git-fsck-objects: Free tree entries after use
[git/git.git] / fsck-objects.c
CommitLineData
20222118
LT
1#include <sys/types.h>
2#include <dirent.h>
3
4b182421 4#include "cache.h"
ff5ebe39
DB
5#include "commit.h"
6#include "tree.h"
7#include "blob.h"
c418eda4 8#include "tag.h"
944d8589 9#include "refs.h"
f9253394 10#include "pack.h"
ff5ebe39
DB
11
12#define REACHABLE 0x0001
d9839e03 13
ab7df187 14static int show_root = 0;
889262ea 15static int show_tags = 0;
d9839e03 16static int show_unreachable = 0;
8a498a05
JH
17static int standalone = 0;
18static int check_full = 0;
de2eb7f6 19static int check_strict = 0;
ae7c0c92 20static int keep_cache_objects = 0;
d9839e03
LT
21static unsigned char head_sha1[20];
22
f1f0d088
PB
23
24static void objreport(struct object *obj, const char *severity,
25 const char *err, va_list params)
26{
27 fprintf(stderr, "%s in %s %s: ",
28 severity, obj->type, sha1_to_hex(obj->sha1));
29 vfprintf(stderr, err, params);
30 fputs("\n", stderr);
31}
32
a7928f8e 33static int objerror(struct object *obj, const char *err, ...)
f1f0d088
PB
34{
35 va_list params;
36 va_start(params, err);
37 objreport(obj, "error", err, params);
38 va_end(params);
39 return -1;
40}
41
a7928f8e 42static int objwarning(struct object *obj, const char *err, ...)
f1f0d088
PB
43{
44 va_list params;
45 va_start(params, err);
46 objreport(obj, "warning", err, params);
47 va_end(params);
48 return -1;
49}
50
51
8ba0bbb2
LT
52static void check_connectivity(void)
53{
54 int i;
55
8ba0bbb2 56 /* Look up all the requirements, warn about missing objects.. */
ff5ebe39
DB
57 for (i = 0; i < nr_objs; i++) {
58 struct object *obj = objs[i];
c418eda4 59 struct object_list *refs;
8ba0bbb2 60
3a6a23e6 61 if (!obj->parsed) {
8a498a05
JH
62 if (!standalone && has_sha1_file(obj->sha1))
63 ; /* it is in pack */
64 else
65 printf("missing %s %s\n",
66 obj->type, sha1_to_hex(obj->sha1));
3a6a23e6
LT
67 continue;
68 }
69
70 for (refs = obj->refs; refs; refs = refs->next) {
8a498a05
JH
71 if (refs->item->parsed ||
72 (!standalone && has_sha1_file(refs->item->sha1)))
3a6a23e6
LT
73 continue;
74 printf("broken link from %7s %s\n",
75 obj->type, sha1_to_hex(obj->sha1));
76 printf(" to %7s %s\n",
aa034134 77 refs->item->type, sha1_to_hex(refs->item->sha1));
3a6a23e6
LT
78 }
79
ff5ebe39 80 if (show_unreachable && !(obj->flags & REACHABLE)) {
c4584ae3
JH
81 printf("unreachable %s %s\n",
82 obj->type, sha1_to_hex(obj->sha1));
8ba0bbb2 83 continue;
d9839e03 84 }
8ba0bbb2 85
ff5ebe39
DB
86 if (!obj->used) {
87 printf("dangling %s %s\n", obj->type,
88 sha1_to_hex(obj->sha1));
d9839e03 89 }
8ba0bbb2
LT
90 }
91}
92
85003492
LT
93/*
94 * The entries in a tree are ordered in the _path_ order,
95 * which means that a directory entry is ordered by adding
96 * a slash to the end of it.
97 *
98 * So a directory called "a" is ordered _after_ a file
99 * called "a.c", because "a/" sorts after "a.c".
100 */
a4f35a2d
JH
101#define TREE_UNORDERED (-1)
102#define TREE_HAS_DUPS (-2)
103
85003492
LT
104static int verify_ordered(struct tree_entry_list *a, struct tree_entry_list *b)
105{
106 int len1 = strlen(a->name);
107 int len2 = strlen(b->name);
108 int len = len1 < len2 ? len1 : len2;
109 unsigned char c1, c2;
110 int cmp;
111
112 cmp = memcmp(a->name, b->name, len);
113 if (cmp < 0)
114 return 0;
115 if (cmp > 0)
a4f35a2d 116 return TREE_UNORDERED;
85003492
LT
117
118 /*
119 * Ok, the first <len> characters are the same.
120 * Now we need to order the next one, but turn
121 * a '\0' into a '/' for a directory entry.
122 */
123 c1 = a->name[len];
124 c2 = b->name[len];
a4f35a2d
JH
125 if (!c1 && !c2)
126 /*
127 * git-write-tree used to write out a nonsense tree that has
128 * entries with the same name, one blob and one tree. Make
129 * sure we do not have duplicate entries.
130 */
131 return TREE_HAS_DUPS;
85003492
LT
132 if (!c1 && a->directory)
133 c1 = '/';
134 if (!c2 && b->directory)
135 c2 = '/';
a4f35a2d 136 return c1 < c2 ? 0 : TREE_UNORDERED;
85003492
LT
137}
138
c418eda4 139static int fsck_tree(struct tree *item)
20222118 140{
64071805 141 int retval;
85003492 142 int has_full_path = 0;
64071805
LT
143 int has_zero_pad = 0;
144 int has_bad_modes = 0;
145 int has_dup_entries = 0;
146 int not_properly_sorted = 0;
85003492
LT
147 struct tree_entry_list *entry, *last;
148
149 last = NULL;
150 for (entry = item->entries; entry; entry = entry->next) {
151 if (strchr(entry->name, '/'))
152 has_full_path = 1;
64071805 153 has_zero_pad |= entry->zeropad;
85003492 154
42ea9cb2
LT
155 switch (entry->mode) {
156 /*
157 * Standard modes..
158 */
159 case S_IFREG | 0755:
160 case S_IFREG | 0644:
161 case S_IFLNK:
162 case S_IFDIR:
163 break;
164 /*
165 * This is nonstandard, but we had a few of these
166 * early on when we honored the full set of mode
167 * bits..
168 */
169 case S_IFREG | 0664:
de2eb7f6
LT
170 if (!check_strict)
171 break;
42ea9cb2 172 default:
64071805 173 has_bad_modes = 1;
42ea9cb2
LT
174 }
175
85003492 176 if (last) {
a4f35a2d
JH
177 switch (verify_ordered(last, entry)) {
178 case TREE_UNORDERED:
64071805
LT
179 not_properly_sorted = 1;
180 break;
a4f35a2d 181 case TREE_HAS_DUPS:
64071805
LT
182 has_dup_entries = 1;
183 break;
a4f35a2d
JH
184 default:
185 break;
85003492 186 }
545f229a
SV
187 free(last->name);
188 free(last);
85003492
LT
189 }
190
191 last = entry;
192 }
545f229a
SV
193 if (last) {
194 free(last->name);
195 free(last);
196 }
197 item->entries = NULL;
85003492 198
64071805 199 retval = 0;
85003492 200 if (has_full_path) {
f1f0d088 201 objwarning(&item->object, "contains full pathnames");
1ea34e36 202 }
64071805 203 if (has_zero_pad) {
f1f0d088 204 objwarning(&item->object, "contains zero-padded file modes");
64071805
LT
205 }
206 if (has_bad_modes) {
f1f0d088 207 objwarning(&item->object, "contains bad file modes");
64071805
LT
208 }
209 if (has_dup_entries) {
f1f0d088 210 retval = objerror(&item->object, "contains duplicate file entries");
64071805
LT
211 }
212 if (not_properly_sorted) {
f1f0d088 213 retval = objerror(&item->object, "not properly sorted");
64071805
LT
214 }
215 return retval;
1ea34e36
LT
216}
217
c418eda4 218static int fsck_commit(struct commit *commit)
1ea34e36 219{
de2eb7f6 220 char *buffer = commit->buffer;
f1f0d088 221 unsigned char tree_sha1[20], sha1[20];
de2eb7f6
LT
222
223 if (memcmp(buffer, "tree ", 5))
f1f0d088
PB
224 return objerror(&commit->object, "invalid format - expected 'tree' line");
225 if (get_sha1_hex(buffer+5, tree_sha1) || buffer[45] != '\n')
226 return objerror(&commit->object, "invalid 'tree' line format - bad sha1");
de2eb7f6
LT
227 buffer += 46;
228 while (!memcmp(buffer, "parent ", 7)) {
229 if (get_sha1_hex(buffer+7, sha1) || buffer[47] != '\n')
f1f0d088 230 return objerror(&commit->object, "invalid 'parent' line format - bad sha1");
de2eb7f6
LT
231 buffer += 48;
232 }
233 if (memcmp(buffer, "author ", 7))
f1f0d088 234 return objerror(&commit->object, "invalid format - expected 'author' line");
bd1e17e2
LT
235 free(commit->buffer);
236 commit->buffer = NULL;
ff5ebe39 237 if (!commit->tree)
f1f0d088 238 return objerror(&commit->object, "could not load commit's tree %s", tree_sha1);
ab7df187 239 if (!commit->parents && show_root)
c418eda4 240 printf("root %s\n", sha1_to_hex(commit->object.sha1));
e6948b6d 241 if (!commit->date)
c418eda4
DB
242 printf("bad commit date in %s\n",
243 sha1_to_hex(commit->object.sha1));
4728b861
LT
244 return 0;
245}
246
c418eda4 247static int fsck_tag(struct tag *tag)
ec4465ad 248{
92d4c85d
LT
249 struct object *tagged = tag->tagged;
250
251 if (!tagged) {
f1f0d088 252 return objerror(&tag->object, "could not load tagged object");
92d4c85d 253 }
889262ea
LT
254 if (!show_tags)
255 return 0;
256
92d4c85d
LT
257 printf("tagged %s %s", tagged->type, sha1_to_hex(tagged->sha1));
258 printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
ff5ebe39 259 return 0;
20222118
LT
260}
261
7e8c174a 262static int fsck_sha1(unsigned char *sha1)
20222118 263{
7e8c174a
LT
264 struct object *obj = parse_object(sha1);
265 if (!obj)
f1f0d088 266 return error("%s: object not found", sha1_to_hex(sha1));
7e8c174a
LT
267 if (obj->type == blob_type)
268 return 0;
269 if (obj->type == tree_type)
270 return fsck_tree((struct tree *) obj);
271 if (obj->type == commit_type)
272 return fsck_commit((struct commit *) obj);
273 if (obj->type == tag_type)
274 return fsck_tag((struct tag *) obj);
f1f0d088
PB
275 /* By now, parse_object() would've returned NULL instead. */
276 return objerror(obj, "unknown type '%s' (internal fsck error)", obj->type);
7e8c174a
LT
277}
278
279/*
280 * This is the sorting chunk size: make it reasonably
281 * big so that we can sort well..
282 */
283#define MAX_SHA1_ENTRIES (1024)
284
285struct sha1_entry {
286 unsigned long ino;
20222118 287 unsigned char sha1[20];
7e8c174a
LT
288};
289
290static struct {
291 unsigned long nr;
292 struct sha1_entry *entry[MAX_SHA1_ENTRIES];
293} sha1_list;
294
295static int ino_compare(const void *_a, const void *_b)
296{
297 const struct sha1_entry *a = _a, *b = _b;
298 unsigned long ino1 = a->ino, ino2 = b->ino;
299 return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0;
300}
301
302static void fsck_sha1_list(void)
303{
304 int i, nr = sha1_list.nr;
305
306 qsort(sha1_list.entry, nr, sizeof(struct sha1_entry *), ino_compare);
307 for (i = 0; i < nr; i++) {
308 struct sha1_entry *entry = sha1_list.entry[i];
309 unsigned char *sha1 = entry->sha1;
310
311 sha1_list.entry[i] = NULL;
f1f0d088 312 fsck_sha1(sha1);
7e8c174a 313 free(entry);
20222118 314 }
7e8c174a
LT
315 sha1_list.nr = 0;
316}
317
318static void add_sha1_list(unsigned char *sha1, unsigned long ino)
319{
320 struct sha1_entry *entry = xmalloc(sizeof(*entry));
321 int nr;
322
323 entry->ino = ino;
324 memcpy(entry->sha1, sha1, 20);
325 nr = sha1_list.nr;
326 if (nr == MAX_SHA1_ENTRIES) {
327 fsck_sha1_list();
328 nr = 0;
329 }
330 sha1_list.entry[nr] = entry;
331 sha1_list.nr = ++nr;
20222118
LT
332}
333
334static int fsck_dir(int i, char *path)
335{
336 DIR *dir = opendir(path);
337 struct dirent *de;
338
230f1322
LT
339 if (!dir)
340 return 0;
20222118
LT
341
342 while ((de = readdir(dir)) != NULL) {
343 char name[100];
7e8c174a 344 unsigned char sha1[20];
20222118
LT
345 int len = strlen(de->d_name);
346
347 switch (len) {
348 case 2:
349 if (de->d_name[1] != '.')
350 break;
351 case 1:
352 if (de->d_name[0] != '.')
353 break;
354 continue;
355 case 38:
356 sprintf(name, "%02x", i);
357 memcpy(name+2, de->d_name, len+1);
7e8c174a
LT
358 if (get_sha1_hex(name, sha1) < 0)
359 break;
360 add_sha1_list(sha1, de->d_ino);
361 continue;
20222118
LT
362 }
363 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
364 }
365 closedir(dir);
366 return 0;
367}
368
944d8589
LT
369static int default_refs = 0;
370
371static int fsck_handle_ref(const char *refname, const unsigned char *sha1)
1024932f 372{
1024932f
LT
373 struct object *obj;
374
1024932f 375 obj = lookup_object(sha1);
8a498a05 376 if (!obj) {
659cacf5
LT
377 if (!standalone && has_sha1_file(sha1)) {
378 default_refs++;
944d8589 379 return 0; /* it is in a pack */
659cacf5 380 }
944d8589
LT
381 error("%s: invalid sha1 pointer %s", refname, sha1_to_hex(sha1));
382 /* We'll continue with the rest despite the error.. */
383 return 0;
8a498a05 384 }
944d8589 385 default_refs++;
1024932f
LT
386 obj->used = 1;
387 mark_reachable(obj, REACHABLE);
7c4d07c7 388 return 0;
1024932f
LT
389}
390
1024932f
LT
391static void get_default_heads(void)
392{
944d8589
LT
393 for_each_ref(fsck_handle_ref);
394 if (!default_refs)
477606f5 395 die("No default references");
1024932f
LT
396}
397
8a498a05
JH
398static void fsck_object_dir(const char *path)
399{
400 int i;
401 for (i = 0; i < 256; i++) {
402 static char dir[4096];
403 sprintf(dir, "%s/%02x", path, i);
404 fsck_dir(i, dir);
405 }
406 fsck_sha1_list();
407}
408
c3330383
LT
409static int fsck_head_link(void)
410{
c3330383 411 unsigned char sha1[20];
8098a178
JH
412 const char *git_HEAD = strdup(git_path("HEAD"));
413 const char *git_refs_heads_master = resolve_ref(git_HEAD, sha1, 1);
414 int pfxlen = strlen(git_HEAD) - 4; /* strip .../.git/ part */
415
416 if (!git_refs_heads_master)
417 return error("HEAD is not a symbolic ref");
418 if (strncmp(git_refs_heads_master + pfxlen, "refs/heads/", 11))
419 return error("HEAD points to something strange (%s)",
420 git_refs_heads_master + pfxlen);
421 if (!memcmp(null_sha1, sha1, 20))
c3330383
LT
422 return error("HEAD: not a valid git pointer");
423 return 0;
424}
425
20222118
LT
426int main(int argc, char **argv)
427{
bcee6fd8 428 int i, heads;
20222118 429
889262ea
LT
430 for (i = 1; i < argc; i++) {
431 const char *arg = argv[i];
432
433 if (!strcmp(arg, "--unreachable")) {
434 show_unreachable = 1;
435 continue;
436 }
437 if (!strcmp(arg, "--tags")) {
438 show_tags = 1;
439 continue;
440 }
ab7df187
LT
441 if (!strcmp(arg, "--root")) {
442 show_root = 1;
443 continue;
444 }
ae7c0c92
JH
445 if (!strcmp(arg, "--cache")) {
446 keep_cache_objects = 1;
447 continue;
448 }
8a498a05
JH
449 if (!strcmp(arg, "--standalone")) {
450 standalone = 1;
451 continue;
452 }
453 if (!strcmp(arg, "--full")) {
454 check_full = 1;
455 continue;
456 }
de2eb7f6
LT
457 if (!strcmp(arg, "--strict")) {
458 check_strict = 1;
459 continue;
460 }
889262ea 461 if (*arg == '-')
215a7ad1 462 usage("git-fsck-objects [--tags] [--root] [[--unreachable] [--cache] [--standalone | --full] [--strict] <head-sha1>*]");
889262ea
LT
463 }
464
8a498a05
JH
465 if (standalone && check_full)
466 die("Only one of --standalone or --full can be used.");
467 if (standalone)
e72a7d45 468 putenv("GIT_ALTERNATE_OBJECT_DIRECTORIES=");
8a498a05 469
c3330383 470 fsck_head_link();
8a498a05
JH
471 fsck_object_dir(get_object_directory());
472 if (check_full) {
d5a63b99 473 struct alternate_object_database *alt;
8a498a05
JH
474 struct packed_git *p;
475 prepare_alt_odb();
d5a63b99 476 for (alt = alt_odb_list; alt; alt = alt->next) {
a3eb250f 477 char namebuf[PATH_MAX];
d5a63b99
JH
478 int namelen = alt->name - alt->base;
479 memcpy(namebuf, alt->base, namelen);
a3eb250f
JH
480 namebuf[namelen - 1] = 0;
481 fsck_object_dir(namebuf);
8a498a05
JH
482 }
483 prepare_packed_git();
f9253394
JH
484 for (p = packed_git; p; p = p->next)
485 /* verify gives error messages itself */
f3bf9224 486 verify_pack(p, 0);
f9253394 487
8a498a05
JH
488 for (p = packed_git; p; p = p->next) {
489 int num = num_packed_objects(p);
490 for (i = 0; i < num; i++) {
491 unsigned char sha1[20];
492 nth_packed_object_sha1(p, i, sha1);
f1f0d088 493 fsck_sha1(sha1);
8a498a05
JH
494 }
495 }
bcee6fd8
LT
496 }
497
498 heads = 0;
d9839e03 499 for (i = 1; i < argc; i++) {
889262ea
LT
500 const char *arg = argv[i];
501
502 if (*arg == '-')
d9839e03 503 continue;
889262ea 504
3c249c95 505 if (!get_sha1(arg, head_sha1)) {
770896e5 506 struct object *obj = lookup_object(head_sha1);
e1a1388d 507
770896e5
LT
508 /* Error is printed by lookup_object(). */
509 if (!obj)
e1a1388d
JF
510 continue;
511
ff5ebe39
DB
512 obj->used = 1;
513 mark_reachable(obj, REACHABLE);
bcee6fd8 514 heads++;
d9839e03
LT
515 continue;
516 }
f1f0d088 517 error("invalid parameter: expected sha1, got '%s'", arg);
d9839e03 518 }
d9839e03 519
1024932f 520 /*
d1af002d 521 * If we've not been given any explicit head information, do the
e7bd907d
LT
522 * default ones from .git/refs. We also consider the index file
523 * in this case (ie this implies --cache).
1024932f 524 */
e7bd907d 525 if (!heads) {
1024932f
LT
526 get_default_heads();
527 keep_cache_objects = 1;
528 }
529
ae7c0c92
JH
530 if (keep_cache_objects) {
531 int i;
532 read_cache();
533 for (i = 0; i < active_nr; i++) {
534 struct blob *blob = lookup_blob(active_cache[i]->sha1);
535 struct object *obj;
536 if (!blob)
537 continue;
538 obj = &blob->object;
539 obj->used = 1;
540 mark_reachable(obj, REACHABLE);
541 }
542 }
543
8ba0bbb2 544 check_connectivity();
20222118
LT
545 return 0;
546}