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                         if (ce_stage(ce))
 170                                 fprintf(stderr, "%s: unmerged (%s)\n",
 171                                         ce->name, sha1_to_hex(ce->sha1));
 172                         else
 173                                 fprintf(stderr, "%s: not added yet\n",
 174                                         ce->name);
 175                 }
 176         }
 177         if (funny)
 178                 return -1;
 179
 180         /* Also verify that the cache does not have path and path/file
 181          * at the same time.  At this point we know the cache has only
 182          * stage 0 entries.
 183          */
 184         funny = 0;
 185         for (i = 0; i < entries - 1; i++) {
 186                 /* path/file always comes after path because of the way
 187                  * the cache is sorted.  Also path can appear only once,
 188                  * which means conflicting one would immediately follow.
 189                  */
 190                 const char *this_name = cache[i]->name;
 191                 const char *next_name = cache[i+1]->name;
 192                 int this_len = strlen(this_name);
 193                 if (this_len < strlen(next_name) &&
 194                     strncmp(this_name, next_name, this_len) == 0 &&
 195                     next_name[this_len] == '/') {
 196                         if (10 < ++funny) {
 197                                 fprintf(stderr, "...\n");
 198                                 break;
 199                         }
 200                         fprintf(stderr, "You have both %s and %s\n",
 201                                 this_name, next_name);
 202                 }
 203         }
 204         if (funny)
 205                 return -1;
 206         return 0;
 207 }
 208
 209 static void discard_unused_subtrees(struct cache_tree *it)
 210 {
 211         struct cache_tree_sub **down = it->down;
 212         int nr = it->subtree_nr;
 213         int dst, src;
 214         for (dst = src = 0; src < nr; src++) {
 215                 struct cache_tree_sub *s = down[src];
 216                 if (s->used)
 217                         down[dst++] = s;
 218                 else {
 219                         cache_tree_free(&s->cache_tree);
 220                         free(s);
 221                         it->subtree_nr--;
 222                 }
 223         }
 224 }
 225
 226 int cache_tree_fully_valid(struct cache_tree *it)
 227 {
 228         int i;
 229         if (!it)
 230                 return 0;
 231         if (it->entry_count < 0 || !has_sha1_file(it->sha1))
 232                 return 0;
 233         for (i = 0; i < it->subtree_nr; i++) {
 234                 if (!cache_tree_fully_valid(it->down[i]->cache_tree))
 235                         return 0;
 236         }
 237         return 1;
 238 }
 239
 240 static int update_one(struct cache_tree *it,
 241                       struct cache_entry **cache,
 242                       int entries,
 243                       const char *base,
 244                       int baselen,
 245                       int flags)
 246 {
 247         struct strbuf buffer;
 248         int missing_ok = flags & WRITE_TREE_MISSING_OK;
 249         int dryrun = flags & WRITE_TREE_DRY_RUN;
 250         int i;
 251
 252         if (0 <= it->entry_count && has_sha1_file(it->sha1))
 253                 return it->entry_count;
 254
 255         /*
 256          * We first scan for subtrees and update them; we start by
 257          * marking existing subtrees -- the ones that are unmarked
 258          * should not be in the result.
 259          */
 260         for (i = 0; i < it->subtree_nr; i++)
 261                 it->down[i]->used = 0;
 262
 263         /*
 264          * Find the subtrees and update them.
 265          */
 266         for (i = 0; i < entries; i++) {
 267                 struct cache_entry *ce = cache[i];
 268                 struct cache_tree_sub *sub;
 269                 const char *path, *slash;
 270                 int pathlen, sublen, subcnt;
 271
 272                 path = ce->name;
 273                 pathlen = ce_namelen(ce);
 274                 if (pathlen <= baselen || memcmp(base, path, baselen))
 275                         break; /* at the end of this level */
 276
 277                 slash = strchr(path + baselen, '/');
 278                 if (!slash)
 279                         continue;
 280                 /*
 281                  * a/bbb/c (base = a/, slash = /c)
 282                  * ==>
 283                  * path+baselen = bbb/c, sublen = 3
 284                  */
 285                 sublen = slash - (path + baselen);
 286                 sub = find_subtree(it, path + baselen, sublen, 1);
 287                 if (!sub->cache_tree)
 288                         sub->cache_tree = cache_tree();
 289                 subcnt = update_one(sub->cache_tree,
 290                                     cache + i, entries - i,
 291                                     path,
 292                                     baselen + sublen + 1,
 293                                     flags);
 294                 if (subcnt < 0)
 295                         return subcnt;
 296                 i += subcnt - 1;
 297                 sub->used = 1;
 298         }
 299
 300         discard_unused_subtrees(it);
 301
 302         /*
 303          * Then write out the tree object for this level.
 304          */
 305         strbuf_init(&buffer, 8192);
 306
 307         for (i = 0; i < entries; i++) {
 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 - 1;
 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                 }
 336                 if (mode != S_IFGITLINK && !missing_ok && !has_sha1_file(sha1)) {
 337                         strbuf_release(&buffer);
 338                         return error("invalid object %06o %s for '%.*s'",
 339                                 mode, sha1_to_hex(sha1), entlen+baselen, path);
 340                 }
 341
 342                 if (ce->ce_flags & (CE_REMOVE | CE_INTENT_TO_ADD))
 343                         continue; /* entry being removed or placeholder */
 344
 345                 strbuf_grow(&buffer, entlen + 100);
 346                 strbuf_addf(&buffer, "%o %.*s%c", mode, entlen, path + baselen, '\0');
 347                 strbuf_add(&buffer, sha1, 20);
 348
 349 #if DEBUG
 350                 fprintf(stderr, "cache-tree update-one %o %.*s\n",
 351                         mode, entlen, path + baselen);
 352 #endif
 353         }
 354
 355         if (dryrun)
 356                 hash_sha1_file(buffer.buf, buffer.len, tree_type, it->sha1);
 357         else if (write_sha1_file(buffer.buf, buffer.len, tree_type, it->sha1)) {
 358                 strbuf_release(&buffer);
 359                 return -1;
 360         }
 361
 362         strbuf_release(&buffer);
 363         it->entry_count = i;
 364 #if DEBUG
 365         fprintf(stderr, "cache-tree update-one (%d ent, %d subtree) %s\n",
 366                 it->entry_count, it->subtree_nr,
 367                 sha1_to_hex(it->sha1));
 368 #endif
 369         return i;
 370 }
 371
 372 int cache_tree_update(struct cache_tree *it,
 373                       struct cache_entry **cache,
 374                       int entries,
 375                       int flags)
 376 {
 377         int i;
 378         i = verify_cache(cache, entries, flags);
 379         if (i)
 380                 return i;
 381         i = update_one(it, cache, entries, "", 0, flags);
 382         if (i < 0)
 383                 return i;
 384         return 0;
 385 }
 386
 387 static void write_one(struct strbuf *buffer, struct cache_tree *it,
 388                       const char *path, int pathlen)
 389 {
 390         int i;
 391
 392         /* One "cache-tree" entry consists of the following:
 393          * path (NUL terminated)
 394          * entry_count, subtree_nr ("%d %d\n")
 395          * tree-sha1 (missing if invalid)
 396          * subtree_nr "cache-tree" entries for subtrees.
 397          */
 398         strbuf_grow(buffer, pathlen + 100);
 399         strbuf_add(buffer, path, pathlen);
 400         strbuf_addf(buffer, "%c%d %d\n", 0, it->entry_count, it->subtree_nr);
 401
 402 #if DEBUG
 403         if (0 <= it->entry_count)
 404                 fprintf(stderr, "cache-tree <%.*s> (%d ent, %d subtree) %s\n",
 405                         pathlen, path, it->entry_count, it->subtree_nr,
 406                         sha1_to_hex(it->sha1));
 407         else
 408                 fprintf(stderr, "cache-tree <%.*s> (%d subtree) invalid\n",
 409                         pathlen, path, it->subtree_nr);
 410 #endif
 411
 412         if (0 <= it->entry_count) {
 413                 strbuf_add(buffer, it->sha1, 20);
 414         }
 415         for (i = 0; i < it->subtree_nr; i++) {
 416                 struct cache_tree_sub *down = it->down[i];
 417                 if (i) {
 418                         struct cache_tree_sub *prev = it->down[i-1];
 419                         if (subtree_name_cmp(down->name, down->namelen,
 420                                              prev->name, prev->namelen) <= 0)
 421                                 die("fatal - unsorted cache subtree");
 422                 }
 423                 write_one(buffer, down->cache_tree, down->name, down->namelen);
 424         }
 425 }
 426
 427 void cache_tree_write(struct strbuf *sb, struct cache_tree *root)
 428 {
 429         write_one(sb, root, "", 0);
 430 }
 431
 432 static struct cache_tree *read_one(const char **buffer, unsigned long *size_p)
 433 {
 434         const char *buf = *buffer;
 435         unsigned long size = *size_p;
 436         const char *cp;
 437         char *ep;
 438         struct cache_tree *it;
 439         int i, subtree_nr;
 440
 441         it = NULL;
 442         /* skip name, but make sure name exists */
 443         while (size && *buf) {
 444                 size--;
 445                 buf++;
 446         }
 447         if (!size)
 448                 goto free_return;
 449         buf++; size--;
 450         it = cache_tree();
 451
 452         cp = buf;
 453         it->entry_count = strtol(cp, &ep, 10);
 454         if (cp == ep)
 455                 goto free_return;
 456         cp = ep;
 457         subtree_nr = strtol(cp, &ep, 10);
 458         if (cp == ep)
 459                 goto free_return;
 460         while (size && *buf && *buf != '\n') {
 461                 size--;
 462                 buf++;
 463         }
 464         if (!size)
 465                 goto free_return;
 466         buf++; size--;
 467         if (0 <= it->entry_count) {
 468                 if (size < 20)
 469                         goto free_return;
 470                 hashcpy(it->sha1, (const unsigned char*)buf);
 471                 buf += 20;
 472                 size -= 20;
 473         }
 474
 475 #if DEBUG
 476         if (0 <= it->entry_count)
 477                 fprintf(stderr, "cache-tree <%s> (%d ent, %d subtree) %s\n",
 478                         *buffer, it->entry_count, subtree_nr,
 479                         sha1_to_hex(it->sha1));
 480         else
 481                 fprintf(stderr, "cache-tree <%s> (%d subtrees) invalid\n",
 482                         *buffer, subtree_nr);
 483 #endif
 484
 485         /*
 486          * Just a heuristic -- we do not add directories that often but
 487          * we do not want to have to extend it immediately when we do,
 488          * hence +2.
 489          */
 490         it->subtree_alloc = subtree_nr + 2;
 491         it->down = xcalloc(it->subtree_alloc, sizeof(struct cache_tree_sub *));
 492         for (i = 0; i < subtree_nr; i++) {
 493                 /* read each subtree */
 494                 struct cache_tree *sub;
 495                 struct cache_tree_sub *subtree;
 496                 const char *name = buf;
 497
 498                 sub = read_one(&buf, &size);
 499                 if (!sub)
 500                         goto free_return;
 501                 subtree = cache_tree_sub(it, name);
 502                 subtree->cache_tree = sub;
 503         }
 504         if (subtree_nr != it->subtree_nr)
 505                 die("cache-tree: internal error");
 506         *buffer = buf;
 507         *size_p = size;
 508         return it;
 509
 510  free_return:
 511         cache_tree_free(&it);
 512         return NULL;
 513 }
 514
 515 struct cache_tree *cache_tree_read(const char *buffer, unsigned long size)
 516 {
 517         if (buffer[0])
 518                 return NULL; /* not the whole tree */
 519         return read_one(&buffer, &size);
 520 }
 521
 522 static struct cache_tree *cache_tree_find(struct cache_tree *it, const char *path)
 523 {
 524         if (!it)
 525                 return NULL;
 526         while (*path) {
 527                 const char *slash;
 528                 struct cache_tree_sub *sub;
 529
 530                 slash = strchr(path, '/');
 531                 if (!slash)
 532                         slash = path + strlen(path);
 533                 /* between path and slash is the name of the
 534                  * subtree to look for.
 535                  */
 536                 sub = find_subtree(it, path, slash - path, 0);
 537                 if (!sub)
 538                         return NULL;
 539                 it = sub->cache_tree;
 540                 if (slash)
 541                         while (*slash && *slash == '/')
 542                                 slash++;
 543                 if (!slash || !*slash)
 544                         return it; /* prefix ended with slashes */
 545                 path = slash;
 546         }
 547         return it;
 548 }
 549
 550 int write_cache_as_tree(unsigned char *sha1, int flags, const char *prefix)
 551 {
 552         int entries, was_valid, newfd;
 553         struct lock_file *lock_file;
 554
 555         /*
 556          * We can't free this memory, it becomes part of a linked list
 557          * parsed atexit()
 558          */
 559         lock_file = xcalloc(1, sizeof(struct lock_file));
 560
 561         newfd = hold_locked_index(lock_file, 1);
 562
 563         entries = read_cache();
 564         if (entries < 0)
 565                 return WRITE_TREE_UNREADABLE_INDEX;
 566         if (flags & WRITE_TREE_IGNORE_CACHE_TREE)
 567                 cache_tree_free(&(active_cache_tree));
 568
 569         if (!active_cache_tree)
 570                 active_cache_tree = cache_tree();
 571
 572         was_valid = cache_tree_fully_valid(active_cache_tree);
 573         if (!was_valid) {
 574                 if (cache_tree_update(active_cache_tree,
 575                                       active_cache, active_nr,
 576                                       flags) < 0)
 577                         return WRITE_TREE_UNMERGED_INDEX;
 578                 if (0 <= newfd) {
 579                         if (!write_cache(newfd, active_cache, active_nr) &&
 580                             !commit_lock_file(lock_file))
 581                                 newfd = -1;
 582                 }
 583                 /* Not being able to write is fine -- we are only interested
 584                  * in updating the cache-tree part, and if the next caller
 585                  * ends up using the old index with unupdated cache-tree part
 586                  * it misses the work we did here, but that is just a
 587                  * performance penalty and not a big deal.
 588                  */
 589         }
 590
 591         if (prefix) {
 592                 struct cache_tree *subtree =
 593                         cache_tree_find(active_cache_tree, prefix);
 594                 if (!subtree)
 595                         return WRITE_TREE_PREFIX_ERROR;
 596                 hashcpy(sha1, subtree->sha1);
 597         }
 598         else
 599                 hashcpy(sha1, active_cache_tree->sha1);
 600
 601         if (0 <= newfd)
 602                 rollback_lock_file(lock_file);
 603
 604         return 0;
 605 }
 606
 607 static void prime_cache_tree_rec(struct cache_tree *it, struct tree *tree)
 608 {
 609         struct tree_desc desc;
 610         struct name_entry entry;
 611         int cnt;
 612
 613         hashcpy(it->sha1, tree->object.sha1);
 614         init_tree_desc(&desc, tree->buffer, tree->size);
 615         cnt = 0;
 616         while (tree_entry(&desc, &entry)) {
 617                 if (!S_ISDIR(entry.mode))
 618                         cnt++;
 619                 else {
 620                         struct cache_tree_sub *sub;
 621                         struct tree *subtree = lookup_tree(entry.sha1);
 622                         if (!subtree->object.parsed)
 623                                 parse_tree(subtree);
 624                         sub = cache_tree_sub(it, entry.path);
 625                         sub->cache_tree = cache_tree();
 626                         prime_cache_tree_rec(sub->cache_tree, subtree);
 627                         cnt += sub->cache_tree->entry_count;
 628                 }
 629         }
 630         it->entry_count = cnt;
 631 }
 632
 633 void prime_cache_tree(struct cache_tree **it, struct tree *tree)
 634 {
 635         cache_tree_free(it);
 636         *it = cache_tree();
 637         prime_cache_tree_rec(*it, tree);
 638 }
 639
 640 /*
 641  * find the cache_tree that corresponds to the current level without
 642  * exploding the full path into textual form.  The root of the
 643  * cache tree is given as "root", and our current level is "info".
 644  * (1) When at root level, info->prev is NULL, so it is "root" itself.
 645  * (2) Otherwise, find the cache_tree that corresponds to one level
 646  *     above us, and find ourselves in there.
 647  */
 648 static struct cache_tree *find_cache_tree_from_traversal(struct cache_tree *root,
 649                                                          struct traverse_info *info)
 650 {
 651         struct cache_tree *our_parent;
 652
 653         if (!info->prev)
 654                 return root;
 655         our_parent = find_cache_tree_from_traversal(root, info->prev);
 656         return cache_tree_find(our_parent, info->name.path);
 657 }
 658
 659 int cache_tree_matches_traversal(struct cache_tree *root,
 660                                  struct name_entry *ent,
 661                                  struct traverse_info *info)
 662 {
 663         struct cache_tree *it;
 664
 665         it = find_cache_tree_from_traversal(root, info);
 666         it = cache_tree_find(it, ent->path);
 667         if (it && it->entry_count > 0 && !hashcmp(ent->sha1, it->sha1))
 668                 return it->entry_count;
 669         return 0;
 670 }
 671
 672 int update_main_cache_tree(int flags)
 673 {
 674         if (!the_index.cache_tree)
 675                 the_index.cache_tree = cache_tree();
 676         return cache_tree_update(the_index.cache_tree,
 677                                  the_index.cache, the_index.cache_nr, flags);
 678 }