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 keep_cache_objects = 0;
  17 static unsigned char head_sha1[20];
  18
  19 static void check_connectivity(void)
  20 {
  21         int i;
  22
  23         /* Look up all the requirements, warn about missing objects.. */
  24         for (i = 0; i < nr_objs; i++) {
  25                 struct object *obj = objs[i];
  26                 struct object_list *refs;
  27
  28                 if (!obj->parsed) {
  29                         if (!standalone && has_sha1_file(obj->sha1))
  30                                 ; /* it is in pack */
  31                         else
  32                                 printf("missing %s %s\n",
  33                                        obj->type, sha1_to_hex(obj->sha1));
  34                         continue;
  35                 }
  36
  37                 for (refs = obj->refs; refs; refs = refs->next) {
  38                         if (refs->item->parsed ||
  39                             (!standalone && has_sha1_file(refs->item->sha1)))
  40                                 continue;
  41                         printf("broken link from %7s %s\n",
  42                                obj->type, sha1_to_hex(obj->sha1));
  43                         printf("              to %7s %s\n",
  44                                refs->item->type, sha1_to_hex(refs->item->sha1));
  45                 }
  46
  47                 if (show_unreachable && !(obj->flags & REACHABLE)) {
  48                         printf("unreachable %s %s\n",
  49                                obj->type, sha1_to_hex(obj->sha1));
  50                         continue;
  51                 }
  52
  53                 if (!obj->used) {
  54                         printf("dangling %s %s\n", obj->type,
  55                                sha1_to_hex(obj->sha1));
  56                 }
  57         }
  58 }
  59
  60 /*
  61  * The entries in a tree are ordered in the _path_ order,
  62  * which means that a directory entry is ordered by adding
  63  * a slash to the end of it.
  64  *
  65  * So a directory called "a" is ordered _after_ a file
  66  * called "a.c", because "a/" sorts after "a.c".
  67  */
  68 #define TREE_UNORDERED (-1)
  69 #define TREE_HAS_DUPS  (-2)
  70
  71 static int verify_ordered(struct tree_entry_list *a, struct tree_entry_list *b)
  72 {
  73         int len1 = strlen(a->name);
  74         int len2 = strlen(b->name);
  75         int len = len1 < len2 ? len1 : len2;
  76         unsigned char c1, c2;
  77         int cmp;
  78
  79         cmp = memcmp(a->name, b->name, len);
  80         if (cmp < 0)
  81                 return 0;
  82         if (cmp > 0)
  83                 return TREE_UNORDERED;
  84
  85         /*
  86          * Ok, the first <len> characters are the same.
  87          * Now we need to order the next one, but turn
  88          * a '\0' into a '/' for a directory entry.
  89          */
  90         c1 = a->name[len];
  91         c2 = b->name[len];
  92         if (!c1 && !c2)
  93                 /*
  94                  * git-write-tree used to write out a nonsense tree that has
  95                  * entries with the same name, one blob and one tree.  Make
  96                  * sure we do not have duplicate entries.
  97                  */
  98                 return TREE_HAS_DUPS;
  99         if (!c1 && a->directory)
 100                 c1 = '/';
 101         if (!c2 && b->directory)
 102                 c2 = '/';
 103         return c1 < c2 ? 0 : TREE_UNORDERED;
 104 }
 105
 106 static int fsck_tree(struct tree *item)
 107 {
 108         int has_full_path = 0;
 109         struct tree_entry_list *entry, *last;
 110
 111         last = NULL;
 112         for (entry = item->entries; entry; entry = entry->next) {
 113                 if (strchr(entry->name, '/'))
 114                         has_full_path = 1;
 115
 116                 switch (entry->mode) {
 117                 /*
 118                  * Standard modes..
 119                  */
 120                 case S_IFREG | 0755:
 121                 case S_IFREG | 0644:
 122                 case S_IFLNK:
 123                 case S_IFDIR:
 124                         break;
 125                 /*
 126                  * This is nonstandard, but we had a few of these
 127                  * early on when we honored the full set of mode
 128                  * bits..
 129                  */
 130                 case S_IFREG | 0664:
 131                         break;
 132                 default:
 133                         printf("tree %s has entry %o %s\n",
 134                                 sha1_to_hex(item->object.sha1),
 135                                 entry->mode, entry->name);
 136                 }
 137
 138                 if (last) {
 139                         switch (verify_ordered(last, entry)) {
 140                         case TREE_UNORDERED:
 141                                 fprintf(stderr, "tree %s not ordered\n",
 142                                         sha1_to_hex(item->object.sha1));
 143                                 return -1;
 144                         case TREE_HAS_DUPS:
 145                                 fprintf(stderr, "tree %s has duplicate entries for '%s'\n",
 146                                         sha1_to_hex(item->object.sha1),
 147                                         entry->name);
 148                                 return -1;
 149                         default:
 150                                 break;
 151                         }
 152                 }
 153
 154                 last = entry;
 155         }
 156
 157         if (has_full_path) {
 158                 fprintf(stderr, "warning: git-fsck-cache: tree %s "
 159                         "has full pathnames in it\n",
 160                         sha1_to_hex(item->object.sha1));
 161         }
 162
 163         return 0;
 164 }
 165
 166 static int fsck_commit(struct commit *commit)
 167 {
 168         free(commit->buffer);
 169         commit->buffer = NULL;
 170         if (!commit->tree)
 171                 return -1;
 172         if (!commit->parents && show_root)
 173                 printf("root %s\n", sha1_to_hex(commit->object.sha1));
 174         if (!commit->date)
 175                 printf("bad commit date in %s\n",
 176                        sha1_to_hex(commit->object.sha1));
 177         return 0;
 178 }
 179
 180 static int fsck_tag(struct tag *tag)
 181 {
 182         struct object *tagged = tag->tagged;
 183
 184         if (!tagged) {
 185                 printf("bad object in tag %s\n", sha1_to_hex(tag->object.sha1));
 186                 return -1;
 187         }
 188         if (!show_tags)
 189                 return 0;
 190
 191         printf("tagged %s %s", tagged->type, sha1_to_hex(tagged->sha1));
 192         printf(" (%s) in %s\n", tag->tag, sha1_to_hex(tag->object.sha1));
 193         return 0;
 194 }
 195
 196 static int fsck_sha1(unsigned char *sha1)
 197 {
 198         struct object *obj = parse_object(sha1);
 199         if (!obj)
 200                 return -1;
 201         if (obj->type == blob_type)
 202                 return 0;
 203         if (obj->type == tree_type)
 204                 return fsck_tree((struct tree *) obj);
 205         if (obj->type == commit_type)
 206                 return fsck_commit((struct commit *) obj);
 207         if (obj->type == tag_type)
 208                 return fsck_tag((struct tag *) obj);
 209         return -1;
 210 }
 211
 212 /*
 213  * This is the sorting chunk size: make it reasonably
 214  * big so that we can sort well..
 215  */
 216 #define MAX_SHA1_ENTRIES (1024)
 217
 218 struct sha1_entry {
 219         unsigned long ino;
 220         unsigned char sha1[20];
 221 };
 222
 223 static struct {
 224         unsigned long nr;
 225         struct sha1_entry *entry[MAX_SHA1_ENTRIES];
 226 } sha1_list;
 227
 228 static int ino_compare(const void *_a, const void *_b)
 229 {
 230         const struct sha1_entry *a = _a, *b = _b;
 231         unsigned long ino1 = a->ino, ino2 = b->ino;
 232         return ino1 < ino2 ? -1 : ino1 > ino2 ? 1 : 0;
 233 }
 234
 235 static void fsck_sha1_list(void)
 236 {
 237         int i, nr = sha1_list.nr;
 238
 239         qsort(sha1_list.entry, nr, sizeof(struct sha1_entry *), ino_compare);
 240         for (i = 0; i < nr; i++) {
 241                 struct sha1_entry *entry = sha1_list.entry[i];
 242                 unsigned char *sha1 = entry->sha1;
 243
 244                 sha1_list.entry[i] = NULL;
 245                 if (fsck_sha1(sha1) < 0)
 246                         fprintf(stderr, "bad sha1 entry '%s'\n", sha1_to_hex(sha1));
 247                 free(entry);
 248         }
 249         sha1_list.nr = 0;
 250 }
 251
 252 static void add_sha1_list(unsigned char *sha1, unsigned long ino)
 253 {
 254         struct sha1_entry *entry = xmalloc(sizeof(*entry));
 255         int nr;
 256
 257         entry->ino = ino;
 258         memcpy(entry->sha1, sha1, 20);
 259         nr = sha1_list.nr;
 260         if (nr == MAX_SHA1_ENTRIES) {
 261                 fsck_sha1_list();
 262                 nr = 0;
 263         }
 264         sha1_list.entry[nr] = entry;
 265         sha1_list.nr = ++nr;
 266 }
 267
 268 static int fsck_dir(int i, char *path)
 269 {
 270         DIR *dir = opendir(path);
 271         struct dirent *de;
 272
 273         if (!dir) {
 274                 return error("missing sha1 directory '%s'", path);
 275         }
 276
 277         while ((de = readdir(dir)) != NULL) {
 278                 char name[100];
 279                 unsigned char sha1[20];
 280                 int len = strlen(de->d_name);
 281
 282                 switch (len) {
 283                 case 2:
 284                         if (de->d_name[1] != '.')
 285                                 break;
 286                 case 1:
 287                         if (de->d_name[0] != '.')
 288                                 break;
 289                         continue;
 290                 case 38:
 291                         sprintf(name, "%02x", i);
 292                         memcpy(name+2, de->d_name, len+1);
 293                         if (get_sha1_hex(name, sha1) < 0)
 294                                 break;
 295                         add_sha1_list(sha1, de->d_ino);
 296                         continue;
 297                 }
 298                 fprintf(stderr, "bad sha1 file: %s/%s\n", path, de->d_name);
 299         }
 300         closedir(dir);
 301         return 0;
 302 }
 303
 304 static int read_sha1_reference(const char *path)
 305 {
 306         char hexname[60];
 307         unsigned char sha1[20];
 308         int fd = open(path, O_RDONLY), len;
 309         struct object *obj;
 310
 311         if (fd < 0)
 312                 return -1;
 313
 314         len = read(fd, hexname, sizeof(hexname));
 315         close(fd);
 316         if (len < 40)
 317                 return -1;
 318
 319         if (get_sha1_hex(hexname, sha1) < 0)
 320                 return -1;
 321
 322         obj = lookup_object(sha1);
 323         if (!obj) {
 324                 if (!standalone && has_sha1_file(sha1))
 325                         return 0; /* it is in pack */
 326                 return error("%s: invalid sha1 pointer %.40s", path, hexname);
 327         }
 328
 329         obj->used = 1;
 330         mark_reachable(obj, REACHABLE);
 331         return 0;
 332 }
 333
 334 static int find_file_objects(const char *base, const char *name)
 335 {
 336         int baselen = strlen(base);
 337         int namelen = strlen(name);
 338         char *path = xmalloc(baselen + namelen + 2);
 339         struct stat st;
 340
 341         memcpy(path, base, baselen);
 342         path[baselen] = '/';
 343         memcpy(path + baselen + 1, name, namelen+1);
 344         if (stat(path, &st) < 0)
 345                 return 0;
 346
 347         /*
 348          * Recurse into directories
 349          */
 350         if (S_ISDIR(st.st_mode)) {
 351                 int count = 0;
 352                 DIR *dir = opendir(path);
 353                 if (dir) {
 354                         struct dirent *de;
 355                         while ((de = readdir(dir)) != NULL) {
 356                                 if (de->d_name[0] == '.')
 357                                         continue;
 358                                 count += find_file_objects(path, de->d_name);
 359                         }
 360                         closedir(dir);
 361                 }
 362                 return count;
 363         }
 364         if (S_ISREG(st.st_mode))
 365                 return read_sha1_reference(path) == 0;
 366         return 0;
 367 }
 368
 369 static void get_default_heads(void)
 370 {
 371         char *git_dir = gitenv(GIT_DIR_ENVIRONMENT) ? : DEFAULT_GIT_DIR_ENVIRONMENT;
 372         int count = find_file_objects(git_dir, "refs");
 373         if (!count)
 374                 die("No default references");
 375 }
 376
 377 int main(int argc, char **argv)
 378 {
 379         int i, heads;
 380         char *sha1_dir;
 381
 382         for (i = 1; i < argc; i++) {
 383                 const char *arg = argv[i];
 384
 385                 if (!strcmp(arg, "--unreachable")) {
 386                         show_unreachable = 1;
 387                         continue;
 388                 }
 389                 if (!strcmp(arg, "--tags")) {
 390                         show_tags = 1;
 391                         continue;
 392                 }
 393                 if (!strcmp(arg, "--root")) {
 394                         show_root = 1;
 395                         continue;
 396                 }
 397                 if (!strcmp(arg, "--cache")) {
 398                         keep_cache_objects = 1;
 399                         continue;
 400                 }
 401                 if (!strcmp(arg, "--standalone")) {
 402                         standalone = 1;
 403                         continue;
 404                 }
 405                 if (*arg == '-')
 406                         usage("git-fsck-cache [--tags] [[--unreachable] [--cache] <head-sha1>*]");
 407         }
 408
 409         sha1_dir = get_object_directory();
 410         for (i = 0; i < 256; i++) {
 411                 static char dir[4096];
 412                 sprintf(dir, "%s/%02x", sha1_dir, i);
 413                 fsck_dir(i, dir);
 414         }
 415         fsck_sha1_list();
 416
 417         heads = 0;
 418         for (i = 1; i < argc; i++) {
 419                 const char *arg = argv[i];
 420
 421                 if (*arg == '-')
 422                         continue;
 423
 424                 if (!get_sha1(arg, head_sha1)) {
 425                         struct object *obj = lookup_object(head_sha1);
 426
 427                         /* Error is printed by lookup_object(). */
 428                         if (!obj)
 429                                 continue;
 430
 431                         obj->used = 1;
 432                         mark_reachable(obj, REACHABLE);
 433                         heads++;
 434                         continue;
 435                 }
 436                 error("expected sha1, got %s", arg);
 437         }
 438
 439         /*
 440          * If we've not been given any explicit head information, do the
 441          * default ones from .git/refs. We also consider the index file
 442          * in this case (ie this implies --cache).
 443          */
 444         if (!heads) {
 445                 get_default_heads();
 446                 keep_cache_objects = 1;
 447         }
 448
 449         if (keep_cache_objects) {
 450                 int i;
 451                 read_cache();
 452                 for (i = 0; i < active_nr; i++) {
 453                         struct blob *blob = lookup_blob(active_cache[i]->sha1);
 454                         struct object *obj;
 455                         if (!blob)
 456                                 continue;
 457                         obj = &blob->object;
 458                         obj->used = 1;
 459                         mark_reachable(obj, REACHABLE);
 460                 }
 461         }
 462
 463         check_connectivity();
 464         return 0;
 465 }