Make "tree_entry" have a SHA1 instead of a union of object pointers
[git] / fsck-objects.c
1 #include <sys/types.h>
2 #include <dirent.h>
3
4 #include "cache.h"
5 #include "commit.h"
6 #include "tree.h"
7 #include "blob.h"
8 #include "tag.h"
9 #include "refs.h"
10 #include "pack.h"
11 #include "cache-tree.h"
12
13 #define REACHABLE 0x0001
14
15 static int show_root = 0;
16 static int show_tags = 0;
17 static int show_unreachable = 0;
18 static int check_full = 0;
19 static int check_strict = 0;
20 static int keep_cache_objects = 0;
21 static unsigned char head_sha1[20];
22
23 #ifdef NO_D_INO_IN_DIRENT
24 #define SORT_DIRENT 0
25 #define DIRENT_SORT_HINT(de) 0
26 #else
27 #define SORT_DIRENT 1
28 #define DIRENT_SORT_HINT(de) ((de)->d_ino)
29 #endif
30
31 static void objreport(struct object *obj, const char *severity,
32                       const char *err, va_list params)
33 {
34         fprintf(stderr, "%s in %s %s: ",
35                 severity, obj->type, sha1_to_hex(obj->sha1));
36         vfprintf(stderr, err, params);
37         fputs("\n", stderr);
38 }
39
40 static int objerror(struct object *obj, const char *err, ...)
41 {
42         va_list params;
43         va_start(params, err);
44         objreport(obj, "error", err, params);
45         va_end(params);
46         return -1;
47 }
48
49 static int objwarning(struct object *obj, const char *err, ...)
50 {
51         va_list params;
52         va_start(params, err);
53         objreport(obj, "warning", err, params);
54         va_end(params);
55         return -1;
56 }
57
58
59 static void check_connectivity(void)
60 {
61         int i;
62
63         /* Look up all the requirements, warn about missing objects.. */
64         for (i = 0; i < obj_allocs; i++) {
65                 struct object *obj = objs[i];
66
67                 if (!obj)
68                         continue;
69
70                 if (!obj->parsed) {
71                         if (has_sha1_file(obj->sha1))
72                                 ; /* it is in pack */
73                         else
74                                 printf("missing %s %s\n",
75                                        obj->type, sha1_to_hex(obj->sha1));
76                         continue;
77                 }
78
79                 if (obj->refs) {
80                         const struct object_refs *refs = obj->refs;
81                         unsigned j;
82                         for (j = 0; j < refs->count; j++) {
83                                 struct object *ref = refs->ref[j];
84                                 if (ref->parsed ||
85                                     (has_sha1_file(ref->sha1)))
86                                         continue;
87                                 printf("broken link from %7s %s\n",
88                                        obj->type, sha1_to_hex(obj->sha1));
89                                 printf("              to %7s %s\n",
90                                        ref->type, sha1_to_hex(ref->sha1));
91                         }
92                 }
93
94                 if (show_unreachable && !(obj->flags & REACHABLE)) {
95                         printf("unreachable %s %s\n",
96                                obj->type, sha1_to_hex(obj->sha1));
97                         continue;
98                 }
99
100                 if (!obj->used) {
101                         printf("dangling %s %s\n", obj->type, 
102                                sha1_to_hex(obj->sha1));
103                 }
104         }
105 }
106
107 /*
108  * The entries in a tree are ordered in the _path_ order,
109  * which means that a directory entry is ordered by adding
110  * a slash to the end of it.
111  *
112  * So a directory called "a" is ordered _after_ a file
113  * called "a.c", because "a/" sorts after "a.c".
114  */
115 #define TREE_UNORDERED (-1)
116 #define TREE_HAS_DUPS  (-2)
117
118 static int verify_ordered(struct tree_entry_list *a, struct tree_entry_list *b)
119 {
120         int len1 = strlen(a->name);
121         int len2 = strlen(b->name);
122         int len = len1 < len2 ? len1 : len2;
123         unsigned char c1, c2;
124         int cmp;
125
126         cmp = memcmp(a->name, b->name, len);
127         if (cmp < 0)
128                 return 0;
129         if (cmp > 0)
130                 return TREE_UNORDERED;
131
132         /*
133          * Ok, the first <len> characters are the same.
134          * Now we need to order the next one, but turn
135          * a '\0' into a '/' for a directory entry.
136          */
137         c1 = a->name[len];
138         c2 = b->name[len];
139         if (!c1 && !c2)
140                 /*
141                  * git-write-tree used to write out a nonsense tree that has
142                  * entries with the same name, one blob and one tree.  Make
143                  * sure we do not have duplicate entries.
144                  */
145                 return TREE_HAS_DUPS;
146         if (!c1 && a->directory)
147                 c1 = '/';
148         if (!c2 && b->directory)
149                 c2 = '/';
150         return c1 < c2 ? 0 : TREE_UNORDERED;
151 }
152
153 static int fsck_tree(struct tree *item)
154 {
155         int retval;
156         int has_full_path = 0;
157         int has_zero_pad = 0;
158         int has_bad_modes = 0;
159         int has_dup_entries = 0;
160         int not_properly_sorted = 0;
161         struct tree_entry_list *entry, *last;
162
163         last = NULL;
164         for (entry = item->entries; entry; entry = entry->next) {
165                 if (strchr(entry->name, '/'))
166                         has_full_path = 1;
167                 has_zero_pad |= entry->zeropad;
168
169                 switch (entry->mode) {
170                 /*
171                  * Standard modes.. 
172                  */
173                 case S_IFREG | 0755:
174                 case S_IFREG | 0644:
175                 case S_IFLNK:
176                 case S_IFDIR:
177                         break;
178                 /*
179                  * This is nonstandard, but we had a few of these
180                  * early on when we honored the full set of mode
181                  * bits..
182                  */
183                 case S_IFREG | 0664:
184                         if (!check_strict)
185                                 break;
186                 default:
187                         has_bad_modes = 1;
188                 }
189
190                 if (last) {
191                         switch (verify_ordered(last, entry)) {
192                         case TREE_UNORDERED:
193                                 not_properly_sorted = 1;
194                                 break;
195                         case TREE_HAS_DUPS:
196                                 has_dup_entries = 1;
197                                 break;
198                         default:
199                                 break;
200                         }
201                         free(last);
202                 }
203
204                 last = entry;
205         }
206         if (last)
207                 free(last);
208         item->entries = NULL;
209         free(item->buffer);
210         item->buffer = NULL;
211
212         retval = 0;
213         if (has_full_path) {
214                 objwarning(&item->object, "contains full pathnames");
215         }
216         if (has_zero_pad) {
217                 objwarning(&item->object, "contains zero-padded file modes");
218         }
219         if (has_bad_modes) {
220                 objwarning(&item->object, "contains bad file modes");
221         }
222         if (has_dup_entries) {
223                 retval = objerror(&item->object, "contains duplicate file entries");
224         }
225         if (not_properly_sorted) {
226                 retval = objerror(&item->object, "not properly sorted");
227         }
228         return retval;
229 }
230
231 static int fsck_commit(struct commit *commit)
232 {
233         char *buffer = commit->buffer;
234         unsigned char tree_sha1[20], sha1[20];
235
236         if (memcmp(buffer, "tree ", 5))
237                 return objerror(&commit->object, "invalid format - expected 'tree' line");
238         if (get_sha1_hex(buffer+5, tree_sha1) || buffer[45] != '\n')
239                 return objerror(&commit->object, "invalid 'tree' line format - bad sha1");
240         buffer += 46;
241         while (!memcmp(buffer, "parent ", 7)) {
242                 if (get_sha1_hex(buffer+7, sha1) || buffer[47] != '\n')
243                         return objerror(&commit->object, "invalid 'parent' line format - bad sha1");
244                 buffer += 48;
245         }
246         if (memcmp(buffer, "author ", 7))
247                 return objerror(&commit->object, "invalid format - expected 'author' line");
248         free(commit->buffer);
249         commit->buffer = NULL;
250         if (!commit->tree)
251                 return objerror(&commit->object, "could not load commit's tree %s", tree_sha1);
252         if (!commit->parents && show_root)
253                 printf("root %s\n", sha1_to_hex(commit->object.sha1));
254         if (!commit->date)
255                 printf("bad commit date in %s\n", 
256                        sha1_to_hex(commit->object.sha1));
257         return 0;
258 }
259
260 static int fsck_tag(struct tag *tag)
261 {
262         struct object *tagged = tag->tagged;
263
264         if (!tagged) {
265                 return objerror(&tag->object, "could not load tagged object");
266         }
267         if (!show_tags)
268                 return 0;
269
270         printf("tagged %s %s", tagged->type, sha1_to_hex(tagged->sha1));
271         printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
272         return 0;
273 }
274
275 static int fsck_sha1(unsigned char *sha1)
276 {
277         struct object *obj = parse_object(sha1);
278         if (!obj)
279                 return error("%s: object not found", sha1_to_hex(sha1));
280         if (obj->type == blob_type)
281                 return 0;
282         if (obj->type == tree_type)
283                 return fsck_tree((struct tree *) obj);
284         if (obj->type == commit_type)
285                 return fsck_commit((struct commit *) obj);
286         if (obj->type == tag_type)
287                 return fsck_tag((struct tag *) obj);
288         /* By now, parse_object() would've returned NULL instead. */
289         return objerror(obj, "unknown type '%s' (internal fsck error)", obj->type);
290 }
291
292 /*
293  * This is the sorting chunk size: make it reasonably
294  * big so that we can sort well..
295  */
296 #define MAX_SHA1_ENTRIES (1024)
297
298 struct sha1_entry {
299         unsigned long ino;
300         unsigned char sha1[20];
301 };
302
303 static struct {
304         unsigned long nr;
305         struct sha1_entry *entry[MAX_SHA1_ENTRIES];
306 } sha1_list;
307
308 static int ino_compare(const void *_a, const void *_b)
309 {
310         const struct sha1_entry *a = _a, *b = _b;
311         unsigned long ino1 = a->ino, ino2 = b->ino;
312         return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0;
313 }
314
315 static void fsck_sha1_list(void)
316 {
317         int i, nr = sha1_list.nr;
318
319         if (SORT_DIRENT)
320                 qsort(sha1_list.entry, nr,
321                       sizeof(struct sha1_entry *), ino_compare);
322         for (i = 0; i < nr; i++) {
323                 struct sha1_entry *entry = sha1_list.entry[i];
324                 unsigned char *sha1 = entry->sha1;
325
326                 sha1_list.entry[i] = NULL;
327                 fsck_sha1(sha1);
328                 free(entry);
329         }
330         sha1_list.nr = 0;
331 }
332
333 static void add_sha1_list(unsigned char *sha1, unsigned long ino)
334 {
335         struct sha1_entry *entry = xmalloc(sizeof(*entry));
336         int nr;
337
338         entry->ino = ino;
339         memcpy(entry->sha1, sha1, 20);
340         nr = sha1_list.nr;
341         if (nr == MAX_SHA1_ENTRIES) {
342                 fsck_sha1_list();
343                 nr = 0;
344         }
345         sha1_list.entry[nr] = entry;
346         sha1_list.nr = ++nr;
347 }
348
349 static int fsck_dir(int i, char *path)
350 {
351         DIR *dir = opendir(path);
352         struct dirent *de;
353
354         if (!dir)
355                 return 0;
356
357         while ((de = readdir(dir)) != NULL) {
358                 char name[100];
359                 unsigned char sha1[20];
360                 int len = strlen(de->d_name);
361
362                 switch (len) {
363                 case 2:
364                         if (de->d_name[1] != '.')
365                                 break;
366                 case 1:
367                         if (de->d_name[0] != '.')
368                                 break;
369                         continue;
370                 case 38:
371                         sprintf(name, "%02x", i);
372                         memcpy(name+2, de->d_name, len+1);
373                         if (get_sha1_hex(name, sha1) < 0)
374                                 break;
375                         add_sha1_list(sha1, DIRENT_SORT_HINT(de));
376                         continue;
377                 }
378                 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
379         }
380         closedir(dir);
381         return 0;
382 }
383
384 static int default_refs = 0;
385
386 static int fsck_handle_ref(const char *refname, const unsigned char *sha1)
387 {
388         struct object *obj;
389
390         obj = lookup_object(sha1);
391         if (!obj) {
392                 if (has_sha1_file(sha1)) {
393                         default_refs++;
394                         return 0; /* it is in a pack */
395                 }
396                 error("%s: invalid sha1 pointer %s", refname, sha1_to_hex(sha1));
397                 /* We'll continue with the rest despite the error.. */
398                 return 0;
399         }
400         default_refs++;
401         obj->used = 1;
402         mark_reachable(obj, REACHABLE);
403         return 0;
404 }
405
406 static void get_default_heads(void)
407 {
408         for_each_ref(fsck_handle_ref);
409         if (!default_refs)
410                 die("No default references");
411 }
412
413 static void fsck_object_dir(const char *path)
414 {
415         int i;
416         for (i = 0; i < 256; i++) {
417                 static char dir[4096];
418                 sprintf(dir, "%s/%02x", path, i);
419                 fsck_dir(i, dir);
420         }
421         fsck_sha1_list();
422 }
423
424 static int fsck_head_link(void)
425 {
426         unsigned char sha1[20];
427         const char *git_HEAD = strdup(git_path("HEAD"));
428         const char *git_refs_heads_master = resolve_ref(git_HEAD, sha1, 1);
429         int pfxlen = strlen(git_HEAD) - 4; /* strip .../.git/ part */
430
431         if (!git_refs_heads_master)
432                 return error("HEAD is not a symbolic ref");
433         if (strncmp(git_refs_heads_master + pfxlen, "refs/heads/", 11))
434                 return error("HEAD points to something strange (%s)",
435                              git_refs_heads_master + pfxlen);
436         if (!memcmp(null_sha1, sha1, 20))
437                 return error("HEAD: not a valid git pointer");
438         return 0;
439 }
440
441 static int fsck_cache_tree(struct cache_tree *it)
442 {
443         int i;
444         int err = 0;
445
446         if (0 <= it->entry_count) {
447                 struct object *obj = parse_object(it->sha1);
448                 if (!obj) {
449                         error("%s: invalid sha1 pointer in cache-tree",
450                               sha1_to_hex(it->sha1));
451                         return 1;
452                 }
453                 mark_reachable(obj, REACHABLE);
454                 obj->used = 1;
455                 if (obj->type != tree_type)
456                         err |= objerror(obj, "non-tree in cache-tree");
457         }
458         for (i = 0; i < it->subtree_nr; i++)
459                 err |= fsck_cache_tree(it->down[i]->cache_tree);
460         return err;
461 }
462
463 int main(int argc, char **argv)
464 {
465         int i, heads;
466
467         track_object_refs = 1;
468         setup_git_directory();
469
470         for (i = 1; i < argc; i++) {
471                 const char *arg = argv[i];
472
473                 if (!strcmp(arg, "--unreachable")) {
474                         show_unreachable = 1;
475                         continue;
476                 }
477                 if (!strcmp(arg, "--tags")) {
478                         show_tags = 1;
479                         continue;
480                 }
481                 if (!strcmp(arg, "--root")) {
482                         show_root = 1;
483                         continue;
484                 }
485                 if (!strcmp(arg, "--cache")) {
486                         keep_cache_objects = 1;
487                         continue;
488                 }
489                 if (!strcmp(arg, "--full")) {
490                         check_full = 1;
491                         continue;
492                 }
493                 if (!strcmp(arg, "--strict")) {
494                         check_strict = 1;
495                         continue;
496                 }
497                 if (*arg == '-')
498                         usage("git-fsck-objects [--tags] [--root] [[--unreachable] [--cache] [--full] [--strict] <head-sha1>*]");
499         }
500
501         fsck_head_link();
502         fsck_object_dir(get_object_directory());
503         if (check_full) {
504                 struct alternate_object_database *alt;
505                 struct packed_git *p;
506                 prepare_alt_odb();
507                 for (alt = alt_odb_list; alt; alt = alt->next) {
508                         char namebuf[PATH_MAX];
509                         int namelen = alt->name - alt->base;
510                         memcpy(namebuf, alt->base, namelen);
511                         namebuf[namelen - 1] = 0;
512                         fsck_object_dir(namebuf);
513                 }
514                 prepare_packed_git();
515                 for (p = packed_git; p; p = p->next)
516                         /* verify gives error messages itself */
517                         verify_pack(p, 0);
518
519                 for (p = packed_git; p; p = p->next) {
520                         int num = num_packed_objects(p);
521                         for (i = 0; i < num; i++) {
522                                 unsigned char sha1[20];
523                                 nth_packed_object_sha1(p, i, sha1);
524                                 fsck_sha1(sha1);
525                         }
526                 }
527         }
528
529         heads = 0;
530         for (i = 1; i < argc; i++) {
531                 const char *arg = argv[i]; 
532
533                 if (*arg == '-')
534                         continue;
535
536                 if (!get_sha1(arg, head_sha1)) {
537                         struct object *obj = lookup_object(head_sha1);
538
539                         /* Error is printed by lookup_object(). */
540                         if (!obj)
541                                 continue;
542
543                         obj->used = 1;
544                         mark_reachable(obj, REACHABLE);
545                         heads++;
546                         continue;
547                 }
548                 error("invalid parameter: expected sha1, got '%s'", arg);
549         }
550
551         /*
552          * If we've not been given any explicit head information, do the
553          * default ones from .git/refs. We also consider the index file
554          * in this case (ie this implies --cache).
555          */
556         if (!heads) {
557                 get_default_heads();
558                 keep_cache_objects = 1;
559         }
560
561         if (keep_cache_objects) {
562                 int i;
563                 read_cache();
564                 for (i = 0; i < active_nr; i++) {
565                         struct blob *blob = lookup_blob(active_cache[i]->sha1);
566                         struct object *obj;
567                         if (!blob)
568                                 continue;
569                         obj = &blob->object;
570                         obj->used = 1;
571                         mark_reachable(obj, REACHABLE);
572                 }
573                 if (active_cache_tree)
574                         fsck_cache_tree(active_cache_tree);
575         }
576
577         check_connectivity();
578         return 0;
579 }