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 #include "list-objects-filter-options.h"
  16 #include "config.h"
  17
  18 /*
  19  * An entry on the bitmap index, representing the bitmap for a given
  20  * commit.
  21  */
  22 struct stored_bitmap {
  23         struct object_id oid;
  24         struct ewah_bitmap *root;
  25         struct stored_bitmap *xor;
  26         int flags;
  27 };
  28
  29 /*
  30  * The active bitmap index for a repository. By design, repositories only have
  31  * a single bitmap index available (the index for the biggest packfile in
  32  * the repository), since bitmap indexes need full closure.
  33  *
  34  * If there is more than one bitmap index available (e.g. because of alternates),
  35  * the active bitmap index is the largest one.
  36  */
  37 struct bitmap_index {
  38         /* Packfile to which this bitmap index belongs to */
  39         struct packed_git *pack;
  40
  41         /*
  42          * Mark the first `reuse_objects` in the packfile as reused:
  43          * they will be sent as-is without using them for repacking
  44          * calculations
  45          */
  46         uint32_t reuse_objects;
  47
  48         /* mmapped buffer of the whole bitmap index */
  49         unsigned char *map;
  50         size_t map_size; /* size of the mmaped buffer */
  51         size_t map_pos; /* current position when loading the index */
  52
  53         /*
  54          * Type indexes.
  55          *
  56          * Each bitmap marks which objects in the packfile  are of the given
  57          * type. This provides type information when yielding the objects from
  58          * the packfile during a walk, which allows for better delta bases.
  59          */
  60         struct ewah_bitmap *commits;
  61         struct ewah_bitmap *trees;
  62         struct ewah_bitmap *blobs;
  63         struct ewah_bitmap *tags;
  64
  65         /* Map from object ID -> `stored_bitmap` for all the bitmapped commits */
  66         kh_oid_map_t *bitmaps;
  67
  68         /* Number of bitmapped commits */
  69         uint32_t entry_count;
  70
  71         /* If not NULL, this is a name-hash cache pointing into map. */
  72         uint32_t *hashes;
  73
  74         /*
  75          * Extended index.
  76          *
  77          * When trying to perform bitmap operations with objects that are not
  78          * packed in `pack`, these objects are added to this "fake index" and
  79          * are assumed to appear at the end of the packfile for all operations
  80          */
  81         struct eindex {
  82                 struct object **objects;
  83                 uint32_t *hashes;
  84                 uint32_t count, alloc;
  85                 kh_oid_pos_t *positions;
  86         } ext_index;
  87
  88         /* Bitmap result of the last performed walk */
  89         struct bitmap *result;
  90
  91         /* "have" bitmap from the last performed walk */
  92         struct bitmap *haves;
  93
  94         /* Version of the bitmap index */
  95         unsigned int version;
  96 };
  97
  98 static struct ewah_bitmap *lookup_stored_bitmap(struct stored_bitmap *st)
  99 {
 100         struct ewah_bitmap *parent;
 101         struct ewah_bitmap *composed;
 102
 103         if (st->xor == NULL)
 104                 return st->root;
 105
 106         composed = ewah_pool_new();
 107         parent = lookup_stored_bitmap(st->xor);
 108         ewah_xor(st->root, parent, composed);
 109
 110         ewah_pool_free(st->root);
 111         st->root = composed;
 112         st->xor = NULL;
 113
 114         return composed;
 115 }
 116
 117 /*
 118  * Read a bitmap from the current read position on the mmaped
 119  * index, and increase the read position accordingly
 120  */
 121 static struct ewah_bitmap *read_bitmap_1(struct bitmap_index *index)
 122 {
 123         struct ewah_bitmap *b = ewah_pool_new();
 124
 125         ssize_t bitmap_size = ewah_read_mmap(b,
 126                 index->map + index->map_pos,
 127                 index->map_size - index->map_pos);
 128
 129         if (bitmap_size < 0) {
 130                 error("Failed to load bitmap index (corrupted?)");
 131                 ewah_pool_free(b);
 132                 return NULL;
 133         }
 134
 135         index->map_pos += bitmap_size;
 136         return b;
 137 }
 138
 139 static int load_bitmap_header(struct bitmap_index *index)
 140 {
 141         struct bitmap_disk_header *header = (void *)index->map;
 142         size_t header_size = sizeof(*header) - GIT_MAX_RAWSZ + the_hash_algo->rawsz;
 143
 144         if (index->map_size < header_size + the_hash_algo->rawsz)
 145                 return error("Corrupted bitmap index (too small)");
 146
 147         if (memcmp(header->magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE)) != 0)
 148                 return error("Corrupted bitmap index file (wrong header)");
 149
 150         index->version = ntohs(header->version);
 151         if (index->version != 1)
 152                 return error("Unsupported version for bitmap index file (%d)", index->version);
 153
 154         /* Parse known bitmap format options */
 155         {
 156                 uint32_t flags = ntohs(header->options);
 157                 size_t cache_size = st_mult(index->pack->num_objects, sizeof(uint32_t));
 158                 unsigned char *index_end = index->map + index->map_size - the_hash_algo->rawsz;
 159
 160                 if ((flags & BITMAP_OPT_FULL_DAG) == 0)
 161                         return error("Unsupported options for bitmap index file "
 162                                 "(Git requires BITMAP_OPT_FULL_DAG)");
 163
 164                 if (flags & BITMAP_OPT_HASH_CACHE) {
 165                         if (cache_size > index_end - index->map - header_size)
 166                                 return error("corrupted bitmap index file (too short to fit hash cache)");
 167                         index->hashes = (void *)(index_end - cache_size);
 168                         index_end -= cache_size;
 169                 }
 170         }
 171
 172         index->entry_count = ntohl(header->entry_count);
 173         index->map_pos += header_size;
 174         return 0;
 175 }
 176
 177 static struct stored_bitmap *store_bitmap(struct bitmap_index *index,
 178                                           struct ewah_bitmap *root,
 179                                           const struct object_id *oid,
 180                                           struct stored_bitmap *xor_with,
 181                                           int flags)
 182 {
 183         struct stored_bitmap *stored;
 184         khiter_t hash_pos;
 185         int ret;
 186
 187         stored = xmalloc(sizeof(struct stored_bitmap));
 188         stored->root = root;
 189         stored->xor = xor_with;
 190         stored->flags = flags;
 191         oidcpy(&stored->oid, oid);
 192
 193         hash_pos = kh_put_oid_map(index->bitmaps, stored->oid, &ret);
 194
 195         /* a 0 return code means the insertion succeeded with no changes,
 196          * because the SHA1 already existed on the map. this is bad, there
 197          * shouldn't be duplicated commits in the index */
 198         if (ret == 0) {
 199                 error("Duplicate entry in bitmap index: %s", oid_to_hex(oid));
 200                 return NULL;
 201         }
 202
 203         kh_value(index->bitmaps, hash_pos) = stored;
 204         return stored;
 205 }
 206
 207 static inline uint32_t read_be32(const unsigned char *buffer, size_t *pos)
 208 {
 209         uint32_t result = get_be32(buffer + *pos);
 210         (*pos) += sizeof(result);
 211         return result;
 212 }
 213
 214 static inline uint8_t read_u8(const unsigned char *buffer, size_t *pos)
 215 {
 216         return buffer[(*pos)++];
 217 }
 218
 219 #define MAX_XOR_OFFSET 160
 220
 221 static int load_bitmap_entries_v1(struct bitmap_index *index)
 222 {
 223         uint32_t i;
 224         struct stored_bitmap *recent_bitmaps[MAX_XOR_OFFSET] = { NULL };
 225
 226         for (i = 0; i < index->entry_count; ++i) {
 227                 int xor_offset, flags;
 228                 struct ewah_bitmap *bitmap = NULL;
 229                 struct stored_bitmap *xor_bitmap = NULL;
 230                 uint32_t commit_idx_pos;
 231                 struct object_id oid;
 232
 233                 if (index->map_size - index->map_pos < 6)
 234                         return error("corrupt ewah bitmap: truncated header for entry %d", i);
 235
 236                 commit_idx_pos = read_be32(index->map, &index->map_pos);
 237                 xor_offset = read_u8(index->map, &index->map_pos);
 238                 flags = read_u8(index->map, &index->map_pos);
 239
 240                 if (nth_packed_object_id(&oid, index->pack, commit_idx_pos) < 0)
 241                         return error("corrupt ewah bitmap: commit index %u out of range",
 242                                      (unsigned)commit_idx_pos);
 243
 244                 bitmap = read_bitmap_1(index);
 245                 if (!bitmap)
 246                         return -1;
 247
 248                 if (xor_offset > MAX_XOR_OFFSET || xor_offset > i)
 249                         return error("Corrupted bitmap pack index");
 250
 251                 if (xor_offset > 0) {
 252                         xor_bitmap = recent_bitmaps[(i - xor_offset) % MAX_XOR_OFFSET];
 253
 254                         if (xor_bitmap == NULL)
 255                                 return error("Invalid XOR offset in bitmap pack index");
 256                 }
 257
 258                 recent_bitmaps[i % MAX_XOR_OFFSET] = store_bitmap(
 259                         index, bitmap, &oid, xor_bitmap, flags);
 260         }
 261
 262         return 0;
 263 }
 264
 265 static char *pack_bitmap_filename(struct packed_git *p)
 266 {
 267         size_t len;
 268
 269         if (!strip_suffix(p->pack_name, ".pack", &len))
 270                 BUG("pack_name does not end in .pack");
 271         return xstrfmt("%.*s.bitmap", (int)len, p->pack_name);
 272 }
 273
 274 static int open_pack_bitmap_1(struct bitmap_index *bitmap_git, struct packed_git *packfile)
 275 {
 276         int fd;
 277         struct stat st;
 278         char *idx_name;
 279
 280         if (open_pack_index(packfile))
 281                 return -1;
 282
 283         idx_name = pack_bitmap_filename(packfile);
 284         fd = git_open(idx_name);
 285         free(idx_name);
 286
 287         if (fd < 0)
 288                 return -1;
 289
 290         if (fstat(fd, &st)) {
 291                 close(fd);
 292                 return -1;
 293         }
 294
 295         if (bitmap_git->pack) {
 296                 warning("ignoring extra bitmap file: %s", packfile->pack_name);
 297                 close(fd);
 298                 return -1;
 299         }
 300
 301         bitmap_git->pack = packfile;
 302         bitmap_git->map_size = xsize_t(st.st_size);
 303         bitmap_git->map = xmmap(NULL, bitmap_git->map_size, PROT_READ, MAP_PRIVATE, fd, 0);
 304         bitmap_git->map_pos = 0;
 305         close(fd);
 306
 307         if (load_bitmap_header(bitmap_git) < 0) {
 308                 munmap(bitmap_git->map, bitmap_git->map_size);
 309                 bitmap_git->map = NULL;
 310                 bitmap_git->map_size = 0;
 311                 return -1;
 312         }
 313
 314         return 0;
 315 }
 316
 317 static int load_pack_bitmap(struct bitmap_index *bitmap_git)
 318 {
 319         assert(bitmap_git->map);
 320
 321         bitmap_git->bitmaps = kh_init_oid_map();
 322         bitmap_git->ext_index.positions = kh_init_oid_pos();
 323         if (load_pack_revindex(bitmap_git->pack))
 324                 goto failed;
 325
 326         if (!(bitmap_git->commits = read_bitmap_1(bitmap_git)) ||
 327                 !(bitmap_git->trees = read_bitmap_1(bitmap_git)) ||
 328                 !(bitmap_git->blobs = read_bitmap_1(bitmap_git)) ||
 329                 !(bitmap_git->tags = read_bitmap_1(bitmap_git)))
 330                 goto failed;
 331
 332         if (load_bitmap_entries_v1(bitmap_git) < 0)
 333                 goto failed;
 334
 335         return 0;
 336
 337 failed:
 338         munmap(bitmap_git->map, bitmap_git->map_size);
 339         bitmap_git->map = NULL;
 340         bitmap_git->map_size = 0;
 341
 342         kh_destroy_oid_map(bitmap_git->bitmaps);
 343         bitmap_git->bitmaps = NULL;
 344
 345         kh_destroy_oid_pos(bitmap_git->ext_index.positions);
 346         bitmap_git->ext_index.positions = NULL;
 347
 348         return -1;
 349 }
 350
 351 static int open_pack_bitmap(struct repository *r,
 352                             struct bitmap_index *bitmap_git)
 353 {
 354         struct packed_git *p;
 355         int ret = -1;
 356
 357         assert(!bitmap_git->map);
 358
 359         for (p = get_all_packs(r); p; p = p->next) {
 360                 if (open_pack_bitmap_1(bitmap_git, p) == 0)
 361                         ret = 0;
 362         }
 363
 364         return ret;
 365 }
 366
 367 struct bitmap_index *prepare_bitmap_git(struct repository *r)
 368 {
 369         struct bitmap_index *bitmap_git = xcalloc(1, sizeof(*bitmap_git));
 370
 371         if (!open_pack_bitmap(r, bitmap_git) && !load_pack_bitmap(bitmap_git))
 372                 return bitmap_git;
 373
 374         free_bitmap_index(bitmap_git);
 375         return NULL;
 376 }
 377
 378 struct include_data {
 379         struct bitmap_index *bitmap_git;
 380         struct bitmap *base;
 381         struct bitmap *seen;
 382 };
 383
 384 struct ewah_bitmap *bitmap_for_commit(struct bitmap_index *bitmap_git,
 385                                       struct commit *commit)
 386 {
 387         khiter_t hash_pos = kh_get_oid_map(bitmap_git->bitmaps,
 388                                            commit->object.oid);
 389         if (hash_pos >= kh_end(bitmap_git->bitmaps))
 390                 return NULL;
 391         return lookup_stored_bitmap(kh_value(bitmap_git->bitmaps, hash_pos));
 392 }
 393
 394 static inline int bitmap_position_extended(struct bitmap_index *bitmap_git,
 395                                            const struct object_id *oid)
 396 {
 397         kh_oid_pos_t *positions = bitmap_git->ext_index.positions;
 398         khiter_t pos = kh_get_oid_pos(positions, *oid);
 399
 400         if (pos < kh_end(positions)) {
 401                 int bitmap_pos = kh_value(positions, pos);
 402                 return bitmap_pos + bitmap_git->pack->num_objects;
 403         }
 404
 405         return -1;
 406 }
 407
 408 static inline int bitmap_position_packfile(struct bitmap_index *bitmap_git,
 409                                            const struct object_id *oid)
 410 {
 411         uint32_t pos;
 412         off_t offset = find_pack_entry_one(oid->hash, bitmap_git->pack);
 413         if (!offset)
 414                 return -1;
 415
 416         if (offset_to_pack_pos(bitmap_git->pack, offset, &pos) < 0)
 417                 return -1;
 418         return pos;
 419 }
 420
 421 static int bitmap_position(struct bitmap_index *bitmap_git,
 422                            const struct object_id *oid)
 423 {
 424         int pos = bitmap_position_packfile(bitmap_git, oid);
 425         return (pos >= 0) ? pos : bitmap_position_extended(bitmap_git, oid);
 426 }
 427
 428 static int ext_index_add_object(struct bitmap_index *bitmap_git,
 429                                 struct object *object, const char *name)
 430 {
 431         struct eindex *eindex = &bitmap_git->ext_index;
 432
 433         khiter_t hash_pos;
 434         int hash_ret;
 435         int bitmap_pos;
 436
 437         hash_pos = kh_put_oid_pos(eindex->positions, object->oid, &hash_ret);
 438         if (hash_ret > 0) {
 439                 if (eindex->count >= eindex->alloc) {
 440                         eindex->alloc = (eindex->alloc + 16) * 3 / 2;
 441                         REALLOC_ARRAY(eindex->objects, eindex->alloc);
 442                         REALLOC_ARRAY(eindex->hashes, eindex->alloc);
 443                 }
 444
 445                 bitmap_pos = eindex->count;
 446                 eindex->objects[eindex->count] = object;
 447                 eindex->hashes[eindex->count] = pack_name_hash(name);
 448                 kh_value(eindex->positions, hash_pos) = bitmap_pos;
 449                 eindex->count++;
 450         } else {
 451                 bitmap_pos = kh_value(eindex->positions, hash_pos);
 452         }
 453
 454         return bitmap_pos + bitmap_git->pack->num_objects;
 455 }
 456
 457 struct bitmap_show_data {
 458         struct bitmap_index *bitmap_git;
 459         struct bitmap *base;
 460 };
 461
 462 static void show_object(struct object *object, const char *name, void *data_)
 463 {
 464         struct bitmap_show_data *data = data_;
 465         int bitmap_pos;
 466
 467         bitmap_pos = bitmap_position(data->bitmap_git, &object->oid);
 468
 469         if (bitmap_pos < 0)
 470                 bitmap_pos = ext_index_add_object(data->bitmap_git, object,
 471                                                   name);
 472
 473         bitmap_set(data->base, bitmap_pos);
 474 }
 475
 476 static void show_commit(struct commit *commit, void *data)
 477 {
 478 }
 479
 480 static int add_to_include_set(struct bitmap_index *bitmap_git,
 481                               struct include_data *data,
 482                               struct commit *commit,
 483                               int bitmap_pos)
 484 {
 485         struct ewah_bitmap *partial;
 486
 487         if (data->seen && bitmap_get(data->seen, bitmap_pos))
 488                 return 0;
 489
 490         if (bitmap_get(data->base, bitmap_pos))
 491                 return 0;
 492
 493         partial = bitmap_for_commit(bitmap_git, commit);
 494         if (partial) {
 495                 bitmap_or_ewah(data->base, partial);
 496                 return 0;
 497         }
 498
 499         bitmap_set(data->base, bitmap_pos);
 500         return 1;
 501 }
 502
 503 static int should_include(struct commit *commit, void *_data)
 504 {
 505         struct include_data *data = _data;
 506         int bitmap_pos;
 507
 508         bitmap_pos = bitmap_position(data->bitmap_git, &commit->object.oid);
 509         if (bitmap_pos < 0)
 510                 bitmap_pos = ext_index_add_object(data->bitmap_git,
 511                                                   (struct object *)commit,
 512                                                   NULL);
 513
 514         if (!add_to_include_set(data->bitmap_git, data, commit, bitmap_pos)) {
 515                 struct commit_list *parent = commit->parents;
 516
 517                 while (parent) {
 518                         parent->item->object.flags |= SEEN;
 519                         parent = parent->next;
 520                 }
 521
 522                 return 0;
 523         }
 524
 525         return 1;
 526 }
 527
 528 static int should_include_obj(struct object *obj, void *_data)
 529 {
 530         struct include_data *data = _data;
 531         int bitmap_pos;
 532
 533         bitmap_pos = bitmap_position(data->bitmap_git, &obj->oid);
 534         if (bitmap_pos < 0)
 535                 return 1;
 536         if ((data->seen && bitmap_get(data->seen, bitmap_pos)) ||
 537              bitmap_get(data->base, bitmap_pos)) {
 538                 obj->flags |= SEEN;
 539                 return 0;
 540         }
 541         return 1;
 542 }
 543
 544 static int add_commit_to_bitmap(struct bitmap_index *bitmap_git,
 545                                 struct bitmap **base,
 546                                 struct commit *commit)
 547 {
 548         struct ewah_bitmap *or_with = bitmap_for_commit(bitmap_git, commit);
 549
 550         if (!or_with)
 551                 return 0;
 552
 553         if (*base == NULL)
 554                 *base = ewah_to_bitmap(or_with);
 555         else
 556                 bitmap_or_ewah(*base, or_with);
 557
 558         return 1;
 559 }
 560
 561 static struct bitmap *find_objects(struct bitmap_index *bitmap_git,
 562                                    struct rev_info *revs,
 563                                    struct object_list *roots,
 564                                    struct bitmap *seen,
 565                                    struct list_objects_filter_options *filter)
 566 {
 567         struct bitmap *base = NULL;
 568         int needs_walk = 0;
 569
 570         struct object_list *not_mapped = NULL;
 571
 572         /*
 573          * Go through all the roots for the walk. The ones that have bitmaps
 574          * on the bitmap index will be `or`ed together to form an initial
 575          * global reachability analysis.
 576          *
 577          * The ones without bitmaps in the index will be stored in the
 578          * `not_mapped_list` for further processing.
 579          */
 580         while (roots) {
 581                 struct object *object = roots->item;
 582                 roots = roots->next;
 583
 584                 if (object->type == OBJ_COMMIT &&
 585                     add_commit_to_bitmap(bitmap_git, &base, (struct commit *)object)) {
 586                         object->flags |= SEEN;
 587                         continue;
 588                 }
 589
 590                 object_list_insert(object, &not_mapped);
 591         }
 592
 593         /*
 594          * Best case scenario: We found bitmaps for all the roots,
 595          * so the resulting `or` bitmap has the full reachability analysis
 596          */
 597         if (not_mapped == NULL)
 598                 return base;
 599
 600         roots = not_mapped;
 601
 602         /*
 603          * Let's iterate through all the roots that don't have bitmaps to
 604          * check if we can determine them to be reachable from the existing
 605          * global bitmap.
 606          *
 607          * If we cannot find them in the existing global bitmap, we'll need
 608          * to push them to an actual walk and run it until we can confirm
 609          * they are reachable
 610          */
 611         while (roots) {
 612                 struct object *object = roots->item;
 613                 int pos;
 614
 615                 roots = roots->next;
 616                 pos = bitmap_position(bitmap_git, &object->oid);
 617
 618                 if (pos < 0 || base == NULL || !bitmap_get(base, pos)) {
 619                         object->flags &= ~UNINTERESTING;
 620                         add_pending_object(revs, object, "");
 621                         needs_walk = 1;
 622                 } else {
 623                         object->flags |= SEEN;
 624                 }
 625         }
 626
 627         if (needs_walk) {
 628                 struct include_data incdata;
 629                 struct bitmap_show_data show_data;
 630
 631                 if (base == NULL)
 632                         base = bitmap_new();
 633
 634                 incdata.bitmap_git = bitmap_git;
 635                 incdata.base = base;
 636                 incdata.seen = seen;
 637
 638                 revs->include_check = should_include;
 639                 revs->include_check_obj = should_include_obj;
 640                 revs->include_check_data = &incdata;
 641
 642                 if (prepare_revision_walk(revs))
 643                         die("revision walk setup failed");
 644
 645                 show_data.bitmap_git = bitmap_git;
 646                 show_data.base = base;
 647
 648                 traverse_commit_list_filtered(filter, revs,
 649                                               show_commit, show_object,
 650                                               &show_data, NULL);
 651
 652                 revs->include_check = NULL;
 653                 revs->include_check_obj = NULL;
 654                 revs->include_check_data = NULL;
 655         }
 656
 657         return base;
 658 }
 659
 660 static void show_extended_objects(struct bitmap_index *bitmap_git,
 661                                   struct rev_info *revs,
 662                                   show_reachable_fn show_reach)
 663 {
 664         struct bitmap *objects = bitmap_git->result;
 665         struct eindex *eindex = &bitmap_git->ext_index;
 666         uint32_t i;
 667
 668         for (i = 0; i < eindex->count; ++i) {
 669                 struct object *obj;
 670
 671                 if (!bitmap_get(objects, bitmap_git->pack->num_objects + i))
 672                         continue;
 673
 674                 obj = eindex->objects[i];
 675                 if ((obj->type == OBJ_BLOB && !revs->blob_objects) ||
 676                     (obj->type == OBJ_TREE && !revs->tree_objects) ||
 677                     (obj->type == OBJ_TAG && !revs->tag_objects))
 678                         continue;
 679
 680                 show_reach(&obj->oid, obj->type, 0, eindex->hashes[i], NULL, 0);
 681         }
 682 }
 683
 684 static void init_type_iterator(struct ewah_iterator *it,
 685                                struct bitmap_index *bitmap_git,
 686                                enum object_type type)
 687 {
 688         switch (type) {
 689         case OBJ_COMMIT:
 690                 ewah_iterator_init(it, bitmap_git->commits);
 691                 break;
 692
 693         case OBJ_TREE:
 694                 ewah_iterator_init(it, bitmap_git->trees);
 695                 break;
 696
 697         case OBJ_BLOB:
 698                 ewah_iterator_init(it, bitmap_git->blobs);
 699                 break;
 700
 701         case OBJ_TAG:
 702                 ewah_iterator_init(it, bitmap_git->tags);
 703                 break;
 704
 705         default:
 706                 BUG("object type %d not stored by bitmap type index", type);
 707                 break;
 708         }
 709 }
 710
 711 static void show_objects_for_type(
 712         struct bitmap_index *bitmap_git,
 713         enum object_type object_type,
 714         show_reachable_fn show_reach)
 715 {
 716         size_t i = 0;
 717         uint32_t offset;
 718
 719         struct ewah_iterator it;
 720         eword_t filter;
 721
 722         struct bitmap *objects = bitmap_git->result;
 723
 724         init_type_iterator(&it, bitmap_git, object_type);
 725
 726         for (i = 0; i < objects->word_alloc &&
 727                         ewah_iterator_next(&filter, &it); i++) {
 728                 eword_t word = objects->words[i] & filter;
 729                 size_t pos = (i * BITS_IN_EWORD);
 730
 731                 if (!word)
 732                         continue;
 733
 734                 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
 735                         struct object_id oid;
 736                         uint32_t hash = 0, index_pos;
 737                         off_t ofs;
 738
 739                         if ((word >> offset) == 0)
 740                                 break;
 741
 742                         offset += ewah_bit_ctz64(word >> offset);
 743
 744                         index_pos = pack_pos_to_index(bitmap_git->pack, pos + offset);
 745                         ofs = pack_pos_to_offset(bitmap_git->pack, pos + offset);
 746                         nth_packed_object_id(&oid, bitmap_git->pack, index_pos);
 747
 748                         if (bitmap_git->hashes)
 749                                 hash = get_be32(bitmap_git->hashes + index_pos);
 750
 751                         show_reach(&oid, object_type, 0, hash, bitmap_git->pack, ofs);
 752                 }
 753         }
 754 }
 755
 756 static int in_bitmapped_pack(struct bitmap_index *bitmap_git,
 757                              struct object_list *roots)
 758 {
 759         while (roots) {
 760                 struct object *object = roots->item;
 761                 roots = roots->next;
 762
 763                 if (find_pack_entry_one(object->oid.hash, bitmap_git->pack) > 0)
 764                         return 1;
 765         }
 766
 767         return 0;
 768 }
 769
 770 static struct bitmap *find_tip_objects(struct bitmap_index *bitmap_git,
 771                                        struct object_list *tip_objects,
 772                                        enum object_type type)
 773 {
 774         struct bitmap *result = bitmap_new();
 775         struct object_list *p;
 776
 777         for (p = tip_objects; p; p = p->next) {
 778                 int pos;
 779
 780                 if (p->item->type != type)
 781                         continue;
 782
 783                 pos = bitmap_position(bitmap_git, &p->item->oid);
 784                 if (pos < 0)
 785                         continue;
 786
 787                 bitmap_set(result, pos);
 788         }
 789
 790         return result;
 791 }
 792
 793 static void filter_bitmap_exclude_type(struct bitmap_index *bitmap_git,
 794                                        struct object_list *tip_objects,
 795                                        struct bitmap *to_filter,
 796                                        enum object_type type)
 797 {
 798         struct eindex *eindex = &bitmap_git->ext_index;
 799         struct bitmap *tips;
 800         struct ewah_iterator it;
 801         eword_t mask;
 802         uint32_t i;
 803
 804         /*
 805          * The non-bitmap version of this filter never removes
 806          * objects which the other side specifically asked for,
 807          * so we must match that behavior.
 808          */
 809         tips = find_tip_objects(bitmap_git, tip_objects, type);
 810
 811         /*
 812          * We can use the blob type-bitmap to work in whole words
 813          * for the objects that are actually in the bitmapped packfile.
 814          */
 815         for (i = 0, init_type_iterator(&it, bitmap_git, type);
 816              i < to_filter->word_alloc && ewah_iterator_next(&mask, &it);
 817              i++) {
 818                 if (i < tips->word_alloc)
 819                         mask &= ~tips->words[i];
 820                 to_filter->words[i] &= ~mask;
 821         }
 822
 823         /*
 824          * Clear any blobs that weren't in the packfile (and so would not have
 825          * been caught by the loop above. We'll have to check them
 826          * individually.
 827          */
 828         for (i = 0; i < eindex->count; i++) {
 829                 uint32_t pos = i + bitmap_git->pack->num_objects;
 830                 if (eindex->objects[i]->type == type &&
 831                     bitmap_get(to_filter, pos) &&
 832                     !bitmap_get(tips, pos))
 833                         bitmap_unset(to_filter, pos);
 834         }
 835
 836         bitmap_free(tips);
 837 }
 838
 839 static void filter_bitmap_blob_none(struct bitmap_index *bitmap_git,
 840                                     struct object_list *tip_objects,
 841                                     struct bitmap *to_filter)
 842 {
 843         filter_bitmap_exclude_type(bitmap_git, tip_objects, to_filter,
 844                                    OBJ_BLOB);
 845 }
 846
 847 static unsigned long get_size_by_pos(struct bitmap_index *bitmap_git,
 848                                      uint32_t pos)
 849 {
 850         struct packed_git *pack = bitmap_git->pack;
 851         unsigned long size;
 852         struct object_info oi = OBJECT_INFO_INIT;
 853
 854         oi.sizep = &size;
 855
 856         if (pos < pack->num_objects) {
 857                 off_t ofs = pack_pos_to_offset(pack, pos);
 858                 if (packed_object_info(the_repository, pack, ofs, &oi) < 0) {
 859                         struct object_id oid;
 860                         nth_packed_object_id(&oid, pack,
 861                                              pack_pos_to_index(pack, pos));
 862                         die(_("unable to get size of %s"), oid_to_hex(&oid));
 863                 }
 864         } else {
 865                 struct eindex *eindex = &bitmap_git->ext_index;
 866                 struct object *obj = eindex->objects[pos - pack->num_objects];
 867                 if (oid_object_info_extended(the_repository, &obj->oid, &oi, 0) < 0)
 868                         die(_("unable to get size of %s"), oid_to_hex(&obj->oid));
 869         }
 870
 871         return size;
 872 }
 873
 874 static void filter_bitmap_blob_limit(struct bitmap_index *bitmap_git,
 875                                      struct object_list *tip_objects,
 876                                      struct bitmap *to_filter,
 877                                      unsigned long limit)
 878 {
 879         struct eindex *eindex = &bitmap_git->ext_index;
 880         struct bitmap *tips;
 881         struct ewah_iterator it;
 882         eword_t mask;
 883         uint32_t i;
 884
 885         tips = find_tip_objects(bitmap_git, tip_objects, OBJ_BLOB);
 886
 887         for (i = 0, init_type_iterator(&it, bitmap_git, OBJ_BLOB);
 888              i < to_filter->word_alloc && ewah_iterator_next(&mask, &it);
 889              i++) {
 890                 eword_t word = to_filter->words[i] & mask;
 891                 unsigned offset;
 892
 893                 for (offset = 0; offset < BITS_IN_EWORD; offset++) {
 894                         uint32_t pos;
 895
 896                         if ((word >> offset) == 0)
 897                                 break;
 898                         offset += ewah_bit_ctz64(word >> offset);
 899                         pos = i * BITS_IN_EWORD + offset;
 900
 901                         if (!bitmap_get(tips, pos) &&
 902                             get_size_by_pos(bitmap_git, pos) >= limit)
 903                                 bitmap_unset(to_filter, pos);
 904                 }
 905         }
 906
 907         for (i = 0; i < eindex->count; i++) {
 908                 uint32_t pos = i + bitmap_git->pack->num_objects;
 909                 if (eindex->objects[i]->type == OBJ_BLOB &&
 910                     bitmap_get(to_filter, pos) &&
 911                     !bitmap_get(tips, pos) &&
 912                     get_size_by_pos(bitmap_git, pos) >= limit)
 913                         bitmap_unset(to_filter, pos);
 914         }
 915
 916         bitmap_free(tips);
 917 }
 918
 919 static void filter_bitmap_tree_depth(struct bitmap_index *bitmap_git,
 920                                      struct object_list *tip_objects,
 921                                      struct bitmap *to_filter,
 922                                      unsigned long limit)
 923 {
 924         if (limit)
 925                 BUG("filter_bitmap_tree_depth given non-zero limit");
 926
 927         filter_bitmap_exclude_type(bitmap_git, tip_objects, to_filter,
 928                                    OBJ_TREE);
 929         filter_bitmap_exclude_type(bitmap_git, tip_objects, to_filter,
 930                                    OBJ_BLOB);
 931 }
 932
 933 static void filter_bitmap_object_type(struct bitmap_index *bitmap_git,
 934                                       struct object_list *tip_objects,
 935                                       struct bitmap *to_filter,
 936                                       enum object_type object_type)
 937 {
 938         if (object_type < OBJ_COMMIT || object_type > OBJ_TAG)
 939                 BUG("filter_bitmap_object_type given invalid object");
 940
 941         if (object_type != OBJ_TAG)
 942                 filter_bitmap_exclude_type(bitmap_git, tip_objects, to_filter, OBJ_TAG);
 943         if (object_type != OBJ_COMMIT)
 944                 filter_bitmap_exclude_type(bitmap_git, tip_objects, to_filter, OBJ_COMMIT);
 945         if (object_type != OBJ_TREE)
 946                 filter_bitmap_exclude_type(bitmap_git, tip_objects, to_filter, OBJ_TREE);
 947         if (object_type != OBJ_BLOB)
 948                 filter_bitmap_exclude_type(bitmap_git, tip_objects, to_filter, OBJ_BLOB);
 949 }
 950
 951 static int filter_bitmap(struct bitmap_index *bitmap_git,
 952                          struct object_list *tip_objects,
 953                          struct bitmap *to_filter,
 954                          struct list_objects_filter_options *filter)
 955 {
 956         if (!filter || filter->choice == LOFC_DISABLED)
 957                 return 0;
 958
 959         if (filter->choice == LOFC_BLOB_NONE) {
 960                 if (bitmap_git)
 961                         filter_bitmap_blob_none(bitmap_git, tip_objects,
 962                                                 to_filter);
 963                 return 0;
 964         }
 965
 966         if (filter->choice == LOFC_BLOB_LIMIT) {
 967                 if (bitmap_git)
 968                         filter_bitmap_blob_limit(bitmap_git, tip_objects,
 969                                                  to_filter,
 970                                                  filter->blob_limit_value);
 971                 return 0;
 972         }
 973
 974         if (filter->choice == LOFC_TREE_DEPTH &&
 975             filter->tree_exclude_depth == 0) {
 976                 if (bitmap_git)
 977                         filter_bitmap_tree_depth(bitmap_git, tip_objects,
 978                                                  to_filter,
 979                                                  filter->tree_exclude_depth);
 980                 return 0;
 981         }
 982
 983         if (filter->choice == LOFC_OBJECT_TYPE) {
 984                 if (bitmap_git)
 985                         filter_bitmap_object_type(bitmap_git, tip_objects,
 986                                                   to_filter,
 987                                                   filter->object_type);
 988                 return 0;
 989         }
 990
 991         if (filter->choice == LOFC_COMBINE) {
 992                 int i;
 993                 for (i = 0; i < filter->sub_nr; i++) {
 994                         if (filter_bitmap(bitmap_git, tip_objects, to_filter,
 995                                           &filter->sub[i]) < 0)
 996                                 return -1;
 997                 }
 998                 return 0;
 999         }
1000
1001         /* filter choice not handled */
1002         return -1;
1003 }
1004
1005 static int can_filter_bitmap(struct list_objects_filter_options *filter)
1006 {
1007         return !filter_bitmap(NULL, NULL, NULL, filter);
1008 }
1009
1010 struct bitmap_index *prepare_bitmap_walk(struct rev_info *revs,
1011                                          struct list_objects_filter_options *filter,
1012                                          int filter_provided_objects)
1013 {
1014         unsigned int i;
1015
1016         struct object_list *wants = NULL;
1017         struct object_list *haves = NULL;
1018
1019         struct bitmap *wants_bitmap = NULL;
1020         struct bitmap *haves_bitmap = NULL;
1021
1022         struct bitmap_index *bitmap_git;
1023
1024         /*
1025          * We can't do pathspec limiting with bitmaps, because we don't know
1026          * which commits are associated with which object changes (let alone
1027          * even which objects are associated with which paths).
1028          */
1029         if (revs->prune)
1030                 return NULL;
1031
1032         if (!can_filter_bitmap(filter))
1033                 return NULL;
1034
1035         /* try to open a bitmapped pack, but don't parse it yet
1036          * because we may not need to use it */
1037         CALLOC_ARRAY(bitmap_git, 1);
1038         if (open_pack_bitmap(revs->repo, bitmap_git) < 0)
1039                 goto cleanup;
1040
1041         for (i = 0; i < revs->pending.nr; ++i) {
1042                 struct object *object = revs->pending.objects[i].item;
1043
1044                 if (object->type == OBJ_NONE)
1045                         parse_object_or_die(&object->oid, NULL);
1046
1047                 while (object->type == OBJ_TAG) {
1048                         struct tag *tag = (struct tag *) object;
1049
1050                         if (object->flags & UNINTERESTING)
1051                                 object_list_insert(object, &haves);
1052                         else
1053                                 object_list_insert(object, &wants);
1054
1055                         object = parse_object_or_die(get_tagged_oid(tag), NULL);
1056                         object->flags |= (tag->object.flags & UNINTERESTING);
1057                 }
1058
1059                 if (object->flags & UNINTERESTING)
1060                         object_list_insert(object, &haves);
1061                 else
1062                         object_list_insert(object, &wants);
1063         }
1064
1065         /*
1066          * if we have a HAVES list, but none of those haves is contained
1067          * in the packfile that has a bitmap, we don't have anything to
1068          * optimize here
1069          */
1070         if (haves && !in_bitmapped_pack(bitmap_git, haves))
1071                 goto cleanup;
1072
1073         /* if we don't want anything, we're done here */
1074         if (!wants)
1075                 goto cleanup;
1076
1077         /*
1078          * now we're going to use bitmaps, so load the actual bitmap entries
1079          * from disk. this is the point of no return; after this the rev_list
1080          * becomes invalidated and we must perform the revwalk through bitmaps
1081          */
1082         if (load_pack_bitmap(bitmap_git) < 0)
1083                 goto cleanup;
1084
1085         object_array_clear(&revs->pending);
1086
1087         if (haves) {
1088                 revs->ignore_missing_links = 1;
1089                 haves_bitmap = find_objects(bitmap_git, revs, haves, NULL,
1090                                             filter);
1091                 reset_revision_walk();
1092                 revs->ignore_missing_links = 0;
1093
1094                 if (haves_bitmap == NULL)
1095                         BUG("failed to perform bitmap walk");
1096         }
1097
1098         wants_bitmap = find_objects(bitmap_git, revs, wants, haves_bitmap,
1099                                     filter);
1100
1101         if (!wants_bitmap)
1102                 BUG("failed to perform bitmap walk");
1103
1104         if (haves_bitmap)
1105                 bitmap_and_not(wants_bitmap, haves_bitmap);
1106
1107         filter_bitmap(bitmap_git, (filter && filter_provided_objects) ? NULL : wants,
1108                       wants_bitmap, filter);
1109
1110         bitmap_git->result = wants_bitmap;
1111         bitmap_git->haves = haves_bitmap;
1112
1113         object_list_free(&wants);
1114         object_list_free(&haves);
1115
1116         return bitmap_git;
1117
1118 cleanup:
1119         free_bitmap_index(bitmap_git);
1120         object_list_free(&wants);
1121         object_list_free(&haves);
1122         return NULL;
1123 }
1124
1125 static void try_partial_reuse(struct bitmap_index *bitmap_git,
1126                               size_t pos,
1127                               struct bitmap *reuse,
1128                               struct pack_window **w_curs)
1129 {
1130         off_t offset, header;
1131         enum object_type type;
1132         unsigned long size;
1133
1134         if (pos >= bitmap_git->pack->num_objects)
1135                 return; /* not actually in the pack */
1136
1137         offset = header = pack_pos_to_offset(bitmap_git->pack, pos);
1138         type = unpack_object_header(bitmap_git->pack, w_curs, &offset, &size);
1139         if (type < 0)
1140                 return; /* broken packfile, punt */
1141
1142         if (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA) {
1143                 off_t base_offset;
1144                 uint32_t base_pos;
1145
1146                 /*
1147                  * Find the position of the base object so we can look it up
1148                  * in our bitmaps. If we can't come up with an offset, or if
1149                  * that offset is not in the revidx, the pack is corrupt.
1150                  * There's nothing we can do, so just punt on this object,
1151                  * and the normal slow path will complain about it in
1152                  * more detail.
1153                  */
1154                 base_offset = get_delta_base(bitmap_git->pack, w_curs,
1155                                              &offset, type, header);
1156                 if (!base_offset)
1157                         return;
1158                 if (offset_to_pack_pos(bitmap_git->pack, base_offset, &base_pos) < 0)
1159                         return;
1160
1161                 /*
1162                  * We assume delta dependencies always point backwards. This
1163                  * lets us do a single pass, and is basically always true
1164                  * due to the way OFS_DELTAs work. You would not typically
1165                  * find REF_DELTA in a bitmapped pack, since we only bitmap
1166                  * packs we write fresh, and OFS_DELTA is the default). But
1167                  * let's double check to make sure the pack wasn't written with
1168                  * odd parameters.
1169                  */
1170                 if (base_pos >= pos)
1171                         return;
1172
1173                 /*
1174                  * And finally, if we're not sending the base as part of our
1175                  * reuse chunk, then don't send this object either. The base
1176                  * would come after us, along with other objects not
1177                  * necessarily in the pack, which means we'd need to convert
1178                  * to REF_DELTA on the fly. Better to just let the normal
1179                  * object_entry code path handle it.
1180                  */
1181                 if (!bitmap_get(reuse, base_pos))
1182                         return;
1183         }
1184
1185         /*
1186          * If we got here, then the object is OK to reuse. Mark it.
1187          */
1188         bitmap_set(reuse, pos);
1189 }
1190
1191 int reuse_partial_packfile_from_bitmap(struct bitmap_index *bitmap_git,
1192                                        struct packed_git **packfile_out,
1193                                        uint32_t *entries,
1194                                        struct bitmap **reuse_out)
1195 {
1196         struct bitmap *result = bitmap_git->result;
1197         struct bitmap *reuse;
1198         struct pack_window *w_curs = NULL;
1199         size_t i = 0;
1200         uint32_t offset;
1201
1202         assert(result);
1203
1204         while (i < result->word_alloc && result->words[i] == (eword_t)~0)
1205                 i++;
1206
1207         /* Don't mark objects not in the packfile */
1208         if (i > bitmap_git->pack->num_objects / BITS_IN_EWORD)
1209                 i = bitmap_git->pack->num_objects / BITS_IN_EWORD;
1210
1211         reuse = bitmap_word_alloc(i);
1212         memset(reuse->words, 0xFF, i * sizeof(eword_t));
1213
1214         for (; i < result->word_alloc; ++i) {
1215                 eword_t word = result->words[i];
1216                 size_t pos = (i * BITS_IN_EWORD);
1217
1218                 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
1219                         if ((word >> offset) == 0)
1220                                 break;
1221
1222                         offset += ewah_bit_ctz64(word >> offset);
1223                         try_partial_reuse(bitmap_git, pos + offset, reuse, &w_curs);
1224                 }
1225         }
1226
1227         unuse_pack(&w_curs);
1228
1229         *entries = bitmap_popcount(reuse);
1230         if (!*entries) {
1231                 bitmap_free(reuse);
1232                 return -1;
1233         }
1234
1235         /*
1236          * Drop any reused objects from the result, since they will not
1237          * need to be handled separately.
1238          */
1239         bitmap_and_not(result, reuse);
1240         *packfile_out = bitmap_git->pack;
1241         *reuse_out = reuse;
1242         return 0;
1243 }
1244
1245 int bitmap_walk_contains(struct bitmap_index *bitmap_git,
1246                          struct bitmap *bitmap, const struct object_id *oid)
1247 {
1248         int idx;
1249
1250         if (!bitmap)
1251                 return 0;
1252
1253         idx = bitmap_position(bitmap_git, oid);
1254         return idx >= 0 && bitmap_get(bitmap, idx);
1255 }
1256
1257 void traverse_bitmap_commit_list(struct bitmap_index *bitmap_git,
1258                                  struct rev_info *revs,
1259                                  show_reachable_fn show_reachable)
1260 {
1261         assert(bitmap_git->result);
1262
1263         show_objects_for_type(bitmap_git, OBJ_COMMIT, show_reachable);
1264         if (revs->tree_objects)
1265                 show_objects_for_type(bitmap_git, OBJ_TREE, show_reachable);
1266         if (revs->blob_objects)
1267                 show_objects_for_type(bitmap_git, OBJ_BLOB, show_reachable);
1268         if (revs->tag_objects)
1269                 show_objects_for_type(bitmap_git, OBJ_TAG, show_reachable);
1270
1271         show_extended_objects(bitmap_git, revs, show_reachable);
1272 }
1273
1274 static uint32_t count_object_type(struct bitmap_index *bitmap_git,
1275                                   enum object_type type)
1276 {
1277         struct bitmap *objects = bitmap_git->result;
1278         struct eindex *eindex = &bitmap_git->ext_index;
1279
1280         uint32_t i = 0, count = 0;
1281         struct ewah_iterator it;
1282         eword_t filter;
1283
1284         init_type_iterator(&it, bitmap_git, type);
1285
1286         while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
1287                 eword_t word = objects->words[i++] & filter;
1288                 count += ewah_bit_popcount64(word);
1289         }
1290
1291         for (i = 0; i < eindex->count; ++i) {
1292                 if (eindex->objects[i]->type == type &&
1293                         bitmap_get(objects, bitmap_git->pack->num_objects + i))
1294                         count++;
1295         }
1296
1297         return count;
1298 }
1299
1300 void count_bitmap_commit_list(struct bitmap_index *bitmap_git,
1301                               uint32_t *commits, uint32_t *trees,
1302                               uint32_t *blobs, uint32_t *tags)
1303 {
1304         assert(bitmap_git->result);
1305
1306         if (commits)
1307                 *commits = count_object_type(bitmap_git, OBJ_COMMIT);
1308
1309         if (trees)
1310                 *trees = count_object_type(bitmap_git, OBJ_TREE);
1311
1312         if (blobs)
1313                 *blobs = count_object_type(bitmap_git, OBJ_BLOB);
1314
1315         if (tags)
1316                 *tags = count_object_type(bitmap_git, OBJ_TAG);
1317 }
1318
1319 struct bitmap_test_data {
1320         struct bitmap_index *bitmap_git;
1321         struct bitmap *base;
1322         struct progress *prg;
1323         size_t seen;
1324 };
1325
1326 static void test_show_object(struct object *object, const char *name,
1327                              void *data)
1328 {
1329         struct bitmap_test_data *tdata = data;
1330         int bitmap_pos;
1331
1332         bitmap_pos = bitmap_position(tdata->bitmap_git, &object->oid);
1333         if (bitmap_pos < 0)
1334                 die("Object not in bitmap: %s\n", oid_to_hex(&object->oid));
1335
1336         bitmap_set(tdata->base, bitmap_pos);
1337         display_progress(tdata->prg, ++tdata->seen);
1338 }
1339
1340 static void test_show_commit(struct commit *commit, void *data)
1341 {
1342         struct bitmap_test_data *tdata = data;
1343         int bitmap_pos;
1344
1345         bitmap_pos = bitmap_position(tdata->bitmap_git,
1346                                      &commit->object.oid);
1347         if (bitmap_pos < 0)
1348                 die("Object not in bitmap: %s\n", oid_to_hex(&commit->object.oid));
1349
1350         bitmap_set(tdata->base, bitmap_pos);
1351         display_progress(tdata->prg, ++tdata->seen);
1352 }
1353
1354 void test_bitmap_walk(struct rev_info *revs)
1355 {
1356         struct object *root;
1357         struct bitmap *result = NULL;
1358         size_t result_popcnt;
1359         struct bitmap_test_data tdata;
1360         struct bitmap_index *bitmap_git;
1361         struct ewah_bitmap *bm;
1362
1363         if (!(bitmap_git = prepare_bitmap_git(revs->repo)))
1364                 die("failed to load bitmap indexes");
1365
1366         if (revs->pending.nr != 1)
1367                 die("you must specify exactly one commit to test");
1368
1369         fprintf(stderr, "Bitmap v%d test (%d entries loaded)\n",
1370                 bitmap_git->version, bitmap_git->entry_count);
1371
1372         root = revs->pending.objects[0].item;
1373         bm = bitmap_for_commit(bitmap_git, (struct commit *)root);
1374
1375         if (bm) {
1376                 fprintf(stderr, "Found bitmap for %s. %d bits / %08x checksum\n",
1377                         oid_to_hex(&root->oid), (int)bm->bit_size, ewah_checksum(bm));
1378
1379                 result = ewah_to_bitmap(bm);
1380         }
1381
1382         if (result == NULL)
1383                 die("Commit %s doesn't have an indexed bitmap", oid_to_hex(&root->oid));
1384
1385         revs->tag_objects = 1;
1386         revs->tree_objects = 1;
1387         revs->blob_objects = 1;
1388
1389         result_popcnt = bitmap_popcount(result);
1390
1391         if (prepare_revision_walk(revs))
1392                 die("revision walk setup failed");
1393
1394         tdata.bitmap_git = bitmap_git;
1395         tdata.base = bitmap_new();
1396         tdata.prg = start_progress("Verifying bitmap entries", result_popcnt);
1397         tdata.seen = 0;
1398
1399         traverse_commit_list(revs, &test_show_commit, &test_show_object, &tdata);
1400
1401         stop_progress(&tdata.prg);
1402
1403         if (bitmap_equals(result, tdata.base))
1404                 fprintf(stderr, "OK!\n");
1405         else
1406                 die("mismatch in bitmap results");
1407
1408         free_bitmap_index(bitmap_git);
1409 }
1410
1411 int test_bitmap_commits(struct repository *r)
1412 {
1413         struct bitmap_index *bitmap_git = prepare_bitmap_git(r);
1414         struct object_id oid;
1415         MAYBE_UNUSED void *value;
1416
1417         if (!bitmap_git)
1418                 die("failed to load bitmap indexes");
1419
1420         kh_foreach(bitmap_git->bitmaps, oid, value, {
1421                 printf("%s\n", oid_to_hex(&oid));
1422         });
1423
1424         free_bitmap_index(bitmap_git);
1425
1426         return 0;
1427 }
1428
1429 int rebuild_bitmap(const uint32_t *reposition,
1430                    struct ewah_bitmap *source,
1431                    struct bitmap *dest)
1432 {
1433         uint32_t pos = 0;
1434         struct ewah_iterator it;
1435         eword_t word;
1436
1437         ewah_iterator_init(&it, source);
1438
1439         while (ewah_iterator_next(&word, &it)) {
1440                 uint32_t offset, bit_pos;
1441
1442                 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
1443                         if ((word >> offset) == 0)
1444                                 break;
1445
1446                         offset += ewah_bit_ctz64(word >> offset);
1447
1448                         bit_pos = reposition[pos + offset];
1449                         if (bit_pos > 0)
1450                                 bitmap_set(dest, bit_pos - 1);
1451                         else /* can't reuse, we don't have the object */
1452                                 return -1;
1453                 }
1454
1455                 pos += BITS_IN_EWORD;
1456         }
1457         return 0;
1458 }
1459
1460 uint32_t *create_bitmap_mapping(struct bitmap_index *bitmap_git,
1461                                 struct packing_data *mapping)
1462 {
1463         uint32_t i, num_objects;
1464         uint32_t *reposition;
1465
1466         num_objects = bitmap_git->pack->num_objects;
1467         CALLOC_ARRAY(reposition, num_objects);
1468
1469         for (i = 0; i < num_objects; ++i) {
1470                 struct object_id oid;
1471                 struct object_entry *oe;
1472
1473                 nth_packed_object_id(&oid, bitmap_git->pack,
1474                                      pack_pos_to_index(bitmap_git->pack, i));
1475                 oe = packlist_find(mapping, &oid);
1476
1477                 if (oe)
1478                         reposition[i] = oe_in_pack_pos(mapping, oe) + 1;
1479         }
1480
1481         return reposition;
1482 }
1483
1484 void free_bitmap_index(struct bitmap_index *b)
1485 {
1486         if (!b)
1487                 return;
1488
1489         if (b->map)
1490                 munmap(b->map, b->map_size);
1491         ewah_pool_free(b->commits);
1492         ewah_pool_free(b->trees);
1493         ewah_pool_free(b->blobs);
1494         ewah_pool_free(b->tags);
1495         kh_destroy_oid_map(b->bitmaps);
1496         free(b->ext_index.objects);
1497         free(b->ext_index.hashes);
1498         bitmap_free(b->result);
1499         bitmap_free(b->haves);
1500         free(b);
1501 }
1502
1503 int bitmap_has_oid_in_uninteresting(struct bitmap_index *bitmap_git,
1504                                     const struct object_id *oid)
1505 {
1506         return bitmap_git &&
1507                 bitmap_walk_contains(bitmap_git, bitmap_git->haves, oid);
1508 }
1509
1510 static off_t get_disk_usage_for_type(struct bitmap_index *bitmap_git,
1511                                      enum object_type object_type)
1512 {
1513         struct bitmap *result = bitmap_git->result;
1514         struct packed_git *pack = bitmap_git->pack;
1515         off_t total = 0;
1516         struct ewah_iterator it;
1517         eword_t filter;
1518         size_t i;
1519
1520         init_type_iterator(&it, bitmap_git, object_type);
1521         for (i = 0; i < result->word_alloc &&
1522                         ewah_iterator_next(&filter, &it); i++) {
1523                 eword_t word = result->words[i] & filter;
1524                 size_t base = (i * BITS_IN_EWORD);
1525                 unsigned offset;
1526
1527                 if (!word)
1528                         continue;
1529
1530                 for (offset = 0; offset < BITS_IN_EWORD; offset++) {
1531                         size_t pos;
1532
1533                         if ((word >> offset) == 0)
1534                                 break;
1535
1536                         offset += ewah_bit_ctz64(word >> offset);
1537                         pos = base + offset;
1538                         total += pack_pos_to_offset(pack, pos + 1) -
1539                                  pack_pos_to_offset(pack, pos);
1540                 }
1541         }
1542
1543         return total;
1544 }
1545
1546 static off_t get_disk_usage_for_extended(struct bitmap_index *bitmap_git)
1547 {
1548         struct bitmap *result = bitmap_git->result;
1549         struct packed_git *pack = bitmap_git->pack;
1550         struct eindex *eindex = &bitmap_git->ext_index;
1551         off_t total = 0;
1552         struct object_info oi = OBJECT_INFO_INIT;
1553         off_t object_size;
1554         size_t i;
1555
1556         oi.disk_sizep = &object_size;
1557
1558         for (i = 0; i < eindex->count; i++) {
1559                 struct object *obj = eindex->objects[i];
1560
1561                 if (!bitmap_get(result, pack->num_objects + i))
1562                         continue;
1563
1564                 if (oid_object_info_extended(the_repository, &obj->oid, &oi, 0) < 0)
1565                         die(_("unable to get disk usage of %s"),
1566                             oid_to_hex(&obj->oid));
1567
1568                 total += object_size;
1569         }
1570         return total;
1571 }
1572
1573 off_t get_disk_usage_from_bitmap(struct bitmap_index *bitmap_git,
1574                                  struct rev_info *revs)
1575 {
1576         off_t total = 0;
1577
1578         total += get_disk_usage_for_type(bitmap_git, OBJ_COMMIT);
1579         if (revs->tree_objects)
1580                 total += get_disk_usage_for_type(bitmap_git, OBJ_TREE);
1581         if (revs->blob_objects)
1582                 total += get_disk_usage_for_type(bitmap_git, OBJ_BLOB);
1583         if (revs->tag_objects)
1584                 total += get_disk_usage_for_type(bitmap_git, OBJ_TAG);
1585
1586         total += get_disk_usage_for_extended(bitmap_git);
1587
1588         return total;
1589 }
1590
1591 const struct string_list *bitmap_preferred_tips(struct repository *r)
1592 {
1593         return repo_config_get_value_multi(r, "pack.preferbitmaptips");
1594 }