git.oblomov.eu Git - git/blob - pack-bitmap-write.c

   1 #include "cache.h"
   2 #include "object-store.h"
   3 #include "commit.h"
   4 #include "tag.h"
   5 #include "diff.h"
   6 #include "revision.h"
   7 #include "list-objects.h"
   8 #include "progress.h"
   9 #include "pack-revindex.h"
  10 #include "pack.h"
  11 #include "pack-bitmap.h"
  12 #include "sha1-lookup.h"
  13 #include "pack-objects.h"
  14 #include "commit-reach.h"
  15 #include "prio-queue.h"
  16
  17 struct bitmapped_commit {
  18         struct commit *commit;
  19         struct ewah_bitmap *bitmap;
  20         struct ewah_bitmap *write_as;
  21         int flags;
  22         int xor_offset;
  23         uint32_t commit_pos;
  24 };
  25
  26 struct bitmap_writer {
  27         struct ewah_bitmap *commits;
  28         struct ewah_bitmap *trees;
  29         struct ewah_bitmap *blobs;
  30         struct ewah_bitmap *tags;
  31
  32         kh_oid_map_t *bitmaps;
  33         struct packing_data *to_pack;
  34
  35         struct bitmapped_commit *selected;
  36         unsigned int selected_nr, selected_alloc;
  37
  38         struct progress *progress;
  39         int show_progress;
  40         unsigned char pack_checksum[GIT_MAX_RAWSZ];
  41 };
  42
  43 static struct bitmap_writer writer;
  44
  45 void bitmap_writer_show_progress(int show)
  46 {
  47         writer.show_progress = show;
  48 }
  49
  50 /**
  51  * Build the initial type index for the packfile
  52  */
  53 void bitmap_writer_build_type_index(struct packing_data *to_pack,
  54                                     struct pack_idx_entry **index,
  55                                     uint32_t index_nr)
  56 {
  57         uint32_t i;
  58
  59         writer.commits = ewah_new();
  60         writer.trees = ewah_new();
  61         writer.blobs = ewah_new();
  62         writer.tags = ewah_new();
  63         ALLOC_ARRAY(to_pack->in_pack_pos, to_pack->nr_objects);
  64
  65         for (i = 0; i < index_nr; ++i) {
  66                 struct object_entry *entry = (struct object_entry *)index[i];
  67                 enum object_type real_type;
  68
  69                 oe_set_in_pack_pos(to_pack, entry, i);
  70
  71                 switch (oe_type(entry)) {
  72                 case OBJ_COMMIT:
  73                 case OBJ_TREE:
  74                 case OBJ_BLOB:
  75                 case OBJ_TAG:
  76                         real_type = oe_type(entry);
  77                         break;
  78
  79                 default:
  80                         real_type = oid_object_info(to_pack->repo,
  81                                                     &entry->idx.oid, NULL);
  82                         break;
  83                 }
  84
  85                 switch (real_type) {
  86                 case OBJ_COMMIT:
  87                         ewah_set(writer.commits, i);
  88                         break;
  89
  90                 case OBJ_TREE:
  91                         ewah_set(writer.trees, i);
  92                         break;
  93
  94                 case OBJ_BLOB:
  95                         ewah_set(writer.blobs, i);
  96                         break;
  97
  98                 case OBJ_TAG:
  99                         ewah_set(writer.tags, i);
 100                         break;
 101
 102                 default:
 103                         die("Missing type information for %s (%d/%d)",
 104                             oid_to_hex(&entry->idx.oid), real_type,
 105                             oe_type(entry));
 106                 }
 107         }
 108 }
 109
 110 /**
 111  * Compute the actual bitmaps
 112  */
 113
 114 static inline void push_bitmapped_commit(struct commit *commit)
 115 {
 116         if (writer.selected_nr >= writer.selected_alloc) {
 117                 writer.selected_alloc = (writer.selected_alloc + 32) * 2;
 118                 REALLOC_ARRAY(writer.selected, writer.selected_alloc);
 119         }
 120
 121         writer.selected[writer.selected_nr].commit = commit;
 122         writer.selected[writer.selected_nr].bitmap = NULL;
 123         writer.selected[writer.selected_nr].flags = 0;
 124
 125         writer.selected_nr++;
 126 }
 127
 128 static uint32_t find_object_pos(const struct object_id *oid)
 129 {
 130         struct object_entry *entry = packlist_find(writer.to_pack, oid);
 131
 132         if (!entry) {
 133                 die("Failed to write bitmap index. Packfile doesn't have full closure "
 134                         "(object %s is missing)", oid_to_hex(oid));
 135         }
 136
 137         return oe_in_pack_pos(writer.to_pack, entry);
 138 }
 139
 140 static void compute_xor_offsets(void)
 141 {
 142         static const int MAX_XOR_OFFSET_SEARCH = 10;
 143
 144         int i, next = 0;
 145
 146         while (next < writer.selected_nr) {
 147                 struct bitmapped_commit *stored = &writer.selected[next];
 148
 149                 int best_offset = 0;
 150                 struct ewah_bitmap *best_bitmap = stored->bitmap;
 151                 struct ewah_bitmap *test_xor;
 152
 153                 for (i = 1; i <= MAX_XOR_OFFSET_SEARCH; ++i) {
 154                         int curr = next - i;
 155
 156                         if (curr < 0)
 157                                 break;
 158
 159                         test_xor = ewah_pool_new();
 160                         ewah_xor(writer.selected[curr].bitmap, stored->bitmap, test_xor);
 161
 162                         if (test_xor->buffer_size < best_bitmap->buffer_size) {
 163                                 if (best_bitmap != stored->bitmap)
 164                                         ewah_pool_free(best_bitmap);
 165
 166                                 best_bitmap = test_xor;
 167                                 best_offset = i;
 168                         } else {
 169                                 ewah_pool_free(test_xor);
 170                         }
 171                 }
 172
 173                 stored->xor_offset = best_offset;
 174                 stored->write_as = best_bitmap;
 175
 176                 next++;
 177         }
 178 }
 179
 180 struct bb_commit {
 181         struct commit_list *reverse_edges;
 182         struct bitmap *commit_mask;
 183         struct bitmap *bitmap;
 184         unsigned selected:1,
 185                  maximal:1;
 186         unsigned idx; /* within selected array */
 187 };
 188
 189 define_commit_slab(bb_data, struct bb_commit);
 190
 191 struct bitmap_builder {
 192         struct bb_data data;
 193         struct commit **commits;
 194         size_t commits_nr, commits_alloc;
 195 };
 196
 197 static void bitmap_builder_init(struct bitmap_builder *bb,
 198                                 struct bitmap_writer *writer)
 199 {
 200         struct rev_info revs;
 201         struct commit *commit;
 202         unsigned int i, num_maximal;
 203
 204         memset(bb, 0, sizeof(*bb));
 205         init_bb_data(&bb->data);
 206
 207         reset_revision_walk();
 208         repo_init_revisions(writer->to_pack->repo, &revs, NULL);
 209         revs.topo_order = 1;
 210
 211         for (i = 0; i < writer->selected_nr; i++) {
 212                 struct commit *c = writer->selected[i].commit;
 213                 struct bb_commit *ent = bb_data_at(&bb->data, c);
 214
 215                 ent->selected = 1;
 216                 ent->maximal = 1;
 217                 ent->idx = i;
 218
 219                 ent->commit_mask = bitmap_new();
 220                 bitmap_set(ent->commit_mask, i);
 221
 222                 add_pending_object(&revs, &c->object, "");
 223         }
 224         num_maximal = writer->selected_nr;
 225
 226         if (prepare_revision_walk(&revs))
 227                 die("revision walk setup failed");
 228
 229         while ((commit = get_revision(&revs))) {
 230                 struct commit_list *p;
 231                 struct bb_commit *c_ent;
 232
 233                 parse_commit_or_die(commit);
 234
 235                 c_ent = bb_data_at(&bb->data, commit);
 236
 237                 if (c_ent->maximal) {
 238                         if (!c_ent->selected) {
 239                                 bitmap_set(c_ent->commit_mask, num_maximal);
 240                                 num_maximal++;
 241                         }
 242
 243                         ALLOC_GROW(bb->commits, bb->commits_nr + 1, bb->commits_alloc);
 244                         bb->commits[bb->commits_nr++] = commit;
 245                 }
 246
 247                 for (p = commit->parents; p; p = p->next) {
 248                         struct bb_commit *p_ent = bb_data_at(&bb->data, p->item);
 249                         int c_not_p, p_not_c;
 250
 251                         if (!p_ent->commit_mask) {
 252                                 p_ent->commit_mask = bitmap_new();
 253                                 c_not_p = 1;
 254                                 p_not_c = 0;
 255                         } else {
 256                                 c_not_p = bitmap_is_subset(c_ent->commit_mask, p_ent->commit_mask);
 257                                 p_not_c = bitmap_is_subset(p_ent->commit_mask, c_ent->commit_mask);
 258                         }
 259
 260                         if (!c_not_p)
 261                                 continue;
 262
 263                         bitmap_or(p_ent->commit_mask, c_ent->commit_mask);
 264
 265                         if (p_not_c)
 266                                 p_ent->maximal = 1;
 267                         else {
 268                                 p_ent->maximal = 0;
 269                                 free_commit_list(p_ent->reverse_edges);
 270                                 p_ent->reverse_edges = NULL;
 271                         }
 272
 273                         if (c_ent->maximal) {
 274                                 commit_list_insert(commit, &p_ent->reverse_edges);
 275                         } else {
 276                                 struct commit_list *cc = c_ent->reverse_edges;
 277
 278                                 for (; cc; cc = cc->next) {
 279                                         if (!commit_list_contains(cc->item, p_ent->reverse_edges))
 280                                                 commit_list_insert(cc->item, &p_ent->reverse_edges);
 281                                 }
 282                         }
 283                 }
 284
 285                 bitmap_free(c_ent->commit_mask);
 286                 c_ent->commit_mask = NULL;
 287         }
 288
 289         trace2_data_intmax("pack-bitmap-write", the_repository,
 290                            "num_selected_commits", writer->selected_nr);
 291         trace2_data_intmax("pack-bitmap-write", the_repository,
 292                            "num_maximal_commits", num_maximal);
 293 }
 294
 295 static void bitmap_builder_clear(struct bitmap_builder *bb)
 296 {
 297         clear_bb_data(&bb->data);
 298         free(bb->commits);
 299         bb->commits_nr = bb->commits_alloc = 0;
 300 }
 301
 302 static void fill_bitmap_tree(struct bitmap *bitmap,
 303                              struct tree *tree)
 304 {
 305         uint32_t pos;
 306         struct tree_desc desc;
 307         struct name_entry entry;
 308
 309         /*
 310          * If our bit is already set, then there is nothing to do. Both this
 311          * tree and all of its children will be set.
 312          */
 313         pos = find_object_pos(&tree->object.oid);
 314         if (bitmap_get(bitmap, pos))
 315                 return;
 316         bitmap_set(bitmap, pos);
 317
 318         if (parse_tree(tree) < 0)
 319                 die("unable to load tree object %s",
 320                     oid_to_hex(&tree->object.oid));
 321         init_tree_desc(&desc, tree->buffer, tree->size);
 322
 323         while (tree_entry(&desc, &entry)) {
 324                 switch (object_type(entry.mode)) {
 325                 case OBJ_TREE:
 326                         fill_bitmap_tree(bitmap,
 327                                          lookup_tree(the_repository, &entry.oid));
 328                         break;
 329                 case OBJ_BLOB:
 330                         bitmap_set(bitmap, find_object_pos(&entry.oid));
 331                         break;
 332                 default:
 333                         /* Gitlink, etc; not reachable */
 334                         break;
 335                 }
 336         }
 337
 338         free_tree_buffer(tree);
 339 }
 340
 341 static void fill_bitmap_commit(struct bb_commit *ent,
 342                                struct commit *commit,
 343                                struct prio_queue *queue,
 344                                struct prio_queue *tree_queue,
 345                                struct bitmap_index *old_bitmap,
 346                                const uint32_t *mapping)
 347 {
 348         if (!ent->bitmap)
 349                 ent->bitmap = bitmap_new();
 350
 351         prio_queue_put(queue, commit);
 352
 353         while (queue->nr) {
 354                 struct commit_list *p;
 355                 struct commit *c = prio_queue_get(queue);
 356
 357                 if (old_bitmap && mapping) {
 358                         struct ewah_bitmap *old = bitmap_for_commit(old_bitmap, c);
 359                         /*
 360                          * If this commit has an old bitmap, then translate that
 361                          * bitmap and add its bits to this one. No need to walk
 362                          * parents or the tree for this commit.
 363                          */
 364                         if (old && !rebuild_bitmap(mapping, old, ent->bitmap))
 365                                 continue;
 366                 }
 367
 368                 /*
 369                  * Mark ourselves and queue our tree. The commit
 370                  * walk ensures we cover all parents.
 371                  */
 372                 bitmap_set(ent->bitmap, find_object_pos(&c->object.oid));
 373                 prio_queue_put(tree_queue, get_commit_tree(c));
 374
 375                 for (p = c->parents; p; p = p->next) {
 376                         int pos = find_object_pos(&p->item->object.oid);
 377                         if (!bitmap_get(ent->bitmap, pos)) {
 378                                 bitmap_set(ent->bitmap, pos);
 379                                 prio_queue_put(queue, p->item);
 380                         }
 381                 }
 382         }
 383
 384         while (tree_queue->nr)
 385                 fill_bitmap_tree(ent->bitmap, prio_queue_get(tree_queue));
 386 }
 387
 388 static void store_selected(struct bb_commit *ent, struct commit *commit)
 389 {
 390         struct bitmapped_commit *stored = &writer.selected[ent->idx];
 391         khiter_t hash_pos;
 392         int hash_ret;
 393
 394         stored->bitmap = bitmap_to_ewah(ent->bitmap);
 395
 396         hash_pos = kh_put_oid_map(writer.bitmaps, commit->object.oid, &hash_ret);
 397         if (hash_ret == 0)
 398                 die("Duplicate entry when writing index: %s",
 399                     oid_to_hex(&commit->object.oid));
 400         kh_value(writer.bitmaps, hash_pos) = stored;
 401 }
 402
 403 void bitmap_writer_build(struct packing_data *to_pack)
 404 {
 405         struct bitmap_builder bb;
 406         size_t i;
 407         int nr_stored = 0; /* for progress */
 408         struct prio_queue queue = { compare_commits_by_gen_then_commit_date };
 409         struct prio_queue tree_queue = { NULL };
 410         struct bitmap_index *old_bitmap;
 411         uint32_t *mapping;
 412
 413         writer.bitmaps = kh_init_oid_map();
 414         writer.to_pack = to_pack;
 415
 416         if (writer.show_progress)
 417                 writer.progress = start_progress("Building bitmaps", writer.selected_nr);
 418         trace2_region_enter("pack-bitmap-write", "building_bitmaps_total",
 419                             the_repository);
 420
 421         old_bitmap = prepare_bitmap_git(to_pack->repo);
 422         if (old_bitmap)
 423                 mapping = create_bitmap_mapping(old_bitmap, to_pack);
 424         else
 425                 mapping = NULL;
 426
 427         bitmap_builder_init(&bb, &writer);
 428         for (i = bb.commits_nr; i > 0; i--) {
 429                 struct commit *commit = bb.commits[i-1];
 430                 struct bb_commit *ent = bb_data_at(&bb.data, commit);
 431                 struct commit *child;
 432                 int reused = 0;
 433
 434                 fill_bitmap_commit(ent, commit, &queue, &tree_queue,
 435                                    old_bitmap, mapping);
 436
 437                 if (ent->selected) {
 438                         store_selected(ent, commit);
 439                         nr_stored++;
 440                         display_progress(writer.progress, nr_stored);
 441                 }
 442
 443                 while ((child = pop_commit(&ent->reverse_edges))) {
 444                         struct bb_commit *child_ent =
 445                                 bb_data_at(&bb.data, child);
 446
 447                         if (child_ent->bitmap)
 448                                 bitmap_or(child_ent->bitmap, ent->bitmap);
 449                         else if (reused)
 450                                 child_ent->bitmap = bitmap_dup(ent->bitmap);
 451                         else {
 452                                 child_ent->bitmap = ent->bitmap;
 453                                 reused = 1;
 454                         }
 455                 }
 456                 if (!reused)
 457                         bitmap_free(ent->bitmap);
 458                 ent->bitmap = NULL;
 459         }
 460         clear_prio_queue(&queue);
 461         clear_prio_queue(&tree_queue);
 462         bitmap_builder_clear(&bb);
 463         free(mapping);
 464
 465         trace2_region_leave("pack-bitmap-write", "building_bitmaps_total",
 466                             the_repository);
 467
 468         stop_progress(&writer.progress);
 469
 470         compute_xor_offsets();
 471 }
 472
 473 /**
 474  * Select the commits that will be bitmapped
 475  */
 476 static inline unsigned int next_commit_index(unsigned int idx)
 477 {
 478         static const unsigned int MIN_COMMITS = 100;
 479         static const unsigned int MAX_COMMITS = 5000;
 480
 481         static const unsigned int MUST_REGION = 100;
 482         static const unsigned int MIN_REGION = 20000;
 483
 484         unsigned int offset, next;
 485
 486         if (idx <= MUST_REGION)
 487                 return 0;
 488
 489         if (idx <= MIN_REGION) {
 490                 offset = idx - MUST_REGION;
 491                 return (offset < MIN_COMMITS) ? offset : MIN_COMMITS;
 492         }
 493
 494         offset = idx - MIN_REGION;
 495         next = (offset < MAX_COMMITS) ? offset : MAX_COMMITS;
 496
 497         return (next > MIN_COMMITS) ? next : MIN_COMMITS;
 498 }
 499
 500 static int date_compare(const void *_a, const void *_b)
 501 {
 502         struct commit *a = *(struct commit **)_a;
 503         struct commit *b = *(struct commit **)_b;
 504         return (long)b->date - (long)a->date;
 505 }
 506
 507 void bitmap_writer_select_commits(struct commit **indexed_commits,
 508                                   unsigned int indexed_commits_nr,
 509                                   int max_bitmaps)
 510 {
 511         unsigned int i = 0, j, next;
 512
 513         QSORT(indexed_commits, indexed_commits_nr, date_compare);
 514
 515         if (writer.show_progress)
 516                 writer.progress = start_progress("Selecting bitmap commits", 0);
 517
 518         if (indexed_commits_nr < 100) {
 519                 for (i = 0; i < indexed_commits_nr; ++i)
 520                         push_bitmapped_commit(indexed_commits[i]);
 521                 return;
 522         }
 523
 524         for (;;) {
 525                 struct commit *chosen = NULL;
 526
 527                 next = next_commit_index(i);
 528
 529                 if (i + next >= indexed_commits_nr)
 530                         break;
 531
 532                 if (max_bitmaps > 0 && writer.selected_nr >= max_bitmaps) {
 533                         writer.selected_nr = max_bitmaps;
 534                         break;
 535                 }
 536
 537                 if (next == 0) {
 538                         chosen = indexed_commits[i];
 539                 } else {
 540                         chosen = indexed_commits[i + next];
 541
 542                         for (j = 0; j <= next; ++j) {
 543                                 struct commit *cm = indexed_commits[i + j];
 544
 545                                 if ((cm->object.flags & NEEDS_BITMAP) != 0) {
 546                                         chosen = cm;
 547                                         break;
 548                                 }
 549
 550                                 if (cm->parents && cm->parents->next)
 551                                         chosen = cm;
 552                         }
 553                 }
 554
 555                 push_bitmapped_commit(chosen);
 556
 557                 i += next + 1;
 558                 display_progress(writer.progress, i);
 559         }
 560
 561         stop_progress(&writer.progress);
 562 }
 563
 564
 565 static int hashwrite_ewah_helper(void *f, const void *buf, size_t len)
 566 {
 567         /* hashwrite will die on error */
 568         hashwrite(f, buf, len);
 569         return len;
 570 }
 571
 572 /**
 573  * Write the bitmap index to disk
 574  */
 575 static inline void dump_bitmap(struct hashfile *f, struct ewah_bitmap *bitmap)
 576 {
 577         if (ewah_serialize_to(bitmap, hashwrite_ewah_helper, f) < 0)
 578                 die("Failed to write bitmap index");
 579 }
 580
 581 static const unsigned char *sha1_access(size_t pos, void *table)
 582 {
 583         struct pack_idx_entry **index = table;
 584         return index[pos]->oid.hash;
 585 }
 586
 587 static void write_selected_commits_v1(struct hashfile *f,
 588                                       struct pack_idx_entry **index,
 589                                       uint32_t index_nr)
 590 {
 591         int i;
 592
 593         for (i = 0; i < writer.selected_nr; ++i) {
 594                 struct bitmapped_commit *stored = &writer.selected[i];
 595
 596                 int commit_pos =
 597                         sha1_pos(stored->commit->object.oid.hash, index, index_nr, sha1_access);
 598
 599                 if (commit_pos < 0)
 600                         BUG("trying to write commit not in index");
 601
 602                 hashwrite_be32(f, commit_pos);
 603                 hashwrite_u8(f, stored->xor_offset);
 604                 hashwrite_u8(f, stored->flags);
 605
 606                 dump_bitmap(f, stored->write_as);
 607         }
 608 }
 609
 610 static void write_hash_cache(struct hashfile *f,
 611                              struct pack_idx_entry **index,
 612                              uint32_t index_nr)
 613 {
 614         uint32_t i;
 615
 616         for (i = 0; i < index_nr; ++i) {
 617                 struct object_entry *entry = (struct object_entry *)index[i];
 618                 hashwrite_be32(f, entry->hash);
 619         }
 620 }
 621
 622 void bitmap_writer_set_checksum(unsigned char *sha1)
 623 {
 624         hashcpy(writer.pack_checksum, sha1);
 625 }
 626
 627 void bitmap_writer_finish(struct pack_idx_entry **index,
 628                           uint32_t index_nr,
 629                           const char *filename,
 630                           uint16_t options)
 631 {
 632         static uint16_t default_version = 1;
 633         static uint16_t flags = BITMAP_OPT_FULL_DAG;
 634         struct strbuf tmp_file = STRBUF_INIT;
 635         struct hashfile *f;
 636
 637         struct bitmap_disk_header header;
 638
 639         int fd = odb_mkstemp(&tmp_file, "pack/tmp_bitmap_XXXXXX");
 640
 641         f = hashfd(fd, tmp_file.buf);
 642
 643         memcpy(header.magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE));
 644         header.version = htons(default_version);
 645         header.options = htons(flags | options);
 646         header.entry_count = htonl(writer.selected_nr);
 647         hashcpy(header.checksum, writer.pack_checksum);
 648
 649         hashwrite(f, &header, sizeof(header) - GIT_MAX_RAWSZ + the_hash_algo->rawsz);
 650         dump_bitmap(f, writer.commits);
 651         dump_bitmap(f, writer.trees);
 652         dump_bitmap(f, writer.blobs);
 653         dump_bitmap(f, writer.tags);
 654         write_selected_commits_v1(f, index, index_nr);
 655
 656         if (options & BITMAP_OPT_HASH_CACHE)
 657                 write_hash_cache(f, index, index_nr);
 658
 659         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC | CSUM_CLOSE);
 660
 661         if (adjust_shared_perm(tmp_file.buf))
 662                 die_errno("unable to make temporary bitmap file readable");
 663
 664         if (rename(tmp_file.buf, filename))
 665                 die_errno("unable to rename temporary bitmap file to '%s'", filename);
 666
 667         strbuf_release(&tmp_file);
 668 }