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