pack-bitmap-write.c

   1 #include "cache.h"
   2 #include "commit.h"
   3 #include "tag.h"
   4 #include "diff.h"
   5 #include "revision.h"
   6 #include "list-objects.h"
   7 #include "progress.h"
   8 #include "pack-revindex.h"
   9 #include "pack.h"
  10 #include "pack-bitmap.h"
  11 #include "sha1-lookup.h"
  12 #include "pack-objects.h"
  13
  14 struct bitmapped_commit {
  15         struct commit *commit;
  16         struct ewah_bitmap *bitmap;
  17         struct ewah_bitmap *write_as;
  18         int flags;
  19         int xor_offset;
  20         uint32_t commit_pos;
  21 };
  22
  23 struct bitmap_writer {
  24         struct ewah_bitmap *commits;
  25         struct ewah_bitmap *trees;
  26         struct ewah_bitmap *blobs;
  27         struct ewah_bitmap *tags;
  28
  29         khash_sha1 *bitmaps;
  30         khash_sha1 *reused;
  31         struct packing_data *to_pack;
  32
  33         struct bitmapped_commit *selected;
  34         unsigned int selected_nr, selected_alloc;
  35
  36         struct progress *progress;
  37         int show_progress;
  38         unsigned char pack_checksum[20];
  39 };
  40
  41 static struct bitmap_writer writer;
  42
  43 void bitmap_writer_show_progress(int show)
  44 {
  45         writer.show_progress = show;
  46 }
  47
  48 /**
  49  * Build the initial type index for the packfile
  50  */
  51 void bitmap_writer_build_type_index(struct pack_idx_entry **index,
  52                                     uint32_t index_nr)
  53 {
  54         uint32_t i;
  55
  56         writer.commits = ewah_new();
  57         writer.trees = ewah_new();
  58         writer.blobs = ewah_new();
  59         writer.tags = ewah_new();
  60
  61         for (i = 0; i < index_nr; ++i) {
  62                 struct object_entry *entry = (struct object_entry *)index[i];
  63                 enum object_type real_type;
  64
  65                 entry->in_pack_pos = i;
  66
  67                 switch (entry->type) {
  68                 case OBJ_COMMIT:
  69                 case OBJ_TREE:
  70                 case OBJ_BLOB:
  71                 case OBJ_TAG:
  72                         real_type = entry->type;
  73                         break;
  74
  75                 default:
  76                         real_type = sha1_object_info(entry->idx.oid.hash,
  77                                                      NULL);
  78                         break;
  79                 }
  80
  81                 switch (real_type) {
  82                 case OBJ_COMMIT:
  83                         ewah_set(writer.commits, i);
  84                         break;
  85
  86                 case OBJ_TREE:
  87                         ewah_set(writer.trees, i);
  88                         break;
  89
  90                 case OBJ_BLOB:
  91                         ewah_set(writer.blobs, i);
  92                         break;
  93
  94                 case OBJ_TAG:
  95                         ewah_set(writer.tags, i);
  96                         break;
  97
  98                 default:
  99                         die("Missing type information for %s (%d/%d)",
 100                             oid_to_hex(&entry->idx.oid), real_type,
 101                             entry->type);
 102                 }
 103         }
 104 }
 105
 106 /**
 107  * Compute the actual bitmaps
 108  */
 109 static struct object **seen_objects;
 110 static unsigned int seen_objects_nr, seen_objects_alloc;
 111
 112 static inline void push_bitmapped_commit(struct commit *commit, struct ewah_bitmap *reused)
 113 {
 114         if (writer.selected_nr >= writer.selected_alloc) {
 115                 writer.selected_alloc = (writer.selected_alloc + 32) * 2;
 116                 REALLOC_ARRAY(writer.selected, writer.selected_alloc);
 117         }
 118
 119         writer.selected[writer.selected_nr].commit = commit;
 120         writer.selected[writer.selected_nr].bitmap = reused;
 121         writer.selected[writer.selected_nr].flags = 0;
 122
 123         writer.selected_nr++;
 124 }
 125
 126 static inline void mark_as_seen(struct object *object)
 127 {
 128         ALLOC_GROW(seen_objects, seen_objects_nr + 1, seen_objects_alloc);
 129         seen_objects[seen_objects_nr++] = object;
 130 }
 131
 132 static inline void reset_all_seen(void)
 133 {
 134         unsigned int i;
 135         for (i = 0; i < seen_objects_nr; ++i) {
 136                 seen_objects[i]->flags &= ~(SEEN | ADDED | SHOWN);
 137         }
 138         seen_objects_nr = 0;
 139 }
 140
 141 static uint32_t find_object_pos(const unsigned char *sha1)
 142 {
 143         struct object_entry *entry = packlist_find(writer.to_pack, sha1, NULL);
 144
 145         if (!entry) {
 146                 die("Failed to write bitmap index. Packfile doesn't have full closure "
 147                         "(object %s is missing)", sha1_to_hex(sha1));
 148         }
 149
 150         return entry->in_pack_pos;
 151 }
 152
 153 static void show_object(struct object *object, const char *name, void *data)
 154 {
 155         struct bitmap *base = data;
 156         bitmap_set(base, find_object_pos(object->oid.hash));
 157         mark_as_seen(object);
 158 }
 159
 160 static void show_commit(struct commit *commit, void *data)
 161 {
 162         mark_as_seen((struct object *)commit);
 163 }
 164
 165 static int
 166 add_to_include_set(struct bitmap *base, struct commit *commit)
 167 {
 168         khiter_t hash_pos;
 169         uint32_t bitmap_pos = find_object_pos(commit->object.oid.hash);
 170
 171         if (bitmap_get(base, bitmap_pos))
 172                 return 0;
 173
 174         hash_pos = kh_get_sha1(writer.bitmaps, commit->object.oid.hash);
 175         if (hash_pos < kh_end(writer.bitmaps)) {
 176                 struct bitmapped_commit *bc = kh_value(writer.bitmaps, hash_pos);
 177                 bitmap_or_ewah(base, bc->bitmap);
 178                 return 0;
 179         }
 180
 181         bitmap_set(base, bitmap_pos);
 182         return 1;
 183 }
 184
 185 static int
 186 should_include(struct commit *commit, void *_data)
 187 {
 188         struct bitmap *base = _data;
 189
 190         if (!add_to_include_set(base, commit)) {
 191                 struct commit_list *parent = commit->parents;
 192
 193                 mark_as_seen((struct object *)commit);
 194
 195                 while (parent) {
 196                         parent->item->object.flags |= SEEN;
 197                         mark_as_seen((struct object *)parent->item);
 198                         parent = parent->next;
 199                 }
 200
 201                 return 0;
 202         }
 203
 204         return 1;
 205 }
 206
 207 static void compute_xor_offsets(void)
 208 {
 209         static const int MAX_XOR_OFFSET_SEARCH = 10;
 210
 211         int i, next = 0;
 212
 213         while (next < writer.selected_nr) {
 214                 struct bitmapped_commit *stored = &writer.selected[next];
 215
 216                 int best_offset = 0;
 217                 struct ewah_bitmap *best_bitmap = stored->bitmap;
 218                 struct ewah_bitmap *test_xor;
 219
 220                 for (i = 1; i <= MAX_XOR_OFFSET_SEARCH; ++i) {
 221                         int curr = next - i;
 222
 223                         if (curr < 0)
 224                                 break;
 225
 226                         test_xor = ewah_pool_new();
 227                         ewah_xor(writer.selected[curr].bitmap, stored->bitmap, test_xor);
 228
 229                         if (test_xor->buffer_size < best_bitmap->buffer_size) {
 230                                 if (best_bitmap != stored->bitmap)
 231                                         ewah_pool_free(best_bitmap);
 232
 233                                 best_bitmap = test_xor;
 234                                 best_offset = i;
 235                         } else {
 236                                 ewah_pool_free(test_xor);
 237                         }
 238                 }
 239
 240                 stored->xor_offset = best_offset;
 241                 stored->write_as = best_bitmap;
 242
 243                 next++;
 244         }
 245 }
 246
 247 void bitmap_writer_build(struct packing_data *to_pack)
 248 {
 249         static const double REUSE_BITMAP_THRESHOLD = 0.2;
 250
 251         int i, reuse_after, need_reset;
 252         struct bitmap *base = bitmap_new();
 253         struct rev_info revs;
 254
 255         writer.bitmaps = kh_init_sha1();
 256         writer.to_pack = to_pack;
 257
 258         if (writer.show_progress)
 259                 writer.progress = start_progress("Building bitmaps", writer.selected_nr);
 260
 261         init_revisions(&revs, NULL);
 262         revs.tag_objects = 1;
 263         revs.tree_objects = 1;
 264         revs.blob_objects = 1;
 265         revs.no_walk = 0;
 266
 267         revs.include_check = should_include;
 268         reset_revision_walk();
 269
 270         reuse_after = writer.selected_nr * REUSE_BITMAP_THRESHOLD;
 271         need_reset = 0;
 272
 273         for (i = writer.selected_nr - 1; i >= 0; --i) {
 274                 struct bitmapped_commit *stored;
 275                 struct object *object;
 276
 277                 khiter_t hash_pos;
 278                 int hash_ret;
 279
 280                 stored = &writer.selected[i];
 281                 object = (struct object *)stored->commit;
 282
 283                 if (stored->bitmap == NULL) {
 284                         if (i < writer.selected_nr - 1 &&
 285                             (need_reset ||
 286                              !in_merge_bases(writer.selected[i + 1].commit,
 287                                              stored->commit))) {
 288                             bitmap_reset(base);
 289                             reset_all_seen();
 290                         }
 291
 292                         add_pending_object(&revs, object, "");
 293                         revs.include_check_data = base;
 294
 295                         if (prepare_revision_walk(&revs))
 296                                 die("revision walk setup failed");
 297
 298                         traverse_commit_list(&revs, show_commit, show_object, base);
 299
 300                         object_array_clear(&revs.pending);
 301
 302                         stored->bitmap = bitmap_to_ewah(base);
 303                         need_reset = 0;
 304                 } else
 305                         need_reset = 1;
 306
 307                 if (i >= reuse_after)
 308                         stored->flags |= BITMAP_FLAG_REUSE;
 309
 310                 hash_pos = kh_put_sha1(writer.bitmaps, object->oid.hash, &hash_ret);
 311                 if (hash_ret == 0)
 312                         die("Duplicate entry when writing index: %s",
 313                             oid_to_hex(&object->oid));
 314
 315                 kh_value(writer.bitmaps, hash_pos) = stored;
 316                 display_progress(writer.progress, writer.selected_nr - i);
 317         }
 318
 319         bitmap_free(base);
 320         stop_progress(&writer.progress);
 321
 322         compute_xor_offsets();
 323 }
 324
 325 /**
 326  * Select the commits that will be bitmapped
 327  */
 328 static inline unsigned int next_commit_index(unsigned int idx)
 329 {
 330         static const unsigned int MIN_COMMITS = 100;
 331         static const unsigned int MAX_COMMITS = 5000;
 332
 333         static const unsigned int MUST_REGION = 100;
 334         static const unsigned int MIN_REGION = 20000;
 335
 336         unsigned int offset, next;
 337
 338         if (idx <= MUST_REGION)
 339                 return 0;
 340
 341         if (idx <= MIN_REGION) {
 342                 offset = idx - MUST_REGION;
 343                 return (offset < MIN_COMMITS) ? offset : MIN_COMMITS;
 344         }
 345
 346         offset = idx - MIN_REGION;
 347         next = (offset < MAX_COMMITS) ? offset : MAX_COMMITS;
 348
 349         return (next > MIN_COMMITS) ? next : MIN_COMMITS;
 350 }
 351
 352 static int date_compare(const void *_a, const void *_b)
 353 {
 354         struct commit *a = *(struct commit **)_a;
 355         struct commit *b = *(struct commit **)_b;
 356         return (long)b->date - (long)a->date;
 357 }
 358
 359 void bitmap_writer_reuse_bitmaps(struct packing_data *to_pack)
 360 {
 361         if (prepare_bitmap_git() < 0)
 362                 return;
 363
 364         writer.reused = kh_init_sha1();
 365         rebuild_existing_bitmaps(to_pack, writer.reused, writer.show_progress);
 366 }
 367
 368 static struct ewah_bitmap *find_reused_bitmap(const unsigned char *sha1)
 369 {
 370         khiter_t hash_pos;
 371
 372         if (!writer.reused)
 373                 return NULL;
 374
 375         hash_pos = kh_get_sha1(writer.reused, sha1);
 376         if (hash_pos >= kh_end(writer.reused))
 377                 return NULL;
 378
 379         return kh_value(writer.reused, hash_pos);
 380 }
 381
 382 void bitmap_writer_select_commits(struct commit **indexed_commits,
 383                                   unsigned int indexed_commits_nr,
 384                                   int max_bitmaps)
 385 {
 386         unsigned int i = 0, j, next;
 387
 388         QSORT(indexed_commits, indexed_commits_nr, date_compare);
 389
 390         if (writer.show_progress)
 391                 writer.progress = start_progress("Selecting bitmap commits", 0);
 392
 393         if (indexed_commits_nr < 100) {
 394                 for (i = 0; i < indexed_commits_nr; ++i)
 395                         push_bitmapped_commit(indexed_commits[i], NULL);
 396                 return;
 397         }
 398
 399         for (;;) {
 400                 struct ewah_bitmap *reused_bitmap = NULL;
 401                 struct commit *chosen = NULL;
 402
 403                 next = next_commit_index(i);
 404
 405                 if (i + next >= indexed_commits_nr)
 406                         break;
 407
 408                 if (max_bitmaps > 0 && writer.selected_nr >= max_bitmaps) {
 409                         writer.selected_nr = max_bitmaps;
 410                         break;
 411                 }
 412
 413                 if (next == 0) {
 414                         chosen = indexed_commits[i];
 415                         reused_bitmap = find_reused_bitmap(chosen->object.oid.hash);
 416                 } else {
 417                         chosen = indexed_commits[i + next];
 418
 419                         for (j = 0; j <= next; ++j) {
 420                                 struct commit *cm = indexed_commits[i + j];
 421
 422                                 reused_bitmap = find_reused_bitmap(cm->object.oid.hash);
 423                                 if (reused_bitmap || (cm->object.flags & NEEDS_BITMAP) != 0) {
 424                                         chosen = cm;
 425                                         break;
 426                                 }
 427
 428                                 if (cm->parents && cm->parents->next)
 429                                         chosen = cm;
 430                         }
 431                 }
 432
 433                 push_bitmapped_commit(chosen, reused_bitmap);
 434
 435                 i += next + 1;
 436                 display_progress(writer.progress, i);
 437         }
 438
 439         stop_progress(&writer.progress);
 440 }
 441
 442
 443 static int hashwrite_ewah_helper(void *f, const void *buf, size_t len)
 444 {
 445         /* hashwrite will die on error */
 446         hashwrite(f, buf, len);
 447         return len;
 448 }
 449
 450 /**
 451  * Write the bitmap index to disk
 452  */
 453 static inline void dump_bitmap(struct hashfile *f, struct ewah_bitmap *bitmap)
 454 {
 455         if (ewah_serialize_to(bitmap, hashwrite_ewah_helper, f) < 0)
 456                 die("Failed to write bitmap index");
 457 }
 458
 459 static const unsigned char *sha1_access(size_t pos, void *table)
 460 {
 461         struct pack_idx_entry **index = table;
 462         return index[pos]->oid.hash;
 463 }
 464
 465 static void write_selected_commits_v1(struct hashfile *f,
 466                                       struct pack_idx_entry **index,
 467                                       uint32_t index_nr)
 468 {
 469         int i;
 470
 471         for (i = 0; i < writer.selected_nr; ++i) {
 472                 struct bitmapped_commit *stored = &writer.selected[i];
 473
 474                 int commit_pos =
 475                         sha1_pos(stored->commit->object.oid.hash, index, index_nr, sha1_access);
 476
 477                 if (commit_pos < 0)
 478                         die("BUG: trying to write commit not in index");
 479
 480                 hashwrite_be32(f, commit_pos);
 481                 hashwrite_u8(f, stored->xor_offset);
 482                 hashwrite_u8(f, stored->flags);
 483
 484                 dump_bitmap(f, stored->write_as);
 485         }
 486 }
 487
 488 static void write_hash_cache(struct hashfile *f,
 489                              struct pack_idx_entry **index,
 490                              uint32_t index_nr)
 491 {
 492         uint32_t i;
 493
 494         for (i = 0; i < index_nr; ++i) {
 495                 struct object_entry *entry = (struct object_entry *)index[i];
 496                 uint32_t hash_value = htonl(entry->hash);
 497                 hashwrite(f, &hash_value, sizeof(hash_value));
 498         }
 499 }
 500
 501 void bitmap_writer_set_checksum(unsigned char *sha1)
 502 {
 503         hashcpy(writer.pack_checksum, sha1);
 504 }
 505
 506 void bitmap_writer_finish(struct pack_idx_entry **index,
 507                           uint32_t index_nr,
 508                           const char *filename,
 509                           uint16_t options)
 510 {
 511         static uint16_t default_version = 1;
 512         static uint16_t flags = BITMAP_OPT_FULL_DAG;
 513         struct strbuf tmp_file = STRBUF_INIT;
 514         struct hashfile *f;
 515
 516         struct bitmap_disk_header header;
 517
 518         int fd = odb_mkstemp(&tmp_file, "pack/tmp_bitmap_XXXXXX");
 519
 520         f = hashfd(fd, tmp_file.buf);
 521
 522         memcpy(header.magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE));
 523         header.version = htons(default_version);
 524         header.options = htons(flags | options);
 525         header.entry_count = htonl(writer.selected_nr);
 526         hashcpy(header.checksum, writer.pack_checksum);
 527
 528         hashwrite(f, &header, sizeof(header));
 529         dump_bitmap(f, writer.commits);
 530         dump_bitmap(f, writer.trees);
 531         dump_bitmap(f, writer.blobs);
 532         dump_bitmap(f, writer.tags);
 533         write_selected_commits_v1(f, index, index_nr);
 534
 535         if (options & BITMAP_OPT_HASH_CACHE)
 536                 write_hash_cache(f, index, index_nr);
 537
 538         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC | CSUM_CLOSE);
 539
 540         if (adjust_shared_perm(tmp_file.buf))
 541                 die_errno("unable to make temporary bitmap file readable");
 542
 543         if (rename(tmp_file.buf, filename))
 544                 die_errno("unable to rename temporary bitmap file to '%s'", filename);
 545
 546         strbuf_release(&tmp_file);
 547 }