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         kh_oid_map_t *reused;
  34         struct packing_data *to_pack;
  35
  36         struct bitmapped_commit *selected;
  37         unsigned int selected_nr, selected_alloc;
  38
  39         struct progress *progress;
  40         int show_progress;
  41         unsigned char pack_checksum[GIT_MAX_RAWSZ];
  42 };
  43
  44 static struct bitmap_writer writer;
  45
  46 void bitmap_writer_show_progress(int show)
  47 {
  48         writer.show_progress = show;
  49 }
  50
  51 /**
  52  * Build the initial type index for the packfile
  53  */
  54 void bitmap_writer_build_type_index(struct packing_data *to_pack,
  55                                     struct pack_idx_entry **index,
  56                                     uint32_t index_nr)
  57 {
  58         uint32_t i;
  59
  60         writer.commits = ewah_new();
  61         writer.trees = ewah_new();
  62         writer.blobs = ewah_new();
  63         writer.tags = ewah_new();
  64         ALLOC_ARRAY(to_pack->in_pack_pos, to_pack->nr_objects);
  65
  66         for (i = 0; i < index_nr; ++i) {
  67                 struct object_entry *entry = (struct object_entry *)index[i];
  68                 enum object_type real_type;
  69
  70                 oe_set_in_pack_pos(to_pack, entry, i);
  71
  72                 switch (oe_type(entry)) {
  73                 case OBJ_COMMIT:
  74                 case OBJ_TREE:
  75                 case OBJ_BLOB:
  76                 case OBJ_TAG:
  77                         real_type = oe_type(entry);
  78                         break;
  79
  80                 default:
  81                         real_type = oid_object_info(to_pack->repo,
  82                                                     &entry->idx.oid, NULL);
  83                         break;
  84                 }
  85
  86                 switch (real_type) {
  87                 case OBJ_COMMIT:
  88                         ewah_set(writer.commits, i);
  89                         break;
  90
  91                 case OBJ_TREE:
  92                         ewah_set(writer.trees, i);
  93                         break;
  94
  95                 case OBJ_BLOB:
  96                         ewah_set(writer.blobs, i);
  97                         break;
  98
  99                 case OBJ_TAG:
 100                         ewah_set(writer.tags, i);
 101                         break;
 102
 103                 default:
 104                         die("Missing type information for %s (%d/%d)",
 105                             oid_to_hex(&entry->idx.oid), real_type,
 106                             oe_type(entry));
 107                 }
 108         }
 109 }
 110
 111 /**
 112  * Compute the actual bitmaps
 113  */
 114
 115 static inline void push_bitmapped_commit(struct commit *commit, struct ewah_bitmap *reused)
 116 {
 117         if (writer.selected_nr >= writer.selected_alloc) {
 118                 writer.selected_alloc = (writer.selected_alloc + 32) * 2;
 119                 REALLOC_ARRAY(writer.selected, writer.selected_alloc);
 120         }
 121
 122         writer.selected[writer.selected_nr].commit = commit;
 123         writer.selected[writer.selected_nr].bitmap = reused;
 124         writer.selected[writer.selected_nr].flags = 0;
 125
 126         writer.selected_nr++;
 127 }
 128
 129 static uint32_t find_object_pos(const struct object_id *oid)
 130 {
 131         struct object_entry *entry = packlist_find(writer.to_pack, oid);
 132
 133         if (!entry) {
 134                 die("Failed to write bitmap index. Packfile doesn't have full closure "
 135                         "(object %s is missing)", oid_to_hex(oid));
 136         }
 137
 138         return oe_in_pack_pos(writer.to_pack, entry);
 139 }
 140
 141 static void compute_xor_offsets(void)
 142 {
 143         static const int MAX_XOR_OFFSET_SEARCH = 10;
 144
 145         int i, next = 0;
 146
 147         while (next < writer.selected_nr) {
 148                 struct bitmapped_commit *stored = &writer.selected[next];
 149
 150                 int best_offset = 0;
 151                 struct ewah_bitmap *best_bitmap = stored->bitmap;
 152                 struct ewah_bitmap *test_xor;
 153
 154                 for (i = 1; i <= MAX_XOR_OFFSET_SEARCH; ++i) {
 155                         int curr = next - i;
 156
 157                         if (curr < 0)
 158                                 break;
 159
 160                         test_xor = ewah_pool_new();
 161                         ewah_xor(writer.selected[curr].bitmap, stored->bitmap, test_xor);
 162
 163                         if (test_xor->buffer_size < best_bitmap->buffer_size) {
 164                                 if (best_bitmap != stored->bitmap)
 165                                         ewah_pool_free(best_bitmap);
 166
 167                                 best_bitmap = test_xor;
 168                                 best_offset = i;
 169                         } else {
 170                                 ewah_pool_free(test_xor);
 171                         }
 172                 }
 173
 174                 stored->xor_offset = best_offset;
 175                 stored->write_as = best_bitmap;
 176
 177                 next++;
 178         }
 179 }
 180
 181 struct bb_commit {
 182         struct commit_list *reverse_edges;
 183         struct bitmap *commit_mask;
 184         struct bitmap *bitmap;
 185         unsigned selected:1,
 186                  maximal:1;
 187         unsigned idx; /* within selected array */
 188 };
 189
 190 define_commit_slab(bb_data, struct bb_commit);
 191
 192 struct bitmap_builder {
 193         struct bb_data data;
 194         struct commit **commits;
 195         size_t commits_nr, commits_alloc;
 196 };
 197
 198 static void bitmap_builder_init(struct bitmap_builder *bb,
 199                                 struct bitmap_writer *writer)
 200 {
 201         struct rev_info revs;
 202         struct commit *commit;
 203         unsigned int i, num_maximal;
 204
 205         memset(bb, 0, sizeof(*bb));
 206         init_bb_data(&bb->data);
 207
 208         reset_revision_walk();
 209         repo_init_revisions(writer->to_pack->repo, &revs, NULL);
 210         revs.topo_order = 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         num_maximal = writer->selected_nr;
 226
 227         if (prepare_revision_walk(&revs))
 228                 die("revision walk setup failed");
 229
 230         while ((commit = get_revision(&revs))) {
 231                 struct commit_list *p;
 232                 struct bb_commit *c_ent;
 233
 234                 parse_commit_or_die(commit);
 235
 236                 c_ent = bb_data_at(&bb->data, commit);
 237
 238                 if (c_ent->maximal) {
 239                         if (!c_ent->selected) {
 240                                 bitmap_set(c_ent->commit_mask, num_maximal);
 241                                 num_maximal++;
 242                         }
 243
 244                         ALLOC_GROW(bb->commits, bb->commits_nr + 1, bb->commits_alloc);
 245                         bb->commits[bb->commits_nr++] = commit;
 246                 }
 247
 248                 for (p = commit->parents; p; p = p->next) {
 249                         struct bb_commit *p_ent = bb_data_at(&bb->data, p->item);
 250                         int c_not_p, p_not_c;
 251
 252                         if (!p_ent->commit_mask) {
 253                                 p_ent->commit_mask = bitmap_new();
 254                                 c_not_p = 1;
 255                                 p_not_c = 0;
 256                         } else {
 257                                 c_not_p = bitmap_is_subset(c_ent->commit_mask, p_ent->commit_mask);
 258                                 p_not_c = bitmap_is_subset(p_ent->commit_mask, c_ent->commit_mask);
 259                         }
 260
 261                         if (!c_not_p)
 262                                 continue;
 263
 264                         bitmap_or(p_ent->commit_mask, c_ent->commit_mask);
 265
 266                         if (p_not_c)
 267                                 p_ent->maximal = 1;
 268                         else {
 269                                 p_ent->maximal = 0;
 270                                 free_commit_list(p_ent->reverse_edges);
 271                                 p_ent->reverse_edges = NULL;
 272                         }
 273
 274                         if (c_ent->maximal) {
 275                                 commit_list_insert(commit, &p_ent->reverse_edges);
 276                         } else {
 277                                 struct commit_list *cc = c_ent->reverse_edges;
 278
 279                                 for (; cc; cc = cc->next) {
 280                                         if (!commit_list_contains(cc->item, p_ent->reverse_edges))
 281                                                 commit_list_insert(cc->item, &p_ent->reverse_edges);
 282                                 }
 283                         }
 284                 }
 285
 286                 bitmap_free(c_ent->commit_mask);
 287                 c_ent->commit_mask = NULL;
 288         }
 289
 290         trace2_data_intmax("pack-bitmap-write", the_repository,
 291                            "num_selected_commits", writer->selected_nr);
 292         trace2_data_intmax("pack-bitmap-write", the_repository,
 293                            "num_maximal_commits", num_maximal);
 294 }
 295
 296 static void bitmap_builder_clear(struct bitmap_builder *bb)
 297 {
 298         clear_bb_data(&bb->data);
 299         free(bb->commits);
 300         bb->commits_nr = bb->commits_alloc = 0;
 301 }
 302
 303 static void fill_bitmap_tree(struct bitmap *bitmap,
 304                              struct tree *tree)
 305 {
 306         uint32_t pos;
 307         struct tree_desc desc;
 308         struct name_entry entry;
 309
 310         /*
 311          * If our bit is already set, then there is nothing to do. Both this
 312          * tree and all of its children will be set.
 313          */
 314         pos = find_object_pos(&tree->object.oid);
 315         if (bitmap_get(bitmap, pos))
 316                 return;
 317         bitmap_set(bitmap, pos);
 318
 319         if (parse_tree(tree) < 0)
 320                 die("unable to load tree object %s",
 321                     oid_to_hex(&tree->object.oid));
 322         init_tree_desc(&desc, tree->buffer, tree->size);
 323
 324         while (tree_entry(&desc, &entry)) {
 325                 switch (object_type(entry.mode)) {
 326                 case OBJ_TREE:
 327                         fill_bitmap_tree(bitmap,
 328                                          lookup_tree(the_repository, &entry.oid));
 329                         break;
 330                 case OBJ_BLOB:
 331                         bitmap_set(bitmap, find_object_pos(&entry.oid));
 332                         break;
 333                 default:
 334                         /* Gitlink, etc; not reachable */
 335                         break;
 336                 }
 337         }
 338
 339         free_tree_buffer(tree);
 340 }
 341
 342 static void fill_bitmap_commit(struct bb_commit *ent,
 343                                struct commit *commit,
 344                                struct prio_queue *queue)
 345 {
 346         if (!ent->bitmap)
 347                 ent->bitmap = bitmap_new();
 348
 349         bitmap_set(ent->bitmap, find_object_pos(&commit->object.oid));
 350         prio_queue_put(queue, commit);
 351
 352         while (queue->nr) {
 353                 struct commit_list *p;
 354                 struct commit *c = prio_queue_get(queue);
 355
 356                 bitmap_set(ent->bitmap, find_object_pos(&c->object.oid));
 357                 fill_bitmap_tree(ent->bitmap, get_commit_tree(c));
 358
 359                 for (p = c->parents; p; p = p->next) {
 360                         int pos = find_object_pos(&p->item->object.oid);
 361                         if (!bitmap_get(ent->bitmap, pos)) {
 362                                 bitmap_set(ent->bitmap, pos);
 363                                 prio_queue_put(queue, p->item);
 364                         }
 365                 }
 366         }
 367 }
 368
 369 static void store_selected(struct bb_commit *ent, struct commit *commit)
 370 {
 371         struct bitmapped_commit *stored = &writer.selected[ent->idx];
 372         khiter_t hash_pos;
 373         int hash_ret;
 374
 375         /*
 376          * the "reuse bitmaps" phase may have stored something here, but
 377          * our new algorithm doesn't use it. Drop it.
 378          */
 379         if (stored->bitmap)
 380                 ewah_free(stored->bitmap);
 381
 382         stored->bitmap = bitmap_to_ewah(ent->bitmap);
 383
 384         hash_pos = kh_put_oid_map(writer.bitmaps, commit->object.oid, &hash_ret);
 385         if (hash_ret == 0)
 386                 die("Duplicate entry when writing index: %s",
 387                     oid_to_hex(&commit->object.oid));
 388         kh_value(writer.bitmaps, hash_pos) = stored;
 389 }
 390
 391 void bitmap_writer_build(struct packing_data *to_pack)
 392 {
 393         struct bitmap_builder bb;
 394         size_t i;
 395         int nr_stored = 0; /* for progress */
 396         struct prio_queue queue = { compare_commits_by_gen_then_commit_date };
 397
 398         writer.bitmaps = kh_init_oid_map();
 399         writer.to_pack = to_pack;
 400
 401         if (writer.show_progress)
 402                 writer.progress = start_progress("Building bitmaps", writer.selected_nr);
 403         trace2_region_enter("pack-bitmap-write", "building_bitmaps_total",
 404                             the_repository);
 405
 406         bitmap_builder_init(&bb, &writer);
 407         for (i = bb.commits_nr; i > 0; i--) {
 408                 struct commit *commit = bb.commits[i-1];
 409                 struct bb_commit *ent = bb_data_at(&bb.data, commit);
 410                 struct commit *child;
 411                 int reused = 0;
 412
 413                 fill_bitmap_commit(ent, commit, &queue);
 414
 415                 if (ent->selected) {
 416                         store_selected(ent, commit);
 417                         nr_stored++;
 418                         display_progress(writer.progress, nr_stored);
 419                 }
 420
 421                 while ((child = pop_commit(&ent->reverse_edges))) {
 422                         struct bb_commit *child_ent =
 423                                 bb_data_at(&bb.data, child);
 424
 425                         if (child_ent->bitmap)
 426                                 bitmap_or(child_ent->bitmap, ent->bitmap);
 427                         else if (reused)
 428                                 child_ent->bitmap = bitmap_dup(ent->bitmap);
 429                         else {
 430                                 child_ent->bitmap = ent->bitmap;
 431                                 reused = 1;
 432                         }
 433                 }
 434                 if (!reused)
 435                         bitmap_free(ent->bitmap);
 436                 ent->bitmap = NULL;
 437         }
 438         clear_prio_queue(&queue);
 439         bitmap_builder_clear(&bb);
 440
 441         trace2_region_leave("pack-bitmap-write", "building_bitmaps_total",
 442                             the_repository);
 443
 444         stop_progress(&writer.progress);
 445
 446         compute_xor_offsets();
 447 }
 448
 449 /**
 450  * Select the commits that will be bitmapped
 451  */
 452 static inline unsigned int next_commit_index(unsigned int idx)
 453 {
 454         static const unsigned int MIN_COMMITS = 100;
 455         static const unsigned int MAX_COMMITS = 5000;
 456
 457         static const unsigned int MUST_REGION = 100;
 458         static const unsigned int MIN_REGION = 20000;
 459
 460         unsigned int offset, next;
 461
 462         if (idx <= MUST_REGION)
 463                 return 0;
 464
 465         if (idx <= MIN_REGION) {
 466                 offset = idx - MUST_REGION;
 467                 return (offset < MIN_COMMITS) ? offset : MIN_COMMITS;
 468         }
 469
 470         offset = idx - MIN_REGION;
 471         next = (offset < MAX_COMMITS) ? offset : MAX_COMMITS;
 472
 473         return (next > MIN_COMMITS) ? next : MIN_COMMITS;
 474 }
 475
 476 static int date_compare(const void *_a, const void *_b)
 477 {
 478         struct commit *a = *(struct commit **)_a;
 479         struct commit *b = *(struct commit **)_b;
 480         return (long)b->date - (long)a->date;
 481 }
 482
 483 void bitmap_writer_reuse_bitmaps(struct packing_data *to_pack)
 484 {
 485         struct bitmap_index *bitmap_git;
 486         if (!(bitmap_git = prepare_bitmap_git(to_pack->repo)))
 487                 return;
 488
 489         writer.reused = kh_init_oid_map();
 490         rebuild_existing_bitmaps(bitmap_git, to_pack, writer.reused,
 491                                  writer.show_progress);
 492         /*
 493          * NEEDSWORK: rebuild_existing_bitmaps() makes writer.reused reference
 494          * some bitmaps in bitmap_git, so we can't free the latter.
 495          */
 496 }
 497
 498 static struct ewah_bitmap *find_reused_bitmap(const struct object_id *oid)
 499 {
 500         khiter_t hash_pos;
 501
 502         if (!writer.reused)
 503                 return NULL;
 504
 505         hash_pos = kh_get_oid_map(writer.reused, *oid);
 506         if (hash_pos >= kh_end(writer.reused))
 507                 return NULL;
 508
 509         return kh_value(writer.reused, hash_pos);
 510 }
 511
 512 void bitmap_writer_select_commits(struct commit **indexed_commits,
 513                                   unsigned int indexed_commits_nr,
 514                                   int max_bitmaps)
 515 {
 516         unsigned int i = 0, j, next;
 517
 518         QSORT(indexed_commits, indexed_commits_nr, date_compare);
 519
 520         if (writer.show_progress)
 521                 writer.progress = start_progress("Selecting bitmap commits", 0);
 522
 523         if (indexed_commits_nr < 100) {
 524                 for (i = 0; i < indexed_commits_nr; ++i)
 525                         push_bitmapped_commit(indexed_commits[i], NULL);
 526                 return;
 527         }
 528
 529         for (;;) {
 530                 struct ewah_bitmap *reused_bitmap = NULL;
 531                 struct commit *chosen = NULL;
 532
 533                 next = next_commit_index(i);
 534
 535                 if (i + next >= indexed_commits_nr)
 536                         break;
 537
 538                 if (max_bitmaps > 0 && writer.selected_nr >= max_bitmaps) {
 539                         writer.selected_nr = max_bitmaps;
 540                         break;
 541                 }
 542
 543                 if (next == 0) {
 544                         chosen = indexed_commits[i];
 545                         reused_bitmap = find_reused_bitmap(&chosen->object.oid);
 546                 } else {
 547                         chosen = indexed_commits[i + next];
 548
 549                         for (j = 0; j <= next; ++j) {
 550                                 struct commit *cm = indexed_commits[i + j];
 551
 552                                 reused_bitmap = find_reused_bitmap(&cm->object.oid);
 553                                 if (reused_bitmap || (cm->object.flags & NEEDS_BITMAP) != 0) {
 554                                         chosen = cm;
 555                                         break;
 556                                 }
 557
 558                                 if (cm->parents && cm->parents->next)
 559                                         chosen = cm;
 560                         }
 561                 }
 562
 563                 push_bitmapped_commit(chosen, reused_bitmap);
 564
 565                 i += next + 1;
 566                 display_progress(writer.progress, i);
 567         }
 568
 569         stop_progress(&writer.progress);
 570 }
 571
 572
 573 static int hashwrite_ewah_helper(void *f, const void *buf, size_t len)
 574 {
 575         /* hashwrite will die on error */
 576         hashwrite(f, buf, len);
 577         return len;
 578 }
 579
 580 /**
 581  * Write the bitmap index to disk
 582  */
 583 static inline void dump_bitmap(struct hashfile *f, struct ewah_bitmap *bitmap)
 584 {
 585         if (ewah_serialize_to(bitmap, hashwrite_ewah_helper, f) < 0)
 586                 die("Failed to write bitmap index");
 587 }
 588
 589 static const unsigned char *sha1_access(size_t pos, void *table)
 590 {
 591         struct pack_idx_entry **index = table;
 592         return index[pos]->oid.hash;
 593 }
 594
 595 static void write_selected_commits_v1(struct hashfile *f,
 596                                       struct pack_idx_entry **index,
 597                                       uint32_t index_nr)
 598 {
 599         int i;
 600
 601         for (i = 0; i < writer.selected_nr; ++i) {
 602                 struct bitmapped_commit *stored = &writer.selected[i];
 603
 604                 int commit_pos =
 605                         sha1_pos(stored->commit->object.oid.hash, index, index_nr, sha1_access);
 606
 607                 if (commit_pos < 0)
 608                         BUG("trying to write commit not in index");
 609
 610                 hashwrite_be32(f, commit_pos);
 611                 hashwrite_u8(f, stored->xor_offset);
 612                 hashwrite_u8(f, stored->flags);
 613
 614                 dump_bitmap(f, stored->write_as);
 615         }
 616 }
 617
 618 static void write_hash_cache(struct hashfile *f,
 619                              struct pack_idx_entry **index,
 620                              uint32_t index_nr)
 621 {
 622         uint32_t i;
 623
 624         for (i = 0; i < index_nr; ++i) {
 625                 struct object_entry *entry = (struct object_entry *)index[i];
 626                 hashwrite_be32(f, entry->hash);
 627         }
 628 }
 629
 630 void bitmap_writer_set_checksum(unsigned char *sha1)
 631 {
 632         hashcpy(writer.pack_checksum, sha1);
 633 }
 634
 635 void bitmap_writer_finish(struct pack_idx_entry **index,
 636                           uint32_t index_nr,
 637                           const char *filename,
 638                           uint16_t options)
 639 {
 640         static uint16_t default_version = 1;
 641         static uint16_t flags = BITMAP_OPT_FULL_DAG;
 642         struct strbuf tmp_file = STRBUF_INIT;
 643         struct hashfile *f;
 644
 645         struct bitmap_disk_header header;
 646
 647         int fd = odb_mkstemp(&tmp_file, "pack/tmp_bitmap_XXXXXX");
 648
 649         f = hashfd(fd, tmp_file.buf);
 650
 651         memcpy(header.magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE));
 652         header.version = htons(default_version);
 653         header.options = htons(flags | options);
 654         header.entry_count = htonl(writer.selected_nr);
 655         hashcpy(header.checksum, writer.pack_checksum);
 656
 657         hashwrite(f, &header, sizeof(header) - GIT_MAX_RAWSZ + the_hash_algo->rawsz);
 658         dump_bitmap(f, writer.commits);
 659         dump_bitmap(f, writer.trees);
 660         dump_bitmap(f, writer.blobs);
 661         dump_bitmap(f, writer.tags);
 662         write_selected_commits_v1(f, index, index_nr);
 663
 664         if (options & BITMAP_OPT_HASH_CACHE)
 665                 write_hash_cache(f, index, index_nr);
 666
 667         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC | CSUM_CLOSE);
 668
 669         if (adjust_shared_perm(tmp_file.buf))
 670                 die_errno("unable to make temporary bitmap file readable");
 671
 672         if (rename(tmp_file.buf, filename))
 673                 die_errno("unable to rename temporary bitmap file to '%s'", filename);
 674
 675         strbuf_release(&tmp_file);
 676 }