freshlybakedca.ke
Git fork
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 "odb.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 = odb_read_object_info(the_repository->objects,
75 oid, NULL);
76
77 ret = type_name(type);
78 if (!ret)
79 ret = _("unknown");
80
81 return ret;
82}
83
84static int objerror(struct object *obj, const char *err)
85{
86 errors_found |= ERROR_OBJECT;
87 /* TRANSLATORS: e.g. error in tree 01bfda: <more explanation> */
88 fprintf_ln(stderr, _("error in %s %s: %s"),
89 printable_type(&obj->oid, obj->type),
90 describe_object(&obj->oid), err);
91 return -1;
92}
93
94static int fsck_objects_error_func(struct fsck_options *o UNUSED,
95 void *fsck_report,
96 enum fsck_msg_type msg_type,
97 enum fsck_msg_id msg_id UNUSED,
98 const char *message)
99{
100 struct fsck_object_report *report = fsck_report;
101 const struct object_id *oid = report->oid;
102 enum object_type object_type = report->object_type;
103
104 switch (msg_type) {
105 case FSCK_WARN:
106 /* TRANSLATORS: e.g. warning in tree 01bfda: <more explanation> */
107 fprintf_ln(stderr, _("warning in %s %s: %s"),
108 printable_type(oid, object_type),
109 describe_object(oid), message);
110 return 0;
111 case FSCK_ERROR:
112 /* TRANSLATORS: e.g. error in tree 01bfda: <more explanation> */
113 fprintf_ln(stderr, _("error in %s %s: %s"),
114 printable_type(oid, object_type),
115 describe_object(oid), message);
116 return 1;
117 default:
118 BUG("%d (FSCK_IGNORE?) should never trigger this callback",
119 msg_type);
120 }
121}
122
123static struct object_array pending;
124
125static int mark_object(struct object *obj, enum object_type type,
126 void *data, struct fsck_options *options UNUSED)
127{
128 struct object *parent = data;
129
130 /*
131 * The only case data is NULL or type is OBJ_ANY is when
132 * mark_object_reachable() calls us. All the callers of
133 * that function has non-NULL obj hence ...
134 */
135 if (!obj) {
136 /* ... these references to parent->fld are safe here */
137 printf_ln(_("broken link from %7s %s"),
138 printable_type(&parent->oid, parent->type),
139 describe_object(&parent->oid));
140 printf_ln(_("broken link from %7s %s"),
141 (type == OBJ_ANY ? _("unknown") : type_name(type)),
142 _("unknown"));
143 errors_found |= ERROR_REACHABLE;
144 return 1;
145 }
146
147 if (type != OBJ_ANY && obj->type != type)
148 /* ... and the reference to parent is safe here */
149 objerror(parent, _("wrong object type in link"));
150
151 if (obj->flags & REACHABLE)
152 return 0;
153 obj->flags |= REACHABLE;
154
155 if (is_promisor_object(the_repository, &obj->oid))
156 /*
157 * Further recursion does not need to be performed on this
158 * object since it is a promisor object (so it does not need to
159 * be added to "pending").
160 */
161 return 0;
162
163 if (!(obj->flags & HAS_OBJ)) {
164 if (parent && !odb_has_object(the_repository->objects, &obj->oid, 1)) {
165 printf_ln(_("broken link from %7s %s\n"
166 " to %7s %s"),
167 printable_type(&parent->oid, parent->type),
168 describe_object(&parent->oid),
169 printable_type(&obj->oid, obj->type),
170 describe_object(&obj->oid));
171 errors_found |= ERROR_REACHABLE;
172 }
173 return 1;
174 }
175
176 add_object_array(obj, NULL, &pending);
177 return 0;
178}
179
180static void mark_object_reachable(struct object *obj)
181{
182 mark_object(obj, OBJ_ANY, NULL, NULL);
183}
184
185static int traverse_one_object(struct object *obj)
186{
187 int result = fsck_walk(obj, obj, &fsck_walk_options);
188
189 if (obj->type == OBJ_TREE) {
190 struct tree *tree = (struct tree *)obj;
191 free_tree_buffer(tree);
192 }
193 return result;
194}
195
196static int traverse_reachable(void)
197{
198 struct progress *progress = NULL;
199 unsigned int nr = 0;
200 int result = 0;
201 if (show_progress)
202 progress = start_delayed_progress(the_repository,
203 _("Checking connectivity"), 0);
204 while (pending.nr) {
205 result |= traverse_one_object(object_array_pop(&pending));
206 display_progress(progress, ++nr);
207 }
208 stop_progress(&progress);
209 return !!result;
210}
211
212static int mark_used(struct object *obj, enum object_type type UNUSED,
213 void *data UNUSED, struct fsck_options *options UNUSED)
214{
215 if (!obj)
216 return 1;
217 obj->flags |= USED;
218 return 0;
219}
220
221static void mark_unreachable_referents(const struct object_id *oid)
222{
223 struct fsck_options options = FSCK_OPTIONS_DEFAULT;
224 struct object *obj = lookup_object(the_repository, oid);
225
226 if (!obj || !(obj->flags & HAS_OBJ))
227 return; /* not part of our original set */
228 if (obj->flags & REACHABLE)
229 return; /* reachable objects already traversed */
230
231 /*
232 * Avoid passing OBJ_NONE to fsck_walk, which will parse the object
233 * (and we want to avoid parsing blobs).
234 */
235 if (obj->type == OBJ_NONE) {
236 enum object_type type = odb_read_object_info(the_repository->objects,
237 &obj->oid, NULL);
238 if (type > 0)
239 object_as_type(obj, type, 0);
240 }
241
242 options.walk = mark_used;
243 fsck_walk(obj, NULL, &options);
244 if (obj->type == OBJ_TREE)
245 free_tree_buffer((struct tree *)obj);
246}
247
248static int mark_loose_unreachable_referents(const struct object_id *oid,
249 const char *path UNUSED,
250 void *data UNUSED)
251{
252 mark_unreachable_referents(oid);
253 return 0;
254}
255
256static int mark_packed_unreachable_referents(const struct object_id *oid,
257 struct packed_git *pack UNUSED,
258 uint32_t pos UNUSED,
259 void *data UNUSED)
260{
261 mark_unreachable_referents(oid);
262 return 0;
263}
264
265/*
266 * Check a single reachable object
267 */
268static void check_reachable_object(struct object *obj)
269{
270 /*
271 * We obviously want the object to be parsed,
272 * except if it was in a pack-file and we didn't
273 * do a full fsck
274 */
275 if (!(obj->flags & HAS_OBJ)) {
276 if (is_promisor_object(the_repository, &obj->oid))
277 return;
278 if (has_object_pack(the_repository, &obj->oid))
279 return; /* it is in pack - forget about it */
280 printf_ln(_("missing %s %s"),
281 printable_type(&obj->oid, obj->type),
282 describe_object(&obj->oid));
283 errors_found |= ERROR_REACHABLE;
284 return;
285 }
286}
287
288/*
289 * Check a single unreachable object
290 */
291static void check_unreachable_object(struct object *obj)
292{
293 /*
294 * Missing unreachable object? Ignore it. It's not like
295 * we miss it (since it can't be reached), nor do we want
296 * to complain about it being unreachable (since it does
297 * not exist).
298 */
299 if (!(obj->flags & HAS_OBJ))
300 return;
301
302 /*
303 * Unreachable object that exists? Show it if asked to,
304 * since this is something that is prunable.
305 */
306 if (show_unreachable) {
307 printf_ln(_("unreachable %s %s"),
308 printable_type(&obj->oid, obj->type),
309 describe_object(&obj->oid));
310 return;
311 }
312
313 /*
314 * "!USED" means that nothing at all points to it, including
315 * other unreachable objects. In other words, it's the "tip"
316 * of some set of unreachable objects, usually a commit that
317 * got dropped.
318 *
319 * Such starting points are more interesting than some random
320 * set of unreachable objects, so we show them even if the user
321 * hasn't asked for _all_ unreachable objects. If you have
322 * deleted a branch by mistake, this is a prime candidate to
323 * start looking at, for example.
324 */
325 if (!(obj->flags & USED)) {
326 if (show_dangling)
327 printf_ln(_("dangling %s %s"),
328 printable_type(&obj->oid, obj->type),
329 describe_object(&obj->oid));
330 if (write_lost_and_found) {
331 char *filename = repo_git_path(the_repository, "lost-found/%s/%s",
332 obj->type == OBJ_COMMIT ? "commit" : "other",
333 describe_object(&obj->oid));
334 FILE *f;
335
336 if (safe_create_leading_directories_const(the_repository, filename)) {
337 error(_("could not create lost-found"));
338 free(filename);
339 return;
340 }
341 f = xfopen(filename, "w");
342 if (obj->type == OBJ_BLOB) {
343 if (stream_blob_to_fd(fileno(f), &obj->oid, NULL, 1))
344 die_errno(_("could not write '%s'"), filename);
345 } else
346 fprintf(f, "%s\n", describe_object(&obj->oid));
347 if (fclose(f))
348 die_errno(_("could not finish '%s'"),
349 filename);
350 free(filename);
351 }
352 return;
353 }
354
355 /*
356 * Otherwise? It's there, it's unreachable, and some other unreachable
357 * object points to it. Ignore it - it's not interesting, and we showed
358 * all the interesting cases above.
359 */
360}
361
362static void check_object(struct object *obj)
363{
364 if (verbose)
365 fprintf_ln(stderr, _("Checking %s"), describe_object(&obj->oid));
366
367 if (obj->flags & REACHABLE)
368 check_reachable_object(obj);
369 else
370 check_unreachable_object(obj);
371}
372
373static void check_connectivity(void)
374{
375 int i, max;
376
377 /* Traverse the pending reachable objects */
378 traverse_reachable();
379
380 /*
381 * With --connectivity-only, we won't have actually opened and marked
382 * unreachable objects with USED. Do that now to make --dangling, etc
383 * accurate.
384 */
385 if (connectivity_only && (show_dangling || write_lost_and_found)) {
386 /*
387 * Even though we already have a "struct object" for each of
388 * these in memory, we must not iterate over the internal
389 * object hash as we do below. Our loop would potentially
390 * resize the hash, making our iteration invalid.
391 *
392 * Instead, we'll just go back to the source list of objects,
393 * and ignore any that weren't present in our earlier
394 * traversal.
395 */
396 for_each_loose_object(the_repository->objects,
397 mark_loose_unreachable_referents, NULL, 0);
398 for_each_packed_object(the_repository,
399 mark_packed_unreachable_referents,
400 NULL,
401 0);
402 }
403
404 /* Look up all the requirements, warn about missing objects.. */
405 max = get_max_object_index(the_repository);
406 if (verbose)
407 fprintf_ln(stderr, _("Checking connectivity (%d objects)"), max);
408
409 for (i = 0; i < max; i++) {
410 struct object *obj = get_indexed_object(the_repository, i);
411
412 if (obj)
413 check_object(obj);
414 }
415}
416
417static int fsck_obj(struct object *obj, void *buffer, unsigned long size)
418{
419 int err;
420
421 if (obj->flags & SEEN)
422 return 0;
423 obj->flags |= SEEN;
424
425 if (verbose)
426 fprintf_ln(stderr, _("Checking %s %s"),
427 printable_type(&obj->oid, obj->type),
428 describe_object(&obj->oid));
429
430 if (fsck_walk(obj, NULL, &fsck_obj_options))
431 objerror(obj, _("broken links"));
432 err = fsck_object(obj, buffer, size, &fsck_obj_options);
433 if (err)
434 goto out;
435
436 if (obj->type == OBJ_COMMIT) {
437 struct commit *commit = (struct commit *) obj;
438
439 if (!commit->parents && show_root)
440 printf_ln(_("root %s"),
441 describe_object(&commit->object.oid));
442 }
443
444 if (obj->type == OBJ_TAG) {
445 struct tag *tag = (struct tag *) obj;
446
447 if (show_tags && tag->tagged) {
448 printf_ln(_("tagged %s %s (%s) in %s"),
449 printable_type(&tag->tagged->oid, tag->tagged->type),
450 describe_object(&tag->tagged->oid),
451 tag->tag,
452 describe_object(&tag->object.oid));
453 }
454 }
455
456out:
457 if (obj->type == OBJ_TREE)
458 free_tree_buffer((struct tree *)obj);
459 return err;
460}
461
462static int fsck_obj_buffer(const struct object_id *oid, enum object_type type,
463 unsigned long size, void *buffer, int *eaten)
464{
465 /*
466 * Note, buffer may be NULL if type is OBJ_BLOB. See
467 * verify_packfile(), data_valid variable for details.
468 */
469 struct object *obj;
470 obj = parse_object_buffer(the_repository, oid, type, size, buffer,
471 eaten);
472 if (!obj) {
473 errors_found |= ERROR_OBJECT;
474 return error(_("%s: object corrupt or missing"),
475 oid_to_hex(oid));
476 }
477 obj->flags &= ~(REACHABLE | SEEN);
478 obj->flags |= HAS_OBJ;
479 return fsck_obj(obj, buffer, size);
480}
481
482static int default_refs;
483
484static void fsck_handle_reflog_oid(const char *refname, struct object_id *oid,
485 timestamp_t timestamp)
486{
487 struct object *obj;
488
489 if (!is_null_oid(oid)) {
490 obj = lookup_object(the_repository, oid);
491 if (obj && (obj->flags & HAS_OBJ)) {
492 if (timestamp)
493 fsck_put_object_name(&fsck_walk_options, oid,
494 "%s@{%"PRItime"}",
495 refname, timestamp);
496 obj->flags |= USED;
497 mark_object_reachable(obj);
498 } else if (!is_promisor_object(the_repository, oid)) {
499 error(_("%s: invalid reflog entry %s"),
500 refname, oid_to_hex(oid));
501 errors_found |= ERROR_REACHABLE;
502 }
503 }
504}
505
506static int fsck_handle_reflog_ent(const char *refname,
507 struct object_id *ooid, struct object_id *noid,
508 const char *email UNUSED,
509 timestamp_t timestamp, int tz UNUSED,
510 const char *message UNUSED, void *cb_data UNUSED)
511{
512 if (verbose)
513 fprintf_ln(stderr, _("Checking reflog %s->%s"),
514 oid_to_hex(ooid), oid_to_hex(noid));
515
516 fsck_handle_reflog_oid(refname, ooid, 0);
517 fsck_handle_reflog_oid(refname, noid, timestamp);
518 return 0;
519}
520
521static int fsck_handle_reflog(const char *logname, void *cb_data)
522{
523 struct strbuf refname = STRBUF_INIT;
524
525 strbuf_worktree_ref(cb_data, &refname, logname);
526 refs_for_each_reflog_ent(get_main_ref_store(the_repository),
527 refname.buf, fsck_handle_reflog_ent,
528 NULL);
529 strbuf_release(&refname);
530 return 0;
531}
532
533static int fsck_handle_ref(const char *refname, const char *referent UNUSED, const struct object_id *oid,
534 int flag UNUSED, void *cb_data UNUSED)
535{
536 struct object *obj;
537
538 obj = parse_object(the_repository, oid);
539 if (!obj) {
540 if (is_promisor_object(the_repository, oid)) {
541 /*
542 * Increment default_refs anyway, because this is a
543 * valid ref.
544 */
545 default_refs++;
546 return 0;
547 }
548 error(_("%s: invalid sha1 pointer %s"),
549 refname, oid_to_hex(oid));
550 errors_found |= ERROR_REACHABLE;
551 /* We'll continue with the rest despite the error.. */
552 return 0;
553 }
554 if (obj->type != OBJ_COMMIT && is_branch(refname)) {
555 error(_("%s: not a commit"), refname);
556 errors_found |= ERROR_REFS;
557 }
558 default_refs++;
559 obj->flags |= USED;
560 fsck_put_object_name(&fsck_walk_options,
561 oid, "%s", refname);
562 mark_object_reachable(obj);
563
564 return 0;
565}
566
567static int fsck_head_link(const char *head_ref_name,
568 const char **head_points_at,
569 struct object_id *head_oid);
570
571static void get_default_heads(void)
572{
573 struct worktree **worktrees, **p;
574 const char *head_points_at;
575 struct object_id head_oid;
576
577 refs_for_each_rawref(get_main_ref_store(the_repository),
578 fsck_handle_ref, NULL);
579
580 worktrees = get_worktrees();
581 for (p = worktrees; *p; p++) {
582 struct worktree *wt = *p;
583 struct strbuf ref = STRBUF_INIT;
584
585 strbuf_worktree_ref(wt, &ref, "HEAD");
586 fsck_head_link(ref.buf, &head_points_at, &head_oid);
587 if (head_points_at && !is_null_oid(&head_oid))
588 fsck_handle_ref(ref.buf, NULL, &head_oid, 0, NULL);
589 strbuf_release(&ref);
590
591 if (include_reflogs)
592 refs_for_each_reflog(get_worktree_ref_store(wt),
593 fsck_handle_reflog, wt);
594 }
595 free_worktrees(worktrees);
596
597 /*
598 * Not having any default heads isn't really fatal, but
599 * it does mean that "--unreachable" no longer makes any
600 * sense (since in this case everything will obviously
601 * be unreachable by definition.
602 *
603 * Showing dangling objects is valid, though (as those
604 * dangling objects are likely lost heads).
605 *
606 * So we just print a warning about it, and clear the
607 * "show_unreachable" flag.
608 */
609 if (!default_refs) {
610 fprintf_ln(stderr, _("notice: No default references"));
611 show_unreachable = 0;
612 }
613}
614
615struct for_each_loose_cb
616{
617 struct progress *progress;
618};
619
620static int fsck_loose(const struct object_id *oid, const char *path,
621 void *data UNUSED)
622{
623 struct object *obj;
624 enum object_type type = OBJ_NONE;
625 unsigned long size;
626 void *contents = NULL;
627 int eaten;
628 struct object_info oi = OBJECT_INFO_INIT;
629 struct object_id real_oid = *null_oid(the_hash_algo);
630 int err = 0;
631
632 oi.sizep = &size;
633 oi.typep = &type;
634
635 if (read_loose_object(the_repository, path, oid, &real_oid, &contents, &oi) < 0) {
636 if (contents && !oideq(&real_oid, oid))
637 err = error(_("%s: hash-path mismatch, found at: %s"),
638 oid_to_hex(&real_oid), path);
639 else
640 err = error(_("%s: object corrupt or missing: %s"),
641 oid_to_hex(oid), path);
642 }
643 if (err < 0) {
644 errors_found |= ERROR_OBJECT;
645 free(contents);
646 return 0; /* keep checking other objects */
647 }
648
649 if (!contents && type != OBJ_BLOB)
650 BUG("read_loose_object streamed a non-blob");
651
652 obj = parse_object_buffer(the_repository, oid, type, size,
653 contents, &eaten);
654
655 if (!obj) {
656 errors_found |= ERROR_OBJECT;
657 error(_("%s: object could not be parsed: %s"),
658 oid_to_hex(oid), path);
659 if (!eaten)
660 free(contents);
661 return 0; /* keep checking other objects */
662 }
663
664 obj->flags &= ~(REACHABLE | SEEN);
665 obj->flags |= HAS_OBJ;
666 if (fsck_obj(obj, contents, size))
667 errors_found |= ERROR_OBJECT;
668
669 if (!eaten)
670 free(contents);
671 return 0; /* keep checking other objects, even if we saw an error */
672}
673
674static int fsck_cruft(const char *basename, const char *path,
675 void *data UNUSED)
676{
677 if (!starts_with(basename, "tmp_obj_"))
678 fprintf_ln(stderr, _("bad sha1 file: %s"), path);
679 return 0;
680}
681
682static int fsck_subdir(unsigned int nr, const char *path UNUSED, void *data)
683{
684 struct for_each_loose_cb *cb_data = data;
685 struct progress *progress = cb_data->progress;
686 display_progress(progress, nr + 1);
687 return 0;
688}
689
690static void fsck_source(struct odb_source *source)
691{
692 struct progress *progress = NULL;
693 struct for_each_loose_cb cb_data = {
694 .progress = progress,
695 };
696
697 if (verbose)
698 fprintf_ln(stderr, _("Checking object directory"));
699
700 if (show_progress)
701 progress = start_progress(the_repository,
702 _("Checking object directories"), 256);
703
704 for_each_loose_file_in_source(source, fsck_loose,
705 fsck_cruft, fsck_subdir, &cb_data);
706 display_progress(progress, 256);
707 stop_progress(&progress);
708}
709
710static int fsck_head_link(const char *head_ref_name,
711 const char **head_points_at,
712 struct object_id *head_oid)
713{
714 int null_is_error = 0;
715
716 if (verbose)
717 fprintf_ln(stderr, _("Checking %s link"), head_ref_name);
718
719 *head_points_at = refs_resolve_ref_unsafe(get_main_ref_store(the_repository),
720 head_ref_name, 0, head_oid,
721 NULL);
722 if (!*head_points_at) {
723 errors_found |= ERROR_REFS;
724 return error(_("invalid %s"), head_ref_name);
725 }
726 if (!strcmp(*head_points_at, head_ref_name))
727 /* detached HEAD */
728 null_is_error = 1;
729 else if (!starts_with(*head_points_at, "refs/heads/")) {
730 errors_found |= ERROR_REFS;
731 return error(_("%s points to something strange (%s)"),
732 head_ref_name, *head_points_at);
733 }
734 if (is_null_oid(head_oid)) {
735 if (null_is_error) {
736 errors_found |= ERROR_REFS;
737 return error(_("%s: detached HEAD points at nothing"),
738 head_ref_name);
739 }
740 fprintf_ln(stderr,
741 _("notice: %s points to an unborn branch (%s)"),
742 head_ref_name, *head_points_at + 11);
743 }
744 return 0;
745}
746
747static int fsck_cache_tree(struct cache_tree *it, const char *index_path)
748{
749 int i;
750 int err = 0;
751
752 if (verbose)
753 fprintf_ln(stderr, _("Checking cache tree of %s"), index_path);
754
755 if (0 <= it->entry_count) {
756 struct object *obj = parse_object(the_repository, &it->oid);
757 if (!obj) {
758 error(_("%s: invalid sha1 pointer in cache-tree of %s"),
759 oid_to_hex(&it->oid), index_path);
760 errors_found |= ERROR_REFS;
761 return 1;
762 }
763 obj->flags |= USED;
764 fsck_put_object_name(&fsck_walk_options, &it->oid, ":");
765 mark_object_reachable(obj);
766 if (obj->type != OBJ_TREE)
767 err |= objerror(obj, _("non-tree in cache-tree"));
768 }
769 for (i = 0; i < it->subtree_nr; i++)
770 err |= fsck_cache_tree(it->down[i]->cache_tree, index_path);
771 return err;
772}
773
774static int fsck_resolve_undo(struct index_state *istate,
775 const char *index_path)
776{
777 struct string_list_item *item;
778 struct string_list *resolve_undo = istate->resolve_undo;
779
780 if (!resolve_undo)
781 return 0;
782
783 for_each_string_list_item(item, resolve_undo) {
784 const char *path = item->string;
785 struct resolve_undo_info *ru = item->util;
786 int i;
787
788 if (!ru)
789 continue;
790 for (i = 0; i < 3; i++) {
791 struct object *obj;
792
793 if (!ru->mode[i] || !S_ISREG(ru->mode[i]))
794 continue;
795
796 obj = parse_object(the_repository, &ru->oid[i]);
797 if (!obj) {
798 error(_("%s: invalid sha1 pointer in resolve-undo of %s"),
799 oid_to_hex(&ru->oid[i]),
800 index_path);
801 errors_found |= ERROR_REFS;
802 continue;
803 }
804 obj->flags |= USED;
805 fsck_put_object_name(&fsck_walk_options, &ru->oid[i],
806 ":(%d):%s", i, path);
807 mark_object_reachable(obj);
808 }
809 }
810 return 0;
811}
812
813static void fsck_index(struct index_state *istate, const char *index_path,
814 int is_current_worktree)
815{
816 unsigned int i;
817
818 /* TODO: audit for interaction with sparse-index. */
819 ensure_full_index(istate);
820 for (i = 0; i < istate->cache_nr; i++) {
821 unsigned int mode;
822 struct blob *blob;
823 struct object *obj;
824
825 mode = istate->cache[i]->ce_mode;
826 if (S_ISGITLINK(mode))
827 continue;
828 blob = lookup_blob(the_repository,
829 &istate->cache[i]->oid);
830 if (!blob)
831 continue;
832 obj = &blob->object;
833 obj->flags |= USED;
834 fsck_put_object_name(&fsck_walk_options, &obj->oid,
835 "%s:%s",
836 is_current_worktree ? "" : index_path,
837 istate->cache[i]->name);
838 mark_object_reachable(obj);
839 }
840 if (istate->cache_tree)
841 fsck_cache_tree(istate->cache_tree, index_path);
842 fsck_resolve_undo(istate, index_path);
843}
844
845static void mark_object_for_connectivity(const struct object_id *oid)
846{
847 struct object *obj = lookup_unknown_object(the_repository, oid);
848 obj->flags |= HAS_OBJ;
849}
850
851static int mark_loose_for_connectivity(const struct object_id *oid,
852 const char *path UNUSED,
853 void *data UNUSED)
854{
855 mark_object_for_connectivity(oid);
856 return 0;
857}
858
859static int mark_packed_for_connectivity(const struct object_id *oid,
860 struct packed_git *pack UNUSED,
861 uint32_t pos UNUSED,
862 void *data UNUSED)
863{
864 mark_object_for_connectivity(oid);
865 return 0;
866}
867
868static int check_pack_rev_indexes(struct repository *r, int show_progress)
869{
870 struct packfile_store *packs = r->objects->packfiles;
871 struct progress *progress = NULL;
872 uint32_t pack_count = 0;
873 int res = 0;
874
875 if (show_progress) {
876 for (struct packed_git *p = packfile_store_get_all_packs(packs); p; p = p->next)
877 pack_count++;
878 progress = start_delayed_progress(the_repository,
879 "Verifying reverse pack-indexes", pack_count);
880 pack_count = 0;
881 }
882
883 for (struct packed_git *p = packfile_store_get_all_packs(packs); p; p = p->next) {
884 int load_error = load_pack_revindex_from_disk(p);
885
886 if (load_error < 0) {
887 error(_("unable to load rev-index for pack '%s'"), p->pack_name);
888 res = ERROR_PACK_REV_INDEX;
889 } else if (!load_error &&
890 !load_pack_revindex(r, p) &&
891 verify_pack_revindex(p)) {
892 error(_("invalid rev-index for pack '%s'"), p->pack_name);
893 res = ERROR_PACK_REV_INDEX;
894 }
895 display_progress(progress, ++pack_count);
896 }
897 stop_progress(&progress);
898
899 return res;
900}
901
902static void fsck_refs(struct repository *r)
903{
904 struct child_process refs_verify = CHILD_PROCESS_INIT;
905 struct progress *progress = NULL;
906
907 if (show_progress)
908 progress = start_progress(r, _("Checking ref database"), 1);
909
910 if (verbose)
911 fprintf_ln(stderr, _("Checking ref database"));
912
913 child_process_init(&refs_verify);
914 refs_verify.git_cmd = 1;
915 strvec_pushl(&refs_verify.args, "refs", "verify", NULL);
916 if (verbose)
917 strvec_push(&refs_verify.args, "--verbose");
918 if (check_strict)
919 strvec_push(&refs_verify.args, "--strict");
920
921 if (run_command(&refs_verify))
922 errors_found |= ERROR_REFS;
923
924 display_progress(progress, 1);
925 stop_progress(&progress);
926}
927
928static char const * const fsck_usage[] = {
929 N_("git fsck [--tags] [--root] [--unreachable] [--cache] [--no-reflogs]\n"
930 " [--[no-]full] [--strict] [--verbose] [--lost-found]\n"
931 " [--[no-]dangling] [--[no-]progress] [--connectivity-only]\n"
932 " [--[no-]name-objects] [--[no-]references] [<object>...]"),
933 NULL
934};
935
936static struct option fsck_opts[] = {
937 OPT__VERBOSE(&verbose, N_("be verbose")),
938 OPT_BOOL(0, "unreachable", &show_unreachable, N_("show unreachable objects")),
939 OPT_BOOL(0, "dangling", &show_dangling, N_("show dangling objects")),
940 OPT_BOOL(0, "tags", &show_tags, N_("report tags")),
941 OPT_BOOL(0, "root", &show_root, N_("report root nodes")),
942 OPT_BOOL(0, "cache", &keep_cache_objects, N_("make index objects head nodes")),
943 OPT_BOOL(0, "reflogs", &include_reflogs, N_("make reflogs head nodes (default)")),
944 OPT_BOOL(0, "full", &check_full, N_("also consider packs and alternate objects")),
945 OPT_BOOL(0, "connectivity-only", &connectivity_only, N_("check only connectivity")),
946 OPT_BOOL(0, "strict", &check_strict, N_("enable more strict checking")),
947 OPT_BOOL(0, "lost-found", &write_lost_and_found,
948 N_("write dangling objects in .git/lost-found")),
949 OPT_BOOL(0, "progress", &show_progress, N_("show progress")),
950 OPT_BOOL(0, "name-objects", &name_objects, N_("show verbose names for reachable objects")),
951 OPT_BOOL(0, "references", &check_references, N_("check reference database consistency")),
952 OPT_END(),
953};
954
955int cmd_fsck(int argc,
956 const char **argv,
957 const char *prefix,
958 struct repository *repo UNUSED)
959{
960 int i;
961 struct odb_source *source;
962
963 /* fsck knows how to handle missing promisor objects */
964 fetch_if_missing = 0;
965
966 errors_found = 0;
967 disable_replace_refs();
968 save_commit_buffer = 0;
969
970 argc = parse_options(argc, argv, prefix, fsck_opts, fsck_usage, 0);
971
972 fsck_walk_options.walk = mark_object;
973 fsck_obj_options.walk = mark_used;
974 fsck_obj_options.error_func = fsck_objects_error_func;
975 if (check_strict)
976 fsck_obj_options.strict = 1;
977
978 if (show_progress == -1)
979 show_progress = isatty(2);
980 if (verbose)
981 show_progress = 0;
982
983 if (write_lost_and_found) {
984 check_full = 1;
985 include_reflogs = 0;
986 }
987
988 if (name_objects)
989 fsck_enable_object_names(&fsck_walk_options);
990
991 repo_config(the_repository, git_fsck_config, &fsck_obj_options);
992 prepare_repo_settings(the_repository);
993
994 if (check_references)
995 fsck_refs(the_repository);
996
997 if (connectivity_only) {
998 for_each_loose_object(the_repository->objects,
999 mark_loose_for_connectivity, NULL, 0);
1000 for_each_packed_object(the_repository,
1001 mark_packed_for_connectivity, NULL, 0);
1002 } else {
1003 struct packfile_store *packs = the_repository->objects->packfiles;
1004
1005 odb_prepare_alternates(the_repository->objects);
1006 for (source = the_repository->objects->sources; source; source = source->next)
1007 fsck_source(source);
1008
1009 if (check_full) {
1010 struct packed_git *p;
1011 uint32_t total = 0, count = 0;
1012 struct progress *progress = NULL;
1013
1014 if (show_progress) {
1015 for (p = packfile_store_get_all_packs(packs); p;
1016 p = p->next) {
1017 if (open_pack_index(p))
1018 continue;
1019 total += p->num_objects;
1020 }
1021
1022 progress = start_progress(the_repository,
1023 _("Checking objects"), total);
1024 }
1025 for (p = packfile_store_get_all_packs(packs); p;
1026 p = p->next) {
1027 /* verify gives error messages itself */
1028 if (verify_pack(the_repository,
1029 p, fsck_obj_buffer,
1030 progress, count))
1031 errors_found |= ERROR_PACK;
1032 count += p->num_objects;
1033 }
1034 stop_progress(&progress);
1035 }
1036
1037 if (fsck_finish(&fsck_obj_options))
1038 errors_found |= ERROR_OBJECT;
1039 }
1040
1041 for (i = 0; i < argc; i++) {
1042 const char *arg = argv[i];
1043 struct object_id oid;
1044 if (!repo_get_oid(the_repository, arg, &oid)) {
1045 struct object *obj = lookup_object(the_repository,
1046 &oid);
1047
1048 if (!obj || !(obj->flags & HAS_OBJ)) {
1049 if (is_promisor_object(the_repository, &oid))
1050 continue;
1051 error(_("%s: object missing"), oid_to_hex(&oid));
1052 errors_found |= ERROR_OBJECT;
1053 continue;
1054 }
1055
1056 obj->flags |= USED;
1057 fsck_put_object_name(&fsck_walk_options, &oid,
1058 "%s", arg);
1059 mark_object_reachable(obj);
1060 continue;
1061 }
1062 error(_("invalid parameter: expected sha1, got '%s'"), arg);
1063 errors_found |= ERROR_OBJECT;
1064 }
1065
1066 /*
1067 * If we've not been given any explicit head information, do the
1068 * default ones from .git/refs. We also consider the index file
1069 * in this case (ie this implies --cache).
1070 */
1071 if (!argc) {
1072 get_default_heads();
1073 keep_cache_objects = 1;
1074 }
1075
1076 if (keep_cache_objects) {
1077 struct worktree **worktrees, **p;
1078
1079 verify_index_checksum = 1;
1080 verify_ce_order = 1;
1081
1082 worktrees = get_worktrees();
1083 for (p = worktrees; *p; p++) {
1084 struct worktree *wt = *p;
1085 struct index_state istate =
1086 INDEX_STATE_INIT(the_repository);
1087 char *path, *wt_gitdir;
1088
1089 /*
1090 * Make a copy since the buffer is reusable
1091 * and may get overwritten by other calls
1092 * while we're examining the index.
1093 */
1094 path = xstrdup(worktree_git_path(the_repository, wt, "index"));
1095 wt_gitdir = get_worktree_git_dir(wt);
1096
1097 read_index_from(&istate, path, wt_gitdir);
1098 fsck_index(&istate, path, wt->is_current);
1099
1100 discard_index(&istate);
1101 free(wt_gitdir);
1102 free(path);
1103 }
1104 free_worktrees(worktrees);
1105 }
1106
1107 errors_found |= check_pack_rev_indexes(the_repository, show_progress);
1108 if (verify_bitmap_files(the_repository))
1109 errors_found |= ERROR_BITMAP;
1110
1111 check_connectivity();
1112
1113 if (the_repository->settings.core_commit_graph) {
1114 struct child_process commit_graph_verify = CHILD_PROCESS_INIT;
1115
1116 odb_prepare_alternates(the_repository->objects);
1117 for (source = the_repository->objects->sources; source; source = source->next) {
1118 child_process_init(&commit_graph_verify);
1119 commit_graph_verify.git_cmd = 1;
1120 strvec_pushl(&commit_graph_verify.args, "commit-graph",
1121 "verify", "--object-dir", source->path, NULL);
1122 if (show_progress)
1123 strvec_push(&commit_graph_verify.args, "--progress");
1124 else
1125 strvec_push(&commit_graph_verify.args, "--no-progress");
1126 if (run_command(&commit_graph_verify))
1127 errors_found |= ERROR_COMMIT_GRAPH;
1128 }
1129 }
1130
1131 if (the_repository->settings.core_multi_pack_index) {
1132 struct child_process midx_verify = CHILD_PROCESS_INIT;
1133
1134 odb_prepare_alternates(the_repository->objects);
1135 for (source = the_repository->objects->sources; source; source = source->next) {
1136 child_process_init(&midx_verify);
1137 midx_verify.git_cmd = 1;
1138 strvec_pushl(&midx_verify.args, "multi-pack-index",
1139 "verify", "--object-dir", source->path, NULL);
1140 if (show_progress)
1141 strvec_push(&midx_verify.args, "--progress");
1142 else
1143 strvec_push(&midx_verify.args, "--no-progress");
1144 if (run_command(&midx_verify))
1145 errors_found |= ERROR_MULTI_PACK_INDEX;
1146 }
1147 }
1148
1149 return errors_found;
1150}