Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 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.sha1, NULL); |
| 77 | break; |
| 78 | } |
| 79 | |
| 80 | switch (real_type) { |
| 81 | case OBJ_COMMIT: |
| 82 | ewah_set(writer.commits, i); |
| 83 | break; |
| 84 | |
| 85 | case OBJ_TREE: |
| 86 | ewah_set(writer.trees, i); |
| 87 | break; |
| 88 | |
| 89 | case OBJ_BLOB: |
| 90 | ewah_set(writer.blobs, i); |
| 91 | break; |
| 92 | |
| 93 | case OBJ_TAG: |
| 94 | ewah_set(writer.tags, i); |
| 95 | break; |
| 96 | |
| 97 | default: |
| 98 | die("Missing type information for %s (%d/%d)", |
| 99 | sha1_to_hex(entry->idx.sha1), real_type, entry->type); |
| 100 | } |
| 101 | } |
| 102 | } |
| 103 | |
| 104 | /** |
| 105 | * Compute the actual bitmaps |
| 106 | */ |
| 107 | static struct object **seen_objects; |
| 108 | static unsigned int seen_objects_nr, seen_objects_alloc; |
| 109 | |
| 110 | static inline void push_bitmapped_commit(struct commit *commit, struct ewah_bitmap *reused) |
| 111 | { |
| 112 | if (writer.selected_nr >= writer.selected_alloc) { |
| 113 | writer.selected_alloc = (writer.selected_alloc + 32) * 2; |
René Scharfe | 2756ca4 | 2014-09-16 20:56:57 +0200 | [diff] [blame] | 114 | REALLOC_ARRAY(writer.selected, writer.selected_alloc); |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 115 | } |
| 116 | |
| 117 | writer.selected[writer.selected_nr].commit = commit; |
| 118 | writer.selected[writer.selected_nr].bitmap = reused; |
| 119 | writer.selected[writer.selected_nr].flags = 0; |
| 120 | |
| 121 | writer.selected_nr++; |
| 122 | } |
| 123 | |
| 124 | static inline void mark_as_seen(struct object *object) |
| 125 | { |
| 126 | ALLOC_GROW(seen_objects, seen_objects_nr + 1, seen_objects_alloc); |
| 127 | seen_objects[seen_objects_nr++] = object; |
| 128 | } |
| 129 | |
| 130 | static inline void reset_all_seen(void) |
| 131 | { |
| 132 | unsigned int i; |
| 133 | for (i = 0; i < seen_objects_nr; ++i) { |
| 134 | seen_objects[i]->flags &= ~(SEEN | ADDED | SHOWN); |
| 135 | } |
| 136 | seen_objects_nr = 0; |
| 137 | } |
| 138 | |
| 139 | static uint32_t find_object_pos(const unsigned char *sha1) |
| 140 | { |
| 141 | struct object_entry *entry = packlist_find(writer.to_pack, sha1, NULL); |
| 142 | |
| 143 | if (!entry) { |
| 144 | die("Failed to write bitmap index. Packfile doesn't have full closure " |
| 145 | "(object %s is missing)", sha1_to_hex(sha1)); |
| 146 | } |
| 147 | |
| 148 | return entry->in_pack_pos; |
| 149 | } |
| 150 | |
| 151 | static void show_object(struct object *object, const struct name_path *path, |
| 152 | const char *last, void *data) |
| 153 | { |
| 154 | struct bitmap *base = data; |
| 155 | bitmap_set(base, find_object_pos(object->sha1)); |
| 156 | mark_as_seen(object); |
| 157 | } |
| 158 | |
| 159 | static void show_commit(struct commit *commit, void *data) |
| 160 | { |
| 161 | mark_as_seen((struct object *)commit); |
| 162 | } |
| 163 | |
| 164 | static int |
| 165 | add_to_include_set(struct bitmap *base, struct commit *commit) |
| 166 | { |
| 167 | khiter_t hash_pos; |
| 168 | uint32_t bitmap_pos = find_object_pos(commit->object.sha1); |
| 169 | |
| 170 | if (bitmap_get(base, bitmap_pos)) |
| 171 | return 0; |
| 172 | |
| 173 | hash_pos = kh_get_sha1(writer.bitmaps, commit->object.sha1); |
| 174 | if (hash_pos < kh_end(writer.bitmaps)) { |
| 175 | struct bitmapped_commit *bc = kh_value(writer.bitmaps, hash_pos); |
| 176 | bitmap_or_ewah(base, bc->bitmap); |
| 177 | return 0; |
| 178 | } |
| 179 | |
| 180 | bitmap_set(base, bitmap_pos); |
| 181 | return 1; |
| 182 | } |
| 183 | |
| 184 | static int |
| 185 | should_include(struct commit *commit, void *_data) |
| 186 | { |
| 187 | struct bitmap *base = _data; |
| 188 | |
| 189 | if (!add_to_include_set(base, commit)) { |
| 190 | struct commit_list *parent = commit->parents; |
| 191 | |
| 192 | mark_as_seen((struct object *)commit); |
| 193 | |
| 194 | while (parent) { |
| 195 | parent->item->object.flags |= SEEN; |
| 196 | mark_as_seen((struct object *)parent->item); |
| 197 | parent = parent->next; |
| 198 | } |
| 199 | |
| 200 | return 0; |
| 201 | } |
| 202 | |
| 203 | return 1; |
| 204 | } |
| 205 | |
| 206 | static void compute_xor_offsets(void) |
| 207 | { |
| 208 | static const int MAX_XOR_OFFSET_SEARCH = 10; |
| 209 | |
| 210 | int i, next = 0; |
| 211 | |
| 212 | while (next < writer.selected_nr) { |
| 213 | struct bitmapped_commit *stored = &writer.selected[next]; |
| 214 | |
| 215 | int best_offset = 0; |
| 216 | struct ewah_bitmap *best_bitmap = stored->bitmap; |
| 217 | struct ewah_bitmap *test_xor; |
| 218 | |
| 219 | for (i = 1; i <= MAX_XOR_OFFSET_SEARCH; ++i) { |
| 220 | int curr = next - i; |
| 221 | |
| 222 | if (curr < 0) |
| 223 | break; |
| 224 | |
| 225 | test_xor = ewah_pool_new(); |
| 226 | ewah_xor(writer.selected[curr].bitmap, stored->bitmap, test_xor); |
| 227 | |
| 228 | if (test_xor->buffer_size < best_bitmap->buffer_size) { |
| 229 | if (best_bitmap != stored->bitmap) |
| 230 | ewah_pool_free(best_bitmap); |
| 231 | |
| 232 | best_bitmap = test_xor; |
| 233 | best_offset = i; |
| 234 | } else { |
| 235 | ewah_pool_free(test_xor); |
| 236 | } |
| 237 | } |
| 238 | |
| 239 | stored->xor_offset = best_offset; |
| 240 | stored->write_as = best_bitmap; |
| 241 | |
| 242 | next++; |
| 243 | } |
| 244 | } |
| 245 | |
| 246 | void bitmap_writer_build(struct packing_data *to_pack) |
| 247 | { |
| 248 | static const double REUSE_BITMAP_THRESHOLD = 0.2; |
| 249 | |
| 250 | int i, reuse_after, need_reset; |
| 251 | struct bitmap *base = bitmap_new(); |
| 252 | struct rev_info revs; |
| 253 | |
| 254 | writer.bitmaps = kh_init_sha1(); |
| 255 | writer.to_pack = to_pack; |
| 256 | |
| 257 | if (writer.show_progress) |
| 258 | writer.progress = start_progress("Building bitmaps", writer.selected_nr); |
| 259 | |
| 260 | init_revisions(&revs, NULL); |
| 261 | revs.tag_objects = 1; |
| 262 | revs.tree_objects = 1; |
| 263 | revs.blob_objects = 1; |
| 264 | revs.no_walk = 0; |
| 265 | |
| 266 | revs.include_check = should_include; |
| 267 | reset_revision_walk(); |
| 268 | |
| 269 | reuse_after = writer.selected_nr * REUSE_BITMAP_THRESHOLD; |
| 270 | need_reset = 0; |
| 271 | |
| 272 | for (i = writer.selected_nr - 1; i >= 0; --i) { |
| 273 | struct bitmapped_commit *stored; |
| 274 | struct object *object; |
| 275 | |
| 276 | khiter_t hash_pos; |
| 277 | int hash_ret; |
| 278 | |
| 279 | stored = &writer.selected[i]; |
| 280 | object = (struct object *)stored->commit; |
| 281 | |
| 282 | if (stored->bitmap == NULL) { |
| 283 | if (i < writer.selected_nr - 1 && |
| 284 | (need_reset || |
| 285 | !in_merge_bases(writer.selected[i + 1].commit, |
| 286 | stored->commit))) { |
| 287 | bitmap_reset(base); |
| 288 | reset_all_seen(); |
| 289 | } |
| 290 | |
| 291 | add_pending_object(&revs, object, ""); |
| 292 | revs.include_check_data = base; |
| 293 | |
| 294 | if (prepare_revision_walk(&revs)) |
| 295 | die("revision walk setup failed"); |
| 296 | |
| 297 | traverse_commit_list(&revs, show_commit, show_object, base); |
| 298 | |
| 299 | revs.pending.nr = 0; |
| 300 | revs.pending.alloc = 0; |
| 301 | revs.pending.objects = NULL; |
| 302 | |
| 303 | stored->bitmap = bitmap_to_ewah(base); |
| 304 | need_reset = 0; |
| 305 | } else |
| 306 | need_reset = 1; |
| 307 | |
| 308 | if (i >= reuse_after) |
| 309 | stored->flags |= BITMAP_FLAG_REUSE; |
| 310 | |
| 311 | hash_pos = kh_put_sha1(writer.bitmaps, object->sha1, &hash_ret); |
| 312 | if (hash_ret == 0) |
| 313 | die("Duplicate entry when writing index: %s", |
| 314 | sha1_to_hex(object->sha1)); |
| 315 | |
| 316 | kh_value(writer.bitmaps, hash_pos) = stored; |
| 317 | display_progress(writer.progress, writer.selected_nr - i); |
| 318 | } |
| 319 | |
| 320 | bitmap_free(base); |
| 321 | stop_progress(&writer.progress); |
| 322 | |
| 323 | compute_xor_offsets(); |
| 324 | } |
| 325 | |
| 326 | /** |
| 327 | * Select the commits that will be bitmapped |
| 328 | */ |
| 329 | static inline unsigned int next_commit_index(unsigned int idx) |
| 330 | { |
| 331 | static const unsigned int MIN_COMMITS = 100; |
| 332 | static const unsigned int MAX_COMMITS = 5000; |
| 333 | |
| 334 | static const unsigned int MUST_REGION = 100; |
| 335 | static const unsigned int MIN_REGION = 20000; |
| 336 | |
| 337 | unsigned int offset, next; |
| 338 | |
| 339 | if (idx <= MUST_REGION) |
| 340 | return 0; |
| 341 | |
| 342 | if (idx <= MIN_REGION) { |
| 343 | offset = idx - MUST_REGION; |
| 344 | return (offset < MIN_COMMITS) ? offset : MIN_COMMITS; |
| 345 | } |
| 346 | |
| 347 | offset = idx - MIN_REGION; |
| 348 | next = (offset < MAX_COMMITS) ? offset : MAX_COMMITS; |
| 349 | |
| 350 | return (next > MIN_COMMITS) ? next : MIN_COMMITS; |
| 351 | } |
| 352 | |
| 353 | static int date_compare(const void *_a, const void *_b) |
| 354 | { |
| 355 | struct commit *a = *(struct commit **)_a; |
| 356 | struct commit *b = *(struct commit **)_b; |
| 357 | return (long)b->date - (long)a->date; |
| 358 | } |
| 359 | |
| 360 | void bitmap_writer_reuse_bitmaps(struct packing_data *to_pack) |
| 361 | { |
| 362 | if (prepare_bitmap_git() < 0) |
| 363 | return; |
| 364 | |
| 365 | writer.reused = kh_init_sha1(); |
| 366 | rebuild_existing_bitmaps(to_pack, writer.reused, writer.show_progress); |
| 367 | } |
| 368 | |
| 369 | static struct ewah_bitmap *find_reused_bitmap(const unsigned char *sha1) |
| 370 | { |
| 371 | khiter_t hash_pos; |
| 372 | |
| 373 | if (!writer.reused) |
| 374 | return NULL; |
| 375 | |
| 376 | hash_pos = kh_get_sha1(writer.reused, sha1); |
| 377 | if (hash_pos >= kh_end(writer.reused)) |
| 378 | return NULL; |
| 379 | |
| 380 | return kh_value(writer.reused, hash_pos); |
| 381 | } |
| 382 | |
| 383 | void bitmap_writer_select_commits(struct commit **indexed_commits, |
| 384 | unsigned int indexed_commits_nr, |
| 385 | int max_bitmaps) |
| 386 | { |
| 387 | unsigned int i = 0, j, next; |
| 388 | |
| 389 | qsort(indexed_commits, indexed_commits_nr, sizeof(indexed_commits[0]), |
| 390 | date_compare); |
| 391 | |
| 392 | if (writer.show_progress) |
| 393 | writer.progress = start_progress("Selecting bitmap commits", 0); |
| 394 | |
| 395 | if (indexed_commits_nr < 100) { |
| 396 | for (i = 0; i < indexed_commits_nr; ++i) |
| 397 | push_bitmapped_commit(indexed_commits[i], NULL); |
| 398 | return; |
| 399 | } |
| 400 | |
| 401 | for (;;) { |
| 402 | struct ewah_bitmap *reused_bitmap = NULL; |
| 403 | struct commit *chosen = NULL; |
| 404 | |
| 405 | next = next_commit_index(i); |
| 406 | |
| 407 | if (i + next >= indexed_commits_nr) |
| 408 | break; |
| 409 | |
| 410 | if (max_bitmaps > 0 && writer.selected_nr >= max_bitmaps) { |
| 411 | writer.selected_nr = max_bitmaps; |
| 412 | break; |
| 413 | } |
| 414 | |
| 415 | if (next == 0) { |
| 416 | chosen = indexed_commits[i]; |
| 417 | reused_bitmap = find_reused_bitmap(chosen->object.sha1); |
| 418 | } else { |
| 419 | chosen = indexed_commits[i + next]; |
| 420 | |
| 421 | for (j = 0; j <= next; ++j) { |
| 422 | struct commit *cm = indexed_commits[i + j]; |
| 423 | |
| 424 | reused_bitmap = find_reused_bitmap(cm->object.sha1); |
| 425 | if (reused_bitmap || (cm->object.flags & NEEDS_BITMAP) != 0) { |
| 426 | chosen = cm; |
| 427 | break; |
| 428 | } |
| 429 | |
| 430 | if (cm->parents && cm->parents->next) |
| 431 | chosen = cm; |
| 432 | } |
| 433 | } |
| 434 | |
| 435 | push_bitmapped_commit(chosen, reused_bitmap); |
| 436 | |
| 437 | i += next + 1; |
| 438 | display_progress(writer.progress, i); |
| 439 | } |
| 440 | |
| 441 | stop_progress(&writer.progress); |
| 442 | } |
| 443 | |
| 444 | |
| 445 | static int sha1write_ewah_helper(void *f, const void *buf, size_t len) |
| 446 | { |
| 447 | /* sha1write will die on error */ |
| 448 | sha1write(f, buf, len); |
| 449 | return len; |
| 450 | } |
| 451 | |
| 452 | /** |
| 453 | * Write the bitmap index to disk |
| 454 | */ |
| 455 | static inline void dump_bitmap(struct sha1file *f, struct ewah_bitmap *bitmap) |
| 456 | { |
| 457 | if (ewah_serialize_to(bitmap, sha1write_ewah_helper, f) < 0) |
| 458 | die("Failed to write bitmap index"); |
| 459 | } |
| 460 | |
| 461 | static const unsigned char *sha1_access(size_t pos, void *table) |
| 462 | { |
| 463 | struct pack_idx_entry **index = table; |
| 464 | return index[pos]->sha1; |
| 465 | } |
| 466 | |
| 467 | static void write_selected_commits_v1(struct sha1file *f, |
| 468 | struct pack_idx_entry **index, |
| 469 | uint32_t index_nr) |
| 470 | { |
| 471 | int i; |
| 472 | |
| 473 | for (i = 0; i < writer.selected_nr; ++i) { |
| 474 | struct bitmapped_commit *stored = &writer.selected[i]; |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 475 | |
| 476 | int commit_pos = |
| 477 | sha1_pos(stored->commit->object.sha1, index, index_nr, sha1_access); |
| 478 | |
| 479 | if (commit_pos < 0) |
| 480 | die("BUG: trying to write commit not in index"); |
| 481 | |
Karsten Blees | b500721 | 2014-11-27 00:24:01 -0500 | [diff] [blame] | 482 | sha1write_be32(f, commit_pos); |
| 483 | sha1write_u8(f, stored->xor_offset); |
| 484 | sha1write_u8(f, stored->flags); |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 485 | |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 486 | dump_bitmap(f, stored->write_as); |
| 487 | } |
| 488 | } |
| 489 | |
Vicent Marti | ae4f07f | 2013-12-21 09:00:45 -0500 | [diff] [blame] | 490 | static void write_hash_cache(struct sha1file *f, |
| 491 | struct pack_idx_entry **index, |
| 492 | uint32_t index_nr) |
| 493 | { |
| 494 | uint32_t i; |
| 495 | |
| 496 | for (i = 0; i < index_nr; ++i) { |
| 497 | struct object_entry *entry = (struct object_entry *)index[i]; |
| 498 | uint32_t hash_value = htonl(entry->hash); |
| 499 | sha1write(f, &hash_value, sizeof(hash_value)); |
| 500 | } |
| 501 | } |
| 502 | |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 503 | void bitmap_writer_set_checksum(unsigned char *sha1) |
| 504 | { |
| 505 | hashcpy(writer.pack_checksum, sha1); |
| 506 | } |
| 507 | |
| 508 | void bitmap_writer_finish(struct pack_idx_entry **index, |
| 509 | uint32_t index_nr, |
Vicent Marti | ae4f07f | 2013-12-21 09:00:45 -0500 | [diff] [blame] | 510 | const char *filename, |
| 511 | uint16_t options) |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 512 | { |
| 513 | static char tmp_file[PATH_MAX]; |
| 514 | static uint16_t default_version = 1; |
| 515 | static uint16_t flags = BITMAP_OPT_FULL_DAG; |
| 516 | struct sha1file *f; |
| 517 | |
| 518 | struct bitmap_disk_header header; |
| 519 | |
| 520 | int fd = odb_mkstemp(tmp_file, sizeof(tmp_file), "pack/tmp_bitmap_XXXXXX"); |
| 521 | |
| 522 | if (fd < 0) |
| 523 | die_errno("unable to create '%s'", tmp_file); |
| 524 | f = sha1fd(fd, tmp_file); |
| 525 | |
| 526 | memcpy(header.magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE)); |
| 527 | header.version = htons(default_version); |
Vicent Marti | ae4f07f | 2013-12-21 09:00:45 -0500 | [diff] [blame] | 528 | header.options = htons(flags | options); |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 529 | header.entry_count = htonl(writer.selected_nr); |
Sun He | 50546b1 | 2014-03-03 17:39:59 +0800 | [diff] [blame] | 530 | hashcpy(header.checksum, writer.pack_checksum); |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 531 | |
| 532 | sha1write(f, &header, sizeof(header)); |
| 533 | dump_bitmap(f, writer.commits); |
| 534 | dump_bitmap(f, writer.trees); |
| 535 | dump_bitmap(f, writer.blobs); |
| 536 | dump_bitmap(f, writer.tags); |
| 537 | write_selected_commits_v1(f, index, index_nr); |
| 538 | |
Vicent Marti | ae4f07f | 2013-12-21 09:00:45 -0500 | [diff] [blame] | 539 | if (options & BITMAP_OPT_HASH_CACHE) |
| 540 | write_hash_cache(f, index, index_nr); |
| 541 | |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 542 | sha1close(f, NULL, CSUM_FSYNC); |
| 543 | |
| 544 | if (adjust_shared_perm(tmp_file)) |
| 545 | die_errno("unable to make temporary bitmap file readable"); |
| 546 | |
| 547 | if (rename(tmp_file, filename)) |
| 548 | die_errno("unable to rename temporary bitmap file to '%s'", filename); |
| 549 | } |