git.oblomov.eu Git - git/blob - pack-bitmap.c

   1 #include "cache.h"
   2 #include "commit.h"
   3 #include "tag.h"
   4 #include "diff.h"
   5 #include "revision.h"
   6 #include "progress.h"
   7 #include "list-objects.h"
   8 #include "pack.h"
   9 #include "pack-bitmap.h"
  10 #include "pack-revindex.h"
  11 #include "pack-objects.h"
  12 #include "packfile.h"
  13 #include "repository.h"
  14 #include "object-store.h"
  15
  16 /*
  17  * An entry on the bitmap index, representing the bitmap for a given
  18  * commit.
  19  */
  20 struct stored_bitmap {
  21         unsigned char sha1[20];
  22         struct ewah_bitmap *root;
  23         struct stored_bitmap *xor;
  24         int flags;
  25 };
  26
  27 /*
  28  * The active bitmap index for a repository. By design, repositories only have
  29  * a single bitmap index available (the index for the biggest packfile in
  30  * the repository), since bitmap indexes need full closure.
  31  *
  32  * If there is more than one bitmap index available (e.g. because of alternates),
  33  * the active bitmap index is the largest one.
  34  */
  35 struct bitmap_index {
  36         /* Packfile to which this bitmap index belongs to */
  37         struct packed_git *pack;
  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
  95 static struct ewah_bitmap *lookup_stored_bitmap(struct stored_bitmap *st)
  96 {
  97         struct ewah_bitmap *parent;
  98         struct ewah_bitmap *composed;
  99
 100         if (st->xor == NULL)
 101                 return st->root;
 102
 103         composed = ewah_pool_new();
 104         parent = lookup_stored_bitmap(st->xor);
 105         ewah_xor(st->root, parent, composed);
 106
 107         ewah_pool_free(st->root);
 108         st->root = composed;
 109         st->xor = NULL;
 110
 111         return composed;
 112 }
 113
 114 /*
 115  * Read a bitmap from the current read position on the mmaped
 116  * index, and increase the read position accordingly
 117  */
 118 static struct ewah_bitmap *read_bitmap_1(struct bitmap_index *index)
 119 {
 120         struct ewah_bitmap *b = ewah_pool_new();
 121
 122         ssize_t bitmap_size = ewah_read_mmap(b,
 123                 index->map + index->map_pos,
 124                 index->map_size - index->map_pos);
 125
 126         if (bitmap_size < 0) {
 127                 error("Failed to load bitmap index (corrupted?)");
 128                 ewah_pool_free(b);
 129                 return NULL;
 130         }
 131
 132         index->map_pos += bitmap_size;
 133         return b;
 134 }
 135
 136 static int load_bitmap_header(struct bitmap_index *index)
 137 {
 138         struct bitmap_disk_header *header = (void *)index->map;
 139
 140         if (index->map_size < sizeof(*header) + 20)
 141                 return error("Corrupted bitmap index (missing header data)");
 142
 143         if (memcmp(header->magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE)) != 0)
 144                 return error("Corrupted bitmap index file (wrong header)");
 145
 146         index->version = ntohs(header->version);
 147         if (index->version != 1)
 148                 return error("Unsupported version for bitmap index file (%d)", index->version);
 149
 150         /* Parse known bitmap format options */
 151         {
 152                 uint32_t flags = ntohs(header->options);
 153
 154                 if ((flags & BITMAP_OPT_FULL_DAG) == 0)
 155                         return error("Unsupported options for bitmap index file "
 156                                 "(Git requires BITMAP_OPT_FULL_DAG)");
 157
 158                 if (flags & BITMAP_OPT_HASH_CACHE) {
 159                         unsigned char *end = index->map + index->map_size - 20;
 160                         index->hashes = ((uint32_t *)end) - index->pack->num_objects;
 161                 }
 162         }
 163
 164         index->entry_count = ntohl(header->entry_count);
 165         index->map_pos += sizeof(*header);
 166         return 0;
 167 }
 168
 169 static struct stored_bitmap *store_bitmap(struct bitmap_index *index,
 170                                           struct ewah_bitmap *root,
 171                                           const unsigned char *sha1,
 172                                           struct stored_bitmap *xor_with,
 173                                           int flags)
 174 {
 175         struct stored_bitmap *stored;
 176         khiter_t hash_pos;
 177         int ret;
 178
 179         stored = xmalloc(sizeof(struct stored_bitmap));
 180         stored->root = root;
 181         stored->xor = xor_with;
 182         stored->flags = flags;
 183         hashcpy(stored->sha1, sha1);
 184
 185         hash_pos = kh_put_sha1(index->bitmaps, stored->sha1, &ret);
 186
 187         /* a 0 return code means the insertion succeeded with no changes,
 188          * because the SHA1 already existed on the map. this is bad, there
 189          * shouldn't be duplicated commits in the index */
 190         if (ret == 0) {
 191                 error("Duplicate entry in bitmap index: %s", sha1_to_hex(sha1));
 192                 return NULL;
 193         }
 194
 195         kh_value(index->bitmaps, hash_pos) = stored;
 196         return stored;
 197 }
 198
 199 static inline uint32_t read_be32(const unsigned char *buffer, size_t *pos)
 200 {
 201         uint32_t result = get_be32(buffer + *pos);
 202         (*pos) += sizeof(result);
 203         return result;
 204 }
 205
 206 static inline uint8_t read_u8(const unsigned char *buffer, size_t *pos)
 207 {
 208         return buffer[(*pos)++];
 209 }
 210
 211 #define MAX_XOR_OFFSET 160
 212
 213 static int load_bitmap_entries_v1(struct bitmap_index *index)
 214 {
 215         uint32_t i;
 216         struct stored_bitmap *recent_bitmaps[MAX_XOR_OFFSET] = { NULL };
 217
 218         for (i = 0; i < index->entry_count; ++i) {
 219                 int xor_offset, flags;
 220                 struct ewah_bitmap *bitmap = NULL;
 221                 struct stored_bitmap *xor_bitmap = NULL;
 222                 uint32_t commit_idx_pos;
 223                 const unsigned char *sha1;
 224
 225                 commit_idx_pos = read_be32(index->map, &index->map_pos);
 226                 xor_offset = read_u8(index->map, &index->map_pos);
 227                 flags = read_u8(index->map, &index->map_pos);
 228
 229                 sha1 = nth_packed_object_sha1(index->pack, commit_idx_pos);
 230
 231                 bitmap = read_bitmap_1(index);
 232                 if (!bitmap)
 233                         return -1;
 234
 235                 if (xor_offset > MAX_XOR_OFFSET || xor_offset > i)
 236                         return error("Corrupted bitmap pack index");
 237
 238                 if (xor_offset > 0) {
 239                         xor_bitmap = recent_bitmaps[(i - xor_offset) % MAX_XOR_OFFSET];
 240
 241                         if (xor_bitmap == NULL)
 242                                 return error("Invalid XOR offset in bitmap pack index");
 243                 }
 244
 245                 recent_bitmaps[i % MAX_XOR_OFFSET] = store_bitmap(
 246                         index, bitmap, sha1, xor_bitmap, flags);
 247         }
 248
 249         return 0;
 250 }
 251
 252 static char *pack_bitmap_filename(struct packed_git *p)
 253 {
 254         size_t len;
 255
 256         if (!strip_suffix(p->pack_name, ".pack", &len))
 257                 BUG("pack_name does not end in .pack");
 258         return xstrfmt("%.*s.bitmap", (int)len, p->pack_name);
 259 }
 260
 261 static int open_pack_bitmap_1(struct bitmap_index *bitmap_git, struct packed_git *packfile)
 262 {
 263         int fd;
 264         struct stat st;
 265         char *idx_name;
 266
 267         if (open_pack_index(packfile))
 268                 return -1;
 269
 270         idx_name = pack_bitmap_filename(packfile);
 271         fd = git_open(idx_name);
 272         free(idx_name);
 273
 274         if (fd < 0)
 275                 return -1;
 276
 277         if (fstat(fd, &st)) {
 278                 close(fd);
 279                 return -1;
 280         }
 281
 282         if (bitmap_git->pack) {
 283                 warning("ignoring extra bitmap file: %s", packfile->pack_name);
 284                 close(fd);
 285                 return -1;
 286         }
 287
 288         bitmap_git->pack = packfile;
 289         bitmap_git->map_size = xsize_t(st.st_size);
 290         bitmap_git->map = xmmap(NULL, bitmap_git->map_size, PROT_READ, MAP_PRIVATE, fd, 0);
 291         bitmap_git->map_pos = 0;
 292         close(fd);
 293
 294         if (load_bitmap_header(bitmap_git) < 0) {
 295                 munmap(bitmap_git->map, bitmap_git->map_size);
 296                 bitmap_git->map = NULL;
 297                 bitmap_git->map_size = 0;
 298                 return -1;
 299         }
 300
 301         return 0;
 302 }
 303
 304 static int load_pack_bitmap(struct bitmap_index *bitmap_git)
 305 {
 306         assert(bitmap_git->map && !bitmap_git->loaded);
 307
 308         bitmap_git->bitmaps = kh_init_sha1();
 309         bitmap_git->ext_index.positions = kh_init_sha1_pos();
 310         load_pack_revindex(bitmap_git->pack);
 311
 312         if (!(bitmap_git->commits = read_bitmap_1(bitmap_git)) ||
 313                 !(bitmap_git->trees = read_bitmap_1(bitmap_git)) ||
 314                 !(bitmap_git->blobs = read_bitmap_1(bitmap_git)) ||
 315                 !(bitmap_git->tags = read_bitmap_1(bitmap_git)))
 316                 goto failed;
 317
 318         if (load_bitmap_entries_v1(bitmap_git) < 0)
 319                 goto failed;
 320
 321         bitmap_git->loaded = 1;
 322         return 0;
 323
 324 failed:
 325         munmap(bitmap_git->map, bitmap_git->map_size);
 326         bitmap_git->map = NULL;
 327         bitmap_git->map_size = 0;
 328         return -1;
 329 }
 330
 331 static int open_pack_bitmap(struct bitmap_index *bitmap_git)
 332 {
 333         struct packed_git *p;
 334         int ret = -1;
 335
 336         assert(!bitmap_git->map && !bitmap_git->loaded);
 337
 338         for (p = get_packed_git(the_repository); p; p = p->next) {
 339                 if (open_pack_bitmap_1(bitmap_git, p) == 0)
 340                         ret = 0;
 341         }
 342
 343         return ret;
 344 }
 345
 346 struct bitmap_index *prepare_bitmap_git(void)
 347 {
 348         struct bitmap_index *bitmap_git = xcalloc(1, sizeof(*bitmap_git));
 349
 350         if (!open_pack_bitmap(bitmap_git) && !load_pack_bitmap(bitmap_git))
 351                 return bitmap_git;
 352
 353         return NULL;
 354 }
 355
 356 struct include_data {
 357         struct bitmap_index *bitmap_git;
 358         struct bitmap *base;
 359         struct bitmap *seen;
 360 };
 361
 362 static inline int bitmap_position_extended(struct bitmap_index *bitmap_git,
 363                                            const unsigned char *sha1)
 364 {
 365         khash_sha1_pos *positions = bitmap_git->ext_index.positions;
 366         khiter_t pos = kh_get_sha1_pos(positions, sha1);
 367
 368         if (pos < kh_end(positions)) {
 369                 int bitmap_pos = kh_value(positions, pos);
 370                 return bitmap_pos + bitmap_git->pack->num_objects;
 371         }
 372
 373         return -1;
 374 }
 375
 376 static inline int bitmap_position_packfile(struct bitmap_index *bitmap_git,
 377                                            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->pack, offset);
 384 }
 385
 386 static int bitmap_position(struct bitmap_index *bitmap_git,
 387                            const unsigned char *sha1)
 388 {
 389         int pos = bitmap_position_packfile(bitmap_git, sha1);
 390         return (pos >= 0) ? pos : bitmap_position_extended(bitmap_git, sha1);
 391 }
 392
 393 static int ext_index_add_object(struct bitmap_index *bitmap_git,
 394                                 struct object *object, const char *name)
 395 {
 396         struct eindex *eindex = &bitmap_git->ext_index;
 397
 398         khiter_t hash_pos;
 399         int hash_ret;
 400         int bitmap_pos;
 401
 402         hash_pos = kh_put_sha1_pos(eindex->positions, object->oid.hash, &hash_ret);
 403         if (hash_ret > 0) {
 404                 if (eindex->count >= eindex->alloc) {
 405                         eindex->alloc = (eindex->alloc + 16) * 3 / 2;
 406                         REALLOC_ARRAY(eindex->objects, eindex->alloc);
 407                         REALLOC_ARRAY(eindex->hashes, eindex->alloc);
 408                 }
 409
 410                 bitmap_pos = eindex->count;
 411                 eindex->objects[eindex->count] = object;
 412                 eindex->hashes[eindex->count] = pack_name_hash(name);
 413                 kh_value(eindex->positions, hash_pos) = bitmap_pos;
 414                 eindex->count++;
 415         } else {
 416                 bitmap_pos = kh_value(eindex->positions, hash_pos);
 417         }
 418
 419         return bitmap_pos + bitmap_git->pack->num_objects;
 420 }
 421
 422 struct bitmap_show_data {
 423         struct bitmap_index *bitmap_git;
 424         struct bitmap *base;
 425 };
 426
 427 static void show_object(struct object *object, const char *name, void *data_)
 428 {
 429         struct bitmap_show_data *data = data_;
 430         int bitmap_pos;
 431
 432         bitmap_pos = bitmap_position(data->bitmap_git, object->oid.hash);
 433
 434         if (bitmap_pos < 0)
 435                 bitmap_pos = ext_index_add_object(data->bitmap_git, object,
 436                                                   name);
 437
 438         bitmap_set(data->base, bitmap_pos);
 439 }
 440
 441 static void show_commit(struct commit *commit, void *data)
 442 {
 443 }
 444
 445 static int add_to_include_set(struct bitmap_index *bitmap_git,
 446                               struct include_data *data,
 447                               const unsigned char *sha1,
 448                               int bitmap_pos)
 449 {
 450         khiter_t hash_pos;
 451
 452         if (data->seen && bitmap_get(data->seen, bitmap_pos))
 453                 return 0;
 454
 455         if (bitmap_get(data->base, bitmap_pos))
 456                 return 0;
 457
 458         hash_pos = kh_get_sha1(bitmap_git->bitmaps, sha1);
 459         if (hash_pos < kh_end(bitmap_git->bitmaps)) {
 460                 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, hash_pos);
 461                 bitmap_or_ewah(data->base, lookup_stored_bitmap(st));
 462                 return 0;
 463         }
 464
 465         bitmap_set(data->base, bitmap_pos);
 466         return 1;
 467 }
 468
 469 static int should_include(struct commit *commit, void *_data)
 470 {
 471         struct include_data *data = _data;
 472         int bitmap_pos;
 473
 474         bitmap_pos = bitmap_position(data->bitmap_git, commit->object.oid.hash);
 475         if (bitmap_pos < 0)
 476                 bitmap_pos = ext_index_add_object(data->bitmap_git,
 477                                                   (struct object *)commit,
 478                                                   NULL);
 479
 480         if (!add_to_include_set(data->bitmap_git, data, commit->object.oid.hash,
 481                                 bitmap_pos)) {
 482                 struct commit_list *parent = commit->parents;
 483
 484                 while (parent) {
 485                         parent->item->object.flags |= SEEN;
 486                         parent = parent->next;
 487                 }
 488
 489                 return 0;
 490         }
 491
 492         return 1;
 493 }
 494
 495 static struct bitmap *find_objects(struct bitmap_index *bitmap_git,
 496                                    struct rev_info *revs,
 497                                    struct object_list *roots,
 498                                    struct bitmap *seen)
 499 {
 500         struct bitmap *base = NULL;
 501         int needs_walk = 0;
 502
 503         struct object_list *not_mapped = NULL;
 504
 505         /*
 506          * Go through all the roots for the walk. The ones that have bitmaps
 507          * on the bitmap index will be `or`ed together to form an initial
 508          * global reachability analysis.
 509          *
 510          * The ones without bitmaps in the index will be stored in the
 511          * `not_mapped_list` for further processing.
 512          */
 513         while (roots) {
 514                 struct object *object = roots->item;
 515                 roots = roots->next;
 516
 517                 if (object->type == OBJ_COMMIT) {
 518                         khiter_t pos = kh_get_sha1(bitmap_git->bitmaps, object->oid.hash);
 519
 520                         if (pos < kh_end(bitmap_git->bitmaps)) {
 521                                 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, pos);
 522                                 struct ewah_bitmap *or_with = lookup_stored_bitmap(st);
 523
 524                                 if (base == NULL)
 525                                         base = ewah_to_bitmap(or_with);
 526                                 else
 527                                         bitmap_or_ewah(base, or_with);
 528
 529                                 object->flags |= SEEN;
 530                                 continue;
 531                         }
 532                 }
 533
 534                 object_list_insert(object, &not_mapped);
 535         }
 536
 537         /*
 538          * Best case scenario: We found bitmaps for all the roots,
 539          * so the resulting `or` bitmap has the full reachability analysis
 540          */
 541         if (not_mapped == NULL)
 542                 return base;
 543
 544         roots = not_mapped;
 545
 546         /*
 547          * Let's iterate through all the roots that don't have bitmaps to
 548          * check if we can determine them to be reachable from the existing
 549          * global bitmap.
 550          *
 551          * If we cannot find them in the existing global bitmap, we'll need
 552          * to push them to an actual walk and run it until we can confirm
 553          * they are reachable
 554          */
 555         while (roots) {
 556                 struct object *object = roots->item;
 557                 int pos;
 558
 559                 roots = roots->next;
 560                 pos = bitmap_position(bitmap_git, object->oid.hash);
 561
 562                 if (pos < 0 || base == NULL || !bitmap_get(base, pos)) {
 563                         object->flags &= ~UNINTERESTING;
 564                         add_pending_object(revs, object, "");
 565                         needs_walk = 1;
 566                 } else {
 567                         object->flags |= SEEN;
 568                 }
 569         }
 570
 571         if (needs_walk) {
 572                 struct include_data incdata;
 573                 struct bitmap_show_data show_data;
 574
 575                 if (base == NULL)
 576                         base = bitmap_new();
 577
 578                 incdata.bitmap_git = bitmap_git;
 579                 incdata.base = base;
 580                 incdata.seen = seen;
 581
 582                 revs->include_check = should_include;
 583                 revs->include_check_data = &incdata;
 584
 585                 if (prepare_revision_walk(revs))
 586                         die("revision walk setup failed");
 587
 588                 show_data.bitmap_git = bitmap_git;
 589                 show_data.base = base;
 590
 591                 traverse_commit_list(revs, show_commit, show_object,
 592                                      &show_data);
 593         }
 594
 595         return base;
 596 }
 597
 598 static void show_extended_objects(struct bitmap_index *bitmap_git,
 599                                   show_reachable_fn show_reach)
 600 {
 601         struct bitmap *objects = bitmap_git->result;
 602         struct eindex *eindex = &bitmap_git->ext_index;
 603         uint32_t i;
 604
 605         for (i = 0; i < eindex->count; ++i) {
 606                 struct object *obj;
 607
 608                 if (!bitmap_get(objects, bitmap_git->pack->num_objects + i))
 609                         continue;
 610
 611                 obj = eindex->objects[i];
 612                 show_reach(&obj->oid, obj->type, 0, eindex->hashes[i], NULL, 0);
 613         }
 614 }
 615
 616 static void show_objects_for_type(
 617         struct bitmap_index *bitmap_git,
 618         struct ewah_bitmap *type_filter,
 619         enum object_type object_type,
 620         show_reachable_fn show_reach)
 621 {
 622         size_t pos = 0, i = 0;
 623         uint32_t offset;
 624
 625         struct ewah_iterator it;
 626         eword_t filter;
 627
 628         struct bitmap *objects = bitmap_git->result;
 629
 630         if (bitmap_git->reuse_objects == bitmap_git->pack->num_objects)
 631                 return;
 632
 633         ewah_iterator_init(&it, type_filter);
 634
 635         while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
 636                 eword_t word = objects->words[i] & filter;
 637
 638                 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
 639                         struct object_id oid;
 640                         struct revindex_entry *entry;
 641                         uint32_t hash = 0;
 642
 643                         if ((word >> offset) == 0)
 644                                 break;
 645
 646                         offset += ewah_bit_ctz64(word >> offset);
 647
 648                         if (pos + offset < bitmap_git->reuse_objects)
 649                                 continue;
 650
 651                         entry = &bitmap_git->pack->revindex[pos + offset];
 652                         nth_packed_object_oid(&oid, bitmap_git->pack, entry->nr);
 653
 654                         if (bitmap_git->hashes)
 655                                 hash = get_be32(bitmap_git->hashes + entry->nr);
 656
 657                         show_reach(&oid, object_type, 0, hash, bitmap_git->pack, entry->offset);
 658                 }
 659
 660                 pos += BITS_IN_EWORD;
 661                 i++;
 662         }
 663 }
 664
 665 static int in_bitmapped_pack(struct bitmap_index *bitmap_git,
 666                              struct object_list *roots)
 667 {
 668         while (roots) {
 669                 struct object *object = roots->item;
 670                 roots = roots->next;
 671
 672                 if (find_pack_entry_one(object->oid.hash, bitmap_git->pack) > 0)
 673                         return 1;
 674         }
 675
 676         return 0;
 677 }
 678
 679 struct bitmap_index *prepare_bitmap_walk(struct rev_info *revs)
 680 {
 681         unsigned int i;
 682
 683         struct object_list *wants = NULL;
 684         struct object_list *haves = NULL;
 685
 686         struct bitmap *wants_bitmap = NULL;
 687         struct bitmap *haves_bitmap = NULL;
 688
 689         struct bitmap_index *bitmap_git = xcalloc(1, sizeof(*bitmap_git));
 690         /* try to open a bitmapped pack, but don't parse it yet
 691          * because we may not need to use it */
 692         if (open_pack_bitmap(bitmap_git) < 0)
 693                 return NULL;
 694
 695         for (i = 0; i < revs->pending.nr; ++i) {
 696                 struct object *object = revs->pending.objects[i].item;
 697
 698                 if (object->type == OBJ_NONE)
 699                         parse_object_or_die(&object->oid, NULL);
 700
 701                 while (object->type == OBJ_TAG) {
 702                         struct tag *tag = (struct tag *) object;
 703
 704                         if (object->flags & UNINTERESTING)
 705                                 object_list_insert(object, &haves);
 706                         else
 707                                 object_list_insert(object, &wants);
 708
 709                         if (!tag->tagged)
 710                                 die("bad tag");
 711                         object = parse_object_or_die(&tag->tagged->oid, NULL);
 712                 }
 713
 714                 if (object->flags & UNINTERESTING)
 715                         object_list_insert(object, &haves);
 716                 else
 717                         object_list_insert(object, &wants);
 718         }
 719
 720         /*
 721          * if we have a HAVES list, but none of those haves is contained
 722          * in the packfile that has a bitmap, we don't have anything to
 723          * optimize here
 724          */
 725         if (haves && !in_bitmapped_pack(bitmap_git, haves))
 726                 return NULL;
 727
 728         /* if we don't want anything, we're done here */
 729         if (!wants)
 730                 return NULL;
 731
 732         /*
 733          * now we're going to use bitmaps, so load the actual bitmap entries
 734          * from disk. this is the point of no return; after this the rev_list
 735          * becomes invalidated and we must perform the revwalk through bitmaps
 736          */
 737         if (!bitmap_git->loaded && load_pack_bitmap(bitmap_git) < 0)
 738                 return NULL;
 739
 740         object_array_clear(&revs->pending);
 741
 742         if (haves) {
 743                 revs->ignore_missing_links = 1;
 744                 haves_bitmap = find_objects(bitmap_git, revs, haves, NULL);
 745                 reset_revision_walk();
 746                 revs->ignore_missing_links = 0;
 747
 748                 if (haves_bitmap == NULL)
 749                         BUG("failed to perform bitmap walk");
 750         }
 751
 752         wants_bitmap = find_objects(bitmap_git, revs, wants, haves_bitmap);
 753
 754         if (!wants_bitmap)
 755                 BUG("failed to perform bitmap walk");
 756
 757         if (haves_bitmap)
 758                 bitmap_and_not(wants_bitmap, haves_bitmap);
 759
 760         bitmap_git->result = wants_bitmap;
 761
 762         bitmap_free(haves_bitmap);
 763         return bitmap_git;
 764 }
 765
 766 int reuse_partial_packfile_from_bitmap(struct bitmap_index *bitmap_git,
 767                                        struct packed_git **packfile,
 768                                        uint32_t *entries,
 769                                        off_t *up_to)
 770 {
 771         /*
 772          * Reuse the packfile content if we need more than
 773          * 90% of its objects
 774          */
 775         static const double REUSE_PERCENT = 0.9;
 776
 777         struct bitmap *result = bitmap_git->result;
 778         uint32_t reuse_threshold;
 779         uint32_t i, reuse_objects = 0;
 780
 781         assert(result);
 782
 783         for (i = 0; i < result->word_alloc; ++i) {
 784                 if (result->words[i] != (eword_t)~0) {
 785                         reuse_objects += ewah_bit_ctz64(~result->words[i]);
 786                         break;
 787                 }
 788
 789                 reuse_objects += BITS_IN_EWORD;
 790         }
 791
 792 #ifdef GIT_BITMAP_DEBUG
 793         {
 794                 const unsigned char *sha1;
 795                 struct revindex_entry *entry;
 796
 797                 entry = &bitmap_git->reverse_index->revindex[reuse_objects];
 798                 sha1 = nth_packed_object_sha1(bitmap_git->pack, entry->nr);
 799
 800                 fprintf(stderr, "Failed to reuse at %d (%016llx)\n",
 801                         reuse_objects, result->words[i]);
 802                 fprintf(stderr, " %s\n", sha1_to_hex(sha1));
 803         }
 804 #endif
 805
 806         if (!reuse_objects)
 807                 return -1;
 808
 809         if (reuse_objects >= bitmap_git->pack->num_objects) {
 810                 bitmap_git->reuse_objects = *entries = bitmap_git->pack->num_objects;
 811                 *up_to = -1; /* reuse the full pack */
 812                 *packfile = bitmap_git->pack;
 813                 return 0;
 814         }
 815
 816         reuse_threshold = bitmap_popcount(bitmap_git->result) * REUSE_PERCENT;
 817
 818         if (reuse_objects < reuse_threshold)
 819                 return -1;
 820
 821         bitmap_git->reuse_objects = *entries = reuse_objects;
 822         *up_to = bitmap_git->pack->revindex[reuse_objects].offset;
 823         *packfile = bitmap_git->pack;
 824
 825         return 0;
 826 }
 827
 828 void traverse_bitmap_commit_list(struct bitmap_index *bitmap_git,
 829                                  show_reachable_fn show_reachable)
 830 {
 831         assert(bitmap_git->result);
 832
 833         show_objects_for_type(bitmap_git, bitmap_git->commits,
 834                 OBJ_COMMIT, show_reachable);
 835         show_objects_for_type(bitmap_git, bitmap_git->trees,
 836                 OBJ_TREE, show_reachable);
 837         show_objects_for_type(bitmap_git, bitmap_git->blobs,
 838                 OBJ_BLOB, show_reachable);
 839         show_objects_for_type(bitmap_git, bitmap_git->tags,
 840                 OBJ_TAG, show_reachable);
 841
 842         show_extended_objects(bitmap_git, show_reachable);
 843
 844         bitmap_free(bitmap_git->result);
 845         bitmap_git->result = NULL;
 846 }
 847
 848 static uint32_t count_object_type(struct bitmap_index *bitmap_git,
 849                                   enum object_type type)
 850 {
 851         struct bitmap *objects = bitmap_git->result;
 852         struct eindex *eindex = &bitmap_git->ext_index;
 853
 854         uint32_t i = 0, count = 0;
 855         struct ewah_iterator it;
 856         eword_t filter;
 857
 858         switch (type) {
 859         case OBJ_COMMIT:
 860                 ewah_iterator_init(&it, bitmap_git->commits);
 861                 break;
 862
 863         case OBJ_TREE:
 864                 ewah_iterator_init(&it, bitmap_git->trees);
 865                 break;
 866
 867         case OBJ_BLOB:
 868                 ewah_iterator_init(&it, bitmap_git->blobs);
 869                 break;
 870
 871         case OBJ_TAG:
 872                 ewah_iterator_init(&it, bitmap_git->tags);
 873                 break;
 874
 875         default:
 876                 return 0;
 877         }
 878
 879         while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
 880                 eword_t word = objects->words[i++] & filter;
 881                 count += ewah_bit_popcount64(word);
 882         }
 883
 884         for (i = 0; i < eindex->count; ++i) {
 885                 if (eindex->objects[i]->type == type &&
 886                         bitmap_get(objects, bitmap_git->pack->num_objects + i))
 887                         count++;
 888         }
 889
 890         return count;
 891 }
 892
 893 void count_bitmap_commit_list(struct bitmap_index *bitmap_git,
 894                               uint32_t *commits, uint32_t *trees,
 895                               uint32_t *blobs, uint32_t *tags)
 896 {
 897         assert(bitmap_git->result);
 898
 899         if (commits)
 900                 *commits = count_object_type(bitmap_git, OBJ_COMMIT);
 901
 902         if (trees)
 903                 *trees = count_object_type(bitmap_git, OBJ_TREE);
 904
 905         if (blobs)
 906                 *blobs = count_object_type(bitmap_git, OBJ_BLOB);
 907
 908         if (tags)
 909                 *tags = count_object_type(bitmap_git, OBJ_TAG);
 910 }
 911
 912 struct bitmap_test_data {
 913         struct bitmap_index *bitmap_git;
 914         struct bitmap *base;
 915         struct progress *prg;
 916         size_t seen;
 917 };
 918
 919 static void test_show_object(struct object *object, const char *name,
 920                              void *data)
 921 {
 922         struct bitmap_test_data *tdata = data;
 923         int bitmap_pos;
 924
 925         bitmap_pos = bitmap_position(tdata->bitmap_git, object->oid.hash);
 926         if (bitmap_pos < 0)
 927                 die("Object not in bitmap: %s\n", oid_to_hex(&object->oid));
 928
 929         bitmap_set(tdata->base, bitmap_pos);
 930         display_progress(tdata->prg, ++tdata->seen);
 931 }
 932
 933 static void test_show_commit(struct commit *commit, void *data)
 934 {
 935         struct bitmap_test_data *tdata = data;
 936         int bitmap_pos;
 937
 938         bitmap_pos = bitmap_position(tdata->bitmap_git,
 939                                      commit->object.oid.hash);
 940         if (bitmap_pos < 0)
 941                 die("Object not in bitmap: %s\n", oid_to_hex(&commit->object.oid));
 942
 943         bitmap_set(tdata->base, bitmap_pos);
 944         display_progress(tdata->prg, ++tdata->seen);
 945 }
 946
 947 void test_bitmap_walk(struct rev_info *revs)
 948 {
 949         struct object *root;
 950         struct bitmap *result = NULL;
 951         khiter_t pos;
 952         size_t result_popcnt;
 953         struct bitmap_test_data tdata;
 954         struct bitmap_index *bitmap_git;
 955
 956         if (!(bitmap_git = prepare_bitmap_git()))
 957                 die("failed to load bitmap indexes");
 958
 959         if (revs->pending.nr != 1)
 960                 die("you must specify exactly one commit to test");
 961
 962         fprintf(stderr, "Bitmap v%d test (%d entries loaded)\n",
 963                 bitmap_git->version, bitmap_git->entry_count);
 964
 965         root = revs->pending.objects[0].item;
 966         pos = kh_get_sha1(bitmap_git->bitmaps, root->oid.hash);
 967
 968         if (pos < kh_end(bitmap_git->bitmaps)) {
 969                 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, pos);
 970                 struct ewah_bitmap *bm = lookup_stored_bitmap(st);
 971
 972                 fprintf(stderr, "Found bitmap for %s. %d bits / %08x checksum\n",
 973                         oid_to_hex(&root->oid), (int)bm->bit_size, ewah_checksum(bm));
 974
 975                 result = ewah_to_bitmap(bm);
 976         }
 977
 978         if (result == NULL)
 979                 die("Commit %s doesn't have an indexed bitmap", oid_to_hex(&root->oid));
 980
 981         revs->tag_objects = 1;
 982         revs->tree_objects = 1;
 983         revs->blob_objects = 1;
 984
 985         result_popcnt = bitmap_popcount(result);
 986
 987         if (prepare_revision_walk(revs))
 988                 die("revision walk setup failed");
 989
 990         tdata.bitmap_git = bitmap_git;
 991         tdata.base = bitmap_new();
 992         tdata.prg = start_progress("Verifying bitmap entries", result_popcnt);
 993         tdata.seen = 0;
 994
 995         traverse_commit_list(revs, &test_show_commit, &test_show_object, &tdata);
 996
 997         stop_progress(&tdata.prg);
 998
 999         if (bitmap_equals(result, tdata.base))
1000                 fprintf(stderr, "OK!\n");
1001         else
1002                 fprintf(stderr, "Mismatch!\n");
1003
1004         bitmap_free(result);
1005 }
1006
1007 static int rebuild_bitmap(uint32_t *reposition,
1008                           struct ewah_bitmap *source,
1009                           struct bitmap *dest)
1010 {
1011         uint32_t pos = 0;
1012         struct ewah_iterator it;
1013         eword_t word;
1014
1015         ewah_iterator_init(&it, source);
1016
1017         while (ewah_iterator_next(&word, &it)) {
1018                 uint32_t offset, bit_pos;
1019
1020                 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
1021                         if ((word >> offset) == 0)
1022                                 break;
1023
1024                         offset += ewah_bit_ctz64(word >> offset);
1025
1026                         bit_pos = reposition[pos + offset];
1027                         if (bit_pos > 0)
1028                                 bitmap_set(dest, bit_pos - 1);
1029                         else /* can't reuse, we don't have the object */
1030                                 return -1;
1031                 }
1032
1033                 pos += BITS_IN_EWORD;
1034         }
1035         return 0;
1036 }
1037
1038 int rebuild_existing_bitmaps(struct bitmap_index *bitmap_git,
1039                              struct packing_data *mapping,
1040                              khash_sha1 *reused_bitmaps,
1041                              int show_progress)
1042 {
1043         uint32_t i, num_objects;
1044         uint32_t *reposition;
1045         struct bitmap *rebuild;
1046         struct stored_bitmap *stored;
1047         struct progress *progress = NULL;
1048
1049         khiter_t hash_pos;
1050         int hash_ret;
1051
1052         num_objects = bitmap_git->pack->num_objects;
1053         reposition = xcalloc(num_objects, sizeof(uint32_t));
1054
1055         for (i = 0; i < num_objects; ++i) {
1056                 const unsigned char *sha1;
1057                 struct revindex_entry *entry;
1058                 struct object_entry *oe;
1059
1060                 entry = &bitmap_git->pack->revindex[i];
1061                 sha1 = nth_packed_object_sha1(bitmap_git->pack, entry->nr);
1062                 oe = packlist_find(mapping, sha1, NULL);
1063
1064                 if (oe)
1065                         reposition[i] = oe_in_pack_pos(mapping, oe) + 1;
1066         }
1067
1068         rebuild = bitmap_new();
1069         i = 0;
1070
1071         if (show_progress)
1072                 progress = start_progress("Reusing bitmaps", 0);
1073
1074         kh_foreach_value(bitmap_git->bitmaps, stored, {
1075                 if (stored->flags & BITMAP_FLAG_REUSE) {
1076                         if (!rebuild_bitmap(reposition,
1077                                             lookup_stored_bitmap(stored),
1078                                             rebuild)) {
1079                                 hash_pos = kh_put_sha1(reused_bitmaps,
1080                                                        stored->sha1,
1081                                                        &hash_ret);
1082                                 kh_value(reused_bitmaps, hash_pos) =
1083                                         bitmap_to_ewah(rebuild);
1084                         }
1085                         bitmap_reset(rebuild);
1086                         display_progress(progress, ++i);
1087                 }
1088         });
1089
1090         stop_progress(&progress);
1091
1092         free(reposition);
1093         bitmap_free(rebuild);
1094         return 0;
1095 }