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