git.oblomov.eu Git - git/blob - 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 = strchrnul(path, '/');
 125         namelen = slash - path;
 126         it->entry_count = -1;
 127         if (!*slash) {
 128                 int pos;
 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         down = find_subtree(it, path, namelen, 0);
 147         if (down)
 148                 cache_tree_invalidate_path(down->cache_tree, slash + 1);
 149 }
 150
 151 static int verify_cache(const struct cache_entry * const *cache,
 152                         int entries, int flags)
 153 {
 154         int i, funny;
 155         int silent = flags & WRITE_TREE_SILENT;
 156
 157         /* Verify that the tree is merged */
 158         funny = 0;
 159         for (i = 0; i < entries; i++) {
 160                 const struct cache_entry *ce = cache[i];
 161                 if (ce_stage(ce)) {
 162                         if (silent)
 163                                 return -1;
 164                         if (10 < ++funny) {
 165                                 fprintf(stderr, "...\n");
 166                                 break;
 167                         }
 168                         fprintf(stderr, "%s: unmerged (%s)\n",
 169                                 ce->name, sha1_to_hex(ce->sha1));
 170                 }
 171         }
 172         if (funny)
 173                 return -1;
 174
 175         /* Also verify that the cache does not have path and path/file
 176          * at the same time.  At this point we know the cache has only
 177          * stage 0 entries.
 178          */
 179         funny = 0;
 180         for (i = 0; i < entries - 1; i++) {
 181                 /* path/file always comes after path because of the way
 182                  * the cache is sorted.  Also path can appear only once,
 183                  * which means conflicting one would immediately follow.
 184                  */
 185                 const char *this_name = cache[i]->name;
 186                 const char *next_name = cache[i+1]->name;
 187                 int this_len = strlen(this_name);
 188                 if (this_len < strlen(next_name) &&
 189                     strncmp(this_name, next_name, this_len) == 0 &&
 190                     next_name[this_len] == '/') {
 191                         if (10 < ++funny) {
 192                                 fprintf(stderr, "...\n");
 193                                 break;
 194                         }
 195                         fprintf(stderr, "You have both %s and %s\n",
 196                                 this_name, next_name);
 197                 }
 198         }
 199         if (funny)
 200                 return -1;
 201         return 0;
 202 }
 203
 204 static void discard_unused_subtrees(struct cache_tree *it)
 205 {
 206         struct cache_tree_sub **down = it->down;
 207         int nr = it->subtree_nr;
 208         int dst, src;
 209         for (dst = src = 0; src < nr; src++) {
 210                 struct cache_tree_sub *s = down[src];
 211                 if (s->used)
 212                         down[dst++] = s;
 213                 else {
 214                         cache_tree_free(&s->cache_tree);
 215                         free(s);
 216                         it->subtree_nr--;
 217                 }
 218         }
 219 }
 220
 221 int cache_tree_fully_valid(struct cache_tree *it)
 222 {
 223         int i;
 224         if (!it)
 225                 return 0;
 226         if (it->entry_count < 0 || !has_sha1_file(it->sha1))
 227                 return 0;
 228         for (i = 0; i < it->subtree_nr; i++) {
 229                 if (!cache_tree_fully_valid(it->down[i]->cache_tree))
 230                         return 0;
 231         }
 232         return 1;
 233 }
 234
 235 static int update_one(struct cache_tree *it,
 236                       const struct cache_entry * const *cache,
 237                       int entries,
 238                       const char *base,
 239                       int baselen,
 240                       int *skip_count,
 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 to_invalidate = 0;
 247         int i;
 248
 249         *skip_count = 0;
 250
 251         if (0 <= it->entry_count && has_sha1_file(it->sha1))
 252                 return it->entry_count;
 253
 254         /*
 255          * We first scan for subtrees and update them; we start by
 256          * marking existing subtrees -- the ones that are unmarked
 257          * should not be in the result.
 258          */
 259         for (i = 0; i < it->subtree_nr; i++)
 260                 it->down[i]->used = 0;
 261
 262         /*
 263          * Find the subtrees and update them.
 264          */
 265         i = 0;
 266         while (i < entries) {
 267                 const struct cache_entry *ce = cache[i];
 268                 struct cache_tree_sub *sub;
 269                 const char *path, *slash;
 270                 int pathlen, sublen, subcnt, subskip;
 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                         i++;
 280                         continue;
 281                 }
 282                 /*
 283                  * a/bbb/c (base = a/, slash = /c)
 284                  * ==>
 285                  * path+baselen = bbb/c, sublen = 3
 286                  */
 287                 sublen = slash - (path + baselen);
 288                 sub = find_subtree(it, path + baselen, sublen, 1);
 289                 if (!sub->cache_tree)
 290                         sub->cache_tree = cache_tree();
 291                 subcnt = update_one(sub->cache_tree,
 292                                     cache + i, entries - i,
 293                                     path,
 294                                     baselen + sublen + 1,
 295                                     &subskip,
 296                                     flags);
 297                 if (subcnt < 0)
 298                         return subcnt;
 299                 i += subcnt;
 300                 sub->count = subcnt; /* to be used in the next loop */
 301                 *skip_count += subskip;
 302                 sub->used = 1;
 303         }
 304
 305         discard_unused_subtrees(it);
 306
 307         /*
 308          * Then write out the tree object for this level.
 309          */
 310         strbuf_init(&buffer, 8192);
 311
 312         i = 0;
 313         while (i < entries) {
 314                 const struct cache_entry *ce = cache[i];
 315                 struct cache_tree_sub *sub;
 316                 const char *path, *slash;
 317                 int pathlen, entlen;
 318                 const unsigned char *sha1;
 319                 unsigned mode;
 320
 321                 path = ce->name;
 322                 pathlen = ce_namelen(ce);
 323                 if (pathlen <= baselen || memcmp(base, path, baselen))
 324                         break; /* at the end of this level */
 325
 326                 slash = strchr(path + baselen, '/');
 327                 if (slash) {
 328                         entlen = slash - (path + baselen);
 329                         sub = find_subtree(it, path + baselen, entlen, 0);
 330                         if (!sub)
 331                                 die("cache-tree.c: '%.*s' in '%s' not found",
 332                                     entlen, path + baselen, path);
 333                         i += sub->count;
 334                         sha1 = sub->cache_tree->sha1;
 335                         mode = S_IFDIR;
 336                         if (sub->cache_tree->entry_count < 0)
 337                                 to_invalidate = 1;
 338                 }
 339                 else {
 340                         sha1 = ce->sha1;
 341                         mode = ce->ce_mode;
 342                         entlen = pathlen - baselen;
 343                         i++;
 344                 }
 345                 if (mode != S_IFGITLINK && !missing_ok && !has_sha1_file(sha1)) {
 346                         strbuf_release(&buffer);
 347                         return error("invalid object %06o %s for '%.*s'",
 348                                 mode, sha1_to_hex(sha1), entlen+baselen, path);
 349                 }
 350
 351                 /*
 352                  * CE_REMOVE entries are removed before the index is
 353                  * written to disk. Skip them to remain consistent
 354                  * with the future on-disk index.
 355                  */
 356                 if (ce->ce_flags & CE_REMOVE) {
 357                         *skip_count = *skip_count + 1;
 358                         continue;
 359                 }
 360
 361                 /*
 362                  * CE_INTENT_TO_ADD entries exist on on-disk index but
 363                  * they are not part of generated trees. Invalidate up
 364                  * to root to force cache-tree users to read elsewhere.
 365                  */
 366                 if (ce->ce_flags & CE_INTENT_TO_ADD) {
 367                         to_invalidate = 1;
 368                         continue;
 369                 }
 370
 371                 strbuf_grow(&buffer, entlen + 100);
 372                 strbuf_addf(&buffer, "%o %.*s%c", mode, entlen, path + baselen, '\0');
 373                 strbuf_add(&buffer, sha1, 20);
 374
 375 #if DEBUG
 376                 fprintf(stderr, "cache-tree update-one %o %.*s\n",
 377                         mode, entlen, path + baselen);
 378 #endif
 379         }
 380
 381         if (dryrun)
 382                 hash_sha1_file(buffer.buf, buffer.len, tree_type, it->sha1);
 383         else if (write_sha1_file(buffer.buf, buffer.len, tree_type, it->sha1)) {
 384                 strbuf_release(&buffer);
 385                 return -1;
 386         }
 387
 388         strbuf_release(&buffer);
 389         it->entry_count = to_invalidate ? -1 : i - *skip_count;
 390 #if DEBUG
 391         fprintf(stderr, "cache-tree update-one (%d ent, %d subtree) %s\n",
 392                 it->entry_count, it->subtree_nr,
 393                 sha1_to_hex(it->sha1));
 394 #endif
 395         return i;
 396 }
 397
 398 int cache_tree_update(struct cache_tree *it,
 399                       const struct cache_entry * const *cache,
 400                       int entries,
 401                       int flags)
 402 {
 403         int i, skip;
 404         i = verify_cache(cache, entries, flags);
 405         if (i)
 406                 return i;
 407         i = update_one(it, cache, entries, "", 0, &skip, flags);
 408         if (i < 0)
 409                 return i;
 410         return 0;
 411 }
 412
 413 static void write_one(struct strbuf *buffer, struct cache_tree *it,
 414                       const char *path, int pathlen)
 415 {
 416         int i;
 417
 418         /* One "cache-tree" entry consists of the following:
 419          * path (NUL terminated)
 420          * entry_count, subtree_nr ("%d %d\n")
 421          * tree-sha1 (missing if invalid)
 422          * subtree_nr "cache-tree" entries for subtrees.
 423          */
 424         strbuf_grow(buffer, pathlen + 100);
 425         strbuf_add(buffer, path, pathlen);
 426         strbuf_addf(buffer, "%c%d %d\n", 0, it->entry_count, it->subtree_nr);
 427
 428 #if DEBUG
 429         if (0 <= it->entry_count)
 430                 fprintf(stderr, "cache-tree <%.*s> (%d ent, %d subtree) %s\n",
 431                         pathlen, path, it->entry_count, it->subtree_nr,
 432                         sha1_to_hex(it->sha1));
 433         else
 434                 fprintf(stderr, "cache-tree <%.*s> (%d subtree) invalid\n",
 435                         pathlen, path, it->subtree_nr);
 436 #endif
 437
 438         if (0 <= it->entry_count) {
 439                 strbuf_add(buffer, it->sha1, 20);
 440         }
 441         for (i = 0; i < it->subtree_nr; i++) {
 442                 struct cache_tree_sub *down = it->down[i];
 443                 if (i) {
 444                         struct cache_tree_sub *prev = it->down[i-1];
 445                         if (subtree_name_cmp(down->name, down->namelen,
 446                                              prev->name, prev->namelen) <= 0)
 447                                 die("fatal - unsorted cache subtree");
 448                 }
 449                 write_one(buffer, down->cache_tree, down->name, down->namelen);
 450         }
 451 }
 452
 453 void cache_tree_write(struct strbuf *sb, struct cache_tree *root)
 454 {
 455         write_one(sb, root, "", 0);
 456 }
 457
 458 static struct cache_tree *read_one(const char **buffer, unsigned long *size_p)
 459 {
 460         const char *buf = *buffer;
 461         unsigned long size = *size_p;
 462         const char *cp;
 463         char *ep;
 464         struct cache_tree *it;
 465         int i, subtree_nr;
 466
 467         it = NULL;
 468         /* skip name, but make sure name exists */
 469         while (size && *buf) {
 470                 size--;
 471                 buf++;
 472         }
 473         if (!size)
 474                 goto free_return;
 475         buf++; size--;
 476         it = cache_tree();
 477
 478         cp = buf;
 479         it->entry_count = strtol(cp, &ep, 10);
 480         if (cp == ep)
 481                 goto free_return;
 482         cp = ep;
 483         subtree_nr = strtol(cp, &ep, 10);
 484         if (cp == ep)
 485                 goto free_return;
 486         while (size && *buf && *buf != '\n') {
 487                 size--;
 488                 buf++;
 489         }
 490         if (!size)
 491                 goto free_return;
 492         buf++; size--;
 493         if (0 <= it->entry_count) {
 494                 if (size < 20)
 495                         goto free_return;
 496                 hashcpy(it->sha1, (const unsigned char*)buf);
 497                 buf += 20;
 498                 size -= 20;
 499         }
 500
 501 #if DEBUG
 502         if (0 <= it->entry_count)
 503                 fprintf(stderr, "cache-tree <%s> (%d ent, %d subtree) %s\n",
 504                         *buffer, it->entry_count, subtree_nr,
 505                         sha1_to_hex(it->sha1));
 506         else
 507                 fprintf(stderr, "cache-tree <%s> (%d subtrees) invalid\n",
 508                         *buffer, subtree_nr);
 509 #endif
 510
 511         /*
 512          * Just a heuristic -- we do not add directories that often but
 513          * we do not want to have to extend it immediately when we do,
 514          * hence +2.
 515          */
 516         it->subtree_alloc = subtree_nr + 2;
 517         it->down = xcalloc(it->subtree_alloc, sizeof(struct cache_tree_sub *));
 518         for (i = 0; i < subtree_nr; i++) {
 519                 /* read each subtree */
 520                 struct cache_tree *sub;
 521                 struct cache_tree_sub *subtree;
 522                 const char *name = buf;
 523
 524                 sub = read_one(&buf, &size);
 525                 if (!sub)
 526                         goto free_return;
 527                 subtree = cache_tree_sub(it, name);
 528                 subtree->cache_tree = sub;
 529         }
 530         if (subtree_nr != it->subtree_nr)
 531                 die("cache-tree: internal error");
 532         *buffer = buf;
 533         *size_p = size;
 534         return it;
 535
 536  free_return:
 537         cache_tree_free(&it);
 538         return NULL;
 539 }
 540
 541 struct cache_tree *cache_tree_read(const char *buffer, unsigned long size)
 542 {
 543         if (buffer[0])
 544                 return NULL; /* not the whole tree */
 545         return read_one(&buffer, &size);
 546 }
 547
 548 static struct cache_tree *cache_tree_find(struct cache_tree *it, const char *path)
 549 {
 550         if (!it)
 551                 return NULL;
 552         while (*path) {
 553                 const char *slash;
 554                 struct cache_tree_sub *sub;
 555
 556                 slash = strchrnul(path, '/');
 557                 /* between path and slash is the name of the
 558                  * subtree to look for.
 559                  */
 560                 sub = find_subtree(it, path, slash - path, 0);
 561                 if (!sub)
 562                         return NULL;
 563                 it = sub->cache_tree;
 564                 if (*slash)
 565                         while (*slash && *slash == '/')
 566                                 slash++;
 567                 if (!*slash)
 568                         return it; /* prefix ended with slashes */
 569                 path = slash;
 570         }
 571         return it;
 572 }
 573
 574 int write_cache_as_tree(unsigned char *sha1, int flags, const char *prefix)
 575 {
 576         int entries, was_valid, newfd;
 577         struct lock_file *lock_file;
 578
 579         /*
 580          * We can't free this memory, it becomes part of a linked list
 581          * parsed atexit()
 582          */
 583         lock_file = xcalloc(1, sizeof(struct lock_file));
 584
 585         newfd = hold_locked_index(lock_file, 1);
 586
 587         entries = read_cache();
 588         if (entries < 0)
 589                 return WRITE_TREE_UNREADABLE_INDEX;
 590         if (flags & WRITE_TREE_IGNORE_CACHE_TREE)
 591                 cache_tree_free(&(active_cache_tree));
 592
 593         if (!active_cache_tree)
 594                 active_cache_tree = cache_tree();
 595
 596         was_valid = cache_tree_fully_valid(active_cache_tree);
 597         if (!was_valid) {
 598                 if (cache_tree_update(active_cache_tree,
 599                                       (const struct cache_entry * const *)active_cache,
 600                                       active_nr, flags) < 0)
 601                         return WRITE_TREE_UNMERGED_INDEX;
 602                 if (0 <= newfd) {
 603                         if (!write_cache(newfd, active_cache, active_nr) &&
 604                             !commit_lock_file(lock_file))
 605                                 newfd = -1;
 606                 }
 607                 /* Not being able to write is fine -- we are only interested
 608                  * in updating the cache-tree part, and if the next caller
 609                  * ends up using the old index with unupdated cache-tree part
 610                  * it misses the work we did here, but that is just a
 611                  * performance penalty and not a big deal.
 612                  */
 613         }
 614
 615         if (prefix) {
 616                 struct cache_tree *subtree =
 617                         cache_tree_find(active_cache_tree, prefix);
 618                 if (!subtree)
 619                         return WRITE_TREE_PREFIX_ERROR;
 620                 hashcpy(sha1, subtree->sha1);
 621         }
 622         else
 623                 hashcpy(sha1, active_cache_tree->sha1);
 624
 625         if (0 <= newfd)
 626                 rollback_lock_file(lock_file);
 627
 628         return 0;
 629 }
 630
 631 static void prime_cache_tree_rec(struct cache_tree *it, struct tree *tree)
 632 {
 633         struct tree_desc desc;
 634         struct name_entry entry;
 635         int cnt;
 636
 637         hashcpy(it->sha1, tree->object.sha1);
 638         init_tree_desc(&desc, tree->buffer, tree->size);
 639         cnt = 0;
 640         while (tree_entry(&desc, &entry)) {
 641                 if (!S_ISDIR(entry.mode))
 642                         cnt++;
 643                 else {
 644                         struct cache_tree_sub *sub;
 645                         struct tree *subtree = lookup_tree(entry.sha1);
 646                         if (!subtree->object.parsed)
 647                                 parse_tree(subtree);
 648                         sub = cache_tree_sub(it, entry.path);
 649                         sub->cache_tree = cache_tree();
 650                         prime_cache_tree_rec(sub->cache_tree, subtree);
 651                         cnt += sub->cache_tree->entry_count;
 652                 }
 653         }
 654         it->entry_count = cnt;
 655 }
 656
 657 void prime_cache_tree(struct cache_tree **it, struct tree *tree)
 658 {
 659         cache_tree_free(it);
 660         *it = cache_tree();
 661         prime_cache_tree_rec(*it, tree);
 662 }
 663
 664 /*
 665  * find the cache_tree that corresponds to the current level without
 666  * exploding the full path into textual form.  The root of the
 667  * cache tree is given as "root", and our current level is "info".
 668  * (1) When at root level, info->prev is NULL, so it is "root" itself.
 669  * (2) Otherwise, find the cache_tree that corresponds to one level
 670  *     above us, and find ourselves in there.
 671  */
 672 static struct cache_tree *find_cache_tree_from_traversal(struct cache_tree *root,
 673                                                          struct traverse_info *info)
 674 {
 675         struct cache_tree *our_parent;
 676
 677         if (!info->prev)
 678                 return root;
 679         our_parent = find_cache_tree_from_traversal(root, info->prev);
 680         return cache_tree_find(our_parent, info->name.path);
 681 }
 682
 683 int cache_tree_matches_traversal(struct cache_tree *root,
 684                                  struct name_entry *ent,
 685                                  struct traverse_info *info)
 686 {
 687         struct cache_tree *it;
 688
 689         it = find_cache_tree_from_traversal(root, info);
 690         it = cache_tree_find(it, ent->path);
 691         if (it && it->entry_count > 0 && !hashcmp(ent->sha1, it->sha1))
 692                 return it->entry_count;
 693         return 0;
 694 }
 695
 696 int update_main_cache_tree(int flags)
 697 {
 698         if (!the_index.cache_tree)
 699                 the_index.cache_tree = cache_tree();
 700         return cache_tree_update(the_index.cache_tree,
 701                                  (const struct cache_entry * const *)the_index.cache,
 702                                  the_index.cache_nr, flags);
 703 }