pack-bitmap.con commit Merge branch 'jk/pack-revindex' (4fd1359)
   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
  13/*
  14 * An entry on the bitmap index, representing the bitmap for a given
  15 * commit.
  16 */
  17struct stored_bitmap {
  18        unsigned char sha1[20];
  19        struct ewah_bitmap *root;
  20        struct stored_bitmap *xor;
  21        int flags;
  22};
  23
  24/*
  25 * The currently active bitmap index. By design, repositories only have
  26 * a single bitmap index available (the index for the biggest packfile in
  27 * the repository), since bitmap indexes need full closure.
  28 *
  29 * If there is more than one bitmap index available (e.g. because of alternates),
  30 * the active bitmap index is the largest one.
  31 */
  32static struct bitmap_index {
  33        /* Packfile to which this bitmap index belongs to */
  34        struct packed_git *pack;
  35
  36        /*
  37         * Mark the first `reuse_objects` in the packfile as reused:
  38         * they will be sent as-is without using them for repacking
  39         * calculations
  40         */
  41        uint32_t reuse_objects;
  42
  43        /* mmapped buffer of the whole bitmap index */
  44        unsigned char *map;
  45        size_t map_size; /* size of the mmaped buffer */
  46        size_t map_pos; /* current position when loading the index */
  47
  48        /*
  49         * Type indexes.
  50         *
  51         * Each bitmap marks which objects in the packfile  are of the given
  52         * type. This provides type information when yielding the objects from
  53         * the packfile during a walk, which allows for better delta bases.
  54         */
  55        struct ewah_bitmap *commits;
  56        struct ewah_bitmap *trees;
  57        struct ewah_bitmap *blobs;
  58        struct ewah_bitmap *tags;
  59
  60        /* Map from SHA1 -> `stored_bitmap` for all the bitmapped commits */
  61        khash_sha1 *bitmaps;
  62
  63        /* Number of bitmapped commits */
  64        uint32_t entry_count;
  65
  66        /* Name-hash cache (or NULL if not present). */
  67        uint32_t *hashes;
  68
  69        /*
  70         * Extended index.
  71         *
  72         * When trying to perform bitmap operations with objects that are not
  73         * packed in `pack`, these objects are added to this "fake index" and
  74         * are assumed to appear at the end of the packfile for all operations
  75         */
  76        struct eindex {
  77                struct object **objects;
  78                uint32_t *hashes;
  79                uint32_t count, alloc;
  80                khash_sha1_pos *positions;
  81        } ext_index;
  82
  83        /* Bitmap result of the last performed walk */
  84        struct bitmap *result;
  85
  86        /* Version of the bitmap index */
  87        unsigned int version;
  88
  89        unsigned loaded : 1;
  90
  91} bitmap_git;
  92
  93static struct ewah_bitmap *lookup_stored_bitmap(struct stored_bitmap *st)
  94{
  95        struct ewah_bitmap *parent;
  96        struct ewah_bitmap *composed;
  97
  98        if (st->xor == NULL)
  99                return st->root;
 100
 101        composed = ewah_pool_new();
 102        parent = lookup_stored_bitmap(st->xor);
 103        ewah_xor(st->root, parent, composed);
 104
 105        ewah_pool_free(st->root);
 106        st->root = composed;
 107        st->xor = NULL;
 108
 109        return composed;
 110}
 111
 112/*
 113 * Read a bitmap from the current read position on the mmaped
 114 * index, and increase the read position accordingly
 115 */
 116static struct ewah_bitmap *read_bitmap_1(struct bitmap_index *index)
 117{
 118        struct ewah_bitmap *b = ewah_pool_new();
 119
 120        int bitmap_size = ewah_read_mmap(b,
 121                index->map + index->map_pos,
 122                index->map_size - index->map_pos);
 123
 124        if (bitmap_size < 0) {
 125                error("Failed to load bitmap index (corrupted?)");
 126                ewah_pool_free(b);
 127                return NULL;
 128        }
 129
 130        index->map_pos += bitmap_size;
 131        return b;
 132}
 133
 134static int load_bitmap_header(struct bitmap_index *index)
 135{
 136        struct bitmap_disk_header *header = (void *)index->map;
 137
 138        if (index->map_size < sizeof(*header) + 20)
 139                return error("Corrupted bitmap index (missing header data)");
 140
 141        if (memcmp(header->magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE)) != 0)
 142                return error("Corrupted bitmap index file (wrong header)");
 143
 144        index->version = ntohs(header->version);
 145        if (index->version != 1)
 146                return error("Unsupported version for bitmap index file (%d)", index->version);
 147
 148        /* Parse known bitmap format options */
 149        {
 150                uint32_t flags = ntohs(header->options);
 151
 152                if ((flags & BITMAP_OPT_FULL_DAG) == 0)
 153                        return error("Unsupported options for bitmap index file "
 154                                "(Git requires BITMAP_OPT_FULL_DAG)");
 155
 156                if (flags & BITMAP_OPT_HASH_CACHE) {
 157                        unsigned char *end = index->map + index->map_size - 20;
 158                        index->hashes = ((uint32_t *)end) - index->pack->num_objects;
 159                }
 160        }
 161
 162        index->entry_count = ntohl(header->entry_count);
 163        index->map_pos += sizeof(*header);
 164        return 0;
 165}
 166
 167static struct stored_bitmap *store_bitmap(struct bitmap_index *index,
 168                                          struct ewah_bitmap *root,
 169                                          const unsigned char *sha1,
 170                                          struct stored_bitmap *xor_with,
 171                                          int flags)
 172{
 173        struct stored_bitmap *stored;
 174        khiter_t hash_pos;
 175        int ret;
 176
 177        stored = xmalloc(sizeof(struct stored_bitmap));
 178        stored->root = root;
 179        stored->xor = xor_with;
 180        stored->flags = flags;
 181        hashcpy(stored->sha1, sha1);
 182
 183        hash_pos = kh_put_sha1(index->bitmaps, stored->sha1, &ret);
 184
 185        /* a 0 return code means the insertion succeeded with no changes,
 186         * because the SHA1 already existed on the map. this is bad, there
 187         * shouldn't be duplicated commits in the index */
 188        if (ret == 0) {
 189                error("Duplicate entry in bitmap index: %s", sha1_to_hex(sha1));
 190                return NULL;
 191        }
 192
 193        kh_value(index->bitmaps, hash_pos) = stored;
 194        return stored;
 195}
 196
 197static inline uint32_t read_be32(const unsigned char *buffer, size_t *pos)
 198{
 199        uint32_t result = get_be32(buffer + *pos);
 200        (*pos) += sizeof(result);
 201        return result;
 202}
 203
 204static inline uint8_t read_u8(const unsigned char *buffer, size_t *pos)
 205{
 206        return buffer[(*pos)++];
 207}
 208
 209#define MAX_XOR_OFFSET 160
 210
 211static int load_bitmap_entries_v1(struct bitmap_index *index)
 212{
 213        uint32_t i;
 214        struct stored_bitmap *recent_bitmaps[MAX_XOR_OFFSET] = { NULL };
 215
 216        for (i = 0; i < index->entry_count; ++i) {
 217                int xor_offset, flags;
 218                struct ewah_bitmap *bitmap = NULL;
 219                struct stored_bitmap *xor_bitmap = NULL;
 220                uint32_t commit_idx_pos;
 221                const unsigned char *sha1;
 222
 223                commit_idx_pos = read_be32(index->map, &index->map_pos);
 224                xor_offset = read_u8(index->map, &index->map_pos);
 225                flags = read_u8(index->map, &index->map_pos);
 226
 227                sha1 = nth_packed_object_sha1(index->pack, commit_idx_pos);
 228
 229                bitmap = read_bitmap_1(index);
 230                if (!bitmap)
 231                        return -1;
 232
 233                if (xor_offset > MAX_XOR_OFFSET || xor_offset > i)
 234                        return error("Corrupted bitmap pack index");
 235
 236                if (xor_offset > 0) {
 237                        xor_bitmap = recent_bitmaps[(i - xor_offset) % MAX_XOR_OFFSET];
 238
 239                        if (xor_bitmap == NULL)
 240                                return error("Invalid XOR offset in bitmap pack index");
 241                }
 242
 243                recent_bitmaps[i % MAX_XOR_OFFSET] = store_bitmap(
 244                        index, bitmap, sha1, xor_bitmap, flags);
 245        }
 246
 247        return 0;
 248}
 249
 250static char *pack_bitmap_filename(struct packed_git *p)
 251{
 252        size_t len;
 253
 254        if (!strip_suffix(p->pack_name, ".pack", &len))
 255                die("BUG: pack_name does not end in .pack");
 256        return xstrfmt("%.*s.bitmap", (int)len, p->pack_name);
 257}
 258
 259static int open_pack_bitmap_1(struct packed_git *packfile)
 260{
 261        int fd;
 262        struct stat st;
 263        char *idx_name;
 264
 265        if (open_pack_index(packfile))
 266                return -1;
 267
 268        idx_name = pack_bitmap_filename(packfile);
 269        fd = git_open_noatime(idx_name);
 270        free(idx_name);
 271
 272        if (fd < 0)
 273                return -1;
 274
 275        if (fstat(fd, &st)) {
 276                close(fd);
 277                return -1;
 278        }
 279
 280        if (bitmap_git.pack) {
 281                warning("ignoring extra bitmap file: %s", packfile->pack_name);
 282                close(fd);
 283                return -1;
 284        }
 285
 286        bitmap_git.pack = packfile;
 287        bitmap_git.map_size = xsize_t(st.st_size);
 288        bitmap_git.map = xmmap(NULL, bitmap_git.map_size, PROT_READ, MAP_PRIVATE, fd, 0);
 289        bitmap_git.map_pos = 0;
 290        close(fd);
 291
 292        if (load_bitmap_header(&bitmap_git) < 0) {
 293                munmap(bitmap_git.map, bitmap_git.map_size);
 294                bitmap_git.map = NULL;
 295                bitmap_git.map_size = 0;
 296                return -1;
 297        }
 298
 299        return 0;
 300}
 301
 302static int load_pack_bitmap(void)
 303{
 304        assert(bitmap_git.map && !bitmap_git.loaded);
 305
 306        bitmap_git.bitmaps = kh_init_sha1();
 307        bitmap_git.ext_index.positions = kh_init_sha1_pos();
 308        load_pack_revindex(bitmap_git.pack);
 309
 310        if (!(bitmap_git.commits = read_bitmap_1(&bitmap_git)) ||
 311                !(bitmap_git.trees = read_bitmap_1(&bitmap_git)) ||
 312                !(bitmap_git.blobs = read_bitmap_1(&bitmap_git)) ||
 313                !(bitmap_git.tags = read_bitmap_1(&bitmap_git)))
 314                goto failed;
 315
 316        if (load_bitmap_entries_v1(&bitmap_git) < 0)
 317                goto failed;
 318
 319        bitmap_git.loaded = 1;
 320        return 0;
 321
 322failed:
 323        munmap(bitmap_git.map, bitmap_git.map_size);
 324        bitmap_git.map = NULL;
 325        bitmap_git.map_size = 0;
 326        return -1;
 327}
 328
 329static int open_pack_bitmap(void)
 330{
 331        struct packed_git *p;
 332        int ret = -1;
 333
 334        assert(!bitmap_git.map && !bitmap_git.loaded);
 335
 336        prepare_packed_git();
 337        for (p = packed_git; p; p = p->next) {
 338                if (open_pack_bitmap_1(p) == 0)
 339                        ret = 0;
 340        }
 341
 342        return ret;
 343}
 344
 345int prepare_bitmap_git(void)
 346{
 347        if (bitmap_git.loaded)
 348                return 0;
 349
 350        if (!open_pack_bitmap())
 351                return load_pack_bitmap();
 352
 353        return -1;
 354}
 355
 356struct include_data {
 357        struct bitmap *base;
 358        struct bitmap *seen;
 359};
 360
 361static inline int bitmap_position_extended(const unsigned char *sha1)
 362{
 363        khash_sha1_pos *positions = bitmap_git.ext_index.positions;
 364        khiter_t pos = kh_get_sha1_pos(positions, sha1);
 365
 366        if (pos < kh_end(positions)) {
 367                int bitmap_pos = kh_value(positions, pos);
 368                return bitmap_pos + bitmap_git.pack->num_objects;
 369        }
 370
 371        return -1;
 372}
 373
 374static inline int bitmap_position_packfile(const unsigned char *sha1)
 375{
 376        off_t offset = find_pack_entry_one(sha1, bitmap_git.pack);
 377        if (!offset)
 378                return -1;
 379
 380        return find_revindex_position(bitmap_git.pack, offset);
 381}
 382
 383static int bitmap_position(const unsigned char *sha1)
 384{
 385        int pos = bitmap_position_packfile(sha1);
 386        return (pos >= 0) ? pos : bitmap_position_extended(sha1);
 387}
 388
 389static int ext_index_add_object(struct object *object, const char *name)
 390{
 391        struct eindex *eindex = &bitmap_git.ext_index;
 392
 393        khiter_t hash_pos;
 394        int hash_ret;
 395        int bitmap_pos;
 396
 397        hash_pos = kh_put_sha1_pos(eindex->positions, object->oid.hash, &hash_ret);
 398        if (hash_ret > 0) {
 399                if (eindex->count >= eindex->alloc) {
 400                        eindex->alloc = (eindex->alloc + 16) * 3 / 2;
 401                        REALLOC_ARRAY(eindex->objects, eindex->alloc);
 402                        REALLOC_ARRAY(eindex->hashes, eindex->alloc);
 403                }
 404
 405                bitmap_pos = eindex->count;
 406                eindex->objects[eindex->count] = object;
 407                eindex->hashes[eindex->count] = pack_name_hash(name);
 408                kh_value(eindex->positions, hash_pos) = bitmap_pos;
 409                eindex->count++;
 410        } else {
 411                bitmap_pos = kh_value(eindex->positions, hash_pos);
 412        }
 413
 414        return bitmap_pos + bitmap_git.pack->num_objects;
 415}
 416
 417static void show_object(struct object *object, const struct name_path *path,
 418                        const char *last, void *data)
 419{
 420        struct bitmap *base = data;
 421        int bitmap_pos;
 422
 423        bitmap_pos = bitmap_position(object->oid.hash);
 424
 425        if (bitmap_pos < 0) {
 426                char *name = path_name(path, last);
 427                bitmap_pos = ext_index_add_object(object, name);
 428                free(name);
 429        }
 430
 431        bitmap_set(base, bitmap_pos);
 432}
 433
 434static void show_commit(struct commit *commit, void *data)
 435{
 436}
 437
 438static int add_to_include_set(struct include_data *data,
 439                              const unsigned char *sha1,
 440                              int bitmap_pos)
 441{
 442        khiter_t hash_pos;
 443
 444        if (data->seen && bitmap_get(data->seen, bitmap_pos))
 445                return 0;
 446
 447        if (bitmap_get(data->base, bitmap_pos))
 448                return 0;
 449
 450        hash_pos = kh_get_sha1(bitmap_git.bitmaps, sha1);
 451        if (hash_pos < kh_end(bitmap_git.bitmaps)) {
 452                struct stored_bitmap *st = kh_value(bitmap_git.bitmaps, hash_pos);
 453                bitmap_or_ewah(data->base, lookup_stored_bitmap(st));
 454                return 0;
 455        }
 456
 457        bitmap_set(data->base, bitmap_pos);
 458        return 1;
 459}
 460
 461static int should_include(struct commit *commit, void *_data)
 462{
 463        struct include_data *data = _data;
 464        int bitmap_pos;
 465
 466        bitmap_pos = bitmap_position(commit->object.oid.hash);
 467        if (bitmap_pos < 0)
 468                bitmap_pos = ext_index_add_object((struct object *)commit, NULL);
 469
 470        if (!add_to_include_set(data, commit->object.oid.hash, bitmap_pos)) {
 471                struct commit_list *parent = commit->parents;
 472
 473                while (parent) {
 474                        parent->item->object.flags |= SEEN;
 475                        parent = parent->next;
 476                }
 477
 478                return 0;
 479        }
 480
 481        return 1;
 482}
 483
 484static struct bitmap *find_objects(struct rev_info *revs,
 485                                   struct object_list *roots,
 486                                   struct bitmap *seen)
 487{
 488        struct bitmap *base = NULL;
 489        int needs_walk = 0;
 490
 491        struct object_list *not_mapped = NULL;
 492
 493        /*
 494         * Go through all the roots for the walk. The ones that have bitmaps
 495         * on the bitmap index will be `or`ed together to form an initial
 496         * global reachability analysis.
 497         *
 498         * The ones without bitmaps in the index will be stored in the
 499         * `not_mapped_list` for further processing.
 500         */
 501        while (roots) {
 502                struct object *object = roots->item;
 503                roots = roots->next;
 504
 505                if (object->type == OBJ_COMMIT) {
 506                        khiter_t pos = kh_get_sha1(bitmap_git.bitmaps, object->oid.hash);
 507
 508                        if (pos < kh_end(bitmap_git.bitmaps)) {
 509                                struct stored_bitmap *st = kh_value(bitmap_git.bitmaps, pos);
 510                                struct ewah_bitmap *or_with = lookup_stored_bitmap(st);
 511
 512                                if (base == NULL)
 513                                        base = ewah_to_bitmap(or_with);
 514                                else
 515                                        bitmap_or_ewah(base, or_with);
 516
 517                                object->flags |= SEEN;
 518                                continue;
 519                        }
 520                }
 521
 522                object_list_insert(object, &not_mapped);
 523        }
 524
 525        /*
 526         * Best case scenario: We found bitmaps for all the roots,
 527         * so the resulting `or` bitmap has the full reachability analysis
 528         */
 529        if (not_mapped == NULL)
 530                return base;
 531
 532        roots = not_mapped;
 533
 534        /*
 535         * Let's iterate through all the roots that don't have bitmaps to
 536         * check if we can determine them to be reachable from the existing
 537         * global bitmap.
 538         *
 539         * If we cannot find them in the existing global bitmap, we'll need
 540         * to push them to an actual walk and run it until we can confirm
 541         * they are reachable
 542         */
 543        while (roots) {
 544                struct object *object = roots->item;
 545                int pos;
 546
 547                roots = roots->next;
 548                pos = bitmap_position(object->oid.hash);
 549
 550                if (pos < 0 || base == NULL || !bitmap_get(base, pos)) {
 551                        object->flags &= ~UNINTERESTING;
 552                        add_pending_object(revs, object, "");
 553                        needs_walk = 1;
 554                } else {
 555                        object->flags |= SEEN;
 556                }
 557        }
 558
 559        if (needs_walk) {
 560                struct include_data incdata;
 561
 562                if (base == NULL)
 563                        base = bitmap_new();
 564
 565                incdata.base = base;
 566                incdata.seen = seen;
 567
 568                revs->include_check = should_include;
 569                revs->include_check_data = &incdata;
 570
 571                if (prepare_revision_walk(revs))
 572                        die("revision walk setup failed");
 573
 574                traverse_commit_list(revs, show_commit, show_object, base);
 575        }
 576
 577        return base;
 578}
 579
 580static void show_extended_objects(struct bitmap *objects,
 581                                  show_reachable_fn show_reach)
 582{
 583        struct eindex *eindex = &bitmap_git.ext_index;
 584        uint32_t i;
 585
 586        for (i = 0; i < eindex->count; ++i) {
 587                struct object *obj;
 588
 589                if (!bitmap_get(objects, bitmap_git.pack->num_objects + i))
 590                        continue;
 591
 592                obj = eindex->objects[i];
 593                show_reach(obj->oid.hash, obj->type, 0, eindex->hashes[i], NULL, 0);
 594        }
 595}
 596
 597static void show_objects_for_type(
 598        struct bitmap *objects,
 599        struct ewah_bitmap *type_filter,
 600        enum object_type object_type,
 601        show_reachable_fn show_reach)
 602{
 603        size_t pos = 0, i = 0;
 604        uint32_t offset;
 605
 606        struct ewah_iterator it;
 607        eword_t filter;
 608
 609        if (bitmap_git.reuse_objects == bitmap_git.pack->num_objects)
 610                return;
 611
 612        ewah_iterator_init(&it, type_filter);
 613
 614        while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
 615                eword_t word = objects->words[i] & filter;
 616
 617                for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
 618                        const unsigned char *sha1;
 619                        struct revindex_entry *entry;
 620                        uint32_t hash = 0;
 621
 622                        if ((word >> offset) == 0)
 623                                break;
 624
 625                        offset += ewah_bit_ctz64(word >> offset);
 626
 627                        if (pos + offset < bitmap_git.reuse_objects)
 628                                continue;
 629
 630                        entry = &bitmap_git.pack->revindex[pos + offset];
 631                        sha1 = nth_packed_object_sha1(bitmap_git.pack, entry->nr);
 632
 633                        if (bitmap_git.hashes)
 634                                hash = ntohl(bitmap_git.hashes[entry->nr]);
 635
 636                        show_reach(sha1, object_type, 0, hash, bitmap_git.pack, entry->offset);
 637                }
 638
 639                pos += BITS_IN_EWORD;
 640                i++;
 641        }
 642}
 643
 644static int in_bitmapped_pack(struct object_list *roots)
 645{
 646        while (roots) {
 647                struct object *object = roots->item;
 648                roots = roots->next;
 649
 650                if (find_pack_entry_one(object->oid.hash, bitmap_git.pack) > 0)
 651                        return 1;
 652        }
 653
 654        return 0;
 655}
 656
 657int prepare_bitmap_walk(struct rev_info *revs)
 658{
 659        unsigned int i;
 660        unsigned int pending_nr = revs->pending.nr;
 661        struct object_array_entry *pending_e = revs->pending.objects;
 662
 663        struct object_list *wants = NULL;
 664        struct object_list *haves = NULL;
 665
 666        struct bitmap *wants_bitmap = NULL;
 667        struct bitmap *haves_bitmap = NULL;
 668
 669        if (!bitmap_git.loaded) {
 670                /* try to open a bitmapped pack, but don't parse it yet
 671                 * because we may not need to use it */
 672                if (open_pack_bitmap() < 0)
 673                        return -1;
 674        }
 675
 676        for (i = 0; i < pending_nr; ++i) {
 677                struct object *object = pending_e[i].item;
 678
 679                if (object->type == OBJ_NONE)
 680                        parse_object_or_die(object->oid.hash, NULL);
 681
 682                while (object->type == OBJ_TAG) {
 683                        struct tag *tag = (struct tag *) object;
 684
 685                        if (object->flags & UNINTERESTING)
 686                                object_list_insert(object, &haves);
 687                        else
 688                                object_list_insert(object, &wants);
 689
 690                        if (!tag->tagged)
 691                                die("bad tag");
 692                        object = parse_object_or_die(tag->tagged->oid.hash, NULL);
 693                }
 694
 695                if (object->flags & UNINTERESTING)
 696                        object_list_insert(object, &haves);
 697                else
 698                        object_list_insert(object, &wants);
 699        }
 700
 701        /*
 702         * if we have a HAVES list, but none of those haves is contained
 703         * in the packfile that has a bitmap, we don't have anything to
 704         * optimize here
 705         */
 706        if (haves && !in_bitmapped_pack(haves))
 707                return -1;
 708
 709        /* if we don't want anything, we're done here */
 710        if (!wants)
 711                return -1;
 712
 713        /*
 714         * now we're going to use bitmaps, so load the actual bitmap entries
 715         * from disk. this is the point of no return; after this the rev_list
 716         * becomes invalidated and we must perform the revwalk through bitmaps
 717         */
 718        if (!bitmap_git.loaded && load_pack_bitmap() < 0)
 719                return -1;
 720
 721        revs->pending.nr = 0;
 722        revs->pending.alloc = 0;
 723        revs->pending.objects = NULL;
 724
 725        if (haves) {
 726                revs->ignore_missing_links = 1;
 727                haves_bitmap = find_objects(revs, haves, NULL);
 728                reset_revision_walk();
 729                revs->ignore_missing_links = 0;
 730
 731                if (haves_bitmap == NULL)
 732                        die("BUG: failed to perform bitmap walk");
 733        }
 734
 735        wants_bitmap = find_objects(revs, wants, haves_bitmap);
 736
 737        if (!wants_bitmap)
 738                die("BUG: failed to perform bitmap walk");
 739
 740        if (haves_bitmap)
 741                bitmap_and_not(wants_bitmap, haves_bitmap);
 742
 743        bitmap_git.result = wants_bitmap;
 744
 745        bitmap_free(haves_bitmap);
 746        return 0;
 747}
 748
 749int reuse_partial_packfile_from_bitmap(struct packed_git **packfile,
 750                                       uint32_t *entries,
 751                                       off_t *up_to)
 752{
 753        /*
 754         * Reuse the packfile content if we need more than
 755         * 90% of its objects
 756         */
 757        static const double REUSE_PERCENT = 0.9;
 758
 759        struct bitmap *result = bitmap_git.result;
 760        uint32_t reuse_threshold;
 761        uint32_t i, reuse_objects = 0;
 762
 763        assert(result);
 764
 765        for (i = 0; i < result->word_alloc; ++i) {
 766                if (result->words[i] != (eword_t)~0) {
 767                        reuse_objects += ewah_bit_ctz64(~result->words[i]);
 768                        break;
 769                }
 770
 771                reuse_objects += BITS_IN_EWORD;
 772        }
 773
 774#ifdef GIT_BITMAP_DEBUG
 775        {
 776                const unsigned char *sha1;
 777                struct revindex_entry *entry;
 778
 779                entry = &bitmap_git.reverse_index->revindex[reuse_objects];
 780                sha1 = nth_packed_object_sha1(bitmap_git.pack, entry->nr);
 781
 782                fprintf(stderr, "Failed to reuse at %d (%016llx)\n",
 783                        reuse_objects, result->words[i]);
 784                fprintf(stderr, " %s\n", sha1_to_hex(sha1));
 785        }
 786#endif
 787
 788        if (!reuse_objects)
 789                return -1;
 790
 791        if (reuse_objects >= bitmap_git.pack->num_objects) {
 792                bitmap_git.reuse_objects = *entries = bitmap_git.pack->num_objects;
 793                *up_to = -1; /* reuse the full pack */
 794                *packfile = bitmap_git.pack;
 795                return 0;
 796        }
 797
 798        reuse_threshold = bitmap_popcount(bitmap_git.result) * REUSE_PERCENT;
 799
 800        if (reuse_objects < reuse_threshold)
 801                return -1;
 802
 803        bitmap_git.reuse_objects = *entries = reuse_objects;
 804        *up_to = bitmap_git.pack->revindex[reuse_objects].offset;
 805        *packfile = bitmap_git.pack;
 806
 807        return 0;
 808}
 809
 810void traverse_bitmap_commit_list(show_reachable_fn show_reachable)
 811{
 812        assert(bitmap_git.result);
 813
 814        show_objects_for_type(bitmap_git.result, bitmap_git.commits,
 815                OBJ_COMMIT, show_reachable);
 816        show_objects_for_type(bitmap_git.result, bitmap_git.trees,
 817                OBJ_TREE, show_reachable);
 818        show_objects_for_type(bitmap_git.result, bitmap_git.blobs,
 819                OBJ_BLOB, show_reachable);
 820        show_objects_for_type(bitmap_git.result, bitmap_git.tags,
 821                OBJ_TAG, show_reachable);
 822
 823        show_extended_objects(bitmap_git.result, show_reachable);
 824
 825        bitmap_free(bitmap_git.result);
 826        bitmap_git.result = NULL;
 827}
 828
 829static uint32_t count_object_type(struct bitmap *objects,
 830                                  enum object_type type)
 831{
 832        struct eindex *eindex = &bitmap_git.ext_index;
 833
 834        uint32_t i = 0, count = 0;
 835        struct ewah_iterator it;
 836        eword_t filter;
 837
 838        switch (type) {
 839        case OBJ_COMMIT:
 840                ewah_iterator_init(&it, bitmap_git.commits);
 841                break;
 842
 843        case OBJ_TREE:
 844                ewah_iterator_init(&it, bitmap_git.trees);
 845                break;
 846
 847        case OBJ_BLOB:
 848                ewah_iterator_init(&it, bitmap_git.blobs);
 849                break;
 850
 851        case OBJ_TAG:
 852                ewah_iterator_init(&it, bitmap_git.tags);
 853                break;
 854
 855        default:
 856                return 0;
 857        }
 858
 859        while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
 860                eword_t word = objects->words[i++] & filter;
 861                count += ewah_bit_popcount64(word);
 862        }
 863
 864        for (i = 0; i < eindex->count; ++i) {
 865                if (eindex->objects[i]->type == type &&
 866                        bitmap_get(objects, bitmap_git.pack->num_objects + i))
 867                        count++;
 868        }
 869
 870        return count;
 871}
 872
 873void count_bitmap_commit_list(uint32_t *commits, uint32_t *trees,
 874                              uint32_t *blobs, uint32_t *tags)
 875{
 876        assert(bitmap_git.result);
 877
 878        if (commits)
 879                *commits = count_object_type(bitmap_git.result, OBJ_COMMIT);
 880
 881        if (trees)
 882                *trees = count_object_type(bitmap_git.result, OBJ_TREE);
 883
 884        if (blobs)
 885                *blobs = count_object_type(bitmap_git.result, OBJ_BLOB);
 886
 887        if (tags)
 888                *tags = count_object_type(bitmap_git.result, OBJ_TAG);
 889}
 890
 891struct bitmap_test_data {
 892        struct bitmap *base;
 893        struct progress *prg;
 894        size_t seen;
 895};
 896
 897static void test_show_object(struct object *object,
 898                             const struct name_path *path,
 899                             const char *last, void *data)
 900{
 901        struct bitmap_test_data *tdata = data;
 902        int bitmap_pos;
 903
 904        bitmap_pos = bitmap_position(object->oid.hash);
 905        if (bitmap_pos < 0)
 906                die("Object not in bitmap: %s\n", oid_to_hex(&object->oid));
 907
 908        bitmap_set(tdata->base, bitmap_pos);
 909        display_progress(tdata->prg, ++tdata->seen);
 910}
 911
 912static void test_show_commit(struct commit *commit, void *data)
 913{
 914        struct bitmap_test_data *tdata = data;
 915        int bitmap_pos;
 916
 917        bitmap_pos = bitmap_position(commit->object.oid.hash);
 918        if (bitmap_pos < 0)
 919                die("Object not in bitmap: %s\n", oid_to_hex(&commit->object.oid));
 920
 921        bitmap_set(tdata->base, bitmap_pos);
 922        display_progress(tdata->prg, ++tdata->seen);
 923}
 924
 925void test_bitmap_walk(struct rev_info *revs)
 926{
 927        struct object *root;
 928        struct bitmap *result = NULL;
 929        khiter_t pos;
 930        size_t result_popcnt;
 931        struct bitmap_test_data tdata;
 932
 933        if (prepare_bitmap_git())
 934                die("failed to load bitmap indexes");
 935
 936        if (revs->pending.nr != 1)
 937                die("you must specify exactly one commit to test");
 938
 939        fprintf(stderr, "Bitmap v%d test (%d entries loaded)\n",
 940                bitmap_git.version, bitmap_git.entry_count);
 941
 942        root = revs->pending.objects[0].item;
 943        pos = kh_get_sha1(bitmap_git.bitmaps, root->oid.hash);
 944
 945        if (pos < kh_end(bitmap_git.bitmaps)) {
 946                struct stored_bitmap *st = kh_value(bitmap_git.bitmaps, pos);
 947                struct ewah_bitmap *bm = lookup_stored_bitmap(st);
 948
 949                fprintf(stderr, "Found bitmap for %s. %d bits / %08x checksum\n",
 950                        oid_to_hex(&root->oid), (int)bm->bit_size, ewah_checksum(bm));
 951
 952                result = ewah_to_bitmap(bm);
 953        }
 954
 955        if (result == NULL)
 956                die("Commit %s doesn't have an indexed bitmap", oid_to_hex(&root->oid));
 957
 958        revs->tag_objects = 1;
 959        revs->tree_objects = 1;
 960        revs->blob_objects = 1;
 961
 962        result_popcnt = bitmap_popcount(result);
 963
 964        if (prepare_revision_walk(revs))
 965                die("revision walk setup failed");
 966
 967        tdata.base = bitmap_new();
 968        tdata.prg = start_progress("Verifying bitmap entries", result_popcnt);
 969        tdata.seen = 0;
 970
 971        traverse_commit_list(revs, &test_show_commit, &test_show_object, &tdata);
 972
 973        stop_progress(&tdata.prg);
 974
 975        if (bitmap_equals(result, tdata.base))
 976                fprintf(stderr, "OK!\n");
 977        else
 978                fprintf(stderr, "Mismatch!\n");
 979
 980        bitmap_free(result);
 981}
 982
 983static int rebuild_bitmap(uint32_t *reposition,
 984                          struct ewah_bitmap *source,
 985                          struct bitmap *dest)
 986{
 987        uint32_t pos = 0;
 988        struct ewah_iterator it;
 989        eword_t word;
 990
 991        ewah_iterator_init(&it, source);
 992
 993        while (ewah_iterator_next(&word, &it)) {
 994                uint32_t offset, bit_pos;
 995
 996                for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
 997                        if ((word >> offset) == 0)
 998                                break;
 999
1000                        offset += ewah_bit_ctz64(word >> offset);
1001
1002                        bit_pos = reposition[pos + offset];
1003                        if (bit_pos > 0)
1004                                bitmap_set(dest, bit_pos - 1);
1005                        else /* can't reuse, we don't have the object */
1006                                return -1;
1007                }
1008
1009                pos += BITS_IN_EWORD;
1010        }
1011        return 0;
1012}
1013
1014int rebuild_existing_bitmaps(struct packing_data *mapping,
1015                             khash_sha1 *reused_bitmaps,
1016                             int show_progress)
1017{
1018        uint32_t i, num_objects;
1019        uint32_t *reposition;
1020        struct bitmap *rebuild;
1021        struct stored_bitmap *stored;
1022        struct progress *progress = NULL;
1023
1024        khiter_t hash_pos;
1025        int hash_ret;
1026
1027        if (prepare_bitmap_git() < 0)
1028                return -1;
1029
1030        num_objects = bitmap_git.pack->num_objects;
1031        reposition = xcalloc(num_objects, sizeof(uint32_t));
1032
1033        for (i = 0; i < num_objects; ++i) {
1034                const unsigned char *sha1;
1035                struct revindex_entry *entry;
1036                struct object_entry *oe;
1037
1038                entry = &bitmap_git.pack->revindex[i];
1039                sha1 = nth_packed_object_sha1(bitmap_git.pack, entry->nr);
1040                oe = packlist_find(mapping, sha1, NULL);
1041
1042                if (oe)
1043                        reposition[i] = oe->in_pack_pos + 1;
1044        }
1045
1046        rebuild = bitmap_new();
1047        i = 0;
1048
1049        if (show_progress)
1050                progress = start_progress("Reusing bitmaps", 0);
1051
1052        kh_foreach_value(bitmap_git.bitmaps, stored, {
1053                if (stored->flags & BITMAP_FLAG_REUSE) {
1054                        if (!rebuild_bitmap(reposition,
1055                                            lookup_stored_bitmap(stored),
1056                                            rebuild)) {
1057                                hash_pos = kh_put_sha1(reused_bitmaps,
1058                                                       stored->sha1,
1059                                                       &hash_ret);
1060                                kh_value(reused_bitmaps, hash_pos) =
1061                                        bitmap_to_ewah(rebuild);
1062                        }
1063                        bitmap_reset(rebuild);
1064                        display_progress(progress, ++i);
1065                }
1066        });
1067
1068        stop_progress(&progress);
1069
1070        free(reposition);
1071        bitmap_free(rebuild);
1072        return 0;
1073}