Commit | Line | Data |
---|---|---|
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" |
f9253394 | 9 | #include "pack.h" |
ff5ebe39 DB |
10 | |
11 | #define REACHABLE 0x0001 | |
d9839e03 | 12 | |
ab7df187 | 13 | static int show_root = 0; |
889262ea | 14 | static int show_tags = 0; |
d9839e03 | 15 | static int show_unreachable = 0; |
8a498a05 JH |
16 | static int standalone = 0; |
17 | static int check_full = 0; | |
ae7c0c92 | 18 | static int keep_cache_objects = 0; |
d9839e03 LT |
19 | static unsigned char head_sha1[20]; |
20 | ||
8ba0bbb2 LT |
21 | static void check_connectivity(void) |
22 | { | |
23 | int i; | |
24 | ||
8ba0bbb2 | 25 | /* Look up all the requirements, warn about missing objects.. */ |
ff5ebe39 DB |
26 | for (i = 0; i < nr_objs; i++) { |
27 | struct object *obj = objs[i]; | |
c418eda4 | 28 | struct object_list *refs; |
8ba0bbb2 | 29 | |
3a6a23e6 | 30 | if (!obj->parsed) { |
8a498a05 JH |
31 | if (!standalone && has_sha1_file(obj->sha1)) |
32 | ; /* it is in pack */ | |
33 | else | |
34 | printf("missing %s %s\n", | |
35 | obj->type, sha1_to_hex(obj->sha1)); | |
3a6a23e6 LT |
36 | continue; |
37 | } | |
38 | ||
39 | for (refs = obj->refs; refs; refs = refs->next) { | |
8a498a05 JH |
40 | if (refs->item->parsed || |
41 | (!standalone && has_sha1_file(refs->item->sha1))) | |
3a6a23e6 LT |
42 | continue; |
43 | printf("broken link from %7s %s\n", | |
44 | obj->type, sha1_to_hex(obj->sha1)); | |
45 | printf(" to %7s %s\n", | |
aa034134 | 46 | refs->item->type, sha1_to_hex(refs->item->sha1)); |
3a6a23e6 LT |
47 | } |
48 | ||
ff5ebe39 | 49 | if (show_unreachable && !(obj->flags & REACHABLE)) { |
c4584ae3 JH |
50 | printf("unreachable %s %s\n", |
51 | obj->type, sha1_to_hex(obj->sha1)); | |
8ba0bbb2 | 52 | continue; |
d9839e03 | 53 | } |
8ba0bbb2 | 54 | |
ff5ebe39 DB |
55 | if (!obj->used) { |
56 | printf("dangling %s %s\n", obj->type, | |
57 | sha1_to_hex(obj->sha1)); | |
d9839e03 | 58 | } |
8ba0bbb2 LT |
59 | } |
60 | } | |
61 | ||
85003492 LT |
62 | /* |
63 | * The entries in a tree are ordered in the _path_ order, | |
64 | * which means that a directory entry is ordered by adding | |
65 | * a slash to the end of it. | |
66 | * | |
67 | * So a directory called "a" is ordered _after_ a file | |
68 | * called "a.c", because "a/" sorts after "a.c". | |
69 | */ | |
a4f35a2d JH |
70 | #define TREE_UNORDERED (-1) |
71 | #define TREE_HAS_DUPS (-2) | |
72 | ||
85003492 LT |
73 | static int verify_ordered(struct tree_entry_list *a, struct tree_entry_list *b) |
74 | { | |
75 | int len1 = strlen(a->name); | |
76 | int len2 = strlen(b->name); | |
77 | int len = len1 < len2 ? len1 : len2; | |
78 | unsigned char c1, c2; | |
79 | int cmp; | |
80 | ||
81 | cmp = memcmp(a->name, b->name, len); | |
82 | if (cmp < 0) | |
83 | return 0; | |
84 | if (cmp > 0) | |
a4f35a2d | 85 | return TREE_UNORDERED; |
85003492 LT |
86 | |
87 | /* | |
88 | * Ok, the first <len> characters are the same. | |
89 | * Now we need to order the next one, but turn | |
90 | * a '\0' into a '/' for a directory entry. | |
91 | */ | |
92 | c1 = a->name[len]; | |
93 | c2 = b->name[len]; | |
a4f35a2d JH |
94 | if (!c1 && !c2) |
95 | /* | |
96 | * git-write-tree used to write out a nonsense tree that has | |
97 | * entries with the same name, one blob and one tree. Make | |
98 | * sure we do not have duplicate entries. | |
99 | */ | |
100 | return TREE_HAS_DUPS; | |
85003492 LT |
101 | if (!c1 && a->directory) |
102 | c1 = '/'; | |
103 | if (!c2 && b->directory) | |
104 | c2 = '/'; | |
a4f35a2d | 105 | return c1 < c2 ? 0 : TREE_UNORDERED; |
85003492 LT |
106 | } |
107 | ||
c418eda4 | 108 | static int fsck_tree(struct tree *item) |
20222118 | 109 | { |
85003492 LT |
110 | int has_full_path = 0; |
111 | struct tree_entry_list *entry, *last; | |
112 | ||
113 | last = NULL; | |
114 | for (entry = item->entries; entry; entry = entry->next) { | |
115 | if (strchr(entry->name, '/')) | |
116 | has_full_path = 1; | |
117 | ||
42ea9cb2 LT |
118 | switch (entry->mode) { |
119 | /* | |
120 | * Standard modes.. | |
121 | */ | |
122 | case S_IFREG | 0755: | |
123 | case S_IFREG | 0644: | |
124 | case S_IFLNK: | |
125 | case S_IFDIR: | |
126 | break; | |
127 | /* | |
128 | * This is nonstandard, but we had a few of these | |
129 | * early on when we honored the full set of mode | |
130 | * bits.. | |
131 | */ | |
132 | case S_IFREG | 0664: | |
133 | break; | |
134 | default: | |
135 | printf("tree %s has entry %o %s\n", | |
136 | sha1_to_hex(item->object.sha1), | |
137 | entry->mode, entry->name); | |
138 | } | |
139 | ||
85003492 | 140 | if (last) { |
a4f35a2d JH |
141 | switch (verify_ordered(last, entry)) { |
142 | case TREE_UNORDERED: | |
85003492 LT |
143 | fprintf(stderr, "tree %s not ordered\n", |
144 | sha1_to_hex(item->object.sha1)); | |
145 | return -1; | |
a4f35a2d JH |
146 | case TREE_HAS_DUPS: |
147 | fprintf(stderr, "tree %s has duplicate entries for '%s'\n", | |
148 | sha1_to_hex(item->object.sha1), | |
149 | entry->name); | |
150 | return -1; | |
151 | default: | |
152 | break; | |
85003492 LT |
153 | } |
154 | } | |
155 | ||
156 | last = entry; | |
157 | } | |
158 | ||
159 | if (has_full_path) { | |
667bb59b | 160 | fprintf(stderr, "warning: git-fsck-cache: tree %s " |
c418eda4 DB |
161 | "has full pathnames in it\n", |
162 | sha1_to_hex(item->object.sha1)); | |
1ea34e36 | 163 | } |
85003492 | 164 | |
59c1e249 | 165 | return 0; |
1ea34e36 LT |
166 | } |
167 | ||
c418eda4 | 168 | static int fsck_commit(struct commit *commit) |
1ea34e36 | 169 | { |
bd1e17e2 LT |
170 | free(commit->buffer); |
171 | commit->buffer = NULL; | |
ff5ebe39 | 172 | if (!commit->tree) |
1ea34e36 | 173 | return -1; |
ab7df187 | 174 | if (!commit->parents && show_root) |
c418eda4 | 175 | printf("root %s\n", sha1_to_hex(commit->object.sha1)); |
e6948b6d | 176 | if (!commit->date) |
c418eda4 DB |
177 | printf("bad commit date in %s\n", |
178 | sha1_to_hex(commit->object.sha1)); | |
4728b861 LT |
179 | return 0; |
180 | } | |
181 | ||
c418eda4 | 182 | static int fsck_tag(struct tag *tag) |
ec4465ad | 183 | { |
92d4c85d LT |
184 | struct object *tagged = tag->tagged; |
185 | ||
186 | if (!tagged) { | |
187 | printf("bad object in tag %s\n", sha1_to_hex(tag->object.sha1)); | |
188 | return -1; | |
189 | } | |
889262ea LT |
190 | if (!show_tags) |
191 | return 0; | |
192 | ||
92d4c85d LT |
193 | printf("tagged %s %s", tagged->type, sha1_to_hex(tagged->sha1)); |
194 | printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1)); | |
ff5ebe39 | 195 | return 0; |
20222118 LT |
196 | } |
197 | ||
7e8c174a | 198 | static int fsck_sha1(unsigned char *sha1) |
20222118 | 199 | { |
7e8c174a LT |
200 | struct object *obj = parse_object(sha1); |
201 | if (!obj) | |
202 | return -1; | |
203 | if (obj->type == blob_type) | |
204 | return 0; | |
205 | if (obj->type == tree_type) | |
206 | return fsck_tree((struct tree *) obj); | |
207 | if (obj->type == commit_type) | |
208 | return fsck_commit((struct commit *) obj); | |
209 | if (obj->type == tag_type) | |
210 | return fsck_tag((struct tag *) obj); | |
211 | return -1; | |
212 | } | |
213 | ||
214 | /* | |
215 | * This is the sorting chunk size: make it reasonably | |
216 | * big so that we can sort well.. | |
217 | */ | |
218 | #define MAX_SHA1_ENTRIES (1024) | |
219 | ||
220 | struct sha1_entry { | |
221 | unsigned long ino; | |
20222118 | 222 | unsigned char sha1[20]; |
7e8c174a LT |
223 | }; |
224 | ||
225 | static struct { | |
226 | unsigned long nr; | |
227 | struct sha1_entry *entry[MAX_SHA1_ENTRIES]; | |
228 | } sha1_list; | |
229 | ||
230 | static int ino_compare(const void *_a, const void *_b) | |
231 | { | |
232 | const struct sha1_entry *a = _a, *b = _b; | |
233 | unsigned long ino1 = a->ino, ino2 = b->ino; | |
234 | return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0; | |
235 | } | |
236 | ||
237 | static void fsck_sha1_list(void) | |
238 | { | |
239 | int i, nr = sha1_list.nr; | |
240 | ||
241 | qsort(sha1_list.entry, nr, sizeof(struct sha1_entry *), ino_compare); | |
242 | for (i = 0; i < nr; i++) { | |
243 | struct sha1_entry *entry = sha1_list.entry[i]; | |
244 | unsigned char *sha1 = entry->sha1; | |
245 | ||
246 | sha1_list.entry[i] = NULL; | |
247 | if (fsck_sha1(sha1) < 0) | |
248 | fprintf(stderr, "bad sha1 entry '%s'\n", sha1_to_hex(sha1)); | |
249 | free(entry); | |
20222118 | 250 | } |
7e8c174a LT |
251 | sha1_list.nr = 0; |
252 | } | |
253 | ||
254 | static void add_sha1_list(unsigned char *sha1, unsigned long ino) | |
255 | { | |
256 | struct sha1_entry *entry = xmalloc(sizeof(*entry)); | |
257 | int nr; | |
258 | ||
259 | entry->ino = ino; | |
260 | memcpy(entry->sha1, sha1, 20); | |
261 | nr = sha1_list.nr; | |
262 | if (nr == MAX_SHA1_ENTRIES) { | |
263 | fsck_sha1_list(); | |
264 | nr = 0; | |
265 | } | |
266 | sha1_list.entry[nr] = entry; | |
267 | sha1_list.nr = ++nr; | |
20222118 LT |
268 | } |
269 | ||
270 | static int fsck_dir(int i, char *path) | |
271 | { | |
272 | DIR *dir = opendir(path); | |
273 | struct dirent *de; | |
274 | ||
275 | if (!dir) { | |
2de381f9 | 276 | return error("missing sha1 directory '%s'", path); |
20222118 LT |
277 | } |
278 | ||
279 | while ((de = readdir(dir)) != NULL) { | |
280 | char name[100]; | |
7e8c174a | 281 | unsigned char sha1[20]; |
20222118 LT |
282 | int len = strlen(de->d_name); |
283 | ||
284 | switch (len) { | |
285 | case 2: | |
286 | if (de->d_name[1] != '.') | |
287 | break; | |
288 | case 1: | |
289 | if (de->d_name[0] != '.') | |
290 | break; | |
291 | continue; | |
292 | case 38: | |
293 | sprintf(name, "%02x", i); | |
294 | memcpy(name+2, de->d_name, len+1); | |
7e8c174a LT |
295 | if (get_sha1_hex(name, sha1) < 0) |
296 | break; | |
297 | add_sha1_list(sha1, de->d_ino); | |
298 | continue; | |
20222118 LT |
299 | } |
300 | fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name); | |
301 | } | |
302 | closedir(dir); | |
303 | return 0; | |
304 | } | |
305 | ||
7c4d07c7 | 306 | static int read_sha1_reference(const char *path) |
1024932f LT |
307 | { |
308 | char hexname[60]; | |
309 | unsigned char sha1[20]; | |
310 | int fd = open(path, O_RDONLY), len; | |
311 | struct object *obj; | |
312 | ||
313 | if (fd < 0) | |
7c4d07c7 | 314 | return -1; |
1024932f LT |
315 | |
316 | len = read(fd, hexname, sizeof(hexname)); | |
317 | close(fd); | |
318 | if (len < 40) | |
7c4d07c7 | 319 | return -1; |
1024932f LT |
320 | |
321 | if (get_sha1_hex(hexname, sha1) < 0) | |
7c4d07c7 | 322 | return -1; |
1024932f LT |
323 | |
324 | obj = lookup_object(sha1); | |
8a498a05 JH |
325 | if (!obj) { |
326 | if (!standalone && has_sha1_file(sha1)) | |
327 | return 0; /* it is in pack */ | |
7c4d07c7 | 328 | return error("%s: invalid sha1 pointer %.40s", path, hexname); |
8a498a05 | 329 | } |
7c4d07c7 | 330 | |
1024932f LT |
331 | obj->used = 1; |
332 | mark_reachable(obj, REACHABLE); | |
7c4d07c7 | 333 | return 0; |
1024932f LT |
334 | } |
335 | ||
477606f5 | 336 | static int find_file_objects(const char *base, const char *name) |
1024932f LT |
337 | { |
338 | int baselen = strlen(base); | |
339 | int namelen = strlen(name); | |
340 | char *path = xmalloc(baselen + namelen + 2); | |
341 | struct stat st; | |
342 | ||
343 | memcpy(path, base, baselen); | |
344 | path[baselen] = '/'; | |
345 | memcpy(path + baselen + 1, name, namelen+1); | |
346 | if (stat(path, &st) < 0) | |
477606f5 | 347 | return 0; |
1024932f LT |
348 | |
349 | /* | |
350 | * Recurse into directories | |
351 | */ | |
352 | if (S_ISDIR(st.st_mode)) { | |
477606f5 | 353 | int count = 0; |
1024932f LT |
354 | DIR *dir = opendir(path); |
355 | if (dir) { | |
356 | struct dirent *de; | |
357 | while ((de = readdir(dir)) != NULL) { | |
358 | if (de->d_name[0] == '.') | |
359 | continue; | |
477606f5 | 360 | count += find_file_objects(path, de->d_name); |
1024932f LT |
361 | } |
362 | closedir(dir); | |
363 | } | |
477606f5 | 364 | return count; |
1024932f | 365 | } |
477606f5 LT |
366 | if (S_ISREG(st.st_mode)) |
367 | return read_sha1_reference(path) == 0; | |
368 | return 0; | |
1024932f LT |
369 | } |
370 | ||
371 | static void get_default_heads(void) | |
372 | { | |
373 | char *git_dir = gitenv(GIT_DIR_ENVIRONMENT) ? : DEFAULT_GIT_DIR_ENVIRONMENT; | |
477606f5 LT |
374 | int count = find_file_objects(git_dir, "refs"); |
375 | if (!count) | |
376 | die("No default references"); | |
1024932f LT |
377 | } |
378 | ||
8a498a05 JH |
379 | static void fsck_object_dir(const char *path) |
380 | { | |
381 | int i; | |
382 | for (i = 0; i < 256; i++) { | |
383 | static char dir[4096]; | |
384 | sprintf(dir, "%s/%02x", path, i); | |
385 | fsck_dir(i, dir); | |
386 | } | |
387 | fsck_sha1_list(); | |
388 | } | |
389 | ||
20222118 LT |
390 | int main(int argc, char **argv) |
391 | { | |
bcee6fd8 | 392 | int i, heads; |
20222118 | 393 | |
889262ea LT |
394 | for (i = 1; i < argc; i++) { |
395 | const char *arg = argv[i]; | |
396 | ||
397 | if (!strcmp(arg, "--unreachable")) { | |
398 | show_unreachable = 1; | |
399 | continue; | |
400 | } | |
401 | if (!strcmp(arg, "--tags")) { | |
402 | show_tags = 1; | |
403 | continue; | |
404 | } | |
ab7df187 LT |
405 | if (!strcmp(arg, "--root")) { |
406 | show_root = 1; | |
407 | continue; | |
408 | } | |
ae7c0c92 JH |
409 | if (!strcmp(arg, "--cache")) { |
410 | keep_cache_objects = 1; | |
411 | continue; | |
412 | } | |
8a498a05 JH |
413 | if (!strcmp(arg, "--standalone")) { |
414 | standalone = 1; | |
415 | continue; | |
416 | } | |
417 | if (!strcmp(arg, "--full")) { | |
418 | check_full = 1; | |
419 | continue; | |
420 | } | |
889262ea | 421 | if (*arg == '-') |
8a498a05 | 422 | usage("git-fsck-cache [--tags] [[--unreachable] [--cache] [--standalone | --full] <head-sha1>*]"); |
889262ea LT |
423 | } |
424 | ||
8a498a05 JH |
425 | if (standalone && check_full) |
426 | die("Only one of --standalone or --full can be used."); | |
427 | if (standalone) | |
428 | unsetenv("GIT_ALTERNATE_OBJECT_DIRECTORIES"); | |
429 | ||
430 | fsck_object_dir(get_object_directory()); | |
431 | if (check_full) { | |
432 | int j; | |
433 | struct packed_git *p; | |
434 | prepare_alt_odb(); | |
435 | for (j = 0; alt_odb[j].base; j++) { | |
436 | alt_odb[j].name[-1] = 0; /* was slash */ | |
437 | fsck_object_dir(alt_odb[j].base); | |
438 | alt_odb[j].name[-1] = '/'; | |
439 | } | |
440 | prepare_packed_git(); | |
f9253394 JH |
441 | for (p = packed_git; p; p = p->next) |
442 | /* verify gives error messages itself */ | |
f3bf9224 | 443 | verify_pack(p, 0); |
f9253394 | 444 | |
8a498a05 JH |
445 | for (p = packed_git; p; p = p->next) { |
446 | int num = num_packed_objects(p); | |
447 | for (i = 0; i < num; i++) { | |
448 | unsigned char sha1[20]; | |
449 | nth_packed_object_sha1(p, i, sha1); | |
450 | if (fsck_sha1(sha1) < 0) | |
451 | fprintf(stderr, "bad sha1 entry '%s'\n", sha1_to_hex(sha1)); | |
452 | ||
453 | } | |
454 | } | |
bcee6fd8 LT |
455 | } |
456 | ||
457 | heads = 0; | |
d9839e03 | 458 | for (i = 1; i < argc; i++) { |
889262ea LT |
459 | const char *arg = argv[i]; |
460 | ||
461 | if (*arg == '-') | |
d9839e03 | 462 | continue; |
889262ea | 463 | |
3c249c95 | 464 | if (!get_sha1(arg, head_sha1)) { |
770896e5 | 465 | struct object *obj = lookup_object(head_sha1); |
e1a1388d | 466 | |
770896e5 LT |
467 | /* Error is printed by lookup_object(). */ |
468 | if (!obj) | |
e1a1388d JF |
469 | continue; |
470 | ||
ff5ebe39 DB |
471 | obj->used = 1; |
472 | mark_reachable(obj, REACHABLE); | |
bcee6fd8 | 473 | heads++; |
d9839e03 LT |
474 | continue; |
475 | } | |
889262ea | 476 | error("expected sha1, got %s", arg); |
d9839e03 | 477 | } |
d9839e03 | 478 | |
1024932f | 479 | /* |
d1af002d | 480 | * If we've not been given any explicit head information, do the |
e7bd907d LT |
481 | * default ones from .git/refs. We also consider the index file |
482 | * in this case (ie this implies --cache). | |
1024932f | 483 | */ |
e7bd907d | 484 | if (!heads) { |
1024932f LT |
485 | get_default_heads(); |
486 | keep_cache_objects = 1; | |
487 | } | |
488 | ||
ae7c0c92 JH |
489 | if (keep_cache_objects) { |
490 | int i; | |
491 | read_cache(); | |
492 | for (i = 0; i < active_nr; i++) { | |
493 | struct blob *blob = lookup_blob(active_cache[i]->sha1); | |
494 | struct object *obj; | |
495 | if (!blob) | |
496 | continue; | |
497 | obj = &blob->object; | |
498 | obj->used = 1; | |
499 | mark_reachable(obj, REACHABLE); | |
500 | } | |
501 | } | |
502 | ||
8ba0bbb2 | 503 | check_connectivity(); |
20222118 LT |
504 | return 0; |
505 | } |