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 #include "delta.h"
  10
  11 #define REACHABLE 0x0001
  12
  13 static int show_root = 0;
  14 static int show_tags = 0;
  15 static int show_unreachable = 0;
  16 static int show_max_delta_depth = 0;
  17 static int keep_cache_objects = 0;
  18 static unsigned char head_sha1[20];
  19
  20 static void expand_deltas(void)
  21 {
  22         int i, max_depth = 0;
  23
  24         /*
  25          * To be as efficient as possible we look for delta heads and
  26          * recursively process them going backward, and parsing
  27          * resulting objects along the way.  This allows for processing
  28          * each delta objects only once regardless of the delta depth.
  29          */
  30         for (i = 0; i < nr_objs; i++) {
  31                 struct object *obj = objs[i];
  32                 if (obj->parsed && !obj->delta && obj->attached_deltas) {
  33                         int depth = 0;
  34                         char type[10];
  35                         unsigned long size;
  36                         void *buf = read_sha1_file(obj->sha1, type, &size);
  37                         if (!buf)
  38                                 continue;
  39                         depth = process_deltas(buf, size, obj->type,
  40                                                obj->attached_deltas);
  41                         if (max_depth < depth)
  42                                 max_depth = depth;
  43                 }
  44         }
  45         if (show_max_delta_depth)
  46                 printf("maximum delta depth = %d\n", max_depth);
  47 }
  48
  49 static void check_connectivity(void)
  50 {
  51         int i;
  52
  53         /* Look up all the requirements, warn about missing objects.. */
  54         for (i = 0; i < nr_objs; i++) {
  55                 struct object *obj = objs[i];
  56                 struct object_list *refs;
  57
  58                 if (!obj->parsed) {
  59                         if (obj->delta)
  60                                 printf("unresolved delta %s\n",
  61                                        sha1_to_hex(obj->sha1));
  62                         else
  63                                 printf("missing %s %s\n",
  64                                        obj->type, sha1_to_hex(obj->sha1));
  65                         continue;
  66                 }
  67
  68                 for (refs = obj->refs; refs; refs = refs->next) {
  69                         if (refs->item->parsed)
  70                                 continue;
  71                         printf("broken link from %7s %s\n",
  72                                obj->type, sha1_to_hex(obj->sha1));
  73                         printf("              to %7s %s\n",
  74                                refs->item->type, sha1_to_hex(refs->item->sha1));
  75                 }
  76
  77                 /* Don't bother with tag reachability. */
  78                 if (obj->type == tag_type)
  79                         continue;
  80
  81                 if (show_unreachable && !(obj->flags & REACHABLE)) {
  82                         if (obj->attached_deltas)
  83                                 printf("foreign delta reference %s\n",
  84                                        sha1_to_hex(obj->sha1));
  85                         else
  86                                 printf("unreachable %s %s\n",
  87                                        obj->type, sha1_to_hex(obj->sha1));
  88                         continue;
  89                 }
  90
  91                 if (!obj->used) {
  92                         printf("dangling %s %s\n", obj->type,
  93                                sha1_to_hex(obj->sha1));
  94                 }
  95         }
  96 }
  97
  98 /*
  99  * The entries in a tree are ordered in the _path_ order,
 100  * which means that a directory entry is ordered by adding
 101  * a slash to the end of it.
 102  *
 103  * So a directory called "a" is ordered _after_ a file
 104  * called "a.c", because "a/" sorts after "a.c".
 105  */
 106 #define TREE_UNORDERED (-1)
 107 #define TREE_HAS_DUPS  (-2)
 108
 109 static int verify_ordered(struct tree_entry_list *a, struct tree_entry_list *b)
 110 {
 111         int len1 = strlen(a->name);
 112         int len2 = strlen(b->name);
 113         int len = len1 < len2 ? len1 : len2;
 114         unsigned char c1, c2;
 115         int cmp;
 116
 117         cmp = memcmp(a->name, b->name, len);
 118         if (cmp < 0)
 119                 return 0;
 120         if (cmp > 0)
 121                 return TREE_UNORDERED;
 122
 123         /*
 124          * Ok, the first <len> characters are the same.
 125          * Now we need to order the next one, but turn
 126          * a '\0' into a '/' for a directory entry.
 127          */
 128         c1 = a->name[len];
 129         c2 = b->name[len];
 130         if (!c1 && !c2)
 131                 /*
 132                  * git-write-tree used to write out a nonsense tree that has
 133                  * entries with the same name, one blob and one tree.  Make
 134                  * sure we do not have duplicate entries.
 135                  */
 136                 return TREE_HAS_DUPS;
 137         if (!c1 && a->directory)
 138                 c1 = '/';
 139         if (!c2 && b->directory)
 140                 c2 = '/';
 141         return c1 < c2 ? 0 : TREE_UNORDERED;
 142 }
 143
 144 static int fsck_tree(struct tree *item)
 145 {
 146         int has_full_path = 0;
 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;
 153
 154                 switch (entry->mode) {
 155                 /*
 156                  * Standard modes..
 157                  */
 158                 case S_IFREG | 0755:
 159                 case S_IFREG | 0644:
 160                 case S_IFLNK:
 161                 case S_IFDIR:
 162                         break;
 163                 /*
 164                  * This is nonstandard, but we had a few of these
 165                  * early on when we honored the full set of mode
 166                  * bits..
 167                  */
 168                 case S_IFREG | 0664:
 169                         break;
 170                 default:
 171                         printf("tree %s has entry %o %s\n",
 172                                 sha1_to_hex(item->object.sha1),
 173                                 entry->mode, entry->name);
 174                 }
 175
 176                 if (last) {
 177                         switch (verify_ordered(last, entry)) {
 178                         case TREE_UNORDERED:
 179                                 fprintf(stderr, "tree %s not ordered\n",
 180                                         sha1_to_hex(item->object.sha1));
 181                                 return -1;
 182                         case TREE_HAS_DUPS:
 183                                 fprintf(stderr, "tree %s has duplicate entries for '%s'\n",
 184                                         sha1_to_hex(item->object.sha1),
 185                                         entry->name);
 186                                 return -1;
 187                         default:
 188                                 break;
 189                         }
 190                 }
 191
 192                 last = entry;
 193         }
 194
 195         if (has_full_path) {
 196                 fprintf(stderr, "warning: git-fsck-cache: tree %s "
 197                         "has full pathnames in it\n",
 198                         sha1_to_hex(item->object.sha1));
 199         }
 200
 201         return 0;
 202 }
 203
 204 static int fsck_commit(struct commit *commit)
 205 {
 206         free(commit->buffer);
 207         commit->buffer = NULL;
 208         if (!commit->tree)
 209                 return -1;
 210         if (!commit->parents && show_root)
 211                 printf("root %s\n", sha1_to_hex(commit->object.sha1));
 212         if (!commit->date)
 213                 printf("bad commit date in %s\n",
 214                        sha1_to_hex(commit->object.sha1));
 215         return 0;
 216 }
 217
 218 static int fsck_tag(struct tag *tag)
 219 {
 220         struct object *tagged = tag->tagged;
 221
 222         if (!tagged) {
 223                 printf("bad object in tag %s\n", sha1_to_hex(tag->object.sha1));
 224                 return -1;
 225         }
 226         if (!show_tags)
 227                 return 0;
 228
 229         printf("tagged %s %s", tagged->type, sha1_to_hex(tagged->sha1));
 230         printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
 231         return 0;
 232 }
 233
 234 static int fsck_sha1(unsigned char *sha1)
 235 {
 236         struct object *obj = parse_object(sha1);
 237         if (!obj)
 238                 return -1;
 239         if (obj->type == blob_type)
 240                 return 0;
 241         if (obj->type == tree_type)
 242                 return fsck_tree((struct tree *) obj);
 243         if (obj->type == commit_type)
 244                 return fsck_commit((struct commit *) obj);
 245         if (obj->type == tag_type)
 246                 return fsck_tag((struct tag *) obj);
 247         if (!obj->type && obj->delta)
 248                 return 0;
 249         return -1;
 250 }
 251
 252 /*
 253  * This is the sorting chunk size: make it reasonably
 254  * big so that we can sort well..
 255  */
 256 #define MAX_SHA1_ENTRIES (1024)
 257
 258 struct sha1_entry {
 259         unsigned long ino;
 260         unsigned char sha1[20];
 261 };
 262
 263 static struct {
 264         unsigned long nr;
 265         struct sha1_entry *entry[MAX_SHA1_ENTRIES];
 266 } sha1_list;
 267
 268 static int ino_compare(const void *_a, const void *_b)
 269 {
 270         const struct sha1_entry *a = _a, *b = _b;
 271         unsigned long ino1 = a->ino, ino2 = b->ino;
 272         return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0;
 273 }
 274
 275 static void fsck_sha1_list(void)
 276 {
 277         int i, nr = sha1_list.nr;
 278
 279         qsort(sha1_list.entry, nr, sizeof(struct sha1_entry *), ino_compare);
 280         for (i = 0; i < nr; i++) {
 281                 struct sha1_entry *entry = sha1_list.entry[i];
 282                 unsigned char *sha1 = entry->sha1;
 283
 284                 sha1_list.entry[i] = NULL;
 285                 if (fsck_sha1(sha1) < 0)
 286                         fprintf(stderr, "bad sha1 entry '%s'\n", sha1_to_hex(sha1));
 287                 free(entry);
 288         }
 289         sha1_list.nr = 0;
 290 }
 291
 292 static void add_sha1_list(unsigned char *sha1, unsigned long ino)
 293 {
 294         struct sha1_entry *entry = xmalloc(sizeof(*entry));
 295         int nr;
 296
 297         entry->ino = ino;
 298         memcpy(entry->sha1, sha1, 20);
 299         nr = sha1_list.nr;
 300         if (nr == MAX_SHA1_ENTRIES) {
 301                 fsck_sha1_list();
 302                 nr = 0;
 303         }
 304         sha1_list.entry[nr] = entry;
 305         sha1_list.nr = ++nr;
 306 }
 307
 308 static int fsck_dir(int i, char *path)
 309 {
 310         DIR *dir = opendir(path);
 311         struct dirent *de;
 312
 313         if (!dir) {
 314                 return error("missing sha1 directory '%s'", path);
 315         }
 316
 317         while ((de = readdir(dir)) != NULL) {
 318                 char name[100];
 319                 unsigned char sha1[20];
 320                 int len = strlen(de->d_name);
 321
 322                 switch (len) {
 323                 case 2:
 324                         if (de->d_name[1] != '.')
 325                                 break;
 326                 case 1:
 327                         if (de->d_name[0] != '.')
 328                                 break;
 329                         continue;
 330                 case 38:
 331                         sprintf(name, "%02x", i);
 332                         memcpy(name+2, de->d_name, len+1);
 333                         if (get_sha1_hex(name, sha1) < 0)
 334                                 break;
 335                         add_sha1_list(sha1, de->d_ino);
 336                         continue;
 337                 }
 338                 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
 339         }
 340         closedir(dir);
 341         return 0;
 342 }
 343
 344 static int read_sha1_reference(const char *path)
 345 {
 346         char hexname[60];
 347         unsigned char sha1[20];
 348         int fd = open(path, O_RDONLY), len;
 349         struct object *obj;
 350
 351         if (fd < 0)
 352                 return -1;
 353
 354         len = read(fd, hexname, sizeof(hexname));
 355         close(fd);
 356         if (len < 40)
 357                 return -1;
 358
 359         if (get_sha1_hex(hexname, sha1) < 0)
 360                 return -1;
 361
 362         obj = lookup_object(sha1);
 363         if (!obj)
 364                 return error("%s: invalid sha1 pointer %.40s", path, hexname);
 365
 366         obj->used = 1;
 367         mark_reachable(obj, REACHABLE);
 368         return 0;
 369 }
 370
 371 static void find_file_objects(const char *base, const char *name)
 372 {
 373         int baselen = strlen(base);
 374         int namelen = strlen(name);
 375         char *path = xmalloc(baselen + namelen + 2);
 376         struct stat st;
 377
 378         memcpy(path, base, baselen);
 379         path[baselen] = '/';
 380         memcpy(path + baselen + 1, name, namelen+1);
 381         if (stat(path, &st) < 0)
 382                 return;
 383
 384         /*
 385          * Recurse into directories
 386          */
 387         if (S_ISDIR(st.st_mode)) {
 388                 DIR *dir = opendir(path);
 389                 if (dir) {
 390                         struct dirent *de;
 391                         while ((de = readdir(dir)) != NULL) {
 392                                 if (de->d_name[0] == '.')
 393                                         continue;
 394                                 find_file_objects(path, de->d_name);
 395                         }
 396                         closedir(dir);
 397                 }
 398                 return;
 399         }
 400         if (S_ISREG(st.st_mode)) {
 401                 read_sha1_reference(path);
 402                 return;
 403         }
 404 }
 405
 406 static void get_default_heads(void)
 407 {
 408         char *git_dir = gitenv(GIT_DIR_ENVIRONMENT) ? : DEFAULT_GIT_DIR_ENVIRONMENT;
 409         find_file_objects(git_dir, "refs");
 410 }
 411
 412 int main(int argc, char **argv)
 413 {
 414         int i, heads;
 415         char *sha1_dir;
 416
 417         for (i = 1; i < argc; i++) {
 418                 const char *arg = argv[i];
 419
 420                 if (!strcmp(arg, "--unreachable")) {
 421                         show_unreachable = 1;
 422                         continue;
 423                 }
 424                 if (!strcmp(arg, "--tags")) {
 425                         show_tags = 1;
 426                         continue;
 427                 }
 428                 if (!strcmp(arg, "--root")) {
 429                         show_root = 1;
 430                         continue;
 431                 }
 432                 if (!strcmp(arg, "--delta-depth")) {
 433                         show_max_delta_depth = 1;
 434                         continue;
 435                 }
 436                 if (!strcmp(arg, "--cache")) {
 437                         keep_cache_objects = 1;
 438                         continue;
 439                 }
 440                 if (*arg == '-')
 441                         usage("git-fsck-cache [--tags] [[--unreachable] [--cache] <head-sha1>*]");
 442         }
 443
 444         sha1_dir = get_object_directory();
 445         for (i = 0; i < 256; i++) {
 446                 static char dir[4096];
 447                 sprintf(dir, "%s/%02x", sha1_dir, i);
 448                 fsck_dir(i, dir);
 449         }
 450         fsck_sha1_list();
 451
 452         expand_deltas();
 453
 454         heads = 0;
 455         for (i = 1; i < argc; i++) {
 456                 const char *arg = argv[i];
 457
 458                 if (*arg == '-')
 459                         continue;
 460
 461                 if (!get_sha1(arg, head_sha1)) {
 462                         struct object *obj = lookup_object(head_sha1);
 463
 464                         /* Error is printed by lookup_object(). */
 465                         if (!obj)
 466                                 continue;
 467
 468                         obj->used = 1;
 469                         mark_reachable(obj, REACHABLE);
 470                         heads++;
 471                         continue;
 472                 }
 473                 error("expected sha1, got %s", arg);
 474         }
 475
 476         /*
 477          * If we've not been given any explicit head information, do the
 478          * default ones from .git/refs. We also consider the index file
 479          * in this case (ie this implies --cache).
 480          */
 481         if (!heads) {
 482                 get_default_heads();
 483                 keep_cache_objects = 1;
 484         }
 485
 486         if (keep_cache_objects) {
 487                 int i;
 488                 read_cache();
 489                 for (i = 0; i < active_nr; i++) {
 490                         struct blob *blob = lookup_blob(active_cache[i]->sha1);
 491                         struct object *obj;
 492                         if (!blob)
 493                                 continue;
 494                         obj = &blob->object;
 495                         obj->used = 1;
 496                         mark_reachable(obj, REACHABLE);
 497                 }
 498         }
 499
 500         check_connectivity();
 501         return 0;
 502 }