cache-tree.c

   1 #include "cache.h"
   2 #include "tree.h"
   3 #include "tree-walk.h"
   4 #include "cache-tree.h"
   5
   6 #ifndef DEBUG
   7 #define DEBUG 0
   8 #endif
   9
  10 struct cache_tree *cache_tree(void)
  11 {
  12         struct cache_tree *it = xcalloc(1, sizeof(struct cache_tree));
  13         it->entry_count = -1;
  14         return it;
  15 }
  16
  17 void cache_tree_free(struct cache_tree **it_p)
  18 {
  19         int i;
  20         struct cache_tree *it = *it_p;
  21
  22         if (!it)
  23                 return;
  24         for (i = 0; i < it->subtree_nr; i++)
  25                 if (it->down[i]) {
  26                         cache_tree_free(&it->down[i]->cache_tree);
  27                         free(it->down[i]);
  28                 }
  29         free(it->down);
  30         free(it);
  31         *it_p = NULL;
  32 }
  33
  34 static int subtree_name_cmp(const char *one, int onelen,
  35                             const char *two, int twolen)
  36 {
  37         if (onelen < twolen)
  38                 return -1;
  39         if (twolen < onelen)
  40                 return 1;
  41         return memcmp(one, two, onelen);
  42 }
  43
  44 static int subtree_pos(struct cache_tree *it, const char *path, int pathlen)
  45 {
  46         struct cache_tree_sub **down = it->down;
  47         int lo, hi;
  48         lo = 0;
  49         hi = it->subtree_nr;
  50         while (lo < hi) {
  51                 int mi = (lo + hi) / 2;
  52                 struct cache_tree_sub *mdl = down[mi];
  53                 int cmp = subtree_name_cmp(path, pathlen,
  54                                            mdl->name, mdl->namelen);
  55                 if (!cmp)
  56                         return mi;
  57                 if (cmp < 0)
  58                         hi = mi;
  59                 else
  60                         lo = mi + 1;
  61         }
  62         return -lo-1;
  63 }
  64
  65 static struct cache_tree_sub *find_subtree(struct cache_tree *it,
  66                                            const char *path,
  67                                            int pathlen,
  68                                            int create)
  69 {
  70         struct cache_tree_sub *down;
  71         int pos = subtree_pos(it, path, pathlen);
  72         if (0 <= pos)
  73                 return it->down[pos];
  74         if (!create)
  75                 return NULL;
  76
  77         pos = -pos-1;
  78         if (it->subtree_alloc <= it->subtree_nr) {
  79                 it->subtree_alloc = alloc_nr(it->subtree_alloc);
  80                 it->down = xrealloc(it->down, it->subtree_alloc *
  81                                     sizeof(*it->down));
  82         }
  83         it->subtree_nr++;
  84
  85         down = xmalloc(sizeof(*down) + pathlen + 1);
  86         down->cache_tree = NULL;
  87         down->namelen = pathlen;
  88         memcpy(down->name, path, pathlen);
  89         down->name[pathlen] = 0;
  90
  91         if (pos < it->subtree_nr)
  92                 memmove(it->down + pos + 1,
  93                         it->down + pos,
  94                         sizeof(down) * (it->subtree_nr - pos - 1));
  95         it->down[pos] = down;
  96         return down;
  97 }
  98
  99 struct cache_tree_sub *cache_tree_sub(struct cache_tree *it, const char *path)
 100 {
 101         int pathlen = strlen(path);
 102         return find_subtree(it, path, pathlen, 1);
 103 }
 104
 105 void cache_tree_invalidate_path(struct cache_tree *it, const char *path)
 106 {
 107         /* a/b/c
 108          * ==> invalidate self
 109          * ==> find "a", have it invalidate "b/c"
 110          * a
 111          * ==> invalidate self
 112          * ==> if "a" exists as a subtree, remove it.
 113          */
 114         const char *slash;
 115         int namelen;
 116         struct cache_tree_sub *down;
 117
 118 #if DEBUG
 119         fprintf(stderr, "cache-tree invalidate <%s>\n", path);
 120 #endif
 121
 122         if (!it)
 123                 return;
 124         slash = strchr(path, '/');
 125         it->entry_count = -1;
 126         if (!slash) {
 127                 int pos;
 128                 namelen = strlen(path);
 129                 pos = subtree_pos(it, path, namelen);
 130                 if (0 <= pos) {
 131                         cache_tree_free(&it->down[pos]->cache_tree);
 132                         free(it->down[pos]);
 133                         /* 0 1 2 3 4 5
 134                          *       ^     ^subtree_nr = 6
 135                          *       pos
 136                          * move 4 and 5 up one place (2 entries)
 137                          * 2 = 6 - 3 - 1 = subtree_nr - pos - 1
 138                          */
 139                         memmove(it->down+pos, it->down+pos+1,
 140                                 sizeof(struct cache_tree_sub *) *
 141                                 (it->subtree_nr - pos - 1));
 142                         it->subtree_nr--;
 143                 }
 144                 return;
 145         }
 146         namelen = slash - path;
 147         down = find_subtree(it, path, namelen, 0);
 148         if (down)
 149                 cache_tree_invalidate_path(down->cache_tree, slash + 1);
 150 }
 151
 152 static int verify_cache(struct cache_entry **cache,
 153                         int entries, int flags)
 154 {
 155         int i, funny;
 156         int silent = flags & WRITE_TREE_SILENT;
 157
 158         /* Verify that the tree is merged */
 159         funny = 0;
 160         for (i = 0; i < entries; i++) {
 161                 struct cache_entry *ce = cache[i];
 162                 if (ce_stage(ce)) {
 163                         if (silent)
 164                                 return -1;
 165                         if (10 < ++funny) {
 166                                 fprintf(stderr, "...\n");
 167                                 break;
 168                         }
 169                         fprintf(stderr, "%s: unmerged (%s)\n",
 170                                 ce->name, sha1_to_hex(ce->sha1));
 171                 }
 172         }
 173         if (funny)
 174                 return -1;
 175
 176         /* Also verify that the cache does not have path and path/file
 177          * at the same time.  At this point we know the cache has only
 178          * stage 0 entries.
 179          */
 180         funny = 0;
 181         for (i = 0; i < entries - 1; i++) {
 182                 /* path/file always comes after path because of the way
 183                  * the cache is sorted.  Also path can appear only once,
 184                  * which means conflicting one would immediately follow.
 185                  */
 186                 const char *this_name = cache[i]->name;
 187                 const char *next_name = cache[i+1]->name;
 188                 int this_len = strlen(this_name);
 189                 if (this_len < strlen(next_name) &&
 190                     strncmp(this_name, next_name, this_len) == 0 &&
 191                     next_name[this_len] == '/') {
 192                         if (10 < ++funny) {
 193                                 fprintf(stderr, "...\n");
 194                                 break;
 195                         }
 196                         fprintf(stderr, "You have both %s and %s\n",
 197                                 this_name, next_name);
 198                 }
 199         }
 200         if (funny)
 201                 return -1;
 202         return 0;
 203 }
 204
 205 static void discard_unused_subtrees(struct cache_tree *it)
 206 {
 207         struct cache_tree_sub **down = it->down;
 208         int nr = it->subtree_nr;
 209         int dst, src;
 210         for (dst = src = 0; src < nr; src++) {
 211                 struct cache_tree_sub *s = down[src];
 212                 if (s->used)
 213                         down[dst++] = s;
 214                 else {
 215                         cache_tree_free(&s->cache_tree);
 216                         free(s);
 217                         it->subtree_nr--;
 218                 }
 219         }
 220 }
 221
 222 int cache_tree_fully_valid(struct cache_tree *it)
 223 {
 224         int i;
 225         if (!it)
 226                 return 0;
 227         if (it->entry_count < 0 || !has_sha1_file(it->sha1))
 228                 return 0;
 229         for (i = 0; i < it->subtree_nr; i++) {
 230                 if (!cache_tree_fully_valid(it->down[i]->cache_tree))
 231                         return 0;
 232         }
 233         return 1;
 234 }
 235
 236 static int update_one(struct cache_tree *it,
 237                       struct cache_entry **cache,
 238                       int entries,
 239                       const char *base,
 240                       int baselen,
 241                       int flags)
 242 {
 243         struct strbuf buffer;
 244         int missing_ok = flags & WRITE_TREE_MISSING_OK;
 245         int dryrun = flags & WRITE_TREE_DRY_RUN;
 246         int i;
 247
 248         if (0 <= it->entry_count && has_sha1_file(it->sha1))
 249                 return it->entry_count;
 250
 251         /*
 252          * We first scan for subtrees and update them; we start by
 253          * marking existing subtrees -- the ones that are unmarked
 254          * should not be in the result.
 255          */
 256         for (i = 0; i < it->subtree_nr; i++)
 257                 it->down[i]->used = 0;
 258
 259         /*
 260          * Find the subtrees and update them.
 261          */
 262         i = 0;
 263         while (i < entries) {
 264                 struct cache_entry *ce = cache[i];
 265                 struct cache_tree_sub *sub;
 266                 const char *path, *slash;
 267                 int pathlen, sublen, subcnt;
 268
 269                 path = ce->name;
 270                 pathlen = ce_namelen(ce);
 271                 if (pathlen <= baselen || memcmp(base, path, baselen))
 272                         break; /* at the end of this level */
 273
 274                 slash = strchr(path + baselen, '/');
 275                 if (!slash) {
 276                         i++;
 277                         continue;
 278                 }
 279                 /*
 280                  * a/bbb/c (base = a/, slash = /c)
 281                  * ==>
 282                  * path+baselen = bbb/c, sublen = 3
 283                  */
 284                 sublen = slash - (path + baselen);
 285                 sub = find_subtree(it, path + baselen, sublen, 1);
 286                 if (!sub->cache_tree)
 287                         sub->cache_tree = cache_tree();
 288                 subcnt = update_one(sub->cache_tree,
 289                                     cache + i, entries - i,
 290                                     path,
 291                                     baselen + sublen + 1,
 292                                     flags);
 293                 if (subcnt < 0)
 294                         return subcnt;
 295                 i += subcnt;
 296                 sub->used = 1;
 297         }
 298
 299         discard_unused_subtrees(it);
 300
 301         /*
 302          * Then write out the tree object for this level.
 303          */
 304         strbuf_init(&buffer, 8192);
 305
 306         i = 0;
 307         while (i < entries) {
 308                 struct cache_entry *ce = cache[i];
 309                 struct cache_tree_sub *sub;
 310                 const char *path, *slash;
 311                 int pathlen, entlen;
 312                 const unsigned char *sha1;
 313                 unsigned mode;
 314
 315                 path = ce->name;
 316                 pathlen = ce_namelen(ce);
 317                 if (pathlen <= baselen || memcmp(base, path, baselen))
 318                         break; /* at the end of this level */
 319
 320                 slash = strchr(path + baselen, '/');
 321                 if (slash) {
 322                         entlen = slash - (path + baselen);
 323                         sub = find_subtree(it, path + baselen, entlen, 0);
 324                         if (!sub)
 325                                 die("cache-tree.c: '%.*s' in '%s' not found",
 326                                     entlen, path + baselen, path);
 327                         i += sub->cache_tree->entry_count;
 328                         sha1 = sub->cache_tree->sha1;
 329                         mode = S_IFDIR;
 330                 }
 331                 else {
 332                         sha1 = ce->sha1;
 333                         mode = ce->ce_mode;
 334                         entlen = pathlen - baselen;
 335                         i++;
 336                 }
 337                 if (mode != S_IFGITLINK && !missing_ok && !has_sha1_file(sha1)) {
 338                         strbuf_release(&buffer);
 339                         return error("invalid object %06o %s for '%.*s'",
 340                                 mode, sha1_to_hex(sha1), entlen+baselen, path);
 341                 }
 342
 343                 if (ce->ce_flags & (CE_REMOVE | CE_INTENT_TO_ADD))
 344                         continue; /* entry being removed or placeholder */
 345
 346                 strbuf_grow(&buffer, entlen + 100);
 347                 strbuf_addf(&buffer, "%o %.*s%c", mode, entlen, path + baselen, '\0');
 348                 strbuf_add(&buffer, sha1, 20);
 349
 350 #if DEBUG
 351                 fprintf(stderr, "cache-tree update-one %o %.*s\n",
 352                         mode, entlen, path + baselen);
 353 #endif
 354         }
 355
 356         if (dryrun)
 357                 hash_sha1_file(buffer.buf, buffer.len, tree_type, it->sha1);
 358         else if (write_sha1_file(buffer.buf, buffer.len, tree_type, it->sha1)) {
 359                 strbuf_release(&buffer);
 360                 return -1;
 361         }
 362
 363         strbuf_release(&buffer);
 364         it->entry_count = i;
 365 #if DEBUG
 366         fprintf(stderr, "cache-tree update-one (%d ent, %d subtree) %s\n",
 367                 it->entry_count, it->subtree_nr,
 368                 sha1_to_hex(it->sha1));
 369 #endif
 370         return i;
 371 }
 372
 373 int cache_tree_update(struct cache_tree *it,
 374                       struct cache_entry **cache,
 375                       int entries,
 376                       int flags)
 377 {
 378         int i;
 379         i = verify_cache(cache, entries, flags);
 380         if (i)
 381                 return i;
 382         i = update_one(it, cache, entries, "", 0, flags);
 383         if (i < 0)
 384                 return i;
 385         return 0;
 386 }
 387
 388 static void write_one(struct strbuf *buffer, struct cache_tree *it,
 389                       const char *path, int pathlen)
 390 {
 391         int i;
 392
 393         /* One "cache-tree" entry consists of the following:
 394          * path (NUL terminated)
 395          * entry_count, subtree_nr ("%d %d\n")
 396          * tree-sha1 (missing if invalid)
 397          * subtree_nr "cache-tree" entries for subtrees.
 398          */
 399         strbuf_grow(buffer, pathlen + 100);
 400         strbuf_add(buffer, path, pathlen);
 401         strbuf_addf(buffer, "%c%d %d\n", 0, it->entry_count, it->subtree_nr);
 402
 403 #if DEBUG
 404         if (0 <= it->entry_count)
 405                 fprintf(stderr, "cache-tree <%.*s> (%d ent, %d subtree) %s\n",
 406                         pathlen, path, it->entry_count, it->subtree_nr,
 407                         sha1_to_hex(it->sha1));
 408         else
 409                 fprintf(stderr, "cache-tree <%.*s> (%d subtree) invalid\n",
 410                         pathlen, path, it->subtree_nr);
 411 #endif
 412
 413         if (0 <= it->entry_count) {
 414                 strbuf_add(buffer, it->sha1, 20);
 415         }
 416         for (i = 0; i < it->subtree_nr; i++) {
 417                 struct cache_tree_sub *down = it->down[i];
 418                 if (i) {
 419                         struct cache_tree_sub *prev = it->down[i-1];
 420                         if (subtree_name_cmp(down->name, down->namelen,
 421                                              prev->name, prev->namelen) <= 0)
 422                                 die("fatal - unsorted cache subtree");
 423                 }
 424                 write_one(buffer, down->cache_tree, down->name, down->namelen);
 425         }
 426 }
 427
 428 void cache_tree_write(struct strbuf *sb, struct cache_tree *root)
 429 {
 430         write_one(sb, root, "", 0);
 431 }
 432
 433 static struct cache_tree *read_one(const char **buffer, unsigned long *size_p)
 434 {
 435         const char *buf = *buffer;
 436         unsigned long size = *size_p;
 437         const char *cp;
 438         char *ep;
 439         struct cache_tree *it;
 440         int i, subtree_nr;
 441
 442         it = NULL;
 443         /* skip name, but make sure name exists */
 444         while (size && *buf) {
 445                 size--;
 446                 buf++;
 447         }
 448         if (!size)
 449                 goto free_return;
 450         buf++; size--;
 451         it = cache_tree();
 452
 453         cp = buf;
 454         it->entry_count = strtol(cp, &ep, 10);
 455         if (cp == ep)
 456                 goto free_return;
 457         cp = ep;
 458         subtree_nr = strtol(cp, &ep, 10);
 459         if (cp == ep)
 460                 goto free_return;
 461         while (size && *buf && *buf != '\n') {
 462                 size--;
 463                 buf++;
 464         }
 465         if (!size)
 466                 goto free_return;
 467         buf++; size--;
 468         if (0 <= it->entry_count) {
 469                 if (size < 20)
 470                         goto free_return;
 471                 hashcpy(it->sha1, (const unsigned char*)buf);
 472                 buf += 20;
 473                 size -= 20;
 474         }
 475
 476 #if DEBUG
 477         if (0 <= it->entry_count)
 478                 fprintf(stderr, "cache-tree <%s> (%d ent, %d subtree) %s\n",
 479                         *buffer, it->entry_count, subtree_nr,
 480                         sha1_to_hex(it->sha1));
 481         else
 482                 fprintf(stderr, "cache-tree <%s> (%d subtrees) invalid\n",
 483                         *buffer, subtree_nr);
 484 #endif
 485
 486         /*
 487          * Just a heuristic -- we do not add directories that often but
 488          * we do not want to have to extend it immediately when we do,
 489          * hence +2.
 490          */
 491         it->subtree_alloc = subtree_nr + 2;
 492         it->down = xcalloc(it->subtree_alloc, sizeof(struct cache_tree_sub *));
 493         for (i = 0; i < subtree_nr; i++) {
 494                 /* read each subtree */
 495                 struct cache_tree *sub;
 496                 struct cache_tree_sub *subtree;
 497                 const char *name = buf;
 498
 499                 sub = read_one(&buf, &size);
 500                 if (!sub)
 501                         goto free_return;
 502                 subtree = cache_tree_sub(it, name);
 503                 subtree->cache_tree = sub;
 504         }
 505         if (subtree_nr != it->subtree_nr)
 506                 die("cache-tree: internal error");
 507         *buffer = buf;
 508         *size_p = size;
 509         return it;
 510
 511  free_return:
 512         cache_tree_free(&it);
 513         return NULL;
 514 }
 515
 516 struct cache_tree *cache_tree_read(const char *buffer, unsigned long size)
 517 {
 518         if (buffer[0])
 519                 return NULL; /* not the whole tree */
 520         return read_one(&buffer, &size);
 521 }
 522
 523 static struct cache_tree *cache_tree_find(struct cache_tree *it, const char *path)
 524 {
 525         if (!it)
 526                 return NULL;
 527         while (*path) {
 528                 const char *slash;
 529                 struct cache_tree_sub *sub;
 530
 531                 slash = strchr(path, '/');
 532                 if (!slash)
 533                         slash = path + strlen(path);
 534                 /* between path and slash is the name of the
 535                  * subtree to look for.
 536                  */
 537                 sub = find_subtree(it, path, slash - path, 0);
 538                 if (!sub)
 539                         return NULL;
 540                 it = sub->cache_tree;
 541                 if (slash)
 542                         while (*slash && *slash == '/')
 543                                 slash++;
 544                 if (!slash || !*slash)
 545                         return it; /* prefix ended with slashes */
 546                 path = slash;
 547         }
 548         return it;
 549 }
 550
 551 int write_cache_as_tree(unsigned char *sha1, int flags, const char *prefix)
 552 {
 553         int entries, was_valid, newfd;
 554         struct lock_file *lock_file;
 555
 556         /*
 557          * We can't free this memory, it becomes part of a linked list
 558          * parsed atexit()
 559          */
 560         lock_file = xcalloc(1, sizeof(struct lock_file));
 561
 562         newfd = hold_locked_index(lock_file, 1);
 563
 564         entries = read_cache();
 565         if (entries < 0)
 566                 return WRITE_TREE_UNREADABLE_INDEX;
 567         if (flags & WRITE_TREE_IGNORE_CACHE_TREE)
 568                 cache_tree_free(&(active_cache_tree));
 569
 570         if (!active_cache_tree)
 571                 active_cache_tree = cache_tree();
 572
 573         was_valid = cache_tree_fully_valid(active_cache_tree);
 574         if (!was_valid) {
 575                 if (cache_tree_update(active_cache_tree,
 576                                       active_cache, active_nr,
 577                                       flags) < 0)
 578                         return WRITE_TREE_UNMERGED_INDEX;
 579                 if (0 <= newfd) {
 580                         if (!write_cache(newfd, active_cache, active_nr) &&
 581                             !commit_lock_file(lock_file))
 582                                 newfd = -1;
 583                 }
 584                 /* Not being able to write is fine -- we are only interested
 585                  * in updating the cache-tree part, and if the next caller
 586                  * ends up using the old index with unupdated cache-tree part
 587                  * it misses the work we did here, but that is just a
 588                  * performance penalty and not a big deal.
 589                  */
 590         }
 591
 592         if (prefix) {
 593                 struct cache_tree *subtree =
 594                         cache_tree_find(active_cache_tree, prefix);
 595                 if (!subtree)
 596                         return WRITE_TREE_PREFIX_ERROR;
 597                 hashcpy(sha1, subtree->sha1);
 598         }
 599         else
 600                 hashcpy(sha1, active_cache_tree->sha1);
 601
 602         if (0 <= newfd)
 603                 rollback_lock_file(lock_file);
 604
 605         return 0;
 606 }
 607
 608 static void prime_cache_tree_rec(struct cache_tree *it, struct tree *tree)
 609 {
 610         struct tree_desc desc;
 611         struct name_entry entry;
 612         int cnt;
 613
 614         hashcpy(it->sha1, tree->object.sha1);
 615         init_tree_desc(&desc, tree->buffer, tree->size);
 616         cnt = 0;
 617         while (tree_entry(&desc, &entry)) {
 618                 if (!S_ISDIR(entry.mode))
 619                         cnt++;
 620                 else {
 621                         struct cache_tree_sub *sub;
 622                         struct tree *subtree = lookup_tree(entry.sha1);
 623                         if (!subtree->object.parsed)
 624                                 parse_tree(subtree);
 625                         sub = cache_tree_sub(it, entry.path);
 626                         sub->cache_tree = cache_tree();
 627                         prime_cache_tree_rec(sub->cache_tree, subtree);
 628                         cnt += sub->cache_tree->entry_count;
 629                 }
 630         }
 631         it->entry_count = cnt;
 632 }
 633
 634 void prime_cache_tree(struct cache_tree **it, struct tree *tree)
 635 {
 636         cache_tree_free(it);
 637         *it = cache_tree();
 638         prime_cache_tree_rec(*it, tree);
 639 }
 640
 641 /*
 642  * find the cache_tree that corresponds to the current level without
 643  * exploding the full path into textual form.  The root of the
 644  * cache tree is given as "root", and our current level is "info".
 645  * (1) When at root level, info->prev is NULL, so it is "root" itself.
 646  * (2) Otherwise, find the cache_tree that corresponds to one level
 647  *     above us, and find ourselves in there.
 648  */
 649 static struct cache_tree *find_cache_tree_from_traversal(struct cache_tree *root,
 650                                                          struct traverse_info *info)
 651 {
 652         struct cache_tree *our_parent;
 653
 654         if (!info->prev)
 655                 return root;
 656         our_parent = find_cache_tree_from_traversal(root, info->prev);
 657         return cache_tree_find(our_parent, info->name.path);
 658 }
 659
 660 int cache_tree_matches_traversal(struct cache_tree *root,
 661                                  struct name_entry *ent,
 662                                  struct traverse_info *info)
 663 {
 664         struct cache_tree *it;
 665
 666         it = find_cache_tree_from_traversal(root, info);
 667         it = cache_tree_find(it, ent->path);
 668         if (it && it->entry_count > 0 && !hashcmp(ent->sha1, it->sha1))
 669                 return it->entry_count;
 670         return 0;
 671 }
 672
 673 int update_main_cache_tree(int flags)
 674 {
 675         if (!the_index.cache_tree)
 676                 the_index.cache_tree = cache_tree();
 677         return cache_tree_update(the_index.cache_tree,
 678                                  the_index.cache, the_index.cache_nr, flags);
 679 }