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