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
  17 /*
  18  * An entry on the bitmap index, representing the bitmap for a given
  19  * commit.
  20  */
  21 struct stored_bitmap {
  22         struct object_id oid;
  23         struct ewah_bitmap *root;
  24         struct stored_bitmap *xor;
  25         int flags;
  26 };
  27
  28 /*
  29  * The active bitmap index for a repository. By design, repositories only have
  30  * a single bitmap index available (the index for the biggest packfile in
  31  * the repository), since bitmap indexes need full closure.
  32  *
  33  * If there is more than one bitmap index available (e.g. because of alternates),
  34  * the active bitmap index is the largest one.
  35  */
  36 struct bitmap_index {
  37         /* Packfile to which this bitmap index belongs to */
  38         struct packed_git *pack;
  39
  40         /*
  41          * Mark the first `reuse_objects` in the packfile as reused:
  42          * they will be sent as-is without using them for repacking
  43          * calculations
  44          */
  45         uint32_t reuse_objects;
  46
  47         /* mmapped buffer of the whole bitmap index */
  48         unsigned char *map;
  49         size_t map_size; /* size of the mmaped buffer */
  50         size_t map_pos; /* current position when loading the index */
  51
  52         /*
  53          * Type indexes.
  54          *
  55          * Each bitmap marks which objects in the packfile  are of the given
  56          * type. This provides type information when yielding the objects from
  57          * the packfile during a walk, which allows for better delta bases.
  58          */
  59         struct ewah_bitmap *commits;
  60         struct ewah_bitmap *trees;
  61         struct ewah_bitmap *blobs;
  62         struct ewah_bitmap *tags;
  63
  64         /* Map from object ID -> `stored_bitmap` for all the bitmapped commits */
  65         kh_oid_map_t *bitmaps;
  66
  67         /* Number of bitmapped commits */
  68         uint32_t entry_count;
  69
  70         /* If not NULL, this is a name-hash cache pointing into map. */
  71         uint32_t *hashes;
  72
  73         /*
  74          * Extended index.
  75          *
  76          * When trying to perform bitmap operations with objects that are not
  77          * packed in `pack`, these objects are added to this "fake index" and
  78          * are assumed to appear at the end of the packfile for all operations
  79          */
  80         struct eindex {
  81                 struct object **objects;
  82                 uint32_t *hashes;
  83                 uint32_t count, alloc;
  84                 kh_oid_pos_t *positions;
  85         } ext_index;
  86
  87         /* Bitmap result of the last performed walk */
  88         struct bitmap *result;
  89
  90         /* "have" bitmap from the last performed walk */
  91         struct bitmap *haves;
  92
  93         /* Version of the bitmap index */
  94         unsigned int version;
  95 };
  96
  97 static struct ewah_bitmap *lookup_stored_bitmap(struct stored_bitmap *st)
  98 {
  99         struct ewah_bitmap *parent;
 100         struct ewah_bitmap *composed;
 101
 102         if (st->xor == NULL)
 103                 return st->root;
 104
 105         composed = ewah_pool_new();
 106         parent = lookup_stored_bitmap(st->xor);
 107         ewah_xor(st->root, parent, composed);
 108
 109         ewah_pool_free(st->root);
 110         st->root = composed;
 111         st->xor = NULL;
 112
 113         return composed;
 114 }
 115
 116 /*
 117  * Read a bitmap from the current read position on the mmaped
 118  * index, and increase the read position accordingly
 119  */
 120 static struct ewah_bitmap *read_bitmap_1(struct bitmap_index *index)
 121 {
 122         struct ewah_bitmap *b = ewah_pool_new();
 123
 124         ssize_t bitmap_size = ewah_read_mmap(b,
 125                 index->map + index->map_pos,
 126                 index->map_size - index->map_pos);
 127
 128         if (bitmap_size < 0) {
 129                 error("Failed to load bitmap index (corrupted?)");
 130                 ewah_pool_free(b);
 131                 return NULL;
 132         }
 133
 134         index->map_pos += bitmap_size;
 135         return b;
 136 }
 137
 138 static int load_bitmap_header(struct bitmap_index *index)
 139 {
 140         struct bitmap_disk_header *header = (void *)index->map;
 141
 142         if (index->map_size < sizeof(*header) + the_hash_algo->rawsz)
 143                 return error("Corrupted bitmap index (missing header data)");
 144
 145         if (memcmp(header->magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE)) != 0)
 146                 return error("Corrupted bitmap index file (wrong header)");
 147
 148         index->version = ntohs(header->version);
 149         if (index->version != 1)
 150                 return error("Unsupported version for bitmap index file (%d)", index->version);
 151
 152         /* Parse known bitmap format options */
 153         {
 154                 uint32_t flags = ntohs(header->options);
 155
 156                 if ((flags & BITMAP_OPT_FULL_DAG) == 0)
 157                         return error("Unsupported options for bitmap index file "
 158                                 "(Git requires BITMAP_OPT_FULL_DAG)");
 159
 160                 if (flags & BITMAP_OPT_HASH_CACHE) {
 161                         unsigned char *end = index->map + index->map_size - the_hash_algo->rawsz;
 162                         index->hashes = ((uint32_t *)end) - index->pack->num_objects;
 163                 }
 164         }
 165
 166         index->entry_count = ntohl(header->entry_count);
 167         index->map_pos += sizeof(*header) - GIT_MAX_RAWSZ + the_hash_algo->rawsz;
 168         return 0;
 169 }
 170
 171 static struct stored_bitmap *store_bitmap(struct bitmap_index *index,
 172                                           struct ewah_bitmap *root,
 173                                           const unsigned char *hash,
 174                                           struct stored_bitmap *xor_with,
 175                                           int flags)
 176 {
 177         struct stored_bitmap *stored;
 178         khiter_t hash_pos;
 179         int ret;
 180
 181         stored = xmalloc(sizeof(struct stored_bitmap));
 182         stored->root = root;
 183         stored->xor = xor_with;
 184         stored->flags = flags;
 185         oidread(&stored->oid, hash);
 186
 187         hash_pos = kh_put_oid_map(index->bitmaps, stored->oid, &ret);
 188
 189         /* a 0 return code means the insertion succeeded with no changes,
 190          * because the SHA1 already existed on the map. this is bad, there
 191          * shouldn't be duplicated commits in the index */
 192         if (ret == 0) {
 193                 error("Duplicate entry in bitmap index: %s", hash_to_hex(hash));
 194                 return NULL;
 195         }
 196
 197         kh_value(index->bitmaps, hash_pos) = stored;
 198         return stored;
 199 }
 200
 201 static inline uint32_t read_be32(const unsigned char *buffer, size_t *pos)
 202 {
 203         uint32_t result = get_be32(buffer + *pos);
 204         (*pos) += sizeof(result);
 205         return result;
 206 }
 207
 208 static inline uint8_t read_u8(const unsigned char *buffer, size_t *pos)
 209 {
 210         return buffer[(*pos)++];
 211 }
 212
 213 #define MAX_XOR_OFFSET 160
 214
 215 static int load_bitmap_entries_v1(struct bitmap_index *index)
 216 {
 217         uint32_t i;
 218         struct stored_bitmap *recent_bitmaps[MAX_XOR_OFFSET] = { NULL };
 219
 220         for (i = 0; i < index->entry_count; ++i) {
 221                 int xor_offset, flags;
 222                 struct ewah_bitmap *bitmap = NULL;
 223                 struct stored_bitmap *xor_bitmap = NULL;
 224                 uint32_t commit_idx_pos;
 225                 const unsigned char *sha1;
 226
 227                 commit_idx_pos = read_be32(index->map, &index->map_pos);
 228                 xor_offset = read_u8(index->map, &index->map_pos);
 229                 flags = read_u8(index->map, &index->map_pos);
 230
 231                 sha1 = nth_packed_object_sha1(index->pack, commit_idx_pos);
 232
 233                 bitmap = read_bitmap_1(index);
 234                 if (!bitmap)
 235                         return -1;
 236
 237                 if (xor_offset > MAX_XOR_OFFSET || xor_offset > i)
 238                         return error("Corrupted bitmap pack index");
 239
 240                 if (xor_offset > 0) {
 241                         xor_bitmap = recent_bitmaps[(i - xor_offset) % MAX_XOR_OFFSET];
 242
 243                         if (xor_bitmap == NULL)
 244                                 return error("Invalid XOR offset in bitmap pack index");
 245                 }
 246
 247                 recent_bitmaps[i % MAX_XOR_OFFSET] = store_bitmap(
 248                         index, bitmap, sha1, xor_bitmap, flags);
 249         }
 250
 251         return 0;
 252 }
 253
 254 static char *pack_bitmap_filename(struct packed_git *p)
 255 {
 256         size_t len;
 257
 258         if (!strip_suffix(p->pack_name, ".pack", &len))
 259                 BUG("pack_name does not end in .pack");
 260         return xstrfmt("%.*s.bitmap", (int)len, p->pack_name);
 261 }
 262
 263 static int open_pack_bitmap_1(struct bitmap_index *bitmap_git, struct packed_git *packfile)
 264 {
 265         int fd;
 266         struct stat st;
 267         char *idx_name;
 268
 269         if (open_pack_index(packfile))
 270                 return -1;
 271
 272         idx_name = pack_bitmap_filename(packfile);
 273         fd = git_open(idx_name);
 274         free(idx_name);
 275
 276         if (fd < 0)
 277                 return -1;
 278
 279         if (fstat(fd, &st)) {
 280                 close(fd);
 281                 return -1;
 282         }
 283
 284         if (bitmap_git->pack) {
 285                 warning("ignoring extra bitmap file: %s", packfile->pack_name);
 286                 close(fd);
 287                 return -1;
 288         }
 289
 290         bitmap_git->pack = packfile;
 291         bitmap_git->map_size = xsize_t(st.st_size);
 292         bitmap_git->map = xmmap(NULL, bitmap_git->map_size, PROT_READ, MAP_PRIVATE, fd, 0);
 293         bitmap_git->map_pos = 0;
 294         close(fd);
 295
 296         if (load_bitmap_header(bitmap_git) < 0) {
 297                 munmap(bitmap_git->map, bitmap_git->map_size);
 298                 bitmap_git->map = NULL;
 299                 bitmap_git->map_size = 0;
 300                 return -1;
 301         }
 302
 303         return 0;
 304 }
 305
 306 static int load_pack_bitmap(struct bitmap_index *bitmap_git)
 307 {
 308         assert(bitmap_git->map);
 309
 310         bitmap_git->bitmaps = kh_init_oid_map();
 311         bitmap_git->ext_index.positions = kh_init_oid_pos();
 312         if (load_pack_revindex(bitmap_git->pack))
 313                 goto failed;
 314
 315         if (!(bitmap_git->commits = read_bitmap_1(bitmap_git)) ||
 316                 !(bitmap_git->trees = read_bitmap_1(bitmap_git)) ||
 317                 !(bitmap_git->blobs = read_bitmap_1(bitmap_git)) ||
 318                 !(bitmap_git->tags = read_bitmap_1(bitmap_git)))
 319                 goto failed;
 320
 321         if (load_bitmap_entries_v1(bitmap_git) < 0)
 322                 goto failed;
 323
 324         return 0;
 325
 326 failed:
 327         munmap(bitmap_git->map, bitmap_git->map_size);
 328         bitmap_git->map = NULL;
 329         bitmap_git->map_size = 0;
 330         return -1;
 331 }
 332
 333 static int open_pack_bitmap(struct repository *r,
 334                             struct bitmap_index *bitmap_git)
 335 {
 336         struct packed_git *p;
 337         int ret = -1;
 338
 339         assert(!bitmap_git->map);
 340
 341         for (p = get_all_packs(r); p; p = p->next) {
 342                 if (open_pack_bitmap_1(bitmap_git, p) == 0)
 343                         ret = 0;
 344         }
 345
 346         return ret;
 347 }
 348
 349 struct bitmap_index *prepare_bitmap_git(struct repository *r)
 350 {
 351         struct bitmap_index *bitmap_git = xcalloc(1, sizeof(*bitmap_git));
 352
 353         if (!open_pack_bitmap(r, bitmap_git) && !load_pack_bitmap(bitmap_git))
 354                 return bitmap_git;
 355
 356         free_bitmap_index(bitmap_git);
 357         return NULL;
 358 }
 359
 360 struct include_data {
 361         struct bitmap_index *bitmap_git;
 362         struct bitmap *base;
 363         struct bitmap *seen;
 364 };
 365
 366 static inline int bitmap_position_extended(struct bitmap_index *bitmap_git,
 367                                            const struct object_id *oid)
 368 {
 369         kh_oid_pos_t *positions = bitmap_git->ext_index.positions;
 370         khiter_t pos = kh_get_oid_pos(positions, *oid);
 371
 372         if (pos < kh_end(positions)) {
 373                 int bitmap_pos = kh_value(positions, pos);
 374                 return bitmap_pos + bitmap_git->pack->num_objects;
 375         }
 376
 377         return -1;
 378 }
 379
 380 static inline int bitmap_position_packfile(struct bitmap_index *bitmap_git,
 381                                            const struct object_id *oid)
 382 {
 383         off_t offset = find_pack_entry_one(oid->hash, bitmap_git->pack);
 384         if (!offset)
 385                 return -1;
 386
 387         return find_revindex_position(bitmap_git->pack, offset);
 388 }
 389
 390 static int bitmap_position(struct bitmap_index *bitmap_git,
 391                            const struct object_id *oid)
 392 {
 393         int pos = bitmap_position_packfile(bitmap_git, oid);
 394         return (pos >= 0) ? pos : bitmap_position_extended(bitmap_git, oid);
 395 }
 396
 397 static int ext_index_add_object(struct bitmap_index *bitmap_git,
 398                                 struct object *object, const char *name)
 399 {
 400         struct eindex *eindex = &bitmap_git->ext_index;
 401
 402         khiter_t hash_pos;
 403         int hash_ret;
 404         int bitmap_pos;
 405
 406         hash_pos = kh_put_oid_pos(eindex->positions, object->oid, &hash_ret);
 407         if (hash_ret > 0) {
 408                 if (eindex->count >= eindex->alloc) {
 409                         eindex->alloc = (eindex->alloc + 16) * 3 / 2;
 410                         REALLOC_ARRAY(eindex->objects, eindex->alloc);
 411                         REALLOC_ARRAY(eindex->hashes, eindex->alloc);
 412                 }
 413
 414                 bitmap_pos = eindex->count;
 415                 eindex->objects[eindex->count] = object;
 416                 eindex->hashes[eindex->count] = pack_name_hash(name);
 417                 kh_value(eindex->positions, hash_pos) = bitmap_pos;
 418                 eindex->count++;
 419         } else {
 420                 bitmap_pos = kh_value(eindex->positions, hash_pos);
 421         }
 422
 423         return bitmap_pos + bitmap_git->pack->num_objects;
 424 }
 425
 426 struct bitmap_show_data {
 427         struct bitmap_index *bitmap_git;
 428         struct bitmap *base;
 429 };
 430
 431 static void show_object(struct object *object, const char *name, void *data_)
 432 {
 433         struct bitmap_show_data *data = data_;
 434         int bitmap_pos;
 435
 436         bitmap_pos = bitmap_position(data->bitmap_git, &object->oid);
 437
 438         if (bitmap_pos < 0)
 439                 bitmap_pos = ext_index_add_object(data->bitmap_git, object,
 440                                                   name);
 441
 442         bitmap_set(data->base, bitmap_pos);
 443 }
 444
 445 static void show_commit(struct commit *commit, void *data)
 446 {
 447 }
 448
 449 static int add_to_include_set(struct bitmap_index *bitmap_git,
 450                               struct include_data *data,
 451                               const struct object_id *oid,
 452                               int bitmap_pos)
 453 {
 454         khiter_t hash_pos;
 455
 456         if (data->seen && bitmap_get(data->seen, bitmap_pos))
 457                 return 0;
 458
 459         if (bitmap_get(data->base, bitmap_pos))
 460                 return 0;
 461
 462         hash_pos = kh_get_oid_map(bitmap_git->bitmaps, *oid);
 463         if (hash_pos < kh_end(bitmap_git->bitmaps)) {
 464                 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, hash_pos);
 465                 bitmap_or_ewah(data->base, lookup_stored_bitmap(st));
 466                 return 0;
 467         }
 468
 469         bitmap_set(data->base, bitmap_pos);
 470         return 1;
 471 }
 472
 473 static int should_include(struct commit *commit, void *_data)
 474 {
 475         struct include_data *data = _data;
 476         int bitmap_pos;
 477
 478         bitmap_pos = bitmap_position(data->bitmap_git, &commit->object.oid);
 479         if (bitmap_pos < 0)
 480                 bitmap_pos = ext_index_add_object(data->bitmap_git,
 481                                                   (struct object *)commit,
 482                                                   NULL);
 483
 484         if (!add_to_include_set(data->bitmap_git, data, &commit->object.oid,
 485                                 bitmap_pos)) {
 486                 struct commit_list *parent = commit->parents;
 487
 488                 while (parent) {
 489                         parent->item->object.flags |= SEEN;
 490                         parent = parent->next;
 491                 }
 492
 493                 return 0;
 494         }
 495
 496         return 1;
 497 }
 498
 499 static struct bitmap *find_objects(struct bitmap_index *bitmap_git,
 500                                    struct rev_info *revs,
 501                                    struct object_list *roots,
 502                                    struct bitmap *seen)
 503 {
 504         struct bitmap *base = NULL;
 505         int needs_walk = 0;
 506
 507         struct object_list *not_mapped = NULL;
 508
 509         /*
 510          * Go through all the roots for the walk. The ones that have bitmaps
 511          * on the bitmap index will be `or`ed together to form an initial
 512          * global reachability analysis.
 513          *
 514          * The ones without bitmaps in the index will be stored in the
 515          * `not_mapped_list` for further processing.
 516          */
 517         while (roots) {
 518                 struct object *object = roots->item;
 519                 roots = roots->next;
 520
 521                 if (object->type == OBJ_COMMIT) {
 522                         khiter_t pos = kh_get_oid_map(bitmap_git->bitmaps, object->oid);
 523
 524                         if (pos < kh_end(bitmap_git->bitmaps)) {
 525                                 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, pos);
 526                                 struct ewah_bitmap *or_with = lookup_stored_bitmap(st);
 527
 528                                 if (base == NULL)
 529                                         base = ewah_to_bitmap(or_with);
 530                                 else
 531                                         bitmap_or_ewah(base, or_with);
 532
 533                                 object->flags |= SEEN;
 534                                 continue;
 535                         }
 536                 }
 537
 538                 object_list_insert(object, &not_mapped);
 539         }
 540
 541         /*
 542          * Best case scenario: We found bitmaps for all the roots,
 543          * so the resulting `or` bitmap has the full reachability analysis
 544          */
 545         if (not_mapped == NULL)
 546                 return base;
 547
 548         roots = not_mapped;
 549
 550         /*
 551          * Let's iterate through all the roots that don't have bitmaps to
 552          * check if we can determine them to be reachable from the existing
 553          * global bitmap.
 554          *
 555          * If we cannot find them in the existing global bitmap, we'll need
 556          * to push them to an actual walk and run it until we can confirm
 557          * they are reachable
 558          */
 559         while (roots) {
 560                 struct object *object = roots->item;
 561                 int pos;
 562
 563                 roots = roots->next;
 564                 pos = bitmap_position(bitmap_git, &object->oid);
 565
 566                 if (pos < 0 || base == NULL || !bitmap_get(base, pos)) {
 567                         object->flags &= ~UNINTERESTING;
 568                         add_pending_object(revs, object, "");
 569                         needs_walk = 1;
 570                 } else {
 571                         object->flags |= SEEN;
 572                 }
 573         }
 574
 575         if (needs_walk) {
 576                 struct include_data incdata;
 577                 struct bitmap_show_data show_data;
 578
 579                 if (base == NULL)
 580                         base = bitmap_new();
 581
 582                 incdata.bitmap_git = bitmap_git;
 583                 incdata.base = base;
 584                 incdata.seen = seen;
 585
 586                 revs->include_check = should_include;
 587                 revs->include_check_data = &incdata;
 588
 589                 if (prepare_revision_walk(revs))
 590                         die("revision walk setup failed");
 591
 592                 show_data.bitmap_git = bitmap_git;
 593                 show_data.base = base;
 594
 595                 traverse_commit_list(revs, show_commit, show_object,
 596                                      &show_data);
 597         }
 598
 599         return base;
 600 }
 601
 602 static void show_extended_objects(struct bitmap_index *bitmap_git,
 603                                   struct rev_info *revs,
 604                                   show_reachable_fn show_reach)
 605 {
 606         struct bitmap *objects = bitmap_git->result;
 607         struct eindex *eindex = &bitmap_git->ext_index;
 608         uint32_t i;
 609
 610         for (i = 0; i < eindex->count; ++i) {
 611                 struct object *obj;
 612
 613                 if (!bitmap_get(objects, bitmap_git->pack->num_objects + i))
 614                         continue;
 615
 616                 obj = eindex->objects[i];
 617                 if ((obj->type == OBJ_BLOB && !revs->blob_objects) ||
 618                     (obj->type == OBJ_TREE && !revs->tree_objects) ||
 619                     (obj->type == OBJ_TAG && !revs->tag_objects))
 620                         continue;
 621
 622                 show_reach(&obj->oid, obj->type, 0, eindex->hashes[i], NULL, 0);
 623         }
 624 }
 625
 626 static void init_type_iterator(struct ewah_iterator *it,
 627                                struct bitmap_index *bitmap_git,
 628                                enum object_type type)
 629 {
 630         switch (type) {
 631         case OBJ_COMMIT:
 632                 ewah_iterator_init(it, bitmap_git->commits);
 633                 break;
 634
 635         case OBJ_TREE:
 636                 ewah_iterator_init(it, bitmap_git->trees);
 637                 break;
 638
 639         case OBJ_BLOB:
 640                 ewah_iterator_init(it, bitmap_git->blobs);
 641                 break;
 642
 643         case OBJ_TAG:
 644                 ewah_iterator_init(it, bitmap_git->tags);
 645                 break;
 646
 647         default:
 648                 BUG("object type %d not stored by bitmap type index", type);
 649                 break;
 650         }
 651 }
 652
 653 static void show_objects_for_type(
 654         struct bitmap_index *bitmap_git,
 655         enum object_type object_type,
 656         show_reachable_fn show_reach)
 657 {
 658         size_t pos = 0, i = 0;
 659         uint32_t offset;
 660
 661         struct ewah_iterator it;
 662         eword_t filter;
 663
 664         struct bitmap *objects = bitmap_git->result;
 665
 666         if (bitmap_git->reuse_objects == bitmap_git->pack->num_objects)
 667                 return;
 668
 669         init_type_iterator(&it, bitmap_git, object_type);
 670
 671         while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
 672                 eword_t word = objects->words[i] & filter;
 673
 674                 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
 675                         struct object_id oid;
 676                         struct revindex_entry *entry;
 677                         uint32_t hash = 0;
 678
 679                         if ((word >> offset) == 0)
 680                                 break;
 681
 682                         offset += ewah_bit_ctz64(word >> offset);
 683
 684                         if (pos + offset < bitmap_git->reuse_objects)
 685                                 continue;
 686
 687                         entry = &bitmap_git->pack->revindex[pos + offset];
 688                         nth_packed_object_oid(&oid, bitmap_git->pack, entry->nr);
 689
 690                         if (bitmap_git->hashes)
 691                                 hash = get_be32(bitmap_git->hashes + entry->nr);
 692
 693                         show_reach(&oid, object_type, 0, hash, bitmap_git->pack, entry->offset);
 694                 }
 695
 696                 pos += BITS_IN_EWORD;
 697                 i++;
 698         }
 699 }
 700
 701 static int in_bitmapped_pack(struct bitmap_index *bitmap_git,
 702                              struct object_list *roots)
 703 {
 704         while (roots) {
 705                 struct object *object = roots->item;
 706                 roots = roots->next;
 707
 708                 if (find_pack_entry_one(object->oid.hash, bitmap_git->pack) > 0)
 709                         return 1;
 710         }
 711
 712         return 0;
 713 }
 714
 715 static struct bitmap *find_tip_blobs(struct bitmap_index *bitmap_git,
 716                                      struct object_list *tip_objects)
 717 {
 718         struct bitmap *result = bitmap_new();
 719         struct object_list *p;
 720
 721         for (p = tip_objects; p; p = p->next) {
 722                 int pos;
 723
 724                 if (p->item->type != OBJ_BLOB)
 725                         continue;
 726
 727                 pos = bitmap_position(bitmap_git, &p->item->oid);
 728                 if (pos < 0)
 729                         continue;
 730
 731                 bitmap_set(result, pos);
 732         }
 733
 734         return result;
 735 }
 736
 737 static void filter_bitmap_blob_none(struct bitmap_index *bitmap_git,
 738                                     struct object_list *tip_objects,
 739                                     struct bitmap *to_filter)
 740 {
 741         struct eindex *eindex = &bitmap_git->ext_index;
 742         struct bitmap *tips;
 743         struct ewah_iterator it;
 744         eword_t mask;
 745         uint32_t i;
 746
 747         /*
 748          * The non-bitmap version of this filter never removes
 749          * blobs which the other side specifically asked for,
 750          * so we must match that behavior.
 751          */
 752         tips = find_tip_blobs(bitmap_git, tip_objects);
 753
 754         /*
 755          * We can use the blob type-bitmap to work in whole words
 756          * for the objects that are actually in the bitmapped packfile.
 757          */
 758         for (i = 0, init_type_iterator(&it, bitmap_git, OBJ_BLOB);
 759              i < to_filter->word_alloc && ewah_iterator_next(&mask, &it);
 760              i++) {
 761                 if (i < tips->word_alloc)
 762                         mask &= ~tips->words[i];
 763                 to_filter->words[i] &= ~mask;
 764         }
 765
 766         /*
 767          * Clear any blobs that weren't in the packfile (and so would not have
 768          * been caught by the loop above. We'll have to check them
 769          * individually.
 770          */
 771         for (i = 0; i < eindex->count; i++) {
 772                 uint32_t pos = i + bitmap_git->pack->num_objects;
 773                 if (eindex->objects[i]->type == OBJ_BLOB &&
 774                     bitmap_get(to_filter, pos) &&
 775                     !bitmap_get(tips, pos))
 776                         bitmap_unset(to_filter, pos);
 777         }
 778
 779         bitmap_free(tips);
 780 }
 781
 782 static unsigned long get_size_by_pos(struct bitmap_index *bitmap_git,
 783                                      uint32_t pos)
 784 {
 785         struct packed_git *pack = bitmap_git->pack;
 786         unsigned long size;
 787         struct object_info oi = OBJECT_INFO_INIT;
 788
 789         oi.sizep = &size;
 790
 791         if (pos < pack->num_objects) {
 792                 struct revindex_entry *entry = &pack->revindex[pos];
 793                 if (packed_object_info(the_repository, pack,
 794                                        entry->offset, &oi) < 0) {
 795                         struct object_id oid;
 796                         nth_packed_object_oid(&oid, pack, entry->nr);
 797                         die(_("unable to get size of %s"), oid_to_hex(&oid));
 798                 }
 799         } else {
 800                 struct eindex *eindex = &bitmap_git->ext_index;
 801                 struct object *obj = eindex->objects[pos - pack->num_objects];
 802                 if (oid_object_info_extended(the_repository, &obj->oid, &oi, 0) < 0)
 803                         die(_("unable to get size of %s"), oid_to_hex(&obj->oid));
 804         }
 805
 806         return size;
 807 }
 808
 809 static void filter_bitmap_blob_limit(struct bitmap_index *bitmap_git,
 810                                      struct object_list *tip_objects,
 811                                      struct bitmap *to_filter,
 812                                      unsigned long limit)
 813 {
 814         struct eindex *eindex = &bitmap_git->ext_index;
 815         struct bitmap *tips;
 816         struct ewah_iterator it;
 817         eword_t mask;
 818         uint32_t i;
 819
 820         tips = find_tip_blobs(bitmap_git, tip_objects);
 821
 822         for (i = 0, init_type_iterator(&it, bitmap_git, OBJ_BLOB);
 823              i < to_filter->word_alloc && ewah_iterator_next(&mask, &it);
 824              i++) {
 825                 eword_t word = to_filter->words[i] & mask;
 826                 unsigned offset;
 827
 828                 for (offset = 0; offset < BITS_IN_EWORD; offset++) {
 829                         uint32_t pos;
 830
 831                         if ((word >> offset) == 0)
 832                                 break;
 833                         offset += ewah_bit_ctz64(word >> offset);
 834                         pos = i * BITS_IN_EWORD + offset;
 835
 836                         if (!bitmap_get(tips, pos) &&
 837                             get_size_by_pos(bitmap_git, pos) >= limit)
 838                                 bitmap_unset(to_filter, pos);
 839                 }
 840         }
 841
 842         for (i = 0; i < eindex->count; i++) {
 843                 uint32_t pos = i + bitmap_git->pack->num_objects;
 844                 if (eindex->objects[i]->type == OBJ_BLOB &&
 845                     bitmap_get(to_filter, pos) &&
 846                     !bitmap_get(tips, pos) &&
 847                     get_size_by_pos(bitmap_git, pos) >= limit)
 848                         bitmap_unset(to_filter, pos);
 849         }
 850
 851         bitmap_free(tips);
 852 }
 853
 854 static int filter_bitmap(struct bitmap_index *bitmap_git,
 855                          struct object_list *tip_objects,
 856                          struct bitmap *to_filter,
 857                          struct list_objects_filter_options *filter)
 858 {
 859         if (!filter || filter->choice == LOFC_DISABLED)
 860                 return 0;
 861
 862         if (filter->choice == LOFC_BLOB_NONE) {
 863                 if (bitmap_git)
 864                         filter_bitmap_blob_none(bitmap_git, tip_objects,
 865                                                 to_filter);
 866                 return 0;
 867         }
 868
 869         if (filter->choice == LOFC_BLOB_LIMIT) {
 870                 if (bitmap_git)
 871                         filter_bitmap_blob_limit(bitmap_git, tip_objects,
 872                                                  to_filter,
 873                                                  filter->blob_limit_value);
 874                 return 0;
 875         }
 876
 877         /* filter choice not handled */
 878         return -1;
 879 }
 880
 881 static int can_filter_bitmap(struct list_objects_filter_options *filter)
 882 {
 883         return !filter_bitmap(NULL, NULL, NULL, filter);
 884 }
 885
 886 struct bitmap_index *prepare_bitmap_walk(struct rev_info *revs,
 887                                          struct list_objects_filter_options *filter)
 888 {
 889         unsigned int i;
 890
 891         struct object_list *wants = NULL;
 892         struct object_list *haves = NULL;
 893
 894         struct bitmap *wants_bitmap = NULL;
 895         struct bitmap *haves_bitmap = NULL;
 896
 897         struct bitmap_index *bitmap_git;
 898
 899         /*
 900          * We can't do pathspec limiting with bitmaps, because we don't know
 901          * which commits are associated with which object changes (let alone
 902          * even which objects are associated with which paths).
 903          */
 904         if (revs->prune)
 905                 return NULL;
 906
 907         if (!can_filter_bitmap(filter))
 908                 return NULL;
 909
 910         /* try to open a bitmapped pack, but don't parse it yet
 911          * because we may not need to use it */
 912         bitmap_git = xcalloc(1, sizeof(*bitmap_git));
 913         if (open_pack_bitmap(revs->repo, bitmap_git) < 0)
 914                 goto cleanup;
 915
 916         for (i = 0; i < revs->pending.nr; ++i) {
 917                 struct object *object = revs->pending.objects[i].item;
 918
 919                 if (object->type == OBJ_NONE)
 920                         parse_object_or_die(&object->oid, NULL);
 921
 922                 while (object->type == OBJ_TAG) {
 923                         struct tag *tag = (struct tag *) object;
 924
 925                         if (object->flags & UNINTERESTING)
 926                                 object_list_insert(object, &haves);
 927                         else
 928                                 object_list_insert(object, &wants);
 929
 930                         object = parse_object_or_die(get_tagged_oid(tag), NULL);
 931                 }
 932
 933                 if (object->flags & UNINTERESTING)
 934                         object_list_insert(object, &haves);
 935                 else
 936                         object_list_insert(object, &wants);
 937         }
 938
 939         /*
 940          * if we have a HAVES list, but none of those haves is contained
 941          * in the packfile that has a bitmap, we don't have anything to
 942          * optimize here
 943          */
 944         if (haves && !in_bitmapped_pack(bitmap_git, haves))
 945                 goto cleanup;
 946
 947         /* if we don't want anything, we're done here */
 948         if (!wants)
 949                 goto cleanup;
 950
 951         /*
 952          * now we're going to use bitmaps, so load the actual bitmap entries
 953          * from disk. this is the point of no return; after this the rev_list
 954          * becomes invalidated and we must perform the revwalk through bitmaps
 955          */
 956         if (load_pack_bitmap(bitmap_git) < 0)
 957                 goto cleanup;
 958
 959         object_array_clear(&revs->pending);
 960
 961         if (haves) {
 962                 revs->ignore_missing_links = 1;
 963                 haves_bitmap = find_objects(bitmap_git, revs, haves, NULL);
 964                 reset_revision_walk();
 965                 revs->ignore_missing_links = 0;
 966
 967                 if (haves_bitmap == NULL)
 968                         BUG("failed to perform bitmap walk");
 969         }
 970
 971         wants_bitmap = find_objects(bitmap_git, revs, wants, haves_bitmap);
 972
 973         if (!wants_bitmap)
 974                 BUG("failed to perform bitmap walk");
 975
 976         if (haves_bitmap)
 977                 bitmap_and_not(wants_bitmap, haves_bitmap);
 978
 979         filter_bitmap(bitmap_git, wants, wants_bitmap, filter);
 980
 981         bitmap_git->result = wants_bitmap;
 982         bitmap_git->haves = haves_bitmap;
 983
 984         object_list_free(&wants);
 985         object_list_free(&haves);
 986
 987         return bitmap_git;
 988
 989 cleanup:
 990         free_bitmap_index(bitmap_git);
 991         object_list_free(&wants);
 992         object_list_free(&haves);
 993         return NULL;
 994 }
 995
 996 int reuse_partial_packfile_from_bitmap(struct bitmap_index *bitmap_git,
 997                                        struct packed_git **packfile,
 998                                        uint32_t *entries,
 999                                        off_t *up_to)
1000 {
1001         /*
1002          * Reuse the packfile content if we need more than
1003          * 90% of its objects
1004          */
1005         static const double REUSE_PERCENT = 0.9;
1006
1007         struct bitmap *result = bitmap_git->result;
1008         uint32_t reuse_threshold;
1009         uint32_t i, reuse_objects = 0;
1010
1011         assert(result);
1012
1013         for (i = 0; i < result->word_alloc; ++i) {
1014                 if (result->words[i] != (eword_t)~0) {
1015                         reuse_objects += ewah_bit_ctz64(~result->words[i]);
1016                         break;
1017                 }
1018
1019                 reuse_objects += BITS_IN_EWORD;
1020         }
1021
1022 #ifdef GIT_BITMAP_DEBUG
1023         {
1024                 const unsigned char *sha1;
1025                 struct revindex_entry *entry;
1026
1027                 entry = &bitmap_git->reverse_index->revindex[reuse_objects];
1028                 sha1 = nth_packed_object_sha1(bitmap_git->pack, entry->nr);
1029
1030                 fprintf(stderr, "Failed to reuse at %d (%016llx)\n",
1031                         reuse_objects, result->words[i]);
1032                 fprintf(stderr, " %s\n", hash_to_hex(sha1));
1033         }
1034 #endif
1035
1036         if (!reuse_objects)
1037                 return -1;
1038
1039         if (reuse_objects >= bitmap_git->pack->num_objects) {
1040                 bitmap_git->reuse_objects = *entries = bitmap_git->pack->num_objects;
1041                 *up_to = -1; /* reuse the full pack */
1042                 *packfile = bitmap_git->pack;
1043                 return 0;
1044         }
1045
1046         reuse_threshold = bitmap_popcount(bitmap_git->result) * REUSE_PERCENT;
1047
1048         if (reuse_objects < reuse_threshold)
1049                 return -1;
1050
1051         bitmap_git->reuse_objects = *entries = reuse_objects;
1052         *up_to = bitmap_git->pack->revindex[reuse_objects].offset;
1053         *packfile = bitmap_git->pack;
1054
1055         return 0;
1056 }
1057
1058 void traverse_bitmap_commit_list(struct bitmap_index *bitmap_git,
1059                                  struct rev_info *revs,
1060                                  show_reachable_fn show_reachable)
1061 {
1062         assert(bitmap_git->result);
1063
1064         show_objects_for_type(bitmap_git, OBJ_COMMIT, show_reachable);
1065         if (revs->tree_objects)
1066                 show_objects_for_type(bitmap_git, OBJ_TREE, show_reachable);
1067         if (revs->blob_objects)
1068                 show_objects_for_type(bitmap_git, OBJ_BLOB, show_reachable);
1069         if (revs->tag_objects)
1070                 show_objects_for_type(bitmap_git, OBJ_TAG, show_reachable);
1071
1072         show_extended_objects(bitmap_git, revs, show_reachable);
1073 }
1074
1075 static uint32_t count_object_type(struct bitmap_index *bitmap_git,
1076                                   enum object_type type)
1077 {
1078         struct bitmap *objects = bitmap_git->result;
1079         struct eindex *eindex = &bitmap_git->ext_index;
1080
1081         uint32_t i = 0, count = 0;
1082         struct ewah_iterator it;
1083         eword_t filter;
1084
1085         init_type_iterator(&it, bitmap_git, type);
1086
1087         while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
1088                 eword_t word = objects->words[i++] & filter;
1089                 count += ewah_bit_popcount64(word);
1090         }
1091
1092         for (i = 0; i < eindex->count; ++i) {
1093                 if (eindex->objects[i]->type == type &&
1094                         bitmap_get(objects, bitmap_git->pack->num_objects + i))
1095                         count++;
1096         }
1097
1098         return count;
1099 }
1100
1101 void count_bitmap_commit_list(struct bitmap_index *bitmap_git,
1102                               uint32_t *commits, uint32_t *trees,
1103                               uint32_t *blobs, uint32_t *tags)
1104 {
1105         assert(bitmap_git->result);
1106
1107         if (commits)
1108                 *commits = count_object_type(bitmap_git, OBJ_COMMIT);
1109
1110         if (trees)
1111                 *trees = count_object_type(bitmap_git, OBJ_TREE);
1112
1113         if (blobs)
1114                 *blobs = count_object_type(bitmap_git, OBJ_BLOB);
1115
1116         if (tags)
1117                 *tags = count_object_type(bitmap_git, OBJ_TAG);
1118 }
1119
1120 struct bitmap_test_data {
1121         struct bitmap_index *bitmap_git;
1122         struct bitmap *base;
1123         struct progress *prg;
1124         size_t seen;
1125 };
1126
1127 static void test_show_object(struct object *object, const char *name,
1128                              void *data)
1129 {
1130         struct bitmap_test_data *tdata = data;
1131         int bitmap_pos;
1132
1133         bitmap_pos = bitmap_position(tdata->bitmap_git, &object->oid);
1134         if (bitmap_pos < 0)
1135                 die("Object not in bitmap: %s\n", oid_to_hex(&object->oid));
1136
1137         bitmap_set(tdata->base, bitmap_pos);
1138         display_progress(tdata->prg, ++tdata->seen);
1139 }
1140
1141 static void test_show_commit(struct commit *commit, void *data)
1142 {
1143         struct bitmap_test_data *tdata = data;
1144         int bitmap_pos;
1145
1146         bitmap_pos = bitmap_position(tdata->bitmap_git,
1147                                      &commit->object.oid);
1148         if (bitmap_pos < 0)
1149                 die("Object not in bitmap: %s\n", oid_to_hex(&commit->object.oid));
1150
1151         bitmap_set(tdata->base, bitmap_pos);
1152         display_progress(tdata->prg, ++tdata->seen);
1153 }
1154
1155 void test_bitmap_walk(struct rev_info *revs)
1156 {
1157         struct object *root;
1158         struct bitmap *result = NULL;
1159         khiter_t pos;
1160         size_t result_popcnt;
1161         struct bitmap_test_data tdata;
1162         struct bitmap_index *bitmap_git;
1163
1164         if (!(bitmap_git = prepare_bitmap_git(revs->repo)))
1165                 die("failed to load bitmap indexes");
1166
1167         if (revs->pending.nr != 1)
1168                 die("you must specify exactly one commit to test");
1169
1170         fprintf(stderr, "Bitmap v%d test (%d entries loaded)\n",
1171                 bitmap_git->version, bitmap_git->entry_count);
1172
1173         root = revs->pending.objects[0].item;
1174         pos = kh_get_oid_map(bitmap_git->bitmaps, root->oid);
1175
1176         if (pos < kh_end(bitmap_git->bitmaps)) {
1177                 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, pos);
1178                 struct ewah_bitmap *bm = lookup_stored_bitmap(st);
1179
1180                 fprintf(stderr, "Found bitmap for %s. %d bits / %08x checksum\n",
1181                         oid_to_hex(&root->oid), (int)bm->bit_size, ewah_checksum(bm));
1182
1183                 result = ewah_to_bitmap(bm);
1184         }
1185
1186         if (result == NULL)
1187                 die("Commit %s doesn't have an indexed bitmap", oid_to_hex(&root->oid));
1188
1189         revs->tag_objects = 1;
1190         revs->tree_objects = 1;
1191         revs->blob_objects = 1;
1192
1193         result_popcnt = bitmap_popcount(result);
1194
1195         if (prepare_revision_walk(revs))
1196                 die("revision walk setup failed");
1197
1198         tdata.bitmap_git = bitmap_git;
1199         tdata.base = bitmap_new();
1200         tdata.prg = start_progress("Verifying bitmap entries", result_popcnt);
1201         tdata.seen = 0;
1202
1203         traverse_commit_list(revs, &test_show_commit, &test_show_object, &tdata);
1204
1205         stop_progress(&tdata.prg);
1206
1207         if (bitmap_equals(result, tdata.base))
1208                 fprintf(stderr, "OK!\n");
1209         else
1210                 fprintf(stderr, "Mismatch!\n");
1211
1212         free_bitmap_index(bitmap_git);
1213 }
1214
1215 static int rebuild_bitmap(uint32_t *reposition,
1216                           struct ewah_bitmap *source,
1217                           struct bitmap *dest)
1218 {
1219         uint32_t pos = 0;
1220         struct ewah_iterator it;
1221         eword_t word;
1222
1223         ewah_iterator_init(&it, source);
1224
1225         while (ewah_iterator_next(&word, &it)) {
1226                 uint32_t offset, bit_pos;
1227
1228                 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
1229                         if ((word >> offset) == 0)
1230                                 break;
1231
1232                         offset += ewah_bit_ctz64(word >> offset);
1233
1234                         bit_pos = reposition[pos + offset];
1235                         if (bit_pos > 0)
1236                                 bitmap_set(dest, bit_pos - 1);
1237                         else /* can't reuse, we don't have the object */
1238                                 return -1;
1239                 }
1240
1241                 pos += BITS_IN_EWORD;
1242         }
1243         return 0;
1244 }
1245
1246 int rebuild_existing_bitmaps(struct bitmap_index *bitmap_git,
1247                              struct packing_data *mapping,
1248                              kh_oid_map_t *reused_bitmaps,
1249                              int show_progress)
1250 {
1251         uint32_t i, num_objects;
1252         uint32_t *reposition;
1253         struct bitmap *rebuild;
1254         struct stored_bitmap *stored;
1255         struct progress *progress = NULL;
1256
1257         khiter_t hash_pos;
1258         int hash_ret;
1259
1260         num_objects = bitmap_git->pack->num_objects;
1261         reposition = xcalloc(num_objects, sizeof(uint32_t));
1262
1263         for (i = 0; i < num_objects; ++i) {
1264                 struct object_id oid;
1265                 struct revindex_entry *entry;
1266                 struct object_entry *oe;
1267
1268                 entry = &bitmap_git->pack->revindex[i];
1269                 nth_packed_object_oid(&oid, bitmap_git->pack, entry->nr);
1270                 oe = packlist_find(mapping, &oid);
1271
1272                 if (oe)
1273                         reposition[i] = oe_in_pack_pos(mapping, oe) + 1;
1274         }
1275
1276         rebuild = bitmap_new();
1277         i = 0;
1278
1279         if (show_progress)
1280                 progress = start_progress("Reusing bitmaps", 0);
1281
1282         kh_foreach_value(bitmap_git->bitmaps, stored, {
1283                 if (stored->flags & BITMAP_FLAG_REUSE) {
1284                         if (!rebuild_bitmap(reposition,
1285                                             lookup_stored_bitmap(stored),
1286                                             rebuild)) {
1287                                 hash_pos = kh_put_oid_map(reused_bitmaps,
1288                                                           stored->oid,
1289                                                           &hash_ret);
1290                                 kh_value(reused_bitmaps, hash_pos) =
1291                                         bitmap_to_ewah(rebuild);
1292                         }
1293                         bitmap_reset(rebuild);
1294                         display_progress(progress, ++i);
1295                 }
1296         });
1297
1298         stop_progress(&progress);
1299
1300         free(reposition);
1301         bitmap_free(rebuild);
1302         return 0;
1303 }
1304
1305 void free_bitmap_index(struct bitmap_index *b)
1306 {
1307         if (!b)
1308                 return;
1309
1310         if (b->map)
1311                 munmap(b->map, b->map_size);
1312         ewah_pool_free(b->commits);
1313         ewah_pool_free(b->trees);
1314         ewah_pool_free(b->blobs);
1315         ewah_pool_free(b->tags);
1316         kh_destroy_oid_map(b->bitmaps);
1317         free(b->ext_index.objects);
1318         free(b->ext_index.hashes);
1319         bitmap_free(b->result);
1320         bitmap_free(b->haves);
1321         free(b);
1322 }
1323
1324 int bitmap_has_oid_in_uninteresting(struct bitmap_index *bitmap_git,
1325                                     const struct object_id *oid)
1326 {
1327         int pos;
1328
1329         if (!bitmap_git)
1330                 return 0; /* no bitmap loaded */
1331         if (!bitmap_git->haves)
1332                 return 0; /* walk had no "haves" */
1333
1334         pos = bitmap_position_packfile(bitmap_git, oid);
1335         if (pos < 0)
1336                 return 0;
1337
1338         return bitmap_get(bitmap_git->haves, pos);
1339 }