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         char *idx_name;
 256         int len;
 257
 258         len = strlen(p->pack_name) - strlen(".pack");
 259         idx_name = xmalloc(len + strlen(".bitmap") + 1);
 260
 261         memcpy(idx_name, p->pack_name, len);
 262         memcpy(idx_name + len, ".bitmap", strlen(".bitmap") + 1);
 263
 264         return idx_name;
 265 }
 266
 267 static int open_pack_bitmap_1(struct packed_git *packfile)
 268 {
 269         int fd;
 270         struct stat st;
 271         char *idx_name;
 272
 273         if (open_pack_index(packfile))
 274                 return -1;
 275
 276         idx_name = pack_bitmap_filename(packfile);
 277         fd = git_open_noatime(idx_name);
 278         free(idx_name);
 279
 280         if (fd < 0)
 281                 return -1;
 282
 283         if (fstat(fd, &st)) {
 284                 close(fd);
 285                 return -1;
 286         }
 287
 288         if (bitmap_git.pack) {
 289                 warning("ignoring extra bitmap file: %s", packfile->pack_name);
 290                 close(fd);
 291                 return -1;
 292         }
 293
 294         bitmap_git.pack = packfile;
 295         bitmap_git.map_size = xsize_t(st.st_size);
 296         bitmap_git.map = xmmap(NULL, bitmap_git.map_size, PROT_READ, MAP_PRIVATE, fd, 0);
 297         bitmap_git.map_pos = 0;
 298         close(fd);
 299
 300         if (load_bitmap_header(&bitmap_git) < 0) {
 301                 munmap(bitmap_git.map, bitmap_git.map_size);
 302                 bitmap_git.map = NULL;
 303                 bitmap_git.map_size = 0;
 304                 return -1;
 305         }
 306
 307         return 0;
 308 }
 309
 310 static int load_pack_bitmap(void)
 311 {
 312         assert(bitmap_git.map && !bitmap_git.loaded);
 313
 314         bitmap_git.bitmaps = kh_init_sha1();
 315         bitmap_git.ext_index.positions = kh_init_sha1_pos();
 316         bitmap_git.reverse_index = revindex_for_pack(bitmap_git.pack);
 317
 318         if (!(bitmap_git.commits = read_bitmap_1(&bitmap_git)) ||
 319                 !(bitmap_git.trees = read_bitmap_1(&bitmap_git)) ||
 320                 !(bitmap_git.blobs = read_bitmap_1(&bitmap_git)) ||
 321                 !(bitmap_git.tags = read_bitmap_1(&bitmap_git)))
 322                 goto failed;
 323
 324         if (load_bitmap_entries_v1(&bitmap_git) < 0)
 325                 goto failed;
 326
 327         bitmap_git.loaded = 1;
 328         return 0;
 329
 330 failed:
 331         munmap(bitmap_git.map, bitmap_git.map_size);
 332         bitmap_git.map = NULL;
 333         bitmap_git.map_size = 0;
 334         return -1;
 335 }
 336
 337 static int open_pack_bitmap(void)
 338 {
 339         struct packed_git *p;
 340         int ret = -1;
 341
 342         assert(!bitmap_git.map && !bitmap_git.loaded);
 343
 344         prepare_packed_git();
 345         for (p = packed_git; p; p = p->next) {
 346                 if (open_pack_bitmap_1(p) == 0)
 347                         ret = 0;
 348         }
 349
 350         return ret;
 351 }
 352
 353 int prepare_bitmap_git(void)
 354 {
 355         if (bitmap_git.loaded)
 356                 return 0;
 357
 358         if (!open_pack_bitmap())
 359                 return load_pack_bitmap();
 360
 361         return -1;
 362 }
 363
 364 struct include_data {
 365         struct bitmap *base;
 366         struct bitmap *seen;
 367 };
 368
 369 static inline int bitmap_position_extended(const unsigned char *sha1)
 370 {
 371         khash_sha1_pos *positions = bitmap_git.ext_index.positions;
 372         khiter_t pos = kh_get_sha1_pos(positions, sha1);
 373
 374         if (pos < kh_end(positions)) {
 375                 int bitmap_pos = kh_value(positions, pos);
 376                 return bitmap_pos + bitmap_git.pack->num_objects;
 377         }
 378
 379         return -1;
 380 }
 381
 382 static inline int bitmap_position_packfile(const unsigned char *sha1)
 383 {
 384         off_t offset = find_pack_entry_one(sha1, bitmap_git.pack);
 385         if (!offset)
 386                 return -1;
 387
 388         return find_revindex_position(bitmap_git.reverse_index, offset);
 389 }
 390
 391 static int bitmap_position(const unsigned char *sha1)
 392 {
 393         int pos = bitmap_position_packfile(sha1);
 394         return (pos >= 0) ? pos : bitmap_position_extended(sha1);
 395 }
 396
 397 static int ext_index_add_object(struct object *object, const char *name)
 398 {
 399         struct eindex *eindex = &bitmap_git.ext_index;
 400
 401         khiter_t hash_pos;
 402         int hash_ret;
 403         int bitmap_pos;
 404
 405         hash_pos = kh_put_sha1_pos(eindex->positions, object->sha1, &hash_ret);
 406         if (hash_ret > 0) {
 407                 if (eindex->count >= eindex->alloc) {
 408                         eindex->alloc = (eindex->alloc + 16) * 3 / 2;
 409                         REALLOC_ARRAY(eindex->objects, eindex->alloc);
 410                         REALLOC_ARRAY(eindex->hashes, eindex->alloc);
 411                 }
 412
 413                 bitmap_pos = eindex->count;
 414                 eindex->objects[eindex->count] = object;
 415                 eindex->hashes[eindex->count] = pack_name_hash(name);
 416                 kh_value(eindex->positions, hash_pos) = bitmap_pos;
 417                 eindex->count++;
 418         } else {
 419                 bitmap_pos = kh_value(eindex->positions, hash_pos);
 420         }
 421
 422         return bitmap_pos + bitmap_git.pack->num_objects;
 423 }
 424
 425 static void show_object(struct object *object, const char *name, void *data)
 426 {
 427         struct bitmap *base = data;
 428         int bitmap_pos;
 429
 430         bitmap_pos = bitmap_position(object->sha1);
 431
 432         if (bitmap_pos < 0)
 433                 bitmap_pos = ext_index_add_object(object, name);
 434
 435         bitmap_set(base, bitmap_pos);
 436 }
 437
 438 static void show_commit(struct commit *commit, void *data)
 439 {
 440 }
 441
 442 static int add_to_include_set(struct include_data *data,
 443                               const unsigned char *sha1,
 444                               int bitmap_pos)
 445 {
 446         khiter_t hash_pos;
 447
 448         if (data->seen && bitmap_get(data->seen, bitmap_pos))
 449                 return 0;
 450
 451         if (bitmap_get(data->base, bitmap_pos))
 452                 return 0;
 453
 454         hash_pos = kh_get_sha1(bitmap_git.bitmaps, sha1);
 455         if (hash_pos < kh_end(bitmap_git.bitmaps)) {
 456                 struct stored_bitmap *st = kh_value(bitmap_git.bitmaps, hash_pos);
 457                 bitmap_or_ewah(data->base, lookup_stored_bitmap(st));
 458                 return 0;
 459         }
 460
 461         bitmap_set(data->base, bitmap_pos);
 462         return 1;
 463 }
 464
 465 static int should_include(struct commit *commit, void *_data)
 466 {
 467         struct include_data *data = _data;
 468         int bitmap_pos;
 469
 470         bitmap_pos = bitmap_position(commit->object.sha1);
 471         if (bitmap_pos < 0)
 472                 bitmap_pos = ext_index_add_object((struct object *)commit, NULL);
 473
 474         if (!add_to_include_set(data, commit->object.sha1, bitmap_pos)) {
 475                 struct commit_list *parent = commit->parents;
 476
 477                 while (parent) {
 478                         parent->item->object.flags |= SEEN;
 479                         parent = parent->next;
 480                 }
 481
 482                 return 0;
 483         }
 484
 485         return 1;
 486 }
 487
 488 static struct bitmap *find_objects(struct rev_info *revs,
 489                                    struct object_list *roots,
 490                                    struct bitmap *seen)
 491 {
 492         struct bitmap *base = NULL;
 493         int needs_walk = 0;
 494
 495         struct object_list *not_mapped = NULL;
 496
 497         /*
 498          * Go through all the roots for the walk. The ones that have bitmaps
 499          * on the bitmap index will be `or`ed together to form an initial
 500          * global reachability analysis.
 501          *
 502          * The ones without bitmaps in the index will be stored in the
 503          * `not_mapped_list` for further processing.
 504          */
 505         while (roots) {
 506                 struct object *object = roots->item;
 507                 roots = roots->next;
 508
 509                 if (object->type == OBJ_COMMIT) {
 510                         khiter_t pos = kh_get_sha1(bitmap_git.bitmaps, object->sha1);
 511
 512                         if (pos < kh_end(bitmap_git.bitmaps)) {
 513                                 struct stored_bitmap *st = kh_value(bitmap_git.bitmaps, pos);
 514                                 struct ewah_bitmap *or_with = lookup_stored_bitmap(st);
 515
 516                                 if (base == NULL)
 517                                         base = ewah_to_bitmap(or_with);
 518                                 else
 519                                         bitmap_or_ewah(base, or_with);
 520
 521                                 object->flags |= SEEN;
 522                                 continue;
 523                         }
 524                 }
 525
 526                 object_list_insert(object, &not_mapped);
 527         }
 528
 529         /*
 530          * Best case scenario: We found bitmaps for all the roots,
 531          * so the resulting `or` bitmap has the full reachability analysis
 532          */
 533         if (not_mapped == NULL)
 534                 return base;
 535
 536         roots = not_mapped;
 537
 538         /*
 539          * Let's iterate through all the roots that don't have bitmaps to
 540          * check if we can determine them to be reachable from the existing
 541          * global bitmap.
 542          *
 543          * If we cannot find them in the existing global bitmap, we'll need
 544          * to push them to an actual walk and run it until we can confirm
 545          * they are reachable
 546          */
 547         while (roots) {
 548                 struct object *object = roots->item;
 549                 int pos;
 550
 551                 roots = roots->next;
 552                 pos = bitmap_position(object->sha1);
 553
 554                 if (pos < 0 || base == NULL || !bitmap_get(base, pos)) {
 555                         object->flags &= ~UNINTERESTING;
 556                         add_pending_object(revs, object, "");
 557                         needs_walk = 1;
 558                 } else {
 559                         object->flags |= SEEN;
 560                 }
 561         }
 562
 563         if (needs_walk) {
 564                 struct include_data incdata;
 565
 566                 if (base == NULL)
 567                         base = bitmap_new();
 568
 569                 incdata.base = base;
 570                 incdata.seen = seen;
 571
 572                 revs->include_check = should_include;
 573                 revs->include_check_data = &incdata;
 574
 575                 if (prepare_revision_walk(revs))
 576                         die("revision walk setup failed");
 577
 578                 traverse_commit_list(revs, show_commit, show_object, base);
 579         }
 580
 581         return base;
 582 }
 583
 584 static void show_extended_objects(struct bitmap *objects,
 585                                   show_reachable_fn show_reach)
 586 {
 587         struct eindex *eindex = &bitmap_git.ext_index;
 588         uint32_t i;
 589
 590         for (i = 0; i < eindex->count; ++i) {
 591                 struct object *obj;
 592
 593                 if (!bitmap_get(objects, bitmap_git.pack->num_objects + i))
 594                         continue;
 595
 596                 obj = eindex->objects[i];
 597                 show_reach(obj->sha1, obj->type, 0, eindex->hashes[i], NULL, 0);
 598         }
 599 }
 600
 601 static void show_objects_for_type(
 602         struct bitmap *objects,
 603         struct ewah_bitmap *type_filter,
 604         enum object_type object_type,
 605         show_reachable_fn show_reach)
 606 {
 607         size_t pos = 0, i = 0;
 608         uint32_t offset;
 609
 610         struct ewah_iterator it;
 611         eword_t filter;
 612
 613         if (bitmap_git.reuse_objects == bitmap_git.pack->num_objects)
 614                 return;
 615
 616         ewah_iterator_init(&it, type_filter);
 617
 618         while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
 619                 eword_t word = objects->words[i] & filter;
 620
 621                 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
 622                         const unsigned char *sha1;
 623                         struct revindex_entry *entry;
 624                         uint32_t hash = 0;
 625
 626                         if ((word >> offset) == 0)
 627                                 break;
 628
 629                         offset += ewah_bit_ctz64(word >> offset);
 630
 631                         if (pos + offset < bitmap_git.reuse_objects)
 632                                 continue;
 633
 634                         entry = &bitmap_git.reverse_index->revindex[pos + offset];
 635                         sha1 = nth_packed_object_sha1(bitmap_git.pack, entry->nr);
 636
 637                         if (bitmap_git.hashes)
 638                                 hash = ntohl(bitmap_git.hashes[entry->nr]);
 639
 640                         show_reach(sha1, object_type, 0, hash, bitmap_git.pack, entry->offset);
 641                 }
 642
 643                 pos += BITS_IN_EWORD;
 644                 i++;
 645         }
 646 }
 647
 648 static int in_bitmapped_pack(struct object_list *roots)
 649 {
 650         while (roots) {
 651                 struct object *object = roots->item;
 652                 roots = roots->next;
 653
 654                 if (find_pack_entry_one(object->sha1, bitmap_git.pack) > 0)
 655                         return 1;
 656         }
 657
 658         return 0;
 659 }
 660
 661 int prepare_bitmap_walk(struct rev_info *revs)
 662 {
 663         unsigned int i;
 664         unsigned int pending_nr = revs->pending.nr;
 665         struct object_array_entry *pending_e = revs->pending.objects;
 666
 667         struct object_list *wants = NULL;
 668         struct object_list *haves = NULL;
 669
 670         struct bitmap *wants_bitmap = NULL;
 671         struct bitmap *haves_bitmap = NULL;
 672
 673         if (!bitmap_git.loaded) {
 674                 /* try to open a bitmapped pack, but don't parse it yet
 675                  * because we may not need to use it */
 676                 if (open_pack_bitmap() < 0)
 677                         return -1;
 678         }
 679
 680         for (i = 0; i < pending_nr; ++i) {
 681                 struct object *object = pending_e[i].item;
 682
 683                 if (object->type == OBJ_NONE)
 684                         parse_object_or_die(object->sha1, NULL);
 685
 686                 while (object->type == OBJ_TAG) {
 687                         struct tag *tag = (struct tag *) object;
 688
 689                         if (object->flags & UNINTERESTING)
 690                                 object_list_insert(object, &haves);
 691                         else
 692                                 object_list_insert(object, &wants);
 693
 694                         if (!tag->tagged)
 695                                 die("bad tag");
 696                         object = parse_object_or_die(tag->tagged->sha1, NULL);
 697                 }
 698
 699                 if (object->flags & UNINTERESTING)
 700                         object_list_insert(object, &haves);
 701                 else
 702                         object_list_insert(object, &wants);
 703         }
 704
 705         /*
 706          * if we have a HAVES list, but none of those haves is contained
 707          * in the packfile that has a bitmap, we don't have anything to
 708          * optimize here
 709          */
 710         if (haves && !in_bitmapped_pack(haves))
 711                 return -1;
 712
 713         /* if we don't want anything, we're done here */
 714         if (!wants)
 715                 return -1;
 716
 717         /*
 718          * now we're going to use bitmaps, so load the actual bitmap entries
 719          * from disk. this is the point of no return; after this the rev_list
 720          * becomes invalidated and we must perform the revwalk through bitmaps
 721          */
 722         if (!bitmap_git.loaded && load_pack_bitmap() < 0)
 723                 return -1;
 724
 725         revs->pending.nr = 0;
 726         revs->pending.alloc = 0;
 727         revs->pending.objects = NULL;
 728
 729         if (haves) {
 730                 revs->ignore_missing_links = 1;
 731                 haves_bitmap = find_objects(revs, haves, NULL);
 732                 reset_revision_walk();
 733                 revs->ignore_missing_links = 0;
 734
 735                 if (haves_bitmap == NULL)
 736                         die("BUG: failed to perform bitmap walk");
 737         }
 738
 739         wants_bitmap = find_objects(revs, wants, haves_bitmap);
 740
 741         if (!wants_bitmap)
 742                 die("BUG: failed to perform bitmap walk");
 743
 744         if (haves_bitmap)
 745                 bitmap_and_not(wants_bitmap, haves_bitmap);
 746
 747         bitmap_git.result = wants_bitmap;
 748
 749         bitmap_free(haves_bitmap);
 750         return 0;
 751 }
 752
 753 int reuse_partial_packfile_from_bitmap(struct packed_git **packfile,
 754                                        uint32_t *entries,
 755                                        off_t *up_to)
 756 {
 757         /*
 758          * Reuse the packfile content if we need more than
 759          * 90% of its objects
 760          */
 761         static const double REUSE_PERCENT = 0.9;
 762
 763         struct bitmap *result = bitmap_git.result;
 764         uint32_t reuse_threshold;
 765         uint32_t i, reuse_objects = 0;
 766
 767         assert(result);
 768
 769         for (i = 0; i < result->word_alloc; ++i) {
 770                 if (result->words[i] != (eword_t)~0) {
 771                         reuse_objects += ewah_bit_ctz64(~result->words[i]);
 772                         break;
 773                 }
 774
 775                 reuse_objects += BITS_IN_EWORD;
 776         }
 777
 778 #ifdef GIT_BITMAP_DEBUG
 779         {
 780                 const unsigned char *sha1;
 781                 struct revindex_entry *entry;
 782
 783                 entry = &bitmap_git.reverse_index->revindex[reuse_objects];
 784                 sha1 = nth_packed_object_sha1(bitmap_git.pack, entry->nr);
 785
 786                 fprintf(stderr, "Failed to reuse at %d (%016llx)\n",
 787                         reuse_objects, result->words[i]);
 788                 fprintf(stderr, " %s\n", sha1_to_hex(sha1));
 789         }
 790 #endif
 791
 792         if (!reuse_objects)
 793                 return -1;
 794
 795         if (reuse_objects >= bitmap_git.pack->num_objects) {
 796                 bitmap_git.reuse_objects = *entries = bitmap_git.pack->num_objects;
 797                 *up_to = -1; /* reuse the full pack */
 798                 *packfile = bitmap_git.pack;
 799                 return 0;
 800         }
 801
 802         reuse_threshold = bitmap_popcount(bitmap_git.result) * REUSE_PERCENT;
 803
 804         if (reuse_objects < reuse_threshold)
 805                 return -1;
 806
 807         bitmap_git.reuse_objects = *entries = reuse_objects;
 808         *up_to = bitmap_git.reverse_index->revindex[reuse_objects].offset;
 809         *packfile = bitmap_git.pack;
 810
 811         return 0;
 812 }
 813
 814 void traverse_bitmap_commit_list(show_reachable_fn show_reachable)
 815 {
 816         assert(bitmap_git.result);
 817
 818         show_objects_for_type(bitmap_git.result, bitmap_git.commits,
 819                 OBJ_COMMIT, show_reachable);
 820         show_objects_for_type(bitmap_git.result, bitmap_git.trees,
 821                 OBJ_TREE, show_reachable);
 822         show_objects_for_type(bitmap_git.result, bitmap_git.blobs,
 823                 OBJ_BLOB, show_reachable);
 824         show_objects_for_type(bitmap_git.result, bitmap_git.tags,
 825                 OBJ_TAG, show_reachable);
 826
 827         show_extended_objects(bitmap_git.result, show_reachable);
 828
 829         bitmap_free(bitmap_git.result);
 830         bitmap_git.result = NULL;
 831 }
 832
 833 static uint32_t count_object_type(struct bitmap *objects,
 834                                   enum object_type type)
 835 {
 836         struct eindex *eindex = &bitmap_git.ext_index;
 837
 838         uint32_t i = 0, count = 0;
 839         struct ewah_iterator it;
 840         eword_t filter;
 841
 842         switch (type) {
 843         case OBJ_COMMIT:
 844                 ewah_iterator_init(&it, bitmap_git.commits);
 845                 break;
 846
 847         case OBJ_TREE:
 848                 ewah_iterator_init(&it, bitmap_git.trees);
 849                 break;
 850
 851         case OBJ_BLOB:
 852                 ewah_iterator_init(&it, bitmap_git.blobs);
 853                 break;
 854
 855         case OBJ_TAG:
 856                 ewah_iterator_init(&it, bitmap_git.tags);
 857                 break;
 858
 859         default:
 860                 return 0;
 861         }
 862
 863         while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
 864                 eword_t word = objects->words[i++] & filter;
 865                 count += ewah_bit_popcount64(word);
 866         }
 867
 868         for (i = 0; i < eindex->count; ++i) {
 869                 if (eindex->objects[i]->type == type &&
 870                         bitmap_get(objects, bitmap_git.pack->num_objects + i))
 871                         count++;
 872         }
 873
 874         return count;
 875 }
 876
 877 void count_bitmap_commit_list(uint32_t *commits, uint32_t *trees,
 878                               uint32_t *blobs, uint32_t *tags)
 879 {
 880         assert(bitmap_git.result);
 881
 882         if (commits)
 883                 *commits = count_object_type(bitmap_git.result, OBJ_COMMIT);
 884
 885         if (trees)
 886                 *trees = count_object_type(bitmap_git.result, OBJ_TREE);
 887
 888         if (blobs)
 889                 *blobs = count_object_type(bitmap_git.result, OBJ_BLOB);
 890
 891         if (tags)
 892                 *tags = count_object_type(bitmap_git.result, OBJ_TAG);
 893 }
 894
 895 struct bitmap_test_data {
 896         struct bitmap *base;
 897         struct progress *prg;
 898         size_t seen;
 899 };
 900
 901 static void test_show_object(struct object *object, const char *name,
 902                              void *data)
 903 {
 904         struct bitmap_test_data *tdata = data;
 905         int bitmap_pos;
 906
 907         bitmap_pos = bitmap_position(object->sha1);
 908         if (bitmap_pos < 0)
 909                 die("Object not in bitmap: %s\n", sha1_to_hex(object->sha1));
 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.sha1);
 921         if (bitmap_pos < 0)
 922                 die("Object not in bitmap: %s\n", sha1_to_hex(commit->object.sha1));
 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->sha1);
 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                         sha1_to_hex(root->sha1), (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", sha1_to_hex(root->sha1));
 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 }