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

   1 #include "cache.h"
   2 #include "commit.h"
   3 #include "tag.h"
   4 #include "diff.h"
   5 #include "revision.h"
   6 #include "progress.h"
   7 #include "list-objects.h"
   8 #include "pack.h"
   9 #include "pack-bitmap.h"
  10 #include "pack-revindex.h"
  11 #include "pack-objects.h"
  12 #include "packfile.h"
  13 #include "repository.h"
  14 #include "object-store.h"
  15
  16 /*
  17  * An entry on the bitmap index, representing the bitmap for a given
  18  * commit.
  19  */
  20 struct stored_bitmap {
  21         struct object_id oid;
  22         struct ewah_bitmap *root;
  23         struct stored_bitmap *xor;
  24         int flags;
  25 };
  26
  27 /*
  28  * The active bitmap index for a repository. By design, repositories only have
  29  * a single bitmap index available (the index for the biggest packfile in
  30  * the repository), since bitmap indexes need full closure.
  31  *
  32  * If there is more than one bitmap index available (e.g. because of alternates),
  33  * the active bitmap index is the largest one.
  34  */
  35 struct bitmap_index {
  36         /* Packfile to which this bitmap index belongs to */
  37         struct packed_git *pack;
  38
  39         /*
  40          * Mark the first `reuse_objects` in the packfile as reused:
  41          * they will be sent as-is without using them for repacking
  42          * calculations
  43          */
  44         uint32_t reuse_objects;
  45
  46         /* mmapped buffer of the whole bitmap index */
  47         unsigned char *map;
  48         size_t map_size; /* size of the mmaped buffer */
  49         size_t map_pos; /* current position when loading the index */
  50
  51         /*
  52          * Type indexes.
  53          *
  54          * Each bitmap marks which objects in the packfile  are of the given
  55          * type. This provides type information when yielding the objects from
  56          * the packfile during a walk, which allows for better delta bases.
  57          */
  58         struct ewah_bitmap *commits;
  59         struct ewah_bitmap *trees;
  60         struct ewah_bitmap *blobs;
  61         struct ewah_bitmap *tags;
  62
  63         /* Map from object ID -> `stored_bitmap` for all the bitmapped commits */
  64         kh_oid_map_t *bitmaps;
  65
  66         /* Number of bitmapped commits */
  67         uint32_t entry_count;
  68
  69         /* If not NULL, this is a name-hash cache pointing into map. */
  70         uint32_t *hashes;
  71
  72         /*
  73          * Extended index.
  74          *
  75          * When trying to perform bitmap operations with objects that are not
  76          * packed in `pack`, these objects are added to this "fake index" and
  77          * are assumed to appear at the end of the packfile for all operations
  78          */
  79         struct eindex {
  80                 struct object **objects;
  81                 uint32_t *hashes;
  82                 uint32_t count, alloc;
  83                 kh_oid_pos_t *positions;
  84         } ext_index;
  85
  86         /* Bitmap result of the last performed walk */
  87         struct bitmap *result;
  88
  89         /* "have" bitmap from the last performed walk */
  90         struct bitmap *haves;
  91
  92         /* Version of the bitmap index */
  93         unsigned int version;
  94 };
  95
  96 static struct ewah_bitmap *lookup_stored_bitmap(struct stored_bitmap *st)
  97 {
  98         struct ewah_bitmap *parent;
  99         struct ewah_bitmap *composed;
 100
 101         if (st->xor == NULL)
 102                 return st->root;
 103
 104         composed = ewah_pool_new();
 105         parent = lookup_stored_bitmap(st->xor);
 106         ewah_xor(st->root, parent, composed);
 107
 108         ewah_pool_free(st->root);
 109         st->root = composed;
 110         st->xor = NULL;
 111
 112         return composed;
 113 }
 114
 115 /*
 116  * Read a bitmap from the current read position on the mmaped
 117  * index, and increase the read position accordingly
 118  */
 119 static struct ewah_bitmap *read_bitmap_1(struct bitmap_index *index)
 120 {
 121         struct ewah_bitmap *b = ewah_pool_new();
 122
 123         ssize_t bitmap_size = ewah_read_mmap(b,
 124                 index->map + index->map_pos,
 125                 index->map_size - index->map_pos);
 126
 127         if (bitmap_size < 0) {
 128                 error("Failed to load bitmap index (corrupted?)");
 129                 ewah_pool_free(b);
 130                 return NULL;
 131         }
 132
 133         index->map_pos += bitmap_size;
 134         return b;
 135 }
 136
 137 static int load_bitmap_header(struct bitmap_index *index)
 138 {
 139         struct bitmap_disk_header *header = (void *)index->map;
 140
 141         if (index->map_size < sizeof(*header) + the_hash_algo->rawsz)
 142                 return error("Corrupted bitmap index (missing header data)");
 143
 144         if (memcmp(header->magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE)) != 0)
 145                 return error("Corrupted bitmap index file (wrong header)");
 146
 147         index->version = ntohs(header->version);
 148         if (index->version != 1)
 149                 return error("Unsupported version for bitmap index file (%d)", index->version);
 150
 151         /* Parse known bitmap format options */
 152         {
 153                 uint32_t flags = ntohs(header->options);
 154
 155                 if ((flags & BITMAP_OPT_FULL_DAG) == 0)
 156                         return error("Unsupported options for bitmap index file "
 157                                 "(Git requires BITMAP_OPT_FULL_DAG)");
 158
 159                 if (flags & BITMAP_OPT_HASH_CACHE) {
 160                         unsigned char *end = index->map + index->map_size - the_hash_algo->rawsz;
 161                         index->hashes = ((uint32_t *)end) - index->pack->num_objects;
 162                 }
 163         }
 164
 165         index->entry_count = ntohl(header->entry_count);
 166         index->map_pos += sizeof(*header) - GIT_MAX_RAWSZ + the_hash_algo->rawsz;
 167         return 0;
 168 }
 169
 170 static struct stored_bitmap *store_bitmap(struct bitmap_index *index,
 171                                           struct ewah_bitmap *root,
 172                                           const unsigned char *hash,
 173                                           struct stored_bitmap *xor_with,
 174                                           int flags)
 175 {
 176         struct stored_bitmap *stored;
 177         khiter_t hash_pos;
 178         int ret;
 179
 180         stored = xmalloc(sizeof(struct stored_bitmap));
 181         stored->root = root;
 182         stored->xor = xor_with;
 183         stored->flags = flags;
 184         oidread(&stored->oid, hash);
 185
 186         hash_pos = kh_put_oid_map(index->bitmaps, stored->oid, &ret);
 187
 188         /* a 0 return code means the insertion succeeded with no changes,
 189          * because the SHA1 already existed on the map. this is bad, there
 190          * shouldn't be duplicated commits in the index */
 191         if (ret == 0) {
 192                 error("Duplicate entry in bitmap index: %s", hash_to_hex(hash));
 193                 return NULL;
 194         }
 195
 196         kh_value(index->bitmaps, hash_pos) = stored;
 197         return stored;
 198 }
 199
 200 static inline uint32_t read_be32(const unsigned char *buffer, size_t *pos)
 201 {
 202         uint32_t result = get_be32(buffer + *pos);
 203         (*pos) += sizeof(result);
 204         return result;
 205 }
 206
 207 static inline uint8_t read_u8(const unsigned char *buffer, size_t *pos)
 208 {
 209         return buffer[(*pos)++];
 210 }
 211
 212 #define MAX_XOR_OFFSET 160
 213
 214 static int load_bitmap_entries_v1(struct bitmap_index *index)
 215 {
 216         uint32_t i;
 217         struct stored_bitmap *recent_bitmaps[MAX_XOR_OFFSET] = { NULL };
 218
 219         for (i = 0; i < index->entry_count; ++i) {
 220                 int xor_offset, flags;
 221                 struct ewah_bitmap *bitmap = NULL;
 222                 struct stored_bitmap *xor_bitmap = NULL;
 223                 uint32_t commit_idx_pos;
 224                 const unsigned char *sha1;
 225
 226                 commit_idx_pos = read_be32(index->map, &index->map_pos);
 227                 xor_offset = read_u8(index->map, &index->map_pos);
 228                 flags = read_u8(index->map, &index->map_pos);
 229
 230                 sha1 = nth_packed_object_sha1(index->pack, commit_idx_pos);
 231
 232                 bitmap = read_bitmap_1(index);
 233                 if (!bitmap)
 234                         return -1;
 235
 236                 if (xor_offset > MAX_XOR_OFFSET || xor_offset > i)
 237                         return error("Corrupted bitmap pack index");
 238
 239                 if (xor_offset > 0) {
 240                         xor_bitmap = recent_bitmaps[(i - xor_offset) % MAX_XOR_OFFSET];
 241
 242                         if (xor_bitmap == NULL)
 243                                 return error("Invalid XOR offset in bitmap pack index");
 244                 }
 245
 246                 recent_bitmaps[i % MAX_XOR_OFFSET] = store_bitmap(
 247                         index, bitmap, sha1, xor_bitmap, flags);
 248         }
 249
 250         return 0;
 251 }
 252
 253 static char *pack_bitmap_filename(struct packed_git *p)
 254 {
 255         size_t len;
 256
 257         if (!strip_suffix(p->pack_name, ".pack", &len))
 258                 BUG("pack_name does not end in .pack");
 259         return xstrfmt("%.*s.bitmap", (int)len, p->pack_name);
 260 }
 261
 262 static int open_pack_bitmap_1(struct bitmap_index *bitmap_git, struct packed_git *packfile)
 263 {
 264         int fd;
 265         struct stat st;
 266         char *idx_name;
 267
 268         if (open_pack_index(packfile))
 269                 return -1;
 270
 271         idx_name = pack_bitmap_filename(packfile);
 272         fd = git_open(idx_name);
 273         free(idx_name);
 274
 275         if (fd < 0)
 276                 return -1;
 277
 278         if (fstat(fd, &st)) {
 279                 close(fd);
 280                 return -1;
 281         }
 282
 283         if (bitmap_git->pack) {
 284                 warning("ignoring extra bitmap file: %s", packfile->pack_name);
 285                 close(fd);
 286                 return -1;
 287         }
 288
 289         bitmap_git->pack = packfile;
 290         bitmap_git->map_size = xsize_t(st.st_size);
 291         bitmap_git->map = xmmap(NULL, bitmap_git->map_size, PROT_READ, MAP_PRIVATE, fd, 0);
 292         bitmap_git->map_pos = 0;
 293         close(fd);
 294
 295         if (load_bitmap_header(bitmap_git) < 0) {
 296                 munmap(bitmap_git->map, bitmap_git->map_size);
 297                 bitmap_git->map = NULL;
 298                 bitmap_git->map_size = 0;
 299                 return -1;
 300         }
 301
 302         return 0;
 303 }
 304
 305 static int load_pack_bitmap(struct bitmap_index *bitmap_git)
 306 {
 307         assert(bitmap_git->map);
 308
 309         bitmap_git->bitmaps = kh_init_oid_map();
 310         bitmap_git->ext_index.positions = kh_init_oid_pos();
 311         if (load_pack_revindex(bitmap_git->pack))
 312                 goto failed;
 313
 314         if (!(bitmap_git->commits = read_bitmap_1(bitmap_git)) ||
 315                 !(bitmap_git->trees = read_bitmap_1(bitmap_git)) ||
 316                 !(bitmap_git->blobs = read_bitmap_1(bitmap_git)) ||
 317                 !(bitmap_git->tags = read_bitmap_1(bitmap_git)))
 318                 goto failed;
 319
 320         if (load_bitmap_entries_v1(bitmap_git) < 0)
 321                 goto failed;
 322
 323         return 0;
 324
 325 failed:
 326         munmap(bitmap_git->map, bitmap_git->map_size);
 327         bitmap_git->map = NULL;
 328         bitmap_git->map_size = 0;
 329
 330         kh_destroy_oid_map(bitmap_git->bitmaps);
 331         bitmap_git->bitmaps = NULL;
 332
 333         kh_destroy_oid_pos(bitmap_git->ext_index.positions);
 334         bitmap_git->ext_index.positions = NULL;
 335
 336         return -1;
 337 }
 338
 339 static int open_pack_bitmap(struct repository *r,
 340                             struct bitmap_index *bitmap_git)
 341 {
 342         struct packed_git *p;
 343         int ret = -1;
 344
 345         assert(!bitmap_git->map);
 346
 347         for (p = get_all_packs(r); p; p = p->next) {
 348                 if (open_pack_bitmap_1(bitmap_git, p) == 0)
 349                         ret = 0;
 350         }
 351
 352         return ret;
 353 }
 354
 355 struct bitmap_index *prepare_bitmap_git(struct repository *r)
 356 {
 357         struct bitmap_index *bitmap_git = xcalloc(1, sizeof(*bitmap_git));
 358
 359         if (!open_pack_bitmap(r, bitmap_git) && !load_pack_bitmap(bitmap_git))
 360                 return bitmap_git;
 361
 362         free_bitmap_index(bitmap_git);
 363         return NULL;
 364 }
 365
 366 struct include_data {
 367         struct bitmap_index *bitmap_git;
 368         struct bitmap *base;
 369         struct bitmap *seen;
 370 };
 371
 372 static inline int bitmap_position_extended(struct bitmap_index *bitmap_git,
 373                                            const struct object_id *oid)
 374 {
 375         kh_oid_pos_t *positions = bitmap_git->ext_index.positions;
 376         khiter_t pos = kh_get_oid_pos(positions, *oid);
 377
 378         if (pos < kh_end(positions)) {
 379                 int bitmap_pos = kh_value(positions, pos);
 380                 return bitmap_pos + bitmap_git->pack->num_objects;
 381         }
 382
 383         return -1;
 384 }
 385
 386 static inline int bitmap_position_packfile(struct bitmap_index *bitmap_git,
 387                                            const struct object_id *oid)
 388 {
 389         off_t offset = find_pack_entry_one(oid->hash, bitmap_git->pack);
 390         if (!offset)
 391                 return -1;
 392
 393         return find_revindex_position(bitmap_git->pack, offset);
 394 }
 395
 396 static int bitmap_position(struct bitmap_index *bitmap_git,
 397                            const struct object_id *oid)
 398 {
 399         int pos = bitmap_position_packfile(bitmap_git, oid);
 400         return (pos >= 0) ? pos : bitmap_position_extended(bitmap_git, oid);
 401 }
 402
 403 static int ext_index_add_object(struct bitmap_index *bitmap_git,
 404                                 struct object *object, const char *name)
 405 {
 406         struct eindex *eindex = &bitmap_git->ext_index;
 407
 408         khiter_t hash_pos;
 409         int hash_ret;
 410         int bitmap_pos;
 411
 412         hash_pos = kh_put_oid_pos(eindex->positions, object->oid, &hash_ret);
 413         if (hash_ret > 0) {
 414                 if (eindex->count >= eindex->alloc) {
 415                         eindex->alloc = (eindex->alloc + 16) * 3 / 2;
 416                         REALLOC_ARRAY(eindex->objects, eindex->alloc);
 417                         REALLOC_ARRAY(eindex->hashes, eindex->alloc);
 418                 }
 419
 420                 bitmap_pos = eindex->count;
 421                 eindex->objects[eindex->count] = object;
 422                 eindex->hashes[eindex->count] = pack_name_hash(name);
 423                 kh_value(eindex->positions, hash_pos) = bitmap_pos;
 424                 eindex->count++;
 425         } else {
 426                 bitmap_pos = kh_value(eindex->positions, hash_pos);
 427         }
 428
 429         return bitmap_pos + bitmap_git->pack->num_objects;
 430 }
 431
 432 struct bitmap_show_data {
 433         struct bitmap_index *bitmap_git;
 434         struct bitmap *base;
 435 };
 436
 437 static void show_object(struct object *object, const char *name, void *data_)
 438 {
 439         struct bitmap_show_data *data = data_;
 440         int bitmap_pos;
 441
 442         bitmap_pos = bitmap_position(data->bitmap_git, &object->oid);
 443
 444         if (bitmap_pos < 0)
 445                 bitmap_pos = ext_index_add_object(data->bitmap_git, object,
 446                                                   name);
 447
 448         bitmap_set(data->base, bitmap_pos);
 449 }
 450
 451 static void show_commit(struct commit *commit, void *data)
 452 {
 453 }
 454
 455 static int add_to_include_set(struct bitmap_index *bitmap_git,
 456                               struct include_data *data,
 457                               const struct object_id *oid,
 458                               int bitmap_pos)
 459 {
 460         khiter_t hash_pos;
 461
 462         if (data->seen && bitmap_get(data->seen, bitmap_pos))
 463                 return 0;
 464
 465         if (bitmap_get(data->base, bitmap_pos))
 466                 return 0;
 467
 468         hash_pos = kh_get_oid_map(bitmap_git->bitmaps, *oid);
 469         if (hash_pos < kh_end(bitmap_git->bitmaps)) {
 470                 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, hash_pos);
 471                 bitmap_or_ewah(data->base, lookup_stored_bitmap(st));
 472                 return 0;
 473         }
 474
 475         bitmap_set(data->base, bitmap_pos);
 476         return 1;
 477 }
 478
 479 static int should_include(struct commit *commit, void *_data)
 480 {
 481         struct include_data *data = _data;
 482         int bitmap_pos;
 483
 484         bitmap_pos = bitmap_position(data->bitmap_git, &commit->object.oid);
 485         if (bitmap_pos < 0)
 486                 bitmap_pos = ext_index_add_object(data->bitmap_git,
 487                                                   (struct object *)commit,
 488                                                   NULL);
 489
 490         if (!add_to_include_set(data->bitmap_git, data, &commit->object.oid,
 491                                 bitmap_pos)) {
 492                 struct commit_list *parent = commit->parents;
 493
 494                 while (parent) {
 495                         parent->item->object.flags |= SEEN;
 496                         parent = parent->next;
 497                 }
 498
 499                 return 0;
 500         }
 501
 502         return 1;
 503 }
 504
 505 static struct bitmap *find_objects(struct bitmap_index *bitmap_git,
 506                                    struct rev_info *revs,
 507                                    struct object_list *roots,
 508                                    struct bitmap *seen)
 509 {
 510         struct bitmap *base = NULL;
 511         int needs_walk = 0;
 512
 513         struct object_list *not_mapped = NULL;
 514
 515         /*
 516          * Go through all the roots for the walk. The ones that have bitmaps
 517          * on the bitmap index will be `or`ed together to form an initial
 518          * global reachability analysis.
 519          *
 520          * The ones without bitmaps in the index will be stored in the
 521          * `not_mapped_list` for further processing.
 522          */
 523         while (roots) {
 524                 struct object *object = roots->item;
 525                 roots = roots->next;
 526
 527                 if (object->type == OBJ_COMMIT) {
 528                         khiter_t pos = kh_get_oid_map(bitmap_git->bitmaps, object->oid);
 529
 530                         if (pos < kh_end(bitmap_git->bitmaps)) {
 531                                 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, pos);
 532                                 struct ewah_bitmap *or_with = lookup_stored_bitmap(st);
 533
 534                                 if (base == NULL)
 535                                         base = ewah_to_bitmap(or_with);
 536                                 else
 537                                         bitmap_or_ewah(base, or_with);
 538
 539                                 object->flags |= SEEN;
 540                                 continue;
 541                         }
 542                 }
 543
 544                 object_list_insert(object, &not_mapped);
 545         }
 546
 547         /*
 548          * Best case scenario: We found bitmaps for all the roots,
 549          * so the resulting `or` bitmap has the full reachability analysis
 550          */
 551         if (not_mapped == NULL)
 552                 return base;
 553
 554         roots = not_mapped;
 555
 556         /*
 557          * Let's iterate through all the roots that don't have bitmaps to
 558          * check if we can determine them to be reachable from the existing
 559          * global bitmap.
 560          *
 561          * If we cannot find them in the existing global bitmap, we'll need
 562          * to push them to an actual walk and run it until we can confirm
 563          * they are reachable
 564          */
 565         while (roots) {
 566                 struct object *object = roots->item;
 567                 int pos;
 568
 569                 roots = roots->next;
 570                 pos = bitmap_position(bitmap_git, &object->oid);
 571
 572                 if (pos < 0 || base == NULL || !bitmap_get(base, pos)) {
 573                         object->flags &= ~UNINTERESTING;
 574                         add_pending_object(revs, object, "");
 575                         needs_walk = 1;
 576                 } else {
 577                         object->flags |= SEEN;
 578                 }
 579         }
 580
 581         if (needs_walk) {
 582                 struct include_data incdata;
 583                 struct bitmap_show_data show_data;
 584
 585                 if (base == NULL)
 586                         base = bitmap_new();
 587
 588                 incdata.bitmap_git = bitmap_git;
 589                 incdata.base = base;
 590                 incdata.seen = seen;
 591
 592                 revs->include_check = should_include;
 593                 revs->include_check_data = &incdata;
 594
 595                 if (prepare_revision_walk(revs))
 596                         die("revision walk setup failed");
 597
 598                 show_data.bitmap_git = bitmap_git;
 599                 show_data.base = base;
 600
 601                 traverse_commit_list(revs, show_commit, show_object,
 602                                      &show_data);
 603         }
 604
 605         return base;
 606 }
 607
 608 static void show_extended_objects(struct bitmap_index *bitmap_git,
 609                                   show_reachable_fn show_reach)
 610 {
 611         struct bitmap *objects = bitmap_git->result;
 612         struct eindex *eindex = &bitmap_git->ext_index;
 613         uint32_t i;
 614
 615         for (i = 0; i < eindex->count; ++i) {
 616                 struct object *obj;
 617
 618                 if (!bitmap_get(objects, bitmap_git->pack->num_objects + i))
 619                         continue;
 620
 621                 obj = eindex->objects[i];
 622                 show_reach(&obj->oid, obj->type, 0, eindex->hashes[i], NULL, 0);
 623         }
 624 }
 625
 626 static void show_objects_for_type(
 627         struct bitmap_index *bitmap_git,
 628         struct ewah_bitmap *type_filter,
 629         enum object_type object_type,
 630         show_reachable_fn show_reach)
 631 {
 632         size_t i = 0;
 633         uint32_t offset;
 634
 635         struct ewah_iterator it;
 636         eword_t filter;
 637
 638         struct bitmap *objects = bitmap_git->result;
 639
 640         ewah_iterator_init(&it, type_filter);
 641
 642         for (i = 0; i < objects->word_alloc &&
 643                         ewah_iterator_next(&filter, &it); i++) {
 644                 eword_t word = objects->words[i] & filter;
 645                 size_t pos = (i * BITS_IN_EWORD);
 646
 647                 if (!word)
 648                         continue;
 649
 650                 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
 651                         struct object_id oid;
 652                         struct revindex_entry *entry;
 653                         uint32_t hash = 0;
 654
 655                         if ((word >> offset) == 0)
 656                                 break;
 657
 658                         offset += ewah_bit_ctz64(word >> offset);
 659
 660                         entry = &bitmap_git->pack->revindex[pos + offset];
 661                         nth_packed_object_oid(&oid, bitmap_git->pack, entry->nr);
 662
 663                         if (bitmap_git->hashes)
 664                                 hash = get_be32(bitmap_git->hashes + entry->nr);
 665
 666                         show_reach(&oid, object_type, 0, hash, bitmap_git->pack, entry->offset);
 667                 }
 668         }
 669 }
 670
 671 static int in_bitmapped_pack(struct bitmap_index *bitmap_git,
 672                              struct object_list *roots)
 673 {
 674         while (roots) {
 675                 struct object *object = roots->item;
 676                 roots = roots->next;
 677
 678                 if (find_pack_entry_one(object->oid.hash, bitmap_git->pack) > 0)
 679                         return 1;
 680         }
 681
 682         return 0;
 683 }
 684
 685 struct bitmap_index *prepare_bitmap_walk(struct rev_info *revs)
 686 {
 687         unsigned int i;
 688
 689         struct object_list *wants = NULL;
 690         struct object_list *haves = NULL;
 691
 692         struct bitmap *wants_bitmap = NULL;
 693         struct bitmap *haves_bitmap = NULL;
 694
 695         struct bitmap_index *bitmap_git = xcalloc(1, sizeof(*bitmap_git));
 696         /* try to open a bitmapped pack, but don't parse it yet
 697          * because we may not need to use it */
 698         if (open_pack_bitmap(revs->repo, bitmap_git) < 0)
 699                 goto cleanup;
 700
 701         for (i = 0; i < revs->pending.nr; ++i) {
 702                 struct object *object = revs->pending.objects[i].item;
 703
 704                 if (object->type == OBJ_NONE)
 705                         parse_object_or_die(&object->oid, NULL);
 706
 707                 while (object->type == OBJ_TAG) {
 708                         struct tag *tag = (struct tag *) object;
 709
 710                         if (object->flags & UNINTERESTING)
 711                                 object_list_insert(object, &haves);
 712                         else
 713                                 object_list_insert(object, &wants);
 714
 715                         object = parse_object_or_die(get_tagged_oid(tag), NULL);
 716                 }
 717
 718                 if (object->flags & UNINTERESTING)
 719                         object_list_insert(object, &haves);
 720                 else
 721                         object_list_insert(object, &wants);
 722         }
 723
 724         /*
 725          * if we have a HAVES list, but none of those haves is contained
 726          * in the packfile that has a bitmap, we don't have anything to
 727          * optimize here
 728          */
 729         if (haves && !in_bitmapped_pack(bitmap_git, haves))
 730                 goto cleanup;
 731
 732         /* if we don't want anything, we're done here */
 733         if (!wants)
 734                 goto cleanup;
 735
 736         /*
 737          * now we're going to use bitmaps, so load the actual bitmap entries
 738          * from disk. this is the point of no return; after this the rev_list
 739          * becomes invalidated and we must perform the revwalk through bitmaps
 740          */
 741         if (load_pack_bitmap(bitmap_git) < 0)
 742                 goto cleanup;
 743
 744         object_array_clear(&revs->pending);
 745
 746         if (haves) {
 747                 revs->ignore_missing_links = 1;
 748                 haves_bitmap = find_objects(bitmap_git, revs, haves, NULL);
 749                 reset_revision_walk();
 750                 revs->ignore_missing_links = 0;
 751
 752                 if (haves_bitmap == NULL)
 753                         BUG("failed to perform bitmap walk");
 754         }
 755
 756         wants_bitmap = find_objects(bitmap_git, revs, wants, haves_bitmap);
 757
 758         if (!wants_bitmap)
 759                 BUG("failed to perform bitmap walk");
 760
 761         if (haves_bitmap)
 762                 bitmap_and_not(wants_bitmap, haves_bitmap);
 763
 764         bitmap_git->result = wants_bitmap;
 765         bitmap_git->haves = haves_bitmap;
 766
 767         return bitmap_git;
 768
 769 cleanup:
 770         free_bitmap_index(bitmap_git);
 771         return NULL;
 772 }
 773
 774 static void try_partial_reuse(struct bitmap_index *bitmap_git,
 775                               size_t pos,
 776                               struct bitmap *reuse,
 777                               struct pack_window **w_curs)
 778 {
 779         struct revindex_entry *revidx;
 780         off_t offset;
 781         enum object_type type;
 782         unsigned long size;
 783
 784         if (pos >= bitmap_git->pack->num_objects)
 785                 return; /* not actually in the pack */
 786
 787         revidx = &bitmap_git->pack->revindex[pos];
 788         offset = revidx->offset;
 789         type = unpack_object_header(bitmap_git->pack, w_curs, &offset, &size);
 790         if (type < 0)
 791                 return; /* broken packfile, punt */
 792
 793         if (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA) {
 794                 off_t base_offset;
 795                 int base_pos;
 796
 797                 /*
 798                  * Find the position of the base object so we can look it up
 799                  * in our bitmaps. If we can't come up with an offset, or if
 800                  * that offset is not in the revidx, the pack is corrupt.
 801                  * There's nothing we can do, so just punt on this object,
 802                  * and the normal slow path will complain about it in
 803                  * more detail.
 804                  */
 805                 base_offset = get_delta_base(bitmap_git->pack, w_curs,
 806                                              &offset, type, revidx->offset);
 807                 if (!base_offset)
 808                         return;
 809                 base_pos = find_revindex_position(bitmap_git->pack, base_offset);
 810                 if (base_pos < 0)
 811                         return;
 812
 813                 /*
 814                  * We assume delta dependencies always point backwards. This
 815                  * lets us do a single pass, and is basically always true
 816                  * due to the way OFS_DELTAs work. You would not typically
 817                  * find REF_DELTA in a bitmapped pack, since we only bitmap
 818                  * packs we write fresh, and OFS_DELTA is the default). But
 819                  * let's double check to make sure the pack wasn't written with
 820                  * odd parameters.
 821                  */
 822                 if (base_pos >= pos)
 823                         return;
 824
 825                 /*
 826                  * And finally, if we're not sending the base as part of our
 827                  * reuse chunk, then don't send this object either. The base
 828                  * would come after us, along with other objects not
 829                  * necessarily in the pack, which means we'd need to convert
 830                  * to REF_DELTA on the fly. Better to just let the normal
 831                  * object_entry code path handle it.
 832                  */
 833                 if (!bitmap_get(reuse, base_pos))
 834                         return;
 835         }
 836
 837         /*
 838          * If we got here, then the object is OK to reuse. Mark it.
 839          */
 840         bitmap_set(reuse, pos);
 841 }
 842
 843 int reuse_partial_packfile_from_bitmap(struct bitmap_index *bitmap_git,
 844                                        struct packed_git **packfile_out,
 845                                        uint32_t *entries,
 846                                        struct bitmap **reuse_out)
 847 {
 848         struct bitmap *result = bitmap_git->result;
 849         struct bitmap *reuse;
 850         struct pack_window *w_curs = NULL;
 851         size_t i = 0;
 852         uint32_t offset;
 853
 854         assert(result);
 855
 856         while (i < result->word_alloc && result->words[i] == (eword_t)~0)
 857                 i++;
 858
 859         /* Don't mark objects not in the packfile */
 860         if (i > bitmap_git->pack->num_objects / BITS_IN_EWORD)
 861                 i = bitmap_git->pack->num_objects / BITS_IN_EWORD;
 862
 863         reuse = bitmap_word_alloc(i);
 864         memset(reuse->words, 0xFF, i * sizeof(eword_t));
 865
 866         for (; i < result->word_alloc; ++i) {
 867                 eword_t word = result->words[i];
 868                 size_t pos = (i * BITS_IN_EWORD);
 869
 870                 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
 871                         if ((word >> offset) == 0)
 872                                 break;
 873
 874                         offset += ewah_bit_ctz64(word >> offset);
 875                         try_partial_reuse(bitmap_git, pos + offset, reuse, &w_curs);
 876                 }
 877         }
 878
 879         unuse_pack(&w_curs);
 880
 881         *entries = bitmap_popcount(reuse);
 882         if (!*entries) {
 883                 bitmap_free(reuse);
 884                 return -1;
 885         }
 886
 887         /*
 888          * Drop any reused objects from the result, since they will not
 889          * need to be handled separately.
 890          */
 891         bitmap_and_not(result, reuse);
 892         *packfile_out = bitmap_git->pack;
 893         *reuse_out = reuse;
 894         return 0;
 895 }
 896
 897 int bitmap_walk_contains(struct bitmap_index *bitmap_git,
 898                          struct bitmap *bitmap, const struct object_id *oid)
 899 {
 900         int idx;
 901
 902         if (!bitmap)
 903                 return 0;
 904
 905         idx = bitmap_position(bitmap_git, oid);
 906         return idx >= 0 && bitmap_get(bitmap, idx);
 907 }
 908
 909 void traverse_bitmap_commit_list(struct bitmap_index *bitmap_git,
 910                                  show_reachable_fn show_reachable)
 911 {
 912         assert(bitmap_git->result);
 913
 914         show_objects_for_type(bitmap_git, bitmap_git->commits,
 915                 OBJ_COMMIT, show_reachable);
 916         show_objects_for_type(bitmap_git, bitmap_git->trees,
 917                 OBJ_TREE, show_reachable);
 918         show_objects_for_type(bitmap_git, bitmap_git->blobs,
 919                 OBJ_BLOB, show_reachable);
 920         show_objects_for_type(bitmap_git, bitmap_git->tags,
 921                 OBJ_TAG, show_reachable);
 922
 923         show_extended_objects(bitmap_git, show_reachable);
 924 }
 925
 926 static uint32_t count_object_type(struct bitmap_index *bitmap_git,
 927                                   enum object_type type)
 928 {
 929         struct bitmap *objects = bitmap_git->result;
 930         struct eindex *eindex = &bitmap_git->ext_index;
 931
 932         uint32_t i = 0, count = 0;
 933         struct ewah_iterator it;
 934         eword_t filter;
 935
 936         switch (type) {
 937         case OBJ_COMMIT:
 938                 ewah_iterator_init(&it, bitmap_git->commits);
 939                 break;
 940
 941         case OBJ_TREE:
 942                 ewah_iterator_init(&it, bitmap_git->trees);
 943                 break;
 944
 945         case OBJ_BLOB:
 946                 ewah_iterator_init(&it, bitmap_git->blobs);
 947                 break;
 948
 949         case OBJ_TAG:
 950                 ewah_iterator_init(&it, bitmap_git->tags);
 951                 break;
 952
 953         default:
 954                 return 0;
 955         }
 956
 957         while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
 958                 eword_t word = objects->words[i++] & filter;
 959                 count += ewah_bit_popcount64(word);
 960         }
 961
 962         for (i = 0; i < eindex->count; ++i) {
 963                 if (eindex->objects[i]->type == type &&
 964                         bitmap_get(objects, bitmap_git->pack->num_objects + i))
 965                         count++;
 966         }
 967
 968         return count;
 969 }
 970
 971 void count_bitmap_commit_list(struct bitmap_index *bitmap_git,
 972                               uint32_t *commits, uint32_t *trees,
 973                               uint32_t *blobs, uint32_t *tags)
 974 {
 975         assert(bitmap_git->result);
 976
 977         if (commits)
 978                 *commits = count_object_type(bitmap_git, OBJ_COMMIT);
 979
 980         if (trees)
 981                 *trees = count_object_type(bitmap_git, OBJ_TREE);
 982
 983         if (blobs)
 984                 *blobs = count_object_type(bitmap_git, OBJ_BLOB);
 985
 986         if (tags)
 987                 *tags = count_object_type(bitmap_git, OBJ_TAG);
 988 }
 989
 990 struct bitmap_test_data {
 991         struct bitmap_index *bitmap_git;
 992         struct bitmap *base;
 993         struct progress *prg;
 994         size_t seen;
 995 };
 996
 997 static void test_show_object(struct object *object, const char *name,
 998                              void *data)
 999 {
1000         struct bitmap_test_data *tdata = data;
1001         int bitmap_pos;
1002
1003         bitmap_pos = bitmap_position(tdata->bitmap_git, &object->oid);
1004         if (bitmap_pos < 0)
1005                 die("Object not in bitmap: %s\n", oid_to_hex(&object->oid));
1006
1007         bitmap_set(tdata->base, bitmap_pos);
1008         display_progress(tdata->prg, ++tdata->seen);
1009 }
1010
1011 static void test_show_commit(struct commit *commit, void *data)
1012 {
1013         struct bitmap_test_data *tdata = data;
1014         int bitmap_pos;
1015
1016         bitmap_pos = bitmap_position(tdata->bitmap_git,
1017                                      &commit->object.oid);
1018         if (bitmap_pos < 0)
1019                 die("Object not in bitmap: %s\n", oid_to_hex(&commit->object.oid));
1020
1021         bitmap_set(tdata->base, bitmap_pos);
1022         display_progress(tdata->prg, ++tdata->seen);
1023 }
1024
1025 void test_bitmap_walk(struct rev_info *revs)
1026 {
1027         struct object *root;
1028         struct bitmap *result = NULL;
1029         khiter_t pos;
1030         size_t result_popcnt;
1031         struct bitmap_test_data tdata;
1032         struct bitmap_index *bitmap_git;
1033
1034         if (!(bitmap_git = prepare_bitmap_git(revs->repo)))
1035                 die("failed to load bitmap indexes");
1036
1037         if (revs->pending.nr != 1)
1038                 die("you must specify exactly one commit to test");
1039
1040         fprintf(stderr, "Bitmap v%d test (%d entries loaded)\n",
1041                 bitmap_git->version, bitmap_git->entry_count);
1042
1043         root = revs->pending.objects[0].item;
1044         pos = kh_get_oid_map(bitmap_git->bitmaps, root->oid);
1045
1046         if (pos < kh_end(bitmap_git->bitmaps)) {
1047                 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, pos);
1048                 struct ewah_bitmap *bm = lookup_stored_bitmap(st);
1049
1050                 fprintf(stderr, "Found bitmap for %s. %d bits / %08x checksum\n",
1051                         oid_to_hex(&root->oid), (int)bm->bit_size, ewah_checksum(bm));
1052
1053                 result = ewah_to_bitmap(bm);
1054         }
1055
1056         if (result == NULL)
1057                 die("Commit %s doesn't have an indexed bitmap", oid_to_hex(&root->oid));
1058
1059         revs->tag_objects = 1;
1060         revs->tree_objects = 1;
1061         revs->blob_objects = 1;
1062
1063         result_popcnt = bitmap_popcount(result);
1064
1065         if (prepare_revision_walk(revs))
1066                 die("revision walk setup failed");
1067
1068         tdata.bitmap_git = bitmap_git;
1069         tdata.base = bitmap_new();
1070         tdata.prg = start_progress("Verifying bitmap entries", result_popcnt);
1071         tdata.seen = 0;
1072
1073         traverse_commit_list(revs, &test_show_commit, &test_show_object, &tdata);
1074
1075         stop_progress(&tdata.prg);
1076
1077         if (bitmap_equals(result, tdata.base))
1078                 fprintf(stderr, "OK!\n");
1079         else
1080                 fprintf(stderr, "Mismatch!\n");
1081
1082         free_bitmap_index(bitmap_git);
1083 }
1084
1085 static int rebuild_bitmap(uint32_t *reposition,
1086                           struct ewah_bitmap *source,
1087                           struct bitmap *dest)
1088 {
1089         uint32_t pos = 0;
1090         struct ewah_iterator it;
1091         eword_t word;
1092
1093         ewah_iterator_init(&it, source);
1094
1095         while (ewah_iterator_next(&word, &it)) {
1096                 uint32_t offset, bit_pos;
1097
1098                 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
1099                         if ((word >> offset) == 0)
1100                                 break;
1101
1102                         offset += ewah_bit_ctz64(word >> offset);
1103
1104                         bit_pos = reposition[pos + offset];
1105                         if (bit_pos > 0)
1106                                 bitmap_set(dest, bit_pos - 1);
1107                         else /* can't reuse, we don't have the object */
1108                                 return -1;
1109                 }
1110
1111                 pos += BITS_IN_EWORD;
1112         }
1113         return 0;
1114 }
1115
1116 int rebuild_existing_bitmaps(struct bitmap_index *bitmap_git,
1117                              struct packing_data *mapping,
1118                              kh_oid_map_t *reused_bitmaps,
1119                              int show_progress)
1120 {
1121         uint32_t i, num_objects;
1122         uint32_t *reposition;
1123         struct bitmap *rebuild;
1124         struct stored_bitmap *stored;
1125         struct progress *progress = NULL;
1126
1127         khiter_t hash_pos;
1128         int hash_ret;
1129
1130         num_objects = bitmap_git->pack->num_objects;
1131         reposition = xcalloc(num_objects, sizeof(uint32_t));
1132
1133         for (i = 0; i < num_objects; ++i) {
1134                 struct object_id oid;
1135                 struct revindex_entry *entry;
1136                 struct object_entry *oe;
1137
1138                 entry = &bitmap_git->pack->revindex[i];
1139                 nth_packed_object_oid(&oid, bitmap_git->pack, entry->nr);
1140                 oe = packlist_find(mapping, &oid);
1141
1142                 if (oe)
1143                         reposition[i] = oe_in_pack_pos(mapping, oe) + 1;
1144         }
1145
1146         rebuild = bitmap_new();
1147         i = 0;
1148
1149         if (show_progress)
1150                 progress = start_progress("Reusing bitmaps", 0);
1151
1152         kh_foreach_value(bitmap_git->bitmaps, stored, {
1153                 if (stored->flags & BITMAP_FLAG_REUSE) {
1154                         if (!rebuild_bitmap(reposition,
1155                                             lookup_stored_bitmap(stored),
1156                                             rebuild)) {
1157                                 hash_pos = kh_put_oid_map(reused_bitmaps,
1158                                                           stored->oid,
1159                                                           &hash_ret);
1160                                 kh_value(reused_bitmaps, hash_pos) =
1161                                         bitmap_to_ewah(rebuild);
1162                         }
1163                         bitmap_reset(rebuild);
1164                         display_progress(progress, ++i);
1165                 }
1166         });
1167
1168         stop_progress(&progress);
1169
1170         free(reposition);
1171         bitmap_free(rebuild);
1172         return 0;
1173 }
1174
1175 void free_bitmap_index(struct bitmap_index *b)
1176 {
1177         if (!b)
1178                 return;
1179
1180         if (b->map)
1181                 munmap(b->map, b->map_size);
1182         ewah_pool_free(b->commits);
1183         ewah_pool_free(b->trees);
1184         ewah_pool_free(b->blobs);
1185         ewah_pool_free(b->tags);
1186         kh_destroy_oid_map(b->bitmaps);
1187         free(b->ext_index.objects);
1188         free(b->ext_index.hashes);
1189         bitmap_free(b->result);
1190         bitmap_free(b->haves);
1191         free(b);
1192 }
1193
1194 int bitmap_has_oid_in_uninteresting(struct bitmap_index *bitmap_git,
1195                                     const struct object_id *oid)
1196 {
1197         return bitmap_git &&
1198                 bitmap_walk_contains(bitmap_git, bitmap_git->haves, oid);
1199 }