 free(filename);
350 }
351 return;
352 }
353
354 /*
355 * Otherwise? It's there, it's unreachable, and some other unreachable
356 * object points to it. Ignore it - it's not interesting, and we showed
357 * all the interesting cases above.
358 */
359}
360
361static void check_object(struct object *obj)
362{
363 if (verbose)
364 fprintf_ln(stderr, _("Checking %s"), describe_object(&obj->oid));
365
366 if (obj->flags & REACHABLE)
367 check_reachable_object(obj);
368 else
369 check_unreachable_object(obj);
370}
371
372static void check_connectivity(void)
373{
374 int i, max;
375
376 /* Traverse the pending reachable objects */
377 traverse_reachable();
378
379 /*
380 * With --connectivity-only, we won't have actually opened and marked
381 * unreachable objects with USED. Do that now to make --dangling, etc
382 * accurate.
383 */
384 if (connectivity_only && (show_dangling || write_lost_and_found)) {
385 /*
386 * Even though we already have a "struct object" for each of
387 * these in memory, we must not iterate over the internal
388 * object hash as we do below. Our loop would potentially
389 * resize the hash, making our iteration invalid.
390 *
391 * Instead, we'll just go back to the source list of objects,
392 * and ignore any that weren't present in our earlier
393 * traversal.
394 */
395 for_each_loose_object(mark_loose_unreachable_referents, NULL, 0);
396 for_each_packed_object(the_repository,
397 mark_packed_unreachable_referents,
398 NULL,
399 0);
400 }
401
402 /* Look up all the requirements, warn about missing objects.. */
403 max = get_max_object_index(the_repository);
404 if (verbose)
405 fprintf_ln(stderr, _("Checking connectivity (%d objects)"), max);
406
407 for (i = 0; i < max; i++) {
408 struct object *obj = get_indexed_object(the_repository, i);
409
410 if (obj)
411 check_object(obj);
412 }
413}
414
415static int fsck_obj(struct object *obj, void *buffer, unsigned long size)
416{
417 int err;
418
419 if (obj->flags & SEEN)
420 return 0;
421 obj->flags |= SEEN;
422
423 if (verbose)
424 fprintf_ln(stderr, _("Checking %s %s"),
425 printable_type(&obj->oid, obj->type),
426 describe_object(&obj->oid));
427
428 if (fsck_walk(obj, NULL, &fsck_obj_options))
429 objerror(obj, _("broken links"));
430 err = fsck_object(obj, buffer, size, &fsck_obj_options);
431 if (err)
432 goto out;
433
434 if (obj->type == OBJ_COMMIT) {
435 struct commit *commit = (struct commit *) obj;
436
437 if (!commit->parents && show_root)
438 printf_ln(_("root %s"),
439 describe_object(&commit->object.oid));
440 }
441
442 if (obj->type == OBJ_TAG) {
443 struct tag *tag = (struct tag *) obj;
444
445 if (show_tags && tag->tagged) {
446 printf_ln(_("tagged %s %s (%s) in %s"),
447 printable_type(&tag->tagged->oid, tag->tagged->type),
448 describe_object(&tag->tagged->oid),
449 tag->tag,
450 describe_object(&tag->object.oid));
451 }
452 }
453
454out:
455 if (obj->type == OBJ_TREE)
456 free_tree_buffer((struct tree *)obj);
457 return err;
458}
459
460static int fsck_obj_buffer(const struct object_id *oid, enum object_type type,
461 unsigned long size, void *buffer, int *eaten)
462{
463 /*
464 * Note, buffer may be NULL if type is OBJ_BLOB. See
465 * verify_packfile(), data_valid variable for details.
466 */
467 struct object *obj;
468 obj = parse_object_buffer(the_repository, oid, type, size, buffer,
469 eaten);
470 if (!obj) {
471 errors_found |= ERROR_OBJECT;
472 return error(_("%s: object corrupt or missing"),
473 oid_to_hex(oid));
474 }
475 obj->flags &= ~(REACHABLE | SEEN);
476 obj->flags |= HAS_OBJ;
477 return fsck_obj(obj, buffer, size);
478}
479
480static int default_refs;
481
482static void fsck_handle_reflog_oid(const char *refname, struct object_id *oid,
483 timestamp_t timestamp)
484{
485 struct object *obj;
486
487 if (!is_null_oid(oid)) {
488 obj = lookup_object(the_repository, oid);
489 if (obj && (obj->flags & HAS_OBJ)) {
490 if (timestamp)
491 fsck_put_object_name(&fsck_walk_options, oid,
492 "%s@{%"PRItime"}",
493 refname, timestamp);
494 obj->flags |= USED;
495 mark_object_reachable(obj);
496 } else if (!is_promisor_object(the_repository, oid)) {
497 error(_("%s: invalid reflog entry %s"),
498 refname, oid_to_hex(oid));
499 errors_found |= ERROR_REACHABLE;
500 }
501 }
502}
503
504static int fsck_handle_reflog_ent(struct object_id *ooid, struct object_id *noid,
505 const char *email UNUSED,
506 timestamp_t timestamp, int tz UNUSED,
507 const char *message UNUSED, void *cb_data)
508{
509 const char *refname = cb_data;
510
511 if (verbose)
512 fprintf_ln(stderr, _("Checking reflog %s->%s"),
513 oid_to_hex(ooid), oid_to_hex(noid));
514
515 fsck_handle_reflog_oid(refname, ooid, 0);
516 fsck_handle_reflog_oid(refname, noid, timestamp);
517 return 0;
518}
519
520static int fsck_handle_reflog(const char *logname, void *cb_data)
521{
522 struct strbuf refname = STRBUF_INIT;
523
524 strbuf_worktree_ref(cb_data, &refname, logname);
525 refs_for_each_reflog_ent(get_main_ref_store(the_repository),
526 refname.buf, fsck_handle_reflog_ent,
527 refname.buf);
528 strbuf_release(&refname);
529 return 0;
530}
531
532static int fsck_handle_ref(const char *refname, const char *referent UNUSED, const struct object_id *oid,
533 int flag UNUSED, void *cb_data UNUSED)
534{
535 struct object *obj;
536
537 obj = parse_object(the_repository, oid);
538 if (!obj) {
539 if (is_promisor_object(the_repository, oid)) {
540 /*
541 * Increment default_refs anyway, because this is a
542 * valid ref.
543 */
544 default_refs++;
545 return 0;
546 }
547 error(_("%s: invalid sha1 pointer %s"),
548 refname, oid_to_hex(oid));
549 errors_found |= ERROR_REACHABLE;
550 /* We'll continue with the rest despite the error.. */
551 return 0;
552 }
553 if (obj->type != OBJ_COMMIT && is_branch(refname)) {
554 error(_("%s: not a commit"), refname);
555 errors_found |= ERROR_REFS;
556 }
557 default_refs++;
558 obj->flags |= USED;
559 fsck_put_object_name(&fsck_walk_options,
560 oid, "%s", refname);
561 mark_object_reachable(obj);
562
563 return 0;
564}
565
566static int fsck_head_link(const char *head_ref_name,
567 const char **head_points_at,
568 struct object_id *head_oid);
569
570static void get_default_heads(void)
571{
572 struct worktree **worktrees, **p;
573 const char *head_points_at;
574 struct object_id head_oid;
575
576 refs_for_each_rawref(get_main_ref_store(the_repository),
577 fsck_handle_ref, NULL);
578
579 worktrees = get_worktrees();
580 for (p = worktrees; *p; p++) {
581 struct worktree *wt = *p;
582 struct strbuf ref = STRBUF_INIT;
583
584 strbuf_worktree_ref(wt, &ref, "HEAD");
585 fsck_head_link(ref.buf, &head_points_at, &head_oid);
586 if (head_points_at && !is_null_oid(&head_oid))
587 fsck_handle_ref(ref.buf, NULL, &head_oid, 0, NULL);
588 strbuf_release(&ref);
589
590 if (include_reflogs)
591 refs_for_each_reflog(get_worktree_ref_store(wt),
592 fsck_handle_reflog, wt);
593 }
594 free_worktrees(worktrees);
595
596 /*
597 * Not having any default heads isn't really fatal, but
598 * it does mean that "--unreachable" no longer makes any
599 * sense (since in this case everything will obviously
600 * be unreachable by definition.
601 *
602 * Showing dangling objects is valid, though (as those
603 * dangling objects are likely lost heads).
604 *
605 * So we just print a warning about it, and clear the
606 * "show_unreachable" flag.
607 */
608 if (!default_refs) {
609 fprintf_ln(stderr, _("notice: No default references"));
610 show_unreachable = 0;
611 }
612}
613
614struct for_each_loose_cb
615{
616 struct progress *progress;
617};
618
619static int fsck_loose(const struct object_id *oid, const char *path,
620 void *data UNUSED)
621{
622 struct object *obj;
623 enum object_type type = OBJ_NONE;
624 unsigned long size;
625 void *contents = NULL;
626 int eaten;
627 struct object_info oi = OBJECT_INFO_INIT;
628 struct object_id real_oid = *null_oid(the_hash_algo);
629 int err = 0;
630
631 oi.sizep = &size;
632 oi.typep = &type;
633
634 if (read_loose_object(path, oid, &real_oid, &contents, &oi) < 0) {
635 if (contents && !oideq(&real_oid, oid))
636 err = error(_("%s: hash-path mismatch, found at: %s"),
637 oid_to_hex(&real_oid), path);
638 else
639 err = error(_("%s: object corrupt or missing: %s"),
640 oid_to_hex(oid), path);
641 }
642 if (err < 0) {
643 errors_found |= ERROR_OBJECT;
644 free(contents);
645 return 0; /* keep checking other objects */
646 }
647
648 if (!contents && type != OBJ_BLOB)
649 BUG("read_loose_object streamed a non-blob");
650
651 obj = parse_object_buffer(the_repository, oid, type, size,
652 contents, &eaten);
653
654 if (!obj) {
655 errors_found |= ERROR_OBJECT;
656 error(_("%s: object could not be parsed: %s"),
657 oid_to_hex(oid), path);
658 if (!eaten)
659 free(contents);
660 return 0; /* keep checking other objects */
661 }
662
663 obj->flags &= ~(REACHABLE | SEEN);
664 obj->flags |= HAS_OBJ;
665 if (fsck_obj(obj, contents, size))
666 errors_found |= ERROR_OBJECT;
667
668 if (!eaten)
669 free(contents);
670 return 0; /* keep checking other objects, even if we saw an error */
671}
672
673static int fsck_cruft(const char *basename, const char *path,
674 void *data UNUSED)
675{
676 if (!starts_with(basename, "tmp_obj_"))
677 fprintf_ln(stderr, _("bad sha1 file: %s"), path);
678 return 0;
679}
680
681static int fsck_subdir(unsigned int nr, const char *path UNUSED, void *data)
682{
683 struct for_each_loose_cb *cb_data = data;
684 struct progress *progress = cb_data->progress;
685 display_progress(progress, nr + 1);
686 return 0;
687}
688
689static void fsck_object_dir(const char *path)
690{
691 struct progress *progress = NULL;
692 struct for_each_loose_cb cb_data = {
693 .progress = progress,
694 };
695
696 if (verbose)
697 fprintf_ln(stderr, _("Checking object directory"));
698
699 if (show_progress)
700 progress = start_progress(the_repository,
701 _("Checking object directories"), 256);
702
703 for_each_loose_file_in_objdir(path, fsck_loose, fsck_cruft, fsck_subdir,
704 &cb_data);
705 display_progress(progress, 256);
706 stop_progress(&progress);
707}
708
709static int fsck_head_link(const char *head_ref_name,
710 const char **head_points_at,
711 struct object_id *head_oid)
712{
713 int null_is_error = 0;
714
715 if (verbose)
716 fprintf_ln(stderr, _("Checking %s link"), head_ref_name);
717
718 *head_points_at = refs_resolve_ref_unsafe(get_main_ref_store(the_repository),
719 head_ref_name, 0, head_oid,
720 NULL);
721 if (!*head_points_at) {
722 errors_found |= ERROR_REFS;
723 return error(_("invalid %s"), head_ref_name);
724 }
725 if (!strcmp(*head_points_at, head_ref_name))
726 /* detached HEAD */
727 null_is_error = 1;
728 else if (!starts_with(*head_points_at, "refs/heads/")) {
729 errors_found |= ERROR_REFS;
730 return error(_("%s points to something strange (%s)"),
731 head_ref_name, *head_points_at);
732 }
733 if (is_null_oid(head_oid)) {
734 if (null_is_error) {
735 errors_found |= ERROR_REFS;
736 return error(_("%s: detached HEAD points at nothing"),
737 head_ref_name);
738 }
739 fprintf_ln(stderr,
740 _("notice: %s points to an unborn branch (%s)"),
741 head_ref_name, *head_points_at + 11);
742 }
743 return 0;
744}
745
746static int fsck_cache_tree(struct cache_tree *it, const char *index_path)
747{
748 int i;
749 int err = 0;
750
751 if (verbose)
752 fprintf_ln(stderr, _("Checking cache tree of %s"), index_path);
753
754 if (0 <= it->entry_count) {
755 struct object *obj = parse_object(the_repository, &it->oid);
756 if (!obj) {
757 error(_("%s: invalid sha1 pointer in cache-tree of %s"),
758 oid_to_hex(&it->oid), index_path);
759 errors_found |= ERROR_REFS;
760 return 1;
761 }
762 obj->flags |= USED;
763 fsck_put_object_name(&fsck_walk_options, &it->oid, ":");
764 mark_object_reachable(obj);
765 if (obj->type != OBJ_TREE)
766 err |= objerror(obj, _("non-tree in cache-tree"));
767 }
768 for (i = 0; i < it->subtree_nr; i++)
769 err |= fsck_cache_tree(it->down[i]->cache_tree, index_path);
770 return err;
771}
772
773static int fsck_resolve_undo(struct index_state *istate,
774 const char *index_path)
775{
776 struct string_list_item *item;
777 struct string_list *resolve_undo = istate->resolve_undo;
778
779 if (!resolve_undo)
780 return 0;
781
782 for_each_string_list_item(item, resolve_undo) {
783 const char *path = item->string;
784 struct resolve_undo_info *ru = item->util;
785 int i;
786
787 if (!ru)
788 continue;
789 for (i = 0; i < 3; i++) {
790 struct object *obj;
791
792 if (!ru->mode[i] || !S_ISREG(ru->mode[i]))
793 continue;
794
795 obj = parse_object(the_repository, &ru->oid[i]);
796 if (!obj) {
797 error(_("%s: invalid sha1 pointer in resolve-undo of %s"),
798 oid_to_hex(&ru->oid[i]),
799 index_path);
800 errors_found |= ERROR_REFS;
801 continue;
802 }
803 obj->flags |= USED;
804 fsck_put_object_name(&fsck_walk_options, &ru->oid[i],
805 ":(%d):%s", i, path);
806 mark_object_reachable(obj);
807 }
808 }
809 return 0;
810}
811
812static void fsck_index(struct index_state *istate, const char *index_path,
813 int is_current_worktree)
814{
815 unsigned int i;
816
817 /* TODO: audit for interaction with sparse-index. */
818 ensure_full_index(istate);
819 for (i = 0; i < istate->cache_nr; i++) {
820 unsigned int mode;
821 struct blob *blob;
822 struct object *obj;
823
824 mode = istate->cache[i]->ce_mode;
825 if (S_ISGITLINK(mode))
826 continue;
827 blob = lookup_blob(the_repository,
828 &istate->cache[i]->oid);
829 if (!blob)
830 continue;
831 obj = &blob->object;
832 obj->flags |= USED;
833 fsck_put_object_name(&fsck_walk_options, &obj->oid,
834 "%s:%s",
835 is_current_worktree ? "" : index_path,
836 istate->cache[i]->name);
837 mark_object_reachable(obj);
838 }
839 if (istate->cache_tree)
840 fsck_cache_tree(istate->cache_tree, index_path);
841 fsck_resolve_undo(istate, index_path);
842}
843
844static void mark_object_for_connectivity(const struct object_id *oid)
845{
846 struct object *obj = lookup_unknown_object(the_repository, oid);
847 obj->flags |= HAS_OBJ;
848}
849
850static int mark_loose_for_connectivity(const struct object_id *oid,
851 const char *path UNUSED,
852 void *data UNUSED)
853{
854 mark_object_for_connectivity(oid);
855 return 0;
856}
857
858static int mark_packed_for_connectivity(const struct object_id *oid,
859 struct packed_git *pack UNUSED,
860 uint32_t pos UNUSED,
861 void *data UNUSED)
862{
863 mark_object_for_connectivity(oid);
864 return 0;
865}
866
867static int check_pack_rev_indexes(struct repository *r, int show_progress)
868{
869 struct progress *progress = NULL;
870 uint32_t pack_count = 0;
871 int res = 0;
872
873 if (show_progress) {
874 for (struct packed_git *p = get_all_packs(r); p; p = p->next)
875 pack_count++;
876 progress = start_delayed_progress(the_repository,
877 "Verifying reverse pack-indexes", pack_count);
878 pack_count = 0;
879 }
880
881 for (struct packed_git *p = get_all_packs(r); p; p = p->next) {
882 int load_error = load_pack_revindex_from_disk(p);
883
884 if (load_error < 0) {
885 error(_("unable to load rev-index for pack '%s'"), p->pack_name);
886 res = ERROR_PACK_REV_INDEX;
887 } else if (!load_error &&
888 !load_pack_revindex(r, p) &&
889 verify_pack_revindex(p)) {
890 error(_("invalid rev-index for pack '%s'"), p->pack_name);
891 res = ERROR_PACK_REV_INDEX;
892 }
893 display_progress(progress, ++pack_count);
894 }
895 stop_progress(&progress);
896
897 return res;
898}
899
900static void fsck_refs(struct repository *r)
901{
902 struct child_process refs_verify = CHILD_PROCESS_INIT;
903 struct progress *progress = NULL;
904
905 if (show_progress)
906 progress = start_progress(r, _("Checking ref database"), 1);
907
908 if (verbose)
909 fprintf_ln(stderr, _("Checking ref database"));
910
911 child_process_init(&refs_verify);
912 refs_verify.git_cmd = 1;
913 strvec_pushl(&refs_verify.args, "refs", "verify", NULL);
914 if (verbose)
915 strvec_push(&refs_verify.args, "--verbose");
916 if (check_strict)
917 strvec_push(&refs_verify.args, "--strict");
918
919 if (run_command(&refs_verify))
920 errors_found |= ERROR_REFS;
921
922 display_progress(progress, 1);
923 stop_progress(&progress);
924}
925
926static char const * const fsck_usage[] = {
927 N_("git fsck [--tags] [--root] [--unreachable] [--cache] [--no-reflogs]\n"
928 " [--[no-]full] [--strict] [--verbose] [--lost-found]\n"
929 " [--[no-]dangling] [--[no-]progress] [--connectivity-only]\n"
930 " [--[no-]name-objects] [--[no-]references] [<object>...]"),
931 NULL
932};
933
934static struct option fsck_opts[] = {
935 OPT__VERBOSE(&verbose, N_("be verbose")),
936 OPT_BOOL(0, "unreachable", &show_unreachable, N_("show unreachable objects")),
937 OPT_BOOL(0, "dangling", &show_dangling, N_("show dangling objects")),
938 OPT_BOOL(0, "tags", &show_tags, N_("report tags")),
939 OPT_BOOL(0, "root", &show_root, N_("report root nodes")),
940 OPT_BOOL(0, "cache", &keep_cache_objects, N_("make index objects head nodes")),
941 OPT_BOOL(0, "reflogs", &include_reflogs, N_("make reflogs head nodes (default)")),
942 OPT_BOOL(0, "full", &check_full, N_("also consider packs and alternate objects")),
943 OPT_BOOL(0, "connectivity-only", &connectivity_only, N_("check only connectivity")),
944 OPT_BOOL(0, "strict", &check_strict, N_("enable more strict checking")),
945 OPT_BOOL(0, "lost-found", &write_lost_and_found,
946 N_("write dangling objects in .git/lost-found")),
947 OPT_BOOL(0, "progress", &show_progress, N_("show progress")),
948 OPT_BOOL(0, "name-objects", &name_objects, N_("show verbose names for reachable objects")),
949 OPT_BOOL(0, "references", &check_references, N_("check reference database consistency")),
950 OPT_END(),
951};
952
953int cmd_fsck(int argc,
954 const char **argv,
955 const char *prefix,
956 struct repository *repo UNUSED)
957{
958 int i;
959 struct object_directory *odb;
960
961 /* fsck knows how to handle missing promisor objects */
962 fetch_if_missing = 0;
963
964 errors_found = 0;
965 disable_replace_refs();
966 save_commit_buffer = 0;
967
968 argc = parse_options(argc, argv, prefix, fsck_opts, fsck_usage, 0);
969
970 fsck_walk_options.walk = mark_object;
971 fsck_obj_options.walk = mark_used;
972 fsck_obj_options.error_func = fsck_objects_error_func;
973 if (check_strict)
974 fsck_obj_options.strict = 1;
975
976 if (show_progress == -1)
977 show_progress = isatty(2);
978 if (verbose)
979 show_progress = 0;
980
981 if (write_lost_and_found) {
982 check_full = 1;
983 include_reflogs = 0;
984 }
985
986 if (name_objects)
987 fsck_enable_object_names(&fsck_walk_options);
988
989 git_config(git_fsck_config, &fsck_obj_options);
990 prepare_repo_settings(the_repository);
991
992 if (check_references)
993 fsck_refs(the_repository);
994
995 if (connectivity_only) {
996 for_each_loose_object(mark_loose_for_connectivity, NULL, 0);
997 for_each_packed_object(the_repository,
998 mark_packed_for_connectivity, NULL, 0);
999 } else {
1000 prepare_alt_odb(the_repository);
1001 for (odb = the_repository->objects->odb; odb; odb = odb->next)
1002 fsck_object_dir(odb->path);
1003
1004 if (check_full) {
1005 struct packed_git *p;
1006 uint32_t total = 0, count = 0;
1007 struct progress *progress = NULL;
1008
1009 if (show_progress) {
1010 for (p = get_all_packs(the_repository); p;
1011 p = p->next) {
1012 if (open_pack_index(p))
1013 continue;
1014 total += p->num_objects;
1015 }
1016
1017 progress = start_progress(the_repository,
1018 _("Checking objects"), total);
1019 }
1020 for (p = get_all_packs(the_repository); p;
1021 p = p->next) {
1022 /* verify gives error messages itself */
1023 if (verify_pack(the_repository,
1024 p, fsck_obj_buffer,
1025 progress, count))
1026 errors_found |= ERROR_PACK;
1027 count += p->num_objects;
1028 }
1029 stop_progress(&progress);
1030 }
1031
1032 if (fsck_finish(&fsck_obj_options))
1033 errors_found |= ERROR_OBJECT;
1034 }
1035
1036 for (i = 0; i < argc; i++) {
1037 const char *arg = argv[i];
1038 struct object_id oid;
1039 if (!repo_get_oid(the_repository, arg, &oid)) {
1040 struct object *obj = lookup_object(the_repository,
1041 &oid);
1042
1043 if (!obj || !(obj->flags & HAS_OBJ)) {
1044 if (is_promisor_object(the_repository, &oid))
1045 continue;
1046 error(_("%s: object missing"), oid_to_hex(&oid));
1047 errors_found |= ERROR_OBJECT;
1048 continue;
1049 }
1050
1051 obj->flags |= USED;
1052 fsck_put_object_name(&fsck_walk_options, &oid,
1053 "%s", arg);
1054 mark_object_reachable(obj);
1055 continue;
1056 }
1057 error(_("invalid parameter: expected sha1, got '%s'"), arg);
1058 errors_found |= ERROR_OBJECT;
1059 }
1060
1061 /*
1062 * If we've not been given any explicit head information, do the
1063 * default ones from .git/refs. We also consider the index file
1064 * in this case (ie this implies --cache).
1065 */
1066 if (!argc) {
1067 get_default_heads();
1068 keep_cache_objects = 1;
1069 }
1070
1071 if (keep_cache_objects) {
1072 struct worktree **worktrees, **p;
1073
1074 verify_index_checksum = 1;
1075 verify_ce_order = 1;
1076
1077 worktrees = get_worktrees();
1078 for (p = worktrees; *p; p++) {
1079 struct worktree *wt = *p;
1080 struct index_state istate =
1081 INDEX_STATE_INIT(the_repository);
1082 char *path, *wt_gitdir;
1083
1084 /*
1085 * Make a copy since the buffer is reusable
1086 * and may get overwritten by other calls
1087 * while we're examining the index.
1088 */
1089 path = xstrdup(worktree_git_path(the_repository, wt, "index"));
1090 wt_gitdir = get_worktree_git_dir(wt);
1091
1092 read_index_from(&istate, path, wt_gitdir);
1093 fsck_index(&istate, path, wt->is_current);
1094
1095 discard_index(&istate);
1096 free(wt_gitdir);
1097 free(path);
1098 }
1099 free_worktrees(worktrees);
1100 }
1101
1102 errors_found |= check_pack_rev_indexes(the_repository, show_progress);
1103 if (verify_bitmap_files(the_repository))
1104 errors_found |= ERROR_BITMAP;
1105
1106 check_connectivity();
1107
1108 if (the_repository->settings.core_commit_graph) {
1109 struct child_process commit_graph_verify = CHILD_PROCESS_INIT;
1110
1111 prepare_alt_odb(the_repository);
1112 for (odb = the_repository->objects->odb; odb; odb = odb->next) {
1113 child_process_init(&commit_graph_verify);
1114 commit_graph_verify.git_cmd = 1;
1115 strvec_pushl(&commit_graph_verify.args, "commit-graph",
1116 "verify", "--object-dir", odb->path, NULL);
1117 if (show_progress)
1118 strvec_push(&commit_graph_verify.args, "--progress");
1119 else
1120 strvec_push(&commit_graph_verify.args, "--no-progress");
1121 if (run_command(&commit_graph_verify))
1122 errors_found |= ERROR_COMMIT_GRAPH;
1123 }
1124 }
1125
1126 if (the_repository->settings.core_multi_pack_index) {
1127 struct child_process midx_verify = CHILD_PROCESS_INIT;
1128
1129 prepare_alt_odb(the_repository);
1130 for (odb = the_repository->objects->odb; odb; odb = odb->next) {
1131 child_process_init(&midx_verify);
1132 midx_verify.git_cmd = 1;
1133 strvec_pushl(&midx_verify.args, "multi-pack-index",
1134 "verify", "--object-dir", odb->path, NULL);
1135 if (show_progress)
1136 strvec_push(&midx_verify.args, "--progress");
1137 else
1138 strvec_push(&midx_verify.args, "--no-progress");
1139 if (run_command(&midx_verify))
1140 errors_found |= ERROR_MULTI_PACK_INDEX;
1141 }
1142 }
1143
1144 return errors_found;
1145}