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
  13 /*
  14  * An entry on the bitmap index, representing the bitmap for a given
  15  * commit.
  16  */
  17 struct 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  */
  32 static struct bitmap_index {
  33         /* Packfile to which this bitmap index belongs to */
  34         struct packed_git *pack;
  35
  36         /* reverse index for the packfile */
  37         struct pack_revindex *reverse_index;
  38
  39         /*
  40          * Mark the first `reuse_objects` in the packfile as reused:
  41          * they will be sent as-is without using them for repacking
  42          * calculations
  43          */
  44         uint32_t reuse_objects;
  45
  46         /* mmapped buffer of the whole bitmap index */
  47         unsigned char *map;
  48         size_t map_size; /* size of the mmaped buffer */
  49         size_t map_pos; /* current position when loading the index */
  50
  51         /*
  52          * Type indexes.
  53          *
  54          * Each bitmap marks which objects in the packfile  are of the given
  55          * type. This provides type information when yielding the objects from
  56          * the packfile during a walk, which allows for better delta bases.
  57          */
  58         struct ewah_bitmap *commits;
  59         struct ewah_bitmap *trees;
  60         struct ewah_bitmap *blobs;
  61         struct ewah_bitmap *tags;
  62
  63         /* Map from SHA1 -> `stored_bitmap` for all the bitmapped commits */
  64         khash_sha1 *bitmaps;
  65
  66         /* Number of bitmapped commits */
  67         uint32_t entry_count;
  68
  69         /* Name-hash cache (or NULL if not present). */
  70         uint32_t *hashes;
  71
  72         /*
  73          * Extended index.
  74          *
  75          * When trying to perform bitmap operations with objects that are not
  76          * packed in `pack`, these objects are added to this "fake index" and
  77          * are assumed to appear at the end of the packfile for all operations
  78          */
  79         struct eindex {
  80                 struct object **objects;
  81                 uint32_t *hashes;
  82                 uint32_t count, alloc;
  83                 khash_sha1_pos *positions;
  84         } ext_index;
  85
  86         /* Bitmap result of the last performed walk */
  87         struct bitmap *result;
  88
  89         /* Version of the bitmap index */
  90         unsigned int version;
  91
  92         unsigned loaded : 1;
  93
  94 } bitmap_git;
  95
  96 static struct ewah_bitmap *lookup_stored_bitmap(struct stored_bitmap *st)
  97 {
  98         struct ewah_bitmap *parent;
  99         struct ewah_bitmap *composed;
 100
 101         if (st->xor == NULL)
 102                 return st->root;
 103
 104         composed = ewah_pool_new();
 105         parent = lookup_stored_bitmap(st->xor);
 106         ewah_xor(st->root, parent, composed);
 107
 108         ewah_pool_free(st->root);
 109         st->root = composed;
 110         st->xor = NULL;
 111
 112         return composed;
 113 }
 114
 115 /*
 116  * Read a bitmap from the current read position on the mmaped
 117  * index, and increase the read position accordingly
 118  */
 119 static struct ewah_bitmap *read_bitmap_1(struct bitmap_index *index)
 120 {
 121         struct ewah_bitmap *b = ewah_pool_new();
 122
 123         int bitmap_size = ewah_read_mmap(b,
 124                 index->map + index->map_pos,
 125                 index->map_size - index->map_pos);
 126
 127         if (bitmap_size < 0) {
 128                 error("Failed to load bitmap index (corrupted?)");
 129                 ewah_pool_free(b);
 130                 return NULL;
 131         }
 132
 133         index->map_pos += bitmap_size;
 134         return b;
 135 }
 136
 137 static int load_bitmap_header(struct bitmap_index *index)
 138 {
 139         struct bitmap_disk_header *header = (void *)index->map;
 140
 141         if (index->map_size < sizeof(*header) + 20)
 142                 return error("Corrupted bitmap index (missing header data)");
 143
 144         if (memcmp(header->magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE)) != 0)
 145                 return error("Corrupted bitmap index file (wrong header)");
 146
 147         index->version = ntohs(header->version);
 148         if (index->version != 1)
 149                 return error("Unsupported version for bitmap index file (%d)", index->version);
 150
 151         /* Parse known bitmap format options */
 152         {
 153                 uint32_t flags = ntohs(header->options);
 154
 155                 if ((flags & BITMAP_OPT_FULL_DAG) == 0)
 156                         return error("Unsupported options for bitmap index file "
 157                                 "(Git requires BITMAP_OPT_FULL_DAG)");
 158
 159                 if (flags & BITMAP_OPT_HASH_CACHE) {
 160                         unsigned char *end = index->map + index->map_size - 20;
 161                         index->hashes = ((uint32_t *)end) - index->pack->num_objects;
 162                 }
 163         }
 164
 165         index->entry_count = ntohl(header->entry_count);
 166         index->map_pos += sizeof(*header);
 167         return 0;
 168 }
 169
 170 static struct stored_bitmap *store_bitmap(struct bitmap_index *index,
 171                                           struct ewah_bitmap *root,
 172                                           const unsigned char *sha1,
 173                                           struct stored_bitmap *xor_with,
 174                                           int flags)
 175 {
 176         struct stored_bitmap *stored;
 177         khiter_t hash_pos;
 178         int ret;
 179
 180         stored = xmalloc(sizeof(struct stored_bitmap));
 181         stored->root = root;
 182         stored->xor = xor_with;
 183         stored->flags = flags;
 184         hashcpy(stored->sha1, sha1);
 185
 186         hash_pos = kh_put_sha1(index->bitmaps, stored->sha1, &ret);
 187
 188         /* a 0 return code means the insertion succeeded with no changes,
 189          * because the SHA1 already existed on the map. this is bad, there
 190          * shouldn't be duplicated commits in the index */
 191         if (ret == 0) {
 192                 error("Duplicate entry in bitmap index: %s", sha1_to_hex(sha1));
 193                 return NULL;
 194         }
 195
 196         kh_value(index->bitmaps, hash_pos) = stored;
 197         return stored;
 198 }
 199
 200 static inline uint32_t read_be32(const unsigned char *buffer, size_t *pos)
 201 {
 202         uint32_t result = get_be32(buffer + *pos);
 203         (*pos) += sizeof(result);
 204         return result;
 205 }
 206
 207 static inline uint8_t read_u8(const unsigned char *buffer, size_t *pos)
 208 {
 209         return buffer[(*pos)++];
 210 }
 211
 212 #define MAX_XOR_OFFSET 160
 213
 214 static int load_bitmap_entries_v1(struct bitmap_index *index)
 215 {
 216         uint32_t i;
 217         struct stored_bitmap *recent_bitmaps[MAX_XOR_OFFSET] = { NULL };
 218
 219         for (i = 0; i < index->entry_count; ++i) {
 220                 int xor_offset, flags;
 221                 struct ewah_bitmap *bitmap = NULL;
 222                 struct stored_bitmap *xor_bitmap = NULL;
 223                 uint32_t commit_idx_pos;
 224                 const unsigned char *sha1;
 225
 226                 commit_idx_pos = read_be32(index->map, &index->map_pos);
 227                 xor_offset = read_u8(index->map, &index->map_pos);
 228                 flags = read_u8(index->map, &index->map_pos);
 229
 230                 sha1 = nth_packed_object_sha1(index->pack, commit_idx_pos);
 231
 232                 bitmap = read_bitmap_1(index);
 233                 if (!bitmap)
 234                         return -1;
 235
 236                 if (xor_offset > MAX_XOR_OFFSET || xor_offset > i)
 237                         return error("Corrupted bitmap pack index");
 238
 239                 if (xor_offset > 0) {
 240                         xor_bitmap = recent_bitmaps[(i - xor_offset) % MAX_XOR_OFFSET];
 241
 242                         if (xor_bitmap == NULL)
 243                                 return error("Invalid XOR offset in bitmap pack index");
 244                 }
 245
 246                 recent_bitmaps[i % MAX_XOR_OFFSET] = store_bitmap(
 247                         index, bitmap, sha1, xor_bitmap, flags);
 248         }
 249
 250         return 0;
 251 }
 252
 253 static char *pack_bitmap_filename(struct packed_git *p)
 254 {
 255         size_t len;
 256
 257         if (!strip_suffix(p->pack_name, ".pack", &len))
 258                 die("BUG: pack_name does not end in .pack");
 259         return xstrfmt("%.*s.bitmap", (int)len, p->pack_name);
 260 }
 261
 262 static int open_pack_bitmap_1(struct packed_git *packfile)
 263 {
 264         int fd;
 265         struct stat st;
 266         char *idx_name;
 267
 268         if (open_pack_index(packfile))
 269                 return -1;
 270
 271         idx_name = pack_bitmap_filename(packfile);
 272         fd = git_open_noatime(idx_name);
 273         free(idx_name);
 274
 275         if (fd < 0)
 276                 return -1;
 277
 278         if (fstat(fd, &st)) {
 279                 close(fd);
 280                 return -1;
 281         }
 282
 283         if (bitmap_git.pack) {
 284                 warning("ignoring extra bitmap file: %s", packfile->pack_name);
 285                 close(fd);
 286                 return -1;
 287         }
 288
 289         bitmap_git.pack = packfile;
 290         bitmap_git.map_size = xsize_t(st.st_size);
 291         bitmap_git.map = xmmap(NULL, bitmap_git.map_size, PROT_READ, MAP_PRIVATE, fd, 0);
 292         bitmap_git.map_pos = 0;
 293         close(fd);
 294
 295         if (load_bitmap_header(&bitmap_git) < 0) {
 296                 munmap(bitmap_git.map, bitmap_git.map_size);
 297                 bitmap_git.map = NULL;
 298                 bitmap_git.map_size = 0;
 299                 return -1;
 300         }
 301
 302         return 0;
 303 }
 304
 305 static int load_pack_bitmap(void)
 306 {
 307         assert(bitmap_git.map && !bitmap_git.loaded);
 308
 309         bitmap_git.bitmaps = kh_init_sha1();
 310         bitmap_git.ext_index.positions = kh_init_sha1_pos();
 311         bitmap_git.reverse_index = revindex_for_pack(bitmap_git.pack);
 312
 313         if (!(bitmap_git.commits = read_bitmap_1(&bitmap_git)) ||
 314                 !(bitmap_git.trees = read_bitmap_1(&bitmap_git)) ||
 315                 !(bitmap_git.blobs = read_bitmap_1(&bitmap_git)) ||
 316                 !(bitmap_git.tags = read_bitmap_1(&bitmap_git)))
 317                 goto failed;
 318
 319         if (load_bitmap_entries_v1(&bitmap_git) < 0)
 320                 goto failed;
 321
 322         bitmap_git.loaded = 1;
 323         return 0;
 324
 325 failed:
 326         munmap(bitmap_git.map, bitmap_git.map_size);
 327         bitmap_git.map = NULL;
 328         bitmap_git.map_size = 0;
 329         return -1;
 330 }
 331
 332 static int open_pack_bitmap(void)
 333 {
 334         struct packed_git *p;
 335         int ret = -1;
 336
 337         assert(!bitmap_git.map && !bitmap_git.loaded);
 338
 339         prepare_packed_git();
 340         for (p = packed_git; p; p = p->next) {
 341                 if (open_pack_bitmap_1(p) == 0)
 342                         ret = 0;
 343         }
 344
 345         return ret;
 346 }
 347
 348 int prepare_bitmap_git(void)
 349 {
 350         if (bitmap_git.loaded)
 351                 return 0;
 352
 353         if (!open_pack_bitmap())
 354                 return load_pack_bitmap();
 355
 356         return -1;
 357 }
 358
 359 struct include_data {
 360         struct bitmap *base;
 361         struct bitmap *seen;
 362 };
 363
 364 static inline int bitmap_position_extended(const unsigned char *sha1)
 365 {
 366         khash_sha1_pos *positions = bitmap_git.ext_index.positions;
 367         khiter_t pos = kh_get_sha1_pos(positions, sha1);
 368
 369         if (pos < kh_end(positions)) {
 370                 int bitmap_pos = kh_value(positions, pos);
 371                 return bitmap_pos + bitmap_git.pack->num_objects;
 372         }
 373
 374         return -1;
 375 }
 376
 377 static inline int bitmap_position_packfile(const unsigned char *sha1)
 378 {
 379         off_t offset = find_pack_entry_one(sha1, bitmap_git.pack);
 380         if (!offset)
 381                 return -1;
 382
 383         return find_revindex_position(bitmap_git.reverse_index, offset);
 384 }
 385
 386 static int bitmap_position(const unsigned char *sha1)
 387 {
 388         int pos = bitmap_position_packfile(sha1);
 389         return (pos >= 0) ? pos : bitmap_position_extended(sha1);
 390 }
 391
 392 static int ext_index_add_object(struct object *object, const char *name)
 393 {
 394         struct eindex *eindex = &bitmap_git.ext_index;
 395
 396         khiter_t hash_pos;
 397         int hash_ret;
 398         int bitmap_pos;
 399
 400         hash_pos = kh_put_sha1_pos(eindex->positions, object->oid.hash, &hash_ret);
 401         if (hash_ret > 0) {
 402                 if (eindex->count >= eindex->alloc) {
 403                         eindex->alloc = (eindex->alloc + 16) * 3 / 2;
 404                         REALLOC_ARRAY(eindex->objects, eindex->alloc);
 405                         REALLOC_ARRAY(eindex->hashes, eindex->alloc);
 406                 }
 407
 408                 bitmap_pos = eindex->count;
 409                 eindex->objects[eindex->count] = object;
 410                 eindex->hashes[eindex->count] = pack_name_hash(name);
 411                 kh_value(eindex->positions, hash_pos) = bitmap_pos;
 412                 eindex->count++;
 413         } else {
 414                 bitmap_pos = kh_value(eindex->positions, hash_pos);
 415         }
 416
 417         return bitmap_pos + bitmap_git.pack->num_objects;
 418 }
 419
 420 static void show_object(struct object *object, const struct name_path *path,
 421                         const char *last, void *data)
 422 {
 423         struct bitmap *base = data;
 424         int bitmap_pos;
 425
 426         bitmap_pos = bitmap_position(object->oid.hash);
 427
 428         if (bitmap_pos < 0) {
 429                 char *name = path_name(path, last);
 430                 bitmap_pos = ext_index_add_object(object, name);
 431                 free(name);
 432         }
 433
 434         bitmap_set(base, bitmap_pos);
 435 }
 436
 437 static void show_commit(struct commit *commit, void *data)
 438 {
 439 }
 440
 441 static int add_to_include_set(struct include_data *data,
 442                               const unsigned char *sha1,
 443                               int bitmap_pos)
 444 {
 445         khiter_t hash_pos;
 446
 447         if (data->seen && bitmap_get(data->seen, bitmap_pos))
 448                 return 0;
 449
 450         if (bitmap_get(data->base, bitmap_pos))
 451                 return 0;
 452
 453         hash_pos = kh_get_sha1(bitmap_git.bitmaps, sha1);
 454         if (hash_pos < kh_end(bitmap_git.bitmaps)) {
 455                 struct stored_bitmap *st = kh_value(bitmap_git.bitmaps, hash_pos);
 456                 bitmap_or_ewah(data->base, lookup_stored_bitmap(st));
 457                 return 0;
 458         }
 459
 460         bitmap_set(data->base, bitmap_pos);
 461         return 1;
 462 }
 463
 464 static int should_include(struct commit *commit, void *_data)
 465 {
 466         struct include_data *data = _data;
 467         int bitmap_pos;
 468
 469         bitmap_pos = bitmap_position(commit->object.oid.hash);
 470         if (bitmap_pos < 0)
 471                 bitmap_pos = ext_index_add_object((struct object *)commit, NULL);
 472
 473         if (!add_to_include_set(data, commit->object.oid.hash, bitmap_pos)) {
 474                 struct commit_list *parent = commit->parents;
 475
 476                 while (parent) {
 477                         parent->item->object.flags |= SEEN;
 478                         parent = parent->next;
 479                 }
 480
 481                 return 0;
 482         }
 483
 484         return 1;
 485 }
 486
 487 static struct bitmap *find_objects(struct rev_info *revs,
 488                                    struct object_list *roots,
 489                                    struct bitmap *seen)
 490 {
 491         struct bitmap *base = NULL;
 492         int needs_walk = 0;
 493
 494         struct object_list *not_mapped = NULL;
 495
 496         /*
 497          * Go through all the roots for the walk. The ones that have bitmaps
 498          * on the bitmap index will be `or`ed together to form an initial
 499          * global reachability analysis.
 500          *
 501          * The ones without bitmaps in the index will be stored in the
 502          * `not_mapped_list` for further processing.
 503          */
 504         while (roots) {
 505                 struct object *object = roots->item;
 506                 roots = roots->next;
 507
 508                 if (object->type == OBJ_COMMIT) {
 509                         khiter_t pos = kh_get_sha1(bitmap_git.bitmaps, object->oid.hash);
 510
 511                         if (pos < kh_end(bitmap_git.bitmaps)) {
 512                                 struct stored_bitmap *st = kh_value(bitmap_git.bitmaps, pos);
 513                                 struct ewah_bitmap *or_with = lookup_stored_bitmap(st);
 514
 515                                 if (base == NULL)
 516                                         base = ewah_to_bitmap(or_with);
 517                                 else
 518                                         bitmap_or_ewah(base, or_with);
 519
 520                                 object->flags |= SEEN;
 521                                 continue;
 522                         }
 523                 }
 524
 525                 object_list_insert(object, &not_mapped);
 526         }
 527
 528         /*
 529          * Best case scenario: We found bitmaps for all the roots,
 530          * so the resulting `or` bitmap has the full reachability analysis
 531          */
 532         if (not_mapped == NULL)
 533                 return base;
 534
 535         roots = not_mapped;
 536
 537         /*
 538          * Let's iterate through all the roots that don't have bitmaps to
 539          * check if we can determine them to be reachable from the existing
 540          * global bitmap.
 541          *
 542          * If we cannot find them in the existing global bitmap, we'll need
 543          * to push them to an actual walk and run it until we can confirm
 544          * they are reachable
 545          */
 546         while (roots) {
 547                 struct object *object = roots->item;
 548                 int pos;
 549
 550                 roots = roots->next;
 551                 pos = bitmap_position(object->oid.hash);
 552
 553                 if (pos < 0 || base == NULL || !bitmap_get(base, pos)) {
 554                         object->flags &= ~UNINTERESTING;
 555                         add_pending_object(revs, object, "");
 556                         needs_walk = 1;
 557                 } else {
 558                         object->flags |= SEEN;
 559                 }
 560         }
 561
 562         if (needs_walk) {
 563                 struct include_data incdata;
 564
 565                 if (base == NULL)
 566                         base = bitmap_new();
 567
 568                 incdata.base = base;
 569                 incdata.seen = seen;
 570
 571                 revs->include_check = should_include;
 572                 revs->include_check_data = &incdata;
 573
 574                 if (prepare_revision_walk(revs))
 575                         die("revision walk setup failed");
 576
 577                 traverse_commit_list(revs, show_commit, show_object, base);
 578         }
 579
 580         return base;
 581 }
 582
 583 static void show_extended_objects(struct bitmap *objects,
 584                                   show_reachable_fn show_reach)
 585 {
 586         struct eindex *eindex = &bitmap_git.ext_index;
 587         uint32_t i;
 588
 589         for (i = 0; i < eindex->count; ++i) {
 590                 struct object *obj;
 591
 592                 if (!bitmap_get(objects, bitmap_git.pack->num_objects + i))
 593                         continue;
 594
 595                 obj = eindex->objects[i];
 596                 show_reach(obj->oid.hash, obj->type, 0, eindex->hashes[i], NULL, 0);
 597         }
 598 }
 599
 600 static void show_objects_for_type(
 601         struct bitmap *objects,
 602         struct ewah_bitmap *type_filter,
 603         enum object_type object_type,
 604         show_reachable_fn show_reach)
 605 {
 606         size_t pos = 0, i = 0;
 607         uint32_t offset;
 608
 609         struct ewah_iterator it;
 610         eword_t filter;
 611
 612         if (bitmap_git.reuse_objects == bitmap_git.pack->num_objects)
 613                 return;
 614
 615         ewah_iterator_init(&it, type_filter);
 616
 617         while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
 618                 eword_t word = objects->words[i] & filter;
 619
 620                 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
 621                         const unsigned char *sha1;
 622                         struct revindex_entry *entry;
 623                         uint32_t hash = 0;
 624
 625                         if ((word >> offset) == 0)
 626                                 break;
 627
 628                         offset += ewah_bit_ctz64(word >> offset);
 629
 630                         if (pos + offset < bitmap_git.reuse_objects)
 631                                 continue;
 632
 633                         entry = &bitmap_git.reverse_index->revindex[pos + offset];
 634                         sha1 = nth_packed_object_sha1(bitmap_git.pack, entry->nr);
 635
 636                         if (bitmap_git.hashes)
 637                                 hash = ntohl(bitmap_git.hashes[entry->nr]);
 638
 639                         show_reach(sha1, object_type, 0, hash, bitmap_git.pack, entry->offset);
 640                 }
 641
 642                 pos += BITS_IN_EWORD;
 643                 i++;
 644         }
 645 }
 646
 647 static int in_bitmapped_pack(struct object_list *roots)
 648 {
 649         while (roots) {
 650                 struct object *object = roots->item;
 651                 roots = roots->next;
 652
 653                 if (find_pack_entry_one(object->oid.hash, bitmap_git.pack) > 0)
 654                         return 1;
 655         }
 656
 657         return 0;
 658 }
 659
 660 int prepare_bitmap_walk(struct rev_info *revs)
 661 {
 662         unsigned int i;
 663         unsigned int pending_nr = revs->pending.nr;
 664         struct object_array_entry *pending_e = revs->pending.objects;
 665
 666         struct object_list *wants = NULL;
 667         struct object_list *haves = NULL;
 668
 669         struct bitmap *wants_bitmap = NULL;
 670         struct bitmap *haves_bitmap = NULL;
 671
 672         if (!bitmap_git.loaded) {
 673                 /* try to open a bitmapped pack, but don't parse it yet
 674                  * because we may not need to use it */
 675                 if (open_pack_bitmap() < 0)
 676                         return -1;
 677         }
 678
 679         for (i = 0; i < pending_nr; ++i) {
 680                 struct object *object = pending_e[i].item;
 681
 682                 if (object->type == OBJ_NONE)
 683                         parse_object_or_die(object->oid.hash, NULL);
 684
 685                 while (object->type == OBJ_TAG) {
 686                         struct tag *tag = (struct tag *) object;
 687
 688                         if (object->flags & UNINTERESTING)
 689                                 object_list_insert(object, &haves);
 690                         else
 691                                 object_list_insert(object, &wants);
 692
 693                         if (!tag->tagged)
 694                                 die("bad tag");
 695                         object = parse_object_or_die(tag->tagged->oid.hash, NULL);
 696                 }
 697
 698                 if (object->flags & UNINTERESTING)
 699                         object_list_insert(object, &haves);
 700                 else
 701                         object_list_insert(object, &wants);
 702         }
 703
 704         /*
 705          * if we have a HAVES list, but none of those haves is contained
 706          * in the packfile that has a bitmap, we don't have anything to
 707          * optimize here
 708          */
 709         if (haves && !in_bitmapped_pack(haves))
 710                 return -1;
 711
 712         /* if we don't want anything, we're done here */
 713         if (!wants)
 714                 return -1;
 715
 716         /*
 717          * now we're going to use bitmaps, so load the actual bitmap entries
 718          * from disk. this is the point of no return; after this the rev_list
 719          * becomes invalidated and we must perform the revwalk through bitmaps
 720          */
 721         if (!bitmap_git.loaded && load_pack_bitmap() < 0)
 722                 return -1;
 723
 724         revs->pending.nr = 0;
 725         revs->pending.alloc = 0;
 726         revs->pending.objects = NULL;
 727
 728         if (haves) {
 729                 revs->ignore_missing_links = 1;
 730                 haves_bitmap = find_objects(revs, haves, NULL);
 731                 reset_revision_walk();
 732                 revs->ignore_missing_links = 0;
 733
 734                 if (haves_bitmap == NULL)
 735                         die("BUG: failed to perform bitmap walk");
 736         }
 737
 738         wants_bitmap = find_objects(revs, wants, haves_bitmap);
 739
 740         if (!wants_bitmap)
 741                 die("BUG: failed to perform bitmap walk");
 742
 743         if (haves_bitmap)
 744                 bitmap_and_not(wants_bitmap, haves_bitmap);
 745
 746         bitmap_git.result = wants_bitmap;
 747
 748         bitmap_free(haves_bitmap);
 749         return 0;
 750 }
 751
 752 int reuse_partial_packfile_from_bitmap(struct packed_git **packfile,
 753                                        uint32_t *entries,
 754                                        off_t *up_to)
 755 {
 756         /*
 757          * Reuse the packfile content if we need more than
 758          * 90% of its objects
 759          */
 760         static const double REUSE_PERCENT = 0.9;
 761
 762         struct bitmap *result = bitmap_git.result;
 763         uint32_t reuse_threshold;
 764         uint32_t i, reuse_objects = 0;
 765
 766         assert(result);
 767
 768         for (i = 0; i < result->word_alloc; ++i) {
 769                 if (result->words[i] != (eword_t)~0) {
 770                         reuse_objects += ewah_bit_ctz64(~result->words[i]);
 771                         break;
 772                 }
 773
 774                 reuse_objects += BITS_IN_EWORD;
 775         }
 776
 777 #ifdef GIT_BITMAP_DEBUG
 778         {
 779                 const unsigned char *sha1;
 780                 struct revindex_entry *entry;
 781
 782                 entry = &bitmap_git.reverse_index->revindex[reuse_objects];
 783                 sha1 = nth_packed_object_sha1(bitmap_git.pack, entry->nr);
 784
 785                 fprintf(stderr, "Failed to reuse at %d (%016llx)\n",
 786                         reuse_objects, result->words[i]);
 787                 fprintf(stderr, " %s\n", sha1_to_hex(sha1));
 788         }
 789 #endif
 790
 791         if (!reuse_objects)
 792                 return -1;
 793
 794         if (reuse_objects >= bitmap_git.pack->num_objects) {
 795                 bitmap_git.reuse_objects = *entries = bitmap_git.pack->num_objects;
 796                 *up_to = -1; /* reuse the full pack */
 797                 *packfile = bitmap_git.pack;
 798                 return 0;
 799         }
 800
 801         reuse_threshold = bitmap_popcount(bitmap_git.result) * REUSE_PERCENT;
 802
 803         if (reuse_objects < reuse_threshold)
 804                 return -1;
 805
 806         bitmap_git.reuse_objects = *entries = reuse_objects;
 807         *up_to = bitmap_git.reverse_index->revindex[reuse_objects].offset;
 808         *packfile = bitmap_git.pack;
 809
 810         return 0;
 811 }
 812
 813 void traverse_bitmap_commit_list(show_reachable_fn show_reachable)
 814 {
 815         assert(bitmap_git.result);
 816
 817         show_objects_for_type(bitmap_git.result, bitmap_git.commits,
 818                 OBJ_COMMIT, show_reachable);
 819         show_objects_for_type(bitmap_git.result, bitmap_git.trees,
 820                 OBJ_TREE, show_reachable);
 821         show_objects_for_type(bitmap_git.result, bitmap_git.blobs,
 822                 OBJ_BLOB, show_reachable);
 823         show_objects_for_type(bitmap_git.result, bitmap_git.tags,
 824                 OBJ_TAG, show_reachable);
 825
 826         show_extended_objects(bitmap_git.result, show_reachable);
 827
 828         bitmap_free(bitmap_git.result);
 829         bitmap_git.result = NULL;
 830 }
 831
 832 static uint32_t count_object_type(struct bitmap *objects,
 833                                   enum object_type type)
 834 {
 835         struct eindex *eindex = &bitmap_git.ext_index;
 836
 837         uint32_t i = 0, count = 0;
 838         struct ewah_iterator it;
 839         eword_t filter;
 840
 841         switch (type) {
 842         case OBJ_COMMIT:
 843                 ewah_iterator_init(&it, bitmap_git.commits);
 844                 break;
 845
 846         case OBJ_TREE:
 847                 ewah_iterator_init(&it, bitmap_git.trees);
 848                 break;
 849
 850         case OBJ_BLOB:
 851                 ewah_iterator_init(&it, bitmap_git.blobs);
 852                 break;
 853
 854         case OBJ_TAG:
 855                 ewah_iterator_init(&it, bitmap_git.tags);
 856                 break;
 857
 858         default:
 859                 return 0;
 860         }
 861
 862         while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
 863                 eword_t word = objects->words[i++] & filter;
 864                 count += ewah_bit_popcount64(word);
 865         }
 866
 867         for (i = 0; i < eindex->count; ++i) {
 868                 if (eindex->objects[i]->type == type &&
 869                         bitmap_get(objects, bitmap_git.pack->num_objects + i))
 870                         count++;
 871         }
 872
 873         return count;
 874 }
 875
 876 void count_bitmap_commit_list(uint32_t *commits, uint32_t *trees,
 877                               uint32_t *blobs, uint32_t *tags)
 878 {
 879         assert(bitmap_git.result);
 880
 881         if (commits)
 882                 *commits = count_object_type(bitmap_git.result, OBJ_COMMIT);
 883
 884         if (trees)
 885                 *trees = count_object_type(bitmap_git.result, OBJ_TREE);
 886
 887         if (blobs)
 888                 *blobs = count_object_type(bitmap_git.result, OBJ_BLOB);
 889
 890         if (tags)
 891                 *tags = count_object_type(bitmap_git.result, OBJ_TAG);
 892 }
 893
 894 struct bitmap_test_data {
 895         struct bitmap *base;
 896         struct progress *prg;
 897         size_t seen;
 898 };
 899
 900 static void test_show_object(struct object *object,
 901                              const struct name_path *path,
 902                              const char *last, void *data)
 903 {
 904         struct bitmap_test_data *tdata = data;
 905         int bitmap_pos;
 906
 907         bitmap_pos = bitmap_position(object->oid.hash);
 908         if (bitmap_pos < 0)
 909                 die("Object not in bitmap: %s\n", oid_to_hex(&object->oid));
 910
 911         bitmap_set(tdata->base, bitmap_pos);
 912         display_progress(tdata->prg, ++tdata->seen);
 913 }
 914
 915 static void test_show_commit(struct commit *commit, void *data)
 916 {
 917         struct bitmap_test_data *tdata = data;
 918         int bitmap_pos;
 919
 920         bitmap_pos = bitmap_position(commit->object.oid.hash);
 921         if (bitmap_pos < 0)
 922                 die("Object not in bitmap: %s\n", oid_to_hex(&commit->object.oid));
 923
 924         bitmap_set(tdata->base, bitmap_pos);
 925         display_progress(tdata->prg, ++tdata->seen);
 926 }
 927
 928 void test_bitmap_walk(struct rev_info *revs)
 929 {
 930         struct object *root;
 931         struct bitmap *result = NULL;
 932         khiter_t pos;
 933         size_t result_popcnt;
 934         struct bitmap_test_data tdata;
 935
 936         if (prepare_bitmap_git())
 937                 die("failed to load bitmap indexes");
 938
 939         if (revs->pending.nr != 1)
 940                 die("you must specify exactly one commit to test");
 941
 942         fprintf(stderr, "Bitmap v%d test (%d entries loaded)\n",
 943                 bitmap_git.version, bitmap_git.entry_count);
 944
 945         root = revs->pending.objects[0].item;
 946         pos = kh_get_sha1(bitmap_git.bitmaps, root->oid.hash);
 947
 948         if (pos < kh_end(bitmap_git.bitmaps)) {
 949                 struct stored_bitmap *st = kh_value(bitmap_git.bitmaps, pos);
 950                 struct ewah_bitmap *bm = lookup_stored_bitmap(st);
 951
 952                 fprintf(stderr, "Found bitmap for %s. %d bits / %08x checksum\n",
 953                         oid_to_hex(&root->oid), (int)bm->bit_size, ewah_checksum(bm));
 954
 955                 result = ewah_to_bitmap(bm);
 956         }
 957
 958         if (result == NULL)
 959                 die("Commit %s doesn't have an indexed bitmap", oid_to_hex(&root->oid));
 960
 961         revs->tag_objects = 1;
 962         revs->tree_objects = 1;
 963         revs->blob_objects = 1;
 964
 965         result_popcnt = bitmap_popcount(result);
 966
 967         if (prepare_revision_walk(revs))
 968                 die("revision walk setup failed");
 969
 970         tdata.base = bitmap_new();
 971         tdata.prg = start_progress("Verifying bitmap entries", result_popcnt);
 972         tdata.seen = 0;
 973
 974         traverse_commit_list(revs, &test_show_commit, &test_show_object, &tdata);
 975
 976         stop_progress(&tdata.prg);
 977
 978         if (bitmap_equals(result, tdata.base))
 979                 fprintf(stderr, "OK!\n");
 980         else
 981                 fprintf(stderr, "Mismatch!\n");
 982
 983         bitmap_free(result);
 984 }
 985
 986 static int rebuild_bitmap(uint32_t *reposition,
 987                           struct ewah_bitmap *source,
 988                           struct bitmap *dest)
 989 {
 990         uint32_t pos = 0;
 991         struct ewah_iterator it;
 992         eword_t word;
 993
 994         ewah_iterator_init(&it, source);
 995
 996         while (ewah_iterator_next(&word, &it)) {
 997                 uint32_t offset, bit_pos;
 998
 999                 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
1000                         if ((word >> offset) == 0)
1001                                 break;
1002
1003                         offset += ewah_bit_ctz64(word >> offset);
1004
1005                         bit_pos = reposition[pos + offset];
1006                         if (bit_pos > 0)
1007                                 bitmap_set(dest, bit_pos - 1);
1008                         else /* can't reuse, we don't have the object */
1009                                 return -1;
1010                 }
1011
1012                 pos += BITS_IN_EWORD;
1013         }
1014         return 0;
1015 }
1016
1017 int rebuild_existing_bitmaps(struct packing_data *mapping,
1018                              khash_sha1 *reused_bitmaps,
1019                              int show_progress)
1020 {
1021         uint32_t i, num_objects;
1022         uint32_t *reposition;
1023         struct bitmap *rebuild;
1024         struct stored_bitmap *stored;
1025         struct progress *progress = NULL;
1026
1027         khiter_t hash_pos;
1028         int hash_ret;
1029
1030         if (prepare_bitmap_git() < 0)
1031                 return -1;
1032
1033         num_objects = bitmap_git.pack->num_objects;
1034         reposition = xcalloc(num_objects, sizeof(uint32_t));
1035
1036         for (i = 0; i < num_objects; ++i) {
1037                 const unsigned char *sha1;
1038                 struct revindex_entry *entry;
1039                 struct object_entry *oe;
1040
1041                 entry = &bitmap_git.reverse_index->revindex[i];
1042                 sha1 = nth_packed_object_sha1(bitmap_git.pack, entry->nr);
1043                 oe = packlist_find(mapping, sha1, NULL);
1044
1045                 if (oe)
1046                         reposition[i] = oe->in_pack_pos + 1;
1047         }
1048
1049         rebuild = bitmap_new();
1050         i = 0;
1051
1052         if (show_progress)
1053                 progress = start_progress("Reusing bitmaps", 0);
1054
1055         kh_foreach_value(bitmap_git.bitmaps, stored, {
1056                 if (stored->flags & BITMAP_FLAG_REUSE) {
1057                         if (!rebuild_bitmap(reposition,
1058                                             lookup_stored_bitmap(stored),
1059                                             rebuild)) {
1060                                 hash_pos = kh_put_sha1(reused_bitmaps,
1061                                                        stored->sha1,
1062                                                        &hash_ret);
1063                                 kh_value(reused_bitmaps, hash_pos) =
1064                                         bitmap_to_ewah(rebuild);
1065                         }
1066                         bitmap_reset(rebuild);
1067                         display_progress(progress, ++i);
1068                 }
1069         });
1070
1071         stop_progress(&progress);
1072
1073         free(reposition);
1074         bitmap_free(rebuild);
1075         return 0;
1076 }