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 *children;
 183         struct bitmap *bitmap;
 184         unsigned selected:1;
 185         unsigned idx; /* within selected array */
 186 };
 187
 188 define_commit_slab(bb_data, struct bb_commit);
 189
 190 struct bitmap_builder {
 191         struct bb_data data;
 192         struct commit **commits;
 193         size_t commits_nr, commits_alloc;
 194 };
 195
 196 static void bitmap_builder_init(struct bitmap_builder *bb,
 197                                 struct bitmap_writer *writer)
 198 {
 199         struct rev_info revs;
 200         struct commit *commit;
 201         unsigned int i;
 202
 203         memset(bb, 0, sizeof(*bb));
 204         init_bb_data(&bb->data);
 205
 206         reset_revision_walk();
 207         repo_init_revisions(writer->to_pack->repo, &revs, NULL);
 208         revs.topo_order = 1;
 209
 210         for (i = 0; i < writer->selected_nr; i++) {
 211                 struct commit *c = writer->selected[i].commit;
 212                 struct bb_commit *ent = bb_data_at(&bb->data, c);
 213                 ent->selected = 1;
 214                 ent->idx = i;
 215                 add_pending_object(&revs, &c->object, "");
 216         }
 217
 218         if (prepare_revision_walk(&revs))
 219                 die("revision walk setup failed");
 220
 221         while ((commit = get_revision(&revs))) {
 222                 struct commit_list *p;
 223
 224                 parse_commit_or_die(commit);
 225
 226                 ALLOC_GROW(bb->commits, bb->commits_nr + 1, bb->commits_alloc);
 227                 bb->commits[bb->commits_nr++] = commit;
 228
 229                 for (p = commit->parents; p; p = p->next) {
 230                         struct bb_commit *ent = bb_data_at(&bb->data, p->item);
 231                         commit_list_insert(commit, &ent->children);
 232                 }
 233         }
 234 }
 235
 236 static void bitmap_builder_clear(struct bitmap_builder *bb)
 237 {
 238         clear_bb_data(&bb->data);
 239         free(bb->commits);
 240         bb->commits_nr = bb->commits_alloc = 0;
 241 }
 242
 243 static void fill_bitmap_tree(struct bitmap *bitmap,
 244                              struct tree *tree)
 245 {
 246         uint32_t pos;
 247         struct tree_desc desc;
 248         struct name_entry entry;
 249
 250         /*
 251          * If our bit is already set, then there is nothing to do. Both this
 252          * tree and all of its children will be set.
 253          */
 254         pos = find_object_pos(&tree->object.oid);
 255         if (bitmap_get(bitmap, pos))
 256                 return;
 257         bitmap_set(bitmap, pos);
 258
 259         if (parse_tree(tree) < 0)
 260                 die("unable to load tree object %s",
 261                     oid_to_hex(&tree->object.oid));
 262         init_tree_desc(&desc, tree->buffer, tree->size);
 263
 264         while (tree_entry(&desc, &entry)) {
 265                 switch (object_type(entry.mode)) {
 266                 case OBJ_TREE:
 267                         fill_bitmap_tree(bitmap,
 268                                          lookup_tree(the_repository, &entry.oid));
 269                         break;
 270                 case OBJ_BLOB:
 271                         bitmap_set(bitmap, find_object_pos(&entry.oid));
 272                         break;
 273                 default:
 274                         /* Gitlink, etc; not reachable */
 275                         break;
 276                 }
 277         }
 278
 279         free_tree_buffer(tree);
 280 }
 281
 282 static void fill_bitmap_commit(struct bb_commit *ent,
 283                                struct commit *commit,
 284                                struct prio_queue *queue)
 285 {
 286         if (!ent->bitmap)
 287                 ent->bitmap = bitmap_new();
 288
 289         bitmap_set(ent->bitmap, find_object_pos(&commit->object.oid));
 290         prio_queue_put(queue, commit);
 291
 292         while (queue->nr) {
 293                 struct commit_list *p;
 294                 struct commit *c = prio_queue_get(queue);
 295
 296                 bitmap_set(ent->bitmap, find_object_pos(&c->object.oid));
 297                 fill_bitmap_tree(ent->bitmap, get_commit_tree(c));
 298
 299                 for (p = c->parents; p; p = p->next) {
 300                         int pos = find_object_pos(&p->item->object.oid);
 301                         if (!bitmap_get(ent->bitmap, pos)) {
 302                                 bitmap_set(ent->bitmap, pos);
 303                                 prio_queue_put(queue, p->item);
 304                         }
 305                 }
 306         }
 307 }
 308
 309 static void store_selected(struct bb_commit *ent, struct commit *commit)
 310 {
 311         struct bitmapped_commit *stored = &writer.selected[ent->idx];
 312         khiter_t hash_pos;
 313         int hash_ret;
 314
 315         /*
 316          * the "reuse bitmaps" phase may have stored something here, but
 317          * our new algorithm doesn't use it. Drop it.
 318          */
 319         if (stored->bitmap)
 320                 ewah_free(stored->bitmap);
 321
 322         stored->bitmap = bitmap_to_ewah(ent->bitmap);
 323
 324         hash_pos = kh_put_oid_map(writer.bitmaps, commit->object.oid, &hash_ret);
 325         if (hash_ret == 0)
 326                 die("Duplicate entry when writing index: %s",
 327                     oid_to_hex(&commit->object.oid));
 328         kh_value(writer.bitmaps, hash_pos) = stored;
 329 }
 330
 331 void bitmap_writer_build(struct packing_data *to_pack)
 332 {
 333         struct bitmap_builder bb;
 334         size_t i;
 335         int nr_stored = 0; /* for progress */
 336         struct prio_queue queue = { compare_commits_by_gen_then_commit_date };
 337
 338         writer.bitmaps = kh_init_oid_map();
 339         writer.to_pack = to_pack;
 340
 341         if (writer.show_progress)
 342                 writer.progress = start_progress("Building bitmaps", writer.selected_nr);
 343         trace2_region_enter("pack-bitmap-write", "building_bitmaps_total",
 344                             the_repository);
 345
 346         bitmap_builder_init(&bb, &writer);
 347         for (i = bb.commits_nr; i > 0; i--) {
 348                 struct commit *commit = bb.commits[i-1];
 349                 struct bb_commit *ent = bb_data_at(&bb.data, commit);
 350                 struct commit *child;
 351                 int reused = 0;
 352
 353                 fill_bitmap_commit(ent, commit, &queue);
 354
 355                 if (ent->selected) {
 356                         store_selected(ent, commit);
 357                         nr_stored++;
 358                         display_progress(writer.progress, nr_stored);
 359                 }
 360
 361                 while ((child = pop_commit(&ent->children))) {
 362                         struct bb_commit *child_ent =
 363                                 bb_data_at(&bb.data, child);
 364
 365                         if (child_ent->bitmap)
 366                                 bitmap_or(child_ent->bitmap, ent->bitmap);
 367                         else if (reused)
 368                                 child_ent->bitmap = bitmap_dup(ent->bitmap);
 369                         else {
 370                                 child_ent->bitmap = ent->bitmap;
 371                                 reused = 1;
 372                         }
 373                 }
 374                 if (!reused)
 375                         bitmap_free(ent->bitmap);
 376                 ent->bitmap = NULL;
 377         }
 378         clear_prio_queue(&queue);
 379         bitmap_builder_clear(&bb);
 380
 381         trace2_region_leave("pack-bitmap-write", "building_bitmaps_total",
 382                             the_repository);
 383
 384         stop_progress(&writer.progress);
 385
 386         compute_xor_offsets();
 387 }
 388
 389 /**
 390  * Select the commits that will be bitmapped
 391  */
 392 static inline unsigned int next_commit_index(unsigned int idx)
 393 {
 394         static const unsigned int MIN_COMMITS = 100;
 395         static const unsigned int MAX_COMMITS = 5000;
 396
 397         static const unsigned int MUST_REGION = 100;
 398         static const unsigned int MIN_REGION = 20000;
 399
 400         unsigned int offset, next;
 401
 402         if (idx <= MUST_REGION)
 403                 return 0;
 404
 405         if (idx <= MIN_REGION) {
 406                 offset = idx - MUST_REGION;
 407                 return (offset < MIN_COMMITS) ? offset : MIN_COMMITS;
 408         }
 409
 410         offset = idx - MIN_REGION;
 411         next = (offset < MAX_COMMITS) ? offset : MAX_COMMITS;
 412
 413         return (next > MIN_COMMITS) ? next : MIN_COMMITS;
 414 }
 415
 416 static int date_compare(const void *_a, const void *_b)
 417 {
 418         struct commit *a = *(struct commit **)_a;
 419         struct commit *b = *(struct commit **)_b;
 420         return (long)b->date - (long)a->date;
 421 }
 422
 423 void bitmap_writer_reuse_bitmaps(struct packing_data *to_pack)
 424 {
 425         struct bitmap_index *bitmap_git;
 426         if (!(bitmap_git = prepare_bitmap_git(to_pack->repo)))
 427                 return;
 428
 429         writer.reused = kh_init_oid_map();
 430         rebuild_existing_bitmaps(bitmap_git, to_pack, writer.reused,
 431                                  writer.show_progress);
 432         /*
 433          * NEEDSWORK: rebuild_existing_bitmaps() makes writer.reused reference
 434          * some bitmaps in bitmap_git, so we can't free the latter.
 435          */
 436 }
 437
 438 static struct ewah_bitmap *find_reused_bitmap(const struct object_id *oid)
 439 {
 440         khiter_t hash_pos;
 441
 442         if (!writer.reused)
 443                 return NULL;
 444
 445         hash_pos = kh_get_oid_map(writer.reused, *oid);
 446         if (hash_pos >= kh_end(writer.reused))
 447                 return NULL;
 448
 449         return kh_value(writer.reused, hash_pos);
 450 }
 451
 452 void bitmap_writer_select_commits(struct commit **indexed_commits,
 453                                   unsigned int indexed_commits_nr,
 454                                   int max_bitmaps)
 455 {
 456         unsigned int i = 0, j, next;
 457
 458         QSORT(indexed_commits, indexed_commits_nr, date_compare);
 459
 460         if (writer.show_progress)
 461                 writer.progress = start_progress("Selecting bitmap commits", 0);
 462
 463         if (indexed_commits_nr < 100) {
 464                 for (i = 0; i < indexed_commits_nr; ++i)
 465                         push_bitmapped_commit(indexed_commits[i], NULL);
 466                 return;
 467         }
 468
 469         for (;;) {
 470                 struct ewah_bitmap *reused_bitmap = NULL;
 471                 struct commit *chosen = NULL;
 472
 473                 next = next_commit_index(i);
 474
 475                 if (i + next >= indexed_commits_nr)
 476                         break;
 477
 478                 if (max_bitmaps > 0 && writer.selected_nr >= max_bitmaps) {
 479                         writer.selected_nr = max_bitmaps;
 480                         break;
 481                 }
 482
 483                 if (next == 0) {
 484                         chosen = indexed_commits[i];
 485                         reused_bitmap = find_reused_bitmap(&chosen->object.oid);
 486                 } else {
 487                         chosen = indexed_commits[i + next];
 488
 489                         for (j = 0; j <= next; ++j) {
 490                                 struct commit *cm = indexed_commits[i + j];
 491
 492                                 reused_bitmap = find_reused_bitmap(&cm->object.oid);
 493                                 if (reused_bitmap || (cm->object.flags & NEEDS_BITMAP) != 0) {
 494                                         chosen = cm;
 495                                         break;
 496                                 }
 497
 498                                 if (cm->parents && cm->parents->next)
 499                                         chosen = cm;
 500                         }
 501                 }
 502
 503                 push_bitmapped_commit(chosen, reused_bitmap);
 504
 505                 i += next + 1;
 506                 display_progress(writer.progress, i);
 507         }
 508
 509         stop_progress(&writer.progress);
 510 }
 511
 512
 513 static int hashwrite_ewah_helper(void *f, const void *buf, size_t len)
 514 {
 515         /* hashwrite will die on error */
 516         hashwrite(f, buf, len);
 517         return len;
 518 }
 519
 520 /**
 521  * Write the bitmap index to disk
 522  */
 523 static inline void dump_bitmap(struct hashfile *f, struct ewah_bitmap *bitmap)
 524 {
 525         if (ewah_serialize_to(bitmap, hashwrite_ewah_helper, f) < 0)
 526                 die("Failed to write bitmap index");
 527 }
 528
 529 static const unsigned char *sha1_access(size_t pos, void *table)
 530 {
 531         struct pack_idx_entry **index = table;
 532         return index[pos]->oid.hash;
 533 }
 534
 535 static void write_selected_commits_v1(struct hashfile *f,
 536                                       struct pack_idx_entry **index,
 537                                       uint32_t index_nr)
 538 {
 539         int i;
 540
 541         for (i = 0; i < writer.selected_nr; ++i) {
 542                 struct bitmapped_commit *stored = &writer.selected[i];
 543
 544                 int commit_pos =
 545                         sha1_pos(stored->commit->object.oid.hash, index, index_nr, sha1_access);
 546
 547                 if (commit_pos < 0)
 548                         BUG("trying to write commit not in index");
 549
 550                 hashwrite_be32(f, commit_pos);
 551                 hashwrite_u8(f, stored->xor_offset);
 552                 hashwrite_u8(f, stored->flags);
 553
 554                 dump_bitmap(f, stored->write_as);
 555         }
 556 }
 557
 558 static void write_hash_cache(struct hashfile *f,
 559                              struct pack_idx_entry **index,
 560                              uint32_t index_nr)
 561 {
 562         uint32_t i;
 563
 564         for (i = 0; i < index_nr; ++i) {
 565                 struct object_entry *entry = (struct object_entry *)index[i];
 566                 hashwrite_be32(f, entry->hash);
 567         }
 568 }
 569
 570 void bitmap_writer_set_checksum(unsigned char *sha1)
 571 {
 572         hashcpy(writer.pack_checksum, sha1);
 573 }
 574
 575 void bitmap_writer_finish(struct pack_idx_entry **index,
 576                           uint32_t index_nr,
 577                           const char *filename,
 578                           uint16_t options)
 579 {
 580         static uint16_t default_version = 1;
 581         static uint16_t flags = BITMAP_OPT_FULL_DAG;
 582         struct strbuf tmp_file = STRBUF_INIT;
 583         struct hashfile *f;
 584
 585         struct bitmap_disk_header header;
 586
 587         int fd = odb_mkstemp(&tmp_file, "pack/tmp_bitmap_XXXXXX");
 588
 589         f = hashfd(fd, tmp_file.buf);
 590
 591         memcpy(header.magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE));
 592         header.version = htons(default_version);
 593         header.options = htons(flags | options);
 594         header.entry_count = htonl(writer.selected_nr);
 595         hashcpy(header.checksum, writer.pack_checksum);
 596
 597         hashwrite(f, &header, sizeof(header) - GIT_MAX_RAWSZ + the_hash_algo->rawsz);
 598         dump_bitmap(f, writer.commits);
 599         dump_bitmap(f, writer.trees);
 600         dump_bitmap(f, writer.blobs);
 601         dump_bitmap(f, writer.tags);
 602         write_selected_commits_v1(f, index, index_nr);
 603
 604         if (options & BITMAP_OPT_HASH_CACHE)
 605                 write_hash_cache(f, index, index_nr);
 606
 607         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC | CSUM_CLOSE);
 608
 609         if (adjust_shared_perm(tmp_file.buf))
 610                 die_errno("unable to make temporary bitmap file readable");
 611
 612         if (rename(tmp_file.buf, filename))
 613                 die_errno("unable to rename temporary bitmap file to '%s'", filename);
 614
 615         strbuf_release(&tmp_file);
 616 }