read-cache.c

   1 /*
   2  * GIT - The information manager from hell
   3  *
   4  * Copyright (C) Linus Torvalds, 2005
   5  */
   6 #include "cache.h"
   7
   8 int trust_executable_bit = 1;
   9 struct cache_entry **active_cache = NULL;
  10 unsigned int active_nr = 0, active_alloc = 0, active_cache_changed = 0;
  11
  12 /*
  13  * This only updates the "non-critical" parts of the directory
  14  * cache, ie the parts that aren't tracked by GIT, and only used
  15  * to validate the cache.
  16  */
  17 void fill_stat_cache_info(struct cache_entry *ce, struct stat *st)
  18 {
  19         ce->ce_ctime.sec = htonl(st->st_ctime);
  20         ce->ce_mtime.sec = htonl(st->st_mtime);
  21 #ifdef USE_NSEC
  22         ce->ce_ctime.nsec = htonl(st->st_ctim.tv_nsec);
  23         ce->ce_mtime.nsec = htonl(st->st_mtim.tv_nsec);
  24 #endif
  25         ce->ce_dev = htonl(st->st_dev);
  26         ce->ce_ino = htonl(st->st_ino);
  27         ce->ce_uid = htonl(st->st_uid);
  28         ce->ce_gid = htonl(st->st_gid);
  29         ce->ce_size = htonl(st->st_size);
  30 }
  31
  32 int ce_match_stat(struct cache_entry *ce, struct stat *st)
  33 {
  34         unsigned int changed = 0;
  35
  36         switch (ntohl(ce->ce_mode) & S_IFMT) {
  37         case S_IFREG:
  38                 changed |= !S_ISREG(st->st_mode) ? TYPE_CHANGED : 0;
  39                 /* We consider only the owner x bit to be relevant for
  40                  * "mode changes"
  41                  */
  42                 if (trust_executable_bit &&
  43                     (0100 & (ntohl(ce->ce_mode) ^ st->st_mode)))
  44                         changed |= MODE_CHANGED;
  45                 break;
  46         case S_IFLNK:
  47                 changed |= !S_ISLNK(st->st_mode) ? TYPE_CHANGED : 0;
  48                 break;
  49         default:
  50                 die("internal error: ce_mode is %o", ntohl(ce->ce_mode));
  51         }
  52         if (ce->ce_mtime.sec != htonl(st->st_mtime))
  53                 changed |= MTIME_CHANGED;
  54         if (ce->ce_ctime.sec != htonl(st->st_ctime))
  55                 changed |= CTIME_CHANGED;
  56
  57 #ifdef USE_NSEC
  58         /*
  59          * nsec seems unreliable - not all filesystems support it, so
  60          * as long as it is in the inode cache you get right nsec
  61          * but after it gets flushed, you get zero nsec.
  62          */
  63         if (ce->ce_mtime.nsec != htonl(st->st_mtim.tv_nsec))
  64                 changed |= MTIME_CHANGED;
  65         if (ce->ce_ctime.nsec != htonl(st->st_ctim.tv_nsec))
  66                 changed |= CTIME_CHANGED;
  67 #endif
  68
  69         if (ce->ce_uid != htonl(st->st_uid) ||
  70             ce->ce_gid != htonl(st->st_gid))
  71                 changed |= OWNER_CHANGED;
  72         if (ce->ce_ino != htonl(st->st_ino))
  73                 changed |= INODE_CHANGED;
  74
  75 #ifdef USE_STDEV
  76         /*
  77          * st_dev breaks on network filesystems where different
  78          * clients will have different views of what "device"
  79          * the filesystem is on
  80          */
  81         if (ce->ce_dev != htonl(st->st_dev))
  82                 changed |= INODE_CHANGED;
  83 #endif
  84
  85         if (ce->ce_size != htonl(st->st_size))
  86                 changed |= DATA_CHANGED;
  87         return changed;
  88 }
  89
  90 static int ce_compare_data(struct cache_entry *ce, struct stat *st)
  91 {
  92         int match = -1;
  93         int fd = open(ce->name, O_RDONLY);
  94
  95         if (fd >= 0) {
  96                 unsigned char sha1[20];
  97                 if (!index_fd(sha1, fd, st, 0, NULL))
  98                         match = memcmp(sha1, ce->sha1, 20);
  99                 close(fd);
 100         }
 101         return match;
 102 }
 103
 104 static int ce_compare_link(struct cache_entry *ce, unsigned long expected_size)
 105 {
 106         int match = -1;
 107         char *target;
 108         void *buffer;
 109         unsigned long size;
 110         char type[10];
 111         int len;
 112
 113         target = xmalloc(expected_size);
 114         len = readlink(ce->name, target, expected_size);
 115         if (len != expected_size) {
 116                 free(target);
 117                 return -1;
 118         }
 119         buffer = read_sha1_file(ce->sha1, type, &size);
 120         if (!buffer) {
 121                 free(target);
 122                 return -1;
 123         }
 124         if (size == expected_size)
 125                 match = memcmp(buffer, target, size);
 126         free(buffer);
 127         free(target);
 128         return match;
 129 }
 130
 131 int ce_modified(struct cache_entry *ce, struct stat *st)
 132 {
 133         int changed;
 134         changed = ce_match_stat(ce, st);
 135         if (!changed)
 136                 return 0;
 137
 138         /*
 139          * If the mode or type has changed, there's no point in trying
 140          * to refresh the entry - it's not going to match
 141          */
 142         if (changed & (MODE_CHANGED | TYPE_CHANGED))
 143                 return changed;
 144
 145         /* Immediately after read-tree or update-index --cacheinfo,
 146          * the length field is zero.  For other cases the ce_size
 147          * should match the SHA1 recorded in the index entry.
 148          */
 149         if ((changed & DATA_CHANGED) && ce->ce_size != htonl(0))
 150                 return changed;
 151
 152         switch (st->st_mode & S_IFMT) {
 153         case S_IFREG:
 154                 if (ce_compare_data(ce, st))
 155                         return changed | DATA_CHANGED;
 156                 break;
 157         case S_IFLNK:
 158                 if (ce_compare_link(ce, st->st_size))
 159                         return changed | DATA_CHANGED;
 160                 break;
 161         default:
 162                 return changed | TYPE_CHANGED;
 163         }
 164         return 0;
 165 }
 166
 167 int base_name_compare(const char *name1, int len1, int mode1,
 168                       const char *name2, int len2, int mode2)
 169 {
 170         unsigned char c1, c2;
 171         int len = len1 < len2 ? len1 : len2;
 172         int cmp;
 173
 174         cmp = memcmp(name1, name2, len);
 175         if (cmp)
 176                 return cmp;
 177         c1 = name1[len];
 178         c2 = name2[len];
 179         if (!c1 && S_ISDIR(mode1))
 180                 c1 = '/';
 181         if (!c2 && S_ISDIR(mode2))
 182                 c2 = '/';
 183         return (c1 < c2) ? -1 : (c1 > c2) ? 1 : 0;
 184 }
 185
 186 int cache_name_compare(const char *name1, int flags1, const char *name2, int flags2)
 187 {
 188         int len1 = flags1 & CE_NAMEMASK;
 189         int len2 = flags2 & CE_NAMEMASK;
 190         int len = len1 < len2 ? len1 : len2;
 191         int cmp;
 192
 193         cmp = memcmp(name1, name2, len);
 194         if (cmp)
 195                 return cmp;
 196         if (len1 < len2)
 197                 return -1;
 198         if (len1 > len2)
 199                 return 1;
 200         if (flags1 < flags2)
 201                 return -1;
 202         if (flags1 > flags2)
 203                 return 1;
 204         return 0;
 205 }
 206
 207 int cache_name_pos(const char *name, int namelen)
 208 {
 209         int first, last;
 210
 211         first = 0;
 212         last = active_nr;
 213         while (last > first) {
 214                 int next = (last + first) >> 1;
 215                 struct cache_entry *ce = active_cache[next];
 216                 int cmp = cache_name_compare(name, namelen, ce->name, ntohs(ce->ce_flags));
 217                 if (!cmp)
 218                         return next;
 219                 if (cmp < 0) {
 220                         last = next;
 221                         continue;
 222                 }
 223                 first = next+1;
 224         }
 225         return -first-1;
 226 }
 227
 228 /* Remove entry, return true if there are more entries to go.. */
 229 int remove_cache_entry_at(int pos)
 230 {
 231         active_cache_changed = 1;
 232         active_nr--;
 233         if (pos >= active_nr)
 234                 return 0;
 235         memmove(active_cache + pos, active_cache + pos + 1, (active_nr - pos) * sizeof(struct cache_entry *));
 236         return 1;
 237 }
 238
 239 int remove_file_from_cache(const char *path)
 240 {
 241         int pos = cache_name_pos(path, strlen(path));
 242         if (pos < 0)
 243                 pos = -pos-1;
 244         while (pos < active_nr && !strcmp(active_cache[pos]->name, path))
 245                 remove_cache_entry_at(pos);
 246         return 0;
 247 }
 248
 249 int ce_same_name(struct cache_entry *a, struct cache_entry *b)
 250 {
 251         int len = ce_namelen(a);
 252         return ce_namelen(b) == len && !memcmp(a->name, b->name, len);
 253 }
 254
 255 int ce_path_match(const struct cache_entry *ce, const char **pathspec)
 256 {
 257         const char *match, *name;
 258         int len;
 259
 260         if (!pathspec)
 261                 return 1;
 262
 263         len = ce_namelen(ce);
 264         name = ce->name;
 265         while ((match = *pathspec++) != NULL) {
 266                 int matchlen = strlen(match);
 267                 if (matchlen > len)
 268                         continue;
 269                 if (memcmp(name, match, matchlen))
 270                         continue;
 271                 if (matchlen && name[matchlen-1] == '/')
 272                         return 1;
 273                 if (name[matchlen] == '/' || !name[matchlen])
 274                         return 1;
 275                 if (!matchlen)
 276                         return 1;
 277         }
 278         return 0;
 279 }
 280
 281 /*
 282  * Do we have another file that has the beginning components being a
 283  * proper superset of the name we're trying to add?
 284  */
 285 static int has_file_name(const struct cache_entry *ce, int pos, int ok_to_replace)
 286 {
 287         int retval = 0;
 288         int len = ce_namelen(ce);
 289         int stage = ce_stage(ce);
 290         const char *name = ce->name;
 291
 292         while (pos < active_nr) {
 293                 struct cache_entry *p = active_cache[pos++];
 294
 295                 if (len >= ce_namelen(p))
 296                         break;
 297                 if (memcmp(name, p->name, len))
 298                         break;
 299                 if (ce_stage(p) != stage)
 300                         continue;
 301                 if (p->name[len] != '/')
 302                         continue;
 303                 retval = -1;
 304                 if (!ok_to_replace)
 305                         break;
 306                 remove_cache_entry_at(--pos);
 307         }
 308         return retval;
 309 }
 310
 311 /*
 312  * Do we have another file with a pathname that is a proper
 313  * subset of the name we're trying to add?
 314  */
 315 static int has_dir_name(const struct cache_entry *ce, int pos, int ok_to_replace)
 316 {
 317         int retval = 0;
 318         int stage = ce_stage(ce);
 319         const char *name = ce->name;
 320         const char *slash = name + ce_namelen(ce);
 321
 322         for (;;) {
 323                 int len;
 324
 325                 for (;;) {
 326                         if (*--slash == '/')
 327                                 break;
 328                         if (slash <= ce->name)
 329                                 return retval;
 330                 }
 331                 len = slash - name;
 332
 333                 pos = cache_name_pos(name, ntohs(create_ce_flags(len, stage)));
 334                 if (pos >= 0) {
 335                         retval = -1;
 336                         if (ok_to_replace)
 337                                 break;
 338                         remove_cache_entry_at(pos);
 339                         continue;
 340                 }
 341
 342                 /*
 343                  * Trivial optimization: if we find an entry that
 344                  * already matches the sub-directory, then we know
 345                  * we're ok, and we can exit.
 346                  */
 347                 pos = -pos-1;
 348                 while (pos < active_nr) {
 349                         struct cache_entry *p = active_cache[pos];
 350                         if ((ce_namelen(p) <= len) ||
 351                             (p->name[len] != '/') ||
 352                             memcmp(p->name, name, len))
 353                                 break; /* not our subdirectory */
 354                         if (ce_stage(p) == stage)
 355                                 /* p is at the same stage as our entry, and
 356                                  * is a subdirectory of what we are looking
 357                                  * at, so we cannot have conflicts at our
 358                                  * level or anything shorter.
 359                                  */
 360                                 return retval;
 361                         pos++;
 362                 }
 363         }
 364         return retval;
 365 }
 366
 367 /* We may be in a situation where we already have path/file and path
 368  * is being added, or we already have path and path/file is being
 369  * added.  Either one would result in a nonsense tree that has path
 370  * twice when git-write-tree tries to write it out.  Prevent it.
 371  *
 372  * If ok-to-replace is specified, we remove the conflicting entries
 373  * from the cache so the caller should recompute the insert position.
 374  * When this happens, we return non-zero.
 375  */
 376 static int check_file_directory_conflict(const struct cache_entry *ce, int pos, int ok_to_replace)
 377 {
 378         /*
 379          * We check if the path is a sub-path of a subsequent pathname
 380          * first, since removing those will not change the position
 381          * in the array
 382          */
 383         int retval = has_file_name(ce, pos, ok_to_replace);
 384         /*
 385          * Then check if the path might have a clashing sub-directory
 386          * before it.
 387          */
 388         return retval + has_dir_name(ce, pos, ok_to_replace);
 389 }
 390
 391 int add_cache_entry(struct cache_entry *ce, int option)
 392 {
 393         int pos;
 394         int ok_to_add = option & ADD_CACHE_OK_TO_ADD;
 395         int ok_to_replace = option & ADD_CACHE_OK_TO_REPLACE;
 396         int skip_df_check = option & ADD_CACHE_SKIP_DFCHECK;
 397         pos = cache_name_pos(ce->name, ntohs(ce->ce_flags));
 398
 399         /* existing match? Just replace it. */
 400         if (pos >= 0) {
 401                 active_cache_changed = 1;
 402                 active_cache[pos] = ce;
 403                 return 0;
 404         }
 405         pos = -pos-1;
 406
 407         /*
 408          * Inserting a merged entry ("stage 0") into the index
 409          * will always replace all non-merged entries..
 410          */
 411         if (pos < active_nr && ce_stage(ce) == 0) {
 412                 while (ce_same_name(active_cache[pos], ce)) {
 413                         ok_to_add = 1;
 414                         if (!remove_cache_entry_at(pos))
 415                                 break;
 416                 }
 417         }
 418
 419         if (!ok_to_add)
 420                 return -1;
 421
 422         if (!skip_df_check &&
 423             check_file_directory_conflict(ce, pos, ok_to_replace)) {
 424                 if (!ok_to_replace)
 425                         return -1;
 426                 pos = cache_name_pos(ce->name, ntohs(ce->ce_flags));
 427                 pos = -pos-1;
 428         }
 429
 430         /* Make sure the array is big enough .. */
 431         if (active_nr == active_alloc) {
 432                 active_alloc = alloc_nr(active_alloc);
 433                 active_cache = xrealloc(active_cache, active_alloc * sizeof(struct cache_entry *));
 434         }
 435
 436         /* Add it in.. */
 437         active_nr++;
 438         if (active_nr > pos)
 439                 memmove(active_cache + pos + 1, active_cache + pos, (active_nr - pos - 1) * sizeof(ce));
 440         active_cache[pos] = ce;
 441         active_cache_changed = 1;
 442         return 0;
 443 }
 444
 445 static int verify_hdr(struct cache_header *hdr, unsigned long size)
 446 {
 447         SHA_CTX c;
 448         unsigned char sha1[20];
 449
 450         if (hdr->hdr_signature != htonl(CACHE_SIGNATURE))
 451                 return error("bad signature");
 452         if (hdr->hdr_version != htonl(2))
 453                 return error("bad index version");
 454         SHA1_Init(&c);
 455         SHA1_Update(&c, hdr, size - 20);
 456         SHA1_Final(sha1, &c);
 457         if (memcmp(sha1, (void *)hdr + size - 20, 20))
 458                 return error("bad index file sha1 signature");
 459         return 0;
 460 }
 461
 462 int read_cache(void)
 463 {
 464         int fd, i;
 465         struct stat st;
 466         unsigned long size, offset;
 467         void *map;
 468         struct cache_header *hdr;
 469
 470         errno = EBUSY;
 471         if (active_cache)
 472                 return active_nr;
 473
 474         errno = ENOENT;
 475         fd = open(get_index_file(), O_RDONLY);
 476         if (fd < 0) {
 477                 if (errno == ENOENT)
 478                         return 0;
 479                 die("index file open failed (%s)", strerror(errno));
 480         }
 481
 482         size = 0; // avoid gcc warning
 483         map = MAP_FAILED;
 484         if (!fstat(fd, &st)) {
 485                 size = st.st_size;
 486                 errno = EINVAL;
 487                 if (size >= sizeof(struct cache_header) + 20)
 488                         map = mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0);
 489         }
 490         close(fd);
 491         if (map == MAP_FAILED)
 492                 die("index file mmap failed (%s)", strerror(errno));
 493
 494         hdr = map;
 495         if (verify_hdr(hdr, size) < 0)
 496                 goto unmap;
 497
 498         active_nr = ntohl(hdr->hdr_entries);
 499         active_alloc = alloc_nr(active_nr);
 500         active_cache = calloc(active_alloc, sizeof(struct cache_entry *));
 501
 502         offset = sizeof(*hdr);
 503         for (i = 0; i < active_nr; i++) {
 504                 struct cache_entry *ce = map + offset;
 505                 offset = offset + ce_size(ce);
 506                 active_cache[i] = ce;
 507         }
 508         return active_nr;
 509
 510 unmap:
 511         munmap(map, size);
 512         errno = EINVAL;
 513         die("index file corrupt");
 514 }
 515
 516 #define WRITE_BUFFER_SIZE 8192
 517 static unsigned char write_buffer[WRITE_BUFFER_SIZE];
 518 static unsigned long write_buffer_len;
 519
 520 static int ce_write(SHA_CTX *context, int fd, void *data, unsigned int len)
 521 {
 522         while (len) {
 523                 unsigned int buffered = write_buffer_len;
 524                 unsigned int partial = WRITE_BUFFER_SIZE - buffered;
 525                 if (partial > len)
 526                         partial = len;
 527                 memcpy(write_buffer + buffered, data, partial);
 528                 buffered += partial;
 529                 if (buffered == WRITE_BUFFER_SIZE) {
 530                         SHA1_Update(context, write_buffer, WRITE_BUFFER_SIZE);
 531                         if (write(fd, write_buffer, WRITE_BUFFER_SIZE) != WRITE_BUFFER_SIZE)
 532                                 return -1;
 533                         buffered = 0;
 534                 }
 535                 write_buffer_len = buffered;
 536                 len -= partial;
 537                 data += partial;
 538         }
 539         return 0;
 540 }
 541
 542 static int ce_flush(SHA_CTX *context, int fd)
 543 {
 544         unsigned int left = write_buffer_len;
 545
 546         if (left) {
 547                 write_buffer_len = 0;
 548                 SHA1_Update(context, write_buffer, left);
 549         }
 550
 551         /* Flush first if not enough space for SHA1 signature */
 552         if (left + 20 > WRITE_BUFFER_SIZE) {
 553                 if (write(fd, write_buffer, left) != left)
 554                         return -1;
 555                 left = 0;
 556         }
 557
 558         /* Append the SHA1 signature at the end */
 559         SHA1_Final(write_buffer + left, context);
 560         left += 20;
 561         if (write(fd, write_buffer, left) != left)
 562                 return -1;
 563         return 0;
 564 }
 565
 566 int write_cache(int newfd, struct cache_entry **cache, int entries)
 567 {
 568         SHA_CTX c;
 569         struct cache_header hdr;
 570         int i, removed;
 571
 572         for (i = removed = 0; i < entries; i++)
 573                 if (!cache[i]->ce_mode)
 574                         removed++;
 575
 576         hdr.hdr_signature = htonl(CACHE_SIGNATURE);
 577         hdr.hdr_version = htonl(2);
 578         hdr.hdr_entries = htonl(entries - removed);
 579
 580         SHA1_Init(&c);
 581         if (ce_write(&c, newfd, &hdr, sizeof(hdr)) < 0)
 582                 return -1;
 583
 584         for (i = 0; i < entries; i++) {
 585                 struct cache_entry *ce = cache[i];
 586                 if (!ce->ce_mode)
 587                         continue;
 588                 if (ce_write(&c, newfd, ce, ce_size(ce)) < 0)
 589                         return -1;
 590         }
 591         return ce_flush(&c, newfd);
 592 }