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 = 0;
 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         revs.first_parent_only = 1;
 211
 212         for (i = 0; i < writer->selected_nr; i++) {
 213                 struct commit *c = writer->selected[i].commit;
 214                 struct bb_commit *ent = bb_data_at(&bb->data, c);
 215
 216                 ent->selected = 1;
 217                 ent->maximal = 1;
 218                 ent->idx = i;
 219
 220                 ent->commit_mask = bitmap_new();
 221                 bitmap_set(ent->commit_mask, i);
 222
 223                 add_pending_object(&revs, &c->object, "");
 224         }
 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 = commit->parents;
 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                         num_maximal++;
 239                         ALLOC_GROW(bb->commits, bb->commits_nr + 1, bb->commits_alloc);
 240                         bb->commits[bb->commits_nr++] = commit;
 241                 }
 242
 243                 if (p) {
 244                         struct bb_commit *p_ent = bb_data_at(&bb->data, p->item);
 245                         int c_not_p, p_not_c;
 246
 247                         if (!p_ent->commit_mask) {
 248                                 p_ent->commit_mask = bitmap_new();
 249                                 c_not_p = 1;
 250                                 p_not_c = 0;
 251                         } else {
 252                                 c_not_p = bitmap_is_subset(c_ent->commit_mask, p_ent->commit_mask);
 253                                 p_not_c = bitmap_is_subset(p_ent->commit_mask, c_ent->commit_mask);
 254                         }
 255
 256                         if (!c_not_p)
 257                                 continue;
 258
 259                         bitmap_or(p_ent->commit_mask, c_ent->commit_mask);
 260
 261                         if (p_not_c)
 262                                 p_ent->maximal = 1;
 263                         else {
 264                                 p_ent->maximal = 0;
 265                                 free_commit_list(p_ent->reverse_edges);
 266                                 p_ent->reverse_edges = NULL;
 267                         }
 268
 269                         if (c_ent->maximal) {
 270                                 commit_list_insert(commit, &p_ent->reverse_edges);
 271                         } else {
 272                                 struct commit_list *cc = c_ent->reverse_edges;
 273
 274                                 for (; cc; cc = cc->next) {
 275                                         if (!commit_list_contains(cc->item, p_ent->reverse_edges))
 276                                                 commit_list_insert(cc->item, &p_ent->reverse_edges);
 277                                 }
 278                         }
 279                 }
 280
 281                 bitmap_free(c_ent->commit_mask);
 282                 c_ent->commit_mask = NULL;
 283         }
 284
 285         trace2_data_intmax("pack-bitmap-write", the_repository,
 286                            "num_selected_commits", writer->selected_nr);
 287         trace2_data_intmax("pack-bitmap-write", the_repository,
 288                            "num_maximal_commits", num_maximal);
 289 }
 290
 291 static void bitmap_builder_clear(struct bitmap_builder *bb)
 292 {
 293         clear_bb_data(&bb->data);
 294         free(bb->commits);
 295         bb->commits_nr = bb->commits_alloc = 0;
 296 }
 297
 298 static void fill_bitmap_tree(struct bitmap *bitmap,
 299                              struct tree *tree)
 300 {
 301         uint32_t pos;
 302         struct tree_desc desc;
 303         struct name_entry entry;
 304
 305         /*
 306          * If our bit is already set, then there is nothing to do. Both this
 307          * tree and all of its children will be set.
 308          */
 309         pos = find_object_pos(&tree->object.oid);
 310         if (bitmap_get(bitmap, pos))
 311                 return;
 312         bitmap_set(bitmap, pos);
 313
 314         if (parse_tree(tree) < 0)
 315                 die("unable to load tree object %s",
 316                     oid_to_hex(&tree->object.oid));
 317         init_tree_desc(&desc, tree->buffer, tree->size);
 318
 319         while (tree_entry(&desc, &entry)) {
 320                 switch (object_type(entry.mode)) {
 321                 case OBJ_TREE:
 322                         fill_bitmap_tree(bitmap,
 323                                          lookup_tree(the_repository, &entry.oid));
 324                         break;
 325                 case OBJ_BLOB:
 326                         bitmap_set(bitmap, find_object_pos(&entry.oid));
 327                         break;
 328                 default:
 329                         /* Gitlink, etc; not reachable */
 330                         break;
 331                 }
 332         }
 333
 334         free_tree_buffer(tree);
 335 }
 336
 337 static void fill_bitmap_commit(struct bb_commit *ent,
 338                                struct commit *commit,
 339                                struct prio_queue *queue,
 340                                struct prio_queue *tree_queue,
 341                                struct bitmap_index *old_bitmap,
 342                                const uint32_t *mapping)
 343 {
 344         if (!ent->bitmap)
 345                 ent->bitmap = bitmap_new();
 346
 347         prio_queue_put(queue, commit);
 348
 349         while (queue->nr) {
 350                 struct commit_list *p;
 351                 struct commit *c = prio_queue_get(queue);
 352
 353                 if (old_bitmap && mapping) {
 354                         struct ewah_bitmap *old = bitmap_for_commit(old_bitmap, c);
 355                         /*
 356                          * If this commit has an old bitmap, then translate that
 357                          * bitmap and add its bits to this one. No need to walk
 358                          * parents or the tree for this commit.
 359                          */
 360                         if (old && !rebuild_bitmap(mapping, old, ent->bitmap))
 361                                 continue;
 362                 }
 363
 364                 /*
 365                  * Mark ourselves and queue our tree. The commit
 366                  * walk ensures we cover all parents.
 367                  */
 368                 bitmap_set(ent->bitmap, find_object_pos(&c->object.oid));
 369                 prio_queue_put(tree_queue, get_commit_tree(c));
 370
 371                 for (p = c->parents; p; p = p->next) {
 372                         int pos = find_object_pos(&p->item->object.oid);
 373                         if (!bitmap_get(ent->bitmap, pos)) {
 374                                 bitmap_set(ent->bitmap, pos);
 375                                 prio_queue_put(queue, p->item);
 376                         }
 377                 }
 378         }
 379
 380         while (tree_queue->nr)
 381                 fill_bitmap_tree(ent->bitmap, prio_queue_get(tree_queue));
 382 }
 383
 384 static void store_selected(struct bb_commit *ent, struct commit *commit)
 385 {
 386         struct bitmapped_commit *stored = &writer.selected[ent->idx];
 387         khiter_t hash_pos;
 388         int hash_ret;
 389
 390         stored->bitmap = bitmap_to_ewah(ent->bitmap);
 391
 392         hash_pos = kh_put_oid_map(writer.bitmaps, commit->object.oid, &hash_ret);
 393         if (hash_ret == 0)
 394                 die("Duplicate entry when writing index: %s",
 395                     oid_to_hex(&commit->object.oid));
 396         kh_value(writer.bitmaps, hash_pos) = stored;
 397 }
 398
 399 void bitmap_writer_build(struct packing_data *to_pack)
 400 {
 401         struct bitmap_builder bb;
 402         size_t i;
 403         int nr_stored = 0; /* for progress */
 404         struct prio_queue queue = { compare_commits_by_gen_then_commit_date };
 405         struct prio_queue tree_queue = { NULL };
 406         struct bitmap_index *old_bitmap;
 407         uint32_t *mapping;
 408
 409         writer.bitmaps = kh_init_oid_map();
 410         writer.to_pack = to_pack;
 411
 412         if (writer.show_progress)
 413                 writer.progress = start_progress("Building bitmaps", writer.selected_nr);
 414         trace2_region_enter("pack-bitmap-write", "building_bitmaps_total",
 415                             the_repository);
 416
 417         old_bitmap = prepare_bitmap_git(to_pack->repo);
 418         if (old_bitmap)
 419                 mapping = create_bitmap_mapping(old_bitmap, to_pack);
 420         else
 421                 mapping = NULL;
 422
 423         bitmap_builder_init(&bb, &writer);
 424         for (i = bb.commits_nr; i > 0; i--) {
 425                 struct commit *commit = bb.commits[i-1];
 426                 struct bb_commit *ent = bb_data_at(&bb.data, commit);
 427                 struct commit *child;
 428                 int reused = 0;
 429
 430                 fill_bitmap_commit(ent, commit, &queue, &tree_queue,
 431                                    old_bitmap, mapping);
 432
 433                 if (ent->selected) {
 434                         store_selected(ent, commit);
 435                         nr_stored++;
 436                         display_progress(writer.progress, nr_stored);
 437                 }
 438
 439                 while ((child = pop_commit(&ent->reverse_edges))) {
 440                         struct bb_commit *child_ent =
 441                                 bb_data_at(&bb.data, child);
 442
 443                         if (child_ent->bitmap)
 444                                 bitmap_or(child_ent->bitmap, ent->bitmap);
 445                         else if (reused)
 446                                 child_ent->bitmap = bitmap_dup(ent->bitmap);
 447                         else {
 448                                 child_ent->bitmap = ent->bitmap;
 449                                 reused = 1;
 450                         }
 451                 }
 452                 if (!reused)
 453                         bitmap_free(ent->bitmap);
 454                 ent->bitmap = NULL;
 455         }
 456         clear_prio_queue(&queue);
 457         clear_prio_queue(&tree_queue);
 458         bitmap_builder_clear(&bb);
 459         free(mapping);
 460
 461         trace2_region_leave("pack-bitmap-write", "building_bitmaps_total",
 462                             the_repository);
 463
 464         stop_progress(&writer.progress);
 465
 466         compute_xor_offsets();
 467 }
 468
 469 /**
 470  * Select the commits that will be bitmapped
 471  */
 472 static inline unsigned int next_commit_index(unsigned int idx)
 473 {
 474         static const unsigned int MIN_COMMITS = 100;
 475         static const unsigned int MAX_COMMITS = 5000;
 476
 477         static const unsigned int MUST_REGION = 100;
 478         static const unsigned int MIN_REGION = 20000;
 479
 480         unsigned int offset, next;
 481
 482         if (idx <= MUST_REGION)
 483                 return 0;
 484
 485         if (idx <= MIN_REGION) {
 486                 offset = idx - MUST_REGION;
 487                 return (offset < MIN_COMMITS) ? offset : MIN_COMMITS;
 488         }
 489
 490         offset = idx - MIN_REGION;
 491         next = (offset < MAX_COMMITS) ? offset : MAX_COMMITS;
 492
 493         return (next > MIN_COMMITS) ? next : MIN_COMMITS;
 494 }
 495
 496 static int date_compare(const void *_a, const void *_b)
 497 {
 498         struct commit *a = *(struct commit **)_a;
 499         struct commit *b = *(struct commit **)_b;
 500         return (long)b->date - (long)a->date;
 501 }
 502
 503 void bitmap_writer_select_commits(struct commit **indexed_commits,
 504                                   unsigned int indexed_commits_nr,
 505                                   int max_bitmaps)
 506 {
 507         unsigned int i = 0, j, next;
 508
 509         QSORT(indexed_commits, indexed_commits_nr, date_compare);
 510
 511         if (writer.show_progress)
 512                 writer.progress = start_progress("Selecting bitmap commits", 0);
 513
 514         if (indexed_commits_nr < 100) {
 515                 for (i = 0; i < indexed_commits_nr; ++i)
 516                         push_bitmapped_commit(indexed_commits[i]);
 517                 return;
 518         }
 519
 520         for (;;) {
 521                 struct commit *chosen = NULL;
 522
 523                 next = next_commit_index(i);
 524
 525                 if (i + next >= indexed_commits_nr)
 526                         break;
 527
 528                 if (max_bitmaps > 0 && writer.selected_nr >= max_bitmaps) {
 529                         writer.selected_nr = max_bitmaps;
 530                         break;
 531                 }
 532
 533                 if (next == 0) {
 534                         chosen = indexed_commits[i];
 535                 } else {
 536                         chosen = indexed_commits[i + next];
 537
 538                         for (j = 0; j <= next; ++j) {
 539                                 struct commit *cm = indexed_commits[i + j];
 540
 541                                 if ((cm->object.flags & NEEDS_BITMAP) != 0) {
 542                                         chosen = cm;
 543                                         break;
 544                                 }
 545
 546                                 if (cm->parents && cm->parents->next)
 547                                         chosen = cm;
 548                         }
 549                 }
 550
 551                 push_bitmapped_commit(chosen);
 552
 553                 i += next + 1;
 554                 display_progress(writer.progress, i);
 555         }
 556
 557         stop_progress(&writer.progress);
 558 }
 559
 560
 561 static int hashwrite_ewah_helper(void *f, const void *buf, size_t len)
 562 {
 563         /* hashwrite will die on error */
 564         hashwrite(f, buf, len);
 565         return len;
 566 }
 567
 568 /**
 569  * Write the bitmap index to disk
 570  */
 571 static inline void dump_bitmap(struct hashfile *f, struct ewah_bitmap *bitmap)
 572 {
 573         if (ewah_serialize_to(bitmap, hashwrite_ewah_helper, f) < 0)
 574                 die("Failed to write bitmap index");
 575 }
 576
 577 static const unsigned char *sha1_access(size_t pos, void *table)
 578 {
 579         struct pack_idx_entry **index = table;
 580         return index[pos]->oid.hash;
 581 }
 582
 583 static void write_selected_commits_v1(struct hashfile *f,
 584                                       struct pack_idx_entry **index,
 585                                       uint32_t index_nr)
 586 {
 587         int i;
 588
 589         for (i = 0; i < writer.selected_nr; ++i) {
 590                 struct bitmapped_commit *stored = &writer.selected[i];
 591
 592                 int commit_pos =
 593                         sha1_pos(stored->commit->object.oid.hash, index, index_nr, sha1_access);
 594
 595                 if (commit_pos < 0)
 596                         BUG("trying to write commit not in index");
 597
 598                 hashwrite_be32(f, commit_pos);
 599                 hashwrite_u8(f, stored->xor_offset);
 600                 hashwrite_u8(f, stored->flags);
 601
 602                 dump_bitmap(f, stored->write_as);
 603         }
 604 }
 605
 606 static void write_hash_cache(struct hashfile *f,
 607                              struct pack_idx_entry **index,
 608                              uint32_t index_nr)
 609 {
 610         uint32_t i;
 611
 612         for (i = 0; i < index_nr; ++i) {
 613                 struct object_entry *entry = (struct object_entry *)index[i];
 614                 hashwrite_be32(f, entry->hash);
 615         }
 616 }
 617
 618 void bitmap_writer_set_checksum(unsigned char *sha1)
 619 {
 620         hashcpy(writer.pack_checksum, sha1);
 621 }
 622
 623 void bitmap_writer_finish(struct pack_idx_entry **index,
 624                           uint32_t index_nr,
 625                           const char *filename,
 626                           uint16_t options)
 627 {
 628         static uint16_t default_version = 1;
 629         static uint16_t flags = BITMAP_OPT_FULL_DAG;
 630         struct strbuf tmp_file = STRBUF_INIT;
 631         struct hashfile *f;
 632
 633         struct bitmap_disk_header header;
 634
 635         int fd = odb_mkstemp(&tmp_file, "pack/tmp_bitmap_XXXXXX");
 636
 637         f = hashfd(fd, tmp_file.buf);
 638
 639         memcpy(header.magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE));
 640         header.version = htons(default_version);
 641         header.options = htons(flags | options);
 642         header.entry_count = htonl(writer.selected_nr);
 643         hashcpy(header.checksum, writer.pack_checksum);
 644
 645         hashwrite(f, &header, sizeof(header) - GIT_MAX_RAWSZ + the_hash_algo->rawsz);
 646         dump_bitmap(f, writer.commits);
 647         dump_bitmap(f, writer.trees);
 648         dump_bitmap(f, writer.blobs);
 649         dump_bitmap(f, writer.tags);
 650         write_selected_commits_v1(f, index, index_nr);
 651
 652         if (options & BITMAP_OPT_HASH_CACHE)
 653                 write_hash_cache(f, index, index_nr);
 654
 655         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC | CSUM_CLOSE);
 656
 657         if (adjust_shared_perm(tmp_file.buf))
 658                 die_errno("unable to make temporary bitmap file readable");
 659
 660         if (rename(tmp_file.buf, filename))
 661                 die_errno("unable to rename temporary bitmap file to '%s'", filename);
 662
 663         strbuf_release(&tmp_file);
 664 }